Fundamental Desgin Control Central Chilled Water Plant Ip

Fundamentals of
Design and Control of

Central Chilled-Water Plants
Steven T. Taylor, PE I-P

A Course Book for
Self-Directed or Group Learning
Includes Skill Development Exercises

for PDH or LU Credits
CHW Plants_covers_I_P.indd 1 9/14/2017 10:43:25 AM

Fundamentals of
Central Chilled-Water
Plants
Steven T. Taylor, PE
A Course Book for Self-Directed or Group Learning
Atlanta
Fundamentals of Design and Control of Central Chilled-Water Plants (I-P)
A Course Book for Self-Directed or Group Learning
ISBN 978-1-939200-66-2 (paperback)
ISBN 978-1-939200-67-9 (PDF)
SDL Course Number: 00662
2017 ASHRAE
All rights reserved.
ASHRAE is a registered trademark in the U.S. Patent and Trademark Office, owned by the American Society of Heat-
ing, Refrigerating and Air-Conditioning Engineers, Inc.
No part of this publication may be reproduced without permission in writing from ASHRAE, except by a reviewer who
may quote brief passages or reproduce illustrations in a review with appropriate credit, nor may any part of this publica-
tion be reproduced, stored in a retrieval system, or transmitted in any way or by any meanselectronic, photocopying,
recording, or otherwithout permission in writing from ASHRAE. Requests for permission should be submitted at
www.ashrae.org/permissions.
ASHRAE has compiled this publication with care, but ASHRAE has not investigated, and ASHRAE expressly disclaims
any duty to investigate, any product, service, process, procedure, design or the like that may be described herein. The
appearance of any technical data or editorial material in this publication does not constitute endorsement, warranty, or
guaranty by ASHRAE of any product, service, process, procedure, design or the like. ASHRAE does not warrant that the
information in this publication is free of errors. The entire risk of the use of any information in this publication is assumed
by the user.
ASHRAE STAFF
ASHRAE Learning Institute Special Publications Publisher
Karen Murray Mark Owen W. Stephen Comstock
Manager of Professional Editor/Group Manager of
Development Handbook and Special Publications
Sarah Boyle Cindy Sheffield Michaels
Managing Editor of Managing Editor
Professional Development Lauren Ramsdell
Kelly Arnold Assistant Editor
Professional Development Mary Bolton
Editorial Assistant
Michshell Phillips
Editorial Coordinator
For course information or to order additional materials, please contact:
ASHRAE Learning Institute Telephone: 404/636-8400

1791 Tullie Circle, NE Fax: 404/321-5478
Atlanta, GA 30329 Web: www.ashrae.org/ali
E-mail: edu@ashrae.org
Errors or omissions in the data should be brought to the attention of Special Publications via SDLcorrections@ashrae.org.
Updates and errata for this publication will be posted on the

ASHRAE website at www.ashrae.org/publicationupdates.
Your Source for HVAC&R Professional Development
1791 Tullie Circle, NE Atlanta, GA 30329-2305 Phone: 678.539.1146 Fax 678.539.2146 www.ashrae.org
Karen M. Murray kmurray@ashrae.org

Manager of Professional Development
Dear Student,
Welcome to this ASHRAE Learning Institute (ALI) self-directed or group learning course. We look forward
to working with you to help you achieve maximum results from this course.
You may take this course on a self-testing basis (no continuing education credits awarded) or on an ALI-
monitored basis with credits (PDHs or LUs) awarded. ALI staff will provide support, and you will have
access to technical experts who can answer inquiries about the course material. For questions or technical
assistance, contact us at 404-636-8400 or edu@ashrae.org.
Skill Development Exercises at the end of each chapter will gauge your comprehension of the course mate-
rial. If you take this course for credit via the ALI online-monitoring system, please complete the exercises
in the workbook then submit your answers at www.ashrae.org/sdlonline (preferred method) or email cop-
ies from each chapter to edu@ashrae.org.
To log in, please enter your student ID number and the SDL number. Your student ID number can be the
last five digits of your Social Security number or another unique five-digit number you create when first
registering online. The SDL course number is located near the top of the copyright page of this book.
Please keep copies of your completed Skill Development Exercises for your records. When you finish all
exercises, you will receive a Certificate of Completion indicating 35 PDHs/LUs of continuing education
credit. The ALI does not award partial credit for self-directed or group learning courses. All exercises must
be completed to receive full continuing education credit. You will have two years from the date of purchase
to complete each course.
We hope your educational experience is satisfying and successful.
Sincerely,
Karen M. Murray
Manager of Professional Development
Continuing Education Opportunities
from ASHRAE Learning Institute
ASHRAE Learning Institute (ALI) provides professional development through in-depth train-
ing that is timely, practical, and targeted to engineers in consulting practices, facility management,
or supplier support with instruction on applying ASHRAE standards and employing new technol-
ogies essential for advanced building performance.
HVAC Design Essentials and Applications TrainingInstructor Led at Approved Locations

Expand your knowledge and understanding of the fundamentals and technical aspects to
design and maintain HVAC systems. Level I covers essentials. Level II instructs on use of
ASHRAE Standards 55, 62.1, 90.1, and 189.1. A companion course explains improving existing
building operations. www.ashrae.org/hvactraining
Online CoursesInstructor Led on the Web

ALI offers high-quality, instructor-led online courses that allow attendees to learn from any-
where with an Internet connection. Course categories include Commissioning, Energy Efficiency,
Environmental Quality, HVAC&R Applications, and Standards and Guidelines. www.ashrae.org/
onlinecourses
ASHRAE Chapter and In-Company TrainingInstructor Led at Your Location

ALI offers a wide range of instruction that helps chapters and companies close the gap
between entry-level engineers and seasoned practitioners. ASHRAEs courses bring your team up
to speed on current standards and explain how to apply new technologies with real-world, bottom-
line emphasis. ASHRAE will arrange for an instructor to visit your location or license use of edu-
cational materials. www.ashrae.org/chaptercourses and www.ashrae.org/companycourses
eLearningWeb-Based Instruction on Demand

ASHRAE eLearning focuses on key skills and practical applications in HVAC&R and related
areas. Because it is web based, students can train from any computer with Internet access. This
makes it ideal for both individual and corporate training. www.ashrae.org/elearning
Self-Directed Learning TextsSelf Study or Texts for Group Instruction

For those seeking traditional book-based instruction, ASHRAE offers Learning Texts for self-
study or group training with instructor materials. Texts cover the basics of what a practicing engi-
neer needs for real-world HVAC&R applications. Skill Development Exercises are included to
evaluate progress. Students receive a course completion certificate designating continuing educa-
tion credits. www.ashrae.org/sdl
ASHRAE Learning Institute www.ashrae.org/education

Steven T. Taylor, PE, Fellow ASHRAE, is the founding principal of
Taylor Engineering, Alameda, CA. He is a registered mechanical engi-
neer specializing in HVAC system design, control system design, indoor
air quality engineering, computerized building energy analysis, and
HVAC system commissioning. Mr. Taylor graduated from Stanford Uni-
versity with a BS in Physics and a MS in Mechanical Engineering and
has 40 years of commercial HVAC system design and construction
experience. He was one of the primary authors of the HVAC sections of
ASHRAE Standard 90.1, Energy Standard for Buildings Except Low-
Rise Residential Buildings and Californias Title 24 energy standards
and ventilation standards. Other ASHRAE projects and technical com-
mittees Mr. Taylor has participated in include ASHRAE Standard 62.1
on indoor air quality (chair), ASHRAE Standard 55 on thermal comfort
(member), Guideline 13 on specifying DDC (chair), Guideline 16 on
economizer dampers (chair), Guideline 36 on advanced control
sequences (founder and member), the TC 1.4 on controls (chair), and the
TC 4.3 on ventilation (chair). He is past vice-chair of the U.S. Green
Building Council (USGBC) Leadership in Energy and Environmental
Design (LEED) Indoor Environmental Quality Technical Advisory
Group, a member of the CSU Mechanical Review Board, and a 16-year
member of the International Association of Plumbing and Mechanical
Officials (IAPMO) Mechanical Technical Committee administering the
Uniform Mechanical Code.
Contents
Preface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ix
Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .xii
Acronyms. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xv
Chapter 1: Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
How to Use This Course . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
Organization of Material . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
Chapter 2: Chilled-Water Plant Loads. . . . . . . . . . . . . . . . . . . . . . . . . 3
Instructions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
Understanding Loads and Their Impact on Design . . . . . . . . . . . 3
Determining Peak Loads . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
Determining Hourly Load Profiles . . . . . . . . . . . . . . . . . . . . . . 11
References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
Skill Development Exercises for Chapter 2 . . . . . . . . . . . . . . . . . . . 14
Chapter 3: Chilled-Water Plant Equipment. . . . . . . . . . . . . . . . . . . . 17
Instructions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
Water Chillers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
Water Chiller Components . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
Heat Rejection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
Pumps . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
Variable-Frequency Drives (VFDs) . . . . . . . . . . . . . . . . . . . . . . 71
References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77
Skill Development Exercises for Chapter 3 . . . . . . . . . . . . . . . . . . . 79
Chapter 4: Hydronic Distribution Systems . . . . . . . . . . . . . . . . . . . . 81
Instructions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
Chilled-Water Distribution Systems . . . . . . . . . . . . . . . . . . . . . 81
Condenser Water Systems . . . . . . . . . . . . . . . . . . . . . . . . . . 119
Plant Layout . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 133
References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 135
Skill Development Exercises for Chapter 4 . . . . . . . . . . . . . . . . . . 138
Chapter 5: Optimizing Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141
Instructions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141
Design Procedure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 142
Selecting Chilled-Water Distribution System Flow Arrangement. . 143
Optimizing Piping Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 154
Optimizing Chilled-Water Design Temperatures . . . . . . . . . . . . . 159
Optimizing Condenser Water Design Temperatures . . . . . . . . . . 164
viii Contents
Selecting Cooling Towers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .167

Water-Side Economizers (WSEs) . . . . . . . . . . . . . . . . . . . . . . . . .173
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .178
Skill Development Exercises for Chapter 5. . . . . . . . . . . . . . . . . . 179
Chapter 6: Chiller Procurement . . . . . . . . . . . . . . . . . . . . . . . . . . . 181
Instructions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 181
Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 181
Chiller Procurement Procedures . . . . . . . . . . . . . . . . . . . . . . 181
Case Study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 199
Simplified Procurement Procedure . . . . . . . . . . . . . . . . . . . . . 201
References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 204
Chapter 7: Controls. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 207
Instructions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 207
Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 207
Sensors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 209
Control Valves . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 226
Controllers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 230
Network Interfaces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 231
Performance Monitoring . . . . . . . . . . . . . . . . . . . . . . . . . . . . 233
Control Schematics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 235
Control Sequences . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 239
Appendix ATOPP Model Coefficients. . . . . . . . . . . . . . . . . . . . 264
Appendix BDetailed Sequence of Operation (SOO) . . . . . . . . . 268
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 277
Chapter 8: Commissioning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 283
Instructions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 283
Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 283
Commissioning Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . 283
Commissioning Focus . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 284
Sequence of Operation (SOO) Review . . . . . . . . . . . . . . . . . 285
Point-to-Point Checkout . . . . . . . . . . . . . . . . . . . . . . . . . . . . 286
Functional Testing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 288
Trend Review . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 290
References. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 292
Online Supplemental Files: This SDL is accompanied by Excel spreadsheets, which
can be found at ashrae.org/CHWSDL. These files include a chiller bid form, a simpli-
fied chiller bid form, and a pipe size optimization tool. If the files or information at the
link are not accessible, please contact the publisher.
Preface
Chilled-water plants are typically the most costly part of large

building or campus HVAC systems and the largest energy user. Opti-
mizing the design and control of chilled water plants can therefore have
a large reduction in HVAC system life-cycle costs. But true optimiza-
tion requires extensive analysis, for which few system designers have
the time or funding. This course is intended to improve on the state of
the art by providing updated design techniques based on rigorous life-
cycle cost analysis that can provide near-optimum chilled-water plant
life-cycle costs with little or no more engineering time than current
practice. Recommended control sequences, also based on rigorous
analysis, can improve plant performance with no more complexity than
typical current practice. In addition to these design techniques, this
course includes practical tips for laying out and piping chilled-water
plants. The guidance applies to small plants serving small buildings as
well as to district cooling plants.
This SDL is accompanied by Microsoft Excel spreadsheets,
which can be found at ashrae.org/CHWSDL. These files include a
chiller bid form, a simplified chiller bid form, and a pipe size optimiza-
tion tool. If the files or information at the link are not accessible, please
contact the publisher.
Acknowledgments
The author would like to thank Pacific Gas and Electric Company for
allowing ASHRAE to use its 1999 CoolTools Chilled Water Plant
Design Guide as the basis of this course. The CoolTools guide was
co-authored by Steve Taylor (author of this course), Mark Hydeman,
Paul DuPont, and Tom Hartman.
Others who provided review and input to this course include:
Brandon Gill, Taylor Engineering
Mick Schwedler, Trane
Steve Duda, Ross & Baruzzini
Bryson Borzini, P2S Engineering
Tony Mueller, P2S Engineering
Anna Zhou, Taylor Engineering
Steven T. Taylor, PE
Taylor Engineering
September 6, 2017
Fundamentals of
Central Chilled-Water
Plants
Acronyms
A/D = analog to digital

AFLV = automatic flow-limiting valve
AHU = air-handling unit
ATS = automatic transfer switch
BAS = building automation system
BEP = best efficiency point
CAD = computer-aided design and drafting
CBV = calibrated balancing valve
CFC = chlorofluorocarbon
CHW = chilled water
CHWST = chilled-water supply temperature
COP = coefficient of performance
COV = change of value
CS = constant speed
CT = current transformer
CVRMSE = coefficient of variation of root mean squared error
CW = condenser water
CWFd = design CW flow rate
CWFR = condenser water flow ratio
CWFSP = condenser water flow set point
CWRT = condenser return temperature
Cx = commissioning
CxA = commissioning authority
D/A = digital to analog
DDC = direct digital control
DI = digital input
DO = digital output
DP = differential pressure
DV/DT = derivative of voltage with respect to time
DX = direct expansion
EMI = electromagnetic interference
EMT = electrical metallic tubing
EOR = engineer of record
EPDM = ethylene propylene diene monomer
FCU = fan-coil unit
xvi Acronyms
GWP = global warming potential

HBM = heat balance method
HCFC = hydrochlorofluorocarbon
HFC = hydrofluorocarbon
HFO = hydrofluoroolefin
HGBP = hot-gas bypass
HOA = hand-off-auto
HX = heat exchanger
HXLWT = heat exchanger leaving water temperature
I/O = input/output
IPLV = integrated part-load value
LCCA = life-cycle cost analysis
LOT = lockout temperature
MBE = mean bias error
NPLV = nonstandard part-load value
NPSHA = net positive suction head available
NPSHR = net positive suction head required
OAT = outdoor air temperature
ODP = ozone depletion potential
OEM = original equipment manufacturer
PHXLWT = predicted heat exchanger leaving water temperature
PICCV = pressure-independent characterized control valve
PID = proportional integral differential
PLC = programmable logic controller
PLR = part-load ratio
PVC = polyvinyl chloride
PWM = pulse width modulation
RFI = radio frequency interference
RMS = root mean square
RTD = resistance temperature detector
RTS = radiant time series
SAT = supply air temperature
SOO = sequence of operation
SPLR = staging part-load ratio
TAB = testing, adjusting, and balancing
TDH = total dynamic head
TES = thermal energy storage
THHN = thermoplastic high heat resistant nylon coated
TOPP = theoretical optimum plant performance
TXV = thermal expansion valve
UPS = uninterruptible power source
VAV = variable air volume
VFD = variable-frequency drive
Fundamentals of Design and Control of Central Chilled-Water Plants I-P xvii
VS = variable speed
VSD = variable-speed drive
WSE = water-side economizer
XHHW-2 = cross-linked polyethylene high heat-resistant water-resistant
XLPE = cross-linked polyethylene
Overview
How to Use This Course

The purpose of this course is to provide guidance to designers and operators of
new and existing central chilled-water (CHW) plants ranging from small, single-
chiller plants to large, district-cooling plants. While design engineers are the pri-
mary audience, the guide also provides useful information for operation and main-
tenance personnel, mechanical contractors, and building managers. Upon
completion of this course, the student should have a thorough understanding of
CHW plant fundamentals and principles that will be useful in conjunction with
plant design or operation.
The course is divided into chapters, each addressing a specific topic. It is
important that you understand each topic before going on. At the end of each
chapter there are questions that are intended to reinforce certain topics and to
test your level of understanding. Your responses should be given to ASHRAE
at www.ashrae.org/sdlonline in order to receive credit and to obtain the answer
sheets.
Introduction
Many large buildings, campuses, and other facilities have plants that make
chilled water and distribute it to air-handling units (AHUs) and other cooling
equipment. The design, operation, and maintenance of these CHW plants has a
very large impact on building energy use and energy operating cost. The intent
of this course is to provide tools and guidance to engineers so that the plants
they design have a near optimum balance of first costs and future operating
costs. The course can also be used by plant operators to understand and resolve
operational problems and improve energy efficiency through controls optimi-
zation.
Organization of Material
The course is organized in eight chapters. The first chapter is this overview.
Loads. Chapter 2 discusses the nature of CHW loads and how they should
be considered in the design of CHW plants. In the past, most engineers have
only estimated the peak or maximum load. However, accounting for the time
pattern of loads can be just as important. Methods of calculating peak loads
2 Chapter 1 Overview
and hourly loads are reviewed. These include site measurements (for existing
facilities), computer simulations, rules of thumb, and prototype buildings.
Equipment. Chapter 3 reviews some basics on chillers, cooling towers,
pumps, and other plant equipment. This chapter discusses the basic refrigera-
tion cycle, water chillers, cooling towers, air-cooled condensers, pumps, and
variable-speed drives.
Distribution Systems. Chapter 4 discusses different ways of arranging
CHW equipment in the system to meet loads while achieving energy efficiency
and operational simplicity. The pros and cons of constant-flow and variable-
flow systems are discussed along with different primary-only and primary/sec-
ondary pumping systems.
Optimizing Design. Chapter 5 provides procedures and analysis tech-
niques for optimizing CHW plant design. Topics include optimizing the selec-
tion of distribution systems and optimizing the selection of CHW and
condenser water design temperatures and pipe sizes. A spreadsheet for sizing
piping and calculating pump head is provided at ashrae.org/CHWSDL (Pipe
Size Optimization Tool spreadsheet). Recommendations were developed from
in-depth life-cycle cost analysis of typical chiller plants and are provided as
easy-to-use rules of thumb and procedures to simplify plant design while still
achieving near-optimum life-cycle performance.
Chiller Procurement. Chapter 6 discusses strategies for evaluating chiller
options and selecting and procuring an energy-efficient and cost-effective
chiller. Case studies of the chiller selection process are provided. Sample
chiller bid forms are also provided (Chiller Bid Form and Simplified Chiller
Bid Form).
Controls. Chapter 7 explores the many design and performance issues
related to controls and instrumentation of CHW plants. Topics include types of
flow and temperature sensors, styles of and selection criteria for control valves,
controller requirements and interfacing issues, performance monitoring, and
recommended near-optimum control sequences for CHW plants, including all-
variable-speed plants where all components have variable-speed drives.
Commissioning. Chapter 8 discusses key elements of the commissioning
process, addressing in detail sequence of operation review, point-to-point
checkout, functional testing, and trend reviews.
Supplemental Files. Supplemental material for this SDL is available at
ashrae.org/CHWSDL. These files include a chiller bid form, a simplified
chiller bid form, and a pipe size optimization tool.
Chilled-Water Plant
Loads
Instructions
Read the material in Chapter 2. Verify the examples presented in the chapter
with your own calculations. At the end of the chapter, complete the Skill
Development Exercises without referring to the text. Review those sections of
the chapter as needed to complete the exercises.
Introduction
This chapter discusses CHW plant peak loads and annual cooling load pro-
files and how they affect plant design and equipment capacity. Fundamental to
the design process is a keen understanding of the chiller plant cooling loads
and how they vary with time. If an existing plant is being modified or
expanded, it is possible to monitor the cooling load and obtain an accurate esti-
mate of both the peak load and the cooling load profile. A great many plants,
however, are designed with only preliminary information available about the
buildings design and function. Getting accurate peak load and cooling load
profile information for these plants is much more difficult. This chapter dis-
cusses the uncertainties involved with predicting chiller plant loads and the
impact of these uncertainties on the design process.
Understanding Loads and Their Impact on Design

To provide an optimum CHW plant design, the designer must determine both
a design (peak) load and a cooling load profile that describes how the load varies
over time. The design load defines the overall installed plant capacity including
the chillers, pumps, piping, and towers. The cooling load profile is required to
design the plant to handle often widely variable loads stably and efficiently. This
includes design decisions such as the unloading mechanisms of the chillers; the
application of variable-frequency drives (VFDs) on the chillers, towers, and
pumps; and the relative sizes of each piece of equipment.
Certain key load parameters affect the cooling load profile and conse-
quently the nature of the plant design. These parameters include the following:
The use of outdoor air economizers and 100% outdoor air units
4 Chapter 2 Chilled-Water Plant Loads
The climate in which the plant is located

Hours of building or facility operation
Base (24/7) loads such as computer rooms
For example, the cooling load profile of a San Francisco office building
that operates five days per week was analyzed with and without economizers.
As Figure 2-1 shows, the number of hours that the plant operates increases dra-
matically when an economizer is not used. Additionally, the shape of the pro-
file changes dramatically. The profile influences the optimum selection of the
number and capacity of the chillers as well as the full-load and part-load
energy efficiency of the machines.
What happens if this same CHW plant serving the office with economizers
must also serve a relatively small data center without an economizer? Figure 2-
2 shows the resulting load profile. This profile is also typical of district and
campus cooling plants that serve relatively small 24/7 loads continuously,
along with much larger peak summer loads. The plant clearly will need to
operate efficiently at low loads. The plant must be designed very differently
than the one serving the office building alone.
Figure 2-1 Cooling load profiles, five-day office in San Francisco.

Fundamentals of Design and Control of Central Chilled-Water Plants I-P 5
Figure 2-2 Cooling load profile of CHW plant serving office building plus a small data center.
Peak Loads Overview

The process for estimating peak cooling loads in new construction is
explained thoroughly in 2017 ASHRAE HandbookFundamentals
(Chapter 18). The basic variables for peak load calculations include weather
conditions, building envelope design, internal heat gain, ventilation, and, to a
lesser extent, infiltration.
Less obvious but nonetheless important is the diversity between the various
load components. The diversity of loads is the probability of simultaneous
occurrence of dynamic peak loads. In other words, diversity accounts for the
fact that the envelope, occupancy, lighting, and plug loads will not each peak at
the same time in all spaces simultaneously. Diversity is often underestimated
by designers, particularly for large central plants. It is not uncommon for peak
central plant loads to be less than half of the connected building design peak
loads.
There is also inherent uncertainty in peak load calculations. Any number of
elements can make the actual load differ from the calculated load. For instance,
the following may occur:
Weather conditions can vary over a period of time as a result of increasing

urbanization, climate change, and changes in land use.
Building envelope elements do not always perform as expected due to

issues such as thermal short-circuits of structural members and poor air
barriers, among others.
Changes may occur in operation (such as tenants moving into or out of the
building).
Internal loads (lighting, plug loads, and people) can be significantly differ-
ent from those estimated in load calculations and can vary over time.
Often the characteristics of the loads served are not clear at the time of the
plant design. This is often the case with district or campus systems where the
designer must essentially guess at system and infrastructure capacity to support
future growth. Simulation tools (discussed in the Determining Peak Loads and
Determining Hourly Load Profiles sections) and budgets based on measured
existing buildings usage can be quite helpful.
A plant expansion or remodel provides the opportunity to monitor the
existing plant for peak and operating loads. Most building automation systems
(BASs) have the capability of supporting trend logs. Of course, the plant must
also be provided with instrumentation (such as flowmeters and temperature
sensors) to provide useful load information. Also, a good operator can often
accurately report on the percent of full load that the plant sees during peak
weather conditions.
For most designers the perceived risks of understating the peak load condi-
tion (and undersizing the cooling plant) are much greater than overstating the
peak load. An undersized cooling plant may not meet the owners expectations
for comfort and may affect the owners ability to manufacture products or pro-
vide essential services. Oversizing the cooling plant, on the other hand, carries
an incremental first-cost penalty and can have a positive or negative energy
impact depending on the piece of equipment and how it is controlled. Over-
sized cooling towers and pipes tend to reduce the energy costs of operating the
plant. Oversized pumps and chillers often run inefficiently at low loads,
although the use of variable-frequency drives (VFDs) mitigates this to a great
extent. Because oversizing always carries a first-cost premium, it is prudent to
not oversize plants. Where actual loads, future growth, or diversity are uncer-
tain, starting small with provisions (space and piping manifold sizes) for the
addition of future pumps, towers, and chillers is advisable.
Annual Load Profiles Overview

A cooling load profile is a time series of cooling plant loads along with
concurrent weather data. The primary role of a cooling load profile is to facili-
tate the correct relative evaluation of competing design options.
An accurate understanding of the cooling load profile affects the plant con-
figuration. For example, a plant that serves a hotel complex with long periods
of very low loads would be designed differently than a plant that serves widely
varying loads only in mild and warm weather during the daytime, such as a
plant serving an office building.
If the actual cooling loads are closely related to weather data, then tempera-
ture bin estimating techniques may produce satisfactory analysis results. How-
ever, in most cases the load is not strongly correlated with outdoor air
temperature due to the predominance of internal and solar loads, which are not
dependent on outdoor air temperature and humidity. Using bin weather data
alone for optimization calculations will seldom provide the accuracy needed
for a truly optimized plant. Therefore, to accurately address the impact of the
expected load profile in a chiller plants design, it is necessary to have hourly
load data for an entire typical year.
As noted above, the designers ability to accurately project hourly load pro-
files into the future includes significant uncertainty. Designers should address
uncertainty in the development of the annual load profiles. One approach is to
consider multiple load profiles, representing a reasonable range of changes in
operating conditions, when designing a plant.
Oversizing/Undersizing Considerations
Because of the uncertainty inherent in design parameters and the risks
associated with undersizing the plant, most CHW plants are larger than needed
to meet maximum load conditions. Some impacts of oversizing a CHW plant
are as follows:
Oversized plants always cost more to build. While a plants cost may not
vary linearly with its total capacity, larger plants have more expensive
chillers, larger pumps, and possibly larger piping.
When operating at part loads, an oversized fixed-speed chiller may not per-
form as efficiently as a smaller machine. Conversely, a variable-speed
chiller at part load and reduced lift may operate more efficiently than a
smaller machine at full load.
Oversized chillers have larger CHW and condenser water (CW) pumps that
consume more energy if the pumps are constant speed. This penalty can be
significantly reduced if the pumps have VFDs or if the CHW plant consists
of multiple smaller pumps.
Oversized chillers can result in greater wear and tear and greater fluctuations
in CHW supply temperature because chillers can only turn down so much
before they must cycle off their compressors and then wait to restart them.
The larger piping in an oversized plant will have less pressure drop than
that of a plant whose piping is rightsized. Rightsized piping will reduce
pumping energy if pumps have VFDs. For campuses where future loads are
extremely uncertain, oversizing piping is usually a very good investment.
An oversized plants cooling towers may save energy by allowing the tower
fans to run slower if fans have VFDs. Also, they may produce lower CW
temperatures for more efficient part-load operation of the chillers. Con-
versely, oversized cooling towers may have flow turndown problems that
force the operators to use fewer cells at higher fan speeds, which can
increase plant energy use.
The owners criteria may call for incorporating redundant chillers, pumps,
and other equipment to reduce exposure to equipment failure. Redundant or
spare equipment is a separate issue from oversizing, because it does not reduce
the ability of the plant to adjust capacity to match the load.
To mitigate problems with oversizing, a CHW plant must run efficiently at
low loads. Chapter 5 discusses strategies for achieving optimum selection of
chiller configurations. The following example from a computer simulation
model helps demonstrate the issue of oversizing. In this case, an 800 ton cool-
ing plant serves an office complex that operates on a basic five days per week
schedule. Typical load profiles were scaled for peak cooling load of exactly
450 tons. The plant was modeled with the following scenarios:
A single 800 ton centrifugal chiller with inlet vane control
The same 800 ton centrifugal chiller with VFDs
Two 400 ton centrifugal chillers with inlet vane control
The same two 400 ton centrifugal chillers each with VFDs
Figure 2-3 shows the results of this simulation. Note the dramatic reduction
in annual cooling energy consumption when the VFD is added to the 800 ton
machine and also when multiple machines are used.
Although other scenarios may produce similar or better results, this exam-
ple illustrates that the energy penalty for an oversized plant can be dramatically
reduced if efficient turndown is incorporated into the design. By either adding
a VFD on a single chiller or providing two smaller fixed-speed chillers, the
annual energy is reduced by approximately one third. Combining these mea-
sures (two chillers with VFDs) reduces the annual energy by nearly one half.
Figure 2-3 Cooling energy usage for four design alternatives.

The following is some guidance with respect to loads and their impact on
system design:
Avoid serving small 24/7 loads without economizers, such as a server room
fan-coil, with a system that primarily serves large intermittent loads (such
as office building air handlers with outdoor air economizers). The plant
would have to run at very low loads, which is not only inefficient but can
also decrease equipment life due to cycling. These auxiliary loads may be
better served by separate cooling systems, such as direct-expansion (DX)
units.
If many hours at low loads are unavoidable, provide multiple chillers and
perhaps unequally sized chillers to improve low-load performance.
Where a wide range of loads can be expected (which is true of most plants,
even those serving typical data centers), use VFDs on all equipment (all
pumps, tower fans, and chillers), creating what is called an all-variable-speed
plant. As discussed in Chapter 5, VFDs are almost always cost-effective on
all equipment (with the possible exception of CW pumps), and they substan-
tially mitigate the energy impact of oversized plants.
The benefits of rightsizing or tightsizing (sizing the plant precisely for
the expected loads) are often overstated, particularly for multi-chiller all-
variable-speed plants, which can operate efficiently over a wide range of
loads. There are also disadvantages to tightsizing, such as having to retro-
fit additional or larger equipment at extremely high cost if any of the
many assumptions made about future loads are wrong. Owners expect
and deserve flexibility to handle future loads within reason, so aggressive
sizing may not be the best approach despite some first-cost and efficiency
benefits.
Determining Peak Loads

Calculations/Simulations
ASHRAE HandbookFundamentals, Chapter 18, defines accepted meth-
ods and procedures for cooling load calculations. These well-known proce-
dures include information on ventilation and infiltration, climatic design
information, residential and nonresidential load calculations, fenestration, and
energy estimating methods. In discussing cooling load principles, the Hand-
book emphasizes the importance of analyzing each variable that may affect
cooling load calculations:
The variables affecting cooling load calculations are numerous, often
difficult to define precisely, and always intricately interrelated. Many
cooling load components vary in magnitude over a wide range during
a 24-h period. Because these cyclic changes in load components are
often not in phase with each other, each must be analyzed to establish
the resultant maximum cooling load for a building or zone. (18.1)
Starting in the 2001 edition, the Handbook supports only two methods of
load calculation: the heat balance method (HBM) (a fundamental first-principles
approach) and the radiant time series (RTS) method (an approximation of the
heat balance method). For all practical purposes, both of these methods require
computer simulation to analyze. Although these calculation techniques have
worked very well over the years, designers must be aware of the limitations of
these techniques and recognize that the methods do not all predict the same
loads. Because of the uncertainties previously discussed, the design load calcula-
tions may be different than the actual chiller plant peak load. Selecting the maxi-
mum capacity of the plant is important, but it is perhaps even more important to
consider the plants part-load performance.
Site Measurements
When an existing chiller plant is being remodeled or expanded, it is possi-
ble to monitor the actual peak cooling load to obtain invaluable information.
The monitoring can be short term (several months) to establish peak load and
daily trends or can be long term (one year or longer) to determine annual load
profiles. Successfully measuring energy and load performance of a cooling
plant requires rigorous monitoring protocols (see ASHRAE Guideline 22,
Instrumentation for Monitoring Central Chilled-Water Plant Efficiency [2012],
for example). These monitoring protocols comprise four stages:
1. Survey of monitoring sites: Conduct a complete audit of the CHW plant.

Develop a comprehensive systems diagram.
2. Monitoring plan: From the comprehensive systems diagrams prepare a plan
for determining the data to be monitored, the monitoring equipment
needed, and the duration of monitoring. Typical monitoring equipment
includes data loggers, flow measurement devices, temperature measure-
ment devices, and power measurement devices. Also required are concur-
rent measurements of weather data, including dry-bulb temperature and
wet-bulb temperature. In many modern plants, the necessary instrumenta-
tion for measuring and trending load and weather is permanently installed
as part of the plants controls system infrastructure, obviating the need for
additional short-term sensors and data loggers. Weather data may also be
available online from nearby government weather stations.
3. Field installation: Install instrumentation in accordance with the monitor-
ing plan and the installation instructions. Take spot measurements to ensure
that the equipment is calibrated properly and that all sensors and instru-
ments are working correctly. Provide guidelines to operators. Have a plan
for removal of instrumentation and patching of insulation, etc.
4. Data collection and analysis: Obtain data and provide validation. Perform
analysis on both a basic level (for example, simple temperature logs of
chiller energy usage) and a more detailed level (for example, chiller plant
energy performance as a function of various elements such as time and
weather). If the weather in the monitoring period does not reach the design
weather conditions, then it will be necessary to extrapolate the measured

load as a function of weather to determine the peak load. This should be
done with care as load curves regressed from available data may not be an
accurate representation of real-world loads when extrapolated to tempera-
tures well outside the monitoring range.
From this procedure, the peak loads will emerge, as well as the relation-
ships and interactions of the various components. The quality of the monitor-
ing protocol will determine the accuracy and usefulness of the results.
Determining Hourly Load Profiles

There are several methods for determining annual cooling load profiles
depending on what stage the project is in and the resources available for analy-
sis. The following are common methods for determining annual cooling load
profiles:
Computer simulation models (customized and prototypical)

Site measurements
For new construction, custom computer simulations and prototypical simu-

lations are the two most common methods. Customized simulations have the
greatest potential for accuracy but can be costly to develop and are subject to
modeling error. Prototypical simulations offer quick and relatively inexpensive
analysis but may not be as accurate as customized simulations.
For retrofit and expansion of existing plants, site visits may be conducted to
measure profiles. This technique yields the most accurate results but requires
special planning, technical expertise, equipment, budget, and time.
Each of these methods can be combined with statistical and mathematical
techniques from a variety of sources including short-term measurements, site
data, and billing data. These hybrid approaches offer the best possibility to bal-
ance accuracy and effort. The following sections discuss each technique.
Computer Simulation Models

Computer simulation models customized for a specific project can take
between a few hours to several person-weeks of time to develop, depending on
the complexity of the building geometry and the effort spent on making the
model accurate. With recent advances in simulation tool data exchange, the
effort to build these models has significantly decreased. For example, building
geometry can be imported from computer-aided design and drafting (CAD)
programs into some load or simulation tools. Examples include EnergyPlus
(which supports both IAI IFCs and GBXML), DOE2, and several commercial
load and energy programs that support GBXML. Although these tools are far
from plug and play they still dramatically reduce the time required to create
models, and they reduce modeling errors.
For projects that are early in design and evaluations for campus systems,
prototypical models are a useful tool. Many of the simulation tools now incor-
porate wizards that enable designers to develop a typical building for analysis
in a matter of minutes. Examples include eQUEST, a free, front-end interface
to DOE2.2 and DOE2.3.
Computer simulation models require experienced modelers for inputting
data and checking results. To assess the impact of uncertainties, the modeler
should consider a range of input variations representing the best estimate, pos-
sible but likely low loads, and possible but likely high loads.
Site Measurements
Site monitoring to determine peak loads was discussed earlier in this chap-
ter. The same site monitoring protocol can be used for determining cooling
load profiles based on either short-term or long-term measurements. Long-
term monitoring is not common because it is costly and time-consuming to
obtain the data. Long-term trend data of plant performance are sometimes
available from BAS trends, but the data are often inaccurate or incomplete.
Experience with long-term data indicates that due to weather and other
variables, a single years measurement would not match the second years data
and as a result is not deterministically exact. The utility of long-term monitor-
ing is maximized by ensuring that the monitoring period captures the full range
of anticipated weather conditions (often necessitating four to six months of
data centered on a swing season, depending on climate zone) and all unique
seasonal operating profiles. For example, a college central plant will have
unique load profiles during in-session and out-of-session periods that must
both be monitored. If these weather and schedule range criteria are met, then
the data can used to create a robust regression model accounting for seasonal
factors, day type (weekday, weekend, holiday, etc.), time of day, and ambient
weather conditions. The resulting model can then be applied to a prototypical
weather year to generate an expected annual load profile for the plant. If the
weather and schedule criteria are not met, then the model runs the risk of gen-
erating invalid results when extrapolated outside of the conditions observed
during the monitoring period.
When long-term data are not available, or are not practical to capture,
short-term data can potentially be used to define the basic shape of a typical
24-hour load profile by season or month. However, such data are climate sensi-
tive and the associated weather/load profiles are difficult to record, especially
considering the solar aspect of the load.
When a few weeks of continuous short-term load and weather data are
available, but are insufficient to generate a robust time- and temperature-
dependent regression model as discussed above, these data may instead be used
to calibrate a computer simulation model. This hybrid modeling and site mea-
surement approach is fairly laborious. The modeler must use the weather data
collected during the short-term monitoring period to create a custom simula-
tion weather file for the site corresponding to the monitoring period. The simu-
lation model is then run with that weather file and calibrated to approximate
the site load data over the monitoring period by adjusting schedule and load
profile variables. The validity of such calibrations is measured using statistical
metrics including mean bias error (MBE) and coefficient of variation of root
mean squared error (CVRMSE). Refer to International Performance Measure-
ment & Verification Protocol (EVO 2002) for further discussion of these cali-
bration metrics.
References
ASHRAE. 2012. ASHRAE Guideline 22, Instrumentation for monitoring cen-
tral chilled-water plant efficiency. Atlanta: ASHRAE.
ASHRAE. 2017. Chapter 18, ASHRAE HandbookFundamentals. Atlanta:
ASHRAE.
EVO. 2002. International Performance Measurement & Verification Protocol.
Washington, DC: Efficiency Valuation Organization.
14 Fundamentals of Design and Control of Central Chilled-Water Plants I-P
Skill Development Exercises for Chapter 2

2-1 Why is the shape of a CHW plants cooling load profile a critical factor in plant
design?
a. It dictates the conditions under which the plant must operate efficiently
to minimize energy costs.
b. It impacts the selection of chillers because the plant must be able to
handle the full range of expected load conditions stably.
c. It drives the peak capacity required of the plant.
d. Both (a) and (b).
e. All of the above.
2-2 Which of the following are true regarding the impact of air-side economizing
on the annual load profile of a plant serving an office building?
i. It reduces the total annual ton hours served by the plant.
ii. It shifts the most common load percentage to a lower value.
iii. It reduces the peak load of the plant.
iv. It reduces the plants run hours.
a. (i), (ii), (iv)
b. (i), (ii)
c. (i), (iii), (iv)
d. (i), (ii), (iii), (iv)
2-3 ASHRAE HandbookFundamentals supports which of the following load cal-
culations methodologies?
a. RTS
b. HBM
c. Transfer function method
d. Only (a) and (b)
e. All of the above
2-4 Oversizing CHW plants
a. Typically yields more efficient pumping in variable-speed applications
due to lower friction losses.
b. Usually leads to more efficient chiller operation.
c. May cause controllability issues if chillers are not properly selected for
stable low-load operation.
d. Is problematic when the condenser and CHW pumps are variable
speed.
e. Both (a) and (c).
2-5 You are replacing oversized chillers in an existing CHW plant with modern
direct digital control (DDC) controls, trending capabilities, and recently cali-
brated instrumentation. Which of the following is the recommended approach
for determining peak load to size the new chillers?
a. Develop a load model of the facility using a simulation tool and utilize
the peak load estimated therefrom.
b. Develop a load model of the facility using a simulation tool and cali-
brate the model on an annual basis using utility billing data, then assess
peak load with the model.
c. Utilize the DDC systems primary CHW loop flowmeter and supply-
and return-temperature sensors to trend load. Use a few months of
trended load and local weather data from the summer and/or swing sea-
sons to develop a load profile and predict peak load therefrom.
d. Install temporary National Institute of Standards and Technology
(NIST)-calibrated instrumentation, including an ultrasonic flowmeter
and supply- and return-temperature sensors to trend load. Use the same
approach as option (c) to predict peak load.
2-6 True or False: In early design, developing a prototypical model of the proposed
building is usually too cost prohibitive to assist in plant design development.
a. True
b. False
Chilled-Water Plant
Equipment
Instructions
Introduction
Design engineers seeking to maximize the performance and economic ben-
efits of upgraded or new CHW plants need a thorough understanding of the
major equipment used in these plants. This chapter provides an overview of the
primary equipment, as well as essential information on how the components
relate to one another, how they are controlled, and what their physical and
operational limitations are. This chapter discusses the following:
The basic vapor compression refrigeration cycle

The components commonly used in commercial water chillers
Methods of heat rejection, such as cooling towers and air-cooled refrigerant
condensers
The characteristics of different types of pumps, pump and system curves
The application and efficiency of VFDs
The intent is to familiarize the reader with basic components. For addi-
tional and more in-depth information, consult with equipment manufacturers,
references such as ASHRAE HandbookHVAC Systems and Equipment
(2016d), and other ASHRAE self-directed learning courses such as Fundamen-
tals of Water System Design (2015).
Water Chillers
This section presents an overview of the current water chiller technologies.
Technology changes rapidly, so students are encouraged to browse manufactur-
ers websites for the most current information on technologies and refrigerants.
18 Chapter 3 Chilled-Water Plant Equipment
Vapor Compression Refrigeration Cycle

The vapor compression refrigeration cycle is the fundamental thermody-
namic basis for removing heat from buildings and rejecting it to the outdoors.
(Absorption chillers, which use a completely different technology, are dis-
cussed in the Absorption Chillers section.) The refrigeration cycle requires
four basic components:
Compressor
Evaporator
Condenser
Expansion device
The vapor compression refrigeration cycle diagram (Figure 3-1) shows the
relationship of these components, as does the pressure-enthalpy chart, also
known as a P-h diagram (Figure 3-2). These diagrams cover the liquid-vapor
regions specific to the cycle refrigerant. The following is a description of the
refrigeration cycle using the points noted on Figure 3-2:
Starting at Point A, the refrigerant is a liquid at high pressure. As it passes

through the expansion device to Point B, the pressure drops. At Point B the
refrigerant is a mixture of liquid and gas. At this point the gas is called flash
gas. Alternatively, the liquid could be subcooled to Point A , which is
below the saturation temperature. If this is done, the liquid would pass
through the expansion device, resulting in less flash gas present at Point B .
Figure 3-1 The refrigeration cycle.

Figure 3-2 Pressure-enthalpy chart.
From Point B to Point D, the liquid is converted to a gas by absorbing heat

(refrigeration effect). Notice the gas leaving the evaporator at Point D has
been heated to a level greater than saturation as shown by Point C. The heat
from Point C to D is called superheat. Superheating in the evaporator ensures
that there is no liquid in the refrigerant as it moves into the compressor.
From Point D the refrigerant is drawn into the suction of the compressor
where the gas is compressed, as shown by Point E. At Point E, the tempera-
ture and pressure of the gas have been increased. The refrigerant is now
called hot gas. Notice that this point is to the right of the saturation curve,
which also represents a superheated state.
The hot gas, Point E, moves into the condenser where the condensing
medium (either air or water) absorbs heat and changes the refrigerant from
a gas back to a liquid as shown by Point A. At Point A the liquid is at an
elevated temperature and pressure. The liquid is forced through the liquid
line to the throttling device and the cycle is repeated.
The difference between the condensing temperature and evaporating tem-

perature is called the lift. The lift is a primary driver of the efficiency of the
chiller, discussed in the following sections, including Water Chiller Components.
Refrigerants
To address safety and environmental concerns, refrigerants must have low
toxicity, low flammability, and a long atmospheric life. They also must have
zero or minimal impact on stratospheric ozone and on global warming via
greenhouse effects.
The relative ability of a refrigerant to destroy stratospheric ozone is called

its ozone depletion potential (ODP). Older refrigerantsparticularly chloro-
fluorocarbons (CFCs)are known to destroy stratospheric ozone; CFCs have
been phased out according to the 1987 Montreal Protocol (see Table 3-1). The
production of CFCs in developed countries ceased in 1995, and another com-
mon refrigerant type, hydrochlorofluorocarbon (HCFCs), have been or are due
to be phased out in a few years. HCFC-22, commonly used for small air condi-
tioners and chillers, has already been phased out for use in new equipment. The
most common HCFC refrigerant used in chillers is HCFC-123, which is sched-
uled to be phased out of production for new equipment in 2020 and production
will be banned in 2030 in developed countries. Chillers installed now can be
expected to be operational well past the production ban date. However, HCFC-
123 will likely be available well into the middle of the twenty-first century and
certainly within the lifetimes of machines currently being manufactured due to
stockpiling, recovery, and recycling of HCFC-123 from existing chillers as
they are replaced.
In response to the Montreal Protocol, several zero-ODP hydrofluorocarbon
(HFC) refrigerants were developed, the most common of which is HFC-134a.
The global warming potential (GWP) of refrigerants is another significant
environmental issue. Gases that absorb infrared energy enhance the greenhouse
effect in the atmosphere, leading to the warming of the earth. Refrigerants have
been identified as greenhouse gases. A chart showing the ODP versus GWP of
various refrigerants is shown in Table 3-2. Theoretically, the best refrigerants
would have zero ODP and zero GWP. Unfortunately, many of the refrigerants
with zero or low ODP and GWP, such as R717 (ammonia), are flammable or
toxic or both.
Table 3-1 Montreal Protocol

The 1987 Montreal Protocol, and subsequent revisions, established the following timeline for the phase-
out of chlorofluorocarbons (CFC) and hydrochlorofluorocarbon (HCFC).
Refrigerant Year Restrictions
CFC-11 1996 Ban on production
CFC-12 1996 Ban on production
HCFC-22 2010 Production freeze and ban on use in new equipment
2020 Ban on production
HCFC-123 2015 Production freeze
2020 Ban on use in new equipment in developed countries
2030 Ban on production in developed countries
HFC-134a No restrictions at this point in time*
* As of this date there are no restrictions in North America on the use of R-134a. This could change, so the reader is advised to
seek out the most recent information. HFCs have been or are proposed to be banned in many European countries.
Table 3-2 ODP Versus GWP for Common Refrigerants

Ozone Global
Theoretical
Depletion Warming
Refrigerant Efficiency,
Potential Potential*
kW/ton**
(ODP) (GWP)
R-11 Trichlorofluoromethane 1 4750 6.6
R-12 Dichlorodifluoromethane 1 10,900 6.3
R-22 Chlorodifluoromethane 0.04 1810 6.2
R-123 Dichlorotrifluoroethane 0.02 77 6.5
R-1234yf 0 4 6.4
R-134a Tetrafluoroethane 0 1430 6.3
R290 Propane 0 3 6.2
R407C (23% R-32, 25% R-125,
0 1770 6.0
52% R-134a)
R-410A (50% R-32, 50% R-125) 0 2080 5.9
R717 AmmoniaNH3 0 0 6.3
* GWP values from IPCC (2007).
** Theoretical efficiencies from Calm (2005).
Table 3-2 also shows the theoretical efficiencies of each refrigerant for a
typical cooling application. Refrigerant type is not the only factor that deter-
mines actual chiller efficiency; factors such as compressor type and design and
the heat transfer effectiveness of the evaporator and condenser also play major
roles. So, from a users perspective, refrigerant theoretical efficiency is not
important with respect to chiller selection; the actual efficiency of the equip-
ment is what matters.
In the constant effort to simultaneously minimize GWP and ODP, hydro-
fluoroolefin (HFO) refrigerants have been developed, including R-1234yf
(already used in automobile air conditioning in place of R-134a), R-1234ze
(another R-134a replacement), and R-1233zd (a low-pressure refrigerant com-
parable to R-123). Some of these are slightly flammable, which has given rise
to a new flammability class 2L, for which application regulations are currently
under development (see ASHRAE Standard 34-2016 for more information on
flammability classes).
Water Chiller Components

Compressors
The four most common types of compressors used in packaged water
chillers are
Reciprocating
Scroll
Screw
Centrifugal
The first three types are called positive displacement compressors because
they compress the refrigerant by trapping a fixed amount and forcing (displac-
ing) the trapped vapor into smaller and smaller volumes.
Reciprocating compressors are almost nonexistent in modern chillers,
replaced primarily by scroll and screw compressors. Hence they are not dis-
cussed here; additional information about them is available in ASHRAE Hand-
bookHVAC Systems and Equipment (2016d).
Scroll
Scroll compressors (Figure 3-3) are the most common compressor type on
for smaller chiller sizes, although there are scroll machines available up to
400 tons in capacity. They are mostly used in outdoor air-cooled chillers.
Scroll compressors used in chillers typically range from 5 to 50 tons and
are single speed without unloading capability; compressors are cycled to con-
trol capacity. Some advanced scroll compressors achieve variable unloading
capacity by rapidly engaging and disengaging the scrolls. These compressors
run at constant speed and have unloading efficiencies similar to cycling com-
pressors but with much finer temperature control (smaller temperature swings).
Variable-speed scroll compressors are also available and are beginning to be
applied to chiller applications.
Figure 3-3 Scroll compressor.

Figure 3-4 Single screw compressor.
Screw
There are two common screw compressor types: single screw and twin
screw.
Single Screw
The single screw (Figure 3-4) consists of a single cylindrical main rotor that
works with a pair of gate rotors. The compressor is driven through the main rotor
shaft, and the gate rotors, followed by direct meshing action. As the main rotor
turns, the teeth of the gate rotor, the sides of the screw, and the casing trap refrig-
erant. As rotation continues, the groove volume decreases and compression
occurs. Because there are two gate rotors, each side of the screw acts inde-
pendently. Single-screw compressors are noted for long bearing life, as the bear-
ing loads are inherently balanced. Some single-screw compressors have a
centrifugal economizer built into them. This economizer has an intermediate
pressure chamber that takes the flash gas (via a centrifugal separator) from the
liquid and injects it into a closed groove in the compression cycle, which
increases efficiency.
The capacity of the single screw compressor is typically controlled from a
slide valve in the compressor casing that changes the location where the refrig-
erant is introduced into the compression zone. This causes a reduction in
groove volume, and hence the volume of gas compressed varies (variable com-
pressor displacement). These compressors are fully modulating. The single
screw has slide valves on each side that can be operated independently. This
allows the machine to have a very low turndown with good part-load energy
performance.
Twin Screw
The twin screw (see Figure 3-5) is also known as a double helical rotary
screw. The twin screw consists of two mating helically grooved rotors, one
male and the other female. Either the male or female rotor can be driven. The
other rotor either follows the driven rotor on a light oil film or is driven with
synchronized timing gears. At the suction side of the compressor, the gas fills a
void between the male and female rotors. As the rotors turn, the male and
female rotors mesh and work with the casing to trap the gas. Continued rota-
tion decreases the space between lobes, and the gas is compressed. The gas is
discharged at the end of the rotors.
The twin screw has a slide valve for capacity control located near the dis-
charge side of the rotors, which bypasses a portion of the trapped gas back to
the suction side of the compressor.
Some manufacturers offer screw chillers with VFDs. In addition to excel-
lent part-load and part-lift performance, these chillers offer significantly
reduced noise and wear at off-design conditions. Variable-speed screw chillers,
unlike centrifugal chillers, do not have surge issues (discussed below) and thus
can operate at lower speeds at higher lifts.
Centrifugal
Centrifugal compressors are dynamic (as opposed to positive displace-
ment) compression devices that on a continuous basis exchange angular
momentum between a rotating mechanical element and a steadily flowing
fluid. Like centrifugal pumps, centrifugal chillers have an impeller that rotates
at high speed. The refrigerant enters the rotating impeller in the axial direction
and is discharged radially at a higher velocity. The dynamic pressure (kinetic
energy) of the refrigerant obtained by the higher velocity is converted to static
pressure through a diffusion process that occurs in the stationary discharge or
diffuser portion of the compressor just outside the impeller.
A centrifugal compressor (see Figure 3-6) can be single stage (having only
one impeller) or multistage (having two or more impellers). On a multistage
centrifugal compressor, the discharge gas from the first impeller is directed to
Figure 3-5 Twin screw compressor.

Figure 3-6 Hermetic centrifugal compressor.
the suction of the second impeller and so on for each stage provided. Like the
rotary compressor, multiple stage centrifugal chillers can incorporate econo-
mizers, which take flash gas from the liquid line at intermediate pressures and
feed this into the suction at various stages of compression. The result is a sig-
nificant increase in energy efficiency.
Centrifugal compressors can be either open or hermetic. Open centrifugal
compressors have the motors located outside the casings with the shaft pene-
trating the casing through a seal. Hermetic centrifugal compressors have the
motor and compressor fully contained within the same housing, with the motor
in direct contact with the refrigerant.
Because the discharge pressure developed by the compressor is a function
of the velocity of the tip of the impeller, for a given pressure, smaller-diameter
impellers result in faster impeller speeds. Similarly, for a given pressure, the
more stages of compression there are, the smaller the impeller diameter needs
to be. With these variables in mind, some manufacturers have chosen to use
gear drives to increase the speed of a smaller impeller, while other manufactur-
ers use direct drives with larger impellers and/or multiple stages. High-speed
directly coupled motor-impeller compressors are also available.
Recently, centrifugal chillers from some manufacturers have become avail-
able with oil-free bearings, either magnetic frictionless bearings or ceramic
bearings. This improves efficiency by almost eliminating bearing losses, and
the removal of oil from the system improves heat transfer efficiency. The elim-
ination of oil also substantially reduces the minimum differential pressure (DP)
across the condenser and evaporator (head pressure) (see the Chapter 7 section
Control Schematic for a Typical Plant for more information on head pressure).
Chillers requiring oil must maintain a minimum head pressure to ensure that
oil can circulate through the system. This limits how much the plant controls
can take advantage of mild weather to reduce condensing temperatures and
chiller lift. As is discussed in more detail below, the lower the lift, the higher
the efficiency.
One of the characteristics of the centrifugal compressor is that it can surge.

Surge is a condition that occurs when the compressor is required to produce
high lift at low volumetric flow. Centrifugal compressors must be controlled to
prevent surge, and this is a limit on part-load performance. During a surge con-
dition, the refrigerant alternately moves backward and forward through the
impeller, creating noise, vibration, and heat. Prolonged operation of the
machine in surge condition can lead to failure. Surge is relatively easy to detect
in that the electrical current to the compressor will alternately increase and
decrease with the changing refrigerant flow. Just before entering surge, the
compressor may exhibit a property called incipient surge, in which the
machine gurgles and churns. This is not harmful to the compressor but may
create unwanted vibration. The electrical current does not vary during incipient
surge.
The capacity of centrifugal compressors may be controlled by two meth-
ods. The most common is to use inlet guide vanes or pre-rotation vanes (see
Figure 3-7). The adjustable vanes are located in the compressors suction at the
eye of the impeller and swirl the entering refrigerant in the direction of rota-
tion. This changes the volumetric flow characteristics of the impeller, provid-
ing the basis for unloading.
A second control method is to vary the speed of the impeller in conjunc-
tion with using inlet guide vanes. As with a variable-speed fan or pump,
reducing the impeller speed produces extremely efficient part-load efficiency.
But with fans and pumps, the required flow and pressure vary together; as the
flow rate falls, the pressure required falls as well, roughly as the square of the
flow rate (see subsequent discussion on pumps). But chillers must maintain a
minimum speed that does not necessarily vary with refrigerant flow and
chiller capacity. Rather minimum speed depends on the following:
Minimum speed required to move the refrigerant from the low-pressure side
(evaporator) to the high-pressure side (condenser): Condenser and evaporator
DP can vary with chiller load somewhat, depending on the application. For
an office building, the condensing temperature can be reduced in mild
Figure 3-7 Inlet guide vanes.

weather. The evaporator temperature can, likewise, be raised in mild weather

due to reduced ventilation and heat transfer loads on the plant. The same
would not be the case for a data center, because the load is weather indepen-
dent. A minimum DP also must be maintained for oil circulation unless the
chillers are oil free.
Minimum speed required to avoid surge: The chiller is most efficient when
operating at its lowest speed just before going into surge, but efficiency
falls and damage can occur if the surge line is crossed. So the controls must
dynamically determine the surge line with a fairly robust strategy to stay
close to but out of the surge region. Some controls use a chiller map of the
surge line as a function of load and DP built into the controller. Others will
lower speed until current spikes are sensed as the compressor enters surge,
then respond by increasing speed.
When the impeller is at the minimum speed, further reductions in capacity

are obtained by using the inlet guide vanes.
Variable-speed centrifugal compressors can produce the most energy-efficient
part-load performance of any compressor type. But to do so the minimum speed
must be as low as possible, which in turn requires that the condenser and evapo-
rator DP and temperature (lift) be as low as possible. Minimizing lift requires
aggressive water temperature reset strategies. Without these strategies, which are
discussed in Chapter 6, variable-speed centrifugal chillers can be no more effi-
cient than fixed-speed chillers. In fact, due to the inefficiency of the drive, if the
lift is not reduced, a variable-speed chiller may be less efficient than a constant-
speed chiller.
Absorption Chillers
The absorption process is another way to evaporate and condense refriger-
ants, but the process is thermal/chemical rather than mechanical. Though
appearing quite complex, absorption chillers use the same refrigeration process
discussed for mechanical compression except that phase change is achieved
with an absorber, generator, pump, and recuperative heat exchanger (HX). The
design used by almost all commercial absorption chillers uses lithium bromide
as the absorbent and water as the refrigerant. See Chapter 18 of ASHRAE
HandbookRefrigeration (2014) for a description of how the absorption cycle
works.
A single-effect absorption process (Figure 3-8) is similar to a double-effect
absorption process (Figure 3-9), except that a generator, condenser, and HX are
added for the double-effect absorption process. The refrigerant vapor from the
primary generator runs through a HX (secondary generator) before entering the
condenser. The secondary generator with the hot vapor on one side of the HX
boils some of the lithium bromide and refrigerant solution, creating the double
effect. The double-effect absorption process is significantly more energy effi-
cient than the single-effect absorption process, but it requires a higher tempera-
ture heat source.
Figure 3-8 Single-effect absorption.
Figure 3-9 Double-effect absorption.

Absorption machines can be direct fired or indirect fired. The direct-fired

absorber has an integral combustion heat source that is used in the primary
generator. An indirect-fired absorber uses steam or hot water from a remote
source.
Absorption machines are controlled by modulating the firing rate of a
direct-fired machine or modulating the flow of steam or hot water in an
indirect-fired machine. Variable-speed refrigerant and solution pumps
greatly enhance the controllability of the absorption machine.
Evaporators
Two types of evaporators are used in water chillersthe flooded shell
and tube and the DX. DX evaporators may be shell-and-tube type or brazed-
flat-plate type. Flooded shell and tube HXs are typically used with large
screw and centrifugal chillers, while DX evaporators are usually used with
positive displacement chillers like the scroll and reciprocating machines.
While water is the most common fluid cooled in the evaporator, other fluids
are also used. These include a variety of antifreeze solutions, the most com-
mon of which are mixtures of ethylene glycol or propylene glycol and water.
The use of antifreeze solutions significantly negatively affects the perfor-
mance of the evaporator but may be needed for low-temperature applications.
The fluid creates different heat transfer characteristics within the tubes and
has different pressure drop characteristics. Machine performance is generally
derated when using fluids other than water.
Flooded Shell and Tube

The flooded shell and tube HX has the cooled fluid (chilled water) inside
the tubes and the refrigerant on the shell side outside the tubes. The liquid
refrigerant is uniformly distributed along the bottom of the HX over the full
length. The tubes are partially submerged in the liquid. Distributors are used as
a means to ensure uniform distribution of vapor along the entire tube length,
and eliminators prevent the violently boiling liquid refrigerant from entering
the compressor suction line. The eliminators are made from parallel plates bent
into a Z shape, wire mesh screens, or both plates and screens. An expansion
valve, float valve, or orifice maintains the level of the refrigerant. The tubes for
the HX are usually both internally and externally enhanced (ribbed) to improve
heat transfer effectiveness.
Manufacturers typically limit water flow on the high end to prevent erosion
of the piping and on the low end (typically around 3 ft/s with smooth tubes and
much lower with enhanced tubes) to maintain Reynolds numbers above the
laminar flow regime to maintain high heat transfer coefficients. It is best to
check with the manufacturers for their specific flow rate limitations on each
chiller. Flooded shell and tube HXs are available with multiple passes, with
two being the most common for temperature differences from roughly 8F to
18F and three passes for 18F to 25F temperature differences. The greater the
number of passes, the lower the minimum flow requirements.
Figure 3-10 Direct expansion (DX).
Direct Expansion (DX)

The DX shell and tube evaporator (Figure 3-10) has the refrigerant inside
the tubes and the cooled fluid (chilled water) on the shell side (outside the
tubes). Larger DX evaporators have two separate refrigeration circuits that help
return oil to the positive displacement compressors during part load. DX cool-
ers have internally enhanced (ribbed) tubes to improve heat transfer effective-
ness. The tubes are supported on a series of polypropylene internal baffles,
which are used to direct the water flow in an up-and-down motion from one
end of the tubes to the other. DX evaporators often are limited to 15F to 18F
temperature differences; where a high temperature difference is desired (see
Chapter 5), chillers must be piped in series.
Condensers
There are a number of different kinds of condensers manufactured for pack-
aged water chillers. These include water-cooled, air-cooled, and evaporative-
cooled condensers. (Air-cooled and evaporative condensers are discussed later in
this chapter with cooling towers and heat rejection devices.) Numerous types of
water-cooled condensers are available including shell and tube, double pipe,
brazed flat plate, and shell and coil. This discussion focuses on the condenser
most commonly used on packaged water chillersthe shell and tube HX.
A horizontal shell and tube condenser (Figure 3-11) has straight tubes
through which water is circulated while the refrigerant surrounds the tubes on
the outside. Hot gas from the compressor enters the condenser at the top where
it strikes a baffle. The baffle distributes the hot gas along the entire length of
the condenser. The refrigerant condenses on the surface of the tubes and falls to
the bottom where it is collected and directed back to the expansion device then
to the evaporator. The bottom tubes are usually the first pass (coldest) of the
condenser water and are used to subcool the refrigerant. Often the condenser is
used as the refrigerant receiver where the refrigerant is stored when not in use.
The tubes can be enhanced (ribbed) on both the inside and outside. How-
ever, because the condenser water often comes from an open cooling tower, the
inside of the condenser tubes may become fouled and require mechanical
Figure 3-11 Polypropylene internal baffles.
cleaning. Inside enhancementusually with straight or spiral groovesmay

be problematic because the grooves will be the first areas to become fouled.
Fouling can become a problem when concentrations of dissolved solids
increase greatly above recommendations and when tube velocities drop into
the laminar flow regime (below about 3 ft/s) for a significant amount of time.
Even considering decreased performance of the enhanced condenser tube due
to fouling, the heat exchange effectiveness with the enhanced tube may still be
greater than a smooth bore tube.
Design condenser water velocities range from about 3 to 12 fps. Lower
speeds are acceptable for short-term conditions, such as for head pressure con-
trol during start-up, but many manufacturers recommend higher velocities for
most run hours to reduce the risk of fouling.
Water-cooled condensers are usually multiple pass, with two pass being
most common. The condenser water side can be split into two separate tube
bundles to accommodate a heat recovery mode or to add a level of redundancy
in the event that the tubes need cleaning while the machine is still operational.
Accessories and Common Options

Purge Units
Centrifugal chillers that use low-pressure refrigerants such as R-123 oper-
ate below atmospheric pressure. When they leak, air and moisture are drawn
into the machine. Purge units remove the noncondensable gases that collect in
the condenser during normal operation and ultimately reduce the heat transfer
effectiveness, causing greater refrigerant head pressures. Moisture inside the
unit causes the formation of acids that break down the oil and increase internal
corrosion. Purge units consist of compressors, motors, separators, and con-
densers that can be automatic or manual. Automatic purge units are preferred
because they maintain the highest chiller efficiencies possible. Purge units that
reduce refrigerant losses during operation should always be used. Discharge
from purge units must be piped outdoors.
Oil Coolers
Lubricants must be cooled, especially those used with screw machines. A
small HX is provided for this purpose. The heat can be rejected through a city
water connection or a CHW connection, or it may be air cooled or internally
cooled by the refrigerant.
Hot-Gas Bypass (HGBP)

Hot-gas bypass (HGBP) is a means of false-loading the chiller to reduce
short-cycling compressors and oscillating water temperatures that occur once
the chiller has reached its minimum unloading capacity. The minimum stable
operating load ratio varies with chiller design. With most scroll compressor
chillers, the minimum load is typically that of the smallest compressor. Some
scroll chillers are available with variable-capacity compressors that can
unload stably to very low loads. Screw and centrifugal chillers typically can
unload to about 10% to 15% of design capacity. Below the minimum capac-
ity, the compressor must be cycled off. If the chiller experiences many hours
at loads below its minimum unloading capacity, the compressor can cycle
excessively, which reduces the longevity of the equipment, particularly for
fixed-speed chillers. To mitigate this problem, HGBP can be used to unload a
machine to near-zero load by directing the hot gas from the compressor dis-
charge back into the suction. There are no part-load energy savings with
HGBPchiller energy remains at that required for the minimum unloading
capacity regardless of actual load. HGBP is a fairly inexpensive option, so it
may be a good investment for screw and centrifugal chillers to prevent short
cycling should loads be unexpectedly low, and it wastes no energy if loads
turn out to be above the minimum. HGBP is usually not needed for scroll
chillers with variable-capacity compressors because they have very low min-
imum loads. For scroll chillers with constant capacity compressors, HGBP
should be avoided because of the energy waste; instead a storage tank should
be added to the CHW loop to provide sufficient thermal mass to minimize
cycling. The chiller manufacturer generally provides guidance for sizing the
tank.
Heat Recovery
Heat recovered from chillers can be used to heat buildings, domestic hot
water, or a wide variety of low-temperature heating applications. Two types of
heat recovery can be applied to chillers: a desuperheater condenser placed
immediately at the discharge of the compressor and in series with the chillers
main condenser and parallel condensers called a double bundle condenser. In
some locales, condensers used to heat potable hot water have to be double wall
so that any refrigerant leaks cannot contaminate the domestic water circuit.
The economics of applying heat recovery condensers must consider the load
profile of the source to be heated.
Desuperheater condensers are generally applied on reciprocating chillers,
particularly air-cooled machines. These condensers roughly recover 30% of a
chillers heat rejection capacity and can often generate hot water 140F,
depending on load and condensing temperatures. Desuperheaters can slightly
improve chiller efficiencies. As these condensers are located in series with the
units main condenser, they must be designed for a very low refrigerant pres-
sure drop. Often desuperheaters are field retrofitted to chillers.
Double bundle condensers are usually applied to centrifugal chillers and
can recover the machines entire heat rejection capacity. A double bundle con-
denser is typically split into two separate tube bundles, or two separate con-
densers, with the heating water piped to one side and the cooling tower water
piped to the other side. Heat is first rejected to the heating bundle and when the
heating requirement decreases, the extra heat is rejected to the cooling tower.
Double bundle condensers can be inefficient, as the condensing temperature
will be elevated to achieve even the smallest amount of heat recovery and
HGBP is often needed on centrifugal chillers to avoid surge at even medium
loads due to the high lift. Due to the low temperatures recovered with double
bundled condenser and the load matching requirements to recover heat effi-
ciently, double bundle condensers are rarely applied.
Marine Water Boxes

An accessory for shell and tube HXs is the marine water box, which is a
header assembly that allows mechanical cleaning of the tubes without disas-
sembling the connecting piping. However, they add to first costs and pressure
drop and, because mechanical cleaning is so seldom required (particularly on
the CHW side), the future maintenance cost savings seldom justify the added
first costs and pump energy costs. The labor cost of tube pull and cleaning can
be reduced by using flanged or mechanical joints on the fittings at the chiller
and condenser connections for ease of temporarily removing the piping.
Performance Characteristics and Efficiency Ratings

Performance Issues
There are a number of variables that determine the operational characteris-
tics and energy performance of water chillers. A chiller is selected to meet a
specific maximum capacity requirement at certain design conditions, to meet
this capacity at specific (maximum) power draw, and to have specific part-load
operating characteristics. To design chillers that meet the performance specifi-
cations, manufacturers of packaged water chillers must consider a very wide
range of variables. These variables include the following:
Compressor design
Internal refrigerant pressure drops
Heat gains: motors, oil pumps, casings

Over/under-compression
Motor efficiency
Use of refrigerant economizers
Surface area of evaporators/condensers
Tube heat transfer coefficients: fouling, tube enhancement, velocity of fluids
Refrigerant
Each design decision has first-cost implications. Because of this complex-

ity, products on the market have a wide variety of performance characteristics.
Chiller Efficiency Ratings

The efficiency of water chillers is characterized by the coefficient of perfor-
mance (COP). The COP is the ratio of the rate of heat removal to the rate of
energy input in consistent units for a complete refrigerating system or some
specific portion of that system under designated operating conditions. The for-
mula for COP is
Net Useful Refrigerating Effect

COP = ----------------------------------------------------------------------------------------------- (3-1)
Energy Supplied from External Sources
The higher the number, the more energy efficient the machine. ANSI/
ASHRAE/IES Standard 90.1-2016 and Californias Title 24 energy standards
(CBSC 2016) provide minimum energy efficiency standards for water chillers.
The theoretical limit of efficiency is the Carnot efficiency:
TE
COP = -------------------
- (3-2)
TC TE
where TE is the evaporation temperature and TC is the condensing temperature,

both measured in absolute degrees (R or K). So, for example, the theoretical
maximum efficiency of a chiller operating at 40F (500R) evaporation tem-
perature and 100F (560R) condensing temperature is
500
COP = ------------------------ = 8.3 (3-3)
560 500
Chiller efficiencies are also characterized in terms of kW/ton, which is
essentially the inverse of COP (kW/ton = 3.517/COP) and is more commonly
used in the U.S. than COP. The lower the kW/ton, the more energy efficient the
machine. In the example in Equation 3-3, the theoretical lowest kW/ton at
these conditions is 0.42.
Equation 3-2 also demonstrates how reduced lift (difference between con-
denser and evaporator temperatures) improves efficiency. The closer the two
temperatures are to each other, the higher the COP.
Standard chiller ratings are based on Air-Conditioning, Heating, and

Refrigeration Institute (AHRI) standard conditions, which set parameters for
the rating capacity of different machines. These parameters are established in
AHRI Standards 550/590 (2011) (vapor-compression chillers) and AHRI Stan-
dard 560 (2000) (absorption chillers). For water chillers the AHRI rating con-
ditions are as listed in Table 3-3.
Another energy efficiency rating is the integrated part-load value (IPLV).
The IPLV is a single-number figure of merit based on part-load kW/ton. Part-
load efficiency for equipment is based on the weighted operation at various
load capacities for the equipment. The equipment kW/ton is derived for 100%,
75%, 50%, and 25% loads (with consideration for condenser water relief) and
is based on a weighted percentage of operational hours (assumed) at each con-
dition. A weighted average is determined to express a single part-load/part-lift
efficiency number. The weighting factors are as follows: 1% at 100% load,
42% at 75% load, 45% at 50% load, and 12% at 25% load. For water-cooled
chillers, condenser water relief assumes that the temperature of the water enter-
ing the condenser declines as a straight line from 85F at 100% load to 65F at
50% load and below, implying a correlation between weather and cooling load.
This represents a 4F decline for a 10% change in load.
The nonstandard part-load value (NPLV) is another useful energy effi-
ciency rating. This is used to customize the IPLV when some value in the IPLV
calculation is different than standard, such as using 42F leaving chilled water
in lieu of 44F.
While IPLV and NPLV are useful energy performance indicators for indi-
vidual chillers, particularly for equipment efficiency standards and regulations,
the large majority of chillers are installed in multiple-chiller plants. Individual
chillers operating in a multiple-chiller plant may be more heavily loaded than
single chillers within single-chiller systems and operate at different condenser
water temperatures than those assumed. When evaluating a multiple-chiller
plant, a comprehensive analysis must be used to predict the CHW system per-
formance. This is discussed in detail in Chapter 6.
Table 3-3 AHRI 550/590-2011 and 560-2000 Rating Conditions for

Water Chillers
Leaving CHW Temperature 44F
Evaporator Water Flow Rate 2.4 gpm/ton
Entering Condenser Water Temperature 85F
Condenser Water Flow Rate (Electric) 3.0 gpm/ton
3.6 gpm/ton (single stage)
Condenser Water Flow Rate (Absorber)
4.5 gpm/ton (two stage)
Ambient Air (for Air-Cooled) 95F
0.00010 hft2F/Btu(evaporator)
Fouling Factors
0.00025 hft2F/Btu (condenser)
Scroll Chillers
These chillers are widely used in tonnage ranges from 50 to 230 tons,
although they are available up to much larger sizes (400 tons and up). Capacity
modulation is typically achieved through staging of multiple compressors that
are grouped (piped in parallel) in several circuits. This creates some redun-
dancy should a compressor fail. Some manufacturers offer variable-capacity
(also called digital) scroll compressors that can unload down to 10% using a
pulse width modulation (PWM) approach opening and closing scroll plates. A
growing number offer variable-speed scroll compressors, which reduce both
minimum turndown ratio and energy use. As positive displacement machines,
they retain near-full cooling capacity even when operated at temperatures
above the design conditions, and they are, therefore, very suitable for air-
cooled applications. For the same reason, they are also suitable for use as heat
recovery machines.
Screw Chillers
Rotary screw chillers are also positive displacement machines. Like scroll
chillers, they are particularly suitable as air-cooled chillers but are popular in both
air- and water-cooled configurations. Screw chillers tend to be most cost competi-
tive in the 100 to 300 ton range, although they are available in a wider range of
capacities. In the low capacities, they compete less successfully with scroll chillers,
and, in the high capacities, centrifugal chillers tend to be more cost-effective.
Most screw chillers have excellent turndown capability. Some chillers incor-
porate multiple compressors. This provides additional efficiency advantages
during part load and allows unloading below 10%. Screw chillers are inherently
more efficient than scroll compressors because they incorporate refrigerant econo-
mizers (discussed in the Performance Issues section). They have very few moving
parts and have balanced forces on the main bearings. As a result, these machines
are very reliable and generally have the lowest maintenance costs. Screw
machines are usually controlled with a slide valve and are fully modulating,
although some less expensive models use multiple discrete injection ports with
stepped controls. Variable-speed control is also now being offered on single-
compressor machines and on one or more compressors on dual-compressor
machines. Screw chillers tend to be noisy at design conditions due to the high
speed of operation. The variable-speed-driven screws offer significant acoustical
benefits at low loads and have less wear and tear on the bearings.
Centrifugal Chillers
Centrifugal chillers have the highest efficiency ratings of all the chillers. They
are available in sizes from 80 to 10,000 tons, but the most common factory-built
sizes are from 200 to 3000 tons. Above 3000 tons, they are generally field erected.
They are available in both air-cooled and water-cooled versions, but, because of
very low kW/tons and very high initial cost, air-cooled centrifugal chillers are
uncommon.
Centrifugal chillers are controlled with inlet guide vanes, which allow for
full modulation to as low as 10% to 15% capacity (with condenser water
relief). Note that chiller efficiency drops off severely at low loads. VFDs can be
added to enhance the part-load/part-lift operation characteristics and are usu-
ally cost-effective when evaluated through life-cycle cost analysis (LCCA). In
addition to the energy savings, centrifugal chillers with VFDs are quieter at
part load.
Because of the economics of centrifugal chiller manufacturing, there are
product differences among all the major manufacturers. There are countless
pros and cons to the various features of these products; the following discus-
sion presents some of the main differences.
Direct Drive Versus Gear Drive

Direct-drive chillers typically operate at 3600 rpm. Gears allow impellers
to rotate at speeds up to 35,000 rpm. This allows smaller impellers to be used,
reducing the machines size and first cost. There is an efficiency loss in the
gear train of 1.5% to 2%. Also, the gears have additional bearings and require
regular maintenance, whereas direct-drive machines do not.
The proper selection of impeller diameter and gear ratio allows the
machines to be selected very near their highest performance level or sweet
spot, whereas the direct-drive machines, because of limited impeller diameter
choices, sometimes are selected several efficiency points away from their
sweet spot. Direct-drive machines sometimes have multiple stages (more than
one impeller). In this situation, economizers can be added to enhance the
energy performance of the machine.
Open Drive Versus Hermetic

Open-drive machines have the motor located outside the casing. Efficiency
ratings do not include motor losses (4% to 5% on larger machines). The heat
from an open-drive motor must be removed from the machine room, which
usually requires additional mechanical cooling. Open-drive machines have
seals that leak and are subject to failure. On high-pressure machines refrigerant
can leak out with dire consequences, and on low-pressure machines air can
leak in, causing more purge compressor time and loss of efficiency.
In the event of a catastrophic motor failure, an open-drive machine can be
repaired and placed back in service relatively easily, whereas a hermetic
machine will require significantly more attention. Motor failures in hermetic
machines are almost always catastrophic. Fortunately, motor failures are rare.
Hermetic centrifugal chillers have the motor totally enclosed within the
chiller casing. The motors are kept clean and are cooled by the refrigerant
stream. Hermetic machines have a lower likelihood of refrigerant leakage than
open-drive machines.
Fixed Orifice Versus Variable Orifice or Float Valve

When a fixed orifice is used as the thermal expansion device, a minimum DP
must be held between the condenser and evaporator to ensure proper refrigerant
flow. This may limit the degree of condenser water relief that can be obtained
during off-peak time, with the consequence that the machine will not have as
good a part-load performance as a machine with a variable orifice or float valve.
Oil Return
Chillers may have an oil pump, but most require a minimum DP between
the condenser and evaporator to be maintained to ensure proper refrigerant
flow. This condition is often exerted by the manufacturer requiring a minimum
15F to 25F between the leaving CHW temperature and the leaving condenser
water temperature (an indicator of refrigerant lift and often called lift). Using
an oil pump can reduce the minimum lift to about 5.5C, resulting in improved
chiller efficiencies at low condenser water temperatures, particularly with vari-
able-speed chillers. With oil-free chillers, the minimum lift need only be a few
degrees and, in some chillers, may be zero or even negative.
Absorption Chillers
Absorption chillers can be either single or double effect. Single-effect
chillers have COPs of 0.60 to 0.70 and double-effect chillers have COPs of
0.92 to 1.20. Because the double-effect machines are 50% to 100% more effi-
cient than the single-effect chillers, there is little doubt about which to choose
if absorption is being considered. Single-effect chillers are beneficial where
waste steam is available or where hot-water temperatures are not high enough
to fuel a double-effect absorption chiller. Triple- and quadruple-effect
machines are being developed but are not yet on the market.
Absorption machines can be direct or indirect fired. Direct-fired machines
have the advantage that they can also be used to heat the building and/or
domestic hot water. If a direct-fired absorption machine is also used as a heater,
the avoided cost of a separate boiler and boiler room (space) may help offset
some of the added cost of the machine.
Sizes for absorption chillers range from 100 to 1700 tons. Absorption
machines typically cost two or more times that of an electric-driven chiller.
Because of absorption chillers low, the heat rejection system must be about
50% larger than with a compression chiller plant, increasing the cost of con-
denser water pumps, piping, and cooling towers.
Commercial absorption chillers have additional operating disadvantages
that should be considered:
They cannot produce water at temperatures as low as those of electric
chillers. The minimum CHW supply temperature is typically 43F or 44F,
which limits their use with thermal energy storage (TES) systemscertainly
ice-storage systems but also CHW storage tanks where 39F water is desired
because water at that temperature has the lowest density, enhancing tank
stratification and increasing storage capacity.
They take longer to start up and to shut down, thus requiring longer time
between cycles than electric chillers. The slow start-up is in part due to the
capacitance of their refrigerant. The cycling is due to the chemistry. As
CHW flow is maintained through the chiller during start-up and shut-down
periods, at lower or no produced cooling capacity, maintenance of system
CHW supply temperature can be an issue. This limits their use in plants
such as those for data centers, where rapid deployment is an issue.
They cannot abide low flows or temperatures on the condenser water side.
This limitation can hamper the performance of plants with water-side econ-
omizers (WSEs) and hybrid plants where variable-speed-driven electric
centrifugal chillers might be optimized by low condenser flows and tem-
peratures at part-load conditions. Primary/secondary condenser water
pumping may be required for most efficient plant operation.
They are significantly larger than electric chillers and require larger towers.
They may not last as long as electric chillers and are subject to failure if not
properly maintained. The absorption chillers chemistry is corrosive and
will destroy the chiller if inhibitors are not properly maintained.
Because of these operating disadvantages, much higher first costs, and much
higher operating costs, absorption chillers are seldom the best choice. However,
there are a few applications where an absorption chiller may make sense:
Very high electrical costs, including demand and low natural gas cost
Electrical service not available or too costly to upgrade
Low-cost gas from landfill, solar, or biomass available
Waste or very low-cost steam or hot water available (e.g., from a cogenera-
tion plant or solar thermal panels)
Turbine-Driven and Engine-Driven Chillers

While not a large segment of the chiller market, turbine-driven and engine-
driven chillers are sometimes economically viable. Both use the same vapor
compression cycle as an electric machine except they use either a reciprocating
engine or a gas- or steam-driven turbine as the prime mover. For larger applica-
tions, the refrigeration component is usually an open-screw or centrifugal
chiller. Because these chillers use variable-speed technology, the part-load
characteristics are comparable to variable-speed electric chillers.
Engines use natural gas or diesel fuel. Some hybrid units have both an engine
and an electric motor so that the fuel may be switched depending on the utility
rates at the time. Engines require heat rejection from the jacket water. Heat can be
rejected out the cooling tower (through a HX) or smaller units can be air cooled.
The jacket water heat is available for heat recovery of domestic water or other
loads occurring at the same time as the engine runs. Heat recovery water tempera-
tures at 180F to 200F are easily produced, availing heat recovery to a wider
range of loads, which if amply available can significantly impact the economics.
Engines need additional maintenance, with top-end overhauls required every
12,000 hours and complete overhauls at 35,000 hours. Reciprocating engines are
much louder than electric-driven or absorption machines and may require special
enclosures or acoustical abatement. Natural gas and steam turbines are a very
small part of the market and are used in very large plants (up to 10,000 tons).
As there are limited manufacturers of these products, care is required in
procuring them. A flat specification for a turbine-driven chiller on a large plant
can give a single manufacturer an unfair advantage on bidding the entire plant
including the turbine and electric chillers.
Heat Rejection
The primary means of heat rejection in the HVAC industry are cooling tow-
ers, air-cooled refrigerant condensers, and evaporative refrigerant condensers.
Cooling Towers
The conversion of liquid water to a gaseous phase requires an amount of
energy called the latent heat of vaporization. Cooling towers use the internal
heat from water to vaporize the water in a near-adiabatic saturation process. A
cooling towers purpose is to expose as much water surface area to air as possi-
ble to promote the evaporation of the water. In a cooling tower, approximately
1% of the total flow is evaporated for each 12.5F temperature change. There
are two important terms used in the discussion of cooling towers:
Range: The temperature difference between the water entering the cooling
tower and the temperature leaving the tower
Approach: The temperature difference between the water leaving the cool-
ing tower and the ambient wet-bulb temperature
The performance of a cooling tower is a function of the ambient wet-bulb tem-
perature, entering water temperature, airflow and water flow. The dry-bulb tem-
perature has an insignificant effect on the performance of a cooling tower.
Nominal cooling tower tons are the capacity based on a 3 gpm flow, 95F entering
water temperature, 85F leaving water temperature, and 78F entering wet-bulb
temperature. For these conditions the range is 10F (9585) and the approach is
7F (8578). Significant confusion in the industry has been caused because cool-
ing tower tons and chiller tons use the same units (tons) but have different values;
accordingly, the use of the term cooling tower tons has been waning and is no lon-
ger common. This is beneficial because the heat rejection capacity of a cooling
tower varies widely depending on flow and temperatures, so the term was also
misleading.
Types of Cooling Towers

Cooling towers come in a variety of shapes and configurations. A direct tower
is one in which the fluid being cooled is in direct contact with the air. This is also
known as an open tower. An indirect tower is one in which the fluid being cooled is
contained within an HX or coil and the evaporating water cascades over the outside
of the tubes. This is also known as a closed-circuit cooling tower or a fluid cooler.
The tower airflow can be driven by a fan (mechanical draft) or can be induced
by a high-pressure water spray, although the spray type is rarely used. The
mechanical draft units can blow the air through the tower (forced draft) or can pull
the air through the tower (induced draft). The water invariably flows vertically
from the top down, but the air can be moved horizontally through the water (cross
flow) or can be drawn vertically upward against the flow (counterflow).
Water-to-air surface area is increased by using fill. Fill can be splash type or
film type. Film-type fill is most commonly used and consists of closely spaced
sheets of corrugated polyvinyl chloride (PVC) arranged vertically. Splash-type
fill uses bars to break up the water as it cascades through staggered rows.
Typically in the HVAC industry, cooling towers are packaged towers that
are factory fabricated and shipped intact to a site. Field-erected towers mostly
serve very large chiller plants and industrial/utility projects. When aesthetics
play a role in the selection of the type of tower, custom-designed field-erected
cooling towers are sometimes used. In these towers, the splash-type fill is often
made of ceramic or concrete blocks.
The following is a discussion of the most common types of cooling towers
encountered in HVAC CHW plants.
Forced-Draft Cooling Towers

Forced-draft towers (Figure 3-12) can be of the cross-flow or counterflow
type, with axial or centrifugal fans. Forward-curved centrifugal fans are com-
monly used in forced-draft cooling towers. The primary advantage of a centrif-
ugal fan is that it has capability to overcome high static pressures that might be
encountered if the tower were located within a building or if sound attenuators
were located on the inlet and/or outlet of the tower to reduce ambient noise, as
might be needed for towers located in noise-sensitive residential areas. Cross-
flow towers with centrifugal fans are also used where low-profile towers are
needed. These towers are relatively quieter than other types of towers in the
Figure 3-12 Forced-draft cooling tower.

high-frequency bands. However, towers with centrifugal fans are not energy
efficient. The energy to operate this tower is more than twice that required for a
tower with an axial fan. Another disadvantage of the forced-draft tower is that,
because of low discharge air velocities, they are more susceptible to recircula-
tion than induced-draft towers. This is discussed in further detail in the section
Induced-Draft Cooling Towers.
Induced-Draft Cooling Towers

The induced-draft tower is by far the most widely used cooling tower
available in the HVAC industry. These towers can be cross flow or counter-
flow and use axial fans (Figure 3-13). Most field-erected cooling towers are
the induced-draft type. Because the air discharges at a high velocity, they are
not as susceptible to recirculation as forced-draft towers. The large blades of
Figure 3-13 Induced-draft cooling towers.

an axial fan can create noise that is difficult to attenuate and, depending on
the location on the property, could cause problems. Many manufacturers
offer low-sound blades that reduce noise levels but often reduce airflow rates
and efficiency as well. Low-sound propeller fan towers can be as quiet as
centrifugal fan towers without sound attenuators.
Closed-Circuit Fluid Coolers

With a closed-circuit fluid cooler, the fluid is located within a coil
(rows of tubes) rather than being open to the environment. A pump draws
water from a sump and delivers it to a header where the water is sprayed
over the coil. With proper initial chemical treatment, the fluid does not foul
the condenser tubes, so chiller maintenance is reduced and energy effi-
ciency is always at peak. Because of the additional heat exchange process,
for the same capacity as an open tower, a closed-circuit fluid cooler is
physically much larger and significantly more expensive than conventional
open towers.
Cooling-Tower Performance
Given a fan selection, flow rate, range, entering wet-bulb temperature,
and fill volume, cooling towers have a wide range of performance charac-
teristics. Typical performance curves (see Figure 3-14) show the relation-
ship between these variables at different operating conditions. In reviewing
the typical performance curve, one feature not well understood is that for a
given range, as the entering wet-bulb temperature decreases, the approach
increases. As entering wet-bulb temperature drops, it is likely that the load
(range) will also decrease for the same flow rate. Yet even at this condition,
the approach usually increases over design condition. This is particularly
important when considering the selection of cooling towers for use with
WSEs. To obtain the maximum effectiveness at low wet-bulb temperatures,
a cooling tower used in a WSE system should often be larger (selected for
lower approach) than a tower selected just for maximum peak duty.
Tower efficiency is defined in ANSI/ASHRAE/IES Standard 90.1 as the
maximum flow rate (gpm) the tower can cool from 95F to 85F at 75F
entering wet-bulb temperature, divided by the motor horsepower (2016b).
Typical efficiencies range from 20 to 50 gpm/hp for centrifugal fan towers
and from 40 to 120 gpm/hp for propeller fan towers. Higher-efficiency tow-
ers usually are physically larger (more fill) with smaller fan motors operat-
ing at lower speeds. Cooling towers are relatively inexpensive when
compared to the total cost of a chiller plant and incremental increases in
tower efficiency can be purchased at a relatively low cost. More efficient
towers also tend to be quieter due to lower fan speeds. Optimum tower effi-
ciency for various applications is discussed further in Chapter 5.
Figure 3-14 Typical tower performance at constant heat rejection load.
Application Issues
Siting and Recirculation
When the saturated air leaving the cooling tower is drawn back into the
intake of the tower, the recirculation that occurs degrades the performance of
the tower. Wind forces create a low-pressure zone on the downwind (leeward)
side of the tower that causes this phenomenon. Wind forces on the leeward side
of the building can also create downward air movement. When cooling towers
are located in such a way that the discharge from one tower is directed into the
intake of an adjacent tower, recirculation can also occur. Recirculation is a
greater problem when cooling towers are confined within pits or have screen
walls surrounding them, typically to hide them for architectural reasons. If the
tower is sited in a pit or well, it is essential to follow the tower manufacturers

guidelines to determine the proper location of the outlet and minimum clear-
ances for the air intake. Typically the manufacturer will require that the tower
outlet be no less than flush with the height of the walls of the enclosure. Cross-
flow side discharge towers should never be used in pits or wells, because recir-
culation is almost assured.
The Cooling Technology Institute (CTI) recommends that recirculation
effects be accounted for in the selection of the tower. Their tests show that as
much as 8% of the discharge air could be recirculated back into the intake and
that the worst conditions occur with winds of 8 to 10 mph. Where recirculation
is a concern, a rule of thumb is that the entering wet-bulb temperature used to
select the tower should be increased by 1F above the ambient temperature to
account for recirculation effects.
Tower height is often a consideration in selection due to the architectural
impact. The lowest profile towers are usually blow-through centrifugal type,
but they are more expensive and less efficient than other options and thus
should be used only in extreme cases. Double inlet cross-flow induced-draft
towers also have low profiles, but they have a large footprint and thus require
more plan space. Counterflow-induced-draft towers have the smallest footprint
but tend to require the most height. However, sometimes the height of these
towers is an advantage when the tower is located in a well and the height of the
well is determined, for example, by other tall penthouse elements, such as a
traction elevator machine room. Because the tower discharge should be at least
flush with the walls of the well, tall cooling towers may avoid the need for and
cost of high support pedestals (which also make maintenance access more dif-
ficult) or fan discharge duct extensions.
Capacity Control
Like most air-conditioning equipment, cooling towers are selected to
deliver peak capacity at design weather conditions, but most of the time they
operate at well less than peak capacity. There are a number of methods used to
control the temperature of the water leaving the cooling tower, including the
following:
On/off: Cycling fans is a viable method but leads to increased wear on belts
and gear drives (if used) and can lead to premature motor failure. It is also
the least energy-efficient control option and can result in large variations in
condenser water temperature, which can cause unstable chiller operation.
Cycling is therefore the least favorable method of controlling temperature.
Two-speed motors: Multiple wound motors or reduced-voltage starters can
be used to change the speed of the fan for capacity control. This method is
cost-effective and well proven. Because of basic fan laws, there are signifi-
cant energy savings when the fans are run at low speed. One pitfall with
two-speed fans is that when switching from high to low speed, the fan rpm
must reduce to below low speed before energizing the low-speed step to
avoid motor overload trips.
Pony motors: This is another version of the two-speed approach. A second,

smaller motor is belted to a common fan shaft. For low-speed operation,
the larger motor is de-energized and the smaller motor energized for a
lower speed. Again, when going from high speed to low speed, the fan
must slow down sufficiently before energizing the low-speed motor. This
option has fallen out of favor as VFD costs have fallen, making it more
expensive than alternative options. It is also typically an option only avail-
able on centrifugal fan blow-through towers.
Modulating discharge dampers: Used exclusively with centrifugal fans, dis-
charge dampers built into the fan scroll can be modulated for capacity con-
trol. Although it does save energy by riding the fan curve, other methods of
capacity control provide better energy savings results at lower costs.
Hence, this option is seldom used in modern towers.
Variable-frequency drive (VFD): Adjustable-frequency VFDs can be added
to the motors for speed control. This method provides the tightest tempera-
ture control performance and is the most energy-efficient method. One pit-
fall to avoid with VFDs is to not run the fans at critical speeds, which are
speeds that result in resonant frequency vibrations and can severely dam-
age the fans. Critical speeds are typically determined empirically post-
installation. Minimum fan speeds are discussed in the sections Belt Versus
Gear Drive and VFD Accessories and Application Considerations: Mini-
mum Speed Setpoint.
Selecting the best control option is discussed in Chapter 5.
Chemical Treatment and Cleaning

Cooling towers are notorious for having high maintenance costs. Unfortu-
nately, cooling towers are very good air scrubbers. A 200 ton open cooling
tower can remove 600 lb of particulate matter in 100 hours of operation.
Because tower water is open to the atmosphere, the water is oxygen saturated,
which can cause corrosion in the tower and associated piping. Towers evapo-
rate water, leaving behind dissolved solids such as calcium carbonate that can
precipitate out on piping and condenser tubes and decrease heat transfer and
energy efficiency.
To avoid these problems, towers must have water treatment systems and
should be inspected and cleaned regularly. It is best to contract with a cooling
tower water treatment specialist to assist in determining the appropriate water
treatment program and to provide regular monitoring.
The following are some of the strategies to consider in a good chemical
treatment program:
Blowdown: To control dissolved solids, a portion of the flow of the tower
must be bled from the system. Depending on the quality of the water (e.g.,
silica and other dissolved solid content) and water treatment approach
(chemical versus nonchemical), the cycles of concentration of dissolved
solids (ratio of blowdown to incoming water concentration) can vary from
2 to 10, equal to a blowdown rate of about 0.06 to 0.03gpm/ton, respec-

tively. This is typically controlled by sensing water conductivity (mhos)
and bleeding water to maintain a certain conductivity set point. This set
point should be determined by a water treatment specialist.
Corrosion control: Corrosion can be caused by high oxygen content, car-
bon dioxide (carbonic acid), low pH, or high dissolved solids. Regular
injection of corrosion inhibitor chemicals is a common solution.
Biological growth: Slime and algae are handled with shock treatments of
chlorine or chlorine compounds. It is best to alternate between two differ-
ent compounds so that organisms do not develop a tolerance to the chemi-
cals. At least one of the biocides should be effective against Legionella, the
organism responsible for Legionnaires disease.
Scale prevention: Control of the pH (acid levels) is extremely important in
areas with very hard water. Usually acids, inorganic phosphates, or similar
compounds are used to control pH. Blowdown is usually effective in areas
with neutral or soft water.
Technologies that purport to eliminate the need for inhibitors and/or bio-
cides have come and gone over the years, usually with mixed or poor results.
One promising technology employs pulsed electromagnetism to remove dis-
solved solids and inhibit biological growth. The appropriate use of these sys-
tems depends on the local makeup water quality and other local conditions. A
local water treatment specialist should be consulted. To get an unbiased recom-
mendation, the specialist should represent or operate both chemical and
nonchemical water treatment systems.
Another water treatment option is particulate filtration. The filter can be
mounted in-line with the primary condenser water flow but more commonly it
is mounted in a small sidestream configuration with its own pump to reduce the
energy penalty of the filter pressure drop. Sidestream filters generally circulate
about 10% of the system flow. Common filters are centrifugal separators or
sand filters; the latter remove finer particles but require higher pump energy
and more frequent backwash and associated water use. A common accessory is
a basin eductor distribution system: the sidestream filter pump draws water
from the basin, pumps it through the filter, and then discharges it through an
array of eductors mounted on the tower basin designed to stir up settling parti-
cles so they can be effectively removed by the filter. The advantage of this
design is that it reduces tower maintenance costs by reducing how frequently
the basin must be isolated and cleaned of dirt that precipitates out in the basin.
But the first costs and energy costs are highin fact it is not unusual for the
basin pump to use more energy than the cooling tower fan. Claims that these
filters reduce fouling of condenser water tubes and thus improve energy effi-
ciency have been made by filter manufacturers but to date have never been
demonstrated with unbiased research. Sidestream filters may be desirable
where the tower is located adjacent to ambient air with high particle concentra-
tions, such as near farming, and possibly where nonchemical water treatment
systems are used, because they are designed to have particles coagulate and
precipitate out in the tower basin. To limit energy use, filters should be oper-
ated on a time schedule for a few hours per day on days when the towers have
operated, preferably during off-peak energy rate periods; the exact time period
required must be determined empirically by the plant operator.
Flow Limits
For a given design of a cooling tower, the manufacturer will specify max-
imum and minimum flow rates through the tower. The maximum flow is usu-
ally based on the capacity of the water distribution system within the tower to
adequately distribute the water over the fill. Too much flow will overflow the
tower distribution pans and create a situation where the tower does not get
adequate mixing of air and water to perform properly. Below the minimum
flow rate, the water may not distribute evenly across the entire fill face. This
creates voids where there is no water in the fill. At the boundary where the
wet and dry portions of the fill meet, dissolved solids can drop out of solution
and plate out on the fill. Prolonged operation below the minimum water flow
can thus cause significant scaling to occur. However, this may not be a prob-
lem in areas with excellent water quality. ANSI/ASHRAE/IES Standard 90.1
(2016b) and Californias Title 24 (CBSC 2016) require that in plants with
multiple condenser water pumps, the tower minimum flow rate must be low
enough to handle flow from the smallest pump down to 50% of the total
design flow rate. This is to allow more cells to be active even when the plant
is at part load; as discussed in Chapters 4, 5, and 6, energy efficiency is max-
imized by running as many tower cells as possible. Low minimum flow is
generally achieved with weir dams in the distribution and/or adjusting distri-
bution nozzle type and size.
Cooling Tower Accessories and Options

The following is a list of accessories and options that should be considered
when purchasing a cooling tower:
Vibration switch: This stops the fan if vibration exceeds a certain limit. It
could prevent catastrophic failure of the fan. Codes in some areas require
the installation of a vibration switch.
Side inlet and internal distribution: Cross-flow towers are often supplied
with field-erected overhead piping to the gravity basins. But this distribu-
tion can be factory installed within the tower to reduce overall height and
substantially reduce installed costs. This is because field-erected overhead
piping cannot be supported off of the towers and thus must have expensive
field-installed support frames connected to adjacent structures. It is also not
uncommon for overhead piping to be self-vented and drain when pumps
are stopped, sometimes causing cold-water basin overflow. Air locks can
also form that starve one cell while overflowing the hot-water basins of the
other(s). This is a highly recommended option.
Ladders and railings: Ladders to the tower supply basins and associated
safety railings may be a convenience to maintenance personnel, although
they add to the cost and sometimes to the height of the tower. Using the
side inlet and internal distribution piping option eliminates the need for
accessing supply shutoff valves on the top of the tower, reducing the need
for this option for cross-flow towers. For tall counterflow towers with
external motors, ladders and railings are recommended for safe access to
motor and drive maintenance.
Stainless steel or protective coatings: In most applications, using stainless
steel or coated galvanized steel hot- and cold-water basins is recommended.
The entire tower can also be built from these materials but at very high cost,
usually justified only where future tower replacement will be very expensive.
Basin heaters: In freezing climates where the tower cannot easily be win-
terized (drained) during the cold-weather season, thermostatically con-
trolled electric basin heaters can be provided. Typically in these instances
piping must also be protected with heat tracing and insulation.
Electronic fill controls: The standard water level controller is a float valve,
much like a toilet fill valve. Electronic fill controls using an electronic level
sensor controlling an electric motorized fill valve will improve reliability.
(An even better approach is to use a separate level sensor mounted in an
equalizer standpipe controlling a central makeup water valve located inside
the chiller room. See Chapter 4 for design details.)
Calibrated balancing valves (CBVs): Balancing valves are almost never
needed on cooling towers. Most plants are near self-balancing simply
because their compact size does not result in large differences in pressure
drop across each tower circuit. Cooling tower performance is also very for-
giving to flow imbalance: the cells with excess flow will create warmer
water and the ones with low flow will create colder water, but when they
are mixed, the resulting temperature is almost exactly the same as it would
be if the cells had equal flow.
Choosing the Type of Cooling Tower

When choosing which cooling tower is most appropriate for a particular
application, the following factors should be considered.
Packaged Versus Field-Erected Cooling Towers
The type of cooling tower selected will be determined to a great extent by
the required capacity.
Packaged cooling towers are manufactured to be cost-effective and to ship
on standard-size carriers. Typically, a single cell of a packaged tower will han-
dle a maximum cooling capacity of 650 to 1000 tons at nominal conditions.
Larger plants will require multiple cells.
If the CHW plant is very large, field-erected cooling tower, cells may be more
cost-effective than a packaged cooling tower. Field-erected cooling towers also
offer a greater degree of energy efficiency because the design flexibility makes it
possible to match lower-horsepower fans with larger fill volume. Although field-
erected cooling towers offer greater flexibility when the site has physical con-
straints, they may have longer procurement times than packaged cooling towers
and they can be much more expensive than multiple packaged towers.
Open Versus Closed Circuit

Open-circuit cooling towers are the most prevalent type of cooling tower
used in the HVAC industry.
Closed-circuit cooling towers (also called fluid coolers) are at a disadvan-
tage due to their higher first cost, additional energy cost, and larger physical
size. Closed-circuit cooling towers are therefore only used where
the condenser water pumps must be located remotely from the tower,
the cooling tower is located below the condensers,
it is necessary to keep the condenser water free from contamination with
dirt or impurities due to poor local water quality, or
the condenser water is mixed with fluids from other closed systems (like
the chilled water or hot water) such as on hydronic heat pump systems.
Forced Versus Induced Draft

Most axial fan towers are induced draft for the same reason that most air
handlers are draw through: the air distribution through the fill is inherently
more uniform when air is drawn through it rather than blown through it. Blow-
through designs benefit from the fact that motors and fans are not in the wet,
more corrosive atmosphere of the tower effluent, but the cost benefits of the
induced-draft design generally outweigh these considerations. Induced-draft
towers also are less likely to recirculate discharge air due to the higher velocity.
Centrifugal Versus Axial Fans

Centrifugal fans are significantly less energy efficient in cooling tower
applications than axial fans and generally more expensive. The use of centrifu-
gal fans in cooling towers should be limited to situations where
low-profile towers are absolutely required architecturally or

sound attenuators or other acoustical considerations make a centrifugal fan
the only choice.
Cross Flow Versus Counterflow

Cross-flow and counterflow induced-draft towers are the most common
cooling towers used in CHW plants. Advantages and disadvantages include the
following:
Minimum flow. Cross-flow towers can typically achieve lower minimum

flow rates than counterflow towers because they use gravity-fed distribution
pans that can be readily modified to provide minimum flow rates as low as
30% to 50% using weir dams and flow nozzle extensions. Counterflow tow-
ers use pressurized nozzles that typically have much more limited turndown;
to meet the 50% turndown limit required by Californias Title 24 (CBSC
2016) and ANSI/ASHRAE/IES Standard 90.1, special low-turndown noz-
zles must be specified. (Note that new nozzle designs available from some
counterflow tower manufacturers can provide performance comparable to
counterflow towers.) The lower minimum flow rate can improve efficiency
and eliminate the need for expensive cell automatic isolation valves. This is
discussed in more detail in Chapter 5.
Maintenance access. Cross-flow towers generally are easier to maintain.
The distribution pans are on the top where they can be easily accessed and
cleaned, whereas access to counterflow spray nozzles requires removal of
mist eliminators.
Efficiency and cost. The counterflow design is usually slightly more effi-
cient for the same cost, or said another way, slightly less expensive for tow-
ers with similar efficiency.
Height and footprint. As discussed above under Siting and Recirculation,
counterflow towers tend to be tall but with a small footprint while cross-
flow towers are the opposite. Either can be an advantage, depending on the
architectural constraints. This factor alone can often drive the selection
between the two types.
Belt Versus Gear Drive

Induced-draft axial fans can have either a belt drive or a direct-drive con-
nection to the motor. Direct-drive fans use gear reducers to maintain the low
speeds of the fan. Gear drives cost more but reduce maintenance frequency and
may reduce lifetime maintenance costs compared to belt drives. But the
increasing popularity of VFDs has also increased the popularity of belt drives;
the belts last longer due to the soft start feature of VFDs, and belt drives allow
near-zero minimum speeds while many gear drives require on the order of 20%
minimum speed to ensure adequate lubrication. Lower minimum speed
reduces fan cycling, which reduces wear and tear, reduces noise levels and
abrupt changes in noise levels, reduces condenser water temperature fluctua-
tions, and (slightly) improves energy efficiency.
Air-Cooled Refrigerant Condensers

Types
Another method of heat rejection commonly used in chiller plants is the air-
cooled refrigerant condenser. This can be coupled with the compressor and evapo-
rator in a packaged air-cooled chiller (Figure 3-15) or can be located remotely.
Remote air-cooled condensers are usually located outdoors and have propeller fans
and finned refrigerant coils housed in a weatherproof casing. Some remote air-
cooled condensers have centrifugal fans and finned refrigerant coils and are
installed indoors in what amounts to an AHU. Indoor condensers are only used on
Figure 3-15 Packaged air-cooled chiller.
small chillers and will not be discussed further here. Air-cooled condensers,
whether remote or packaged within an air-cooled chiller, normally operate with a
temperature difference between the refrigerant and the ambient air of 10F to 30F,
with fan power consumption of less than 0.08 hp/ton cooling. Maximum size for
remote air-cooled refrigerant condensers is about 500 tons, with a more common
maximum of 250 ton. Air-cooled chillers are available up to about 400 tons.
There are a number of reasons air-cooled chillers are used. These include
Water shortages or quality problems
Lower cost than water-cooled equipment
With packaged air-cooled chillers, no need for machine rooms with safety
monitoring, venting, etc.
Less maintenance required than cooling towers
Air-cooled chillers are not as energy efficient as water-cooled chillers in

most applications. When comparing the energy efficiency of air-cooled to
water-cooled chillers, care must be taken to include in the water-cooled chiller
the energy consumed by the condenser water pump and cooling tower. Air-
cooled chillers can have very good part-load performance if properly con-
trolled; as the outdoor air temperature drops, the kW/ton improves significantly
if condensing pressure is allowed to float downward. Remote air-cooled refrig-
erant condensers in CHW plants are very seldom used because of the physical
size for the larger capacity machines. Air-cooled chillers are more often used in
smaller chiller plants, generally below 300 tons, as space, water treatment, and
the additional maintenance cost associated with cooling towers or evaporative
condensers outweighs the energy benefit. Californias Title 24 limits the size of
air-cooled chiller plants to 300 tons (CBSC 2016). ANSI/ASHRAE/IES Stan-
dard 90.1 has no current limit on the use of air-cooled chillers.
Air-cooled chillers require little maintenance but they do need to have coils
cleaned regularly, they require standard lubrication, and the refrigerant charge
needs to be periodically checked. If leaves from trees or other debris become a
problem, permanent air filters are available to protect the coils. However, air fil-
ters slightly degrade system performance and require additional maintenance.
As with siting cooling towers, air-cooled chillers can potentially recirculate

the warm discharge air, especially when multiple condensers are located adjacent
to one another or condensers are located within a pit or screen wall. Consult the
manufacturers location guidelines for multiple machines or pit locations.
Controls
When air-cooled condensers operate, typically the fan runs continually in
conjunction with the compressor. When the outdoor temperature falls, it is pos-
sible to decrease the liquid refrigerant pressure too much to adequately over-
come the thermal expansion valve (TXV) pressure drop. In this case, controls
are required to limit the heat rejection. These controls include the following:
Flooded coil: Control valves back up liquid refrigerant into the condenser
to limit the heat transfer surface. This requires a receiver and a large refrig-
erant charge.
Fan cycling: Usually need multiple fans with one or more cycling on and
off to maintain minimum head pressure.
Dampers: Discharge dampers on condenser fan restrict airflow. This option
has become less popular as VFD costs have fallen.
Variable-speed fans: Fan speed modulates airflow.
For systems not intended to run at cold temperatures (less than 40F), fan
cycling is usually the most appropriate choice for control. For systems
intended to run at temperatures down to 0F, fan speed control is the most
common and most efficient.
Evaporative Condensers
Evaporative condensers (Figure 3-16) are similar to closed-circuit fluid coolers
but with refrigerant rather than closed-circuit water in the HX tubes. A pump draws
water from a sump and sprays it on the outside of a coil. Air is blown (drawn)
Figure 3-16 Evaporative condenser.

across the coil and some of the water evaporates, causing heat transfer. The hot gas
from the compressor condenses inside the tubes. Evaporative condensers are pri-
marily used with DX air-conditioning units, although a few manufacturers produce
small packaged water chillers with evaporative condensers.
The effectiveness of the evaporation of the water and the refrigerant in the
heat transfer process means that for a given load, evaporative condensers can
have the smallest footprint of any heat rejection method. The evaporative con-
denser causes lower condensing temperatures and, as a result, is far more effi-
cient than air-cooled condensing. Maintenance and control of evaporative
condensers is similar to that of a closed-circuit fluid cooler. As with cooling
towers, the style of the tower can significantly impact fan power.
Pumps
In the HVAC industry, most pumps are single-stage (one impeller) centrifu-
gal pumps that have either a single inlet (end suction) or a double inlet (double
suction). Vertical turbine pumps are sometimes used in a cooling tower sump
application.
Most pumps in the HVAC industry are bronze fitted, meaning they have a
bronze impeller and wear rings, a bronze or stainless steel shaft sleeve, a stain-
less steel shaft, and a cast iron casing. Centrifugal pumps come with mechani-
cal seals (most common) or packing gland seals. Packing gland seals are
sometimes (but infrequently) used in condenser water systems, where an accu-
mulation of dirt can damage mechanical seals.
Pump Types
The following is a brief discussion of the various types of pumps used in a
CHW plant.
Single (End) Suction

Single- or end-suction pumps can be either flexible coupled (also known as
base mounted [Figure 3-17]) or close coupled (Figure 3-18). Close-coupled
pumps use a special motor that has an extended shaft to which the pump impeller
is directly connected. The motor and pump cannot be misaligned, and they take up
Figure 3-17 Base-mounted end-suction pump.

Figure 3-18 Close-coupled end-suction pump.
Figure 3-19 In-line pump.
less floor space than flexible coupled pumps. However, replacement motors can
have a long lead time and can be difficult to obtain after a breakdown, especially
for larger sizes (25 hp). Flexible coupled pumps allow the motor or pump to be
removed without disturbing the other. The flexible coupling requires very careful
alignment and a coupling guard. Standard EPDM couplings have been known to
fail when used with VFDs at low speeds; flexible polyurethane couplings are rec-
ommended where there will be many hours of operation below about 50% speed.
The flexible coupled pump is usually less expensive than the close-coupled pump
for pumps with motors 20 hp and larger. Usually end-suction pumps are the most
cost-effective for use up to 1500 gpm to 2000 gpm but are available from some
manufacturers up to 4000 gpm.
In-line pumps (Figure 3-19) are end-suction pumps with a suction fitting
designed so that they can be inserted directly into a pipe. They can also be mounted
on a base like other pumps. In the past, these pumps were used almost exclusively
for small pumps, but now they are available in the full range of sizes, with larger
pumps generally mounted on a base for ease of maintenance. Inline pumps are not
quite as efficient as end-suction pumps due to the inlet fitting. These pumps can
save considerable space when mounted in-line with the piping, but maintenance
access is worse when the pump and motor are well above the floor.
Figure 3-20 Double-suction pump.
Figure 3-21 Vertical turbine pump.
Double Suction
In the double-suction pump (Figure 3-20), water is introduced on each side of
the impeller and the pump is flexibly connected to the motor. These pumps are
preferred for larger flow systems (typically greater than 1500 gpm), because they
are very efficient and can be opened, inspected, and serviced without disturbing
the motor, impeller, or the piping connections. Typically, the pumps are mounted
horizontally but can be mounted vertically. The pump case can be split axially
(parallel to shaft) or vertically for servicing. This pump takes more floor space
than end-suction pumps, particularly with the traditional horizontal inlet and dis-
charge (as shown in Figure 3-20). However, some manufacturers offer a style
with vertical inlet and/or discharge to reduce space requirements.
Vertical Turbine
Vertical turbine pumps (Figure 3-21) are axial-type pumps that are used almost
exclusively for cooling tower site-built sump applications. These pumps can be
purchased with enclosures or cans around the bowls when not sump mounted.
Pump Performance Curves

For a given impeller size and rotational speed, the performance of a pump
can be represented on a head-capacity curve of total head (in ft of water) versus
flow (in gallons per minute). Total dynamic head (TDH) is the difference
between suction and discharge pressure and includes the difference between
the velocity head at the suction and discharge connection. Starting from zero
flow, as the pump delivers more water, the mechanical efficiency of the pump
increases until a best efficiency point (BEP) is reached. Increasing the flow fur-
ther decreases the efficiency until a point where the manufacturer no longer
publishes the performance (end of curve). Pump performance data are gener-
ally shown as a family of curves for different size impellers (Figure 3-22).
Notice as the impellers get smaller, the pump efficiency decreases. The power
(hp) requirements are also shown on the performance curve; notice that the
power lines cross the pump curve until one value does not cross. This value is
called the non-overloading horsepower, because operation at any point on the
published pump curve will not overload the motor. Finally, information on the
net positive suction head required (NPSHR) is shown on the pump curve. This
is discussed in greater detail in the Pump Inlet Limitations section.
Figure 3-22 Pump head capacity curve.

Pump curves are also characterized as steep or flat. A flat-curve pump

is defined as a pump where the pressure from shutoff head (head at zero flow)
to the pressure at the BEP does not vary more than a factor of 1.1 to 1.2.
Parallel and Series Pumping

When two or more pumps are operated in parallel (Figure 3-23), a combined
parallel pump curve can be drawn that holds the head constant and adds the flow.
Similarly, a series pump curve can be drawn that holds the flow constant and
adds the head. Pumps are rarely placed in series in HVAC applications.
Variable-Speed Pumping
For a given impeller size, a family of curves can be drawn to represent the
variable-speed performance of a pump (see Figure 3-24). Notice that the BEP
Figure 3-23 Parallel pump curve.
Figure 3-24 Variable-speed performance curve.

follows a parabolic curve just like an ideal system curve (which is discussed in
greater detail in the System Curves section). Also notice that the NPSHR lines
follow fairly closely with the published end-of-curve lines for the various
speeds. The power lines decrease rapidly as the speed decreases, which graphi-
cally demonstrates the potential power savings of variable-speed operation in
variable-flow systems. For a more detailed example of variable-flow applica-
tions, refer to Chapters 4 and 6.
Variable-speed pumps should, in general, not be piped in parallel with
constant-speed pumps if both are expected to operate together. The constant-
speed pump will overpower the variable-speed pump, keeping its check valve
closed until the variable speed is increased sufficiently high to meet the pres-
sure created by the constant-speed pump. Thus the VFD provides much
lower energy savings when both pumps are operating. This can be seen by
the example in Figure 3-25, where two variable-speed pumps in parallel will
use 7.6 bhp while a variable-speed pump in parallel with a constant-speed
pump will use 8.5 bhp.
ANSI/ASHRAE/IES Standard 90.1-2016 and California Title 24 (CBSC
2016) requires variable-flow design for all CHW systems with more than three
control valves. The standards also require VFDs on all variable-flow CHW sys-
tems with pump motors greater than 5 hp.
Figure 3-25 Performance of two VFD pumps versus one VFD and one constant-speed pump.
Figure 3-26 Variable-speed performance at varying DP set points.

Source: Taylor 2012.
ANSI/ASHRAE/IES Standard 90.1-2016 also requires that the DP sensor

used to control pump speed be located at the most remote coil or HX. This is
because the lower the DP set point, the lower pump energy will be, as shown
in Figure 3-26. If the DP sensor were located at the pump discharge, the set
point would have to be set for the design pump head downstream of the pump
to ensure adequate flow at design conditions. The result will be that the pump
runs at near full speed most of the time and pump energy is not much better
than if the pump had no VFD and simply rode its curve as flow reduced.
Selecting Pumps
Pump Type
Table 3-4 summarizes the advantages and disadvantages of the most com-
monly used pumps for CHW plants. For plants with variable-speed pumps
expected to operate at low speeds, flexible-coupled pumps should be avoided
unless special couplings are provided, as discussed previously. From a cost and
space perspective, typical CHW plant pumps with motors 15 hp and less (20 hp
and less with VFDs) should be close-coupled end-suction type. Larger pumps
should be flexible-coupled end-suction type until they reach about 1500 to
2000 gpm, where double-suction pumps become cost competitive and their
lower maintenance costs make them the best option.
Table 3-4 Pump Selection Summary

Type Advantages Disadvantages
High efficiency Highest cost for pump
Hydraulically balanced (<2000 gpm)
(less bearing wear) Requires most space
Double suction
Longest life Highest cost for inertia base
Only pump type available Alignment required
>2500 gpm or so Possible coupling failure
Low cost Higher cost than close-coupled
Uses standard motors <20 hp
End suction More space required than
Base mounted close-coupled
(flexible coupled) Higher cost for inertia base
Alignment required
Possible coupling failure
Lowest cost for pumps for Motors above 25 hp are hard to
End suction 2 to 15 hp or so find and costly
Close coupled Unusual motors of all sizes are
often harder to find
Lowest cost for small pumps Often less efficient
In-line <2 hp or so (due to inlet volute)
Close coupled or Lowest cost to install when mounted Maintenance is more difficult
Flexible coupled in-line if installed in-line
Least space required
Pump Selection
Pumps should be selected between 65% and 115% of the flow rate at the
BEP for the impeller required by the design flow and head (see Figure 3-27)
(not the BEP of the largest impeller). Selecting a pump too close to shutoff
head or too near the end of the curve presents problems with radial thrust and
potential cavitation. This is discussed further in the section Radial Thrust. Con-
stant-speed CHW pumps serving terminals with two-way valves (variable
flow) generally should be flat-curved pumps to reduce the additional head gen-
erated as the pump rides up its curve. If variable speed, efficiency is improved
at low loads if the pump is selected to the right of the BEP.
For multiple constant-speed pumps in parallel, motor sizes should be
selected so that the power curve does not cross the pump curve at any point
(non-overloading). Selecting for non-overloading is not required for single-
pump constant-flow systems, because the system can be balanced with
valves or the impeller can be trimmed to prevent overloading. Selecting for
non-overloading is optional for pumps with VFDs, because the VFD can be
configured to limit speed to prevent overloading, but a non-overloading
motor will allow for greater pump flow when one pump is operating alone,
which may be an advantage in case one pump fails.
Figure 3-27 Pump selection recommendations.
Pumps with VFDs can also be selected with larger impellers to operate at
less than 60 Hz at design conditions. As seen in Figure 3-22, pump efficiency
generally increases with larger impellers, depending on the design flow relative
to the BEP flow. Simply put, a pump with a VFD should be selected with the
largest impeller available that does not increase motor and VFD sizes at full
speed. For instance, assume the pump in Figure 3-25 was selected for 280 gpm
at 50 ft of head. It requires about 5.1 bhp, requiring a 7.5 hp motor. Instead of
trimming the impeller to the design duty (~8 in. as seen from the pump curve),
it could be selected to max out the 7.5 hp motor. Using pump laws,
D2 3
HP2 = HP1 -------
D1
1/3
D2 = D1 -----------
HP2
HP1
1/3
D2 = 8 -------
7.5
= 9.1 in.
5.1
So, a 9.1 in. impeller should be selected. The VFD would then automatically
slow the pump to 8/9.1 60 = 53 Hz to achieve the desired duty. By oversizing
the impeller in this way, pump efficiency increases from about 70% to about 75%
at no additional cost. If the application allowed the pump to ride out its curve, the
VFD could be configured to limit pump speed to prevent overloading.
A common error is to take this one step further: if a 9.1 in. impeller is better
than 8 in., why not select the maximum impeller for the pump, 9.5 in. in this
case, and still use the 7.5 hp motor, again relying on the VFD to limit speed to
prevent overloading? Unfortunately, this will not work. With a 9.5 in.impeller,
the pump would have to slow down to 8/9.5 60 = 50 Hz and at that speed the
delivered hp would drop to 7.5 (50/60)3 = 4.5 bhp (assuming the VFD is con-
figured to a quadratic V/Hz relationship typically used for pump applications),
less than the required 5.1 bhp. So, the impeller can only be maxed out within
the capability of the selected motor size. The motor and VFD could have been
increased to 10 hp with the 9.5 in. impeller, but overall efficiency usually gets
worse with such an increase, because at low loads, motor and VFD efficiencies
fall quickly, as discussed in the section Drive System Efficiency.
System Curves
The system curve describes how the pressure drop in the system varies with
flow. For a given system geometry, the head can be calculated using Equation 3-4:
x
H = Q (3-4)
where the exponent x varies from 2 for turbulent flow to 1 for laminar flow. The
pressure drop of fittings, such as valves and elbows, can be modeled well with a
constant exponent of 2. Straight piping and coil tubing seldom have velocities high
enough for fully developed turbulent flow and usually are in the mixed region
between turbulent and laminar. An exponent of about 1.8 is often used to describe
the average relationship of flow versus pressure drop in common HVAC systems.
In open systems (cooling towers), the static head of the tower is a constant, as is the
DP set point in a variable-flow system where pump speed is controlled to maintain
DP in the distribution system. These constant pressures are represented by raising
the starting point of the curve at the zero flow line to the pressure that remains con-
stant (Figure 3-28).
Figure 3-28 System curve with fixed head.

The system will operate where the pump curve and the system curve inter-
sect (Figure 3-29). This point (gpm and head) is usually different from the
design gpm and calculated head used to select the pump, due to the inaccura-
cies of pump head calculations and/or field changes (e.g., greater or fewer
elbows installed than assumed).
For constant-speed pumps, if the actual head is lower than the calculated
head, one of two actions can be taken: the most efficient option is to trim the
impeller to match the actual requirements. The system curve can be used to
size the new impeller diameter (Figure 3-30). The second option is to partially
Figure 3-29 System curve and pump curve.
Figure 3-30 System curve (for sizing new impeller).

close balancing valves, which simply raises the actual system curve back up to
the design system curve with obvious increases in pump energy usage. Trim-
ming the impeller is required by ANSI/ASHRAE/IES Standard 90.1 with some
exceptions that allow valve throttling (2016). For pumps with VFDs, neither
impeller trimming nor valve throttling is required because the pump speed will
automatically adjust to the reduced head.
If the calculated head is below the actual head for fixed-speed pumps, the
impeller must be replaced with a larger impeller. This may require a larger
motor as well. For pumps with VFDs, the VFD can be operated at higher than
60 Hz to increase flow until the motor and VFD are at maximum current, in
which case they will need to be replaced with a larger motor and drive. There is
no value to increasing the pump impeller diameter in this case; speeding up the
pump with the VFD is less expensive and the result is basically the same.
The system curve modeled with a fixed exponent is reasonably accurate as long
as nothing in the system changes (the geometry is fixed). When two-way valves are
incorporated into the system, the variable pressure drop created changes the system
curve dynamically. So, in a variable-flow system with multiple two-way valves,
there are an infinite number of system curves that may represent the operating con-
dition at any one time. Also, as the flow decreases, the velocities of the water in
piping and coils can fall closer to the laminar flow regime (exponent = 1), reducing
the average exponent and making the system curve more linear. So a two-way
valve system can best be described as a range of system curves (Figure 3-31).
However, for simplicity in energy modeling, the system curve is generally modeled
as having a fixed exponent as if it were a constant-geometry system. It is not known
how accurate this assumption is in representing real two-way valve systems.
Pump Inlet Limitations

The boiling temperature of water is a function of the absolute pressure sur-
rounding the water. In a pump, the pressure at the eye of the impeller can be the
lowest in the system and, depending on the temperature, the water can boil
(vaporize). As the liquid moves through the impeller and gains pressure, the
water vapor collapses back into liquid. This process is called cavitation and
can be very harmful to the impeller. In the field, the pump sounds like it is
pumping gravel. The pressure at the inlet of the pump must therefore be high
enough to prevent the water from boiling.
Manufacturers publish NPSHR curves with pump curves (e.g., Figure 3-25).
Notice that the NPSHR increases when the flow gets higher. The designer must
ensure that the system will have enough net positive suction head available
(NPSHA) to prevent cavitation. NPSHA (in ft) is calculated using Equation 3-5:
0.102 2 2
NPSHA = ------------- P a P vp + V a V s 193 + z a z s H f ,a s (3-5)
s
where s is the specific gravity of the fluid (= 1 for water), P is absolute pressure
(psia), V is velocity (ft/min), and Hf is head loss (ft) due to friction. These are
defined at two points: point a, which is a reference point in the system where
conditions are known, and point s, which is the pump suction. For closed sys-
tems, point a is typically the expansion tank connection; for open systems,
point a is the top of the tower cold-water basin. Pvp is the vapor pressure of the
fluid at the pump suction, which is a function of fluid temperature. For water in
chiller plants, it ranges from 0.25 psia at 60F (typical of a CHW pump suc-
tion) to 0.5 psia at 80F (typical of a condenser water pump suction).
Maintaining a high NPSHA is easily done in closed systems by adjusting
the precharge pressure of the expansion tank (Pa). Because the vapor pressure
is so low in CHW systems, all that is required to ensure adequate NPSHA is to
maintain a positive gage pressure at the pump suction.
In open systems, such as open cooling tower condenser water systems,
many engineers get concerned about maintaining NPSHA and consequently
insist on elevating the tower basin well above the pump suction. However, this
is, in fact, not necessary because atmospheric pressure (Pa) is so high
(14.7 psia at sea level). For instance, assuming the suction line pressure drop
Hf is 2 ft and the pump suction elevation is the same as the basin elevation, the
equation above indicates
2
2.31 0 12 2
NPSHA = ---------- 14.7 0.5 + ------------------------ + 0 2
1 61.3 (3-6)
= 28.5 ft
This is well above NPSHR for common pumps. So, cavitation is seldom an
issue as long as the pressure drop between the basin and the pump suction Hf is
reasonable. See additional discussion in the Pump Installation section that fol-
lows.
What is often mistaken as cavitation is another phenomenon that occurs
when the pump is not elevated well below the basin: air that is dissolved in the
water in the tower basin comes out of solution as the pressure drops when
water flows into the pump suction. These air bubbles can make a gurgling
sound that is similar to (but not as loud as) the sound of cavitation. It is gener-
ally harmless to the pump, although the pump efficiency can be slightly
reduced. To completely prevent this from occurring, the pump must be below
the tower basin by an elevation equal to or larger than the NPSHR plus the
pressure drop of the inlet piping Hf. This is almost never practical when the
tower and pumps are located on the same floor of the building (e.g., the roof).
Because the air bubbles do no harm to the pump, it does not make sense to
spend a lot of money raising the tower elevation to ensure the bubbles do not
occur.
Another pump inlet problem to be avoided is vortexing. Any time water is
drawn from an open tank or sump, there is a potential that a vortex will occur.
Vortexing causes air to enter the pump suction line and decreases the effective-
ness of the pump. Vortexing occurs even with very deep sumps. Any time
water is drawn from a sump or open tank, anti-vortexing devices should be
installed. These are standard devices supplied with cooling towers.
Figure 3-31 Range of system curves for two-way valve system.

When using vertical turbine pumps, care must be taken to ensure that man-
ufacturer recommendations are followed to maintain a minimum submergence
distance above the inlet bell. Also, adequate clearance must be maintained
from the tanks bottom to the pumps inlet.
Radial Thrust
When pumps operate at points on the pump curve other than BEP, non-
uniform pressures can develop on the impeller. This is called radial thrust and
can cause shaft deflection, excessive wear on pump bearings, and even shaft
failure if high radial thrust is maintained for long periods of time. Radial
thrust is greatest when pumps are operated at or near the shutoff pressures or
near the end of the curve (Figure 3-32). This is one reason pumps should be
selected close to the BEP (Figure 3-27). The use of VFDs on variable-flow
systems also reduces radial thrust, because they keep the pump operating
near the BEP; when constant-speed pumps ride up their curves, radial thrust
increases as shown in Figure 3-32.
Pump Installation
Locating Condenser Water Pumps
Figure 3-32 Radial thrust.
When using an open-circuit cooling tower, the water falls by gravity into
the collection basin and sump. Typically, on a packaged cooling tower, the col-
lection basin and sump are an integral part of the tower. At the outlet to the
pump suction there should be a screen for collecting larger debris and a device
for breaking the vortex created by the suction. These are standard components
provided with packaged towers. From the cooling tower outlet, the water flows
into the suction of the pump with these application considerations:
The pump must be at the same level as or below the tower basin water level. As
discussed in the previous section, the pump need not be located well below the
tower basin to avoid cavitation, because atmospheric pressure is well above net
positive suction head requirements.
Piping from the basin to the pump inlet must have a low pressure drop if the
pump and tower elevation are nearly the same. Cavitation can only occur if the
pressure drop from the tower to the pump inlet has a very high-pressure drop.
That is why no strainer is shown in the piping at the pump inlet in Figure 3-34.
The strainer can get clogged and generate a high enough pressure drop that
cavitation can occur and the pump could be damaged. So, if the pump is
located near the same elevation as the tower basin, strainers should be located
on the discharge side of the pump, typically at the inlets to each chiller con-
denser. The pump impellers will be adequately protected by the screen pro-
vided with the tower at the basin outlet. That will keep any large debris from
impacting the impeller; small particles will pass through the pump without
causing any damage.
Piping from the basin to the pump inlet should be at or below the basin water
level elevation. If the piping rises above the water level, any air leakage into the
piping can cause it to vent and drain when the pump shuts off. The pump prime
is thus lost and when the pump restarts it will operate dry and be damaged.
The pump must include a check valve if the pump discharge piping rises
above the basin elevation. This will prevent backflow and basin overflow
when the pump shuts off. The piping is vented from the supply nozzles on
the tower supply distribution side.
To avoid trapping air bubbles, reducers should be eccentric with the top flat.
Pump Piping Considerations

The following are installation and operation guidelines that help ensure
proper operation of pumps:
The minimum flow through a pump should be sufficient to remove the heat of
compression (motor input power) with a limited temperature rise to avoid dam-
age to seals or excessive warming of supply water. This is never an issue with
primary CHW and condenser water pumps because minimum evaporator and
condenser flows are so high. It can be an issue on secondary CHW pumps serv-
ing only a few coils, but not for those with VFDs with low minimum speed (see
Variable-Frequency Drives [VFDs] section below) due to low pump power.
A minimum of 4 to 6 diameters of straight pipe should be located
upstream of the pump suction, or a suction diffuser should be installed.
Suction diffusers are essentially turning vanes and can have strainers
installed in them and thus provide double duty. Typically condenser
water pumps follow the straight pipe recommendation because, as noted
above, they should not have strainers at their inlets unless their eleva-
tion is well below the tower basin. CHW pumps typically are fitted with
suction diffusers to reduce costs and space requirements.
Variable-flow pumps should never have balance valves installed on the dis-
charge, as flow balancing is accomplished automatically by varying the
speed of the pump.
When using a combination shutoff, balance, and check valve (also called a
triple duty valve), install an additional shutoff valve downstream so that the
check valve can be maintained. This is also advisable because most triple
duty valves make poor shutoff valves because they require hand wrenches
and multiple turns to close. Alternatively, simply do not use combination
valves, per Chapter 5.
A single pump gage should be used and piped with shutoff cocks to the taps
that are provided with the pump at the suction and discharge. This allows
the gage reading to be more accurately compared to the pump curves to
determine pump operating point, because any error in the gage reading is
canceled out when taking the difference in readings.
Temperature gages and test plugs at each pump connection are of little
value because pumps generate almost no temperature rise across the pump
(only that due to the inefficiency of the pump impeller).
Vibration isolation must be considered for pumps located above or below
noise-sensitive occupied spaces. Twin-sphere EPDM-style isolators provide
vibration dampening from the pump as well as the water flowing through the
pumpstiff, braided-type flex connections are ineffective at dampening
vibrations due to impeller-generated water pulsations. Pumps in remote, ded-

icated plant buildings typically do not need vibration isolation.
Figure 3-33 and Figure 3-34 show typical CHW and condenser water pump
piping details following these guidelines.
Figure 3-33 Typical CHW pump piping detail.
Figure 3-34 Typical condenser water pump piping detail.

Variable-Frequency Drives (VFDs)

One of the greatest improvements in the efficiency of chiller plants is the
result of VSDs, also known as VFDs. (The more commonly used acronym,
VFD, is used herein.) The advent of a cost-effective means to vary the speed of
chiller impellers, tower fans, and pump impellers has meant that greater operat-
ing efficiencies are now possible. Systems are also quieter, provide more stable
temperature control, and have lower maintenance costs due to reduced or elim-
inated cycling and reduced radial thrust.
The adjustable-frequency VFD is an electronic device that works by con-
verting a fixed three-phase voltage and 60 Hz frequency source into a variable
voltage and frequency source. Frequency of the source to the motor controls
the motor speed. In order to keep the required torque of the motor, the voltage
and frequency relationship must be maintained. This is called the volts to hertz
relationship. Most HVAC applications require a variable torque because as the
speed of the motor decreases, the load (torque) also decreases. For these appli-
cations, the drive is most efficient if it has a quadratic volts to hertz relation-
ship, although the linear relationship still works.
The large majority of the drives currently in HVAC systems use PWM
technology. The PWM drive has a fixed diode rectifier that converts the AC
input voltage to fixed DC voltage. The DC voltage is filtered and sent to the
inverter section that changes the fixed DC voltage to variable AC voltage and
changes fixed frequency to variable frequency. The inverter uses power transis-
tors to chop the DC voltage to create the variable output. The transistors are
turned on and off at a variable rate (carrier frequency) to create the variable
output voltage and frequency. PWM VFDs have very high efficiency with little
motor heating, have a constant input power factor, can run at low speeds, and
have reduced audible motor noise (because of high carrier frequency).
VFDs have many benefits:
Energy savings: For CHW and condenser water pumps, the savings is mul-
tiplied because the reduced pump heat also reduces the load on the chiller
or tower. It is also possible to select larger pump impellers to improve
pump efficiency, as noted previously.
Soft start: This reduces bearing wear and belt wear on cooling tower fans.
Reduced radial thrust: Reduces bearing wear.
Reduced noise from fans and impellers: VFDs can create a whiny noise
from the motor windings, but this can be minimized by increasing the car-
rier frequency of the drive. See the Motor Noise section.
Better temperature control: Due to reduced cycling on cooling tower fans
and due to better controllability of control valves on variable-flow pumping
systems.
Pump power and status available to DDC system via network cards: Most
VFDs include network cards (e.g., BACnet MS/TP [ASHRAE 2016c])
that allow them to be integrated into the DDC system for remote monitor-
ing and control without hardwiring input/output (I/O) points. One benefit is
that VFDs include a built-in power meter so pump power is known without
having to install a separate meter.
VFDs do have some disadvantages, including a possible negative impact on

power quality, motor noise, electromagnetic interference (EMI), radio frequency
interference (RFI), potential motor bearing pitting and failure, high-frequency
motor noise (whining from vibrations in the motor windings), and nuisance trip-
ping. The advantages far outweigh the disadvantages in most applications.
Drive System Efficiency

The efficiency of the drive system takes into account electrical losses from
the VFD, the connected motor, and the combination of the two devices. VFDs
have losses in the form of thermal power from the inverter (60%) and rectifier
(30%) and leaking current and power lost in the cooling equipment (10%). The
inverter and rectifier losses are proportional to the motor shaft speed, and the
other losses are fairly constant. Drive losses are 2% to 3% of the nominal
power of the drive (i.e., VFD efficiency is 97% to 98%). The motor losses are
from rotating losses, including friction and iron losses, and resistance losses
caused by the resistance in the internal wiring in both the stator and the rotor.
Motor efficiencies can be obtained from motor manufacturers or simply
assumed to be the minimum efficiencies prescribed by the National Electrical
Manufacturers Association (NEMA) for the motor type.
The actual efficiency of the drive and motor operating together can be cal-
culated if both the motor and drive losses are known over the entire speed
range. The typical performance of the drive and motor, plus a combined curve,
are shown in Figure 3-35. For this figure, data were provided by two popular
drive and motor manufacturers and averaged. The y-axis is percent of full-load
efficiency. So, the combined efficiency of a premium-efficiency 25 hp motor
(93.6% efficiency at 25 bhp) with a VFD (~97% efficiency at 25 bhp) at 20%
load (5 bhp) is about 0.9360.970.8 (from curve) = ~73%.
Fan power as measured by the VFD is shown in Figure 3-36, along with
predicted power, assuming power varies as the cube of the flow adjusted by the
combined drive and motor efficiency curves shown in Figure 3-35. The plot
shows that the motor/drive efficiency curves are conservative at low motor
loads but reasonably accurate above about 40% motor loads. The primary rea-
son for the discrepancy is that the power measured by the VFDs internal
power meter is outgoing power, not incoming power, so it does not include
VFD inefficiency. As shown in Figure 3-35, VFD efficiency is high at higher
loads but falls at lower loads.
Figure 3-36 show that there is little energy saved below about 25% speed;
the cube law savings is offset by the reduced efficiency of the VFD and motor.
However, the low minimum speed does improve temperature control and limit
cycling, so it is beneficial to set the minimum speed set point as low as possible
anyway. See the Minimum Speed Set Point section.
Figure 3-35 Typical VFD and motor efficiency at part load.
Figure 3-36 Fan performancemeasured versus model.
VFD Accessories and Application Considerations

Bypass Starter
A VFD can be equipped with a manual or automatic bypass switch and
starter. Normally this is provided to allow the operation of the motor across the
line (with no speed control) if the VFD fails or while it is being serviced. This
requires two sets of contacts, one on the power side and one on the load side to
isolate the drive for servicing. Care must be taken when operating the motor
across the line. Larger motors (60 hp and above) are usually not started across
the line due to limitations on inrush current. When operating in the bypass
mode, the system may be overpressurized, causing control valves to control

more erratically or to be pushed open if their shutoff capability is less than the
pump head. Bypasses cost about half as much as the VFD itself, and because
VFDs have become very reliable, the cost/benefit of bypasses is questionable
in noncritical plants. Even in more critical applications, bypasses might not add
much value if the equipment served by the VFD (e.g., pump, tower cell fan) is
itself redundant; modern VFDs are probably no more likely to fail than the
equipment they serve.
Motors
Motors powered through VFDs must be able to withstand spike and tran-
sient voltages induced by VFDs. For common pump and fan applications,
motors called VFD ready or inverter ready that comply with NEMA MG-1
(2016), part 31, are sufficient. It is not necessary to use more expensive inverter
duty motors, which are designed for constant torque applications.
It is no longer necessary to specify premium-efficiency motors in the
United States; they are mandated by federal law.
Care must be taken when applying a new VFD to an existing motor. In
some older motors, Class B insulation windings may not be sufficient to handle
the voltage surges and additional overheating caused by the VFD. In these
cases, the motors should be replaced with inverter duty motors or the existing
motors can have a Megger test that will verify the condition of the insulation
on the windings. Contact the VFD supplier; some maintain a database of exist-
ing motors and the expected failure rate when applying a VFD.
Shaft Grounding
VFDs have been shown to create voltage differences between the rotating
shaft and the casing of the motor. Electrical current passes from the shaft
through the bearings into the casing. This electrical current can cause damage
in the form of pitting of the bearing cases. A pattern known as fluting can form,
which will eventually cause premature failure of the bearings.
Experience suggests that the effects of bearing currents from common
mode voltages are seldom appreciable, and, therefore, shaft grounding protec-
tion is not needed provided the following are true:
Motors are inverter ready (i.e., fully compliant with NEMA MG-1 [2016],
part 31).
Motors are 75 hp and smaller.
The installation adheres strictly to the motor and drive manufacturers rec-
ommendations regarding the cabling and grounding, the latter being the
key.
The PWM switching (also called carrier) frequency is below 8 kHz.
VFD speed is not fixed at or near one frequency.
Where these criteria do not apply, shaft ground kits can be installed on the
shaft to neutralize the shaft voltage differences. Installing ceramic bearings in
the non-drive end of motors is further recommended in applications requiring
motors greater than 100 hp in order to mitigate high-frequency circulating cur-
rents.
Minimum Speed Set Point

There are many misconceptions about minimum speed set points on VFDs.
Some designers are concerned about maintaining a minimum speed for motor
cooling. But because pump and fan power falls roughly with the cube of flow,
the motor heat generated at very low loads is small, and the temperature ratings
of motors are very high (about 390F). So, for variable-torque applications like
pumps and fans, there simply is no minimum speed required for motor cooling.
The minimum speed should be set in the field as that required to cause the
pump or fan to barely move: simply manually raise the VFD speed output until
the pump or fan can be seen to move. This is typically about 5% to 10% speed.
If concerns remain about having minimum speeds this low, consider using 10%
as the minimum, because almost all major motor manufacturers advertise 10:1
turndown for VFD-ready motors used in variable-torque applications. (As
noted previously, for gear-drive cooling towers, a higher minimum speed may
be required to maintain crankcase oil flow.)
For typical CHW and condenser water pumps, the minimum speed is never
reached because minimum flow requirements through evaporators and con-
densers are so high. But for pumps used to control flow through cooling coils
(see Chapter 4) and for cooling tower fans, more stable control is provided if
minimum speed is set as low as possible, because it minimizes the need to
cycle the motor.
VFD Location and Enclosure Type

VFDs should always be located indoors wherever possible. Those that must
be located outdoors require NEMA 3R enclosures. But even with these enclo-
sures, VFD service life will not be as long if located outdoors. Capacity and
efficiency can also degrade due to very high temperatures that can occur on a
sunny rooftop location. Concerns about the length of wiring between the drive
and motor are seldom valid arguments for locating VFDs outdoors; the maxi-
mum length recommended by most manufacturers is 300 ft or more, based on
manufacturer.
Motor Noise
PWM VFDs cause motors to hum due to vibrations in the motor windings.
This can be mitigated by increasing the PWM switching (also called carrier)
frequency. Most VFD manufacturers offer multiple switching frequency set
points. But high switching frequency has several negative impacts: it increases
the likelihood of shaft grounding issues and motor insulation failures, maxi-
mum current is reduced, maximum conductor length between the VFD and the
motor is reduced, and motor/VFD efficiency is reduced. So, raising carrier fre-
quency should be the last resort for solving motor noise problems. Here is a
suggested procedure for setting switching frequency:
1. Set to the standard frequency recommended by the manufacturer (typically

4 kHz), then check for motor noise in nearby occupiable spaces.
2. If motor noise is audible in occupied spaces, enable the noise smoothing

feature offered by some VFD manufacturers. This feature randomly
changes the switching frequency, resulting in more of a white noise effect
rather than a continuous tonal hum.
3. If noise is still a problem, raise switching frequency to 8 kHz.
4. Raising the switching frequency above 8 kHz should be done only as a last
resort and only if motor shaft grounding is provided.
Input Line Reactors

Most modern VFDs include line reactors or DC chokes to reduce harmon-
ics to the power line and as protection from AC line transients. They are ade-
quate at controlling harmonics in almost all applications. The exception may
be dedicated chiller plant buildings housing all-variable-speed plants. In this
case, almost every motor has a VFD so harmonic noise may be an issue. In this
case, the primary VFD supplier or the electrical engineer should perform an
IEEE 519 (2014) analysis. The study should consider not only the utility point
of common coupling but also other portions of the electrical distribution sys-
tem and the effects on any alternative power sources (e.g., diesel generators).
In addition to the nonlinear load characteristics of VFDs, other nonlinear load
types (e.g., uninterruptible power supplies [UPSs]) should be taken into con-
sideration. If it is determined that line reactors are required, they should be
included in VFD specifications.
Output Load Reactors and Derivative of Voltage

with Respect to Time (DV/DT) Filters
The maximum conductor length between the VFD and the motor recom-
mended by the VFD manufacturer should not be exceeded to prevent excessive
electrical conditions such as voltage reflection and or capacitive coupling.
Exceeding the maximum cable length value can cause damage to the VFD,
motor, and cables. The maximum cable length varies depending on the VFD
manufacturer/model and the selected switching frequency. If the length must
exceed the VFD manufacturers recommended maximum length, discuss the
application with the VFD manufacturer and consider a DV/DT filter in lieu of
an output load reactor, because output load reactors may reduce available
motor torque at full load.
Output Power Wiring

For most applications, wiring between the VFD and the motor should be
separate conductors with XHHW-2 insulation field installed in a single
metallic raceway (e.g., electrical metallic tubing [EMT]), including an inter-
nal equipment ground conductor. Standard PVC-insulated conductors (e.g.,
THHN) should not be used in new VFD installations due to inconsistencies
in the manufacturing process (e.g., air voids) or nicks/cuts in the insulation
during the conductor pulling/installation process, which may lead to corona
and insulation degradation. XHHW-2 insulation is a high-temperature XLPE
insulation that handles the high-voltage spikes from VFD outputs well, pro-
vides superior corona resistance, and will last longer than THHN PVC insu-
lation.
Generators
An IEEE 519 (2014) study should be performed when generators are sup-
porting VFD loads. As a rule of thumb, VFD loads on a generator must be
less than approximately 50% of generator capacity to limit total harmonic
distortion to less than 15%. In addition to the IEEE 519 study, a generator
sizing study should be performed to confirm generator compatibility with the
VFD nonlinear load and other loads that the generator may be supporting
(e.g., UPSs, medical imaging, elevators with regenerative loads). The genera-
tor sizing study should address voltage and frequency transient response,
dips, overshoot, recovery time, leading power factor, and other sizing criteria
at various loading conditions (not just full load) per the application require-
ments for the specific installation. Many of the generator manufacturers can
assist with generator sizing studies with custom software for their specific
generators.
Automatic transfer switches (ATSs) that switch loads from utility power to
generator power and back again should be selected with a programmable tran-
sition delay feature such that the ATS can be programmed to be placed in a
neutral position (not connected to either source) for a minimum of 7 to 10 sec-
onds (or as recommended by the VFD manufacturer) when transferring
between two sources. This will ensure the VFD is not damaged when power is
switched while VFDs are under power. This is recommended even when VFDs
have a flying start feature that allow them to start into a rotating load without
tripping.
References
AHRI. 2011. AHRI Standards 550/590, 2011 Standard for performance rating
of water-chilling and heat pump water-heating packages using the vapor
compression cycle. Arlington, VA: Air-Conditioning, Heating, and Refrig-
eration Institute.
AHRI. 2000. AHRI Standard 560, 2000 Standard for Absorption Water Chill-
ing and Water Heating Packages. Arlington, VA: Air-Conditioning, Heat-
ing, and Refrigeration Institute.
ASHRAE. 2014. Chapter 18, ASHRAE HandbookRefrigeration. Atlanta:
ASHRAE.
ASHRAE. 2015. Fundamentals of water system design. Atlanta: ASHRAE.
ASHRAE. 2016a. ANSI/ASHRAE Standard 34, Designation and safety classi-
fication of refrigerants. Atlanta: ASHRAE.
ASHRAE. 2016b. ANSI/ASHRAE/IES Standard 90.1-2016, Energy standard
for buildings except low-rise residential buildings. Atlanta: ASHRAE.
ASHRAE. 2016c. ANSI/ASHRAE Standard 135-2012, BACnetA data com-
munication protocol for building automation and control networks.
Atlanta: ASHRAE.
ASHRAE. 2016d. ASHRAE HandbookHVAC systems and equipment.
Atlanta: ASHRAE.
IEEE. 2014. IEEE Standard 519-2014, IEEE recommended practice and
requirements for harmonic control in electric power systems. Piscataway,
NJ: Institute of Electrical and Electronics Engineers.
IPCC. 2007. Table 2.14 (Errata). IPCC Fourth Assessment Report: Climate
Change 2007. https://www.ipcc.ch/publications_and_data/ar4/wg1/en/
errataserrata-errata.html#table214.
Calm, James M. 2005. Comparative efficiencies and implications for greenhouse
gas emissions of chiller refrigerants. Non-CO2 Greenhouse Gases (NCGG-4).
Rotterdam: Millpress. http://jamesmcalm.com/pubs/Calm%20JM,%20
2005.%20Comparative%20Efficiencies%20and%20Implications%20for
%20GHG%20Emissions%20of%20Chiller%20Refrigerants.%20NCGG4.pdf.
CBSC. 2016. 2016 California Building Standards Code. California Code of
Regulations, Title 24. Sacramento, CA: California Building Standards
Commission.
NEMA. 2016. NEMA MG-1 2016. Washington, DC: National Electrical Manu-
facturers Association.
Taylor, S. 2012. Optimizing design & control of chilled water plants: Part 5:
Optimized control sequences. ASHRAE Journal 6:5674.

3-1 In the vapor compression refrigeration cycle, refrigerant flows from the com-
pressor discharge, then
a. through the condenser, evaporator, expansion valve, and back to the
compressor.
b. through the condenser, expansion valve, evaporator, and back to the
compressor.
c. through the expansion valve, condenser, evaporator, and back to the
compressor.
d. through the evaporator, expansion valve, condenser, and back to the
compressor.
3-2 Which of the following are valid reasons for a building owner to prefer select-
ing a chiller utilizing R-134a instead of R-123 refrigerant?
i. ODP
ii. Purge unit requirements
iii. GWP
iv. Future refrigerant availability
a. (i), (ii)
b. (i), (ii), (iv)
c. (i), (ii), (iii), (iv)
d. (iii), (iv)
3-3 You are selecting a 300 ton water-cooled chiller. The available compressor
types in this size range include which of the following?
a. Scroll, screw, and centrifugal
b. Scroll and screw
c. Screw and centrifugal
d. Centrifugal
3-4 Manufacturers typically limit evaporator flow rates on the low end to avoid
_________ and on the high end to avoid _________.
a. Surge; noise
b. Deposit formation; erosion
c. Laminar flow; erosion
d. Deposit formation; noise
3-5 Benefits of gear-driven centrifugal chillers relative to direct-drive centrifugal

chillers include the following:
a. Smaller physical footprint
b. The more common incorporation of multiple stages, and thus refriger-
ant economizers
c. Greater flexibility in optimizing compressor efficiency
d. All of the above

e. (b) and (c)
f. (a) and (c)
3-6 You are designing a 300 ton primary-only CHW plant with a 20F CHW T.
The pump type with the best combination of efficiency and cost for this appli-
cation is most likely to be
a. Close coupled, end suction
b. Base mounted, double suction
c. Base mounted, end suction
d. Close coupled, in-line
3-7 A customer expresses concern that the condenser water pumps in a plant
located in San Diego, CA make a gurgling noise. The pumps are installed 3 ft
below the basin water level. The strainers are located downstream of the pumps
and the towers have vortex shedders. What is the most likely cause of the
issue?
a. Cavitation
b. Dissolved air coming out of solution
c. Vortices forming at the tower suction
d. None of the above
3-8 You are designing a 700 ton central plant for a 20-story high-rise building. The
cooling towers are to be located in a well on the roof where the primary con-
straint is roof area. Acoustics are not a concern. The type of tower best suited
to this application is
a. Field erected, induced draft, counterflow
b. Packaged, induced draft, cross flow
c. Packaged, induced draft, counterflow
d. Packaged, forced draft, counterflow
3-9 A cooling tower has been selected for a 15F range with a 6F approach when
the ambient wet-bulb temperature is 72F. At design flow rate and 15F range,
but with an ambient wet-bulb temperature of 62F, the tower is capable of pro-
viding
a. A better approach
b. A worse approach
c. The same approach
3-10 Pump and tower VFD minimum speed set points of 6 Hz or less
a. Result in significant energy savings relative to a minimum speed set
point of 12 Hz.
b. Are not feasible, because they cause motors to overheat.
c. Improve controllability under low-load situations.
d. None of the above.
Hydronic
Distribution
Systems
Instructions
Read the material in Chapter 4. Verify the examples presented in the chap-
ter with your own calculations. At the end of the chapter, complete the Skill
Introduction
This chapter addresses piping layouts and design issues related to CHW
distribution systems and condenser water systems. The first section (Chilled-
Water Distribution Systems) addresses the CHW (evaporator) side of the
chillers, CHW pumps, and cooling coils. The second section (Condenser Water
Systems) addresses the condenser water side of the chillers, including cooling
towers, condenser water pumps, WSEs, and other design options. The third
section addresses optimal plant layout approaches. This chapter also introduces
distribution system fundamentals. Chapter 5 addresses how these systems are
applied to various applications and how components are selected for minimum
life-cycle cost.
Chilled-Water Distribution Systems

The chilled-water distribution system consists of chillers, pumps, piping,
cooling coils, controls and other components on the evaporator side of the
chillers. Understanding how hydronic distribution systems react to varying
loads and how their components interact is essential for designing a reliable,
energy-efficient, and cost-effective CHW plant. This section introduces funda-
mental distribution system types followed by a discussion of system design
considerations. A brief outline follows:
Constant-flow CHW systems
Single or multiple chillers serving a single cooling load
Single chiller serving multiple cooling loads
Multiple chillers (in parallel or series) serving multiple cooling loads
82 Chapter 4 Hydronic Distribution Systems
Variable-flow CHW systems

Concerns about variable flow in evaporators
Primary/secondary variable-flow design
Primary-only variable-flow design
Distributed systems in larger plants
Coil pumping strategies
CHW system design considerations
Primary pump configuration
Balancing variable-flow systems
Degrading T syndrome
Connecting heat recovery chillers
Connecting Thermal Energy Storage (TES)
Constant-Flow Chilled-Water Systems

Single Chiller Serving a Single Cooling Load
With one or more chillers serving a single cooling coil (Figure 4-1), the sim-
plest design strategy is to eliminate the traditional three-way control valve at the
coil and to use a constant-volume pump to circulate water between the evapora-
tor and the coil. Supply air temperature control is provided by resetting the tem-
perature of the chilled water leaving the chiller. While constant CHW flow
results in constant pump energy, chiller performance is improved when the leav-
ing CHW temperature is reset to be as high as possible. A VFD could also be
added to the pump to make the system partially variable flow (minimum flow is
limited by the chiller minimum flow rate), but that adds cost and complexity and
Figure 4-1 Constant-flow system, single chiller, single coil.

Source: Taylor 2011a.
may not be cost-effective, depending on the chillers minimum flow rate and sys-
tem head requirements. Typically in a single-chiller, single-coil plant, pump head
is small because the chiller and coil are close-coupled (physically close together).
Many engineers have concerns about causing high space humidity with this
design because CHW temperatures must be aggressively reset to maintain supply
air temperature at set point under low-load conditions. However, in fact, that is
never a concern, because the leaving supply air condition is about the same
regardless of CHW temperature. For example, if the supply air temperature leav-
ing the coil is 55F, the air leaving the coil is close to saturated whether the CHW
supply temperature is 42F or 50F. It is the supply air temperature set point that
determines supply air humidity conditions, not the CHW temperature.
Chillers must have a sufficient volume of water in the piping system to prevent
unstable temperature swings. This may be an issue with single-chiller, single-coil
systems. Often a small storage tank is required if the chiller is closely coupled to
the coil. The minimum water volume varies with the chiller unloading capabilities
and should be verified with the chiller manufacturer.
Figure 4-1 shows a single chiller, but any number of chillers can be used.
When two chillers are used, this is a good application for piping chillers in
series rather than in parallel. Series chiller piping is discussed further below in
the section Multiple Series Chillers Serving Multiple Cooling Loads.
Single Chiller Serving Multiple Cooling Loads

In a constant-flow system serving multiple coils (Figure 4-2), three-way
valves are used at the cooling coils to modulate the load at each air handler.
Figure 4-2 Constant-flow system, single chiller, multiple coils.

Despite the use of the term constant flow with three-way valve systems, three-
way valves do not actually provide constant flow. This is demonstrated in
Figure 4-3. The first four rows of the table show the system pressure drop by com-
ponent at a fixed flow rate of 100 gpm to the branch circuit. The resulting pressure
drop at the point of connection to the branch circuit varies from 20 ft of head at
both the full (100%) flow and no (0%) flow conditions. The bypass balance valve
is provided in the coil bypass for this reason. At the 50% flow condition, the pres-
sure drops to 11.5 ft of head because there are now two paths for the water to take
and the pressure drop varies roughly as the square of the flow. In the bottom row
of the table, we see what the flow is through the branch if the system pressure is
held at 20 ft of head at the branch circuit point of connection. At both the 100%
flow and 0% flow conditions through the coil, the flow is 100 gpm (design). But at
the 50% valve position, the flow increases to 132 gpm, 32% over design flow.
Thus, at part-load conditions, systems with three-way valves can experience
starved coils at the most remote parts of the system. Using automatic flow-limiting
Pressure Drop (ft) @ 100 gpm

Item
100% to Coil 50% to Coil 0% to Coil
Pipe/valves 2 2 2
Coil and/or bypass 8 2 6
Globe control valve 10 7.5 12
Total 20 11.5 20
gpm @ 20 ft P* 100 132 100
Actual P available may change.
Figure 4-3 Flow variation as a function of valve position.

valves (AFLVs) at each coil is one way to maintain system balance. However, a
better solution is to provide a few two-way valves in the system, as discussed in
the Variable-Flow Chilled-Water Systems section.
Multiple Parallel Chillers Serving Multiple Cooling Loads

Figure 4-4 is the same as Figure 4-2, except there are now multiple chillers.
When the system operates near full load, performance is satisfactory because
all chillers and pumps are operating. However, constant-flow systems can have
problems during part-load or off-peak conditions depending on how coil loads
vary. To demonstrate the point, consider a system like that shown in Figure 4-4,
with two equally sized chillers that serve two equally sized coils with each coil
in turn serving a hotel meeting room. If there are functions in both rooms and
both rooms are at or near full-load, the system operates well: both chillers with
their associated pumps are running and each function space is receiving its
design flow. But a problem can occur if the loads are significantly out of bal-
ance. For instance, what happens when only one of the two function spaces is
occupied (say Coil A) and the other (Coil B) is vacant. The system as a whole
is at half load, so in theory only one chiller and pump could satisfy the load.
However, Coil B will also take its design flow, although it will merely bypass
Figure 4-4 Constant-flow system, multiple parallel chillers, multiple coils.

this flow from the supply to the return. If the plant operates with only one
chiller and pump, it has sufficient load capacity but it cannot meet the flow
demands. One half of the water will flow through Coil A, which is less than it
needs to meet the load, and the other half of the water will flow through Coil B.
To avoid starving Coil A, both pumps will have to operate and both chillers
will have to operate at 50% load. This problem is one of the reasons designers
migrated to variable-flow designs, as discussed in the section Variable-Flow
Chilled-Water Systems.
But this system can work well as long as all coil loads tend to vary in the same
proportion, as they might if all coils served similar occupancies (e.g., all serve
offices on the same schedule). For instance, if the coils served are below half load
and only one chiller and pump are operating, all coils will be capable of meeting
their loads. The system is thus a quasi-variable-flow system in that pumps and
chillers can be staged. Also, because the loads vary similarly, CHW temperature
may be reset aggressively, which allows the plant to be about as efficient as one of
the true variable-flow systems discussed in the section Variable-Flow Chilled-
Water Systems. So, this system is a reasonable choice for small applications with
only a few coils serving similar loads; it is simple and inexpensive and avoids all
of the complexities of variable-flow systems. Also, small systems like this are typ-
ically close-coupled so there is not much pump energy to save. Note that ANSI/
ASHRAE/IES Standard 90.1 only allows this approach for systems with three
coils or fewer or a total CHW pump system power of 10 hp and less (2016).
Multiple Series Chillers Serving Multiple Cooling Loads

One solution to the staging problems associated with constant-flow systems
with multiple chillers is to put the chillers in series where the entire flow passes
through each machine (Figure 4-5). This design was popular in the 1960s and
1970s because it simplified controls. The design eventually lost favor to the par-
allel arrangement (Figure 4-4), which has lower first coststhe series arrange-
ment requires additional piping and valves to allow one chiller to be offline for
service while the other is operational, and parallel flow allows pipe sizes to be
reduced at each chiller connection. Series chillers are now coming back into
favor as an energy conservation measure, in part because of the movement
toward high CHW temperature differences (see Chapter 5). With the series
arrangement, the upstream chiller reduces the water temperature to roughly half
of the overall range (e.g., from 62F to 52F) and the downstream chiller does the
rest (e.g., 52F to 42F). Full-load chiller power is primarily a function of the
temperature leaving the chiller; the entering temperature has almost no impact on
efficiency. Hence, the upstream chiller will be more efficient than the down-
stream chiller, and so the overall chiller power will be less than with two parallel
chillers. However, this savings is at least partly offset by the increased pump
energy because the evaporators are in series. The pump head through the evapo-
rator can be minimized by judiciously selecting the chillers and/or reducing the
number of passes through the evaporators from two (or three) to one (or two),
with some reduction in the downstream chiller efficiency.
Figure 4-5 Multiple series chillers, multiple coils.
Variable-Flow Chilled-Water Systems

Introduction
Variable flow has many advantages in large CHW systems with multiple
chillers and multiple loads or coils. Varying the flow allows the load and flow
to better track, making chiller staging more efficient. Significant pumping
energy can also be saved. This section discusses many important aspects of
variable-flow CHW systems, including the following:
The effect of variable flow through the evaporator

How primary/secondary distribution systems work
The emergence of primary-only variable-flow pumping
Energy efficiency opportunities with distributed pumping techniques
Variable Flow in the Evaporator of a Chiller

Flow in the evaporator can be dynamically varied but within several con-
straints. If a chiller is operating in a stable condition and flow in the evaporator
is reduced, the leaving CHW temperature will drop. If the flow reduction
occurs slowly, the controls will have adequate time to respond and the system
will remain stable. But a rapid change in flow will cause the leaving water tem-
perature to drop quickly. If the controls react too slowly, the chiller may shut
down on low-temperature safety. This is a significant nuisance because the
operator must manually reset the safety control and the chiller must remain off
for a minimum period of time before restarting. Fortunately, modern digital
controls are much more robust than older controllers and can accommodate
changes in evaporator flow as long as they are not too abrupt.
Another issue is avoiding laminar flow in the evaporator tube. A fluid
velocity resulting in ~3000 Reynolds number and above (roughly equal to
about 3 ft per second and above with smooth tubes) is recommended to main-
tain good heat transfer. Tubes enhanced with rifling (grooves) on the inside can
operate at even lower velocities before laminar flow sets in. Consult the manu-
facturers literature for chiller minimum flow rates. The distribution system
design must maintain flow rates through the chiller greater than the minimum.
Because of these issues, designers traditionally chose constant-flow rates
through the evaporators, leading to primary/secondary distribution systems.
However, due to cost and energy advantages (discussed in Chapter 5) coupled
with the success of robust new chiller controllers, variable flow through the
evaporator is rapidly becoming standard practice.
Primary/Secondary Variable-Flow Design

Primary/secondary refers to two hydraulic loops, each with its own pump,
that share a common piping element called the common pipe or common leg. If
the pressure drop in the common leg is very low (Figure 4-6), the two loops are
hydraulically independent, meaning that flow through one loop has little or no
impact on flow in the other loop.
This concept was ideal for early CHW systems. To accommodate wide-rang-
ing coil loads and reduce pumping costs, two-way valve variable-flow distribution
is desirable. However, chillers required stable and (at the time) near-constant flow.
Figure 4-6 Primary/secondary hydraulic independence.

Primary/secondary pumping solved that problem: the primary loop could provide
constant flow through the chillers while the secondary loop could provide variable
flow through the coils (Figure 4-7). Until recently, primary/secondary was the
standard design for central CHW plants with multiple chillers and multiple cool-
ing loads, and it still has many applications, as discussed in Chapter 5.
Primary pumps are typically constant-volume, low-head pumps intended to
provide a constant flow through the chillers evaporator. The secondary pumps
deliver the chilled water from the common leg to coils and then back to the
common leg.
Secondary pumps are typically fitted with VFDs. VFDs are typically cost-
effective except on very small systems. Note that ANSI/ASHRAE/IES Stan-
dard 90.1 requires VFDs on CHW pumps exceeding 5 hp (2016). The VFDs
are controlled by a DP sensor located near the most remote coil so that the DP
set point can be as low as possible; this is also a requirement of ANSI/
ASHRAE/IES Standard 90.1. Locating the sensor near the pump requires a
high DP set point and eliminates most of the energy savings from the VFD. See
additional discussion on pump controls in Chapter 7.
Figure 4-7 Primary/secondary variable-flow piping, multiple parallel chillers, and multiple coils.
It is essential to have the flow rate in the primary loop equal to or greater
than the flow rate in the secondary loop. When the secondary flow exceeds the
primary, return water from the system flows back through the common pipe
and mixes with the supply water from the chillers. This increases the tempera-
ture of the supply water to the secondary system. The warmer supply tempera-
ture in turn causes the control valves at each cooling coil to open even more
(more water is required to meet the coil load when it is warmer), creating an
ever-increasing demand for secondary system flow. Eventually, the secondary
system will be at full flow and coils will be starved. This phenomenon (often
referred to as the death spiral) is well documented in literature, as it has
plagued many CHW systems that suffer from degrading T syndrome. See the
discussion under Degrading T Syndrome for mitigation measures.
The arrangement of the primary/secondary common leg piping connections
is considered important by some engineers. The concern is the migration of the
secondary system return water back to the supply piping, increasing supply
water temperature. When the flow enters into a bullhead tee, the velocity pres-
sure of the water can cause the flow to be forced into the supply. As shown in
Figure 4-8, the tee used in the secondary piping system return should be
arranged so that flow from the bypass is induced into the return system. If this
is not possible, separate the return connection bullhead tee from the secondary
supply connection by three or four pipe diameters. However, this issue is only
important when the primary and secondary flow rates are nearly equal, which
is not a common situation, so in practice the configuration of the piping seldom
has a significant impact on plant performance.
Another issue in primary/secondary piping is the size of the common leg.
In a multiple-chiller system that is properly controlled, the maximum flow in
the bypass should not exceed 110% to 115% of the flow from one chiller. How-
ever, in many cases the bypass line may be the same size as the supply and
return headers for the following reasons:
Figure 4-8 Water flow through tee in common piping.

The first cost of saving material by decreasing the size of the bypass line
can be more than offset by the labor for the extra fittings required.
Lower pressure drop in the common leg makes the two circuits more
hydraulically independent.
In very large systems (e.g., pipes larger than 20 in.), particularly if the com-
mon leg is long, the cost savings achieved by downsizing the common pipe
should be considered.
Another issue is where to locate the common leg. Figure 4-9 shows the com-
mon leg in the traditional front-loaded position (top) and back-loaded position (bot-
tom). The former always results in the two chillers being equally loaded (assuming
they each are controlled to the same leaving water temperature) because they have
the same entering water temperature. But the back-loaded location causes the
(a)
(b)
Figure 4-9 (a) Front-loaded versus (b) back-loaded common leg.

chillers to be unequally loaded; the chiller closest to the return will be fully loaded,
while the one closest to the common leg will handle the rest of the load, often a
very low load at which chillers operate very inefficiently. So the back-loaded com-
mon leg results in less efficient operation for most chiller plant applications.
Primary/secondary systems can also have chillers piped in series. This
could be done using the approach shown in Figure 4-5, but a more flexible and
more energy-efficient design is shown in Figure 4-10. In this case each chiller
has its own common leg. Pumping energy is lower in this scheme than with the
design used in Figure 4-5 at low loads when only one chiller is on, because the
inactive chiller is bypassed, creating no added pressure drop effectively. This
design also allows chillers to be unequally sized, which can improve perfor-
mance at very low loads, common to plants serving AHUs with outdoor air
economizers.
One application for piping chillers in series is coupling an absorption or
engine-driven chiller with an electric chiller. This arrangement allows the oper-
ator flexibility to choose which machine to load based on utility rates or other
criteria. In addition, a series configuration allows the absorption machine to
operate at higher outlet temperatures, which increases its efficiency and allows
it to operate in a cold-water chiller plant, often used in large distribution sys-
Figure 4-10 Primary/secondary variable-flow piping, multiple series chillers, and multiple coils.
tems to increase T and reduce piping costs. Absorption chillers generally can-
not produce colder than 40F chilled water and thus are incompatible with a
plant designed for 38F40F chilled water when piped in a parallel configura-
tion. The series configuration also allows the chillers to be unequally loaded
(if, for example, you wanted to preferentially load the thermal machine during
on-peak times). This is done by adjusting CHW set points; the lower the set
point, the more the chiller is loaded. With this control, either the upstream or
downstream chiller can be preferentially loaded.
Primary-Only Variable-Flow Design

The simplest variable-flow distribution system is shown in Figure 4-11. Two-
way valves are installed at most coils with just enough three-way valves installed to
maintain the minimum flow required by the chiller. This minimum rate, which can
be obtained from the manufacturer, varies with design CHW flow rate and the
chiller type, size, and manufacturer but is typically 25% to 50% of the design flow.
Locating the three-way valves near the chiller minimizes pump energy if
the pump has a VFD. Locating them remotely increases flow to the extremes of
the system, which increases the pump pressure and power required. There is
usually no benefit to locating three-way valves remotely to keep the system
Figure 4-11 Primary-only variable-flow system with three-way valve minimum flow, single
chiller, multiple coils.
cold so that chilled water is instantly available to coils; it typically takes just
seconds or perhaps minutes for water to travel from the chiller to the most
remote coil, and the load will not be lost in that short time. Nevertheless, the
best location for the three-valves is usually at the most remote coils because
this engages the water volume (heat capacity) of the system, which minimizes
chiller short cycling. The design also will retain the self-balancing nature of
two-way valve systems (see Balancing Variable-Flow Systems). These opera-
tional considerations are more important than the small pump energy penalty.
The energy efficiency of a single-chiller primary-only variable-flow system
could be enhanced by providing two-way valves at each coil and maintaining
minimum flow using a flowmeter and bypass valve as shown in Figure 4-12.
Flow and, therefore, pump energy are lower with this design, but with a signif-
icant increase in control complexity of the bypass (discussed further below in
this section and in Chapter 5).
Figure 4-13 shows a primary-only variable-flow system with multiple
chillers in parallel and a minimum flow bypass. Unlike with a single chiller, the
use of three-way valves for minimum chiller flow control is not a good choice
with multiple chillers because of likely chiller staging problems discussed pre-
viously for constant-flow multiple-chiller systems. With multiple chillers, per-
formance is enhanced when flow and load track each other, so all coil valves
should be two-way.
Figure 4-12 Primary-only variable-flow piping with bypass, single chiller, multiple coils.
Figure 4-13 Primary-only variable-flow piping, multiple parallel chillers, multiple coils.
The type and size of the minimum flow bypass valve is critical to the stabil-
ity of the system. A common mistake is to oversize the bypass as if it were the
common leg of a primary/secondary system. This results in unstable, fluctuat-
ing flow through the bypass that can cause fluctuating water temperatures to
the chillers, making their controllers unstable and possibly causing chiller short
cycling. For systems with constant-speed pumps, the valve should be sized for
the high pump head that will occur as the pump rides up its curve at minimum
flow. For variable-speed pumps, the head available at the bypass may be about
as low as the DP set point used to control the pump due to the low flow out to
coils. In both cases, the valve should usually be sized for the minimum flow
rate of the lead chiller; no bypass is likely to be needed when multiple chillers
are operating, except perhaps if they are very unequally sized. The valve
should be a ball, globe, or pressure-independent valve, not a butterfly valve
(which does not have the desired flow characteristic). Flow rate is usually mea-
sured using a flowmeter in the primary line serving the chillers. It is also possi-
ble to deduce flow using a DP sensor across the chillers and correlating
pressure drop to flow using chiller manufacturers curves. This is usually less
expensive, but a flowmeter is more common because flow is typically desirable
for determining plant load used for chiller and pump staging and plant perfor-
mance calculations, as discussed in Chapter 7.
If the bypass valve is controlled only to maintain the minimum flow rate
recommended by the chiller manufacturer, which typically is 25% to 50% of
the design flow rate, pump energy will be lower than for primary/secondary
systems with constant-speed primary pumps. See more discussion of relative
pump energy in Chapter 5.
The bypass valve can be located either adjacent to the chiller as shown in
Figure 4-13 or far out in the system. Locating it close to the chillers provides
the best energy performance because it reduces flow in the distribution piping,
hence reducing pump energy. It also allows the chiller flowmeter and the
bypass valve control to be from the same control panel, ensuring that DDC sys-
tem network delays or failures do not affect control of the bypass valve. On the
other hand, a remote location results in more consistent pressure drop across
the valve, which makes control more stable, and any instabilities are less likely
to cause fluctuating chiller entering water temperature, which can cause chiller
cycling. For campus situations, a more remote location also ensures that the
water in the distribution system is kept cold. For a large campus loop, the time
required to cool down the mass of water in the system can be substantial. How-
ever, keeping the loop cold can be accomplished by other means (e.g., a three-
way valve at the end of the system) and is generally not necessary in single
building systems where chilled water can reach a remote coil in seconds or (at
worst) a few minutes. Because of the energy and control system cost advan-
tages of the close location, it is usually preferred; however, proper valve sizing
and loop tuning are essential.
Large Plant or Building Distribution Systems

Distributed Pumping System
See Figure 4-14. The primary/secondary pumping arrangements described
above have the secondary pumps located near the common piping (within the
central plant) and serving all of the secondary system. While this strategy is rea-
sonable for most systems, it uses more pump energy than necessary on systems
serving coils with widely differing head requirements, such as those serving
large hospitals, airports, and university campuses. For these applications, distrib-
uted secondary pumps as shown in Figure 4-14 can reduce pump energy. In the
traditional primary/secondary arrangement, pressure created by the secondary
pumps must be sufficient to deliver the chilled water to the most remote load or
coil. As a result, the coils located closest to the secondary pumps have excess DP
that is simply throttled by the control valve. This overpressurization not only rep-
resents energy waste, but it strains the ability of the control valve to accurately
modulate water flow and may even cause valves to lift off their seats.
Figure 4-14 Primary/secondary variable-flow piping, distributed pumping.
A distributed pumping system moves the secondary pumps from within the
plant and locates them remotely nearer to the loads they serve. In a single large
building these pumps could be located either at the coils or on distribution
branches that serve a group of smaller loads. In a large system, the distributed
secondary pumps would be located in each building served by the plant. In
either case, the distributed secondary pumps are sized for the pressure drop
needed to move the water from the common pipe at the plant to their most
remote coil within the building and back to the common pipe.
Tertiary Pumping System

See Figure 4-15. A variation of the standard primary/secondary pumping
strategy is to provide standard secondary pumps within the central plant and
locate tertiary pumps remotely within the building, at the loads. Traditionally, a
crossover bridge with a two-way valve would be used at the connection point
between the main distribution piping connections and the tertiary pump
(Figure 4-15, Building A). The crossover bridge hydraulically isolates the ter-
tiary pumping system in the building from the main secondary system. The
two-way valve ensures that the secondary flow through the crossover bridge is
equal to or less than the building flow. This valve is typically controlled to pro-
vide building chilled-water supply temperatures (CHWSTs) 1F warmer than
the secondary loop supply temperature (to ensure that all of the secondary flow
goes to the building).
Figure 4-15 Primary/secondary variable-flow piping, tertiary pumping.
With modern variable-speed technology and control, however, the cross-

over bridge is not necessary with the tertiary pump piped in series with the sec-
ondary pumps (Figure 4-15, Building B). Without the crossover bridge, the
closest buildings can use available DP, with the tertiary pumps simply supple-
menting this when needed. When using a tertiary pump without the crossover
bridge, a bypass line with check valve can be added around the tertiary pump
so that during periods of high DP in the secondary piping, the tertiary pump
can be de-energized. Series-piped tertiary pumps must be controlled with
VFDs with speed controlled to maintain the building coil pressure. This will
ensure that total pressure from the pumps in series does not overpressurize the
building, and it ensures that the tertiary pump does not disrupt the flow in the
tertiary loops of adjacent buildings.
Coil Pumping Strategies

Four coil pumping strategies are discussed here and shown in Figure 4-16.
Figure 4-16 (Coil Pumping Scheme A) shows a variable-speed coil pump
used in place of a two-way control valve; rather than controlling flow by mod-
ulating a valve, pump speed is used. This reduces pump energy and can reduce
first costs when applied to large AHUs in a distributed design approach. This
design is discussed further in Chapter 5.
Figure 4-16 (Coil Pumping Scheme B) shows a primary/secondary coil
pump. This design is used to maintain a constant flow through the coil, which
can be an effective means of coil freeze protection if the coil is exposed to sub-
freezing outdoor air. In this design, the coil pump and plant distribution pumps
Figure 4-16 Coil pumping strategies.

are hydraulically independentflow through the coil is unaffected by the oper-

ation of the distribution pumps or the two-way control valve. The distribution
pumps must have sufficient head to deliver water through the common leg and
back to the plant.
Figure 4-16 (Coil Pumping Scheme C) is an option when the distribution
pumps do not have sufficient head to deliver water to the coil. When the three-
way valve is open to the distribution system, the coil pump is in series with the
distribution pumps and can pull water to the coil. When the valve is closed to
the distribution system, the pump maintains constant flow through the coil
(e.g., for freeze protection). Caution must be applied to this approach because
the pump may reduce or even reverse the DP across the distribution supply and
return lines, possibly starving nonpumped coils.
In Figure 4-16 (Coil Pumping Scheme D), the coil pump is in parallel with
the main distribution pumps. Again, this design is primarily used for coil freeze
protection as in Figure 4-16 (Coil Pumping Scheme B), but it has the advantage
that either the main distribution pumps or the coil pump can maintain flow
through the coil should either of them fail. The pump is usually controlled to
operate when the two-way valve has closed below 50% or so then turn off at
higher primary flow rates.
Coil Pumping Schemes B, C, and D have been used also to ensure the coil
flow rate never falls into the laminar region. The theory was that coil perfor-
mance (T) would degrade as the flow through the coil becomes laminar,
reducing plant T. In fact, with fully circuited coils, T improves at lower
flows and the T never drops below design even in the laminar flow region (see
Figure 4-20). Furthermore, simulations have shown coil pumps use more
energy than they could possibly save even if the T is assumed to degrade dras-
tically at low flows. So, other than perhaps to provide freeze protection (not
typically needed with CHW coils), these configurations should be avoided;
they increase both installed costs and operating costs.
Chilled-Water System Design Considerations

Primary Pump Arrangements
There are three common options for piping primary pumps on a CHW
system:
Option A: Dedicate a pump for each evaporator. See Figure 4-17a.

Option B: Provide a common header at the pump discharge and two-way
automatic isolation valves for each evaporator. See Figure 4-17b.
Option C: Provide a common header with normally closed (NC) manual
isolation valves in the header between pumps. See Figure 4-17c.
The advantages of dedicated pumps for each chiller (Option A) include the
following:
(a)
(b)
(c)
Figure 4-17 Piping options for primary pumps: (a) Option A: dedicated pumps,
(b) Option B: headered pumps with automatic-isolation valves, and
(c) Option C: headered pumps with manual isolation valves.
The pump can be custom-selected for the chiller it serves. Pump selection
can then take into account variations in evaporator pressure drop and flow
rates when chillers are not identical. This can reduce pump energy com-
pared to Option B, where the head of each pump must be the same and
sized for the evaporator with the highest pressure drop; balance valves at
the other evaporators must be throttled to generate this same pressure drop.
Controls are a bit simpler because the pump can be controlled using the
contact provided with the chiller controller. This ensures that the pump
starts and stops when the chiller wants it to. With Option B, the control of
the isolation valves and pumps is by the DDC system and must be coordi-
nated with the needs of the chiller controller to avoid nuisance trips. For
instance, the pumps generally must run for several minutes after the com-
mand for the chiller to stop so that the chiller can pump down the refrig-
erant.
Pump failures do not cause multiple-chiller trips. With dedicated pumps, if
a pump fails, only the chiller it serves will see a flow disruption and trip.
With Option B, all operating chillers will see a flow reduction when a
pump fails, possibly causing more than one chiller to trip due to low flow
or a sudden drop in evaporator temperature. This has become less of an
issue with modern, robust chiller controllers.
The advantages of headered (manifolded) pumps (Option B) include the

following:
Redundancy is improved. With Option A, if a pump fails and a chiller other

than the one it serves also fails (albeit, this is a rare event), then two chillers
will be inoperative. With Option B, any pump can serve any chiller and
under many conditions one pump can provide enough flow for two chillers
to operate near full capacity.
Including a standby pump is much simpler. Adding a standby pump to
Option A is cumbersome and expensive because it requires extensive pip-
ing and manual or automatic isolation valves. If standby pumps are desired,
Option B is the best option.
Isolation valves can be modulating or slow-acting to reduce sudden flow
variations when starting and stopping pumps and chillers. With dedicated
pumps, when a chiller and pump are enabled, there is no flow through the
pump or chiller until the pump speed matches that of the other operating
primary pumpsotherwise, the pump check valve is closed. So, flow
changes can be sudden, which can cause chiller trips with primary-only
systems, as discussed in Chapter 5.
Headered pumps with manual isolation valves (Option C) can have the
advantages of Option A (although this option works best with identical
chillers). It also mitigates the redundancy disadvantage of Option A, although
accommodating a pump failure requires manual manipulation of valves versus
the automatic response in Option B. Including a standby pump is possible with
Option C, but it only works (depending on which pump fails) with the header
isolation valves open, and chillers must be staged by manually opening and
closing their isolation valves.
First costs are usually lowest with Option A if the chiller and pump pairs
are close coupled (physically close together) and the manual isolation valves
between the two are eliminated (each chiller-pump pair is isolated for service
as a pair). Costs can be higher with Option A if pumps are grouped together
and not adjacent to the chillers due to long piping runs from the pumps to their
dedicated chillers. Option C is usually less expensive than Option B, but
Option B is usually the best choice on primary-only variable-flow systems and
where standby pumps are required.
Balancing Variable-Flow Systems

Balance valves are often provided to deliver design flow rates at each coil
or other heat exchange device in hydronic systems. There is no debate about
the need to balance flow in constant-flow (three-way valve) systems. However,
the need to balance variable-flow (two-way valve) hydronic systems is contro-
versial and the subject of many articles and papers.
Balancing of variable-flow systems is intended to do the following:
Ensure adequate flow available at all coils to meet loads. Note that ade-
quate flow may be less than the design flow depending on coil loads.
Ensure DP across control valves is not so high as to cause erratic control
such as two-positioning, where the valve cracks open, provides too much
flow due to the high DP, and then immediately closes after overshooting set
point.
The discussion that follows is based on a detailed analysis (Taylor and

Stein 2002) of a real design of variable-flow chilled- and hot-water systems
using a hydronic analysis program to evaluate the following design options:
1. No balancing (relying on two-way control valves to automatically provide

balancing)
2. Manual balancing, most commonly using CBVs to measure and adjust
flow
3. AFLVs
4. Reverse return
5. Oversized main piping
6. Undersized branch piping
7. Undersized control valves
At the time the referenced study was made, pressure-independent control

valves were only available from one manufacturer, so they were not included in
the study. They are now made by several manufacturers and have become more
cost competitive. Pressure-independent valves generally have two components:
a standard ball- or globe-type control valve controlled by the digital control

system and a self-powered pressure control section that maintains a constant
4 to 5 psi DP across the control valve section regardless of the available DP.
The latter component also typically provides flow-limiting duty, preventing the
valve from using more than design flow. While not included in the detailed
study, pressure-independent valves are included in the discussion and recom-
mendations below.
Table 4-1 summarizes the design valve pressure drops and transient flow
rates of the balancing methods studied in the article.
The columns with maximum pressure drop data are for the condition of all
coils at design flow. The data indicate how much DP the worst-case control
valve has to shed to get design flow. This is an indirect indication of valve con-
trollability (the higher the pressure, the more closed the valve is under design
conditions). Standard electric/electronic control valves typically have 50:1 to
100:1 turndown ratio. This allows the valve to control reasonably stably with a
DP on the order of 30 psi (70 ft), based on the authors experience. Still, lower
DPs are desirable.
The results show that all options result in sufficiently low DPs to provide
reasonable controllability. Reverse return (Option 4) provides the lowest DP
while Option 1 (No balancing) results in the highest DP. Option 3 (AFLVs)
Table 4-1 Pressure Drop and Flow Extremes of Balancing Options

Percent of Design Flow
(Percent of Design Coil Sensible Capacity)
Maximum Pressure
with All Control Valves 100% Open
Drop of Control
Valve Required for
Balancing Method Design Flow, Maximum Flow Minimum Flow
feet through through
Closest Coil Most Remote Coil
CHW HW CHW HW CHW HW

1 No balancing 20.5 44.4 143% 212% 73% 75%
Manual balance using
2 calibrated balancing 0 0 100% 100% 100% 100%
valves
Automatic flow limiting
3 Note 6 Note 6 100% 100% 100% 100%
valves
4 Reverse-return 1.2 10.4 103% 150% 99% 85%
5 Oversized main piping 7.0 20.9 122% 173% 94% 82%
6 Undersized branch piping 19.5 NA 142% NA 73% NA
7 Undersized control valves 8.0 NA 120% NA 86% NA
also results in the same high DPs as Option 1. This is because these valves only
act to prevent flow rates above design; when the flow is below design, as it
always is, other than during rare design load conditions, the valves do nothing.
The last four columns indicate the condition of a transient warm-up or
cooldown scenario where all the control valves are completely open. Propo-
nents of balancing point to this condition as the justification for balancing:
without it, parts of the building may warm up or cool down much more slowly
than others due to flow imbalances. The numbers in each column indicate the
percentage design flow. The numbers in parentheses indicate the percentage
sensible coil capacity at this flow. Coils are very nonlinear so percent flow and
percent capacity do not track each other. For the CHW coil with the lowest cir-
cuit pressure drop, in Option 1 (No balancing), the flow was 212% of the
design flow, but this represented only 119% of design coil capacity. The worst-
case cooling coil had only 73% of the design flow but 89% of its design capac-
ity. The resulting capacities are so close to design (arguably within the accu-
racy of load calculation programs) that it is clear that system warm-up and or
cooldown time will not be significantly affected by these flow imbalances.
Table 4-2 summarizes the pump head, annual pump energy costs, and
incremental first costs for each of the options. The first costs are relative to
Option 1 (No balancing).
Table 4-3 ranks each option with respect to controllability, pump energy
costs, and first costs from best (1) to worst (8).
Table 4-2 Energy and Installed Cost of Balancing Options
Annual Pump Incremental First Costs vs. Option 1

Pump head,
Energy,
Balancing Method ft
$/yr $ $ per design gpm
CHW HW CHW HW CHW HW CHW HW
1 No balancing 58.5 82.7 $1910 $3930
Manual balance
2 using calibrated 60.3 83.6 $1970 $3970 $7960 $47,530 $6.60 $88.00
balancing valves
Automatic flow
3 66.6 90.8 $2170 $4310 $11,420 $50,750 $9.50 $94.00
limiting valves
4 Reverse-return 55.3 80.0 $1810 $3800 $28,460 $17,290 $23.70 $32.00
Oversized main
5 45.0 59.3 $1470 $2820 $12,900 $7040 $10.80 $13.00
piping
Undersized branch
6 58.5 NA $1910 NA ($250) NA ($0.20) NA
piping
Undersized control
7 58.5 NA $1910 NA ($2340) NA ($2.00) NA
valves
Table 4-3 Summary of Balancing Alternatives

Controllability Pump Energy First
Balancing Method
(All Conditions) Costs Costs
No balancing 7 3 3
Manual balance using
4 6 6
calibrated balancing valves
Automatic flow limiting valves 7 7 7
Reverse-return 2 2 5
Oversized main piping 3 1 4
Undersized branch piping 6 4 2
Undersized control valves 5 4 1
Pressure independent control valve 1 7 8
Recommendations for balancing of variable-flow systems are as follows:
For other than very large distribution systems with high pump head,
Option 1 (No balancing) appears to be the best option. It has a very low
first costs, excellent energy performance, and minimal or insignificant
operational problems.
Using AFLVs is one of the most expensive options, yet it offers only the
benefit of reducing imbalances during transients which, as shown in
Table 4-1, has almost no functional impact on performance. Accordingly,
this is generally the worst balancing option for variable-flow systems.
Manual balancing with calibrated balance valves also adds to first costs
with little benefit. They can also introduce coil performance problems
under certain operating conditions (refer to Taylor and Stein [2002] for a
detailed discussion). However, calibrated balance valves can be very handy
for diagnosing flow problems because flow can be readily measured. If
they are provided for this purpose, they should not be used for balancing
(i.e., all valves should be wide open).
For systems with long hours of operation, the added cost of oversized
mains may be cost-effective, based on pump energy savings, but also offers
considerable flexibility for future changes because future loads can be
added anywhere to the system. This is an excellent option for campuses
where future additions can never be well predicted.
Reverse return is effective at reducing DP across control valves and also
can reduce pump head, but the cost can be high and this option is unlikely
to be cost-effective. However, reverse return is sometimes almost free (e.g.,
for floors with a loop distribution, reverse return can be achieved by loop-
ing the CHW supply in one direction, for example, clockwise, while loop-
ing the CHW return in the other, for example, counterclockwise). In these
cases, reverse return should be used.
Undersizing piping and valves on the noncritical runs can reduce first costs
but requires significant additional engineering time. If the branches are
reduced too far, they could become the index runs and cause increased
energy usage. The index run also may vary with varying time schedules
(e.g., a coil may become the index run when other more remote coils are
scheduled off).
For coils that will experience high and varying DPs (e.g., coils nearest to
high-head distribution pumps), consider using pressure-independent control
valves. They will provide more stable control and should reduce the time
required to tune control loops. By providing more stable control, CHW flow
rate can be decreased for the same sensible load (i.e., the same supply air
temperature set point). Theoretically, there is exactly one flow rate (and one
T) that will meet the desired set point given the coil entering air and water
conditions at any moment in time. So, if standard valves could be tuned to
provide stable control, pressure-independent valves would provide no
improvement in flow and T. But because coils are nonlinear, fluctuating
control that results in an average supply air temperature around set point will
result in an average flow rate that is higher than the rate resulting from stable
control, and thus average T will decrease. This can be seen in Figure 4-18,
which shows pressure-independent characterized control valve (PICCV) per-
formance versus a standard pressure-dependent globe valve. Notice on the
right side how the average T of the globe valve is lower than that of the
more stable PICCV. (This figure also shows improved T on the left side
where control for both valve types is stable; this is not supported by physics
and probably reflects some bias by the PICCV manufacturer who performed
these tests.) Figure 4-18 shows a T improvement of about 10%. Coil model-
ing shows that to achieve that savings, supply air temperature of a typical
cooling coil would have to vary by 5F, which is very unstable control. So,
only those valves that are hard to tune due to high and varying DP greater
Figure 4-18 Pressure-independent valve performance (as rated by manufacturers).

than approximately 35 psi (80 ft)as a rule of thumbwill benefit from

pressure-independent control. The first-cost premium for these valves, cur-
rently very high but falling, is hard to justify in other applications.
Degrading T Syndrome
The load is directly proportional to flow rate and T (the difference
between return and supply CHW temperatures):
Q = m c p T (4-1)
In most variable-flow CHW plants, it is assumed that T will remain rela-

tively constant. If T is constant, it follows that the flow rate must vary propor-
tionally with the load.
However, in almost every real-world chiller plant, T often falls well short
of design levels, particularly at low loads. Figure 4-19 shows data from a large
central plant during cool weather (January through March); T is well below
the 10F design T.
As T falls, flow rate must increase and no longer proportionally tracks the
load. This phenomenon, often called degrading (or low) T syndrome, will
impact plant energy performance in several ways:
Pump energy increases as flow increases.
Chillers may have to be prematurely staged on in primary/secondary sys-
tems. As noted previously, primary flow must exceed secondary flow to
Figure 4-19 Example of degrading T syndrome.

avoid the death spiral that results in starved coils. If the secondary flow is
high due to low T, primary pumps and chillers must be staged on prior to
being fully loaded. If the primary pumps and chillers do not have VFDs,
this will increase energy usage.
The plant may be unable to meet coil demands despite adequate chiller
capacity; it may run out of flow capability before running out of chiller
capacity.
For plants with CHW TES, TES storage capacity is reduced, possibly caus-
ing chillers to run in partial or peak demand periods (full storage) or
increasing their load in these periods (partial storage).
Causes of degrading T can be broken into three categories:
Causes that can be avoided by proper design or operation of the CHW

system
Causes that can be mitigated but through measures that may not result in
overall energy savings
Causes that are inevitable and simply cannot be avoided
These are summarized in the following subsections. More details can be

found in Degrading Chilled Water Plant Delta-T: Causes and Mitigation
(Taylor 2002).
Degrading TCauses that Can Be Eliminated by Design/Operation

Improper set points or calibration: For example, dropping coil supply air
temperature set point by only 2F can double the flow rate and halve the T.
Use of three-way valves: Three-way valves should pretty much never be used
in variable-flow CHW systems. Some designers use them to ensure that
water is instantly available at coils, but there is seldom a need: cooling
loads change slowly, and, in a typical building, chilled water will reach a coil
when its valve is opened after seconds or perhaps minutes, but definitely fast
enough to avoid overheating. Three-way valves simply increase flow unnec-
essarily and reduce plant T. They also cost more than two-way valves.
No control valve interlock: It is essential that control valves and associated
control loops be disabled when the associated AHU is shut off. Otherwise
the valve will naturally open as the supply air or room temperature cannot
meet set point. This is often overlooked in DDC system programming.
Coils piped backwards: CHW coils must be piped in an overall counterflow
arrangement entering at the air discharge side of the coil. If piped back-
wards, the coil effectiveness is significantly reduced and the supply air tem-
perature set point can almost never be reached.
Uncontrolled process loads: Plants sometimes provide cooling to process
loads such as medical and biological laboratory freezers. These devices
must have automatic valves that shut off flow when the equipment does not
need it.
Incorrectly selected control valves: Valves must have sufficient actuator

power to shut off against the DP created by the pumps. This is less of a
problem now with electric actuators and the common use of ball valves
instead of globe valvessmall pneumatic globe valves typically have only
about 30 psi of shutoff head capability versus 200 ft for an electric ball
valve. Oversized valves can also be an issue because they result in hunting;
as shown in Figure 4-17, oscillating flow can increase average T. Again,
modern control valves mitigate this problem; they have a rangeability of
50:1 up to 100:1, compared to 10:1 up to 15:1 for large pneumatic valves.
Two-position (on/off) control valves, such as those used to control small
fan-coil units (FCUs), are often blamed for low T problems. If these
valves are not equipped with flow-limiting valves, or piped in a reverse
return arrangement, they may consume more water flow when open than
the design calls for. With full flow through the coil at partial loads, the T
will invariably be lower than design. However, because the air temperature
entering the FCU is fairly constant and is usually not subject to outdoor air
conditions, the T will not degrade significantly.
Incorrectly selected coils: When connecting to a central plant, it is essential
for the designer to select coils that meet the minimum design T for which
the plant was designed. It is not uncommon for engineers to select coils for
a 10F T simply because it is the basis of chiller AHRI standard ratings,
instead of the 20F or more assumed in the plant design. On campuses
where there may be many engineering firms designing buildings over the
years, it can be hard to police their coil selections. Consequently, under-
sized coils are often the most common source of degrading T in campus
CHW plants.
Improper bridge connection and control: Tertiary pump bridge connections
(Figure 4-15) should be controlled off of the CHW supply temperature to
the building at a set point that is a few degrees above the temperature of the
water supplied by the central plant. If the set point is at or lower than the
plant temperature, the bridge control valve will be wide open and water
will recirculate directly to the return, substantially reducing T. Nonethe-
less, a proper set point will not help improve T if the coils downstream are
not maintaining a high T. To resolve this problem, some designers move
the temperature sensor controlling the two-way valve from the supply line
to the return water line. The control valve is then modulated to maintain the
return water temperature at design levels. If return water is too cold, it is
recirculated back into the building in an attempt to make it absorb more
heat. While instinctively it may make sense that recirculating water will
increase the return water temperature, the return water temperature is
driven primarily by the entering air temperature and the coil effectiveness,
not the entering water temperature. Table 4-4 (based on an eight row, 96 fpf
coil designed for 77F entering dry-bulb, 62F entering wet-bulb, and 55F
leaving dry-bulb temperatures) shows that for a given load, increasing
entering CHW temperature results in a lower, not higher, leaving return
water temperature. So, recirculating water not only increases the flow in
Table 4-4 Coil Performance with Increasing CHWST

Entering CHW Leaving CHW
Flow Rate, T,
Temperature, Temperature,
gpm F
F F
42 30 16.7 58.7
44 34.5 14.7 58.7
46 41 12.3 58.3
48 53 9.5 57.5
the building tertiary loop, it also slightly increases flow in the secondary
loop. Furthermore, the same death spiral that occurs with primary/second-
ary systems can occur here: as water is recirculated the T gets worse so
more water is recirculated until the valve is closed and the system is fully
recirculating water. Eventually, in this mode, the return water temperature
will rise but only because coils are starved. Once return water temperature
does rise above set point, the CHW valve will open, but it will soon close as
the return water temperature once again drops. Clearly, this is not a good
control strategy.
Low TMeasures that Improve T but with an Energy Trade-Off

It must be remembered that degrading T is not a problem in and of
itselfit is a problem because it increases plant energy usage. Therefore, solu-
tions that improve T but do not result in an overall reduction in plant energy
seldom make any sense. Here are two examples:
Reducing CHW temperature: Reducing CHW temperature will definitely

improve T; typically a 1F drop in supply water temperature will increase
T by 1.5F to 2F. That will reduce pump energy and perhaps improve
efficiency due to improved chiller staging, but it will increase chiller energy
because chillers are more efficient at higher leaving water temperatures.
Simulations have shown that resetting CHW temperature downwards
always results in a net increase in plant energy, even when pump head is
high (greater than 100 ft). See Chapter 7 for more discussion. So, it is sel-
dom a reasonable strategy to reduce CHW temperature to increase T in an
existing plant. (With a new plant, there are substantial first-cost benefits to
designing for large Ts and subsequent lower CHW temperatures; see
Chapter 5.)
Coil pumps: At design conditions, flow through cooling coils is typically
in the fully turbulent flow regime. But as load falls and flow rate is
reduced, flow will quickly fall into the mixed region and down into the
laminar flow region. Theory suggests that in the laminar flow region, heat
transfer rates will fall dramatically, resulting in reduced coil effectiveness
and reduced T. In fact, most coils do not behave that way. First, laminar
flow is interrupted due to tube bends at the end of each row. Second, there
is another factor occurring at the same time that more than offsets this
rise in heat transfer resistance: at low flow rates, the coil is effectively
oversized (i.e., the ratio of flow rate to coil surface area of is low). The
water stays in the coil longer, and more heat transfer occurs, which
causes temperature rise to increase rather than decrease. Figure 4-20
shows T as a function of space sensible load in a variable-air-volume
(VAV) system for three coil types. The effect of increasing coil effective-
ness due to the reduced flow rates overcomes the increase in film resis-
tance so that coil T remains above design T except at the transition
point into the laminar flow regime. Contrary to conventional thinking, T
below the onset of laminar flow increases rather than decreases. But even
if laminar flow did negatively affect coil performance causing degrading
T, adding coil pumps would not be a good mitigation: Figure 4-21
shows pumping energy for a three-chiller plant first using primary/sec-
ondary pumping assuming T degraded proportionally with load, and
second with coil pumps added assuming a constant T. The added energy
of the constant-speed coil pumps more than offsets the reduced secondary
pump energy resulting from improved T. So, coil pumps should never be
used for the sole purpose of improving T.
Figure 4-20 Cooling coil T at part load.

Figure 4-21 CHW pump energythree-chiller plant.
Degrading TCauses that Cannot be Eliminated

Despite the best design, most plants experience some T degradation, par-
ticularly in mild weather. Two reasons are as follows:
Coil heat transfer degradation: Coil heat transfer effectiveness is reduced

by water-side fouling (e.g., slime, scale, or corrosion on the inside of coil
tubes), air-side fouling (e.g., dirt buildup on coil fins), air-side deterioration
(e.g., deteriorating fins), nonuniform air distribution across the cooling coil,
and coil bypass air. Any reduction in coil effectiveness increases the flow
rate of water required to deliver the desired leaving water temperature, thus
reducing T. Coil cleaning, particularly on the air side, should be per-
formed regularly and when fouling is visible.
Air economizers and 100% outdoor air systems: Air-handling systems
designed for high T (above about 14F) that have integrated outdoor air-side
economizers or supply 100% outdoor air will experience degrading T when
the weather is cool but not cold enough to provide 100% of the systems
cooling load. Under these conditions, the air temperature entering the coil is
low, causing correspondingly low return water temperatures. For instance, a
coil might be designed for 80F entering air temperature with a CHW return
temperature of 60F. When the outdoor air temperature is 60F, it is clearly
impossible to maintain a 60F return water temperature. A coil on a VAV sys-
tem designed for 44F chilled water and an 18F T would only be able to
achieve a 11F to 15F T at 55F to 65F outdoor air temperatures.
Designing Chiller Plants to Accommodate Low T

The previous discussion shows that degrading T syndrome is caused by
many conditions, most of which can be avoided by careful design and mainte-
nance practices. There are, however, situations where degrading T syndrome
is inevitable. Therefore, the plant must be designed to accommodate some T
degradation while still operating efficiently.
Coil effectiveness degradation can be mitigated by selecting chillers so that
they can operate at ~1F lower water temperature than that used to size the
coils. For instance, if coils are selected for 43F (see Chapter 5 for how this is
determined), the chiller could be selected for 42F supply temperature at the
same T. This will allow the chiller to provide colder water on peak days with-
out surging or tripping on high lift.
The increase in pump energy caused by reduced T cannot be avoided. So, the
issue is: how can the plant be designed to avoid the energy impact of premature
staging of chillers and primary pumps? Common solutions include the following:
Put a check valve in the common leg of primary/secondary systems.
Use variable-speed chillers and pumps.
Use primary-only distribution systems.
Each of these is discussed in more detail in the following subsections.
Mitigating Degrading TCheck Valve in Common Leg

When a chiller plant has multiple machines piped in parallel and the T is
lower than design, the result is that one or more additional chillers must be acti-
vated, not because of load but because of flow. This causes an increase in primary
CHW pump energy and condenser water pump energy as pumps are added to serve
the newly activated chiller, and it results in more chillers operating at lower load,
which reduces chiller efficiency for fixed-speed chillers. To mitigate this phenome-
non, flow through the evaporators on the primary side must be increased to match
the secondary flow. One way to do this is to insert a check valve in the common leg
(Figure 4-22). If a low T situation occurs and the secondary flow increases above
the primary flow of one chiller, the check valve will close and cause the secondary
pump to drive the water through the primary pump (i.e., the primary and secondary
pumps will be in series). The increase in additional flow through the primary pump
depends on where the primary pump was operating on its pump curve, the steep-
ness of the pump curve, and the excess horsepower of the secondary pumps. A pri-
mary flow-rate increase of 25% to 40% is usually possible. The pressure drop in
the primary circuit increases roughly as the square of the flow, and pump power
increases roughly with the cube of the flow, so flow increases are generally limited
by available secondary pump horsepower. Adding a check valve is an excellent ret-
rofit opportunity for existing plants with fixed-speed chillers that suffer from
degrading T syndrome.
Using a check valve in this manner is controversial. Some concerns are
largely unsubstantiated, such as the following:
Figure 4-22 Check valve in common pipe.
The check valve makes the primary and secondary circuits hydraulically
dependent when the secondary flow rate exceeds the primary flow rate.
Some purists find this heresy, but there is no intrinsic value to pressure
independence in modern variable-speed, variable-flow systems.
Evaporator flow may exceed the maximum limit published by chiller manu-
facturers. This is very unlikely to be true in practice, particularly if chillers
are selected for a high T as recommended in Chapter 5. The amount of flow
increase is limited by the cube law increase in pump power, as noted previ-
ously. But even if the maximum is exceeded, it should be noted that the max-
imum flow rate established by chiller manufacturers is largely an arbitrary
limit based on rules of thumb to limit erosion; the chiller will not suddenly
fail when that limit is exceeded. Because the hours that the system will be
operating above the maximum will be few, erosion should not be an issue.
When the primary and secondary pumps are in series, the additional pres-
sure will force control valves open, overcooling spaces and wasting energy.
This is simply not true when the secondary pumps have VFDs, as they
always should. The pump controller will simply back off on the pump
speed to maintain the desired DP in the system.
Primary pumps may ride out beyond the end of their pump curves as the
flow increases through them, causing cavitation. Again, due to the limita-
tion in flow increase, this is seldom a problem and is one that can be
avoided by picking pumps with lots of room on the right side of their
curves so that they do not fall off the end even at 40% excess flow.
However, there are some real issues with the check valve:
Deadheading secondary pumps: When primary CHW pumps are piped in a
headered arrangement (Figure 4-17, Option B), if primary pumps are off and
chiller isolation valves are closed, the check valve will stop flow through the
secondary pumps and they will be deadheaded. Without flow, an operating
pump will soon cause water within the pump to get very hot, damaging seals.
The solution is to logically interlock secondary pumps to primary pumps in
DDC system software; if primary pumps are off for more than 5 minutes
when secondary pumps are on, shut the secondary pumps off.
Ghost flow through inactive chillers: When primary CHW pumps are
piped in a dedicated arrangement (Figure 4-17, Option A or C), the check
valve can cause flow to be pushed through inactive primary pumps and
chillers because the primary pump check valves allow flow in that direc-
tion. This water bypasses the operating chillers, raising the overall leaving
water temperature, which can exacerbate the degrading T problem. The
solution is to always use headered pumps (Figure 4-17, Option B) when
using a check valve in the common leg.
Ghost flow through inactive coils with coil pumps: When coils are con-
trolled by variable-speed coil pumps in lieu of control valves (Figure 4-16
[Coil Pumping Scheme A]), the pressure drop across the check valve can
induce a positive DP across the secondary loop, pushing water through
inactive coil pumps and coils. This can add a load to the CHW system if
these coils are part of air handlers operating in air economizer mode. The
solution is to use low pressure drop swing check valves in the common leg,
not spring-loaded silent check valves, which have a higher pressure drop.
Mitigating Degrading TVariable-Speed Chillers and Pumps

As noted in the previous subsection, degrading T can cause an increase in pri-
mary pump energy, condenser water pump energy, and chiller energy if chillers are
staged on prematurely due to high secondary flow rather than load. But this
assumes constant-speed pumps and chillers. If primary CHW pumps and con-
denser water pumps have VFDs, there will be very little increase in pump power
when a new chiller and associated pumps are staged on if these pumps are con-
trolled by flow, as recommended in Chapter 7. The speed of the primary pumps is
varied to match the flow of the secondary system so that the flow in the common
pipe is near zero. And as also explained in Chapter 7, variable-speed chillers are
more efficient at low loads (as long as lift is reduced) than at high loads, thus stag-
ing chillers on prior to their being at full load is deliberate, not premature. Hence,
efficiently accommodating degrading T is inherent in an all variable-speed chiller
plants controlled using the sequences recommended in Chapter 7.
Mitigating Degrading TPrimary-Only Distribution

With primary-only distribution systems, the whole issue of backwards flow
through the common leg disappears because there is no common leg. The chillers
are generally staged on load alone, not flow. If degrading T is severe, it is possible
the chillers will have to be staged on prematurely because the very high-pressure
drop in the operating chillers (remember pressure drop increases as the square of
the flow) can reduce the DP available in the distribution system, potentially starv-
ing coils. However, this should not be the case if the system is well designed (the
inevitable causes of degrading T should not result in such severe degradation),
and the problem is resolved by using variable-speed chillers and possibly variable-
speed condenser water pumps, as recommended in Chapter 5.
Connecting Heat Recovery Chillers

The heat rejected from the condenser of a chiller can be used for many pur-
poses, including domestic water preheating, process heating, and building heating.
Heat recovery chillers are usually sized for a small portion of the total cooling
load because of the need to have a simultaneous mechanical cooling load and
heating load and because of the lower cooling efficiency of heat recovery chillers.
An unsatisfactory strategy for incorporating a heat recovery machine into a chiller
plant is to pipe the chiller in parallel with the other chillers in the primary loop.
The problem with this approach is that constant-flow primary chillers will almost
always have a percentage of the cold supply water bypassed into the return,
thereby decreasing the temperature of the water entering the chiller. This
decreased entering temperature can diminish the heat recovery potential (cooling
load) of the machine. In primary-only variable-flow systems where the flow
through the evaporator is allowed to vary, this is not as much of a concern.
Another method of dealing with a heat recovery chiller is to pipe it for prefer-
ential loading. Figure 4-23 shows a heat recovery machine piped in parallel with
other chillers, but the location of the heat recovery primary pump suction pipe is
such that it receives only the warmest return water from the system. Any bypass
flow will go to the other chillers unless the heat recovery machine is the only one
on. A problem with this approach is that there is no way to effectively unload the
heat recovery machine during times when the heating load is low. The preferen-
tially loaded machine will be required to cool its full volume of warm return
water and, because the need for recovered heat is low, most of the heat will be
rejected out the cooling tower. Mitigation includes providing a variable-speed
primary CHW pump, which will allow the heat recovery chiller to be partly
unloaded down to the minimum flow allowed by the chiller.
Figure 4-24 shows a configuration for a heat recovery machine that puts it
in series with the remaining chillers in the plant. Warm return water is pumped
to the chiller, then back into the return, thereby precooling the inlet water to the
other chillers. The heat recovery machine can remove as much or as little heat
as is needed for the heating load.
Figure 4-23 Preferentially loaded heat recovery chiller.
Figure 4-24 Heat recovery chiller in series.

Figure 4-25 Thermal energy storage (TES).
Connecting Thermal Energy Storage (TES)

Chilled water can be generated during utility off-peak periods and stored
for use during on-peak periods to reduce energy costs. This is called chilled-
water TES. Figure 4-25 shows an elegant way to include a stratified TES tank
into a system: locate the tank in the common leg. In this position, the tank can
be charged and discharged nonsimultaneously (full storage) or simultaneously
(partial storage) without any opening and closing automatic valves.
Condenser Water Systems

Introduction
In designing energy-efficient central CHW plants, it is important to select
the proper condenser water system. The efficiency of the chillers is affected not
only by the operation of the cooling towers and associated pumps but also by
the temperature and quality of the condenser water. In the next sections the fol-
lowing aspects of condenser water systems are discussed:
Options for piping multiple chillers and cooling towers (Piping Multiple
Chillers and Cooling Towers)
Integration of WSEs (Piping for Water-Side Economizers [WSEs])
Piping heat recovery chillers (Piping Heat Recovery Options)
Piping Multiple Chillers and Cooling Towers

Condenser Water Pump Arrangements
The common piping options for condenser water pumps are the same as for
CHW pumps. See Options A, B, and C in Figure 4-17. The advantages and dis-
advantages are also similar. For ease of use, duplicate advantages and disad-
vantages are repeated here.
The advantages of dedicated pumps for each condenser (Option A) include
the following:
The pump can be custom-selected for the condenser it serves. Pump selec-
tion can then take into account variations in condenser pressure drop and
flow rates when chillers are not identical. This can reduce pump energy
compared to Option B, where the head of each pump must be the same and
sized for the condenser with the highest pressure drop; balance valves at the
other condensers must be throttled to generate this same pressure drop.
Controls are a bit simpler because the pump can be controlled using the con-
tact provided with the chiller controller. This ensures that the pump starts and
stops when the chiller wants it to. With Option B, the control of the isolation
valves and pumps is by the DDC system and must be coordinated with the
needs of the chiller controller to avoid nuisance trips. For instance, the pumps
generally must run for several minutes after the command for the chiller to
stop so that the chiller can pump down the refrigerant.
Pump failures do not cause multiple-chiller trips. With dedicated pumps, if
a pump fails, only the chiller it serves will see a flow disruption and trip.
With Option B, all operating chillers will see a flow reduction when a
pump fails, possibly causing more than one chiller to trip due to low flow
or high refrigerant head. However, if there is a lag or standby pump with
Option B that can be started quickly, trips can usually be avoided because it
takes some time for refrigerant head to rise.
The advantages of headered (manifolded) pumps (Option B) include the
following:
Redundancy is improved. With Option A, if a pump fails and a chiller other
than the one it serves also fails (albeit this is a rare event), then two chillers
will be inoperative. With Option B, any pump can serve any chiller and
under many conditions one pump can provide enough flow for two chillers
to operate near full capacity.
Including a standby pump is much simpler. Adding a standby pump to
Option A is cumbersome and expensive because it requires extensive pip-
ing and manual or automatic isolation valves. If standby pumps are desired,
Option B is the best option.
Isolation valves can double as head pressure control valves. See the discus-
sion on head pressure control below in the section Refrigerant Head Pres-
sure Control. For Option A, head pressure control would require the
addition of VFDs on condenser water pumps or tower bypass valves.
It is easier to integrate a WSE. See the discussion on WSEs below in the
section Piping for Water-Side Economizers (WSEs). Because WSEs are
only operational in cold weather when loads are generally low, the con-
denser water side can use one (or more) of the condenser water pumps
serving chillers rather than a dedicated pump. This reduces first costs.
Headered pumps with manual isolation valves (Option C) can have the
advantages of Option A (although this option works best with identical
chillers). It also mitigates the redundancy disadvantage of Option A, although
accommodating a pump failure requires manual manipulation of valves versus
the automatic response in Option B. Including a standby pump is possible with
Option C, but it only works (depending on which pump fails) with the header
isolation valves open, and chillers must be staged by manually opening and
closing their isolation valves.
First costs are usually lowest with Option A if the chiller and pump pairs
are close coupled and the manual isolation valves between the two are elimi-
nated (each chiller/pump pair is isolated for service as a pair). Costs can be
higher with Option A if pumps are grouped together and not adjacent to the
chillers due to long piping runs from the pumps to their dedicated chillers.
Option C is usually less expensive than Option B, but Option B is usually the
best choice where head pressure control and standby pumps are required.
If pumps are constant speed with Option B, some engineers have concerns
about overpumping condensers as pumps ride out their curves due to reduced
losses in common piping. This is almost never a consideration because the
maximum condenser flow rate is seldom reached, particularly if pumps are
selected for a high T as recommended in Chapter 5, and the small increase in
pump energy as the pump rides out its curve is offset by the improved chiller
efficiency due to lower condenser-leaving water temperatures. There is, there-
fore, no need to limit condenser water flow at chiller condensers, and use of
flow-limiting valves is strongly discouraged because they can easily become
clogged in this open-circuit environment.
Equalizer Piping and Maintaining Sump Levels

When piping multiple cooling towers, the water flow rate drawn from the
sump is never exactly equal to the amount distributed into the inlet. This can
lead to an overflowing collection basin with makeup water supplied to other
basins and possibly to air entrainment into the suction piping if water levels get
extremely low. It is virtually impossible to balance flow accurately enough to
prevent over/underflow from occurringeven a few gpm imbalance is enough
over time. To prevent this problem an equalizer must be provided between
sumps, typically one of these two designs:
Equalizer flume gates installed between tower basins: This requires that the
tower basins be physically attached to each other. A removable cover is
usually provided to allow one basin to be separated from the others for
maintenance (e.g., cleaning the basin). Because the cover requires that the
operator climb into the tower basin, this option is not as convenient for
maintenance.
Equalizer piping installed between tower basins: The equalizer line allows
flow by gravity from one basin to the next. Because the force that moves
the water through the equalizer line is the difference in water level between
the sumps (which is sometimes just several inches), it is essential that the
equalizer line be sized for a very low pressure drop. The equalizer line must
be independent of the suction piping due to the various pressure differen-
tials in the suction piping.
Cooling Tower Makeup and Level Control

Typically, makeup water to cooling towers is provided by domestic or
industrial water piped to a float valve in each tower basin. Float valves, which
are similar to the valves in tank-type toilets, are not very reliablethey tend to
get stuck open and leak. An alternative is to install an electronic level switch in
each tower basin wired to an electric slow-closing solenoid makeup water
valve.
Another novel makeup water system design is summarized as follows:
No makeup water connection or assembly is provided with the cooling

towers.
The cooling tower equalizers are fitted with shutoff valves at each cell and a
standpipe is installed in the common equalizer pipe as shown in Figure 4-26.
The standpipe is open at the top and is fitted with an analog level sensor;
Figure 4-26 Cooling tower equalizer standpipe.

capacitance type sensors are typically the most cost-effective. This is tied to
the BAS. The level of water in the standpipe is the same as that in the cooling
tower basins for any basin open to the equalizer. A basin that is valved off
and drained (e.g., for cleaning) has no impact on the level sensor reading.
Makeup water is piped through a slow-closing valve to the condenser water
system in either the supply or return piping to the tower; it does not matter
which, provided the makeup water pressure is higher than that in the piping
and there are no automatic valves or check valves between the connection
point and the towers. Typically, the connection will be indoors near the
chillers. The valve is wired to the BAS.
The level sensor is programmed in the BAS to open the makeup water
valve as required to maintain the tower basin water level within the tower
manufacturers recommended range whenever the condenser water pumps
are on. It can also be programmed to provide high- and low-level alarms.
The advantages of this design include the following:
Makeup water piping need not be piped to each cell and generally need not
be piped outdoors. This is a particular advantage in freezing climates
because it obviates the need for heat tracing and insulating the makeup
water piping.
The analog sensor provides control as well as high- and low-level alarms
with set points easily adjusted in software. False alarms are virtually elimi-
nated; when level switches are provided at each cell for alarming, a false
alarm is generated if the cell is drained for cleaning. (California Title 24
energy standards [CBSC 2016] require a high-level alarm to reduce water
waste; this design meets that requirement.)
Float valves are eliminated.
Costs are lower than electronic makeup water controls, especially if they
include high- and low-level alarms to the BAS. Costs are also typically
lower than float valve makeup water when the system has more than two
cells or is located in a freezing climate.
Maximum and Minimum Flow Rates

When water enters the cooling tower, it is distributed uniformly across the fill
by means of spray nozzles or a gravity distribution basin with properly sized noz-
zles. Each cell of a cooling tower has a maximum and a minimum flow rate. The
maximum flow rate is that required to prevent overflow of gravity distribution
systems or excessive spray through nozzles. The minimum flow rate is that
required to ensure that tower fill is fully wetted. See Chapter 3 for more details.
In plants with multiple cooling towers and chillers, it is desirable to stage
condenser water pumps with the chillers (Chapter 7) so there will be times
when one condenser water pump will operate alone. This will reduce the
flow rate through cooling towers. Options for maintaining minimum flow
rates (Figure 4-27) include the following:
(a)
(b)
(c)
Figure 4-27 Cooling tower cell isolation options: (a) Option A, weir dams and/or low-flow
nozzles; (b) Option B, automatic-isolation valves on supply only; and
(c) Option C, automatic-isolation valves on supply and suction.
Source: Taylor 2011b.
Option A: Select tower weir dams and/or nozzles to allow one pump to
serve all towers. For systems with two or three tower cells this can elimi-
nate the need for isolation valves, which cost much more than the weir
dams and nozzles. This option is also the most efficient; tower energy
usage is minimized by operating as many cells as possible, particularly
when tower fans are controlled by VFDs. This is because fan speed is
reduced (reducing fan power by almost the cube of the speed) and cooling
is achieved through tower cells even when fans are off. With most manufac-
turers and tower types, nozzles and dams are available to reduce flow by
50%, and many can go down to 33% or even 25%, depending on the selec-
tion and design flow rate. Because of low cost and high efficiency, this
option should always be the first choice. When a plant has many tower cells
and automatic isolation valves are unavoidable, the dams and nozzles
should still be selected to allow as many cells to operate as possible.
Option B: Install automatic isolation valves on supply lines only. This
option uses the equalizer to keep basin levels between overflow and fill
lines and will require that equalizers be oversized from that required for
normal duty. For example, assume there are three tower cells and only one
is active; supply flow to the others is shut off. However, water is drawn out
of all three cell basins because the suction lines have no automatic isolation
valves. The water level in the basin of the cell that is supplied will rise
while the other two basin levels will fall. The difference in the two eleva-
tions must provide enough head for water to transfer from the supplied cell
to the others through the equalizer. If the equalizer is undersized, water will
overflow in the supplied cell and the others will be drawn so low that
makeup water valves open, wasting water and water treatment chemicals.
There are only a few inches of elevation difference between the overflow
and fill lines, so it is imperative that the equalizer be properly sized for this
option to work. Another option is to eliminate the basins at each tower and
use a common sump, often located indoors in cold climates. This avoids the
need for equalizer lines entirely but is much more expensive.
Option C: Install automatic isolation valves on both supply and suction
lines. This is usually the most expensive option because automatic valves
are expensive relative to an incremental increase in equalizer size. This
design also increases exposure to a valve failurean oversized equalizer
line has no failure modes. It also increases the risk of freezing (or increases
the energy used by basin heaters) in the basins of inactive cells in systems
that must operate in cold weather. However, this is often the best option
when there are many tower cells that are not located close together (i.e.,
when equalizer lines would be too long for Option B).
Start-Up Conditions
After a condenser water system has been shut down, water will drain by
gravity from the inlet piping into the cooling tower basin. Depending on the
size of the lines, this volume could be enough to overflow the sump, thereby
wasting valuable treated water. Conversely, when the pump starts, there needs
to be enough water in the sump to fill the empty piping without drawing the
volume so low that air is entrained into the suction piping. The amount of water
moved back and forth during start-up and shutdown must be minimized by one
or more of the following options:
Locate the tower at the highest elevation of the system. This prevents water
from draining back to the tower when pumps shut off.
Install inverted traps at the tower return inlet. This is recommended if there
is a significant amount of piping at or just above the elevation of tower
return. The return connection is typically self-venting, so any water in pip-
ing at that level will drain into the tower when the pump stops. An inverted
trap with the top above the highest piping elevation prevents any water
upstream of the trap from draining back to the tower.
Use automatic valves that shut off when the pumps shut off. Valves will be
necessary if there is significant piping and equipment located at a higher
elevation than the towers.
Properly size the sump volume. This is a practical option when the sump is
field built. It is not an option on factory-built towers.
Refrigerant Head Pressure Control

All chillers require a minimum refrigerant head (lift) between the evapora-
tor and condenser. This can be quite high for most screw chillers and some her-
metic centrifugal chillers and very low for magnetic bearing or ceramic bearing
chillers, which have no oil return considerations. There are two common rea-
sons low refrigerant head pressure can occur:
Cold water in the cooling tower basins at start-up: Some chillers can oper-
ate for a short period of time with low start-up head while others will trip
on low head pressure safeties almost immediately. To determine if head
pressure control is required for cold starts, consult with the chiller manu-
facturer.
Integrated WSE operation (discussed below in Piping for Water-Side Econ-
omizers [WSEs]): Head pressure control is almost always mandatory
because cooling tower water temperatures are deliberately kept very cold
for long periods.
Options to avoid low head pressure problems include the following:
Tower three-way bypass valves: The bypass water is diverted around the
tower fill into the cooling tower sump or into the suction piping, thus avoid-
ing the natural cooling that occurs across the tower fill even when tower
fans are off. Piping the bypass to the suction line also avoids the mass of
water in the basin for an even faster warm-up, but the design can be prob-
lematic: unless the bypass line is balanced to create a pressure drop equal to
the height of the cooling tower, air will be drawn into the system backwards
from the spray nozzles because piping above the basin will fall below atmo-
spheric pressure. For staged or variable condenser water flow systems, the
bypass must be balanced at the lowest expected flow rate. This creates a
high-pressure drop and reduced flow if more pumps operate. However,
reduced flow is acceptable when the intent of the bypass is to raise head
pressure. The bypass valve is controlled by supply water temperature typi-

cally with a low limit set point well below the normal set point used to con-
trol tower fan on/off and speed. Tower bypass is most commonly used
where towers must operate in very cold weather to avoid freezing in the fill
itself. The other options below are less expensive and therefore preferred in
other applications.
For systems with dedicated condenser water pumps (Option A or C,
Figure 4-17), VFDs on the pumps can be used to reduce water flow to the
chiller. Head pressure can be maintained even with very cold supply water
as long as the flow rate can be reduced so that the condenser refrigerant
pressure can be high enough (head pressure depends on the condenser
water temperature leaving the chiller, not entering the chiller). Pump speed
can be controlled by the temperature leaving the condenser at a set point
that corresponds to minimum condenser pressure or (preferably) by a sig-
nal from the chiller controller indicating head pressure needsmost chiller
controllers have an analog output dedicated for this purpose.
For systems with headered pumps (Option B, Figure 4-17), the isolation valves
can double as head pressure control valves by converting them from two posi-
tion to modulating. Valve position is typically controlled by the chiller control-
ler head pressure control analog output, either directly or through the digital
control system. This signal will close the valve when the chiller shuts off.
The second two options above reduce flow through the condenser. Many engi-
neers have concerns about low condenser water flow contributing to fouling of the
condenser tubes, but there is little definitive evidence to support the concept that
high velocity keeps tubes clean; strainers and sidestream filters that prevent parti-
cles from entering the condenser in the first place are preferred. However, even if
this is an issue, for most head pressure control applications there are few hours at
reduced flowonly during cold startsso the impact on tube fouling should not
be significant. Low flow through the cooling tower may also be an issue (see dis-
cussion in Chapter 3) but, again, it should not be given the short duration.
Piping for Water-Side Economizers (WSEs)

WSEs are an alternative to air-side economizers. Air-side economizers are
usually more energy efficient, but they are not always practical and can be
much more expensive. Applications where WSEs are often preferred include
floor-by-floor air handlers in a high-rise office building or computer room air
handlers serving a large data center. A WSE uses cold water generated at the
cooling tower to produce chilled water without, or with reduced, mechanical
refrigeration. This is accomplished by running the cooling towers to produce
water temperatures typically 45F and less during periods of low ambient wet-
bulb temperatures. The cold water is pumped through a high-effectiveness
water-to-water HX, usually a plate and frame type, to produce chilled water at
temperatures of 50F or less. The HX protects the CHW system from the cor-
rosion, dirt, and debris typical of open-circuit condenser water.
For detailed design guidance on sizing WSE HXs and flow rates, see Chap-
ter 5.
Figure 4-28 shows a non-integrated WSE where the economizer HX is
piped in parallel with the chiller evaporators on the CHW side. This design
allows the economizer to operate only if the chillers are not operating and vice
versathey cannot operate together. This design was the most common when
WSEs first became popular in the 1980s, but it is not very efficient and is no
longer allowed to be used by energy standards such as ANSI/ASHRAE/IES
Standard 90.1. Instead, WSEs must use an integrated piping arrangement
shown in Figure 4-29 for a primary/secondary system and in Figure 4-30 for a
primary-only system. Integrated systems, which cost only slightly more than
non-integrated systems, allow simultaneous operation of the chillers and the
economizer because the HX is piped in series with the chiller evaporators on
the CHW side. The economizer can provide some precooling of the return
CHW temperature even if it cannot provide all of the cooling. This substan-
tially extends the number of hours the economizer can be operational. Controls
are also much simpler for integrated economizers.
Figure 4-28 Water-side economizer (WSE), non-integrated.

Figure 4-29 WSE, integrated, primary/secondary.

Figure 4-29 shows two options for how to provide flow through the HX. The
least expensive option is to place a two-position valve in the CHW return line. The
valve closes when the economizer is enabled and is open otherwise. This option
requires that secondary pumps have VFDs so that they can slow down when the
HX is out of the circuit and vice versa. The secondary pumps generally do not
need to be sized for the added head of the HX because the HX will be in the loop
only when the economizer is active and cooling loads (and flows) are low. If sec-
ondary pumps are constant speed (rarely true in modern plants) or if the design
flow rate through the HX is much lower than the expected CHW flow during econ-
omizer operation, a sidestream pump should be used instead of the two-position
valve. This sidestream pump is sized with enough head to draw water out of the
secondary return, pump it through the HX, then back to the return.
In both the integrated and non-integrated designs, the HX is generally not
provided with its own condenser water pumps. Because the load will be low
when the weather is cold enough for the towers to deliver cold water, it should
Figure 4-30 WSE, integrated, primary only.

not be necessary to run all chillers, so one of the chiller condenser water pumps
can serve the HX. The HX should be selected so that its pressure drop is simi-
lar to the pressure drop across chiller condensers.
Figure 4-30 includes a bypass WSE-only valve. It is opened when the WSE
is able to provide cold enough water that none of the chillers need to operate,
thus reducing pressure drop through the system.
When using WSEs, refrigerant head pressure control is required (except for
with some magnetic bearing chillers) because of the cold water coming off the
cooling tower. See the discussion in the section Refrigerant Head Pressure
Control regarding head pressure control options.
Piping Heat Recovery Options

Heat rejected from chillers can be used in numerous ways, including pre-
heating domestic hot water andwith the use of double-bundle heat recovery
chillersheating buildings. In the case of preheating domestic hot water, the
condenser water is routed through a double-wall HX that is either an integral

part of a storage tank or is remotely located with a circulation pump to the stor-
age tank (see Figure 4-31).
For double-bundle heat recovery chillers, the piping of the cooling tower
side of the double-bundle condenser involves using a three-way valve that con-
trols the water temperature leaving the heat recovery side of the double-bundle
condenser (see Figure 4-32). In a double-bundle heat recovery condenser, the
hot gas from the compressor first enters the heat recovery side of the condenser
where the buildings heating system removes the heat at a suitable temperature
(105F to 130F). If all of the heat from the chiller is not rejected in the heat
recovery bundle, the leaving heating water temperature (and refrigerant pres-
sure) will rise above set temperature. This will cause the temperature controller
to modulate the three-way valve on the cooling tower side of the condenser to
maintain set temperature. A balance valve must be provided on the bypass line
that goes back to the pump suction.
Centrifugal heat recovery chillers have limited unloading capability when in
the heat recovery mode due to the high condensing temperature level. If the cool-
ing load is small relative to the design chiller capacity, HGBP must be used to
prevent surge, severely increasing energy usage. Therefore, where heat recovery
Figure 4-31 Preheat of domestic hot water.

Figure 4-32 Piping for double-bundle heat recovery chiller.
chillers are used, it is important to install one or more non-heat-recovery chillers.

The more efficient non-heat-recovery chillers can be operated at low loads when
there is insufficient load to keep the heat recovery chiller on without HGBP. They
can also be run when the cooling load exceeds the capacity of the heat recovery
chiller.
Using chiller heat recovery for space heating and using economizers (air
or water) are generally mutually exclusive because the economizers will keep
the chillers from operating in cold weather, so there is no condenser heat to
recover. During cold weather when the heating load is equal to or greater
than the amount of heat rejected from the chillers, it can be shown that using
heat recovery chillers is more energy efficient than air-side economizers and
gas-fired heating systems. In most commercial buildings, the cooling load
will be small in cold weather because only the interior zones need cooling
and their loads are usually relatively small. In this case, heat recovery will be
more efficient than economizers. But as the weather gets milder, heating
loads get smaller and cooling loads can get larger as some sunny perimeter
zones switch from requiring heating to requiring cooling. At this point, econ-
omizers begin to outperform heat recovery systems. On an annual basis,
economizer systems tend to be more energy efficient in mild climates
because of the following:
The heating season is relatively short and mild.

Integrated economizers reduce energy usage even when heating is not
required. For instance, in mild weather (55F to 65F), integrated econo-
mizers reduce the cooling load, which can be substantial because both
interior and perimeter zones require cooling in this case, while heat
recovery systems will do little because heating loads are very small or
nonexistent. In Californias mild climate, for instance, a substantial num-
ber of building operating hours fall into this temperature range. (In very
mild climates like San Franciscos, more than 85% of the operating hours
fall into this range.)
As noted previously, heat recovery chillers have limited unloading capabil-

ity when in the heat recovery mode, often requiring the use of HGBP. When
using HGBP, the chiller is essentially acting like a costly electric resistance
heater.
Heating systems using recovered heat must be designed for low tempera-
tures (e.g., 110F to 130F) and low-temperature differences because of the
limits of the heat recovery condenser. This will increase hot-water flow rates
and may require larger heating coils, increasing both air- and water-side pres-
sure drops and thus increasing fan and pump energy.
If there is a large constant heating load, such as that for domestic hot water
in a hotel, heat recovery will probably outperform economizer systems. A
detailed computer analysis would be required to evaluate the two design
options in this application. It is important to include maintenance costs in the
analysis because heat recovery systems require the chiller to operate all day
long, all year long, increasing the maintenance costs and reducing the service
life of this expensive machine.
It is possible to combine economizers and heat recovery to maximize
energy savings. The heat recovery mode is used, with economizers locked out,
when the heating load is large enough (e.g., when outdoor air temperature
[OAT] < 45F) and the economizer mode is used during milder weather. How-
ever, the first costs of this design can be prohibitive.
Plant Layout
Figure 4-33 shows a CHW plant with primary-only pumping. Conceptually
the design is fine, but the designer of this plant unnecessarily increased first
costs.
First, reverse-return piping was provided for both condenser water and CHW
systems. Reverse return, where the first device supplied is the last returned, is a pip-
ing scheme intended to self-balance hydronic systems but, more importantly, to
keep the pressure drop across modulating two-way control valves relatively low
and equal among all valves as they open and close with changing loads (see the
Balancing Variable-Flow Systems section). However, in this case, the valves at
each chiller are essentially two-position and the difference in pressure across each
valve is very small even if the system is direct return (first supplied, first returned)
and was not manually balanced. A balanced direct-return piping system will result
in the same flow rates across each chiller as the reverse-return design, regardless of
how many chillers are enabled. So, there is no value in performance to reverse
return in this application, yet it substantially increases first costs.
Another often expensive design concept is to group pumps together and
pipe all the pumps first to a larger common pipe before distributing the supply
water to the chillers. In this example, the common pipe size is 12 in. on the
CHW side and 14 in. on the condenser water side. It is not uncommon to see all
the pumps squished into a corner of the chiller room. There is little synergy to
grouping pumps next to each other, and doing so can increase first costs and
reduce space around the pumps for maintenance.
Figure 4-33 Expensive CHW plant.
Figure 4-34 shows the same plant as Figure 4-33 but instead of reverse-return
and grouped pumps, the pumps are aligned with chillers and are piped into a
common header on the discharge side of the pump and also on the discharge side
of the condensers. (The discharge side of the evaporators is the same as that in
Figure 4-33.) This design still allows any pump to serve any chiller, but it short-
ens piping runs and it reduces pipe sizes because there is no common pipe that
serves the total system flow. On the condenser water side, all pipes are 10 in.; the
14 in. common pipe is eliminated. Similarly, the 12 in. CHW piping at the dis-
charge of the pump to the evaporators is eliminated; all pipes are 8 in. First costs
are substantially reduced with no impact on performance.
This layout also usually reduces the footprint of the plant, reducing the
floor area required for the chiller room. In fact, locating the equipment as
close together as possible is key to the first-cost savings. Figure 4-35 shows a
Figure 4-34 Less expensive CHW plant.
floor plan of the plant shown schematically in Figure 4-34. The CHW and
condenser water pumps align with the chillers but are offset to provide tube
pull space and piping connections between the pumps. This is a very compact
layout, but it still provides adequate maintenance access for all equipment.
References
ASHRAE. 2016. ANSI/ASHRAE/IES Standard 90.1-2013, Energy standard
CBSC. 2016. 2016 California building standards code. California Code of Regula-
tions, Title 24. Sacramento, CA: California Building Standards Commission.
Figure 4-35
CHW plant
floor plan.
Taylor, S. 2002. Degrading chilled water plant Delta-T: Causes and mitigation.
ASHRAE Transactions 108(1).
Taylor, S., and J. Stein. 2002. Balancing variable-flow hydronic systems.
ASHRAE Journal 10.
Taylor, S. 2011a. Optimizing design & control of chilled water plants: Part 1:
Chilled water distribution system selection. ASHRAE Journal 6:1425.
Taylor, S. 2011b. Optimizing design & control of chilled water plants: Part 2:
Condenser water system design. ASHRAE Journal 9:1425.

4-1 Aggressive CHW temperature reset
a. Causes issues with thermal comfort because higher supply water tem-
perature yields significantly higher supply air moisture content for a
given set of entering air conditions and thus higher space humidity.
b. Typically increases pump energy usage significantly enough to offset
the benefit of the decrease in chiller energy use.
c. Has little to no impact on space humidity control.
4-2 A plant consists of two identical fixed-speed centrifugal chillers, each with a
dedicated constant-speed primary CHW pump. The chillers supply chilled
water to one large built-up air handler that primarily serves daytime commer-
cial office space loads and another large air handler that serves an auditorium
space most frequently occupied in the evening. Both air handlers have three-
way CHW control valves and are of approximately equal size. Which of the
following are true?
i. The design will require two chillers to operate when the auditorium
air handler is at full load, even if the office air handler is off.
ii. The plant will operate least efficiently in the rare instances that both
the office air handler and auditorium air handler are at near design
load.
iii. The plant will operate most efficiently in the rare instances that
both the office air handler and auditorium air handler are at near
design load.
iv. Headering the primary pumps would increase the controls complex-
ity with no benefit in system redundancy.
a. (i), (iii), (iv)

b. (i), (iii)
c. (ii), (iv)
d. (i), (ii)
4-3 For three-way valve systems

a. The flow rate through the branch serving the coil is constant, irrespec-
tive of valve position.
b. The flow rate through the branch serving the coil peaks when the valve
is fully open to the coil.
c. The flow rate through the branch serving the coil peaks when the valve
is 50% open.
d. Balancing of the bypass leg is never necessary.
4-4 The back-loaded position of the common leg shown below

a. Causes one chiller to be almost fully loaded with the remainder of the
load handled by the other chiller.
b. Causes unbalanced flow through the two chillers.
c. Results in the same energy performance as a system with a common leg
located in the normal position just upstream of the secondary pumps.
d. Is a reasonable location for most plants if it is less expensive due to the
physical layout of the plant.
4-5 Construction of a variable primary CHW plant is just about complete when it is
discovered that the design includes only two-way valves and no means to
maintain minimum flow. Which of the following last-minute design change
options will resolve the problem at minimum cost?
a. Install a CHW bypass locally at the CHW plant. Measure DP across the
chillers to indirectly measure flow.
b. Install a CHW bypass locally at the CHW plant. Install a flowmeter in
the main return line at the plant to measure flow.
c. Install a CHW bypass at the end of the line. Install a flowmeter in the
main return line at the plant to measure flow.
d. Install enough three-way valves at end-of-line coils to maintain mini-
mum flow.
4-6 True or false?

Both headered and dedicated pump per chiller configurations are equally
appropriate for plants requiring a standby pump.
a. True
b. False
4-7 Which balancing method is most appropriate for all but very large distribution
systems?
a. Manual balancing using CBVs to measure and adjust flow.
b. No balancing.
c. AFLVs at all coils.
d. Reverse-return piping.
4-8 True or False: Air-side economizing systems can contribute to low CHW T
issues in systems with high T designs.
a. True
b. False
4-9 What factors constrain the number of cooling towers that can be operated with
a given number of constant-speed condenser water pumps enabled?
a. Minimum per-tower flow requirements.
b. Maximum per-tower flow requirements.
c. Neither (a) nor (b).
4-10 Isolating cooling towers by means of isolation valves on the tower supply pip-
ing only
a. Requires that the equalizer be oversized.
b. Does not require the equalizer to be oversized.
c. Requires that all towers be operated whenever the plant is enabled. The
isolation valves are only installed to prevent tower overflow upon plant
shutdown.
d. Is not a viable design option.
4-11 True or false?
Most modern centrifugal chillers can operate with an integrated WSE without
any means of head pressure control.
a. True
b. False
Optimizing Design
Instructions
Introduction
Previous chapters discussed the basic principles behind central CHW plant
components and distribution system design. This chapter provides procedures
and analysis techniques for optimizing the design to minimize first costs and
operating costs (in particular, energy costs) over the plants life cycle. This
chapter primarily applies to new CHW plants, but many of the techniques can
be used for retrofit projects as well.
To rigorously optimize a central plant design would be a Herculean task
due to the almost infinite number of design decisions that affect energy costs
and first costs. For instance, energy costs are determined by the
full-load and part-load/part-lift efficiency of each piece of plant equipment

(e.g., chillers, towers, pumps),
quantity and staging of each type of equipment,
design of the distribution system (e.g., variable flow versus constant flow,
primary only versus primary/secondary),
control sequences,
pipe and valve sizing,
flow rate sizing,
and more!
First costs can be even more complex to account for during initial design.
There are many reasons for this, including the fact that
costs are not a continuous function of capacity,

capacity for some equipment and materials is only available in discrete sizes,
costs vary by manufacturer and by market conditions, and
costs vary widely depending on the physical layout of the plant and other
design details.
142 Chapter 5 Optimizing Design
Rather than trying to account for every design variable, this chapter sug-
gests a chiller plant design approach that combines detailed analysis and rule-
of-thumb recommendations. This approach provides much better results in
terms of plant performance and cost compared to traditional design procedures
with little or no more effort.
Design Procedure
Rigorous optimization of a CHW system would require a mathematical
model of each possible system component and design option, accurately
describing and defining its operating performance and first costs. Unfortu-
nately, this approach is not practical; there are simply too many options and
their cost and performance cannot always be described by continuous mathe-
matical functions.
Nevertheless, it is possible to partially optimize the chiller plant with a rea-
sonable engineering effort. The key is to break the chiller plant into subsystems
and then optimize within those subsystems. The plant will not be completely
optimized due to the complex interactions between subsystems, but the result
should be close to optimum.
Before beginning the detailed design process, however, it is necessary to
first develop plant load profiles as described in Chapter 2. It is also essential
that the designer be very knowledgeable about chiller plant equipment (see
Chapter 3) and hydronic distribution systems (see Chapter 4).
For most chiller plants, near-optimum plant design can be achieved by the
following step-by-step procedure:
1. Select CHW distribution system.

2. Select CHW temperatures, flow rate, and primary pipe sizes.
3. Select condenser water distribution system.
4. Select condenser water temperatures, flow rate, and primary pipe sizes.
5. Select cooling tower type, speed control option, efficiency, and approach
temperature, and make cooling tower selection.
6. Select chillers.
7. Finalize piping system design, calculate pump head, and select pumps.
8. Develop and optimize control sequences.
Steps 1 to 5 are discussed in this chapter. Chiller selection, Step 6, is dis-

cussed in Chapter 6. Step 7 is discussed in this chapter and Chapter 4. Control
system design, Step 8, is discussed in Chapter 7.
Selecting Chilled-Water Distribution System

Flow Arrangement
Simplified Distribution System Selection
Table 5-1 lists recommendations for life-cycle cost optimum distribution
systems based on the size and number of loads served and the number of chillers.
Life-cycle cost optimum is in quotation marks because these recommendations
are generalizations that should apply to the majority of typical HVAC applica-
tions but may not prove to be optimum for every application and have not been
rigorously proven as the best choice. The recommendations are based on the
authors design and commissioning experience, analysis that was done in con-
junction with development of this manual, work done on an earlier CHW plant
design manual (EDR 1999), and the prescriptive requirements of ANSI/
ASHRAE/IES Standard 90.1 (2016b). The intent is to allow designers to select
the system that is most often the best choice from a life-cycle cost perspective for
a given application without having to perform any lengthy analyses.
Figures of CHW distribution systems are repeated from Chapter 4 for conve-
nience. In each figure, multiple chillers are shown in parallel. For most applica-
tions with twoor a multiple of twoequally sized chillers, the chillers could
alternatively be piped in series. This results in lower chiller energy usage, partly
Table 5-1 Chilled-Water Distribution System

Number of
Application Number of Size of Coils/ Control Recommended
Coils/Loads
Number Chillers Loads Served Valves Distribution Type
Served
Primary only/single coil
1 One Any Any None
(Figure 5-1)
Two way and Primary only/single
2 More than one One Small (100 gpm)
three way chiller (Figure 5-2)
Primary only/
Few coils
More than multiple chillers/few
3 serving Small (100 gpm) Three way
one coils with similar loads
similar loads
(Figure 5-3)
Many coils
Primary-only variable
serving similar
More than flow (Figure 5-4) or
4 loads or Small (100 gpm) Two way
one Primary/secondary
any serving
(Figure 5-5)
dissimilar loads
Primary/distributed
5 More than one Any Large campus Two way
secondary (Figure 5-6)
Primary/coil secondary
6 More than one Any Large coils (100 gpm) None
(Figure 5-7)
offset by higher pump energy usage. However, first costs are usually higher with
series piping due to larger piping and pumps and bypass piping typically pro-
vided to allow one chiller to operate while the other is down for maintenance.
Because of limited funding, the life-cycle costs of series piping were not evalu-
ated, and thus series piping is not included in the recommendations in Table 5-1.
This option will be evaluated for cost-effectiveness in future versions of this text-
book. Note that air-cooled chillers using DX evaporators are not able to achieve
the high Ts recommended below in the section Optimizing Chilled-Water
Design Temperatures. For those applications, series chillers may have lower first
costs than chillers piped in parallel due to lower piping costs and, thus, are likely
to have lower life-cycle costs than a system using parallel piping.
Primary Only/Single Coil

With one or more chillers serving a single cooling coil (Figure 5-1), the sim-
plest design strategy is to not use any control valves at the coil. Instead, a constant-
volume pump circulates water between the chiller and the coil, and supply air
temperature is controlled by resetting the temperature of the chilled water leaving
the chiller. While constant CHW flow results in constant pump energy, chiller per-
formance is improved when the leaving CHW temperature is reset to be as high as
possible. A VFD could also be added to the pump to make the system variable
flow, but that adds cost and complexity. VFDs are seldom cost-effective because
pump power is typically small in a single-coil plant because the chiller and coil
are usually close coupled and it is more efficient to increase CHW temperature
than to reduce pump speed and pump energy. (Chapter 7 further discusses the
trade-off between resetting CHW temperature and pump energy.)
Figure 5-1 shows a single chiller, but any number of chillers can be used.
When two chillers are used, this is a good application for piping chillers in
series rather than in parallel.
Figure 5-1 Primary only/single coil.

Primary Only/Single Chiller

Small CHW plants commonly have a single chiller, typically air cooled.
Single-chiller plants do not have to deal with flow and staging problems com-
mon to multiple-chiller plants and thus can have a simple distribution and con-
trol system. The recommended design is shown in Figure 5-2. It is the simplest
variable-flow primary-only system. Two-way valves are installed at most coils
with just enough three-way valves installed to maintain the minimum flow
required by the chiller. This minimum rate, which can be obtained from the
manufacturer, will vary with design CHW flow rate and the chiller type, size,
and manufacturer but is typically 25% to 50% of the design flow. A VFD is
shown in Figure 5-2; VFDs are typically cost-effective, except on very small
systems. Note that ANSI/ASHRAE/IES Standard 90.1 requires VFDs on
CHW pumps exceeding 5 hp (2016b). The VFD is controlled by a DP sensor
located near the most remote coil so that the DP set point can be as low as pos-
sible; this is also a requirement of ANSI/ASHRAE/IES Standard 90.1. Locat-
ing the sensor near the pump requires a high DP set point and eliminates most
of the energy savings from the VFD.
See Chapter 4 for a discussion of considerations for locating the three-way
and two-way valves.
Figure 5-2 Primary only/single chiller.

Primary Only/Multiple Chillers/Few Coils with Similar Loads

When systems have multiple chillers, chiller staging can be a problem
when flow and load do not track, and they generally do not when three-way
valves are used. But three-way valves (Figure 5-3) can work well with this type
of system so long as all coil loads tend to vary in the same proportion, as they
might if all coils serve similar occupancies (e.g., all serve offices on the same
schedule). For instance, if the coils served are below half load and only one
chiller and pump are operating, all coils will be capable of meeting their loads.
The system is thus a quasi-variable-flow system in that pumps and chillers can
be staged. Also, because the loads vary similarly, CHW temperature may be
reset aggressively, which allows the plant to be about as efficient as one of the
true variable-flow systems discussed below in the next section Primary-Only
Variable-Flow and Primary/Secondary Systems. So, this system is a reasonable
choice for small applications with only a few coils serving similar loadsit is
simple and inexpensive and avoids all of the complexities of variable-flow sys-
tems. Also, small systems like this are typically close coupled, so there is not
much pump energy to save. Note that ANSI/ASHRAE/IES Standard 90.1 only
allows this approach for systems with three coils or fewer or a total CHW
pump system power of 10 hp and less (2016b).
Figure 5-3 Primary only/multiple chillers/few coils with similar loads.

Primary-Only Variable-Flow and Primary/Secondary Systems

Application 4 in Table 5-1 is probably the most common. It applies to sys-
tems serving many small coils (or a few coils with dissimilar loads) and more
than one chiller. In this case, either of two systems is recommended: primary-
only variable flow (Figure 5-4) or primary/secondary (Figure 5-5). Both sys-
tems have positives and negatives, as summarized in Table 5-2.
Primary-only systems always cost less and take up less space than primary/
secondary systems, and, with VFDs, primary-only systems also always use less
pump energy than traditional primary/secondary systems. The pump energy
savings are due to the following:
Reduced system head as a result of the elimination of the extra set of
pumps and related piping and devices (shutoff valves, strainers, suction dif-
fusers, check valves, etc.)
Figure 5-4 Primary-only variable flow.

Figure 5-5 Primary/secondary.

Table 5-2 Advantages and Disadvantages of Primary-Only

Versus Primary/Secondary Systems
Advantages of Primary Only Disadvantages of Primary Only
Lower first costs Complexity of bypass control
Less plant space required Complexity of staging chillers
Reduced pump peak power
Lower pump annual energy usage
More efficient pumps. The primary pumps in the primary/secondary sys-

tem will be inherently less efficient due to their high flow and low head.
This can be partially mitigated by using larger pumps running at lower
speed but with an increase in first costs.
Variable flow through the evaporator, which allows flow to drop below
design flow to some minimum flow rate prescribed by the chiller manufac-
turer. VFDs can be added to the primary pumps of a primary/secondary
system and controlled to track secondary flow down to the chiller mini-
mum flow rate but with an increase in first costs and control complexity.
(This is, however, a cost-effective way to improve the performance of exist-
ing primary/secondary systems and less expensive than converting the sys-
tem to primary only.)
The lower energy costs and lower first costs of the primary-only system
often make it an easy choice versus primary/secondary, but the system does
have two significant disadvantages, as discussed in the following subsections.
Bypass Control Complexities
A bypass valve (shown in Figure 5-4)is required to ensure that minimum flow
rates are maintained through operating chillers. The valve must be automatically
controlled by flow, typically using a flowmeter in the primary circuit (as shown in
Figure 5-4) or DP sensors across chillers correlated to flow. The flowmeter is more
costly but is more easily adapted into plant load calculations, which are necessary
for optimum chiller and CHW pump staging (discussed in Chapter 7).
Selecting the bypass control valve and tuning the control loop is sometimes
difficult because of the widely ranging DPs across the valve caused by its loca-
tion near the pumps. The valve must be large enough to bypass the minimum
chiller flow through it with a pressure drop as low as the DP set point used to
control CHW pump VFDs. This is necessary because if only a few valves are
open in the system, the pressure at the DP sensor location will be nearly equal
to the DP at the plant because there is little pressure drop between these two
points due to the low flow rate. This constraint makes the valve oversized for
other flow scenarios that can occur, so tuning can be difficult. If the control
loop is unstable, cold CHW supply can be fed back into the return intermit-
tently and cause chillers to cycle off due to low load or cold supply water tem-
peratures. However, if the loop is too slow, it may not respond quickly enough
to sudden changes in flow (e.g., when a large number of AHUs shut off at the
same time), resulting in insufficient flow through the chillers, thereby causing
them to trip on low flow or low temperature.
Complex control systems are prone to failure, so at some point in the life of
the plant one can expect the bypass control to fail. A failure of the bypass sys-
tem can cause nuisance chiller trips, which generally require a manual reset. If
an operator is not present to reset the chiller, the plant can be out of service for
some time.
Staging Control Complexities
When one or more chillers are operating and another chiller is started by
abruptly opening its isolation valve (or starting its pump for dedicated pump
configurations), flow through the operating chillers will abruptly drop. The
reason for this is simple: flow is determined by the demand of the CHW coils
as controlled by their control valves. Starting another chiller will not create an
increase in required flow, so flow will be split among the active machines. If
this occurs suddenly, the drop in flow will cause operating chillers to trip.
To stage the chillers without a trip, active chillers must first be temporarily
unloaded (demand limited or set point raised), then flow must be slowly
increased through the new chiller by slowly opening its isolation valve. Then all
chillers can be allowed to ramp up to the required load together. During the stag-
ing sequence, CHW temperatures will rise somewhat. This is seldom a problem
in comfort applications but may be an issue for some industrial applications.
Given these considerations, primary-only systems are most appropriate for
the following situations:
Plants with many chillers (more than three) and with fairly high base loads,
as might be expected in an industrial or data center application. For these
plants, the need for bypass is minimal or nil due to the high base loads, and
flow fluctuations during staging are small due to the large number of
chillers.
Plants where design engineers and future on-site operators understand the
complexity of the controls and the need to maintain them.
The primary/secondary system may be a better choice for buildings where

fail-safe operation is essential or on-site operating staff is unsophisticated or
nonexistent.
Primary/Distributed Secondary
For plants serving groups of large loads such as buildings on a college
campus or terminals in an airport, the primary/distributed secondary system
(Figure 5-6) is usually the best option. Starting from a typical primary/sec-
ondary system as discussed above, the secondary pumps at the central plant
are eliminated and variable-speed pumps are added at each building. The
building pumps are controlled by DP sensors at the most remote coil in each
building. Building pump heads are sized for the pressure drop of the loop
from the plant, to the building, through the buildings coils, then back to the
plant through the common leg. Thus, each pump has a different head custom-
ized for the building.
The advantages of this design compared to conventional primary/secondary
and primary/secondary/tertiary systems include the following:
Overall pump horsepower is reduced. With a conventional primary/second-
ary system, secondary pump head must be sized for the most remote build-
ing (say 100 ft), whereas with a distributed secondary system, the building
pumps close to the central plant can have much smaller heads (say 50 ft).
The system is self-balancing via the speed controls on the secondary
pumps. There is no need to throttle pressure at close buildings and flow
self-adjusts over time as additional buildings are connected to the system.
Overpressurization of control valves located near the central plant is elimi-
nated. With large, high-head secondary systems, these valves must operate
against excess DP, which can reduce controllability and may even force
flow through the valve if it does not have sufficient shutoff head.
Figure 5-6 Primary/distributed secondary.
Pump energy is reduced because of the custom pump heads and the more
precise control of the VFDs. With a conventional primary/secondary sys-
tem, the secondary pumps are typically controlled to maintain DP at the
entry to the most remote building. Thus, the DP set point must be higher
than that for the distributed pump system, which is controlled by the DP at
the most remote coil in each building. At part load, the pumps can therefore
operate at slower speeds.
With primary/secondary/tertiary systems, the tertiary pumps are generally
piped with a bridge and a two-way control valve. Control of the bridge is
always difficult and, if done incorrectly, is often the cause of degrading T
(Taylor 2002). With this distributed pumping system, bridge connections
are eliminated.
The system will be less expensive, more energy efficient, and have lower
maintenance costs than a primary/secondary/tertiary system.
Disadvantages include the following:

Expansion tank pressurization may have to be increased to maintain positive
suction pressure at building pumps if the pumps are located at the top of cam-
pus buildings. This has only a minor cost impact on the expansion tank.
Primary/distributed secondary systems usually cost more than conventional
primary/secondary systems because there are more pumps and space is
required to house the pumps in each building.
Primary/Coil Secondary
For plants serving large individual air-handling systems, using distributed vari-
able-speed-driven coil secondary pumps (Figure 5-7) is usually the best option.
The advantages of this design compared to a conventional primary/second-
ary system include the following:
Connected pump motor horsepower is reduced. This is due in part because

of the customized heads for each pump but also because the control valve is
eliminated. Two-way control valves are typically selected for a wide-open
pressure drop of 27 to 34 kPa, about 10 ft. This is a substantial savings.
The system is self balancing. There is no need for balancing valves of any
kind nor are there any advantages to self-balancing designs such as reverse-
return arrangements.
Pump energy is significantly lower with this design. This is due mostly to the
reduced pump heads but also because there is no need to maintain a minimum
DP in the system as there is with conventional secondary pumps. Because of
this minimum DP and because of the throttling caused by partially closed con-
Figure 5-7 Primary/coil secondary.

trol valves, conventional secondary pumps will not follow the theoretical para-
bolic system curve. Hence, pump efficiency will generally get worse,
particularly at low load. With the variable-speed coil pump design, there are no
control valves or minimum DP, so pump efficiency will be nearly constant.
Control of large control valves is inherently slow due to their size and slow
responsiveness. With the coil pump design, flow can be controlled almost
instantaneously with the VFD, so control is precise. There is also no fear of
overpressurizing control valves with the coil pump design. Such overpressur-
ization can lead to reduced controllability and may negatively impact plant T.
Because of the eliminated control valves and lower pump horsepower, this
system generally has lower costs than a conventional primary/secondary
system. It is usually a little more expensive than a primary-only system.
Control valves can be thought of as brakes on a car while pumps are the car
engines; from an energy perspective, it never makes sense to press both the
brake and the accelerator pedals at the same time, but that is effectively what
systems with control valves do. The primary/coil secondary system is therefore
ideal from a pumping perspective: it has no brakes.
Unfortunately, there are a few disadvantages with this system:
All coils must have a pump. If a coil were connected to the secondary circuit
without a pump, flow through the coil will be backwards from the return to the
supply. For a building that has a mixture of small coils and large coils, pumps
for the small coils will most likely have to be expensive multistage pumps.
Exposure to equipment failure is increased. A control valve is extremely
reliablethe pump and VFD in this design are more likely to fail. Duplex
pumps could be used to improve redundancy, but the cost is prohibitive in
most situations. A good design philosophy is to provide the same level of
redundancy as the rest of the system served. For instance, if the air handler
has only a single fan, then it makes sense to provide only a single pump.
For more critical applications, redundant pumps or an alternative distribu-
tion system design should be considered.
For this design to be energy efficient, coils must be large due to the inher-
ent inefficiency of small pumps, particularly low-flow/high-head pumps.
For instance, a typical pump at 60 ft of head will have an efficiency on the
order of 20% at 10 gpm, 40% at 20 gpm, 50% at 50 gpm, 60% at 100 gpm,
70% at 200 gpm, and 75% to 85% at higher flow rates. That is why this
system is recommended only for coils with flows greater than 100 gpm in
Table 5-1. This flow limit is obviously a rough rule of thumb because effi-
ciency will vary over a range, not drop abruptly below 100 gpm.
If a project includes both small and large coils, a hybrid system of both dis-
tributed coil pumps and conventional secondary pumps to serve small coils is
possible. See Figure 5-8 for an example hybrid plant.
Figure 5-8 Hybrid primary/coil secondary and primary/secondary system.
Optimizing Piping Design

Pipe Sizing
Selecting the optimum pipe size for a given design flow rate is a function of
the location of the pipe in the system (whether or not it is in the critical cir-
cuit, i.e., the circuit that determines pump head);
the first costs of installed piping;
the pump energy costs, which in turn depend on pump and motor effi-
ciency, distribution system type (constant or variable flow), annual flow
profile through the system as well as the pipe in question, type of pump
control (variable speed or riding pump curve), etc.;
erosion considerations (high velocities can contribute to hastening of pipe
wall deterioration);
noise considerations, such as velocity limits to minimize noise caused by
turbulence and the proximity of the pipe to noise-sensitive areas;
physical constraints; and
budget constraints.
Traditionally, most designers size piping using rules of thumb, such as limit-
ing friction rate (e.g., 4 ft per 100 ft of pipe), water velocity (e.g., 10 ft per sec-
ond), or a combination of the two. These methods are expedient and reproducible,
but they seldom result in an optimum design from either a first-cost or a life-cycle-
cost perspective. A better approach is to size piping based on LCCA. This method
is discussed in a 2008 ASHRAE Journal article (Taylor and McGuire 2008) and
can easily be performed using the pipe size optimization spreadsheet provided
free with this textbook (available at ashrae.org/CHWSDL). The spreadsheet is
very easy to use, but it is primarily intended to analyze piping systems that are
completely laid out with all valves, fittings, and other appurtenances fully identi-
fied. It is therefore not as handy to use during the early design phase when these
details are not yet known. An easier tool to use in early design is a simple lookup
table showing maximum flow rates for each pipe size, such as Table 5-3. This
table, which was extracted from ANSI/ASHRAE/IES Standard 90.1 (2016b), was
developed from the LCCA spreadsheet (Pipe Size Optimization Tool v2.0.6
(available at ashrae.org/CHWSDL) assuming a typical distribution system and
ANSI/ASHRAE/IES Standard 90.1 life-cycle cost parameters (see Taylor and
McGuire [2008] for details). The flow rates listed are the maximum allowed by
ANSI/ASHRAE/IES Standard 90.1 for each pipe size using the prescriptive com-
pliance approach. Tables 5-4 and 5-5 are similar tables that the author uses for pre-
Table 5-3 Piping System Design Maximum Flow Rate in gpm

(Table 6.5.4.6 from ANSI/ASHRAE/IES Standard 90.1-2016)
Operating h/yr 2000 h/yr >2000 and 4400 h/yr >4400 and 8760 h/yr
Variable Variable Variable
Nominal Pipe Size, Flow/ Flow/ Flow/
Other Other Other
in. Variable Variable Variable
Speed Speed Speed
2 1/2 120 180 85 130 68 110
3 180 270 140 210 110 170
4 350 530 260 400 210 320
5 410 620 310 470 250 370
6 740 1100 570 860 440 680
8 1200 1800 900 1400 700 1100
10 1800 2700 1300 2000 1000 1600
12 2500 3800 1900 2900 1500 2300
Maximum velocity
for pipes over 8.5 fps 13.0 fps 6.5 fps 9.5 fps 5.0 fps 7.5 fps
12 in. size
liminary design of variable-flow, variable-speed and constant-flow, and constant-

speed pumping systems, respectively. These tables were also developed from the
LCCA spreadsheet assuming California utility rates and fairly aggressive life-
cycle cost parameters for discount rates, energy rate escalation, etc., which the
author believes are appropriate for green buildings ($0.15/kWh, 1% energy esca-
lation rate, 5% discount rate, 30 year life). The maximum flow rates are a bit
lower than those in Table 5-3 accordingly. Tables 5-4 and 5-5 also include a set of
flow limits for piping located in acoustically sensitive areasagain, see Taylor
and McGuire (2008) for details. Table 5-4 and Table 5-5 are useful for selecting
pipe sizes in the early design phase; once the design is more complete, the LCCA
spreadsheet can be used to select final pipe sizes.
Table 5-4 Piping System Design Maximum Flow Rate in gpm for
Variable-Flow, Variable-Speed Pumping Systems
(Developed from LCCA spreadsheet assuming green life-cycle cost parameters)
Non-noise Sensitive Noise Sensitive
Pipe Diameter
2000 4400 8760 2000 4400 8760
1/2 7.8 5.9 4.6 1.8 1.8 1.8
3/4 18 14 11 4.6 4.6 4.6
1 29 22 17 8.9 8.9 8.9
1 1/4 51 39 30 15 15 15
1 1/2 88 67 52 24 24 24
2 120 84 67 51 51 51
2 1/2 160 120 91 81 81 81
3 270 210 160 140 140 140
4 480 360 290 280 280 280
5 670 510 390 490 490 390
6 1100 800 630 770 770 630
8 1800 1400 1100 1500 1400 1100
10 2900 2200 1800 2700 2200 1800
12 4400 3300 2600 4200 3300 2600
14 6000 4600 3600 5400 4600 3600
16 7400 5700 4500 7200 5700 4500
18 10,000 8000 6300 9200 8000 6300
20 11,000 8800 7000 11,000 8800 7000
24 17,000 13,000 11,000 17,000 13,000 11,000
26 21,000 16,000 13,000 20,000 16,000 13,000
Table 5-5 Piping System Design Maximum Flow Rate in gpm for
Constant-Flow, Constant-Speed Pumping Systems
(Developed from LCCA spreadsheet assuming green life-cycle cost parameters)
Non-Noise Sensitive Noise Sensitive
Pipe Diameter
2000 4400 8760 2000 4400 8760
1/2 5.0 3.9 3.0 1.8 1.8 1.8
3/4 12 9.0 7.0 4.6 4.6 4.6
1 19 14 11 8.9 8.9 8.9
1 1/4 34 26 20 15 15 15
1 1/2 57 43 34 24 24 24
2 73 55 44 51 51 44
2 1/2 100 77 60 81 77 60
3 180 140 110 140 140 110
4 320 240 190 280 240 190
5 430 330 260 430 330 260
6 700 530 420 700 530 420
8 1200 900 720 1200 900 720
10 1900 1500 1200 1900 1500 1200
12 2900 2200 1700 2900 2200 1700
14 4000 3000 2400 4000 3000 2400
16 4900 3800 3000 4900 3800 3000
18 7000 5300 4200 7000 5300 4200
20 7700 5800 4600 7700 5800 4600
24 12,000 8900 7100 12,000 8900 7100
26 14,000 11,000 8500 14,000 11,000 8500
Selecting Valves and Fittings

To minimize the energy use of piping systems, piping system pressure drop
must be minimized. The procedures described in the Pipe Sizing section should
result in near-optimum pipe size for straight pipe and fittings. However, acces-
sories such as valves and strainers are also major contributors to system pres-
sure and must be optimized by proper selection. The following
recommendations should be considered.
Valves at Chillers and Towers

CHW plants in general should have only two types of valves for flow isola-
tion and balance: butterfly valves for large piping (typically 3 in. and larger)
and ball valves for smaller piping. These are not only the least expensive types
of valves, but they also are the easiest to use because they require only a quar-
ter turn to open and close (unlike globe and gate valves) and they may be used
for balancing (unlike gate valves). They are also physically smaller than other
valve types. Standard two-piece ball valves come in two types: full port (the
opening in the ball is the same as the pipe size) and standard-port (the hole is
smaller than the pipe size). Full-port ball valves have a lower pressure drop but
cost more than standard-port ball valves. It is cost-effective to use the full-port
valve for valves located in the critical circuit (the circuit that has the highest
pressure drop and that determines the pump head). Otherwise, standard-port
ball valves are the most cost-effective.
Valves to avoid and/or that are not needed include the following:
Globe and gate valves: They provide no advantages over butterfly and ball
valves, and they cost more and require more space.
CBVs: Most plants are almost self-balancing simply because their com-
pact size does not result in large differences in pressure drop across each
evaporator or condenser. Even if no balancing is done, the plant generally
will work wella small difference in CHW or condenser water flow
among chillers and towers has only a small impact on plant performance.
What little balancing is needed can be accomplished by modulating the
condenser/evaporator isolation butterfly valves to cause the pressure drop
across the condenser/evaporator to be the same among the chillers (or
proportionally the same if chillers are not all the same). Cooling tower
balancing is almost never an issue, and there is no need to create equal
flow to each cell; the cells with excess flow will create warmer water and
the ones with low flow will create colder water, but when they are mixed,
the resulting temperature is almost exactly the same as it would be if the
cells had equal flow.
Flow-limiting valves (also called flow control valves): These valves
should never be used in a CHW plant. There are times when exceeding
design flow is actually desired. For example, a chiller plant with
degrading CHW T can be more efficient if higher flow is allowed to be
forced through operating chillers rather than start another. Exceeding
the chiller manufacturers maximum flow rate is rarely an issue because
pressure drop increases with the square of the flow and pumps seldom
have the capability to force this much water through the evaporator or
condenser.
Multipurpose valves (also called triple duty): These valves, typically
located at pump discharges, provide three functions in one: shutoff valve,
check valve, and flow measurement/balance. However, they are hard to
use as shutoff valves because they are multi-turn rather than quarter turn
and a wrench is required; they generally do not include handles. They can
also be hard to repair because they cannot isolate themselves. When sized
the same as the pipe size, as is typical, flow measurement is often not
possible; they must be undersized to provide accurate flow readings,
which adds to pressure drop. Flow balancing is also not needed at the
pump; it can and should be done at the HXs, and throttling flow for bal-
ance is not required at all if the pump has a VFD. Finally, cost savings
versus a check and butterfly valve are minor or nonexistent. Therefore,
triple-duty valves are unnecessary in almost all applications.
Balancing Devices at Coils

As discussed in Chapter 4, two-way valve variable-flow systems do not
require balancing. AFLVs should never be used with variable-flow CHW sys-
tems. Manual CBVs are also not needed, but they can be handy for diagnos-
ing problems because they can be used to measure how much flow is going
through a coil. Where used, CBVs that use ball valves are preferred because
they have a lower pressure drop than those that use globe valves.
Strainers at Coils
Strainers are often installed at coils to protect control valves from fouling.
But strainers located at pumps usually eliminate sufficient debris from the sys-
tem so that problems seldom occur at coils even without strainers. Fine debris
can pass through coils and control valves without causing damage. In fact,
strainers at coils may cause flow problems because they often are not readily
accessible for maintenance and therefore get clogged (particularly during start-
up) and restrict flow.
Optimizing Chilled-Water Design Temperatures

Table 5-6 shows the typical range of CHW temperature difference (com-
monly referred to as delta-T or T) and the general impact on energy usage
and first costs. The table shows that there are significant benefits to increas-
ing T from a first-cost standpoint, and there may be energy cost savings as
well, depending on the relative size of the fan energy increase (due to
increased air-side pressure drop with deeper CHW coils) versus pump energy
decrease as T increases. Chiller energy usage is largely unaffected by T
Table 5-6 Impact on First Costs and Energy Costs of Chilled-Water Temperature Difference
(Assuming Constant Chilled-Water Supply Temperature)
T
Low High
Typical range 8F 25F
Smaller pipe
First-cost impact Smaller coil Smaller pump
Smaller pump motor
Energy-cost impact Lower fan energy Lower pump energy
for a given CHW supply temperature. The leaving CHW temperature drives
the evaporator temperature, which in turn drives chiller efficiency; entering
water temperature has almost no impact on efficiency.
Intuitively, one might think that fan energy would dominate in the energy
balance between fan and CHW pump because fan energy is so much larger
than pump energy annually, and the fan sees the coil pressure drop under all
conditions, while the CHW pump typically only runs in warmer weather
(assuming the system has an air-side economizer). However, detailed analysis
has shown that not to be the case: the impact on the air side of the system is sel-
dom significant. Table 5-7 shows a typical cooling coils performance over a
range of CHW Ts. While the example in the table will not be true of all appli-
cations, it does suggest that air-side pressure will not increase very much as
CHW T rises, while water-side pressure drop falls significantly. For VAV sys-
tems, the impact on annual fan energy is even less significant because any full-
load air-side pressure drop penalty will fall rapidly as airflow decreases.
Figure 5-9 shows the impact of CHW T on energy usage for a typical Oak-
land, CA, office building served by a VAV air-distribution system with VFD and
an air-side economizer. Fan energy rises only slightly as T increases. If pipe
size is left unchanged as T increases, CHW pump energy will fall substantially
due to reduced flow and reduced piping losses. In real applications, pipe sizes are
generally reduced to decrease first costs, but pump energy will still fall due to
reduced flow rates and reduced coil and evaporator pressure drops, although not
as dramatically as in Figure 5-9.
Reducing CHW temperature can eliminate the fan energy penalty.
Figure 5-10 shows the same system as Figure 5-9, but instead of holding CHW
temperature constant and increasing coil size to increase T, CHW temperature
is lowered and coil size is held constant to keep air-side pressure drop (and
therefore fan energy) constant as T increases. Dropping CHW temperature
increases chiller energy, but pump energy savings more than make up the dif-
ference. As with Figure 5-9, the pump energy shown in Figure 5-10 assumes
that pipe sizes remain constant, which is not always the case.
Table 5-7 Typical Coil Performance Versus Chilled-Water Temperature Difference

Chilled-Water T, F 5.5 7.2 8.9 10.6 12.2 14
Coil water pressure drop, ft H2O 7.2 4.2 2.8 2.5 2.0 1.4
Coil air-side pressure drop, in. H2O 12.2 12.7 13.2 15.2 16.0 19.8
Rows 6 6 6 8 8 8
Fins per in. (fpi) 2.9 3.3 3.7 3.0 3.4 4.6
Cooling coil pressure air- and water-side drops were determined from a manufacturers AHRI-certified selection program assum-
ing 500 fpm coil face velocity, smooth tubes, maximum 12 fpi fin spacing, 43F CHW supply temperature, 78F/63F entering air
temperature, and 53F leaving air temperature.
Figure 5-9 Typical annual energy usage versus CHW T with a constant CHW supply tem-
perature and constant pipe sizes.
Figure 5-10 Typical annual energy usage with coils selected for constant air-side pressure
drop.
Table 5-8 compares three cooling coils with 4, 6, and 8 rows that result in
about 10F, 18F, and 25F T, respectively. Pipe sizes were selected from
Table 5-4 assuming an acoustically sensitive location with about 2000 h/yr of
operation. Coil first costs were obtained from the manufacturers representative
Table 5-8 Cooling Coil and Associated Piping Costs

(For 20,000 cfm coil sized at 500 fpm, 42F CHW supply temperature, 78F entering dry-bulb
temperature, 62F entering wet-bulb temperature, and 53F leaving dry-bulb temperature)
Coil Piping
Air Fluid
Fins per Pressure Fluid T, Fluid Pressure Pipe Size, Coil Total
Rows Drop, in. Coil Cost
in. F Flow,gpm Drop, in. Connection Cost
H2O ft H2O
10 4 17.8 10.1 118.7 9.1 $3598 3 $4551 $8149
11 6 16.51 18.2 66.0 7.6 $4845 2.5 $3581 $8426
10 8 20.32 24.9 47.0 5.7 $5956 2 $2101 $8057
and piping costs (including typical valve train and 20 ft of branch piping) were
obtained from the LCCA spreadsheet piping cost data. Table 5-8 shows that the
added cost of the deeper coil is more than offset by the savings in the cost of
piping the coil. And there are additional first-cost savings from the reduced
piping main, pump, pump motor, and VFD sizes.
So, increasing T reduces both first costs and energy costs. Clearly life-
cycle costs will be lower the higher the T. We were unable in our analysis to
find a point where the negative impact on fan energy or coil costs caused life-
cycle costs to start to rise with increasing T; within the range of our analysis
(up to 25F T), higher T was always better. Energy savings from high T are
even greater with systems that have WSEs or CHW TES. To reiterate: our anal-
ysis suggests that it never makes sense to use the traditional 10F or 12F Ts
that are commonly used in standard practice.
As T is increased, eventually the ever-deepening coil will run into the
ASHRAE Standard 62.1 coil pressure drop limit (2016a). ASHRAE
Standard 62.1 uses dry coil pressure as a surrogate for the cleanability of the
coil. Section 5.11.12 of the standard requires that dry coil pressure drop at
500 fpm face velocity must not exceed 0.75 in. This is roughly the pressure
drop of an 8 row, 12 fpi coil (depending on the details of the fin and tube
design).
So, the design procedure for selecting CHW coils is simple: rather than
arbitrarily selecting CHW temperatures and then selecting coils that deliver
those temperatures, reverse the logic: always use the biggest (highest effec-
tiveness) coil available without exceeding the ASHRAE Standard 62.1 pres-
sure drop limits and let the CHW T be determined by the coil and design air
conditions.
Based on this logic, the recommended procedure for sizing CHW coils and
selecting CHW design temperatures is as follows. The intent of this procedure
is to achieve all of the piping first cost savings resulting from a high T but
with as warm a CHW temperature as possible to improve chiller efficiency.
1. Calculate the CHW flow rate for all coils assuming a 25F T.1
2. Pick primary pipe sizes (at pumps, headers, main risers, and main branch
lines) in the critical circuit (that which determines pump head) using
Table 5-4 or the LCCA spreadsheet.
3. With pipe sizes selected, use Table 5-4 or work backwards using the LCCA
spreadsheet to find the maximum flow for each pipe size, and then recalcu-
late the T in each pipe using these flow rates. This is the minimum aver-
age T for this leg of the circuit.
4. Use the coil manufacturers selection program to find the maximum coil
size that complies with the ASHRAE Standard 62.1 cleanability limit, typ-
ically 8 row/12 fpi.2 Use this for all coils so that T is maximized. (For
some smaller fan-coils, 8 row coils may not be an option. If so, use the larg-
est coil available but no less than 6 rows. If that is not an option with the
selected manufacturer, find another manufacturer.)
5. With the coil manufacturers selection program, iterate on coil selections to
determine the CHW supply temperature that results in the selected T on
average for each leg of the critical circuit, starting with the coil at the end of
the circuit and working back to the plant. It is not necessary that all Ts be
the same (and in fact they definitely will not be the same with this approach),
just that the flow through the circuit equals the maximum flow determined in
Step 3. The recommended minimum CHW supply temperature is 42F.2 If
this minimum is reached in any leg and the flow exceeds the maximum, then
the process must be started over with a smaller T assumption in Step 1.
6. The lowest required CHW supply temperature for any coil in the circuit is
then the design temperature.
7. Determine actual flow and T in other coils in other circuits using the coil
selection program with this design CHW supply temperature, again maxi-
mizing coil size within ASHRAE Standard 62.1 limits (e.g., 8 rows, 12 fpi)
and letting the program determine return water temperature.
8. The plant flow is that calculated using the diversified (concurrent) load and
the gpm-weighted average return water temperature of all coils.
1. Some engineers may be concerned that a 25F T is nonconservative and reduces future
flexibility for load changes. Designing around large Ts results in large coils and small
pipes and pumps. Designing around small Ts results in the opposite, small coils and
large pipes and pumps. Both are equally forgiving with respect to possible coil load
changesone is no more conservative than the other. If excess capacity is desired for
future flexibility, it should be explicitly built into the design rather than relying on acci-
dental flexibility from design parameters.
2. In our analyses, the lowest CHW supply temperature resulting from this technique was
about 42F; we do not know if lower CHW temperatures will start to affect the life-
cycle cost due to reduced chiller efficiency. So, unless the designer performs additional
LCCA, we suggest limiting the design CHW supply temperature to no colder than 42F.
For most applications, this low temperature will not be required to achieve the target
25F T. Limiting the supply temperature to 42F also provides some conservatism in
the design; should there be a miscalculation in loads or unexpectedly high loads at a cer-
tain coil, CHW temperature can be lowered below 42F to increase coil capacity,
although with a resultant loss in overall chiller capacity and efficiency.
If this procedure seems too long and complicated, here is a shortcut proce-
dure: skip Steps 1 to 5 and just assume a CHW supply temperature of 42F in
Step 6. This will provide basically the same result except that the design CHW
temperature may be lower than it needed to be to achieve the pipe size savings
from high T, so the chiller design efficiency may be worse than it needed to
be. The energy impact will be minimal, however, because CHW temperature
should be aggressively reset, as discussed in Chapter 7. This simplified
approach also results in a somewhat lower CHW flow rate, so pump size and
power will be reduced.
Optimizing Condenser Water Design Temperatures

Selecting optimum condenser water temperatures is more complex than
selecting CHW temperatures due to the complex interactions between cooling
towers and chillers. As with chilled water, there can be significant first-cost
savings using high condenser water Ts (also known as cooling tower range).
But with chilled water, the supply fan energy impact was small, so increasing
T was found to always reduce total system energy costs. With condenser
water, the energy impact on the chiller of increasing T and return condenser
water temperature is not small (in fact it is very large), and T also signifi-
cantly affects the energy used by the cooling tower. So, optimum condenser
water temperatures are not as easily determined as those for chilled water.
Table 5-9 shows the first-cost and energy impacts of condenser water tem-
perature difference within the ranges commonly used in practice. Higher Ts
reduce first costs (because pipes, pumps, and cooling towers are smaller), but
the net energy-cost impact may be higher or lower depending on the specific
design of the chillers and towers.
Figure 5-11 shows chiller, tower, and condenser water pump energy usage for
the example Oakland office building introduced in Figures 5-9 and 5-10. The
condenser water temperature and T were selected so that the cooling tower
Table 5-9 Impact on First Costs and Energy Costs of Condenser Water Temperature
Difference Assuming Constant Condenser Water Supply Temperature
T
Low High
Typical range 8F 18F
Smaller pipe
Smaller pump
First-cost impact Smaller condenser Smaller pump motor
Smaller cooling tower
Smaller cooling tower motor
Lower pump energy
Energy-cost impact Lower chiller energy
Lower cooling tower energy
Figure 5-11 Oakland office building annual energy usage versus with condenser water supply
temperature and T selected for constant tower size and constant pipe sizes.
size and fan power do not change. As the T decreases, the temperature of the
water returning to the cooling tower decreases and the tower becomes less effi-
cient. This requires the condenser water temperature leaving the tower to rise
(or the tower size must be increased). The most energy-efficient combination
in this case was a 14F T. But this assumes pipe sizing is constant; however,
the pipe sizes could have been reduced for the larger T designs, reducing first
costs but increasing pump energy costs.
Figure 5-12 shows life-cycle costs for a large office building CHW plant that
was analyzed as part of the development of this SDL. Utility costs and life-cycle
cost assumptions are those used in the evaluation of energy conservation mea-
sures for ANSI/ASHRAE/IES Standard 90.1 ($0.094/kWh average electricity
costs and 14 scalar ratiothe scalar ratio is essentially the simple payback
period). The plant was modeled in great detail (including real cooling tower and
piping costs) for three climates: Oakland, CA; Albuquerque, NM; and Chicago,
IL. Figure 5-12 shows results for Chicago but the trend was the same in all three
climate zones: life-cycle costs were minimized at the largest of the three Ts ana-
lyzed, about 15F. This was true for both office buildings and data centers and
for both single-stage centrifugal chillers and two-stage centrifugal chillers. It was
also true for low-, medium-, and high-approach cooling towers (high-approach
tower data are shown Figure 5-12). (The optimum approach temperature is dis-
cussed under the section Selecting Cooling Towers, but it had no impact on the
optimum T). In all cases, pipe, pump, pump motor, and pump VFD sizes
reduced as T increased, and these cost differences were the primary driver in
life-cycle cost differences, as shown in Figure 5-12. The differences in energy
Figure 5-12 Life-cycle costs of 1000 ton chilled plant in Chicago as a function of condenser
water T.
use among the options is not as significant because savings in pump and tower
energy largely (though not completely) offset the increase in chiller energy use.
Other studies have also found that 15F condenser water T is optimum and can
even reduce annual energy costs (Trane 2005, 2011). The plant analyzed had a
relatively short distance between the towers and chillers; high Ts would have an
even larger life-cycle cost advantage for plants that have a large distance between
them, such as a plant with chillers in the basement and towers on the roof.
Based on this analysis, the following procedure is suggested to choose the
condenser water T (cooling tower range):
Calculate the condenser water flow rate for all pipe sections assuming a
range of 15F.
Pick primary pipe sizes (at pumps, headers, main risers, main branch lines) in
the critical circuit (that which determines pump head) using the LCCA
spreadsheet.
With pipe sizes selected, work backwards using the LCCA spreadsheet to
find the maximum flow for each pipe size and then recalculate the T in
each pipe using these flow rates.
The largest calculated T in any pipe segment is the design plant T.
Recalculate all flow rates using this T.
This procedure attempts to minimize cost by reducing pipe size as much as

possible. However, it then takes full advantage of the resulting pipe size to min-
imize T to reduce chiller energy. Pump energy will be a bit higher than if a
15F T were simply used, but pump energy is small relative to the impact of
high T on chiller energy use.
Table 5-10 Example Condenser Water Pipe Sizing

1 2 3 4 5 6
Pipe Size Maximum T at gpm at
Piping gpm at 15F
Application in. per gpm per Maximum Maximum
Section T
Table 5-5 Table 5-5 gpm T
Common Constant flow/constant speed,
1850 10 1900 14.6 1900
pipe 2000 h, non-noise sensitive
To each Constant flow/constant speed,
925 8 1200 11.6 950
equipment 2000 h, non-noise sensitive
Example Plant
Take, for example, a 1000 ton plant serving an office building in Oakland.
Each pump, chiller, and tower is sized for half the load.
At the initial chiller selection COP of 6.28, the rejected heat is about
13.9 million Btu/h. So at a15F T, the total condenser water flow rate is
about 1850 gpm and the flow rate to each individual piece of equipment is
925 gpm (Table 5-10, Column 2).
For an Oakland office served by a system with an air-side economizer, the
chiller plant will operate for about 2000 h/yr. Assuming constant flow/con-
stant speed, we can use Table 5-5 for pipe sizing. Because the chiller room
is not noise sensitive, the pipe sizes on the left side of the table are used.
The selected pipe sizes for each of the two piping sections are shown in
Table 5-10, Column 3.
Next, the selected pipe sizes are maxed out using the maximum flow rates
for each from Table 5-5 (Table 5-10, Column 4) and the T for each piping
section is recalculated using this flow rate.
The highest T in Column 5 is selected (14.6F) and flow rates are recalcu-
lated using this T (Table 5-10, Column 6). These flow rates would be used
to select chillers, towers, and pumps.
Selecting Cooling Towers

Cooling tower selection and the strategies used to control the tower fans
have a significant impact on CHW plant performance. Sizing towers for a
close approach to ambient wet-bulb temperature will improve chiller effi-
ciency, but it increases tower fan energy and first costs. Other factors to con-
sider are the efficiency of the tower itself and the tower fan speed control
options.
Tower Fan Speed Control

Tower fan control is required to maintain tower leaving water temperature
under all load and weather conditions. The common options (listed roughly in
ascending order of first-cost premium) are as follows:
One-speed motor (cycling)

Two-speed motor where low speed is half of full speed (1800 rpm/900 rpm)
Two -speed motor where low speed is two-thirds of full speed (1800 rpm/
1200 rpm)
Dual-motor drives (the smaller pony motor is typically designed for two-
thirds of full speed)
VFDs
Figure 5-13 shows the energy performance of these options. The perfor-
mance of the pony motor option will typically be close to the 100%/67%
(1800/1200 rpm) two-speed motor option. The performance of the one-speed
fan is not linear because of the free cooling that occurs in the off cycle due to
the stack-effect-induced natural draft through the tower.
The optimum fan control option depends primarily on the number of hours
the tower operates under various load conditions. This, in turn, is a function of
the application (e.g., office building, data center), climate, and the tower con-
trol set point strategy (see Chapter 7).
Figure 5-13 Cooling tower fan model at part load: percent power as a function of percent
load.
Despite these complex interactions, the following generalizations can be

drawn from our studies of typical plants:
One-speed control is almost never the optimum strategy for the typical
chiller plant, regardless of building occupancy type, climate, or tower con-
trol set point reset strategy. Energy savings from at least one of the two-
speed options result in very short payback periods. In addition, the two- or
variable-speed options reduce noise and wear and tear on belts.
The 100%/50% (1800 rpm/900 rpm) two-speed motor is somewhat more
energy efficient than the 100%/67% option, particularly when controlled as
recommended in Chapter 7. The 1800/900 rpm motor is available in a one-
winding motor, which is less expensive than the two-winding motor required
for 1800/1200 rpm operation, although this savings is partly offset by the
higher cost of the starter. At low speed, towers driven by 1800/900 motors
are also quieter than those driven by 1800/1200 motors.
Pony motors are usually more expensive than standard two-speed motors,
but the additional motor reduces exposure to motor failure. Note that this
option is not available on all towers.
For plants with multiple towers or multiple cells, two- or variable-speed
control must be provided on all cells, not just the lead cells. The towers are
most efficient when all cells are running at low speed rather than some at
full speed and some off. For instance, two cells operating at half speed will
use about 25% of full power compared to 50% of full power when one cell
is on and the other is off.
Because VFD efficiency is not that much better than two-speed motor effi-
ciency (see Figure 5-13), VFDs were not found to be cost-effective when
we first performed a LCCA in the late 1990s. Since that time, however,
VFD costs have plummeted, and they are now the clear best option. In
addition to energy savings, VFDs offer these other benefits:
Similar or lower first cost to two-speed motors
Tighter temperature control than two-speed motors
Reduced noise at low loads and less abrupt changes in noise generation
Reduced belt maintenance costs for belt-driven fans
Operating information available from the VFD, such as motor power
Tower Efficiency
Tower efficiency (as defined in California Title 24 [CBSC 2016] and
ANSI/ASHRAE/IES Standard 90.1 [2016b]) is the ratio of the maximum
tower flow rate (gpm) to the motor horsepower () at standard Cooling Tower
Institute (CTI) rating conditions (95F to 85F at 75F wet bulb). The optimum
tower efficiency depends on the number of hours the tower operates under var-
ious load and weather conditions.
Two primary factors affect tower efficiency:
Fan type: Towers are commonly available with either propeller fans or cen-
trifugal blowers. The latter require about twice the fan power and are no
less expensive. From an energy and first-cost perspective, propeller fan
towers (whether draw through or blow through) should always be used.
There are a few advantages to centrifugal fan towers: they are generally
quieter than propeller fan towers and they can operate against a larger
external static pressure drop, such as that caused by louvers when towers
are located indoors. However, the design can usually be modified to accom-
modate propeller fan towers, such as by oversizing the tower to reduce fan
speed and noise, using ultraquiet fan blades available from most tower
manufacturers, or oversizing intake louvers to reduce pressure drop.
Because of the severe energy penalty associated with centrifugal blowers,
every effort should be made to accommodate a propeller fan tower before
considering a centrifugal fan tower.
Fan pressure drop through the fill: Pressure drop is primarily a function of
the fills size and design but can also be affected by fan inlet and discharge
configuration. Most engineers use the least expensive tower selections
available in manufacturers selection software. However, the tower and fill
can be oversized to reduce pressure drop, thereby allowing the fan to be
slowed down, which reduces motor power. Whether this is cost-effective
depends on the application, climate, and the added cost to oversize the
tower and to accommodate the larger tower footprint and weight.
To determine optimum efficiency and approach temperature, a large office

building CHW plant was analyzed using utility costs and life-cycle cost
assumptions as described above. The plant was modeled in great detail for
three climates: Oakland, CA; Albuquerque, NM; and Chicago, IL. Details
included real equipment and piping costs. Additional analyses for optimum
approach temperature were made for Miami, FL; Las Vegas, NV; and Atlanta,
GA. The condenser water system was designed, cost estimated, and modeled at
all permutations of the following design parameters:
Three ranges of tower efficiencies: low was the least efficient available for the
cross-flow propeller fan tower series analyzed, with efficiencies ranging from
45 to 60 gpm/hp; medium, with efficiencies ranging from 65 to 75 gpm/hp;
and high, with efficiencies ranging from 80 to 100 gpm/hp. Note that even the
low-efficiency towers are significantly more efficient than the ANSI/
ASHRAE/IES Standard 90.1 minimum of 38.2 gpm/hp (2016b).
Tower approach temperatures ranging from 2.5F to 11F based on actual
tower selections for a cross-flow propeller fan tower series.
Tower costs were based on manufacturers price plus sales tax and contrac-
tor markup, plus a 50% premium that is intended to estimate the secondary
installed cost impact of larger towers. Tower size (both footprint and height)
and weight increase with increasing efficiency and decreasing approach. Both
can impact tower installed costs depending on tower location. The actual pre-
mium can vary from close to nothing for a tower located on grade to a signifi-
cant premium if the tower is located on the roof and requires architectural
screening to hide the tower from view, because larger towers will require addi-
tional structural work and larger screens. The 50% premium is probably con-
servativein most cases the premium will be lower.
Figure 5-14 shows life-cycle costs for the three ranges of cooling tower
efficiencies for an office building in Miami with a tower range of 15F; life-
cycle costs were minimized with the high-efficiency towers. This was true in
all climate zones, ranges, and approach temperatures analyzed. The analyses
were based on a fairly aggressive scalar ratio (maximum simple payback
period) of 14, but high-efficiency towers were found to be cost-effective down
to a scalar ratio of about 5 (i.e., they will have a simple payback of five years
compared to the next best tower option, even in the mildest climates). The rea-
son is that the net cost premium for increasing tower efficiency is relatively
small; the tower physical size and fill area increase but motor and VFD size
and cost decrease, partially offsetting the tower cost increase. For example, the
net installed first cost add for the high-efficiency tower versus the low-efficiency
tower for the 1000 ton plant in Miami was only about $9000, a 6% increase,
while annual energy savings were about $5500 and life-cycle energy savings
were $77,000. The magnitude of the savings is smaller in milder climates,
but the high-efficiency towers were found to be cost-effective in all climates
analyzed.
Figure 5-14 Life-cycle costs of 1000 ton chiller plant serving a Miami office building as a func-
tion of tower efficiency range.
Accordingly, cooling towers should be selected for an efficiency of at least

80 gpm/hp. Those serving data centers and other 24/7 applications should be
selected for 100 gpm/hp. Plants with WSEs should have even higher efficien-
cies as discussed in the Water-Side Economizers (WSEs) section.
Tower Approach Temperature

While increasing tower efficiency from low to high is relatively inexpen-
sive (about 6% to 12% of tower costs in this study), the same cannot be said for
reducing tower approach. As shown in Figure 5-15, the cost of a tower that pro-
vides a low approach of 2F to 3Fcan be 60% higher than a tower providing a
10F to 12F approach.
A reasonable correlation was found between the sum of the life-cycle cost
optimum tower approach (TA) (again using ANSI/ASHRAE/IES Standard 90.1
energy costs and scalar ratio) and tower range (also called condenser water
temperature difference, TCW), and cooling degree-days base 50 (CDD50).
This is shown in Figure 5-16 for each of the climate zones tested. The straight
line curve fit is approximately
T A + T CW = 15 0.0006CDD 50 (5-1)
Figure 5-15 Cooling tower installed costs for a 1000 ton chiller plant as a function of tower
approach.
Figure 5-16 Life-cycle cost optimum approach plus range as a function of cooling degree
days base 50F.
Solving for approach,
T A = 15 T CW 0.0006CDD 50 (5-2)
Based on the limited data, the approach determined from this equation
should be limited to no more than 9.5Fand no less than 2.5F.
In summary, based on this analysis the following design criteria are recom-
mended for selecting cooling towers for office buildings and buildings with
similar load profiles:
Tower efficiency should be 80 gpm/hp or greater.
Tower approach should be selected using Equation 5-2, with a minimum of
2.5F and a maximum of 9.5F.
Water-Side Economizers (WSEs)

As discussed in Chapter 4, WSEs should always be piped in an integrated
precooling position, as shown in Figure 5-17.
As shown in Figure 5-17, the same cooling towers and condenser water
pumps can be used to serve both the economizer HX and the chiller condens-
ers. Some designers provide separate towers and pumps, at considerable
expense, because of concerns about low chiller head pressure due to low con-
denser water supply temperatures. However, head pressure control is easily and
Figure 5-17 Integrated economizer.
inexpensively addressed by making the condenser isolation valves modulating

and controlling them off the head pressure control signal output that is standard
on most chiller controllers. The valves throttle flow through the condenser as
needed to maintain chiller minimum lift regardless of how cold the condenser
water supply temperature is in economizer mode. (See Chapter 4 for more
details on head pressure control.)
The capacity and quantity of cooling towers and condenser water pumps
remain the same as they would without the economizer. For office building
applications, this is intuitively clear: we know that when the economizer is
on, weather will be cold so loads will be well below design loads; hence,
only one of the two chillers (in Figure 5-17) will be needed, freeing the other
condenser water pump to supply tower water to the economizer HX. This is
also true for data centers where the load may require all chillers to run even
in cold weather. The reason is that the load on the towers is actually reduced
by the economizer because compressor heat is reduced, and condenser water
flow to the chiller condensers may be less than design because the water tem-
perature is colder than design (and in fact may be reduced by the throttling of
the head pressure control valves discussed in Chapter 4 in the section Refrig-
erant Head Pressure Control), thus making water available to the economizer
HX without the need to add pumps.
While cooling tower capacity is not affected by the economizer, it may be
necessary to reduce the design approach temperature in order to meet ANSI/
ASHRAE/IES Standard 90.1 Section 6.5.1.2.1 (2016b) WSE requirements,
particularly for plants with high loads in cold weather. This is discussed further
in the example below in Example Design Procedure.
It is critical that cooling towers be very efficient because they will be run-
ning at full speed many hours of the year. A minimum of 90 gpm/hp is recom-
mended for office-type applications and 110 gpm/hp for 24/7 applications,
such as data centers. These efficiencies are 10% above those shown to be cost-
effective for non-economizer applications.
Cooling towers should be selected so that as many tower cells as possible
can be enabled when the economizer is enabled to maximize efficiency and
capacity while maintaining the minimum flow rates required by the tower man-
ufacturer to prevent scaling. Low minimum flow rates can be achieved using
weir dams and special nozzles in the hot-water distribution pans.
CHW pump head increases due to the pressure drop of the HX when in
economizer mode. However, in applications like offices, where the loads are
low when the economizer is on, pump head may not need to increase above
design head when the economizer is off; excess head may be available for the
HX when the economizer is active due to the reduced CHW flow to coils.
The HX should be a plate-and-frame type and selected for an approach of
about 3F (i.e., the temperature of the chilled water leaving the HX is equal to
3F above the temperature of the condenser water entering the HX) in office
applications. HX cost increases exponentially with approach temperature, so
very close approaches should be tested for cost-effectiveness. Plate-and-frame
HX cost typically increases dramatically with an increase in frame size,
whereas adding additional plates to a smaller frame usually results in a linear
increase in cost. As such, one effective design approach in applications such as
data centers, where a close approach may prove cost-effective, is to maximize
the plate capacity of a given frame. For instance, if the design condition is for a
3F approach, but the selection leaves 20% additional frame capacity, it may
prove cost-effective to max out the frame. The HX pressure drop on the con-
denser water side should be similar to that of the condensers so the flow rate
will be similar when serving either the condensers or the HX. On the CHW
side, pressure drop is typically limited to about 5 or 6 psi to minimize the
CHW pump energy impact. The HX performance must be certified per AHRI
400 (2001), as required by ANSI/ASHRAE/IES Standard 90.1.
To maximize economizer performance, and also the performance of the
system even when not economizing, the CHW system must be designed for a
very high temperature rise (T) using the procedure described above.
Example Design Procedure
1. Calculate the CHW load at 50F dry-bulb temperature and 45F wet-bulb
temperature. This is the performance test condition prescribed by ANSI/
ASHRAE/IES Standard 90.1 for buildings other than data centers (2013).
This can be done using standard load calculation software by selecting a
spring or fall month and overriding design outdoor air temperatures. All
other load assumptions remain the same. Loads should be reduced from
design loads due to reduced conduction and outdoor air conditioning loads.
2. Use coil selection software or other coil models to determine the warmest
CHW supply temperature that can meet 100% of the CHW load deter-
mined above for all coils. To do this, first determine the warmest supply
water temperature that can meet the load at the design flow for each coil,
then select the coldest of these and use it to determine the CHW flow
through all the other coils. The coil software will also determine the CHW
return temperature from each coil. Typically the CHW supply temperature
can be reset 5F or more above the design CHW supply temperature. This
can be true even for data centers, despite the consistently high load, by tak-
ing advantage of redundant air handlers to effectively increase available
coil area.
3. Select the condenser water flow rate equal to a multiple of design con-
denser water pump flow rates as required to closely match the CHW flow
rate determined above.
4. Determine the HX condenser water supply temperature equal to the
required reset CHW supply temperature determined above less the 3F
approach temperature.
5. Calculate the condenser water return temperature to match the CHW load
based on the condenser water flow and supply temperature determined
above.
6. Use cooling tower selection software to verify that the cooling towers can
provide the required condenser water supply temperature at 45F wet-bulb
temperature. If not, then cooling towers (and/or HXs) will need to be rese-
lected for closer approach temperatures.
For instance, assume the plant in Figure 5-17 serves an office building with
floor-by-floor air handlers. The design conditions of the two chillers at the
design cooling peak are shown in Table 5-11.
The pump flow rates match the chiller design rates and cooling towers are
selected at 68F wet-bulb temperature for a 7F approach.
The cooling loads are then recalculated at 50F/45F outdoor- air tempera-
ture conditions; the load drops from 600 to 350 tons. Coil selection programs
are then used to determine the CHW conditions required at the air handlers,
and condenser water conditions are determined using the steps above. The
resulting HX design conditions are shown in Table 5-12.
Table 5-11 Chiller Design Conditions (Each)

CHW Side Condenser Water Side
Capacity, tons 300 345
Flow, gpm 290 550
Entering water, F 69.0 75
Leaving water, F 44.0 90
P, psi 6.5 6.7
Table 5-12 WSE HX Design Conditions

CHW (Hot) Side Condenser Water (Cold) Side
Total load, tons 350 350
Flow, gpm 560 550
Entering water, F 65.0 47.0
Leaving water, F 50.0 62.3
P, psi 5 4.8
Cooling tower selection software is then used to see if the selected cooling
towers could cool 550 gpm (275 gpm across each tower) from 62.3F to 47F,
a 2F approach to the 45F wet-bulb temperature. However, the software indi-
cates that the towers are only able to deliver 48F. So, the towers are reselected
for a 73F leaving water temperature (5F approach to design wet-bulb tem-
perature) in order to achieve a 2F approach at 45F ambient wet-bulb tem-
perature and the load conditions prescribed in Table 5-12. Note that HX
approach also could be reduced to deliver the desired CHW temperature, but it
is usually more cost-effective to invest in larger cooling towers because they
also improve efficiency when the economizer is off.
Selecting Type, Number, and Size of Chillers

Design and selection procedures for chillers are discussed in Chapter 6.
The procedures are included in a separate chapter to make them easier for engi-
neers to use for retrofits and other projects that do not include the design of the
CHW and condenser water systems addressed in this chapter.
Thermal Storage
Thermal storage systems can offer significant energy cost savings where
energy costs are based on time-of-day or real-time pricing. The design of these
systems is beyond the scope of this SDL. For more information, refer to
ASHRAEs Design Guide for Cool Thermal Storage (Dorgan and Elleson 1993).
References
AHRI. 2001. AHRI 400-2001, Liquid-to-liquid heat exchangers. Arlington,
VA: Air Conditioning, Heating and Refrigeration Institute.
ASHRAE. 2016a. ANSI/ASHRAE Standard 62.1-2016, Ventilation for accept-
able indoor air quality. Atlanta: ASHRAE.
ASHRAE. 2016b. ANSI/ASHRAE/IES Standard 90.1-2016, Energy standard
CBSC. 2016. 2016 California building standards code. California Code of
Regulations, Title 24. Sacramento, CA: California Building Standards
Commission.
Dorgan, Charles E., and James S. Elleson. 1993. Design guide for cool thermal
storage. Atlanta: ASHRAE.
EDR. 1999. CoolTools chilled water plant design guide. Sonoma, CA:
Energy Design Resources.
McBride, M. 1995. Development of economic scalar ratios for Standard 90.1.
Proceedings of Thermal Performance of the Exterior Envelopes of Build-
ings VI. Clearwater Beach, FL.
Taylor, S. 2002. Degrading chilled water plant Delta-T: Causes and mitigation.
ASHRAE Transactions 108(1).
Taylor, S., and M. McGuire. 2008. Sizing pipe using life-cycle costs. ASHRAE
Journal 50(10).
Taylor, S. 2011a. Optimizing design & control of chilled water plants: Part 1:
Chilled water distribution system selection. ASHRAE Journal 6:1425.
Taylor, S. 2011b. Optimizing design & control of chilled water plants: Part 3:
Pipe sizing and optimizing T. ASHRAE Journal 12:2234.
Chiller & cooling tower selection. ASHRAE Journal 3:6070.
Trane. 2005. Trane Engineers Newsletter 34(1). ADM-APN014-EN.
Trane. 2011. Chiller system design and control: Applications manual. Dublin:
Trane.

Complete these questions by writing your answers on the worksheets at the back of this book.
5-1 You are responsible for developing the master plan for a new college campus
that will include multiple buildings. Which of the following CHW distribution
approaches will be the most energy efficient?
a. Variable primary flow
b. Primary, variable secondary
c. Primary, variable secondary with tertiary bridge connected pumps at
the buildings
d. Primary, distributed secondary where the secondary pumps are sized
for the pressure drop from and back to the primary loop
5-2 Your customer is building a sprawling six-story, 600,000 ft2 building with three
wings, each of which is served by a large AHU. The building also has multiple
computer rooms served by FCUs. All of the high density rooms are localized in
the same area of the building. What is the likely to be the most energy-efficient
CHW distribution design approach?
a. Variable primary with a plant bypass leg and two-way valves at all
AHU and FCU coils
b. Constant primary, variable secondary with two-way valves at all AHU
and FCU coils
c. Variable primary with distributed variable secondary coil pumps for all
AHUs and a separate variable secondary pump serving all FCUs with
two-way valves
d. Variable primary with distributed variable secondary pumps at all
AHUs and all FCUs
5-3 When sizing piping to minimize life-cycle costs, which of the following are
critical considerations?
a. Expected system operating hours
b. Variable-speed versus constant-speed operation
c. Utility rates
d. All of the above
5-4 Which of the following are benefits resulting from selecting condenser water
loop T for a larger range?
i.Reduced chiller energy use due to lower chiller lift
ii.Reduced pumping energy use
iii.Increased cooling tower capacity for a given tower selection
iv.Reduced cooling tower scaling potential
a. (ii.) and (iii.)
b. (i.), (ii.) and (iii.)
c. (ii.), (iii.), and (iv.)
d. (i.), (ii.), and (iv.)
5-5 Which of the following are true when selecting cooling tower fan controls?
a. Constant-speed towers often prove life-cycle cost-effective because of
their low upfront costs.
b. VFD fan speed controls are prohibitively expensive relative to the small
energy benefit gained relative to two-speed motor controls.
c. To achieve the energy benefits provided by variable-speed tower con-
trol, it is only necessary to install a VFD on the lead cooling tower.
d. VFDs on all tower cells generally result in lowest life-cycle costs.
5-6 You are retrofitting an existing central plant serving a large office building with
a WSE in a climate where freeze conditions are not a concern. The plant cur-
rently maintains minimum chiller head pressure in cool dry weather using a
cooling tower bypass. How does this arrangement need to be modified to
accommodate the WSE?
a. No changes are needed.
b. A separate cooling tower without condenser water bypass needs to be
added to accommodate the WSE HX.
c. Modulating actuators should be added to the condenser isolation valves
on each chiller and the cooling tower bypass should be permanently
shut.
d. The WSE HX should be piped in series with the cooling towers and the
cooling tower bypass should be shut permanently.
Chiller
Procurement
Instructions
Introduction
This chapter discusses a novel approach to selecting chillers using life-
cycle cost techniques. There are many chiller options that apply to a given proj-
ect, and, in the case of centrifugal chillers, there are hundreds of options
because these custom-built products include so many options (condenser and
evaporator sizes and quantity of tubes, compressor and impeller sizes, motor
drive and bearing options, etc.) The typical selection process is largely subjec-
tive, which seldom results in the optimum selection. A case study of the chiller
selection process is provided for a new office building. Also provided are two
sample chiller bid forms in spreadsheet format, available online at ashrae.org/
CHWSDL.
Chiller Procurement Procedures

Table 6-1 compares typical versus recommended approaches for procuring
chillers. The typical approach, shown in the tables left column, is expedient,
but it seldom leads to an optimum selection. There are simply too many selec-
tion options on large chillers to make a good choice by inspection. Not only are
there often major design options such as centrifugal versus screw compressors,
refrigerants R-123 versus R-134a (or, now, next-generation low-GWP refriger-
ants), constant versus variable speed, and conventional versus magnetic or
ceramic bearings, there can be hundreds of combinations of evaporator, con-
denser, and compressor options for a given capacity that can radically affect
chiller price and performance. The recommended approach to chiller selection
is outlined in the right column of Table 6-1 and is discussed in more detail in
the subsequent sections of this chapter. This method ensures selection is based
on objective performance rather than subjective issues.
182 Chapter 6 Chiller Procurement
Table 6-1 Chiller Procurement Approaches

Typical Approach Recommended Approach
1. Calculate or estimate the required plant total 1. Calculate the required plant total capacity and
capacity. design temperatures and flow rates.
2. Pick number of chillers, usually arbitrarily or as 2. Develop a bid list of chiller vendors based on
limited by program or space constraints. past experience, local representation, etc.
3. Take plant load and divide by number of chillers to 3. Request chiller bids based on a performance
get chiller size (all equal). specification. Multiple options encouraged.
4. Pick favorite vendor, often based on who is most 4. Adjust bids for other first-cost impacts.
attentive to the engineering firm. 5. Estimate utility costs of options based on
5. Have vendor suggest one or two chiller options. detailed computer model simulations of the
6. Select one option based on minimal or no analysis. building/plant.
7. Bid the chillers along with the rest of the job and let 6. Estimate maintenance cost differences between
market forces determine which chillers are actually options.
installed. 7. Calculate life-cycle costs.
8. Select the chiller option with the lowest life-
cycle cost.
9. Hard-spec the selected chiller (no substitutions)
and include contractor price in specifications.
The recommended performance bid approach should take place once plant
capacity is well defined. This is typically at the end of the design development
phase.
The recommended approach has many advantages:
The owner generally benefits from lower life-cycle costs.

Arbitrary selection of chiller vendor and model is eliminated, potentially
lowering chiller costs due to a more competitive bid process. In traditional
design/bid/build projects, the specified chiller is commonly provided in
contractors bids even if it is not the lowest cost because of the uncertainty
of the secondary cost impacts that might occur from a substitution.
The procedure generally results in a more energy-efficient chiller selection.
The traditional approach often leads to a low-cost mentality where the least
expensive, and often least efficient, chiller is selected.
Chiller vendors can make proposals that take advantage of their systems
strengths or sweet spots both for cost and efficiency. The conventional
approach, where size and efficiency are more arbitrarily selected, usually
favors one vendor (inadvertently or intentionally) who happens to have a
sweet spot for the selected piece of equipment.
Chiller selections are finalized in the design stage. Selection at this time
allows the designer to customize the design of the plant, including physical
layout and chiller-specific design parameters such as minimum flow
(which affects minimum flow bypass line and valve size), knowing that the
chiller selection will not be changed at bid time. The traditional approach
can result in substitutions at bid time, which can lead to coordination issues
(and costs) due to changes in size, weight, peak power, minimum flow
rates, etc.
There are also disadvantages to this approach:
It takes more time, both for the engineer and the equipment vendors.1
Assumptions made in the energy calculations may prove to be incorrect. For
instance, utility rates, internal loads, and occupancy patterns assumed in the
analysis may not be correct or may change over time. Energy escalation and
inflation rates used in the analysis also have considerable uncertainty.
The computer model used to calculate energy usage is imperfect in the way
it models building and system loads, the chiller plant and pumping systems,
and plant control strategies. Optimum plant control strategies are often
complex and can vary from one chiller option to another. These strategies
cannot always be modeled accurately by existing energy simulation tools.
The benefits of long-term product reliability and vendor support are seldom
included in the life-cycle cost calculations because their cost benefits are
difficult to estimate, although they may be significant.
Despite these disadvantages, the recommended chiller procurement approach

represents an improvement over conventional approaches and has been success-
fully used on several projects. The case study at the end of this chapter is based
on an actual project that followed this recommended approach.
Note that this approach has been used successfully on projects that require
competitive bidding, such as most state and General Services Administration
projects. This is because most of the statutes mandate only that competitive
bids occur, not that they occur at the traditional bid time when the design is
complete. Most allow metrics such as life-cycle cost (not first costs) to be the
basis of selection provided the selection criteria are well defined.
Procurement Step #1: Calculate Plant Capacity and Design Conditions

See Chapter 2 for an in-depth overview of available methods of calculating
or estimating the required plant total capacity (tonnage).
The chiller procurement procedures described below assume that chiller
plant design parameters, such as condenser and CHW entering and leaving
temperatures and flow rates, have been determined. For retrofit applications
these design parameters are often set by the constraints of the existing system
design. Chapter 5 describes procedures for selecting these design parameters
for new chiller plants.
1. Note: ASHRAE is developing a new standard, ASHRAE Standard 205, Standard Representation of
Performance Simulation Data for HVAC&R and Other Facility Equipment, that will define standard data
models and formats for use in creating accurate equipment energy models. This will substantially reduce
or eliminate the time burden on equipment vendorsthe performance data will be generated by chiller
factory engineersand it will improve the accuracy of chiller models. The standard is expected to be
published in 2017.
Procurement Step #2: Develop Vendor Bid List

Pick a list of chiller vendors based on past experience, local representation,
and other factors. Not all of the major vendors need to be on the bid list; how-
ever, the more vendors on the bid list, the better the competition will be, and,
therefore, the lower the chiller energy costs and first costs are likely to be.
However, low first costs and energy costs are not the only selection criteria:
low maintenance costs, system reliability, vendor support, and other issues are
also considerations.
This approach also works when the owner has a specific chiller manufac-
turer they want to use based on past experience or to maintain consistency with
chillers already installed at their building or campus. Optimizing chiller selec-
tion with a single vendor is still valuable because within their product line there
are many options, such as larger heat transfer areas in condensers and evapora-
tors and VFDs. It is not uncommon for centrifugal chiller selection software to
generate hundreds of selections for a single set of capacity and design condi-
tions.
Procurement Step #3: Obtain Chiller Bids

Performance Specification
First, develop a performance bid specification that includes all design con-
straints for the chillers. In the conventional procurement approach, a chiller is
typically specified by capacity, efficiency, construction details, etc. These
details, however, are not usually appropriate for a performance specification.
Vendors should be encouraged to propose as many chiller options as possible
without constraintsthe LCCA will determine which option is the best, so
constraints are not necessary and may inadvertently eliminate better choices.
As a rule, do not include the following in the performance specification,
instead leaving them up to the bidders:
Number of chillers and capacity of each chiller. (For some projects, the
engineer may decide to fix chiller quantity and size based on other factors
or for simplicity.)
Chiller efficiency metrics (kW/ton, IPLV, etc.), although the specification
should state that all selections must meet all full-load and part-load energy
code requirements.
Variable speed or fixed speed.
Conventional or magnetic or ceramic bearings.
Refrigerant type.
Encourage multiple options from each vendor. They may, for instance, pro-
pose one design that uses several equally sized chillers and another that
includes unequally sized chillers or a pony chiller. Another option may include
VFDs on one or more of the chillers. Encourage the vendors to be imaginative.
Also eliminate unnecessary boilerplate requirements. Conventional specifi-

cations often contain a long list of construction details that are sometimes pro-
prietary and usually not very important from a quality and reliability
perspective. If a chiller vendor has been approved for the bid list, then that
should mean that the vendors standard construction details are acceptable.
Including endless boilerplate requirements in the specifications is usually a
waste of time and can be counterproductive if they mask truly important speci-
fication details.
The following details should be included in the performance specification:
Total plant capacity. In general, specify only total plant capacity, not the
capacity of each chiller. See Chapter 2 for an overview of methods of deter-
mining plant total tonnage.
The minimum number of chillers or required level of redundancy. For
redundancy and reliability, the specification may require, for instance, that
there be at least two chillers in the plant, or possibly at least two compres-
sors to allow for dual-compressor options. Alternatively, for critical plants,
such as those serving an industrial process or data center, the specification
may require N+1 redundancy, meaning that the chiller plant must be able to
handle the load even with the failure of the largest chiller.
Design conditions. Design condenser and evaporator flow rates and enter-
ing and leaving water temperatures of the plant as a whole must be speci-
fied. For retrofit applications these factors are usually fixed by the
constraints of the existing design. Chapter 5 describes how to determine
these parameters for new buildings.
Energy sources available, such as electricity or gas. This procurement pro-
cedure is not limited to electric chillers, although mixing gas and electric
chillers will require more cost adjustments to account for the larger heat
rejection required by absorption chillers and for differences in gas and elec-
tric utility service sizes and distribution systems.
Typical cooling load profile. In order to offer intelligent proposals, the
chiller vendors need to know the cooling load profile, typically developed
from a computer model of the systems served by the plant (see Chapter 2).
For instance, the plant with the load profile shown in Figure 2-1 for a typi-
cal office building will require much different chiller sizing for best effi-
ciency than the plant with the load profile shown in Figure 2-2, which has a
relatively small data center that provides a constant base load. The plant
serving Figure 2-2 loads will most likely require a small pony chiller or
unequally sized chillers for best performance; so, it is important that the
vendors bidding the chillers be aware of this need.
Limited mechanical room space. If this is a factor, include a plan of the
mechanical room in the specification.
Available voltage and, for retrofit projects, available capacity of the electri-
cal service.
Type of allowed refrigerants. For instance, the owner may disallow R-123
as an option because it is to be phased out of production in the near future
in developed countries, or the owner may prefer R-123 because it has a
lower GWP. (See Chapter 3 for additional comments on refrigerant issues.)
Sound power limits. If noise is a design constraint because the chiller is
located adjacent to noise-sensitive spaces, an acoustical engineer should
back-calculate the maximum chiller sound power levels needed to meet
noise criteria levels in occupied spaces. These limits should then be
included in the performance specification. Either the proposed chillers have
to meet these sound power limits inherently or the vendor has to include a
chiller sound enclosure in their proposal.
Performance Verification
The following are optional requirements in the specification to ensure that
the performance data from the chiller manufacturer are accurate.
Factory Tests
Factory testing is an option offered by almost all manufacturers. It can be
witnessed by the owners representative (often the design engineer) or non-
witnessed. The latter is usually much less expensive because testing can
occur right when the chiller comes off the assembly line; witnessed tests
must be arranged well in advance and can cause production inefficiencies, in
addition to the cost of travel for the witness.
It is highly recommended that factory tests be required, despite their cost,
for these reasons:
Factory tests make it possible to reject the chiller if it does not meet perfor-
mance specifications. Once a chiller is installed in the field, rejecting the
chiller is no longer a practical option.
Testing is more accurate in the factory than in the field, resulting in less con-
troversy about whether and to what extent chillers fail to meet performance
requirements. The factory test stand is fitted with calibrated laboratory-grade
flowmeters and temperature sensors. Even if accurate instruments were avail-
able in the field, their accuracy is easily questioned by the chiller manufac-
turer should the tests indicate chiller underperformance.
Field tests cannot supply the necessary load and ambient conditions to test
the chillers at all the specified conditions.
Factory tests obviate the need for field tests for commissioning or perfor-
mance verification purposes. Tests are more easily and less expensively
done in the factory than in the field, and often field tests are not practical
because of incomplete instrumentation (e.g., lack of a condenser water
flowmeter).
Liquidated Damages
If performance tests are required, it is recommended that monetary penal-
ties apply if a chiller fails to meet performance requirements. Without teeth
in the specification, vendors will be more likely to exaggerate chiller perfor-
mance in their proposals. An example liquidated damages clause follows:
Chiller manufacturer shall repair or replace equipment, at no cost to
the Buyer, until equipment is certified by AHRI 550/590 test proce-
dures to meet the performance indicated in the manufacturers pro-
posal. In the event that these revisions do not achieve submitted
performance or require longer than 2 weeks, the following penalties
will be imposed.
1. Capacity penalty: For each ton below the proposed design capacity,
one thousand dollars per ton will be deducted from the contract
price.
2. Power consumption penalty: The power consumption penalty for
each tested load point shall be
= [Measured kW (Measured Tons Proposed kW/Ton)]
$2000/kW.
3. Total penalty: The total performance penalty will be the sum of the
capacity penalty and power consumption penalty at each tested
load point for each chiller tested.
Zero AHRI Tolerance

AHRI Standard 550/590 (2015) allows a tolerance in the measured versus
actual chiller capacity and efficiency. The magnitude of the tolerance varies as
a function of load, increasing as load decreases, as shown in Figure 6-1.
The tolerance curve, established over 30 years ago, was intended to
account for variations in manufacturing, but over time manufacturers have
found that their manufacturing tolerances are smaller than AHRI standard tol-
erances. Market forces then caused manufacturers to take credit for the toler-
ances in the chiller performance reported by their rating programs. In other
words, the power reported by manufacturers rating programs for a given
chiller has been decreased from the actual power required by an amount equal,
or almost equal, to the allowed AHRI tolerance. Obviously, to get an accurate
picture of predicted chiller performance in the LCCA, actual chiller perfor-
mance data must be obtained from chiller vendors. Figure 6-2 shows predicted
chiller performance for the same chiller with and without AHRI tolerance
assuming standard AHRI condenser water relief. The AHRI tolerance data
were from the chiller manufacturers selection program without adjustment.
The zero tolerance data were from the same manufacturer for the same chiller
but with the stipulation that the chiller be factory tested and must meet pre-
dicted performance claims with zero tolerance. It is clear that low-load chiller
kW/ton is significantly underpredicted when AHRI tolerance is allowed. This
can completely skew chiller selection. For instance, the manufacturers who
take the most advantage of the AHRI tolerances (not all push the limit) will
Figure 6-1 AHRI Standard 550/590 tolerance.
Figure 6-2 Chiller-predicted performance with and without AHRI Standard 550/590 tol-
erance.
have a significant advantage. If the analysis was in part to determine how many
chillers or what size they should be for plants with unequally sized chillers, the
AHRI tolerance can result in an incorrect answer because it makes it appear
that large chillers are efficient at low loads. It is strongly recommended, there-
fore, that the performance claims by manufacturers be required to have zero
AHRI tolerance. The bid specification should state that no tolerance, such as
that normally allowed by AHRI Standard 550/590, will be accepted on pro-
posed efficiency and capacity data. Note that obtaining zero tolerance data
from vendors will often elicit complaints because it can be difficult due to
selection software constraints, but it is absolutely essential for accurate analy-
sis and proper chiller selection.
Other Bid Inclusions

The project construction documents will ultimately include contractor pric-
ing for the chillers in the bid specifications. For pricing to be complete, costs
normally included in vendor proposals must be clearly identified and included
in the performance bid. This includes cost of freight to the job site. Including
local sales taxes is also a good idea because some part of the chiller bid may
not be taxed, such as freight and some start-up services.
Performance Bid Forms

Bid forms are used to collect both chiller pricing and chiller performance
data. It should be made clear to chiller vendors that chiller pricing and selec-
tion using this procedure are final; there will be no opportunity to reprice
chillers when the project is bid at the 100% construction documents phase.
To create chiller simulation models using common simulation programs
such as DOE-2.1E, DOE-2.2, DOE-2.3, and EnergyPlus, chiller performance
data over a wide range of operating conditions must be collected. From these
data, regression models of the chillers can be made so that the chillers perfor-
mance can be accurately modeled.
A sample performance spreadsheet (Chiller Bid Form) is included as sup-
plemental material for this SDL (available at ashrae.org/CHWSDL). The
form includes pricing and also generates full-load and part-load performance
data tabs based on design conditions that the chiller vendor must complete. The
performance data tabs are designed to ensure that all operating conditions
expected for the particular project are within the bounds of the collected data
so that regression models do not need to extrapolate. For instance, part-load
data are collected at the chillers minimum and maximum lift conditions.
The spreadsheet includes a macro that generates the regression coefficients
for DOE-2.1E, DOE-2.2, DOE-2.3, and EnergyPlus chiller models once the
data fields have been completed. The regression coefficients can then be
entered directly into the programs to create accurate models of the proposed
chillers.
Procurement Step #4: Adjust for Other First-Cost Impacts

After obtaining chiller bids, adjustments must be made to account for other
installation cost impacts. Chiller prices obtained through the bidding process
are not necessarily directly comparable. Here are some reasons prices may not
be apples to apples:
One option may have more chillers than another (for instance, three versus
two). The added cost of rigging, piping, wiring, and controls for the addi-
tional chiller must be accounted for.
Three-pass evaporators can increase piping costs compared to two-pass
evaporators. Sometimes, depending on the plant layout, the opposite may
be true.
Open-drive compressors require more chiller room cooling than hermetic
motor options. In many climates, cooling is not needed at all for chiller
rooms housing hermetic chillers because the room has very few heat
sources (only those due to the inefficiency of pump motors) and substantial
heat sinks (e.g., heat flow into CHW piping and evaporators). But heat
gains from open-drive chiller motors are substantial, requiring mechanical
cooling of the chiller room in almost all climates. (The energy used by this
mechanical cooling also must be addressed in the energy calculations.)
One option may have less efficient chillers, which can affect electrical ser-
vice costs.
One proposal may include a control system that can be directly connected
to the BAS network, while another may require a field-installed gateway.
For the first pass of the LCCA, these secondary first-cost impacts can be
roughly estimated. Then, if the chillers for which rough estimates were made
end up with the lowest life-cycle costs, a more accurate estimate of the second-
ary costs should be made. This is best handled by a contractor. If a contractor is
not part of the design team, a local mechanical contractor who is likely to bid
the project should be consulted. Many will provide rough estimating services
at no cost to the projectthey write off the costs internally as a marketing
expense, and they also have the advantage of gaining project familiarity prior
to bid.
Procurement Step #5: Estimate Utility Costs

One of the most important elements of the CHW plant procurement pro-
cess is accurately estimating the utility costs of each chiller option. Utility
costs include electricity, gas (if applicable), and water (especially if both water-
and air-cooled chillers are being considered). The level of accuracy and detail
necessary for energy calculations depends on the projects size and engineering
budget. The primary engineering cost of the computer model is not creating the
chiller plant model itself but creating the model of the building and systems
generating the plant load.
Many modern projects include energy modeling in order to meet energy

code requirements (e.g., the energy cost budget approach to ANSI/ASHRAE/
IES Standard 90.1 compliance [2016]), utility incentive programs, or green
building rating programs such as U.S. Green Building Councils (USGBCs)
Leadership for Energy and Environmental Design (LEED). In this case, the
added cost to perform chiller life-cycle costing is relatively low, particularly
when the design team becomes more adept at the process over time.
For projects that are not using energy modeling for other reasons, the cost
of creating the plant profile can be substantially reduced by using prototypical
profiles developed from generic models of buildings with similar occupancies
and HVAC systems. For instance, the eQUEST program (a free front-end inter-
face to DOE-2.2) has a wizard that can create a prototypical building in less
than an hour.
For existing projects, measured performance data may be used.
See Chapter 2 for a more detailed discussion of load profiles and energy
simulation programs.
Once performance bids from chiller vendors have been obtained and a
computer model of the building and HVAC systems is complete, create regres-
sion model coefficients of each of the proposed chillers using the macro within
the performance spreadsheet (Chiller Bid Form) provided with this SDL
(available at ashrae.org/CHWSDL). Then enter the regression coefficients into
the simulation program.
The following factors should be considered in the simulation.
Utility Rates
Utility rates vary over time and are difficult to predict, particularly given the
current transition from regulated to unregulated utility markets. There are a num-
ber of energy forecasts that project electricity and natural gas costs up to 30 years
in the future (e.g., those of the Department of Energy/Energy Information
Administration [DOE/EIA], American Gas Association [AGA], Gas Technology
Institute [GTI]) that attempt to account for the impact of deregulation as well as
world oil and gas reserves. However, these reports are typically generalized for
the nation or the world as a whole and may not reflect local utility rate trends.
Given this uncertainty and for the sake of simplicity, it is typically assumed that
current rate structures will be in effect during the chiller plants life cycle. The
LCCA (discussed below [Procurement Step #7: Calculate Life-Cycle Costs])
includes utility escalation rates that can roughly address future rate changes.
Virtually all utilities charge both for energy consumption as well as for
power demand. Because chillers are one of the largest energy users in typical
buildings, it is essential that demand charges be properly taken into account.
This is particularly true when demand charges are ratcheted, meaning the
owner pays some percentage of the maximum peak demand over the year,
regardless of actual monthly demand. Most hourly simulation programs allow
rate schedules to be entered explicitlythis is usually much more accurate
than using effective annual rates.
Control Logic and Set Points

The optimum control logic will vary for different chillers. It is therefore
essential to iterate on control logic and set points in order to simulate the
chillers as they will perform (presuming of course that sequences are custom-
ized to the chiller as recommended in Chapter 7).
Unfortunately, most simulation programs are not able to precisely model
the sequences recommended in Chapter 7. But they can get close. The follow-
ing are some control variables and logic that should be optimized in the model.
Tower Fan Control (Condenser Water Set Point Reset)

As noted previously, condenser water temperature/tower fan control
options in energy programs are limited. It is recommended that various tower
control strategies that may be available in the simulation tool be modeled to see
which option is best. The actual logic recommended in Chapter 7 will (accord-
ing to theoretical optimum plant performance [TOPP] models) be better, but
optimizing within the available program model options should be close enough
for an apples-to-apples comparison of the chiller options. For instance, one or
more of the following control models for condenser supply temperature set
point should be available in the simulation program; all of them should be tried
to see which results in the lowest chiller energy usage.
Constant set point: Iterate on the set point from the minimum allowed by
the chiller manufacturer up to the design set point. The optimum is likely to
be closer to the minimum for most centrifugal chillers, particularly those
with VFDs. A set point equal to the design wet-bulb temperature has
proven to be a surprisingly effective set point for constant-speed chillers.
Wet-bulb reset: This is usually modeled as a constant approach to outdoor
air wet-bulb temperature within limits. The offset and limits should be
adjusted to minimize energy usage. As noted in Chapter 7, this strategy
requires that a humidity or wet-bulb sensor be installed in the control sys-
tem, which increases maintenance costs and reduces reliability.
Load reset: This is usually modeled as a linear reset from a given tempera-
ture at 100% load to some lower set point at a lower load percentage. The
various set points should be adjusted to find best performance. This logic is
the closest to the sequences recommended in Chapter 7.
Optimum tower set point: Some energy analysis tools dynamically deter-
mine the tower set point to optimize the sum of chiller plus tower energy
use.
Staging
Fixed-speed chillers are almost always optimally controlled by running the
fewest chillers necessary to meet the load. This is easily modeled by most plant
simulation tools and is usually the default control strategy.
Variable-speed chillers, however, use less energy when they are operating
at low-lift conditions, except at very low loads. Therefore, a plant with multiple
variable-speed chillers will be more efficient when more chillers are operating
at part load than if fewer are operating near full load, despite the additional
energy used by CHW and condenser water pumps. (See the Control Sequences
section in Chapter 7 for an additional discussion of optimizing CHW plant
design.) For estimating energy usage of chiller plants with multiple variable-
speed chillers, the staging on/off point should be tested to find the optimum.
Chilled-Water Reset
As noted in Chapter 7, CHW temperature reset is almost always the best
strategy versus a fixed set point, even for plants with variable-flow variable-
speed CHW pumping systems. Reset strategies available to most simulation
tools are limited. Here are a few common models:
Reset by load (valve position): This is the most effective reset logic, as
noted in Chapter 7.
Reset by outdoor air temperature: This model inversely resets CHW tem-
perature proportionally to outdoor air temperature.
The simulation programs ability to accurately simulate reset is a signifi-

cant issue. The program must be able to model the effect reset has on cooling-
coil effectiveness; it requires an accurate coil model and most programs
(including DOE-2.1 and 2.2) have poor models that do not properly account for
variations in coil performance with flow and entering water temperature. If all
the chillers being evaluated are of a similar type (for example, all are centrifu-
gal chillers), it may be best not to model reset to avoid skewing the results due
to coil modeling errors.
Variable-Flow Pumping Systems

Most plant computer models are limited in their ability to accurately model
variable-flow pumping systems, particularly primary-only systems. It is often
necessary to fake modeling the actual system in the program by changing
default. performance curves or adjusting pump heads. This can require consid-
erable judgment on the part of the engineer doing the modeling. Fortunately, in
most cases, the pumping scheme is the same for all the chiller options being
considered, so errors in the model tend to cancel out when options are com-
pared.
The important differences between chiller options that must be accounted for
are variations in condenser and evaporator pressure drops and, for variable-flow
primary systems, the minimum and maximum evaporator flow rates. Pressure
drop differences can usually be accurately accounted for by adjusting pump
heads in the computer model and, in any case, have a small impact on overall
energy. Differences in minimum and maximum flow rates with variable-flow pri-
mary systems cannot be modeled well using current simulation programs. How-
ever, the impact of varying minimum flow rates is usually small unless there are
large differences between options and many hours when the plant operates at low
loads. The impact of varying maximum flow rates is seldom an issue because the
maximum rates seldom occur unless the system experiences very bad T degra-
dation.
Variable-flow condenser water systems are also not modeled well with
some programs (including DOE-2.1) because they use condenser water supply
temperature in chiller regressions. This is accurate for constant condenser
water flow but not for variable flow because it is condenser leaving water tem-
perature that correlates to condenser temperature and pressure (which is the
drive for chiller efficiency), and leaving condenser water temperature rises as
condenser water flow rate falls. DOE-2.3 and EnergyPlus include improved
models that account for condenser water flow variation.
Procurement Step #6: Estimate Maintenance Costs

Maintenance costs are more difficult to estimate accurately than energy
costs. There are few data available indicating the relative maintenance costs of
various chiller types (for example, screw, centrifugal, absorption, engine-
driven) and among the various manufacturers of each chiller type. Manufactur-
ers make claims about their products advantages, but they seldom have hard,
independently collected data to support those claims. To further complicate the
issue, annual maintenance costs are not constant. The costs are low for the first
few years, jump during years when a complete overhaul is required, and
increase gradually as equipment wears. The length of time that different pieces
of equipment last before they must be replaced also varies, although usually
this is not an issue in chiller selection unless very long life cycles are analyzed.
Because maintenance costs are difficult to estimate, they are often ignored
in the chiller selection life-cycle costing and considered only as a soft issue
when making the final chiller selection. This is probably a reasonable approach
when the number and types of chillers are the same. However, when the num-
ber of chillers in each option varies, both air- and water-cooled options are
being considered, or chillers of different types (e.g., electric and gas-engine
driven) are being considered, maintenance costs need to be included explicitly
in the life-cycle cost calculation for best results.
One method to estimate long-term maintenance costs is to request a five- or
ten-year full maintenance contract bid, perhaps including refrigerant replace-
ment, from each manufacturer. This gives some measure of the manufacturers
costs for maintenance. This approach should be used with care; sometimes
quoted costs and actual maintenance costs may not jibe, depending on the will-
ingness of the vendors (who are not insurance or actuarial professionals) to
take on risk.
Factors that affect maintenance costs include the following:
Gear versus direct drive: Gear-drive machines have slightly higher mainte-
nance costs than direct-drive machines.
Open-drive versus hermetic motors: Open-drive machines have slightly

higher maintenance costs than hermetic machines, although they are less
costly to repair should there be a motor burnout. (In chillers with hermetic
motors, motor burnout typically contaminates the refrigerant.)
Variable-frequency drives (VFDs): VFDs introduce another component
subject to failure into the system. Few data are available on their reliability
because the latest generation of VFDs is fairly new. It is also likely that
VFDs will need to be replaced one or more times in the chillers lifetime.
Accounting for VFD replacement cost should be considered if the years of
analysis exceeds 15 years, the typical expected service life of VFDs.
Varying manufacturer quality: There are many claims and anecdotes but
few data on who makes the best chiller. For the purpose of maintenance
cost analysis, it is generally assumed that if the chiller manufacturer is on
the bid list, they make a reliable product.
Compressor type: Screw chillers usually require less maintenance than cen-
trifugal chillers, although this varies from manufacturer to manufacturer.
Number of chillers (for options where the number of chillers differs): The
more chillers in the system, the higher the maintenance costs will be, even
for the same total plant capacity. The costs for maintaining a chiller are not
strongly dependent on the chillers size. If costs are not available, a reason-
able estimate of annual maintenance cost per water-cooled chiller is about
$2000 to $4000 for the first few years, with an additional $1000 per year in
repairs after the first five years or so. There is a slight increase in the per-
machine maintenance cost at about 1000 tons, and there may be another
incremental jump at 2000 tons.
Air versus water cooled: Whether the system is air or water cooled has by
far the most significant impact on maintenance costs among the issues
listed here. Water-cooled systems always cost more to maintain due to the
constant water treatment requirements and the need for regular tube clean-
ing. Water-cooled chillers also generally last longer, particularly in harsh
environments such as near oceans where salt in the air can significantly
shorten the life of air-cooled condensers.
To estimate the differences in maintenance costs between air- and
water-cooled systems, request input from local HVAC service companies.
ASHRAE Handbooks (particularly ASHRAE HandbookHVAC Applica-
tions, Chapter 37 [2015]) also provide some maintenance cost information.
Absent other information, a reasonable estimate of annual maintenance
cost savings is $1000 to $2000 per year per chiller for the first five to ten
years (plus the cost of cooling tower maintenance and chemicals).
Gas-engine-driven chillers: Gas-engine-driven chillers are known to have
much higher maintenance costs than either electric or absorption chillers.
These chillers have frequent engine-related maintenance requirements
(e.g., for spark plugs, oil and oil filters, air cleaners, belts) that are not
required for other chiller types. Manufacturers of these chillers can provide
estimates of the frequency and cost of this work.
Procurement Step #7: Calculate Life-Cycle Costs

The life-cycle cost of a chiller plant is the present value of the total cost of
owning and operating the plant over a specified period of time. A detailed
description of the parameters used in calculating life-cycle cost is beyond the
scope of this SDL. For more details, refer to the 2015 ASHRAE Handbook
HVAC Applications, Chapter 37 and other engineering manuals. Below is a
brief summary of the relevant variables and formulas for calculating life-cycle
costs. These formulas should be entered into a spreadsheet so that the sensitiv-
ity of various assumptions can be evaluated. (See the Case Study section for an
example that uses these equations.)
The life-cycle cost can be calculated using the following equation:
N
UC j + MC j
LCC = FC + ---------------------------
1 + d j
- (6-1)
j=1
where
LCC = present value of the owning and operating costs of the chiller
plant
FC = first costs of the plant. In this case, this is the cost of the chillers
as proposed by the vendor adjusted for associated installation
factors
UCj = plant utility costs for year j. If there is more than one utility type
(e.g., electricity, gas, water), this component is duplicated for
each
MCj = relative maintenance costs for year j
d = discount rate, also called the cost of capital or the minimum rate
of return. This rate is used to discount future cash flows,
converting them to present costs. The higher the rate, the less
future energy savings will help offset the first-cost penalty of a
more expensive chiller plant. The discount rate for a typical
business owner might be 8% to 15%, reflecting the cost of
borrowing money plus a few percentage points to reflect the
investment risk. A more progressive owner or government entity
may be less conservative and use rates near 5%
N = number of years of analysis, or life cycle. While called the life
cycle, N is seldom equal to the actual number of years the chiller
plant will operate. Most studies only look at the first 10 to 15
years because there are so many uncertainties in the utility costs
and overall energy usage the further into the future one looks.
Also, most businesses will expect a return on investment in well
less than 10 years
The life-cycle cost of Equation 6-1 does not include tax impacts, deprecia-
tion, investment tax credits, financing costs, or salvage value. Please refer to
more detailed texts on life-cycle costing if these factors are considered signifi-
cant enough to take into account.
If energy and maintenance costs are constant, Equation 6-1 can be simpli-
fied to
LCC = FC + PWFe UC + PWFm MC (6-2)
where
1 + e' N 1
PWF e = ------------------------------
e' 1 + e' N
d e
e' = ----------------
1 + e
1 + m' N 1
PWF m = --------------------------------
m' 1 + m' N
d m
m' = ------------------
1 + m
UC = utility costs at current rates

MC = maintenance costs at current prices
e = escalation (inflation) rate for electricity (or whatever fuel type is
being used) above current rates. Estimating escalation rates is very
difficult in this transitional market between regulated and
unregulated utilities. See the discussion under the Utility Rates
section above
m = maintenance cost escalation (inflation) rate. This can typically be
assumed to be equal to the consumer price index (CPI), although,
like energy escalation rate, this is difficult to predict with any
certainty. A 1% escalation is usually reasonable
PWF = present worth factor
Procurement Step #8: Final Chiller Selection

Evaluating chiller options using the procedures recommended in this SDL
requires making many assumptions and simplifications. To pick the best
chiller plant option, it is important to test the sensitivity of various assump-
tions. For instance:
If there is uncertainty about the loads the plant will need to handle, develop
energy costs under various load profiles. For example, a plant may be
expected to ultimately handle a large data center load, but it is possible,
even likely, that the actual load will be smaller. Be sure that the plant can
efficiently operate at low loads and that the plant that proved to be optimum
under high-load conditions is also near optimum under low-load condi-
tions.
Utility rates may change dramatically when utility markets are fully dereg-
ulated. Experiment with rates that include very high on-peak energy
charges and demand charges. For mixed-fuel plants, experiment with esca-
lating rates for one utility and deescalating rates for the other.
Life-cycle cost assumptions, such as discount rate and years of analysis,
affect how much future energy savings are weighted. Experiment with var-
ious assumptions to see how they affect the results.
More than likely, after calculating results over a range of assumptions, the
life-cycle costs of several chiller options will be close to the optimum option
(the one with the lowest life-cycle cost). It is also possible that the optimum
choice will vary depending on the assumptions made. In this case, the final
selection must also consider soft factors (those to which a dollar amount can-
not be easily attached) to break the tie. The final selection should be made by
the entire design and ownership team, not just by the engineer. This will ensure
that all soft issues have been considered and everyone had a fair chance to
express any vendor preferences.
These soft factors include the following:
Reliability and reputation of the manufacturer and local representation:

Most facility owners and operators will have a favorite chiller vendor based
on past experience. This can be a very important tie breaker when making
the final selection. If the owners favorite vendor offers an option that has
close to the optimum life-cycle cost, it may be a politically sensible idea to
choose that option.
Refrigerant type: Even if a range of refrigerant options is allowed in the bid
specifications, the owner or engineer may have a preference for a given
refrigerant type based on the refrigerants impact on the ozone layer and
global warming, or impending production phase-out dates.
Redundancy: Typically, an analysis of chiller options assumes that the
chiller plant operates normally. But what happens if a chiller fails? Differ-
ent chiller options may accommodate outages better than others. Options
offering the most chillers will generally offer the least exposure to chiller
failure. However, other more complex failure considerations should be con-
sidered. For example, if a plant with unequally sized chillers loses the large
CHW pump, will the flow ranges allow the large chiller to stay on-line
using the small CHW pump?
Procurement Step #9: Chiller Specification

Once the final chiller selection is made, it must be made clear to contrac-
tors bidding the project at the 100% construction documents phase that the
chiller selection is final. The following are guidelines to achieve this.
Final equipment schedules and plans must, of course, reference the
selected chillers. Add NO EQUAL to the remarks column to make it
clear this is the final selection.
Strip the chiller specification of the normal Part 2 Materials Section and
most of Part 1 General; retain Part 3 Installation. In Part 1, state that the
chiller must be the one scheduled and include chiller prices. For example:
a. Provide the scheduled chillers, including the scheduled options for the
following prices:
1. CH-1: $101,406
2. CH-2 and CH-3: $189,450 each
3. Price includes freight but does not include taxes
b. Bids for other chillers will not be accepted.
Because this procedure may be unusual to some contractors, contractor
bids should be reviewed to ensure compliance.
Case Study
This case study illustrates the recommended chiller procurement approach
discussed in this chapter. The case study building is 15 stories high, enclosing
540,000 ft2 in San Francisco, CA. The building primarily contains offices, with
some assembly and retail space on the ground floor, a 5000 ft2 data center, and
a large cafeteria. The data center and various small server rooms operate con-
tinuously and place roughly a 50 ton base load on the plant. Total plant load
was calculated to be 1100 tons. Figure 2-2 in Chapter 2 shows the load profile
for the building generated by a DOE-2.1 simulation.
After the primary design temperatures and flow rates were determined (see
Chapter 5), the performance specification and bid forms were developed (see the
sample spreadsheet bid form [Chiller Bid Form] included with this SDL, available
at ashrae.org/CHWSDL). Four chiller vendors were invited to bid. The data from
the bid forms were used to generate regression coefficients of each chiller in each
bid option. These coefficients were then input into the DOE-2.1 model of the build-
ing. Energy savings were calculated using various control scenarios such as con-
denser water set point reset and staging points. The lowest energy costs for each
option were selected and life-cycle costs were calculated as shown in Table 6-2.
Note that first-cost adjustments were made for additional piping and pump-
ing system costs caused by the use of three-pass evaporators in some cases and
multiple chillers in others. Additional exhaust fan capacity was added for
open-drive machines due to the higher heat load they generate in the chiller
room. (In more severe climates, this high heat load may require that mechani-
cal cooling be added to the room.)
Table 6-2 Life-Cycle Cost Summary
LCC Premium
Total Building
1st Cost Rank
Energy Costs,
Exhaust Fan,
Energy Cost
Description
Other First
Contractor
LCC Rank
Total Cost,
Life-Cycle
Cost Add/
Cost with
Markup,
vs. Base,
Deduct,
Option
Added
Rank
Cost,
$
$
$
$
$
400 ton, 0.50 kW/ton;
1 268,235 6000 0 274,235 6 712,293 9 6,371,092 142,016 10
700 ton, 0.55 kW/ton
400 ton with VFD, 0.50 kW/ton;
2 301,235 6000 0 307,235 9 694,427 2 6,251,168 22,092 4
365 ton, 0.56 kW/ton;
3 199,980 3000 0 202,980 1 724,350 12 6,403,038 173,962 12
4 240,130 0 0 240,130 3 702,168 5 6,250,322 21,246 3
5 735 ton with VFD, 289,190 0 0 289,190 8 694,854 4 6,236,778 7702 2
0.50 kW/ton
200 ton,
6 0.50 kW/ton; 900 ton dual 212,031 0 0 212,031 2 712,100 8 6,307,235 78,159 5
0.54 kW/ton
550 ton dual, 0.56 kW/ton;
7 256,485 0 0 256,485 5 714,269 11 6,370,255 141,179 9
550 ton dual, 0.56 kW/ton
400 ton dual, 0.53 kW/ton;
8 254,533 0 0 254,533 4 711,137 7 6,341,495 112,419 8
700 ton dual 0.53 kW/ton
200 ton, 0.53 kW/ton;
9 350 ton dual 0.57 kW/ton; 252,485 25,000 0 277,485 7 712,547 10 6,376,516 147,440 11
550 ton dual 0.59 kW/ton
10 307,106 6000 3000 316,106 11 702,969 6 6,333,154 104,078 7
11 305,151 6000 3000 314,151 10 691,038 1 6,229,076 0 1
12 366 ton, 0.51 kW/ton 338,320 33,000 3000 374,320 12 694,806 3 6,321,497 92,421 6
The total building energy cost column was taken from the DOE-2 model.
Life-cycle costs were calculated based on a 15-year life, 8% discount rate, and
0% energy escalation rate.
Option 11 has the lowest life-cycle cost, but Option 5 is very close and even
Options 2 and 4 have life-cycle costs close enough to that of Option 11 that
they need to be considered. Ultimately, Option 5 was selected for the project
based on the following soft considerations:
This option uses R-134a, whereas Option 11 uses R-123. R-123 was con-
sidered acceptable, but R-134a was preferred by the owner because it has
zero ODP and was not scheduled to be phased of production at the time of
design for this project.
Both chillers in Option 5 have VFDs, whereas only the small chiller in
Option 11 has a VFD. Having two VFD chillers improves redundancy
because either chiller should be able to handle the low nighttime data cen-
ter loads.
Both chillers in Option 5 have design CHW and condenser water flow rates
and minimum flow rates that overlap so that the small CHW pump is large
enough to keep the big chiller on-line. That is not the case with Option 11.
In case of any pump failure, the large chiller can remain on-line with
Option 5, whereas if one of the large pumps fails, only the small chiller can
remain on-line with Option 11.
Option 5 uses hermetic motors whereas Option 11 uses an open drive. Both
have advantages and disadvantages, but the owners engineer preferred her-
metic motors based on past seal problems experienced with open-drive
machines. The hermetic motors also made the chiller room have a net neu-
tral load so no conditioning was required.
Option 5 is less expensive by about $15,000, making it easier to reach first-
cost budgets.
After the bid, chiller vendors were given a modified version of Table 6-2
that included only the columns for first-cost rank, energy cost rank, life-cycle
cost savings versus base, and life-cycle cost rank. This allowed the vendors to
see how their proposals compared to their competitors without seeing actual
pricing.
Simplified Procurement Procedure

The selection procedure described in this chapter takes considerable time
and cost on the part of both the design engineer and chiller vendors. For many
projects, the schedule may not allow time for this rigorous analysis or the cost
may not be in the design budget. On some projects, the design team may not
have the expertise to perform the required energy modeling and life-cycle cost
calculations. In these cases, instead of falling back on the traditional, almost
arbitrary, selection process, we suggest using the following simplified proce-
dure:
1. Once loads are confirmed (typically at the end of the design development
phase), complete the simplified chiller bid form (available at ashrae.org/
CHWSDL) included with this SDL with project design data. In most
cases, the designer would predetermine the size of each chiller, but that is
not mandatory. For instance, pricing for a two-chiller plant could be
solicited for both equally sized chillers and for 1/3- and 2/3-sized
chillers.
2. Send the simplified bid form to vendors for pricing and performance data.
3. Select chillers based on first costs and a subjective judgment of energy
costs based on AHRI rated data.
4. Hard-spec the chiller per Procurement Step #9: Chiller Specification
above.
Example
An office building chiller plant is composed of two 300 ton chillers serving
a primary variable-flow distribution system. The simplified chiller bid form
was completed by four chiller vendors with a total of seven options. Results are
summarized in Table 6-3.
From these options, by inspection based on first cost, energy efficiency,
historical maintenance, acoustics, and experience with similar chillers on other
projects, Chillers 2, 3, and 6 were deemed the best three options. Each has
strengths and weaknesses:
Option 2 is the most expensive and loudest but also has the highest pub-
lished efficiency and, because it has a screw compressor, can have lower
maintenance costs.
Option 3 is the least expensive despite having magnetic bearings. Magnetic
bearings reduce noise and maintenance costs and provide good low-load
performance and stability. This chiller also needs no head pressure control
because it can operate with negative lift. While its published efficiency is
not as good as the other two options, it has the lowest minimum CHW flow
rate (35%), which will result in lower CHW pump energy.
Option 6 has good efficiency but has higher pressure drop (higher pump
energy) and is physically much larger than the other two options. It also
uses R-123, a refrigerant scheduled to be phased out of production during
the chillers service life.
Ultimately, the design team chose Option 3, in part because it had lower
first costs.
This example demonstrates the following:
This simplified approach is obviously not as comprehensive as the recom-

mended procurement procedure, but it should be more effective than arbi-
trary chiller selection. Just by inspection, for example, it may be clear that
the superefficient chiller the design team was considering (e.g., Option 7) is
simply too expensive and not likely to be cost-effective.
It also allows the design team to be able to design around a specific chiller.
For instance, premium-efficiency chillers are often physically larger; if the
premium-efficiency option is selected, the mechanical room can be sized
accordingly. Conversely, if a smaller chiller is selected, the mechanical
room can be reduced in size.
Table 6-3 Simplified Chiller Bid Summary

Option 1 2 3 4 5 6 7
Price for one chiller
$118,750 $116,850 $99,595 $81,222 $108,850 $105,270 $154,760
(including freight)
Magnetic Ceramic
Tri-rotor Tri-rotor Twin
Compressor type bearing Centrifugal Centrifugal bearing
screw screw screw
centrifugal centrifugal
Refrigerant type R-134a R-134a R-134a R-134a R-123 R-123 R-123
Operating weight,
16,554 15,988 11,275 13,386 16,708 16,394 19,831
lb
Refrigerant weight, lb 950 840 558 574 500 500 750
Delivery lead time, weeks 9 9 10 9 7 7 11
VFD, Y/N Y Y Y Y Y Y Y
% of design capacity
10% 10% 10% 10% 10% 10% 6%
below which chiller cycles
Minimum lift (condenser
return temperature
[CWRT] chilled-water
12.6 12.6 6.0 15.0 13.0 13.0 13.0
supply temperature
[CHWST]) at minimum
load, F
CHW pressure drop,
8.5 22.5 14.4 19.7 4.97 16.48 7.72
ft
Minimum CHW flow
43% 29% 35% 53% 46% 31% 36%
rate, %
Condenser water (CW)
pressure drop, 7.6 6.8 11.8 9.38 16.91 16.91 17.03
ft
Sound power dB(A) 82 82 74 88 77 77 70
NPLV 0.331 0.335 0.362 0.415 0.358 0.348 0.341
kW/ton at 25% load 0.346 0.347 0.364 0.375 0.373 0.356 0.342
Design kW/ton 0.537 0.542 0.525 0.609 0.552 0.554 0.487
Even a subjective review like this can help make a better decision than the
conventional arbitrary process. For instance, in this example, the three
finalists were very different chillers (one screw, one magnetic bearing cen-
trifugal, and one R-123 centrifugal). Using a conventional arbitrary selec-
tion, one type of chiller would have been specified, typically eliminating
the other options.
Prior to bid, the design team thought magnetic bearing chillers would be
prohibitively expensive. That was not the case. Again, using an arbitrary
selection process, the chiller that was ultimately selected might not have
even been considered.
The bid process is fairly painless for the vendors because all of the data in
the bid form are available from their standard selection software reports.
Efficiency ratings are based on AHRI Standard 550/590 and not adjusted
for zero tolerance, which greatly simplifies the completion of the form.
There is little value to getting zero tolerance efficiency data because energy
costs are based on judgment, not rigorous analysis.
The lack of objective energy and life-cycle cost calculations can make the
final selection more difficult, as shown in the example where it was not
easy to make a selection among the three finalists. The more comprehen-
sive procurement procedure is therefore the better option when possible
within project time and budget constraints.
References
AHRI. 2015. AHRI Standard 550/590-2015, Performance rating of water
chilling packages using the vapor compression cycle. Arlington, VA: Air-
Conditioning, Heating, and Refrigeration Institute.
2015. ASHRAE. Chapter 37, ASHRAE HandbookHVAC applications.
Atlanta: ASHRAE.
ASHRAE. 2016. ANSI/ASHRAE/IES Standard 90.1-2016, Energy standard

6-1 The benefits of a performance-based chiller bid approach include which of the
following:
a. It can minimize life-cycle costs for the owner.
b. It solidifies selections in the design phase, allowing the designer to
accurately lay out the plant and avoid design rework by the contractor
during the construction phase.
c. The chiller selection process becomes based on objective criteria rather
than subjective criteria that contractors may use when including chillers
in their project bids.
d. All of the above.
6-2 Which of the following parameters is not appropriate to include in a performance-

based bid approach:
a. The anticipated cooling load profile for the plant.
b. Total required plant capacity.
c. Chiller efficiency metrics.
d. Sound power limits.
6-3 The following three selections are provided by a chiller manufacturer.
Chiller 1 Chiller 2 Chiller 3

200 ton centrifugal 200 ton screw 200 ton centrifugal
Two-pass evaporator Two-pass evaporator Three-pass evaporator
Open-drive compressor Hermetic compressor Hermetic compressor
CHW pressure drop: 5 ft CHW pressure drop: 10 ft CHW pressure drop: 15 ft
Full-load kW/ton: 0.5 Full-load kW/ton: 0.6 Full-load kW/ton: 0.55
Native BACNet Native BACNet BACNet gateway required
Variable speed Variable speed Variable speed
Which option is likely to have the highest first cost, not including the cost
of the chiller itself?
a. Chiller 1
b. Chiller 2
c. Chiller 3
d. Indeterminate based on the provided data
6-4 When estimating annual plant utility costs with simulation modeling
a. Using a flat effective annual rate that accounts for both electricity and
demand is usually sufficiently accurate.
b. Annual escalation of utility costs should be ignored because of the
unknown future variation of utility prices.
c. Using real rate schedules that account for both electricity and demand
independently should be used.
d. Annual escalation of utility costs should be estimated based on fore-
casts provided by DOE/EIA.
6-5 Using zero tolerance chiller performance data in bid forms
a. Greatly increases the time needed to complete the forms by vendors.
b. Better ensures the best chillers are selected, particularly for plants
expected to operate many hours at low load.
c. Is not recommended when using the simplified procurement procedure.
Controls
Instructions
Introduction
This chapter begins with a discussion of some general factors that should
be considered when designing and installing a CHW plant BAS and concludes
with suggested control sequences. It is assumed that the BAS is a DDC system.
The two terms (BAS and DDC system) are used interchangeably in this chapter.
To maintain stable and effective plant operation, designers must ensure that
the control systems are of high quality and are sufficiently accurate and reli-
able, balanced with cost considerations. Also, the CHW plants control system
must have sufficient programming and data manipulation and trending capabil-
ities. Most importantly, the control system must be programmed with control
sequences that optimize, or nearly optimize, energy performance while provid-
ing reliable operation. The following topics related to controls and instrumen-
tation in CHW plants are covered:
Types of sensors available for energy monitoring and control

Styles of and selection criteria for control valves
Controller requirements and interfacing issues
Performance monitoring
Types and configuration of local instrumentation
Control sequences for CHW plants
Industrial Versus Commercial Controls

Many older CHW plants were designed using industrial control systems,
including programmable logic controllers (PLCs) and industrial sensors and
instrumentation. That was justified by the lack of robust and reliable commer-
cial DDC systems in the 1980s and early 1990s. But almost all modern com-
mercial DDC systems have the capability to optimally control and monitor
CHW plants of virtually any size and complexity. Commercial systems are
208 Chapter 7 Controls
usually easier to program and maintain (they are simpler and more operators
are familiar with them) and always cost less to install, typically half the cost or
less. But industrial controls do have one key advantage: the ability to integrate
redundant controllers to allow seamless operation should a controller fail. It is
possible to design commercial control systems with redundant controls, but it
is complex and difficult to implement, in part because it is not common prac-
tice. The complexity of programming multiple controllers to provide backup
operation in case of controller failure can ironically lead to unreliable perfor-
mance. Typically with commercial control systems, controller failure is
addressed solely by manual interventiona building operator must be made
aware of the failure (in itself a problem if alarms and alarm broadcasting are
also provided by the failed controller) and have time to manually control
pumps, chillers, valves, etc., before the system loses control of the cooling
load. Manual intervention is a reasonable approach for noncritical plants such
as those serving offices and other commercial occupancies. However, it may
not be reasonable for plants serving critical data centers where loads are so
high that failure of the chiller plant for even minutes may cause loss of tem-
perature control and computer equipment failure. Industrial control systems
may be appropriate for plants serving these and other very critical applications.
Commercial controls are generally preferred otherwise.
Choosing Control and Monitor Points

If the control and monitor points are not carefully chosen, the system may
suffer from either under- or overinstrumentation. An underinstrumented sys-
tem may be difficult to control optimally, whereas an overinstrumented system
may be confusing to the operations staff, expensive, and difficult to maintain.
The selection of control and monitor points should be based on a careful
analysis of the chiller plants control and operating requirements. Each point
must meet at least one of the following criteria:
1. Essential: The point is necessary for effective control of the chiller plant as
required by the sequence of operations (SOO) established for the plant.
2. Desirable: The point provides information such as status and alarms that
can be used by plant operators to ensure that the plant is operating properly
or to indicate a problem has occurred or may soon occur.
3. Optional: The point may be used to gather accounting or administrative
information such as energy use, efficiency, or runtime, if such monitoring
is desired by the plant operator or owner.
Because each plant has individual operating requirements for control logic,
staffing, accounting, and operations and maintenance, there is no single stan-
dard for determining exactly what instrumentation is required for any plant.
The decisions must be made by the designer with owner input and concur-
rence.
The types and selection of sensors and control devices are discussed first.
This is followed by an example of essential, desirable, and optional points.
Sensors
Sensors are used in CHW plants for a variety of measurements, including:
Temperature
Humidity
Liquid flow
Pressure
Electric current
Electric power
Gas flow
The sections that follow discuss the end-to-end accuracy requirements and
the types of sensors available for each type of measurement, their measuring
characteristics, and special factors to consider. Installation and calibration
information is also provided.
End-to-End Accuracy
Both accuracy (the ability to measure an actual value) and resolution (the
ability to sense changes in the value) need to be considered together in select-
ing sensors. If accuracy is considered without looking at resolution, or vice
versa, the measurement objective may not be satisfied.
The inherent accuracy of the sensing device is the primary factor in deter-
mining the overall accuracy of the measurement. Other factors are
the means of transferring the signal from the measurement device to the
control system and
the signal span and number of bits used in the control systems analog-to-
digital (A/D) converter.
Whenever possible, it is useful to specify end-to-end measurement property

requirements. Typical and suggested end-to-end accuracies for common chiller
plant control points are shown in Table 7-1. These accuracies are readily
achievable with common commercial DDC systems and are usually all that is
required for acceptable plant performance. Increased accuracy may be
achieved but usually at added cost. Additional discussion on accuracy require-
ments is included in specific sensor sections.
Table 7-1 Typical End-Use Accuracy

Measured Variable Reported Accuracy
Outdoor air dry-bulb temperature 1F
CHW and condenser water temperatures at
0.2F
central plant mains
CHW and condenser water temperatures
0.5F
elsewhere
Water T (supply to return) 0.15F
Relative humidity 5% rh
Water flow 1% of full scale
Water pressure 2% of full scale
Electrical power 1% of reading, 3 kHz response for VFD-driven equipment
Temperature Sensors
Types of SensorsTemperature Sensors
The most common temperature sensing devices used in hydronic applica-
tions are
Thermistors
Resistance temperature detectors (RTDs)
Integrated circuit (solid-state) temperature sensors
These all use materials whose resistance or impedance changes with tem-
perature.
Thermistor signals are nonlinear (Figure 7-1) and thus must be converted to a
temperature reading from the resistance reading. Modern DDC systems include
the conversion of standard thermistors either in firmware or software so that
thermistors may be directly connected to them without a transmitter. The connec-
tion is then inexpensive and accurate because the preset scaling ranges on these
systems can automatically adjust for the nonlinear signal. Thermistors generally
cannot be calibrated on the device itself, although some DDC manufacturers
allow a slope/intercept calibration within the point configuration software.
RTDs, on the other hand, have linear resistances to temperature signals that
do not require any conversion, but many RTDs have very low electrical resis-
tance (e.g., 100 to 1000 ohms) such that the voltage drop in the wiring from the
sensor to the DDC input can skew the signal. For these RTDs, a transmitter is
required. The output of the transmitter is a linear current signal (e.g., 420 mA)
that is independent of wiring length. The transmitters must be calibrated and
limit the range of the output to a fixed temperature range.
Figure 7-1 Typical thermistor resistance versus temperature curve.
Integrated circuit temperature sensors are not commonly used in HVAC

applications except with Btu meters. They are not very accurate (~1.0F) but
they are extremely repeatable and linear. They also do not require calibration.
Hence they are excellent for differential temperature measurement once the
two sensors are matched and calibrated at the factory.
Issues and RecommendationsTemperature Sensors

Thermistors and RTDs each have advantages and disadvantages, but when
properly applied either may be used to measure temperature. The following
issues may impact the choice of temperature sensor.
Cost
RTDs generally are more expensive than thermistors, particularly those for
which transmitters are required. Some high-resistance thin-film RTDs can have
lower costs because transmitters are not required, but these types of RTDs are
also much less accurate.
Accuracy
Depending on the material (platinum is most common) and construction,
RTDs can be purchased with a wide range of accuracies from very accurate
Class A RTDs (as low as 0.02F) to modestly accurate thin film RTDs (typi-
cally 1F) over typical chiller plant temperature ranges. Matched-pair (sen-
sor/transmitter) sensors are available to improve overall accuracy; they use the
tight tolerance of Class A RTDs and a NIST traceable factory calibration pro-
cedure. Thermistors typically are offered with at least two options: standard
and extra-precision. Standard thermistors have an accuracy of 0.4F over
standard temperature ranges while extra-precision thermistors have an accu-
racy of about 0.2F.
Long-Term Stability
RTDs are more stable in the long term than thermistors. However, the
external transmitter circuits on most RTDs can cause readings to drift, so
routine calibration is required. In the past, thermistors were known for drift-
ing over short time periods, but modern thermistors are very stable. Most
manufacturers guarantee drift of no more than 0.05F over a five- or ten-year
span.
Because of their lower first costs, similar or higher accuracy, and reduced
need for routine calibration, thermistors are usually preferred over RTDs for
most temperature sensors in CHW plants. The exception may be those used for
supply and return temperatures that are also used for plant load calculations. To
achieve the 0.15F accuracy suggested in Table 7-1 for T measurement,
each sensor must have an error of no more than 0.1F. This requires a high-
precision, matched-pair RTD transmitter. Alternatively, a Btu meter can be
used in lieu of these sensors and a standalone flowmeter; Btu meters are dis-
cussed in a later section.
Sensor InstallationTemperature Sensors

Indirect well temperature sensors are recommended for measuring water
temperature, so that the sensor may be removed from the well for calibration or
testing. The wells should be brass or stainless steel. Recommended guidelines
for installing well temperature sensors are as follows:
Place the temperature sensor in a well that penetrates the pipe by the lesser
of half the pipe diameter or 4 in.
Install the sensor in the well with a thermal-conducting grease or mastic.
Use a closed-cell insulation patch that is integrated into the pipe insulation
system to isolate the top of the well from ambient conditions but allows
easy access to the sensor.
Locate wells far enough downstream from regions of thermal stratification
or mixing so that the fluids temperature is uniform at the well.
For field calibration and testing, a test plug should be installed adjacent to
the sensor. This will allow the sensor to be tested without removal using a
handheld sensor connected to the test plug.
CalibrationTemperature Sensors
There are two approaches to calibration; each has merit depending on how
critical the measurement applications are.
Factory Calibrated
For noncritical measurement applications, it is usually satisfactory to rely
on the manufacturers temperature tolerance limit for the sensor supplied.
Factory-established temperature tolerance limits generally range about
0.5F for RTD and thermistor sensors. Some manufacturers use the term
pre-calibrated to refer to the tolerance limit. However, this pre-calibration

does not account for all the factors involved in the end-to-end measurement
characteristics of the sensor (see End-to-End Accuracy above).
Field Calibrated
It is recommended for more critical measurement applications (such as
supply and return water temperatures used for plant load calculations) that
each temperature sensor be field calibrated as part of the commissioning
and start-up process (see Chapter 8 for information on commissioning).
Because of the limited rangeability of most chiller plant temperature-sensing
requirements, a single-point calibration is often adequate. An end-to-end
field calibration of all critical devices can enhance the overall accuracy of
the system. However, it is imperative that the temperature-sensing device
used for calibration be at least twice as accurate as the respective field
device. For example, if the temperature sensor is 0.4F accurate, test
equipment must be 0.2F accurate over the same range. For highly precise
sensors such as those used for CHW supply/return temperatures, a dry-well
bath is typically required because handheld sensors with 0.05F accuracy
are rare.
Liquid Flow Sensors

Types of SensorsLiquid Flow Sensors
Fluid flow measurement can be a difficult and costly instrumentation item
for a chiller plant. However, it can also be a critical measurement depending on
control sequences and performance monitoring goals.
Table 7-2 compares various types of common liquid flow sensing devices.
Table 7-2 Comparison of Liquid Flow Sensing Devices

Range of Flow Relative First Measurement Maintenance
Flowmeter Type Configuration
(Turndown Ratio) Cost Accuracy Costs
Orifice In-line Low (5:1) Low 5% Medium
Insertion turbine Insertion Medium (30:1) Medium 2% Medium
Vortex shedding Insertion Medium (30:1) High 2% Medium/low
Single-point magnetic Insertion High (50:1) High 1% Low
Transit time ultrasonic External High (100:1) High 0.5% Low
High up to
Full-bore magnetic In-line Highest (1000:1) 12 in. pipe; 0.5% Lowest
Very high > 12 in.
Issues and RecommendationsLiquid Flow Sensors

Generally, the most important flow measuring characteristics are range and
accuracy. The design engineer must fully understand the expected error at vari-
ous flow levels to be certain the device chosen will meet all requirements.
For CHW and condenser water applications, full-bore magnetic flowmeters
are often the best choice. Where costs are prohibitive on large pipes, single-point
insertion magnetic flowmeters can be used. Full-bore magnetic meters are the
most accurate, require the least maintenance, and are the least susceptible to
errors from variations in flow profile (they require the least amount of space
upstream and downstream of the meter) and the presence of particulates or air in
the pipe. They also have no moving parts in the water stream that can get fouled,
a critical feature with respect to open condenser water flow measurement.
Ultrasonic meters share many of the positive qualities of magnetic flowme-
ters, but they can provide erroneous readings if air or other particles pass
through the meter, such as on open-circuit condenser water systems. They are
also more prone to installation errors; accuracy significantly depends on align-
ment of sensors and thickness of piping. However, they can be a good alterna-
tive to single-point magnetic flowmeters in large CHW piping where full-bore
magnetic flowmeters are cost prohibitive.
Turbine flowmeters are often used to reduce costs where budgets demand,
but they should not be used on open condenser systems where fouling can ren-
der the sensor ineffective in a very short period of time.
InstallationLiquid Flow Sensors

The manufacturers requirements for placement and installation must be
carefully observed to ensure an accurate flow measurement. Because flow
measurements often require specific upstream and downstream lengths of
undisturbed straight piping, it may be necessary to pay special attention to the
piping layout during plant design and construction. Full-bore magnetic flow-
meters are the most forgiving in this regard; they will read flow accurately
unless turbulence is so bad that flow reversal occurs within the bore. On the
other hand, single-point magnetic and turbine insertion flowmeters can require
as much as 15 to 30 pipe diameters upstream of the meter, which is difficult to
achieve in most plants.
Where flowmeters are intended to measure flow that may be bi-directional
(e.g., installations in a common leg or for a TES tank), sensors must be capable
of reading bi-directionally as well. All of the sensors in Table 7-2 are either
inherently bi-directional or available in a bi-directional option, except for vor-
tex shedding meters.
CalibrationLiquid Flow Sensors

Calibrating flowmeters in the field can be very difficult and, more often
than not, is impractical or impossible. Calibration of any device requires field
measurement of the measured variable with a device that is substantially
more accurate than the device being calibrated. With flow measurements, it
is often hard to find an acceptable location for the flowmeter, one with ade-
quate length of straight piping both up- and downstream of the device. It is
therefore very seldom possible to locate an additional, temporary flow sensor
in the system in order to calibrate the installed sensor. Strap-on ultrasonic
meters are often considered, but they can be difficult to properly use and, at
least with current technology, are not usually more accurate than the installed
flowmeter and therefore not appropriate for use in calibration. Use of pump
curves, DP across known devices (such as chillers), or CBVs at pumps or
coils also are not accurate enough for calibration. However, all or some of
these can be used as a reality check to verify that the flowmeter is reading
in the proper range, and often that is all that can be done in the field. There-
fore it is important that flowmeters be accurately factory calibrated and pref-
erably NIST traceable.
Btu Meters
Types of SensorsBtu Meters
Measuring CHW load (in Btu/h, generically abbreviated Btu) is often required
for chiller staging, chiller plant performance monitoring, or submetering CHW
usage, such as at each building of a college campus. The load calculation can be
performed by the BAS from flow and temperature sensors, or it can be done by a
device called a Btu meter. The Btu meter generally is configured to send calculated
Btu data, optionally along with individual temperature and flow measurement data,
to the BAS or other data collection system for monitoring. It may also have a dis-
play for manual reading of internally stored energy usage data.
Btu meters are generally designed to work with any flowmeter and tempera-
ture sensors, but more commonly the Btu meter manufacturer provides the meter,
flowmeter, and temperature sensors as a package as shown in Figure 7-2. With
this style, which is used for larger piping, the flowmeter and temperature sensors
are all field mounted; with some smaller Btu meters, the flowmeter and one tem-
perature sensor are built into the main meter housing. The temperature sensors
are provided with the Btu meter so that they can be factory matched and cali-
brated for improved accuracy. The flowmeter can be any type depending on the
desired accuracy (see the previous Liquid Flow Sensor section). The output of
the Btu meter can be a pulse or analog output connected to a BAS or other data
collection system. Modern Btu meters also include the ability to directly connect
to common control networks such as BACnet/MSTP, BACnet/IP, Modbus/EIA-
485, LonWorks, and various proprietary networks.
Issues and RecommendationsBtu Meters

The main advantage of the Btu meter is that temperature sensors are factory
matched to minimize temperature difference calculation error. However, they
generally cost more than using individual sensors connected to the BAS. Nev-
ertheless, Btu meters are recommended due to their improved temperature
Figure 7-2 Typical thermal energy meter.
measurement accuracy and stability and ease of data collection, particularly if

energy is being metered for revenue purposes (e.g., allocating costs of CHW or
hot-water usage per building).
Installation and CalibrationBtu Meters

See the preceeding Temperature Sensors and Liquid Flow Sensors sections
above for installation and calibration information.
Humidity Sensors
Types of SensorsHumidity Sensors
The most common humidity sensors are either capacitance or resistive type
measuring relative humidity. Common nominal accuracies are 1%, 3%, or 5%
including hysteresis, linearity, and repeatability. These are manufacturer-listed

accuracies. Actual accuracy will vary depending on the quality of the sensor
and how well and how frequently the sensor has been calibrated. Temperature
sensors tend to be very stable and remain accurate for many years (Edwards
1983; Lawton and Patterson 2002). Humidity sensors, on the other hand, are
notorious for being difficult to maintain in calibration. A test of commercial
humidity sensors (NBCIP 2004, 2005) showed that few of the sensors met
manufacturers claimed accuracy levels out of the box and were even worse in
real applications. Figure 7-3 and Figure 7-4 show the results of the NBCIP one
year in situ tests of two brands of humidity sensors among the six brands
tested. There were two sensors tested for each brand, represented by the dark
and light gray dots. Figure 7-3 shows sensor data from the best performing
manufacturer in the study, although even these top quality sensors did not meet
the manufacturers claim of 3% accuracy. Figure 7-4 shows sensor data from
the worst performing manufacturer; both sensors generated almost random
humidity readings.
Issues and RecommendationsHumidity Sensors

In chiller plants, humidity sensing is generally not required unless the
system has a WSE (see sequences below in the section Water-Side Econo-
mizer [WSE] Control), in which case outdoor wet-bulb temperature is
needed to enable the economizer. Wet-bulb temperature is not typically
measured directly; rather, relative humidity and temperature are measured
and wet-bulb temperature is calculated from these readings in firmware or
software. Relative humidity sensors are subject to error due to sensing
Figure 7-3 Iowa Energy Center NBCIP studybest humidity sensor.

Figure 7-4 Iowa Energy Center NBCIP studyone of the worst humidity sensors.
technology, hysteresis, and drift. They are much more likely to drift out of
calibration than temperature sensors and are often out of calibration when
first installed.
The humidity sensor should be specified for 3% rh or lower. Because
only one humidity sensor is needed for CHW plants, and because its accu-
racy will have a significant impact on WSE performance, a very high-quality
sensor from a high-quality manufacturer should be used. Humidity sensors
offered by major control system manufacturers, generally devices manufac-
tured by original equipment manufacturer (OEM) mass production sensor
companies, often are not of sufficient high quality despite their specifica-
tions; sensors from companies that specialize in humidity measurements are
recommended instead. The NBCIP studies (2004, 2005) in the References
should be consulted.
InstallationHumidity Sensors
Outdoor air temperature and humidity must be measured in a location well
protected from direct sunlight and also from exhaust outlets from building fans
and exhaust from cooling tower fans, all of which can skew the readings.
Where no such location can be found (as is typical), a fan-aspirated assembly
can be used and located on the roof far from exhaust plumes. The fan assembly
obviates the need to avoid direct solar exposure.
CalibrationHumidity Sensors
The outdoor air humidity sensor must be field tested initially (see Chapter 8
for information on commissioning) and at intervals recommended by the man-
ufacturer (but no less than semi-annually) using a portable humidity-sensing
device. Because humidity sensors are so prone to drift, it is recommended that
DDC systems with Internet connectivity test sensor calibration by connecting
to the closest National Oceanic and Atmospheric Administration (NOAA)
website and comparing their dew-point temperature reading to that calculated
at the site using outdoor air temperature and relative humidity. The dew point
(absolute humidity) should be similar between the building and NOAA sites if
they are not too far apart geographically; if they are not, an alarm can be gener-
ated alerting building engineers to possible sensor calibration problems.
Pressure Sensors
Types of SensorsPressure Sensors
Pressure is always measured as a DP, either the difference between the
pressures of two fluids or the difference in pressure between a fluid and a refer-
ence pressure. When the reference pressure is atmospheric pressure, the sensor
is referred to as a gage pressure sensor.
The most common means of sensing pressure for fluid conditions are fast-
response capacitance type. Standard commercial-grade sensors offer excellent
accuracy, usually 1% or less of the specified pressure range.
Issues and RecommendationsPressure Sensors

The following are common pressure sensor applications in CHW plants:
Control of variable-speed pumps: The sensors should be located at the

extreme ends of the system as discussed below. Pump speed is modulated
to maintain DP at set point.
Control of flow: DP across a device of fixed geometry can be correlated to
flow using known flow versus pressure drop curves. As discussed in Chap-
ter 4, this is an alternative way to control minimum flow through chillers,
rather than using a flowmeter. However, a flowmeter, which can also more
accurately be used for plant load calculation, is a more common and usu-
ally preferred option.
Monitoring of system pressure: A gage pressure sensor can be installed at
the CHW system expansion tank connection to sense a drop in pressure that
can occur when there is a leak, generating an alarm accordingly. (Note that
makeup water connections to CHW systems should be shut off after air is
removed from the system to avoid exacerbating the damage from a leak.)
Pressure need not be measured extremely accurately in these applications,

so standard commercial sensors may be used without paying a premium for
premium accuracy.
Sensors must be purchased for the pressure range required by the applica-
tion. As a general rule, the pressure range should be about two times the
expected set point.
InstallationPressure Sensors
The sensor should be mounted in a location where it is accessible for main-
tenance and that is not subject to physical harm, such as from vibration or
water damage. Often this means that the sensor is located in a temperature con-
trol panel and is piped to the remote piping taps rather than located remotely
where it may not be readily accessible.
DP transmitters are often installed in systems with pressures much higher
than the DP being monitored. During installation, start-up, or shut-down, the
pressure differential may exceed the transmitter DP rating, resulting in severe
damage to the transmitter. A three-valve bypass assembly (Figure 7-5) should
be used to minimize this possibility. The valve located in parallel with the DP
transmitter is left opened until the sensor is ready for use, ensuring the DP
across it is minimal. Test ports should be provided to allow a handheld gage to
be used for field calibration. A five-valve assembly is similar but also includes
two bleed valves for air removal; these additional ports can be used in lieu of
test ports for field calibration.
CalibrationPressure Sensors
Generally, factory calibration of pressure sensors is adequate for most pres-
sure measurement applications. It is recommended that each pressure-sensing
device be field tested during start-up and balancing to confirm its accuracy at
both zero pressure and at least one typical pressure condition.
Figure 7-5 Three-valve DP transmitter assembly with test ports.

Flow Switches and Indicators

Types of SensorsFlow Switches and Indicators
Historically, the most common flow indicating switch was a paddle switch.
The switch contact is closed when fluid flow deflects the paddle. Another com-
mon sensor is the DP switch, which is activated when DP, such as across the
evaporator, condenser, or pump, exceeds a set point.
A relatively new flow indicator uses thermal or calorimetric principles. The
sensor tip is heated to a few degrees above the liquid temperature. As the liquid
flows across the tip, it is cooled by the liquid proportional to liquid velocity.
The indicated liquid velocity is then compared to the set point programmed
into the device and the contact is closed when the set point is exceeded.
Issues and RecommendationsFlow Switches and Indicators

Pump status in the past was commonly determined using DP switches
mounted across the pump. But the switch would provide a false flow indication
if a valve in the system shut off flowhigh DP would still exist across the
switch. They are also expensive to install. Less expensive and more reliable
status indicators for pumps are current switches for fixed-speed pumps and
VFD status for variable-speed pumps, discussed in the section Issues and Rec-
ommendationsElectric Current Sensors.
Chiller manufacturers often require flow indicators across evaporators and,
to a lesser extent, condensers. Evaporator flow switches used to be essential
because a sudden loss in flow could cause significant damage, such as freezing
of the evaporator if the chiller was not shut off immediately upon loss of flow.
Some manufacturers no longer require evaporator flow switches, relying
instead on robust chiller controls to sense flow loss by the sudden drop in sup-
ply water temperature. Even more manufacturers have eliminated the require-
ment for condenser water flow indication, instead relying on high head
pressure alarms to disable the chiller upon loss of flow. The one disadvantage
of this approach is that loss-of-flow indication alarms are typically automatic
reset alarms should flow resume; by contrast, high head pressure alarms some-
times require manual reset. This can often be avoided by the BAS simply not
enabling chillers when flow indication at pumps or isolation valve positions or
end switches indicate flow is not available.
Flow indication switches should be avoided unless required by the chiller
manufacturer because they can be a source of false trips, particularly for paddle-
type switches, which are sensitive to physical damage, dirt, and corrosion. DP
switches are more reliable, but they can give false negative flow indication with
variable-flow systems, particularly on the condenser water side on systems that
vary flow for head pressure control (discussed in more detail below in the section
Control Schematics). Calorimetric flow switches have no moving parts, are resis-
tant to fouling, and can be set to very low-flow set points and are thus recom-
mended, particularly on the condenser side.
Flow indicators are not required by some manufacturers, are factory

installed by others, and are required to be field installed by a few. The latter can
be a hidden cost to the controls and piping contractors if the specified chiller
had factory-installed flow indicators or none at all. To level the playing field in
chiller specifications, flow indicators should be required to be factory installed
or the chiller manufacturer must carry the cost of their installation.
InstallationFlow Switches and Indicators

Flow switches, depending on the type, should be located well away from
elbows or other sources of turbulence that may cause them to flutter or be
damaged.
CalibrationFlow Switches and Indicators

Flow indicators do not need to be calibrated per se, but they do have a set
point that must be adjusted. On variable-flow systems, the set point must be
adjusted so that flow is indicated even at the minimum flow rate.
Electric Current Sensors

Types of SensorsElectric Current Sensors
The universal means of sensing AC current is the current transformer (CT).
As current passes through this device, a small voltage is generated proportional
to the current that is being measured. There are both digital and analog versions
of electric current sensors. The digital sensor (usually called a current switch)
provides a binary signal (contact closure) as long as the current is above a fixed
or adjustable set point. The analog sensor (usually called a current sensor or cur-
rent transducer) provides an analog signal (usually 05 Vdc or 420 mA) that
can be scaled to read the current draw. Current sensing alone is not very accurate
for power measurement because it does not include the impact of simultaneous
voltage and phase offsets. For power measurements, a true power meter should
be used as discussed in the Electric Power Meters section.
Issues and RecommendationsElectric Current Sensors

Adjustable set point current switches are very commonly used for on/off
status of equipment. They are almost always better status indicators than DP
sensors and flow switches for several reasons:
Current switches are less expensive due to substantially lower costs for
installation and wiring. DP and flow switches require installation into the
piping and are usually a long way from the DDC panel. Current switches
are more easily installed, particularly those with split core CTs, and are
mounted in the starter panel, which is usually close to the DDC panel and
also must be wired to anyway for on/off control of the motor starter. (For
this reason, current switches are available packaged into the same enclosure
with a pilot relay for on/off control of the starter.)
A DP switch for a pump may indicate pressure when in fact there is no

flow. This can happen when a valve is closed but the pump is still on. The
current switch set point can be set to a current higher than the power draw
when the pump is deadheaded (no flow).
DP switches cannot be used for cooling tower fan status because they do
not generate enough DP to make a DP switch work reliably.
The current switch set point can be set to a current higher than the motors
no-load current so that it still can be a reliable indicator of a pump coupling
or cooling tower fan belt failure.
Current switches are solid-state devices with no moving parts or dia-
phragms, and are therefore much more reliable than DP switches or flow
switches. They also require no maintenance and last longer.
Because of these advantages, current switches are recommended for constant-

speed motor status.
For variable-speed motors, current switches are usually not the best choice.
For some applications such as cooling tower fans, current switches can indicate
false failures due to very low current draw at low speed. However, a less expen-
sive and more reliable status indicator for variable-speed motors is simply to
use the status point that comes standard with the VFD. Like a current switch, it
can be programmed to indicate failure at no-load power (broken coupling or
belt) or deadhead power (closed valve).
InstallationElectric Current Sensors

A current switch used for status of constant-speed motors should be located
in the motor starter, mounted so that it is accessible and does not block access
to other devices. In three-phase applications, a CT is mounted only on one leg
of the power. The motor starter provides single-phase protection and will auto-
matically shut down the motor if one phase is lost.
CalibrationElectric Current Sensors

Factory calibration of current switches is adequate. However, as noted
above, current switch set points must be field adjusted to indicate failure at no-
load and deadhead currents.
Electric Power Meters

Types of SensorsElectric Power Meters
Two major types of meters are often used for power monitoring. The kW
demand sensor provides an analog output (usually 0-5 Vdc or 4-20 ma) that
indicates the instantaneous rate of electricity use. The kWh consumption sen-
sor provides a pulse signal that indicates the number of kilowatt hours of elec-
tricity that have flowed since the last pulse. Both types of sensors have the
same components. For three-phase applications, these sensors include CTs and
voltage measurement taps for each leg.
Modern power meters often have an option to transmit monitored informa-
tion over a network such as BACnet/IP. Networked meters provide both kW
and kWh information and commonly also provide power factor, kVAR, VA,
and harmonic information. These data may be very useful in many facilities,
especially if power quality is a concern.
Issues and RecommendationsElectric Power Meters

Most kW and kWh meters provide better than 2% accuracy, which is suit-
able for verifying the plants energy use. However, in applications that involve
monitoring the power of motors that are operated by VFDs or other wave dis-
torting equipment, accuracy may be reduced unless the power sensor provides
true root mean square (RMS) power sensing. Although there is no industry
standard for true RMS sensing, there is agreement in the community that a
minimum sampling or response rate of 3 kHz is required to get accurate mea-
surement of nonlinear loads like VFDs.
In most CHW plants where power monitoring is for performance monitor-
ing (as opposed to monitoring for revenue purposes), there is no need for field-
installed power meters because almost every motor has a VFD, and VFDs
include power monitoring inherently. The BAS simply needs to connect to the
VFD (or chiller controller) network interface or hardwire to the power output
point. For fixed-speed condenser water pumps, power meters are usually not
needed because power draw is generally fairly constant, so power can be mea-
sured once with a handheld device during commissioning and calculated based
on this measurement and pump status.
While energy monitoring using VFD onboard meters and motor status for
fixed-speed pumps is generally adequate for performance monitoring, it is not
accurate enough for revenue metering. VFD power meters have on the order of
3% to 5% accuracy at higher currents and many are less accurate at low cur-
rents. They also measure power output to the motor, so they do not include the
inefficiency of the VFD; VFD efficiency is very high at full load but falls off at
low loads (see Chapter 3). If very accurate power monitoring is required, a true
RMS revenue-grade meter of the equipment or the power service to the whole
plant is required.
InstallationElectric Power Meters

Power meters are generally located in the motor starter panels or electrical
distribution panels. Care must also be taken to ensure that sufficient ventilation
is provided so that the manufacturers temperature limits for the equipment are
not exceeded.
CalibrationElectric Power Meters

Factory calibration of electric kW and kWh sensors is nearly always ade-
quate. However, reality checks should always be done during commissioning
to verify that power meter scaling factors have been correctly configured.
Gas Flow Sensors

Types of SensorsGas Flow Sensors
There are several types of gas flow sensors that can monitor the natural gas-
flow of absorption or gas engine-driven chillers. The diaphragm gas meter is
most widely used (this is the type that most utilities install as site meters).
Other gas flow sensors are rotary and turbine meters; these are generally used
when the maximum gas flow requirements exceed the capacity of a diaphragm
meter.
Table 7-3 compares various options for gas flow-sensing devices.
Issues and RecommendationsGas Flow Sensors

Volumetric (diaphragm and rotary) meters are generally recommended for
most chiller plant applications. An important factor in choosing a gas flow
meter is range of the flows that the meter must measure. Another consideration
is available gas pressure; some diaphragm meters have high-pressure drops, so
the available gas pressure must be high enough to accommodate the meter and
still provide enough pressure for the chiller.
The metering application should always be discussed with the utility sup-
plying the gas. It is sometimes necessary to monitor the pressure in order to
improve the measurement accuracy. Temperature compensation options should
always be selected.
Where the gas is serving only the chiller plant, the best option is to use the
utilitys gas meter with an auxiliary transmitter to provide data (usually a pulse
signal) to the BAS. This is usually the most economical and most accurate
choice. The auxiliary transmitter is often an option, so coordination with the
utility is recommended.
Table 7-3 Comparison of Gas Flow-Sensing Devices

Range of Flow, Rangeability
Type of Gas Meter First Cost
Standard ft3/h (Turndown Ratio)
Diaphragm Up to 5000 Low 100:1
Rotary 100 to 50,000 High 40:1
Turbine 1500 to 200,000 Medium 15:1
InstallationGas Flow Sensors

Because diaphragm and rotary gas flowmeters measure volume, not veloc-
ity, their placement is far less critical than for turbine meters. For turbine
meters, the manufacturers requirements for placement and installation must be
carefully observed.
CalibrationGas Flow Sensors

Typically diaphragm and rotary gas meters are fully factory tested and cali-
brated and require no further calibration.
Control Valves
Valve Types
Ball Valves
Modern ball valves designed for control applications are inexpensive,
effective, and reliable in smaller chiller plant piping. Ball valves are now avail-
able in up to 6 in. pipe sizes. Ball valves are well suited for isolation valves
because they can be ported for full pipe size (i.e., the opening in the ball valve
is the same as the inside diameter of the pipe, reducing pressure drop). Ball
valves are also well suited for modulating control because they act with an
equal percentage characteristic (see the Valve Selection Criteria section that
follows) when fully ported or in special porting configurations. They usually
have lower first costs than the globe valves that have been traditionally used for
modulating duty. However, ball valves must be specifically designed for con-
trol applications; standard ball valve designs are not adequate for the continu-
ous movement required for modulating control duty and usually suffer seal
failures in a short period of time.
Butterfly Valves
Butterfly valves are the most popular large-diameter control valves in
chiller plants. Like ball valves, butterfly valves make excellent isolation valves
because they offer nearly full pipe bore when open and thus have low pressure
drop. Butterfly valves also have valve characteristics similar to equal percent-
age valves when used in modulating valve applications. However, they have
very low pressure drop and thus have low valve authority, usually making them
inappropriate for use in two-way modulating duty at cooling coils.
Globe Valves
Globe valves have lost almost all their market share to ball valves for small
modulating duty control valves, but for many years they were still the most
common control valves for large (>3 in.) modulating control valves. They have
relatively high-pressure drops and provide good authority for improved con-
trollability. They are thus the valve of choice (along with ball valves for smaller
coils) in two-way valve, variable-flow systems. However, now that ball valves
are available up to 6 in. in size, globe valves will further fall from favor. Where
both ball and globe valves are available, ball valves are preferred because they
are less expensive and more reliable and have higher shutoff pressures while
still providing equal percentage characteristics.
Pressure-Independent Valves
Generally, pressure-independent valves are not required in typical chiller
plant configurations but may be of use in solving special problems or system
features. See Chapter 4 for a discussion of the use of pressure-independent
valves.
Valve Selection Criteria

Valve Sizing and Flow Coefficient
Valve sizing in two-position (on/off) applications is straightforward: the
valve is simply the same size as the piping it is installed in. However, valve siz-
ing in modulating applications is more difficult and a fairly controversial sub-
ject. The valve size is based on its full open pressure drop, which in turn
determines the valves authority and the ability of the control system to func-
tion as desired and expected. It is probably intuitively clear that an oversized
valve will not be able to control flow well. As an extreme example, imagine
trying to pour a single glass of water using a giant sluice gate at the Hoover
Dam. However, undersizing a valve increases the system pressure drop, which
leads to higher pump costs and higher energy costs. These two considerations
must be balanced when making valve selections.
The size of a valve is determined by its pressure drop when it is at full open.
The question then is: what pressure drop should be used? Unfortunately, there is
no right answer to this question and there are differing opinions and rules of
thumb expressed by controls experts and manufacturers (discussion of which is
beyond the scope of this SDL). While there is disagreement about the exact value
of the desired pressure drop among these authorities, there is general agreement
that the control valve pressure drop, whatever it is, must be a substantial fraction
of the overall system pressure drop in order for stable control to be possible.
With the advent of more sophisticated control algorithms such as proportional-
integral-differential (PID) and fuzzy logic, some designers have questioned the
need for high valve pressure drops. However, while a well-tuned controller can cer-
tainly compensate for some valve oversizing, there is clearly a point where no con-
trol algorithm will help. For instance, getting a single glass of water out of a sluice
gate will be impossible no matter how clever the control algorithm may be. Over-
sizing will also result in the valve operating near close-off most of the time. This
can increase noise from flow turbulence and may accelerate wear on the valve
seats. Therefore, relaxing old rules of thumb on valve selection is not recom-
mended.
One old rule of thumb that has been used successfully in valve sizing at
coils is to select valve full open pressure drop between a minimum of half the
coil pressure drop and a maximum of 5 psi. Two-way valves should be selected
to result in a pressure drop near the maximum while three-way valves should
be selected for pressure drops near the minimum. See also ASHRAE Hand-
bookHVAC Systems and Equipment (2016b), Chapter 47, Valves; Funda-
mentals of HVAC Control Systems; and valve manufacturers selection guides
for valve sizing guidance.
Once the pressure drop is determined, the valve can be selected using a rat-
ing called the valve flow coefficient, Cv. The valve flow coefficient is defined as
the number of gallons per minute of fluid that will flow through the valve at a
pressure drop of one psi with the valve in its wide-open position, expressed
mathematically as
s
C v = Q ------- (7-1)
P
where
Q = flow rate in gpm
s = specific gravity of the fluid (the ratio of the density of fluid to that
of pure water at 60F)
P = pressure drop in psi
Specific gravity for water below about 200F is nearly equal to 1.0, so this
variable need not be considered for most HVAC applications other than those
using brines and other freeze-protection solutions. Valve coefficients, which
are a function primarily of valve size but also of the design of the valve body
and plug, can be found in manufacturers catalogs.
Valve Characteristics
For a more complete discussion of valve characteristics, refer to the
ASHRAE HandbookHVAC Systems and Equipment (2016b), Chapter 47,
Valves; and Fundamentals of HVAC Control Systems. As a summary of the
typical recommendations, see the following:
Select equal percentage characteristics for
Any two-way valve

Hot-water three-way valves
Select linear characteristics for
CHW and condenser water three-way valves
Not all manufacturers offer choices of valve characteristics, so it may not

be possible to always follow these recommendations.
Valve Shut Off

Modern valve seat materials provide zero leakage for many valves and
pressure conditions. However, substantial variances in close-off rating exist
between valves. For both ball and butterfly valves, the fluid pressure does
not affect the closing force required, but fluid pressure is a factor with
globe valves. With globe valves, it is important to carefully consider the
valve operating conditions to ensure that the valve has adequate close-off
capability. Many valves will have two close-off ratings, one for two-position
duty and another for modulating duty that is sometimes called the dynamic
close-off rating. The dynamic rating, which is always lower than the two-
position rating, is the maximum DP allowed for smooth modulation of the
valve, particularly near shut off. Above this DP, the design turndown ratio
will not be achieved. This is the rating that should be used when selecting a
valve for modulating applications. In two-way valve systems, a common
practice is to require that valves be capable of modulating and/or shutting
off against the pump shutoff head plus a safety factor (typically 25% to
50%). This is conservative for systems with variable-speed driven pumps
but still advisable because the pumps may be operated at fixed speed in
case of VFD failure.
Valve Actuators
Control valves and actuators for chiller plants should be purchased as a sin-
gle unit that is designed to meet the specified requirements. Variations in
breakaway torque as well as closing force variations due to fluid pressure may
affect the size of the actuator required. Purchasing the valve and actuator as an
assembly ensures that the performance of the assembly will meet specifica-
tions.
Knowing the valve position of modulating valves is required if advanced
reset logic is desired for CHW supply temperature set point and DP set point.
Analog actuators, those controlled by analog outputs from the DDC system
using 010 Vdc or 420 mA signals, are generally preferred because valve
position is equal to the signal sent to the valve, other than the time delay for
the motor as it adjusts the position to match the signal. Floating point actua-
tors, which are controlled by two digital outputs (one to open the valve and
one to close it), are less expensive, but actual valve position is not known
unless feedback from the actuator is available and wired to a DDC analog
input. With this feedback signal added, the cost is generally the same as for
analog actuators. Hence, it is recommended that modulating valves have ana-
log actuators.
The actual position of two-position actuators is also desired to ensure fail-
safe operation. These actuators can be specified to include end switches that
indicate full-open or full-closed position wired to DDC system digital input
points.
Controllers
Minimum DDC Controller Requirements
Programmability
Designers should ensure that any DDC system that is to be used for chiller
plant control have a powerful and flexible programming language. This lan-
guage should permit a nearly unrestrained capacity to incorporate logic, math-
ematics, timing, and other functions. This is important not only for the initial
development of the plants operations but also for future improvements to the
plants performance.
Variables
Chiller plants must be able to operate automatically under various operat-
ing conditions, including those caused by equipment failure and manual opera-
tor override. It is essential that the DDC system have the capacity for a large
number of variables (sometimes called pseudo, virtual, or software points) so
that features such as lead/lag sequences, automatic failure remedies, and opera-
tor disabling of equipment for maintenance can be implemented simply and
effectively.
Flexible Input/Output (I/O) Point Capacity

The DDC system must interface to I/O devices that use industry-standard
interfaces. Also, each chiller plant may have a unique mix of digital and analog
inputs and digital and analog outputs. Therefore, DDC controllers that have
universal I/O points (those that can be configured as analog or digital) and have
the capability of expanding point count with added I/O modules may provide
lower cost and greater flexibility for future changes than those with dedicated
I/O hardware.
Analog-to-Digital (A/D) and Digital-to-Analog (D/A) Resolution

The A/D conversion must provide adequate resolution to read all analog
inputs accurately to the number of significant digits desired. A 12-bit A/D res-
olution for analog inputs (4096 segments for the device span) is recommended
for analog inputs. Similarly, D/A resolution must be high enough on analog
outputs to provide good turndown control of valves and other devices; 8-bit or
10-bit D/A resolution for analog outputs is usually acceptable.
Automatic Network
Each DDC controller used in a chiller plant must have the capacity to auto-
matically and seamlessly share all point and variable information with other
controllers in the plant. It must also be able to preserve the required analog
value precision for those points whose value must be transmitted across the
network. Also, the network characteristics must require that
the maximum change of value (COV) (where employed) for all points be
the same as the specified precision or accuracy of those points, and
the maximum scan time (where employed) be less than 30 seconds (this
means that all controllers will use point data that is within at least 30 sec-
onds of the current value).
Trend Logs
For commissioning and long-term performance monitoring, controllers
must have the capability to record historic data on the status of control points,
to analyze this data, and produce trend logs that show the behavior of the con-
trol point relative to other variables. Plant operators should have the capability
to identify the control points for which trend logs are generated, to set the time
interval for taking data, and otherwise configure trend logs. It is very desirable
to be able to trend all hardware points and key software points, such as set
points and loop output points, in the system at short intervals (e.g., 1 minute)
during commissioning and at longer intervals (e.g., 5 minutes) for long-term
performance monitoring and diagnostics. The network must be designed to be
able to pass this data robustly from controllers back up to the control system
server where historical trend data may be stored for later analysis. Data should
be stored in a format, such as an SQL database, that can allow for post-analysis
by external performance analysis software. Proprietary data storage formats
should be avoided because they can limit how the trend data may be used.
Trending at short intervals can result in many megabytes of data, so a very
large disk capacity at the control system server or operator workstation is rec-
ommended.
Network Interfaces
Network Connections to Equipment Instrumentation
Some equipment, such as chillers and VFDs, have built-in controls. These
controls should include an interface, such as a BACnet gateway, that can be
connected to the DDC system to allow it to access the built-in control and mon-
itoring points and avoid the cost of installing redundant control points as part
of the BAS. For instance, chillers have controls that include CHW and con-
denser water temperature sensors. If data from these sensors is accessible to the
BAS system, the installation of additional sensors can be avoided.
In general it is good practice to limit these network connections to monitor-
ing and not control of critical components, as network connections often drop
out or experience momentary loss of connection. This is particularly important
for plants that serve mission-critical loads like manufacturing, hospitals, and
data centers. Also, transferring control loop outputs (such as pump speed)
across networks can result in unstable operation because of slow or inconsis-
tent network speed. In general, controlled points (e.g., DP), controlled devices
(e.g., VFD speed), and the associated PID loop should be connected to/reside
in the same controller. Specifications should be specific as to which points
should be hardwired and which can be transferred via the network.
With all network connections it is important to make sure to coordinate the
work of the equipment manufacturer and the BAS contractor in the specifica-
tions. Critical details include the network communication language (e.g., BAC-
net), physical link (e.g., RS 485), and points to be mapped from the network
device to the BAS. Be aware that many gateways allow only a limited number
of the available points to be mapped to the external BAS.
Chiller Interface
The start/stop point should be hardwired for all plants.
For critical plants such as those serving data centers, status and alarm
points should also be hardwired. Status can often be deduced from other vari-
ables, but that can lead to time delays in starting backup equipment, which can
result in critical loss of plant capacity or supply temperature control.
Even in systems where CHW supply temperature set point is actively reset, it
is usually not a critical point, so network speed and reliability concerns do not
mandate that the point be hardwired. The exception is chiller plants serving a sin-
gle coil where supply air temperature is controlled by CHW temperature directly.
The following are the minimum chiller monitoring points that should be
accessible from the chiller controller to the BAS via the network:
Supply (leaving) CHW temperature
Return CHW temperature
Supply (entering) condenser water temperature
Leaving condenser water temperature
Evaporator refrigerant pressure
Evaporator refrigerant temperature
Condenser refrigerant pressure
Condenser refrigerant temperature
Compressor discharge refrigerant temperature
Oil temperature
Oil pressure
Chiller electrical demand (power)
Chiller electrical current
CHW flow status
Condenser water flow status
Chiller operating status
Chiller alarm status
Inlet vane (centrifugal) or slide valve (screw) position
Compressor speed (if variable speed)
The following are the minimum chiller control points that should be con-
trollable from the BAS to the chiller controller via the network (if not hard-
wired):
Chiller start/stop
Chiller demand limit set point
CHW supply temperature set point
VFD Interface
The following points should be hardwired for all plants:
Start/stop
Speed
As indicated in the section Network Connections to Equipment Instrumen-

tation, time delays in transmitting the speed signals over the network can cause
control loops to be unstable or at least difficult to tune. Hence, speed should
always be a hardwired point.
For critical plants, such as those serving data centers, VFD status should
also be hardwired to avoid time delays in starting backup equipment.
The following are the minimum VFD monitoring points that should be
accessible directly from each VFD to the BAS via the network:
Status
Fault or alarm status
Actual speed
Power
Performance Monitoring
Integrating Chiller-Plant Efficiency Monitoring with Control
The Benefit of Performance Monitoring
In many climates, chiller plants are responsible for a major portion of a
facilitys energy use. Performance monitoring can help identify energy effi-
ciency opportunities. Many chiller plants are not fully automated, and nearly
all plants require ongoing maintenance to achieve top operating efficiencies.
Integrating chiller plant monitoring with the control system helps the plant
operating staff determine the most efficient equipment configuration and set-
tings for various load conditions. It also helps the staff schedule maintenance
activities at proper intervals so that maintenance is frequent enough to ensure
the highest levels of efficiency but not so frequent that it incurs unnecessary
expense.
Monitoring Considerations
It does not have to be expensive to integrate energy efficiency monitoring
with chiller plant control. Most DDC systems capable of operating chiller
plants effectively are well suited to provide monitoring capabilities. Because
chiller plant efficiency is calculated by comparing the CHW energy output to
the energy (electric, gas, or other) required to produce the chilled water, effi-
ciency monitoring requires only the following three items:
CHW output: Most plants will have the instrumentation needed for
measuring plant capacity (CHW flow and supply and return tempera-
tures) without additional cost because these sensors are needed for nor-
mal control.
Energy input: Again, most plants will already have the necessary power
metering because power meters are inherent in VFDs and constant-speed
condenser water pump energy can be estimated from on/off status, as dis-
cussed under the Electric Power Meters section. Fixed-speed chillers gen-
erally do not include power meters, so one must be added for each chiller
or to the plant as a whole in this case.
DDC math and trend capabilities: The DDC system must have math
functions so that the instrumentation readings can be easily scaled, con-
verted, calculated, displayed, and stored in trend logs for future refer-
ence. Again, this is inherent in almost all DDC systems at no added
cost.
In most cases, the only cost premium for performance monitoring is the
programming required to calculate performance metrics (such as kW/ton) over
various time intervals and to configure graphical displays and trends.
Data Displays
For the performance data to be useful to plant operators, the data must be
collected over various time intervals so the current and prior performance can
be compared. For instance, a graphical display might show monthly and
year-to-date data for plant output, energy, and average/minimum/maximum
efficiency (kW/ton) compared to the same data for prior years. Then, by
inspection, the operator can see trends, such as plant efficiency degrading
over time.
For systems with interfaces to the chiller controllers, this performance
graphic should also include trends for condenser and evaporator refrigerant-to-
water approach temperature differences, which are an indicator of tube fouling.
Alarms should be generated when approach temperatures stray a few degrees
from design values.
Control Schematics
Control Schematic for a Typical Plant
Control diagrams and schematics are represented in various ways with var-
ious levels of detail. Their primary purposes are to do the following:
Indicate the required control system sensors and their locations in the system
Show the system schematically as a whole so that the relationship of the
various components and equipment is clear
The latter goal is best achieved by limiting clutter in the schematic by

showing only control devices; other system components that are unrelated to
control, such as isolation valves, check valves, makeup water connections,
expansion tanks, air separators, and pressure relief valves, should be shown on
separate piping diagrams and details. Instrumentation such as thermometers,
pressure gages, non-control water meters (such as at tower water makeup con-
nections), test plugs, etc., should also be shown on piping diagrams (not con-
trol diagrams) because they are not dynamic control devices and they are also
not provided by the BAS contractor.
Figure 7-6 is a typical schematic of a primary-only CHW plant serving an
office building. Some of the key elements of the design are as follows:
Condenser water pumps are constant speed. As discussed in more detail
below in the Condenser Water Pumps section, use of VFDs on condenser
water pumps appears to be only marginally cost-effective if pump speed is
optimally controlled. Non-optimal control logic can easily increase plant
energy usage. So, VFDs on condenser water pumps are recommended only
where sequences will be optimized through rigorous analysis or via real-
time optimization add-on modules.
Because CW pumps are constant speed, kilowatt meters are shown on them
for overall plant performance monitoring. Meters are not required on other
equipment because they have VFDs, which inherently include power
meters. Pump power could also be estimated based on pump status and a
one-time power measurement (performed after plant commissioning) when
each pump is on alone and when both are on together.
The VFDs to pumps and tower fans have hardwired start/stop and speed
points with a network connection to the VFD interface for all other points.
As noted previously, if the plant were serving critical systems such as data
center air-handling systems, status points would also have been hardwired
to allow faster and more reliable response to failed equipment.
Towers do not include any isolation valves to shut off flow to allow one
tower to operate alone. As noted in Chapter 2, towers can generally be
selected with nozzles and dams to allow half flow while still providing full
coverage of fill, and it is always most efficient to run as many cells as pos-
sible.
Figure 7-6
Typical
CHW plant
control
schematic.
Makeup water to the condenser water system is shown to be controlled by

the BAS using a level sensor mounted on a standpipe connected to the
tower equalizer. This allows the tower level to include high- and low-level
alarms and other benefits. See discussion in Chapter 3.
The condenser water isolation valves at each chiller are wired to the head
pressure control output of the chiller controller, an output that is standard
on most chiller controllers. (The term head pressure, a historical term for
condenser pressure, is used here because it is common. In actuality, mini-
mum lift, the difference between condenser and evaporator pressure, is
what is being controlled by the chiller controller.) This allows the valve to
do double duty: it can shut off flow when the chiller is off and also modu-
late flow to maintain head pressure when the system is first started and
tower basin water may be cold. Head pressure control is usually required
on screw chillers unless they have some separate means for oil control.
Most centrifugal chillers can operate at low head long enough for the tower
water to warm up so head pressure control is not needed unless the tower
basin contains a very large volume of water or the chiller must operate in
cold weather. Note that controlling head pressure in this manner obviates
the need for a tower bypass control valve except in locations where the
chiller is required to operate in freezing weather. Because valve position is
not DDC controlled, a valve position feedback signal is wired to the DDC
system so that position is known. Another option is to wire the chiller head
pressure demand signal as an analog input to the DDC system, then control
the condenser water valve as an analog output from the DDC system. This
allows the DDC system to filter the logic of the chiller controller to ensure
it is performing the desired function (including shutting off flow when the
chiller is off). A similar approach can be used for chillers that do not have
direct head pressure control outputs. Either actual refrigerant lift (differ-
ence between condenser and evaporator refrigerant pressure) can be passed
to the DDC system from the chiller controller via the network interface or a
lift indicator, such as the difference between leaving condenser water tem-
perature and leaving CHW temperature, can be used as the controlled vari-
able.
Flow switches are shown on the CHW side of the system but not on the
condenser water side. Flow switches on the condenser water side are not
needed and can cause false trips when head pressure control is active, as
explained in Chapter 3.
Chillers are controlled by only a single hardwired point for on/off. All
other points are mapped through the chiller controller network interface.
That includes CHW and condenser water temperatures; separate field-
mounted temperature sensors hardwired to the DDC system are not
required.
CHW isolation valves are modulating rather than two position. This
ensures the chillers can be sequenced smoothly without rapid flow changes
to prevent chiller trips when staging from one chiller to two. See the section
Control Sequences for further explanation. As chiller controllers become

more robust, modulating valves may no longer be needed (two-position
valves with electric actuators are already fairly slow moving), but the cost
increase to use modulating rather than two-position valves is relatively
small and can be justified by the flexibility they modulating valves add to
control the speed at which valves open and close.
CHW flow and supply and return water temperatures are shown to be part
of a Btu meter package to improve the accuracy of plant load calculation.
The Btu meter is shown connected to the BAS via a network connection
through which flow, supply and return temperatures, and calculated Btu/h
data are available. The flowmeter is also shown hardwired to a BAS ana-
log input. This is required because the flowmeter is also being used by
the BAS for minimum chiller flow control and controlled variables
should be connected to the same controller as the controlled device (the
minimum flow bypass valve) to avoid instability caused by slow and
inconsistent network speeds. Minimum chiller flow could also be con-
trolled by measuring DP across each chiller and correlating DP to flow,
but a flowmeter is more accurate. The flowmeter also can be used directly
for CHW pump staging and (with Btu calculation) for chiller staging and
cooling tower set point reset. Therefore, a flowmeter is highly recom-
mended and assumed to be installed in the control sequences discussed in
the next section.
The minimum flow bypass valve used to ensure minimum chiller flow is
maintained is located so that the flowmeter measures the flow through the
chillers so it can be used for valve control. A separate CHW return tem-
perature sensor is added upstream of the valve so that the temperature of
the water from the coils can be monitored.
A DP sensor is shown remote from the plant. As noted in Chapter 3, the
further out into the system the sensor is located, the lower the DP set point
can be, which results in the lowest pump energy. In many plants where
chilled water is piped to several different branches, multiple DP sensors are
required, one for each branch. Each could have its own set point as deter-
mined by the balancing contractor to ensure adequate pressure is available
downstream in each branch. Separate control loops would be executed for
each sensor and the largest loop output would be used to control pump
speed. These sensors should all be hardwired back to the plant controller
rather than passed through the network to avoid control instability caused
by slow and inconsistent network speeds.
Both CW supply and return temperature sensors are shown hardwired to
the DDC system. Both temperatures are available individually for each
chiller through the network interface. Whichever sensor is used for control
(we recommend the return temperature in sequences below in the section
Tower Fan Speed Control) must be hardwired to the DDC system because
controlled points and controlled devices (tower fan speed in this case)
should not rely on the network for control, as explained above in the sec-
tion Network Connections to Equipment Instrumentation. The other sensor

is optional and could be eliminated because it is for monitoring only and
the temperatures are available through the chiller interface.
No CW flowmeter is shown. As discussed below in the section Variable-
Speed Condenser Water Pumps, CW flow is desirable as a controlled vari-
able for optimum control when VFDs are used on condenser water pumps.
CW flow is not needed for constant-speed CW pump control. However, a
side benefit of having a CW flowmeter is that it can also be used as a real-
time instrumentation check: condenser heat rejection can be compared to the
sum of compressor power (adjusted for motor losses if open drive) plus evap-
orator load with any discrepancy indicating that a sensor is faulty or out of
calibration. Where CW flowmeters are used, full-bore magnetic or ultrasonic
types are recommended to avoid measurement errors and added maintenance
due to fouling and corrosion.
The plant is instrumented sufficiently for total plant efficiency measure-
ment, typically characterized as kW/ton . The thermal energy meter pro-
vides accurate plant load data. All variable-speed chillers (assumed in this
example, Figure 7-6) and many constant-speed chillers include power
meters. Pump and tower fan VFDs also include power meters. Chiller and
VFD power data can be transmitted through the network interface. (Note
that power measured by VFDs is output power, not input power, so VFD
inefficiency is not included. However, VFD efficiency is very high except
at low speeds where power is also very low, so the error should be small;
see Chapter 3.) If condenser water pumps are constant speed, either power
meters can be installed (as shown in this example) or pump power can be
calculated based on pump status as described above in the section Electric
Power Meters.
Control Sequences
Determining Optimal Control Sequences
CHW plants have many characteristics that make each plant unique so that
generalized sequences of control that maximize plant efficiency are not readily
determined. Equipment and system variables (see Chapter 5) that affect perfor-
mance include the following:
Chillers: Each chiller has unique characteristics that affect full-load and
part-load efficiency, such as compressor design, evaporator and condenser
heat transfer characteristics, unloading devices (such as VFDs, slide valves,
and inlet guide vanes), and internal control logic.
Cooling towers: Tower efficiency (gpm/hp) varies significantly by almost
an order of magnitude between a compact centrifugal fan tower to an over-
sized propeller fan tower. Towers can also be selected for a wide range of
approach temperatures.
CHW and condenser water pumps: Pumps and piping systems can be
selected for a broad range of Ts and may or may not include VFDs.
Pump efficiency also varies by pump type and size, and pump head varies
significantly depending on physical arrangement and pipe sizing stan-
dards.
CHW distribution systems: Distribution system arrangements, such as pri-
mary/secondary versus primary-only variable flow, significantly affect
plant control logic.
Weather: Changes in outdoor air conditions affect loads and the ability of
cooling towers to reject energy.
Load profile: The size and consistency of loads affect optimum sequences.
For instance, control sequences that are optimum for an office building
served by air-handling systems with air-side economizers may not be opti-
mum for a data center served by systems without economizers.
It is clear that no single control sequence will maximize the plant efficiency
of all plants in all climates.
There are a number of articles (Hartman 2005; Hydeman and Zhou 2007)
on techniques to optimize control sequences for CHW plants. Almost all
require some level of computer modeling of the system and system compo-
nents and an associated amount of engineering time that most plant designers
do not have.
In developing this SDL, significant modeling was performed in an effort to
determine generalized control sequences that accounted for the variation in
plant design parameters summarized above. The technique used to determine
optimized performance is described in a June 2007 ASHRAE Journal Article
(Hydeman and Zhou 2007). In brief, the technique involves developing a cali-
brated simulation model of the plant and plant equipment that is run against an
annual hourly CHW load profile with coincident weather data while parametri-
cally modeling all of the potential modes of operation at each hour using multi-
ple nested iterative loops. See Figure 7-7. The operating mode requiring the
least amount of energy for each is determined. The minimum hourly energy
usage is summed for the yearthis is the TOPP. Because all modes of opera-
tion were simulated, the plant performance cannot be better than the TOPP
within the accuracy of the component models. The operating modes (e.g., num-
ber of chillers, condenser water flow and pump speed, tower fan speed and
related condenser water temperatures) that result in the TOPP are then studied
using scatter plots, frequency charts, etc., to see how they relate to independent
variables such as plant load, operating temperatures, and wet-bulb temperature
in order to find trends that can be used to develop simple sequences to control
the plant in real applications through the DDC system. Ideally, equipment
should be controlled as simply as possiblecomplex sequences are less likely
to be sustained because operators are more likely to disable them at the first
sign of perceived improper operation.
Figure 7-7
Theoretical optimum plant
performance (TOPP) model
flow chart.
Equipment models include the following:
Chillers
Regression-based electric chiller model used in EnergyPlus (Hydeman
et al. 2002)
Multipoint calibration using zero tolerance manufacturers data (see
Chapter 6 of this course book)
Towers
DOE-2.2 model
Calibrated using manufacturers data for real tower selections
Pumps
Multiple piping sections assuming near turbulent flow (P = C
gpm1.8)
Pump efficiency based on regression models of manufacturers data for
real pump selections
VFD and motor efficiency
Part-load curves from manufacturers data (see Chapter 3 of this course
book)
The sequences described in the sections below were developed from the
TOPP modeling for the all-variable-speed CHW plant shown in Figure 7-6
serving a typical office building for a wide range of plant design options for
tower range, approach, and efficiency; different chiller types and chiller effi-
ciencies; and varying climates (see Appendix A for details). Also included
are control sequences for constant-speed chillers and pumps developed from
past experience.
Cooling Towers
Tower Staging
As noted in Chapter 4, cooling towers are most efficient when the most
cells are operated within the flow limits of the towers. Plants with two, and
sometimes three, tower cells can be designed so that all cells are active under
all load conditions by selecting tower distribution pans and nozzles for the flow
of only one condenser water pump. Where cells must be staged using isolation
valves (discussed in Chapter 4), the maximum number of cells possible should
be enabled while still maintaining minimum flow through each active cell.
Tower Fan Speed Control

A common approach to controlling cooling towers is to reset condenser
water supply temperature based on outdoor air wet-bulb temperature. However,
our simulations seldom indicated a good fit; as shown in Figure 7-8, the cor-
relation of optimum condenser water supply temperature versus wet-bulb tem-
perature was fairly good in Miami but not in Oakland or most other climates.
Figure 7-8 TOPP optimum condenser water supply temperature versus wet-bulb tem-
perature.
For plants serving typical office buildings,1 good correlations were found
in all TOPP simulations between plant part-load ratio (PLR, actual plant load
divided by total plant design capacity) and the difference between the CWRT
(leaving the condenser) and the CHW supply temperature (CHWST, leaving
the evaporator). Examples are shown in Figure 7-9. The CWRT CHWST dif-
ference is a direct indicator of the refrigerant lift (the condenser and evaporator
leaving water temperatures are determined by the condenser and evaporator
temperatures), which drives chiller efficiency.
The data in Figure 7-9 can be fit to a straight line:
LIFT = CWRT CHWST = A PLR + B (7-2a)
where A and B are coefficients that vary with climate and plant design (see
Appendix A).
LIFT calculated from Equation 2a must be no larger than the maximum lift
at design conditions (design CWRT minus design CHWST from equipment
schedules) and must be no less than the minimum lift required at minimum load
(obtained from the chiller manufacturer). The latter varies among chiller types
and manufacturers. For frictionless (magnetic) bearing chillers, the minimum
lift is typically less than 5F because these chillers have no oil, eliminating one
of the primary reasons for maintaining minimum lift. Some manufacturers of
these chillers even claim to have 0F minimum lift requirement. The minimum
lift at minimum load for standard bearing variable-speed centrifugal chillers
(e.g., for the chiller modeled in Figure 7-9) is typically around 9F to 20F.
1. For plants with more consistent loads that do not vary with weather, such as those serving
data centers and those located in consistently humid climates, such as Miami, correlation of
load with CWRT/CHWST temperature difference is poor. For these plants, optimum CWST
versus wet-bulb temperature was found to have better correlation. However, for office build-
ings in general, the correlations in Figure 7-9 were more consistent.
Figure 7-9 TOPP (CWRT CHWST) versus plant load ratio.

Minimum lift is typically highest for screw chillers at 25F to 30F, but it may
be lower for variable-speed screw chillers, which generally have separate oil
separators and pumps to avoid having to rely on refrigerant lift for oil return.
The lower this minimum is, the lower the annual chiller plant energy will be,
particularly in mild climates.
Equation 2a can be solved for the optimum CWRT set point given the cur-
rent CHWST:
CWRT = CHWST + A PLR + B (7-2b)
Cooling tower fans are then modulated to maintain condenser water return
temperature at this set point.
Controlling tower fan speed based on return temperature rather than supply
temperature is nonconventional, but it makes sense because it is the temperature
leaving the condenser that determines chiller lift, not the entering (supply) water
temperature. Chiller efficiency is not sensitive to entering chilled- or condenser
water temperature. Controlling fans off of return temperature rather than supply
temperature is even more critical with variable-flow condenser water pumps
because CWRT rises as CW flow falls. See the Variable-Speed CW Pump section.
Controlling tower fans off of CWRT can be unstable if the towers and
chillers are greatly separated, such as towers located on the roof and chillers
located in the basement. This is due to time delays caused by the relatively
slow-moving water mass in the piping and is exacerbated by control network
delays if the sensors are not all wired to the same controller. Where towers and
chillers are separated, cascading control loops are recommended: the tower
fans would be controlled by the water temperature leaving the tower (CWST)
using the controller near the towers, but the set point for this loop would be
reset by a second control loop maintaining the water temperature leaving the
chillers at the set point from Equation 2a using the controller located near the
chillers. As with all cascading loops, the loop resetting the CWST set point
must be slower than the loop controlling tower fan speed for stable operation.
Chilled-Water Pumps
Primary CHW Pumps on Primary/Secondary Systems
Constant-speed primary pumps on primary/secondary systems generally
are staged with chillers: if a chiller is started, then a pump is started; if a chiller
is stopped, a pump is stopped.
Where primary pumps have VFDs, they must be controlled to maintain pri-
mary flow at least equal to the secondary flow (to avoid the death spiral dis-
cussed in Chapter 4) and above the chiller minimum flow rate. One of two
sequences is commonly used:
If the primary and secondary circuits both have flowmeters, primary pump
speed can be controlled to maintain primary flow ~10% larger than second-
ary flow but no lower than the sum of the minimum flow rates of all active
chillers. This logic requires reliable flowmeters, such as full-bore magnetic
flowmeters discussed above in the section Liquid Flow Sensors.
If there are no flowmeters, primary pump speed can be controlled to
maintain secondary CHW supply temperature equal to the primary CHW
supply temperature down to a minimum speed that provides minimum
chiller flow as determined during the test and balance phase. This is best
done using trim and respond logic (Taylor 2015): primary pump speed is
steadily dropped until secondary CHW temperature rises above primary
CHW temperature, which generates a request for higher speed. Exam-
ples of trim and respond logic are provided in the example sequences in
Appendix B.
Secondary CHW Pumps on Primary/Secondary Systems

The speed of secondary pumps in primary/secondary systems is typically
controlled to maintain DP measured far out in the system at set point. The DP
sensor should be as far out in the system as possible, as discussed in
Chapter 3.
The design DP set point for variable-speed pumps should be determined by

the system balancer using one of the following procedures depending on
whether or not the system is balanced (see Balancing Variable-Flow Systems in
Chapter 4):
For systems that are fully balanced, the set point is simply the DP reading
when all coils are balanced and operating at design flow rates. The most
remote coil should have its balance valve fully open if the system is prop-
erly balanced.
For systems that are not balanced, the set point is determined by first clos-
ing all control valves except those downstream of the DP sensor, then,
manually decreasing pump speed until flow through one of these down-
stream coils just reaches its design rate. The DP reading at this condition
becomes the set point.
The DP set point can be reset to minimize pump energy usage under low-
load conditions. This is done by monitoring valve position and resetting the set
point as required to maintain the most open valve near full wide open. See the
discussion in the Chilled-Water Temperature and Differential Pressure (DP)
Set Point Reset section.
For large buildings or campuses, DP sensors may be located far from the
plant. Pump control in this case can be unstable due to time delays between a
change in pump speed and the resulting DP change, and greatly exacerbated by
control network delays if the DP sensors are not wired to the controller con-
trolling pump speed. Where DP remote sensors are required, cascading control
loops are recommended: the pumps would be controlled by a DP sensor
located at the plant tied to the controller controlling pump speed, but the set
point for this loop would be reset by the output of a second control loop main-
taining the remote DP sensor at its set point with the loop residing in the con-
troller to which the DP sensor is connected. This logic allows the use of
multiple DP sensors; the highest output of each remote loop would be used to
control the pump. As with all cascading loops, the loops resetting the set point
must be slower than the loop controlling pump speed for stable operation.
Pump Staging
Figure 7-10 shows the optimum number of CHW pumps as a function of
CHW flow ratio (CHWFR, actual flow divided by design flow) and as a func-
tion of pump speed for a two chiller constant primary/variable secondary sys-
tem with two secondary pumps plant based on TOPP modeling. Unlike cooling
towers, the optimum sequence is not to run as many pumps as possible. This is
because the pumps all pump through the same circuit (other than the pipes into
and out of each pump between headers), so there are not cube law energy ben-
efits for each pump individually. Because of the minimum DP being main-
tained at coils (which causes the system curve to bend off of the ideal curve at
low flow, reducing pump efficiency) and because motor efficiency falls rapidly
at low loads, running excess pumps will increase energy use. The optimum
(a)
(b)
Figure 7-10 (a) TOPP CHW pump staging versus (b) CHWFR and pump speed.
staging in Figure 7-10 is about 47% flow as opposed to 50% flow because of
the piping dedicated to each pump as noted above; the pressure drop in these
piping sections is minimized when more pumps are operating.
As suggested by Figure 7-10, the optimum number of CHW pumps should
be staged as a function of CHW flow, not CHW speed as is common practice.
Specifically, stage up from one pump to two pumps above 47% CHWFR; stage
down to one pump below 47% CHWFR, where CHWFR = actual flow divided
by design flow. Provide a time delay (e.g., 10 minutes) between each stage to
prevent short cycling.
The above logic applies to a two-pump plant but it can be extended to an N-
pump plant:
Stage up from one pump to two pumps at (1/N 3%) CHWFR and vice versa
Stage up from two pumps to three pumps at 2 (1/N 3%) CHWFR and
vice versa
Stage up from three pumps to four pumps at 3 (1/N 3%) CHWFR and
vice versa
Etc.
All operating pumps run at the same speed. Time delays must be provided
between stages. (The time delay could also be provided by creating a deadband
between staging up and staging down set points, but using actual timers is more
direct and can be more energy efficient.)
Using flow to control pump staging, rather than speed, is also beneficial for
plants with three or more pumps because using pump speed alone can result in
running too few pumps, which in turn can result in pumps running off the ends
of their curves and cavitating.
For very large pumps ( 100 hp), it may be worth the effort to determine
the actual pump operating point (flow versus head) and optimize staging based
on pump efficiency determined by flow and pressure drop readings mapped to
pump curves duplicated mathematically in the DDC system (Rishel 2001).
This can allow pumps to operate closer to their design efficiency as the system
operating curve varies from the ideal parabolic curve due to modulating valves
and minimum DP set point. However, the small potential energy savings is not
worth the effort for most CHW plants.
Primary-Only CHW Pumps

Staging primary pumps in a primary-only variable-flow system is identical to
staging secondary pumps as described above in the Pump Staging section. The
pumps must respond to the flow and pressure requirements of the system, not to the
load. For headered variable-speed pumps (see Figure 4-17b), it is not necessary to
start a pump when a chiller starts. For instance, two pumps may be able to meet the
flow requirements for three chillers over a wide flow range. Conversely, if there is
significant T degradation, three pumps could operate to meet flow requirements
while running only two chillers to meet the load. This is one of the advantages of
this design, and it is therefore recommended for primary-only variable-flow sys-
tems over dedicated pumps piped directly to each chiller (see Figure 4-17a).
Condenser Water Pumps

Constant-Speed CW Pumps
Constant-speed condenser water pumps generally are staged with chillers:
if a chiller is started, then a pump is started; if a chiller is stopped, a pump is
stopped. In larger plants with three or more pumps, efficiency can be improved
by operating fewer CW pumps than chillers (a quasi-variable-flow system), but
actual sequences must be developed from energy models to avoid increasing
plant energy. See the Variable-Speed CW Pumps section that follows.
Variable-Speed CW Pumps
Optimum control of variable-speed CW pumps is challenging because flow
reduction reduces pump and tower fan energy but simultaneously causes an
increase in chiller energy due to higher lift. Because of these offsetting factors,
energy savings even with TOPP optimization are small, as shown in Figure 7-11.
(The x-axis of this figure is made up of codes for various simulations with differ-
ent climates, tower efficiencies, and chiller types.)
The savings are even smaller when real sequences are used based on the TOPP
models because we could not find any strong correlation between the TOPP opti-
mized flow rate and any independent variables from which a good real sequence
Figure 7-11 Energy use of plants with constant speed versus variable speed.
could be developed. Optimum condenser water pump speed and flow were plotted
against various parameters such as PLR, wet-bulb temperature, chiller percent
power, and lift with no consistent relationships. The best correlation was flow ver-
sus PLR. Figure 7-12 shows optimized CW flow versus plant load, both expressed
as a percentage of design flow and load. (Pump flow was varied in 10% increments
so the data points are not continuous.) However, the correlations were seldom
strong (R2 was typically less than 0.85 and some were as low as 0.5). The correla-
tions were significantly weaker for pump speed than for flow, so a condenser water
flowmeter should be added if one is not already part of the design.
The data in Figure 7-12 can be fit to a straight line:
CWFR = C PLR + D (7-3a)
where CWFR is condenser water flow ratio (percentage of design flow), and C
an D are coefficients that vary with climate and plant design (see Appendix A).
Pump flow set point, condenser water flow set point (CWFSP), can then be
calculated from the design condenser water flow rate, CWFd:
CWFSP = CWFd CWFR (7-3b)
This set point must be bounded by the minimum required CW flow rate
obtained from the chiller manufacturer. The minimum flow from most manu-
facturers correlates to the onset of laminar flow and will be on the order of 40%
to 70% of design flow depending on the design T, number of tubes, number of
passes, and tube design (e.g., smooth versus enhanced). Higher rates are
reputed to discourage fouling of condenser tubes, but, to our knowledge, no
studies have been done to support that notion (Li and Webb 2001). Pump speed
is then modulated to maintain measured flow at this set point.
Optimum staging for variable-speed CW pumps was found to correlate very
well to CW flow with 60% of the total design flow as the staging point (i.e., one
pump should operate when the CWFR is below 60% and two pumps should
operate when CWFR is above 60%, with a time delay to prevent short cycling).
Figure 7-12 TOPP optimum percent CW loop flow versus percent plant load.
When C and D coefficients determined for specific plants were fed back
into the energy model, actual performance ranged from 101% to 110% of the
TOPP. This performance gets worse when C and D are determined from the
regression equations based on plant design (see Appendix A) rather than from
actual plant performance modeling (e.g., Figure 7-12).
Figure 7-13 shows life-cycle costs (based on LCCA parameters
described in Chapter 5essentially a 14-year simple payback period) for an
Oakland, CA, office building with both constant-speed and variable-speed
CW pumps with pump speed both optimally controlled and controlled using
Equation 3b with coefficients C and D determined from curve-fits in
Figure 7-12. The VFDs are barely cost-effective even with perfect controls
and even assuming life-cycle cost parameters that equate to a 14-year simple
payback period. In other words, even with perfect controls, the VFDs will
barely pay for themselves in their typical 15-year service life. And with the
reduced performance of real control sequences, CW pump VFDs are not
cost-effective, as shown on the right side of Figure 7-13. Cost-effectiveness
was even worse in more humid climates like Miami and a bit better in dry
Figure 7-13 Life-cycle costs for an Oakland office building with constant-speed (CS) and
variable-speed (VS) CW pumps.
climates like Albuquerque. For plants seeing higher loads over more widely
varying weather conditions, such as 24/7 data centers, variable-speed con-
denser water pumps can be cost-effective but only if the controls are opti-
mized using plant simulation.
Of even greater concern, our studies found that variable-speed CW pumps
can increase the energy usage of the plant if not optimally controlled. For
example, Figure 7-14 shows energy usage for a plant serving an office building
in Denver, CO, and Figure 7-15 shows the same plant and building in Miami,
FL, each using three control strategies:
TOPP: This is the theoretical optimum plant performance of the plant with
variable-speed condenser water pumps determined using the technique
described in the section Determining Optimal Control Sequences). This is
the theoretical best performance possible.
Standard (STD): This is the performance of the plant with constant-speed
condenser water pumps and cooling tower fans controlled to reset supply
water temperature per AHRI Standard 550/590 condenser water relief
curves (2015). This is most indicative of conventional practice.
Oakland, CA (OAK): This is the performance of the plant with variable-
speed pumps controlled using control sequences that were optimized for the
same plant located in Oakland, CA (see Appendix A), instead of Denver, CO.
The figures show that energy usage is highest using control sequences that
provided near-ideal performance for the same plant in another climate zone,
significantly higher energy usage than the plant without VFDs on condenser
Figure 7-14 Denver CHW plant energy usage using three control strategies.
Figure 7-15 Miami CHW plant energy usage using three control strategies.
water pumps. This demonstrates how sensitive plant performance is to the

details of the control logic.
Based on this analysis, VFDs on CW pumps should not be used for most
commercial applications for the following reasons:
Energy use may be increased if not optimally controlled.
VFDs are unlikely to be cost-effective even if optimally controlled.
A condenser water flowmeter is required for optimized control.
Higher lift can cause chillers to operate in surge (see in the section Variable-
Speed Chiller Staging for more details).
Low flow rates may cause fouling of condenser tubes due to low tube
velocities.
Low flow rates may cause scaling in towers if rates are below minimum
flow or tower isolation valves plus low-flow nozzles must be provided.
For plants that operate 24/7, VFDs may be cost-effective but should only be
considered if the control logic is optimized using TOPP type simulations.
Chilled-Water Temperature and Differential Pressure (DP)

Set Point Reset
Chillers are more efficient at higher CHWST because this reduces lift.
Resetting the CHW temperature set point upward when loads are low is always
an effective energy-saving strategy for constant-flow systems. Reset may or
may not be an energy-saving strategy in variable-flow systems depending on
the plant design. High CHW temperature reduces coil performance, so coils in
two-way valve systems will demand more chilled water for the same load,
degrading T and increasing flow and pump energy requirements. Whether the
net energy savings (chiller energy decrease less pump energy increase) is posi-
tive and sufficient to offset the cost of implementing the reset strategy depends
on chiller performance characteristics and the nature of coil loads. This is dis-
cussed further below.
CHWST set point reset strategies include the following:
Resetting inversely proportional to outdoor air temperature

Resetting from return water temperature, either indirectly by maintaining a
constant return water temperature or resetting the set point proportional to
return water temperature
Resetting from CHW valve position. See a detailed discussion on how this
is implemented below
The last strategy (reset of valve demand) was once impractical with pneu-
matic controls and distributed controls in large campus buildings. However, it
is readily done in systems with DDC for all control valves. It is by far the most
reliable and efficient strategy in that it ensures that no coil is starved. The other
strategies are indirect and cannot assure all coils will be satisfied unless they
are very conservative (i.e., will yield little actual reset). Using valve position
also ensures that humidity control will be maintained. Contrary to conventional
wisdom, the impact of reset on the dehumidification capability of air handlers
is quite small and should not be a concern. Space humidity is a function of the
supply air humidity ratio, which in turn is a function of the coil leaving dry-
bulb temperature set point. Regardless of CHWST, the air leaving a wet cool-
ing coil is nearly saturated; lowering CHWST only slightly reduces the supply
air humidity ratio. So as long as the supply air temperature can be maintained
at the desired set point, as can be indicated by valve position, resetting
CHWST will not impact space humidity.
Valve position can also be used to reset the DP set point used to control
pump speed. In fact this is required by ANSI/ASHRAE/IES Standard 90.1
(2016a). (ANSI/ASHRAE/IES Standard 90.1-2016 allows either CHW tem-
perature set point reset, DP set point reset, or both.) The logic is similar to
CHWST set point reset: the DP set point is reset upwards until the valve con-
trolling the coil that requires the highest DP is wide open.
So, we have a dilemma: valve position can be used to reset either CHWST
set point or DP set point but not both independentlyit is not possible to know
if the valve is starved from lack of pressure or from lack of cold enough water.
We must decide which of the two set points to favor.
While resetting CHWST set point upward reduces chiller energy use, it
increases pump energy use in variable-flow systems. Higher CHW temperature
causes coils to require more chilled water for the same load, degrading CHW
T and increasing flow and pump energy requirements. Degrading T can also
affect optimum chiller staging; however, this is not generally an issue in pri-
mary-only plants with variable-speed chillers (see Chapter 4). Furthermore,
our simulations have shown that the positive impact of resetting CHW tem-
perature on chiller efficiency is much greater than the negative impact on pump
energy even when distribution losses are high for plants that have variable-
speed chillers. Figure 7-16 shows a DOE2.2 simulation of a primary-only plant
with variable-speed chillers and CHW pumps with high pump head (150 ft)
using three reset strategies based on valve position: reset of CHW temperature
alone, reset of DP set point alone, and a combination of the two that first resets
CHW temperature then resets DP set point. The simulations were done in sev-
eral climate zones (Houston and Oakland results are shown in the figure) and
in all cases, resetting CHW temperature was a more efficient strategy than
resetting DP set point. Sequencing the two (resetting CHW temperature first
then DP set point) was the best approach, although only slightly better than
resetting CHW temperature alone.
Figure 7-17 shows how this sequenced reset strategy can be implemented.
The x-axis is a software point called CHW Plant Reset that varies from 0% to
100% using trim and respond logic. The coil valve controllers generate
requests for colder CHW temperature or higher pump pressure when the
valve is full open. When valves are generating requests, CHW Plant Reset
increases; when they are not, CHW Plant Reset steadily decreases. When
CHW Plant Reset is 100%, the CHWST set point is at Tmin and the DP set
Figure 7-16 Plant energy with CHWST set point reset, CW DP set point reset, and a combi-
nation of the two.
Figure 7-17 CHWST set point reset sequenced with CW DP set point reset off of CHW valves.
Figure 7-18 CW DP set point reset sequenced with CHWST set point reset off of CHW valves.
point is at DPmax. Tmin is typically the design CHW temperature for plants with
variable-speed chillers but should be 1F to 2F lower for constant-speed
plants; this allows the operating chillers to fully max out their capacity before
staging on another chiller (see the Constant-Speed Chiller Staging section).
DPmax is the design DP set point determined as described above for variable-
speed CHW pumps. As the load backs off, the trim and respond logic reduces
the CHW Plant Reset point. As it does, CHW temperature is increased, first up
to a maximum Tmax (equal to the lowest air-handler supply air temperature set
point less 2F), then DP set point is reduced down to a minimum value DPmin
(such as 5 psi). In practice, this logic seldom results in much reset of the DP set
pointthe CHWST reset is aggressive enough to starve the coils firstso it is
important to locate the DP sensor(s) at the most remote coil(s) so that DPmax
can be as low as possible to minimize pump energy.
The opposite was found to be true for plants with constant-speed chillers.
Their efficiency benefits less from the reduced lift, so the increase in CHW
pump energy more than offsets the chiller savings. For these plants, the reset
logic from valve position is the same but the DP set point is reset preferentially
instead of CHWST. This is shown in Figure 7-18.
Chiller Staging
Constant-Speed Chiller Staging
For a plant composed of single-speed chillers, the most efficient logic is to
operate no more chillers than required to meet the load. Chiller efficiency actu-
ally improves as load and lift fall (see Figure 7-19), so it may seem to make
Figure 7-19 Typical chiller part-load performance with and without VFDs (includes con-
denser water relief, as defined in AHRI Standard 550/590).
sense to run, say, two chillers at 40% load rather than one at 80% load, but this
chiller savings is offset by the added energy of an additional CW pump (and an
additional primary CHW pump on primary/secondary systems).
Logic for staging chillers on is straightforward: a new chiller is started when
the operating chillers are no longer able to meet the load, as indicated by plant
leaving water temperature rising above set point by 1F or 2F.
For plants that have the ability to reset CHW temperature, it is important
that the CHW temperature of operating chillers be reset 2F or so below design
CHW temperature to ensure chillers are fully loaded before starting the next
chiller. This reduces the efficiency of the operating chiller, but, for most single-
speed chiller plants, the total plant energy use will be less than if another
chiller were started. See Chilled-Water Temperature Reset logic above in the
section Chilled-Water Temperature and Differential Pressure (DP) Set Point
Reset section.
Logic for staging chillers off is trickier. First, it must be conservative
because staging a chiller off prematurely will cause it to stage back on too
quickly, causing excessive wear on the chiller and starters. The logic must
determine that the load can be comfortably handled with one less chiller. For
primary-only variable-flow plants (which should have a flowmeter for mini-
mum flow control and pump staging), load can be determined from a Btu
meter or calculated from flow and supply/return water temperatures. For con-
stant-volume systems, flow is typically presumed from design data or mea-
sured once during balancing and assumed constant thereafter, thereby

making a flowmeter unnecessary. For primary/secondary systems, stage-
down logic cannot just be based on load; it must also ensure that primary
flow always exceeds secondary flow to avoid the death spiral described in
Chapter 4. This is best accomplished by installing a flowmeter in the second-
ary loop. A chiller can be staged off only if the load determined from this
meter is below the capacity of the remaining operating chillers and the sec-
ondary flow rate will be lower than the primary flow rate (determined from a
flowmeter or as described above for constant-flow systems) with the remain-
ing operating primary CHW pumps.
Variable-Speed Chiller Staging

Figure 7-19 shows the performance of fixed-speed versus variable-speed
chillers with so-called condenser water reliefcondenser water temperatures
fall with the load per AHRI Standard 550/590. Note that without the reduction
in lift provided by the reduced condenser water temperatures, the part-load
efficiency of both constant-speed and variable-speed chillers gets worse at part
load. The efficiency of fixed-speed chillers with condenser water relief
improves as the load falls from peak, but, for most chillers, efficiency will start
to rise above design efficiency at about 40% to 50% load. However, variable-
speed chillers do not suffer from this degradation in efficiency until the load is
very low, about 20% of full load with condenser relief. Because of the opera-
tion of ancillary equipment, such as condenser and primary CHW pumps, the
overall plant efficiency will start to degrade at an even higher part-load point,
as shown in Figure 7-21, but still well below 50%. This figure is for a plant
with fixed-speed condenser water and primary CHW pumps. For systems with
variable-speed primary-only pumps, the staging point is even lower, as in Fig-
ure 7-20. Figure 7-21 also corroborates the conventional wisdom that effi-
ciency when staging fixed-speed chillers is maximized when operating chillers
are maxed out before starting the next stage.
Figure 7-20 shows the optimum number of chillers that should be run plot-
ted against plant load for variable-speed centrifugal chillers in a plant with two
equally sized chillers and variable primary distribution. The graph shows that it
is often optimum to operate two chillers as low as 25% of overall plant load.
This result may seem somewhat counterintuitiveconventional wisdom is to
run as few chillers as possible. That is true for fixed-speed chillers but not for
variable-speed chillers, which are more efficient at low loads when condenser
water temperatures are low.
Figure 7-20 shows that staging chillers based on load alone does not opti-
mize performance because there is a fairly wide range where either one or two
chillers should be operated. There is also another problem with staging based
on load alone: it can cause the chillers to operate in surge. This can be seen in
Figure 7-22, which schematically shows centrifugal chiller load versus lift,
defined as the difference between condenser and evaporator refrigerant tem-
peratures. If two chillers are operated when the lift is high (upper horizontal
Figure 7-20 TOPP variable-speed chiller staging versus plant load ratio (Albuquerque).
Figure 7-21 Two-chiller plant part-load performance with and without VFDs.
Figure 7-22 Possible surge problem staging by load only.

line), the chillers will operate in the surge region. To avoid surge, the chiller
controllers will speed up the compressors and throttle inlet guide vanes to con-
trol capacity. This reduces chiller efficiency so that it would then be more effi-
cient to operate one chiller rather than two. But if the lift is low (lower
horizontal line in Figure 7-22), the chillers would not be in surge, so operating
two chillers would be more efficient than operating one. Therefore, in addition
to load, chiller staging must take chiller lift into account. (This consideration
applies only to centrifugal chillers; surge does not occur with positive displace-
ment chillers, such as those with screw compressors.)
Figure 7-23 shows the optimum number of operating chillers (light gray
dots indicate one chiller while dark grey dots indicate two chillers) as deter-
mined by TOPP simulations. For all plant design options and for all climate
zones, good correlations were found for the optimum staging point described
by a straight line:
SPLR = E (CWRT CHWST) + F (7-4)
where staging part-load ratio (SPLR) is the PLR staging and E and F are coef-
ficients that vary with climate and plant design as shown in Appendix A. If the
actual measured PLR is less than the SPLR, one chiller should operate; if the
PLR is larger than the SPLR then two chillers should operate, with a time delay
to prevent short cycling.
Figure 7-23 Optimum staging versus (CWRT CHWST) and plant PLR.
Primary-only variable-flow plants also require coordinated staging

sequences and minimum flow control to avoid chiller trips. These sequences
and why they are needed are discussed in more detail in Chapters 4 and 5.
Water-Side Economizer (WSE) Control

Recommended control sequences for integrated WSEs are as follows:
Reset CHW supply temperature set point based on valve demand, as

described in the Chilled-Water Temperature and Differential Pressure (DP)
Set Point Reset section.
Enable the economizer if the CHW return temperature is greater than the
predicted HX leaving water temperature (PHXLWT) plus 2F. The 2F dif-
ferential is needed to avoid expending a lot of cooling tower fan energy for
only minimal economizer load reduction. PHXLWT is estimated using the
equation below:
PHXLWT = T WB + P A HX + P A CT
P A HX = D A HX PLR HX
P A CT = m DT WB T WB + D A CT
where
TWB = current wet-bulb temperature
PAHX = predicted heat exchanger approach
PACT = predicted cooling tower approach
DAHX = design heat exchanger approach
PLRHX = predicted heat exchanger part-load ratio (current chilled-
water flow rate divided by design heat exchanger chilled-
water flow rate)
DTWB = design wet-bulb temperature
DACT = design cooling tower approach
m = slope developed from the manufacturers cooling tower
selection program or empirically after the plant is
operational (typical values are 0.2 to 0.5 for near-constant
load applications like data centers; for office type
applications, m is typically in the range of 0.2 to 0)
Disable the WSE if it is not reducing the CHW return temperature by at

least 1F.
Disable chillers when HX leaving water temperature (HXLWT) is at or

below the desired CHW supply temperature set point.
Enable chillers when CHW supply temperature is greater than the desired
set point. Note that multiple chillers may need to be enabled if the current
CHW flow is well above the design flow of a single chiller.
Run as many tower cells as tower minimum flow limits allow.
Control condenser water flow to roughly match the current CHW flow but
reduce flow (within tower minimum flow constraints) as needed to maintain
a minimum 5F range. The lower flow and higher range improves tower effi-
ciency and reduces pump power. Flow can be controlled by staging pumps,
modulating speed on variable-speed pumps, and/or modulating isolation
valves on the HX. Flow rate can be measured directly with a flowmeter (full-
bore magnetic or ultrasonic types are recommended to prevent fouling) or
deduced from HX pressure drop.
Tower speed control logic varies based on WSE and chiller status:
When WSE is disabled: control fan speed to maintain normal con-
denser water temperatures, which should be reset from load or wet-bulb
temperature as described above.
When WSE and chillers are enabled: run tower fans at 100% speed.
When chillers are disabled: control speed to maintain HXLWT at
desired CHW supply temperature set point.
The only complex sequence above is predicting when the economizer
should be enabled. Fortunately, if the prediction calculation is off and the econ-
omizer is enabled prematurely, it will shortly be disabled and the plant will see
no disruptions in CHW flow or supply temperature. This contrasts with nonin-
tegrated economizers where switching from economizers to chillers can be dis-
ruptive and guessing wrong about economizer performance can result in chiller
short cycling and temporary loss of CHW supply temperature control.
Real-Time Optimization
The TOPP modeling technique used to develop the sequences recom-
mended above in this chapter has several disadvantages:
It requires significant modeling effort, which not only increases engineer-

ing time but requires expertise not all design engineers have.
Sequences are only valid for the range of conditions in the model of the
plant load profile and concurrent weather conditions. The sequences may
not be optimum or even stable for unexpected operating conditions.
It relies on equipment models being accurate and cannot account for any
degradation in equipment performance over time.
The simplified correlations and curve fits are not always optimal and, in
fact, can be so poor in the case of variable-speed condenser water pump
control that VFDs are not recommended for these pumps.
An alternative to TOPP modeling is to perform real-time TOPP modeling

in the DDC system or on a computer tied to the DDC system network. For
every time step (e.g., 5 minutes or a bit longer depending on transitions such as
equipment start/stop made in the previous time step) based on the current
weather, load, and other operating conditions, the program would iterate
through all possible operating set points, staging, etc. to determine the opti-
mum. These would be implemented in real time and the process would be
repeated for the next time step. Data on equipment performance could also be
collected in real time to allow the program to automatically adjust equipment
models to maintain their accuracy over time (e.g., to account for reduced
chiller performance as tubes foul). This type of control system would truly pro-
vide the most optimum control possible provided equipment models are accu-
rate and maintain their accuracy over time.
There is currently a very limited number of companies providing real-time,
model-based optimization, but it is expected to be standard practice one day,
given its advantages.
Appendix ATOPP Model Coefficients

The plant in Figure 7-6 (with the optional addition of VFDs on condenser
water pumps) serving a typical office building was modeled with all permuta-
tions of the following design variables:
Weather: Oakland, CA; Albuquerque, NM; Chicago, IL; Atlanta, GA;

Miami, FL; Las Vegas, NV
CHWST: reset by valve position from 42F to 57F
Chillers:
Two styles (two stage and open drive)
Efficiency at 0.35, 0.5, and 0.65 kW/ton at AHRI conditions
Towers:
Approach: 3F, 6F, 9F, and 12F
Tower range: 9F, 12F, and 15F
Efficiency: 50, 70, and 90 gpm/hp
Condenser water pumps: with and without VFDs
The control equation coefficients were determined from each run, then these
coefficients were themselves regressed against various design parameters and
weather indicators. The results are shown in the subsections that follow. The
development of these regressions is ongoing to include more weather sites and
chiller variations. Because of the limited range of variables, the coefficients cal-
culated using the equations that follow can be invalid if any of the variables are
out of range from those used in the regressions, so these equations must be used
with care. In each case, we provide simplified coefficients that we have found to
be stable in most applications, although of course they are not optimized.
These control sequences strictly apply to primary-only plants with centrifu-

gal chillers serving air handlers with outdoor air economizers in a typical office
building. It is not known how well they apply to other applications.
Condenser Water Temperature Control

Control cooling tower fan speed to maintain CW return temperature to the
set point (CWRTsp) determined from Equation A-1
LIFT = A PLR + B (A-1a)
CWRTsp = CHWST + LIFT (A-1b)
Lift calculated from Equation A-1b should be no smaller than LIFTm and
no larger than LIFTd.
Regressed coefficients (use with care):
A = 63 + 0.0053 CDD65 0.0087 WBDD55 + 1.67 WB +

0.52 APPROACH 0.029 gpm/hp
B = 18 0.0033 CDD65 + 0.0053 WBDD55 0.26 WB +

0.15 APPROACH 0.014*gpm/hp
Simplified coefficients (recommended):
A = (LIFTd LIFTm)/0.9
B = LIFTd A
Variable-Speed Condenser Water Pumps

Control CW pump speed to maintain CW flow at the set point determined
from Equation A-2:
CWFR = C PLR + D (A-2a)
CWFSP = CWFd CWFR (A-2b)
C and D coefficients must be determined through modeling. VFDs on CW

pumps are not recommended otherwise.
Chiller Staging
Use one chiller when the PLR is less than the SPLR determined from
Equation A-3; use two chillers otherwise:
SPLR = E (CWRT CHWST) + F (A-3)

Regressed coefficients (use with care):

E = 0.057 0.000569 WB 0.0645 IPLV 0.000233 APPROACH
0.000402 RANGE + 0.0399 kW/ton
F = 1.06 + 0.0145 WB + 2.16 IPLV + 0.0068 APPROACH

+ 0.0117 RANGE 1.33 kW/ton
Simplified coefficients:
E = 0.45/(LIFTd LIFTm)
F = E (0.4 LIFTd 1.4 LIFTm)
Variables
APPROACH = design tower leaving water temperature minus WB, F
CHWFR = chilled-water flow ratio, actual flow divided by total plant
design flow
CHWST = current chilled-water supply temperature (leaving evapora-
tor temperature), F
CWFd = design condenser water flow rate
CWFR = condenser water flow ratio, actual flow divided by total
plant design flow
CWRT = current condenser water return temperature (leaving con-
denser water temperature), F
CDD65 = cooling degree-days base 65F (see Table 7-A)
DP = differential pressure, ft H2O
kW/ton = chiller efficiency at AHRI conditions
T = temperature difference, F
= tower efficiency per ANSI/ASHRAE/IES Standard 90.1
IPLV = integrated part-load value per AHRI 550/590, kW/ton
LIFT = CWRT CHWST
LIFTd = lift at design conditions (from equipment schedule) =
CWRTdesign CHWSTdesign
LIFTm = minimum lift at minimum load (from chiller manufacturer)
NPLV = nonstandard part-load value per AHRI Standard 550/590,
kW/ton
RANGE = design tower entering minus leaving water temperature, F
PLR = plant part-load ratio, current load divided by total plant
design capacity
TOPP = theoretical optimum plant performance
WB = design wet-bulb temperature, ASHRAE 1% design condi-
tion, F
WBDD55 = wet-bulb cooling degree-days base 55F (see Table 7-A)
Table 7-A ASHRAE Climate Zone Weather Data

Location Zone CDD65 WBDD55
Miami 1A 4147 5255
Houston 2A 2898 3842
Phoenix 2B 4290 1554
Atlanta 3A 1646 2100
Los Angeles 3B (LA) 564 1127
Las Vegas 3B (LV) 3308 683
San Francisco 3C 198 214
Baltimore 4A 1272 1608
Albuquerque 4B 1406 418
Seattle 4C 272 248
Chicago 5A 948 1094
Boulder 5B 922 225
Minneapolis 6A 803 964
Helena 6B 561 130
Duluth 7A 284 374
Fairbanks 8A 146 87
Example
TOPP modeling was performed for a plant serving an Oakland, CA office
building. The following slopes and intercepts were determined from curve-fits:
A = 47, B =5.2
C = 1.3, D = 0.13
E = 0.009, F = 0.21
Figure 7-A shows the theoretical optimum performance for both variable-
speed (VS) and constant speed (CS) CW pumps compared to our proposed real
sequences using the coefficients listed above. Despite their simplicity, our
sequences resulted in only about 1% higher energy use than the TOPP. VFDs
on the CW pumps saved 3% versus constant-speed pumps, but this was not
enough savings to make them cost-effective at a 14-year simple payback period
for this plant. Also shown in the figure for comparison is plant performance
using the AHRI Standard 550/590 condenser water relief curve to reset con-
denser water temperature (4% higher energy use than our sequences) and per-
formance assuming CWST set point is fixed at the design temperature (16%
higher than our sequences).
Figure 7-A TOPP versus real sequences for both constant-speed and variable-speed CW
pumps.
Appendix BDetailed Sequence of Operation (SOO)

The following is a detailed sequence of operation for the plant in Figure 7-6.
Sequence of Operation (SOO)

1.01 SEQUENCES OF OPERATION
A. General
1. The term proven (i.e., proven on/proven off) shall mean that the
equipments DI status point matches the state set by the equipments
DO command point.
2. The term PID loop or control loop is used generically for all control
loops and shall not be interpreted as requiring proportional plus integral
plus derivative gains on all loops. Unless specifically indicated other-
wise, the following guidelines shall be followed:
a. Use proportional only (P only) loops for limiting loops (such as
zone CO2 limiting loops, etc.) to ensure there is no integral windup.
b. Do not use the derivative term on any loops unless field tuning is
not possible without it.
3. Trim and respond set point reset logic
a. Trim and respond set point reset logic and zone/system reset
requests where referenced in sequences shall be implemented as
described below.
b. Requests are pressure, cooling, or heating set point reset requests
generated by zones or air-handling systems.
1. For each zone or system, and for each set point reset request
type listed for the zone/system, provide the following soft-
ware points:
a. Importance multiplier (default = 1). This point is
used to scale the number of requests the zone/sys-
tem is generating. A value of zero causes the zone/
systems requests to be ignored. A value greater than
zero can be used to effectively increase the number
of requests from the zone/system based on the criti-
cal nature of the spaces served or to increase the
requests beyond the number of ignored requests
(defined below) in the trim and respond reset block.
b. Request hours
1. This point accumulates the integral of requests
(prior to adjustment of importance multiplier) to
help identify zones/systems that are driving the
reset logic. Every x minutes (adjustable, default
5 minutes), add x/60 times the current number of
requests to this request-hours accumulator point.
2. The request-hours point is reset to zero upon a
global command from the system/plant serving
the zone/systemthis global point simultane-
ously resets the request-hours point for all zones/
systems served by this system/plant.
3. Cumulative percent-request-hours is the zone
request hours divided by the zone run hours (the
hours in any mode other than unoccupied mode)
since the last reset, expressed as a percentage.
4. A Level 4 alarm is generated if the zone impor-
tance multiplier is greater than zero, the zone
percent-request-hours exceeds 70%, and the
total number of zone run hours exceeds 40.
2. See zone and air-handling system control sequences for
logic to generate requests.
3. Multiply the number of requests determined from zone/sys-

tem logic times the importance multiplier and send to the
system/plant that serves the zone/system. See system/plant
logic to see how requests are used in trim and respond logic.
c. Variables. All variables below shall be adjustable from a reset
graphic accessible from a hyperlink on the associated system/plant
graphic. Initial values are defined in system/plant sequences below.
Values for trim, respond, time step, etc. shall be tuned to provide
stable control.
Device = associated device (e.g., fan, pump)
SP0 = initial set point
SPmin = minimum set point
SPmax = maximum set point
Td = delay timer
T = time step
I = number of ignored requests
R = number of requests from zones/systems
SPtrim = trim amount
SPres = respond amount
SPres-max = maximum response per time interval
d. Trim and respond logic shall reset set point within the range SPmin
to SPmax. When the associated device is off, the set point shall be
SP0. The reset logic shall be active while the associated device is
proven on, starting Td after initial device start command. When
active, every time step T, trim the set point by SPtrim. If there are
more than I requests, respond by changing the set point by SPres
times (R I), that is (the number of requests minus the number of
ignored requests). But the net response shall be no more than SPres-
max. The sign of SPtrim must be the opposite of SPres and SPres-max.
For example, if SPtrim = 0.1, SPres = +0.15, SPres-max = +0.35,
R = 3, I = 2, then each time step, the set point change = 0.1 + (3
2) 0.15 = +0.05. If R = 10, then set point change = 0.1 + (10 2)
0.15 = 1.1 but limited to a maximum of 0.35. If R2, the set point
change is 0.1.
4. Lead/lag and lead/standby alternation
a. Even wear
1. Lead/lag. Unless otherwise noted, parallel staged devices
(such as pumps, towers) shall be lead/lag alternated when
more than one is off or more than one is on so that the
device with the most operating hours is made the later stage
device and the one with the least number of hours is made
the earlier stage device. For example, assuming there are
three devices, if all three are off or all are on, the staging
order will simply be based on run hours from lowest to
highest. If two devices are on, the one with the most hours
will be set to be Stage 2 while the other is set to Stage 1; this
may be the reverse of the operating order when the devices
were started. If two devices are off, the one with the most
hours will be set to be Stage 3 while the other is set to
Stage 2; this may be the reverse of the operating order when
the devices were stopped.
2. Lead/standby. Unless otherwise noted, parallel devices
(such as pumps and towers) that are 100% redundant shall
be lead/standby alternated when more than one is off so that
the device with the most operating hours is made the later
stage device and the one with the least number of hours is
made the earlier stage device. For example, assuming there
are three devices, if all three are off, the staging order will
be based on run hours from lowest to highest. If devices run
continuously, lead/standby shall switch at an adjustable run-
time; standby device shall first be started and proven on
before former lead device is changed to standby and shutoff.
b. Exceptions
1. Operators shall be able to manually fix staging order via
software points on graphics overriding the even wear logic
above but not overriding the failure or hand operation logic
below.
2. Failure: If the lead device fails or has been manually
switched off, the device shall be placed into high-level
alarm (Level 2) and set to the last stage position in the lead/
lag order until alarm is reset by operator. Staging position of
remaining devices shall follow the even wear logic. A failed
device in alarm can only automatically move up in the stag-
ing order if another device fails. Note that a device in alarm
will be commanded to run if the sequence calls for it to run.
In this way the BAS will keep trying to run device(s) until it
finds enough that will operate. Failure is determined by:
3. Variable-speed fans and pumps
1. VFD critical fault is ON or
2. Status point not matching its ON/OFF point for
3 seconds after a time delay of 15 seconds when
device is commanded ON or
3. Supervised HOA at control panel in OFF posi-
tion or
4. Loss of power (e.g., VFD DC bus voltage =
zero)
b. Constant-speed fans and pumps

3 seconds after a time delay of 15 seconds when
device is commanded ON or
2. Supervised HOA at control panel in OFF position
c. Chillers
1. Chiller alarm contact or
2. Chiller is manually shut off as indicated by the
status of the local/auto switch from chiller gate-
way or
3. Chiller status remains off 5 minutes after com-
mand to start.
4. Hand operation. If a device is on in hand (for example via an
HOA switch or local control of VFD), the device shall be set
to the lead device and a low-level alarm (Level 4) shall be
generated. The device will remain as lead until the alarm is
reset by the operator. Hand operation is determined by
a. Variable-speed fans and pumps
15 seconds when device is commanded OFF or
2. VFD in local hand mode or
3. Supervised HOA at control panel in ON position.
b. Constant-speed fans and pumps
15 seconds when device is commanded off or
2. Supervised HOA at control panel in ON position.
c. Chillers
1. Chiller is manually turned on as indicated by the
status of the local/auto switch from chiller gate-
way.
B. Chiller plant
1. Chillers shall be lead/lag alternated per Paragraph 1.01A.4. If a chiller

is in alarm, its CHW isolation shall be closed.
2. Chillers are staged in part based on load calculated from thermal
energy meter except for 15 minutes after a stage-up or stage-down tran-
sition, freeze-calculated load at the value at the initiation of the transi-
tion. This allows steady state to be achieved and ensures a minimum ON
and OFF time before changing stages.
3. Staging shall be as follows. Timers shall reset to zero after every stage
change. Each stage shall have a minimum runtime of 15 minutes
(including Stage 0). Plant PLR is calculated load divided by total
chiller design load as scheduled on drawings. Lockout temperature
(LOT) shall be 60F (adjustable). The chiller plant shall include an
enabling schedule that allows operators to lock out the plant during off
hours (e.g., to allow off-hour operation of HVAC systems, except the
chiller plant; the default schedule shall be 24/7 [adjustable]). The stag-
ing part-load ratio shall be calculated every 5 minutes as
SPLR = E(TCWR TCHWS) + F
where E = xx and F = ?? (see Appendix A.)
Chillers Nominal Stage Down to Lower

Stage Stage Up to Next Stage if Either
On Capacity Stage if
Any chiller plant
requests and OAT >
0 All off 0
LOT and schedule is
active
CHW plant reset = No chiller plant
Lead for 15 minutes PLR is 100 for 15 minutes requests for 5 minutes
2 50%
chiller greater than SPLR and PLR greater or OAT < (LOT 5F)
than 30% or schedule is inactive
Both for 15 minutes PLR
3 100%
chillers less than SPLR
4. Whenever there is a stage-up command:

a. Command operating chillers to reduce demand to 50% of their cur-
rent load. Wait until actual demand <55% up to a maximum of
5 minutes before proceeding.
b. Slowly change the minimum bypass controller set point to that
appropriate for the stage as indicated in Paragraph 1.01B.8 below.
(If the bypass valve suddenly opens then the chiller load will sud-
denly drop and the chiller(s) could trip. Also, the DP control loop
will be unstable). After new set point is achieved wait 1 minute to
allow loop to stabilize.
c. Start the next CW pump and when the CW isolation/head pressure
control valve feedback indicates the valve is more than 10% open,
wait 30 seconds.
d. Slowly open CHW isolation valve of the chiller that is to be started.
The purpose of opening slowly is to prevent sudden disruption to
flow through active chillers. Valve timing to be determined in the
field as that required to prevent nuisance trips.
e. Start the next stage chiller after CHW isolation valve is full open.
f. Release the demand limit.
5. Whenever there is a stage-down command:
a. Shut-off last stage chiller.
b. When the controller of the chiller being shut off indicates no
request for chilled-water flow, slowly close the chillers CHW isola-
tion valve to avoid sudden change in flow through other operating
chiller.
c. When the CW isolation/head pressure control valve feedback indi-
cates the valve is fully closed, shut off last-stage condenser water
pump.
d. Change the minimum bypass controller set point to that appropriate
for the stage as indicated in Paragraph B.8 of Appendix B.
6. Condenser water pumps:
a. Condenser water pumps shall be lead/lag alternated per
Paragraph B.4 of Appendix B.
b. See Paragraph B.4 of Appendix B and Paragraph B5 of Appendix B
for ON/OFF staging sequence.
7. Chilled-water pumps:
a. Chilled-water pumps shall be lead/lag alternated per Paragraph B.4
of Appendix B.
b. CHW pumps shall be staged as a function of CHW flow ratio
(CHWFR = actual flow divided by total plant design flow). The
lead pump shall run whenever the plant is enabled. When CHWFR
is above 47% for 10 minutes, start the lag pump. When CHWFR is
below 47% for 15 minutes, shut off the lag pump.
c. When any pump is proven on, pump speed will be controlled by a
PID loop maintaining the DP signal at a set point determined by the
reset scheme described below. All pumps receive the same speed
signal. Minimum speed set point in VFDs shall be 10%.
8. Bypass valve: When any CHW pump is proven ON, the bypass valve
shall be enabled, and opened otherwise. Bypass valve shall be modu-
lated to maintain minimum flow as measured by the flowmeter. Mini-
mum flow rates are as follows (based on manufacturers minimum flow
rates plus 10% to ensure control variations do not cause flow to go
below actual minimum):
Chiller Stage Minimum Flow

0 0
1 xx
2 xx
9. Chilled-water supply temperature set point and pump differential static

pressure set point shall be reset based on the figure below and the value
CHW plant reset determined as described below. DPmax shall be deter-
mined under Section 230593 Testing, Adjusting and Balancing. Tmin is
the design chilled-water temperature as scheduled on Drawings.
Figure 7-B CHW plant reset.
a. CHW Plant Reset Shall be reset using trim and respond logic (see
Paragraph 1.01A.3) with the following parameters:
Variable Value
Device Any CHW pump
SP0 0%
SPmin 0%
SPmax 100%
Td 15 min
T 5 min
I 2
R Cooling CHWST reset requests
SPtrim 2%
SPres +3%
SPres-max +7%
b. CHW plant reset logic shall be disabled and value fixed at its last
value for 15 minutes after the plant stages up or down.
10. Cooling tower
a. Fans are controlled off of CW return temperature (leaving chiller)
rather than supply. Tower fans are enabled when any CW pump is
proven on and CWRT rises above set point by 1F. If CWRT drops
below set point and fans have been at minimum fan speed for
5 minutes, fans shall cycle off for at least 3 minutes and until
CWRT rises above set point by 1F.
b. Condenser water return temperature set point shall be
CWRTsp = CHWST + LIFTx
LIFTx = A PLR + B
where PLR is the plant PLR (actual chiller load divided by total plant
design capacity), A =?? and B =?? (See Appendix A), but in no case
shall LIFTx be less than the minimum lift at low load from the manu-
facturer (12F) nor more than the design lift (45F).
c. Fan speed shall be modulated to maintain CWRT at set point. All
operating fans receive the same speed signal.
11. Tower makeup water
a. Makeup water valve shall cycle based on tower water fill level sen-
sor. The valve shall open when water level falls below the minimum
fill level recommended by the tower manufacturer. It shall close
when the water level goes above the maximum level recommended
by the tower manufacturer.
12. Performance Monitoring
a. Total plant power. Calculate total plant power as the sum of chiller
power, pump power, and cooling tower fan power. For motors with
VFDs, power shall be actual power as indicated by the VFD. For
fixed-speed motors (e.g., CW pumps), power shall be assumed to be
fixed at brake power (from equipment schedule) 0.746/0.93
(approximate motor efficiency).
b. Summary Data. For each chiller and total plant, statistics shall be
retained and displayed on graphic for runtime, average actual effi-
ciency (kW/ton), and average demand (tons) and load (ton-hours).
Show on chiller plant graphic: instantaneous values, year-to-date
totals/averages and previous year totals/averages.
13. Alarms
a. Maintenance interval alarm when pump has operated for more than
1500 hours: Level 5. Reset interval counter when alarm is acknowl-
edged.
b. Maintenance interval alarm when chiller has operated for more than
1000 hours: Level 5. Reset interval counter when alarm is acknowl-
edged.
c. Chiller alarm: Level 2.
d. Tower level:
1. If tower water level sensor indicates low water level, gener-
ate a Level 2 alarm.
2. If tower water level sensor indicates high water level, gener-
ate a Level 3 alarm.
e. High chiller leaving chilled-water temperature (more than 5F
above set point) for more than 15 minutes when chiller has been
enabled for longer than 15 minutes: Level 3.
f. Pump or tower fan alarm is indicated by the status input being dif-
ferent from the output command after a period of 15 seconds after a
change in output status.
1. Commanded on, status off: Level 2
2. Commanded off, status on: Level 4
g. Head pressure CW valve is partly closed and tower fan is on for
more than 10 minutes: Level 3.
h. Excessive CW approach indicating water side fouling: If leaving
condenser water temperature is more than 3F below refrigerant
condensing temperature for 15 minutes at least 15 minutes after
chiller start.
i. Excessive CHW approach indicating water side fouling: If leaving
chilled-water temperature is more than 3F above refrigerant evapo-
rator temperature for 15 minutes at least 15 minutes after chiller
start.
References
AHRI. 2015. AHRI Standard 550/590, 2015 Standard for performance rating
of water-chilling and heat pump water-heating packages using the vapor
compression cycle. Arlington, VA: Air-Conditioning, Heating, and Refrig-
eration Institute.
ASHRAE. 2011. Fundamentals of HVAC control systems. Atlanta: ASHRAE.
ASHRAE. 2012. ASHRAE Guideline 22-2012, Instrumentation for monitor-
ing central chilled-water plant efficiency. Atlanta: ASHRAE.
ASHRAE. 2016a. ANSI/ASHRAE/IES Standard 90.1-2016, Energy standard
ASHRAE. 2016b. Chapter 47, Valves, ASHRAE HandbookHVAC Systems
and Equipment. Atlanta: ASHRAE.
Edwards, T.J. 1983. Observations on the stability of thermistors. Review of Sci-
entific Instruments 54:613; doi:10.1063/1.1137423.
Hartman, T. 2005. Designing efficient systems with the equal marginal perfor-
mance principle. ASHRAE Journal 47(7):6470.
Hydeman, M., and G. Zhou. 2007. Optimizing chilled water plant control.
ASHRAE Journal 48:4554.
Hydeman, M., N. Webb, S. Blanc, and P. Sreedharan. 2002. Development and
testing of a reformulated regression-based electric chiller model. ASHRAE
Transactions 108(1).
Lawton, K.M., and S.R. Patterson. 2002. Long-term relative stability of therm-
istors. Precision Engineering 26(3):34045.
Li, W., and R. Webb. 2001. Fouling characteristics of internal helical-rib
roughness tubes using low-velocity cooling tower water. International
Journal of Heat and Mass Transfer (6).
NBCIP. 2004. Product testing report: Duct-mounted relative humidity trans-
mitters. Iowa Energy Center, April 2004, National Building Controls Infor-
mation Program. http://www.iowaenergycenter.org/wp-content/uploads
/2017/02/PTR_Humidity_Rev.pdf.
NBCIP. 2005. Product testing report supplement: Duct-mounted relative
humidity transmitters. Iowa Energy Center, July 2005, National Building
Controls Information Program. http://www.iowaenergycenter.org/wp
-content/uploads/2017/02/NBCIP_S.pdf.
Condenser water system design. ASHRAE Journal 9:1425.
Optimized control sequences. ASHRAE Journal 6:5674.
Taylor, Steve. 2015. Setpoint reset using trim & respond logic. ASHRAE Jour-
nal 57(6).
Rishel, J.B. 2001. Wire-to-water efficiency of pumping systems. ASHRAE
Journal 43(4).

7-1 Which of the following statements regarding commercial versus industrial con-
trols is not true:
a. Commercial controls are not suitable for most CHW plant applications
due to their inferior reliability.
b. Industrial controls should be used for mission-critical applications
requiring parallel operation of redundant controllers.
c. Commercial control systems have lower first costs.
d. Commercial control systems cost less to maintain.
7-2 You are in the process of specifying control sensors for a CHW plant used to
serve both comfort cooling and mission-critical loads. The plant has variable-
speed primary CHW pumps. Which of the follow options is most appropriate
for monitoring CHW pump status?
a. Hard wire status from the VFD.
b. Read status over the network from the VFD.
c. Install a current transducer on one phase on the line side of the VFD.
d. Install a DP switch across the pump.
7-3 Which of the following statements regarding calibration of water temperature

sensors is true?
i. Field calibration is typically not necessary for noncritical sen-
sors.
ii. Field calibration reference temperature sensors should be at
least twice as accurate as the field-installed sensors.
iii. Dry well baths are always required to field calibrate the tem-
perature sensors used for CHW plant load measurement.
iv. RTDs are more stable than thermistors and therefore require
less frequent repeat calibration.
a. (i), (ii)
b. (i), (ii), (iii), (iv)
c. (i), (ii), (iv)
d. (ii), (iv)
7-4 You are specifying two-way control valves for air-handler CHW coils served
by a primary, variable secondary plant. The valve sizes vary between 1.5 and
4 in. What valve type(s) are most appropriate for this application?
a. Globe valves for all sizes.
b. Ball valves for 3 in. and less; globe valves for all greater than 3 in.
c. Ball valves for all sizes.
d. Ball valves for 3 in. and less; butterfly valves for all greater than
3 in.
7-5 You are designing a variable-speed primary loop that controls pump speed
based on DP measured by a DP sensor installed far out in the loop, remote
from the mechanical room where the pumps are to be installed. A colleague
recommends specifying that the DP sensor be wired to a controller located near
the sensor, then passing the sensor reading to the pump controller over the net-
work. This approach
a. Can save on wiring costs relative to hardwiring the DP feedback to the
pump controller.
b. May result in control loop instability due to network latency or limited
feedback polling frequency.
c. Increases network traffic and may slow control system performance.
d. Will lead to loss of pump feedback control if network communications
between controllers are lost.
7-6 Monitoring CHW plant efficiency using modern DDC control systems
a. Typically requires installation of expensive true-RMS power meters.
b. Typically requires installation of CHW loop instrumentation beyond
that necessary for CHW plant control.
c. Requires additional programming on the part of the DDC controls con-
tractor.
7-7 Plant optimization modeling indicates that the most efficient approach to con-
trol cooling tower fans serving a plant with typical office loads, short of using
real-time optimization, is
a. Controlling fans to maintain a condenser water supply temperature set
point fixed to the minimum allowed by the chiller minimum lift
requirement.
b. Controlling fans to maintain a condenser water supply temperature set
point fixed at the design temperature to minimize tower fan energy.
c. Controlling fans to maintain a condenser water supply temperature
reset as a function of ambient wet-bulb temperature.
d. Controlling fans to maintain a condenser water return temperature set
point reset based on plant load and CHW supply temperature.
7-8 Variable-speed control of condenser water pumps

a. Can easily increase plant energy use if not controlled optimally.
b. May be cost-effective in high load applications, such as data centers
that require operation under a variety of load and ambient wet-bulb
conditions, but modeling or real-time optimization is needed to deter-
mine the optimal sequence.
c. Is highly climate dependent: the optimal strategy in one climate is
unlikely to be optimal in another.
7-9 The optimal approach to resetting CHW supply temperature set point and
CHW DP set point for a plant with variable-speed chillers is:
a. Dependent on climate.
b. Dependent on plant load profile.
c. Resetting CHW supply temperature set point first, then CHW DP set
point.
d. Resetting CHW DP set point first, then resetting CHW supply tempera-
ture set point.
Commissioning
Instructions
Introduction
This chapter focuses on elements of the commissioning (Cx) process that
are key to ensuring that chiller plants meet their design intent.
Commissioning Overview
ASHRAE defines commissioning as a systematic process of ensuring that
systems are designed, installed, functionally tested, and capable of being oper-
ated and maintained to perform in conformity with the design intent
(ASHRAE 2007).
Commissioning is intended to achieve the following objectives:
Ensure that equipment and systems are properly installed and receive ade-
quate operational checkout by the installing contractors.
Verify and document proper operation and performance of equipment and
systems.
Ensure that the design intent for the project is met.
Ensure that the project is thoroughly documented.
Ensure that the facility operating staff is adequately trained.
Commissioning may be considered a quality control process that adds

value by helping to meet the projects design intent and achieve top perfor-
mance of a CHW plant.
This chapter is not intended to be a primer on the commissioning process
and its costs and benefits. The cost-effectiveness of commissioning HVAC sys-
tems, in particular complex systems such as CHW plants, is well documented.
Instead, this chapter focuses on key commissioning activities applied to CHW
plants:
SOO review
Point-to-point checkout
284 Chapter 8 Commissioning
Functional testing
Trend review
Other commissioning activities, such as design and submittal reviews,

operator training, and issues logging and resolution, are not addressed in detail.
For further information on the commissioning process, refer to ASHRAE
Guideline 1.1-2007, HVAC&R Technical Requirements for the Commissioning
Process, and other ASHRAE materials on commissioning.
This chapter also does not address who should serve as the commissioning
authority (CxA), the person or group of people who coordinate and oversee Cx
activities. There are debates about whether the engineer of record (EOR) (sup-
ported by project subcontractors) can successfully serve as the CxA or whether
the CxA should be a third party, independent of the design and construction
teams. The pluses and minuses of each approach are outside the scope of this
SDL. The Cx activities addressed in this chapter apply well to either approach.
Commissioning Focus
While following a comprehensive Cx process (as outlined in the many
ASHRAE Cx documents and guidelines) will improve plant performance, it is
important that the Cx process focus on the most likely source of plant under-
performance: the control system. Occasionally, plant performance may be
compromised by a chiller, pump, or tower that does operate as the manufac-
turer claims, but this is rare. Instead, when a plant has problems, the cause is
almost always due to the controls, primarily due to improper control sequences
and/or improper implementation of the sequences (BAS programming). The
two most effective tools for identifying these control problems are as follows:
Functional tests: Functional tests are intended primarily to ensure that the
control sequences specified by the design engineer have been properly pro-
grammed into the BAS. The tests essentially mimic the control sequences
by simulating an operating scenario and observing the BAS response.
Trend reviews: Trend reviews consist of collecting data of actual system
operation under normal conditions using the BAS trending capabilities then
analyzing the data to see if the BAS is operating properly. In addition to
complimenting functional testing to verify proper BAS programming, trend
reviews can also identify flaws in the control sequences (e.g., when the sys-
tem reacts inappropriately to an operating condition that was not antici-
pated in the control sequences).
Likewise, the two most effective tools for minimizing the potential for control
system issues to arise during functional testing and trend review are as follows:
SOO review: SOO review entails analyzing the specified sequence in detail
to identify any logic flaws that may prevent the sequence from operating
the plant without issue. Because the CxA will ultimately develop functional
tests to validate the sequence, this also provides an early opportunity to

identify any unclear aspects of the written sequence that may ultimately
impact the control system programmers ability to code to the design intent.
Point-to-point checkout: As discussed in Chapter 7, the accuracy required
of temperature sensors used for CHW plant control exceeds that required of
most other commercial HVAC system components. As such, the point-to-
point checkout process is particularly critical to ensure that (1) all points
are mapped into the BAS properly and (2) all points are calibrated properly,
with appropriate documentation, to ensure that proper plant control can be
achieved.
Sequence of Operation (SOO) Review

SOO review provides the CxA an opportunity to identify issues with plant
control logic before it is implemented and tested in the field. The value of finding
issues before they exist cannot be overstated: changing a few lines of text in the
written SOO is much faster than (1) identifying an issue during functional testing
or trend review, (2) getting sign off from the EOR on modifying the sequence,
(3) having the controls contractor implement the sequence revision, (4) repeating
functional tests for the revised portion of the sequence, (5) reviewing trends
again, and (6) tracking the entire process through the issues log.
If the EOR is acting as the CxA for the project, then a second set of eyes,
either from within the EORs firm or a third party hired specifically for a peer
review, should be engaged to complete a review of the SOOs.
Key questions to keep in mind while reviewing the SOOs include:
Does the plant controls system include all of the necessary hardware points
to execute the sequence as specified?
Is the logic too vaguely expressed to be executed as intended by a controls
programmer?
Are there aspects of the sequence that may yield control instability, equip-
ment cycling, or damage?
Can the logic be simplified in any way to achieve the expressed intent with
less complication?
Are there clear sections of the logic that will work, but unnecessarily
waste energy while doing so? If so, how might they be amended?
The first question is typically easiest to answer and simply requires due dil-
igence on the part of the CxA. For example, if the sequence for a variable pri-
mary/variable secondary plant calls for resetting the primary pump speed if the
secondary flow rate exceeds the primary flow rate but does not include a sec-
ondary loop flowmeter, this is an easy red flag to the CxA. Is the design intent
to include a secondary loop flowmeter, or should the SOO instead be modified
to reset primary loop flow based on temperature sensors (e.g., the secondary
and primary return temperature sensors)?
The second question is typically partially answered during an initial read-

through of the SOOs. Going through the thought process of how each section
of the sequence should be functionally tested often further reveals aspects of
the sequence that are unclear but upon initial inspection seemed fine. If there
are aspects of the sequence that are unclear to the CxA, then the controls con-
tractor tasked with implementing the sequence is likely to face the same
problem.
The third question requires more detailed inspection of the sequence and
thought experiments to work through how the sequence will function in prac-
tice. For each control loop and each piece of staging logic called for in the
sequence, the CxA should evaluate how it will work in the field. As an exam-
ple, consider a sequence that calls for staging primary pumps based on flow
rate. Does the sequence include minimum stage runtime delays to prevent short
cycling, or does it include hysteresis on the stage-up/stage-down set points to
achieve the same effect? If neither, then there is a problem. As another exam-
ple, does the sequence stage chillers based on plant load but lack logic to reset
the CWRT set point based on plant load as well? If so, the sequence may pre-
dispose the chillers to surge.
The fourth question should be kept in mind at all times because the written
sequence ultimately has be implemented in programming. The more compli-
cated the programming, the more likely issues will reveal themselves during
functional testing and trend review. The goal of answering this question is not
to simplify the logic just for simplicitys sake but to simplify it as much as pos-
sible while still achieving the design intent.
The fifth question more commonly yields conflict between the CxA and the
EOR than the others because it comes down to design preference, not opera-
tional integrity. However, if the sequence unnecessarily wastes energy by run-
ning equipment suboptimally, it is the CxAs responsibility to bring these
findings to the owners and EORs attention. Does the sequence call for staging
towers even though minimum tower flow can be achieved with all towers run-
ning? If so, tower fan energy is wasted. Does the sequence call for always run-
ning as many primary CHW pumps as possible due to a misapplication of the
affinity laws? If so, pump energy is wasted.
Point-to-Point Checkout
The point-to-point checkout procedure for CHW plants is no different in
implementation than the point-to-point checkout for any other commissioned
HVAC equipmentit is just more critical that it is done correctly. To that end,
the CxA should be thoroughly engaged in the process. Instead of just reviewing
and cataloging point-to-point forms completed by the contractor, the CxA
should be actively involved in crafting the forms to ensure that all necessary
information is documented. At a minimum, point-to-point checkout forms
should include the following:
Sensor calibration entries for all field-calibrated sensors. Recorded data

should include reference instrumentation measured value, BAS connected
sensor value, implemented calibration offset, and implemented calibration
slope (as applicable).
Binary input and output status verification. Regardless of whether an input
device (e.g., an isolation valve end switch) is identified as normally open or
normally closed in submittal data, the actual condition corresponding to a
hot (energized) input should be verified on the point-to-point form. Simi-
larly, for output points, the operational condition corresponding to a hot
output should be recorded.
Analog output configuration. Forms should include fields to record how
controller outputs scale to controlled device outputs. This typically requires
recording the operational state at the low and high ends of the control sig-
nal output range. Using VFD hardwired speed control as an example, a
10 V controller output may correspond to 6 Hz at the drive, while a 10 V
output may correspond to 60 Hz at the drive. Alternatively, 0 V at the con-
troller may correspond to 0 Hz at the drive. The data in this example sce-
nario provide valuable indication to the future operator, and the CxA
during testing and trend review, of whether minimum speed control is
implemented at the drive or through the BAS.
Analog input configuration. As with analog outputs, forms should include
entries for how controller inputs from transducers scale to outputs from
sensors. As an example, for a 020 psi DP sensor, the form should note the
corresponding input rangelikely 4 to 20 mA. For sensors that do not uti-
lize transducers, such as thermocouples, the device type (e.g., Type K)
should be noted for verification of correct output mapping by the BAS.
Trend configuration. For each I/O point, forms should indicate whether a
trend has been configured (Yes/No) and the trending interval or COV con-
figuration as appropriate.
In addition to the above field-completed forms, the CxA should also pre-
pare a list of required factory calibration certifications and NIST-traceable cali-
bration certificates for all sensors called out as such in the controls
specification. Point-to-point checkout should not be considered complete until
the field forms are completed by the controls contractor and reviewed by the
CxA and all required calibration certificates are in hand.
The above data provide a configuration record for all hardwired compo-
nents in the controls system. These data are useful to plant operators and con-
trol technicians tasked with modifying the plant controls in the future. Perhaps
more importantly, requiring detailed point-to-point checkout documentation
inherently requires the controls contractor to execute a detailed point-to-point
checkout procedure. This in turn increases the probability that issues with
faulty sensors, or improperly mapped inputs and outputs, are identified early in
the Cx process before they become bigger issues during plant start-up, func-
tional testing, and trend review.
Functional Testing
Role of Functional Testing
Functional testing is intended to verify that the BAS has been properly pro-
grammed to execute the specified sequence of controls (Chapter 7). Tests
should be performed after the plant has completed traditional start-up and test-
ing, adjusting, and balancing (TAB) procedures and the BAS has been fully
programmed.
All plant control sequences and alarms should be tested. This is done by
creating a test script that challenges the BAS with different operating scenarios
and observing and recording the response. If the actual response matches the
expected response, the test is successful. It is not at all uncommon for tests to
be performed several times before all are completed successfully. Typically, Cx
procedures require the BAS contractor to perform the tests on their own prior
to CxA witness testing. The tests, or a subset of key tests, should also be con-
ducted in the presence of plant operating engineers so that they can learn first-
hand how the plant is expected to operate.
Functional tests also serve a secondary role: they interpret the written con-
trol sequences. Because of the imprecise nature of written language, and per-
haps the lack of detail common to many written sequences, the specifics of
how the system is to respond to various operating scenarios is seldom precisely
clear. The functional tests then can serve as a formal interpretation of the
designers intent. It is with this added function in mind that thought should be
given to how sequences will be tested during the initial SOO review, as noted
above.
Creating Proper Testing Conditions

Functionally testing chiller plants has an added complication that does not
exist with most other mechanical systems: there may not be any actual load
when tests are being performed. For example, servers have not been installed in
a data center, the plant serves a speculative office building that does not yet
have tenants, or testing is being conducted in the winter. Short of waiting until
real loads exist, there are a few strategies that can be used, discussed in the fol-
lowing subsections.
Option 1: Simulated Loads

Plant staging logic and set point reset logic are generally functions of load and
flow, as discussed in Chapter 7. Design and variable-flow rates can generally be
achieved without actual loads by opening coil control valves. (If there are no valves
yet installed [e.g., if all coils are installed by future tenants], then the plant simply
cannot be tested.) Varying load can then be simulated by varying flow by manually
overriding pump speed and overriding supply and return temperatures. The chillers
are locked out from actually starting at their control panels, so no real cooling is
done, but the commands to start the chiller from the BAS can be observed and cor-
rect staging logic can be confirmed. Chiller status may also have to be overridden
to ON when the chiller is commanded on to simulate proper operation. Chiller fail-
ures can be simulated by overriding status and/or leaving water temperatures.
Pump start/stop and staging logic can also be readily tested in this way. The plants
response to degrading T (Chapter 4) can also be simulated by overriding return
CHW temperature to lower values. Cooling tower set point reset and low-load fan
cycling can also be tested, but fan speed control to maintain set point cannot; that
would have to wait for the trend review phase, discussed in a later section.
Option 2: Artificial Loads

It may be possible to create loads on the CHW system using the heating
system. Example strategies include the following:
For a VAV system, cooling loads at AHU CHW coils can be created by
resetting all zones to maximum airflow, resetting zonal discharge air tem-
perature set points to maximum, locking out the outdoor air economizer,
and setting the air-handler supply air temperature (SAT) set points low.
Varying AHU SAT set points and manually modulating AHU fan speed can
be used to achieve the desired load for testing.
Loads on AHUs with preheat coils can be created by overriding the preheat
coil valve to create a load on the CHW coil. Again, SAT set points and
AHU fan speed can be adjusted to achieve the desired plant load.
Similar approaches for achieving simultaneous heating and cooling exist

for most other system types as well, and are effective means of creating the
load necessary for plant testing. In most cases, there is no need to physically
interconnect the CHW and hot-water systems. Doing so can lead to complica-
tions due to varying system pressures or water treatment regimens, as well as
possible equipment damage if entering water temperatures to chillers or boilers
are too extreme.
Simulator Factory Witness Testing

One effective means of increasing the probability that testing works the
way it should in the field is to set up a plant simulator in the controls contrac-
tors facility. Simulator tests are run as part of a factory witness test procedure
conducted before the sequence is ever implemented in the field. The simulator
is constructed by loading the actual programming to be installed in the field on
controllers in the laboratory and recreating the field network of controllers in
the laboratory. Hardwired binary I/Os can be connected to relays such that
command outputs and status feedback can be verified in the laboratory. Analog
output commands can be verified in software as well, but understandably the
010 V and 420 mA outputs for all points cannot reasonably be wired in the
laboratory. Analog inputs (temperatures, pressures, etc.) must be overridden to
fixed values in software that are then adjusted throughout testing to represent
desired field conditions.
The CxA is responsible for generating a factory witness test script that
takes the simulated system through all operational scenarios the same way a
field test script would. This process is necessarily complicated by the fact that
the CxA must define the value of every hardwired analog input during each test
step to ensure that the plant conditions being simulated represent the desired
real-world conditions.
The entire factory witness test process adds considerable overhead to Cx
for both the CxA and the controls contractor that must be accounted for
when budgeting. The CxA must complete another round of testing with a
second script that usually takes longer to assemble than the actual field
script, and the controls contractor must assemble a laboratory test bench and
participate in additional testing. The benefit of all this additional work is
that the control sequence is much more likely to work the first time in the
field because the majority of critical flaws should be identified during fac-
tory testing. This has the dual benefits of minimizing the potential for real-
world downtime resulting from flaws in the controls logic and reducing the
amount of time spent testing and retesting conditions that may only present
themselves occasionally in the field. Simulator factory witness testing is
particularly invaluable for testing sequences that will be implemented in
already live mission-critical facilities, such as data centers and laboratories,
that cannot have any downtime. It generally does not make sense on less
critical facilities such as commercial buildings.
Trend Review
Trend review is an invaluable final step in the central plant Cx process
because it
validates operating conditions that were not observed during functional

testing;
validates logic that only can truly be observed when buildings are operating
under real load conditions, such as optimum start and many set point reset
algorithms;
allows proper tuning and stability of control loops and reset blocks to be
verified; and
reveals control logic flaws due to conditions that were not anticipated when
SOOs were created.
Trend Review Preparation

Before a trend review can be properly executed, the groundwork for that
review needs to be established. In particular, the CxA should provide the con-
trols contractor with a list of desired trend review points during implementa-
tion of the sequence. This ensures that in addition to hardwired points, which
are captured on the point-to-point forms discussed previously, software values
of interest such as set points and trim and respond requests are preconfigured
for trending. (It is best if trending requirements are established in the bid spec-
ifications to avoid BAS contractor change orders; with some BASs, the time to
set up trends can be significant.)
If the EORs control specification does not clearly lay out a point-naming
convention to be used by the control contractor, the CxA should work with the
controls contractor to develop a point-naming standard to ease the identifica-
tion of trend points when a trend database export is provided for trend review.
In addition to helping with the near-term trend review, this step also helps
make trend data more usable for engaged operators, contractors, and engineers
that may complete future work in the building. An effective naming convention
includes the following information:
Building name tag (if more than one apply)

System type (e.g., CHW)
Equipment or system component identifier (e.g., CH-1 or Primary loop)
Property (e.g., CHWST)
The above point might be tagged as 1.CHW.CH1.CHWST, for example.

With all points clearly named, and all points preconfigured upfront, the
overhead involved with later adding missing points and deciphering and clean-
ing up poorly named trends can be avoided.
Trend Review Execution

The first goal of the trend review process should be to assess whether the
plant is achieving its basic intent of supplying chilled water at the temperature
it is commanded to, when it should, and with limited fluctuation in control to
set point. The second goal should be verifying the proper implementation of all
control sequences in much the same way that was done during functional test-
ing, albeit with an added focus on control loop stability and tuning. To ensure
that no sequences go without review, an effective strategy is to step through the
written sequence line by line and identify the type of graph (time series or scat-
ter plot) and points that need to be included therein to validate each sequence.
After the review is completed, the graphical results along with detailed expla-
nations of identified issues and proposed resolutions should be provided to the
owner in the form of a detailed report for review. The trend review graphs serve
as the evidence that is then brought to bear when discussing proposed resolu-
tions with the owner, EOR, and controls contracting team.
The two most common types of graphs used to observe trends are time-
series plots and scatter plots.
Time-series plots are often the easiest because almost all BASs have the
capability built into their interface software. Several points can be plotted
together, such as set point, controlled point, and loop output to the controlled
device. Figure 8-1 is a time-series plot that shows pump speed as it modulates
to control DP set point. There is the typical erratic control on start-up, but the
loop eventually settles.
Figure 8-1 Time-series plot.
Figure 8-2 Scatter plot.
The ability to create scatter plots is not always possible using BAS inter-
face software. If not, the data need to be downloaded and plotted using a
spreadsheet. Scatter plots are also useful in evaluating loop tuning by plotting
the set point versus the controlled point. The desired result is a tight fit along a
45 line. Figure 8-2 is a scatter plot that shows a cooling tower fan system that
is not well tuned.
References
ASHRAE. 2007. Guideline 1.1-2007, HVAC&R technical requirements for the
commissioning process. Atlanta: ASHRAE.

8-1 Which aspect of a central plant is most prone to issues during start-up and
commissioning?
a. Cooling towers
b. Chillers
c. Control system hardware
d. Control system configuration and sequence implementation
8-2 You are reviewing a preliminary control sequence for a plant with two equally
sized variable-speed centrifugal chillers and two cooling towers with the
clauses below. What types of issues do you envision with this sequence?
Cooling Towers: Control cooling towers to maintain a fixed condenser
water supply temperature equal to the design temperature of 85F. Stage tow-
ers optimally to minimize energy use.
Chillers: Stage the lag chiller ON when plant load exceeds 30% of design
plant capacity. Stage the lag chiller OFF when the plant load drops below 30%
of design plant capacity.
a. The sequence predisposes the chillers to surge.
b. The sequence predisposes the chillers to cycling.
c. The sequence for cooling tower staging is too vague to be implemented
in practice.
8-3 In which of the following scenarios is factory witness testing with a simulator
appropriate?
i. Installation of a new plant in a new ten-story commercial office
building.
ii. Updating the sequences of an existing plant serving a data center.
iii. Retrofit of an existing plant serving a college campus.
iv. Installation of a new plant in a new laboratory facility.
a. (i), (ii), (iii), (iv)
b. (ii), (iii), (iv)
c. (ii), (iii)
d. (ii), (iv)
8-4 Which of the following are key benefits of functional testing?
a. Test scripts serve as a formal interpretation of the written SOOs.
b. Test script development provides an opportunity to identify unclear
sequences that are likely to impact the controls contractor as well.
c. Functional testing verifies implementation of the sequence consistent
with the design intent.
d. Functional testing decreases the probability that issues with the plant
occur following commissioning due to controls programming and
point-to-point configuration issues.
8-5 Which of the following plant control issues are more likely to be identified in
trend review than in functional testing?
a. Unstable control loops
b. Improperly implemented staging logic
c. Improperly configured set point resets
d. Proper response to equipment failure
Skill Development
Exercises
To receive full continuing education credit, all questions must be answered and
submitted at www.ashrae.org/sdlonline. Please log in using your student ID
number and the SDL number. Your student ID number is composed of the last
five digits of your Social Security Number or another unique five-digit number
you create when first registering online. The SDL number for this course can
be located near the top of the copyright page of this book.
Fundamentals of Design and Control of Central Chilled-Water Plants I-P
Chapter 2 Skill Development Exercises

Total number of questions: 6
2-1 Why is the shape of a CHW plants cooling load profile a critical factor in plant
design?
a. It dictates the conditions under which the plant must operate efficiently
to minimize energy costs.
b. It impacts the selection of chillers because the plant must be able to
handle the full range of expected load conditions stably.
c. It drives the peak capacity required of the plant.
2-2 Which of the following are true regarding the impact of air-side economizing
on the annual load profile of a plant serving an office building?
i. It reduces the total annual kWh-hours served by the plant.
ii. It shifts the most common load percentage to a lower value.
iii. It reduces the peak load of the plant.
iv. It reduces the plants run hours.
a. (i), (ii), (iv)
b. (i), (ii)
c. (i), (iii), (iv)
d. (i), (ii), (iii), (iv)
2-3 ASHRAE HandbookFundamentals supports which of the following load cal-
culations methodologies?
a. RTS
b. HBM
c. Transfer function method
d. Only (a) and (b)
e. All of the above
2-4 Oversizing CHW plants:
a. Typically yields more efficient pumping in variable-speed applications
due to lower friction losses.
b. Usually leads to more efficient chiller operation.
c. May cause controllability issues if chillers are not properly selected for
stable low-load operation.
d. Is problematic when the condenser and CHW pumps are variable
speed.
e. Both (a) and (c).
2-5 You are replacing oversized chillers in an existing CHW plant with modern
direct digital control (DDC) controls, trending capabilities, and recently cali-
brated instrumentation. Which of the following is the recommended approach
for determining peak load to size the new chillers?
a. Develop a load model of the facility using a simulation tool and utilize
the peak load estimated therefrom.
b. Develop a load model of the facility using a simulation tool and cali-
brate the model on an annual basis using utility billing data, then assess
peak load with the model.
c. Utilize the DDC systems primary CHW loop flowmeter and supply-
and return-temperature sensors to trend load. Use a few months of
trended load and local weather data from the summer and/or swing sea-
sons to develop a load profile and predict peak load therefrom.
d. Install temporary National Institute of Standards and Technology
(NIST)-calibrated instrumentation, including an ultrasonic flowmeter
and supply- and return-temperature sensors to trend load. Use the same
approach as option (c) to predict peak load.
2-6 True or False: In early design, developing a prototypical model of the proposed
building is usually too cost prohibitive to assist in plant design development.
a. True
b. False

3-1 In the vapor compression refrigeration cycle, refrigerant flows from the com-
pressor discharge, then
a. through the condenser, evaporator, expansion valve, and back to the
compressor.
b. through the condenser, expansion valve, evaporator, and back to the
compressor.
c. through the expansion valve, condenser, evaporator, and back to the
compressor.
d. through the evaporator, expansion valve, condenser, and back to the
compressor.
3-2 Which of the following are valid reasons for a building owner to prefer select-
ing a chiller utilizing R-134a instead of R-123 refrigerant?
i. ODP
ii. Purge unit requirements
iii. GWP
iv. Future refrigerant availability
a. (i), (ii)
b. (i), (ii), (iv)
c. (i), (ii), (iii), (iv)
d. (iii), (iv)
3-3 You are selecting a 300 ton water-cooled chiller. The available compressor
types in this size range include which of the following?
a. Scroll, screw, and centrifugal
b. Scroll and screw
c. Screw and centrifugal
d. Centrifugal
3-4 Manufacturers typically limit evaporator flow rates on the low end to avoid
_________ and on the high end to avoid _________.
a. Surge; noise
b. Deposit formation; erosion
c. Laminar flow; erosion
d. Deposit formation; noise
3-5 Benefits of gear-driven centrifugal chillers relative to direct-drive centrifugal
chillers include:
a. Smaller physical footprint
b. The more common incorporation of multiple stages, and thus refriger-
ant economizers
c. Greater flexibility in optimizing compressor efficiency
d. All of the above

e. (b) and (c)
f. (a) and (c)

3-6 You are designing a 300 ton primary-only CHW plant with a 20F CHW -T.
The pump type with the best combination of efficiency and cost for this appli-
cation is most likely to be
a. Close coupled, end suction
b. Base mounted, double suction
c. Base mounted, end suction
d. Close coupled, in-line
3-7 A customer expresses concern that the condenser water pumps in a plant
located in San Diego, CA make a gurgling noise. The pumps are installed 3 ft
below the basin water level. The strainers are located downstream of the pumps
and the towers have vortex shedders. What is the most likely cause of the
issue?
a. Cavitation
b. Dissolved air coming out of solution
c. Vortices forming at the tower suction
d. None of the above
3-8 You are designing a 700 ton central plant for a 20-story high-rise building. The
cooling towers are to be located in a well on the roof where the primary con-
straint is roof area. Acoustics are not a concern. The type of tower best suited
to this application is
a. Field erected, induced draft, counterflow
b. Packaged, induced draft, cross flow
c. Packaged, induced draft, counterflow
d. Packaged, forced draft, counterflow
3-9 A cooling tower has been selected for a 15F range with a 6F approach when
the ambient wet-bulb temperature is 72F. At design flow rate and 15F range,
but with an ambient wetbulb temperature of 62F, the tower is capable of pro-
viding
a. A better approach
b. A worse approach
c. The same approach
3-10 Pump and tower VFD minimum speed set points of 6 Hz or less
a. Result in significant energy savings relative to a minimum speed set
point of 12 Hz.
b. Are not feasible, because they cause motors to overheat.
c. Improve controllability under low-load situations.
d. None of the above.

4-1 Aggressive CHW temperature reset
a. Causes issues with thermal comfort because higher supply water tem-
perature yields significantly higher supply air moisture content for a
given set of entering air conditions and thus higher space humidity.
b. Typically increases pump energy usage significantly enough to offset
the benefit of the decrease in chiller energy use.
c. Has little to no impact on space humidity control.
4-2 A plant consists of two identical fixed-speed centrifugal chillers, each with a
dedicated constant-speed primary CHW pump. The chillers supply chilled
water to one large built-up air handler that primarily serves daytime commer-
cial office space loads and another large air handler that serves an auditorium
space most frequently occupied in the evening. Both air handlers have three-
way CHW control valves and are of approximately equal size. Which of the
following are true?
i. The design will require two chillers to operate when the auditorium
air handler is at full load, even if the office air handler is off.
ii. The plant will operate least efficiently in the rare instances that both
the office air handler and auditorium air handler are at near design
load.
iii. The plant will operate most efficiently in the rare instances that
both the office air handler and auditorium air handler are at near
design load.
iv. Headering the primary pumps would increase the controls complex-
ity with no benefit in system redundancy.
a. (i), (iii), (iv)
b. (i), (iii)
c. (ii), (iv)
d. (i), (ii)
4-3 For three-way valve systems
a. The flow rate through the branch serving the coil is constant, irrespec-
tive of valve position.
b. The flow rate through the branch serving the coil peaks when the valve
is fully open to the coil.
c. The flow rate through the branch serving the coil peaks when the valve
is 50% open.
d. Balancing of the bypass leg is never necessary.
4-4 The back-loaded position of the common leg shown below

a. Causes one chiller to be almost fully loaded with the remainder of the
load handled by the other chiller.

b. Causes unbalanced flow through the two chillers.
c. Results in the same energy performance as a system with a common leg
located in the normal position just upstream of the secondary pumps.
d. Is a reasonable location for most plants if it is less expensive due to the
physical layout of the plant.
4-5 Construction of a variable primary CHW plant is just about complete when it is
discovered that the design includes only two-way valves and no means to
maintain minimum flow. Which of the following last-minute design change
options will resolve the problem at minimum cost?
a. Install a CHW bypass locally at the CHW plant. Measure DP across the
chillers to indirectly measure flow.
b. Install a CHW bypass locally at the CHW plant. Install a flowmeter in
the main return line at the plant to measure flow.
c. Install a CHW bypass at the end of the line. Install a flowmeter in the
main return line at the plant to measure flow.
d. Install enough three-way valves at end-of-line coils to maintain mini-
mum flow.
4-6 True or false?
Both headered and dedicated pump per chiller configurations are equally
appropriate for plants requiring a standby pump.
a. True
b. False
4-7 Which balancing method is most appropriate for all but very large distribution
systems?

a. Manual balancing using CBVs to measure and adjust flow.
b. No balancing.
c. AFLVs at all coils.
d. Reverse-return piping.
4-8 True or False: Air-side economizing systems can contribute to low CHW T
issues in systems with high T designs.
a. True
b. False
4-9 What factors constrain the number of cooling towers that can be operated with
a given number of constant-speed condenser water pumps enabled?
a. Minimum per-tower flow requirements.
b. Maximum per-tower flow requirements.
c. Neither (a) nor (b).
4-10 Isolating cooling towers by means of isolation valves on the tower supply pip-
ing only
a. Requires that the equalizer be oversized.
b. Does not require the equalizer to be oversized.
c. Requires that all towers be operated whenever the plant is enabled. The
isolation valves are only installed to prevent tower overflow upon plant
shutdown.
d. Is not a viable design option.
4-11 True or false?
Most modern centrifugal chillers can operate with an integrated WSE with-
out any means of head pressure control.
a. True
b. False

5-1 You are responsible for developing the master plan for a new college campus
that will include multiple buildings. Which of the following CHW distribution
approaches will be the most energy efficient?
a. Variable primary flow
b. Primary, variable secondary
c. Primary, variable secondary with tertiary bridge connected pumps at
the buildings
d. Primary, distributed secondary where the secondary pumps are sized
for the pressure drop from and back to the primary loop
5-2 Your customer is building a sprawling six-story, 600,000 ft2 building with three
wings, each of which is served by a large AHU. The building also has multiple
computer rooms served by FCUs. All of the high density rooms are localized in
the same area of the building. What is the likely to be the most energy-efficient
CHW distribution design approach?
a. Variable primary with a plant bypass leg and two-way valves at all
AHU and FCU coils
b. Constant primary, variable secondary with two-way valves at all AHU
and FCU coils
c. Variable primary with distributed variable secondary coil pumps for all
AHUs and a separate variable secondary pump serving all FCUs with
two-way valves
d. Variable primary with distributed variable secondary pumps at all
AHUs and all FCUs
5-3 When sizing piping to minimize life-cycle costs, which of the following are
critical considerations?
a. Expected system operating hours
b. Variable speed versus constant-speed operation
c. Utility rates
d. All of the above
5-4 Which of the following are benefits resulting from selecting condenser water
loop T for a larger range?
i.Reduced chiller energy use due to lower chiller lift
ii.Reduced pumping energy use
iii.Increased cooling tower capacity for a given tower selection
iv.Reduced cooling tower scaling potential
a. (ii.) and (iii.)
b. (i.), (ii.) and (iii.)
c. (ii.), (iii.), and (iv.)
d. (i.), (ii.), and (iv.)
5-5 Which of the following are true when selecting cooling tower fan controls?
a. Constant-speed towers often prove life-cycle cost-effective because of
their low upfront costs.

b. VFD fan speed controls are prohibitively expensive relative to the small
energy benefit gained relative to two-speed motor controls.
c. To achieve the energy benefits provided by variable-speed tower con-
trol, it is only necessary to install a VFD on the lead cooling tower.
d. VFDs on all tower cells generally result in lowest life-cycle costs.
5-6 You are retrofitting an existing central plant serving a large office building with
a WSE in a climate where freeze conditions are not a concern. The plant cur-
rently maintains minimum chiller head pressure in cool dry weather using a
cooling tower bypass. How does this arrangement need to be modified to
accommodate the WSE?
a. No changes are needed.
b. A separate cooling tower without condenser water bypass needs to be
added to accommodate the WSE HX.
c. Modulating actuators should be added to the condenser isolation valves
on each chiller and the cooling tower bypass should be permanently
shut.
d. The WSE HX should be piped in series with the cooling towers and the
cooling tower bypass should be shut permanently.

6-1 The benefits of a performance-based chiller bid approach include which of the
following:
a. It can minimize life-cycle costs for the owner.
b. It solidifies selections in the design phase, allowing the designer to
accurately lay out the plant and avoid design rework by the contractor
during the construction phase.
c. The chiller selection process becomes based on objective criteria rather
than subjective criteria that contractors may use when including chillers
in their project bids.
6-2 Which of the following parameters is not appropriate to include in a performance-
based bid approach:
a. The anticipated cooling load profile for the plant.
b. Total required plant capacity.
c. Chiller efficiency metrics.
d. Sound power limits.
6-3 The following three selections are provided by a chiller manufacturer.
Chiller 1 Chiller 2 Chiller 3
200 ton centrifugal 200 ton screw 200 ton centrifugal
Two-pass evaporator Two-pass evaporator Three-pass evaporator
Open-drive compressor Hermetic compressor Hermetic compressor
CHW pressure drop: 5 ft CHW pressure drop: 10 ft CHW pressure drop: 15 ft
Full-load kW/ton: 0.5 Full-load kW/ton: 0.6 Full-load kW/ton: 0.55
Native BACNet Native BACNet BACNet gateway required
Variable speed Variable speed Variable speed
Which option is likely to have the highest first cost, not including the cost
of the chiller itself?
a. Chiller 1
b. Chiller 2
c. Chiller 3
d. Indeterminate based on the provided data
6-4 When estimating annual plant utility costs with simulation modeling
a. Using a flat effective annual rate that accounts for both electricity and
demand is usually sufficiently accurate.
b. Annual escalation of utility costs should be ignored because of the
unknown future variation of utility prices.
c. Using real rate schedules that account for both electricity and demand
independently should be used.
d. Annual escalation of utility costs should be estimated based on fore-

casts provided by DOE/EIA.
6-5 Using zero tolerance chiller performance data in bid forms
a. Greatly increases the time needed to complete the forms by vendors.
b. Better ensures the best chillers are selected, particularly for plants
expected to operate many hours at low load.
c. Is not recommended when using the simplified procurement procedure.

7-1 Which of the following statements regarding commercial versus industrial con-
trols is not true:
a. Commercial controls are not suitable for most CHW plant applications
due to their inferior reliability.
b. Industrial controls should be used for mission-critical applications
requiring parallel operation of redundant controllers.
c. Commercial control systems have lower first costs.
d. Commercial control systems cost less to maintain.
7-2 You are in the process of specifying control sensors for a CHW plant used to
serve both comfort cooling and mission critical loads. The plant has variable-
speed primary CHW pumps. Which of the follow options is most appropriate
for monitoring CHW pump status?
a. Hard wire status from the VFD.
b. Read status over the network from the VFD.
c. Install a current transducer on one phase on the line side of the VFD.
d. Install a DP switch across the pump.
7-3 Which of the following statements regarding calibration of water temperature
sensors is true?
i. Field calibration is typically not necessary for noncritical sensors.
ii. Field calibration reference temperature sensors should be at
least twice as accurate as the field-installed sensors.
iii. Dry well baths are always required to field calibrate the tem-
perature sensors used for CHW plant load measurement.
iv. RTDs are more stable than thermistors and therefore require
less frequent repeat calibration.
a. (i), (ii)
b. (i), (ii), (iii), (iv)
c. (i), (ii), (iv)
d. (ii), (iv)
7-4 You are specifying two-way control valves for air-handler CHW coils served
by a primary, variable secondary plant. The valve sizes vary between 1.5 and
4 in. What valve type(s) are most appropriate for this application?
a. Globe valves for all sizes.
b. Ball valves for 3 in. and less; globe valves for all greater than 3 in.
c. Ball valves for all sizes.
d. Ball valves for 3 in. and less; butterfly valves for all greater than 3 in.
7-5 You are designing a variable-speed primary loop that controls pump speed
based on DP measured by a DP sensor installed far out in the loop, remote
from the mechanical room where the pumps are to be installed. A colleague
recommends specifying that the DP sensor be wired to a controller located near
the sensor, then passing the sensor reading to the pump controller over the net-
work. This approach
a. Can save on wiring costs relative to hardwiring the DP feedback to the
pump controller.
b. May result in control loop instability due to network latency or limited
feedback polling frequency.
c. Increases network traffic and may slow control system performance.
d. Will lead to loss of pump feedback control if network communications
between controllers are lost.
7-6 Monitoring CHW plant efficiency using modern DDC control systems
a. Typically requires installation of expensive true-RMS power meters.
b. Typically requires installation of CHW loop instrumentation beyond
that necessary for CHW plant control.
c. Requires additional programming on the part of the DDC controls con-
tractor.
7-7 Plant optimization modeling indicates that the most efficient approach to con-
trol cooling tower fans serving a plant with typical office loads, short of using
real-time optimization, is
a. Controlling fans to maintain a condenser water supply temperature set
point fixed to the minimum allowed by the chiller minimum lift
requirement.
b. Controlling fans to maintain a condenser water supply temperature set
point fixed at the design temperature to minimize tower fan energy.
c. Controlling fans to maintain a condenser water supply temperature
reset as a function of ambient wet-bulb temperature.
d. Controlling fans to maintain a condenser water return temperature set
point reset based on plant load and CHW supply temperature.
7-8 Variable-speed control of condenser water pumps
a. Can easily increase plant energy use if not controlled optimally.
b. May be cost-effective in high load applications, such as data centers
that require operation under a variety of load and ambient wet-bulb
conditions, but modeling or real-time optimization is needed to deter-
mine the optimal sequence.
c. Is highly climate dependent: the optimal strategy in one climate is
unlikely to be optimal in another.
7-9 The optimal approach to resetting CHW supply temperature set point and
CHW DP set point for a plant with variable-speed chillers is:

a. Dependent on climate.
b. Dependent on plant load profile.
c. Resetting CHW supply temperature set point first, then CHW DP set
point.
d. Resetting CHW DP set point first, then resetting CHW supply tempera-
ture set point.

8-1 Which aspect of a central plant is most prone to issues during start-up and
commissioning?
a. Cooling towers
b. Chillers
c. Control system hardware
d. Control system configuration and sequence implementation
8-2 You are reviewing a preliminary control sequence for a plant with two equally
sized variable-speed centrifugal chillers and two cooling towers with the
clauses below. What types of issues do you envision with this sequence?
Cooling Towers: Control cooling towers to maintain a fixed condenser
water supply temperature equal to the design temperature of 85F. Stage tow-
ers optimally to minimize energy use.
Chillers: Stage the lag chiller ON when plant load exceeds 30% of design
plant capacity. Stage the lag chiller OFF when the plant load drops below 30%
of design plant capacity.
a. The sequence predisposes the chillers to surge.
b. The sequence predisposes the chillers to cycling.
c. The sequence for cooling tower staging is too vague to be implemented
in practice.
8-3 In which of the following scenarios is factory witness testing with a simulator
appropriate?
i. Installation of a new plant in a new ten-story commercial office
building.
ii. Updating the sequences of an existing plant serving a data center.
iii. Retrofit of an existing plant serving a college campus.
iv. Installation of a new plant in a new laboratory facility.
a. (i), (ii), (iii), (iv)
b. (ii), (iii), (iv)
c. (ii), (iii)
d. (ii), (iv)
8-4 Which of the following are key benefits of functional testing?
a. Test scripts serve as a formal interpretation of the written SOOs.
b. Test script development provides an opportunity to identify unclear
sequences that are likely to impact the controls contractor as well.
c. Functional testing verifies implementation of the sequence consistent
with the design intent.
d. Functional testing decreases the probability that issues with the plant
occur following commissioning due to controls programming and
point-to-point configuration issues.

8-5 Which of the following plant control issues are more likely to be identified in
trend review than in functional testing?
a. Unstable control loops
b. Improperly implemented staging logic
c. Improperly configured set point resets
d. Proper response to equipment failure
ASHRAE LEARNING INSTITUTE
Self-Directed Learning Course Evaluation Form
Course Title: Fundamentals of Design and Control of Central Chilled-Water Plants (I-P) (2017)
On a scale of 1 to 5, circle the number that corresponds to your feeling about the statements below.
(1 = strongly agree, 5 = strongly disagree, 3 = undecided)
Strongly Strongly
Course Content Agree Undecided Disagree
1. The objectives of the course were clearly stated. 1 2 3 4 5

2. The course content supported the stated objectives. 1 2 3 4 5
3. The content quality and format of the course material make it valuable as a 1 2 3 4 5
future reference.
4. The quality and clarity of the charts and diagrams enhanced your ability 1 2 3 4 5
to understand the course concepts.
5. The organization of course material supported effective mastery of the topic. 1 2 3 4 5
6. The material presented will be of practical use to you in your work. 1 2 3 4 5
7. The degree of difficulty (level) of this course was correct to meet your 1 2 3 4 5
needs and expectations.
General
1. Which description best characterizes your primary job function?
_____Architect* _____Developer _____Manufacturer _____Sales

_____Code Agency _____Educator/Research _____Marketing _____Specifier
_____Consultant _____Energy Conservation _____Owner _____Student
_____Contractor/Installer _____Facilities Engineer _____Plant Engineer _____Utilities
_____Consumer/User _____Government _____Policy Maker/Regulator
_____Other (please specify) _______________________________________________________________________
*Are you a registered architect? ___No ___Yes, AIA Membership Number (required): _____________________
2. Which describes your educational background?
_____High School _____Master's DegreeEngineering

_____Associates Degree/Certificate Program _____Master's DegreeOther Than Engineering
_____Bachelor's DegreeEngineering Technology _____Doctoral DegreeEngineering
_____Bachelor's DegreeEngineering _____Doctoral DegreeOther Than Engineering
_____Bachelor's DegreeOther Than Engineering
_____Other (please specify) _______________________________________________________________________
3. Approximately how many hours did it take you to complete this course?
_____10 hours _____20 hours _____30 hours _____40 hours _____Other (please specify)___________
4. What topics would you suggest for future courses? ______________________________________________________________
_______________________________________________________________________________________________________
General Comments regarding any aspect of the course, including suggestions for improvement:
_________________________________________________________________________________________________________
_________________________________________________________________________________________________________
_________________________________________________________________________________________________________
_________________________________________________________________________________________________________
_________________________________________________________________________________________________________
_________________________________________________________________________________________________________
_________________________________________________________________________________________________________
_________________________________________________________________________________________________________
_________________________________________________________________________________________________________
_________________________________________________________________________________________________________
_________________________________________________________________________________________________________
_________________________________________________________________________________________________________
_________________________________________________________________________________________________________
_________________________________________________________________________________________________________
_________________________________________________________________________________________________________
_________________________________________________________________________________________________________
_________________________________________________________________________________________________________
_________________________________________________________________________________________________________
_________________________________________________________________________________________________________
_________________________________________________________________________________________________________
_________________________________________________________________________________________________________
Name (optional) __________________________________________________________________________________________
Phone (optional) __________________________________________________________________________________________
E-mail (optional) __________________________________________________________________________________________
Return to: ASHRAE, Education Department, 1791 Tullie Circle NE, Atlanta, GA 30329
Email: edu@ashrae.org Fax: 404-321-5478
Flexible and Effective Continuing Education
for HVAC&R Professionals
ASHRAEs Fundamentals of Design and Control of Central Chilled-Water

Plants, I-P Edition, self-directed learning course book focuses on
optimizing the design and control of chilled-water plants to minimize life-
cycle costs. This work is an invaluable tool for HVAC designers of various
backgrounds and an introduction for those new to chilled-water plants.
Plant operators, energy engineers, and control system designers will also
find information on loads, equipment, distribution, chiller procurement,
controls, and commissioning. Supplemental online files are included to
promote understanding of real-world scenarios.
Skill Development Exercises at the end of each chapter help readers
assess their understanding of the material and apply what they learn to
real-world situations. Answers to these exercises can be submitted online
to earn PDH or LU credits.
1791 Tullie Circle

Atlanta, GA 30329-2305
Telephone: 404/636-8400
Fax: 404/321-5478
E-mail: edu@ashrae.org
www.ashrae.org/ali
ISBN 978-1-939200-66-2 (paperback)

ISBN 978-1-939200-67-9 (PDF)
9 781939 200662
Product Code: 98014 9/17
CHW Plants_covers_I_P.indd 2 9/14/2017 10:43:25 AM

Fundamental Desgin Control Central Chilled Water Plant Ip

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Fundamental Desgin Control Central Chilled Water Plant Ip

Uploaded by

Copyright:

Available Formats

Fundamentals of

Design and Control of

Steven T. Taylor, PE I-P

Includes Skill Development Exercises

CHW Plants_covers_I_P.indd 1 9/14/2017 10:43:25 AM

A Course Book for Self-Directed or Group Learning

For course information or to order additional materials, please contact:

ASHRAE Learning Institute Telephone: 404/636-8400

Updates and errata for this publication will be posted on the

Karen M. Murray kmurray@ashrae.org

We hope your educational experience is satisfying and successful.

HVAC Design Essentials and Applications TrainingInstructor Led at Approved Locations

Online CoursesInstructor Led on the Web

ASHRAE Chapter and In-Company TrainingInstructor Led at Your Location

eLearningWeb-Based Instruction on Demand

Self-Directed Learning TextsSelf Study or Texts for Group Instruction

ASHRAE Learning Institute www.ashrae.org/education

Selecting Cooling Towers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .167

Chilled-water plants are typically the most costly part of large

Others who provided review and input to this course include:

Brandon Gill, Taylor Engineering

Mick Schwedler, Trane

Steve Duda, Ross & Baruzzini

Bryson Borzini, P2S Engineering

Tony Mueller, P2S Engineering

Anna Zhou, Taylor Engineering

A/D = analog to digital

GWP = global warming potential

How to Use This Course

Understanding Loads and Their Impact on Design

The climate in which the plant is located

Figure 2-1 Cooling load profiles, five-day office in San Francisco.

Peak Loads Overview

Weather conditions can vary over a period of time as a result of increasing

Building envelope elements do not always perform as expected due to

Annual Load Profiles Overview

Figure 2-3 Cooling energy usage for four design alternatives.

Determining Peak Loads

1. Survey of monitoring sites: Conduct a complete audit of the CHW plant.

weather conditions, then it will be necessary to extrapolate the measured

Determining Hourly Load Profiles

Computer simulation models (customized and prototypical)

For new construction, custom computer simulations and prototypical simu-

Computer Simulation Models

Skill Development Exercises for Chapter 2

The basic vapor compression refrigeration cycle

Vapor Compression Refrigeration Cycle

Starting at Point A, the refrigerant is a liquid at high pressure. As it passes

Figure 3-1 The refrigeration cycle.

Figure 3-2 Pressure-enthalpy chart.

From Point B to Point D, the liquid is converted to a gas by absorbing heat

The difference between the condensing temperature and evaporating tem-

The relative ability of a refrigerant to destroy stratospheric ozone is called

Table 3-1 Montreal Protocol

Table 3-2 ODP Versus GWP for Common Refrigerants

Water Chiller Components

Figure 3-3 Scroll compressor.

Figure 3-4 Single screw compressor.

Figure 3-5 Twin screw compressor.

Figure 3-6 Hermetic centrifugal compressor.

One of the characteristics of the centrifugal compressor is that it can surge.

Figure 3-7 Inlet guide vanes.

weather. The evaporator temperature can, likewise, be raised in mild weather

When the impeller is at the minimum speed, further reductions in capacity