Analysis of performance measures of flexible manufacturing ... · Abstract Flexible manufacturing system (FMS) is a highly integrated manufacturing system. The relation between its

Journal of King Saud University – Engineering Sciences (2012) 24, 115–129

King Saud University

Journal of King Saud University – Engineering Sciences

www.ksu.edu.sawww.sciencedirect.com

ORIGINAL ARTICLE

Analysis of performance measures of flexible

manufacturing system

Abdulziz M. El-Tamimi a, Mustufa H. Abidi b,*,

S. Hammad Mian b, Javed Aalam b

a Industrial Engineering Department, College of Engineering, King Saud University, P.O. Box 800, Riyadh 11421, Saudi Arabiab Princess Fatima Alnijris Research Chair for Advanced Manufacturing Technology (FARCAMT), Industrial

Engineering Department, College of Engineering, King Saud University, P.O. Box 800, Riyadh 11421, Saudi Arabia

Received 4 November 2010; accepted 21 June 2011Available online 17 August 2011

*

E

10

El

Pe

do

KEYWORDS

Flexible manufacturing

system (FMS);

Petri net;

Simulation;

Bottleneck;

Visual Slam AweSim

Corresponding author. Tel.

-mail address: mustufahaide

18-3639 ª 2011 King Saud

sevier B.V. All rights reserve

er review under responsibilit

i:10.1016/j.jksues.2011.06.00

Production and h

: +966 5

r@yahoo

Universit

d.

y of King

5

osting by E

Abstract Flexible manufacturing system (FMS) is a highly integrated manufacturing system. The

relation between its components is very complex. The mathematical programming approaches are

very difficult to solve for very complex system so the simulation of FMS is widely used to analyze its

performance measures. Also the FMS components are very sophisticated and costly. If FMS has to

be implemented then it is better to analyze its results using simulation which involves no loss of

money, resource and labor time. As a typical discrete event system FMS have been studied in such

aspects as modeling and performance analysis. In this paper, a concept and implementation of the

Petri nets for measuring and analysis of performance measures of FMS is applied. Also the system

has been modeled in Visual Slam software, i.e. AweSim. The other well defined mathematical tech-

nique, i.e. bottleneck technique has also been applied for the purpose of comparison and verifica-

tion of the simulation results. An example FMS has been taken into consideration and its Petri net

model, AweSim model, and mathematical model has been constructed. Several performance mea-

sures have been used to evaluate system performance. And it has been found that the simulation

techniques are easy to analyze the complex flexible manufacturing system.ª 2011 King Saud University. Production and hosting by Elsevier B.V. All rights reserved.

45269756.

.com (M.H. Abidi).

y. Production and hosting by

Saud University.

lsevier

1. Introduction

In the present market scenario, the customer demand and spec-

ification of any product changes very rapidly so it is very impor-tant for a manufacturing system to accommodate these changesas quickly as possible to be able to compete in the market. This

evolution induces often a conflict for a manufacturing systembecause as the variety is increased the productivity decreases.So the flexible manufacturing system (FMS) is a good combina-

tion between variety and productivity. In this system, the mainfocus is on flexibility rather than the system efficiencies. A

mailto:[email protected]

http://dx.doi.org/10.1016/j.jksues.2011.06.005



http://www.sciencedirect.com/science/journal/10183639

116 A.M. El-Tamimi et al.

competitive FMS is expected to be flexible enough to respondto small batches of customer demand and due to the fact thatthe construction of any new production line is a large invest-

ment so the current production line is reconfigured to keepup with the increased frequency of new product design.

The optimal design of FMS is a critical issue and it is a

complex problem. There are various modeling techniques forFMS; the most common one are based on mathematical pro-gramming. FMS is a highly integrated manufacturing system

and the inter-relationships between its various componentsare not well understood for a very complex system. Due to thiscomplexity, it is difficult to accurately calculate the perfor-mance measures of the FMS which leads to its design through

mathematical techniques. Therefore, computer simulation isan extensively used numeric modeling technique for the analy-sis of highly complex flexible manufacturing systems (Cheng,

1985; Jain and Foley, 1986; Kalkunte et al., 1986).Since an FMS can be viewed as a discrete event system, the

methods for modeling and control of such a system have been

developed by using Petri nets (Ezpeleta et al., 1995;Viswanadham et al., 1990) max-plus algebra (Cuninghame,1979) or, by using a matrix description approach (Lewis

et al., 1998; Gurel et al., 2000). There are research activities thatfollow the trend of graphically oriented design of FMS control-lers. Good examples are the programs like Onika (Gertz andKhosla, 1994), Robotica (Nethery and Spong, 1994), OpenRob

(Ge et al., 2000), and the Robotics Toolbox for MATLAB,which are intended for the graphical design of systems for thecontrol of robotized plants and which integrate already existing

software modules for control of robot manipulators.Petri-net has evolved into an elegant and powerful graphi-

cal modeling tool for asynchronous concurrent event-driven

systems. The wide ranging application areas of Petri-net in-clude communication protocols (Berthelot and Terrat, 1982;Diaz, 1982), distributed systems (Ayache et al., 1982), database

(Voss, 1980), complier and operating system (Baer and Ellis,1977; Noe, 1971), formal languages (Mandrioli, 1977; Peter-son, 1976a,b). In additional, Petri-net are very useful tool formodeling (Narakari and Viswanadham, 1985; Chin-Jung

et al., 1992; Kimon, 1990), simulation (Venkatesh et al.,1992), and scheduling of discrete event systems (Bispo et al.,1992), such as flexible manufacturing systems, and also for

designing and implementing controllers for manufacturing(Crockett et al., 1987; Zhou et al., 1990). Modeling and simu-lation of FMS is a field of research for many people now days.

However, they all share a common goal; to search for solutionsto achieve higher speeds and more flexibility and thus increasemanufacturing productivity. From the design point of view,the use of nets has many advantages in modeling, qualitative

analysis, performance evaluation and code generation.In this research work, FMS is modeled with the help of Pet-

ri net and also it is modeled in Visual Slam AweSim to analyze

its performance measures. In addition, the bottlenecktechnique has been applied to compare and verify the resultsobtained from the simulation techniques.

2. Literature survey

Browne et al., 1984 defines FMS as an integrated computer-

controlled system with automated material handling devicesand CNC machine-tools and which can be used to simulta-neously process a medium-sized volume of a variety of parts.

Chan et al. (2007) presented a simulation study usingTaguchi’s method analysis of physical and operating parame-ters of the flexible manufacturing system along with flexibility.

An approach is developed to study the impact of variations inthe physical and operating parameters of an FMS and to iden-tify the level of these variations. The physical and operating

parameters of alternative resources may influence the system’sperformance with the changing levels of flexibility and opera-tional control parameters such as scheduling rules. The results

of simulation study shows that expected benefits may not bepresent when routing flexibility (RF) levels are increased withpresence of the variations in physical and operating parame-ters. The increase in RF level becomes counterproductive

under such environment when variations are above certain lim-its. It may be useful for decision maker to distinguish the levelof flexibility up to which it can be gainfully increased under the

presence of variations.Sarker et al. (1994) have presented a detailed classification

for the types of manufacturing related flexibility as follows:

routing flexibility (RF), machine, flexibility, process flexibility,expansion flexibility, job flexibility, design flexibility, materialhandling flexibility, setup time flexibility, and volume flexibility.

Buitenhek et al. (2002) described that the design of thesesystems is an important issue of the expensive components ofFMS. The design of FMS requires both the physical and thecontrol aspects. Physical aspects includes the issues such as

types and numbers of machines, material handling systems,processing times on a machine, machine setting time, toolchanging time, transportation time, loading, unloading time,

etc. and for the control aspects, the design involves definingthe scheduling rules or algorithms that defines the way the sys-tem is to be operated.

Gupta and Buzacott (1989) explained that the flexibilitydoes not come from the abilities of machine alone; in fact flex-ibility is the result of a combination of factors like physical

characteristics, operating decisions, information integration,and management practice. Flexibility is critical in providingthe effectiveness to manufacturing system under different oper-ating conditions.

Bennett et al. (1992) identifies the factors crucial to thedevelopment of efficient flexible production systems, namely:effective integration of subsystems, development of appropri-

ate controls and performance measures, compatibility betweenproduction system design and organization structure, and ar-gues that the flexibility cannot be potentially exploited if its

objectives are not defined and considered at design stage.Kumar et al. (2003) used an ant colony optimization

approach for scheduling of FMS for a given level of flexibility.Wang and Yen (2001) considered the transportation times

in automated material handling system for simulation studyof dispatching rule performance.

Chan (2001) used the simulation under Taguchi method to

study the effect of RF on an FMS. Bruccoleri et al. (2003) sug-gested the simulation as a tool for defining the configuration ofan FMS or a complex system.

Shnits et al. (2004) used simulation of operating system asa decision support tool for controlling the flexible system toexploit flexibility.

Tuysuz and Kahraman (2009) presented an approachfor modeling and analysis of time critical, dynamic and com-plex systems using stochastic Petri nets together with fuzzysets.

Analysis of performance measures of flexible manufacturing system 117

Santarek and Buseif (1998) introduced a method and toolsfor the design of automated manufacturing systems and theirsequential controllers. The high level system design specifica-

tion was developed using Structured Analysis and Design Tech-nique (SADT) method and Design: IDEF software package.The interface is based on a number of transformation rules

from an IDEF0 specification into a Petri net. Petri net is usedto determine if the manufacturing system will operate in the de-sired manner. Petri nets are proven tools enabling analysis of

concurrent processes performed with the use of shared re-sources (machines, equipment, operators, etc.). SADT: IDEF0represents activity oriented modeling approach. IDEF0 repre-sentation of a manufacturing system consists of an ordered

set of boxes representing activities performed by the system.The activity may be a decision-making, information conver-sion, or material conversion activity. The inputs are those items

which are transformed by the activity; the output is the result ofthe activity. The conditions and rules describing the manner inwhich the activity is performed are represented by control ar-

rows. The mechanism represents resources (machines, comput-ers, operators, etc.) used when performing the activity. IDEF0was found to be a powerful descriptive tool that offers a num-

ber of features which make it easy to apply and easy to under-stand. It allows for a top down step refinement, using agraphical representation with few constructs and simple rules.IDEF0 is able to modify the existing system by adding a new

constrains (machine, buffer, robot, etc.) without any essentialchange in the existing model. Standard PN’s do not includeany time concept. Therefore, with a standard PN it is possible

to describe only the logical structure of systems, and not theirtime evolution.

Viswanadham et al. (1990) investigated the use of Petri net

(PN) models in the prevention and avoidance of deadlocks inflexible manufacturing systems. For deadlock prevention, areachability graph of a Petri net model of the given FMS is

used, whereas for deadlock avoidance, Petri net-based on-linecontroller is proposed. Deadlock handling can take two forms:deadlock prevention in which deadlocks are eliminated by sta-tic resource allocation policies and deadlock avoidance in

which dynamic policies are employed to avert deadlocks justin time.

Tavana (2008) used Petri nets (PNs) for dynamic process

modeling of the emergency management system at a nuclearpower plant. PNs with their graphical and precise nature andtheir firm mathematical foundation are useful in reducing the

number of false evacuations at the plant. Process models arewidely used for decomposing organizational complexities,adapting best business practices, identifying process weak-nesses, training end-users, and designing business blueprints.

As a graphical tool, PNs provide a visual medium for a mod-eler to describe a complex system. As a mathematical tool, PNmodels can be represented by linear algebraic equations,

creating the possibility for the formal analysis of the model(Zurawski and Zhou, 1994). Mathematical properties of PNcan be classified into structural properties that depend on the

net structure and behavioral properties that depend on the ini-tial and subsequent markings. As a graphical tool, PNs are usedto enhance communications and produce accurate and com-

plete specifications while the mathematical properties of PNsare used to detect deadlock, overflow, and irreversiblesituations. Performance evaluation is also possible throughmathematical analysis of the model using stochastic timed

PNs. One of the strengths of PNs is their broad based applica-bility to a wide range of systems. Timed PNs are those in whichplaces or transitions have time durations in their activities (Liu

et al., 2007). Stochastic PNs include the ability to model ran-domness in a situation, and also allow for time as an elementin the PN (Murata, 1989; Lee et al., 2001). Colored PNs allow

the user and developer to witness the changes in places andtransitions through the application of color-specific tokens,and movement through the system can be represented through

the changes in colors (Chen et al., 2001). Fuzzy PNs are used tomodel fuzzy rule-based reasoning to handle uncertain andimprecise information (Huang et al., 2008; Li and Lara-Rosano,2000).

Sun and Fu (1994) proposed a systematic method toconstruct a timed place Petri-net model for modeling anFMS which is as an automatic Petri-net generator, with a

graphic user interface, through which a user can input thesystem information of a practical FMS, such as the numberof automated guided vehicles (AGVs), the number of ma-

chines, and their geometric relationship, as well as processflows of parts to be processed. This generator will then usethe above information to generate the corresponding Petri-

net model.Venkateswaran and Bhat (2006) reviews the difficulties in

evolution of modeling the system, design of a controller andthen their implementation using a programmable logic con-

troller to a discrete event system. A model for investigatingthe development of an algorithm for fuzzy logic controllerfor supervisory control of a discrete event system is described.

Such an algorithm is helpful to efficiently maintain an opti-mum throughput and energy requirements of the drives. Thisresults in concise representation and handling of uncertainty

in discrete event systems.Kovacic et al. (2001) presented a concept to implement

computer integrated tool for design and simulation of flexible

manufacturing systems (FMS). A methodology is described forsetting an FMS configuration that supports automatic genera-tion of a matrix-based FMS description and consequently,enables effective off-site design and simulation of the FMS

by using virtual reality models and web-related technologies.An integrated control system is presented for the intelligentprocessing of the acquired sensor data, graphically oriented

setup of the FMS and dynamically linked automated multi-level control of FMS system.

Delgadillo and Llano (2006) introduced a Petri net-based

integrated approach, for simultaneously modeling and sched-uling manufacturing systems. A prototype that simulates theexecution of the production plan, and implements priority dis-patching rules to solve the eventual conflicts, is presented. Such

an application was tested in a complex flexible job shop-typesystem. Scheduling is a difficult task in most manufacturingsettings due to the complexity of the system. Hence there is a

requirement of powerful tools that can handle both modelingand optimization.

Nandkeolyar andChristy (1989) interfaced a computer sim-

ulation model of an FMS with the Hooke–Jeeves algorithm tosearch an optimum design without full factorial experimenta-tion. Some modifications of the HJ algorithm are carried out

to accommodate the stochastic nature of computer simulation.The inter-relationships between FMS components are not wellunderstood. Consequently, it has not been possible to developclosed form analytic models of FMSs. So, computer


simulation has been extensively applied to study theirperformance.

After reviewing the above set of research papers it can be

said that the design and modeling of the complex FMS is a dif-ficult task using mathematical techniques, so the computersimulation seems to be a better option. Therefore, to check

the accuracy of the results obtained from simulation tech-niques this research work has been carried out.

3. Overview of flexible manufacturing system

Flexible manufacturing system (FMS) is a class of manufactur-ing system that can be quickly configured to produce variety of

products. Over the last few decades, the modeling and theanalysis of FMSs has been closely studied by control theoristsand engineers. An FMS is a production system where a

discrete number of raw parts are processed and assembled bycontrolled machines, computers and/or robots (Ruiz et al.,2009). It generally consists of a number of CNC machine tools,robots, material handling, automated storage and retrieval sys-

tem, and computers or workstations. A typical FMS can fullyprocess the members of one or more part families on a contin-uing basis without human intervention and is flexible enough

to suit changing market conditions and product types withoutbuying other equipment (the concept ‘‘flexible’’ can refer tomachines, processes, products, routings, volume, or produc-

tions). The concept of FMS is credited to David Williamson,a British engineer employed by Molins during the mid 1960s.Molins applied for a patent for the invention that was grantedin 1965. The concept was called System 24 then because it was

believed that the group of machine tools comprising the systemcould operate 24 h a day. One of the first FMS installed in USwas a machining system at Ingersoll-Rand Company (Luggen,

1991; Maleki, 1991; Mejabi, 1988) (Fig. 1). There are threecapabilities that a manufacturing system must possess in orderto be flexible:

(1) The ability to identify and distinguish among differentincoming part or product styles processed by the system.

(2) Quick changeover of operating instructions.(3) Quick changeover of physical setup.

To qualify as being flexible the automated system should

pass these four tests:

(1) Part variety test.

(2) Schedule change test.

Figure 1 Flexible manufactu

(3) Error recovery test.

(4) New part test.

The FMS has emerged as one of the revolutions in the

manufacturing industries in recent years. It has made it possi-ble to produce a variety of parts in less time and cost. Theapplication of FMS in the current market scenario can satisfythe growing demands of variety, quantity and speed at the

same time. The components of the FMS can be classified intotwo categories:

1. Hardware: machine tools, handling systems, guided vehi-cles, inspection center, robots, etc.

2. Software: software for FMS can further be classified into

extrinsic and intrinsic functions.

Extrinsic functions are used to plan and control the func-tions that take place outside the physical boundaries of the

FMS, e.g. production scheduling, process planning, mainte-nance planning, etc.

Intrinsic functions control the components within the phys-

ical boundary, e.g. production control, production monitoring,machine diagnostic, etc.

FMS mainly consists of following flexibilities: machine

flexibility, operational flexibility, process flexibility, produc-tion flexibility, routine flexibility, etc. The performance ofthe system improves as the flexibilities increases. The differ-

ent levels of manufacturing flexibility can be defined asfollows:

(a) Basic flexibilities

Machine flexibility: machine’s ability to adapt to a vari-ety of products.Material handling flexibility: it is a measure of the sys-

tem’s ability with which different part types can betransported and properly positioned at the variousmachine tools.

Operation flexibility: it measures adaptability to alterna-tive operation sequences for processing a part type.

(b) System flexibilities

ring sy

Volume flexibility: it measures system’s capability to

operate efficiently at different volumes of the part types.Routing flexibility: it is the system’s ability to use multi-ple machines to perform the same operation on a part. It

is a measure of the alternative paths that a part caneffectively follow through a system for a given processplan.

stem configuration.


Process flexibility: it is a measure of the volume of the

set of part types that a system can produce withoutchanging any setup.Product flexibility: the volume of the set of part types

that can be manufactured in a system with minor setup.(c) Aggregate flexibilities

Program flexibility: the ability of a system to run forlong periods.

Production flexibility: the volume of the set of part typesthat a system can produce without major investment incapital equipment.

Market flexibility: the ability of a system to efficientlyadapt to changing market conditions.

The case study presented in the paper mainly consists of

two kinds of flexibilities machine flexibility and routine flex-ibility. A key issue in the FMS is the performance evalua-tion of the system. The better performance of the FMSresults in reduced labor costs, increased output, decreased

manufacturing costs, increased flexibility, and reduced pro-duction lead time. The major performance measures usedin the study were machine utilization and overall productiv-

ity. Machine utilization increases as the jobs to be processedon the machine increases and a bottleneck machine has100% utilization in the system. The machine utilization is

measured by the number of hours it operates in the systemto the total available hours in the system. The design andperformance evaluation of the FMS system is a complex

process and requires a thorough investigation. Three differ-ent methods have been proposed and compared in the pre-sented case studies to evaluate the different cases ofdifferent FMS systems. The feasibility of the different tech-

niques has also been reviewed in order to determine theirsuccessful application in the evaluation of FMS system.

4. Petri nets

Petri nets are a class of modeling tools, which were originatedby Petri (Petri,1962, 1976), they have a well-defined mathemat-

ical foundation and easy to understand graphical feature. It is apowerful design tool which facilitates visual communicationbetween people who are engaged in the design process. A Petri

net is a directed graph consisting of three structural compo-nents – places, transitions, and arcs. Places which are drawnas circles represent possible states or conditions of the systemwhile transition, which are shown by bars or boxes, describe

events that may modify the system states. The relationshipsbetween places and transitions are represented by a set of arcswhich are the only connectors between a place and a transition

in either direction. The dynamic behavior of the system can berepresented using tokens which graphically appear as blackdots in places.

Figure 2 A Petri net model

Manufacturing system is a discrete system. Hence any mod-eling has to be based on the concepts of events and activities.An event corresponds to a state change. When using Petri nets,

events are associated with transitions. Activities are associatedto the firing of transitions or/and to the marking of places.Queuing models can also be used for handling events and

activities but synchronizations are difficult in these models.Hence Petri net is preferred. Analysis of the Petri net revealsimportant information about the structure and dynamic

behavior of the modeled system. This information is used toevaluate the modeled system and suggest improvements ofchanges. Petri nets are used to model the occurrence of variousevents and activities in a system (Fig. 2).

4.1. Properties of Petri nets

Petri nets as a mathematical tools possess several properties.

These properties, when construed in the context of modeledsystem allow the system designer to identify the presence or ab-sence of the application domain specific functional properties

of the system under design. There are two types of propertiesfor Petri nets: behavioral and structural. The behavioral prop-erties are those which depend on initial state or marking of

Petri nets while the structure properties do not depend uponthe initial state but on the topology or net structure. Someof the behavioral properties are reachability, boundedness,conservativeness, etc.

4.1.1. Reachability

It is important property when designing a distributed system

that whether a system has reached a particular state or not.A sequence of firings will result in a sequence of markings.

A marking Mn is said to be reachable from a marking M0 ifthere exists a sequence of firings that transforms M0 to Mn. A

firing or occurrence sequence is denoted by r = t1, t2...tn. Inthis case, Mn is reachable from M0 by r and we writeM0 [r > Mn]. The set of all possible markings reachable from

M0 in a net (N, M0) is denoted by R (N, M0) or simply R (M0).The set of all possible firing sequences from M0 in a net (N,M0) is denoted by L (N, M0) or simply L (M0). A marking

Mn is said to be reachable from a marking M0 if there existsa sequence of transitions firings which transforms a markingM0 to Mn (Murata, 1989).

4.1.2. Boundedness

A Petri net (N, M0) is said to be bounded if the number of to-kens in each place does not exceed a finite number k for any

marking reachable from M0, i.e. M(p) 6 k for every place pand every marking M e R(Mo). Places are frequently used torepresent product and tool storage areas in manufacturing sys-tems, etc. In manufacturing systems, attempts to store more

of a manufacturing line.


tools, for instance, in the tool storage area may result in theequipment damage. The Petri net property which helps to iden-tify in the modeled system the existence of overflows is the con-

cept of boundedness.

4.1.3. Conservativeness

In real systems, the number of resources in use is typically re-

stricted by the financial as well as other constraints. If tokensare used to represent resources, the number of which in a sys-tem is typically fixed, then the number of tokens in a Petri net

model of this system should remain unchanged irrespective ofthe marking the net takes on. This follows from the fact thatresources are neither created nor destroyed, unless there is a

provision for this to happen. For instance, a broken toolmay be removed from the manufacturing cell, thus reducingthe number of tools available by one. A Petri net is conserva-

tive if the number of tokens is conserved. From the net struc-tural point of view, this can only happen if the number of inputarcs to each transition is equal to the number of output arcs.

5. Visual Slam AweSim

AweSim is a general-purpose simulation system which takesadvantage of the latest in Windows technology to integrate

programs. AweSim includes the Visual Slam simulation lan-guage to build network, sub network, discrete event, and con-tinuous models. AweSim incorporates the Visual Slam

modeling methodology. The basic component of a Visual Slammodel is a network, or flow diagram, which graphically por-trays the flow of entities (people, parts, information, etc.)

through the system. A network is built interactively in AweSimby selecting symbols from a graphical palette and draggingthem to the desired location with the mouse. On-line error

checking is performed upon completion of the form so that in-put errors can be corrected immediately. AweSim also facili-tates model building by providing context-sensitive help andsearch capabilities (O’ Really, 1999; see Fig. 3).

6. Bottleneck technique

Important aspects of the FMS performance can be mathemat-

ically described by a deterministic model called the bottleneckmodel developed by Solberg (1981). The bottleneck model issimple and intuitive but it has a limitation of a deterministic

approach. It can be used to provide starting estimates ofFMS design parameters such as production rate, number ofwork stations, etc. The term bottleneck refers to the fact that

an output of the production system has an upper limit, giventhat the product mix flowing through the system is fixed. Thismodel can be applied to any production system that possesses

possess this bottleneck feature.

Figure 3 Node

6.1. Terminology and symbols

Part mix, a mix of the various parts or product styles producedby the system is defined by pi. The value of pimust sum to unity.X

i

pi ¼ 1:0

The FMS has a number of distinctly different workstations

n and si is the number of servers at the ith workstation. Oper-ation frequency is defined as the expected number of times agiven operation in the process routing is performed for eachwork unit.

fijk ¼ operation frequency

For each part or product, the process routing defines the

sequence of operations, the workstations where operationsare performed, and the associated processing time.

tijk ¼ processing time for operation

The average workload, WLi

WLi ¼X

j

X

k

tijkfijkpi

The average of transport required completing the process-ing of a work part, n
t
nt ¼X

i

X

j

X

k

fijkpi � 1

The workload of handling system, WLn+1

WLnþ1 ¼ nttnþ1

where tn+1 = Mean Transport time per move, min.The FMS maximum production rate of all part, R�p, Pc/min

R�p ¼ S�=WL�

where WL* is workload min/Pc and S* = Number of

machines at the bottle-neck stationThe part (j) maximum production rate, R�pi, Pc/min.

R�pi ¼ piðR�piÞ ¼ piS�

WL�

Mean utilization of a station (i), Ui

Ui ¼WLi

Si

ðR�pÞ ¼WLi

Si

� S�

WL�

Average Utilization of FMS including Transport system

U ¼Pnþ1

i¼1Ui

nþ 1

Overall FMS utilization

Us ¼Pn

i¼1SiUiPni¼1Si

s in AweSim.


7. Case studies

7.1. System description for case study 1

The system shown in the Fig. 4 consists of three robots, twomachines, i.e. a drilling machine and a milling machine and

an inspection center. The system also consists of two convey-ors, conveyor in and conveyor out. The system has operationalflexibility in the sense that two types of products are produced

in the system using two different process operations. Type Aproduct undergoes milling operation only whereas Type Bproduct undergoes drilling and then milling. Both the partsare inspected before moving to conveyor out. Robot 2 loads

the milling and drilling machine from conveyor in, and Robot1 loads the part B onto the milling machine after it completesthe drilling operation. It also unloads the milling machine. Ro-

bot 3 loads the inspection center.Material handling system consists of three robots, one con-

veyor and three work carriers with mean transport time = 5

min (see Table 1).

7.1.1. Solution methodology

Three types of techniques are applied to find the parameters of

the given FMS; the two are simulation techniques and the one

Figure 4 Block diagram of con

Table 1 List of operations and process time on different machining

Part Part mix Operation Pro

A 0.4 Load 4

Mill 25

lisped 6

Unload 2

B 0.6 Load 4

Drill 20

Mill 30

Inspect 8

Unload 2

is mathematical technique. The simulation techniques are: Vi-sual Slam AweSim and Petri net. The mathematical one is thebottleneck technique. The system is modeled in both simula-

tion techniques and then the results are compared with themathematical technique.

7.1.1.1. Visual Slam AweSim. The system is modeled in VisualSlam AweSim software in order to evaluate its performance.The network is shown in Fig. 10.

The system is using different kinds of nodes such as createnode, assign node, await node, free node, collect node, termi-nate node, goon node, etc. to describe the given system. Thesystem starts with creating entities at an interval of 2 min. A

goon node is used to divide in different sequence accordingto their operational requirements. In the await node, entitieswait for the resources to be available. The system is using three

resources of robots, one resource of milling machine and oneresource of drilling machine each having a capacity of three.An inspection system is also present in the system. The system

runs for 1 week working hours.

7.1.1.2. Modeling in Petri net. The system consists of loading

and unloading station, two process stations, inspection centerand two AVGs.

sidered FMS in case study 1.

centers.

cess time (min) No. of servers Frequency

1 1

3 1

1 1

1 1

1 1

3 1

3 1

1 1

1 1

Figure 5 Block diagram of considered FMS in case study 2.

Table 2 List of operations and process time on different stations.

Part Weekly demand Process sequence Operation time, min

Load Process Inspection station Unload

A B C D Frequency, f A

1 250 A fi C fi D fi A 5 21 14 0.5 3

2 350 A fi B fi D fi A 5 22 14 0.3 3

3 150 A fi B fi C fi D fi A 5 20 22 15 0.5 3

4 250 A fi C fi B fi C fi D 5 15 20 14 0.4 3

Mean travel time of AGV, min 2.5

Table 3 Part mix and number of machines at different

stations.

Part Part mix Station Number of machines

1 0.25 A 2

2 0.35 B 3

3 0.15 C 4

4 0.25 D 1


To start a cycle, raw parts and the AVGs must be available.

Then only firing will take place. The AVG carries a raw partfrom loading station to the process station according to the gi-ven sequence for the different parts. After the completion of anoperation in one station the part is again carried by AVG to its

next required station. At the end the part is carried to theunloading station.

Place representing the stations have tokens according to the

number of machines they have. The Petri net model is simu-lated to get the overall productivity of the given system. ThePetri net model is shown in Figs. 12 and 13.

7.1.1.3. Bottleneck technique. A C++ program has beendeveloped to get the performance measures of the FMS system

using bottleneck technique.

7.2. System description for case study 2

A four stations (A, B, C, and D) FMS is used to produce four

parts (1, 2, 3, and 4) according the data given in Table 2. Sta-tion A is the loading/unloading station, stations B and Care the process stations and station D is the inspection

station. FMS runs for 16 h/day and 6 days a week (see Fig. 5and Table 3).

There are two AGVs for parts movement between different

stations.

7.2.1. Solution methodology

In this case study, the above mentioned three techniques are

applied, i.e. bottleneck technique and two simulation tech-niques; Visual Slam AweSim and Petri net.

7.2.1.1. Visual Slam AweSim. The system is modeled in VisualSlam AweSim software in order to evaluate its performance.The network is shown in Fig. 11.

The system is using different kinds of nodes such as createnode, assign node, await node, free node, collect node, termi-nate node, goon node, etc. to describe the given system. The sys-

tem starts with creating entities at an interval of 0.5 min. A goonnode is used to divide in different sequence according to theiroperational requirements. In the await node, entities wait forthe resources to be available. The system is using two resources

for loading and unloading; three resources for process at stationB, three resources for process at station B, one for inspectionand two AVGs are used. The system runs for 500 min.

7.2.1.2. Modeling in Petri net. The system consists of four sta-tions and two AGVs. Each station consists of several ma-

chines like station A has two machines, station B has threemachines, station C has four machines, and station D hasone machine.

To start a cycle, a raw part and the robot must be available.Then only firing will take place. The robot moves a raw partfrom the incoming conveyor and loads it to the machine

Figure 6 Comparison of utilization from different techniques for

case study 1.

Figure 7 Comparison of overall productivity coming from


according to operation requirement. If the part requires thesecond operation, the second robot loads that part onto thenext station. The operation is performed while the robot re-

turns. The robot unloads the part from the machine anddeposits it on the outgoing conveyor and returns.

Two output places (counters) are introduced to count the

number of Types A and B products generated. Place P1 hastwo tokens representing the two types of products. Each ofdrilling and milling machines has one token each representing

their capacity which is one. Robots 1 and 2 also have capacitiesof one each therefore consisting of one token. The Petri netmodel is simulated to get the overall productivity of the givensystem. The Petri net model is shown in Figs. 14 and 15.

7.2.1.3. Bottleneck technique. A C++ program has beendeveloped to get the performance measures of the FMS system

using bottleneck technique.

7.3. Results and discussion

By the three different techniques the obtained results for casestudy 1 are as follows:

Solution Techniques Utilization

Visual

Slam AweSim

Bottle-neck

model

Petri net model

Operations

Drilling 0.5343 0.4280 0.4237

Milling 0.9903 0.9987 0.9884

Inspection 0.8110 0.7704 0.7493

different techniques for case study 1.

Overall productivity (parts per minute)

Visual Slam AweSim Bottle-neck model Petri net model

0.1180 0.1070 0.1210

Then the Utilization and overall productivity is compared

as and shown in Figs. 6 and 7.By the three different techniques the obtained results for

case study 2 are as follows:

Solution Techniques Utilization

Visual Slam -

AweSim

Bottle-neck

model

Petrinet

Model

Stations

A 0.7845 0.6963 0.7654

B 0.9161 0.8384 0.8968

C 0.7843 0.8072 0.7452

D 0.9812 1.0000 0.9767

Figure 8 Comparison of utilization from different techniques for

case study 2.

Visual Slam AweSim Bottle-neck model Petri net model

Overall productivity (parts per minute)

0.2013 0.1741 0.1976

Then the utilization and overall productivity is compared as(Figs. 8 and 9):

There are large numbers of techniques such as simulationtechniques, modeling techniques, mathematical programmingwhich can be used to evaluate the performance of any FMS.

To find the application and scope of these techniques twocase studies of FMS is modeled and evaluated using Petrinet, AweSim, and bottleneck technique and the results arecompared.

Figure 10 Visual Slam Network of the

Figure 9 Comparison of overall productivity coming from

different techniques for case study 2.


From the results it can be concluded that the performanceobtained for a given system from the three techniques are veryclose to each other in the case study 1. In case study 2, there is

more deviation between the results obtained from the mathe-matical technique and the simulation technique as the FMSconsidered in case study 2 is more complex than in case study

1. Still the difference in results obtained is not very significanthence it can be said that these techniques are applicable to anyFMS system to evaluate and confirms its performance.

But for a complex system it is very difficult to analyze theperformance measures using the mathematical techniques soit can be said that the simulation techniques are the betteralternatives for evaluating the complex systems to save effort

and time.

given FMS system in case study 1.


Acknowledgments

The author would like to thank Princess Fatima Alnijris Re-

search Chair for Advanced Manufacturing Technology (FAR-CAMT) for their support in this work. And the author wouldalso like to thank Professor Abdul Rahman Al-Ahmari for hishelp and guidance.

Figure 11 Visual Slam Network of the

Appendix A

A.1. Visual Slam Network

A.2. Petri net model’s different stages of simulation and results

Figs. 13 and 15

given FMS system in case study 2.

Figure 13 Petri net model of the FMS of case study 1 using MATLAB Petri net toolbox while simulating.

Figure 12 Petri net model of the FMS of case study 1 using MATLAB Petri net toolbox.


Figure 14 Petri net model of the FMS of case study 2 using MATLAB Petri net toolbox.

Figure 15 Petri net model of the FMS of case study 2 using MATLAB Petri net toolbox while simulating.



References

Ayache, J.M., Courtist, J.P., Diaz, M., 1982. REBUS: a fault tolerant

distribution system for industrial real time control. IEEE Trans.

Comput. C-31, 637–674.

Baer, J.L., Ellis, C.A., 1977. Model design and evaluation of a

compiler for a parallel processing environment. IEEE Trans.

Software Eng. SE-3 (6), 394–405.

Bennett, D., Forrester, P., Hassard, J., 1992. Market-driven strategies

and the design of flexible production systems: evidence from the

electronics industry. Int. J. Oper. Prod. Manag. 12 (2), 25–43.

Berthelot, G., Terrat, R., 1982. Petri nets theory for correctness of

protocols. IEEE Trans. Commun. COM-30, 2497–2509.

Bispo, C.F.G., Sentieiro, J.J.S., Hibberd, R.D., 1992. Adaptive

scheduling for high-volume shops. IEEE Trans. Robot Automat.

8, 696–706.

Browne, J., Dubois, D., Rathmill, K., Sethi, P., Steke, KE., 1984.

Classification of flexible manufacturing systems. FMSMag., 14–27.

Bruccoleri, M., Sergio, NL., Perrone, G., 2003. An object-oriented

approach for flexible manufacturing controls systems analysis and

design using the unified modeling language. Int. J. Flexible Manuf.

Syst. 15 (3), 195–216.

Buitenhek, R., Baynat, B., Dallery, Y., 2002. Production capacity of

flexible manufacturing systems with fixed production ratios. Int. J.

Flexible Manuf. Syst. 14 (3), 203–226.

Chan, FTS., 2001. Effects of routing flexibility on a flexible manufac-

turing system. Int. J. Comput. Integrat. Manuf. 14 (5), 431–445.

Chan, F.T.S., Bhagwat, R., Wadhwa, S., 2007. Flexibility perfor-

mance: Taguchi’s method study of physical system and operating

control parameters of FMS. Rob. Comput. Integr. Manuf. 23,

25–37.

Chen, S.C., Ke, Y.L., Wu, J.S., 2001. Coloured petri-nets approach for

solving distribution system contingency. Proc. IEEE 148 (5),

463–470.

Cheng, T.C.E., 1985. Simulation of flexible manufacturing system.

Simulation 45 (6), 299–302.

Chin-Jung, T., Li-Chen, F., Yung Jen, H., (1992). Modeling and

simulation for flexible manufacturing system using Petri net.

Second Int. Conf. Autom. Technol. 4, 31–36.

Crockett, D., Dearochers, A., DiCesare, F., Ward, T. (1987).

Implementation of a Petri net controller for machine workstations.

Proc. of IEEE Robotics Automat. Conf. 1861–1867.

Cuninghame, Green R.A., (1979). Minimax Algebra, Vol. 166 of

Lecture Notes in Economics and Mathematical Systems. Springer-

Verlag, Berlin, Germany.

Delgadillo, G.M., Llano, S.B., 2006. Scheduling application using

petri nets: a case study: intergraficas s.a.’. In: Proceedings

of 19th international conference on production research,

Valparaiso, Chile.

Diaz, M., 1982. Modeling and analysis of communication and

cooperation protocols using Petri net based models. Comput.

Networks 6.

Ezpeleta, S.D., Colom, J.M., Martinez, J.A., 1995. Petri net based

deadlock prevention policy for flexible manufacturing systems.

IEEE Trans. Rob. Autom. 11 (2), 173–184.

Ge, S.S., Lee, T.H., Gu, D.L., Woon, L.C., 2000. A one stop solution

in robotic control system design. IEEE Rob. Autom. Mag. 7,

42–55.

Gertz, M. W., Khosla, P. K., (1994). Onika: A multilevel human–

machine interface for real-time sensor-based robotics systems. In:

Proceedings of SPACE 94: The Fourth International Conference

and Expo. on Engineering Construction, Albuquerque.

Gupta, D., Buzacott, JA., 1989. A framework for understanding

flexibility of manufacturing systems. J. Manuf. Syst. 8 (2), 89–97.

Gurel, A., Bogdan, S., Lewis, F.L., Huff, B., (2000). Matrix approach

to deadlock-free dispatching in multi-class finite buffer flowlines.

IEEE Trans. Autom. Control 45(11), 2086–2090.

Huang, Y-M., Chen, J-N., Huang, T-C., Jeng, Y-L., Kuo, Y-H., 2008.

Standardized course generation process using dynamic fuzzy petri

nets. Expert Syst. Appl. 34, 72–86.

Jain, S., Foley, W.J., (1986). Basis for development of a generic fms

simulator. Proc. Second ORSA/TIMS Conf. Flexible Manuf. Syst.:

Oper. Res. Models Appl., 393–403.

Kalkunte, M.V., Sarin, S.C., Wilhelm, W.E., 1986. Flexible Manu-

facturing Systems: A Review of Modeling Approaches to Design,

Justification and Operations. Flexible Manuf. Syst.: Methods

Stud.. Springer, Berlin, pp. 3-25.

Kimon, P.V., 1990. On the hierarchical modeling analysis and

simulation of flexible manufacturing system with extended Petri

nets. IEEE Trans. Syst. Manuf. Cybernatics 20 (1), 94–109.

Kovacic, Z., Bogdan, S., Reichenbach, T., Smolic-Rocak, N., Bir-

gmajer, B., (2001). FlexMan – A computer-integrated tool for

design and simulation of flexible manufacturing systems. In:

Proceedings of the Ninth Mediterranean Conference on Control

and Automation Control, Dubrovnik.

Kumar, R., Tiwari, MK., Shankar, R., 2003. Scheduling of flexible

manufacturing systems: an ant colony optimization approach.

Proc. Inst. Mech. Eng. 217 (10), 1443–1453.

Lee, J., Pan, J.I., Kuo, J.Y., 2001. Verifying scenarios with time petri-

nets. Inf. Softw. Technol. 43 (13), 769–781.

Lewis, F.L., Gurel, A., Bogdan, S., Doganalp, A., Pastravanu, O., 1998.

Analysis of deadlock and circular waits using a matrix model for

discrete event manufacturing systems. Automatica 34 (9), 1083–1100.

Li, X., Lara-Rosano, F., 2000. Adaptive fuzzy Petri nets for dynamic

knowledge representation and inference. Expert Syst. Appl. 19 (3),

235–241.

Liu, R., Kumar, A., van der Aalst, W., 2007. A formal modelling

approach for supply chain event management. Decis. Support Syst.

43, 761–778.

Luggen, W.W., 1991. Flexible Manufacturing Cells and Systems.

Prentice Hall Inc., Englewood Cliffs, NJ.

Maleki, R.A., 1991. Flexible Manufacturing Systems: The Technology

and Management. Prentice Hall Inc., Englewood Cliffs, NJ.

Mandrioli, D., 1977. A note on Petri net language. Inf. Control 34 (2),

169–171.

Mejabi, O., (1988). Modeling in Flexible Manufacturing Systems

Design. PhD Dissertation, Lehigh University, Bethlehem, PA.

Murata, T., 1989. Petri nets: properties, analysis and applications.

Proc. IEEE 77 (4), 541–580.

Nandkeolyar, U., Christy, D.P., (1989). Using computer simulation to

optimize flexible manufacturing system design. In: Proceedings of

the 1989 Winter Simulation Conference.

Narakari, Y., Viswanadham, N., 1985. A Petri net approach to

modeling and analysis of flexible manufacturing system. Annu.

Oper. Res. 3, 449–472.

Nethery, J., Spong, M.W., 1994. Robotica: a mathematica package for

robot analysis. IEEE Rob. Autom. Mag. 1 (1), 13–20.

Noe, J. D., (1971). A Petri net model of the CDC 6400’. In:

Proceedings of ACM/SIGOPS Workshop on Systems Performance

Evaluation, pp. 362–378.

O’ Really, J.J., Lilegdon, W.R., (1999). Introduction to AweSim. In:

Proceedings of the 1999 Winter Simulation Conference. pp. 196–

200.

Peterson, J.L., 1976a. Computation sequence sets. J. Comput. Syst.

Sci. 13 (1), 1–24.

Peterson, J.L., 1976b. Petri nets. Comput. Surv. 9 (3), 223–252.

Petri, C., (1962). Kommunikation mit Automaten. Ph.D. Dissertation,

University of Bonn, West Germany.

Petri, C., 1976. General net theory. In: Proceedings of the Joint IBM/

University of Newcastle upon Tyne Seminar on Computing System

Design, Springer, Berlin, pp. 131–169.

Ruiz, C.M., Cazorla, D., Cuartero, F., Macia, H., 2009. Improving

performance in flexible manufacturing systems. J. Logic Algebra

Program. 78, 260–273.


Santarek, K., Buseif, I.M., 1998. Modelling and design of flexible

manufacturing systems using SADT and Petri nets tools. J. Mater.

Process. Technol. 76, 212–218.

Sarker, B.R., Krishnamurthy, S., Kuthethur, S.G., 1994. A survey and

critical review of flexibility measures in manufacturing systems.

Prod. Plan Control, 512–523.

Shnits, B., Rubinovittz, J., Sinreich, D., 2004. Multi-criteria dynamic

scheduling methodology for controlling a flexible manufacturing

system. Int. J. Prod. Res. 42, 3457–3472.

Solberg, J.J., 1981. Capacity planning with a stochastic workflow

model. AIIE Trans. 13 (2), 116–122.

Sun, TH., Fu, LC., (1994). An automatic petri-net generator for

modeling a flexible manufacturing system. Proc. of Int. Conf. on

Systems, Man Cybernetics, IEEE, 670–675.

Tavana, M., 2008. Dynamic process modelling using Petri nets with

applications to nuclear power plant emergency management. Int. J.

Simul. Process Mod. 4 (2).

Tuysuz, F., Kahraman, C., 2009. Modeling a flexible manufacturing

cell using stochastic Petri nets with fuzzy parameters. Expert Syst.

Appl.

Venkatesh, K., Masory, O., Wu, J., 1992. Simulation and robot

scheduling of a just-in-time flexible automated system using timed

Petri nets. Second Int. Conf. Autom. Technol. 4, 73–79.

Venkateswaran, P.R., Bhat, J., 2006. Fuzzy Petri net algorithm for

flexible manufacturing systems. ACSE J. 6 (1).

Viswanadham, N., Narahari, Y., Johnson, T.L., 1990. Deadlock

prevention and deadlock avoidance in flexible manufacturing

systems using Petri net models. IEEE Trans. Rob. Autom. 6 (6),

713–723.

Voss, K., 1980. Using predicate/transition nets to model and analyse

distributed database systems. IEEE Trans. Softw. Eng. SE-6, 539–

544.

Wang, FK., Yen, PY., 2001. Simulation analysis of dispatching rules

for an automated interbay material handling system in wafer fab.

Int. J. Prod. Res. 39, 1221–1238.

Zhou, M.C., DiCesare, F., Rudolph, D., 1990. Control of a flexible

manufacturing system using petri nets. Proc. IFAC Conf. 9, 43–48.

Zurawski, R., Zhou, M., 1994. Petri nets and industrial applications: a

tutorial. IEEE Trans. Ind. Electron. 41 (6).

Documents

Analysis of performance measures of flexible manufacturing ... · Abstract Flexible manufacturing system (FMS) is a highly integrated manufacturing system. The relation between its