The case for storing heat instead of electricity

The evening peak that dominates grid planning in cold climates is not purely an electrical problem. Much of it is a heating problem. Space heating, water heating, and commercial thermal loads climb steeply through the late afternoon, peak around dinner, and fall off by late evening. And for a heating problem, the most efficient storage medium is usually not a battery.

The argument turns on thermodynamics.

Why converting heat back to electricity costs you so much

When you charge a lithium cell and discharge it later, the round-trip efficiency is high: most of what went in comes back. When you convert electricity to heat and then convert that heat back to electricity, you are running a heat engine, and heat engines are constrained by the Carnot limit. The fraction of thermal energy you can recover as work depends on the temperature difference you are working with, and practical machines fall far short of the theoretical ceiling. The losses are substantial. Electricity-to-heat-to-electricity is a poor round trip.

But electricity-to-heat-stored-and-used-as-heat is a different calculation entirely. Resistance heating converts electricity to heat with very high efficiency. An insulated tank holds that heat for hours with modest losses. If you want hot water at 7 p.m., and you heat the water at noon when solar generation is cheap and abundant, you have shifted energy across the peak with no battery and no round-trip conversion penalty.

The water heater as a grid asset

Grid-interactive water heaters are the simplest physical instantiation of this idea. The appliance runs on a time-of-use tariff or responds to a utility signal, heating water during midday surplus and sitting quiet during the evening peak. The tank is the storage medium. The round-trip efficiency is far better than any electrochemical system, not because the physics are novel, but because there is no return trip: the stored energy is deployed as heat and the end use is heat.

Scaled up, district heating systems can hold hot water in large, well-insulated tanks or underground pits, accumulating energy over hours and releasing it as needed. Some northern European cities have run systems like this for decades, and the operational records are unremarkable in the best sense: the technology works, it is not exotic, and nothing about it requires a chemistry breakthrough.

Building thermal mass is quieter still. Pre-heating a concrete building slightly before the peak, or pre-cooling it in the morning, stores energy in the structure itself. The building releases that energy slowly, reducing its draw during the constrained hours without any dedicated hardware. Grid programs that enroll buildings for demand flexibility treat this capacity as real and often substantial.

What batteries can do that heat storage cannot

None of this makes batteries optional. Thermal storage is tied to thermal end uses: it does nothing for data centers, EV charging, medical equipment, or the lights. A hot water tank cannot provide frequency regulation or discharge into the grid during a sudden voltage event. Batteries can do all of those things, and their ability to respond in milliseconds and inject power directly into the AC system is genuinely irreplaceable.

The case for thermal storage is narrower and more specific. For the fraction of evening load that is thermally motivated, and that fraction is not small in a cold climate, electrical storage is neither the most efficient path nor typically the cheapest. A storage plan that ignores how much of the peak is really a heat problem will over-specify batteries and under-utilize flexibility that is often cheaper and already partly in place.

The accounting problem

Part of why this gap persists is how storage gets counted. Batteries show up in megawatt-hours, in auctions, in headlines. Thermal flexibility gets classified as demand response and tracked in separate systems by different parts of the utility. The vocabulary of grid planning is battery-shaped, and things that do not fit the vocabulary tend to get undercounted.

A more complete picture of the evening peak, disaggregated by end use, would probably show a meaningful share that can be handled by managed thermal loads at a fraction of the cost of electrochemical storage. The goal is not to choose one approach over the other. It is to use each where it makes sense, and to stop treating the battery as the default answer to every storage question, including the ones where the answer was already sitting in the basement.