Characterising electricity demand through load curve clustering: A case of Karnataka electricity system in India. (July 2021)

• Record Type: Journal Article
• Authors: Balasubramanian, S.; Balachandra, P.
• Is Part Of: Computers & chemical engineering. Volume 150(2021)
• Publication Date: 2021-07
• Subjects: Load curve clustering -- Electricity system modelling -- Renewable energy transition -- Representative load curves; Chemical engineering -- Data processing -- Periodicals; 660.0285
• DOI: 10.1016/j.compchemeng.2021.107316
• Languages: English
• ISSNs: 0098-1354

Abstract

Highlights: Simulation approach for load curve classification and demand representation. Seven Representative Load Curves to characterize 365 load curves. Characterizing demand based on weather, economic, agricultural & socio-religious factors. Knowing factors influencing demand variations helps in prioritizing interventions. 90% of hourly loads fall in the range of ±12% error when compared to actual data. Abstract: Till recently, in the case of electricity systems with dominant conventional generation technologies, the supply was largely driven by dynamic demand and thus, it was characterized by fixed supply and variable demand. However, the ongoing transformations in electricity systems leading to mainstreaming of variable renewable energy sources and technologies have introduced variability even on the supply-side. This variability in both supply and demand for power have presented several challenges for both planning and operating such systems. On the supply-side, these variabilities need to be accounted while planning installed capacity additions in the long-term, resource allocations and generation scheduling in the medium-term, and system operation in the short-term. Similarly, demand-side-management strategies need to consider variations in both demand and supply while moderating the load curves. To perform these tasks effectively, it is important to have a better understanding of the dynamic demand in terms of patterns of demand variability, magnitude of variability, span of variability, factors influencing variability, and temporal variability. In this research, we propose to use simulation-based approach as an alternative method to classify load curves and verify its effectiveness for accurate and simpler representation of variability in electricity demand. Initially, logical clustering is performed to segment daily load curves, and then probability distribution-based Monte Carlo simulation is employed to establish seven load curves that represent seven groups of 365 load curves. The results show that these seven groups of load curves represent different weather patterns, lifestyles, and economic and socio-religious activities.