Heat Pump Water Heater Insights: a line drawing of a heat pump water heater and its piping, over a building floor plan.

HPWH Insights · Issue 2: Oversizing the primary heaters is a design flaw, not a margin


From AbGarm Energy.

Good morning.

This issue is about the parts of a central heat pump plant that tend to be sized last: the storage, the temperature maintenance and the backup. Two field studies were selected to show what happens when they are an afterthought. Our case, for design engineers and energy consultants, is that an oversized primary heater is a design flaw, not a safety margin. And storage deserves the same “love” as the primary heat pumps.


Oversizing the primary heaters is a design flaw, not a margin

A central heat pump water heating plant can be divided into four pieces working as one: primary heat pumps, primary storage, temperature maintenance heater/storage and backup. If sized one at a time, each piece can look reasonable while the whole system falls short.

In two field studies, as we read them, the pieces treated as afterthoughts decided how the plant ran.

Where the electricity went: A field demonstration published by CalNEXT monitored packaged CO2 systems in two new multifamily buildings in Menlo Park, California. Each monitored system serves roughly 56 residences. The plant design used a swing tank, a dedicated electric resistance tank piped in series after the primary storage, to maintain the hot water distribution loop temperature.

The project reports the swing tank energy use at 46 to 68 percent of total system energy, “far exceeding a reasonable target of around 10 to 15 percent.” The loop itself was not the problem. Its loads were “within expectations,” with “a very reasonable seven-degree difference” between hot water supply and recirculation return temperatures.

The swing tank's share of total system energy One monitored packaged CO2 system per building, Menlo Park, California. The swing tank setpoints are the report's estimates, and the target is the report's own. CalNEXT report, swing tank energy fraction Building 1, half occupancy, high setpoint Building 1, half occupancy, high setpoint: 68% 68% Building 1, half occupancy, low setpoint Building 1, half occupancy, low setpoint: 62% 62% Building 2, full occupancy, high setpoint Building 2, full occupancy, high setpoint: 46% 46% The report's target The report's target: 10 to 15 percent 10 to 15 percent Source: AbGarm Energy, after CalNEXT (2025)
The swing tank's share of total system energy in two packaged CO2 systems: 68%, 62% and 46% across three operating regimes, against the report's target of 10 to 15 percent. Source: AbGarm Energy, after CalNEXT (2025).

Why would a tank meant only to hold the loop’s temperature use that much? The report says it “can be exacerbated by certain commissioning and design decisions such as high swing tank setpoints and poor recirculation return water balancing.”

Blaming the operator and installers is the same theme as in issue 1, whose report blamed the setting rather than the design.

The initial design was supposed to send only a small part of the recirculation return flow to the swing tank, through a balancing valve. The final skid design left the valve out. A ball valve stood in, fully open, “allowing unrestricted return water flow back to the swing tank,” the report indicates.

The swing tank’s controls were also reported “rudimentary and low-resolution,” so lowering the setpoint took the manufacturer’s own visit. The reasoning does not make much sense! If the swing tank’s controls were the problem, a simple temperature reading at the outlet of the tank would be enough to reset the tank temperature to a lower setpoint.

Furthermore, using figures annualized from regressions, with no winter data, the report puts Building 2’s plant coefficient of performance at 2.3, against a rated 3.2. It calls that “an example of the energy penalty that can be expected if a swing tank is set too high.”

When the backup is a gas heater: A paper in the ACEEE Summer Study proceedings describes another swing tank design in an office building at Oak Ridge National Laboratory in Tennessee.

Two air-source heat pump water heaters with a capacity of 90 kBtu/hr keep one gas tank, the primary storage, hot without firing it. A second 200 kBtu/hr gas water heater, “rated at 95% efficient,” takes the recirculation return, doing the swing tank’s job.

The high rated efficiency of the gas swing tank indicates that it is a condensing water heater. But if the designers had looked at the whole system, the high expected temperature of the recirculation return could have prevented the “mistake” of selecting a condensing gas water heater as a swing tank.

What happens to the heat pumps’ share when the swing tank burns fossil gas? In all four sampled periods it charts, the plant’s fossil gas energy use was, by our reading, more than the electricity used by the primary electric HPWHs. Gas added more heat than the heat pumps, even under “HP Dominated Control.”

The gas figures are inferred: the paper back-calculated gas from the energy balance and assumed the 95% efficiency value, with 85% and 75% run as a sensitivity study of the cost savings. The higher gas efficiencies are most probably a wrong assumption, and the gas usage can be even more than reported.

The reason the authors give for the plant arrangement is that it was “the recommended configuration by the manufacturer upon start-up of the system.” Their cost comparison uses Oak Ridge rates, where electricity “costs about 3.33 times more per kWh than natural gas.”

The heat pumps “often short cycled, due to minimal storage volume, to a 15 minute run.” With longer time-outs, “a single HPWH could meet the heating demand under tested conditions.” This clearly shows oversizing of the primary heaters compared to the primary storage.

The next steps the authors summarized call for smaller heat pumps and more storage, which is one of the arguments in this issue.

In both plants, as we read them, the trouble sat in pieces designed apart from each other. A swing tank was set and plumbed alone, gas covered whatever the heat pumps left, and too little storage sat behind them.

A generous heater margin is not a safety margin when the storage and temperature maintenance behind it were never sized with it.

A generous heater margin is not a safety margin when the storage and temperature maintenance behind it were never sized with it.

For a facility manager, the point is plainer. A heat pump plant is not a one-for-one swap for a gas water heater. It needs that whole-plant analysis before anyone picks equipment.

A range of designs, not one: What should a designer hand the energy manager, then? Several sized options, each pairing storage, capacity and temperature maintenance differently, because the right one depends on the building. Even the CalNEXT report suggests swing tank systems “may be better suited for buildings that have a high probability of full occupancy.”

Code should also allow that variation rather than force one design on every building.


Capacity and storage are one decision, not two

How much heat pump does a building need? Ecosizer, a widely used free design tool, has been trying to answer this question.

Ecotope’s manual for its Ecosizer sizing tool, revised September 2026, answers with a trade, not a number. Capacity follows from a maximum daily runtime, 16 hours by default, and sweeping that runtime traces a curve of storage against capacity.

The September 2026 revision covers swing tank, parallel loop, and single-pass and multi-pass return-to-primary plants. The tool is going through constant changes, and this revision prints no change log.

The manual’s May 2025 revision did not cover return to primary. Yet it called the single-pass kind “the most efficient” for refrigerants such as R290.

The May 2025 revision also set a design demand of 25 gpdpp (gallons per day per person), the 98th percentile of daily demand measured in Seattle. It called ASHRAE Medium, at 49 gpdpp, “likely an overestimate of any modern multifamily building.”

The September 2026 revision prints no such figure, and takes multifamily demand per person as a user input. That increases the uncertainty of the results.

One building, two accounts: Only the May 2025 revision carries a case study, Elizabeth James House, “a 60-unit low-income multifamily building” in Seattle. It reports “a 63% reduction in energy” and 20 kW of heat pump output, “one-sixth of the pre-retrofit capacity (120kW)” of the original resistive water heaters.

It also calls that capacity “an order of magnitude lower” than traditional methods specify. This was the only field case the manual shared for its design method.

A field study in the journal Sustainability comes from the same group of authors. It cites the same case-study report with more details. It gives four heat pumps of “15,000 BtuH (4.3 kW)” each, a “nominal capacity of 5 tons.”

The old heaters, of “120 kW design,” are “more than five times the current output capacity.” Neither “one-sixth” nor “more than five times” is really “an order of magnitude lower”!

The savings differ too. The paper prints no percentage saving, only a calculated 58,511 kWh/yr in its savings table. The two accounts disagree on capacity and savings, and we treat neither as settled.

More independent studies are needed to evaluate the effectiveness of this design method.


What the research shows

Across the field studies and design guides we follow, heat pumps invert the conventional recipe of fast recovery on small storage. They want large storage, small capacity and long compressor runtime. Yet the usual methods size different things.

ASHRAE’s method sizes peak and recovery, and leaves out temperature maintenance and load shifting. The Title 24 calculation method models energy for compliance and outputs no equipment sizes. Ecosizer sizes storage, capacity and temperature maintenance together, for swing tank, parallel loop and return-to-primary plants.

The research still disagrees on how to size storage, and leaves open what rule keeps a swing tank from becoming an energy penalty. Across the field evaluations of central HPWH systems, lower-than-designed performance is generally reported. We are questioning the current design methods as a source of low performance in real installations.

None of the current design methods say a plant should be designed as a range of options, or that code should allow one. That is our argument.


THE SPEED READ

  • Chapter 51 of the ASHRAE Handbook, the design reference for service water heating, warns that its demand values “are not intended for use as designers’ sole references for hot-water system sizing purposes.” (ASHRAE, paywalled)
  • With loads estimated from hourly gas meter data at eight Southern California multifamily sites, a gas-program study says the ASHRAE method oversized water heaters at all of them, “with an average oversizing of 97%,” a finding about gas systems. (ETCC report)
  • California’s water heating calculation method lists series (swing), parallel and return-to-primary tank arrangements for central heat pump plants, and says the equipment “is sized to meet the domestic hot water load.” (California Energy Commission)

What to check this week

  • In the trend data, look for short cycling of the primary heat pumps. In the gas-backed plant above, it traced to storage too small for them.
  • Compare the temperature at the mixing valve’s supply outlet with the loop return. A small drop points to an oversized recirculation pump or imbalanced flows.
  • On the drawings and at the skid, confirm the balancing valve that limits return flow to the swing tank, and check the swing tank’s setpoint against its commissioning value.

Previously, in issue 1: without metering, a water heating plant’s faults stay invisible.

In issue 3, we turn to multi-pass operation in split heat pump water heating systems. We look at what stirring the storage tanks does to stratification, outlet temperature and efficiency.

Sources