Dynamic and distributed early planning assessment by a hybrid simulated annealing and greedy algorithm

Abstract

This paper describes a novel methodology for optimizing the wide range of early planning choices available for the manufacture of complex products, across a distributed and dynamically changing enterprise. A hybrid simulated annealing and greedy algorithm has been developed to optimize such planning decisions. The hybrid algorithm and associated methods detailed within the paper allow the utilization of key attributes from the core product, process and resource digital models to create an initial valid plan and a dynamic ‘manufacturing phase space’. Within this phase space, the detailed methods, heuristics and algorithms are deployed in order to optimize the initial plan by evaluating alternative processes and resources. The evaluation is based on an alternative's impact on the overall solution obtained by applying user-defined cardinal weightings on quality, cost, delivery and knowledge (QCD + K) metrics. To enable the application of business objectives on QCD + K, methods were developed for the conversion of quality, delivery and knowledge into a cost equivalent. The paper is concluded with an example based upon the collaborating company's aerospace products.