Efficient Rollout of a Dynamic Optimization Algorithm

Abstract

A successful deployment process of a complex software solution, such as optimization algorithms in a dynamic production environment is an extremely complicated and time-consuming task. Over this, the acceptance by those employees where the software solution is to be used later should not be underestimated; it is essential to involve them in the entire development process, as they can be game changers. This paper summarizes some best practices and lessons learned about the full integration of an optimization algorithm in the dynamic steelmaking environment of voestalpine Stahl GmbH. The focus is on the development of a practical test strategy and ultimately a generally valid rollout concept for an arbitrary complex software system for every conceivable industry. Over that, emerging gaps between theory and practice are highlighted, an attempt is made to justify their cause, and, as a result, possible solutions to tackle those challenges are presented.