Robust Dual Sourcing with Two Spot Market Supplies
DescriptionThis research is motivated by the challenge faced by a fast-food chain. Frozen beef (more accurately, a specific beef product called short plate) is the most important raw material to the firm, accounting for over 30% of its operating costs. Consequently, beef price fluctuations often endanger the firm’s profit margin, and a good sourcing strategy is therefore critical to the firm's profitability. Specifically, the firm sources from two suppliers: an overseas supplier, called packer, with a longer lead time and a local supplier, called trader, with a shorter lead time. However, the packer’s price is not necessarily always lower than the trader's price.These two prices are correlated and fluctuate over time. The firm's problem is to find the optimal sourcing policy that minimizes its purchasing cost subject to the no-stockout constraint.The purpose of this project is to study the optimal sourcing policy for such a firm that faces two suppliers with fluctuating and correlated prices. The problem is complicated by the long lead times, which means the curse of dimensionality when applying dynamic programming, and by the stochastic wholesale prices, which lack practically reliable and accurate forecasting models. To address these two challenges, we propose modeling the prices and demand with the central limit theorem uncertainty set and solving the problem with robust optimization. Successful attempts have recently been made in applying a similar idea to some classical inventory control problems and the dual-sourcing problem with random demand. With the proposed approach, we will be able to not only efficiently compute an actionable high-quality sourcing policy but also provide structural insights regarding the optimal policy.
|Effective start/end date||1/01/19 → 27/06/22|