Return to Colloquia & Seminar listing
Solving Multi-Stage Stochastic Capacity-Planning Problems
Optimization| Speaker: | Prof. Andy Philpott, University of Auckland |
| Location: | 2112 MSB |
| Start time: | Fri, Jan 20 2006, 2:10PM |
Description
We describe a general multi-stage stochastic integer-programming model
for planning the discrete capacity expansion of production facilities
in a situation in which a scenario tree represents uncertainty. A split-variable reformulation of the basic model leads to a strong
Dantzig-Wolfe decomposition, whose efficiency we demonstrate on a
capacity-expansion model for electric power distribution.
When a facility can expand at most once over the planning horizon,
a specialised model yields substantial computational savings.
(Joint work with Kavinesh Singh and Kevin Wood)