| This dissertation establishes a practical method of implementing tactical production planning (TPP) in industry. TPP is the mid-term planning (typically one year) of production, inventory, and manpower levels. Numerous TPP models have been developed over the past few decades, but few cases of on-going industry use have been reported in the literature. An interactive decision support system is developed that combines the power of a mathematical model and computer capabilities to provide flexibility for managers to make ad-hoc changes. The model handles seasonal or fluctuating productivity changes and is field tested using a large, multi-plant North American foundry operation.; Existing TPP models, implementation issues, aggregation, and validation are reviewed. General analytical models are examined to understand why managers do not use them. An aggregation procedure is developed to select the best level of consolidating data: by plant, production line, or part number. A method is shown to validate the model using linear programming, through the use of temporary constraints.; The model is developed in two phases: linear and non-linear. The non-linear component is further composed of variables and constraints which can be reasonably modeled, and those which are impractical to model. Some non-linear components of the problem are modeled and tested, but take too long to run on the computer. These factors, as well as those which are not practical to model, are handled interactively by the planning manager.; Five versions of HICAP (Heuristic Interactive Computer Assisted Planning) are developed, which help the manager move the solution towards the optimal and achieve certain targets. The first three versions of HICAP are a combination electronic spreadsheet and linear program which provide the same optimal as the full LP model. HICAP IV and V use a simplified tradeoff of inventory and productivity to assign production on a priority basis. These two models are tested and achieve an average savings of 85% compared to the optimal linear solution.; Although many TPP models exist, this dissertation demonstrates new practical techniques for aggregation and validation, and a decision support system which incorporates the planning manager's experience. |
