AN INTRODUCTION TO USING PROSIM FOR BUSINESS PROCESS SIMULATION AND ANALYSIS

ed. S. Andradóttir, K. J. Healy, D. H. Withers, and B. L. Nelson

AN INTRODUCTION TO USING PROSIM FOR BUSINESS PROCESS SIMULATION AND ANALYSIS

Malay A. DalalMadhav ErraguntlaPerakath Benjamin

Knowledge Based Systems, Inc. (KBSI)College Station, TX 77840, U.S.A.

ABSTRACT

This paper introduces PROSIM, a knowledge-basedsystem for modeling processes, analyzing them usingsimulation and project management tools, and designingimproved processes. Domain experts who are familiarwith the various aspects of the system may beinterviewed and process models can be developed basedon the evidence collected using PROSIM. The tutorialfocuses on the built-in simulation design supportenvironment of PROSIM that enables building simulationmodels automatically from process models. The laterpart of the tutorial briefly covers extensions to PROSIMwhich permit information captured in PROSIM models tobe leveraged for other analysis—project planning in aproject management tool such as Microsoft® Project™.A purchase order processing system example isdescribed to demonstrate the utility of PROSIM.1 INTRODUCTION

PROSIM (KBSI 97a) can be used to support the analysisand design of manufacturing systems, business systems,logistics systems, command, control, communication,and intelligence systems, service systems, andinformation systems. It is a critical tool for realizing themonetary benefits of business process re-engineeringand activity based costing.

Originally developed under a research contract fromthe National Science Foundation (NSF) (KBSI 92a),PROSIM pioneered the concept of process modelsimulation. PROSIM has since evolved into a processengineering workcenter providing assistance during allphases in the lifecycle of process management:

1. Modeling—descriptions of existing process are

captured;

2. Analysis—performance metrics are estimated using

simulation or static analysis;

3. Design—new “to-be” processes are developed,

including integration of existing processes forformation of Virtual Enterprises.

As a complete process-engineering environment,PROSIM empowers the domain expert and reduces theirdependence on highly trained analysts. Additionally, itpermits the domain expert to focus on describing orimproving the process instead of abstracting the processinto a simulation model.

2 PROCESS MODEL GENERATION USING

PROSIMPROSIM enables a modeler to capture processes orsituations as an ordered sequence of events or activities.It is designed to capture process descriptions of theprecedence and causality relations betweenevents/situations in a form that is natural to domainexperts in an environment. One of the primary goals ofthe tool is to provide mechanisms for expressing domainexpert’s knowledge about how a particular processworks.

PROSIM uses “scenarios” as the basic organizingstructure for establishing the focus and boundaryconditions for the process description. This feature ismotivated by the tendency of humans to describe whatthey know in terms of an ordered sequence of observedactivities within the context of a given situation.

The basic syntactic unit of PROSIM graphical processdescriptions within the context of a scenario, is the UnitOf Behavior (UOB) represented by a rectangular box.Each UOB represents a specific view of the world interms of a perceived state of affairs or state of changerelative to the given scenario. Stated simply a UOB mayrepresent an event, activity or process in the system ofinterest. UOBs are connected to one another viajunctions and links. Junctions provide the semanticfacilities for expressing synchronous and asynchronousbehavior among a network of UOBs. Links representtemporal precedence, object flow, and relations between

718

An Introduction to Using ProSim for Business Process Simulation and Analysis 719

UOBs. Each UOB can have associated with it bothdescriptions in terms of other UOBs (a decomposition ofa parent UOB) and a set of participating objects and theirrelations (an elaboration).

Let us consider a situation of an analyst interviewinga purchase department manager to capture mainactivities involved in generating purchase orders at the

3 GENERATING SIMULATION MODELS

USING PROSIMThe intelligent support provided by PROSIM reduces theexpertise required to design effective simulation modelsand perform simulation experiments. It provides anenvironment in which users can develop designs for

Figure 1: A Process Flow Diagram of the Purchase Order Process

company. The purchase department manager is mainlyresponsible for the processing of purchase request forms,finding existing suppliers for the part, inviting bids frompotential suppliers, receiving bids, evaluating bids,placing purchase orders, and acquiring the material.First, the initiating department (typically design orprocess planning) formally requests the needed material.If the requested material has existing suppliers, apurchase order can be placed immediately. On the otherhand, if the requested material is new, or the existingsupplier no longer produces the part, then the purchasedepartment has to invite bids from potential suppliers,receive the bids, evaluate the bids, and place a purchaseorder to the selected supplier. The process flow diagramthat depicts the sequence of activities involved in thepurchase order generation process can be captured usingPROSIM and is shown in Figure 1.

discrete event simulation models from processdescriptions and supports the conceptual modelingactivities at various stages of the model design process.PROSIM can be used to support the analysis and designof manufacturing systems, business systems, logisticssystems, command, control, communication, andintelligence systems, concurrent engineering systems,service systems, and information systems. It is a criticaltool for realizing the monetary benefits of businessprocess re-engineering and activity based costing.

PROSIM has a built-in simulation model designsupport environment. In other words, it has an expertsystem programmed to provide expertise at a levelcomparable to that of a simulation-modeling expert.Process flow diagram constructs, UOBs, objects, andjunctions, are required to be detailed within PROSIM inorder to generate a simulation model. Consider therequest for bids UOB in the example process flowdiagram in Figure 1. The real world objects that

720 Dalal, Erraguntla, and Benjamin

participate in the process of requesting bids are“Technical Package,” “Purchase Department Manager,”“Potential Supplier List,” “Bid Package,” and “Building1.” A “technical package” contains the technical data ofthe part to be ordered. The technical data of the partmay include engineering specifications, tolerances,material to be used, special tools and jigs required tomanufacture the part. A “Purchase DepartmentManager” requests a bid by sending a technical packagealong with the bid package to potential suppliers. The“Potential Supplier List” consists of details of all thepotential suppliers such as “supplier code,” “supplier

Figure 2: Dialog Box for Detailing a Unit of Behaviorname,” “supplier address,” “supplier manufacturingcapabilities.” The “Bid Package” consists ofinformation about the requirements of the bid responses,financial details, and other contractual details. Thisactivity is performed in the location “Building 1.”

In PROSIM, a dedicated resource is categorized aseither a location or a queue. The location label isprovided whenever a non-waiting activity is representedby the UOB. If the UOB is used to represent a waitingactivity, the dedicated resource is associated as a queue.Flow objects that participate in UOBs are called entities.3.1UOB Detailing

The dialog box shown in Figure 2 is used to record theprocess times for the Request Bids UOB. Process timesare recorded for the entities that participate in theprocess. The names of the entity objects that participatein this UOB are New Material Request Form (labeled“New Order”), “Bid Package,” and “TechnicalPackage.” The process time for “New Order” is aconstant of 0.5 days. The process time for “BidPackage” is a normal random variable with a mean of 5days and a standard deviation of 1 day. The processtime for “Technical Package” is a normal randomvariable with a mean of 12 days and a standard deviation

of 2 days. The Objects such as “computers,” “personnelmanager” are categorized as resources in PROSIM.Resource objects, as well as other objects, can be sharedbetween different UOBs in a PROSIM model.3.2Entity Detailing

Entities flow between UOBs in a scenario. The inter-arrival time distribution is an item of informationrequired to simulate the behavior of entities in a system.Figure 3 displays how the details of “New MaterialRequest Form” are recorded in a PROSIM model.Observe that the inter-arrival time for this entity is anegative exponential random variable with a mean of10.0 time units. The arrival point is the first processvisited by the entity as it flows through the scenario.

Figure 3: Entity Detailing

3.3Decision Logic Specification

Junctions in PROSIM provide a mechanism to specify thelogic of process branching. Different junction types aresupported in PROSIM to aid in capturing the semantics ofbranching in real-world processes (Figure 4). Junctionssupport the description of 1) a process that splits intotwo or more process paths, or 2) two or more processpaths converge into a single process. Junctions areclassified in three different ways. First, they areclassified according to the logical semantics involved:AND (&), OR (O), and exclusive OR (X). Junctions arefurther classified as either fan-in or fan-out, based onwhether they represent a convergence or a divergence inthe logic of the process description. They are alsoclassified based on the coordination of the timing of theassociated UOBs as either synchronous or asynchronous(KBSI 1992b).

An exclusive OR (X) junction is used to modeldecision logic. In our purchase order processingexample, an exclusive OR junction is used to representthe logic that activity “request material” is followedexactly by one of the following two activities: 1) “order

An Introduction to Using ProSim for Business Process Simulation and Analysis 721

from an existing supplier”, or 2) “request bids from newsuppliers” with probabilities of 0.8 and 0.2, respectively(Figure 5).

Figure 4: Adding a Junction

Figure 5: Junction Detailing

3.4Performance Metric Specification

The goal in designing a simulation model is to generatedata to answer a set of questions. Examples of questionsinclude, “What is the total process time?”, “What is theresource utilization?”. PROSIM provides support fordesigning instrumentation to help answer thesequestions. In other words, PROSIM helps you selectappropriate performance metrics that will automate thegeneration of output data to answer the questions. Forexample, the performance metric used to answer the firstquestion is Process Time for the UOBs “preparation ofbid package,” “preparation of technical package,”“filling in new material request form,” and “filling in therepeat material request form.” The PROSIM dialog boxesused to specify these performance metrics are shown inFigures 6 and 7.

Figure 6: Simulation Performance Metrics

Specification

Figure 7: Metrics to Compute Process Time

3.5Model Verification

In the context of PROSIM, model verification is definedas the process of ensuring that the minimum amount ofinformation required to generate a WITNESS simulationmodel exists in the process model. Model verification isperformed using the Model Verification option in theDiagram menu in PROSIM. A more detailed descriptionis given in the PROSIM User’s Manual and ReferenceGuide (1997).

3.6Simulation Model Generation

PROSIM simulation model design specifications are nowtransformed to an equivalent representation in theWITNESS Command Language. This is done by usingthe Build Simulation Model option in the PROSIM Filemenu. After the model has been generated, the next stepis to open the application and load the model using the

722 Dalal, Erraguntla, and Benjamin

Read Commands option in the Model menu (see LannerGroup 1997). After the WITNESS model has beenloaded, the user can run the model and generate thedesired performance metrics. WITNESS provides state-of-the-art capabilities for simulation-based analysis andexperimentation. An analyst knowledgeable in theWITNESS language may customize the PROSIM-generated model if necessary.

A unique feature of PROSIM is the ability to generatemultiple simulation models from one process model.The value of this capability is immediately obviouswhen one realizes that there is no such thing as “the”simulation model; each model is built to answer specificquestions. PROSIM not only permits a hierarchicalprocess description but also allows simulation model tobe generated at different levels of abstraction. In otherwords the user can specify whether to include activitiescontained in the decomposition of a high level activity(UOB) in the simulation model. For an initial analysisthe process may be simulated at a high level and finerlevel of granularity may be specified for a more robustsimulation later on.

In addition to simulating a process at different levelsof abstraction, PROSIM allows the user to generate twodifferent types of simulation models: Location-constrained, and Activity-centered. In the former,PROSIM Locations, e.g., desk, department, machine, etc.,are the focus of the simulation and an Entity may visitthe Location several times to participate in the differentProcesses that occur at that Location. In the laterapproach, there is a one-to-one relationship betweenProcesses, i.e., UOB, and Machine in WITNESS. Aunique conceptual Location is created and assigned toeach Process to build a process-centered simulation.During simulation, an Entity makes multiple visits to aMachine only if it participates in the same process morethan once. Physical locations, such as desk, department,and machine, are modeled as Resources if they are aconstraining factor. The Location quantity should reflectthe number of parallel activations of the Process. In thecase of unlimited parallel activations, the Locationquantity should be at least equal to the maximumnumber of Resource available for performing thatProcess. This latter approach is useful in activity-centered analysis, which requires answering questionssuch as “Which activity (or process step) is thebottleneck in the purchasing process?”

4 PROJECT MANAGEMENT USING PROSIM

AND PROJECTLINKPROJECTLINK (KBSI 1997b), an add-on module toPROSIM, enables information in a PROSIM process modelto be leveraged for analysis using a project management

tool, e.g., MS-Project. PROJECTLINK facilitates analternative approach to process analysis since a detailedsimulation-based analysis is often unnecessary in orderto obtain an estimate of metrics such as resourceutilization. Static analysis using the resource levelingfunction in project management tools may be adequatefor identifying resource contention problems resultingfrom the process design. Additionally, Critical Pathanalysis may help identify process steps that need to beperformed in parallel with other activities so that theoverall cycle-time is reduced. A link between processmodeling and project planning is also useful during theimplementation of large, perhaps infrequently occurringprocesses such as a shuttle-launch at NASA.

Since PROJECTLINK is a two-way bridge betweenprocess modeling and project planning, it gives rise toother useful and interesting possibilities. Projectplanning analysis can identify limitations and mandatechanges to the project plan. The resulting changes canthen be brought back to the process model in PROSIM.Alternatively, an existing project plan can be importedinto PROSIM for performing dynamic analysis using thePROSIM-generated WITNESS simulation model.

The implementation of PROJECTLINK makes theexchange of information between a process model andproject plan (in either direction) transparent to the user.In most cases, only one piece of additional informationis needed to go from a process model to a project plan:the user selects one path from the options available ateach “OR” or “XOR” junction. The process model isthen exported using the Export menu item in the Filemenu. PROJECTLINK is then activated from the LaunchApplication menu item. A file containing the projectplan is generated in the Microsoft Project Export format(.MPX file) and the project management tool is launchedusing this file. Figure 8 shows the PROJECTLINK-generated project plan for the Purchase Order processmodel described earlier.

Figure 8: A Project Plan of the Purchase Order ProcessConverting a project plan into a PROSIM processmodel is just as easy to do. The file is saved in the.MPX format and PROJECTLINK is launched by selectingthe “Launch ProjectLink” menu item (appended to the

An Introduction to Using ProSim for Business Process Simulation and Analysis 723

Tools menu in MS-Project during installation.) Whenbringing in a project plan into an existing process modelthe user has the option of merging the changes with theolder process model, or creating a new process model.5 PROCESS IMPROVEMENT USING PROCESS

DESIGN ASSISTANTThe most recent extension to PROSIM is the ProcessDesign Assistant (PDA): a knowledge-based designassistant that facilitates process design by analyzingprocess models for consistency, completeness, andpotential improvement opportunities (KBSI 97b). Aprocess designer can use the PDA to diagnose possibleerrors in the model and identify areas for improvement.Given a process model, PDA will check whether theinformation is complete and consistent with respect tothe process design intent. It performs different kinds ofqualitative and quantitative analysis and produces avariety of metrics that are useful for processimprovement and re-design.

PDA consists of a repository of rules that can beexecuted on selected process models. Each of the ruleschecks for some desired feature of a complete,comprehensive and valid process model. PDA wasdeveloped in a manner to facilitate addition of new rulesand customization of how the rules are organized. Thescreen below provides a means to organize, manipulate,select and execute rules. The rules in PDA areorganized into two classification schemes: based on RuleClass and based on Rule Domain. Rule Domain refersto the domain of application such as BPR, CPI, leanlogistics, concurrent engineering, CALS/EDI, etc. RuleClass refers to the basic rule types such as completeness,consistency, cycle-time improvement, qualityimprovement, coordination, metrics, etc. These twoclassification schemes are used to help browse throughthe rules and select the most appropriate rules forexecution. The PDA rules may be customized fordifferent domains by indicating which rules classes arerelevant.

During execution the user simply selects the processmodel to be analyzed and selects the set of rules to beexecuted as shown in Figure 9. PDA executes allselected rules sequentially on the selected processmodels and identifies the problems or potentialimprovement opportunities with the selected model. Theanalysis results are displayed as shown in Figure 10.PDA also contains an Agenda Manager, which can beused to keep track of the tasks to be performed to fixproblems with different process models. PDA analysisresults can be automatically updated as action items inthe Agenda Manager for the selected process model.

Figure 9: Rule Selection in PDA

Figure 10: Results from a PDA Process AnalysisA demonstration prototype of the PDA has beendeveloped and will be available commercially in the nearfuture.

6 POTENTIAL BENEFITS OF USING PROSIMPROSIM provides an integrated approach to enterpriseprocess engineering through the integration of: 1)modeling methods with performance analysis methods,and 2) qualitative and quantitative techniques. As asimulation model generator, PROSIM has the followingpotential benefits: 1) it enables novice simulationmodelers to develop and use complex simulationmodels, 2) it encourages and promotes the use ofsimulation as a decision support tool among a widegroup of users from a variety of application domains, 3)it substantially reduces time and effort needed to developsimulation models, and 4) it facilitates the capture andretention of simulation model design rationale.

724 Dalal, Erraguntla, and Benjamin

7 SUMMARY

This paper described salient features of PROSIM, aknowledge-based tool for process engineering. Anexample of purchase order processing system isprovided to demonstrate the procedure of buildingprocess models from domain experts’ descriptions andthen generating a simulation model from this processmodel. The paper also described extensions to PROSIMincluding PROJECTLINK a two-way bridge betweenprocess modeling and project management tools, and aProcess Design Assistant to facilitate improved processdesign.REFERENCES

Knowledge Based Systems, Inc. (KBSI). 1992a. IDEF3method report. Prepared for the U.S. Air ForceHuman Resources Laboratory, Contract No. F33615-C-90-0012.

Knowledge Based Systems, Inc. (KBSI). 1992b. Issuesin knowledge-based assistance for simulation modeldesign from IDEF3 descriptions. Technical Report forNSF SBIR Phase II Grant No. III-9123380.

Knowledge Based Systems, Inc. (KBSI). 1997a. PROSIMuser’s manual and reference guide, Version 2.2.1,College Station, TX.

Knowledge Based Systems, Inc. (KBSI). 1997b.PROJECTLINK user’s manual and reference guide,Version 1.1, College Station, TX.

Knowledge Based Systems, Inc. (KBSI). 1997c. VirtualEnterprise Engineering Environment (VE3).Technical Report for DARPA Contact No. F33615-96-C-5601.

Lanner Group Inc. 1997. WITNESS user manual,Release 8, Cleveland, OH.AUTHOR BIOGRAPHIES

MALAY A. DALAL received his M.S. and Ph.D.degrees in Industrial and Systems Engineering in 1991and 1996 respectively, from Virginia PolytechnicInstitute and State University (Virginia Tech). AsResearch Scientist at KBSI, Dr. Dalal develops andapplies software tools for activity, cost, data, function,process and simulation modeling. He has been involvedin the design of the Process Design Assistant for theVirtual Enterprise Engineering Environment (VE3)project, and the implementation of simulation-basedscheduling systems at Tinker AFB and Hill AFB.MADHAV ERRAGUNTLA received his M.S. inIndustrial Engineering from the National Institute forTraining in Industrial Engineering in 19 and Ph.D. in

Industrial Engineering from Texas A&M University in1996. Dr. Erraguntla conducted research and developedapplications in planning, simulation, costing, knowledgebased systems, optimization, neural networks and fuzzylogic. He has numerous journal publications andconference proceedings. He designed and developedPROJECTLINK and process design assistant (PDA)presented in this paper. Currently, Dr. Erraguntla isworking as a Research Scientist at KBSI.

PERAKATH C. BENJAMIN received his Master ofScience degree in Industrial Engineering from theNational Institute for Training in Industrial Engineeringin 1983. He received his Ph.D. in Industrial Engineeringfrom Texas A&M University in May 1991. As VicePresident (Innovation and Engineering) at KnowledgeBased Systems, Inc., Dr. Benjamin manages and directsresearch KBSI’s research and development projects. Dr.Benjamin’s is the principal architect of PROSIM and isresponsible for its continued evolution.