Multi-product, multi-level product control system analysis
Several techniques are applicable to the modelling of production and inventory control systems. In this thesis discrete linear control theory is examined as a method of modelling multi-product, multi-level systems. These systems are categorised and a general discrete linear control model is used to determine system stability and to predict system responses to specific patterns of input information. The response of the system to random variability in input or other system variable is also shown to be predictable. A library of sub-system models is provided and the method is illustrated by examples and a case study. Alternative modelling techniques rely upon sequential simulation, either directly or in solving equations representing the system. The need to include forecasting, inventory and production decision-making procedures makes such models large and their sequential nature imposes the need for complete remodelling for each system modification and for each input pattern. Where random effects are modelled, protracted runs are necessary to achieve statistically acceptable results. In contrast, discrete linear control theory provides a nonsequential model, thereby alleviating these problems. Thus it is possible both to reduce computing expense and increase the range of systems susceptible to manual analysis. The method is limited by the restriction of linearity, but, in many practical situations this restriction poses no insuperable difficulty in the interpretation of results.