You are on page 1of 15

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/257336877

An ISA-95-based manufacturing intelligence system in support of lean


initiatives

Article  in  International Journal of Advanced Manufacturing Technology · March 2012


DOI: 10.1007/s00170-012-4223-z

CITATIONS READS

24 1,828

1 author:

Hakki Ozgur Unver


TOBB University of Economics and Technology
39 PUBLICATIONS   142 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

A Virtual Factory Framework for SMEs View project

Deep Learning for Intelligent Manufacturing View project

All content following this page was uploaded by Hakki Ozgur Unver on 02 July 2016.

The user has requested enhancement of the downloaded file.


Int J Adv Manuf Technol (2013) 65:853–866
DOI 10.1007/s00170-012-4223-z

ORIGINAL ARTICLE

An ISA-95-based manufacturing intelligence system


in support of lean initiatives
Hakki Ozgur Unver

Received: 17 November 2011 / Accepted: 3 May 2012 / Published online: 17 May 2012
# Springer-Verlag London Limited 2012

Abstract Increasingly, lean manufacturing is being applied 1 Introduction


by leading manufacturers throughout the world. As contin-
uous improvement cycles of many lean initiatives focus on Manufacturing is one of the most important value-adding
cost control and improving quality of product, turbulence in pillars of developed and emerging economies. In Europe,
world markets demand more agility and responsiveness manufacturing represents 22 % of gross domestic product
without compromising cost and quality. In order to attain (GDP), and it is estimated that 75 % of GDP and 70 % of
more agility, information and communication technologies employment are somehow related to manufacturing [1].
are utilized by many manufacturers, both at shop floor However today’s markets are very volatile, unpredictable,
systems and enterprise resource planning (ERP) layer. This and subject to intense turbulence. The accelerating pace of
increasing trend created a disconnect that presents an op- innovation, reduced product life cycles, increased product
portunity for manufacturing intelligence (MI) systems. proliferation and various stakeholder pressures for cleaner
Bridging this gap, MI can enhance responsiveness by pro- production create even more uncertainty.
viding visibility into operations and improve quality by Over the last several decades, tremendous progress has
tracking long-term data, hence support the continuous im- been made in manufacturing by use of Information and
provement philosophy of lean manufacturing. This paper Communication Technology (ICT). In Western companies,
presents an ISA-95-based MI framework that can support interest in material requirements planning (MRP) skyrock-
lean manufacturing by contextualizing low-level shop floor eted in the 1970s, a phenomenon known as the “MRP
data using production operation information from ERP sys- crusade” [2]. In the 1980s and 1990s, the technology
tems. Processed data is presented on dashboards via Key evolved into MRP II and to Enterprise Resource Planning
Performance Indicators, which managers can use to deter- (ERP) software provided by giant industry vendors such as
mine appropriate action for their lean initiatives, timely and Oracle, IBM, and SAP. Despite these advances, ERP soft-
effectively. ware integration with shop floor execution and automation
systems remained weak.
Keywords Manufacturing intelligence . Lean In contrast, during the same time period, Japanese com-
manufacturing . ANSI/ISA-95 panies focused on developing lean methods, preferring in-
expensive automation and relatively simple manufacturing
technologies [3]. As a production management philosophy,
in USA the lean concept was first introduced in 1990 by
Womack et al. with their famous book, The Machine that
Changed the World [4]. While describing the global success
H. O. Unver (*) of Japanese vehicle manufacturers, in particular Toyota, the
Mechanical Engineering Department, authors analyzed the rise and fall of mass production in the
TOBB University of Economics and Technology,
twentieth century. In doing so, they defined lean thinking as
Sogutozu,
Ankara 06560, Turkey specifying value while eliminating waste [4]. The five key
e-mail: hounver@etu.edu.tr principles are: (a) specify value, (b) identify value streams,
854 Int J Adv Manuf Technol (2013) 65:853–866

(c) make value flow without disruption, (d) allow customers intelligence (BI) tools and analytics can turn real-time data
to pull value, and (e) pursue perfection [5]. The ultimate into valuable knowledge that allows managers to optimize
goal is to eliminate seven deadly wastes (i.e., overproduc- manufacturing operations.
tion, waiting, transportation, extra processing, inventory, In this paper, we propose a framework for a manufactur-
motion, and defects) and to continuously pursue perfection. ing intelligence system, namely manufacturing operations
Over the last three decades, companies have imple- center (MOC), that incorporates the necessary tools that
mented many systems and solutions with the goal of elim- both support the continuous improvement philosophy of
inating excess inventory and waste from the supply chain. lean manufacturing and responsiveness in operations, while
After the success of many lean strategies, the focus now has providing ERP shop floor integration based on the Interna-
shifted to improving agility in order to adapt to new tional Society of Automation’s ISA-95 standard. This
demands in the marketplace. While lean manufacturing framework is the result of 2 years of work at Oracle, Inc.,
focuses mostly on low cost, quality assurance, and continu- and is based on state-of-the-art BI technologies such as
ous improvement, agile manufacturing aims to complement Oracle Business Intelligence Enterprise Edition (OBIEE),
these concepts with responsiveness and flexibility with in- Oracle Warehouse Builder (OWB), and Oracle Database
tensive use of ICT [6, 7]. However, use of extensive ICT 10 g. MOC was developed in close cooperation with shop
created a gap between manufacturing shop floor sys- floor hardware and software vendors such as Mitsubishi
tems and back office ERP systems (Fig. 1). Although Corp, Kepware, and ILS technologies.
many believe lean manufacturing and ICT to be mutually
exclusive [8], integrating these concepts geared to fill this
gap, presents an enormous opportunity to improve visibil- 2 Using business intelligence for analytics and decision
ity, responsiveness, and agility while supporting continu- making
ous improvement, in a lean, and demand-driven supply
chain. In business, the term “analytics” refers to using data and
Lean philosophy incorporates numerous tools and tech- systematic reasoning to make decisions. Managers utilize
niques. Shah and Ward present a comprehensive review of many tools for analytical decision making and apply them to
lean practices that can support a corporate lean transforma- tasks ranging from statistical analysis to data mining [11]. By
tion, including Single Minute Exchange of Dies (SMED), employing analytical reasoning, organizations can answer
Kanban, Six Sigma, Value Stream Mapping, 5S, Total Qual- fundamental questions about their business. Using a matrix,
ity Management, Just in Time, Total Productive Mainte- Davenport et al. identified the six key questions that data and
nance (TPM), and Visual Management (Andon), among analytics can address in an enterprise (Fig. 2) [11]. In the top
others [9]. A lean planning system can document the impact row of the matrix, the past information cell represents tradi-
of transformation on selected lean metrics, or performance tional business reporting, the present information cell high-
measures for lean manufacturing. Examples include overall lights current monitoring activities, and the future information
equipment effectiveness (OEE), process throughput, total cell represents how the future can be forecasted using extrap-
manufacturing lead time, etc. [10]. However, when tracking olations of past patterns. In the bottom row, the past insight
these metrics over time, manufacturing intelligence (MI) can cell highlights how insight into the past can be gained through
be a powerful tool to create the visibility to improve respon- statistical modeling, the present insight cell represents how
siveness and long term data necessary for continuous im- data can be translated into action (e.g., a price discount rec-
provement efforts within a production facility. Business ommendation based on overstocked items), and the future

Fig. 1 Disconnection between


enterprise and shop floor
systems
Int J Adv Manuf Technol (2013) 65:853–866 855

Fig. 2 Key questions Past Present Future


addressed by analytics [9]

What happened? What is happening What will happen?


Information now?

(Reporting) (Alerts) (Extrapolation)

Howandwhy did it What’sthe next What’sthe best/worst


happen? bestaction? that canhappen?
Insight
(Modeling, (Recommendation) (Prediction,
experimentaldesign) optimization,simulation)

insight cell represents of the use of what-if scenarios to deter- 4. Data mining, the process of sorting through large
mine the most feasible choices. amounts of data and identifying relevant information
In the early 1990s, an analyst at the Gartner Group, Howard by seeking patterns that can aid decision making [13];
Dressner, coined the term “Business Intelligence (BI)”, which 5. Extract, load, transform (ETL), the process of extracting
is now widely used by both practitioners and scholars [12]. BI data from various source systems before storage in a
is defined as the information that people use to support their final data warehouse, typically comprised of validation,
decision making efforts, and encompasses three broad activi- transformation by application of business rules and
ties: (a) collecting information, (b) discerning patterns and cleaning up dangling data [16]; and
meaning in the information, and (c) presenting and responding 6. Dashboards and scorecards, which provide comprehen-
to the resultant information [13]. Most basic BI activities are sive visual representations of corporate performance
represented by the top row of Fig. 2. measures (i.e., Key Performance Indicators, or KPIs),
Although the concept of BI has existed for three decades, trends and exceptions [17] and integrate information
the rise of some key technologies in the 1990s accelerated from multiple business areas, enabling sources of under-
its adoption by many industries, including manufacturing. performance to be isolated by using drill down capabil-
Key technologies include: ities across hierarchical levels of an organization.
1. Enterprise application integration, modeling tools for At the heart of modern business intelligence technologies, a
integration of information systems managing manufac- concept called “Dimensional Modeling” lies. Dimensional
turing or service activities, such as ARIS, CIMOSA, modeling is a logical design technique used with data ware-
TOVE etc.[14]; house technology that attempts to present denormalized data
2. Data warehouses, repositories for electronically stored in an intuitive standard framework to enable high performance
organizational data designed to facilitate reporting and processing. Every dimensional model is composed of one
analysis [15]; table called the fact table and a set of smaller tables called
3. Online analytical processing, an approach that quick- dimension tables [18]. Each dimension table has a single part
ly provides answers to multidimensional analytical primary key that corresponds exactly to one of the compo-
queries [15]; nents of the multipart key in the fact table (Fig. 3).

Fig. 3 Fact and dimension


tables for manufacturing asset
performance
856 Int J Adv Manuf Technol (2013) 65:853–866

3 Manufacturing intelligence based on ISA-95 and quality systems [24]. Based on ISA-95.02, this infor-
mation is structured in UML models, which are the basis for
The term MI is coined by AMR research in order to scope developing standard interfaces between ERP and Manufac-
emerging BI applications in manufacturing and operations turing Execution Control (MES) systems. ISA-95.03
that can provide a multisite view of production performance addresses activity models of manufacturing operations man-
in terms of KPIs on multilayer dashboards [19]. MI systems agement and provides terminology and a consistent set of
can connect to and extract data from a highly diverse set of models for exchanging and processing production informa-
sources ranging from shop floor instruments, to historians tion. Figure 4 presents functional hierarchy enterprise and
(temporal data stores), to operational data stores and other production control system levels as defined by ISA-95, and
relational stores. Important MI capabilities are: (1) aggrega- the scope of each level.
tion of data from a variety of real-time and back-end, ERP data Real-time production information is crucial to the daily
sources; (2) contextualization of critical data elements from manufacturing operations and its availability will improve
such floor such as process variables against ERP level data; agility of lean systems. Manufacturing intelligence can pro-
(3) analysis of data by calculating a range of KPIs using raw vide visibility into the lean operations within production
process performance and cost-based information from ERP; facility, analyze the real-time data, and turn them into ac-
and (4) visualization by providing an intuitive graphical rep- tionable insight for driving continuous improvement. Posi-
resentation of intelligence based on dashboards, enabling tioned at the space between ERP systems (level 4) and
users to drill down from multiplant representations to auto- distributed plant systems (level 3) MI systems can assess
mated shop floor machines as required [20]. numerous conditions, deliver information in as close to “real
Although manufacturing intelligence continues to grow time” as possible, consider the implications of alternative
and gets attention from vendors all over the world, internal response scenarios, and implement corrective actions for
MI architectures still lacks standardization [21]. As multiple continuous improvement. This ability requires a common
vendors exist at both ERP and shop floor layer, there is a plant information repository, access to resources, processes
desperate need for standardization at MI segment. In order and transactions and the ability to contextualize the data.
to help to close this divide, the ISA Committee published Today by use of BI technologies and ERP to shop floor
ANSI/ISA-95 standard for developing automated commu- layer data integration standards such as ISA-95, it is possi-
nication between enterprise planning and shop floor control ble to create manufacturing intelligence solutions which can
systems. The major contribution of the ISA-95 standard is mitigate most manufacturers top challenge; lack of visibility
the formalization of interactions of the manufacturing sys- and improving responsiveness of lean operations. This new
tem to other business processes of the company [22]. In space for manufacturing intelligence solutions within ISA-
other words, the standard aims to specify data flows and 95 hierarchy is depicted in Fig. 5. Even companies that have
interfaces between enterprise business systems and manu- not embraced MES can collect data from the shop floor.
facturing control systems using enterprise modeling techni- Production and performance data can be directly collected
ques [23]. ISA-95.01 describes models and terminology that from automation equipment, often aggregated and contex-
are used to determine which information must be exchanged tualized targeted for various recipients. Manufacturing In-
between sales, finance, logistics, production, maintenance, telligence solutions can leverage web services, SOA, as well

Fig. 4 Production functional


hierarchy as defined in ISA-95
[35]
Int J Adv Manuf Technol (2013) 65:853–866 857

as BI, to rationalize information and present vital, real-time into manufacturing operations and asset performance. It
dashboards containing KPIs. solves the complex problem of connecting fragmented, dis-
As a cutting edge MI solution, MOC was developed by a connected shop floor data to the business context of back
team at Oracle Inc. based on ISA-95 standard, within 2 years office systems. This combination delivers real-time moni-
of effort. The next section describes its architecture, data toring and analysis of shop floor operations—a foundation
model, and features in detail. for running continuous improvement programs in lean ini-
tiatives. The main components of the MOC system archi-
tecture is presented in Fig. 6.
4 Manufacturing operations center MOC has a data model based on the ISA-95 industry
standard and extensible attribute framework. It includes: (a)
We developed MOC as a standalone manufacturing intelli- prebuilt extensible adapters for MES and ERP systems and a
gence solution using an ISA-95 data model distinct from prebuilt communication protocol between leading hardware
ERP, as it is designed to be ERP agnostic. The solution vendors; (b) a robust contextualization rules engine to con-
integrates and creates common ground between the periodic vert raw shop floor data into meaningful business informa-
and transactional ERP world, and the real-time, fragmented tion by merging data from ERP systems; and (c) prebuilt
manufacturing systems world. Real-time data matched with KPIs and analytics, delivered in easy to configure BI tech-
the ERP-based business context enables MOC to deliver nology, namely OBIEE.
visibility, sophisticated operational analytics, and standard-
ized integration between the enterprise and the shop floor. 4.2 Standards-based open data model
MOC is meant to be used as a unified plant data repos-
itory and is not a substitute for another discrete or process The core component of MOC is a generic, open, and extensi-
MES, BI application, middleware, or data historian. The ble data model that can meet the requirements of diverse
system uses integration infrastructure provided by Fusion industries. The data model is based on the ISA-95 reference
Middleware to gather data from data sources such as an model and provides a hierarchical structure for reporting KPIs
MES or quality application. In order to collect real-time that can be analyzed along different dimensions (Fig. 7). The
data from plant equipment and control systems, key partners data model is generic enough to support and fit the distinct
such as Kepware, ILS technologies, and Matrikon provide needs of discrete, flow, and process manufacturing operations.
gateways to MOC [25]. Collected data is processed and
contextualized by Oracle Data Warehouse 10 g, and pre- 4.3 ERP data adaptor
sented in role-based dashboards to plant managers and pro-
duction supervisors as specific KPIs. MOC uses OWB to create data integration maps for extract-
ing data from ERP systems related to organizations, shifts,
4.1 System architecture items, item categories, item costs, resources, resource
groups, resource costs, released work orders/flow schedules,
MOC addresses the need for accurate and timely informa- operations, and material and resource requirements. The
tion about product and process quality, and provides insight integration scheme is illustrated in Fig. 8. In this case, the

Fig. 5 Creating manufacturing


visibility based on ANSI/ISA-
95
858 Int J Adv Manuf Technol (2013) 65:853–866

Fig. 6 MOC system


architecture

ERP adaptor simplifies integration, as the MOC handles communicate with devices. To ensure coverage for different
data cleansing and errors. types of devices, we partnered closely with Kepware, ILS
technologies, and Mitsubishi automation. Kepware and ILS
4.4 Shop floor connectivity built a large library of device drivers for various types of
hardware found on the shop floor. In addition to their
MOC is designed to aggregate real-time data directly from standard OPC servers and clients, Kepware built additional
control systems such as programmable logic controllers capabilities such as complex tags that deliver tighter inte-
(PLCs), supervisory control and data acquisition, and dis- gration to MOC (Fig. 9).
tributed control systems on the shop floor. MOC primarily However, to increase flexibility, we also created an API and
utilizes an object process control (OPC)-based connection to interface table for customers who may want to build or utilize

Fig. 7 ISA-95-based data model


Int J Adv Manuf Technol (2013) 65:853–866 859

Fig. 8 ERP data adaptor

their own connectivity infrastructures on the shop floor. In data along multiple levels of an organization or facility [26].
terms of error handling, MOC provides an infrastructure for MOC has 55 predefined KPIs on 14 dashboards: (a) asset
capturing and processing errors at every stage of data collec- performance (OEE), (b) asset performance (OEE) by equip-
tion. In addition, the OWB tool set allows users to define ment, (c) equipment downtime analysis, (d) equipment
custom business rules for filtering MOC data. downtime reasons, (e) production slippage trend, (f) produc-
tion loss analysis, (g) production loss detail, (h) equipment
4.5 Device data management and contextualization efficiency analysis, (i) equipment scrap analysis, (j) equip-
ment scrap reasons, (k) batch performance, (l) batch perfor-
One of the core contributions of this work is the ability to mance detail, and (m) production performance. A full list of
convert highly granular tag data from devices into meaningful KPIs and metrics related to manufacturing asset perfor-
data for business users such as plant and operations managers, mance is given in Table 1.
as well as quality and manufacturing engineers. MOC has a
contextualization engine with the functionality to define busi-
ness meanings and processing rules for various types of tag 5 Use case I: OEE and analysis of production loss
data received from PLCs and automation equipment (Fig. 10).
Four basic operational contexts can be added to raw data First use case of MOC is implemented to demonstrate the
received from a PLC or automation device: shift, work support to a TPM initiative. The TPM concept was launched
order, work order segment, and product. Contextualization by Seiichi Nakajima, vice-chairman of the Japanese Institute
methods include: (a) tag-based, where tags are grouped, of Plant Engineers, in the 1980s [27]. TPM describes a
assigned an identifier, and contextualized into one of the synergistic relationship among all organizational functions,
four contexts mentioned above; (b) schedule-based, where but particularly between production and maintenance, for
tags are grouped based on the equipment production sched- the continuous improvement of product quality, operational
ule; and (c) manual, where users manually update contexts efficiency, capacity assurance, and safety [28]. This strategic
when tags are received from PLCs. lean tool offers a complete maintenance methodology, em-
phasizing the overall effectiveness and utilization of equip-
4.6 Dashboards and KPIs ment through the active participation of equipment workers
in the setting of teamwork [29].
Dashboards usually include a combination of reports, charts TPM attacks “six big losses”, draining productivity,
and KPIs which provide drill down capabilities to analyze which consists of breakdowns, set-up/adjustment loss,
Fig. 9 Open shop floor
connectivity
860 Int J Adv Manuf Technol (2013) 65:853–866

Fig. 10 Device data


contextualization

idling/minor stoppages, reduced speed, defects/rework, and down in order to investigate the reasons for low OEE
start-up and yield loss [29, 30]. The reduction of these losses performance. For instance, if availability is the lowest
increases OEE the most popular numerical metric in order to OEE component for Mixer1201, he can browse through
evaluate TPM initiatives. [31, 32]: each downtime period of Mixer1021 and investigate
downtime reasons.
OEE ¼ availability  performance  quality
The scenario described above can be realized by use of
Where contextual data supplied from ERP and real-time shop floor
data collected from automation equipment. In this case,
Actual available time
Availability ¼ Planned_available_time is ERP level data supplied from
Planned available time planning system, where as Actual_available_time, Effecti-
ve_run_time, Good_quantity_produced, and Total_quanti-
Effective run time ty_produced are collected from shop floor automation and
Performance ¼ control systems. Further detailed loss analysis is also possi-
Actual available time
ble through a series of screens. During a certain calendar
time (i.e., 24 or 8 h shift), for each equipment user can
Good quantity produced examine contribution of production loss type (e.g., un-
Quality ¼ planned downtime, inefficiency, bad quality) in pie chart
Total quantity produced
format, as well as tabular data (Fig. 13).
Availability captures deleterious effects due to break- As demonstrated in this use case, benefits of data inte-
downs, setups and adjustments, performance captures pro- gration from ERP level with shop floor level manifests itself
ductivity loss due to reduced speed, idling, and minor as improved visibility to performance measures of processes
stoppages, and Quality is the yield of acceptable product and also ability to support continuous improvement by
that captures loss due to defects, rework, and yield facilitation of deep analysis of root cause of problems.
(Fig. 11) [33].
MOC OEE dashboards presents site level aggregated
OEE calculations for a multisite manufacturing compa- 5.1 Use case II: shop floor visibility
ny (Fig. 12). In the given scenario, at the highest level
when plant manager observes that Seattle manufactur- In lean environments, the workers have the authority to stop
ing site overall OEE is close to red zone (75 %), in the production line when they detect defects to work imme-
order to investigate the detail, he can drill down to diately in their solution and attack the problem at its roots
equipment level. At this level, worst five performing [34]. This practice known as “Andon” in Japanese, means
equipment will be available for investigation. He can lantern covered with paper. In a classic lean system, a
check when this trend has started by using trend lantern would lit when a worker pull the trigger, when a
graphs, numerically see components of OEE for worst problem occurs, and line will stop. ICT enable lean produc-
performing equipment. As presented in Fig. 12, if plant tion environment to display real-time information in the
manager decides to investigate the worst performer work cells and in the plant in order to everybody know the
which is Mixer1201 in this case, he can further drill status of processes at a given time. Real time and historic
Int J Adv Manuf Technol (2013) 65:853–866 861

Table 1 List of MOC KPIs

Metric Description

Actual available time (hours) Actual available time for the equipment after accounting for planned and unplanned
downtimes
Actual down time (hours) Unplanned equipment down time
Actual run time (hours) Equipment run time
Effective run time (hours) Effective equipment run time after taking machine efficiency into account
Equipment calendar time (hours) Equipment calendar time, in hours
Equipment first time good quantity Equipment output quantity that has not been reworked or scrapped
Equipment output quantity Total equipment output quantity, including acceptable and unacceptable outputs
Equipment scheduled time (hours) Scheduled available time based on equipment shift availability
Equipment standard efficiency Standard efficiency of equipment, expressed as a percentage
Equipment standard utilization Standard utilization of equipment, expressed as a percentage
First pass yield Percentage of total production that is acceptable the first time, without rejections
or scrap
Machine availability ratio Actual time for which equipment was available for an operation
divided by calendar time,
based on the site calendar
Machine efficiency Actual rate of production compared to the ideal rate of production
Output standard rate –
Overall effective run time (hours) Effective run time after deducting for bad quality
Overall equipment effectiveness –
Performance ratio Ratio of effective run time to actual available time
Production loss—planned (%) –
Production loss—planned (hours) –
Production loss—planned (value) –
Production loss due to bad quality (%) Unplanned equipment production loss due to bad quality, expressed as a percentage
Production loss due to bad quality (hours) Unplanned equipment production loss due to bad quality, expressed in hours
Production loss due to bad quality (value) Unplanned equipment production loss due to bad quality, expressed in cost terms
Production loss due to inefficiency (%) Unplanned equipment production loss due to machine inefficiency compared to
the ideal production rate, expressed as a percentage
Production loss due to inefficiency (hours) Unplanned equipment production loss due to machine inefficiency when compared
to the ideal production rate, expressed in hours
Production loss due to inefficiency (value) Unplanned equipment production loss due to machine inefficiency when compared
to the ideal production rate, expressed in cost terms
Production loss due to other reasons (%) Unplanned equipment production loss due to reasons other than downtime, efficiency
or bad quality, such as lack of demand or material, expressed as a percentage
Production loss due to other reasons (hours) Unplanned equipment production loss due to reasons other than downtime, efficiency
or bad quality, such as lack of demand or material, expressed in hours
Production loss due to other reasons (value) Unplanned equipment production loss due to reasons other than downtime, efficiency
or bad quality, such as lack of demand or material, expressed in cost terms
Production loss due to unplanned downtime (%) Unplanned equipment downtime due to breakdowns or similar reasons, expressed
as a percentage
Production loss due to unplanned downtime (hours) Unplanned equipment downtime due to breakdowns or similar reasons, expressed
in hours
Production loss due to unplanned downtime (value) Unplanned equipment downtime due to breakdowns or similar reasons, expressed
as resource costs
Resource standard efficiency Standard resource efficiency, expressed as a percentage
Resource standard utilization Standard resource utilization, expressed as a percentage

information such as Equipment_status, Scheduled_quantity, in order to inform everyone in plant about current status and
Completed_quantity, Scrap_quantity, and rates such as Pro- improve responsiveness to events such machine breakdown,
duction_rate or Catch_up_rate can be displayed at real-time a quality problems or schedule change.
862 Int J Adv Manuf Technol (2013) 65:853–866

Fig. 11 OEE and production


loss analysis

In order to demonstrate the improvement of real-time an attached e-Factory module to simulate shop floor hard-
visibility which also enhances responsiveness, in a lean ware. The e-Factory module, developed by ILS Corp., can
plant by MOC, we set up a laboratory system which per- send all PLC tag readings to the MOC data warehouse via a
forms real-time data collection and contextualization. The TCP/IP Ethernet port. An HMI screen simulated six ma-
system includes five distributed components: (a) a Mitsu- chine tools, two boring machines, a drilling machine, a CNC
bishi PLC and Human Machine Interface (HMI) screen, (b) machining center, a grinding machine, and an extruder. For
a Linux server running Oracle Database Server 10 g, (c) a each machine tool, it is possible to download work orders
Linux server running OWB, (d) a laptop running ILS- from the ERP system, perform a simulated run of each work
Devicewise workbench software, and (e) a demo laptop order, and monitor machine tool statuses such as UP, IDLE,
running OBIEE. or DOWN. Errors can be generated randomly, which put the
The shop floor validation system configuration is pre- machines in downtime, scrap some quantities in the work
sented in Fig. 14. A Mitsubishi PLC controller is used with order while completing the rest.

Fig. 12 OEE analysis, drill down scenario


Int J Adv Manuf Technol (2013) 65:853–866 863

Fig. 13 Production loss analysis

4. Process Tag Raw Data

2 - Write Tag Data to Loadto Status/Output


MOC Tag Raw Table And Summary table

Oracle DB 10g OWB


Schedule PF
1 -Input to run automatically
Completion Qty
Scrap Qty
Status

3
Scan
IP Address

5. View Dash board


In OBIEE
Devicewise
Workbench
Raw Table
Created in Demo Laptop

Router Demo Laptop


User

Work Order
Item
Planned Qty View Table
Created in Demo Laptop

Fig. 14 Setup configuration for real time data collection


864 Int J Adv Manuf Technol (2013) 65:853–866

Fig. 15 Dashboards and KPIs for shop floor visibility


Int J Adv Manuf Technol (2013) 65:853–866 865

In Fig. 15, a sample screen which displays this production References


line made up of six machine tools. For each machine tool,
Current_equipment_status, Last_status_change_time are dis- 1. Manufuture (2006) Strategic research agenda, manufuture-eu.
played as status information. Total_scheduled_quantity, www.manufuture.org
Scheduled_rate, Standard_rate are metrics pulled from ERP 2. Hopp WJ, Spearman ML (2008) Factory physics, 3rd edn.
layer, contextualized from WorkOrders scheduled for selected McGraw-Hill, Irwin
3. Riezebos J, Klingenberg W, Hicks C (2009) Lean production and
shift. Total_completion_quantity, Actual_production_rate are information technology: connection or contradiction? Comput Ind
metrics collected from devices. Then, Catch_up_rate can be 60:237–247
calculated by use of contextualized Scheduled_rate and 4. Womack JP, Jones DT, Roos D (1990) The machine that changed
Actual_production_rate for a specific shift. A separate server the world, the story of lean production. Harper Perennial, New
York
which runs OWB, an ETL tool to execute necessary process 5. Womack JP, Jones DT (2003) Lean thinking, banish waste and
flows for data contextualization. Further by use of available create wealth in your corporation. Free Press
metrics, breakdown bar chart view of equipment status during 6. Yusuf YY, Adelye EO (2002) A comparative study of lean and
the shift and graphical trends are displayed such as Equi- agile manufacturing with a related survey of current practices in
the UK. Int J Prod Res 40(17):4545–4562
pment_status_trend and Equipment_output_trend. A red ban- 7. Gunasekaran A, Yusuf YY (2002) Agile manufacturing: a taxon-
ner displayed on the top, informs about equipments with an omy of strategic and technological imperatives. Int J Prod Res 40
alert status and in need of immediate attention by the staff to (6):1357–1385
resolve pending issues. 8. Thomas H (2005) ICT and lean management: will they ever
get along? MPRA Paper no: 2502. http://mpra.ub.uni-
muenchen.de/2502/. Accessed 1 Oct 2011
9. Shah R, Ward PT (2003) Lean manufacturing: context, practice
bundles, and performance. J Oper Manag 21:129–149
6 Conclusion
10. Cottyn JC, Van Landeghem HV, Stockman K, Derammelaere S
(2011) A method to align a manufacturing execution system with
In this article, an ISA-95 based manufacturing intelli- lean objectives. Int J Prod Res 49(14):4397–4413
gence system, namely manufacturing operations center 11. Davenport TH, Harris JG, Morison R (2010) Analytics at work,
smarter decisions, better results. Harvard, Boston
has been presented. My colleagues and I developed an
12. Watson HJ, Wixom BH (2007) The current state of business
integrated architecture model with hardware and software intelligence. Computer 40(9):96–99
infrastructure that can support continuous improvement 13. Baltzan P, Phillips A (2009) Essentials of business driven infor-
in lean manufacturing and enhance responsiveness by mation systems. McGraw-Hill, New York
14. Liu X, Zhang WJ, Radhakrishnan R, Tu YL (2012) Manufacturing
providing real-time visibility into shop floor operations.
perspective of enterprise application integration: the state of the art
Using advanced BI technologies, the main contribution review. Int J Prod Res 46(16):4567–4596
of this study lies in the contextualization of low-level 15. Gang T, Kai C, Bei S (2008) The research and application of
shop floor data with ERP level data to dynamically business intelligence system in retail industry. Proceedings of the
IEEE International Conference on Automation and Logistics,
generate KPIs. To illustrate MOC capabilities, two use
Qingdao, China, pp. 87–89
cases are presented. First use case examines demonstra- 16. Kimball R, Ross M, Thornthwaite W, Mundy J, Becker B (2008)
tion of OEE metric in a TPM environment using the drill The data warehouse lifecycle toolkit, 2nd edn. New York, Wiley
down capabilities of OBIEE in order to find the root 17. Turban E, Aronson JE, Liang TP, Sharda R (2007) decision support
and business intelligence systems, 8th edn. Pearson, New York
source of inefficiency in a single OEE component of a
18. Kimball R, Reeves L, Ross M, Thornthwaite W (1998) The data
machine tool. Second use case, aims to improve shop warehouse lifecycle toolkit, expert methods for designing, devel-
floor visibility by a validation setup that demonstrates oping, and deploying data warehouses. Wiley, New York
the real-time data collection capability of MOC and how 19. Smith A (2008) From EMI to operations intelligence, part 1:
business leaders seek key performance drivers, manufacturing
data can be transformed into dashboard KPIs to improve
operations 2008 landscape series, AMR Research
responsiveness. Future work remains to include other 20. Smith A (2008) From EMI to operations intelligence, part 2:
lean performance measures such as value-added/non-val- choosing the best mix of applications for your manufacturing
ue added time, work in process, process throughput, scenario, Manufacturing Operations 2008 Landscape Series,
AMR Research
manufacturing lead time, set-up time, labor productivity
21. Smith A (2008) Enterprise manufacturing intelligence architec-
etc. Further system can be extended into sustainability tures still a work in process, alert article, AMR Research
space by adding metrics and extending the data model to 22. Ugarte BSD, Artiba A, Pellerin R (2009) Manufacturing execution
incorporate energy consumption and greenhouse gas system—a literature review. Prod Plan Control 20(6):525–539
23. Chen D (2005) Enterprise-control system integration—an interna-
emissions of manufacturing systems.
tional standard. Int J Prod Res 43(20):4335–4357
24. ISA-95 Enterprise Control Systems. www.isa-95.com. Accessed 1
Acknowledgments The author is sincerely thankful to Amit Singh, Oct 2011
Ram Palaniappan, and Andy Binsley of Oracle Inc. for their valuable 25. Jacobson S (2009) Oracle’s MOC: enhance value of production
contributions to this work. environment longevity of EBS, AMR Research
866 Int J Adv Manuf Technol (2013) 65:853–866

26. Oracle Business Intelligence Suite Enterprise Edition, an Oracle 31. Muthiah KMN, Huang SH, Mahadevan S (2008) Automating
White Paper (2006) factory performance diagnostics using overall throughput effec-
27. Nakajima S (1988) Introduction to TPM: total productive mainte- tiveness (OTE) metric. Int J Adv Manuf Technol 36:811–824
nance. Productivity Press, Cambridge 32. Mcadam R, Duffner A (1996) Implementation of total productive
28. Sun H, Yam R, Wai-Keung N (2003) The implementation and maintenance in support of an established total quality programme.
evaluation of total productive maintenance (TPM)—an action case Total Qual Manag 7(6):613–630
study in a Hong Kong manufacturing company. Int J Adv Manuf 33. Huang SH, Dismukes JP, Shi J, Su Q, Razzak MA, Bodhale R et al
Technol 22(3–4):224–228 (2003) Manufacturing productivity improvement using effective-
29. Chong MY, Chin JF, Hamzah HS (2012) Transfer of total produc- ness metrics and simulation analysis. Int J Prod Res 41(3):513–527
tive maintenance practice to supply chain. Total Qual Manag Bus 34. Duque DFM, Cadavid LR (2007) Lean manufacturing measure-
Excell. doi:10.1080/14783363.2011.637788 ment: the relationship between lean activities and lean metrics.
30. Brah SA, Chong W-K (2004) Relationship between total produc- Estudios Genrenciales 23(105):69–83
tive maintenance and performance. Int J Prod Res 42(12):2383– 35. American National Standard (2010) ANSI/ISA-95.00.01-2010 En-
2401 terprise control system integration—Part 1: models and terminology

View publication stats

You might also like