site logo

Main Index > Articles Main > Tank management >
8 visitors reading
Undergravel Filters: Conventional, Reverse-Flow, & RTR's Ultimate Over-Engineered RFUG
By: Robert T. Ricketts, a.k.a. RTR




Fish, like any other animal, take in food, water, and oxygen to maintain their metabolism and life.  They also excrete solid and dissolved wastes.  Those wastes are toxic to the animals that produced them.  In the wild wastes are rapidly diluted by the much greater water volume to fish ratios there than those we use in out tanks.  We are effectively concentrating the waste material, so we have to ensure that we can detoxify the waste in some manner.  There are bacteria that are delighted to take fish waste and use it as their own food/energy source.  The primary toxic nitrogen-containing waste product from fish is ammonia (NH3), or ammonium ion (NH4+), dependent on pH and temperature.  Certain bacteria absorb ammonia/ammonium ion and get energy by oxidizing it to nitrite (NO2--).  They release the nitrite as a waste product.  Nitrite is as toxic as ammonia, or more so.  Other bacteria take up nitrite and oxidize it to nitrate (NO3--), releasing the nitrate into the water.  Nitrate is relatively non-toxic at low concentration.  Both of these types of bacteria must live attached to a substrate, a surface.  The largest surface area in our tanks is not the glass, but the gravel.  The desired bacteria also require oxygen in the water, and as they are attached, water currents must bring the oxygen to them.  The same currents will bring them "food" in the form of ammonia and nitrites- a good life if you are a nitrifying bacterium.  Current can only penetrate a small distance into most gravel beds.  Oxygen can penetrate by diffusion, but this is a much slower process than by current.  This limits the oxygen- and food-rich sites to very near the surface.  Okay, suppose we could arrange to have a slow current flowing through the bed, we would have the entire surface of the gravel open for bacterial colonization.  From such thought came the undergravel filter.

under gravel filter




An undergravel filter (UG) is a slotted plate that fits the bottom of the tank, on top of which is a layer of aquarium gravel.  There are several ports on the plate where vertical riser tubes may be placed.  Originally all UGs were driven by airlift. The earliest units used just the rigid airline itself.  A bit later, an airstone was attached to the end of a piece of rigid airline tubing and lowered into the riser.  To operate an air-driven UG, the output from an airpump is directed through a gang valve (to control and balance air volumes) from which individual lines can run to each UG riser.  The airstone in the riser is best placed just above the collar connecting the riser to the UG plate itself.  This avoids the possibility of air being pumped under the plate, which could block operation.  The bubbles from the airstone (the smaller the better) rise to the surface and out across the water surface.  The column of air + water in the tube is lighter, weighs less, than an imaginary equal column of water alone would weigh.  Since the atmosphere is pressing down equally all over the surface of the tank, the air-water mixture in the column moves upward due to atmospheric pressure and gravity trying to equalize the riser air-water mixture and the tank water.  Air is constantly pumped into the riser, so they cannot be equalized.  Water moves from the tank through the gravel and UG plate and up the riser in constant circulation.  Simple and effective, the slow current provides enough oxygen and "food" (ammonia and nitrites, not fishfood) to the bacteria. 

Airstones clog and need regular replacement. Air bubbles are noisy (I object to that). They splash the cover glass, leaving behind mineral and protein films that can eventually etch the glass. In the short term this water and mineral promotes algae growth on the underside of the glass. It can also contribute to salt/mineral creep onto cover glass and the tank rim. I object to all of these things. I object to most things that make noise, and to anything that makes another chore for me.

More modern installations use a powerhead (an electrically driven water pump) on top of the riser and omit the air, airstones, etc. The pump draws water from the riser and pushes it across the water surface, or more commonly, under the water surface. This can increase the flow tremendously, depending on the gallons per hour rating of the pump, the thickness of the gravel bed, and the particle size of the gravel. Please don’t kid yourself that a pump rated at 300 gallons per hour is actually going to move 300 gallons through your gravel every hour. Pump and filter ratings are done with no load and at zero head. They represent the maximum possible output, not a real-world application. The pump has to overcome friction and inertia to move water from the tank down through the gravel bed and the slots in the UG plate, across the tank bottom and up the tube. The type of pumps we use “push” better than they “pull”, and we are asking them to pull in this application. Not a big deal, they do the job quite well. It is highly questionable if large flow rates are even desirable, but I’ll argue that point later. Pumps require much less upkeep than airlifts. They do need to be opened and cleaned at intervals of a few months minimum to maintain their flow rate. This is easy if you have a selection of small “bottle brushes” for cleaning out the impeller housing, intake, and outlet. Many powerheads have venturi openings on the outlet side of the pump. If desired, a piece of tubing can run from the opening to above the water surface. Air is sucked down the tube and into the outlet stream of water (the venturi effect is strongly reduced pressure at right angles to the direction of flow). Remember the noise of the airlift? This is much louder. Remember the films, algae, and mineral creep of the airlift? This covers a larger area, so to me is worse. Do you need to ask if I EVER use the venturi of a powerhead? Well, not for air, thank you. If you have a current of water from a powerhead or other filter return running across under the surface of the water so that it ripples the surface without breaking the surface, you will provide sufficient aeration to the water.

Gravel size does matter for use with an undergravel filter, but it is pretty straightforward. Surface area per unit weight is inversely related to particle size- the smaller the average particle, the more surface there is in a pound of particles. Very small particles (sugar sand) will restrict flow and sift through the plate, so that is out. Somewhat larger sizes are used often by plant growers, as many plants prefer a bit finer particle size than standard aquarium gravel, but plants are not recommended with standard UGs. Standard aquarium gravel is fine for UG. Larger gravels, such as pea gravel, are not good, as you are losing surface area. Large gravel is harder to keep clean also. Food and fish waste can fall between the gravel particles and be difficult to get out (hydrovacuums or gravel washers as part of vacuuming the substrate, like UG plates, were designed for standard gravel). Even worse, it may go through the gravel to the plate and build up on the floor of the tank. A gravel bed of one to two inches depth is standard. Deeper beds may restrict flow.

Most of the controversy associated with UGs is related to maintenance. UGs operate best as biological filters. They unfortunately serve nicely as mechanical filters as well, especially in tanks using high-output powerheads. Such pumps do move a lot of water through the bed, so they can trap most particulates out of the water column. This pleases the tank keeper, because the tank looks clearer and cleaner. Fine, but as is the case with any mechanical filter, the tank is not really cleaner, the debris has been removed from the water column and deposited in the filter. Here, the filter is the gravel. Microorganisms of various sorts will digest the material eventually, but you are building up decomposed waste in and eventually under the gravel bed. Untended gravel beds in UGs are bombs waiting for critical mass. Nitrates in such tanks tend to be elevated or astronomic, but most people don’t test for nitrates. On the other hand, if UGs are maintained properly, they can run for years with no problems. Proper maintenance is not complicated or difficult.

I am a firm believer in water changes. Water is cheaper and easier to work with than carbon, affinity resins, filter cartridges or flocculants. My tanks are partialed routinely. For me, a one-third to one-half change every week is no big deal. Others may prefer some other change level, fine. Pick a level that is comfortable for you, do it until it becomes habit. While you are still developing the habit, test for nitrates - before or after your water change doesn’t matter so long as you are consistent. Record the results - I use calendars: MR1 (name of tank) NO3: 10ppm. The next week: MR1 - NO3 >10ppm. The week following: MR1 - NO3 >>10ppm. The end of the month: MR1 - NO3: 20ppm. Hmm, there is a problem here. The nitrates are rising with time. I need to a) reduce the stock, b) feed less, c) change more water. I’m happy with the tank as is; I don’t think I overfeed, so I’ll start changing 40% per week. The next month I see: MR1 - NO3 <20ppm, MR1 - NO3 >10ppm, MR1 - NO3: 10ppm. Problem solved. I’m happy, the tank is happy, and the tank is cleaner than it was. If this increase had not done the job, I would have moved the percent water change up again. Note - there was no visible change in the tank. If I had continued the smaller changes, the tank would have continued to decline imperceptibly. I could have experienced less active fish, algae outbreaks, perhaps disease outbreaks. It is best to solve your problems before they become problems.

That was the long slow curve. The fast break is that is that water changes are how you clean UGs. While you are siphoning (or Pythoning) water from the tank, you use a hydrovacuum on the gravel. If you do small regular changes you may not be able to vacuum the entire substrate in one partial. Again not a problem- start at the left (or right, your choice), work front to back, and when you have removed as much water as you wanted, mark your place with a small rock. Next time start at the rock and continue. Continue re-marking your place until you’ve covered the tank and start over. This is simple, no-brainer work. If you allow the UG to go without vacuuming for months, then clean it, you will have a major task. You may also reduce the functional capacity of the biofilter. When the gravel is vacuumed regularly, the bacterial film on the gravel will stay young and healthy. A neglected filter may have more bacteria living on the particulates between the grains of gravel than on the gravel itself. This is so because the bacterial film on the gravel will have been smothered by debris, or simply have peeled off the gravel from its own over-growth. Vacuum that bed and you may cut your handling capacity. If you have a neglected bed and I’ve inspired you to clean it before it crashes, clean 1/4 every two weeks, or 1/3 every three weeks. After that you can clean as much as you can cover during your regular water change. The other hazard of neglected UG beds is build-up of debris under the plate itself. With regular maintenance, this is unlikely to occur. It may occur under a bed made of oversized gravel matched with an oversized powerhead. In a neglected bed it is inevitable. You can try siphoning from under the plate with flexible airline tubing fed down the riser and under the plate, but I’m uncertain of the effectiveness of this technique.

What tanks should NOT use conventional undergravel filters? There is a short but clearly defined list:

1. Substrate-spawning cichlids or other fish with major earth-moving tendencies (some large cats) are incompatible with conventional UG. They dig and short-circuit the water flow.

2. Fish needing or requiring fine to very fine sand [rays, flounders, many “eels” (Mastacembelidae), some catfish, elephant-nose] in which to bury themselves or forage for worms and crustaceans are not suitable for UG, as their substrate will go through the UG plate.

3. Plant tanks are not suited for this technique. Most root-nourished plants may not perform as expected over UG plates, and enriched substrates are incompatible with the technique. There are exceptions to this rule, but they would be for advanced hobbyists. Eventually there may be another article on that topic. Plants in pots are unaffected by UG of course.

4. Any tank belonging to a fish keeper who is unable (for whatever reason) to perform regular maintenance. If you are the type of hobbyist who must or does ignore your tanks (except for sporadic feeding) for weeks to months, then goes into a mad flurry of tank cleaning, UGs are not for you.


It is possible to avoid the biggest problem of UGs, their excellent mechanical filtration potential, by reversing the flow. That is, instead of pulling the water down through the gravel, you push it up through the gravel. You also use a sponge pre-filter on the powerhead, so it is pushing pre-filtered water down the riser tube, out under the plate, and up through the gravel. This is reverse-flow undergravel (RFUG) filtration. Several brands of powerheads can be adapted to this, but I only have personal experience with one, Penguin, which offers an accessory kit to convert their conventional pump to reverse flow. The kit has an elbow attachment to mate the pump output tube to the riser tube and a sponge assembly for the intake side of the pump. This is not expensive, and being a lazy man, I haven’t tested any other technique since these came on the market. The sponges are easily removed for rinsing under the tap. There is no concern with preserving any nitrifying bacteria in residence in the sponge, as the gravel bed is to perform that service. The sponge can function purely as a mechanical filter, and be rinsed weekly, biweekly, or whatever your water change interval may be. I like biweekly at least, weekly is better, to get waste out of the tank before it is completely digested in the system. The gravel is still hydrovacuumed, but you will be astonished at how little material comes out of the gravel compared to conventional flow UG, or to conventional substrates.

There are other powerhead brands that have diverters built into the intake side to reverse flow. In my experience, the reversed flow with these units is a small fraction of their normal rated flow. You are running substantial power consumption and not getting much output from it.


Having removed the primary objection to the technique, we have the problem of digging fish to resolve. It can be done, but is DIY and MUCH more setup work than either straightforward UG or RFUG. At the time of development of the particular technique I was breeding several substrate-spawners. I despaired over having the tanks a) look decent, b) be easier to maintain, and c) not have bare glass areas on the bottom.

There had been an article long ago, in FAMA I believe, on using floss under the gravel itself over conventional UG plates (my apologies to the author, I don’t have the citation at hand). The object of the exercise was to increase the sites available for colonization. Okay, there is more surface to a mat of fiber than to a gravel bed, but I was concerned about clogging the mat with gunk over time, compression of the mat by the overlying gravel, and the Cichlids could still uncover the mat, with double or more gravel depth elsewhere. This would result in uneven flow and potential loss of function.

Enter eggcrate light diffuser. This is a plastic grid, 1/2 inch or less thick, used in some commercial fluorescent light fixtures. It is available at hardware stores and plastic suppliers in 2 x 4 foot sheets. So, cut a piece of eggcrate with about I/2 inch clearance from the front and side walls of the tank, cut out notches to allow the UG plate riser collars to protrude. Then cut another piece just like it. Cut a sheet of filter floss with less clearance than allowed for the eggcrate, say 1/4” front and sides. If I am working with a tank that is a close fit for the standard sheet size, I don’t bother to trim the 1/4” inch off, gravel will conceal it. Make short straight cuts in from the back at the location of each riser collar, then a crosscut out from that to the edge of each collar. This is in effect a “+” with one leg anchored at the back of the floss. I use Marineland’s bonded filter pad, which is available in 12 x 24” sheets. It seems the most uncrushable of the types I’ve tested. If you have a different favorite flavor, by all means use it. If larger areas are to be covered, simply butt the pieces together. Finally, cut a sheet of fiberglass window screen 3” longer and 2 1/2” wider than the eggcrate panels you cut previously.

Assembly is simple. The only non-standard technique I use is to assemble with several inches of water already in the tank. Trapped air under the plate will stay there, so I make sure that I have all the air out from the UG plate before I continue. Place the riser tube in whichever collar you wish, but test-fit the reverse-flow powerhead now, before completely filling the tank, to be sure it fits where you want it to be. Commonly the riser tub must be cut down to a lesser height. Remove the pump and its’ riser again to have it out of your way. Cap the leftover collars (without trapping air please). Spread the fiber mat and push it down around the collars. Cover that with one of the panels of eggcrate. Next add the sheet of fiberglass screen. You want about 1 1/2” folded down over the eggcrate front and sides, about 1” folded down at the back wall (less here as there will be no gravel between the grid/plate assembly and the back tank wall). A fist-sized rock sitting on the assembly in the middle of the tank can help hold everything while you make the adjustments. When you are happy with the screen placement, cut an X or + over the chosen riser collar and push down the excess screen. Replace the riser tube. Remove the rock and add the second sheet of eggcrate. Add part of the gravel, mainly around the edges of the tank. Using a flexible spatula, disposable plastic knife, or similar tool, work gravel down at the front and sides, alternating side to side to front frequently so you don’t push the whole assembly askew by over-packing any one area. You are NOT forcing gravel into a crevasse; you are working in just enough to conceal the eggcrate. Add enough more gravel to barely fill the eggcrate grid. If the tank is to have significant rockwork for caves, etc., place this now, solidly on the eggcrate for support. Finally add more gravel to cover the eggcrate about 1/4 -1/2”. Fill the tank and reposition the pump. Make sure it isn’t full of air before plugging it in. You are in business.

Maintenance is easy. Rinse the prefilter sponge, preferably weekly; at least with each water change except when you have small free-swimming fry in the tank- they graze the sponges and I reduce the rinses during this short period. Vacuum the gravel as usual, but you can’t go deeply as you really only have a thin veneer of gravel, your real biofilter is the fiber pad. If your fish are mounding too much gravel in corners, just remove some. By this time you won’t be shocked to see a white grid in your gravel. I like to leave the fish some top gravel to move around, it is part of their spawning instinct after all.

If your tank is taller than a 15 or 20L, and if you may have a higher bioload than a spawning pair of fish and their fry, you can double the floss layer. Cut a second sheet of floss and another eggcrate. Layer from the tank bottom up: UG plate, floss, eggcrate, floss, eggcrate, screen, eggcrate, gravel. I just dismantled a 30XH with double floss set 11 years ago. The floss wasn’t dirty. It was not pristine white, but for 11 years old it was very clean. My mbuna tank with hundreds of pounds of rocks was set the same way. It was older than the 30XH, but it had been moved, and had to be dismantled for the re-location. Both of those tanks were started with canisters feeding the RFUGs, but that is a real waste of a canister. Besides, I don’t like to use canisters as mechanical filters, they are too inconvenient to service as often as I like for mechanicals.

What type of tanks should not use RFUGs? Well, there were four on the conventional UG list. Here I would reduce it to two for the over-engineered RFUG. It still isn’t the best technique for plants to most planted-tank hobbyists, but hey, neither are most substrate-spawning Cichlids. Very fine sand will sift through the fiberglass screen, so it is still out for fish that need that. Even those folk who can’t manage to do maintenance as often as they should are likely to be no worse off with these than with any other filter. The prefilter sponges have a pretty good capacity for holding particulates, but as those are broken down they will contribute to the nitrates in the tank as will any untended mechanical filter.

Just don’t dismiss UG biofiltration out of hand as a cheap technique for novices. It can, properly set and properly maintained, provide as large a reserve capacity as a fluidized bed filter, a wet/dry, or a big Eheim canister, and do so at somewhat to much lower cost. It is unfortunately a DIY technique and not off-the-shelf. However, the skill level required to DIY is much lower than that needed to build a light hood or a tank stand.

I freely admit that I do not use my RFUGs as the sole filter in any of my tanks, nor would I recommend them as sole filter, as they provide no discernible current. On the other hand, neither do I use canisters, power filters, fluidized beds, veggie filters or algae scrubbers, W/Ds or anything else as sole filtration. ALL of my tanks other than the snail tanks (and even most of them) have at least two, and up to four distinct filter types on them (another topic, another time). The mbuna tank mentioned above had a sizable canister thermofilter, RFUGs, two internal canisters, and a W/D (more for increased water volume than filtration to be honest). I do believe in redundancy, not often to that extent, but that one is an awfully nice tank and very easy to keep. I would not consider discontinuing use of UGs. They are too valuable to me in several applications, especially in tanks with lots of rockwork. I’m more than willing to do the extra work up front to remove the risk of anaerobic pockets around my base rocks at some unknown time in the future. The UG is, like any filter, a useful tool. Like any tool it can be misused or abused. Certainly it is often misunderstood.

This article appeared first on another site. It has been edited and updated for this site.

Robert T. Ricketts, a.k.a. RTR





Privacy Policy | Contact Badman's Tropical Fish
Copyright © 1997-2009
All rights reserved. Reproduction of any portion of this website's content is forbidden without written permission.