GBA Logo horizontal Facebook LinkedIn Email Pinterest Twitter Instagram YouTube Icon Navigation Search Icon Main Search Icon Video Play Icon Plus Icon Minus Icon Picture icon Hamburger Icon Close Icon Sorted

Community and Q&A

Electric Water Heater for Occasional-Use Space

washxhouse | Posted in General Questions on

I’m looking at finishing a mother-in-law apartment over a detached garage. It will be used for an actual mother-in-law and occasional guests, probably max 30 days / year.

I’m (newly) all electric on this property and not interested in reintroducing fossil fuels to the mix, so I’m limited to electric options for hot water.

I’ve read a lot of helpful stuff on the forum indicating that natural gas tankless heaters don’t make sense in retrofit situations when used in a regularly-occupied space, but my situation is more suitable, I believe, in that the standby cost differential between a tankless system and a tank would be more significant due to the 9/10 unoccupied days in this space.

So my question is, does (electric) tankless make sense in this context, serving a shower and sink for ~1/10 days of the year for 1-2 people? Or should I get a well-insulated (eg Rheem Marathon) small (eg ~20 gal) tank heater and put a controller on it (eg Aquanta, which I already use in my main house and don’t love but would do the job) to disable the heater remotely when it’s not in use?

GBA Prime

Join the leading community of building science experts

Become a GBA Prime member and get instant access to the latest developments in green building, research, and reports from the field.


  1. PBP1 | | #1

    I'd vote for small tank, the current/power draw for on-demand electric is significant and performance/value depends on the inlet temperature. I have a small 2.4 kW under kitchen sink on-demand electric (dedicated 20 amp circuit), a small 2.5 gal electric tank (plug-in) and an on-demand gas. The least of the worries go to the small electric tank, easy to turn on/off, service and disconnect. The 2.4 kW is fed by room temperature water and is only "good" for low flow (0.5 gal/min or less), including flow low enough to avoid triggering the on-demand gas. So, when I need to wash my hands quickly, it does the job without having to trigger the on-demand gas. For the electric tank option, you may want to have an easy way to drain it when not in use (drain without making a big mess, maybe for yard use?), not sure guests would want to use stagnant water sitting in a tank for 300 days.

  2. walta100 | | #2

    If you want a hot shower, you will need a large tankless electric unit that is going to require 3 maybe 4 40 amp 240 volt circuits unless you had this in your plans it seems unlikely your electric service has that much spare capacity.

    A 20 gallon tank will limit you to one very short shower every 6 hours or so.

    If you go tankless gas make sure to avoid one with a standing pilot light.


    1. washxhouse | | #13

      Thanks Walta!

  3. Expert Member


    I'd go with a variant of your last option. A newly installed electric tank heater filled with cold water gets up to temperature within an hour. If you know in advance the suite will be used, you can simply turn on the heater shortly beforehand, and the rest of the time leave it off, eliminating standby losses. Using it that way, I'm not sure it makes sense to spring for an expensive insulated Marathon over a full-sized conventional tank heater.

    1. washxhouse | | #10

      Oh yeah I do see your point ... I guess the appeal of the Marathon to me was longevity but I have read a bit about anode replacement extending the life of steel tank heaters so I'll do that cost-benefit with that in mind as well. Thanks!

      1. Expert Member
        MALCOLM TAYLOR | | #14

        I'm backtracking. Depending on your water, the absence of an anode might be enough reason to go with the Marathon. I've had terrible luck trying to change anodes out. It's buried in foam, you need a deep socket no one has, and it's usually seized enough you need a bar and a couple of other people to hold the tank from rotating while you apply torque. I'm probably batting a 50% success rate in successfully removing them.

  4. Expert Member
    Akos | | #4

    If you are in warmer climate, an electric tankless might not be all that bad. The two things dirving the sizing of these is how cold the water temperature is in the winter time and how big of a shower head you install.

    For example, around me we have 40F water in the winter time. Looking at a reasonable shower head say 2.2gal/min and want shower water at 100F, so a 60F rise.
    60F*2.2gpm* 500 BTU/(gpm*F)=66000 BTU/h or 19kW

    19kW works out to around a 80A demand so you need at least a 150A service to feed it. Not unreasonable but also not that small.

    If you can install a low flow shower head (~1.5GPM) you can probably squeak by with a smaller 13kW tankless unit so you can use a 100A service.

    The one thing to watch is if you hard water. Hard water is the enemy of tankless units, unless you want an ongoing maintenance item, stick to a tank even if less efficient.

    1. washxhouse | | #11

      Good points, all. I have fairly hard well water before it hits my iron breakers and softener units and I'm also right on the border of zones 4A/5A so not the warmest climate. I'm not super huge on adding ongoing maintenance items so that's a strong tick in the tank column.

  5. Expert Member
    BILL WICHERS | | #5

    I'm going to assume that since you mention this is a detached garage, you probably have at best a relatively small subpanel feeding power to that structure. That means you're probably guaranteed to not have enough capacity to run an electric tankless water heater, since those guzzle down a LOT of electricity while they're running. They are also pretty inefficient in terms of power grid loading too, which is another issue.

    I would recommend you use a tank-type hot water heater, maybe a 20 gallon, and put a drain water heat recovery unit in to help maximize the effective capacity of the relatively small water heater. Since you said the apartment is "over" the garage, you probably have access to the drain line somewhere.

    I recommend you tag the breaker feeding the hot water heater so that you can easily shut it off when the apartment is unoccupied. When the water heater is turned off, there is zero standby losses which eliminates ANY power consumption during the times the apartment is unoccupied, and you don't even need any additional equipment to implement this functionality! Just remember to turn it off when the inlaws leave, and try to remember to turn it on the day before the inlaws arrive. Worst case here you either get an "AHH! NO HOT WATER!" shout to tell you to turn it on, or a little higher electric bill for a while if you forget to turn it off. Pretty minor issues either way. You could add a pilot light somewhere to indicate when there is power to the water heater, which might help with the remembering...

    BTW, it's worth mentioning that in most areas of the country, using electric RESISTANCE heating will actually use MORE natural gas compared to using natural gas directly to heat the water. This is due to inefficiencies in power generation and transmission (pesky laws of physics), and the fact that most new generation installed in the past decade or two has been natural gas fueled. If you still have natural gas available, and your goal is reducing overall emissions, you'd actually be better off to use a natural gas tank-type (or on demand, the difference with natural gas isn't an issue the way it is with an electric on demand unit) water heater here. The only way around this particular issue would be to use a heat pump water heater, but I'm not aware of any of those that are small enough to really be suitable for your application here.


    1. user-2310254 | | #6

      If you decide to go with a gas water heater, watch out for connection charges. My gas utility charges about $40 dollars per month. (The only gas burner I have at the moment is a 40 gallon water heater.) The gas usage charge is generally less than $20 a month.

      1. Expert Member
        BILL WICHERS | | #9

        Ouch! My gas co charges $12.60/mo for a connection charge.

        I do agree that it probably doesn't make sense to go with gas if you don't have the service available already for other things.


    2. washxhouse | | #12

      I like this - had not thought of a drain water heat recovery system but this does feel like an appropriate application!

    3. Expert Member
      Dana Dorsett | | #15

      >"BTW, it's worth mentioning that in most areas of the country, using electric RESISTANCE heating will actually use MORE natural gas compared to using natural gas directly to heat the water. "

      That would only come close to being true in the dubious circumstance of a 100% gas-fired local grid. Most grids in North America already have a growing fraction of the power supplied by renewables.

      In places like Kansas & Oklahoma wind alone (ignoring solar & hydro) more than a third of the annual power generated, in Iowa it's closing in on 2/3 and climbing. S0urce:

      The EIA is projecting that for the US as a whole, 5% of the total power generated (on the utility's side of the meter) in 2022 will be from solar:

      Mind you the forward looking growth of solar (and wind, for that matter) projected by the EIA is always linear, whereas the actual growth of renewables has been exponential, with the installation increasing as the costs fall. That "14% by 2035, 20% by 2050" projection is a joke. By 2035 new utility scale solar power will be half the price of wind, and is already approaching (in some locations below) just the operating cost of gas generators. How likely is it that transition will be a steady ramp as shown in the graphic on that page? (Laughably UN-likely, sez me. :-) )

      And that is just on the utility side of the meter- there is also a LOT of solar installed on the ratepayer's side that shows up as "missing demand" in the daily power use curves plotted by grid operators. Just because it isn't being fully metered doesn't mean it isn't providing real power.

      At current trends, before 2030, well within the typical lifecyle electric tank water heater, and WAY within the lifecycle of a tank water heater used only 30 days out of the year (lifecycle is fairly linear with the cumulative volume of water heated over time), even 100% renewables grid will be cheaper than just the fuel & maintenance to keep an gas fired generator going. Rooftop solar will also be cheaper than residential retail rates (it has already passed that mark in much of the US).

      Even in the gas-heavy ISO-New England grid only a bit less than half of the net-delivered (after losses) annual power from commercial generators is gas, the vast majority of which is from combined cycle gas generators running ~50% thermal efficiency, with 23% sourced by nuclear, another 10%+ sourced by non-hydro renewables (diverse) , 6% hydro, etc. with only tiny fractions from fossil sources other than gas. See:

      There is no WAY that even a best-in-class gas fired tankless uses less gas than a plain old electric tank in that current scenario. The gas water heater would be using at least 50% more gas than the electric tank even now, and is likely to be using 300-1000% more gas than the electric water heater in the last year of it's expected lifecycle.

      Of course, your grid may vary.

      In response #7 below, DC Contrarian writes:

      >"I see no advantage to a small tank. Buy a reasonable sized tank and turn it off when you're not using it."

      Absolutely right!

      Smaller tanks often cost more up than standard 40 & 50 gallon models too. Be sure to turn off the cold water feed to it when walking away from it, in the unlikely (but expensive) event that it springs a leak when nobody is around to notice.

      Any electric water heater that is only delivering the equivalent of ~100 showers/year (3+ showers per day, 30 days/year) is likely to go 25+ years. The wear & tear on a water is predominantly stresses from thermal cycles, and there aren't going to be many at that level of use. Liming & corrosion are predominantly driven by the total amount of water passing though the water heater, which is also going to be very modest compared to a water heater serving up ~1500 showers/year (4+ showers/day, 365 days/year) over a typical 10 year warranty period. Even a 6 year warranty model (either no anode, or a stubby non-replaceable anode) could easily go more than 25 years using it only 30 days/year.

      1. Expert Member
        BILL WICHERS | | #16

        The numbers for an on-demand in terms of overall efficiency may be better when you consider that the tankless has zero standby losses for the ~11 months of the year that the OP says the apartment would be unused. It would take a lot of on-demand water heater inefficiency to make up for 11 months of standby losses! My preferred solution though, as I mentioned above, would be to just shut off a tank-type water heater during the times the suite was vacant as that way you get the best of both worlds.

        I suppose related to the "use more gas with electric", I should probably have just said "more fuel use". Much/most of the country has a mix somewhere around 50-70% from coal and natural gas, with the nuclear the largest next step, and everything else a relatively small slice of what's left. Thinking 'all electric' means fewer emissions (which is really the ultimate goal here, right?) is not necassarily the case. Remember also that there are transmission losses from lines (mostly) and transformers (small) too, which are probably somewhere around 15-20% worst case, since typical design efficiencies for transmission lines are in the 95%(ish) range. Those line losses effectively increase the amount of energy that had to be produced, or fuel consumed, to deliver the same amount of useful "work" at the end.

        I have been surprised at the amount of generation from wind in recent years, but it does vary by region, and it is not constant throughout the year. I have seen in the midwest up to around 20-25% from wind, which really surprised me, but it's usually much less than that, although 22.34% right now when I checked. Even with that much, 62% is coal and natural gas, and another 11.57% nuclear. What's left is "everything else", including solar. If we get up to 50% non-combustion sources, that's still a big chunk of power being sourced from plants that are burning something for fuel. That makes electric RESISTANCE heating a problem if your goal is reduced emissions, since you're consuming at least double the fuel in terms of BTU content than you're getting in BTU content from the electricity. That would make electric resistance heating the equivalent of a 50% efficient (at best) gas water heater, and typical water heaters are at least 80% efficient.

        BTW, good to see you posting again Dana. Long time no read :-)


        1. Expert Member
          Dana Dorsett | | #29

          >"The numbers for an on-demand in terms of overall efficiency may be better when you consider that the tankless has zero standby losses for the ~11 months of the year that the OP says the apartment would be unused. "

          The standby loss of a tank heater that is OFF for the idle 11 months is also zero.

          >"Much/most of the country has a mix somewhere around 50-70% from coal and natural gas, with the nuclear the largest next step, and everything else a relatively small slice of what's left. "

          Close- renewables edge out nuclear in most US markets, even without counting behind-the-meter PV. For the US nation as a whole the EIA numbers for 2o21 for grid sources (behind the meter PV, et al isn't counted) were:

          ... 60.8% fossil fuels

          ... 18.9% nuke,

          ... 20.1% renewables (with wind carrying the lion's share of that at 9.2%)


          That is going to change VERY dramatically over the lifecycle of a new water heater.

          Taking the US national averages, a minimum-legal 0.92 UEF electric tank doesn't have much standby even when it's on. So call it 1/0.92= 1.09 source-energy use, 40% of which is near-zero carb, 60% fossil burned at an average thermal efficiency of 45% (WAG, net of grid losses). The fuel use is then somewhere around (1.09 x 0.60)/0.45= 1.45 times fossil fuel input energy BTU to deliver up a BTU hot water. A 0.95 UEF tankless uses 1/0.95= 1.05 times the fossil fuel input.

          So in 2021 the electric tank would be using only about 1.45/1.05= 1.38 x as much fossil fuel as a gas or propane tankless. The ratio of carbon emissions would be somewhat higher than 1.38x, given that coal is still part of the mix, yet nowhere near 2x.

          Are we going to bet that those numbers will be about the same in 2028, 2032, or 2034 (when the warranty on the tank installed in 2022 would be up)? Growth in renewables is still exponential, not linear, doubling about every 3 years in the US. ( see ). On that historical curve trend (which has been pretty consistent for 40 years) the ~3% share of the pie PV held in 2021 will be more like ...

          5-6% in 2024, likely hitting

          ...15% in 2028%

          ...and 30%+ by 2032,

          ... or about 10x what it was in 2021.

          At some point the logistic curve will begin to flatten and become incrementally linear, but PV is still on the steep part of the ramp, with plenty of running room before it hit's that knee in the curve.

          Growth curves for wind output is a bit lumpier, with some years outstripping others, but as the nascent offshore biz launches and older wind farms get re-fitted with better kit, the 9% in 2021 will likely be more than 20% by 2032. Between just those two by the time a 10 year water heater warranty is up, non-hydro renewables will comprise half of all grid power, and that's just going on a business-as-usual approach. Applying various policy levers could accelerate that by quite a bit (at a cost LOWER than business as usual.)

          >"Remember also that there are transmission losses from lines (mostly) and transformers (small) too, which are probably somewhere around 15-20% worst case, since typical design efficiencies for transmission lines are in the 95%(ish) range."

          That 15-20% grid losses worst case would only be true for the user at the furthest end of the line in a typical third-world grid. In North America even a 12% loss would be the extreme case, with US average being under 10%. The world average grid losses run about 10%, though in India it's still north of 20% (a significant fraction of which is attributable to electricity theft, not simple line losses.)

          But if we are going to count transmission losses on the electric grid, a fair comparison would also have to count gas pipeline pumping energy use (including LNG compression for the NE-ISO grid), as well as fugitive methane from wellheads and transmission pipelines (including distribution pipeline) leakage, all of which are significant.

          Going down those paths require making a much more refined model, and which isn't really adding much value to the discussion over simpler mental models.

          >"I have been surprised at the amount of generation from wind in recent years, but it does vary by region, and it is not constant throughout the year. I have seen in the midwest up to around 20-25% from wind, which really surprised me, but it's usually much less than that, although 22.34% right now when I checked."

          It's true that the wind resources on the MISO grid are both large and highly variable, but almost never zero. New England offshore wind (or even off the Carolinas) will downright embarrass midwestern wind resources in terms of temporal consistency and total capacity factor. Yes, like water heating energy use patterns, wind is variable at any given location, but the transmission resources are already pretty good for moving it around region to region, not that overbuilding the transmission grid is financially optimal in every case compared to local/regional grid storage options.

  6. Expert Member
    NICK KEENAN | | #7

    I see no advantage to a small tank. Buy a reasonable sized tank and turn it off when you're not using it.

    1. irene3 | | #18

      Cheaper, less water to flush out after being unused for a long time, and takes up less space. If a fifty-gallon tank serves a household of five people taking showers and doing dishes and doing laundry, the way it did in our house when there were five at home, I don't see why a twenty-gallon tank shouldn't be fine for one or two people who mostly just use it for showers.

      1. Expert Member
        NICK KEENAN | | #19

        I just looked on the Home Depot site and a 20 gallon water heater is $399 and a 40 gallon is $449. I'll grant you it takes up less space. But 20 gallons is going to be really marginal for a shower for a lot of people.

        1. Expert Member
          MALCOLM TAYLOR | | #21


          I just bought a 20 gal tank at HD up here, as it was the only one what would fit in the crawlspace. It was more expensive than the 40.

          1. Expert Member
            NICK KEENAN | | #22

            Entirely believable. Non-standard is often more than standard.

  7. nynick | | #8

    I have a tankless propane for the exact situation you describe. I turn it on and off as needed. I love it.

  8. PBP1 | | #17

    A seldomly used tank (w/ heat turned off), like a seldomly used line of a home run system, will have stagnant water, which can be quite bad. Unused lines of a home run system (or even some other systems) may be drained or periodically flushed to reduce risks. A seldomly used tank should be drained when not in use or otherwise kept on and above 120 F (like 140 F). Thermostatic mixing valves are quite good when tank/on-demand water temp is high.

    “The water heaters were full of stagnant water, just chock-full of bacteria, and when people would take a shower they would breathe in bacteria,” Wasson says.

    1. Expert Member
      NICK KEENAN | | #20

      I would plumb a system like this so it can be drained just by turning the water off and opening the faucets, and I'd put a floor drain next to the water heater with a hose from the water heater drain going into the floor drain. So when it's time to close up you turn off the water shutoff, turn off the water heater breaker, and open the water heater drain and all the taps. Elapsed time less than five minutes, quicker than taking out the trash after guests. For a lot of reasons you don't want to leave the water on when nobody's there.

      1. washxhouse | | #23

        This makes sense to me. Would need to add another drain at ground level as this heater is likely to live under the stairs that go up into the unit but would save a lot of headache to be able to just open that tap and drain it out.

        1. Expert Member
          NICK KEENAN | | #24

          What I've done in seasonal houses is just have a pipe that goes outside for draining the water heater. It depends on the layout of the house whether it will work.

          1. PBP1 | | #25

            Can double as a gravity driven drip irrigation system ;-)

        2. Expert Member
          Akos | | #26

          Whenever you drain a water tank, eventually you'll forget to turn off the power or to fully fill the tank before turning it on. Inevitably you'll end up burning the heating element out.

          Adding a high visibility tag to both breaker and drain valve helps. Having a spare element and socket wrench or at least knowing where you can pick one up in a hurry is a good backup.

          1. Expert Member
            BILL WICHERS | | #27

            Putting a shutoff switch for the water heater next to the shutoff valve for the water supply to the water heater also helps. If the two shutoffs are right next to each other, it's much more difficult to turn one on and forget the other. Put a tag on the electrical switch that says "TURN ON WATER FIRST!".


          2. washxhouse | | #28

            Sage advice!

          3. Expert Member
            Dana Dorsett | | #30


          4. Expert Member
            MALCOLM TAYLOR | | #31


            What's a good shut off switch to use for these 220v loads?

          5. Expert Member
            BILL WICHERS | | #32

            I use double pole commercial ("spec" grade) switches for this purpose, which install in a regular electrical box. A 30A version is here:

            30A is enough for many water heaters with 5kw heating elements. There is a cheaper 20A version of the same switch that will work for 3kw water heaters. I use these switches for basic shutoff purposes for small water heaters, steam-type humidifiers, and other relatively small electrical things that need a double-pole switch capable of switching a 240V circuit.

            If you need something that can be locked out, the commercial electrical term for that is a "safety switch", which will be integrated with a box and will have a lever on the side to operate the switch. They're much larger and more expensive, and they aren't needed for basic things like we're discussing here, but may be required by code in some situations for other things.

            BTW, while it costs a little more, they make a version of this switch with a red toggle that can help make it obvious it's a special-purpose switch. I like to color code certain things like this to make things simple.


          6. Expert Member
            MALCOLM TAYLOR | | #33


            Thanks! I wish more appliances had external switches like that.

          7. Expert Member
            BILL WICHERS | | #34

            Malcolm, it is code here to include a shutoff switch for maintenance purposes within sight of many types of installed appliances. Steam humidifiers are one that comes to mind. I always install these switches even where not required, since they make service work MUCH easier. I like to install a switch near the dishwasher too, often in the basement next to the water shutoff for the dishwasher. This makes service work easy, and eliminates the need to go figure out which circuit breaker turns off the dishwasher.


Log in or create an account to post an answer.


Recent Questions and Replies

  • |
  • |
  • |
  • |