Sourcing Decision and Inventory Management in Supply Chain

This study is dedicated to supplier selection problem (SSP) with two different aspects: (1) SSP with deterministic demand, and (2), SSP with stochastic demand. For both aspects, we assume that all suppliers were already pre-evaluated according to some criteria, such as financial strength and performance history, and now the buyer needs to further assess the suppliers for order allocation based on some quantitative criteria such as purchasing cost, etc. SSP with deterministic demand: Such problems are usually modeled as a multi-objective optimization problem (MOOP) subject to suppliers’ capacity, buyer’s demand, etc. Number of objectives and constraints vary from one problem to another. For solving the MOOP, three different cases have been considered in the literature: Case 1) the Decision Makers (DMs) determine a goal for each objective and then try to drive achieved objectives towards their goals as close as possible (this case is known as the crisp MOOP); Case 2) the DMs determine the weights of objectives instead of the goals so that the objectives approach their possible ideal solution according to their weights (this case is known as the fuzzy MOOP); Case 3) the DMs determine multiple goals or an interval goal for each objective and then try to drive achieved objectives towards the goals (known as Multi-Choice Goal Programming (MCGP) model). In Case 1, goal programming is the most famous and wildly used approach in the literature. However, the approach is not able to make the achieved objectives consistent with their goals. In chapter 2 (as the first contribution), we develop a normalized goal programming approach to achieve some levels of consistency among different objectives. Then, the proposed approach is extended for solving the fuzzy MOOP of Case 2. Due to uncertainly/imprecision, Case 3 may be more applicable than Case 1 as the DMs can determine an interval goal (or multiple goals) for every objective. In chapter 3 (as the second contribution), we propose an improved MCGP approach providing the DMs with more control over their preferences in comparison with the previous models. SSP with stochastic demand: Such problems can be modeled by using probability distributions. The newsvendor model is one of the most famous models in this area that can be wildly used in reality because of the decline of product life cycle. In Chapter 4, we consider a multi-period SSP where a buyer procures an item (i.e., raw material) from a set of capacitated suppliers to meet the final product stochastic demand in order to maximize his/her expected profit. The suppliers may offer quantity discount as a competitive factor to induce the buyer to purchase more. We first model the problem by mixed integer nonlinear programming, and then propose an algorithm for solving the model (as the third contribution). In Chapter 5, we consider a single period newsvendor problem where a buyer purchases a single item from a set of capacitated suppliers. In this problem, we assume that both the demand and supply are uncertain. Wholesale prices offered by the suppliers and their supply uncertainties are considered as the competitive factors. In order to compensate the supply uncertainties, the suppliers may allow the buyer to return unsold products at the end of season (buyback policy). Therefore, the buyer has to take into account three criteria (suppliers’ wholesale price, suppliers’ unreliability level, and suppliers’ buyback price) for evaluating the suppliers that contributes to the complexity of the problem. In this chapter, we develop an algorithm to solving such problem (as the forth contribution).