Small cabin cooling: how many kWh does your AC use after sunset?

For those cooling a small off-grid cabin, how much electricity does your air conditioner actually use between sunset and bedtime, or overnight? I'm especially interested in measured kWh over a known number of hours, rather than its BTU rating or the compressor's nameplate watts.

For transparency, I maintain SolarCalc, a solar sizing tool. I don't have my own measured cabin AC dataset; I'm trying to understand where simple sizing assumptions miss real conditions.

Here is a hypothetical example to make the question concrete: an AC averaging 600 W over six hours uses 3.6 kWh. With an assumed 90% inverter efficiency and 80% usable battery capacity, that alone calls for 5 kWh of nominal battery capacity (3.6 / 0.9 / 0.8). That excludes other loads, inverter idle consumption, and a reserve for poor solar weather. It is not a claim that any particular AC averages 600 W, or a complete system design.

If you have readings, what were the AC model, cabin size and insulation, approximate outdoor temperature/humidity, indoor setpoint, and hours measured? Was the energy measured at the AC outlet or on the battery side? That distinction matters to avoid counting inverter losses twice.

Has shading, insulation, or cooling earlier while solar is available noticeably reduced your nighttime battery use? Even one hot evening's measurement with that context would be useful.

Which solarcalc do you maintain? There is an app, and a .net . Are they the same?
These days I am usually using the tonnage rating as a wattage determination and assigning it a duty cycle based on perceived insulation of the dwelling. We dont worry as much about it here due to our much longer summer solar hours creating a glut for off grid locations which is perfect for AC. Most people are doing heat pumps here so the fun part become trying to figure out how much they can use it during the shoulder seasons.  I do very little net metering mostly off grid.

I don't have a day's kWh, but I've measured a small window unit and it came in at just under the rated wattage when the compressor was on. How often the compressor actually runs depends on many factors like heat differential, insulation, and so on of course.

In my experience, even an uninsulated house with drafty windows and a pretty suboptimal AC setup in terms of where the fan blows and so on, it only really runs continuously when it's over 90°F out (AC set to 70°C).

So assuming that it's hot cause it's sunny, and you can capture that sun with solar, I'm thinking ~400W of solar should be enough. Won't charge your battery much cause you're passing all of it through to the AC, but then 400W is pretty modest. This is for a very suboptimally-layouted, uninsulated space of about ~200-250sqft with a 400W window unit. (Old unit, too, maybe new ones are more efficient?)

David, thanks for asking: I maintain solarcalc.fyi, the website I meant in my first post. Sorry for the ambiguity. It includes affiliate links on some equipment pages; there is no need to visit it to join this discussion.

Your distinction between summer surplus and shoulder-season heating is useful. I would keep those as separate cases rather than carry a summer cooling assumption into a heating estimate. When you use tonnage as a starting point, do you also check the particular unit's electrical input at the expected outdoor temperature?

M Pearson, thank you for the measured compressor-on reading. Do you know the unit's model and roughly how many minutes per hour it ran during those conditions? I assume the 70 degree setpoint was Fahrenheit? That would help distinguish a 400 W on-cycle reading from the average over an evening. Also, a 400 W panel nameplate would not provide a continuous 400 W at the AC outlet, so I would not treat equal panel and appliance wattages as a complete sizing check.

Paul,
We heat and cool 2640 sq ft or 242 Sq M of very well insulated house at latitude 44.5 degrees using a heat pump central system. I am attaching three screenshots showing a yearly total, an hourly summary on a very hot July day (it was a record busting hot summer for us) and an instantaneous shoulder season heating scenario. The system uses a 2.5 tonne heat pump which when active consumes between 1.2 and 1.35 kWHrs per hour if it was running all the time. On our worst day duty cycles was about 33percent.  For the instantaneous reading shot I cranked the heating temp of the house up to 23 degrees with an outside temp of 13 The house is currently at 21 and I would not usually do that. That does not quite  duplicate what we see as a differential in summer time but its the best I can generate.  Average set point for daytime summer is 23, nighttime 21, winter average is 21 day 19 night. Heat pump heating totals in winter are not accurate since we use some radiant electric at nighttime on nights colder than -15C. Trying to peg down exact efficiency is a near impossible science as energy modelling of this house proved to be extremely inaccurate. Its an ICF build with triple pane windows and extra slab foam so did not fit well into the software so it outperforms. Hope some of this interests you and does not divert from the original question.
Cheers,  David

No AC here in Finland, so I can't add measured numbers, but one thing I'd stress for sizing is inverter overhead. Mine idles around 40W, so an all-night cooling run quietly tacks on another 0.3-0.4kWh that people often forget to budget.

Also, duty cycle matters more than rated watts. A small cabin AC mostly sips once the space is already cold, so pre-cooling during afternoon solar hours genuinely shrinks the nighttime battery number. If you're collecting real data, a cheap plug-in energy meter over a full week tells you more than any calculator.

Mikael Korhonen wrote:No AC here in Finland, so I can't add measured numbers, but one thing I'd stress for sizing is inverter overhead. Mine idles around 40W, so an all-night cooling run quietly tacks on another 0.3-0.4kWh that people often forget to budget.

Also, duty cycle matters more than rated watts. A small cabin AC mostly sips once the space is already cold, so pre-cooling during afternoon solar hours genuinely shrinks the nighttime battery number. If you're collecting real data, a cheap plug-in energy meter over a full week tells you more than any calculator.

That is such a great point Mikael; using the mass of your home  as a battery extension by over cooling your space in the high sun hours makes a significant difference.

I just open the windows after sunset.

Burra Maluca wrote:I just open the windows after sunset.

In my experience this tends to not be enough. Whenever I try this the house doesn't cool down enough to be comfortable until well after 11pm or sometimes even midnight. Tried it in several houses. I don't know if that's just heat stored in the house's mass or what but the AC takes several hours to cool down everything in the house to compensate. Hence running it while the sun is up (free cooldown if running on solar).

David Baillie wrote:On our worst day duty cycles was about 33percent.

David, thank you for the detailed conditions. Your written figures give a useful illustration: if the 1.2-1.35 kW running input and 33% duty cycle refer to the same measurement period, that works out to roughly 0.40-0.45 kWh per elapsed hour for that running load. That is quite different from budgeting the full running input for every hour. I would not extend that average to a whole night without the corresponding nighttime reading.

Does your heat-pump energy measurement include the indoor air handler/blower as well as the outdoor unit? That boundary would help when translating it into a battery budget. Your note that the winter totals exclude some supplemental resistance heating is also important: those totals cannot represent the house's complete heating electricity use.

The insulation and latitude context makes this useful, even though it is a larger house than a cabin. I would keep it as a well-described individual case rather than scale it to cabins by floor area alone.

Paul Grace wrote:

David Baillie wrote:On our worst day duty cycles was about 33percent.

David, thank you for the detailed conditions. Your written figures give a useful illustration: if the 1.2-1.35 kW running input and 33% duty cycle refer to the same measurement period, that works out to roughly 0.40-0.45 kWh per elapsed hour for that running load. That is quite different from budgeting the full running input for every hour. I would not extend that average to a whole night without the corresponding nighttime reading.

Does your heat-pump energy measurement include the indoor air handler/blower as well as the outdoor unit? That boundary would help when translating it into a battery budget. Your note that the winter totals exclude some supplemental resistance heating is also important: those totals cannot represent the house's complete heating electricity use.

The insulation and latitude context makes this useful, even though it is a larger house than a cabin. I would keep it as a well-described individual case rather than scale it to cabins by floor area alone.

That is a great point a out the air handler. In my case it adds 90watt hr per hour of runtime. I took a look at the site Paul. The company I used to work for had a detailed spreadsheet for calculating usage and peak watts. We would walk them through their load profile, build a model and custom design for them. It took time and we would charge for it. The pace has accelerated around solar since then. These days my clients seem much less interested in modelling and more in maximizing what we can deliver based on their budget. The site would be useful to them. I would like to see a much more populated library of loads though.
Cheers, David

Hi Paul,

That’s a really difficult question to answer.  And it is highly specific to your home’s exact setting.

The first house my wife and I lived in had absolutely no western daytime protection/shade.  One of the first things we did as homeowners was to plant backyard shade trees, but really it was other owners who got the benefit of those.

In July and August the sun absolutely beat through our bedroom wall (which was an exterior west facing wall).  And being a spec house, the insulation was marginal at best.  I couldn’t believe how HOT that wall would get!  Even 2 hours after sunset, the wall still radiated heat into the house!

Our current house was specifically built with trees to the west to prevent that intense heat that poured through.

My point is that it’s is going to be a bit difficult to get any kind decent estimate.  Our area is hot and humid which doesn’t help.  But up north it seems like the moment the sun sets the temperature plummets.

I wish I had better news for you,


Eric

David Baillie wrote:I would like to see a much more populated library of loads though.

David, that is specific and useful feedback. Which three to five missing loads would make the biggest difference for the people you help? For example, is the gap well pumps, air handlers, water treatment, communications equipment, or something else?

I would want each entry to distinguish running power, starting demand where relevant, and measured daily energy or hours of use. A generic preset should be clearly labelled as a starting estimate and remain editable, rather than imply that every model uses the same power. Your budget-first point is useful too: separating essentials from optional loads would help show what a constrained system can cover.

Thanks also for separating out the blower. If its 90 W applies only while the compressor runs and the same 33% duty cycle applies, that would add about 0.030 kWh per elapsed hour. If the fan continues between compressor cycles, its own runtime needs to be counted instead.

Eric Hanson wrote:Even 2 hours after sunset, the wall still radiated heat into the house!

Eric, that is useful information rather than bad news. It highlights why a cooler outdoor temperature after sunset does not necessarily mean the cooling load has already fallen. For comparing readings, I would record west-facing exposure/shade, insulation, humidity and the indoor starting temperature alongside the measured hours and kWh.

I would keep a hot, exposed evening and a shaded evening as separate cases rather than combine them into one average based on floor area. Your example also supports measuring from before sunset through bedtime: a short compressor-on reading would miss how long the house continues releasing stored heat. Thank you for explaining the contrast between the two homes.

Paul,

I am glad it was helpful!  I loved that little house, but I learned there were several things I would have changed.  Noteably that would be the insullation!  But anything that could cast shade would have done wonders as well.

NOTE:  We have very hot and humid summers, so consider that (Southern Illinois).


Eric

With our 8000BTU wall unit, starting around 2:00pm, I run it full blast during the day, getting the upstairs as cool as I can, to ~70-72F.  With both South and West-facing solar arrays, I can keep the wall unit running with zero battery depletion till about 6:30pm, at which point the solar resources drop down below the ~750W (maybe 1000W total) needed to run the unit.

At that point, I start adjusting the temperature up to ~74-75F so the unit switches over to cycling on and off because it's reached it's set temperature.  What I've found is if the upstairs is already pre-cooled, I'm only keeping the unit turned on till about midnight.  At midnight, the outside air temperature has dropped below ~74F, and I simply turn off the air-con, open the windows, and run an electric fan instead.  Because the air-con is a wall-mount, it does not interfer with windows opening and closing.

So, no matter how many kWh of ac power I consume during the daytime, I'm only consuming about 2-3 kWh of power till sunrise the next morning.

No matter how hot it's been during the day, I have not yet had to run the air-con after midnight.  This might not be the case though if I was in a more humid SE US location?

Michael, everyone,

Good points about the pre-cooling in the day since you have solar to make it work.  Also good to know about the temperature information that you have for your overnight lows being in the low 70s or lower.

I have not yet seen where the OP stated his overnight lows so that can be make-or-break.  

Perhaps I have a sort of worst-case scenario where my overnight lows can be in the mid 80s and the humidity intense.  There is commonly about a 6-week period in the summer where the temperature is just plain ugly.  Upper 90s highs and mid 80s lows and the humidity just won’t quit.

If you have circumstances different than me then by all means take advantage of them!


Eric

It's probably good to just look at average temperature. If the average doesn't go much below 75-80°F, no amount of trickery/cycling is probably going to do much and you just need to brute-force it with AC. Especially when in a humid climate, some tricks can be done in arid climates by adding moisture to the air I suppose.

But if you're in the South, probably no way around a big solar array and battery storage.

Eric Hanson wrote:I have not yet seen where the OP stated his overnight lows so that can be make-or-break.

Eric, I should clarify that I am not describing a particular cabin or my own AC installation. As disclosed at the start, I maintain a solar-planning website; this thread is collecting operating conditions, not sizing a system for an unstated personal location. There is therefore no single overnight low for my example.

Your mid-80s nights with high humidity belong in a different case from a night that cools enough to open windows. I would keep those cases separate and would not use the cooler-night example to promise the same battery requirement in southern Illinois. The useful comparison is measured energy over stated hours, alongside outdoor temperature and humidity, indoor setpoint, and whether fans or other loads are included. Thank you for catching the missing context.

Michael's numbers show exactly why pre-cooling changes the battery math so much — the heavy work happens while the panels are producing, so the battery only has to cover the overnight simmer. One trick from my own off-grid setup: a freezer makes a perfect opportunity load on a charge controller's diversion output, soaking up surplus solar that would otherwise go to waste once the batteries are full. Same principle applies to pre-cooling a cabin — the coldest thermostat setpoint should coincide with peak sun, not with bedtime.

Paul,

You are right--I missed that info in the beginning--oops!

Just pleased that anything I could offer was of any use.



Eric