Image Credit: Elden Lindamood The electrical panel with all the TED sensors installed, ready to data-log. This pie chart shows energy use over one year. "MTUI" is the house total but does not represent all power consumption on the property. The red line shows the minisplit cycling in and out of defrost mode on a -25° night. The blue and green lines are the electric cove heaters ticking on when the minisplit could no longer keep up at 4 a.m. The yellow line represents total electrical use in the house. This is a 24 hour graph, showing the minisplit (red line) turning off after the sun comes out at 11 a.m., coming on again after sunset, and then shutting off again later on when I started a fire in the wood stove. The yellow line is the whole house usage spiking with the hot water heater and range mostly. High temp that day was -12°. Here, the red line shows the mini-split coming on and ramping up to full speed 20 minutes of every hour. I soon figured out this was happening in conjunction with the HRV coming on despite the minisplit thermostat setting being lower than the room temperature. This was proof of the problem. The fiberglass triple-glazed windows get a bit of condensation on them when it is -25° outside, as it was when this photo was taken. The windows are “outies” and I don’t take the screens out. Interior relative humidity was about 40% in this instance. Note the condensation on the frame as well as the glass edge. The house has performed as planned, and it's a welcoming spot to come home to.
Editor’s note: Architect Elden Lindamood wrote about the construction of his house in northern Minnesota in a series of blogs at GBA in 2016. The first of them was called A Low-Energy House for Northern Minnesota. Here’s his report on the first year of occupancy.
It has been over a year now that Catherine and I have lived in our low-energy house. I’ve continued to chip away at some of the items on the endless “to-do” list, but the house is complete from an operational standpoint. I’ve been compiling experiences and data, and this is what I’ve learned.
First I’d like to revisit the interior humidity issue. If you’ll recall from earlier posts, my indoor relative humidity (RH) was hovering in the 75% range much of the first summer. I simply didn’t have enough cooling load for the air conditioning to run long enough and beat it down. I also suspected I was dealing with a lot of construction moisture.
I found that if I ran a portable dehumidifier in the mechanical room for 12-24 hours, that alone would knock the humidity down to 45% for a while, so I did that periodically that first summer and even occasionally over the first winter. Since then, the indoor humidity seems to have settled down. This last winter it has hovered around 30-35%, all winter, with no mechanical dehumidification. This anecdotally verifies that there was likely a lot of construction moisture banked in the walls and slab.
For those who are wondering, I pretty much set my heat-recovery ventilator (HRV) to run 20-minutes-on/ 40-minutes-off all the time, using the high-speed override when showering. I do realize I could have affected the indoor humidity by fussing with the HRV settings more, but I didn’t want to have to do that in order to maintain a reasonable moisture level. I’d like the house to function with no more inputs than Catherine would readily offer. I could obsess and tinker with things, but I shouldn’t have to.
I installed a TED5000 energy monitoring system on select circuits to see how things were performing, and where there might be room for improvement (see Image #2 below). The monitor has a main set of current transformers (CTs) for the line coming into the house, and then 16 CTs on individual circuits for more specific data collection (two CTs on 240 volt circuits).
This has provided me with a wealth of information, and helped me diagnose and solve a major problem with the minisplit. A pie chart and table showing the usage of individual circuits is below (see Image #3).
I should note here that there is a “farm total” electrical use number above the circuit breakdown. Because of the way we developed our farm, not all power goes through our house. The “MTU1” line on the table is the house total, but that does not include the garage/shop, the well, the greenhouse fans and heaters, or the potting shed which has lots of grow lights on in the spring. The “Farm Other” line at the bottom of the table is the electrical bill less the house usage. The pie chart represents the house circuits only.
I ran two separate energy models of the house prior to construction: One using REM/Design software, and one using Marc Rosenbaum’s calculators. The results of the two energy models were similar, predicting annual space heating load of 17.9 million Btu per year (MMBtu/yr) and 15.75 MMBtu/yr respectively. I was happy that both the calculators were within a reasonable range of similarity.
REM/Design calculated the passive solar contribution, internal gains, and coefficient of performance (COP) to be worth 6.2 MMBtu/yr, giving me a predicted annual consumption (as opposed to load) of 12.6 MMBtu/yr for heat, and 1.3 MMBtu/yr for cooling, for a total of 13.9 MMBtu/yr predicted consumption.
To check my actual consumption I combined the usage log of the CTs on the minisplit circuit, along with the usage of the CTs on my backup electric resistive cove heaters, and arrived at a total consumption of 2,092 kWh, or 6.96 MMBtu for the year for heating and cooling, so at first I was miffed at how the models could have been so far off. I had entered the heat pump modeled COP at 2.0, assuming it would be running in the extreme cold much of the time. Unless it was way better than that, or unless my solar contribution was way better than predicted, I had some explaining to do. Then I started considering the wood stove.
The wood stove factor
My partner and I did use our wood stove, but not religiously. We’d burn a fire in the evening maybe three or four nights a week, sometimes less. We lit it for ambiance rather than out of some desire for great energy savings. Because of our intermittent use of the stove, I didn’t think it could be contributing much, but I figured I should calculate that, roughly at least. I did not measure how much firewood we consumed, but I knew by visual guess that we used less than a cord of semi-rotten deadfall poplar during the year.
It turns out that contained maybe 8 MMBtu when adjusted for wood quality and a 70% efficiency wood stove. When I add that to the electrical usage, that put me pretty much right where the energy models had predicted. Cool.
So for a year’s worth of heating and cooling we used 2,100 kWh, plus some free firewood, totaling about $300 for the year for heating and cooling. This is in northern Minnesota, with 9,500 heating-degree days. Also cool.
Looking at the pie chart reveals that the minisplit was our largest consumer for the year, as I expected, followed closely by the domestic hot water heater (DHW). Although it is a simple electric-resistance storage tank heater, I found that we used quite a bit less energy for hot water than both of the models predicted. The REM and Rosenbaum models predicted use of 2,843 kWh and 3,582 kWh per year for DHW, yet my data logger says we only used 1,501 kWh. Catherine asked if that meant she could take longer showers. I simply looked at her disappointingly.
Out of curiosity I checked my data once when we got back from vacation. With nobody home, the hot water heater blips on for about 3 to 5 minutes every 14 hours or so. Not bad.
The ducted minisplit has exceeded my expectations. I’ve taken the “set it and forget it” stance, leaving it set at 70°F all the time, regardless of outdoor temperature. At -25°F, it is still out there churning away and contributing a small amount of heat. Not quite enough, however, to keep the house warm on a -30°F night.
So, I keep the minisplit thermostat set at 70°F, and I have each of the six backup cove heaters set at 65°F, such that if the minisplit can’t keep up, the backups come on automatically. The line graph below (see Image #4) shows the minisplit cycling in and out of defrost mode (red line), when at about 4:30 in the morning one of the backup cove heaters blips on (green line). Over the next couple hours additional cove heaters blip on more and more to keep the house warm on a -28°F night. This is working just like I planned, and I am actually surprised at how long the minisplit can manage it before needing help.
The next line graph is a longer timeline of the day (see Image #5 below). You can see the red mini-split line running until about 10:30 a.m., when the sun warms the space enough to meet the thermostat and the minisplit shuts off. It remains off until about 6 p.m. despite it being -12°F for a high that day. Then it ticks on for a while, until I started a fire about 8 p.m. By 9 p.m., the stove is working and causes the minisplit to shut off again until about midnight. On the right, you can see the cove heaters kicking on again in the wee hours of the extremely cold night. Incidentally, the yellow line is total house use. It spikes with the use of the range and hot water. The tall spike is when we turned the broiler on to make dinner.
After getting home one evening I pulled up my data and said to Catherine, “It looks like you took a shower about 10 a.m., and had lunch about 12:30 and then did some laundry”. She told me to stop doing that.
Solving a minisplit problem
The two main disappointments about the minisplit are the power use when the fan is on, and the power use when it is off. I’ve determined it is way more efficient to circulate the warm air from the wood stove or sun using the HRV on recirc rather than the minisplit on fan mode. Also, the minisplit uses about 32 kWh a month when it is off. I could throw the breaker to kill it entirely, but I don’t consider that a reasonable expectation.
The data did help me solve a problem with the minisplit, though. In the first few months of the first winter, I felt the minisplit was coming on when it shouldn’t have. It would be 75°F in the living room because of the solar gain, and the minisplit would kick on periodically, hard and fast. I eventually linked the minisplit operation to the HRV operation with the attached graph (see Image #6 below).
Working with the equipment supplier we determined that the HRV was dumping cold air into the upstream side of the minisplit, and the thermistor in the unit was causing it to panic and ramp onto high heat to try to fix it. This caused the compressor coil to ice up badly. It turns out that out-of-the-box, the minisplit was set up to read air temperature inside the fan unit rather than at the thermostat. We went deep into the technical manual and re-programmed the unit to read the air temperatre at the thermostat instead, and it has been running flawlessly ever since. No ice-up even at -25°F.
I had originally wanted the HRV supply to be in dedicated ductwork, but was convinced to supply it via the minisplit ducts during construction. Had I stuck to my guns, this issue would not have come up. As it is, however, it works fine now. Just be aware that sometimes you might need a data graph to prove that your system isn’t working how you need it to, and that you might need the 400-page technical manual from the manufacturer to adjust things the way you want.
Other than that, things are working well, and the house is meeting or exceeding my performance expectations. Catherine has backdrafted the wood stove badly with the range hood twice, neglecting to turn on the makeup air (MUA) unit. I’ve also found that we’ll tend to open a small kitchen window a crack rather than turning on the MUA, even when it is brutally cold out. The MUA is loud, and the small window is near the range hood, so the air doesn’t cool the kitchen much.
The condensing clothes dryer takes an extremely long time to dry clothes, but I’m trying to retrain my mind to accept that. In the summer we use the line outside anyway.
The entry can get a bit cold (64°F) on cold nights, but it has the least efficient, but prettiest, door in the house. I might put more minisplit air there if I were to do it again.
The triple-pane Duxton windows will get a small amount of condensation on them when it is below -15°F out (see Image #7 below). Not bad at all, though, and only in the lower corners. If I removed my screens, that would help. That exceeds my expectations.
Another lesson is that I stupidly put a closet on an outside northwest corner, and then crammed it full of junk. I checked the lower corner wall temperatre one cold night, and the drywall was 55°F behind all that stuff. I “thinned” the junk and now leave the closet door open when it is really cold. I would not locate that closet there if I were to do it again.
Smaller carbon footprint, bigger waistline
Finally, I did a quick and dirty calculation. My previous house was in the city, and I biked most places. The house, built in 1901, and leaked like a sieve and used about 600 gallons of fuel oil a year to heat its 600 square feet with deep nightly thermostat setbacks. Our new 1,575-square-foot “mansion” uses 2,000 kWh to heat (and cool) per year, but I have to drive 26 miles each way to work with my car, which gets about 32 MPG.
Without getting too deep in the weeds, I determined my carbon footprint is now about 30% less for all my activities, even with the commute. I am also about 30% fatter now that I’ve stopped biking. PV will help the carbon footprint at least.
So, to close it all out, I’d give the following advice:
- Passive solar works. Do it.
- Cold climate minisplits work. Do it.
- You can have a wood stove in a super-tight house if you pay attention 99.99% of the time, and don’t mind your house smelling like a campfire .01% of the time.
And it’s nice place to come home to (see Image #8).