Who uses power, who gets microgrids? Uncovering regional trends in energy consumption and microgrid deployment in California

Abstract

This study develops a spatial-statistical framework to examine how the geography of operational microgrid infrastructure co-occurs with residential electricity demand and community context across California. We combine quarterly ZIP-code residential electricity consumption data from major utilities (SCE, PG&E, and SDG&E) with physical, built-environment, and sociodemographic indicators harmonized to a common spatial support through a Geographic Information System (GIS) workflow. Microgrid locations are geocoded to facility-level points from the U.S. Department of Energy registry, enabling kernel density estimation of deployment concentration. Time-series clustering to characterize distinct demand regimes, local hotspot statistics to detect spatial concentrations of energy-use intensity, and multivariate clustering and spatial regression (including geographically weighted regression) were applied to summarize associations between demand, contextual variables, and microgrid presence. Importantly, the analysis is designed to identify spatial “opportunity structures” where existing microgrid assets overlap with elevated residential demand and equity-relevant community characteristics under a socio-technical transition lens and can become a strong lever for building community microgrids for communities in need. Results indicate strong, statewide seasonality in demand with stable cluster ordering over time, and show that associations with homeownership, population density, and racial/ethnic composition are more pronounced than those of temperature or air quality in explaining demand patterns. Microgrids remain concentrated in urban areas and are largely driven by anchor institutions (hospitals, universities, military sites), limiting systematic overlap with community indicators. The proposed framework supports transparent, extensible opportunity mapping for community-oriented microgrid pathways and can be replicated in other states as data availability improves.