[hoppycalif]

Several days ago another member posted an excellent point - canister filters don't pump water "uphill" to an aquarium. They just move water. That got me to thinking about how they work, because I had always assumed that if the filter was 3 feet below the tank, the filter had to produce a pressure of 3 feet of water to get water back to the tank. My thinking led to the sketch and logic shown below. Am I right on this?

 

[niko]

3 things:

1. The canister filter pump has to be powerful enough to move the water up. If you take a tiny powerhead and try to move water up you will see that it can barely can lift the water more than 8 inches.

2. The restrictions (hose diameters, kinks, elbows etc.) play a huge role in the output. I have an old looking Otto canister filter that moves way more water than a comparable size Eheim. The Otto has fat ugly hoses and the Eheim has a thin outflow hose.

3. One important thing about canisters the bypass. A canister MUST have bypass. Meaning when the media gets dirty the water should find it's way around and still flow in/out. Filter designs that try to restrict bypass imply that you will be watching the filter closely and clean it when it gets clogged. There are no such filters for aquarium use really except diatoms. I mean the bypass is not a bad thing, it has to be there.

That being said I still don't exactly understand how the flow of water in a canister filter happens. Her's the source of my confrion:

If you shut the canister filter pump, take the same gph pump and connect it to the filter inflow OR outflow pipe you get a huge increase if flow. My two 315 gph Eheim canisters made only 120 gph when they where under the tank. Now when they are inside the tank the flow is indeed 315. How is that different from having the pump anywhere else in the syphon that we call "canister filter"?

--Nikolay

 

[ghengis]

I think I see your point that the canister pump merely serves to help the water along thru it's circuit from the tank and back again, rather than actually pushing it... However, I'm not sure if I can agree on that. IMO, unless the canister is at a point almost level with the water line, there will need to be some effort required of the pump to force the water, against gravity, into the tank. The amount of effort would have to compound, as the distance between Canister outlet and water line increases.

Having said all that, and as I think of what to type, I can see in my minds eye that if one were to place a sealed vessel, with two hoses in the top, under a table and then siphon water from another vessel on the tabletop, the water would flow down into the sealed vessel, fill that vessel and then continue upward, until it was level with the waterline in the vessel on the table... If this weren't the case, the bucket I use for water changes would not overflow when I wander off and forget about it for two minutes.

Can someone please empty their (spare, if you have one?) Canister filter, sit it below a table, place a bucket on the table (with both the in and out hoses in the bucket), start a siphon and record what happens? Then post it here??

 

[hoppycalif]

Ghengis, what you want tried is exactly how a Rena Filstar filter primes itself. You fill the inlet hose to the empty filter with water, with the shutoff valve at the filter closed. Then attach the hoses to the filter and open the shutoff valves. The water in the inlet hose falls down into the filter, sucking water over the top of the tank and establishing a siphon. The water flows until the filter fills with water, then goes on up the outlet hose back up to the tank again. You end up with the filter and both hoses full of water. The filter pump isn't on during this, so no energy is needed to push the water back up the outlet hose.

Turning the filter on just spins a rotor in the pump to make the water move. No pressure is required to push it up to the tank level, since it is already up there - the needed pressure is provided by the head pressure from the filter inlet line. If there were no restrictions of any kind in the inlet and outlet line the pump would be producing no pressure rise across it at all. It seems weird, but it also seems to be true.

You could test this by removing the outlet hose from the tank, rotating it so the "U" fitting was slightly lower than the one on the inlet hose, and the outlet was aimed at a bucket. All of the tank water would begin siphoning out of the tank, with the filter pump off. You could also measure the head pressure produced by the filter pump by raising that "U" fitting above the inlet "U" fitting until the flow stopped. The difference in height between the inlet and outlet "U"'s would be the filter pump head pressure (minus the losses due to restrictions in the outlet and inlet lines.)EDIT: At the height where the flow stops, there are no losses due to restrictions anywhere, because there is no flow, so that height is the maximum head pressure for the filter.

I'm not going to try these experiments. I would for sure flood my floor if I did.

 

[hoppycalif]

Niko, a pump mounted below the tank, with its inlet and outlet fittings at the same height above the pump should be able to pump water a lot higher than the head rating of the pump, because the head needed to raise the water back up to the tank is provided by the water in the inlet line. Remember, the water in this closed system does not see atmospheric pressure, just the head pressure from being below the tank.

Of course there really are restrictions in both the inlet and the outlet hoses when there is flow through them. So, whatever pressure is needed to maintain that flow through those restrictions has to be provided by the rotor in the pump. The reason canister filters need to be below the tank is that if they aren't, the pressure loss from the flow through the inlet hose would reduce the pressure at the inlet to the pump too far below that required to maintain flow and the pump would cavitate. If the pressure loss in the inlet hose is 20 inches of water at 200 gph, for example, the filter would have to be 20 inches or more below the tank to keep the pressure in the pump above atmospheric pressure, where cavitation would occur.

 

[zer0zax]

If the inlet line and associated connections were fatter than the outlet, would the pump be able to produce more flow? It seems to me that having the inlet hose be the same diameter as the outlet hose would slow down the flow, the pump would have to work a little harder to bring in the same amount of water that it expels, especially if the pump is forcing water up quicker than gravity can bring it down. A larger inlet would flow at the same speed, but allow more water in the line. But then again, if the inlet was lower in the aquarium than the outlet, maybe water would flow 'easier' down the inlet because of more pressure...

I don't own a canister yet, but someday I will! And I will also have a very wet floor, but at least my curiosity will have been satisfied for 5 minutes.

 

[hoppycalif]

Being a retired engineer, with much of my career spent working on hydraulic systems, I find this fascinating to figure out. Some people enjoy fish, I enjoy pumps.

 

[niko]

Zer0zax,

A skinny outflow and a fat inflow does not help the canister filter flow. That is exactly what my Eheim canister filters had - 1" inflow hose (yes one inch diameter) and 1/2" outflow hose. Flow was pathetic.

The Otto canister that I talked about above has 2 identical thick hoses. 3/4" diameter each. This thing moves water much better than an Eheim pump rated the same gph so it maybe the pump is somehow better than an Eheim, although I don't understand how.

--Nikolay

 

[zer0zax]

Crazy! Honestly I thought the principles of how a canister works would be simple, like 2 gravelvacs connected to a pump. The gears in my head start grinding when I add in the sealed container, pump, and the water moving back up part. Niko, your making my head hurt! Maybe the construction of the Otto, how water flows inside the canister itself is what makes it more efficient?

Hoppy-I'm glad you raise up these questions! Most of the stuff you post is pretty complex, but after reading and grinding gears I start learning more and more. This hobby has tons of different aspects, and by focusing on each one we all understand the system as whole that much better.

 

[hoppycalif]

Let's complicate it some more: The filter has a motor that has a power rating of "W" watts. That power is converted to pressure times flow rate by the pump, with an efficiency of "E"% (expressed as a decimal). So, the head pressure the pump can produce will be 117*W/[E*(gpm)]. Now all you need is the power rating of the filter and you can make a chart of head vs gpm the filter will produce. Unfortunately, the efficiency will vary with head presssure.

Fellow engineers, please double check this!!

 

[Logan's Daddy]

Good reasoning hoppy, however the W will vary from model to model as the rating applied to a pump is actually the number of watts it consumes, not how much it puts out. Different pumps are going to be more or less efficient than others.

I think the only positive way to test would be real-world trials with different head-heights on each pump in question.

-on a side note, your diagram at the beginning of the post is exactly right. In reality the only "head height" any canister style filter will ever see is any distance above the water line that the return line ends up so it can clear the edge of the tank, usually about 3" or so.

Anyone that disagrees should think about it like this: in a reef tank if you have a pump in a sump sending water up to a display tank it needs to be a lot more powerful than a pump being used for a "closed loop" system (which is basically a canister filter without the canister - used to increase flow only without equipment in the tank). If both pumps sit under the tank the at the same level you generally need less than half the power on the closed loop to get the same flow you get from the return pump, the only difference? the closed loop does not have to overcome head height, the pressure of the incoming water offsets the pressure needed to push the return water back up.

The easiest way to illustrate this is to simply turn off the pump, the water level in the tubes will equalize near the level of the surface of the display tank, the head height the pump needs to overcome is whatever tubing is left above that level - not much.

 

[hoppycalif]

Yes, the wattage will be different for different filters, and it is a measure of power consumed by the pump. But, all canister filters, and almost all powerheads use the same design pump, with is just a rotor that uses centrifugal force to push water out of the pump. That type of pump doesn't generate much pressure at all, and I suspect that all of them have roughly the same efficiency. Higher power ones just have slightly bigger diameter rotors with more paddle blades. It takes more power to spin the bigger rotors with more blades. The efficiency of that type of pump is zero if you block the outlet, and goes to a maximum value if the outlet is completely open and very short. I have no idea what that efficiency is, but it would be easy to calculate it by looking at flow vs head charts for those pumps. I'm guessing it is never very high.

 

[hoppycalif]

To give proper credit, finally: It was OldMan's post in Ghengis's thread that triggered my starting this thread. That was a very perceptive post he made and I am grateful that it started my sluggish brain to functioning!

 

[stuckintexas]

Yes, the wattage will be different for different filters, and it is a measure of power consumed by the pump. But, all canister filters, and almost all powerheads use the same design pump, with is just a rotor that uses centrifugal force to push water out of the pump. That type of pump doesn't generate much pressure at all, and I suspect that all of them have roughly the same efficiency. Higher power ones just have slightly bigger diameter rotors with more paddle blades. It takes more power to spin the bigger rotors with more blades. The efficiency of that type of pump is zero if you block the outlet, and goes to a maximum value if the outlet is completely open and very short. I have no idea what that efficiency is, but it would be easy to calculate it by looking at flow vs head charts for those pumps. I'm guessing it is never very high.

I know you are an engineer so i just want to clarify something. I believe you are talking about centripetal force because I am under the impression that centrifugal force does not exist...at least that is what my physics professor always said...if I can remember correctly as it was a while ago.

 

[hoppycalif]

I know you are an engineer so i just want to clarify something. I believe you are talking about centripetal force because I am under the impression that centrifugal force does not exist...at least that is what my physics professor always said...if I can remember correctly as it was a while ago.

Physics professors always tell their classes that. Centripetal force is the force required to keep something going around a pivot - the string when you twirl a rock tied on a string. When you take a wet bicycle wheel and spin it, it throws water off. That is what is meant by centrifugal force - it is the lack of the centripetal force holding the water in place that spins it off. If you were riding along on the rim of the bike wheel, you would see the water being accelerated off, by something - centifugal force. I never had the nerve to challenge my Physics professor on that, nor would I have known enough to do so. He was my all time favorite college prof.

 

[stuckintexas]

i thought that "center force" is the force that throws the water off the spinning wheel? heh, i really liked my physics professor too, he was the best ive had...really really cared about getting the students to understand. not just going through the motions.

 

[BryceM]

Hoppy, you're pretty close, but I've got a couple of issues with your original post. The pump in a typical canister is located after the media and before the return pipe. The pump imparts added pressure, so the pressure on the outlet side is 36" + delta P, making it very easy for the pump to push water up the return pipe. It will actually be able to raise a water column to somewhere between 2' and 10' above the aquarium surface level depending on the specific pump. There is no siphon effect on the outlet side. A siphon implies negative pressure. Powerheads manage very little head because their pump vanes are designed to operate in near-zero head conditions.

In actual practice, the head at the inlet side of the pump will be something less than 36" because the situation is not static. There is head loss at the inlet strainer, along the inlet tubing, at fittings, and through the filter media. The effect of same-diameter fittings is minor. By far, most of the energy lost in the system is due to high-velocity flow along a long section of relatively narrow tubing. The velocity of flow in the media portion of the canister is very small, since the diameter is enormous, even when the media is fairly dirty.

It is possible to develop a negative pressure situation if the inlet tubing is too narrow, if it is too long, or if the pump isn't located at least a little bit lower than the aquarium. Eheim inlet tubing sizes are usually larger than outlet sizes to ensure that a negative situation (with resultant cavitation & noise) doesn't develop.

Thought experiment: A canister filter placed 100' below an aquarium will have very low output compared to one placed 3' below the aquarium, but it's only because of the energy lost along 200' of tubing. For comparison, a canister filter located 100' below the aquarium that is serviced by 6" diameter inlet and return pipes (almost zero flow velocity) will develop more flow than one 3' below the tank using typical 1/2" tubes.

Next thought experiment:

The loss of head (pressure) in a horizontal pipe due to friction (turbulence) is expressed by the Darcy-Weisbach equation:

f is a coefficient that comes from a Moody diagram (which accounts for laminar vs turbulent flow)
L is length of the pipe
V is the velocity of flow
D is inner diameter of the pipe
g is the gravitational constant (32.2 feet per second squared)

For our purposes, assume f is constant. Within reasonable limits, this is an ok approximation.

Assume a 1" inner pipe diameter, a 10 foot section of pipe, 500 gallons per hour flow, and an f of 0.03 (a typical value).

Velocity in this setting is 4.7 feet per second and total head loss is 1.25 feet.

Change the pipe to 3/4" diameter and velocity becomes 8.4 fps with a head loss of 5.3 feet.

Just for fun, assume 200' of 6" diameter pipe. Velocity is 0.13 fps and headloss is 0.0032 feet.

Lets go back to our initial values of 10' of pipe (typical) and 1" tubing (generous). Now, lets use 9.8 feet of 1" pipe and 0.2 feet of 1/2" pipe to simulate the fittings. If there is a smooth transition zone into and away from the fittings (to minimize turbulence), this is a reasonable assumption.

Head loss from 9.8 feet of 1" pipe is 1.23 feet.
Head loss from 0.2 feet of 1/2" pipe is 0.80 feet.

The energy loss in a 0.2 foot length of 1/2" diameter tubing is almost as large as the energy lost along 9.8 feet of 1" pipe!

In real life, when you turn the power on, the flow through a canister filter will increase from zero to a value where the pressure (energy) lost from turbulence (friction) equals the pressure (energy) added from the pump. As flow rates increase, pumps have less ability to impart energy (see a pump curve). At zero flow (shutoff head) pumps will produce the most pressure.

As flow picks up, velocity in the pipes also increases. The added velocity produces friction which robs energy from the system (headloss). Once the sum of headloss from each section (pipe, fittings, media compartment) equals the head added from the pump, a static condition is reached. There are a lot of moving targets here. Water acts funny when it transitions from laminar to turbulent flow. Centrifugal pump curves aren't straight lines (despite what Eheim publishes).

Fascinating stuff, fluid mechanics. Pump designers are smart dudes!

 

[BryceM]

Yes, the wattage will be different for different filters, and it is a measure of power consumed by the pump. But, all canister filters, and almost all powerheads use the same design pump, with is just a rotor that uses centrifugal force to push water out of the pump. That type of pump doesn't generate much pressure at all, and I suspect that all of them have roughly the same efficiency. Higher power ones just have slightly bigger diameter rotors with more paddle blades. It takes more power to spin the bigger rotors with more blades. The efficiency of that type of pump is zero if you block the outlet, and goes to a maximum value if the outlet is completely open and very short. I have no idea what that efficiency is, but it would be easy to calculate it by looking at flow vs head charts for those pumps. I'm guessing it is never very high.

The design of various centrifugal pumps is highly variable depending on what you're trying to accomplish. Some of them can develop tremendous pressures. Rotor speed, vane angle, vane geometry, vane number, inlet and outlet geometries, and other factors all come together to produce a pump with a given set of characteristics.

A pump's maximum effeciency is usually somewhere in the middle of it's operating range. Pumps acting against zero head actually aren't very effecient:

 

[hoppycalif]

The pumps used in our filters and powerheads are not typical centrifugal pumps. They are simple devices that just crudely spin a wheel with a few vanes on it, relying on the vanes to sling water out the outlet. It isn't until you start spending a lot more $$$ that you get into more sophisticated designs that are customized to the application. I doubt that any of the powerheads or filters we use have even a 30% efficiency, and any pressure load on them at all reduces that to near zero. That they work at all is a miracle.

By contrast, the Koralia type of "powerhead" doesn't use a centrifugal pump, instead using a propellor to move the water. It is much more efficient, but probably produces much less head pressure. It just moves a load of water.

The lesson I come away with, from all of this, is that in our real life applications, the height of the tank above the filter is not important. It can be 3 feet or 5 feet, and if the filter doesn't spring a leak, it works the same. But, when we add things like reactors, heaters, UV filters, valves, elbows, spray bars, etc. we can use up all of the available pressure output of the filter, and even before we do that, the flow from the filter will have dropped way down. Knowing that, I would not have made my valve manifold, that I described in a DIY article, like I did. I gained convenience, but probably lost half the flow rate from the filter.

 

[BryceM]

Well, I'll grant you that most of the ones in the hobby are straight-vaned, made of plastic, and of a fairly simple design. Their efficiency isn't too high, but it isn't as if the canister filter manufacturers haven't thought about it. They're probably as efficient as they can be given the constraints of cost. They work well enough, no? Raising the price of a canister filter by $200 to make it 80% more efficient is a loose-loose for everyone.

An Eheim (which is fairly low-flow by design) can be made much more efficient by up-sizing the tubing diameter. If you go up even one size beyond the usual recommendation you should be able to run a reactor, a UV unit, or a manifold system with even higher flow rates than with a "clean" installation.