Impact of Input Model Accuracy on Probabilistic Load Flow Outputs

Various methods of probabilistic load flow (PLF), as used for planning and design, accounting for uncertainties in the system variables by applying probability-based input models, statistical computation of the load flow, and characterization of the output distributions. The choice of a probability distribution model to characterize the system inputs has implications for the accuracy and computational efficiency of the probabilistic solutions. The impacts on computational efficiency and complexity are well studied, but accuracy has only been assessed at the input modelling stage and not in terms of the PLF outputs. This paper assesses the effects of input model accuracy on PLF outputs, comparing two forms of modelling: mixed models and standard unimodal parametric distributions. Simulations demonstrate that although mixture models achieve high levels of accuracy in modelling the variability in inputs, the differences in PLF outputs are not always significant. Where the actual input is nearly unimodal, the errors in assuming a unimodal, standard distribution are insignificant.