Abstract
We study the allocation of machines in a job shop environment to groups of jobs so as to obtain a collection of disjoint production lines minimising the overall cycle time. Jobs are often grouped according to similarities based on functionality or production process, but in some settings it makes more sense to group jobs according to batch size to exploit differences in setup and run times among technologies and machine types. We formulate a model to find an optimal group partition of jobs and an optimal assignment of machines to groups. We show that the problem is hard and formulate a heuristic based on genetic algorithms to find approximate solutions.