Trade-off balancing in scheduling for flow shop production and perioperative processes

Abstract

We balance trade-offs between two fundamental and possibly inconsistent objectives, minimization of maximum completion time and minimization of total completion time, in scheduling serial processes. We use a novel approach of current and future deviations (CFD) to model the trade-offs between the two completion times. We also use weights α and β to balance trade-offs at the operation level and at the process level, respectively. Accordingly, we develop a constructive CFD heuristic, and compare its performance with three leading constructive heuristics in scheduling based on three separate datasets: 5400 small-scale instances, 120 Taillard’s benchmark instances, and one-year historical records of operating room scheduling in a university hospital system. We show that minimization of maximum completion time and minimization of total completion time yield inconsistent scheduling sequences, and the two sequences are relatively uncorrelated. We also show that our CFD heuristic can balance trade-offs between these two objectives, outperform the three leading heuristics across different performance measures, and allow larger variations on two fundamental completion times. This means that trade-off balancing by our CFD heuristic enables a serial process to have a larger tolerance on variations of process performance, and keeps the process better under control. This performance improvement can be significant for flow shop scheduling in manufacturing in trading-off production and holding costs, and for operating room scheduling across the perioperative process in healthcare in trading-off hospital cost and patient waiting time.