How long will my fish tank stay warm in a power outage?
Longer than a small tank and shorter than the forums hope, and it is arithmetic rather than folklore: the water loses heat through the same glass, surface and evaporation the heater calculator sizes a heater against, and a liter of water holds 4,186 joules per degree. Divide the one by the other and you have the rate. In the winter case the chart below uses — 77 °F water, a house that has cooled to 59 °F with the furnace off — an open 10 gallon loses its first degree Celsius in about 31 minutes and is down to 68 °F in 3.5 hours. The open 75 takes 6.8 hours to get there, the 120 8.1, the 210 9.7. A tank does not cool at a fixed number of degrees an hour: the rate falls as the gap to the room closes, so the 75 is at 69 °F after six hours, 64 after twelve, and sits at the room's 59 °F by the next evening.
Two things the chart shows that no rule of thumb does. Same gallons, different hours: the 125 gallon cools to 68 °F in 7.1 hours against the 120 gallon's 8.1, because the six-foot tank has more glass around the same water — heat leaves through area, the same fact that makes watts-per-gallon the wrong unit for a heater. And the lid is worth more than the volume: a glass lid takes the 75 from 6.8 to 12.4 hours, because evaporation is 38 % of what an open tank loses; blankets over the top and sides take it to 44 hours, past two nights.
| Tank | Water | Open | Glass lid | Wrapped | First hour |
|---|---|---|---|---|---|
| 5.5 gallon | 5 gal · 19 L | 2.8 h | 4.8 h | 19 h | −4.3 °F · −2.4 °C |
| 10 gallon | 10 gal · 36 L | 3.5 h | 6.1 h | 23 h | −3.4 °F · −1.9 °C |
| 15 gallon | 14 gal · 52 L | 3.8 h | 7.0 h | 25 h | −3.2 °F · −1.8 °C |
| 20 gallon long | 17 gal · 65 L | 3.9 h | 7.3 h | 25 h | −3.1 °F · −1.7 °C |
| 20 gallon high | 19 gal · 71 L | 4.6 h | 7.7 h | 31 h | −2.6 °F · −1.5 °C |
| 29 gallon | 26 gal · 100 L | 5.1 h | 8.5 h | 34 h | −2.4 °F · −1.3 °C |
| 37 gallon | 33 gal · 124 L | 5.7 h | 9.0 h | 39 h | −2.1 °F · −1.2 °C |
| 40 gallon breeder | 42 gal · 159 L | 5.4 h | 10.5 h | 35 h | −2.2 °F · −1.2 °C |
| 55 gallon | 54 gal · 205 L | 6.1 h | 10.3 h | 41 h | −2.0 °F · −1.1 °C |
| 65 gallon | 65 gal · 244 L | 7.2 h | 12.2 h | 47 h | −1.7 °F · −0.9 °C |
| 75 gallon | 75 gal · 283 L | 6.8 h | 12.4 h | 44 h | −1.8 °F · −1.0 °C |
| 90 gallon | 86 gal · 326 L | 7.4 h | 13.1 h | 49 h | −1.6 °F · −0.9 °C |
| 120 gallon | 115 gal · 434 L | 8.1 h | 15.1 h | 52 h | −1.5 °F · −0.8 °C |
| 125 gallon | 112 gal · 425 L | 7.1 h | 13.3 h | 46 h | −1.7 °F · −0.9 °C |
| 150 gallon | 151 gal · 573 L | 8.6 h | 14.8 h | 56 h | −1.4 °F · −0.8 °C |
| 180 gallon | 180 gal · 680 L | 8.7 h | 16.9 h | 55 h | −1.4 °F · −0.8 °C |
| 210 gallon | 209 gal · 793 L | 9.7 h | 18.0 h | 62 h | −1.2 °F · −0.7 °C |
Standard tanks filled an inch below the rim, glass of the typical thickness for the height, 50 % humidity, ordinary room air, no pumps. Loss from the heater page's model (ASHRAE still-air film in series with the pane, 2 W/m²K through the stand, 9.5 W/m²K off an open surface, evaporation by the Carrier equation), integrated at the water's current temperature. "Wrapped" is a folded fleece blanket (1 m²K/W) over the top and all four sides; a bath towel is about a third of that. "First hour" is the initial cooling rate. Water only: glass and gravel add 4 % or so of thermal mass to a 75, in your favor.
The same open 75 has 9.7 hours to 68 °F in a room still at 64 °F, 6.8 at 59 and 4.6 at 50 °F. A summer outage in a warm house is an oxygen problem and not a heat problem at all; in a January ice storm the house itself is cooling toward the outdoors and the tank is following it. Enter the temperature the room will actually be, not the thermostat setting.
What to do in the first hour: air first, then heat
Oxygen runs out before warmth does in any well-stocked tank. Still water stops absorbing oxygen at the surface and the filter's bacteria, which use a lot of it, keep breathing. So the first move is a battery air pump — an airstone at the surface is agitation, which is what matters — or, failing that, scooping a jug of tank water and pouring it back from a height every half hour. Stop feeding: food is ammonia and oxygen demand, and digestion is both. Then the temperature: put the lid on, drape blankets or a duvet over the top and sides, and keep the room's door shut. Don't open the tank to check on it. Everything a blanket buys is in the chart above, and it is a lot: the open 10 gallon's 3.5 hours become 6.5 with a blanket over the top alone and 23 wrapped.
The filter media is the other worry. Left in a canister with no flow it goes anaerobic within hours and its water turns foul; that is a reason to rinse the media in tank water and restart the filter carefully when the power returns, not a reason to fear the bacteria have died. They go dormant and recover over a day or two; the nitrogen cycle guide explains where they actually live. Sponge filters and hang-on-backs stay wet and open to the air and come through an outage better than a sealed canister does.
How long will a UPS or battery run my aquarium?
Hours = the battery's watt-hours × what you can use, ÷ (the load ÷ the inverter's efficiency + the unit's own draw). The number on a UPS box is volt-amps, which is how big a load it can switch, and says nothing about how long; the battery inside says everything. A "1500 VA" unit holds two 12 V 9 Ah batteries — 216 Wh — and a "900 VA" unit two 12 V 7 Ah, 168 Wh. Lead-acid gives up about 80 % of that before it is spent, the inverter loses 15 % turning it into mains, and then the part nobody prints: a UPS on battery draws power just to run itself.
Bulk Reef Supply ran an 18 W powerhead on both units. The 216 Wh unit lasted 5.25 hours; the 168 Wh unit 5.75. Work the model backward and the two units' own draws are 12 W and 2 W — the 900 W inverter costs more to keep alive than its extra 48 Wh of battery is worth at an aquarium's kind of load. At 8 W of overhead, the model's default, a UPS spends 27 % of an 18 W job on itself. The same pump on EcoTech's DC backup, dropped to 20 % speed on battery, ran 81.5 hours from the same 216 Wh; the affinity law says the motor is drawing a tenth of a watt there and the driver electronics about 2, which gives 81 hours. Nothing to invert, nothing of its own to run.
What that means device by device, on the same 216 Wh battery, for the open 75 in the 59 °F room:
| Running | Draw | On a UPS | On a DC backup at ⅓ speed |
|---|---|---|---|
| Air pump (airstone or sponge filter) | 2.5 W | 15.8 h | — |
| Hang-on-back filter | 8 W | 9.9 h | — |
| Canister filter | 15 W | 6.7 h | — |
| DC return pump | 32 W | 3.8 h | 56 h |
| DC powerhead | 14 W | 7.1 h | 71 h |
| Heater | 200 W (flat out — the tank loses more than it makes) | 43 min | — |
216 Wh at 80 % usable, 85 % inverter, 8 W overhead for the UPS; DC: 80 % usable, no inverter, no overhead, pump at the affinity law plus a 2 W driver floor. The heater's draw is the tank's actual heat loss (323 W here), capped at its rating. Every figure is editable in the tool.
So: the heater is the one thing not to put on a UPS. It empties 216 Wh in 43 minutes and buys the 75 less than an hour of warmth that a blanket gives it for free. An air pump on the same battery runs 16 hours. A DC return pump turned down to a third runs 56 hours on a DC backup — days — which is why the reef-specific battery boxes exist and why the answer to "which backup" is a DC pump plus a DC battery, not a bigger inverter. Two D cells in a battery air pump are about 32 hours. And a generator runs the heater; nothing smaller does.
Will my sump overflow when the power goes out?
The display drains until two things stop it: the overflow stops taking water once the surface reaches the weir crest, and the return line stops back-siphoning once its outlet sucks air. Whichever is deeper sets the drop, and the drop times the inside footprint is the water that arrives in the sump. On the 120 gallon (48 × 24 in) a one-inch drop to the crest is 18 L, 4.7 gallons. Put the return nozzle three inches under — common, because it looks tidier — and the siphon keeps pulling after the overflow has stopped: 55 L, 14.4 gallons, with the return pipe's own contents added. A 36 × 18 in sump needs 1.9 in of freeboard for the first and 5.9 for the second, with a 15 % margin for the snail on the siphon-break hole.
The return nozzle's depth is usually the lever, not the overflow, and the fix costs nothing: a small hole drilled in the return just below the running surface breaks the siphon there. A check valve is the wrong tool — it fails silently, and it costs the pump three elbows' worth of head; the return pump page makes that case with the flow numbers. "Unplug it and watch" is still the right test once it is built; the arithmetic is for the day you are choosing a sump and setting its running level, when there is nothing to unplug yet.
Keeping the tank warm without power
- Blankets first. They are the biggest lever in this whole page: 6.5× the hours on the 75, 6.7× on the 10 gallon. Top and sides, leave the stand alone, and tuck them.
- Hot water, floated, never poured. A two-liter bottle filled at 140 °F carries 81 Wh into the tank as it equalizes — about 15 minutes of the open 75's 323 W loss, so you would need 4 an hour to hold it; on the wrapped tank, losing 50 W, the same bottle is 1.6 hours. Float sealed bottles or jugs; a gas stove or a camp stove can keep filling them. Pouring hot water in makes a hot layer and a temperature swing the fish did not need.
- Chemical heat packs taped to the outside of the glass under the blanket are a few watts each for hours — real, but small; count them as blanket, not as heater.
- Move the small tank. A 10 gallon's whole problem is its surface-to-volume ratio; carrying it (or its fish, in a bucket) to the warmest room in the house does more than anything you can wrap it in.
When the power comes back
Check the heater is under water before it switches on. Restart a canister only after tipping out its stale water and rinsing the media in tank water; run the return pump and watch the sump level come back up. Feed lightly for a day. And test ammonia the next morning: a filter that sat still for twelve hours, plus fish that were stressed and a tank that was cold, is the classic setup for a small spike — the free ammonia calculator turns the reading into whether it matters at your pH. If the tank cooled a long way, let the heater bring it back at its own pace rather than adding hot water; a few degrees an hour is fine, a jump is not.
Common questions
How long can fish survive without a filter? The filter is not what they run out of; oxygen and, later, ammonia are. A lightly stocked tank with a large surface is fine for a day with no filter and nothing else done; a heavily stocked one can be in trouble in hours without aeration. The battery air pump is the cheapest insurance on this page.
Should I do a water change during an outage? Not for temperature — you have no way to heat the new water and a cold change is worse than a slow drift. For oxygen, pouring water back in from a height does the same job with the water you have.
Generator, UPS or DC backup? A generator runs the heater and everything else; a DC backup runs one DC pump for days; a UPS runs an air pump for a night and is the wrong tool for anything hotter than that. Match the backup to the outage your area actually gets.