IntroductionEnhancing simulationDesigning manufacturing systems beforesoftware for use inimplementing them in practice is a task thatmanufacturingsimulation modelling is ideally suited to,given the complexity and dynamic andstochastic behaviour of such systems. One ofL.P. Baldwinthe advantages of using simulation is that itT. Eldabican be used to provide a high level ofV. Hlupic andutilisation of manufacturing facilities in bothSMEs and larger organisations by ensuringZ. Iranithat any bottlenecks or problems which mightprevent a smooth and uninterrupted flow ofThe authorswork are reduced or eliminated. Anotherbenefit is that simulation provides a betterL.P. Baldwin and T. Eldabi are Lecturers, and V. Hlupicunderstanding of the behaviour of theand z. Irani are Senior Lecturers, all in the Department ofcompletemanufacturing system within anInformation Systems and Computing at Brunel University,SME (or larger organisation), or particularUxbridge, UK.parts of it. As many problems relating to thedesign and operation of manufacturingKeywordssystems are usually too complex to allowManufacturing systems, Computer simulation,rigorous mathematical treatment, it is notSoftware development, Europe, Logisticspossible to represent such problems bymathematical relationships without having toAbstractmake a very large number of assumptions.The mathematical relationships can thereforeThe design of manufacturing systems is a complex andno longer represent, in a realistic manner, theexpensive task for both SMEs and large organisationsproblem being investigated.Such problemsalike. Many problems relating to the design and operationcan, however, be investigated byof manufacturing systems are too complex to allow forexperimentation, although in the workplace itmathematical treatment,andasaresultthereisis very difficult, or often impossible, to carryincreasing recognition of the applicability of simulation inoutexperiments on theactual manufacturingthe manufacturing environment. This has in turn resultedsystems themselves. Simulation can usefullyin a growth in simulation software products available inbe employed to carry out such experimentsthe marketplace, and manufacturers of such products are(Kochhar, 1989).Pegden et al. (1990)naturallykeentoextol thevirtues of whattheysell.Inahighlight a further benefit of using simulationrapidlychangingworkplace,thereissomesuggestionthatmodelling techniques, namely that they cansimulation tools may not have developed to deal with thebeused to predict the behaviour of complexnew demands, and so it is timely to review what is onsystems by calculating the movement andoffer. This paper presents the results of a survey ofinteractionofsvstemcomponents.PegdenetEuropean simulation specialists, both working in industryal.(1990)go onto saythat simulationallowsand in an academic environment, on the use of simulationfordetailed analysisandevaluationofnotsoftware. The main purpose of this survey was toonly physical layouts but also equipmentdetermine how satisfied users are with simulationselections and operating procedures as well.software and to highlight the most positive and negativeEvaluation of the movement of parts throughfeatures. The findings offer an insight into currentthe machines and workstations, together withsoftware, and how this can be further enhanced.examination of the conflicting demands oflimited resources, at thedesign stage, haveElectronic accessclear benefits in terms of increased efficiencyThe current issue and full text archive of this journal isand lower costs.It is true to say that computeravailable atsimulation has now become one of the mosthttp://www.emerald-library.comwidely used techniques in manufacturingsystems design, as it enables both decisionmakersandmanufacturing engineers in bothLogistics Information ManagementSMEs and large organisations to study theVolume 13 - Number 5 - 2000 - pp. 263-270MCB University Press -ISSN 0957-6053complexityoftheirsystemsandhowchanges263
! " # $%&%! ' $%%(! ) ' $%%(! ) * ) ! " # $ " $ ! # ! ! $ $ ! !$ ! ! ! ! # " !$ ! # ! ! ! ! # ! ! # % ! & '( ) * +, +-(.+/, 0 ) ,1*/2-,*(
Enhancing simulation software for use in manufacturingLogistics Information ManagementLP. Baldwin, IT.Eldabi, V. Hlupic and Z. IraniVolume 13 - Number 5 - 2000 :263270in system configuration or operating strategylanguage. Although this approach providescan affect overall performance.flexibility,it is time consuming and expensive.Thegrowingpopularity of simulationhasSMEs, as with all organisations, need to makeresulted in a growth in the number ofeffective use of limited time and resources,simulation languages and simulators availableand costs incurred of buying the softwareon the softwaremarket (Hlupic, 1994; Lawneed to be balanced against the assumedandMcComas,1997).Although spacedoesbenefits.A simulatorallows themodelling ofnot permit here to list all of the latest products,the problem with little or no programming,a few ofthem are now described, as these give awhich might significantlyreduce the"flavour"of the literature.With the currentmodellingtime,butonlyifthesystemunderinterest in object oriented design, Healy andconsideration fits the domain of the package.Kilgore(1997)promotethevirtuesofmodelGiven theparticular difficulties of SMEs,itsimplicity and reusability of Silk, a generalmay well be useful for them to further explorepurpose language which has its roots in thelinks with academic institutions, asprogramminglanguage Java.They go on to sayinformation systems orcomputersciencethat Java and other Internet relateddepartments will be able to provide usefultechnologies have the ability to deal with thesupport. It is likely that SMEs do not haveshortcomingsofsimulationtoolsalreadyinsufficient numbers of modellers, orexistence, and that Silk will encourage bettersufficientlywideexperienceofmodelling,sosimulation. Quite how this is achieved is notthey may benefit from closer links withexplained, unfortunately.Pritsker and O'Reillyacademic institutions in that such expertise ismore readily accessible, and available.With(1997) describe AweSim, a new generalpurpose simulation system which has thesuch a wide choice of simulation toolsadvantage of "openness and interconnectivity"available, choosing one is not an easy task,Particularly for an SME with limited(p.545)to standard word processingprograms, spreadsheets and databases.Yetmodelling experience. As the aboveanother new simulation system is described bydemonstrates, the claims that the creators ofEhrlich and Lilegdon (1997,p. 565), who extolsuch tools make are persuasive, with eachthe virtues of FACTOR/AIM, a simulationpromising efficiency and effectiveness and allsystem specifically designed to be used in thefor less time and money.decision support of manufacturing. ThisThis paper reports not on the claims of thedesigners of these new, or not so new,system, say the authors,makes modelling easierand faster and allows decision makers moresystems, but on those who use them. It will betime to think about their decisions, and thusinteresting to see whether the claims made forcome up with better decisions that can onlythe softwaremeet the expectations of thoseserve to enhance the manufacturing processes.who buy them. Given the high cost of many ofSLX is, according to its designers, “a well-these, choosing the appropriate tool for thetask is important for both SMEs and largeconceived, layered simulation system"whichhas all the convenience of a high-level packageorganisations alike, not only in terms of(Ehrlieh and Lilegdon, 1997, p.565). Incapital outlay but in time and effort as well.contrast to the novelty of Silk, FACTOR/AIMThis paper presents and discusses the resultsand SLX, Crain (1997)iskeen to point out theof a survey on the use of simulation software.superior qualities of GPSS/H which, althoughA survey of a number of European simulationnot new, is"one of the most general, flexiblespecialists, both in the industrial andacademic environments, was carried out toand powerful" simulation tools on themarketplace today, and is used in a wide varietydiscover whether users are satisfied with theof working environments (p.567).Particularlysimulation software they use, and how thisuseful for manufacturing is SiMPLE++, say itssoftwarecan befurther improved.Despitethecreators,Kalasky and Levasseur (1997),whoupbeat tone of the creators of the toolsclaim that it is “a highly efficient, highly flexibledescribed earlier in this section,tool" which has been used successfully in manyimprovements are clearly needed as,organisations throughout the world.according to Pratt andBeasley (1997),Whether using this language or any other,although the needs of the workplace havewhen a model is developed using a simulationchanged, simulation tools have not altered atlanguage, the simulation analyst has to writeathe same pace, and more efficient andprogram usingthemodelling constructs oftheeffective ones are required, both in SMEs and264
" + $%%,- . / $%%0! 11 22 3 4 + # $%%0! 5 " 5 * 6 ' 27 $%%0! 11 22 8,8! 9 . $%%0 8:8! ; /"7< * " .= 11 > 22 > . $%%0 8:8! * ; /"7< * .= / $%%0! ) ?'<+ 11 22 8:0! ' '. @@ # . $%%0! 11 22 3 ? * 3 " * ? " ' A $%%0! ) " ! 0 ! # " & 3 4 & '( ) * +, +-(.+/,
Enhancing simulation software for use in manufacturingLogistics Information ManagementVolume 13 - Number 5 - 2000 : 263270LP. Baldwin, T. Eldabi, V. Hlupic and Z. Iranilarger organisations.Kreutzer et al. (1997,sample of simulation packages according tovarious criteria. On the basis of the responses,p.483)concur, saying that commercialsimulation tools have not developed alongsidethey classified the software evaluated intothe rapid pace of technological change andclusters according to the main softwarefeatures.Kirkpatrick and Bell (1989)used athat, as a result, many popular simulationtools today “are riddled with non-intuitivesurvey approach to investigate the issuesinterfaces and inconsistent and unnecessaryrelatedto visual interactivesimulation inrestrictions". In their view, it is not possible toindustry, reporting that these issues includeanticipate everything that any given user willthe types of problems being addressed,need, and that good tools must allow for safereasons for using visual interactive modelling,and convenient extensions, refinements andand the ways in which this type of modellingadditional layers.Following a brief review ofaffectsproblemsolving.Theresultsrevealedsimulation research in the area of simulationthat although some of the participants weresoftware analysis using the survey method, aaware of the significant set-up costs and thedescription of the actual survey is presented,demands associated with learning newand the results obtained are examined andsoftware and a new methodology, mostdiscussed. The conclusions outline the mainparticipants agreed that visual interactivemodelling provides enhanced interaction withfindings of this research.decision makers, more useful and easier-to-understand models, and better decisions.The main findings to be drawn from theSurveys of users in the simulationanalysis of simulation research studies dealingliteraturewith users' surveys reveal that although themajority of the survey studies investigateBefore the questionnaires for the survey usedissues related to simulation software, none ofin this research were designed, a literaturethem have examined the opinions oftheusersreview was undertaken in order to findthemselves about possibleways in which theexamples of any similar previous studies.Thissoftware could be improved.review found several examples of userssurveys in the simulation literature, althoughmany were, like the survey carried out byKleine (1970, 1971), somewhat dated.A survey of simulation software forKleine's research examined users'views of 11manufacturingdiscrete simulation languages. The results ofThe main purpose of the survey was tothis survey showed that it was difficult tointerpret the results, and that this was mainlyinvestigatetherequirementsofEuropeandue to the fact that there was a limitedusers of simulation software, and especially ofnumberofrespondents whowereproficient insoftwareused for manufacturing simulation,with a view to soliciting their opinions aboutmore than one language.In addition, theexpertise of some respondents was difficult towaysof improvingcurrentsimulationsoftware tools to better satisfy their needs.specify.Christy and Watson (1983)used asurvey of non-academic users to exploreThe questionnaire distributed to theissues such as thefunctional areas that useparticipantsinthesurveyconsistsof ninequestions dealing with the type of simulationsimulation, the method of selectingsoftwareused (Figure 1),the specification ofsimulation software, the popularity of variousparticular packages used (WITNESS,softwaretools for simulation applications andSIMFACTORY II.5, SIMAN/CINEMA,other suchfactors.TheiranalysisrevealedProModelPC,XCELL+,INSTRATA orthat ofthetotal applications of simulation,59 per cent were in the area ofmanufacturingFigure 1 Type of software usedsystems.For simulation software, the resultsshowed that there is a widespread reluctanceto implement and learn new programming38%DSIMULATCRlanguages for simulation applications. VanBSIM&LANGLAGEBreedam (1990)conducted a survey in order57%DLANGUAGESto evaluate several simulation software tools.Experienced users of simulation were given a5%questionnaire to complete in order to rate a265
# B $%%0 ,&C! 11 > 22 * ; " A ) " 2 # $%0( $%0$! #2 2 $$ " * / 3 $%&C! > " 8% ; D A $%%(! ) # A $%&%! " > > > " 2 4 " ) " ) ) ; $! 3*"E * ; /"79 **8 * E</*E ' '/ =/ .@ *E"7 " "# $ # ! " ! 0 ! # " & 3 4 & '( ) * +, +-(.+/,
Enhancing simulation software for use in manufacturingLogistics Information ManagementLP. Baldwin, T. Eldabi, V. Hlupic and Z. IraniVolume 13 - Number 5 -2000 :263270other) (Figure 2),the purpose of usingFigure5Asummaryofusersopinionsaboutthemainlimitationsofthesimulation (Figure 3),general opinions aboutsoftware usedeach software item used (Figure 4) and the36%types of systems being modelled (Figure 5)30%Otherquestionsincludeanestimationofhow25%successful the simulation studies carried out20%were from the point of view of the software15%used (Figure 6). In particular, users had to10%appraise whether substantial approximations5%0%hadto be made due to limitations of theQ1Q2Q304Q5software, or whether all desirable features ofQ607the systems under consideration could beQuestions askedmodelled.The participants were also asked toQI:Restricled BexibilityQ2:SlolistthemainweaknessesandlimitationsoftheQ3:Validaticn difficultiessoftware used (Figure 7), as well as the mostO4:Lack of facilityfoeouputarahysisQS:DiffictittouseDifficult o lenFigure 2 Number of packages usedLack of facility foeouput analysisPoce statistics3%5%Mamufacturing hias and tenminclogy pnoblemQ6:Lackofdataluse linkages13%LimitstosizeofmodelsD1PACKAGEEspensive2PACKAGEQ7:Inodequate graptics5%D3PACKAGELack ofgood suppont foe fluid procesingD4PACKAGELack of good suppont fieobjacl oricmed conceptsBig models are mol understandahle10%SPACKAGE64%D6PACKAGEFigure 6 A summary of users' opinions about the most important positivefeatures of the software usedFigure 3 Purpose of simulation40%8%CMOD.REAL30%39%SYSTEMS25%BEDUCATION% ef U6ers 20%53%15%DMOD.REALSYS10%&EDU%J08%0%Q10293040506CuostionshsdFigure 4 A summary of users' general opinions about the softwareQFGraphics (animgtion)O2Fane ofurseQ:Ease oflearing40%Automatic repont. generationQ4:Imeractivity30%Ulser suppertUser inierface20%Sped ofmodillingQ5:FirsibilityDocuntentation10%Good statisaical analysis6:Imerfiace with other sodtwere0%Sappont foe UNIX plalformsQ1Q2Q3Q4Incorpcealed cod amailysisQ5QBQ7Q8Eaay chst ofwhai-ifquestionCheapQL:Too linited for complexproHemBeing mxnu drivenQ2:Good graphics03:SowQ4:Easy totseimportant positive softwarefeatures (Figure8).Generally very good interactiveQ5:Easy to leamFinally,they were asked to specify the mostBiuscdwosimulationproemQ5:Pocr statistical supportimportantfeaturesthatshouldbeincludedinUar friendlyQ7:Dificuit to validateexisting simulation packages, and that are toInsdeqaate esperimentation facilitiesthebestoftheirknowledgenotyetprovidedDificuk touse Sornon standard problemsQ8:Quick(Figure 9). The majority of the questionsLack of good support for fluid processingregarding opinions about the software, and266
! ; F! ; C! ; ,! ; 8! ) ; :! * " ; 0! ; &! ; ; %! " 4 ) "# % ) $ "# & 0 "# ' 5 6 ! "# ( 5 6 ! "# ) 5 6 ! " ! 0 ! # " & 3 4 & '( ) * +, +-(.+/,
Enhancing simulation software for use in manufacturingLogistics Information ManagementLP. Baldwin, T. Eldabi, V. Hlupic and Z. IraniVolume 13 -Number 5 -2000 .263270The sample frame and response rates ofFigure 7 Type of systems simulatedthe survey23%The survey sample includes a number ofSONGMAN.SYSregular simulation users both in educationalinstitutions and in industrial companiesBMANUFACT&OTHERSaround Europe.The survey sample was notOOTHERselected by any formal statistical method, and41%the participants, known or believed to beregular users of simulation, were selecteddeliberately.It was intended to obtain aFigure 8 The modelling success from simulationssample of users experienced mainly in usingsimulators (referred as simulation packages in31%31the questionnaire) rather than languages.DALLFEATUREMOOELResponse was moderate, with 30 per cent ofMOSTFEATURESMOCELDMJORPROBLEMSthe 120 distributed questionnaires beingreturned. It should be noted that the ratio ofresponses from universities (70 per cent) and38%responsesfromindustry(30per cent)did notmatch the distribution of questionnaires,Figure 9 A summary of users' opinions about the features that should bewhere an equal number of questionnaires wasincluded in simulation softwaredistributed to each group of users. It is18%interestingtonotethatnot onlywasthe16%numberofresponsessignificantlyhigherfrom14%users from universities, but that on average12%each response from a university provided%orUsens 10%more information than the response from8%users in industry.Further research might6%usefully reveal why this was the case. These4%figures raise the question of the statistical26significance of the results obtained. Although0%Q1Q2Q3O4C56Q7there are as yet unanswered questions as to0809Q10Q11Quesiors Asksdthe response rates, the authors neverthelessclaim that the selecting of participantsQ1:Better sodtware compatibility8Link to datatoseexperienced in simulation serves to justifytheLink to spreadshieets04:Lisk to CAD sofisareimportance of the results, as such users areQ5:Link to statistical packagevery much representative of those who useOLinktoMRPsthedalingQ7:Facilityforoatput analysissimulation tools.QS:More flexibilityHep inesperimemal designQ9:Betcrandmoreiteligoton-linehcipResults of the surveyBetcrExperimenttionfaciliticwFigure 1 shows the results obtained regardingSuppoet standard progranming oonceptQt0:Eliminaticnfmemocylinitstionsthe type of software used. A majority of theBetter docunertationEasy model editingusers use only simulators (56.5 per cent), 38.4Dedicated systemsfoespeeifieapplieationsper cent use both simulators and languages,Highcr esecuion spend01lAhility to cresde run-ine seplicitionand 5.1per cent use only simulationlanguages.AuiomotivesaxeAn analysis of the number of simulationMoreprompttosaseHierarchikcal modelbuildingsoftware tools used reveals that a majority (64.1LoweostsoftuareEany cost ofcn-linereportsper cent) of the users use only one softwareAvailahility on slundrd hedwire ind sotlwire systermstool, but the other 35.9 per cent of users usemore than one software, up to six differentpossibleways of improving it (questions 4,6,software packages as shown in Figure 2.7,8 and 9), were open-ended.Open-endedRegarding thepurpose of simulation,38.5perquestions are, in the view of the authors,centofparticipantsusesimulationonlyforpreferable as they are more likely to generatemodelling real systems, 7.7 per cent useless biased responses than can be said to besimulation only for education, whilst thethe case with closed questions.majority of 53.8per cent use simulation both267
) , : 0 & %! > > ) ) " " * ) ! 7 C( $F( ) * 0( ! C( ! ) ) ) * ; " ) ) 4 ; $ 4 8:8 ! C&, 8$ 4 :,$ ! C8% ; F 7 C&8 00 4 8C& "# * # "# + # "# , 5 6 ! " ! 0 ! # " & 3 4 & '( ) * +, +-(.+/,