Impact of massive penetration of intermittent generation and energy storage devices in planning the amount of transmission system usage

Abstract

The increase in the penetration levels of intermittent generators (e.g. photovoltaic generation) and of energy storage devices (e.g. batteries) in the Brazilian distribution networks will significantly affect the demand provided by the distribution company and, consequently, the techniques currently employed to predict the behavior of the load and to determine the Amount of Transmission System Usage (MUST, in Portuguese) at the points of interconnection with the transmission company. If the contracted demand is violated and there is subcontracting (above 110% of the contracted MUST) or over-contracting (below 90% of the contracted MUST), the distribution company must pay fines to the transmission company. The main difficulty and novelty of this new distribution network scenario are the highly stochastic characteristics associated with the intermittent generation and with the operation of energy storage devices in the distribution network. In addition to the stochastic nature inherently associated with these technologies, the distribution company does not have previous knowledge over the number of connections of photovoltaic generators and energy storage devices to the low voltage grid that will occur in a period (e.g. 1 year) or the locations of these connections. In this context, this project fits within the scope of strategic planning, with the objective of developing a new probabilistic methodology based on an optimization model to determine the optimal value of MUST to be contracted by the distribution company in scenarios with high penetration of photovoltaic generation and energy storage. This project will cope with uncertainties through scenario-based stochastic programming, which properly incorporates the uncertainties present in this planning problem. Real data of Brazilian networks from Campinas, São Paulo, will be used and the optimization methodology will be developed from real networks simulations under some different scenarios. Studies evolving the Monte Carlo method will be performed in order to identify the dominant factors of this problem and, thereafter, a generic methodology will be developed and implemented in the mathematical programming language AMPL and solved using commercial solvers. (AU)