Energy Management Approaches for Building-Scale Hybrid Energy Storage Systems: A Comparative Assessment Based on Design Considerations and Performance Metrics

Building-scale hybrid energy storage systems combine electrical, thermal and hydrogen-based storage technologies to manage renewable generation variability, building demand and resilience requirements across time scales. However, the benefits of such systems depend not only on the characteristics of storage technologies but also on the ability of the energy management system to coordinate their operation. This study presents a comparative assessment of energy management system (EMS) methods for building-scale hybrid energy storage systems, including batteries, supercapacitors, flywheels, thermal energy storage and hydrogen-based storage, proposing a framework matching storage characteristics to control objectives and performance metrics. The examined EMS methods are classified as rule-based control, mathematical optimization, model predictive control, metaheuristic optimization and artificial intelligence-based approaches, while centralized, distributed and hierarchical control architectures are evaluated separately. Storage design considerations (response time, power/energy limits, degradation, forecast uncertainty) and comparison criteria (computational complexity, real-time applicability, adaptability, scalability) are grouped by relevance to method selection. Energy, economic, grid-interaction, reliability and sustainability indicators are considered as performance-assessment dimensions. The assessment indicates that no single method satisfies all performance requirements; the reviewed literature consistently indicates that rapid power fluctuations are best allocated to supercapacitors and flywheels, short- to medium-term balancing to batteries, and longer-duration requirements to thermal and hydrogen-based storage. Rule-based approaches offer computational efficiency and straightforward implementation but limited adaptability. Mathematical programming enables constraint handling and multi-objective scheduling but depends on model accuracy and computational capability, whereas metaheuristic optimization offers derivative-free search suited to non-convex, large-scale problems without solution-quality guarantees. Model predictive control supports forecast-based, multi-timescale coordination, whereas artificial intelligence-based methods offer adaptive control under nonlinear conditions but require data and careful validation. Hybrid and hierarchical strategies appear most promising for coordinating storage technologies across time scales. This comparative framework links storage characteristics, control objectives and performance metrics to guide EMS method selection for future buildings.