On the Improvement of representative demand curves via a hierarchical agglomerative clustering for power transmission network investment

• Published in: Energy (Oxford) 2021-05, Vol.222, p.119989, Article 119989
• Main Authors: González-Cabrera, Nestor, Ortiz-Bejar, Jose, Zamora-Mendez, Alejandro, Arrieta Paternina, Mario R.
• Format: Article
• Language: English
• Subjects: Algorithms; Clustering; Demand curves; Elbow; Elbow rule; Electric power; Electric power demand; Electric power systems; Electricity distribution; Hierarchical agglomerative clustering; High-dimensionality data; K-means; Linkage criterion; Strategy; Transmission expansion planning; Transmission loss
• ISSN: 0360-5442; 1873-6785
• DOI: 10.1016/j.energy.2021.119989

This paper introduces an optimal clustering-based strategy to gain representative demand curves from hourly demand data that allow determining the power transmission network investment by solving the transmission expansion planning (TEP) problem. The proposed approach also provides a high-dimensionality data optimal reduction for the representative demand curves that feed the TEP problem. The key idea behind this strategy is to extract demand patterns from the electric power system demand data through the implementation of a hierarchical agglomerative clustering algorithm (HACA) based on the Elbow’s rule and a linkage criterion, such as Ward’s variance. Then, a 24-h demand pattern is provided by following three different grouping strategies: seasonal, monthly, and weekly. As a second stage, this strategy includes the TEP formulation together with the transmission losses’ linearised model aiming to test the representative demand curves achieved by HACA. To illustrate the efficiency, application, and superior functionality of the proposal, this is implemented over the IEEE 118-node network under several case studies. To determine the most appropriate approach, the results are compared with the well-known K-means method. [Display omitted] •Representative demand curves in TEP problems are drawn by hierarchical clustering.•HACA-based demand curves provide operational costs in a precise and accurate way.•The Elbow-based HACA faces with the high-dimensionality data reduction of TEP.•TEP problems solved by HACA strategy achieve an affordable demand separation.