Analysis of purchasing policies with rolling horizon under uncertain price and demand conditions

ABSTRACTPhd. THESISANALYSIS OF PURCHASING POLICIES WITH ROLLING HORIZONUNDER UNCERTAIN PRICE AND DEMAND CONDITIONSErcan ŞENY Ğ TDEPARTMENT OF INDUSTRIAL ENGINEERINGINSTITUTE OF NATURAL AND APPLIED SCIENCESÇUKUROVA UNIVERSITY: Assoc. Prof. Dr. Rızvan EROLSupervisorYear : 2006 Pages : 275: Assoc. Prof. Dr. Rızvan EROLJuryAssoc. Prof. Dr. Türkay DERELAssoc. Prof. Dr. Adil BAYKASOĞLUAssist. Prof. Dr. S. Noyan OĞULATAAssist. Prof. Dr. Adem GÖLEÇPurchase lot-sizing problem is essentially to supply raw materials and parts atthe right time, at the right amount and at the right place with economic price andquality. This problem can be formulated in deterministic or stochastic conditions.However, finding optimal purchasing policies under stochastic conditions such asdemand, availability and purchasing price is a considered very difficult problem.There are a large number of papers dealing with deterministic lot-sizing problem.Analytical solutions for stochastic lot-sizing are possible under very strictassumptions that limit utility of such solutions in practice.This study proposes new purchasing policies that consider demand andpurchasing price uncertainties simultaneously. Moreover, purchasing decisions aremade on a rolling horizon basis rather than fixed planning horizon. The proposedpurchasing policy is basically based on a lost-gain evaluation at decision points.Also, well-known Least Unit Cost and Silver-Meal algorithms are modified for bothtime varying purchasing price and rolling horizon.Simulation analysis is conducted for different values of set-up cost to holdingcost ratio, number of rolling horizon periods, coefficient of variation of demand andprice, backordered demand to unsatisfied demand ratio, service level factors. Policiesare compared with respect to deviation from Wagner-Whitin solution, purchasingcost, holding cost, lost sale cost, backordered cost, set-up cost, positive and negativeinventory levels. Simulation results show that developed lost-gain purchasing policygenerally outperforms the other policies.Key Words: Inventory, stochastic lot-sizing, rolling horizon, simulation.