Heat pump water heaters are one of the most efficient ways to electrify a home, but most of them run on the same simple logic as a basic thermostat. They reheat whenever the tank drops below a set temperature, regardless of whether that happens at 3 a.m. or during the most expensive, carbon intensive hour of the evening. New research presented at the 2026 ACEEE Summer Study on Energy Efficiency in Buildings tested a smarter approach in real homes, and found it can meaningfully cut costs and emissions, but only where the equipment has room to work.
Researchers from UC Davis developed a cloud based supervisory control system that layers on top of a heat pump water heater’s existing controls through its manufacturer app, requiring no new hardware. The system uses model predictive control, meaning it forecasts electricity prices, grid emissions, weather, and expected hot water use, then decides in advance when to heat the tank so it avoids the utility’s costliest hours. Comfort was treated as the top priority throughout, with the system designed to sacrifice cost or emissions savings rather than risk running a household out of hot water.
In lab testing, the system cut utility costs by about 15%, reduced marginal emissions by up to 61%, and cut runtime during the 4 to 9 p.m. peak period by about 79%, compared with a water heater’s standard factory controls.
The real test came in a field demonstration with 17 low-income households across two multifamily properties in Northern California, one in San Jose and one in Woodland. Many residents were formerly unhoused, agricultural workers, or multigenerational families, communities that are often left out of early smart technology deployments. Researchers used a randomized testing design, switching each unit between standard controls and the smart system every few days, so weather and household behavior would not skew the results.
The outcomes varied sharply between the two sites. At the Woodland property, the system cut peak period energy use by 31%, a statistically significant result. At the San Jose property, savings were smaller, about 20%, and not statistically significant. Researchers traced the difference to installation conditions. The San Jose units sat in enclosed interior closets with limited airflow and served higher hot water demand, which pushed the system to lean more on inefficient electric resistance backup heat to protect comfort. Emissions dropped modestly at both sites, by 9% in Woodland and 3% in San Jose. Water heater efficiency held steady overall, with average performance nearly identical to the standard controls.