The social factors shaping community microgrid operation

The study addresses how communities should manage scarce stored energy within microgrids during prolonged, disaster-induced blackouts. It examines social acceptance of a differentiated service model in which residents who pay more receive higher energy quotas, versus equal quotas for all. Two hypotheses guide the work: H1, that respondents would oppose a conventional market-based system and favor equal quotas due to the life-critical nature of electricity during blackouts; and H2, that willingness-to-sell (WTS) energy from personal storage to the microgrid would decline as expected blackout duration increases or as personal storage capacity decreases, reflecting self-preservation. The context includes increasing frequency of extreme weather, growing maturity and deployment of microgrids, and the critical need to integrate social preferences into microgrid operation during scarcity. The purpose is to quantify preferences for differentiated service and to measure WTS under varying conditions to inform socially acceptable, economically viable microgrid operation.

Prior research has extensively measured willingness-to-pay (WTP) for uninterrupted power supply across regions including the US, Europe, Ghana, India, and Indonesia. These studies inform economic planning but do not resolve operational rationing under finite stored energy during blackouts. Literature indicates perceived fairness influences WTP for services, implying that fairness perceptions could affect acceptance of microgrid backup schemes. Some prior work integrates fairness indicators into microgrid control but does not establish how such fairness parameters should be determined socially. Existing analyses also often omit explicit constraints on total available backup energy, limiting their utility for quota design.

Design: A web-based survey (Qualtrics) administered via Amazon Mechanical Turk recruited 1,021 unique respondents (February 2024). Eligibility required age ≥18 and residence in the continental US for the past 5 years. IRB approval (National University of Singapore) and informed consent were obtained. Survey structure: Respondents were educated on consequences of extreme-weather blackouts and increasing outage frequency; reported prior blackout experience, essential loads, and ownership/plans for backup solutions; were introduced to a community microgrid concept and finite stored energy implying zero-sum rationing during blackouts. Differentiated service assessment: Likert-scale (0–10) items gauged perceived fairness of a differentiated service model (quota proportional to payment) versus equal quotas, views on quota-based usage, and strict enforcement (cutoff when quotas are exhausted). Payment plan preferences for backup service were elicited among options with different fixed/variable annual costs ($200 fixed; $100 plus $20 per blackout day; $50 plus dynamic $20–40 per blackout day), along with willingness to subsidize economically weaker residents. WTS experiment: Respondents were randomly assigned to one of three hypothetical ownership cohorts representing different storage capacities and typical household autonomy: Ford F-150 EV (~10 days), Tesla PowerWall (~3 days), or portable diesel generator (~2 days) (cohorts n≈339–342 each). WTS was reported as a percentage of personal storage, for anticipated blackout durations of 2 days and 7 days, to avoid kWh anchoring. Preferences for compensation type when selling to the microgrid (cash, fee reduction, higher usage quota, free sharing, unwilling) and indicative price expectations were collected. Analysis: Because response distributions were not assumed normal, non-parametric tests were primary. Correlations between ordinal variables used Spearman tests. Where relevant, linear regressions were fitted and tested via two-sided t-tests. Paired-sample tests assessed within-respondent changes in WTS between 2-day and 7-day scenarios. Two-sample Kolmogorov–Smirnov tests compared distributions across cohorts to assess storage-capacity effects on WTS differences. Demographic covariates (income, age, education, blackout experience, employment status, sex, ethnicity) were analyzed via linear regression for associations with perceived fairness and WTS.

• Support for differentiated service: 91.8% of respondents rated ≥5/10 in support of a differentiated service paradigm (mean 6.595, s.d. 2.065). Quota-based usage and strict enforcement (cutoff after quota exhaustion) also received strong support, contradicting H1.
• Payment plans for backup: Among those expressing a preference, 46.82% favored $200/year fixed, 27.62% favored $100/year plus $20 per blackout day, 22.14% favored $50/year plus dynamic $20–40 per blackout day, and 3.43% had no preference. Over half preferred variable-cost plans, suggesting expectations of fewer blackout days. Respondents with prior long-duration blackouts showed greater preference for the $200 fixed plan. Willingness to subsidize economically weaker residents clustered around a median $50–$150/year, with a positive correlation (Spearman p≤0.05) between higher personal backup commitment and higher subsidy amounts, indicating an anchoring effect.
• Willingness-to-sell (WTS) participation: 91.4% were willing to participate in selling surplus energy (mean 6.414, s.d. 1.860 on 0–10 Likert).
• WTS magnitude: Mean WTS across cohorts ranged roughly 42–53% of available personal storage. For each cohort, median WTS was 45% for a 2-day and 55% for a 7-day anticipated blackout; min 5%, max 95%. Standard deviations: Ford F-150 (2d 0.187; 7d 0.224), PowerWall (2d 0.168; 7d 0.226), Diesel generator (2d 0.175; 7d 0.216).
• WTS vs blackout duration: Contrary to H2, WTS increased with longer anticipated blackout (paired tests p=0 for all cohorts). Linear regression slopes (2d vs 7d): Ford F-150 0.464 (intercept 32.51%), PowerWall 0.617 (24.53%), Diesel 0.292 (39.59%); all p=0. KS tests comparing WTS-increase distributions across cohorts yielded p-values 0.992, 0.999, 0.978 (>0.05), indicating no significant dependence on storage capacity for the change distribution. Since WTS was in percentage terms, actual energy sold would still scale with capacity, partially aligning with H2.
• Motivation for higher WTS in longer blackouts: Increased WTS correlated negatively with stated willingness to let neighbors charge devices (Spearman −0.098, p=0.0017), arguing against purely communal responsibility. Respondents preferring compensations (fee reduction, cash, higher quota) showed a 7.64% increase in average WTS (paired test p=0), suggesting revenue motivation.
• Compensation preferences for selling: 43.10% preferred higher usage quota, 29.08% microgrid fee reduction, 20.57% cash, 14.20% freely share, 2.06% unwilling to share. Many expected a price higher than the normal electricity rate.
• Interdependence: WTS correlated positively with perceived fairness of differentiated service (regression slope 0.322, t-test p=0), indicating alignment between acceptance of market-based consumption and willingness to sell.
• Demographics: Income positively associated with perceived fairness and WTS; age and education negatively associated with both; prior blackout experience negatively associated with perceived fairness (not significant for WTS). Sex, ethnicity, and employment status showed no significant effects. These patterns align with determinants found in WTP literature.

Findings show strong social acceptance for a market-based, differentiated service approach to rationing scarce microgrid energy during blackouts, refuting H1. Consumers generally endorse quota-based usage with strict enforcement and are willing to pay for backup services, including subsidizing less affluent neighbors. High WTS rates and preferences for compensation indicate readiness to integrate personal storage into community operations. Contrary to H2, WTS increased with longer anticipated blackouts, primarily for monetary reasons rather than communal altruism, though WTS still scales with storage capacity. The positive linkage between support for differentiated service and WTS suggests a coherent acceptance of energy-as-a-service during emergencies. These results support microgrid operators and policymakers in designing socially acceptable market mechanisms—such as tiered quotas, variable pricing, and incentives (fee reductions, higher quotas, cash)—to leverage distributed storage and reduce centralized storage investment, enhancing technical and economic viability.

This study quantifies social preferences for operating community microgrids under energy scarcity. Most respondents support differentiated, market-based rationing and are willing to sell substantial portions of their personal storage, especially with appropriate incentives. The results provide actionable guidance for energy-as-a-service models in microgrids, including quota design, payment plan structuring, and compensation mechanisms for distributed resource participation. Future research should develop and field-test operational market mechanisms for community microgrids, examine behavioral dynamics during real blackout events, assess generalizability across regions and cultures, and track how attitudes evolve with growing blackout experience and microgrid adoption.

• External validity: Respondents were US-based; attitudes may differ elsewhere. The community microgrid concept is not yet widespread in the US, and responses may reflect hypothetical contexts.
• Hypothetical bias: Many respondents do not currently own the specified storage devices; WTS and fairness assessments may differ in real transactions.
• Framing/education effects: The survey included detailed blackout consequence descriptions, which may reduce the influence of personal experience and shape responses.
• Measurement choices: WTS reported as percentages rather than kWh avoids anchoring but limits direct translation to absolute energy quantities without device-specific capacities.
• Temporal dynamics: Preferences for fairness, sharing, and WTS may change over time with evolving blackout experiences or technology adoption.
• Self-report and online panel biases: As with any survey, responses may be subject to self-reporting and panel selection biases.