Title: Inventory policy characterisation methodologies for a single-product recoverable manufacturing system
Authors: Semra Sebnem Ahiska; Russell E. King
Addresses: Industrial Engineering Department, Galatasaray University, Ortakoy, 34349, Istanbul, Turkey ' Edward P. Fitts Department of Industrial and Systems Engineering, North Carolina State University, Daniels Hall 126A, Box 7906, NCSU Campus, Raleigh, NC, 27695, USA
Abstract: We consider the production/inventory control for a recoverable system with stochastic demand and returns where stock is replenished by manufacturing new items or remanufacturing returned items. The optimal control policy found by solving a Markov decision process (MDP) for a given problem instance can be represented by a list of the optimal manufacturing and remanufacturing decisions for each possible inventory state. This list does not provide any general structure or insight for the optimal control nor is it practical for implementation. We propose two heuristic methodologies which can be used together to provide intuitive, easy to implement, near-optimal to optimal policies. This allows the policy for new scenarios to be determined without requiring solution of the MDP model. Results from a numerical experimentation show that the characterisations provided by our proposed methodologies represent the optimal inventory policies well with small deviations from optimal cost. [Received 9 March 2013; Revised 15 July 2013; Revised 20 November 2013; Accepted 1 January 2014]
Keywords: inventory policy; policy characterisation; product recovery; remanufacturing; Markov decision process; MDP; heuristics; neural networks; recoverable manufacturing systems; inventory control; stochastic demand; production control.
European Journal of Industrial Engineering, 2015 Vol.9 No.2, pp.222 - 243
Available online: 02 Apr 2015 *Full-text access for editors Access for subscribers Purchase this article Comment on this article