DEAL: A Heuristic Approach for Collaborative Planning in Detailed Scheduling

Aroused by ongoing globalization, the division of labor steadily increases and future competition will likely take place between whole supply chains and not only between single companies. This compels suppliers and manufacturers to better coordinate their production plans in order to save production and holding costs. Despite this necessity, the willingness to share sensitive production data such as cost factors or resource availability has remained very limited. Usually the companies consider these data as vital to their business. However, today’s production planning models and solvers used in industry are of monolithic type and require full visibility of data in order to compute a solution. Hence, they cannot be used to tackle supply-chain-wide planning problems involving different companies. In recent years, “Collaborative Planning” as a joint decision making process under information asymmetry has received increased attention. The majority of present research in this field breaks with the monolithic approach but assumes that planning problems can be solved to optimality. On the contrary, industry struggles in operational business with large detailed scheduling problems that are only solvable by heuristics – violating this fundamental assumption. However, being computed only shortly before execution, badly aligned detailed schedules most obviously demand a coordinated solution. We propose a decentralized evolutionary algorithm (DEAL) for coordinating large-sized detailed schedules that does not demand the exchange of sensitive data but only transmits delivery dates and ordinal rankings. Experimental results prove that DEAL computes in the same time solutions of similar quality as monolithic heuristics are able to.