Professional Documents
Culture Documents
Table of Contents
1.0 INTRODUCTION 3
1.1 Purpose 3
1.2 Scope . 3
1.3 Intended Users .. 4
1.4 References and Related Documents 4
2.0 GENERAL 5
2.1 Definitions .. 5
2.2 Principles and Concepts 5
2.3 Degrees of Freedom . 5
2.4 Process Variability 6
2.5 HX Design and Simulation .. 9
2.6 Fouling Monitoring & Mitigation ...................... 14
3.0 PROCESS HEARING TRAINS . 19
3.1 Single Process Stream Heated in Series .. 19
3.2 Single Process Stream Heated in Network of Parallel HX Trains .. 21
3.3 Avoid Un-Necessary Heating 23
4.0 PROCESS COOLING TRAINS . 25
4.1 Single Process Stream Cooled in Series 25
4.2 Single Process Stream Cooled in Network of Parallel HX Trains 26
4.3 Multiple process Streams Cooled with a Single Cooling Utility .. 29
4.4 Avoid Un-Necessary Cooling 30
4.5 Load Shedding versus Process Modifications .. 31
5.0 HEAT EXCHANGER NETWORKS .. 33
5.1 Simple HENs 33
5.2 Complex HENs 37
5.3 HEN Operability Considerations and Constraints .. 42
6.0 BOILER NETWORKS 46
6.1 Boiler Sparing Philosophy for Optimum Reliability . 46
6.2 Load Allocation among Multiple Parallel Boilers . 49
6.3 Load Management of Boiler Auxiliaries .. 53
6.4 Steam Balance Optimization . 58
7.0 Furnace Networks 63
7.1 Heater Sparing Philosophy . 64
7.2 Load Allocation among Multiple Parallel Heaters .. 65
7.3 Load Allocation in Hot Oil Loops 65
7.4 Load Management of Furnace Auxiliaries . 69
7.5 WHB Opportunities 73
Page 3 of 84
Document Responsibility: Energy Systems Optimization Standards Committee SABP-A-008
Issue Date: 28 February 2016
Next Planned Update: TBD Load Management for Energy Efficiency: Heat Transfer Equipment
1.0 Introduction
1.1 Purpose
Large industrial plants commonly use multiple parallel equipment trains for improved reliability. Very often,
installed equipment capacity far exceeds normal production requirements. This excess capacity can often
be translated into energy cost savings. Also, the heating and cooling objectives for the process can be
met in a variety of alternative ways, using different types and amounts of utilities. There is a certain utilities
mix that results in the lowest overall operating costs. The purpose of this new Best Practice is to describe
ways in which energy cost can be minimized through optimum allocation of load among existing heat transfer
equipment.
1.2 Scope
Many types of equipment commonly used in Saudi Aramco plants are significant energy consumers
and amenable to operational optimization through Load Management, including:
Steam heaters
Fired Heaters (furnaces)
Process Coolers air, water, refrigerant
Complex Heat Exchanger Networks
Boilers fired and unfired
Pumps and compressors
Steam and Gas Turbines
This Best Practice manual provides guidelines on methods to determine the optimum load management
policies for heat transfer equipment only. Other types of energy consuming and converting equipment are
covered in complementary Best Practice manuals.
Although some of the topics covered in this manual may overlap with those found in the foregoing
texts, they have been included only to provide background, and to make the reader aware that these subjects
have a bearing on the determination of operational optimization via Load Management. For detailed
information on these subjects, the reader should consult the reference books and related documents cited in
section 1.4.
Page 5 of 84
Document Responsibility: Energy Systems Optimization Standards Committee SABP-A-008
Issue Date: 28 February 2016
Next Planned Update: TBD Load Management for Energy Efficiency: Heat Transfer Equipment
SABP-A-007: Steam Trap Management for Energy Efficiency, Saudi Aramco Engineering
Standard (2006)
rd
Handbook of Heat Transfer, 3 ed, W. M. Rohsenow, J. P. Hartnett, and Y. I. Cho, McGraw- Hill
Inc, New York (1998), Ch 18.
Fundamentals of Heat Exchanger Design, R. P. Shah and D. P. Sekulic, John Wiley & Sons Inc.,
Hoboken, NJ (2003), Chs 3 and 11.
th
Heat Transfer, 8 ed., J. P. Holman, McGraw-Hill Inc, New York (1997), Ch 10.
Heat Exchanger Network Synthesis, U. V. Shenoy, Gulf Publishing Co, Houston, TX (1995)
Thermal Design of Heat Exchangers, E. M. Smith, John Wiley & Sons Ltd, Chichester, UK (1997),
Ch 2.
th
Steam Plant Operation, 7 ed, E. B. Woodruff, H. B. Lammers, and T. F. Lammers, McGraw- Hill
Inc., New York (1998).
The Control of Boilers, S. G. Dukelow, ISA Press, Research Triangle Park, NC (1986).
The John Zink Combustion Handbook, .C. E. Baukal, ed., CRC Press, Boca Raton, FL (2001),
Chs 2, 16.
No conflict is expected between the optimum load management policy and other Saudi Aramco standards with
respect to reliability, safety, etc. If any such conflict should arise, the standard shall take precedence.
Page 6 of 84
Document Responsibility: Energy Systems Optimization Standards Committee SABP-A-008
Issue Date: 28 February 2016
Next Planned Update: TBD Load Management for Energy Efficiency: Heat Transfer Equipment
2.0 General
2.1 Definitions
Best Practice: A process or method that, when correctly executed, leads to enhanced system performance.
Load Management: An operating policy that distributes the load among multiple machines or equipment
installed as series-parallel networks in a way that minimizes their energy (fuel + power) consumption,
without compromising safety or reliability.
The analytical procedures outlined in this manual will help establish the quantitative relationship between
operating flexibility and energy costs, thereby enabling plant engineers and foremen to jointly make
intelligent decisions about the optimum operating policy.
Page 7 of 84
Document Responsibility: Energy Systems Optimization Standards Committee SABP-A-008
Issue Date: 28 February 2016
Next Planned Update: TBD Load Management for Energy Efficiency: Heat Transfer Equipment
additional costs elsewhere. An example of a soft constraint is the requirement for redundancy in installed
equipment in order to increase the level of operator comfort. It follows that the range of available
options can be increased by relaxing soft constraints, and by finding some other way to alleviate the problem
that the constraint was intended to prevent/mitigate.
One way of highlighting the nature of a constraint is to set bounds for the acceptable temperature range.
Process plants seldom operate under fixed condition. For instance, as discussed below, throughput can vary
significantly, which gives rise to variations in process temperatures. We can identify the hardness of a
constraint by specifying both a normal (eg. design point) value and maximum and minimum acceptable
values. In this way we can define the flexibility inherent within the plant, and exploit it to reduce operating
costs.
The key to increased flexibility is introducing new Degrees of Freedom, which are parameters or design
features over which one has some control. For example, a process heater that operates under varying load
can be piped up to use two alternative sources of steam LP and HP, so that the cheapest one can be used
depending on operating conditions requirements.
Basically, one must keep an open mind. Think out-of-the-box. Do not accept the existing plant
configuration as inviolate; try to think of the ideal solution, and then systematically add features to the
existing design that will help to reach that ideal solution. Learn to recognize the difference between hard and
soft constraints.
Consider the feed rate profile of experienced by one of our condensate fractionation plants. As you will see
in Figure 1, it is highly variable, with a max/min flow ratio of roughly 630/350 =1.8 and a peak/avg ratio
of about 630/450 = 1.4. The plant is clearly capable of being operated at a throughput of 630 MBD
while still maintaining temperatures within acceptable bounds. This suggests that if the throughput were
controlled closely within narrow bounds, the investment that has already been made would permit a 40%
increase in production. That is the inherent power of reducing process variability.
Now consider the impact of the variable fractionation feed rate on plant energy costs. The steam
production rate will typically follow the feed flow rate. The ratio of peak/average flow is seen to be
approximately 2200/1530 = 1.44, consistent with the feed flow profile.
The benefits of reduced process variability apply to utility systems as well. Consider the case of a plant
whose operating policy is to run N+1 boilers, where N is the minimum number of boilers required to
satisfy peak process demand. Suppose that the boilers are equal sized, with a maximum sustained capacity
of 530 Klb/h each. The estimated potential savings from reducing process variability to 1.0 is shown in
Table 1. Note that these are free savings, requiring minimal capital investment in process revamp.
Page 8 of 84
Document Responsibility: Energy Systems Optimization Standards Committee SABP-A-008
Issue Date: 28 February 2016
Next Planned Update: TBD Load Management for Energy Efficiency: Heat Transfer Equipment
700
C2+ Feed
600 C3+ Feed
Combined Feeds
500
Flow, MBD
400
300
200
100
0
5/1/01 6/8/01 7/16/01 8/23/01 9/30/01 11/7/01 12/15/01 1/22/02 3/1/02 4/8/02
2000
Total HP Stm Gen, Klb/h
1500
1000
500
0
4/1/2001 5/21/2001 7/10/2001 8/29/2001 10/18/2001 12/7/2001 1/26/2002 3/17/2002
Page 9 of 84
Document Responsibility: Energy Systems Optimization Standards Committee SABP-A-008
Issue Date: 28 February 2016
Next Planned Update: TBD Load Management for Energy Efficiency: Heat Transfer Equipment
How can we reduce process variability? Generally by revamping the control strategy and introducing
additional surge capacity at critical pinch points in the process scheme. In the example cited, it is clear
that the data clearly show that swings in steam demand follow (and therefore are most likely a consequence
of) variations in feed flow rate. By increasing the amount of feed surge capacity through addition of a
couple of storage tanks, the feed to fractionation was able to be changed from level control (as obvious in
Figure 3) to flow control, per Figure 4, thus eliminating the hourly swings.
Feed to Fractionation
12000
10000
8000
Total Flow, BPH
6000
4000
0
4/1 4/25 5/19 6/12 7/6 7/30 8/23 9/16 10/10 11/3 11/27 12/21 1/14 2/7 3/3 3/27
Page 10 of 84
Document Responsibility: Energy Systems Optimization Standards Committee SABP-A-008
Issue Date: 28 February 2016
Next Planned Update: TBD Load Management for Energy Efficiency: Heat Transfer Equipment
The most fundamental relationship in heat transfer is the heat balance, which derives from the first law of
thermodynamics, and can be expressed as follows:
q = Mass flow rate of cold stream x specific heat cold stream x (t2 t1)
= Mass flow rate of hot stream x specific heat of hot stream x (T1-T2) The heat
q = U x A x CMTD
Page 11 of 84
Document Responsibility: Energy Systems Optimization Standards Committee SABP-A-008
Issue Date: 28 February 2016
Next Planned Update: TBD Load Management for Energy Efficiency: Heat Transfer Equipment
The overall heat transfer coefficient is related to the film heat transfer coefficients for the hot and cold fluids
by the equation:
where h1 and h2 are the film heat transfer coefficients, inclusive of fouling allowances, for the hot and cold
fluids respectively, and which are a function of the fluid flow regime (proportional to Reynolds number
raised to the power 0.8), exchanger geometry (equivalent diameter of cross-sectional flow area), and fluid
properties (specific heat, thermal conductivity, density, viscosity).
The CMTD is a complex function of the exchanger geometry and fluid flow profiles, and is generally
expressed as
CMTD = Ft x LMTD
where LMTD = log mean temperature difference for true countercurrent flow, and
Ft = correction factor for non-countercurrent temperature profile
Consider Figure 5, which shows the schematic representation of a HX with a true counter- current
temperature profile. The Logarithmic Mean Temperature Difference, or LMTD, is defined as
The correction factor Ft adjusts the LMTD for cases where the temperature profiles of the hot and cold
stream are not true counter-current. Detailed procedures for calculating Ft are beyond the scope of this
Best Practice manual. The reader is instead referred to basic college text books on heat transfer theory and
th
heat exchanger design, such as Holmans Heat Transfer, 8 ed, McGraw-Hill Book Co, New York (1997),
or Kerns Process Heat Transfer, McGraw-Hill Book Co, New York (1950)
Page 12 of 84
Document Responsibility: Energy Systems Optimization Standards Committee SABP-A-008
Issue Date: 28 February 2016
Next Planned Update: TBD Load Management for Energy Efficiency: Heat Transfer Equipment
For economic reasons, the HX is normally designed is such a way that the value of Ft is between 0.75
and 1.0 (Saudi Aramco standard is 0.8). For multi-pass shell & tube HX with t2 < T2 (ie. no temperature
cross), the approximations of Gulyani & Mohanty [Chem Eng, (Nov 1996) p127] can be used with an
error of less than 1%:
T
t 2
Ft = A + B
2
T1
t1
HX Configuration A B
Page 13 of 84
Document Responsibility: Energy Systems Optimization Standards Committee SABP-A-008
Issue Date: 28 February 2016
Next Planned Update: TBD Load Management for Energy Efficiency: Heat Transfer Equipment
HX Effectiveness Relationships
th
Source: J. P. Holman, Heat Transfer, 8 ed, p 581. Note that C = C
min/C
max
, and N = NTU. For a more comprehensive list, see Table 17.6, Handbook of Heat Transfer, Rohsenow et
al, op. cit.
Page 14 of 84
Document Responsibility: Energy Systems Optimization Standards Committee SABP-A-008
Issue Date: 28 February 2016
Next Planned Update: TBD Load Management for Energy Efficiency: Heat Transfer Equipment
Except when C (more commonly denoted as R) = 0, the hot and cold outlet temperatures can
be calculated from the known hot and cold stream inlet temperatures T1 and t1 as:
T2 = T1 (T1 t1), and t2 = t1 + R(T1 t1)
Using the Ft method, the number of minimum shells in series can only be determined by trial and error.
Shenoy [Ref. Heat Exchanger Network Synthesis, Gulf Publishing Co, Houston (1995), pp 260-264] has
described a more direct method using two parameters P and R:
Page 15 of 84
Document Responsibility: Energy Systems Optimization Standards Committee SABP-A-008
Issue Date: 28 February 2016
Next Planned Update: TBD Load Management for Energy Efficiency: Heat Transfer Equipment
The minimum number of shells required (in multiple tube-pass configuration) in order to have Ft
0.75 is obtained by
(NS)min usually turns out to be a fractional number, and must be rounded up to the next integer. The
design is considered acceptable when P PC.
The Ft method can be used only when the inlet and outlet temperatures are specified, or otherwise
known. For heat load management, that is not generally the case. Normally, we know only two of the four
terminal temperatures, with the other two having to be determined by simulation. Because the equations
involved are non-linear, the solution requires trial and error. For simulating the performance of existing
heat exchangers of known geometry under different operating conditions, the effectiveness-NTU method is
much more convenient. The effectiveness is defined as:
Page 16 of 84
Document Responsibility: Energy Systems Optimization Standards Committee SABP-A-008
Issue Date: 28 February 2016
Next Planned Update: TBD Load Management for Energy Efficiency: Heat Transfer Equipment
A quick approximate estimate for the number of shells-in-series required for a given duty can obtained by
stepping down the hot and cold duty lines (see Figure 6). This is equivalent to applying a temperature meet
condition to each shell. Since the number of shells must be an integer, fractional solutions are rounded up.
These are the critical relationships that will be required to perform HEN simulations, from which we
can derive the optimum HX load management policies.
Page 17 of 84
Document Responsibility: Energy Systems Optimization Standards Committee SABP-A-008
Issue Date: 28 February 2016
Next Planned Update: TBD Load Management for Energy Efficiency: Heat Transfer Equipment
Where: TW = tube wall temperature, which can be estimated as = (hoTo + hiti) / (ho + hi)
To = bulk temperature of hot fluid (shell side), and ti
= bulk temperature of cold fluid (tube side).
Fouling rates are determined through HX performance monitoring under actual field conditions, as illustrated in
Figure 7. The activation energy E is a key variable which must be empirically adjusted to fit the data, and
usually falls in the range 16.8 - 18.9 Btu/lbmole (39-44 kJ/kg-mole) for most crude oil streams. The removal
-8
constant k has a value of 3 x 10 .
0.011
0.01
0.009
0.008
Overall Fouling Factor
0.007
0.006
0.005
0.004
0.003
0.002
0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000
Time since last cleaning (hours)
Activation Energy (E) and Removal Constant (k) are best obtained through the analysis of the performance of
an existing plant. Figure 8 shows a screen shot from ESDUs EXPRESS computer program. The plot is
derived for chosen E and k. The upper line (red) shows fouling rate predicted for the hottest point in the
exchanger, the lower line (blue) shows the rate predicted for the crude inlet, the middle line (magenta)
gives a prediction of the overall fouling rate (derived from an integral equation), the short (grey) line is the rate
measured on the plant.
It is possible to get the integral rate and the measured rate to coincide through judicious selection of E
and k. The best values for E and k are obtained by analyzing the performance of a range of exchangers each
operating at different velocity and bulk temperatures.
At high temperatures, the deposition rate increases. At high velocities, the scale removal rate increases. It is
therefore possible to construct a chart, as in Figure 10, that shows the flow regimes and temperatures in
which deposition rates exceed removal rates, and vice versa. Such fouling threshold plots should be
developed for each exchanger (shell-side as well as tube-side), and the HX must be operated in the
appropriate temp-flow regime to minimize fouling. The fouling threshold identifies the flow and
Page 18 of 84
Document Responsibility: Energy Systems Optimization Standards Committee SABP-A-008
Issue Date: 28 February 2016
Next Planned Update: TBD Load Management for Energy Efficiency: Heat Transfer Equipment
temperature conditions at which it becomes impossible to eliminate fouling through exchanger design. Above
this point either a cleaning schedule should be introduced or fouling controlled through the use of tube inserts.
Source: H. Lemke, Fouling in Refinery Equipment an Overview, AIChE Spring Mtg (March 1999)
Page 19 of 84
Document Responsibility: Energy Systems Optimization Standards Committee SABP-A-008
Issue Date: 28 February 2016
Next Planned Update: TBD Load Management for Energy Efficiency: Heat Transfer Equipment
The fouling problem is equally important for fin-fan to air coolers, which are the predominant type of utility
coolers in Saudi Arabia. Fortunately, they are relatively easy to clean.
340
320
300
Temperature (C)
280
260
240
Epstein Model
Polley Model
220 No Fouling
Positive Fouling
Negative Fouling
200
0 0.5 1 1.5 2 2.5 3 3.5
Velocity (m/s)
Page 20 of 84
Document Responsibility: Energy Systems Optimization Standards Committee SABP-A-008
Issue Date: 28 February 2016
Next Planned Update: TBD Load Management for Energy Efficiency: Heat Transfer Equipment
The conventional cleaning method is to use hydro-blasting. However, this tends to damage the
fins, causing a permanently high flow resistance on the air aide, and corresponding loss of
capacity. A newer and preferred method is foam cleaning. This has been successfully used in
some Saudi Aramco refineries, and is recommended as a best practice. Since cleaning costs are
low, the optimum cleaning cycle is much shorter than for shell & tube HX cleaning.
Gentle Foam
Drainage
through Fins
Low Pressure
Foam Aerates
and Protects
Fins
Page 21 of 84
Document Responsibility: Energy Systems Optimization Standards Committee SABP-A-008
Issue Date: 28 February 2016
Next Planned Update: TBD Load Management for Energy Efficiency: Heat Transfer Equipment
Page 22 of 84
Document Responsibility: Energy Systems Optimization Standards Committee SABP-A-008
Issue Date: 28 February 2016
Next Planned Update: TBD Load Management for Energy Efficiency: Heat Transfer Equipment
When a process stream has to be heated from its supply temperature to its target temperature, this can be
accomplished in a variety of ways. For example, all of the heat could be supplied through utilities, or some
could be supplied via heat recovery from other process streams, and the balance through utilities. There is also
the option of allocating load between utilities.
Page 23 of 84
Document Responsibility: Energy Systems Optimization Standards Committee SABP-A-008
Issue Date: 28 February 2016
Next Planned Update: TBD Load Management for Energy Efficiency: Heat Transfer Equipment
Page 19 of 74
Document Responsibility: Energy Systems Optimization Standards Committee SABP-A-008
Issue Date: 28 February 2016
Next Planned Update: TBD Load Management for Energy Efficiency: Heat Transfer Equipment
In case 1, all heat is supplied using high-pressure (HP) steam, at a cost of $58.3 per hour. In case 2, the
heating load is split between low-pressure (LP) and HP steam, for a total cost of
$52 per hour. This reduction in operating cost was possible because we shifted some of the heating duty
from a more expensive (HP steam) utility to a less expensive utility (LP steam).
Now let us consider case 3a. Here, some of the heating duty is provided by recovery of waste heat from
another process stream, one which requires cooling. The cost drops further, and rather dramatically, from
$52 per hour to $37.5 per hour. If we do more heat recovery, as in case 3b, the cost drops to $36 per hour.
Obviously, our goal is to minimize the total utility cost, which corresponds to case 3b. However, unless the
design of the existing train conforms to the configuration shown in case 3, we will not have the flexibility
to trade-off LP steam versus HP steam, and LP steam versus process heat recovery.
These three cases illustrate the essential principle of load management. viz., when we have the option to use
multiple utilities, we should maximize the duty of the cheapest utility, and minimize the duty of the most
expensive utility. Thus, in Cases 2 and 3, the control system should be designed to maximize the process
stream temperature out of the LP steam heater, as illustrated in Figure 14, rather than setting LP steam on
direct flow control.
Figure 14: Control Strategy for Optimal Steam Load Distribution in a Preheat Train
In general, we will not have the freedom to do all of the heating with free waste heat. Nor will we be able to
eliminate use of the most expensive utility. This is because we have certain design constraints. One is
the temperature of the heat source, another is the available quantity of the heat source, and the third is the
capacity of the heat exchange equipment itself.
Page 25 of 84
Document Responsibility: Energy Systems Optimization Standards Committee SABP-A-008
Issue Date: 28 February 2016
Next Planned Update: TBD Load Management for Energy Efficiency: Heat Transfer Equipment
The first two constraints are obvious. It is not possible to heat a cold stream to a temperature higher than the
temperature of the hot stream (heat source). This is known as the driving force constraint. Similarly, the
cold stream cannot be supplied more heat than is available in the heat source. This is known as the heat
balance constraint. The third constraint has to do with heat transfer capacity of the HX, which could become
limiting because at flow rates below design, U, LMTD, and Ft all fall, and the multiplicative effect of the
combined reduction could over-ride the reduced requirement for Q.
While the Case 3 configuration shown in Figure 13 may give the lowest operating cost, it should be
recognized that the capital cost of this design will be higher than that of Case 1. Suppose we are faced
with a Case 1 configuration. What can be done to minimize operating cost at minimal expense of capital cost?
A possible solution is illustrated in Figure 15.
The basic principle is to introduce a new degree of freedom by adding a new LP supply connection,
and blend LP steam with HP steam through an ejector. The discharge pressure from the ejector will be an
intermediate value between that of LP steam and HP steam. The cost savings derive from the fact that
some of the HP steam is substituted with LP steam. Some additional capital will be required for an ejector
and for additional process controls, but it would be substantially less than adding a new LP steam heater.
What is the optimum load management policy when the actual flow rate is less than design? Should we
reduce all flows equally, or should we valve off some of the HX and run the rest at closer to design capacity?
The two options are compared in Table 3.
Page 26 of 84
Document Responsibility: Energy Systems Optimization Standards Committee SABP-A-008
Issue Date: 28 February 2016
Next Planned Update: TBD Load Management for Energy Efficiency: Heat Transfer Equipment
Page 27 of 84
Document Responsibility: Energy Systems Optimization Standards Committee SABP-A-008
Issue Date: 28 February 2016
Next Planned Update: TBD Load Management for Energy Efficiency: Heat Transfer Equipment
Page 22 of 74
Document Responsibility: Energy Systems Optimization Standards Committee SABP-A-008
Issue Date: 28 February 2016
Next Planned Update: TBD Load Management for Energy Efficiency: Heat Transfer Equipment
What is very interesting is that even though the heat transfer duty on the inter-changer goes down
substantially at lower operating rates, the feed outlet temperature stays about the same. This is not an
0.2
aberration; but has a theoretical basis: LMTD is roughly proportional to r , where r = actual throughput
design throughput. Thus LMTD remains substantially constant over the normal range of operating rates. A
50% reduction in throughout will cause LMTD to suffer only a 13% decrease. As LMTD falls, the feed
outlet temperature rises, and steam requirement per unit of feed flow goes down. This observation is
generically true.
The simulation results show that it is better to run the network with all units, although the difference is
very slight. This is based on the assumption that lower velocities will not result in increased fouling rates (per
Figure 9). However, if the lower velocities at reduced operating rates cause the HX to fall into the fouling
zone, then it would be better to run the network with fewer trains, and accept the small increase in steam
costs. The model should check that HX duty q = M.Cp.(T1-T2) is q = U.A.CMTD, as this constraint cannot
be violated.
What the simulation model also shows is that a very small energy cost penalty (< 1%) is incurred if the
plant elects to shut down one train (during low-rate operation) for HX cleaning or repair while the rest of the
trains pick up the slack.
Page 27 of 84
Document Responsibility: Energy Systems Optimization Standards Committee SABP-A-008
Issue Date: 28 February 2016
Next Planned Update: TBD Load Management for Energy Efficiency: Heat Transfer Equipment
Typically, the intermediate product from process A must be cooled before storage, both to prevent
material loss due to evaporation, as well as consequent environmental emissions. Typically this
intermediate product must then be reheated before it is fed to process B. Once we recognize the inter-
relationship, it becomes obvious that both heating and cooling duties could be eliminated (or at least
minimized) by by-passing the intermediate storage tank during regular operation, and using it only to
compensate for flow imbalances between the operating rates of processes A and B.
Page 28 of 84
Document Responsibility: Energy Systems Optimization Standards Committee SABP-A-008
Issue Date: 28 February 2016
Next Planned Update: TBD Load Management for Energy Efficiency: Heat Transfer Equipment
Load management policies for process coolers follow the same general rules as for process heaters.
Page 29 of 84
Document Responsibility: Energy Systems Optimization Standards Committee SABP-A-008
Issue Date: 28 February 2016
Next Planned Update: TBD Load Management for Energy Efficiency: Heat Transfer Equipment
In case 1, all cooling is done with refrigerant, at a cost of $34.9 per hour. In case 2, a significant
portion of the cooling duty has been shifted to lower cost air cooling, for a total cost of only $14.9 per hour.
Process-process heat recovery, while it can reduce total utility cooling duty, does not reduce costs
significantly, because all the saving occurs in low-cost air-cooling duty. Therefore Case 2 will in all
likelihood have better economics than Cases 3a and 3b. The key optimization parameter here would be
the outlet temperature from the air cooler (120F in the example). All efforts should be made to
minimize this temperature. The appropriate operating policy whenever refrigerant is being used would
therefore include some or all of the following:
Run all the air-cooler (fin-fans) at full speed all the time
Clean the fin-fans frequently to maintain high U (must set up a monitoring program)
Use water spray on fin-fans to boost heat transfer capacity through evaporative cooling. Water
quality must be very high, though, to prevent scale formation and/or corrosion. Cooled boiler
blowdown may a suitable candidate for this application.
It would be a mistake to control the air-cooler outlet temperature to some set design value by cutting back on
the fans.
Depending on the economics, it might also be worth introducing another degree of freedom, e.g., cooling
water as an additional utility to off-load the refrigerant duty.
The best way to determine the optimum load management strategy is to construct a simulation model for each
application as illustrated in Table 4, and focus on optimizing the operating parameters that have the greatest
impact on energy total cost.
Page 30 of 84
Document Responsibility: Energy Systems Optimization Standards Committee SABP-A-008
Issue Date: 28 February 2016
Next Planned Update: TBD Load Management for Energy Efficiency: Heat Transfer Equipment
Observe that the previous conclusion about running all trains in preference to fewer trains is no longer
correct. The previous conclusion is valid only if the utility cost is a continuous function of capacity; eg
for steam, for which the cost per MMBtu is constant within narrow bands. In the case of air cooled (fin-
fan) HX that is generally not so. Air coolers are usually equipped with fixed speed fans which can be
either ON or OFF, so that a reduced cooling load does not automatically translate into power savings. Only
when the cooling load drops sufficiently, one of the fans in a fin-fan bank c a n be shut off. In other words,
fan power is a discrete (step-wise) function of cooling load.
Page 31 of 84
Document Responsibility: Energy Systems Optimization Standards Committee SABP-A-008
Issue Date: 28 February 2016
Next Planned Update: TBD Load Management for Energy Efficiency: Heat Transfer Equipment
For the selected design basis conditions, the optimum operating policy is:
It was determined by running the simulation model at various load factors, making sure that the actual
capacity of the heat exchangers never exceeds the maximum capacity.
In the illustrative example, the cold process stream is being heated against the product and fed directly to a
reactor. That may not always be feasible, and a trim heater (or flow bypass) for more precise temperature
control may be required.
If the air cooler motors were equipped with variable frequency drives, then it would be possible to modulate
the fan speed and reduce the air flow rates to exactly match the required duty. In such a case, the cost of
operating the air cooler would be more nearly a constant with respect to capacity, and we would have to do
the economics differently, as in Table 4, with the optimum operating policy probably also being different.
So the message is one can only determine the optimum load management policy by constructing a
simulation model that accurately reflects the actual capabilities and operating characteristics of the
equipment in the field. Generalizations can be dangerously misleading.
Page 32 of 84
Document Responsibility: Energy Systems Optimization Standards Committee SABP-A-008
Issue Date: 28 February 2016
Next Planned Update: TBD Load Management for Energy Efficiency: Heat Transfer Equipment
What Table 5 demonstrates is that it is possible to save a significant portion (25%) of the power cost for
operating this cooling train if the number of trains being operated is adjusted accordingly to the load. The
actual savings potential versus flat out operation with all trains running all the time can be estimated using
historical data on the load profile, as illustrated in Figure 20 and Table 6.
Page 28 of 74
Document Responsibility: Energy Systems Optimization Standards Committee SABP-A-008
Issue Date: 28 February 2016
Next Planned Update: TBD Load Management for Energy Efficiency: Heat Transfer Equipment
Flow Profile
4500
4000
3500
3000
Hours per year
2500
2000
1500
1000
500
0
200 240 280 320 360 400
Flow, MBD
Figure 20: Historical (12 month) Flow Profile of Stream Being Cooled
The potential savings in this case are not especially large, but they are FREE, requiring zero capital
investment, and are a substantial reduction (about 20%) compared to the base case power cost
when running with all four trains all the time. In the example shown, only flow rate variations have
been examined. In fact, many other parameters could also be varying, such as supply temperatures
of the product and cold process stream, U values (due to fouling), and of course, ambient
temperature. These additional complexities can be easily accommodated in a spreadsheet model; the
simulation approach remains the same.
Page 34 of 84
Document Responsibility: Energy Systems Optimization Standards Committee SABP-A-008
Issue Date: 28 February 2016
Next Planned Update: TBD Load Management for Energy Efficiency: Heat Transfer Equipment
The optimum distribution of LMTD in each cooler occurs when the coolant flow is controlled to
achieve the desired target temperature of the hot process stream, which minimizes the power
consumption of the cooling water circulation pumps [Ref. Kumana, A Pre-Design Strategy
for Selecting Near-Optimum Cooling Water Flow Rates, AIChE Symposium Series
no. 236, Vol 80, (1984), pp 117-121].
Most often, all coolers are arranged in parallel, between the CW supplies and return headers. This is
usually not the optimum configuration. Some of the coolers should be in series with respect to coolant
flow. The correct cooler network structure (series/parallel arrangement and sequences) should be determined
by Pinch Analysis, and the appropriate piping modifications should be made to approach the optimum structure
as close as possible.
Observe in particular the location of the cooling water makeup and blowdown locations. The make-up must
be added at the point in the circuit where coolant temperature is lowest, and the blowdown should be taken
from the point in the circuit where the coolant temperature is highest. Usually these are reversed, for
mechanical convenience, by vendors and contractors who neither understand nor care about operating
efficiency.
Page 40 of 74
Document Responsibility: Energy Systems Optimization Standards Committee SABP-A-008
Issue Date: 28 February 2016
Next Planned Update: TBD Load Management for Energy Efficiency: Heat Transfer Equipment
Page 40 of 74
Document Responsibility: Energy Systems Optimization Standards Committee SABP-A-008
Issue Date: 28 February 2016
Next Planned Update: TBD Load Management for Energy Efficiency: Heat Transfer Equipment
The tricky part is being able to determine when exactly we have achieved total condensation, something very
difficult to do. The proposed solution is to have two condensers in series. The main condenser would
condense only about 90-95% of the vapor, and the vent condenser would condense the balance. The
control system would be set up to maintain a fixed flow ratio in the range from 10:1 to 20:1 between the
main flow and the vent flow.
The required discharge pressure in Case 2 is found by successive iteration until the calculated condenser
surface area for cases 1 and 2 are identical.
Although process modifications cannot strictly be classified as Load Management, the subject has been
presented here because it is a way to introduce new degrees of freedom that enable optimal load allocation
between the different energy consumers in the overall system.
Page 40 of 74
Document Responsibility: Energy Systems Optimization Standards Committee SABP-A-008
Issue Date: 28 February 2016
Next Planned Update: TBD Load Management for Energy Efficiency: Heat Transfer Equipment
Process heating and cooling seldom occurs in isolation. Invariably, heating and cooling loads are interlinked
though the heat and material balance. Even the simplified examples in sections
3.2 and 3.3 featured both heating and cooling.
In this section we are going to focus on how to optimize operating loads in more complex Heat Exchanger
Networks (HENs), that are typical of oil, gas and petrochemical processes.
The following design data are given, with the flow rate and feed supply temperature known to deviate
significantly from design. What should the optimum operating policy be when these two parameters
(identified as green-shaded cells) vary over their typical range of values?
Page 40 of 74
Document Responsibility: Energy Systems Optimization Standards Committee SABP-A-008
Issue Date: 28 February 2016
Next Planned Update: TBD Load Management for Energy Efficiency: Heat Transfer Equipment
Once again, we take the simulation approach, and work out what the performance of the HEN would be
under various scenarios. Once we understand the response of the HEN to the expected deviations from
design conditions, it becomes possible to determine the optimum load management policy that gives the
lowest operating cost.
Page 40 of 74
Document Responsibility: Energy Systems Optimization Standards Committee SABP-A-008
Issue Date: 28 February 2016
Next Planned Update: TBD Load Management for Energy Efficiency: Heat Transfer Equipment
What if the operation involves a change in both the feed rate and the feed temperature, as shown below?
The model structure would still be the same, but the optimum set points for operation would be different, as
illustrated in Table 9.
Table 9: Simulation Results for Simple HEN Combined Flow and Temp Deviations
Page 40 of 74
Document Responsibility: Energy Systems Optimization Standards Committee SABP-A-008
Issue Date: 28 February 2016
Next Planned Update: TBD Load Management for Energy Efficiency: Heat Transfer Equipment
The simulation models are not difficult to develop, but the following points should be noted:
(c) There is one drawback of calculating terminal temperatures in this way to prevent circular
references in the algorithm, the hot and cold stream flow rates are required as input data. In the
case of process streams this is generally not a problem, as they are known from the material
balance. In the case of utility streams, however, eg. cooling water, this algorithm does not
accurately represent the way the process is operated. Normally, for utilities, the objective is to
minimize the flowrate required to achieve a particular terminal temperature for the process
stream being heated or cooled, and that is how the control loop is set up. Therefore, instead of
specifying utility stream flow, we should specify the EMAT for each exchanger. We then need to
go through an iterative trial and error procedure to ensure that the specified EMAT values result
in terminal temperatures and heat transfer duties consistent with the HX size and the expected
U values at the new (off-design) operating conditions. The iteration stops when the required U
and available U agree within the specified tolerance limits.
(d) Flow rates for the heating and cooling utilities are calculated from the heat balance.
Page 40 of 74
Document Responsibility: Energy Systems Optimization Standards Committee SABP-A-008
Issue Date: 28 February 2016
Next Planned Update: TBD Load Management for Energy Efficiency: Heat Transfer Equipment
The simulation models tell us that if we change the set points of the HX control loops to achieve the
indicated terminal temperatures and duties, then the energy cost of operating the HEN will drop from $332/h
to $238/h, a saving of 28% in utility costs in Case A and a saving of 26.5% in Case B. The control logic for
the steam heaters should be as shown in Figure 14. The recommended control logic for the trim cooler is
presented in Figure 24.
Page 40 of 74
Document Responsibility: Energy Systems Optimization Standards Committee SABP-A-008
Issue Date: 28 February 2016
Next Planned Update: TBD Load Management for Energy Efficiency: Heat Transfer Equipment
The spreadsheet simulation model for the NHT process (as designed) is shown in Table 10.
Observe that the actual (operating) U values, even for the base case scenario, are significantly lower than the
expected clean U values, almost by an order of magnitude. Why such a huge discrepancy? The reason is that
most EPC contractors tend to use overly conservative fouling factors, which result in excessive heat transfer
area. The fact that actual U values are so low indicates that both shell and tube side the velocities are
extremely low, under which conditions rapid fouling is much more likely. Therefore the exchangers will
have to be cleaned more frequently, defeating the purpose of choosing a high fouling allowance to begin
with. What we need to do is check the shell and tube side velocities for each application, and determine
where the HX operation falls on the fouling threshold chart (see Figure 10). Normally the velocities should
be in the range of 3-10 ft /sec.
Page 40 of 74
Document Responsibility: Energy Systems Optimization Standards Committee SABP-A-008
Issue Date: 28 February 2016
Next Planned Update: TBD Load Management for Energy Efficiency: Heat Transfer Equipment
Table 10: Simulation Model of HEN for NHT Process (Design Case)
Page 40 of 74
Document Responsibility: Energy Systems Optimization Standards Committee SABP-A-008
Issue Date: 28 February 2016
Next Planned Update: TBD Load Management for Energy Efficiency: Heat Transfer Equipment
A detailed HX simulation (as opposed to HEN simulation) should also be done for each HX, as illustrated in
Table 11 and Figure 26. From this analysis, it is possible to determine what mechanical modifications,
e.g. adding another tube-side pass, adding a longitudinal shell-side baffle (conversion from E-shell to
F-shell configuration), etc, would be most practical to achieve the desired performance.
Table 11: Simulation Model for Shell & Tube Heat Exchanger
Page 40 of 74
Document Responsibility: Energy Systems Optimization Standards Committee SABP-A-008
Issue Date: 28 February 2016
Next Planned Update: TBD Load Management for Energy Efficiency: Heat Transfer Equipment
To assess maximum heat transfer capacity of the HX, adjust the duty until the available fouling factor is
approximately 0.0005 (generally between 0.0003 and 0.001).
To goal of making such mechanical modifications is to get additional heat transfer capacity from the HX,
and thereby to achieve better overall heat recovery in the HEN.
HX Temp Profile
160
150
140
130
Temp, F
120
110
100
90
Hot Stream
80 Cold Stream
70
0 5 10 15 20 25 30 35
Q, MMBtu/h
Although all the HEN simulation models of Saudi Aramco plants have thus far been developed internally from
scratch, it is possible to purchase commercial software programs that automate much of the work. One is
Persimmon , that provides a convenient ready-made Excel inter- face for general HEN simulation as well
as some advanced features such as a module for determining optimum HX cleaning schedules. This
software has been extensively used in the oil refining and petrochemical industries, and is available from
Veritech Energy, Virginia, USA. Also, a suite of inter-related programs that include Integrity and Express,
are available from ESDU International plc, London, UK. [Note: Software names have been mentioned here
for the readers convenience only, and should not be construed as a commercial endorsement.]
A very useful tool for assessing whether the pinch HX is in the network is the Driving Force Plot, as in
Figure 27. If the actual temp profile of the HX is far from the ideal driving force profile, the HX is
wasting temperature gradient. If the available temperature gradient within the HX is significantly less than
the ideal, then that HX is the one limiting the heat transfer capability of the HEN. To get improved
performance from the HEN, some corrective action in terms of piping modifications or addition of more
surface area will be required.
Another useful tool is the Exchanger Response (or Sensitivity) Plot, illustrated in Figure 28.
Page 40 of 74
Document Responsibility: Energy Systems Optimization Standards Committee SABP-A-008
Issue Date: 28 February 2016
Next Planned Update: TBD Load Management for Energy Efficiency: Heat Transfer Equipment
Poor Alignment
Good Alignment
Page 47 of 84
Document Responsibility: Energy Systems Optimization Standards Committee SABP-A-008
Issue Date: 28 February 2016
Next Planned Update: TBD Load Management for Energy Efficiency: Heat Transfer Equipment
Unfortunately, the problem of optimum operation of existing HENs has received very little attention
from research scholars, and there is a dearth of published material on the subject. Only four relevant papers
were found after conducting an exhaustive literature search:
K. Li and B. Niemeyer, Optimal Operation of HENs Under Uncertainty, Int J of Heat Exchrs, vol V
(2004), pp 79-94.
B. Glemmestad, S. Skogestad and T. Gundersen, Optimal Operation of HENs, Computers & Chem Eng, vol
23 (1999) pp 509-522.
R. Ratnam and V. S. Patwardhan, Sensitivity Analysis of HENs, Chem Eng Sci, vol 46, no 2 (1991), pp
451-458.
(a) Set up the model to determine the optimum temperature set-points for HEN control, with the
objective of minimizing total utility cost.
(b) Run the model periodically with updated process flow and inlet temperature data, say every 4-8
hours, and adjust the set points accordingly.
A HX may be improperly sized for its duty either because it was incorrectly designed for the specified
service to begin with, or because the actual operating conditions (flow rates, temperatures, heat transfer
coefficients) have deviated from expected design conditions. The consequences of undersized HX are
generally well known, but the pitfalls of over-sized HX are often not recognized.
2
Consider the simple system shown in Figure 29. Exchanger E-1 requires 1622 ft in order to achieve the
2
necessary duty. However, it has been oversized by 10% and 1785 ft have been installed. Similarly
2 2
exchanger E-2, which requires 5155 ft , has also been oversized by 10% with 5671 ft . Detailed simulation
results are summarized in Table 11.
The result of over-sizing E-1 means that it performs better than expected. So, the cold stream leaves the unit at
280F rather than the targeted value of 277F. Since, this does not have a detrimental effect upon the process
downstream of the unit this over-performance would not normally draw attention.
Page 48 of 84
Document Responsibility: Energy Systems Optimization Standards Committee SABP-A-008
Issue Date: 28 February 2016
Next Planned Update: TBD Load Management for Energy Efficiency: Heat Transfer Equipment
However, the hot stream leaves the unit at a temperature of 253F rather than the expected value of 259F.
This reduces the available temperature driving force in E-2 to below that used for design, and results in less
heat recovery than expected for this unit. The cold stream now leaves the E-2 at 210F rather than the
expected 212F. The shortfall in performance occurs despite the exchanger actually being over-sized by 10%,
and has two negative consequences
(a) energy penalty of 0.4 MMBtu/h, and (b) failure to meet the target cold stream temperature of 300F. One
could argue that the target temperature could be achieved by adding more surface area to the heater as
well, but that is additional investment for no energy benefit. A possible solution to avoiding the energy
2
penalty is to add 690 ft more surface area to E-2. We now have traded off a capital cost penalty for an energy
2
penalty. Observe that we would have to add 690 ft more area over and above the 10% over-size already built
2
into the design, for a total of 1205 ft extra (i.e., 23% over-sized). This is more than 7 sq ft extra in E-2 for
each sq ft of excess area in E-1.
Page 49 of 84
Document Responsibility: Energy Systems Optimization Standards Committee SABP-A-008
Issue Date: 28 February 2016
Next Planned Update: TBD Load Management for Energy Efficiency: Heat Transfer Equipment
The proper solution to this problem is to recognize that E-1 is over-performing, and operate a bypass around
it in order to restore the temperature driving force on E-2. These insights cannot be obtained without a
thorough analysis, as illustrated above. This example further demonstrates that in order to employ effective
Heat Load Management, the control system must be designed properly as well, and adequate instrumentation
must be provided.
An important principle for understanding the performance of HENs is the propagation of disturbances.
Recall that the increase in area of E-1 resulted in a disturbance to temperatures at which both hot and cold
streams left the exchanger, and in turn affecting the performance of E-2 located downstream. Reduced heat
transfer in E-2 caused the temperature of the cold stream leaving it to fall, thereby increasing the required
duty on the heater positioned downstream of E-2. Note that both disturbances only affected equipment
that was located downstream along the paths defined by the individual streams and the heat recovery
exchangers (Figure 30).
Now consider what would have happened if the temperature at which the cold stream entered E-2 had been
disturbed (see Figure 31). This disturbance affects the temperature at which the cold stream leaves E-2 and
enters the heater positioned downstream. It also affects the temperature at which the hot stream involved in
the match leaves the exchanger. Given the network structure, it cannot possibly affect the temperature of
the hot stream entering E-2. In short, the disturbance cannot move upstream.
The principle is simple, but general: disturbances can only propagate downstream.
Page 50 of 84
Document Responsibility: Energy Systems Optimization Standards Committee SABP-A-008
Issue Date: 28 February 2016
Next Planned Update: TBD Load Management for Energy Efficiency: Heat Transfer Equipment
The concept of downstream paths though simple is very powerful. For instance, if one process stream
is known to be subject to large disturbances and another stream needs to have a closely controlled
temperature, the designer can prevent control problems by simply ensuring that the sensitive stream is not
on a path downstream of the variable stream. If it is known that one particular exchanger is the subject of
severe fouling but some others in the system are not, then the effects of that fouling on the overall system
can be compensated for by installing extra area to clean exchangers that are on the same path as the dirty
one. The exchanger that is subject to fouling can then be designed with the objective of minimizing the
fouling within it rather than adding extra area to the unit (with resultant lower velocity and higher
fouling), which is the normal practice.
The concept of downstream paths is particularly relevant for Load Management in HENs. The
opportunities for exploiting changes in operating conditions of a given HX only exist along thermal paths
downstream of that location.
The real parameters of significance for developing an optimum HEN load management strategy are not the
unique target temperatures for streams, but the maximum and minimum allowable temperatures. The target
temperatures are useful the plant design; the allowable bounds are useful for heat load management. During
plant operation it is normal to find that the actual process temperature differs from the original target
(design) temperature. It is the relationship between actual temperature and the bounds that is important. For
instance, in the HEN of Exhibit 5-8, rather than simply compensate for the over-performance of E-1 relative
to the design target, the operator can exploit the flexibility inherent in the HEN by operating a bypass
around E-1 such that both the outlet temperatures are close to their optimum values, and result in minimum
steam use.
Page 51 of 84
Document Responsibility: Energy Systems Optimization Standards Committee SABP-A-008
Issue Date: 28 February 2016
Next Planned Update: TBD Load Management for Energy Efficiency: Heat Transfer Equipment
6.0 Boilers
Almost all industrial plants use steam as the principal energy source for process heating. However,
boilers should not be seen as islands, but merely as one of the components of the overall utility system, also
known as the Combined Heat and Power (CHP) system for the facility. Normally, a plant CHP system has
multiple boilers, which may or may not be of equal capacity and operating pressure, and may or may not
use the same fuel or operate at the same efficiencies. Furthermore, some of the process steam may be
generated in Waste Heat Boilers (unfired heat exchangers), while some might be generated in Heat
Recovery Steam Generators (HRSGs) associated with a gas turbine. Also HRSGs can be either supplementary
fired or unfired. Both HRSGs and WHBs are part of the plant CHP system. Although their mechanical
designs are radically different, functionally they are both identical to boilers.
This manual is focused on the predominant type of boilers used in Saudi Aramco plants, viz. natural-
circulation oil- and gas-fired water-tube industrial boilers. Operating pressures of modern boilers can
range from 150 to 1800 psig, with 150 psig and 600 psig being the most common within the company.
However, most of the load management principles described and recommendations made in this manual will
apply to other types of boilers as well, includ- ing smaller fire-tube boilers used in lower pressure applications
such as HVAC.
The policy of running more boilers than needed is expressed as N+, where N is the minimum number of
boilers required to supply the peak process steam demand, and is the number of extra boilers that are kept
in operation to provide instantaneous spare capacity in the event one of the boilers trips.
Some Saudi Aramco plants operate with an N+1 policy, while others follow an N+2 policy; but there is a
cost. When excess numbers of boilers are operated, the average steam generation rate for each boiler will be
significantly less than design, and the energy efficiency will be lower, as shown in Table 13.
The fact is that improved reliability can be achieved in a number of different ways, most of which do not
incur such a high energy cost penalty. An understanding of the factors that govern reliability of steam
supply to the process is therefore critical to the question of optimum load management policies. The crucial
issue is not how many boilers to operate as running spares, but whether adequate steam reserve is available
at all times to accommodate short- term process load swings.
Page 52 of 84
Document Responsibility: Energy Systems Optimization Standards Committee SABP-A-008
Issue Date: 28 February 2016
Next Planned Update: TBD Load Management for Energy Efficiency: Heat Transfer Equipment
Figure 32: Typical Design of Field-erected Water-tube Boiler Fired with Clean Fuels
rd
Source: Singer, ed, Combustion: Fossil Power Systems, 3 ed, Combustion Engg Inc, Windsor, CT (1981), p8-
25
Page 53 of 84
Document Responsibility: Energy Systems Optimization Standards Committee SABP-A-008
Issue Date: 28 February 2016
Next Planned Update: TBD Load Management for Energy Efficiency: Heat Transfer Equipment
One of the ways to reduce the required steam reserve is to minimize process variability (cf. section 2.4).
Another is to develop a load-shedding program during upsets, when non-critical steam users (such as storage
tank coils, or even the de-aerator) can be temporarily shut off. A third might be to increase the amount of
steam generated in the process WHBs. A fourth option would be to keep the spare boiler on hot standby,
rather than full operation; this will enable rapid startup (within 20-30 min at most) to full rate in case one of
the boiler trips, during which time the plant can resort to load shedding or one of the other tactics. In the
worst case, production rate might have to be slowed down somewhat for a few minutes, which can be
easily made up as soon as the spare boiler is up and running; hardly a disaster.
In Table 13, notice the available steam reserve for the N+0 operating policy. It is equivalent to more than 2
full boilers. This is the type of analysis that should be done when determining the optimum boiler load
management policy.
The example cited assumes equal sized boilers at the same pressure and with the same energy efficiency
profiles. In such cases, a CHP model such as that described in section 6.4 can be a more effective analytical
tool.
Page 54 of 84
Document Responsibility: Energy Systems Optimization Standards Committee SABP-A-008
Issue Date: 28 February 2016
Next Planned Update: TBD Load Management for Energy Efficiency: Heat Transfer Equipment
Best Practice depends upon the amount of reliable data available, and the sophistication of the control system.
One is load allocation by efficiency; the other is by least operating cost.
This is the simpler of the two methods. It is close to optimal when the boilers use the same fuel, and have
a congruent efficiency profile. However, it could be far from optimal if these conditions do not hold. The
appropriate operating policy, when we have N parallel boilers using the same fuel and with efficiency
curves of the same shape is simple, and easy to implement:
Use the most efficient (N-1) boilers for base load, and
use the least efficient boiler for swing load.
Boiler Efficiency is non-linear with steam load. However, the fuel consumption varies quasi- linearly with
steam load (the fuel consumption is approximately linear with steam rate for boiler loads of 25% and
higher, with a quadratic multiplier at low steam rates). It is therefore much easier to obtain accurate
correlations, when attempting to determine boiler efficiency from field data, by plotting the total energy
(fuel + power) consumption versus steam rate rather than plotting efficiency versus steam rate. Figures 33
and 34 illustrate this point.
Boiler Eff, %
60 60
50
40 40
30
20 Fuel 20
Efficiency 10
0 0
0 20 40 60 80 100
Steam gen rate, % of design
Figure 33: Boiler Fuel Use and Efficiency versus Steam Rate
Source: Kumana, Correlation of typical boiler performance data (Saudi Aramco plants)
Page 55 of 84
Document Responsibility: Energy Systems Optimization Standards Committee SABP-A-008
Issue Date: 28 February 2016
Next Planned Update: TBD Load Management for Energy Efficiency: Heat Transfer Equipment
Boiler efficiency is traditionally defined according to the ASME code, which considers
it to be the heat absorbed in the boiler (radiant and convection sections) for
converting boiler feedwater into steam, whether saturated or superheated, divided
by the amount of fuel energy supplied. Thus,
Saudi Aramco practice, which follows the ASME boiler code, is to use the higher
heating value (HHV) for calculating fuel input, not the lower heating value (LHV).
Also, boiler efficiency is normally calculated by the energy balance (heat loss)
method; the equation presented above is intended only to illustrate the underlying
concept.
E1
=
12
E i E
0
i2
Page 56 of 84
Document Responsibility: Energy Systems Optimization Standards Committee SABP-A-008
Issue Date: 28 February 2016
Next Planned Update: TBD Load Management for Energy Efficiency: Heat Transfer Equipment
1.1
2
y = 0.75x - 0.555x + 1.0995
1.08 2
R = 0.9991
Fuel multiplier
1.06
1.04
1.02
0.98
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35
% Boiler load
Page 51 of 74
Document Responsibility: Energy Systems Optimization Standards Committee SABP-A-008
Issue Date: 28 February 2016
Next Planned Update: TBD Load Management for Energy Efficiency: Heat Transfer Equipment
The current state of the art technique for boiler load allocation when using different fuels for
different boilers (some could even be dual-fueled) is the least cost algorithm per Figure 35,
which develops a bias signal to raise or lower the firing rate.
Page 51 of 74
Document Responsibility: Energy Systems Optimization Standards Committee SABP-A-008
Issue Date: 28 February 2016
Next Planned Update: TBD Load Management for Energy Efficiency: Heat Transfer Equipment
The computations for load allocation signals include all operating costs fuel consumption, unit fuel
cost, and any special factors (such as O&M differentials). The boiler with the highest efficiency may not
be the best one for adding incremental steam load; the correct policy is to load and unload the boilers at
the most favorable incremental rate.
Figure 36 demonstrates the logic of this approach. Assume that we have two boilers for which the
efficiency curves are known (Figure 36a). Boiler 1 efficiency is always higher than that of Boiler 2 at all
steam rates, which might lead us to conclude that Boiler 1 is always more economical than Boiler 2.
However, this may not be so. The fuel input vs steam output curves in Figure 36b clearly show that in this
case, the least cost load allocation policy for a boiler load range between 0% and 100% is actually as
follows:
Page 59 of 84
Document Responsibility: Energy Systems Optimization Standards Committee SABP-A-008
Issue Date: 28 February 2016
Next Planned Update: TBD Load Management for Energy Efficiency: Heat Transfer Equipment
In order to implement such a policy, accurate data for each boiler is required to determine the slopes of the
efficiency curves, and this must then be converted into an optimum boiler load allocation policy as
illustrated above.
Page 52 of 74
Document Responsibility: Energy Systems Optimization Standards Committee SABP-A-008
Issue Date: 28 February 2016
Next Planned Update: TBD Load Management for Energy Efficiency: Heat Transfer Equipment
6.3.1 Burners
In the past, the principal objective of good burner design was efficient combustion. New and increasingly
stringent environmental regulations have forced a consideration of NOx emissions to the environment as well.
Unfortunately there is usually a penalty associated with low NOx emissions. For example, a proven
technique for reducing NOx emissions is staging, where the primary combustion zone is deficient in
either the fuel or air, with the balance being injected into the secondary combustion zone downstream.
4
Staging reduces both fuel NOx and thermal NOx. However, because flux is roughly proportional to T , lower
flame temperatures also reduce the capacity of the radiant zone. Similarly the flue gas recirculation
approach to reducing NOx emissions increases the amount of power consumption.
Burners are most commonly classified according to the method of fuel-air mixing diffusion, pre-mixed,
and staged. However, they may also be classified according to location floor burners, roof burners, and
wall burners.
The keys to efficient burner operation are to (a) accurately control the air-to-fuel ratio (see Figure 37),
and (b) make sure that the firing rate of each burner is within its operating range. An excellent discussion of
the parameters that govern burner efficiency is given by Khavkin (cf. section 1.4).
Extensive experience with oil and gas fired boilers has shown that it is imperative to achieve proper air
distribution to each burner to control flame shape, flame length, excess air ratio, and overall combustion
efficiency. Most existing boilers with older technology have burners with a turn-down ratio limited to 3:1, so
that when the boiler is operating significantly below design rates, the correct burner management policy is
to selectively shut down burners rather than attempt to reduce the firing rate equally to all burners. That
is why boilers usually come equipped with multiple burners (typically 2-8 for Saudi Aramco plants), to
provide relatively continuous turn-down capability. Equal load distribution overa ll operating burners gives
the best overall results [cf. Baukal, Combustion Handbook, p 564]. This requires independent flow control of
both fuel and air, as shown in Figure 38.
There is an optimum pattern to shutting down burners when steam generation rate is reduced, and restarting
burners when the steam generation rate is increased. This pattern depends on the location of the burners in
relation to the radiant section tubes. This operating procedure is normally specified by the boiler manufacturer,
and should be followed exactly. When burners are loaded properly, the flames will be very similar in
appearance, as in Figure 39.
Newer burner technology is now available that offers improved operating turndown ratios of up to 10:1, while
maintaining flame stability and low NOx emissions. For boilers equipped with these new burners, the best
operating policy would be possible to keep all burners running (ie. no selective shut-downs) but continue to
distribute the load evenly to all.
Page 61 of 84
Document Responsibility: Energy Systems Optimization Standards Committee SABP-A-008
Issue Date: 28 February 2016
Next Planned Update: TBD Load Management for Energy Efficiency: Heat Transfer Equipment
Table 14: Heating Values and Stoichiometric Air Requirement for Common Fuels
* approximate
Page 62 of 84
Document Responsibility: Energy Systems Optimization Standards Committee SABP-A-008
Issue Date: 28 February 2016
Next Planned Update: TBD Load Management for Energy Efficiency: Heat Transfer Equipment
Figure 38: Fuel and Air Distribution Control for Multi-Burner Boilers
Page 63 of 84
Document Responsibility: Energy Systems Optimization Standards Committee SABP-A-008
Issue Date: 28 February 2016
Next Planned Update: TBD Load Management for Energy Efficiency: Heat Transfer Equipment
Source: Baukal, John Zink Combustion Handbook, CRC Press (2001), p566
Page 64 of 84
Document Responsibility: Energy Systems Optimization Standards Committee SABP-A-008
Issue Date: 28 February 2016
Next Planned Update: TBD Load Management for Energy Efficiency: Heat Transfer Equipment
The flow of air and combustion gases through a boiler is achieved by creating draft, defined as the pressure
difference between the flue gases in the furnace section and the atmosphere by one of four methods natural
draft, forced draft, induced draft, or balanced draft. In modern boilers natural draft and induced draft are not
commonly used. In a forced draft system, the fan is on the combustion air supply; in balanced draft
systems, there is an additional induced draft fan is on the hot flue gas at the boiler outlet just prior to
the stack. The furnace is maintained at 0.05-0.10 inches of water (gauge) below atmospheric pressure. The
balanced draft system is standard for modern designs, except for package boilers which use forced draft.
Typically, boiler has its own set of fans (Figure 40a). During turndown operation, therefore, the combustion
air supply from the FD fan has to be regulated with a damper, forcing it to operate at lower than optimum
efficiency. An alternative scheme that minimizes pressure loss across the damper is illustrated in Figure 40b.
The feasibility of such an arrangement depends to a large extent on the physical location of the boilers. If
they are too far apart, it may not be cost-effective to install the required new ductwork for the discharge
header. The analysis, therefore, must be done on a case-by-case basis. An illustrative example is presented in
Figure 41 and Table 15.
Page 65 of 84
Document Responsibility: Energy Systems Optimization Standards Committee SABP-A-008
Issue Date: 28 February 2016
Next Planned Update: TBD Load Management for Energy Efficiency: Heat Transfer Equipment
For ID fans, a similar load management policy would probably not be feasible, as installing a new flue gas
header is unlikely to be practical.
Page 66 of 84
Document Responsibility: Energy Systems Optimization Standards Committee SABP-A-008
Issue Date: 28 February 2016
Next Planned Update: TBD Load Management for Energy Efficiency: Heat Transfer Equipment
The analysis procedure for load management of boiler feedwater pumps is identical. Once again, a new
discharge header will be needed, as illustrated in Figure 42.
Figure 42: BFW Pumps Load Management: (a) 3 running; (b) 2 running
Using such a model, one can easily compare alternative operating policies to meet various process steam
and power demand scenarios, and determine the one that will provide the lowest total operating cost
while still meeting all process requirements and equipment constraints. Three alternative scenarios are
evaluated in Figures 43 through 45.
Page 67 of 84
Document Responsibility: Energy Systems Optimization Standards Committee SABP-A-008
Issue Date: 28 February 2016
Next Planned Update: TBD Load Management for Energy Efficiency: Heat Transfer Equipment
Figure 43: Refinery Steam Balance with Gas Turbines running at Full Capacity
Page 60 of 84
Document Responsibility: Energy Systems Optimization Standards Committee SABP-A-008
Issue Date: 28 February 2016
Next Planned Update: TBD Load Management for Energy Efficiency: Heat Transfer Equipment
Figure 44: Refinery Steam Balance with Gas Turbines Off, and Boilers Equally Loaded
Page 61 of 84
Document Responsibility: Energy Systems Optimization Standards Committee SABP-A-008
Issue Date: 28 February 2016
Next Planned Update: TBD Load Management for Energy Efficiency: Heat Transfer Equipment
Figure 45: Refinery Steam Balance with Gas Turbines Off, and Boilers Optimally Loaded
Page 62 of 84
Document Responsibility: Energy Systems Optimization Standards Committee SABP-A-008
Issue Date: 28 February 2016
Next Planned Update: TBD Load Management for Energy Efficiency: Heat Transfer Equipment
Plant process steam and power requirements are fixed in all cases:
300 Klb/h HP steam, 125 Klb/h MP steam, 75 Klb/h LP steam, and 37 MW.
In scenario 1, all three gas turbines (GTs) are running at their full capacity of 20 MW each, and the HRSGs are
being run without supplementary firing. This leaves only 54 Klb/h of steam deficit, which is made up by
running Boiler #4. Boilers #1 and 2 are kept on hot standby.
In scenario 2a, the gas turbines are all off, and all steam comes from boilers no. 2 and 4, loaded equally
(85% of design capacity). In scenario 2b, the GTs are again off, but Boiler #4 (high efficiency) is loaded
fully, while Boiler #2 (low efficiency) is loaded only 66%. The only difference between scenarios 2a and 2b
is the boiler load allocation. The savings are seen to be about $90,000 per year. While they may represent
only 0.5% of the total site energy bill, they can be achieved at zero capital expense purely by optimal load
management.
Page 63 of 84
Document Responsibility: Energy Systems Optimization Standards Committee SABP-A-008
Issue Date: 28 February 2016
Next Planned Update: TBD Load Management for Energy Efficiency: Heat Transfer Equipment
Fired heaters are used for direct process heating, as opposed to steam generation for indirect process heating.
The reason for using them is that they are able to deliver very high process temperatures typically required
in oil refining processes that are beyond the capability of steam. Normally each fired heater is dedicated
to a particular process heating duty, and it is very rare to see more than two fired heaters in parallel for the
same process duty. If a process duty is too small to justify its own fired heater, a fired heater could be
used to heat up a circulating hot oil loop, which would typically serve several such small high-temp duties;
this is a form of load sharing between one heater and multiple process duties.
The mechanical design of fired heaters is also very different from that of steam boilers. The process fluid is
always inside the tubes; natural draft is the most common type.
Page 64 of 84
Document Responsibility: Energy Systems Optimization Standards Committee SABP-A-008
Issue Date: 28 February 2016
Next Planned Update: TBD Load Management for Energy Efficiency: Heat Transfer Equipment
Many different mechanical arrangements are used, one of the principal considerations being prevention of
damage to the tubes and refractory linings at the high temperatures prevalent in the fire-box. An additional
concern is fouling inside the tubes due to coking.
Page 65 of 84
Document Responsibility: Energy Systems Optimization Standards Committee SABP-A-008
Issue Date: 28 February 2016
Next Planned Update: TBD Load Management for Energy Efficiency: Heat Transfer Equipment
Hot oil loops are normally used only in oil refineries and GOSPs. However, in remote desert areas where
low-cost water supply is not available, they could be a viable alternative to steam boilers as well (eg., Haradh
Gas Plant).
This situation is similar to that for Cooling Water loops (section 4.3). The optimum distribution of LMTD in
each process heater occurs when the hot oil flow is controlled to achieve the desired target temperature of
the hot process stream, which minimizes the power consumption of the circulation pumps. The process duties
are usually arranged in parallel, which is often not the optimum configuration. The optimum network
structure (series/parallel arrangement and sequences) should be determined by Pinch Analysis, and the
appropriate piping and control system modifications should be made to approach the optimum design as
closely as possible.
Page 66 of 84
Document Responsibility: Energy Systems Optimization Standards Committee SABP-A-008
Issue Date: 28 February 2016
Next Planned Update: TBD Load Management for Energy Efficiency: Heat Transfer Equipment
This situation is less common, but does occur, eg. Safaniya Onshore Plants. The basic principles and
engineering approach are best illustrated with an example. Consider the case where there are three desalter
trains at a GOSP two for Arab Heavy crude, and one for Arab Medium crude. All trains were built at
different times, so the designs are not identical. The design includes some process-process heat recovery,
with the remaining feed heating duty being provided by dedicated hot oil loops. The heating load is a
function of flow rate, water cut, and supply temperature from the wells, all of which vary with time.
Step one is to collect PI data for these variables and determine the heat load profile over 1 full year (Figure
50).
Page 67 of 84
Document Responsibility: Energy Systems Optimization Standards Committee SABP-A-008
Issue Date: 28 February 2016
Next Planned Update: TBD Load Management for Energy Efficiency: Heat Transfer Equipment
100
80
Days per year
60
40
20
0
250 300 350 400 450 500 550 600 650
Next, we evaluate the potential savings that can be realized through two complementary load
management strategies - load sharing and load shifting strategies. The basic idea of load sharing is
to see if we can distribute the total process load among the available heaters in a way that
minimizes operating cost. If the different trains are not interconnected on the utility side, then such
jumpover pipes with appropriate controls must be added. This will effectively create a single hot oil
supply system, and provide the flexibility to supply process needs from any one of the operating
heaters. Figure 51 illustrates the piping modifications and additional equipment required for both
strategies. The red lines in represent the new jumpover piping. The green dashed box represents the
changes to the utility (CHP) system required to introduce additional degrees of freedom for load
shifting.
Sample calculations for potential savings from load sharing are presented in Table 18 and 19. For
the sake of simplicity, only two time periods (e.g., summer & winter) have been considered. In
actual practice it would be necessary to perform these calculations for each period over the total
load range shown in Figure 50.
The load shifting approach was described earlier in section 3.1. The idea is to introduce a new heat
exchanger in series which can do a portion of heating duty with a lower cost utility, such as steam.
Typically the marginal cost of effective process heating with steam is about 75-85% that of hot oil,
depending on the steam pressure and the path that it follows through the CHP system. The more
significant benefit of having a steam heater in series, however, is that it can extend the range of
process throughput over which N fired heaters can be operated before starting up the N+1 th. Table
20 presents the estimated savings for this scenario, significantly higher than load sharing alone, though
not without some associated capital cost.
Finally, it should be noted that such a steam heater can be designed to be switchable between LP steam
and HP steam. This extends the run time between shutdowns for HX cleaning.
Page 68 of 84
Document Responsibility: Energy Systems Optimization Standards Committee SABP-A-008
Issue Date: 28 February 2016
Next Planned Update: TBD Load Management for Energy Efficiency: Heat Transfer Equipment
Page 69 of 84
Document Responsibility: Energy Systems Optimization Standards Committee SABP-A-008
Issue Date: 28 February 2016
Next Planned Update: TBD Load Management for Energy Efficiency: Heat Transfer Equipment
Page 70 of 84
Document Responsibility: Energy Systems Optimization Standards Committee SABP-A-008
Issue Date: 28 February 2016
Next Planned Update: TBD Load Management for Energy Efficiency: Heat Transfer Equipment
7.3.3 Burners
The considerations for load management of burners described in section 6.3 for boilers applies for fired heaters
as well. The main difference is that burners in fired heaters tend to be smaller and more numerous (typically
12-24) than in boilers, mainly to provide more accurate control of the temperature profile in the firebox.
Source: Baukal, John Zink Combustion Handbook, CRC Press (2001), p 490
Table 23: Adiabatic Flame Temperatures for Various Fuels with 15% Excess Ambient Air
Source: Baukal, John Zink Combustion Handbook, CRC Press (2001), pp 45, 61
Page 71 of 84
Document Responsibility: Energy Systems Optimization Standards Committee SABP-A-008
Issue Date: 28 February 2016
Next Planned Update: TBD Load Management for Energy Efficiency: Heat Transfer Equipment
One way to get higher temperatures in the fire box (radiant section) is through the use of fuel and air
preheating in the convection section of the furnace. The relationship between air supply temperature and
adiabatic flame can be directly and easily calculated from the combustion-side heat and material balance.
Higher temperatures in the fire box have the advantage of higher capacity due to better heat transfer driving
forces, as explained below. However, there are significant disadvantages and risks as well. One is that NOx
emissions increase exponentially with flame temperature. Coking rates also increase, though not
exponentially. The second more serious disadvantage is that if spot temperatures in the fire box exceed
allowable limits, the heater tubes could rupture, because the mechanical strength of metals drops off
sharply at high temperatures. Nevertheless, the air-preheat control scheme shown in Figure 53 offers an
opportunity to increase the capacity of the fired heater, at least on a short term basis. By enabling a furnace to
temporarily exceed its nominal design capacity, the need to start up a second furnace can be avoided.
The energy efficiency of the heater is also improved, as seen from Table 24. Maintaining the proper
distribution of fuel and air to each burner is critically important. See section 6.3.1, especially Figure 38.
Figure 52: Adiabatic Flame Temperature vs Air Preheat (for Stoichiometric Air)
Page 72 of 84
Document Responsibility: Energy Systems Optimization Standards Committee SABP-A-008
Issue Date: 28 February 2016
Next Planned Update: TBD Load Management for Energy Efficiency: Heat Transfer Equipment
Figure 53: Furnace Capacity Control System Based on Combustion Air Preheat
Page 73 of 84
Document Responsibility: Energy Systems Optimization Standards Committee SABP-A-008
Issue Date: 28 February 2016
Next Planned Update: TBD Load Management for Energy Efficiency: Heat Transfer Equipment
Observe in Table 24 that when operating at lower than design capacity, no air preheating is needed, as the
required firebox temperature to accomplish the process heating duty is lower than would be obtained with
the design excess air ratio. In such a situation, the excess air ratio should be increased (to 20%, in this
example) in order to avoid exceeding the target process temperature (eg. to prevent fouling or evaporation).
Normally only a single air supply fan is used for each heater. If there multiple heaters are located in close
proximity, however, the introduction of a header system similar to that shown in Figure 40 could open up
some opportunity for energy savings through load management. One has to be careful though that the
required air supply pressures and fan performance characteristics are compatible.
Page 74 of 84
Document Responsibility: Energy Systems Optimization Standards Committee SABP-A-008
Issue Date: 28 February 2016
Next Planned Update: TBD Load Management for Energy Efficiency: Heat Transfer Equipment
Normally, only a single fuel supply pump is used even for multiple parallel heaters (although with an
installed spare for emergencies), so the question of load management does not arise.
The addition of a WHB at the back end of a fired heater or thermal oxidizer (as in Claus sulfur plants)
introduces a new degree of freedom, and can have significant implications for load management of the
comprehensive CHP system, as described in Section 6.4.
Page 75 of 84
Document Responsibility: Energy Systems Optimization Standards Committee SABP-A-008
Issue Date: 28 February 2016
Next Planned Update: TBD Load Management for Energy Efficiency: Heat Transfer Equipment
When the convection coil (or even an external HX) is used for steam generation as opposed to air preheating,
furnace efficiency can be maintained at a high level even under low process load conditions, because any heat
that is not absorbed by the process fluid will be recovered as steam, which in turn will help to back off boiler
fuel.
Revision Summary
12 March 2011 Reaffirmed the contents of the document, and reissued with editorial changes.
21 July 2013 Editorial revision to change document responsibility name from P&CSD/Energy Systems
Unit to P&CSD/Energy Systems Division.
31 January 2016 Minor revision to convert the naming of tables and charts from exhibits to table and figure.
Page 76 of 84