EURO-Online login
- New to EURO? Create an account
- I forgot my username and/or my password.
- Help with cookies
(important for IE8 users)
2454. Minimizing maximum regret via robust optimization in a two-stage facility location problem under demand uncertainty and facility cost uncertainty
Invited abstract in session MD-61: Location under uncertainty, stream Locational Analysis.
Monday, 14:30-16:00Room: S10 (building: 101)
Authors (first author is the speaker)
| 1. | Ronald McGarvey
|
| IESEG | |
| 2. | Andreas Thorsen
|
| Jake Jabs College of Business and Entrepreneurship, Montana State University |
Abstract
Consider a situation in which an initial decision is made to prepare a set J of locations for potential future use supporting a set of customers under uncertain demand and uncertain future facility construction costs. Once the demand is revealed, the decision maker then makes a recourse decision selecting the set K of actual facility locations to be opened, where K is a subset of J. We present robust MILP formulations of this problem, assuming an objective function that minimizes the maximum regret. A budgeted uncertainty set is assumed, in which the realized demand is assumed to be equal to a most-likely value plus a potential positive deviation minus a potential negative deviation. Use of a regret-based objective function has a significant impact on the solution time when directly solving with a commercial MILP solver. In part, this is due to the fact that we assume demand could potentially be less than the expected value at any customer, which is relevant for a regret-based objective (but not for a total cost-based objective). We then develop a computationally-efficient solution algorithm, based on a column-and-constraint generation (CCG) approach, and examine numerical test instances to identify the relationship between problem conditions and solution structure.
Keywords
- Location
- Robust Optimization
- Supply Chain Management
Status: accepted
Back to the list of papers