Reef Tank Power Outage Priority: What to Run First When You Cannot Run Everything

EcoTech Marine, VorTech, Vectra, IceCap, CoralVue, Maxspect and Reef Octopus are trademarks of their respective owners; EmergencyPetPrep is not affiliated with or endorsed by any of them.

The lights in the house go out and the room goes quiet in a way that is specific to reef keeping. Not just quiet. The particular absence of a return pump, two powerheads, a skimmer running air through a neck, and the small mechanical noises of a wavemaker changing direction. And then, within about ninety seconds, the arithmetic starts: there is one battery in the cupboard, or one small power station, or one long extension cord to a neighbour who still has power, and there are six things plugged into the stand.

This page is about that arithmetic. Not what a reef outage does in general, which our aquarium and reptile power outage guide already covers in detail, including reef heater wattage, salinity creep, temperature margins and the battery air pumps that work identically in saltwater. This page is the narrower and harder question underneath it: when you cannot power everything, what do you power?

The short version is that the published answer is water movement, that the reasoning behind it is better than most people realise, and that the two best-credentialed sources on the subject decline to state it that plainly. All three of those things belong on the page. So does the fact that nobody publishes an agreed number of hours you have. The publishers who do give a number all land in the same few-hour band and then mean completely different things by it: one is counting hours until stress can begin, one until low oxygen becomes a problem, one until an outage is devastating, and one until fish and corals die. Four clocks wearing the same units is a worse situation than four different numbers would be, because it looks like agreement.

What This Page Answers, and What It Hands Off

The question you arrived with	Where the answer honestly sits
What do I plug the one battery into	Here, in the priority section and the order-under-scarcity section
How long can corals go without light	Here, with the peer-reviewed span, its ceiling, and the two scope caveats that travel with it
How many hours do I have	Here, as a disagreement between four publishers rather than a number, plus the professional reformulation that makes it a property of your tank
How long will a battery run my powerhead	Here, because a direct-current pump on a battery is the one case where manufacturers actually publish hours
How long will a power station run my heater, my chiller or anything through an inverter	Not here. That is alternating-current runtime and it belongs on our pet load power budget calculator, which publishes the manufacturer formula, the four documented reasons the real figure lands below it, and a standing refusal to supply your equipment’s wattage
What does a reef outage do to my tank generally, and what about the heater and salinity	Our aquarium and reptile power outage guide, which carries the reef failure chain, the heater wattage problem, the coral temperature discussion, both the cold and the hot halves of the temperature problem, and salinity creep
What order do I bring the system back up in after power returns	Not here. The sibling carries the restoration step that is time-critical for a reef system, in its salinity section, and its staged timeline is scoped to freshwater rather than to a reef restart; the full sequence is on neither page. The media-and-cycle restart sequence is a longer job than an outage page should carry, and this page adds only one restart-adjacent note, on lighting, which is that we could not locate a published photoperiod ramp
How do I move fish out if the outage is going to outlast my gear	Our fish evacuation and transport guide, which owns the bag-and-bucket numbers and the ammonia clock in transit

Two rules hold this page together, and they are the same two the sibling page runs on. Symptoms outrank clocks, always. And where this page joins two sources that never met, it says so in its own voice rather than dressing the join up as somebody’s published finding.

One more thing to state at the top, because it decides what this page is allowed to be. This site has no credentialed veterinary reviewer. What follows is sourced fact, manufacturer specification, and access-and-timing logistics. It contains no treatment instruction, no dose, no additive protocol and no threshold you should act on alone.

The Priority Question, and Who Actually Answers It

Search for a reef outage priority order and you will find a lot of confident lists. Very few of them cite anything. Here is the evidence, source by source, in descending order of authority, including the two sources that decline to give you the answer the rest of the page arrives at.

The accrediting body for North American public aquariums, which enumerates rather than ranks

The Association of Zoos and Aquariums publishes an accreditation standards document that its member institutions are measured against. The 2026 edition contains a standard written specifically about the class of enclosure a reef tank belongs to.

Standard 10.2.1 is the general life-support requirement:

“Critical life-support systems for the animals, including but not limited to plumbing, heating, cooling, aeration, and filtration, must be equipped with a warning mechanism, and emergency backup systems must be available. Warning mechanisms and emergency backup systems must be tested at least annually.”

Read the qualifier before you read the list, because it is load-bearing in the opposite direction from where a writer would want it: including but not limited to. AZA is explicitly telling you that enumeration is not exhaustive. An argument from what is missing from that list is therefore a weak argument, and this page is not going to lean on it.

The stronger one is the standard directly underneath it, which is scoped to exactly the enclosure in question. Standard 10.2.1.1 opens: “Enclosures (tanks) used to exhibit or maintain fish and/or aquatic invertebrates must have a warning mechanism to alert staff about critical life support failures in a timely manner.” Corals are aquatic invertebrates. Its Explanation, which is where AZA does the reasoning, is the sentence worth carrying:

“Aquatic systems with fish and/or aquatic invertebrates are particularly threatened by life support system failures. The inability of the system to maintain adequate water flow, oxygen, temperature, and gas saturation can result in catastrophic morbidity/mortality in tank occupants.”

Four parameters, named without an open-ended qualifier this time: water flow, oxygen, temperature, and gas saturation. Exhibit lighting is not one of them.

The sentence AZA prints immediately after that one is the one a reef keeper should take personally, because it describes a reef tank rather than a public exhibit: “Shallow, warm water, high biological loaded enclosures, such as stingray touch tanks, are especially vulnerable.” Shallow, warm and heavily stocked is a reasonable description of most reef displays, and it is the same variable Bulk Reef Supply names when it says oxygen falls fastest “especially in tanks with a higher bioload.” AZA then states what the monitoring has to achieve: protocols “must be developed in a manner capable of detecting system failures prior to the onset of untoward effects on the tank occupants.”

And on how that monitoring should be done, AZA states a preference before it states the fallback: “Automated systems are preferable, but not mandatory.” That is the accrediting body saying an alarm beats a person remembering to check, which is worth knowing before you decide your plan is to notice.

Now the discipline this page owes you, because an argument from omission is exactly the shape that has produced false claims on this project before. AZA does not rank those four. AZA does not say coral lighting is unimportant. And AZA addresses lighting twice elsewhere in the same document, both times in ways that are worth knowing rather than hiding. Standard 10.3.1 requires that “Lighting must be sufficient in all indoor facilities, including night houses, so that maintenance can be accomplished and animals can be observed. A means for emergency lighting must be available.” That is facility lighting, for seeing by, and it is a requirement. And Standard 1.5.7, in the animal welfare chapter, states that within an animal’s habitat “there should be particular emphasis on species appropriate physical environments including weather, temperature, sound, vibration, light, and air and water quality”, with an Explanation adding that “Animals should receive lighting suitable to their biology throughout the 24-hour and annual periods, with consideration for providing dark periods at night and attention given to animals and environments where light needs to be provided artificially.”

So the honest reading of AZA is this. The accrediting body treats species-appropriate lighting as a welfare standard phrased as a should, in a chapter about husbandry. It treats water flow, oxygen, temperature and gas saturation as the parameters whose failure causes catastrophic mortality in an invertebrate tank, in a chapter about equipment and emergency backup. Those are two different chapters, two different verbs, and two different failure modes. That is corroboration for a priority order. It is not AZA publishing one.

There is one more line in 10.2.1.1 that this page returns to later, because it is the grown-up version of the question everybody actually wants answered: “In those cases, in which manual monitoring is relied upon, the interval for system evaluation must be less than the survival time for enclosure inhabitants in case of a life support failure.”

The veterinary college that ranks lighting last, and names reef tanks as the exception in the same item

NC State’s College of Veterinary Medicine publishes outage guidance for aquariums and ponds, credited to Gregory Lewbart, a professor of aquatic, wildlife and zoologic medicine. It is the one credentialed source we located that puts lighting in an explicit last place, and its lighting item is four sentences long, so this page takes all four rather than the convenient one. It opens:

“Lighting: This is the least of your worries. Ornamental fish survive indefinitely without fluorescent or supplemental lighting.”

Then, without a paragraph break, it takes a reef tank back out of that ruling:

“More concern is with tropical reef aquariums where live sponges and coral are maintained.”

The sentence after that one says that many of these invertebrate species depend on bright light and on the symbiotic organisms that light supports. It is reported here outside quotation marks rather than reproduced, because the source prints reply where rely is plainly meant, and carrying the typographical error inside the marks would be the same defect as altering it.

So read the item as two findings rather than one. A veterinary aquatic-medicine professor ranks lighting last among outage worries, which is the strongest credentialed corroboration on this page for the order below. And the same professor names a reef tank as the specific case where that ranking gets less comfortable, which is the same caution Hemdal and Yanong raise further down this section and the reason this page does not tell you darkness is free.

Three scope facts travel with everything this page takes from that document, and the first is the author’s own. In the sentences that introduce the guidelines, Lewbart states the scope himself, in both directions: “While this focuses on temperate pond and aquarium species like goldfish and koi, the basic principles apply for many species.” Read both halves of that sentence. Goldfish and koi are not corals, and he does not claim they are, but he also does not fence the document off from anything. The second: it is written about aquariums and ponds. The third: every biological figure in it, its temperature tolerances included, is stated about fish rather than about corals or live rock. This page therefore carries two things from it, both of which cost no power and neither of which is a number: the seasonal split in its temperature item, used in the scarcity order below, and this lighting item. It carries none of its figures across to a reef system.

The reef retailer, which publishes the instruction outright and also disagrees with itself

Bulk Reef Supply is a reef retailer rather than a scientific body, and it is a source the sibling page already cites. Its two outage articles, both credited to Robert Farnsworth, Director of Web Content, and both stating Last updated Dec 19, 2025, publish the priority instruction plainly.

On the mechanism: “The biggest threat during a power outage is the lack of gas exchange. When your pumps stop running, the water in your aquarium stops moving which then stops or drastically reduces the amount of gas exchange that can occur.” And on the consequence: “Without water movement, your fish will quickly breathe all of the available oxygen from the water and begin to suffocate.”

On what to back up: “the most important equipment to backup during a power failure is going to be a powerhead or a battery-powered air pump.” And, on the same page, a ranking of the equipment classes by what is at stake: “Temperature, water quality, and lighting really only become a threat during prolonged outages.”

Its companion article carries the two priority sentences that the preflight for this page flagged as a self-contradiction, and they are worth showing rather than smoothing, because the context resolves more of the apparent conflict than a bare quotation would. Under a heading about investing in an uninterruptible power supply, which is a limited-capacity battery, the instruction is:

“Prioritize running powerheads or air pumps to maintain oxygenation.”

The bullet directly beneath it adds the condition under which the list grows: “Larger UPS units can also run heaters for short periods to maintain temperature.” So the short list is not a claim that heat does not matter. It is what a small supply can carry.

Under a separate heading about portable generators, which can run for days on fuel, the instruction is:

“Prioritize running heaters, powerheads, and filters to maintain water quality and stability.”

The two sentences are not identical and the same document prints both. Reading them as scoped to their headings, so that the short list is what you run on a small battery and the longer list is what you run once you have a generator, is this page’s own reading of the source rather than something the source states. Bulk Reef Supply does not say that. But the two sentences do sit under those two headings, and lighting is absent from both.

That same article lists losing the lights as a real failure mode rather than a non-event, and this page is quoting the whole bullet rather than the convenient half:

“Loss of Light: Corals rely on proper lighting for photosynthesis. While a short outage won’t kill them, prolonged darkness can stress your corals and affect their health.”

And on the clock, the same page states: “Power outages can be devastating for saltwater reef aquariums, even if they only last a few hours.”

The hobby press, which gives the most explicit ranking of all three systems

In April 2016 a Reef Builders reader asked the question this page is built on, in almost these words: which aquarium systems should get top priority during a power outage, assuming it is temperature control, lighting and aeration. Jeff Kurtz answered it directly, and his answer is the most explicit published ranking of the three that we located:

“I agree with prioritizing temperature control and aeration/circulation, but I wouldn’t be overly concerned about the lighting unless you’re facing a prolonged power outage of at least several days. It won’t harm your fish to be in the dark during the outage—in fact, it will help keep them calm and less active so they consume less oxygen.”

On the corals specifically:

“As far as corals are concerned, while it’s true that many of them depend on appropriate lighting to sustain the symbiotic zooxanthellae residing in their tissues, most can easily tolerate a few days of reduced light or darkness. After all, we do ship them all around the world in dark containers with no problem.”

And the practical point that most priority lists skip entirely, which is that reef lighting is a large electrical load and a small battery cannot carry it anyway: “Also, reef-quality lighting draws a lot of power, so if you’re using an emergency backup power supply with limited capacity, running your lights will likely be impractical.”

This is hobby trade press, not a clinical or peer-reviewed source, and it should be weighted accordingly. It is corroboration for a conclusion the better sources support by other routes, not the foundation of one.

The two credentialed sources that name light and decline to rank it

Here is the part a page arguing for flow-before-light would be tempted to leave out, which is the reason it goes here rather than in a footnote.

Jay Hemdal has been an aquarist and diver at a large public aquarium since 1985 and curator of fishes and invertebrates at another since 1989, with more than two hundred articles and seven books to his name. His 2008 article on emergency protocols for home aquariums lists the criteria to manage in an emergency, and light is in the list:

“The life support criteria that must be managed during an emergency include dissolved gasses (primarily oxygen), light (for corals), temperature, and in long-term emergencies, nitrogenous wastes.”

Four criteria, no order given. He does not rank oxygen above light and this page will not put that ranking in his mouth.

He goes further than that later in the same article, in a sentence a page arguing flow-before-light would very much prefer not to print. Introducing his section on chemical additives, he writes: “For example, in a long-term power outage, fish life may be saved by some device, but the corals begin to die due to lack of light or some other factor.” Read what that sentence is and is not. It is a public aquarium curator naming lack of light as a thing that kills corals in an outage, which is the strongest statement against this page’s ordering that we found anywhere. It is also hedged in its own words, “or some other factor”, and it is scoped to a long-term outage rather than to the evening this page is mostly about. It sits comfortably beside a light clock that runs in days. It does not sit comfortably beside anyone telling you that darkness is free, and this page is not going to tell you that.

Dr. Roy Yanong is an Extension Veterinarian and a University of Florida Institute of Food and Agricultural Sciences Professor of Fish Health and Aquaculture. In a question-and-answer piece published by Florida Sea Grant on October 28, 2025, he was asked what is special about saltwater tanks in an outage, and his answer names lighting alongside oxygen rather than beneath it:

“Fish-only saltwater tanks usually have fewer issues. Reef tanks will require extra care and ideally a generator during a power outage, since more oxygen and lighting is needed for the coral and other animals.”

Note what the sentence actually recommends: a generator, which is a supply large enough to run both. That is not the scarcity case this page is about, and it is not an instruction to run lights instead of pumps. But it is a credentialed extension veterinarian putting oxygen and lighting in one clause, and any page that told you every expert ranks flow first would be misreporting him.

On the same reader’s freshwater tank, incidentally, Yanong takes the opposite view about darkness for the fish, and it lines up with Kurtz: “Fish can usually handle being in the dark during a storm, and it’s best to keep the tank or room dark to reduce stress and aggression for the fish.” The exception he names is photosynthetic life. Asked about aquarium plants, he says “If you have live plants, they do need light” and gives 6 to 8 hours a day as the target, adding that if you cannot provide any light at all it is best to remove them, because dying plants rot and pollute the water. That instruction is written about freshwater plants in a freshwater tank and this page is not going to extend it to corals, which are animals with a different biology and a different failure mode. It is quoted here because it shows exactly where an extension veterinarian draws the line: fish are fine in the dark, and the photosynthetic organisms are the ones the question is really about.

What this site concludes, marked as its own

Putting those four bodies of evidence together produces a conclusion none of them states in full, so here it is in this site’s own voice rather than borrowed from any of them.

Under scarcity, spend your power on water movement first. The reasons, in order of how much weight each one carries. AZA’s explanation of its own fish-and-invertebrate standard names four parameters whose failure causes catastrophic mortality and lighting is not among them, while AZA’s lighting standards live in a different chapter with a different verb. The peer-reviewed literature, in the next section, puts the oxygen injury on a scale of hours and the darkness injury on a scale of days. A veterinary college ranks lighting last in its own outage guidance, in the same item where it then names tropical reef aquariums as the case for more concern. Two reef publishers point owners with a limited supply at the pumps: Bulk Reef Supply prints its pumps-first instruction under an uninterruptible-power-supply heading, and Kurtz tells anyone on a backup of limited capacity that running reef lighting will likely be impractical. And that last constraint is the one that decides it in most houses anyway, because reef lighting is a large load and a small battery cannot run it.

Against that: two of the most credentialed sources on the topic name light as a criterion and decline to rank it, the veterinary college that does rank it last carves reef tanks back out in the same breath, and the sole retailer sentence about darkness says prolonged darkness stresses corals rather than saying it is free. The conclusion survives those, but it survives them as a judgment about scarcity, not as a claim that light does not matter.

The Two Clocks, Which Is Where the Reasoning Actually Lives

The priority order above is only as good as the mechanism underneath it. This is the mechanism, and it comes from two peer-reviewed papers that never cite each other.

The light clock runs in days, and it has a ceiling

In 2017, a team from the Australian Institute of Marine Science published a study in Scientific Reports that did the experiment a reef keeper would actually want done. They took adult colonies of two coral species, juveniles of one of them, and a crustose coralline alga, and held them under six light levels for 30 days. The lowest treatment was effectively total darkness.

What happened, in their results:

“Coral colonies in the lower light treatments gradually lost colour though time, with paling observed after 10 d in all groups”

at the lowest light levels, and by day 20 the corals in those treatments were, in the paper’s word, bone white. But colour loss is not tissue loss, and the paper is careful about the difference:

“A. millepora showed no signs of tissue loss regardless of light intensity, even after 30 d of exposure”

Of the two coral species, one did suffer partial mortality. The paper reports it in “juvenile and adult P. acuta”, adds that “P. acuta adults suffered from more partial mortality than juveniles”, and puts the adult tissue loss at 10 days, in the near-zero treatment and in the 0.1 treatment specifically rather than across the whole dark end. It also reports that the juvenile pattern was inconsistent across treatments and may not have been caused by the low light at all.

And now the organism this page would be leaving out if it stopped there, which is the pink and purple crust on the rock and the back glass. The experiment also ran the crustose coralline alga Porolithon onkodes, and the CCA came off worst of everything in the study. In the results: “P. onkodes exhibited the highest level of partial mortality”, with the paper putting the majority of it at 25 days in the darkest four treatments and none at all at the brightest. In the discussion the authors state it plainly: “CCA was more sensitive to the impacts of low irradiance than both adult and juvenile corals, and suffered from higher levels of partial mortality.” They also report that the darkest-treated CCA fragments “discoloured rapidly and had sections of pale tissue and sections of bone white skeleton where cells had been lost.” Twenty-five days is still a days-and-weeks clock rather than an hours clock, so it does not change this page’s conclusion. It does change the sentence people repeat, because the organism this study found most sensitive to darkness was not a coral at all.

There are two different paling figures in that paper and this page is going to show both rather than pick the convenient half. The results section, quoted above, puts paling at 10 days in all groups at the lowest light. The discussion section, describing the same experiment, says something faster:

“In this study corals began noticeably paling after 4–5 days and were heavily bleached after 10 days”

The paper prints both and never says which one supersedes the other. The words “in all groups” in the results sentence may be doing the reconciling, with colour loss starting in some colonies at four to five days and visible in every group by ten. Either way, plan against the shorter one. Four to five days is still far longer than a household outage runs, and treating the faster number as the operative one costs you nothing.

The same paper’s introduction is unusually useful, because it compiles the earlier darkness literature in one place. Across six studies:

Study, as the 2017 paper describes it	Species	What darkness did, and when
DeSalvo and colleagues	Acropora palmata, Montastraea faveolata	Colonies “becoming pale and eventually bleaching after 3–5 d in darkness”
Titlyanov and colleagues	Stylophora pistillata	Bleaching observed after 4 days
Hoegh-Guldberg and Smith	Stylophora pistillata	“observed bleaching of Stylophora pistillata in the dark after 10 d”
Franzisket	Pocillopora elegans, Porites compressa, Montipora verrucosa, Fungia scularia, held 60 days in darkness	“All colonies bleached within 10–20 d and there was no growth observed over the exposure period”; then, critically, “Pocillopora elegans died after 30 d while the remaining species survived over the exposure period”
Yonge and Nicholls	Lobactis scutaria, Psammocora contigua, Galaxea fascicularis	Discolouration in response to darkness over 18, 22 and 19 days respectively
Kevin and Hudson	Plesiastrea urvillei, a temperate coral	“lost algal symbionts after ~40 d in darkness”

Read the Franzisket row twice. Three of four species were bleached inside three weeks of total darkness and were still alive at sixty days. Bleaching is not death, and a page that reported only the bleaching figures would leave you with a worse mental model than the literature actually supports.

Two scope limits have to travel with all of it, and they travel in the same paragraph as the permission rather than three sections later.

The corals were deliberately not fed. The 2017 paper says so in its methods: “In this experiment we did not feed the corals in order to mimic the impacts of an offshore dredging scenario where heterotrophic feeding is less important”. The same paper names heterotrophic feeding as one of the survival routes corals use under prolonged low light, noting that under extended light reduction “some corals could survive by switching from phototrophic to heterotrophic feeding to maintain a positive energy balance”. A keeper who can still feed a tank is not in the position the experiment modelled.

These are two species of coral in laboratory tanks, under a controlled light gradient, for a stated purpose that had nothing to do with power outages. The study exists to set light thresholds for managing dredging near reefs. It is evidence about how corals respond to darkness. It is not a prediction about your tank on Tuesday.

The oxygen clock runs in hours to a day, and its threshold is low

In November 2021, a team led by Maggie Johnson at the Smithsonian Marine Station published the complementary experiment in the same journal. They took two Caribbean species from Mote Marine Laboratory nurseries and held them at four dissolved oxygen levels, from severe deoxygenation up to normal. One detail of the method matters for how far the comparison can be pushed: the two species were run in separate, sequential experiments of different lengths, and the paper’s own figure caption states them, with Acropora cervicornis exposed to the treatment conditions for 5 days and Orbicella faveolata for 11. So the eleven-day figure below belongs to O. faveolata alone, and A. cervicornis was never run out to eleven days at all. That does not weaken the within-a-day result, which happened inside the first of its five, but it does mean the two species were not watched over the same window.

The headline result, in their words: “Coral species demonstrated markedly different temporal resistance to deoxygenation”. Concretely, Acropora cervicornis “suffered tissue loss and mortality within a day of exposure to severe deoxygenation” at approximately 1.0 milligrams per litre of dissolved oxygen, while Orbicella faveolata “remained unaffected after 11 days of continuous exposure” to the same concentration.

Now the part that stops this from becoming a scare number. The two intermediate treatments, at roughly 2.25 and 4.25 milligrams per litre, barely did anything. The paper says they:

“elicited minimal responses in both species, indicating a low oxygen threshold for coral mortality and coral resilience to oxygen concentrations that are lethal for other marine organisms”

So the picture is not that corals die whenever oxygen dips. It is that they tolerate a lot, and then, in at least one species, fail fast once the concentration gets very low. That is a different and more useful shape than a linear decline, and it is a reason to keep the water moving early rather than to watch a number.

The method matters for scope. Oxygen was held down deliberately: “Oxygen concentrations were maintained in each independent tank by bubbling seawater with nitrogen gas and ambient air”. That is a controlled deoxygenation of a stirred, lit, temperature-controlled mesocosm. It is not a living room with the pumps off.

Why putting the two together is fair, and what it does not license

Here is the observation that makes the comparison worth anything, and it is this site’s own join rather than a finding either paper reports, because neither paper cites the other.

Each experiment held the other variable at normal.

The darkness study kept its water moving. Its methods state: “In tank circulation was maintained with a TUNZ pump (EcoTech Marine, PA, US).” Seawater was fed into each tank at a rate the paper describes as producing roughly six complete water turnovers a day. The corals that came through 30 days of darkness without tissue loss did so in a tank with a powerhead running in it and fresh seawater arriving continuously.

The deoxygenation study kept its lights on. Each treatment tank had an aquarium LED “programmed to simulate a diel light cycle over a 12:12 h photoperiod”, at an irradiance the paper says matched the nursery the corals came from and was sufficient to stimulate maximal photosynthesis. The corals that died inside a day of severe oxygen depletion did so with a full normal photoperiod overhead.

Read the two together and the asymmetry is stark. Oxygen with effectively no light bought a month in one species. A full photoperiod with the oxygen driven down to about 1.0 milligram per litre bought a day in another. Note the asymmetry inside that sentence, because it is easy to state it more dramatically than the papers do: the darkness treatment really was near-zero light, but the oxygen treatment was severe depletion and not an absence of oxygen. That is what the flow-before-light conclusion actually rests on.

Three things that comparison specifically does not license, stated plainly because the temptation runs the other way.

It does not license an hours-remaining figure for your tank. The oxygen figures are dissolved oxygen concentrations under controlled bubbling, not elapsed time with a pump off. Converting a mortality-within-a-day finding at about 1.0 milligrams per litre into an hours-remaining figure for your tank requires a model of how fast your specific tank falls to that concentration, and no source we located publishes such a model. Anyone who hands you that number built it themselves.

It does not license a species claim about your corals. The two studies used four species between them, from two ocean basins, and found that species differed enormously within each experiment. Acropora cervicornis died in a day of severe hypoxia while Orbicella faveolata was unbothered for eleven. Acropora millepora took 30 days of darkness without tissue loss while Pocillopora acuta lost tissue at 10.

And it does not license the idea that darkness is free. The 3-day figure at the short end of the compiled darkness literature is real, and the sentence Bulk Reef Supply publishes about prolonged darkness stressing corals is not wrong. The permission this page gives is narrow: hours of darkness during an outage are not the thing that is hurting your corals. That is a very different sentence from saying light does not matter.

The complication that actually strengthens the case

There is one finding in the 2017 paper that looks at first like it undercuts everything above, and on a careful reading does the opposite. Its discussion notes:

“Several microsensor studies have shown that when placed in darkness, coral tissue rapidly (within minutes) enters a hypoxic and then near anaerobic state”

and, immediately after:

“Although corals routinely enter hypoxia at night time, tissue oxygen concentrations also rapidly increase on exposure to light in the early morning”

The paper gives the mechanism in the sentence between those two, and the mechanism is the part that matters here. The hypoxia happens, it says, because of “high metabolic activity of the symbiotic dinoflagellates and polyp tissue, limiting the diffusive supply of” oxygen, in the paper’s phrasing, from the surrounding water through what it calls the diffusion boundary layer. The oxygen the coral cannot make for itself in the dark has to arrive from the water around it, across a layer at the tissue surface.

So a coral in the dark is not idle. Within minutes of the lights going out, its tissue goes hypoxic, because photosynthesis has stopped producing oxygen inside the animal while the animal keeps respiring. Every reef tank does this every night, and every reef tank recovers in the morning. And on how corals get through it, the same paper is candid: “How corals tolerate hypoxia is unknown, although symbiotic anemones have been found to survive through fermentation processes involving glycolysis”. It adds that such fermentation produces energy at roughly a sixth of the yield of aerobic respiration, and the rest of that sentence is the part that has to travel with it: the paper describes the fallback as “a short term, temporary energy source, but not over extended periods in low light”. Short term is the whole offer.

This site’s own reading of that, which the paper does not state because the paper is about dredging rather than about outages: on a normal night, the coral’s tissue goes hypoxic and the water around it stays oxygenated, because the pumps are running and the surface is being disturbed. In a power outage, the lights and the pumps go out in the same instant, so the coral loses its internal oxygen production and its external oxygen supply at the same time. The two systems that normally cover for each other fail together. That is not an argument for running the lights. It is an argument that darkness is the condition under which flow matters most, which is exactly the case for spending your one battery on the pump.

How Long Do I Have? Four Publishers, Four Answers

This is the question everybody actually types, and the honest deliverable is the disagreement rather than a number.

Every cell below is what that source’s own page states, read on August 19, 2026. Where a source does not address something, the cell says so rather than borrowing an answer from another row.

Source (read 2026-08-19)	Source tier	Hours before trouble	Days figure, and what it is conditional on	Ranks flow above light?
AZA, 2026 Accreditation Standards, 10.2.1.1	Accrediting body for North American public aquariums	Not addressed as a number. Instead: a manual monitoring “interval for system evaluation must be less than the survival time for enclosure inhabitants”	Not addressed	Enumerates water flow, oxygen, temperature and gas saturation as the parameters whose failure causes catastrophic mortality in a fish or invertebrate tank. Does not rank them, and does not address exhibit lighting in that standard
Jay Hemdal, reefs.com, 2008	Public aquarium curator of fishes and invertebrates	“Most marine aquariums develop low dissolved oxygen problems within a few hours of the power going out”	Not addressed	No. Lists “dissolved gasses (primarily oxygen), light (for corals), temperature” as criteria in no order
Roy Yanong, via Florida Sea Grant, October 28, 2025	Extension veterinarian, UF/IFAS professor of fish health and aquaculture	Not addressed as an hours figure	“For the first day or two, your system should be stable, especially with proper aeration.” Conditional on aeration, and stated in an interview whose saltwater content is one question long	No. Says reef tanks need “more oxygen and lighting” and recommends a generator
NC State College of Veterinary Medicine (Gregory Lewbart)	Veterinary college; professor of aquatic, wildlife and zoologic medicine	Not addressed as an outage clock. Its hours figures are temperature-tolerance windows plus one transport-survival window, all of them stated about fish, not time-since-the-power-went-out	Not addressed	Yes for lighting, and it is the bluntest ranking on this page: “Lighting: This is the least of your worries.” The same item then names tropical reef aquariums as the case for “More concern”, so it ranks lighting last and calls a reef tank the exception
Bulk Reef Supply	Reef retailer	Outages “can be devastating for saltwater reef aquariums, even if they only last a few hours”	“Your tank should survive for 2-3 days without the filtration, heater, or lighting so long as you have a powerhead moving the water”. Conditional on a powerhead	Yes, under its uninterruptible-power-supply heading: “Prioritize running powerheads or air pumps to maintain oxygenation.” A separate sentence under its generator heading lists heaters, powerheads and filters
CoralVue / IceCap	Manufacturer selling the battery backup	“Fish and corals can perish in as little as 4 hours without water flow”	Not addressed	Implicitly, by product category. This is marketing copy on a product page and the vendor’s interest runs toward a shorter clock
Shane Coleman, Reef Builders, October 24, 2023	Hobby blog; author bio states he is a full-time coral farmer in Cairns, Australia	“You’ve got at least 2-3 hours of a dead still tank before livestock can even start to show signs of stress”	Not addressed	Not addressed directly. Lists flow pumps, return pumps and heating or cooling as the essentials for a generator
Jeff Kurtz, Reef Builders, April 8, 2016	Hobby publication	The average marine aquarium can “shrug off” several hours without power, immediately qualified by tank size, stocking level, fish activity, ambient temperature and tank location	Not addressed	Yes, explicitly. Prioritise temperature control and aeration or circulation; lighting only matters at “at least several days”

What that table is actually telling you

Three things, and none of them is a number.

The 2-3 hour figure has one publisher, and two of the others say the opposite while the third is only compatible with it on a very tight reading. Coleman’s sentence is the only place we located it, it is stated without a source, and it is stated in a stronger form than a summary of it usually carries: the published wording is “at least 2-3 hours” before livestock “can even start” to show stress, which is stronger than a rounded summary of it would be. The sentences immediately after it are worth having too, because they are good advice and they show the author is not being careless in general: “The fish will most likely just hide when the lights go out. Corals might even go into feeding mode (but don’t feed them).” Hemdal and Bulk Reef Supply point the other way, and one of those is a public aquarium professional: a few hours is where Hemdal puts the onset of low dissolved oxygen and where Bulk Reef Supply puts devastation, while Coleman puts it before stress can even begin. CoralVue is the awkward one to characterise honestly, because four hours to death is not strictly incompatible with stress beginning at two or three, so it is listed here as a shorter clock from an interested party rather than as a flat contradiction. Our own sibling page already tells reef keepers to treat an outage as act-sooner rather than wait-and-see. This page publishes no grace window, and if you take one thing from this section, take that the number you were looking for does not have consensus behind it.

The two figures that do run into days both have the pump running. Bulk Reef Supply’s 2 to 3 days is explicitly “so long as you have a powerhead moving the water”. Yanong’s day or two is explicitly “especially with proper aeration.” It carries a second condition in the sentence before it, which is easy to drop: he is describing a tank where “if you’ve prepared well and aren’t feeding, checking water quality hourly isn’t necessary.” Prepared and not feeding are part of the figure. Strip the condition and you have not got a days-scale figure at all, you have got an unsourced one. This is the same finding as the two clocks section arrived at from the peer-reviewed side, reached independently by two publishers who are not scientists, which is about as good as corroboration gets in this category.

The professional version of the question makes it a property of your tank. AZA’s standard for accredited aquariums does not hand institutions an hours figure either. It tells them that if they are checking a tank by hand rather than by alarm, “the interval for system evaluation must be less than the survival time for enclosure inhabitants in case of a life support failure.” That is the same question turned around: the number you need is how long your animals last, and the requirement is that you find out before you need it. A home version of that is not a web page. It is a monitoring alarm, a tested backup, and knowing what your own tank looks like an hour into an outage, which is the last section of this page.

The observation that outranks all of it

Both professionals who address timing hand you a symptom rather than a clock, and they describe the same symptom.

Hemdal, on when to start emergency aeration, writes: “How long to wait is a matter of good estimating. Without a dissolved oxygen test kit, it may be a bit of a guess, although you can certainly observe the fish’s respiration rates, and don’t begin emergency aeration until a slight rise in their breathing rate is noted.”

Yanong, listing the signs of distress in fish for Florida Sea Grant, names “gasping at the surface, otherwise known as “piping”” among the common ones, and says that when a fish shows distress “that’s when you should test for ammonia or nitrite.”

Read Hemdal’s rule for what it is, because it is easy to read it as a grace period and it is not one. His reasoning is entirely about a finite battery: a battery has a fixed amount of energy, running it while oxygen is still at saturation spends some of that energy holding a level the animals did not need held, and delaying spends more of it where it matters. It is a rule about when to flip a switch on a countable resource. It is not a claim that the first hours are harmless.

The sentences he writes immediately after it settle that reading rather than leaving it to inference. He applies the identical logic to ice: “Another application of this idea is when using ice or other finite water cooling method. Delaying the use of ice until needed will extend the length of time that the aquarium’s water temperature is kept within a range that can be survived by the animals.” And he states the objective outright: “Notice that by initiating the emergency procedures only when absolutely necessary, the animals can be kept alive a bit longer.” Longer, not safer. The whole rule is about stretching a finite supply.

And it comes with an exception that covers most modern gear, which is this site’s own observation rather than something Hemdal writes, since automatic aquarium battery backups were not the norm he was writing about in 2008: the whole dilemma only exists for equipment you switch on by hand. A backup that senses the outage and starts itself has already made the decision, and CoralVue states that its unit “turns on immediately after a power failure to protect your aquarium”. If your backup is automatic, Hemdal’s timing question is not yours to answer. If it is a pump you plug in yourself, it is.

And here is this page’s own answer to it, stated rather than left hanging, because two sections of one page giving opposite instructions is worse than either instruction on its own. Hemdal’s rule is a good rule for the thing he was describing, which is a rescue you can only spend once: a bag of ice, a tank of air, a battery you cannot recharge until the grid comes back. It is not a good rule for the first pump. Bulk Reef Supply tells owners with a limited supply to run a powerhead, our own reef section on the aquarium and reptile power outage guide tells reef keepers to act sooner rather than wait and see, and the two clocks above say the oxygen side is the fast one. So: start the water moving, and use Hemdal’s watch-the-breathing rule to decide when to spend anything you cannot get back, not to decide whether to run the pump. If your one battery is genuinely a one-shot and its published hours are shorter than the outage looks like being, that is the case where his logic applies, and the honest version of it is that you are choosing which hours to cover, not buying safety.

And Yanong attaches something to those distress signs that this page is not going to drop, because it is the source’s own next sentence and it is the one that says what to do when watching is no longer enough: “If you notice these signs, seek advice from a veterinarian or an experienced aquarium expert.” He names two routes, and both of them are far easier to find on a quiet afternoon than at hour one of a blackout. This site has no veterinary reviewer and is not one of those two routes.

The Order, Under Scarcity

Here is the decision, assembled from everything above. Each item says what it is sourced to and what is this page’s own reasoning, because the ordering itself is ours.

1. Water movement, at the surface. Sourced: AZA’s four parameters, Bulk Reef Supply’s powerhead-first instruction, Kurtz’s ranking, and the two clocks. If you have exactly one battery and one thing to plug into it, this is the thing. A circulation pump or a battery air pump both do the job; the pump moves more water, the air pump uses less power and most models switch on by themselves.

And the version of item 1 that needs no power at all, because a scarcity page that only tells you what to plug in has skipped the case where there is nothing to plug in. Three of this page’s sources publish it. Bulk Reef Supply, under a heading about having a plan for water movement, writes “Use battery-powered air pumps or manually stir the water every hour.” Kurtz offers the same thing as a substitute for a battery pump: “you can use the very-low-tech aeration method of scooping out water and pouring it back into the tank at regular intervals.” Coleman is the most specific about technique and the most honest about what it costs you: “you can siphon water from the sump and pour it back into the display tank, letting it fall from a little height to make small splashes and oxygenate the water. This will need to be done almost continuously so make sure you’re up to the task of being a human return pump.” Note that the two who give an interval disagree with each other, from every hour to almost continuously, the third only says regular intervals, and none of the three sources any of it. Take the shape rather than the number: the target is the same moving, choppy surface described further down this page.

2. Temperature, but only if you can actually carry the load. Sourced: AZA names temperature in the same four; Kurtz ranks temperature control alongside aeration rather than below it; Bulk Reef Supply’s generator instruction adds heaters once the supply is large enough. The constraint is electrical, and it is the reason temperature sits second here rather than joint first: a reef heater is a load in the hundreds of watts, against a pump load small enough that CoralVue’s own two worked examples use a 0.2 amp draw and a 20 watt draw. Our aquarium and reptile power outage guide works that comparison through with manufacturer wattage figures, and it carries both directions of the temperature problem rather than only the cold one: the insulation and warm-bottle bridges that cost no power at all, and a summer half for the case where the risk is the tank getting too warm. Coleman adds a point about direction that is easy to get backwards, and his sentence is printed here whole because the half of it people quote is not the half that tells you what to do: “Hotter temperatures are more dangerous than cooler temperatures since dissolved oxygen is lower as temperatures become warmer so you probably won’t need to focus on heating too much unless you live in a very cold climate.” In a summer outage the temperature problem and the oxygen problem are the same problem, and Coleman’s own conclusion pushes heating further down a scarcity list rather than up it.

And here is the summer half on this page, rather than only as a handoff, because the paragraph above has just argued it is the more dangerous direction. NC State’s outage guidance splits its temperature item by season and gives the warm half its own no-power instruction: “During summer, remove anything from the water’s surface to increase the area for gas-exchange efficiency.” Further down the same item, in its list of what to do once the temperature moves, it adds: “If the water gets too warm, minimize or eliminate exposure to direct sunlight.” Neither costs a watt, and the first of them is the same surface-gas-exchange mechanism the whole of this page runs on, which is why it belongs in the order rather than only on a sibling. One cost travels with the first one, and it is this page’s own note rather than NC State’s: whatever you clear off the surface was also slowing evaporation, and the auto top-off that would normally replace that water is a powered pump and is off too, so this action and the water-level check further down are the same job rather than two. Read what follows each of them in the source before you take a number from either. The sentence after the direct-sunlight instruction gives a tolerance window in the high 80s or low 90s for a day or more, and its subject is tropical and temperate fish. The one after that names stirring the water with a whisk as a way to raise oxygen, in a sentence that reads as though a word has dropped out of it. This page takes the two no-power actions and leaves the window where the source put it, on fish.

One limit rides with any attempt to move a tank’s temperature at all, and it is the reason this page names no cooling technique beyond the two no-power actions above, one of which clears the surface for gas exchange and one of which takes away a heat input. In the item about warming a chilled tank, NC State publishes a ceiling on the rate: “A 10 percent water temperature change every hour or two should not endanger fish.” Read the subject of that sentence. It is written about fish, in a document written about aquariums and ponds, so this page reports it rather than extending it to corals and live rock, and it does not turn it around into a cooling rate. A fast correction in either direction is its own injury, and what the safe rate is for a reef system is a question for a reef-specialist retailer, a local reef society, or a veterinarian who sees aquatic animals, not for this page.

3. Filtration and the skimmer. Sourced: Bulk Reef Supply describes the loss as waste accumulation rather than immediate suffocation, writing that filtration systems “stop removing organic waste” and that this “allows ammonia and other harmful compounds to accumulate quickly.” Hemdal puts nitrogenous wastes in his criteria list with the qualifier “in long-term emergencies”. Both frame this as a slower clock than gas exchange. This page publishes no ammonia number for a reef tank; see the scope section below for why.

4. Light, last. Sourced: the whole of the priority section. Kurtz’s practical point is the one that usually settles it before the biology does, which is that reef lighting is a heavy load and a limited backup will not run it usefully anyway.

One more rule that is this page’s own, and it is about the shape of the decision rather than its order. Do not split a small supply across several loads to be fair to all of them. A battery driving one pump properly is doing a job the literature supports. The same battery split three ways is running a pump too slowly to break the surface, a heater too briefly to hold temperature, and a light nobody needed. EcoTech’s own manual states that running two pumps off one backup “will reduce the total run-time for each pump by approximately one-half”, which is the manufacturer version of the same arithmetic. Pick the job, do it properly, and add capacity rather than dividing it.

Runtime, Direct Current: The One Place Hours Are Actually Published

Our pet load power budget calculator is the site’s page for runtime arithmetic, and it exists partly to refuse to give you your equipment’s wattage. So why does this page carry any numbers at all?

Because a direct-current aquarium pump on a matched battery backup is a genuinely different case, and it is different in a way that shows up in what the manufacturers publish. There is no inverter in the path. Nothing is converting stored direct current into household alternating current and losing energy doing it. The battery drives the pump directly at low voltage, and because that path is simple and the manufacturer built both ends of it, two manufacturers publish hours here, where no manufacturer we located publishes hours for an alternating-current aquarium load. Say that precisely, because our own siblings do publish alternating-current hour figures: they arrive at them by dividing a pack’s watt-hours by a load and then applying a stated loss allowance, which is arithmetic the site performs rather than a number a manufacturer printed. The moment an inverter enters the picture you are back in the other page’s world, and this section says so at the end.

EcoTech Marine: a table, a footnote, and a warranty that argues with the table

EcoTech Marine’s Battery Backup manual carries the document code 041713 in its footer. Its front page prints this:

Backup Operation Time, HOURS*	1 UNIT	2 UNITS
MP10	72	36
MP40	36	18
MP60	20	10

with the footnote printed directly beneath it: *“Actual hours may vary due to operating conditions and battery cycle times.”

What a UNIT is, and the evidence, because it is possible to read this table exactly backwards. Page two of the same manual describes two setups. Under a heading reading TWO VORTECH PUMPS, ONE BATTERY BACKUP: “You may run two VorTech pumps off one Battery Backup, but doing so will reduce the total run-time for each pump by approximately one-half.” Under a heading reading TWO BATTERY BACKUPS, ONE VORTECH PUMP: “To double the run-time of one VorTech pump, connect two Battery Backups together in a daisy-chain fashion using the provided cables.” Two pumps halves the runtime; two backups doubles it. The 2 UNITS column is exactly half of the 1 UNIT column in every row, so UNITS means pumps. A reader who took UNITS to mean batteries would conclude that two batteries give an MP10 36 hours, when the manual means 72 doubled. That is a factor-of-four error in the wrong direction and it is worth the paragraph.

Date the table. The document code is 041713 and its rows name the MP10, MP40 and MP60. It predates the current VorTech generation. EcoTech’s own parts store, read on August 19, 2026, lists the Battery Backup under SKU BB18AH-US as a current, in-stock product and describes it as “Compatible with all models and generations of VorTech Propeller Pumps and Vectra Centrifugal Pumps” with a note that Vectra pumps need a separately sold booster. That is EcoTech’s compatibility claim in 2026, not a re-measurement of the 2013 table.

And read the manual’s warranty text next to the table, not in a footnote. The same document that publishes those hours also publishes this:

“There are many necessary conditions that aquarium livestock require to survive in a home aquarium which may not be met during periods where power is not being supplied to your aquarium equipment. As a result, a VorTech propeller pump operating in battery backup mode may not produce enough flow to prevent the death of aquarium livestock.”

That is the manufacturer stating, in its own product documentation, that a working battery backup running for its published hours may not be enough. It also recommends who to ask, in the same paragraph: “We strongly recommend that you consult an expert to determine the optimal battery backup and pump arrangement to improve the likelihood of successfully maintaining your livestock during a power outage.” And it states that runtime “will depend upon many variables such as age and usage history of battery, the condition of your VorTech pump, the speed that you run it at while in battery backup mode, the number of pumps that you run off of one battery backup, or the number of batteries that you use to run one pump” and that “These variables cannot be accurately predicted and testing your pump and battery system is a critical part of ensuring the safety of your aquarium against power outages.”

One more instruction from the same document, worth staging now: “We recommend testing your Battery Backup in action once every month.” The test it describes is to unplug both the backup and the pump from the outlet, then read the charge remaining on the pump’s controller. That takes two minutes on a quiet afternoon and cannot be run during the outage it protects you from. The manual also gives an operating temperature it wants the unit kept near, printed as 70 degrees Fahrenheit; the figure is stated here outside quotation marks because the manual renders the degree symbol with a masculine ordinal character rather than a degree sign, and reproducing it inside quotation marks would carry the typographical error.

CoralVue: a published formula, two worked examples, and a contradiction on the manufacturer’s own page

CoralVue’s product page for the IceCap Battery Backup v3.0, SKU IC-BTBK3, does something unusual and useful: under a heading titled How to Calculate Battery Run Time, it publishes the function rather than only the answer. Its stated capacity is 2.5Ah, and the page works two examples.

The current form:

“Using it with a pump that is drawing 0.2 Amps (0.2Amps = 200mA), the time that the battery will last is (2500) ÷ (0.2 x 1000) = (2500) ÷ (200)= 12.5 Hours.”

The wattage form:

“If instead of the current you have voltage and watts, then the formula is (mAh x Volts) ÷ (Watts x 1000) = (hours). If you have a pump that is drawing 12volts and 20 watts from the battery, it will last (2500 x 12) ÷ (20 x 1000) or 30000 ÷ 20000 = 1.5 hrs.”

The contradiction this page is disclosing rather than resolving. The same product page states in its highlights that “This new version is 12V like the original!”, states in its specification block that the main unit’s DC output is 12V at 2.5A maximum, and then states in a note further down: “Depending on the pump and pump controller you are using, the pump may not slow down during a power loss when the battery backup is activated, as you might expect with a 12V battery (this is a 24V battery).”

Twelve volts twice, twenty four volts once, on one page, about one product. And the choice is not cosmetic: the manufacturer’s own worked example runs the formula at 12 volts and gets 1.5 hours for a 20 watt pump. Running CoralVue’s own published formula at the other voltage its own page states gives 3 hours for the same pump. That doubling is this site’s arithmetic on the manufacturer’s published function, not a CoralVue figure, and it is here to show you the size of the ambiguity rather than to settle it. Do not average the two. Ask CoralVue which figure applies to your pump before you plan on either.

Two further limits. CoralVue’s formula assumes the full nameplate capacity is usable and that nothing is lost in the process, which is a ceiling rather than a promise; our pet load power budget calculator sets out in detail the four documented reasons a manufacturer quotient lands above the real figure, and every one of them applies here. And compatibility is narrow: CoralVue states the v3.0 works with “most 24V or less variable speed DC pumps on the market today” and names Maxspect Gyre, IceCap Gyre and Reef Octopus Octo Pulse pumps as examples, while stating in capitals that it is “NOT compatible with Reef Octopus VarioS Pumps, EcoTech pumps, Tunze, Jebao, or AI Nero pumps.” A battery backup that does not drive your pump is not a backup.

One row per issuer, no averaging

Issuer	What it publishes	Form	The caveat the issuer publishes with it
EcoTech Marine, Battery Backup manual, document code 041713	72, 36 and 20 hours for the MP10, MP40 and MP60 respectively on one backup; half those figures with two pumps on one backup; double with two backups on one pump	A table, with named pump models	“*Actual hours may vary due to operating conditions and battery cycle times”, plus a warranty statement that a pump on battery backup “may not produce enough flow to prevent the death of aquarium livestock”
EcoTech Marine, current parts store listing (read 2026-08-19)	No runtime figure at all	None. The listing publishes contents and compatibility only	Not applicable. The absence is the finding: the manual is the only EcoTech document publishing hours
CoralVue / IceCap, Battery Backup v3.0 product page	A formula and two worked examples: 12.5 hours for a 0.2 amp draw, 1.5 hours for a 20 watt pump at 12 volts	A published function you run on your own pump’s draw	The same page states the unit is 12V twice and 24V once, which doubles or halves every result. Compatibility is limited to most 24V-or-less variable speed DC pumps and expressly excludes several named brands
Every alternating-current path	No manufacturer we located publishes an hours figure for an aquarium load on an inverter	Not applicable. Our own portable power stations comparison does publish hour figures for aquarium loads, but it derives them by dividing capacity by load and applying its own loss allowance	See the inverter section below and our power budget page

And the moment an inverter enters, you leave this world

The reason the two manufacturers above can publish hours is that nothing is converting voltage. Put an inverter in the path, so you can run a household-plug pump or a heater off a 12 volt battery, and two things change at once: the arithmetic acquires a conversion loss, and the inverter itself becomes a sizing problem.

Be careful how you state the first one, because this site publishes two conversion-loss figures already and it would be wrong to tell you none exists. Our pet load power budget calculator reproduces a manufacturer’s own planning formula, which it renders as Jackery’s “Working Time = Capacity Wh × 0.85 / Operating wattage of the device”, and that page, not this one, is where that formula was read from its publisher. It is careful about what the 0.85 is worth: it is, in that page’s words, “the only conversion factor any of these manufacturers put in print in front of us, and it appears as a rule of thumb in a consumer article rather than as a specification for a named model.” Our portable power stations comparison works its own examples with a 10 to 15 percent inverter-loss allowance. So a planning figure exists and this page is not going to pretend otherwise. What the calculator page reports as missing is narrower and is the thing you would actually want: “What nobody we checked publishes is the actual figure for a given unit.”

Hemdal publishes the sizing rule: “these devices have power ratings for peak and continuous usage, so be sure that your expected total wattage will be less than 80% of the converter’s continuous rating.” He adds a specific one about the socket people reach for first: “It is often recommended not use an inverter rated higher than 300 watts if it plugs into a cigarette lighter socket as it may cause the vehicle’s fuse to blow.” And he notes that some equipment only works on what he calls true sine wave inverters, so check with your equipment manufacturer. Coleman makes the same point about generators from the other direction, writing that a pure sine wave unit lets you run controllers, LED lighting and aquarium controllers, and that without one “you run the risk of damaging your equipment.”

Note what Hemdal’s rule is and is not. It is a rule about not overloading the inverter. It is not a runtime figure, and his 2008 article’s dollar figures and capacity generalisations are old enough that they should not be relied on. For runtime on the alternating-current side, including the four documented reasons a calculated ceiling overstates reality, go to our pet load power budget calculator; for choosing hardware, our portable power stations for pets comparison has the manufacturer specifications.

The Highest-Value Free Move, and the Thing Nobody Attaches to It

If you read nothing else in this page and you already own a battery backup, read this.

EcoTech Marine’s Battery Backup manual closes its warranty section with an instruction that is not a warranty term at all. It is technique, published by the pump manufacturer, and almost nobody quotes it:

“Finally, we recommend that pumps used on the battery backup system during a power outage be placed as high in the tank as possible, so as to disrupt the surface of the water and improve oxygen diffusion as much as possible.”

It costs nothing, it takes thirty seconds, and it changes what the pump is doing. A powerhead down in the rockwork on battery power is circulating water. The same powerhead near the surface is doing gas exchange, which is the failure AZA and Bulk Reef Supply both name.

Two other publishers describe the same mechanism from their own side, without giving the instruction. Bulk Reef Supply: “Without flow, waste settles, corals cannot feed properly, and oxygen exchange at the water surface is severely reduced.” Yanong, on what good aeration looks like: “Good aeration happens when the surface of the water is moving, ideally with a turbulent or “choppy” appearance. Still water means poor oxygen exchange.” That gives you a target you can see. Not simply that the pump is on, but a moving, choppy surface.

Now the join nobody publishes, added here as this site’s own reasoning rather than as a manufacturer instruction. Our sibling page documents the salinity problem on a long outage: water evaporates, salt stays behind, and the auto top-off that normally replaces the evaporated water is itself a powered pump, so it is off too. The consequence for the instruction above is one nobody seems to have connected. If you set the pump as high in the tank as possible in hour one, the water level is falling underneath it from hour one. On a multi-day outage in a warm room, a pump positioned right at yesterday’s waterline can end up drawing air, cavitating, or running dry. So: set it high, and then check the water level rather than trusting the position. If the level has dropped, lower the pump to follow it, and read the salinity section of the aquarium and reptile power outage guide for what to do about the salinity itself when power returns.

The 12 Volt Car Battery and the Hazard Nobody Attached to It

The single most widely repeated do-it-yourself reef outage suggestion is a car battery and an inverter. Reef Builders publishes it in its simplest form, as a bullet under a heading about deep-cycle batteries:

“A relatively cost-effective solution that you can have on standby would be a 12v battery (the kind you’d have in your car) and a power inverter to scale the voltage up to the input requirement of the equipment you want to run.”

That is the whole entry. There is no safety note attached to it on that page, and this page is not going to reproduce the recommendation the same way.

Here is what other publishers say about the object being recommended. The US Occupational Safety and Health Administration’s construction standard on batteries and battery charging, at 29 CFR 1926.441, requires:

“Batteries of the unsealed type shall be located in enclosures with outside vents or in well ventilated rooms and shall be arranged so as to prevent the escape of fumes, gases, or electrolyte spray into other areas.”

and, in the subdivision immediately after it:

“Ventilation shall be provided to ensure diffusion of the gases from the battery and to prevent the accumulation of an explosive mixture.”

Read the scope before you read the requirement. That is an occupational safety standard written for construction workplaces. It is not a household rule, it does not apply to your living room, and it does not tell you what ventilation a single battery in a house requires. An agency threshold written for one setting is not a threshold for another, and this site has published that lesson before in other contexts. What the standard does tell you, from a federal safety agency, is what the hazard is: unsealed batteries emit gases, and OSHA’s own words for what those gases can form is an explosive mixture. The same standard also requires acid-resistant flooring unless the floor is protected from acid accumulations, plus face shields and eye-drenching facilities within 25 feet for people handling batteries, which is a description of the electrolyte hazard rather than a household requirement.

Now the geometry, which is where a reef tank makes this specifically worse than a battery in a garage, and which is this site’s own reasoning because no source we located addresses the combination. A reef display is tens of gallons of saltwater standing above the cabinet, often with a sump below it, hoses running between them, and a siphon path that a power outage can itself set in motion. Putting an open, vented, acid-filled battery inside or under that cabinet puts it under a standing spill risk and next to a salt-laden atmosphere. Our pet load power budget calculator already publishes the site’s siting rule for the analogous case of a battery power station beside a tank, arrived at the same way and marked the same way: site the pack above and away from any splash, siphon or hose-failure path, on a surface a leak cannot reach, and route its cords where neither water nor an animal can reach them.

So the position this page takes, stated as a position rather than as a citation:

• Prefer a purpose-built aquarium battery backup or a listed portable power station over a loose car battery indoors. One honest limit on that recommendation: neither EcoTech nor CoralVue states a cell chemistry or a sealed-versus-unsealed designation on the pages read for this article, so this page cannot tell you that a purpose-built unit is sealed. What it can tell you is that both are sold as enclosed consumer products with published compatibility and, in EcoTech’s case, a monthly test procedure, and that OSHA’s ventilation requirement is written about “Batteries of the unsealed type” specifically. Sealed or unsealed is therefore the first thing to establish about whatever you put next to the tank, a car battery included, because it is the word OSHA’s requirement turns on and it is not a property you can assume.
• If you use a battery pack of any kind near the tank, site it above and away from every water path, per the rule on our power budget page.
• If your plan is a fuel generator, that is a different hazard with a non-negotiable rule, and it is on our generator carbon monoxide guide for pet households. Never indoors, never in a garage, no exceptions for an open door.
• Improvised heat and light sources belong on their own page. Our guide to outage improvisations that endanger pets covers burning fuel indoors, candles and dry ice, all three of which people reach for in exactly this situation.

One improvisation that appears in the professional literature and is deliberately not reproduced here as a technique: Hemdal’s 2008 article describes rigging an aeration system from a SCUBA cylinder, and notes that a medical or welding oxygen cylinder can be rigged similarly. That is a pressurised-gas improvisation involving a regulator, a needle valve and a rated pressure. It is named here so that you know where it comes from if you encounter it, and it is not a technique this page is going to walk you through.

What This Page Will Not Tell You, and Why

Five specific things, listed rather than quietly omitted.

No hydrogen peroxide oxygenation instruction. Hemdal mentions that some aquarists have experimented with hydrogen peroxide as a supplemental oxygen source, and in the same sentence says why that is not an emergency technique: because the compound is so reactive, it is “difficult, if not impossible to dose correctly during an emergency.” That is the source ruling out its own suggestion, and there is nothing for this page to add.

No ammonia neutraliser dose, and no product protocol. Hemdal names ammonia neutralising chemicals as something to keep on hand for a long outage. Yanong mentions a named product class to Florida Sea Grant and immediately qualifies it, saying such conditioners “may reduce ammonia’s harm but doesn’t remove it.” Reporting that professionals name ammonia management as a long-outage problem is a sourced fact. Telling you what to add to a reef tank, and how much, is a treatment decision this site does not make.

No reef ammonia threshold and no reef water-change figure. Figures of that kind exist in ornamental-fish guidance, and our sibling page carries one from NC State, the same veterinary college this page cites above. Say the scope precisely, because the reason for the refusal has to be checkable: that document is not labelled freshwater-only, and its lighting item names reef tanks explicitly, but its ammonia figure is written about fish in unfiltered water and it addresses corals nowhere except on light. Yanong’s post-outage water-testing instruction is written around hang-on-back and canister filters, which is a freshwater ornamental context. Carrying either figure across to a reef system on this page’s authority would be inventing the transfer. If ammonia in transit is your question rather than ammonia at home, our fish evacuation and transport guide carries an extension service’s published relationship between total ammonia, pH and temperature, with the scope it was written under.

No hours-remaining figure computed from the hypoxia literature. Explained at length in the two clocks section. The literature establishes that the oxygen clock is hours-to-a-day scale in at least one species and that the concentration threshold is low. It does not establish how fast your tank gets there.

No coral temperature threshold, and no coral rate-of-change figure either. Our sibling page already handles the threshold question and already publishes the honest boundary, which is that no primary authority publishes a corals-die-at-X table. This page does not add one. It also does not convert NC State’s tolerance window or its 10 percent-every-hour-or-two rate ceiling into reef figures: both are stated about fish, in a document written about aquariums and ponds, which is the reason and not a claim about what a keeper can measure. The two summer actions this page does carry from that document, clearing the surface and getting the tank out of direct sun, are not numbers and do not need one.

What We Could Not Read

Two retrieval notes, published rather than swallowed, because a source you could not open is not a source that says nothing.

Florida Sea Grant is unreachable live from this environment, and the version quoted on this page is an archive capture. flseagrant.org returned a Cloudflare block page to every route tried on August 19, 2026: plain retrieval with a browser user agent, an automated fetch, and a fully rendered browser session, all of which returned the interstitial reading that you have been blocked rather than a challenge a browser can clear. That is an address-level block rather than a bot check. The Q&A quoted throughout this page was read from the Internet Archive’s capture of October 28, 2025, which carries the article’s own byline, date and interviewee credentials. If you can reach flseagrant.org yourself, read it live: it is the closest thing to a United States extension service voice on this exact question that we located, and its author is a credentialed extension veterinarian. Florida Sea Grant also publishes a visual checklist version of the same material, in English and in Spanish. We could reach the page holding it but not read the content of the graphic itself, so nothing from it is quoted here.

CoralVue’s product page took a second route. It returned a 403 to direct retrieval and was read in a rendered browser session on August 19, 2026, where the formula, both worked examples, the specification block and the 12V-versus-24V note were all confirmed. That is one route failing rather than a dead source.

Two further notes on how the primary documents were handled, since both are the kind of file that can be extracted wrongly. The AZA standards PDF and the EcoTech manual were each extracted twice, once preserving layout and once as raw text, and the two extractions were compared word for word. On the AZA document the two agree word for word across every passage quoted on this page, which is the claim that matters here; across the whole 45,000-word file they do not agree perfectly, because the raw extraction occasionally fuses a word pair across a line break and the two modes place the running header and page number differently. Neither defect touches a quotation used above. On the EcoTech manual the layout extraction scatters the runtime table across the page while the raw extraction prints it in reading order, and the two agree on every figure. And Hemdal’s article body on reefs.com sits outside the page element an extractor would normally scope to, so a naive extraction of that page returns an events widget rather than the article; the full text here came from stripping the entire document.

Your Reef Outage Priority Checklist

The first six items are for the outage. The rest are for the quiet afternoon, and they are the ones that actually decide how the first six go.

• Power water movement first. One pump or one air pump, running properly, beats three loads running badly. If your backup is automatic, confirm it actually started.
• Put the pump high, near the surface. EcoTech’s own instruction, and the target to look for is a moving, choppy surface rather than just a running motor.
• Check the water level again a few hours in. Evaporation continues while the auto top-off is dead, and a pump set at the old waterline will not stay at the new one. This one is our reasoning, not a published instruction.
• If the outage is a hot one, treat the heat as an oxygen problem. Coleman’s point is that warmer water holds less oxygen, so a hot room is working against the one thing this page tells you to protect, and his own conclusion is that heating is the direction you can usually deprioritise. NC State’s two summer instructions cost no power: “During summer, remove anything from the water’s surface to increase the area for gas-exchange efficiency.” and “If the water gets too warm, minimize or eliminate exposure to direct sunlight.” Both come from a document written about aquariums and ponds whose temperature figures are stated about fish, and whose author scopes it to temperate pond and aquarium species while saying the basic principles apply for many species, so this page takes the two actions and not the numbers, and it publishes no cooling rate for a reef tank. Clearing the surface also takes away whatever was slowing evaporation, so this box and the water-level box above are one job. That join is our reasoning, not NC State’s.
• Watch the animals, not the clock. A rise in breathing rate, or fish gasping at the surface, outranks every hours figure on this page and every hours figure you will find elsewhere. Yanong’s instruction for what to do when you see those signs is to “seek advice from a veterinarian or an experienced aquarium expert”, and that is a phone number worth having before the power goes out rather than during.
• Keep the lights off for the length of an ordinary outage. Kurtz writes that darkness will help keep fish calm and less active so they consume less oxygen; Yanong separately says it is best to keep the tank or room dark to reduce stress and aggression, without making the oxygen point. The peer-reviewed darkness literature runs in days rather than hours, and reef lighting is a load a small backup cannot carry. If the outage is running into days rather than hours, stop treating this as free: the shortest published darkness-to-bleaching figure is three days, Bulk Reef Supply says prolonged darkness stresses corals, and Hemdal names lack of light as something that kills corals in a long outage. That is not a reason to run reef lights off a small battery, which will not work anyway. It is the point at which the problem stops being which load to power and becomes a bigger supply or a different plan, which is what Yanong means by recommending a generator for a reef tank and what our weeks-long outage plan is for.
• Do not put an open car battery under the tank. OSHA’s occupational standard describes what unsealed batteries emit and calls it an explosive mixture; a reef stand is a standing spill and siphon path. Site any battery above and away from every water route, and prefer a purpose-built aquarium unit or a listed power station. OSHA’s clauses are about unsealed batteries specifically, so find out which kind yours is.
• Do not run a generator indoors or in a garage, with the door open or otherwise. That rule has its own page and it is not negotiable.
• Test your battery backup monthly. EcoTech publishes the interval and the procedure: unplug both the backup and the pump from the wall and read the charge remaining. It takes two minutes and cannot be done during the outage.
• Check that your backup actually drives your pump. CoralVue’s unit excludes several named pump brands outright; EcoTech’s is sold for its own pumps and needs an extra booster for one of its own lines. Find out now, not at hour one.
• Write down your own pump’s draw, in amps or in watts, from the manufacturer’s specification. CoralVue’s published formula needs that number and nothing else. Without it the formula is useless to you.
• Stage fresh cells, or keep the rechargeable pack topped up. Every published runtime figure on this page is conditional on the state of the battery.
• Put a monitoring alarm on the tank. Bulk Reef Supply’s own preparation advice is to use power monitoring that messages your phone when the aquarium loses power, and AZA’s standard for accredited aquariums is that the monitoring interval has to be shorter than the animals’ survival time. An alarm is the home version of that requirement.
• Find out what your own tank does in an outage before you need to know. Not from a page. From the one time you unplug the pumps for an hour on a Saturday morning, with the lights on and yourself in the room, and watch what the surface, the fish and the thermometer do.

The Version You Do on a Quiet Afternoon

Three preparations turn everything above from a decision into a routine.

Decide what the one battery is for, before there is an outage. The whole of this page is an argument about scarcity, and scarcity is a planning failure you can partly undo in advance. Knowing that the battery is for the circulation pump, that the pump lives high in the tank when it is on battery, and that the lights stay off, is a decision you make once and then never make again at two in the morning with a torch in your teeth.

Test it, because until you have watched it work, the runtime figure is a claim. EcoTech asks for a monthly test and describes it in two steps. Every published number on this page comes with a manufacturer disclaimer about operating conditions, battery age and cycle count, and the only figure that means anything for your battery is the one your battery produces.

Find out what your tank’s actual survival time looks like. This is the AZA requirement translated into a home. Accredited aquariums are told that if they check a system by hand, they have to check it more often than the animals can survive a failure. You cannot know that interval without knowing roughly what your own system does when the power stops. An hour with the pumps off, in daylight, with you present and the thermometer in view, tells you more about your tank than every hours figure in the table above, because those figures are about other people’s tanks and yours is the only one that matters.

Where to Go Next

This page is the priority spoke of our pets and power outages pillar, and it answers exactly one question: what to run when you cannot run everything.

For the full reef outage picture. The aquarium and reptile power outage guide carries the reef failure chain, the heater wattage problem and why battery-backing a reef heater is mostly a myth, the coral temperature discussion, the summer half of that problem as well as the winter one, salinity creep on a long outage, the battery air pumps with their published specifications, and the post-restoration check that catches a heater whose smart plug came back in the off position.

For runtime on anything through an inverter or a household plug. The pet load power budget calculator has the published formula, the four documented reasons the real figure lands below it, and the siting rule for a battery pack next to a tank. For choosing the hardware, portable power stations for pets compares manufacturer specifications, and quiet generator versus portable power station is the fuller version of that decision.

If the outage is going to outlast your gear. The fish evacuation and transport guide has the bag-and-bucket numbers, and the weeks-long outage plan is the version of this problem where the answer stops being a battery.

If you are improvising. Read outage improvisations that endanger pets before you light, burn or freeze anything indoors, and generator carbon monoxide and pets before you start an engine.

Then do the one thing that pays off most, and it takes five minutes. Look up your circulation pump’s power draw in watts or amps, write it on a card, and tape the card inside the cabinet door next to the battery. CoralVue’s published formula needs that one number to tell you how long you have, and the hour you need it is the hour you cannot look it up.