Incoming cold water temperature: 11 Celsius (constant for the example).

Indoor temperature: 20 Celsius (constant for the example).

Let us have two choices:

  1. Tank-based water heater solution: has to heat up daily 120 L water from 11C to 60C. But although it has a little insulation, it losses heat at a rate of 20% per day (I have no precise info about this), so called "standby-loss"

  2. Tankless water heater solution: has to heat up daily 120 L water from 11C to 40C. There is no other heat waste.

The big question: how much energy and electricity does the two consume daily?

Didn't wanted to make it more complex, since the tank based solution is having electricity 30% cheaper vs. the tankless

And the tankless needs a very BIG circuit breaker.

  • 1
    Are split-system heat pump water heaters (often called "EcoCutes") a thing where you live? Jul 16, 2020 at 11:51
  • Heat pump water heaters are much more efficient but don’t recover as quickly as tanked , review the comments on electric tankless most services will need to be upgraded and in colder climates I find additional point of use heaters are required. I had a customer install a new service, purchase the largest tankless he could find added point of use units to supplement the tankless and after ~10k spent went back to a tanked water heater. I have had gas tankless that worked well but there are limits.
    – Ed Beal
    Jul 16, 2020 at 13:44
  • 3
    If you are using water at 40C than you are not going to be using the same volume of water at 60C, since it will be mixed with cold water for non-scalding use. So that's a "method to rig the numbers falsely in favor of tankless."
    – Ecnerwal
    Feb 12, 2022 at 15:46

5 Answers 5


Real-world "standby loss" example with measured data:

Tanked Electric water heater (made in 2021) - 38 gallon nominal size, 35 actual gallons, (132.5 liters) set to 140°F/60°C (tempering valve on output but does not matter for standby losses.) Came with an external fiberglass wrap from the factory as well as the internal foamed insulation. Pipes are insulated.

Presently on but with no use due to work travel, so the standby is right there to be seen on the power use graph. (Why not off? well, we just had a lovely ice storm and power outage and if we get another, then not freezing the water heater beats saving less than 20 cents per day on standby losses. Allows much more time to get the problem resolved if the water starts from hot.)

Kicks on very noticeably in the usage graph about once every 20-24 hours and consumes about 0.8kWh (by comparison to the hours before and after it kicks on - heat pump heating and refrigerator are the other background loads, and neither has the solid bump of a 4.5 KW element engaging for 10-15 minutes that "spikes the graph" as it does.) Updated to include the "15-minute" version of the power use. The first and last ones are clearly spread across two intervals. Second might be. Variation in background is primarily outside temperature influenced.

Usage Graph

Spike times (above 0.5kWh/15 minutes)

  • 2-09-17:15
  • 2-09-17:30
  • 2-10-15:15 (21:45)
  • 2-11-14:00 (22:45)
  • 2-12-12:00 (22:00)

How that compares to use, "percentage wise" depends how much is used, percentage-wise. But that's a hard number for the often wildly estimated standby loss on a well-built (but not fanatically superinsulated) modern tanked heater.

For your use-case, if the tanked heater is 30% cheaper power cost, it will be a win unless you hardly ever use hot water at all. The kWh for resistance-heating the water is exactly the same for the same water use at the same temperature (regardless of storage temperature, since showering in 60°C water will send you to the hospital with second-degree burns, so you won't be doing that, actually.)

  • Nice answer based on a real life measurement. Perhaps add the graph (preferred!), or otherwise could remove any references to it.
    – P2000
    Feb 12, 2022 at 16:37
  • 1
    The easy graph is by the hour for a single day. I'll bother to download the 15 minute data that's available, but only by download, and see if I can get that into a nice graph that covers more than one day, then edit.
    – Ecnerwal
    Feb 12, 2022 at 16:44
  • 1
    Excellent data. 0.8kWh/day is $30 to $60 per year depending on where you live. What's the price of comfort? Should grinding out the maximum possible efficiency to save $45/yr take a back seat to choosing the system that best meets your lifestyle and needs? The huge differences in the way these systems behave seem to overshadow such a small difference in operating costs. What this answer says to me: Pick the one you love.
    – jay613
    Feb 12, 2022 at 17:18
  • I probably could have supported an electric tankless, but it would have meant spending a bunch of the power budget on that, and a large expense in cables/breakers/etc. for small savings. They don't make a hybrid heat pump that fits my space, (plus it'd be into loading my heating heat pumps in the hard part of the year) and straight heat pumps that will do DHW are absurdly priced. The tanked electric gives me a tank I can dump eventual solar into, fits my space, and fit my budget (payback on the fancier systems requires assumptions about trouble-free lifetimes that don't seem realistic.)
    – Ecnerwal
    Feb 12, 2022 at 18:34
  • What's the time scale of the graph? It's impossible to read, even when embiggened.
    – FreeMan
    Feb 14, 2022 at 16:39

It takes 4.18kJ to raise 1kg (or litre) of water by 1 degree Celsius.

From this, we can see that:

  • The tank system uses 4180Jx120kgx49C=24.5MJ, or 6.8kWh, to heat the water.

  • The tankless system uses 4180Jx120kgx29C=14.5MJ or 4kWh.

Estimating loss is harder. If you mean that it loses 20% of the heat stored within per day, we would need to know the tank size. Assuming 10C loss per day, and a 180C tank, it's 2kWh per day.

However, your maths is unlikely to be correct. When you use hot water, the 60C supply temperature instead of 40C doesn't actually change energy usage much - you use the mixer tap to reduce it down to whatever is necessary, so you use less hot water with a higher supply temperature.


The danger of legionella is a very important point to consider. And there is the possibility to install a counterflow heat exchanger, e.g. DIY -ones as proposed by Rob the plumber on YouTube.

From real experience with both types: A tankless heater with heat exchanger needs ca. 35% less energy (compared to tank heater with heat exchanger), is always ready for use (coming back from business travels, holidays etc.), has lower danger of growing germs, but may need rewiring/new breakers. A tank heater can be adapted to the wiring system (heating during night time) and to special lower tariffs (night), but needs a regular heating up to ca. 65 degree Celsius to destroy legionella. The tank insulation can nearly always be improved by old clothes, mineral wool etc.

  • Note also that electric tankless heating is storing power essentially as spinning reserve at the electric utility instead of using cheap baseload energy. Jul 16, 2020 at 11:51
  • @ThreePhaseEel Best way would be a heat pump and a photovoltaic system plus a storage tank with 3 to 10 in and outputs (temperature layers), but in many countries that does not pay off. One important fact not yet mentioned is the loss of energy by micro- or in-tube-circulation. Many running systems do not have a proper backflow preventer and/or a thermo- syphon in order to avoid the heated water to crawl out of the tank.
    – xeeka
    Jul 17, 2020 at 13:31
  • Yeah, the heat traps on most new water heaters in North America are kinda meh at best from what I've heard Jul 17, 2020 at 13:44

The ruling issue is standby heat losses.

For instance, our water heater at the lodge is on a timer. 36 hours after the timer ran out, I had occasion to check it. It was tepid. So that means it fell from 60°C to 30°C in 36 hours, or 0.8°C per hour.

*Except remember: thermal losses are proportional to the difference in temperature, and ambient was 25C. So, skipping a bunch of math, let's just say our losses at operating temperature were 1°C/hr. (closer to 1.75°C according to comments below).

That 30 gallon tank is about 120 kg, so that means we were wasting 4180 J/deg/kg x 120kg x 1°/hr.

Multiply All that and cancel the units and we get 501,600 joules/hour. Or 139 joules/second, which is watts. (or by the other math, more like 850,000 j/hr or over 200 j/s=watts).

The difference is that (our roached old) heater wastes 139 watts when it is in standby. That makes sense; that is simply a matter of insulation losses. A tankless heater would waste 0 watts while in standby.

Our rule of thumb is that a 1-watt load costs 1 USD per year. Probably works for Euros and quid, too. So about $139/yr or about $12/mo. (by the better math: over $200).

Another factor: Piping distance

There will be some length of piping between the water heater and the spigot. After some minutes of disuse, the water in that pipe will have cooled down to ambient. So the common practice is to flow the hot water valve until the water actually warms up, then set to your shower, washing, whatever. When one is finished, one abandons the hot water in the pipe. This cools down within minutes even if insulated. This is a total energy loss. Worse if someone enlarged the pipes for better flow, since more volume of water is abandoned.

This problem invites a wasteful workaround: "recirculation systems" which pre-load the pipes with hot water. That means fast hot water but it also means much greater heat losses.

What does this have to do with the tankless question? Real simple: Electric tankless units are very compact and do not have flues - which means they can fit in all sorts of places a 30-gallon tank would not. That means the tankless(es) can be moved much closer to the points-of-use, and that means less abandoned water and more convenience. At extremes, the British put their "electric shower" heaters right at the showerhead - and those units only take 8500-10,500 watts (35-45 amps).

  • the cooling curve would be exponential so your 0.8°C per hour will be inaccurate. You went from deltaT of 35° to 5° in 36 hours. If my math is correct that translates to 5% deltaT per hour or 1.75°C lost in the first hour. A little bit more than your number. Jul 16, 2020 at 12:33
  • @ratchetfreak yeah, I am not doing the math I should there, but the numbers were unbelievable enough as they were. Jul 16, 2020 at 13:19
  • 2
    But the high cost of “flash heating” and the water in the unit cools when not in use so it’s not actually 0 losses.
    – Ed Beal
    Jul 16, 2020 at 13:47
  • We do put them right at the shower head (1m hose for mine), but I will note that 8.5kW still isn’t really enough during the winter. The water gets hot enough, but at the expense of flow rate / perceived pressure. 10.5kW ones are a lot more expensive. FYI for anyone in the UK in 2022: a 1W load is close to £3/year. We sometimes do the same for taps where it would be hard to get hot water: toolstation.com/ferroli-cubo-undersink-water-heater/p72504
    – Tim
    Dec 3, 2022 at 23:41

Electric tank heaters commonly require a 240V 30A dedicated circuit. Electric tankless need much more power -- I've seen some that need 60A or more. That's not feasible in many homes.

If you have space for a tank (sounds like it since you're considering an electric tank), a better bet would be hybrid heat-pump water heaters. Existing ones still need 240V 30A because they have a backup electric resistance heating element, but newer ones are about to hit the market that dispense with that and only need a 120v circuit. As you may guess, that means recovery times will be slower which isn't great for large families.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Not the answer you're looking for? Browse other questions tagged or ask your own question.