Round-trip efficiency: where the missing ten percent goes
Put 10 kWh into a home battery and you get about 9 back. This article follows the missing kWh through the inverter, the cells and the cooling fan, shows why the loss matters more on some tariffs than others, and explains how to check your own battery's figure from the numbers you already have.
What round-trip means
Round-trip efficiency is the share of the energy you put into a battery that comes back out. Charge with 10 kWh from the mains, discharge until the battery is back where it started, and measure what the house received. If it is 9 kWh, the round-trip efficiency is 90 %.
Tesla quotes around 90 % for the Powerwall, measured at the AC terminals under standard conditions. Real homes see something near that, sometimes a little worse, depending on how hard the battery is worked, how warm it is and how the energy arrived. Once you know where the 10 % goes, the variation makes sense.
Losses on the way in
A Powerwall stores energy as direct current in lithium cells. Your house runs on alternating current. Energy arriving from the grid, or from a separate solar inverter, has to be converted from AC to DC before it can be stored. That conversion happens in the Powerwall's own inverter, and no inverter is perfect. A few percent is lost as heat in the switching electronics and transformers.
Then the cells take a share. Pushing current into a cell meets internal resistance, and resistance makes heat. The faster you charge, the larger the loss, because the heat rises with the square of the current. A gentle charge at 1 kW wastes less in the cells than a hard charge at 5 kW, though the difference is small beside the inverter's share.
Losses on the way out
Discharging reverses the process. Current leaves the cells, meeting the same internal resistance, and the DC is converted back to AC for the house. The inverter loses a few percent again. An inverter is also least efficient at very low power: covering a 100 W standby load overnight costs a higher share of what is delivered than running a 3 kW oven does, because the inverter's own housekeeping is a fixed cost spread over less useful output.
Adding it up, a typical split for an AC-coupled home battery looks like this:
| Stage | Typical loss |
|---|---|
| AC to DC conversion on charge | 2–4 % |
| Cell losses on charge | 1–2 % |
| Cell losses on discharge | 1–2 % |
| DC to AC conversion on discharge | 2–4 % |
| Total round trip | Roughly 10 % |
These are typical ranges for home-battery equipment, not a specification for any one model. The loss is spread across four places, and the two conversions are the biggest.
Standby and thermal management
A battery also uses energy while doing nothing. The control electronics, the communications with the gateway and the cloud, and the contactors all draw a small continuous load, typically a few tens of watts. Over a day that is a few hundred watt-hours, which is small against 13.5 kWh of throughput but not nothing.
Thermal management is the variable part. Lithium cells are happiest in a narrow temperature band, and a Powerwall has a liquid cooling loop and a heater to keep them there. On a hot afternoon when the battery is charging hard, the pump and fan run. On a night below freezing, the heater runs before the battery will charge at all. A Powerwall on a north-facing outside wall in January spends more keeping itself warm than one in a garage. None of this appears as a separate line in any app; it simply widens the gap between energy in and energy out.
AC-coupled versus DC-coupled
How solar reaches the battery changes how many conversions it goes through.
Powerwall 2: AC-coupled
A Powerwall 2 has no solar input of its own. The panels feed a separate solar inverter, which turns their DC into AC for the house. Surplus flows into the Powerwall, which converts it back to DC to store, and later to AC again to use. Solar that passes through the battery makes three conversions.
Powerwall 3: DC-coupled
A Powerwall 3 has the solar inverter built in, with six DC inputs for up to 20 kW of panels. Solar arrives as DC and goes to the cells with only a DC-to-DC step to match voltages, then makes one conversion on the way out. That saves a full AC conversion for every solar kWh stored, worth a few percent on that energy. Grid charging still goes AC to DC and back, exactly as on a Powerwall 2, so a Powerwall 3 that mostly charges overnight gains nothing from the coupling. See Powerwall 2 versus Powerwall 3.
Why a small price gap makes it matter
Every kWh stored costs about a tenth of itself. If the energy was free solar that would otherwise have been exported, the loss costs you a tenth of the export rate, which on most tariffs is small. If the energy was bought from the grid to use later, the sum is different and worth doing carefully.
Buying a kWh at the cheap price and using it at the dear price only pays if the dear price is more than the cheap price divided by the efficiency. At 90 % efficiency, the cheap price has to be below 90 % of the dear price just to break even, and well below it to be worth the wear on the battery. With illustrative round numbers, not any real tariff: say the dear rate is 30p. A kWh bought at 27p returns 0.9 kWh worth 27p, so nothing is gained. Bought at 10p, it returns energy worth 27p, a clear win.
- Overnight tariffs with a deep cheap window (Go, Intelligent Go, the cheap bands of Flux and Cosy) have a gap several times larger than the loss. The efficiency barely changes the decision.
- Agile on a flat day, when the cheapest and dearest half-hours differ by only a fifth or so, can make grid charging worthless or slightly negative once the loss is counted. See Agile with a battery and when grid charging pays.
- Storing bought energy to export later is the hardest case. The export price is usually below the import price already, and the loss comes off the top of that.
The efficiency is the first number to apply to any charge-cheap-use-dear plan.
Measuring your own, and what the app already shows
The Tesla app's energy graphs report energy to and from the battery each day. Add up a month of both, divide out by in, and you have your own round-trip figure. A month averages out days that started full and ended empty, which would skew a single day.
Two cautions. Where the Powerwall's meters sit varies by installation and generation, and on some systems the conversion losses are booked outside the battery's in and out pair, so the ratio can read close to 100 %. That doesn't mean a lossless battery; the loss shows up elsewhere, as slightly more import or less solar than you would expect. And in winter the ratio drifts lower, for the heating reasons above.
What you don't need to do is subtract 10 % from anything the app shows. Every figure in the energy graphs is a measured flow: what the house drew, what the battery gave, what the grid supplied. The loss is already in there, as the gap between the totals, and so is any money figure worked out from those flows.
In SunHarvest
SunHarvest works the efficiency out from your own ledger, battery energy out divided by battery energy in over recent history, and falls back to the Powerwall's specification figure when the meters make the ratio read as no loss at all. That figure goes into the AI planner's context, with the rule that buying to store only pays when the later price beats the buy price divided by the efficiency, so the planner never books a round trip that loses money. The tariff comparison on the Tariffs page uses the same measured loss when it replays your history against each Octopus tariff, which is why a tariff with a shallow price gap ranks lower than its headline rates suggest.
General information, not financial or electrical advice. Tariff terms, prices, warranties and connection rules change and differ by supplier, region and installation; check the current documents for yours. SunHarvest is not affiliated with Tesla or Octopus Energy.