Forecast uncertainty modeling and data management for a cutting-edge security assessment platform

Abstract

The increasing penetration of renewables and the constraints posed by pan-European market make more and more crucial the need to evaluate the dynamic behaviour of the whole grid and to cope with forecast uncertainties from operational planning to online environment. The FP7 EU project iTesla addresses these needs and encompasses several major objectives, including the definition of a platform architecture, a dynamic data structure, and dynamic model validation. The on line security assessment is characterised by a multi-stage filtering process: this includes a “Monte Carlo like approach” which applies the security rules derived from extensive security analyses performed offline to a set of “new base cases” sampled around the power system (PS) forecast state with the aim to discard as many stable contingencies as possible. The paper will focus on the management of historical data - related to stochastic renewable and load snapshots and forecasts-in order to solve some intrinsic criticalities of raw data and to derive a reliable model of the multivariate distributions of renewables and loads conditioned to the specific forecast state of the grid, with the final aim to generate the “uncertainty region” of states around the forecast state.