Problem definition 31

DES and SD are two simulation approaches used to model social or managed systems with the view to understanding the system behaviour. As Sweetser (1999) mentions, both simulation approaches can be used to understand the way systems behave over time to compare their performance under different conditions. Despite the overall common objective, SD is inherently involved in studying the effect of policies on system behaviour. SD is viewed as a dynamic feedback system and studies the interaction of control policies or exogenous events and the model’s feedback structure in producing dynamic behaviour (Mak, 1993). It can also be used as a goal seeking tool in making decisions on a particular variable in the model in order to achieve a desired goal. In DES modelling the system under study is assessed with the view to improving system capacity, resource utilisation or

queuing time in the system. A ‘what if’ philosophy is used to answer questions like: “would additional resources in the system reduce the queue size?” (Mak, 1993)

Another facet tackled in the literature is the nature of problems modelled by each simulation technique, ‘strategic’ vs. ‘tactical/operational’. It is believed that SD focuses mainly on strategic policy analysis, while DES is generally used to study problems at an operational or tactical level (Taylor and Lane, 1998; Sweetser, 1999; Lane, 2000). Based on the differences of discrete and continuous systems,

Richardson (1991) maintains that the choice of one or the other approach depends on the conceptual difference from which one views the problem. The SD approach

Chapter 2

where events and decisions are seen in the form of patterns of behaviour and system structures (Richardson, 1991). Rabelo et al. (2005) point out some of the factors that make SD suitable for high level strategic modelling, including a number of

unsupported claims, which are generally accepted in the comparison literature. These consist of the following:

„ Takes a holistic approach of systems, integrating many subsystems „ Focuses on policies and system structure

„ Use of feedback loops to represent the effects of policy decisions

„ Represents a dynamic view of the cause and effect relationships among the system elements

„ SD has minimal data requirements to build a model.

Several authors suggest that DES is not suitable for strategic modelling, as it does not normally represent models at aggregate levels (Baines and Harrison, 1999; Lee et al., 2002a; Oyarbide, et al., 2003). To cater for this disadvantage, a number of studies (Lee et al., 2002a; Rabelo et al., 2005; Helal et al., 2007) have suggested the use of hybrid simulation approaches combining DES and SD. Rabelo et al. (2005) in their study of an integrated manufacturing enterprise system, used DES to model local production decisions for selected parts of the enterprise, while the SD model captured the long term effects of these decisions on the entire enterprise and the interactions between decisions made at different levels of management. In another study, Lee et al. (2002a) recommended the use of analytical models for modelling at

Chapter 2

operational levels, DES for modelling at tactical level, while for modelling at strategic levels recommended the use of hybrid discrete and continuous simulation models. The same authors created a discrete event and a hybrid discrete-continuous simulation model of a supply chain and concluded that the discrete event model overestimated the outputs of inventory levels. The authors recommended the use of hybrid simulation models to model supply chain simulation models, which were shown to be neither completely discrete nor continuous systems.

On the other hand, various authors have expressed the view that, even though it has not yet been adequately exploited, SD can be successfully used in modelling operational systems. For example, Han et al. (2005) represented an operational SD model of an earth-moving system in a construction management study and

compared it with an equivalent (already existing) DES model. Their study suggests that an SD-based operational model can address the operational aspects of the model as accurately and reliably as a DES-based model. The advantages of using SD at an operational level are discussed. These include modelling of feedback effects, managerial actions and soft variables. Furthermore, while the use of SD has been rarely considered in manufacturing systems modelling, Oyarbide et al. (2003) comment on the potential of using SD modelling in this context. Taking into consideration the inherent characteristics of the two modelling techniques, the authors suggest that SD would be a better choice in the intermediate stages of decision making when less detailed models or results are required. Some of the

Chapter 2

intermediate stages of evaluation are: the simplicity of the data required, ease of building a simulation model and reduced execution time. Obviously, these are statements which represent authors’ opinions and have not been empirically verified for their accuracy.

In a study of a manufacturing plant, Greasley (2005) reports on the successful use of the DES approach to investigate the operational aspects of a production-planning facility. The outcome of the DES study was the recommendation of new production sequencing activities. In addition, as a result of the study, it emerged that the

disruptions in production planning in the manufacturing plant needed to be further considered. In this case, the SD approach was preferred in order to model the softer aspects related to the problem of disruptions. Greasley considered the SD approach useful in modelling the organisational context of the problem and so extended the already created DES model, using SD.

It is clear that the opinions expressed by the authors referred to in this section tend to suggest that SD modelling is more appropriate in modelling at a strategic level, while DES at a tactical/operational level, however, no empirical evidence has been found to verify these opinions.

Chapter 2