A parameter variation modeling approach for enterprise optimization

The past two decades have seen significant improvements in optimization modeling and software solvers for large-scale optimization problems, especially discrete problems. We feel that a critical feature of many of these systems is being overlooked. That is, the process control engineer adjusts process parameters while only considering the local efficiency or not considering process efficiency at all. Production control engineers, while optimizing the global system performance, consider process parameters as given and fixed, i.e., unchangeable. Combining the optimization of the process parameters with a global system view can significantly improve the overall system performance. In practice, "hot jobs" are treated in this ad hoc manner, making sure that all resources are available and operate at peak efficiency (minimum production time) for these critical products. This phenomenon occurs not only in manufacturing but also in many other industries. This modeling part of the optimization problem can be even more important than "optimal versus heuristic"-based solution decisions made. In this paper, we present an aggregative high-fidelity modeling approach and illustrate the formulation of parameter variability in three different domains: manufacturing, air travel, and food processing.