Efficient scheduling of a stochastic no-wait job shop with controllable processing times

Abstract

This work derives a novel effective and efficient algorithm for a stochastic no-wait job-shop scheduling problem with controllable processing times. Some of the processing times are stochastic and the proposed solution effectively minimizes the makespan and increases the robustness of the makespan against deviating processing times. Therefore, a no-wait job shop with controllable deterministic processing times is solved by a decomposition into timetabling and sequencing. During timetabling, extra safety margins are added to the scheduled processing times without delaying jobs. In the sequence optimization subproblem, an extra penalty term is added to the cost function which punishes uncertain tasks at positions that have an impact on the makespan. Simulation results based on real plant data and tailor-made benchmark problems show that these measures can reduce the standard deviation of the makespan dramatically. This significantly improves the prediction accuracy of the scheduling method.