Why a Gravel Vacuum's Mouth Is the Specification

This page contains affiliate links. If you buy through one we may earn a commission, at no cost to you.

Yes, and the tool that exists for doing it properly is not a suction device at all. A gravel vacuum is a separation by terminal velocity, and once you see that, the whole specification sheet reads differently, because the number that decides whether it works is the one nobody publishes.

Three things in a tank fall through water at speeds 1,404 apart: a detritus floc at 0.27 millimetres a second, fine sand at 24.8, and three millimetre gravel at 379. Any upward water velocity between the first and the last lifts the waste and leaves the substrate. That is an enormously wide window, so the real question is why gravel ever gets sucked up the tube. The answer is that the pump sets the flow and the mouth sets the velocity, and velocity goes as one over the square of the mouth diameter.

On a published flow of 1,700 litres an hour, retaining three millimetre gravel needs a mouth 39.8 millimetres across. Retaining fine sand needs one 155.8 millimetres across, which is six inches, and nothing hand-held has that.

The three speeds

A particle falling through water reaches a speed where drag balances its weight less its buoyancy. For the smallest particles that gives the Stokes expression, but sand and gravel are well past the range where Stokes holds, so the figures below are iterated through the standard drag curve rather than taken from the small particle formula. At 20 degrees, with water at 1.002 millipascal seconds:

A detritus floc, taken as a tenth of a millimetre across at 1,050 kilograms a cubic metre, because organic waste is barely denser than the water it sits in, settles at 0.27 millimetres a second. That is one centimetre in thirty seven seconds, which is why a disturbed tank stays cloudy for hours.

Fine sand, two tenths of a millimetre at 2,650, settles at 24.8 millimetres a second.

Gravel, three millimetres at the same density, settles at 379 millimetres a second.

So the sorting job is easy in principle. Pull water upward at anything between a third of a millimetre and a third of a metre a second and the waste leaves while the gravel stays. The practical difficulty is entirely in hitting that window, and the window is set by geometry rather than by the motor.

Where the velocity actually comes from

An electric aquarium vacuum 5-in-1 gravel cleaner at 28 watts publishes a 110 volt 28 watt pump motor, a maximum flow rate of 1,700 litres an hour, four spliced rigid tubes extending from 12.4 to 33.6 inches, a 2 metre water hose, a reusable sponge filter cup that returns clean water to the tank, and a rating for roughly 10 to 100 gallon freshwater, saltwater and turtle tanks.

Seventeen hundred litres an hour is 4.722 times ten to the minus four cubic metres a second. Divide that by the mouth area and you get the upward velocity in the gravel bed:

Through a 25 millimetre mouth, 0.962 metres a second, which is two and a half times the gravel's own settling speed. The gravel goes up the tube.

Through 40 millimetres, 0.376, just under it.

Through 50 millimetres, 0.241. Through 76, 0.104.

Inverting it gives the specification. Retaining three millimetre gravel at this flow needs a mouth 39.8 millimetres across. Retaining two tenths of a millimetre sand needs 155.8 millimetres, which is six inches.

Which settles a question that comes up constantly about sand bottomed tanks. No hand-held tool at these flows can keep fine sand in the tank if you push the mouth into the bed, and that is not a defect in any particular product. It is why sand is cleaned by hovering the mouth a centimetre or two above the surface, where the velocity has already fallen off and only the light material gets entrained. More in our review of the 28 watt electric gravel cleaner, whose own caution matters for the arithmetic above: the handle and switch "are not waterproof and must stay away from water", which limits how low the discharge can go.

That listing also carries a cross-check worth running. It claims it "can drain a 180 gallon tank in roughly 30 minutes under optimal conditions". A hundred and eighty US gallons is 681 litres, which at the rated 1,700 litres an hour is 24.0 minutes. So the stated thirty minutes is the rated flow derated to about 80 percent. That is an honest figure rather than a contradiction, and in a catalogue where headline numbers are usually best cases it is worth saying so.

What the flow dial is worth, which is less than the heads

An AQQA 6-in-1 electric gravel cleaner rated 600 GPH publishes a 34 watt DC motor at 600 gallons an hour with a rotary controller that varies power steplessly down to 15 watts, an in-line switch for quick flow adjustment, an ABS filter cup with biochemical cotton that accepts other media, tool-free connector heads for each of six functions, and a telescoping handle from 16.9 to 33.6 inches.

Six hundred gallons an hour is 2,271 litres, so by the same division it needs a 46.1 millimetre mouth to keep three millimetre gravel and 180.1, seven inches, to keep fine sand.

Now the dial. A centrifugal pump's flow goes roughly as its speed while power goes as the cube of it, so flow goes as the cube root of power. Fifteen watts out of 34 is 0.441, and the cube root of that is 0.761, so the quiet end of the dial delivers about 76 percent of the flow. The required mouth goes as the square root of the flow, so it falls from 46.1 to 40.2 millimetres.

The entire range of the flow control is worth 12.8 percent on the mouth diameter. Which is a genuinely useful thing to know, because the same listing publishes interchangeable heads, and swapping a head changes the mouth by far more than the dial changes the flow. The heads are the important control and the dial is the one with the knob on it. Details in our review of the AQQA 6-in-1 cleaner, including its own maintenance condition: the pump "must be cleaned after every use".

A manual siphon is not gentler, only differently controlled

A SEAOURA 3-in-1 aquarium gravel vacuum and siphon cleaning kit publishes gravel washing, stain scraping and waste removal in one body, dual airbags with an atmospheric cavity for manual pumping, a filter basket net at the intake, and four assembled setups whose lengths it actually states: two tubes at 33.5 inches, two tubes plus head at 35, one tube at 15.74, and one tube plus head at 17, excluding the handle.

Flow through a siphon is not set by effort. It is set by the vertical drop from the tank's surface to where the water comes out, and the velocity goes as the square root of that drop. With a friction dominated discharge coefficient of about 0.45, which is where a long narrow tube with bends lands:

A bucket 40 centimetres below the surface gives 1.26 metres a second in the tube. At 50 centimetres, 1.41. At 89 centimetres, 1.88.

Those are tube velocities rather than mouth velocities, and they are the same order as the powered machines above, which is the point: a manual siphon is not a gentle tool, it is a tool whose single control is the bucket height. And the kit's four published lengths span a ratio of 2.224, which is 1.491 in velocity, because of the square root. The extension tubes are a flow control that nobody calls one.

Both SEAOURA records are admirably straight about the consequence. Their drawbacks read "Water flow has no fixed standard" and "Flow depends on your manual pumping", which is exactly why every manual figure on this page comes from a geometry rather than from a listing. See our review of the SEAOURA 3-in-1 kit.

Two mechanisms for one job, described as one feature

A SEAOURA 4-in-1 aquarium gravel vacuum siphon cleaner publishes four functions, dual built-in airbags to start the siphon, no electricity and no motor noise, a rating for tanks up to 50 gallons, and a filter basket net at the intake that it says "guards against sucking out small fish or fine gravel".

That is true and it is a completely different mechanism from the one this page has been describing, which is worth separating because the records do not. A net stops substrate by size. A wide mouth stops it by velocity. A net with two millimetre holes will pass two millimetre grit at any velocity you like, and a four centimetre mouth will retain three millimetre gravel with no net in it at all. The two protections cover different particles, and a product with both is covered twice over the size the net catches and once below it.

The smallest version of the same idea makes the geometry explicit. An aquarium siphon vacuum and water changer kit publishes a duckbill suction inlet that reaches into gravel to lift waste while leaving the substrate in place, a filter net cover, a brush with sponge, a removable pump head and a rating for tanks up to 10 gallons. A duckbill is a mouth widened in one direction, which is precisely the right shape for this problem: it increases the inlet area without increasing the depth it digs to. That is the hydraulic answer built into a moulding, and the listing sells it as reach. More in our review of the siphon and water changer kit, and our review of the SEAOURA 4-in-1.

So, about stirring it

The arithmetic above says the mechanical half of the question is benign. Stirring a bed lifts detritus at 0.27 millimetres a second and leaves gravel exactly where it was, and a tool with a wide enough mouth removes the first while declining the second. Mixing layers of different grain size is the one mechanical risk, and it is a real one, because the fines then settle through the coarse material and fill its voids, which no vacuum will reverse.

The reason for caution is biological rather than hydraulic, and it is not this page's to settle. A long undisturbed bed can hold anaerobic pockets, and releasing them at once is a different kind of event from clouding the water. Our page on why a fish tank goes cloudy after you clean it owns what happens to the bacterial population after a disturbance, and why an aquarium's pH falls while you change the water owns the chemistry of the water you put back. The practical routine is in how to build a simple aquarium cleaning routine, and the right reading of all four pages together is: stir a section at a time, not the whole bed, and let the filter handle what you lift.

One credit worth giving, because the two arguments look alike and are not. Our page on why a bag is the filter and a cyclone is not works out inertial separation in air, where particles are thrown across streamlines and the governing number is a cut diameter. A siphon sorts by settling against an upward flow instead, which is a balance rather than a deflection, and that is why the answer here is a mouth area and the answer there is an inlet width and a number of turns.

The honest limit, and the one dimension nobody publishes

What is exact here is the structure. Velocity is flow divided by area, so it goes as one over the square of the mouth diameter, and the required mouth goes as the square root of the flow: those two statements carry the 39.8, the 155.8, the 46.1, the 180.1 and the 12.8 percent, and none of them depends on a settling speed being right. Siphon velocity goes as the square root of the drop, so the 1.491 across four published tube lengths is exact. The 24.0 minutes against the stated 30 is arithmetic on two published figures.

What is representative rather than published is every particle. The 0.1 millimetre detritus floc at 1,050 kilograms a cubic metre, the 0.2 millimetre sand and the 3 millimetre gravel at 2,650 are typical values, and real substrate is a distribution rather than a size. The settling speeds follow from the standard drag curve, which is reliable for spheres and optimistic for the irregular particles actually involved: a flat flake settles slower than this says and a rounded pebble a little faster. The 0.45 discharge coefficient on the siphon is a friction-dominated estimate rather than a measurement, and it moves every manual velocity proportionally while leaving the 1.491 untouched. The cube root relation between pump power and flow assumes a centrifugal impeller, which is what these are.

And the gap is the whole point. Not one of these five records publishes a mouth diameter, an inlet width or an inlet area. The single dimension the physics needs is absent from every listing in the category, while tube lengths, wattages, flow rates, handle extensions and function counts are all published. The arithmetic can tell you what mouth a given flow requires. It cannot tell you what mouth you are buying.

Two things this page deliberately does not compute. It does not estimate what fraction of the detritus actually leaves per pass, because that needs the bed's porosity and the flow field inside it, and a single velocity at the mouth says nothing about either. And it does not touch the biological consequence of disturbing an established bed, which is the real content of the question and belongs to the pages linked above rather than to a settling calculation.

Three numbers to carry. Waste and gravel fall through water at speeds 1,404 apart, so the separation is easy and only the geometry is hard. Velocity goes as one over the square of the mouth, so a 40 millimetre mouth keeps gravel at 1,700 litres an hour and a 25 millimetre one does not. And fine sand needs six inches of mouth at that flow, which is why you hover over sand instead of digging into it.