Required sump flow if using closed loop

[das75]

Doing some planning.

Have a 75g with in tank skimmer. Thinking about changes such as adding sump in the basement and because of head loss etc, using a closed loop.

Seen the numbers like 10x, 15x, 20X for turnover for tank but what type of flow for the sump?

 

[Anonymous]

Those x numbers are a little misleading. You don't really need that much flow through the sump, but if you are keeping sps you want more flow in the tank. I prolly run 10x through the sump, but have much more circulation in the tank.

 

[Anonymous]

In reality you only need enough flow through the sump to feed the skimmer. I like 10x flow through the sump and about 50x through my CL.

 

[Mihai]

Can anyone come with an explanation for the need for 10x through the sump? Why not 1x or 2x?

I like 10x flow through the sump

Personal preferences do not qualify as explanations

Mihai

 

[wade1]

I would say the largest reason is for aeration in the main tank. Movement of water down to the sump, thru skimmer, and back provides a great deal of CO2 removal and oxygenation to the tank.

I would also contend that you should flow more than the skimmer can take in, in order to not "waste" effort on the part of the skimmer. (ie- if the skimmer removes what it can from the sump, theoretically of course, then you want that volume being replaced by 'dirty' water from above at least as fast as the skimmer can take it in.

 

[ChrisRD]

In addition to what Wade said... ...most people locate their heaters, top-off, reactors, etc. in or discharging into the sump so turning over the main tank every 5 - 10 minutes makes sense to keep things well mixed and uniform throughout the system. Too little flow through the sump might cause problems with that (and the stuff Wade mentioned above), too much usually results in too much noise/turbulence and makes it harder to keep skimmer bubbles out of the return pump.

The old rule-of-thumb seems to be about 5X to 10X. If you want a scientific study to back that up you might have to conduct your own... :wink:

 

[ChrisRD]

One example I can think of where you'd want to get specific about the flowrate through your sump is if you're trying to design it to function as a settling tank. Then you could calculate exactly how much flow should be coming through the sump based on the cross-sectional area and the size/density of the particles you're trying to settle-out. Needless to say most of us don't look that far into it (although I'm sure someone has! :wink: ).

 

[Mihai]

Both reasons make sense at the first glance. However, I think that they do not stand to a close quantitative scrutiny (see below):

I would say the largest reason is for aeration in the main tank. Movement of water down to the sump, thru skimmer, and back provides a great deal of CO2 removal and oxygenation to the tank.

Tanks can go skimmerless and without a sump as long as you break the surface to allow for the air exchange. This leads to the conclusion that an overflow and a skimmer doesn't improve the situation much irrespective
of the flow rate. If it would be to put some numbers up they will probably look like: sufficient aeration 50%, aeration without overflow: 70%, aeration with sump and skimmer at 1x: 99.5%, aeration with skimmer and sump at 10x: 99.7%.

The point is that the last 0.2% is truly not needed.

I would also contend that you should flow more than the skimmer can take in, in order to not "waste" effort on the part of the skimmer. (ie- if the skimmer removes what it can from the sump, theoretically of course, then you want that volume being replaced by 'dirty' water from above at least as fast as the skimmer can take it in.

I think that this is the same as the previous one: it will probably produce a 0.1% improvement in the skimmer efficiency, but not more. That is unless you get really big spikes (e.g., say you drop a bottle of Windex in the tank and you want all water to go through the skimmer as fast as possible - assuming the skimmer gets the Windex out - which I doubt, but just to make the point). Other than that, the skimmer at 1x will make the water very clean before the return pump sends it back up, so you'll have like a nice "water change", as opposed to just "slightly better water" that results from the 10x. I'm sure I can write the equations if needed, but I think that you get the intuition from what I'm saying.

The argument for CO2 is on a similar rational, but it's stronger than for the waste removal: the CO2 removal has a maximum that you cannot go over, while for the skimmer the assumption is that you can always remove more if you skim more.

Mihai

 

[PJsea]

Maybe someone can come up with the more scientific explanation but I can't rember the exact calculations right now. (There are calculations for this).
On the high end, you cannot have large amounts of flow going over the skimmer( I mean the skimmer for overflow box not Protein Skimmer) without worrying about levels needed to maintain their flow. You may also not want the excessive flows going thru refugiums and sumps.
On the low end you want a flow rate to provide for the aeration and "filtration" going on in your system so the entire tank contents are turned over within a reasonable time.
Look at it like doing water changes. if you did 10% changes it would take about 20 changes to reduce the concentrations of something by 90%. If you did 50% changes you could reduce the concentrations by about 97% in only 5 changes. (These are batch changes but continous processes can be mathematically described also).
I've seen numbers for sump flow that are down at 3-5X the tank volume per hour. These are probably adequate for sump flow.
Total flow in the tank would be determined more by what inhabitants you have in the tank.

 

[Mihai]

On the high end, you cannot have large amounts of flow going over the skimmer without worrying about levels needed to maintain their flow. You may also not want the excessive flows going thru refugiums and sumps.

The assumption here is that the skimmer goes with its own pump and the flow through the skimmer is, thus, independent of the flow through the sump. If you have a skimmer that requires a high flow and the only flow the skimmer gets is directly from the overflow, then sure, you need a high turn-over rate.

On the low end you want a flow rate to provide for the aeration and "filtration" going on in your system so the entire tank contents are turned over within a reasonable time.
Look at it like doing water changes. if you did 10% changes it would take about 20 changes to reduce the concentrations of something by 90%. If you did 50% changes you could reduce the concentrations by about 97% in only 5 changes. (These are batch changes but continous processes can be mathematically described also).

The problem with your logic is that they are not water changes: at high rates most of the water comes down only to go back again, no going through anything but the return pump, and thus not getting any cleaner, just circulated.

M.

 

[ChrisRD]

At lower turnover rates more stuff is going to settle out in the main tank and not make it to the sump (and therefore the skimmer).

FWIW, I don't think a particular number really matters as long as things stay homogenous.

 

[Mihai]

At lower turnover rates more stuff is going to settle out in the main tank and not make it to the sump (and therefore the skimmer).

I agree with you now. This enters the category of lots of stuff put in at once (e.g. food). If you have a high turnover much of the food will get sucked in the overflow and eliminated from the display tank. However, I doubt that this is what you want -> just feed less.

The only case when this would be beneficial is if you blow the detritus and want it in the sump/skimmer rather than back into the tank... then yes, it matters.

I agree that if it's homogenous it doesn't matter.

M.

 

[wade1]

Actually, this is an endless and unsolvable arguement. The reason being that very different mechanisms are in use for each of the facets: total flow in display, filtration, etc etc.

As for the above number crunching, you must maintain a flow into the skimmer of maximally polluted water. Otherwise, you are wasting effort, hence my comment about new water supply at the rate of skimmer input. That also changes drastically based on skimmer type... ie- a beckett turns over 1000gph, whereas a needlewheel is hugely less.

As for gas balance (O2 and CO2) you can't underestimate the consumption and production by critters in the tank. All of which is balanced ok by increased surface contact (skimmer is still 100-500x better at gas exchange, hence I use them in my research). You cannot reach supersaturation with moving water alone near a surface.

 

[ChrisRD]

Actually, this is an endless and unsolvable arguement.

Awh, cummon Wade, we LIVE for those here... :lol:

...you had to go get all reasonable on us and probably just killed 10 pages of debate now... :roll:

 

[wade1]

Ok then. Lets try this tack...


My tank is the best, the rest of you are no where near the reefer I am. And I say you must have 2x the skimmer intake into the sump for it to be effective... and my corals are pretty... and and and....



I think if someone really wanted to try, we could do a mass balance equation for this. If you guys toss me some cash, I'll do a closed system, with a measureable compoun thrown in... then I can do a mass balance based on detection of each fraction in skimmer waste, sump, and main tank.

 

[Mihai]

Well, we'll get numerical then. But not now, I gotta go now. Tonight I'll come up with some numbers to follow on my intuition.

Chris, don't worry, a call for reason never stopped anyone from endless debates, why do you think it would now?

M.

 

[Anonymous]

I just don't understand how a lot of flow through the sump could be bad. I have a very slow turnover rate right now on my sump, but only because it's shared by two tanks--one soft one SPS--and I want the water to be well filtered by the skimmer and carbon before passing between the two. <anecdotal evidence> It seems like the skimmer is not all that efficent compared to how it has performed in the past <off> Beyond the noise issue and the bubble issue (which is pretty easy to solve) why wouldn't you have a higher flow rate (around 10 times or so) through your sump?

 

[Mihai]

I just don't understand how a lot of flow through the sump could be bad. [...] Beyond the noise issue and the bubble issue (which is pretty easy to solve) why wouldn't you have a higher flow rate (around 10 times or so) through your sump?

Exactly this is the issue. You will get more noise from the flow, more bubbles and you'll need a stronger pump -> even more noise, more heat in the system, more energy drain -> more coal, global warming, etc. .

In short, it's wasteful and unnecessary.

If you want a reef-keeping related reason -> to allow the detritus to settle in the sump (so it's broken down close to the skimmer/scrubber that remove them, or even to siphon it).

M.

 

[Anonymous]

From an industrial stand point. Sumps used in service water and cooling towers are kept below 1 ft/sec for proper operation. This would be the max you would want to run. It may be loud. As for min, my toughts are just over 1x the skimmer rate so you do not reskim any water. If you have a heater or chiller in the sump it would depend on the cooling or heating power the flow rate could be optimised for the tank if you feel like doing the heat transfer equations.

 

[Mihai]

OK, here's the inside scoop: I simulated the levels of nasties in the main tank and in the sump with 1x (red) and 10x (blue) flows and that's what I came up with. As you can see it's fairly close except right when the nasties get into the tank (I assumed that bad stuff gets added to the tank once every 24 hours). I was not sure how to model the protein skimmer -> I assumed that it gets out 1% out of all nasties that it goes through it and it has a flow of 10 times the tank volume every hour.

As you can see it's not much difference after all. Furthermore, the simulation is tilted towards 10x: the assumption is that all nasties get into the tank in half an hour, while in reality it takes significantly longer than that, and the spike is avoided.

Regards,
Mihai