EURO-Online login
- New to EURO? Create an account
- I forgot my username and/or my password.
- Help with cookies
(important for IE8 users)
2615. Multistage Sales and Production
Invited abstract in session TD-34: Trends and Open Problems in Robust Optimization, stream Stochastic, Robust and Distributionally Robust Optimization.
Tuesday, 14:30-16:00Room: 43 (building: 303A)
Authors (first author is the speaker)
| 1. | Dominic Keehan
|
| Engineering Science, Univeristy of Auckland | |
| 2. | Andy Philpott
|
| Engineering Science, The University of Auckland | |
| 3. | Andrew J Mason
|
| Dept Engineering Science, University of Auckland |
Abstract
We study a multistage sales and production problem. At each stage a market price is observed, sales decisions are made, and resources are available for a production decision which yields products to be sold at later stages. The prices of each product evolve randomly. To address model ambiguity, we study two distributional robustifications of the sample average version of this problem. The model is applied to a problem arising in New Zealand's dairy industry. We first compare the out-of-sample performance of the sample average policy to that of a model predictive control policy which is shown to be distributionally robust with an ambiguity set of distributions with matched means. We then compare this with a distributional robustification using discrete Wasserstein distance-based ambiguity sets. The models are solved using stochastic dual dynamic programming in SDDP.jl.
Keywords
- Stochastic Optimization
- Robust Optimization
- Optimal Control
Status: accepted
Back to the list of papers