Could the PCE method be a better alternative to the Monte-Carlo method on the Power System Adequacy Assessment problem to make it more efficient computationally?

Details

Supervisors

Contino, Francesco ; De Jaeger, Emmanuel

Faculty

Ecole polytechnique de Louvain

Degree label

Master [120] : ingénieur civil électricien, à finalité spécialisée

Abstract

With the relatively recent wide use of PCE in the energy sector, the PCE method is explored as a potential better alternative than the Monte-Carlo method on Power System Adequacy Assessment problems to make it more efficient computationally. Using Adequacy and PCE tools, a PCE process is created and the results are compared with the initial Monte-Carlo method. With a high Leave-One-Out (LOO) error, new methods are explored in the hope of improving the initial method. First, a transformation of the outputs LOE (Loss-of-Energy) and LOL (Loss-of-Load) into continuous output variables is performed, thanks to the extrapolation of monotonic curves for each profile (set of realizations of the uncertainty). Those extrapolations can be linear or quadratic. The LOO error is still high; profiles are divided in subprofiles (same realizations of uncertainties but different start times), the mean LOE and LOL of the subprofiles being the LOE and LOL of the profile. A certain combination of number of subprofiles, parameters of extrapolation and order of PCE permits to have decent results, according to the LOO error, to compare them with Monte-Carlo and to explore the analysis of a study case through PCE. Thanks to a decision tree, a hierarchy of units impacting the adequacy of the power system can be established and be served for deeper research.