Global parallel genetic algorithm approach applied to long term generation expansion planning

The hydrothermal generation systems expansion is based on the ability of meeting the future energy market through increasing the existing power plants and/or the increase of the ability in transferring energy among the several regions in the country. The optimal investment to be performed is a function of generating ability of new units which are dimensioned according to the energy generation ability, to the impact caused by new interconnections and to the energy supply criterion. Thus, this work aims at obtaining the optimal planning of the existing power plants through the building perspective of new generation units in order to meet the market in a trustful and economic manner. However, the problem involves several expansion programs of thermal and hydro plants and several synthetic series corresponding to the hydro scenarios, giving the problem a combinatorial problem. In order to do so, herein we will use a genetic algorithm, which presents a particular genetic structure and incorporate rules used by the system planner, to search for the best expansion strategy independently. To improve the agility of the genetic algorithm in achieving the convergence, a solution based on global parallel computing was implemented. The algorithm response time decreases in a quite linear rate as we grow the computers network.