Work allocation to stations with various learning slopes in assembly lines for lots

This paper addresses allocating work elements with various learning slopes to stations in an assembly line for lots, to minimize the makespan of a lot of products. The line operates under learning, and no buffers are permitted in between the stations. Due to the nature of work, station's learning slopes can be different. We propose a two stage optimization methodology: (1) NLP optimization with some constraints relaxation; (2) Modified line balancing procedure that finds a non-relaxed solution that is the closest to the unconstrained solution found in the first stage. The savings in the optimal makespan over the balanced loading case are demonstrated.