Generation Portfolio Optimization by NPV formulation, Monte Carlo Estimators and Genetic Algorithms

In this paper we present the Portfolio Optimization Problem in the electricity generation framework. We consider traditional and fully controllable energy sources together with wind source, strongly supported by economical benefits but exposed to intermittent generation volatility. A new formulation in terms of uNPV (unit Net Present Value) is proposed and analysed, due to the statistical uncertainty about parameters, we formalize the optimization problem in a probabilistic sense in terms of Monte Carlo Estimators and structured in terms of Risk Aversion factor. The optimization routine is implemented with a Genetic Algorithm.