A new clustering algorithm for load profiling based on billing data

In open energy markets, the settlement process between distribution operators and traders is made on an hourly (or 15 min) basis, while LV consumers’ billing data continues to result from monthly energy bills. In order to reconcile these two different realities, load profiling is used as a means to redistribute the consumed energy of each trader's portfolio by hourly intervals, according to recorded consumption patterns. This paper presents a new clustering approach to derive typical load diagrams that can be used in the process. The algorithm uses real load diagrams obtained in measurement campaigns to define classes (in the billing information space) that maximize the compactness of the diagrams in each class. The methodology was developed in a project with EDP Distribution (the Portuguese distribution system operator) and the result was approved by the Regulatory Authority that adopted the proposed profiles for market use.

► We propose a method to derive load profiles to be used on market settlement process.
► The clustering approach optimizes load diagrams compactness in each class.
► Consumers’ class assignment process is based on billing data.
► We compute energy and economic deviations to illustrate the adequacy of the profiles.
► The load profiles were approved by the Portuguese regulator for market use.