Workload control release mechanisms: from practice back to theory building

Much Workload Control research has focussed on the order release stage but failed to address practical considerations that impact practical application. Order release mechanisms have been developed through simulations that neglect job size variation effects while empirical evidence suggests groups of small/large jobs are often found in practice. When job sizes vary, it is difficult to release all jobs effectively—small jobs favour a short period between releases and a tight workload bounding while large jobs require a longer period between releases and a slacker workload bounding. This paper represents a return from a case study setting to theory building. Through simulation, the impact of job sizes on overall performance is explored using all three aggregate load approaches. Options tested include: using distinct load capacities for small/large jobs and prioritising based on job size or routing length. Results suggest the best solution is assigning priority based on routing length; this improved performance, especially for large jobs, and allowed a short release period to be applied, as favoured by small jobs. These ideas have also been applied to a second practical problem: how to handle rush orders. Again, prioritisation, given to rush orders, leads to the best overall shop performance.