Characterization of Regulated Electricity Consumption Flexibility Using Variability, Entropy, and Latent Profiling

Energy flexibility in regulated users is examined as a structural property of demand, assessed through variability and disorder metrics derived from smart metering data. Using the coefficient of variation and normalized entropy, the analysis reveals stable routines during weekdays and greater heterogeneity in transitional periods such as evenings and weekends. Non-negative matrix factorization (NMF) is applied to extract latent user pro-files, which are subsequently clustered to uncover representative trajectories of consumption. Groups with bimodal or extended load distributions emerge as the most adaptable, highlighting the role of latent profiling in identifying flexibility potential. Simulations of partial load redistribution demonstrate that, while individual savings remain modest, aggregated benefits and improvements in reliability indicators (SAIDI, SAIFI, ENS) are significant. These findings confirm that flexibility is unevenly distributed across users and time, and that its quantification provides a strategic foundation for differentiated demand response schemes and the design of resilient, user-oriented energy systems.