What size chiller do I need?
One that removes the heat actually entering the water, running about half the time on the hottest afternoon. That heat has three sources and one sink. A standard 120 gallon reef (434 L) at 79 °F in a room that reaches 82 °F gains about 50 W through the glass, bottom and surface — the same U·A the heater page uses, sign reversed — plus 155 W from its pumps (every watt of a submerged pump is heat) and 63 W from 250 W of LED over an open top; meanwhile the 3.8 liters a day that evaporate carry away 106 W. Net, 162 W — 554 BTU/h. To run it at 50 % that wants 1,108 BTU/h, so a nominal 1/10 hp unit (1,270 BTU/h), which would run 44 % of the time in that room with the lights on. The "(gallons × 10) × degrees" formula says 4,129 BTU/h for the same tank, because it counts volume, not surfaces, and knows nothing about the pumps or the evaporation.
Turn the pumps and lights off on the 120 above and the room's 4 °F advantage adds 50 W while evaporation removes 106 W — the water would settle below room temperature on its own (net -55 W). The chiller is there for the 218 W of pumps and lights, which is why the cheapest cooling in the hobby is a DC return pump run slower, a light schedule that skips the hottest hours, and an open top: put a glass lid on this tank and the load rises from 162 W to 242 W, because the lid keeps the evaporation in.
Can a fan cool my aquarium instead of a chiller?
Often, and the number is computable: a fan across the sump or the surface raises evaporation, and every liter that evaporates removes 0.68 kWh of heat. On the open 120 in a 50 % RH room a fan adds about 104 W of cooling for 3.7 more liters of RO a day — 64 % of the chiller's job for the price of a clip fan and a top-off reservoir. What decides it is humidity, and the forums have the limit wrong: a fan is not capped at room temperature, it is capped at the wet-bulb temperature, which in dry air is well below it. The tool finds where evaporation balances the room and the equipment with a fan alone:
| Room humidity | Fan adds | RO it costs | Water settles at (pumps & lights on) | Settles at (equipment off) |
|---|---|---|---|---|
| 30 % RH | 156 W | 5.6 L/day | 25.1 °C · 77 °F | 20.7 °C · 69 °F |
| 50 % RH | 104 W | 3.7 L/day | 27.1 °C · 81 °F | 22.9 °C · 73 °F |
| 70 % RH | 51 W | 1.8 L/day | 28.9 °C · 84 °F | 25 °C · 77 °F |
The 120 above in a 82 °F room, open top, fan on the surface. Red: the fan can't hold the 79 °F target with the equipment on. Evaporation from the same ±50 % model the ATO page uses — treat the watts as ±50 % too.
So in a dry house a fan holds this reef at target with room to spare; in a humid one it holds 28.9 °C (84 °F) and you are buying the chiller. Below about 150 W of net load the tool says "fan" outright. Two costs come with it: the RO water, and the humidity the fan puts into the room — which is the chiller's problem from the other side.
How are chillers rated — what does 1/10 hp mean?
Horsepower is the compressor's nominal size; the number that matters is BTU/h of heat removed, and it varies by maker for the same hp. Below is the table the tool sizes from — JBJ's published figures for the small units, the usual 12,000 BTU/h per hp above — with the watts each removes and the design load it carries at 50 % duty. Makers also print "up to N gallons"; that figure assumes their idea of a room, a lid and a light, and is the least useful line on the box. Titanium exchangers lose some of the nameplate; the duty target is the margin for it.
| Nominal | BTU/h | Removes | Design load at 50 % | Source |
|---|---|---|---|---|
| 1/15 hp | 650 | 191 W | 95 W | typical nameplate |
| 1/10 hp | 1,270 | 372 W | 186 W | JBJ Arctica DBA-075 |
| 1/5 hp | 2,400 | 703 W | 352 W | JBJ Arctica |
| 1/4 hp | 3,000 | 879 W | 440 W | JBJ Arctica DBE-200 |
| 1/3 hp | 4,000 | 1,172 W | 586 W | JBJ Arctica |
| 1/2 hp | 6,000 | 1,758 W | 879 W | 12,000 BTU/h per hp |
| 3/4 hp | 9,000 | 2,638 W | 1,319 W | 12,000 BTU/h per hp |
| 1 hp | 12,000 | 3,517 W | 1,758 W | 12,000 BTU/h per hp |
1 W = 3.412 BTU/h. Highlighted: the 120's unit. Pull-down for that pair — 434 L down 3 °C on 372 W of cooling against the 162 W load — is about 7.2 hours; a chiller is a thermostat, not a sprint.
It moves heat, it doesn't destroy it: everything it takes out of the water plus everything the compressor draws comes out of the condenser as warm air. At a COP of about 2.5 — 2.5 W removed per watt drawn, typical of a small hermetic compressor with a titanium exchanger — the 120's 162 W load puts 227 W into the room, which then warms the tank a little more. In a closed cabinet that loop runs away: the unit short-cycles, the room gains, the exchanger fouls. Duct the exhaust out of the stand, or put the chiller in the next room; it is the single biggest thing between a chiller that runs 44 % and one that never stops.
How much flow does the chiller need through it?
A window, printed on the unit, and both edges matter. Too little flow and the water in the exchanger gets colder than the setpoint while the sensor at the outlet reads warm — the compressor runs on, the titanium coil ices, and the chiller's own thermostat can't see it. Too much and the water leaves before it has given up its heat, so the unit short-cycles against a return that barely changed. Feed it from a dedicated pump sized inside the window, or from a valved branch off the return line, never from the whole return: the return pump page's friction numbers apply to the chiller's hoses too, and a 1/10 hp unit's barbs are usually ¾ in, which is the size that costs head fastest. Put the temperature probe in the sump or display, not on the chiller's outlet, and set the hysteresis wide — a degree or more — so the compressor gets long runs and long rests rather than short-cycling, which is what kills small units. The chiller's flow is not circulation: it does nothing for the corals and is not counted on the powerhead page.
How much does a chiller cost to run?
Its load divided by its COP, for the hours the load is there. The 120's 162 W with the lights on and 100 W with them off averages 121 W across the day — 1.2 kWh a day at COP 2.5, on the hottest days only. The electricity cost calculator takes it as a timed device at the hours it runs; the fan that does 64 % of the same job draws about 5 W.
Why is my tank too warm — and what to try before a chiller
In order of cost: take the lid off (the lid is worth 80 W on this tank); put a fan across the sump; shorten or dim the photoperiod through the afternoon; turn a DC return pump and the powerheads down — every watt of pump is a watt of heat — and move the lights an inch higher; cool the room, since the tank follows it; and only then size the chiller, to the load that is left. The tool takes the same inputs the heater page does: if the room is cooler than the target it says so and sends you back there, because the two problems are the same balance with the sign flipped, and a tank that needs both in one year is a tank in the wrong room.