When we think of energy storage for the electrical grid, we typically picture chemical batteries, pumped hydro, or massive flywheels. However, a massive, ubiquitous, and largely untapped storage medium already exists all around us: the physical structure of our built environment.
Every commercial building, residential home, and industrial cooling plant possesses thermal inertia. When a heating, ventilation, and air conditioning (HVAC) system is throttled or turned off, the indoor temperature does not instantly spike or drop. Instead, it drifts slowly over minutes or hours. This thermal lag behaves exactly like an energy buffer. By intelligently scheduling when we heat or cool spaces, we can transform buildings into "virtual batteries" that absorb excess renewable energy or shed load during grid stress, all without compromising occupant comfort.
The Physics of the Thermal Flywheel
To understand why a building can act as a battery, consider its thermal capacitance. A building is not just a volume of air; it is a complex envelope of concrete walls, steel frames, floors, and furniture.
As reviewed by Ning Zhang et al., this inherent thermal mass acts as a buffer against volatile energy inputs. When an electrical grid experiences a surge in renewable generation, buildings can "over-cool" or "over-heat" their structural mass within safe limits, effectively storing that surplus electricity as thermal energy. Conversely, during peak demand periods, the HVAC systems can be cycled off, allowing the stored thermal energy in the walls and floors to maintain the indoor climate.
To exploit this accurately, control systems must look beyond simple air temperature. Fengji Luo et al. demonstrate the power of a "two-parameter" thermal model that explicitly accounts for both indoor air temperature and the thermal capacitance of structural walls. By treating the physical structure as a storage medium, utilities can predict temperature trajectories during power cycling far more reliably, ensuring that load shedding does not result in rapid comfort degradation.
Mapping Thermal Dynamics to Battery States
How do we translate a thermodynamic system (governed by heat transfer equations) into a resource that a grid operator can dispatch like a battery? The secret lies in changing how we represent operational constraints.
Traditionally, demand response treats appliances as binary switches with instantaneous power limits. However, Altuğ Bitlislioğlu et al. show that modeling thermal systems using integral constraints unlocks significantly higher flexibility. Because thermal mass can average out power adjustments over time, an integral constraint can represent the cumulative energy capacity of a building over a long horizon. This is mathematically equivalent to a battery's State-of-Charge (SoC).
[Physical Battery] ---> State-of-Charge (SoC) ---> Energy Capacity (kWh)
[Virtual Battery] ---> Temperature Deadband ---> Thermal Capacity (kWh)
This abstraction allows diverse thermal systems to speak the same language as electrochemical storage. For instance, Manu Lahariya et al. utilize grey-box modeling to map industrial evaporative cooling systems directly to virtual battery metrics, defining their capacity through an operational state-of-charge and rate-of-charge. Similarly, Bingqing Chen et al. abstract complex thermodynamic behaviors into linear storage models, allowing building-level controllers to solve simple quadratic programs to manage grid response.
Exploiting the Elasticity of Human Comfort
Physical batteries have rigid voltage and capacity limits; pushing past them damages the cell. Virtual thermal batteries, however, possess a unique advantage: their boundaries are defined by human perception, which is inherently elastic.
Zhanbo Xu introduces the concept of "soft comfort," demonstrating that thermal comfort is better modeled as a flexible range rather than a static setpoint. When energy prices fluctuate, the acceptable temperature boundaries can dynamically expand or contract. If electricity prices spike, the system allows the indoor temperature to gradually "drift" toward the edge of the comfort zone, drastically reducing HVAC power draw.
By treating human preferences as an elastic resource, buildings can absorb massive grid fluctuations. Di Xie et al. leverage this by using Lyapunov optimization on "virtual temperature queues," allowing indoor temperatures to float dynamically based on real-time smart grid pricing without requiring complex future weather predictions. When paired with local physical assets like solar arrays or chemical batteries, this virtual storage provides a highly effective, zero-marginal-cost layer of added capacity, as shown by Qing Yang et al..
Real-World Validation: Sub-Second Grid Balancing
Can virtual thermal storage actually match the performance of fast-acting chemical batteries? Empirical evidence shows it can, even at the demanding timescales required for grid frequency regulation.
At the macro level, Wenting Wei et al. prove that hierarchical and distributed control strategies can aggregate thousands of thermostatically controlled appliances into a single Virtual Power Plant (VPP), smoothing out volatile renewable energy injections with sub-second processing times.
In the field, researchers have successfully qualified commercial buildings for strict ancillary service markets. Tomasz T. Gorecki et al. demonstrated this capability within the Swiss electricity market, proving that HVAC systems could sustain frequency regulation over extended periods by leveraging building mass. Following this, Luca Fabietti et al. conducted real-world experiments showing that commercial heating systems could comfortably commit between 25% and 43% of their total installed capacity to high-precision grid services. Because the building mass dampens the rapid cycling of the heating units, the indoor temperature remains stable within a narrow fraction of a degree.
By shifting our perspective from controlling loads to managing virtual storage, this framework turns everyday infrastructure into a critical tool for stabilizing a low-carbon grid.