Industrial Engineering MS Thesis Defense by Mustafa Gökçen Göksel



*******************************************************************

KOÇ UNIVERSITY

GRADUATE SCHOOL OF SCIENCES & ENGINEERING

INDUSTRIAL ENGINEERING

MS THESIS DEFENSE BY MUSTAFA GÖKÇEN GÖKSEL

******************************************************************

 

Title: Procurement Strategies of a Commodity with a Spot Market

 

Speaker: Mustafa Gökçen Göksel

 

Time: August 11, 2017, 10:30

 

Place: ENG 208

Koç University

Rumeli Feneri Yolu

Sariyer, Istanbul

Thesis Committee Members:

Prof. Fikri Karaesmen (Advisor, Koç University)

Prof. Gürhan Kök (Koç University)

Asst. Prof. Bora Çekyay (Doğuş University)

Abstract:

We investigated the single period procurement policy of a commodity which is traded in a spot market. The presence of a spot market differentiates the problem definition from the classical single period inventory models, and requires different solution strategies. We have proposed two data-driven linear programs to find the optimal procurement strategies for two different objectives: (i) expected profit maximization (risk-neutral), (ii) minimization of CVaR (risk-averse). Our proposed solution approaches use historical demand and price data as well as the historical data of exogenous variables which are assumed to be correlated with demand and/or price. We have conducted a simulation to ensure the validity of the models. According to the simulation results, the risk-neutral case is able to yield near-optimal solutions for all price processes and all price-demand dependence cases. The risk-averse model is able to outperform the risk-neutral one in terms of CVaR risk measure for certain cases of price processes and price-demand dependence cases; and the presence of the exogenous variables improved the performance of program. We also investigated a slightly modified version of the initial problem that has multiple procurement opportunities, in order to establish a theoretical basis for our study. The problem is modeled as a Stochastic Dynamic Program. We found the optimal inventory level which is the state dependent base stock policy, and analyzed the structural properties of the optimal solution.