EURO-Online login
- New to EURO? Create an account
- I forgot my username and/or my password.
- Help with cookies
(important for IE8 users)
2526. A hybrid approach for sales and operations planning under demand and supply uncertainty
Invited abstract in session TA-49: Stochastic lot-sizing problems, stream Lot Sizing, Lot Scheduling and Production Planning.
Tuesday, 8:30-10:00Room: M1 (building: 101)
Authors (first author is the speaker)
1. | Thorsten Greil
|
TUM School of Management, Technical University of Munich | |
2. | Martin Grunow
|
TUM School of Management, Technical University of Munich |
Abstract
Sales and operations planning (S&OP) problems are rarely addressed with stochastic methodologies since available real-world data often limits applicability. We address this issue in a case study with a global electronics manufacturer and develop a hybrid approach to an S&OP problem under demand and supply uncertainty for multiple products. Decisions cover the full range of S&OP, i.e. raw materials, production, transport to distribution centers, inventories, and demand fulfillment. To tackle demand uncertainty, we use scenario-based two-stage stochastic programming. Scenarios are created with readily available data such as point forecasts and correlated forecast errors. To overcome data limitations, we employ shrinkage covariance estimation. Furthermore, we address supply lead time uncertainty with parametric cost function approximation for a central raw material’s stock level. We thus capture the extent to which raw material safety stock should be altered beyond the level implied by our demand-based stochastic program. We chose this approach instead of extending our scenarios since companies tend to have no regular lead time review process, hardly store related data in a structured way, and do not forecast lead time. Another reason is the higher computational burden if we extend scenarios by supply uncertainty. While conducting out-of-sample rolling horizon evaluation, we optimize the parametrization online and compare our hybrid approach to its deterministic counterparts.
Keywords
- Stochastic Models
- Programming, Stochastic
- Supply Chain Management
Status: accepted
Back to the list of papers