| In general, capacity planning relies on accurate workload estimation and workload assignment. This praxis presents optimization-based tools for capacity planning in Semiconductor Final Test (SFT) operations. The tools estimate device testing workload using the basic SFT three-stage test model for single-site, single-flow then multi-site, multi-flow processes. The assignment of devices to test cells uses a multi-period network model with side constraints that also accounts for setup time, unmet demand, and lots held for disposition. The models start with deterministic demand but the last capacity planning model presented defines demand as a random variable and the static network assignment model is converted to a two-stage stochastic mixed-integer program with recourse. The results show that large scale SFT capacity planning problems can be solved with guaranteed optimal answers in relatively short run times. Since these models have a basis in realistic capacity planning methodology and parameters they have potential to be very useful as part of a semiconductor manufacturer's supply line management system. |
