Guide to Advanced Empirical

5. Simulation Methods What distinguishes DE simulation from SD simulation is the degree of model detail, the model representation, and the logic underlying the computation of model states. DE simulation modelling is very flexible and easily adaptable when it becomes necessary to add or change attributes of entities. Moreover, in DE simulation it is possible to model the behaviour of distinct real-world entities (e.g., artefacts, resources) of the same type individually, while SD typically models the average behaviour of a large number of entities of the same type. The possibility of subscripting mitigates this limitation of SD only to some extent.
One disadvantage of DE simulation comes as a downside of its ability to capture many details. DE simulation tools like, for example EXTEND, offer a large number of different modelling constructs, often specifically tailored to manufacturing processes. Although these blocks are reusable in several contexts, more training is needed for the modeller to become familiar with the variety of options and they have to be adapted to capture software development processes. While DE simulation is capable to model production processes in greater detail, SD simulation models can capture not only the mechanical aspects of software development processes (which mainly consist of writing and checking different types of documents, but also the cause-effect mechanisms underlying the process behaviour. This includes the flow of information, which is important in software engineering, in contrast to material flows. Typically, information about these cause-effect relationships are part of the (mostly implicit) mental models of managers or decision makers, and contain intangible concepts like learning cf. variable code learning state in the example above, motivation, stress, communication, decision policies, etc.

Download 1.5 Mb.

Share with your friends: