A heuristic for a batch processing machine scheduled to minimise total completion time with non-identical job sizes

This paper studies the problem of scheduling semiconductor burn-in operations, where each job has non-identical lot sizes and an oven (a batch processing machine) that processes several jobs within its capacity limit simultaneously. We present some properties of the problem and an efficient heuristic algorithm. In a computational experiment, pairs of burning operations (alpha,beta) were presented with the arrival time r(i) and processing time p(i) for different (alpha,beta) pairs to examine the effect of arrival time and the processing time on minimising the total completion time. The result shows that a ratio of alpha to beta greater than 1 was superior to other ratios, which can be a guide for schedule planners of burn-in operations. The heuristic obtains a satisfactory average performance rapidly.