My natural gas bill notes how many GJ of energy I used over each billing period. I've converted that to BTU/h and plotted that below:
It's not clear if that is the instantaneous peak BTU/hr at some indoor & outdoor temperature (probably not) which is what we are looking for, or if it's the average over the number of hours in a month(?). Without accurate average heating degree-day temperature data an average use per month isn't going to tell us much.
If I understand correctly, we want to know how many BTU/h my furnace consumes at -26C? To estimate that, I downloaded the thermostat data which tracks indoor and outdoor temperature. When we hit -26C, the temperature difference (dT) would be 47C because indoor temperature is around 21C. None of the average dT's were that low during any of the billing periods.
The BTU/hr at 21C indoors, -26C outdoors is the relevant number. And it has to be scaled by the efficiency of the unit, not the input BTUs, since it's the net output that went into the heating of the house- the rest went up the flue. Eg, a burn rate of 50,000 BTU/hr into an 80% furnace only delivers 0.8 x 50K= 40,000 BTU/hr into the house, and if using ASHRAE's 1.4x scaling factor the right size would be a furnace with 1.4 x 40,000= 56,000 BTU/hr.
To estimate energy consumption at -26C (or a DT of 47), I created a cross plot with data from the above two plots. I then drew a straight line extrapolation to see what energy usage would be at DT of 47:
At a DT of 47C, I would expect the furnace to use about 50,000 BTU/h. If I multiply that by 1.4, then a furnace of 70,000 BTU/h would be good.
BUT, my old furnace only had a 80% efficiency and I'm replacing it with something that is 95% efficient. Thus, I could go down to 60,000 BTU/h. Does that sound right?
Again it's the OUTPUT BTU/hr, not input BTU/hr that's relevant. So scaling a 40K design load by to 56K means a 95% efficiency x 60K-in (= 57K out) furnace would be about right.
Several questions/concerns. On the cross plot, the data point at 40,000 BTU/h looks to be an outlier. However, it happened during the coldest month we had this year, so I don't feel good about ignoring that data point. The other thing I noted is that my thermostat tracks the amount of time the flame is on. I have a single stage furnace with input of 99,000 BTU/h. I figured I could multiply this by the number of hours the furnace is on (according to my thermostat) and that it would line up with the gas bill. It doesn't. Gas bill says I'm using 50% more gas that that calculation. Does it make sense that the input to my furnace could be 50% higher than name plate? What do you do when there is uncertainty? Size up?
That's why you need to run the math on gas use against heating degree-day data from a nearby weather station during the coldest months only (to reduce the distortions of hot water use and passive solar gains). If you keep it ~21C indoors use base 18C HDD weather data. You will find that the BTU per HDD numbers will fall in a fairly narrow range when using colder month only data. Converting BTU/degree-day to BTU per degree-hour is simple arithmetic (divide by 24), which is a linear constant. The zero point is the HDD base (I'm recommending using 18C based on your indoor design temp), since that is the presumptive point where internal gains from mammalian & avian occupants and background energy use (water heater standby, refrigerators, other 24/7 electrical plug loads) keep the place at a fairly constant temperature without direct heating. So you're extrapolating from a linear rate- the heat that needs to be supplied by the furnace increases by your derived constant with every degree below
18C (not the full difference between indoor outdoor temperatures).
While the performance of the building envelope isn't perfectly linear with delta-T, it's close enough. The performance of insulation increases with higher delta-Ts, the performance of windows falls, but within the temperature range of interest it's going to close enough for sizing the equipment. Only very-high performance houses would call for better thermal modeling to be used.
I'm a bit surprised there hasn't been more chatter about the charts/data. I suspect it is pretty rare for someone to have actual data, so perhaps I have thrown everyone for a loop
The loop I've been on is more about binge watching NetFlix on the couch with the dog waiting for the pandemic to subside.
The shorter term thermostat duty cycle and temperature information is useful, but usually not quite as good as accurate as longer term averages. Still it looks pretty reasonable. Looking at ONLY the duty cycle sections
after the house has cooled off to a stable 20C the average outdoor temp is about -28C, and the duty cycle is about 25 minutes/hr = 42%.
The output of an 80% 99KBTU/hr furnace is 79,000 BTU/hr, so a 42% duty cycle means it's using 0.42 x 79K= ~33,200 BTU/hr. With a presumptive balance point of 18C and an outdoor temp of 4C that's 18C- -28C=46C heating degrees and a heating constant of 32,200/46C= 700 BTU/degree-hour indicating a pretty tight house- that's pretty good for a 2x6 framed walkout- I would have expected somewhat more (but not 2x).
So at a design temp of -26C outdoors you would be at (18C - -26C= ) 44 heating degrees, for an implied load at that temperature of 44 x 700 BTU/hr= 30,800BTU/hr. Scaling that by 1.4 (ASHRAE) puts the right size furnace output at 1.4 x 30,800= 43,120 BTU/hr. That means a 50K condensing furnace(x 95% efficiency= 47.5K out) is about right but at (47.5K/30.8K=) 1.67x still slightly oversized, thus 2-stage or fully modulating would be recommended.
Try this method, to gain better psychological comfort and (usually) better accuracy. It's conceivable that it may turn out that 40K furnace would be a better fit, if the longer term analysis shows slightly lower design load numbers.