Planning for production of a set of components when yield is random

Random yield in semiconductor manufacturing poses a major challenge for production planning. The task becomes even more difficult if the demand is in terms of a set of components. Here, model are presented which explicitly take account of yield losses and show how available chip sites should be allocated among various types of chips to meet the demand of a set with the maximum probability. This allocation model also provides tradeoff curves between the number of chip sites assigned for the production of a set and the highest attainable level of confidence in meeting the demand. The models presented here can be used at the aggregate production planning stage where jobs have to be released into the fabrication line based on demands.<>