A model of the California system that consists of 130 generators, 225
buses and 375 lines. This model uses a scenario selection algorithm
inspired by importance sampling and resorts to a synchronous implementation
of the subgradient method. The trade-off between scenario set size and
optimality gap is examined. We report on the economic benefits of
stochastic unit commitment relative to a deterministic nodal pricing model.

* A model of the Central Western European (CWE) system that consists of 656
generators, 679 nodes and 1073 lines. The model uses a scenario selection
algorithm based on sample average approximation and resorts to a parallel
implementation of the asynchrnous subgradient method. Stochastic unit
commitment is compared to current practice in the California power pool. We
report on the economic benefits of stochastic unit commitment relative to
(i) a deterministic nodal pricing model and (ii) the zonal pricing model
that is currently employed in the CWE market.