Solving Manufacturing Problems Using Software for Large-Scale Nonlinear Programming.

Abstract : Our general objectives in this research have been twofold. Our primary objective has always been to apply optimization to real problems in manufacturing. Our secondary objective is to thereby advance research into nonlinear programming algorithms. In order to achieve the first objective we consider it essential to work directly with the personnel in the manufacturing environment. We actively seek out manufacturing groups and we request from them the problems which they consider important. Thus the framework in which we work consists of initially famailiarizing ourselves with a particular production team, followed by identifying suitable problems, in consideration of both the applicability of our expertise and the significance of the problem in the manufacturing environment. The next stage consists of producing an initial model that is simple enough to give rapid feedback (in order to maintain and stimulate cooperation) and realistic enough to indicate useful results (in order to encourage confidence in the approach). This stage also includes the collection of data from the practitioners. Once the initial model has been solved, one goes through a cyde of analyzing results, expanding and modifying the model, resolving, collecting new data, etc. The ultimate goal is to see the solution implemented in the operation of the manufacturing site. During the period of this award we have been able to initiate a considerable number of projects exactly within the model outlined above. Given the current progress we have no doubt that our work in nonlinear programming can make useful contributions to manufacturing and we intend to do our utmost to pursue these approaches. In addition, we have achieved extremely useful input and support for our more theoretical endeavors.