A Data-Driven Approach for Demand Side Flexibility: Reducing Peak Demand and CO2 Emissions in Smart Grid

Abstract

The volatility in energy supply and the flexible demand require more efficient demand management techniques. The residential demand response has the ability to provide a promising solution. It becomes difficult to identify as households differ widely in how they utilize electricity and how flexible their usage patterns are. To address this issue, this study suggests a data-driven dual-clustering framework that utilizes both the electricity consumption behaviour and its flexibility potential for targeting residential consumers. Consumption clusters are formed using cluster validity indices that are used along with a majority voting mechanism, while the flexibility clusters are formed using a novel Demand Response Separability Score that emphasizes operational relevance over relying solely on statistical criteria. The cross-cluster analysis of these clusters reveals how consumption patterns relate to flexibility levels and facilitates the identification of most suitable households for the demand response programs. A simulation to achieve demand response is conducted by shifting loads from peak-hour to off-peak periods for selected flexible groups under a time-of-use pricing scheme. The results show that the suggested strategy achieves $12,928.77 saving in the electricity costs and a reduction of 25,857.54 kg in CO₂ emissions over the study period. The proposed framework offers improved practical feasibility by matching DR participation with household-specific flexibility capabilities. The importance of integrating behavioral and operational characteristics in consumer segmentation provides useful information to design demand response program for the efficient, scalable, and sustainable smart grid.