A multi-objective comparison of dispatching rules in a drum-buffer-rope production control system

The advantages of the TOC (Theory of Constraints) philosophy have been extensively documented in the literature since its introduction during the 80s. At the operational level, TOC is implemented by means of the well-known DBR (Drum-Buffer-Rope) production control system. In a multiproduct manufacturing environment, the performance of DBR is greatly affected by the dispatching rules employed in front of the bottleneck station. Furthermore, it has been proved that no single Dispatching Rule (DR) performs globally better than any others. Therefore, for systems usually influenced by variability conditions, the selection of a robust DR could help practitioners to reach a good system performance. In this paper we propose a methodology to obtain a robust DR (by means of Taguchi signal-to-noise ratio) from a set of previously selected rules according to the performance measures of the system pursued by the practicing managers. We study the performance of different dispatching rules for several conflicting objectives (namely average tardiness, maximum tardiness, and WIP) from a robustness viewpoint and for a range of manufacturing scenarios in a shop floor formed by five stations in line and three different products. Different variability sources, such as processing times, breakdowns and set-ups, are discussed. The results obtained are of special interest for practitioners.