Optimal adaptive inventory control for a multi-location model with redistribution

Köchel considered a multi-location inventory model with redistribution for the discounted cost criterion with full information on the demand characteristics. In the present paper the case that the distribution function of the demand depends on an unknown parameter is investigated. For this results on the discounted cost criterion are carried over to the average cost criterion. The so-called identification scheme from adaptive control theory for discrete-time Markov processes is applied and average optimality for the resulting adaptive inventory policy is shown. Simulation results show the practicability of that approach.