So I've been upgrading my homes insulation and building envelope and have plans to replace the furnace with a smaller efficient one. The home has an 80% 100k BTU forced air furnace. I'm tempted to take the risk on going with a 96% 40k BTU Furnace. I know there's calculations that supposedly come up with the proper size, but here's my logic: last years usage in the coldest month was just over 30 mcf's which I calculated into total BTU of 30,855,000. Average hourly usage would then be 42,854. Adjusting for 80% efficiency becomes 34,283. Keep in mind I have a water heater/dryer using gas so my furnace uses less than that. Has anyone done a similar reduction in furnace size, what were the results? What's your thoughts?

Your logic and approach is correct, but I wonder if your numbers are correct. Do you experience short cycling with your existing furnace which is presumably significantly oversized based on your numbers? We have free software on our website for determining the heat loss of an existing building by measuring the heating fuel usage and using the actual HDDs during the measurement period:

Borst Existing Building Energy Usage Analysis Software

Here’s an excerpt from the software instructions that is relevant to what you are doing and also consistent with your logic and approach:

“When replacing an old inefficient furnace with a new efficient furnace, you need to be cognizant about not allowing the rated efficiency improvement to cause over-sizing the new furnace. This is easier to get wrong than you might think. For example, if your old furnace is rated for 50,000 BTU per hour and 60% Average Fuel Utilization Efficiency (AFUE), it is currently providing 30,000 BTU per hour of heat to the building. If “catalog engineering” is the extent of your HVAC knowledge and you happen to replace it with a new furnace rated for 55,000 BTU per hour and 95% AFUE because the next smaller size is less than 50,000 BTU per hour, this new furnace will provide 52,250 BTU per hour of heat to the building. This is almost 75% more heat than the original furnace provided and this over-sizing will most likely cause short cycling resulting in reduced efficiency and reduced furnace life. Assuming 30,000 BTU per hour is truly the maximum design heat load required for this building, you would be much better served by installing a new 95 AFUE furnace have a rating of 31,500 BTU per hour or perhaps even less. This smaller furnace will cost less to install, will cost less to operate, will have a longer life, and will provide the required heat.”

I'll check out the software. As for numbers I used my utility bill of about 30.25 mcf * 1,020,000 btu/mcf (Google) then divided by 30 days and 24 hours. Last winter on one of the coldest nights I remember timing the cycle to being on about 14 minutes and off for 5 minutes. On more normal days of winter it would stay on for something more like 5 minutes and off for 12. I added an additional 12-15 inches of cellulose in my attic on top of about 3-4 of fiberglass. I had maybe 1/3 of my band joists filled with foam board and spray foam. This winter I'll have all band joists filled and sealed as well as basement walls with foam r 10. I'm also building a utility room for furnace and water heater to separate them from the rest of the basement because i have a combustion air intake that just let's cold air poor in. I have no idea what to expect from these improvements

For a detailed description of how fuel use load calculations work (as opposed to just plugging numbers into the Borst calculator), see this bit o' bloggery:

http://www.greenbuildingadvisor.com/blogs/dept/guest-blogs/out-old-new

Since the heating/cooling balance point base temperature is unknown it's best to run it with both 60F and 65F for a base temp, which will nicely bracket the real heat load, pretty much, more or less.

Where this type of calculation fails is when there is a lot of unaccounted for auxilliary fuel use, and sometimes in homes with an unusually large amount of window area. Using only wintertime fuel use rather than annual fuel use tends to be more accurate, since the heating load is larger and more consistent, diluting the error induced by domestic hot water or other uses, and with generally lower solar gains (which balances other-uses for the fuel to some degree.)

The number you're looking for in sizing a furnace is the heat loss rate at the 99th percentile hourly binned outdoor temperature, and NOT an average hourly use over the coldest month. The average outdoor temperature over a cold month is quite a bit warmer than the 99th percentile temperature bin.

The coldest hour of the year is colder than the 99% design temp too, but not by a lot, and not for extended periods of time. If you take the ASHRAE recommendation of sizing at 1.4x the 99% heat load it will usually cover you for the coldest hour of the quarter-century.

Yes, don’t use the 1.4x ASHRAE recommendation. Not sure why you would since that wasn't your approach or suggested. Anyhow, sharpen a pencil or use our software. Per our software directions, you can use degreedays.net to obtain your precise HDD data for the measurement period at your base temperature which can even be tailored at two degree F increments. If you have an electric heat source, it can be isolated with a power meter such as this:

EKM Metering

With a little bit of care and preparation, you should be able to obtain your building performance within a couple percent of reality which is likely an order of magnitude better than an ACCA Manual J8 analysis. However, the energy usage method doesn't provide the room-by-room heat loss data which the Manual J8 analysis does provide and which is needed if you are transitioning from a hot air furnace to HR floor heating.

I'll be honest, either I didn't use the calculator correctly or didn't understand my results. But going with the idea of heat loss I came up with this "method". So this morning I timed my furnace on and off time. My definition of on is from starting up to when the burner stops(not the fans 2 minutes of cool down). With that I had 9 minutes on and 14 off with thermostat set a constant 72 and outside at 24 (48 deg delta). With an output of 80,000 BTU running for 9 minutes that was 12,000 BTU. So since the overall on/off cycle is 23 minutes that means 12,000BTU/ 23 minutes = 521.7 BTU/min heat loss at delta 48 degrees. Now I've read that even though heat loss is not linear its close enough. Using that logic I divided 521.7 by my ^48 degrees = 10.869 BTU/(^deg*min). Finally if I keep my thermostat @ 72 and the outside temp hits 0 I'd have 72*10.869*60(min)=46,954 BTU output needed.

Fortunately, the forecast has a low of 5 coming up so I can test my numbers by clocking my furnace. I don't know if my numbers are higher due to already being over-sized and losing efficiency. I also read elsewhere that its not so bad if your furnace can't keep up a couple days out of the year.

I had a house once that couldn't keep up on the coldest of days. I just lit a couple stove burns and it worked great. Being that the stove is 100 percent efficient is even better.

Posted By Dan84 on 07 Dec 2017 04:01 PM
I'll be honest, either I didn't use the calculator correctly or didn't understand my results. But going with the idea of heat loss I came up with this "method". So this morning I timed my furnace on and off time. My definition of on is from starting up to when the burner stops(not the fans 2 minutes of cool down). With that I had 9 minutes on and 14 off with thermostat set a constant 72 and outside at 24 (48 deg delta). With an output of 80,000 BTU running for 9 minutes that was 12,000 BTU. So since the overall on/off cycle is 23 minutes that means 12,000BTU/ 23 minutes = 521.7 BTU/min heat loss at delta 48 degrees. Now I've read that even though heat loss is not linear its close enough. Using that logic I divided 521.7 by my ^48 degrees = 10.869 BTU/(^deg*min). Finally if I keep my thermostat @ 72 and the outside temp hits 0 I'd have 72*10.869*60(min)=46,954 BTU output needed.

Fortunately, the forecast has a low of 5 coming up so I can test my numbers by clocking my furnace. I don't know if my numbers are higher due to already being over-sized and losing efficiency. I also read elsewhere that its not so bad if your furnace can't keep up a couple days out of the year.

Logging the duty cycle at a known indoor to outdoor temperature difference and projecting it downward to the 99% outside design temperature works fairly well, but the error-bars are large if it's just a one-time measurement: For the short duty cycle measurement period was the sun streaming through the windows? Was it dead calm or was the wind howling?, etc. One-off duty cycle measurements also don't tell you how much of the load was being covered by the internal heating sources, or what the heating/cooling balance temperature is. Using fuel use calculations does a bit of averaging those errors out. Datalogging duty cycle against temperature over many hours or days would probably be necessary to get the error down to under 10%.

Also, is 0F your actual 99th percentile temperature bin suitable for sizing the equipment, or was that just a nice round number pulled out of a hat?

https://articles.extension.org/sites/default/files/7.%20Outdoor_Design_Conditions_508.pdf

ASHRAE's 1.4x multiplier should be used as a maximum oversize factor (not an optimum), but the load at the 99% outside design temperature (and not some randomly chosen number) is rightly viewed as the minimum. Equipment sizing increments don't always fall neatly into that bracket, but multi-stage or modulating equipment can usually get you there, even if a fixed output unit won't/can't.

With bigger, oversized furnaces with oversized air handlers some of that measured heat load is the parasitic air-handler induced air infiltration. Going with a right-sized furnace and a smaller air handler reduces that parasitic load. So if the measured 99% load was 40K, using a 100K furnace as the measuring instrument, the actual load a right-sized furnace would have to support is somewhat lower than 40K, since the parasitic losses will be lower.

Some locations (NY state is one) have building codes that REQUIRE oversizing the equipment by some minimum factor (which is a bad rule to have IMHO), but lenders for new construction generally require that the heating system be able to heat the house to a minimum of 68F at the 99% outside design temperature (which is code in most of the US.)

I'm about 30 minutes south of Flint, which has a 99% @3. I made it lower because I also agree that a one time cycle analysis is hardly perfect. With the exception of wind, I figure the morning is the best since no other interior heat sources have been used and also less sun effect. I do plan to take more readings especially with the colder days ahead. I hope they suggest I can go smaller on BTU since they will be longer cycles. How much do you think parasitic loses typically add up to? I'm sure I have those as 100k btu surly has to blast the air to keep the heat limit switches happy.

If you had blower door testing information along with a duct leakage test it might be possible to estimate the parasitic air handler driven losses, but even that isn't enough information for better than a WAG. Measuring the room-to-room pressure differences in every doored-off room with a manometer would also be useful but still incomplete data. An Energy Star house would have a verified Manual-D duct design, and only fully commissioned when it measures less than 3 pascals pressure difference between rooms, under all conditions (doors open/closed) with the air handler running. In large in-situ studies of systems in California parasitic duct losses averaged something like 15%, but some where as high as 40%. Systems that lousy in MI would probably get noticed and fixed up a bit.

Even a 1-off duty cycle test indicating 47K @ 0F means the 100K furnace is clearly oversized. With enough data points you'll be able to narrow in on it further. But if you have a heating history on the place try running the fuel-use load calculations as a sanity check.

A reasonably tight reasonably insulated 2x4 house without an excessive amount of clear-glass double-panes (or clear storms over wood sashed single pane) at 0F outside, 68F inside would typically come in between 15-20 BTU/hr per square foot of above-grade conditioned space (not counting basement floor area), with plenty of exceptions to prove the rule. With an insulated foundation &/or very-minimal above grade exposure it'll usually be in the 15 BTU/hr per square foot range. If that describes your house, a measured load of 47K @ 0F would imply a 2400-3000' house, depending on the particulars. If it's a 2x6/R19 type house with 25-30 year old low-E double panes and <15% window/floor area ratio the load/area ratio would normally be between 10-15 BTU/hr per square foot of fully conditioned above grade space. Again, there will be many exceptions on both ends of that bracket.

If you had blower door testing information along with a duct leakage test it might be possible to estimate the parasitic air handler driven losses, but even that isn't enough information for better than a WAG. Measuring the room-to-room pressure differences in every doored-off room with a manometer would also be useful but still incomplete data. An Energy Star house would have a verified Manual-D duct design, and only fully commissioned when it measures less than 3 pascals pressure difference between rooms, under all conditions (doors open/closed) with the air handler running. In large in-situ studies of systems in California parasitic duct losses averaged something like 15%, but some where as high as 40%. Systems that lousy in MI would probably get noticed and fixed up a bit.

Even a 1-off duty cycle test indicating 47K @ 0F means the 100K furnace is clearly oversized. With enough data points you'll be able to narrow in on it further. But if you have a heating history on the place try running the fuel-use load calculations as a sanity check.

A reasonably tight reasonably insulated 2x4 house without an excessive amount of clear-glass double-panes (or clear storms over wood sashed single pane) at 0F outside, 68F inside would typically come in between 15-20 BTU/hr per square foot of above-grade conditioned space (not counting basement floor area), with plenty of exceptions to prove the rule. With an insulated foundation &/or very-minimal above grade exposure it'll usually be in the 15 BTU/hr per square foot range. If that describes your house, a measured load of 47K @ 0F would imply a 2400-3000' house, depending on the particulars. If it's a 2x6/R19 type house with 25-30 year old low-E double panes and <15% window/floor area ratio the load/area ratio would normally be between 10-15 BTU/hr per square foot of fully conditioned above grade space. Again, there will be many exceptions on both ends of that bracket.

I have thought about doing the blower door test. Unfortunately my homes heating history has not been consistent. Before we had our first child we'd turn thermostat way down at night and also close venting upstairs. But now we keep it steady temp all day/night and I also decided to fully heat the upstairs since one of those rooms would become our child's room. So last winter is the only historical data I have with highest use being 30.25 Mcf for 30 days for all gas appliances (dryer, water heater, furnace). I have a basement only foundation of 1400sqft with 2300 sqft total (including one room over garage). Built home built in '99 with 2x4 walls. I have 12-15" cellulose over 3-4" fiberglass batts. 75% completed project of sealing/insulating band joists. 50% completed basement wall insulation. I also did my best to seal outlets and lights and wall seems before adding attic insulation. Eventually I plan to replace my carpet flooring and at that time seal the wall footing and am considering attempting to seal the band joists between main and upper floor. So I got room for improvement. The 15-20 range you said seems about right for sure as at 15 comes in at my average usage from my worst month last year and 20 comes in near my 1 cycle estimate. I'd hate to think with all my work that I'm @20 the top of that range. I better finish my basement insulation project and get some more numbers with the upcoming cold spells.

Taking the highest few months fuel use is going to be more accurate than a one-off duty cycle measurement taking place over a couple of hours or less, so your real load is then in the ~35K range(?). Unless you have 5 people taking endless showers the hot water & dryer use error is going to be counterbalanced by the solar gain error. If your family is populated with heavy hot water (ab)user types the fuel use heat load number will actually be higher than the real load.

With more air sealing and finishing the foundation insulation project will probably put it in the ~30K range, which is pretty good for a 2x4 framed house. Goodman makes a condensing 30K-in 2 stager, but that's probably cutting it a bit close, with a high-fire output of ~29K. Their 40K condensing 2-stager puts out ~27K at low fire, ~38.5K at high fire, and probably a good fit, covering your load now, but not too oversized for the "after upgrades" picture:

https://www.acwholesalers.com/hvac/pdf/goodman/goodman-gas-furnace/GMEC96-sp.pdf

There are surely a few others out there.

If you finish off the insulation projects soon and can get good data from this winter's fuel use indicating a load less than 29K the smaller one might still work.

If you're also going to be installing AC using the same ducts, you might consider getting a heat pump solution instead of gas, but I suspect gas is going to be much cheaper to operate in the near term. A ductless mini-split heat pump for the main living area might still beat condensing gas on operating cost during the shoulder seasons if sized correctly.

Dana1, what are your thoughts on the cfm rating? That's something I don't really have much knowledge on other than I've seen 800-1600 as the range at least with the furnaces I've looked at online. When thinking about it, I assume that the higher the fan power the more parasitic loss as you mentioned before, but then I also wonder at what rate is it not enough (assuming that's a thing).

Most oversized furnaces are delivering ridiculously high cfm in the first place, but there can sometimes be issues with room-to-room temperature balance when backing off the cfm by half or more. The duct design contributes to the as-used cfm, since duct friction determines the amount of backpressure the air handler experiences. Flex duct that flopped into place and not stretched tight is notorious for making actual room flows unpredictable. If it's hard-piped ducts designed for 0.1 water-inches at higher flows the actual flow using a smaller air handler can even exceed spec. For a house your size I would expect that most duct runs are short enough that it barely matters, unless it has lots of twist & turns using sharp-throated ells (with a hard 90 degree bend rather than a radius on the inside of the turn) which add a lot of "equivalent length" to the runs.

https://www.energyvanguard.com/blog/74831/The-2-Primary-Causes-of-Reduced-Air-Flow-in-Ducts