Flexibility of building virtual energy storage: a systematic quantitative review of modeling and controlling heating and cooling systems

Rapid urbanization and building electrification have made building heating and cooling a major contributor to electricity demand and peak load. Virtual energy storage (VES), enabled by building thermal inertia and heating and cooling operation, offers a low-cost option for load shifting and demand response; however, its flexibility remains difficult to quantify with reported performance strongly influenced by modeling assumptions, operating conditions, and control strategies. The flexibility of VES is closely tied to the operation of building heating and cooling systems, and inconsistent characterization of VES has limited the comparability of existing studies and the scalability of control strategies. Developing efficient operational strategies can enhance both energy efficiency and thermal performance in buildings. This paper conducts a systematic quantitative review of optimization and control for VES-integrated building heating and cooling systems, explicitly covering both demand-side and supply-side operation. A structured literature search identifies 124 closely related studies for in-depth analysis. We classify VES modeling approaches into physics-integrated and pure data-driven methods, and compare three categories of control strategies: traditional control, model predictive control, and model-free control (including reinforcement learning). The reviewed studies show that exploiting VES can enable notable peak shaving (up to 62%) and increase renewable energy utilization (up to 93%), while the effective VES capacity may vary by as much as a factor of two under different factors. Finally, a systematic quantitative framework is synthesized to support standardized evaluation of VES flexibility and to facilitate more consistent comparison of modeling and control strategies in building heating and cooling systems.