Many electric utilities have been tapping such energy sources as wind energy or conservation for years. However, the literature shows few attempts to incorporate such non-dispatchable energy sources as decision variables into the long-range planning methodology. In this dissertation, efficient algorithms for electric utility capacity expansion planning with renewable energy are developed.

The algorithms include a deterministic phase which quickly finds a near-optimal expansion plan using derating and a linearized approximation to the time-dependent availability of non-dispatchable energy sources. A probabilistic second phase needs comparatively few computer-time consuming probabilistic simulation iterations to modify this solution towards the optimal expansion plan.

For the deterministic first phase, two algorithms, based on a Lagrangian Dual decomposition and a Generalized Benders Decomposition, are developed. The Lagrangian Dual formulation results in a subproblem which can be separated into single-year plantmix problems that are easily solved using a breakeven analysis. The probabilistic second phase uses a Generalized Benders Decomposition approach. A depth-first Branch and Bound algorithm is superimposed on the two-phase algorithm if conventional equipment types are only available in discrete sizes. In this context, computer time savings accrued through the application of the two-phase method are crucial.

Extensive computational tests of the algorithms are reported. Among the deterministic algorithms, the one based on Lagrangian Duality proves fastest. The two-phase approach is shown to save up to 80 percent in computing time as compared to a purely probabilistic algorithm.

The algorithms are applied to determine the optimal expansion plan for the Tijuana-Mexicali subsystem of the Mexican electric utility system. A strong recommendation to push conservation programs in the desert city of Mexicali I results from this implementation.