Optimization of Recurrent Stochastic Capacity Expansion Models and Generalization to a Non-Recurrent Model.
Abstract
The determination of optimal expansion sizes X*(K)+1, for recurrent capacity expansion strategies of the form (X,K) where X+1 represents the expansion size undertaken whenever excess demand reaches the value K+1 is considered here, for fixed values K. A POlicy Improvement algorithm is derived to determine optimal expansion sizes for general expansion functions. For a certain class of expansion functions, which includes discrete convex functions, it is shown that the expected discounted costs are unimodal in the expansion size and that the optimal expansion size X*(K)+1 increases monotonically with K; an Interval-Bisection algorithm is given to determine X*(K) for this case. The monotonicity of the optimal expansion size is also demonstrated, prior to integer-truncation, for important classes of concave expansion functions which are continuiously differentiable; a simple Function Iteration algorithm is derived to determine optimal expansion size for this case.
Document Details
- Document Type
- Technical Report
- Publication Date
- Oct 11, 1976
- Accession Number
- ADA033429
Entities
People
- R. Scott Shipley
Organizations
- Stanford University