Unlike natural gas heating equipment, it’s vital that heat pumps be accurately sized. New calculation methods take advantage of data we already have to make heat pump sizing more accurate, faster, and cheaper.
About six years ago I replaced my home’s natural gas furnace. The contractor proposed a new unit with the same energy input as my old one. I did a calculation based on my utility bills (using a technique explained below) and found that a smaller furnace, with one-sixth the input, would keep my house warm. I asked my contractor for that size, but the smallest one available was only one-third the input. I agreed to have it installed. Not only does it keep my house warm during winter’s coldest hours, but it’s capable of warming it up from an overnight setback temperature in just a few hours.
The oversizing I discovered seems to be relatively common, at least judging from reports from contractors and researchers. There are several factors driving contractors towards oversizing furnaces. They don’t want to get any complaints about about cold houses, larger furnaces don’t cost a whole lot more than smaller furnaces, and customers like how quickly large furnaces bring houses back to temperature after an overnight setback.
There are two major drawbacks to installing larger than necessary furnaces: they often operate a bit less efficiently than accurately sized ones, and their more frequent cycling on and off can cause components to fail prematurely. But installing contractors aren’t paying the utility bills, few customers actually know the efficiency at which their furnaces are operating, and any equipment failures due to oversizing will come well after any warranty is over. It’s no wonder that the standard calculations contractors use to size furnaces are biased towards oversized units.
While it’s not great practice to oversize gas furnaces, it’s even worse for heat pumps. Larger heat pumps cost significantly more than smaller ones and oversized units can run much less efficiently. How can contractors and homeowners ensure that heat pumps are accurately sized? Although recent advancements in technology and software packages enable contractors to more quickly and rigorously perform the standard calculations, it’s not clear these innovations are producing more accurately sized equipment. The good news is that new calculation methods are emerging that use data freely available in existing buildings, from utility bills and smart thermostats, and have the potential to be not only faster and cheaper than the standard calculations, but also produce more accurate results.
These new data-based calculations could provide an important boost to the drive to electrify space heating, but there are imposing barriers holding back their widespread use. Although a few small studies support their benefits and accuracy, no independent laboratory has published a comprehensive study. Then again, no such study exists to verify the accuracy of the standard calculations either. Also, no national organizations publish standards for data-based calculations, and those standards are needed for them to be incorporated into building codes, as the standard calculations currently are.
Currently, several highly regarded research organizations are studying data-based heat pump sizing calculations. If those organizations ultimately publish results that are consistent with the smaller studies already available, it’s likely that use of these methods will become far more widespread.
How everybody is supposed to size heating equipment now
To size a heat pump, one must first determine a home’s peak heating and cooling loads, or in other words, the amount of energy it takes to keep it comfortable during the coldest and hottest hours of the year. The standard method heating and cooling contractors use was established in 1961 as the Air Conditioning Contractors of America’s Manual J. It’s widely accepted by building codes and incorporated into numerous software applications. It can be applied to new construction or retrofits, and it’s the only rigorous method available for the former. But even after all these years of development and application, there are still numerous problems associated with Manual J.
To apply Manual J software, users must enter the dimensions of all external surfaces, including walls and windows, the building materials incorporated into those surfaces, the heat resistance of any insulation and other building materials, and any measurements or assumptions about air leakage rates. Early in my career, well before there was Manual J software, I did many such calculations by hand, and it was tedious and time consuming. Although using computer software makes this work much easier, just collecting all the information required is sufficiently tedious and time consuming that contractors are notoriously adverse to doing so.
Even with great attention to detail, there are many ways that inaccuracies can creep in. For example, surface measurements may be off, within walls there can be missing insulation, or additional structural elements, like wooden boards with low thermal resistance, and it’s difficult to estimate what air leakage rates will be. These problems may be exacerbated by software products take a lot of shortcuts, such as assuming average amounts of window area, and that building materials are identical to those required by code at the time of construction.
No one has ever done a comprehensive study to determine how accurate Manual J calculations typically are, but my experience lines up with that of Nate Adams, a contractor who does a lot of post-installation monitoring on his projects. He wrote in a recent blog post that “I've consistently found that the US industry standard Manual J load calculation comes up with numbers that are around double what's actually needed, especially if you don't know how leaky the house is.”
The ACCA is aware of these problems and is doing research to improve accuracy and consistency. Also, there are several new software products emerging that have the potential to help. For example, one named Conduit Tech uses pulsed laser light to measure surfaces in existing buildings, and produces three dimensional building models and augmented reality visualizations. Another uses AI to scan building plans to calculate wall and window dimensions, as well as orientations.
While these innovations have driven down the time it takes to do a Manual J calculation from hours to minutes, subscribing to these services costs more than many contractors want to pay, and it’s unknown how accurate the heating and cooling loads they produce are.
Equipment observation is cheaper, faster, and accurate
Yogi Berra once said “You can observe a lot by watching.” The same applies to heating and cooling equipment. You want to know how well matched your furnace is to your house? On an extremely cold night, when it’s nearly as cold as it ever gets where you live, go down to the furnace room with a stopwatch for a few hours. Keep track of how often your furnace turns on and off (This technique is only applicable to single-stage furnaces, which don’t have variable fan speeds or burners).
If the furnace runs constantly, and your house temperature doesn’t drop, it’s perfectly sized. If it runs only half the time, a furnace half the size would get the job done. In general, multiply the portion of time the furnace runs by its input, then multiply by its efficiency, and you’ll know how much heating energy you’ll need to get out of any replacement unit. For your air conditioner, spend a few hours with it on the hottest day of the year, and perform a similar calculation.
Let’s say you’re busy, and you don’t have a few hours to spend hanging out in the furnace room, or you planned on doing it, but never got around to it, and now your furnace failed. What can you do? If you’ve got a smart thermostat, like an Ecobee or Google Nest, you can download your historical data, and take a look at how much your furnace operated during the coldest and hottest hours of the year.
I’ve got an Ecobee thermostat, so I can access 15 months of historical data. Within that time period, I can see, for every 5 minutes, my thermostat set temperature, my measured indoor temperature, if my thermostat called for heat, and the outdoor temperature. I can replicate sitting in the furnace room with a stop watch by selecting the lowest temperature hours, and adding up what portion of those hours my furnace ran.
By definition, equipment observation is the most accurate means of determining heating and cooling loads, because it measures exactly what we want to know: under extreme conditions what size heating equipment do we need to maintain comfort? Don’t fret though, if you don’t have a single stage furnace or a smart thermostat. There is still another method that makes use of data you already have.
Utility bills are good for more than costing you money
If you can get your hands on about a year’s worth of utility bills, you can figure out your peak heating load. First, make a table in a spreadsheet application, like Microsoft Excel or Google Sheets, that lists every month’s natural gas consumption and the corresponding average outdoor temperature. You should be able to find both of those quantities on your bill.
Look at the non-heating months, like July and August, to determine what your non-heating energy consumption is, and subtract that amount from all the heating months’ energy consumption. Then use the regression function in your spreadsheet application to determine the relationship between heating energy consumption and average outdoor temperature.
Next, look up your outdoor heating system design temperature in the American Society of Heating, Refrigerating and Air-Conditioning Engineers Standard 169-2025 Climatic Data for Building Design Standards. ASHRAE defines this value as the ambient temperature that a specific location stays above for 99% of the hours in a year, based on a 30-year historical average (In a nod to climate change, ASHRAE now places more weight on the last ten years). ASHRAE charges $105 for Standard 169, so here’s a US EPA publication that will give you 2017 data for free.
Plug the design outdoor temperature into the relationship developed above to calculate the peak heating load your heat pump has to be able to meet. Confused? No big surprise. This is complicated, at least at first. For all the details on how to process utility bill data, see this presentation by Max Reichlin, a Lead Product Manager with Clean Power Research. On that page you’ll even find a link to an example worksheet. There’s also a link to, and information about, a product that Max’s company put out for awhile that did all these calculations. Unfortunately, that product is no longer available.
This calculation method is largely limited to homes heated solely by either natural gas or electric resistance. If a home is heated by some combination of natural gas and electricity, it can be difficult to parse out what portion of each to allocate to space heating. It can be used to estimate cooling load, but since air conditioner effectiveness varies with outdoor temperature, that adds an additional wrinkle to the calculations. It also can’t be used if the home was unoccupied during the time period the utility bills were collected. There’s also one more obstacle to applying both thermostat and utility bill data, and it’s a biggie.
Building codes can be a problem
To replace an air conditioner or furnace with a heat pump, you’ll need a building permit, and in the case of a planned replacement, most building codes require a Manual J calculation. (For emergency replacements, most building departments won’t make this demand.) The actual language of the 2024 International Residential Code specifies that “Heating and cooling equipment and appliances shall be sized…based on building loads calculated in accordance with ACCA Manual J or other approved heating and cooling calculation methodologies.”
Might calculations based on thermostat or utility bill data qualify under the “other approved methodologies” clause? I asked Brad Smith, a building code official at Fort Collins, CO, if he’d accept such calculations, and he told me he hadn’t considered it yet. He was quick to add, “I’m curious if you’ve established some type of framework for this.” And here’s the rub.
In building codes, the word approved means “Acceptable to the building code official.” Brad’s an innovative code official, and he might well accept data-based calculations if he were persuaded they were well founded. Other officials I’ve dealt with aren’t so open minded. They wouldn’t accept any calculation that wasn’t done according to a national standard published by a group like the ASHRAE.
What would it take get standards for data-based calculations in place? First, a comprehensive study would have to be done to establish their validity in a wide variety of climates, then a major organization would have to issue a standard enshrining an approved calculation method, and lastly, that standard would have to be incorporated into building codes. As it so happens, this process is underway.
The path to standardization
Although no definitive study of data-based calculations has been done, several smaller ones have been published, and others are on the way. I’ve seen several, both published and unpublished, and their results look similar to this study by Brittany Farrell of Clean Power Research. For 17 homes in the Pacific Northwest, Brittany calculated heating load using 3 different techniques: Manual J, regression analysis using historic temperature and metered energy consumption, and heating system energy consumption on the coldest day of the year. She found that for every home, the highest estimate was the one obtained by Manual J, and it was almost always about 2-5 times that obtained by the other methods. The heating loads produced by the other two methods, in nearly every case, were very close, with the heating bill regression method producing estimates that were usually just slightly higher than coldest day energy consumption.
Brittany’s study isn’t sufficient to completely validate utility bill analysis. After all, she only analyzed data from 17 homes, and all of them were in a single region. A definitive study would analyze data from far more homes, and in a variety of climate types. A group of researchers is currently working to produce such a study.
Heather McDiarmid, a consultant specializing in residential heat pumps and energy efficiency, is teaming up with Natural Resources Canada to produce a study similar to Brittany’s, but including many more homes from a variety of climates. Currently she’s collecting residential datasets that include a year’s worth of hourly energy consumption, home energy audits, heating and cooling equipment specifications and runtime, and other information needed to perform a Manual J calculation. Such datasets are hard to come by, so if you’re aware of where to find one, please contact Heather.
Not only do Heather and NRCan hope to validate data-based equipment sizing calculations, but also, to get the CSA Group to recognize them. If the CSA does so, it’s likely that similar US organizations like ASHRAE, and the International Code Council will follow suit.
Before you try this at home
If you intend to use the calculation methods described above to size heat pumps, there are a few additional caveats:
- Before selecting any heat pump for an existing building, check to ensure that the existing electric panel has enough capacity, and that the ductwork can accommodate its airflow. If not, there are ways to expand the capacity of both, but they’re not always feasible, and even if they are, they can be impractically expensive.
- If you’re sizing a heat pump for the heating load, and that load is much bigger than the cooling load, consider installing a dehumidifier or a reheat coil. In a humid climate, oversized cooling equipment can lead to inadequate dehumidification, which can result in mold growth and structural problems.
- Some analysts regress heating energy consumption against degree days. I prefer to use average outdoor temperature. That data is easier to come by, as it’s usually included on utility bills, and also, it’s likely to yield more accurate results.
- If you’re using utility bill data, there is a bit of noise in that data, including temperature setbacks, internal and external gains, and unoccupied periods. To account for these factors, add a small safety adjustment to your calculated load. There isn’t any formal algorithm yet for determining what this factor should be, but some folks I know bump the heat pump to the next size up.
- Once heating and cooling loads are estimated, there’s more to equipment sizing, and that work is detailed in ACCA Manual S, or in Canada, this Toolkit for air source heat pump sizing and selection.
Now you’re ready to use data to size your next heat pump.
And now a word from our sponsor
Want to keep up with the wacky world of energy efficiency technology? You’ll want to subscribe to Energy Technology Revolution. Every time we put out a new post, we’ll send you an email. Don’t worry about us filling up your inbox. We don’t send out that many messages. To subscribe just click here.
Lastly, check out these other recent ETR posts on heat pumps and air conditioners: