Microgrids face a fundamental challenge: balancing variable renewable energy supply with unpredictable demand while encouraging customers to shift consumption during peak periods. Demand response programs offer a solution, but their real-world effectiveness is hampered by two interconnected problems—uncertain renewable generation and uncertain user behavior.
A new research framework addresses both challenges through coordinated day-ahead and intraday scheduling. Rather than treating demand response as a purely technical problem, the model incorporates behavioral economics, specifically modeling the "endowment effect"—the psychological resistance users experience when asked to deviate from habitual consumption patterns.
The approach uses different mathematical strategies across different timescales. For day-ahead planning, scenario-based stochastic optimization handles long-term uncertainty about wind and solar generation. For real-time intraday operations, fuzzy chance-constrained optimization adapts to short-term forecast errors and rapidly changing conditions.
Testing reveals substantial practical improvements. The endowment-effect model reduced psychological costs on customers from approximately 14,914 CNY to 2,347 CNY—meaning fewer users drop out of demand response programs. Importantly, the framework requested only 3.33 MW of demand response, a more realistic quantity than approaches ignoring behavioral factors.
The scenario-generation method also outperformed standard approaches by 67.6% on autocorrelation accuracy and 69.8% on ramp-rate performance, better capturing actual renewable variability patterns. Meanwhile, the two-stage framework cut online computation time by 84%, enabling real-time implementation on actual microgrid controllers.
Comparisons with unified optimization approaches showed the hybrid framework increased intraday revenue by 642 CNY while reducing unnecessary demand curtailment by 0.48 MW. This balance—protecting revenue, minimizing risk, respecting user behavior, and maintaining computational tractability—reflects the engineering reality that optimal solutions must account for technical, economic, and human factors simultaneously.
