Is this a valid way to measure heat load?
Our manual J calculation said that the heat load of the house is about 30K at about 65 degree delta T. (It was more precise than that but I don’t have the report handy, so I am going from memory.) I am hoping to confirm that the actual load is not too far away from that before we get into really cold weather.
Over the weekend, I finally got around to programming a schedule into the thermostat. I set the thermostat to drop back at 11:00 p.m. and go back up at 7:00 a.m. I happened to be awake just before 7:00 a.m. and noticed that the temperature had decreased only 3 degrees so the furnace should not have come on at all during the night. When the setback period ended, the furnace ran continuously until 1:00 p.m. to get the house back up to temperature. After concluding that setbacks are not very useful in this house, I decided to use this info to estimate the heat load. The furnace has an output rating of 39 kbtu/hr on high stage, which it should have been on for most of that time. Over the 14 hour period from 11:00 p.m. to 1:00 p.m., the furnace ran 6 hours. So, I estimate the heat load as 39K*(6/14) = 16.7 kbtu/hr. If I scale that linearly from the roughly 35 delta T at the time to the 65 degree design delta T, I get 31.0 kbtu/hr. Internal gains were about as low as they would ever be (overcast sky, no cooking, etc.).
Is this methodology valid, or does it leave out something significant?
I was surprised by the slow rate of change in both directions. Does that fall within the believable range?
GBA Detail Library
A collection of one thousand construction details organized by climate and house part
Replies
Reid,
Your anecdote raises several questions.
Unless I missed it, you didn't tell us the outdoor temperature on the morning in question. How cold was it?
Assuming that you have a multi-stage furnace, what makes you certain that the furnace was operating at high stage for six hours straight? It seems more likely to me that it was operating at less than full load.
The outdoor temperature averaged about 35 degrees during this time period (for a delta T of about 35 degrees).
Regarding my assumption that the furnace was on high stage most of the time: I have Emerson Sensi thermostats which have a web application. The web interface says how long the current heat call has been in progress. (I wish I had Ecobees so I could download a time history.) One of the three zones had a continuous six hour heating call. I have a single stage zone controller. Based on the literature for the zone controller, it sends a heat call to the furnace whenever any of the three zones is sending it a heating call. Based on the literature for the furnace, it operates on low stage for the first 12 minutes and then switches to high stage until the end of a heat call. Therefore, I concluded that the furnace should have operated on high stage for all but the first 12 minutes of the 6 hours. There is a plenum temperature sensor and the zone controller is set to interrupt a heating call if the plenum temperature goes above 170. I don't know whether that happened.
Your analysis is correct, although there are some limitations, most of which you have mentioned already. In addition to what you have mentioned, the temperatures of the materials mid-way through the walls were not in steady-state through the duration of the experiment. You'd have better accuracy if the indoor and outdoor temperatures were steady during the experiment. But probably your analysis is accurate enough for what you want to accomplish.
Since a large fraction of the sample period was during daylight hours, the amount of passive solar gains received or the amount of wind (or lack thereof) could skew the measurement considerably. With enough repeat experiments you'd be able to sort that out.
The result of any single day's experiment would have large error bars, but the average performance over a wintertime month would narrow the bars a bit. But even a single day's measurement such as that is usually closer than the typical HVAC contractor load estimation methods.
Depending on the house, air infiltration can be a big part of heat load. And infiltration varies greatly with wind.
In leaky houses infiltration can also vary significantly by delta-T, particularly house 2 stories or taller. At IRC code max leakage rates or tighter both factors tend to fade.
But for ballparking the heat load of a reasonably tight house monitoring the duty cycle of the furnace against outdoor temp or delta-T is still pretty good.
The blower door test for this house was 0.82 ACH50, so we are not dealing with a leaky house.
Update:
Last night, between about midnight and sunrise, the outside temperature was within a couple degrees of the design temperature of 8 degrees used in the manual J. Without any thermostat setbacks, the furnace ran continuously from about 4:30 a.m. until 8:30 a.m. That leads me to believe that my actual heat losses at that temperature are close to the furnace's output rating of 39K btu/hr. (Since the sun came up, it has been running about half of the time.) How concerned should I be that this is higher than the 34.8K btu/hr predicted by the manual J? Most of the differences I can think of between the modeled scenario and the actual conditions last night would trend in the opposite direction:
- The manual J assumed "semi-tight construction. The blower door test was 0.82 ACH50.
- The manual J assumed 109 cfm ventilation with no recovery. Actual was 130 with an ERV.
- The manual J assumed no internal gains. We had five people, two dogs, and two cats in the house, plus some electrical loads.
- The manual J assumed the same outside temperature around the full envelope (above ground). A substantial fraction of the exterior walls abut the garage which was about 40 degrees.
It will be interesting to see how the house performs this Sunday night when the outside temperature is forecast to drop to -6.