Minimum Viable Data (MVD) for digital twinning: Underserved community energy representation in data scarcity scenarios

Source abstract

Underserved communities often face disproportionately high energy burdens, yet lack access to detailed audits and sensor-based data collection networks that conventional energy studies require. This paper proposes and evaluates a “Minimum Viable Data” (MVD) framework for community-scale digital twinning under conditions of data scarcity. Using electricity billing data, appliance-level consumption records, and publicly available property assessment records for 17 single-family homes in Atlanta, we identify which minimal inputs are sufficient to generate actionable insights to alleviate energy cost burdens. Findings show that monthly billing data combined with assessor records can reveal household vulnerability and risk profiles, with assessed property value emerging as a stronger predictor of proxy energy burden than physical building attributes. Spatial mapping within the digital twin identifies geographic patterns of energy disparity that are less apparent in tabular datasets. This visualization allows for the differentiation between building-related energy losses and household economic constraints. Although appliance-level disaggregation validated consumption patterns, the results demonstrate that high-frequency monitoring is not essential for effective intervention planning. The study concludes that the MVD framework can effectively isolate high-priority households. Specifically, k-means clustering identified a high burden outlier group comprising 17.6% of the sample (n=3), with proxy energy burdens ranging from 24.8% to 36.7% and energy intensities up to 1.99 kWh/ft2. These findings demonstrate that MVD based digital twins can differentiate between structural inefficiencies and economic vulnerability without the need for invasive sensor based audits.