A Two stage stochastic and risk averse approach to inventory positioning in floricultural auction network

Download
2014
Sezen, Gizem
This study is motivated by a real life problem of an auction company serving in floricultural sector. Company has 6 auction centers. A supplier chooses the auction center that his or her products will be auctioned and the products are delivered to that auction center before the clock. After the auction, sold products are delivered to the buyer's boxes hired at the auction centers. Some ine cient logistic flows are observed in the network of the company due to the positioning of products before the clock. This study aims to find the initial optimal product positioning decisions prior to the auction that minimizes the transportation cost in the network under the uncertainties of buyers and their purchase quantities. Two-stage stochastic integer programming is proposed to model the problem. A scenario based approach is used in the model. In order to solve large sized problems, L-shaped and multicut L-shaped decomposition algorithms are adapted to the problem. In addition, problem is also modeled with risk-averse approach. First-order mean-semideviation is used as the risk measure. Risk-averse model aims to minimize the expected cost of transportation and expected upper deviation from the expected cost. The objective function of the risk-averse model is nonlinear and nondif erentiable. Linearization method and a variation of multicut L-shaped decomposition algorithm for the linearized formulation are proposed for the solution of risk-averse model.
Citation Formats
G. Sezen, “A Two stage stochastic and risk averse approach to inventory positioning in floricultural auction network,” M.S. - Master of Science, Middle East Technical University, 2014.