Grid size dependence of electrical consumer energy consumption fluctuations

ORAL

Abstract

Consumer behavior introduces both random and correlated patterns in electrical energy usage. Random fluctuations arise from uncorrelated individual behaviors, while correlated fluctuations emerge from common patterns such as daily routines, seasonal cycles, and shared energy consumption habits. This duality in behavior creates distinct characteristics in electricity demand variability. As the number of users on a grid increases, random fluctuations tend to average out due to statistical smoothing. However, correlated behaviors, such as synchronized energy usage during peak hours, become more pronounced. This dual effect highlights that population size simultaneously reduces some fluctuations and amplifies others, depending on their origin. Using simulated and real world data sets we analyze how electrical energy consumption fluctuations scale with population size (proxy for grid size). Using simple statistical and spectral methods, we empirically find electrical demand variability exhibits giant fluctuations that only start to smooth beyond a population size of 16000, thereby setting a scale for micro, distributed, and island grids, and underscoring the need for engineered methods to smooth fluctuations for such small networks. Beyond a population size of around 0.2 million, the character of smoothing changes, thereby setting a 2nd population size beyond which super grids operate.

*MMB and SM were supported through subsidy funding from the Cabinet office of the Prime Minister of Japan to OIST Graduate University.

Presenters

  • Mahesh M Bandi

    • Okinawa Institute of Science & Technology

Authors

  • Mahesh M Bandi

    • Okinawa Institute of Science & Technology
  • Sayan Mitra

    • Okinawa Institute of Science & Technology
  • Colm P Connaughton

    • London Mathematical Laboratory
  • Ambarish Nag

    • National Renewable Energy Laboratory
  • Jerome Apt

    • Department of Engineering and Public Policy, Carnegie Mellon University