It’s Hard to Undersize a Heating System
One of the biggest fears of HVAC contractors is to undersize a heating system. I’m not saying it’s impossible, but installing a heating system that won’t keep a house warm is really hard to do. Sometimes the rules of thumb that contractors use assume some pretty extreme conditions. You know, like that the folks in the house are going to set the thermostat at 90 °F (32 °C), throw the windows open, and sit around in their underwear expecting to be warm and cozy. Well, let me show you a couple of real-world examples to prove my point.
Savannah home retrofit
We have a client in Savannah, Georgia who came to us for help with system sizing because she wanted to replace her furnace and air conditioner with a heat pump. Here’s what she started with:
- 2,600 square foot house
- 29 °F (-2 °C) winter design temperature (99%)
- Gas furnace: 110,000 BTU/hr input, 89,000 BTU/hr output
That’s our reference point. Next, we did a Manual J heating and cooling load calculation. For heating, the result was:
- 60,000 BTU/hr
That tells us the 89,000 BTU/hr furnace was about 50 percent larger than it needed to be.
But wait! It gets worse.
After calculating the heating load, we helped the homeowner measure the runtime of her furnace last winter. By looking at how many minutes per hour the furnace ran during those times when the outdoor temperature was close to the design temperature of 29 °F, we estimated the actual amount of heat needed to be:
- 46,000 BTU/hr
OK, so now it seems her furnace was twice as big as it needed to be. Holy shiitake!
But wait! It gets worse…again.
The new heat pump
Last fall she had a new 2-stage heat pump installed. It runs at 24,000 BTU/hr on low and at 37,000 BTU/hr on high. We’re at the end of winter now, and we have some runtime data for the new heat pump. Turns out, it ran mostly on its low stage when the temperature was below 30 °F. And it never hit the auxiliary heat at all (except for a minor thermostat glitch one time).
So, let’s generously say that with mostly low stage and a bit of high stage operation, the heat pump kept the house warm with about 30,000 BTU/hr of heat. Now, the old furnace looks to be oversized by a factor of 3. Here’s a summary:
- 89,000 BTU/hr furnace
- 60,000 BTU/hr Manual J heating load — Furnace 1.5x oversized
- 46,000 BTU/hr furnace runtime estimate — Furnace 2x oversized
- 30,000 BTU/hr new heat pump runtime estimate — Furnace 3x oversized
Wow! I know we’re talking about Savannah here, but it sure seems like it would take a lot to undersize a heating system.
Why was the Manual J so wrong?
Here are three reasons why even the Manual J seems to be giving us a load that’s twice as much as the heat pump was providing.
- The homeowner keeps the thermostat set to 67 °F (20 °C) in winter, whereas we used 70 °F (21 °C) in the load calculation. If she kept the house warmer, the heat pump most likely would have used more stage 2 heat.
- We assumed the furnace and heat pump were operating at their rated capacities. If the furnace were operating at less than its rated capacity, it would have been sized closer to the heating load.
- Manual J generally inflates heating loads. (It can inflate cooling loads, too, but not as much.) We’ve seen this in cold climates as well as milder climates.
One of the big takeaways here is that you don’t need to pad your load calculations. There’s already ample excess capacity built into Manual J results.
My house in Atlanta
Now, let’s look at a different example. I don’t have data from the furnace that was in the house when we bought it in 2019, but I did an accurate Manual J load calculation before putting in my new Mitsubishi heat pump. The photo below shows two of my three heat pumps. The one on the right is the one we had installed in 2019 to serve the main floor and is the one I’m talking about here. On the left is the new one installed in 2024 for our basement.
The heat pump for the main floor is undersized relative to the Manual J heating load. In fact, its heating capacity is about 30 percent lower than the load. And yet, based on performance, it’s sized perfectly. When our outdoor temperature drops to our design temperature of 23°F (-5 °C), it can keep the indoors at our Manual J setpoint of 70 °F (21 °C).
I’ve written about this heat pump a few times before, so read my article on how it performed when we got an arctic blast of well-below design temperature weather if you want the details.
Reasons your house may not feel warm enough
I’m not saying that people never feel cold in their homes with the heat running. I’m not even saying it’s impossible to undersize a heating system. But here are a few other reasons you may not feel warm enough:
- Building enclosure problems. Too much air leakage and not enough or poorly installed insulation add to the heating load. If you have large holes in the air barrier that weren’t accounted for in the heating system size, that could be the problem. Same with insulation that got disturbed or removed later.
- Mean radiant temperature. You may be getting enough heat to get the air temperature up, but your body loses a lot of heat via radiation. Uninsulated floors, walls, and ceilings can suck heat right out of your body. Go read my Naked People Need Building Science article.
- Failing heating system. It could be that your heating system isn’t working the way it should. Call your HVAC company.
Now, some of you in places like Minneapolis or Ottawa may be doubting this because both examples are from Georgia. Yes, we have milder winters here. But we’ve seen similar results in comparing heating load calculation results to existing heating system sizes in cold climates. I’ll work on putting together a more comprehensive set of data for a future article.
So, is it really hard to undersize a heating system? Based on the sample of two houses shown here, yes. But how far does it really extend? If you’re in the HVAC business and have done Manual J load calculations, let me know in the comments what you’ve found.
Allison A. Bailes III, PhD is a speaker, writer, building science consultant, and the founder of Energy Vanguard in Decatur, Georgia. He has a doctorate in physics and is the author of a bestselling book on building science. He also writes the Energy Vanguard Blog. For more updates, you can follow Allison on LinkedIn and subscribe to Energy Vanguard’s weekly newsletter and YouTube channel.
Related Articles
You Don’t Need a Load Calculation
My Undersized Heat Pump in an Arctic Blast
How to Cheat on a Manual J Load Calculation
Naked People Need Building Science
Comments are welcome and moderated. Your comment will appear below after approval. To control spam, we close comments after one year.
Similar experience here (climate zone 3C). “Undersized” is actually right-sized.
When came the time to switch to a heat pump for space heating, we had lived in our current home for 6 years, leaving me with 5 entire heating seasons worth of utility bills. I thought this would be a great way to check whether the sizing suggested by HVAC installers made sense.
I wanted the system to work for us — not the hypothetical very demanding customer installers fear would call them right back if the house isn’t toasty minutes after adjusting the thermostat.
At the time, we burned methane gas to heat our home and water (we had switched to induction cooking; recommended, btw). I took our summer gas usage and subtracted that from the winter months (plus an extra 20% to account for the longer/hotter winter-time showers of other family members) to have a better idea of how much fuel was going into space heating.
I reckoned that an air-source heat pump which could output as much heat (even if running continuously) would be the bare minimum we needed. Obviously this then gets rounded up when picking the actual equipment.
Result:
Previous gas boiler: an absolutely bonkers 160k BTU/h (input rating; the output would still be 120k+). Initially single-stage but the previous owner added a modulating gas valve.
Current heat pump: 20k BTU/h (rated), 30k BTU/h max; variable-speed; no “auxiliary”/”emergency” heat option (besides a space heater in one bedroom, just in case).
And sure enough, it does what the math said it would. The house has remained comfortable since installation 7-some years ago — even though the system’s max capacity is roughly half what I was told I should get.
I did stack the deck in my favor: that heat pump is a cold-climate model, even though our winter temperatures don’t require that.
What I was interested in was its high energy efficiency (HSPF 12.5), heating capacity over cooling capacity, and how well it’s maintained at lower ambient.
Happy with the result. Thank you Allison Bailes and others for educating me on this topic.
The northeast is mostly hydronic and much lower design temps but it holds true, most boilers are oversized by a factor of 2.5-3X
I’m in Canada, near Ottawa, with heating design temperature of -25C (-13F). My F280 calculation gave a load of just over 22,000 Btu/h, but the contemporary version of PHPP put it at just under 8,400 Btu/h . I installed a 1-ton GSHP. I’d have to scour the data, but I think the longest run time may be 12 hours in ambient conditions colder than design temp. Further, I can’t recall it ever running at any time other than at night in any condition. One take-away is that F-280 is over-estimating significantly. Another is that daytime internal gains and even indirect passive gains are probably what is keeping the indoor temperature above the setpoint.
Here in the Minneapolis / St Paul area the heating loads are slightly higher. I have done preliminary Manual J using WrightSoft that were used to help design my retirement house. The actual Manual JDT is currently in the queue at Energy Vanguard which will allow me to compare models. The 2160 sq ft Pretty Good House has a heating load of around 18,000 BTUh and a cooling load of 8000 BTUh. The typical NG boiler modulates between 7,800 BTUh to 78,000 BTUh so it will be running at 20% on design day when the -23F is out of the operating range of the typical heat pump. Several contractors were trying to sell me on boilers that only modulated down to 13,000 BTUh – they were going in the wrong direction by going with oversized systems and then increasing costs by over sizing buffer tanks to deal with the imbalance. Fortunately I found a contractor who will accept a third party Manual J.
One upside of our builder over sizing the boiler that was installed in our original home was putting on an addition that just about doubled the square footage and we had plenty of available btu’s to heat it.
Every HVAC contractor we contacted before getting a well done manual j (Thanks) wanted to add heating capacity to the existing system. It was only by getting the manual-j and specifications for our requirements independently and seeking quotes were we able to avoid getting an oversized system. Even with all our precautions we still had a bumpy road. Our heating air handler was installed first with assurance that we would be able to match it to the correctly sized heat pump the following year when we wanted to add A/C.
Turns out contractors want to install whatever equipment is readily available at their supply house. What was readily available was up sized and I had to insist they install what was specified. The last thing we wanted was an oversized condenser that was short cycling for A/C.
Here in Michigan, if you use the natural gas bills for the three coldest months and the actual heating degree days for those months you can size a system a lot faster than using Manual J and it eliminates things like air infiltration which is always a guess.
Wayne, what is your methodology?
I’ve heard anecdotal evidence of what’s wrong. I know someone who claimed her heat pump isn’t working. No, she just won’t get comfortable until it’s 80 degrees; the heat pump does fine at 70. I told her to point an infrared space heater at herself or sleep with an electric blanket. What about design temperature versus apocalyptic temperature? In the Knoxville metro, design is 15. We saw -2 a few years ago, but the record is -24. What if it happens again? To that I say: A properly sized system will still heat the house enough for safety. Use extra clothes, warm drinks, and personal heaters to keep warm. Most people aren’t going to go outside and fix their building envelope in winter.
For anyone who heats by oil, gas or propane, a couple years’ historic fuel records together with daily weather records from a nearby location provides enough info to reverse-engineer a heating load calculation. Fuel used for water heating and cooking, if applicable, can be estimated and deducted using summer consumption data. The biggest assumptions in this approach will be the true efficiency of the existing system, and making a reasonable estimate of historic thermostat settings.
I’m also in Ottawa. Having real-world data allowed me to be comfortable in accepting (or rejecting!) furnace capacities proposed by rule-of-thumb heating contractors …
100%
I’m convinced the foundational step in heat pump adoption is actually convincing contractors and homeowners how low the loads actually are. Manufacturers are huge part of the problem (or solution?) as well, fully half of the residential furnaces they make are too big for nearly every application. Think – 80k is always too big! (…that’s OK though because furnaces all need phased out of existence anyway 😉
When a customer has an oversized system, I advise them to open some windows until the system runs longer. That way they match the house to the equipment.
Another factor: it may not be necessary to maintain a constant indoor temperature year-round. In winter, I get used to the temperature being generally cooler, and excessive heating may bring RH below comfortable levels (nose bleeds, anyone?). Moderate overnight temperature drops are probably healthy too. In summer, the cooling effects of A/C dehumidification and moving air means you may not need much of a temperature drop to achieve comfort.
Enjoying the comments. Can anyone share the formula or a link that would allow determining the heating load from past oil and electricity usage. We burn oil and have a heat pump and we keep close track of our oil and electricity consumption and I would be curious to see how our manual-J compares to our actual historic heat load. Thanks
I appreciate the Ottawa shout out, and look forward to your future article with a local Ottawa project!
Allison – how often in your sizing routines are you also trying to ensure 350-400 CFM per ton of airflow? And for low load homes of larger volumes – are you ever bringing in supplemental air handlers to ensure sufficient airflow? I’m not sure if Manual J also looks at airflow in addition to the heating and cooling loads.