Combining Statistical Clustering Techniques and Exploratory Data Analysis to Compute Typical Daily Load Profiles - Application to the Expansion and Operational Planning in Brazil

Power system planning demands a huge computational effort and to mitigate this some simplifications are needed, among them the use of approximated load curves or typical daily load profiles expressed by a small number of load levels (e.g., heavy, medium and light). This paper presents a methodology to build typical daily load profiles useful to the expansion and operational planning studies. The proposed approach is based on statistical clustering techniques and on exploratory data analyses (EDA), which are preceded by a filtering procedure. It is also proposed the creation of LoadPlot curves, which consist of putting in the same curve boxplots for each hour of the day together with the aggregated levels. The proposed methodology is illustrated by using real data of the Brazilian interconnected electric power system.