Purdue Field Test Keeps a Fully Electrified Home Under 100 Amps for 31 Winter Days
Dale Resnick
A 30-year veteran of residential HVAC who's crawled through more attics than he can count. Dale writes the 'Duct Tape & Beyond' column and believes every compressor tells a story if you listen close enough.
A control system built at Purdue held an all-electric 1920s house inside the safe range of a 100 amp panel for 31 straight winter days, through a cold snap that bottomed out at -4 °F. It managed that by adjusting the setpoints of exactly two appliances: the heat pump and the water heater. Under electrical codes, the paper notes, the same house nominally requires 200 A.
Panel capacity kills more heat pump retrofits than payback math does.
A version note, because the two versions are not equally easy to get. The journal article ran in Applied Energy, volume 386, article 125528, May 2025, and it is paywalled. Everything below comes from the arXiv preprint, version 2, revised February 5, 2025, which still carries the line "Preprint submitted to Elsevier." The authors are Elias N. Pergantis, Levi D. Reyes Premer, Alex H. Lee, Priyadarshan, Haotian Liu, Eckhard A. Groll, Davide Ziviani and Kevin J. Kircher, all at Purdue's Center for High Performance Buildings.
Test site: a 1920s, 2,240-square-foot detached single-family house in West Lafayette, Indiana, climate zone 5A, renovated to code-minimum insulation. Air-to-air heat pump with resistance backup, resistance water heater, everything else electric. Three people lived there through the test. The vehicle was a plug-in hybrid with an 8.8 kWh battery on Level I charging, pulling 1.8 kW.
Data collection ran December 25, 2023 through January 25, 2024, against a baseline from December 2022 and January 2023, before the controller existed. Mean outdoor temperature was about 27 °F during the test and about 34 °F during the baseline. The harder month was the one with the controller running.
What a 100 Amp Panel Ceiling Looked Like in the Data
At baseline the house often sat in the 80 to 100 A band and sometimes pushed past 115 A. One spike hit 120 A when a defrost cycle lined up with the water heater during a large hot water draw.
With the controller running, whole-home current crossed 90 A 0.61 times per day, against 2.2 times per day at baseline. It went above 100 A exactly once, reaching 104 A for a single five-minute interval, and never got past 105 A. That excursion traced back to a Wi-Fi delay that left the water heater energized through a defrost cycle, a failure mode the authors say needs work.
Comfort held. Indoor temperature averaged 68.4 °F over the baseline and 67.6 °F over the test. Supply water averaged 118.0 °F and 117.7 °F. Hot water dipped under the 98 °F comfort threshold four times, four minutes of lukewarm water total, all of it back-to-back showers colliding with the controller shutting the water heater down. Starting January 15 the team preheated the tank to 130 °F twice a day.
Architecture is two layers. A high-level scenario-based model predictive controller plans setpoints over a rolling forecast horizon. Below it, a rule-based layer watches real-time current, sheds devices in stages, and predicts defrost-driven resistance heat by monitoring supply air temperature. All of it ran through the manufacturers' standard communicating-thermostat APIs, with no factory-level hardware access and no changes to device control logic.
The second-vehicle result is simulation, not measurement. On a modeled day averaging 5 °F, an added EV with a 70 kWh battery charging at 48 A would have pushed the house over 150 A several times without optimization. With optimization, the model kept it under 100 A by interleaving charging with heat pump operation.
What This Means for a Retrofit Bid
Nothing here lets you tell a customer to skip the panel upgrade. The paper is a proof of concept in one house, with one heat pump, in one climate zone, and the authors say so. They list electrical code compliance as open work, and they hedge the abstract's own conclusion with the words "if codes permit." A load calculation and your authority having jurisdiction still decide what gets installed.
What changes is the conversation. A 2023 Department of Energy report led by Lawrence Berkeley National Laboratory with NREL, cited in the paper, puts about 21% of US homes at 100 A or less of panel capacity, and about 44% at two or fewer open slots. Los Angeles Department of Water and Power figures, also cited there, put roughly 46% of single-family homes in disadvantaged LA communities at 100 A or below. Those customers are staring at a $2,000 to $10,000 adder on top of the equipment. Some of them are about to start asking whether controls can do that job instead.
Our read is that the near-term money isn't in model predictive control, but in the cheaper cousins already on the market: 120V heat pump water heaters, breaker-level load management devices, and plug-in heat pumps like the $3,800 unit Merino launched this spring. Field-proven MPC has shown up in commercial work before, including a dual-fuel system that cut utility costs 27%. But the residential path runs through thermostat vendors, not contractors writing control code.
A heat pump calling resistance backup on the way out of defrost is the biggest current spike in an all-electric house, and predicting it is where this controller spent most of its effort. That squares with what cold-climate field studies keep finding about where the electricity actually goes below freezing.
Testing in a bigger, leakier house is on the authors' list.
Sources
Pergantis, E.N., Reyes Premer, L.D., Lee, A.H., Priyadarshan, Liu, H., Groll, E.A., Ziviani, D., Kircher, K.J. (2025). "Protecting residential electrical panels and service through model predictive control: A field study." Applied Energy, 386, 125528. doi:10.1016/j.apenergy.2025.125528
Pergantis, E.N., Reyes Premer, L.D., Lee, A.H., Priyadarshan, Liu, H., Groll, E.A., Ziviani, D., Kircher, K.J. (2025). Preprint of the above: arXiv:2409.04884v2, revised 5 February 2025. arXiv:2409.04884
Ham, S.W., Paul, L., Kim, D., Pritoni, M., Brown, R., & Feng, J. (2024). "Decarbonization of heat pump dual fuel systems using a practical model predictive control: Field demonstration in a small commercial building." Applied Energy, 361, 122237. doi.org/10.1016/j.apenergy.2024.122237
ACHR News. (2026). "Merino Energy Introduces $3,800 Heat Pump, Installed in an Hour or Less." achrnews.com