Abstract
New software development methodologies such as top-down development, structured programming, and modular design have obvious and reported benefits in improving software quality and reducing software development costs. Since computer simulation languages are themselves software languages, and since credibility and modifiability are never more stressed than in models, the application of these methodologies to simulation languages seems appropriate.
The three development methods, top-down development, structured programming, and process construction are considered in their application to computer simulation for program design analysis. While, potentially, each is shown to have significant positive effect in the quality of model product, some also lead to negative effects when used improperly; to implement a new development method which enhances model quality at the cost of model representation can severely degrade the credibility of results. The positive benefits of product quality must be weighed against the negative effects of non-credible results when these techniques are considered for computer system simulation.
By accurately representing the system to be investigated in simulation language terms, assurance of the accuracy of results is maintained. If the process of accurate representation were considered as a development methodology itself, then model quality both as a software product and as an analysis tool is possible by supplementing new development techniques with representative modelling.
The combination of these two techniques takes us one step further “towards a more mature discipline” as discussed by D. J. Lasser in (15) in that we are using structured translation techniques from real-system to model, validating models directly against the real-system, and increasing mutual understanding between the modeller and model customer via are representative model.
Therefore, each of top-down development, structured programming, and process construction are first discussed as they are applicable to general software systems; second, as they are applicable to computer simulation models; and, lastly, how they can be used in conjunction with a representative modelling methodology to enhance both model quality and modelling results.
