Life with a Geothermal Heat Pump
Damn. Air conditioner died again.
This was my reaction on returning home after a weekend of camping in the hot summer sun, and equally hot and sticky nights, and not getting much sleep. So I had been looking forward to a nice cool evening; and instead, the house was hot and stuffy. It was even hotter and stuffier outside, so opening the windows wouldn’t help.
Back when I moved into my new house, it didn’t come with air conditioning. Or curtains. Or toilet paper roll holders. It was the spring, and unlike the previous house, this one didn’t have a giant swimming pool. (A story for another time.) I clearly wasn’t going to make it through the summer without my life-sustaining AC. Something had to be done. And pronto!
I could have installed a traditional air-source heat pump. Those are the ones with the whirring fan, compressor, and heat exchange coils situated right outside, found next to many people’s houses. However, I had read about a new type of geothermal heat pump that could be installed with vertical wells, did not need acres of land, and operated reliably summer and winter - with just a single compressor, and no secondary water-based heat exchange cycle to freeze up or fail at inopportune moments.
Well. Or more precisely, well, well, well, well. It turns out that installing a geothermal heat pump is not cheap. Digging wells is problematic at the best of times, and the compressor and associated heat transfer hardware (coils) need to be sized for both winter heating loads and summer cooling loads — whichever is larger. In my climate, with -30 °C winters and +30 °C summers, winter heating requires more energy than summer cooling, even with a well-insulated house. (It could always be better insulated too, of course, and that’s a whole other kettle of fish as well.) I opted for a 4 ton heat exchanger based on the advice of the installer. This meant four wells had to be dug in the yard, one per ton of cooling.
(Side note: A “ton” of refrigeration is defined as the heat of “fusion” [melting] required to melt 1 short ton [907 kg] of pure water ice, initially at 0 °C, over a period of 24 hours. The same energy can, of course, be used in reverse to provide heating.)
In an ideal situation, the wells for in-ground heat exchangers can be dug before the house foundation is finished, and can therefore be situated right underneath it, out of the way. My builder wouldn’t allow this, so the wells were dug in the back yard. It took a month to dig four of them. They go down 100 feet through clay, bedrock, sand, more bedrock, and a significant volume of underground rivers. Wait, what? Underground rivers? Right under my house?
It turns out that the water flow is important because it enables efficient heat transfer. If there was less natural flow, or none, the wells would have had to include a water circulation system fed from the garden hose. That makes everything more complicated. And expensive. And prone to freezing. So I’m very glad I didn’t need extra water in the system.
Of course, during the well-digging process, some of the underground rivers ended up rivering right up the well hole, across the yard, and into my basement. That’s another story too.
All of this is worth it, though, because once a heat pump is installed, it is dramatically more efficient than any other heating or cooling system you can find — other than pure solar heating. (I looked into that too. Not this time.)
This is because the heat pump has access to an underground reservoir of heat (and cold) that is approximately 4 °C year round. In the summer, this means that instead of trying to pump heat from a 20 °C or 25 °C house to a 30 °C ambient outdoor air environment — which is an uphill battle from a thermodynamic point of view — it can simply encourage the heat to flow “downhill” (both from a gravitational and a thermodynamic standpoint) into the 4 °C ground. Of course, the compressor is still required to move the refrigerant and increase the rate of heat transfer, but it can operate very efficiently, since it is pumping heat in the direction it would normally have gone anyway. Meanwhile, in winter, an air source heat pump is effectively useless when the air source is at -30 °, because the 50 ° uphill climb to a 20 ° indoor space is simply too high for any efficiency whatsoever. A 4 ° ground source on the other hand is quite useable, and represents approximately the same thermodynamic efficiency as summertime cooling.
(An added benefit is that this heat pump includes a domestic hot water heating loop. This is optional, and other heat pumps have these too, but it is still an extremely efficient add-on. In the winter, the heat pump heats a hot water storage tank which is used to feed the gas-powered heating tank. It is more efficient to move heat around than to create it from scratch, so this saves some money compared to pure natural gas (or electric) water heating. But the best efficiency gain occurs in summer. The heat pump takes heat out of the house and heats the water tank with it, before sending the remainder into the ground. Hot water is free! It is a waste by-product of cooling the house. Fan-tastic.)
All of that sounds great on paper. But how well does it work in real life?
The first thing I noticed once I had used the system for a while (other than having a natural gas bill measured in basically pennies) is that the 4 ton system was overpowered. 3 tons would have been plenty. Ideally you want your heat pump (in winter) to provide most of your heating — around 90% — but not all. The remainder, when occasionally needed, can be made up with an electric resistance heater. This way, the heat pump is running efficiently, being on most of the time, rather than cycling on and off for shorter periods.
Not only was the 4 ton system overpowered for winter heating, but because the summer cooling load is significantly less than winter heating, it was *way* overpowered for air conditioning. Most single-family houses typically have a 1.5 ton air conditioner - not 4 tons!
Well, you might think, that’s great — who would not want to have more heating or air conditioning than they actually need? Sometimes you want to warm up or cool down a lot. And the effect of this overpowered system turned out to be not too drastic in winter: the cycle time was shorter than it should be, but still reasonably long, especially when it got particularly cold outside. I disabled the electric resistance heater backup stage entirely. (That stage is effectively a giant 10 kW toaster in the air duct. With some creative engineering, maybe I could turn it into an actual toaster…)
In summer, though, the cooling cycle time, as initially configured, was way too short to be efficient either in terms of electricity usage, or in terms of dehumidification.
(A brief digression about dehumidification: in humid climes like mine, air conditioners serve a dual purpose, and each is equally important: cooling the air, and condensing water out of it. Dry air feels much cooler than humid air at the same temperature. Sometimes it is possible to get away with just a dehumidifier, which is simply an air conditioner that doesn’t bother to vent the hot air it creates outside. And in excessively hot but dry climates, spraying water mist can cool quite effectively as the water evaporates and absorbs heat in the process.)
So the rest of my story —my adventures with this system so far — have been a direct result of creative attempts to compensate for the overpowered heat transfer and consequent short and inefficient cycle times. With one exception…
To wit: Many motor-driven single-phase-AC-powered systems — for instance, ceiling fans or air conditioners — include a run capacitor to provide the correct phase of power in sequence to the motor windings, in order to allow it to turn in one direction or the other. These capacitors seem to be the first part to regularly fail, and are fortunately usually designed to be easy to replace. My heat pump is on its third run capacitor now (and my ceiling fan is about to need a new one too).
But enough about capacitors, and onward to cycle time optimization…
My first step was to replace my standard thermostat with a slightly more programmable one. Most basic thermostats have a fixed temperature window: the amount by which they let the room temperature vary from your desired set point before they turn the furnace or air conditioner on. And this window is usually fixed at 0.5 °C because someone once determined that this was the minimum amount of change that a human could readily perceive.
My requirements are totally different.
I care not at all whether my house temperature varies a couple of degrees, or more, from the set point. That won’t kill me. I do care that my heat pump operates efficiently, lowering my electric bill, which could kill me. That requires long cycle times.
So I found a thermostat (by Robertshaw) that has a programmable set point window. I increased the window to its maximum of 1.5 °C. Not ideal; I would have liked 2 ° or more. But it is a lot better than 0.5 °.
This worked quite well to maintain a fixed temperature with acceptable cycle times. All was good… until my electric utility introduced time-of-day pricing.
I have no beef with time-of-day pricing per se; the underlying generation cost is governed by supply and demand like everything else, and it will be higher during the day and lower at night. (I would even happily live with that, and not worry about optimizing it at all, if we were only paying the underlying electricity cost. But my government in Ontario has tacked on anywhere from 5 to 15 cents per kWh, depending on the time of day, in taxes, to fund their misguided “sustainable energy” pipe dreams. That’s a whole other rant too.)
So all of a sudden my 3 kilowatt compressor started sticking out like a sore thumb in my electricity usage. I am frugal! I can’t have that thing costing me upwards of 28 cents per kilowatt-hour to run (that would be the total cost, including distribution etc.). At 3 kilowatts, that’s nearly a dollar an hour. Inconceivable!
Well, no worries, that thermostat I bought is programmable, right? It allows me to set up a program that would heat or cool the house during cheap times (at night), and let it naturally cool down (or warm up) during expensive times, i.e. the daytime. That would have worked very well - except that it also comes with an “energy efficient recovery” feature. Uh oh… a what now?
When my thermostat started turning on my heat pump at completely inappropriate (expensive) times, I investigated what that meant. The “energy efficient recovery” is designed to (in some way that escapes me) “reduce” energy usage by anticipating the amount of time it will take to reach a given set point, based on its own past experience, and therefore turning on the heating or cooling system early. How early? Well, there’s no way to tell. It tries to be smart about it, but it won’t tell you how smart. And there is no way to disable that feature on this thermostat. And this specific thermostat is the only one I could find that has a programmable set point window — which is critical for extending the heat pump cycle time, and maximizing its efficiency.
So, the obvious solution is that I need a better, i.e. more flexible, thermostat. There doesn’t appear to be an off-the-shelf answer to that problem. Yet. I will probably have to fabricate my own. (And I don’t like the Nest if it won’t work at all when the network is down. That’s not acceptable.)
In the meantime, I was able to work around that issue by simply turning on the thermostat myself at 7 pm, and turning it off at 7 am. That works well as long as I am at home.
What if I am not at home for the night? That situation leads to more excitement...
A few times I tried just leaving the heat pump off while I was away. Then when I came back, the house was 7 or 8 °C warmer than I like. Fine, so far… a 4-ton heat pump would usually have no trouble dealing with that kind of cooling task. But trying to pump all of that heat out, in one continuous run, led to three different problems on three different occasions!
Problem #1: The compressor itself overheated after four or five hours, and shut down in self-defense. This I think is largely due to the system cabinet design. A typical air source heat pump, with the compressor located outdoors in a well-ventilated cabinet with a giant fan blowing on it all the time, can happily run forever. My split cabinet geothermal unit is different. The compressor and associated plumbing and valves are in an enclosed soundproofed box, indoors, sitting on a styrofoam insulating pad. (It’s very quiet.) But there is no ventilation in there. Since it is in the business of pumping heat, you might think that it could take care of keeping itself cool enough in the process, but, while surprisingly it can run continuously for several hours, it cannot run indefinitely.
The manual solution, of course, is to shut it off to cool down periodically. A simple automated solution would be to have, in addition to all my other programmability features, a way to tell my thermostat to shut off the compressor to cool, after, say, two or three hours of running, no matter what else is going on. A more complex automated solution would be to create a second air conditioning system designed to cool only the air conditioner cabinet!
(Alternatively, I could just leave the cabinet open, but then it is very noisy. There may be a way to allow air through and keep noise down. I am still thinking about this possibility.)
Problem #2: The refrigeration coils in the air handler (where the fan, filter, and coils are) frosted over and blocked the airflow — again after three or four hours. This is a common problem with refrigeration equipment and can be seen in any refrigerator, freezer, dehumidifier, etc. This problem would also be solved with the above cycle time limiter algorithm, basically an enforced defrost cycle. When will they invent a frost-free geothermal air conditioner?
Manual solution: same as above, namely, turn the compressor off and let the coils defrost and dry, then turn it on again.
Problem #3: The run capacitor died. (See explanation of this one above.) This may or may not have been related in any way to the cooling demands during the hot summer.
Solution: replace the run capacitor. Alright, that one was easy enough, and doesn’t require an automated solution. (I did think about an automated run-capacitor hot-swap mechanism!)
Update: After experimenting with airflow settings, and air filters, and pressure gauges, we determined that the cause of some of the above problems was an insufficient charge (quantity) of refrigerant in the system. That, oddly enough, makes the evaporation coils too cold (below freezing). When the coils are below freezing, frost builds up inevitably. As a side effect, since the refrigerant is also used to cool the compressor motor, that can overheat while the coils are frosting up, and then it can potentially also overheat anything else in the same cabinet, such as the run capacitor, which is apparently sensitive to heat.
Having increased the refrigerant charge by three pounds, the evaporator coils are now above freezing, thereby avoiding frosting up, and the compressor runs cooler. I have not yet tried re-closing the cabinet to see if the compressor is sufficiently cooled by the refrigerant, however.
In conclusion, I am a huge fan of the geothermal heat pump, but if you want to go this route, make sure not to get an oversized one. And as with any motor-driven device, be prepared to replace the run capacitor (have one or two spares on hand).