Bayes Markovian Decision Models for a Multiperiod Reject Allowance Problem

Two versions of a multiperiod reject allowance problem are formulated as Markovian decision models. In the first, total output capacity is considered fixed; in the second, a maximum production rate is postulated. In both models it is assumed that output may be inspected at intermediate points within the production time span as well as at the conclusion of production. Also in both models the production process is characterized as a Bernoulli process with unknown parameter p, the probability of producing a defective. It is shown that optimal sublot size decision rules may be computed via linear programming methods.