Stok kontrollü konum belirleme-rotalama problemi: Formüle edilmesi, çözüm metodu ve hesaplamalı sonuçlar

ABSTRACTThe objective of this thesis is to develop mathematical models and solution algorithmsfor the multi-depot location-routing problem (MDLRP) in a distribution network withinventory control decisions. In a distribution network with a single plant, one type of productis sent from the plant via warehouses (i.e., distribution centers or DCs) to a set of customerseach with uncertain demand. Each DC plays a direct intermediary role between the plant andthe customers. Safety stock is kept at DCs to satisfy the desired level of customer service.Decisions related to the design of a distribution network include (i) facility location decisionsthat are considered as strategic decisions; (ii) transportation, or routing decisions that areconsidered as operational decisions; and lastly (iii) inventory decisions (e.g., levels of cycleand safety inventories at the depots) that are considered as tactical decisions. MDLRP withinventory control decisions involves facility (depot) location, warehousing, transportation(vehicle routing) and inventory control decisions in order to determine the location of depots,the optimal set of vehicle schedules and routes between depots and customers, quantityshipped from the plant (supply point) to depots and the inventory level at each depot.The problem is formulated as a warehouse location-routing problem based on themodel proposed by Perl and Daskin (1985). Since finding exact solutions for this problem isNP-hard, a heuristic approach, that is, a modified version of a two-phase tabu searchalgorithm which was proposed first for the solution of LRP by Tuzun et al. is used. Finally, aset of test problems are performed to test and evaluate robustness and efficiency of theproposed method based on tabu search parameters, and computational benchmarking resultsand some concluding remarks are presented.