1. Introduction

In recent decades discrete event simulation has become a much-used tool for logistic analysis of a large variety of operations management problems. Its popularity is boosted by the advent of visual interactive simulation (Hurrion, 1976; Bell, 1985, 1990; Au and Paul, 1996; Robinson, 2005). The possibility to visualize alternative system designs and their dynamics is helpful, both as a means of communication between analyst and decision makers, and for model building. It is therefore striking to see how the dominant view on simulation use, as described in many course books and implemented in software, has hardly altered since then. It still considers the use of simulation as just a methodology to analyse design decisions. This conforms to the 'hard' OR paradigm of optimizing systems of operations. In this view, the analyst has a major role in the design of models and solutions, building on representative data on system design and behaviour. Visualization is seen only as supportive for checking model correctness and validity and demonstrating simulation outcomes. It is no topic in most of the course books.

The 'hard' view on simulation use pays little attention to the separate role of visual simulation models in facilitating active user participation, for creating better and accepted solutions (Robinson, 2001). Moreover, uses of simulation other than analysis, which primarily rely on visual interaction, are easily overlooked. Examples include simulation being used as a technique for knowledge elicitation (Robinson et al, 2005), or as a means for staff training or student education in operations management (see, eg Zulch and Brinkmeier, 1995; Adelsberger et al, 1999; Chwif and Barretto, 2003; Van Houten et al, 2005). In this article, we address the latter issue, by studying the way simulation knowledge and tools may be exploited for training industrial workers in new working procedures.

Many researchers indicate the high potential of simulation games for mastering new business concepts and procedures in operations management (see, eg Angelides and Paul, 1993; Chapman and Martin, 1995; Ruohomaki, 1995, 2003; Haapsalo and Hyvonen, 2001; Chwif and Barretto, 2003; Smeds, 2003; Lainema and Hilmola, 2005). Basic reasons for giving gaming a decisive advantage over conventional lecturing are the active involvement of trainees, the possibility to experience the topic as a whole, and its suitability to convey system characteristics (Greenblat, 1988; Faria and Wellington, 2004). In principle, these reasons are also valid for training-on-the-job. Actually, in 'ideal' circumstances this option may be preferred. When considering factors such as visibility, reproducibility, safety, economy, and system availability, however, preferences may shift to gaming (Raser, 1969; Ruohomaki, 1995).

Clearly, the extensive libraries of generic building blocks offered by discrete event simulation tools present a welcome starting point for building game models—allowing for an efficient representation of manufacturing logistics (buffers, machines, goods, etc), and performance measurements. On the other hand, game models set specific requirements to their construction due to their role as a facilitator of learning built on user interaction. Relevant research issues in this respect are:

• Modelling player decision-making: In what way should the conceptual framework or world view underlying simulation models and tools be adapted?
• Implementation: How to implement player decision-making in a discrete event simulation model?
• Learning contents: How to design and validate game scenarios? How to analyse and present scores?

Although literature reports use of simulation tools for gaming purposes (see above) we found few answers to the above questions. Typically, they receive little attention, which is understandable from the research objectives set by the respective authors, as their primary focus is not on game development, but on their meaning for learning. To exploit the potential of discrete event simulation as a tool for gaming, however, the research questions are highly relevant. In this article, we will address them using a case study concerned with the development and use of a tailored simulation game for training assembly line workers in a new concept for line control (Kalk, 2005; Van der Zee and Slomp, 2005; Wind, 2006).

To answer the research questions some specific requirements for game modelling are identified. We study the way they are met in the case context—assuming the use of discrete event simulation modelling and tools. Remark how the purpose of the game (training), the educational background of players (industrial workers), and a pre-determined system (the assembly line) set the starting point for game set up. Clearly, this distinguishes the game from games supporting, for example, engineering education, where students are encouraged to develop critical thinking, the ability to formulate good questions, and creativity in solution finding, building on the notion of a tailored 'imaginary' system (see Jones, 1995; Corsun, 2000; Klabbers, 2003).

Insights obtained from our study may also be helpful in supporting visual interactive simulation for analysis purposes—as they emphasize a full-grown role for decision makers. Also, we stress the meaning of the case study as an example of the way discrete event simulation games may be used for supporting new business concepts in terms of acceptance, understanding, training, and operational support. As such our study presents an illustrative answer to a widely felt need in industry and engineering education for new methods to accelerate the dialogue between knowledge and practice for efficient learning (Smeds, 2003; Kumar and Labib, 2004).

The remainder of the paper is organized as follows: in the next section we will introduce the methodology adopted for game design and operation. Following this, the background to the case study is outlined before describing the application of the methodology. Next, we address our research questions by evaluating the use of simulation modelling for game set up and play. The paper concludes with a summary of the major findings.

2. Game design

In this section we address the methodology adopted for game design. We chose to start from the framework introduced by Greenblat and Duke (1981). Their framework has been enriched by Riis et al (1995) building on their extensive experience in game design and use. The framework covers the overall design process in four phases: Initialization, Design, Construction, and Operation of the game (see Table 1). We only summarize essential elements for each phase here. For a more detailed discussion, we refer to Greenblat and Duke (1981), and Riis et al (1995). Note how Greenblat also introduces a five-phase approach (Greenblat, 1988). However, the contents of these phases largely cover the four-phase approach of their earlier work. The framework by Greenblat and Duke was chosen because of the structured way it describes the design process, and its role as a reference model for simulation and gaming—being much referenced by simulation game designers (cf. Angelides and Paul, 1993; Crookall, 1995; Smeds, 2003).

Table 1 - Framework for game design (cf. Greenblat and Duke, 1981; Riis et al, 1995).

Full table

The outcome of the initial phase in game design should be a clear focus on objectives and scope of the game. Important issues are concerned with the appropriateness of the game relative to alternative approaches, purpose of the game, and its constituting elements. The actual game design covers two phases: the development of a game concept, and the implementation of this game concept in terms of construction. The final phase is concerned with the actual operation of the game.

3. Case study—a manual assembly line

The starting point for game development is a new design for manual assembly lines for mail inserting systems (Kalk, 2005; Van der Zee and Slomp, 2005; Wind, 2006). Basically, mail-inserting systems fold, fill and close envelopes in an automated way (Figure 1). The company operates several assembly lines, each associated with its own product family, that is, type of mail-inserting systems. Here we focus on one assembly line that served as a pilot for redesign based on lean manufacturing principles. It is expected that other assembly lines will be redesigned according to the findings of the pilot.

Figure 1.

Mail-inserting system.

Full figure and legend (96K)

The basis for the design of the new assembly line is the lean manufacturing concept. Essentially, the concept aims at flexible and efficient work cells by reducing waste in all forms, such as production of defective parts, excess inventory, unnecessary processing steps, and unnecessary movements of people or materials (Womack et al, 1990; Goldman et al, 1995). As for flexibility and efficiency the new line should meet a number of goals:

• Volume and mix flexibility: the line should be able to cope with fluctuations in demand volume (range: 20–40 products). Also the product mix may change.
• Product flexibility: the line should be able to cope with new variants of products.
• High worker efficiency.

A novel design for the assembly line was defined by the firm in close cooperation with a consultancy firm. Essential characteristics of the new system are (see Figure 2):

• The line consists of 17 stations in line. Each station can be manned by one worker at the most. Average process times are about 10 min.
• The number of workers in the system depends on the required output.
• Each station is decoupled from its predecessor and successor by a buffer of size 1.

Figure 2.

The new assembly line.

Full figure and legend (168K)

Volume flexibility for the line is realized by varying the number of workers. Hence, workers are responsible for operations at more than one station. In periods of moderate and high demand, temporary workers are hired.

In order to gain a controllable assembly line, the firm was advised to split the line in three segments (Kalk, 2005). A team of workers was assigned to each segment. Team activities are coordinated using the takt-time principle (Baudin, 2002). Essentially, the takt-time principle realizes line pacing by considering a fixed time frame—the takt-time—in which each team should complete assembly operations for one product. The respective product is to be placed in a buffer and serves as an input for a successor segment of the line. The setting of the takt-time is based on demand figures. Basically, it relates planning period and its associated demand. Ideally, implementation of the takt-time principle should contribute to a regular product flow for the line, and reduce the effects of blocking and starvation.

The new line concept, however, does not present a solution to the deployment problem of selecting a next station for a worker who completed a job at an assembly station. At first sight it may seem that this problem can be easily solved by adopting rules like 'down-up', or 'up-down', which send idle workers downstream or upstream to search for a new job. Such rules, however, do not consider job variety and worker differences.

Another issue that has to be solved is concerned with the acceptance of the proposed control solution by the workers and the management. Management was only willing to implement the solution and to make the required investments (eg software, screens for monitoring line operation, training time for workers) after they had been convinced of the major advantages of the new takt-time control concept. At the same time, the takt-time control concept evokes a natural resistance among workers, as working according to takt-time can easily be associated with working-like-robots. Workers need to understand that the takt-time concept does not necessarily degrade their working conditions. In this particular case, the control concept may even upgrade the working conditions, since it puts the teams in control of their subsystem.

In order to cope with the assignment problem and the implementation issues, it was suggested to develop a simulation game. The new game should facilitate (1) the demonstration of the new control concept for management and workers, and (2) worker training on selecting appropriate deployment rules.

4. A new simulation game

In this section we describe game development and use, being guided by the aforementioned framework for game design (see Table 1). Our prime focus is on those elements of the framework that are directly relevant for the set up of the game model and its operation. This is motivated by our research questions, which stress the role and meaning of discrete event simulation for game model development and use. Hence we refrain from issues concerning game management, both with respect to the organization of the game itself and its embedding in the firm. We start from the assumption that gaming is a suitable and viable vehicle for training workers at the assembly line.

4.1. Initialization

Purpose of the game

Before implementing the new assembly line concept (teams of workers, takt-time-controlled), it was necessary to familiarize both management and workers with the new control concept. Both parties set different demands. As for convincing the management, a clear demonstration of the concept would be sufficient, whereas for the workers the incorporation of some practicing would be required. Workers need to learn the meaning of the takt-time-control concept and the importance of making team-based 'where-to-go' control decisions.

Appropriateness of the game: alternative approaches for learning

Comparison with alternative approaches for learning is useful, as it may shed light on relevant demands for game modelling and use. We found that training on the job faces major disadvantages. Firstly, it is difficult to give appropriate and timely feedback to workers on the impact of their control decisions. Typically, the impact can only be seen after some time—after completing several (subsequent) operations. This time span is problematic for a clear understanding of 'cause and effect'. Furthermore, workers have to cope with a complex assembly task, which requires cognitive efforts. Secondly, in case of training on the job, the learning intensity is low. An assembly operation takes about 10 min, whereas control decisions are made in a matter of seconds. Hence, an adequate game model should allow for the isolation of control activities from operations with respect to learning.

To realize the learning effect among the firm's employees, researchers working on the development of the assembly line initially proposed the use of lecturing. This was facilitated by a series of animated sheets using MS Powerpoint™ (Kalk, 2005). It demonstrated the principles underlying takt-time control. Although the lecture contributed to the understanding among workers and management, we found that experience of the attendees is limited—they may watch model dynamics, but are not able to influence model behaviour. Consequently, only a limited contribution is made to their confidence and insights in solution quality. More specific, there is little room for workers to bring in, test and analyse their own views on line control, as well as the effects these may have on team performance. In turn, this does not contribute to team building in terms of a joint acceptance of solutions. In conclusion, a game model should facilitate players in having the freedom to make deployment decisions on their own. This refers both to the contents and the timing of their decisions.

Limitations on resources and equipment

The most important restriction on resources follows from the availability of workers. The duration of the game session should be at most 3 h based on production losses during workers' absence.

General demands on game development

Relevant demands as for game model development are:

• A clear game focus. See the game purpose.
• Adjustment to the learning levels of the participants, following from the diversity of workers' educational backgrounds and their work experience at the assembly line.
• Valid and 'realistic' reflection of real-life settings. The model should enable workers to assign the experiences to their daily life.
• Game results should be measurable. Quantities are preferred.

Constituent elements

Starting from the game purpose, Table 2 shortly characterizes each of the game elements.

Table 2 - Game elements.

Full table

The model refers to just one segment of the line. This is a natural choice as worker teams are related to line segments. It is also a logical choice, as line segments may be considered rather independent of each other, being decoupled by the takt-time control concept and/or a small buffer between the segments. Relevant line elements (symbols) are workers, buffers, stations and products. Scenarios may be defined by setting worker characteristics (working speeds), operation characteristics (norm times) as well as line pacing (yes/no). The latter factor allows for a comparison of a setting assuming takt-time control with a setting where there is no line pacing.

Essentially, decision-making is triggered either by a job completion (a worker becomes idle), or by a product arrival (a new opportunity to start a job). At such events the player having the role of line-manager may decide to start a new job by assigning a worker to a station. Obviously, within the rules of the game, it is only possible to assign (1) idle workers to (2) stations that are not blocked and (3) for which the input buffer is filled.

Indicators of system status are the distribution of products over the line, and the time left to stay in pace (in case of takt-time control). Results of a game run are logistic performance measures: throughput and average flow time of products. Also, the ability to produce according to takt-times is reported, that is, the number of times takt-time is not met, and the cumulative excess time.

The game process starts by introducing the players to the subject of assembly lines. Trade-offs in control decisions are discussed. Also the working of the model is demonstrated. Gaming is concerned with a series of game runs. Each game run corresponds to a scenario. These scenarios reflect real-life situations at the work floor. During a game run the players have to make control decisions. After each run learning experiences and initiatives towards rule construction are recorded by means of open interviews. In turn, these recordings facilitate game evaluation.

An important choice for design is concerned with the roles to be played. In the game a worker has to play the role of a line manager. This 'shift of roles' is meant to prevent a myopic focus on optimizing the worker's own tasks. Instead, it stresses the importance of team coordination through a dedicated allocation of a heterogeneous work force.

4.2. Game design and construction

For the construction of the model the simulation tool EM-Plant™ has been used. The model is concerned with a segment of the assembly line—five stations. It allows a single player to make decisions on worker deployment by dragging worker icons to respective stations (Figure 3). As an input for decision-making a player may consider shop status, that is, the distribution of jobs and workers over the line, and several indicators (see the upper right corner of Figure 3). Alternative scenarios may be chosen by changing the experiment number in the display—top of the model, in the middle. Worker and station settings are reflected by setting their labels accordingly. For example, an employee working at 75% efficiency is labelled as such (see Figure 3). Game runs take about one half hour per scenario. During a run, line performance and decisions are recorded for further analysis.

Figure 3.

Game model.

Full figure and legend (218K)

4.3. Game operation

In this paragraph we reflect on the use of the game model. More specifically, we address the way we tested it for (1) correct working, (2) player appreciation in realizing learning effects, and (3) learning effects aimed at.

So far the game has been used for training four workers of the firm. Also the game has been embedded in two undergraduate courses. Group sizes were 20 and 40 university students, respectively. Within the context of this article educational use of the game will only be considered as far as it is supportive for testing proper working of the training game, player appreciation, and basic learning effects (as implied for training). The initial group of 20 students was asked to review and test the game for proper working. Main elements in the review were:

• Correct working and user friendliness of the model interface.
• Understanding of symbols and indicators.
• Correct and meaningful recording of player decisions, and their effects on line performance (see also next section).
• Time required to complete a game run.

All 20 students were interviewed for their observations on these issues. Based on their observations the game design was improved. Improvements mainly concerned the way the interface deals with player errors in decision-making, such as, for example, the dragging of a worker to a station for which there is no product in buffer. Solutions in cases like these may be found in a combination of error messages, suggesting correct player behaviour and the possibility to restore the initial model settings.

Further, a questionnaire was used for all 60 students to test their appreciation of the game in realizing learning objectives. Their learning was considered during game evaluation, and found satisfactory. Note how regular game use includes testing and fostering learning effects for players. Workers were questioned in open interviews on game set up, game appreciation and their learning. The enthusiasm of both workers and students may indicate game appreciation. Fascinated by the game, they both skipped coffee breaks. Some students even tried to improve their scores by repeating game runs. While students' fascination seems to be motivated by competition, workers seem to be driven by the possibility to reflect upon their daily practice. This includes insights (learning) on the way their own control rules influence team performance.

5. Simulation as a tool for gaming

In the previous section we discussed the development and operation of the game model. Here we evaluate the use of simulation modelling for game set up and play in more detail, starting from our research questions. Essentially, the research questions link the set up and use of the game model to key phases in a classic simulation study: conceptual modelling, model coding, and experimenting (see, eg Law and Kelton, 2000). In this section basic game characteristics are related to specific demands on modelling for each of these phases. Alternative ways to meet these demands are discussed, supported by illustrations from the case study.

5.1. Conceptual modelling—model player decision-making

Intrinsic to gaming are players making decisions, being a part of the model themselves (Jones, 1998). This is clearly reflected in the definition of game elements, like role, and decision (see Table 1). An explicit notion of decision makers (players, roles), their activities, and their interaction in defining game plots seems no more than natural. This contradicts simulation models and tools for system analysis, in which decision-making often receives less attention. Despite their relevance for logistic performance, decision makers, control rules and their interactions are often 'hidden'. Mostly they are hard coded and dispersed throughout the model, following tool facilities and analyst's insights (Pratt et al, 1994; Bodner and McGinnis, 2002; Galland et al, 2003). Particularly so, as far as human decision-making is concerned, its effects are often overlooked or captured in strong simplifications (Baines et al, 2004; Robinson et al, 2005).

Essentially, the modelling of decision-making in simulation analysis can be traced back to the (implicit) reference models underlying simulation tool libraries and the analyst's activities in model building. This practice may be partly caused by a lack of supportive means for conceptual modelling. Existing approaches and methods for conceptual modelling tend to focus largely on model pruning by suggesting guidelines for model simplification (see Robinson, 2007). However, they do not address model creation, in terms of what to model. Attempts to address this gap are scarce. One way to fill the gap is the development of modelling frameworks. These present a procedural approach in detailing a model in terms of its elements, their attributes, and their relationships (Robinson, 2007). The relevance of such structured approaches for project success is confirmed in a recent survey of McHaney et al (2002).

Starting from the above observations, Van der Zee (1997), Van der Zee and Van der Vorst (2005), and Van der Zee (2006a, 2006b, 2007) defined a new modelling framework for manufacturing and supply chain simulation. It was implemented and tested for simulation modelling and analysis of systems for manufacturing planning and control (Van der Zee, 2006a, 2007; Van der Zee et al, 2008), supply chains (Van der Zee, 1997; Van der Vorst et al, 2005; Van der Zee and Van der Vorst, 2005), and transportation networks (Ebben et al, 2004). An intrinsic element of the modelling framework is a well-defined conceptual view on modelling decision logistics for simulation. Here decision logistics refers to the activities associated with decision makers being responsible for planning and control (eg schedulers, workers). Here we apply the modelling framework as a guide in specifying and coding game elements, including player interaction. Essential elements of the modelling framework are:

• Decision makers are modelled as intelligent agents who carry out decision jobs, just like machines carry out physical jobs.
• The definition of a common internal structure for agents, which specifies basic elements and operations. Elements concern:
  • Buffers for storing data, goods, and resources.
  • A job queue containing job definitions issued by a higher-level agent.
  • A transformer reflecting jobs being processed.
  • Local intelligence, specifying the control logic applied by the agent in job execution.
    Handling of incoming and outgoing flows of goods, resources, data, and job definitions are realized by input and output operations.
• Model dynamics is linked to job dynamics as activities only start if both a job definition, and its required input, that is, goods, data and/or resources, are present.

For more details, see Van der Zee and Van der Vorst (2005), and Van der Zee (2006a, 2006b, 2007). Let us illustrate the use of the modelling framework by specifying game elements for the assembly line case (Figure 4):

• A player is related to the role of a line manager, that is, an intelligent agent.
• Decision jobs are triggered by events, that is, product arrivals and the completions of operations.
• Decision jobs call on the players' intelligence (local intelligence) for assigning workers to stations. This may involve the consultation of shop status, being reported by indicators, representing, for example, the time remaining until the start of a new takt-time period, and visual observations of the line.
• Decision-making is realized in real-time. Note that as dynamics is governed by both players and model there is a decoupling of simulation time and real time.
• Decision output is concerned with job definitions specifying assignments of workers to stations.

Figure 4.

Conceptual game model—line manager and assembly line.

Full figure and legend (47K)

In a similar way the assembly line has been defined as an entity executing assembly jobs, involving both products and workers, and providing feedback on shop status to the line manager. Remark how the above specification sets a blueprint for model coding.

5.2. Implementation—player decision-making

Above we considered model contents. When considering the way the model is implemented—building on EM-Plant™ facilities—we will distinguish between the coded model, player interaction, and game dynamics. The model of the assembly line could be built rather fast, relying on the basic library of EM-Plant™. This is also true for the modelling of player interaction in terms of a realization of decisions and obtaining feedback on these decisions. Messages to the player may, for example, be displayed by pop-up boxes. For choosing among options several types of buttons exist. We chose to model worker assignments by the 'dragging' of workers to the respective stations. In doing so the decision (assignment) and its effectuation (worker moves to the station) are not separated.

Game dynamics is governed by both player and model (see above). Real-time player decision-making is decoupled from the physical transformations being represented by the model. Decoupling is important in order to give the player the peace of mind required for making decisions—and learning from them. It also provides opportunities for the game leader to intervene. In addition, the setting of simulation speed should be tuned in such a way that the player is able to observe and interpret the consequences of his decisions, in terms of worker movements, product transformations, and related performance measures, but does not get bored (Slomp et al, 2007).

5.3. Learning contents: choice of game scenarios, analysis of scores

When applying simulation for system analysis, an experimental frame is defined to specify experimental factors, their settings, and resulting model outcomes. For gaming a similar frame is used for specifying scenarios, that is, game plots resulting from alternative system configurations, and results specifying player's performance.

It appeared that the number of scenarios for the assembly line game depended highly on the limitations set on worker availability. The training session was allowed to take 3 h at the most. This puts much pressure on the selection of those few scenarios that are illustrative for system operation, that is, are concerned with a valid representation that appeals to players. After due consultation of the line manager we chose to consider worker efficiency, and the control concept as the experimental factors. The rightness of this choice was confirmed by the workers later on—in terms of their enthusiasm, and learning. Extensions to the model—with real-life relevance—are left to the game evaluation. In this final part of the game we include the possibility of alternative 'what if' scenarios being considered—either being a priori scheduled or brought up by game players. Such scenarios may relate to model extension or a detailing of experimental factors. Examples in the case context are concerned with the consideration of alternative product types, and aspects underlying worker efficiencies like motivation, and learning, respectively. Note that the game model can still play an important role here. However, user involvement tends to shift somewhat from player interaction to (interactive) demonstration by the game leader.

Implementation of scenarios in terms of settings of experimental factors, and model output was straightforward. This appeals to facilities that are common for simulation tools. However, we mention one novel facility we introduced for the game. It is concerned with a 'decision trace'. Such a trace supplies an overview of the sequence and nature of decisions made by players. Where simulation tools generally supply a trace of events using text or a film of model operation, they do not record human decision-making in detail. For example, within the context, it may be interesting to reconsider deployment decisions made by a player, both in individual or plenary evaluations. In that case an overview of decisions made during a game run is required, in terms of a sequential series of assignments of workers to stations. The overview may be used to (partly) replay the game and consider the consequences of alternative decisions. Remark that effects of decisions may only become apparent after some time. The relevance of the decision trace was clearly recognized in the initial game series reported on in the previous section.

6. Summary of main findings

In this article we discussed the use of discrete event simulation as a tool for training industrial workers in new working procedures. To this end we studied the development and use of a new game. The case example used is derived from industry and involves the operation of a manual assembly line. The game was tested and played both at the company and in two engineering courses.

In general, we found that discrete event simulation models and the tool applied—EM-Plant™—are quite suitable for supporting development of game plots and models. Specific requirements that should be met for game modelling are concerned with the representation of decision-making, either by the player or the model. Typically, simulation models stress operations rather than decision-making. Given the tool facilities and analyst's insights, decision-making is typically hard coded and dispersed over the model, or—in case of human decision-making—overlooked or very simplified.

For conceptual modelling of game elements we relied on a novel modelling framework for simulation, which includes an explicit notion of decision makers, their activities (decision jobs) and their interaction (cf. Van der Zee and Van der Vorst, 2005). The framework allowed us to specify a blueprint for the game model, which is both complete and transparent from the perspective of player decision-making.

Model implementation was relatively easy. Relevant facilities for modelling elements of the production system (buffers, machines, etc), its performance (bars, indicators etc) as well as player interactions (dragging, messages etc) are available within EM-Plant™. Yet, we would like to stress that model structure as defined in the conceptual modelling phase is most relevant for both modelling efficiency, and model effectiveness.

Choice of game scenarios had to be made under rather strict limitations on time available for training. As a consequence, four scenarios were chosen concerning two experimental factors: worker efficiency and line control. Game conditions should be recognized by workers in their daily practice. This contributes to player enthusiasm, and learning.

The recording and analysis of player decision-making is of specific concern. We found it worthwhile to include a 'decision trace' in our models giving a detailed account of all player decisions. It allows for a (partial) replay of the game and the consideration of alternative decision sequences.

In sum, we conclude that simulation knowledge and tools present a powerful basis for game design. Basic flaws are, however, in the representation and implementation of player interaction and decision-making. It is suggested to address these flaws, by extending conceptual models using the explicit notion of decision makers and their activities, and the inclusion of a decision trace in the experimental frame.

The case study also raises some interesting issues for further research. Our experiences in game design and use indicate that gaming may support 'classic' use of simulation for systems design. Gaming may be used as a method of inquiry for identifying relevant human factors and their settings. In turn, these data may be used for further modelling and analysis. Also worker's game experiences may contribute to solution finding and acceptance. While the case study assumes a single player game, other games may allow for a multi-player setting. Think of, for example, supply chain or care chain games. This assumes distributed simulation, which receives increasing attention in literature. In this respect efforts by Van Houten and Jacobs (2004), and Verbraeck and Van Houten (2005) should be mentioned. They work on distributed simulation games, starting from web-enabled technologies. Further we point at basic research on standards for model interoperation—such as the High Level Architecture (HLA) (see, eg Taylor et al, 2006). The development, testing, and tailoring of such standards and their implementation for simulation tools are prerequisites for efficient and effective gaming. Finally, we mention researchers trying to link insights from entertainment games with games for training purposes (Zyda, 2004, 2005). They bring in their experience in project setup for game design, and game interfaces—think of immersive qualities. Embedding the various efforts in a softer methodology on simulation game design and use—including conceptual modelling—is among the challenges we face.

References

1. Adelsberger HH, Bick MH, Kraus UF and Pawlowski JM (1999). A simulation game approach for efficient education in enterprise resource planning systems. In: Szczerbicka H (ed). Proceedings of the 13th European Simulation Multiconference, Warsaw. Society for Computer Simulation: Ghent Belgium, pp 454–460.
2. Angelides MC and Paul RJ (1993). Towards a framework for integrating intelligent tutoring systems and gaming simulation. In: Evans GW, Mollaghasemi M, Russell EC and Biles WE (eds). Proceedings of the 1993 Winter Simulation Conference. IEEE: Piscataway, NJ, pp 1281–1289.
3. Au G and Paul RJ (1996). Visual interactive modelling: A pictorial simulation specification system. Eur J Opl Res 91(1): 14–26. | Article |
4. Baines T, Mason S, Siebers P and Ladbrook J (2004). Humans: The missing link in manufacturing simulation? Simul Model Pract Theory 12: 515–526. | Article |
5. Baudin M (2002). Lean Assembly—The Nuts and Bolts of Making Assembly Operations Flow. Productivity Press: New York.
6. Bell PC (1985). Visual interactive modeling in operational research—Successes and opportunities. J Opl Res Soc 36(11): 975–982. | Article |
7. Bell PC (1990). Visual interactive modeling—The past, the present, and the prospects. Eur J Opl Res 54(3): 274–286. | Article |
8. Bodner DA and McGinnis LF (2002). A structured approach to simulation modelling of manufacturing systems. In: Proceedings of the 2002 Industrial Engineering Research Conference. Norcross, GA, IIE: Orlando, Florida.
9. Chapman GM and Martin JF (1995). Computerized business games in engineering education. Comput Educ 25(1/2): 67–73. | Article |
10. Chwif L and Barretto MRP (2003). Simulation models as an aid for the teaching and learning process in operations management. In: Chick S, Sánchez PJ, Ferrin D and Morrice DJ (eds). Proceedings of the 2003 Winter Simulation Conference. IEEE: Piscataway, NJ, pp 1994–2000.
11. Corsun DL (2000). We sail the same ship: Response to "shuffling deck chairs". J Hosp Tour Educ 12(3): 10–11.
12. Crookall D (1995). A guide to literature on simulation/gaming. In: Crookall D and Arai K (eds). Simulation and Gaming across Disciplines and Cultures. Sage Publications: Thousand Oaks, pp 151–171.
13. Ebben MJR, Zee DJ and van der Heijden MC (2004). Dynamic one-way traffic control in automated transportation systems. Transport Res B 38(5): 441–458. | Article |
14. Faria AJ and Wellington WJ (2004). A survey of simulation game users, former users, and never users. Simul Gam 35: 178–207. | Article |
15. Galland S, Grimaud F, Beaune P and Campagne JP (2003). M(A)MA-L: An introduction to a methodological approach for the simulation of distributed industrial systems. Int J Prod Econ 85(1): 11–31. | Article | ISI |
16. Goldman S, Nagel R and Preiss K (1995). Agile Competitors and Virtual Organizations—Strategies for Enriching the Customer. Van Nostrand Reinhold: New York.
17. Greenblat CS (1988). Designing Games and Simulations—An Illustrated Handbook. Newbury Park: Sage Publications.
18. Greenblat CS and Duke RD (1981). Principles and Practices of Gaming Simulation. Sage Publications: London.
19. Haapsalo H and Hyvonen J (2001). Simulating business and operations management—A learning environment for the electronics industry. Int J Prod Econ 73: 261–272. | Article |
20. Hurrion RD (1976). The design, use and required facilities of an interactive computer simulation language to explore production planning problems. PhD thesis, University of London, England.
21. Jones K (1995). Simulations: A Handbook for Teachers and Trainers, 3rd edn. Nichols Publishing Company: East Brunswick, NJ.
22. Jones K (1998). Simulations: Reading for action. Simul Gam 29: 326–327. | Article |
23. Kalk L (2005). Logistieke beheersing van een vraaggestuurde assemblagelijn. Master thesis. University of Groningen, The Netherlands (in Dutch).
24. Klabbers JHG (2003). The gaming landscape: A taxonomy for classifying games and simulations. In: Copier M and Raessens J (eds). Digital Games Research Conference 2003. University of Utrecht: Utrecht, pp 54–67.
25. Kumar A and Labib AW (2004). Applying quality function deployment for the design of a next-generation manufacturing simulation game. Int J Eng Educ 20(5): 787–800.
26. Lainema T and Hilmola OP (2005). Learn more, better and faster: Computer-based simulation gaming of production and operations. Int J Bus Perform Mngt 7(1): 34–59. | Article |
27. Law AM and Kelton WD (2000). Simulation Modeling and Analysis. McGraw-Hill: Singapore.
28. McHaney R, White D and Heilman GE (2002). Simulation project success and failure: Survey findings. Simul Gam 33: 49–66. | Article |
29. Pratt DB, Farrington PA, Basnet CB, Bhuskute HC, Kanath M and Mize JH (1994). The separation of physical, information, and control elements for facilitating reusability in simulation modelling. Int J Comput Simul 4(3): 327–342.
30. Raser JR (1969). Simulation and Society: An Exploration of Scientific Gaming. Allyn and Bacon: Boston.
31. Riis JO, Johansen J and Mikkelsen H (1995). Simulation games in production management—An introduction. In: Riis JO (ed). Simulation Games and Learning in Production Management. Chapman & Hall: London, pp 3–11.
32. Robinson S (2001). Soft with a hard centre: Discrete-event simulation in facilitation. J Opl Res Soc 52(8): 905–915. | Article |
33. Robinson S (2005). Discrete-event simulation: From the pioneers to the present, what next? J Opl Res Soc 56(6): 619–629. | Article |
34. Robinson S (2007). Conceptual modelling for simulation Part I: definition and requirements. J Opl Res Soc 59: 278–290. | Article |
35. Robinson S, Alifantis T, Edwards JS, Ladbrook J and Waller A (2005). Knowledge based improvement: Simulation and artificial intelligence for identifying and improving human decision-making in an operations system. J Opl Res Soc 56(8): 912–921. | Article |
36. Ruohomaki V (1995). Learning and education with simulation games. In: Riis JO (ed). Simulation Games and Learning in Production Management. Chapman & Hall: London, pp 13–25.
37. Ruohomaki V (2003). Simulation gaming for organizational development. Simul Gam 34: 531–549. | Article |
38. Slomp J, van der Zee DJ and Molleman E (2007). Exploring success factors of a computer game—The development and use of a 'Takt Time Control' game. In: Peters V, Van de Westelaken M, Van Gils A and Van Loveren A (eds). Proceedings of the 2007 Conference of The International Simulation and Gaming Association (ISAGA). ISAGA: Nijmegen, The Netherlands.
39. Smeds R (2003). Simulation for accelerated learning and development in industrial management—Guest editorial. Prod Plann Control 14(2): 107–110. | Article |
40. Taylor SJE, Wang X, Turner SJ and Low MYH (2006). Integrating heterogeneous distributed COTS discrete-event simulation packages: An emerging standards-based approach. IEEE Trans Syst Man Cybernet—Part A: Syst Hum 36(1): 109–122.
41. Van der Vorst JGAJ, Tromp S and Van der Zee DJ (2005). A simulation environment for the redesign of food supply chain networks: Integrating quality and logistics modeling. In: Kuhl ME, Steiger NM, Armstrong FB and Joines JA (eds). Proceedings of the 2005 Winter Simulation Conference. IEEE: Piscataway, NJ, pp 1658–1667.
42. Van der Zee DJ (1997). Simulation as a tool for logistics management. Phd thesis. University of Twente, The Netherlands.
43. Van der Zee DJ (2006a). Modeling decision making and control in manufacturing simulation. Int J Prod Econ 100(1): 155–167. | Article | ISI |
44. Van der Zee DJ (2006b). Building communicative models—A job oriented approach to manufacturing simulation. In: J Garnett, Brailsford S, Robinson S and Taylor S (eds). Proceedings of the Operational Research Society Simulation Workshop (SW06). Operational Research Society: Birmingham, pp 185–194.
45. Van der Zee DJ (2007). Developing participative simulation models—Framing decomposition principles for joint understanding. J Simul 1(3): 187–202. | Article |
46. Van der Zee DJ, Pool A and Wijngaard J (2008). Conceptual modelling of planning & control systems for simulation—A case study in industry. In: Kotiadis K, Currie C, Robinson S and Taylor S (eds). Proceedings of the Operational Research Society Simulation Workshop (SW08). Birmingham: Operational Research Society (accepted for publication).
47. Van der Zee DJ and Slomp J (2005). Simulation and gaming as a support tool for lean manufacturing systems—A case example from industry. In: Kuhl ME, Steiger NM, Armstrong FB and Joines JA (eds). Proceedings of the 2005 Winter Simulation Conference. IEEE: Piscataway, NJ, pp 2304–2313.
48. Van der Zee DJ and Van der Vorst JGAJ (2005). A modeling framework for supply chain simulation—Opportunities for improved decision-making. Decis Sci 36(1): 65–95. | Article | ISI |
49. Van Houten SPA and Jacobs PHM (2004). An architecture for distributed simulation games. In: Ingalls RG, Rossetti MD, Smith JS and Peters BA (eds). Proceedings of the 2004 Winter Simulation Conference. IEEE: Piscataway, NJ, pp 2081–2086.
50. Van Houten SPA, Verbraeck A, Boyson S and Corsi T (2005). Training for today's supply chains: An introduction to the distributor game. In: Kuhl ME, Steiger NM, Armstrong FB and Joines JA (eds). Proceedings of the 2005 Winter Simulation Conference. IEEE: Piscataway, NJ, pp 2338–2345.
51. Verbraeck A and Van Houten SPA (2005). From simulation to gaming: An object-oriented supply chain training library. In: Kuhl ME, Steiger NM, Armstrong FB and Joines JA (eds). Proceedings of the 2005 Winter Simulation Conference. IEEE: Piscataway, NJ, pp 2346–2356.
52. Wind H (2006). Ontwerp van een assemblagelijn. Master thesis, University of Groningen, The Netherlands (in Dutch).
53. Womack K, Jones D and Roos D (1990). The Machine that Changed the World. Maxwell Macmillan International: Oxford.
54. Zulch G and Brinkmeier B (1995). Simulation aided planning of work structures. In: Riis JO (ed). Simulation Games and Learning in Production Management. Chapman & Hall: London, pp 91–104.
55. Zyda M (2004). Does the future of modeling and simulation have a game face? In: RG Ingalls, MD Rossetti, JS Smith and BA Peters (eds). Proceedings of the 2004 Winter Simulation Conference. IEEE: Piscataway, NJ, p 3.
56. Zyda M (2005). From visual simulation to virtual reality to games. Computer 38(9): 25–32. | Article |