Existing production/inventory models with random (variable) yield take the yield distribution as given. This work takes a step towards selecting the optimal yield randomness, jointly with lot sizing decisions. First, the authors analyze an EOQ model where yield variance and lot size are to be selected simultaneously. Two different cost structures are considered. Secondly, the authors consider source diversification (‘second sourcing’) as a means of reducing effective yield randomness, and trade its benefits against its costs. Conditions for the superiority of diversification between two sources with distinct yield distributions over a single source are derived. The optimal number of identical sources is also analyzed. Some comments on the congruence of the results with recent JIT practices are provided.
