Ranking dispatching rules in multi-objective dynamic flow shop scheduling: a multi-faceted perspective

In this paper, we present a multi-faceted approach for ranking dispatching rules (DRs) in multi-objective dynamic flow shop scheduling systems using data envelopment analysis (DEA). The merits of the proposed DEA-based approach stem in its ability to (1) integrate explicitly, under the same DEA framework, desirable and undesirable performance criteria of DRs without a priori normalisation or aggregation; (2) guarantee that the best DR preserves its benchmarking status regardless of the production scenario; (3) circumvent potential occurrence of multiple efficient DRs through embedding ordered weighted averaging (OWA) under DEA cross evaluation to produce aggregate ranking scores for the DRs. The evaluation of the new ranking approach is conducted using 18 data instances of 20 DRs each. The results reveal that, whatever the OWA optimism level, the preferred DR shifts away from the Shortest Processing Time (SPT) rule to the Cost Over Time (COVERT) rule as due-date tightness becomes relaxed, which appears consistent with known performance expectations of these DRs under such settings. To demonstrate a possible implementation of these results to support decision making in operations scheduling, we present a basic adaptive rule that switches automatically between the preferred rules based on real-time due-date tightness and machine utilisation levels.

Automated summary

This paper presents a multi-faceted approach for ranking dispatching rules (DRs) in multi-objective dynamic flow shop scheduling systems using data envelopment analysis (DEA). The proposed DEA-based approach integrates desirable and undesirable performance criteria of DRs without normalization or aggregation, ensures the best DR maintains its benchmarking status across production scenarios, and avoids potential occurrence of multiple efficient DRs.