Firstly, I would recommend that you fire up your web browser, as there is going to bequite a few diagrams and graphs presented here. Actually there is a lot, and youwont get quite as much of it if you can't look at the graphs and diagrams.

A couple of weeks ago there was a discussion on the best place to have a skimmer located,and which one was more efficient.

In order to sort out what most people had been saying for quite sometime, I decided tosimulate the skimmer/aquarium system to actually find out what was happening.

The simulator uses a simple stepwise calculations method. What this means itthat is moves the water around the system, then calculates the state of the system, thenmoves move water around. As long as the time step employed is small enough, then thesimulation is close to what actually happens in a continuous system.

Note that most of the results attained from this simulator are nothing special, they arejust confirmation of what is basically common sense. But it is always nice to havesomething to back up thoughts like this.

Systems Simulated

There are seven different system arrangements that the program simulates. Two ofthem are ideal cases, so are not really applicable to real life. The remaining fiveare more realistic, and should not be too far away from what really happens in ouraquariums.

Fully Mixed Tank Only

This simply consists of a tank with a skimmer attached. The tank is fully mixed,which means as soon as some water is added back to it from the skimmer, it is mixedtotally with the rest in the tank. This is an ideal case, so is far away from whatactually happens. The reason I looked at the ideal cases, as they are an easy placeto start, and they do tell you a fair amount of what is happening.

Partially Mixed Tank Only, Return Same End

Now the tank has been divided into four equal parts, with flow between these sections. The skimmer takes water from one end, and then returns it to the same end of thetank. This is basically what most hang on the back type skimmers are arranged like.

Partially Mixed Tank Only, Return Opposite End

Same tank set up, but now the skimmer return goes back into the tank at the oppositeend of the tank. This is the arrangement that most people tell you to have in youraquarium, and I will be able to show you actually why shortly.

Fully Mixed Tank and Sump

The tank and the sump are both fully mixed, this is the ideal case. Water thatenters the tank or sump are instantly mixed with all the other water in that tank. Once again, not a very realistic system, but it does give a starting point.

Partially Mixed Tank and Sump, Return Same End

The tank is again divided into four equal sized compartments, and the sump into two. The water from the sump is returned to the same end of the tank. If theskimmer takes water out of the sump faster that the return rate, then water is simplyrecycled within the sump.

Partially Mixed Tank and Sump, Return Opposite End

This is how a typical tank is arranged, water is draw into the sump at one end, theskimmer takes water from the sump and returns it to the sump, and water is then returnedto the tank at the opposite end of the tank. And the last one is...

Partially Mixed Tank and Sump, Return Opposite End, Tank Fed Skimmer

This arrangement is the same as the last, but instead of water being removed from thesump then fed to the skimmer, it goes straight to the skimmer. If the skimmer rateis larger than the return rate, then water is drawn from the sump to make up for it.

This is what got me into the discussion to start with, where it would be better to takewater directly from the overflow into the skimmer, then into the sump.

Definable Parameters

The following system parameters can be defined within the simulation program, and it isquite flexible with what you can do.

Tank Volume (litres) The amount of water that the display tank holds.

Sump Volume (litres) The amount of water that the sump tank holds.

Return Rate (litres per hour) The flowrate of water from the tank to thesump, and from the sump back to that tank.

Skimmer Rate (litres per hour) The flowrate of water through theskimmer.

Tank Rate (litres per hour) The flowrate of water within the tank,excluding the skimmer or return pumps i.e. circulation pumps/powerheads.

Initial DOC (milligrams per litre) The initial concentration ofdissolved organic material in the system.

Skimmer Efficiency (%) The percentage of the total amount of DOC theskimmer has access in a single pass it can remove.

Skimmer Exit DOC (milligrams per litre) The exit concentration of DOCfrom the skimmer.

Increase DOC in Tank (milligrams per litre) This is a feature that canbe used to see what happens when there is DOC input into the system. The value of thisparameter sets how much the DOC concentration in the tank is increased.

Increase Period (minutes) This determines how often the Increase DOC inTank occurs.

Time Step (minutes) The time between calculations in the simulation.

OK, now I am sure that some of you are wondering about a couple of those parameters...like how do we know what the skimmer efficiency is? Or even what is the level of DOCin the system? Well, the short answer at the moment is we have no idea .... OK, maybe a little with the skimmer efficiency as some initial work has been done on this. But this does not stop us from being able to use the results that we attain. You can still make comparisons, and say something intelligent about what ishappening, even though the actual values might not be real.

Simulator Results

OK, now it is time to go through some of the results that you get out of this simulator. It should all start to make a bit more sense from here on in. Firstly, thebase systems is as follows, i.e. the common values better each of the graphs shown unlessstated otherwise.

Base System

Tank Volume : 1,000 lt

Sump Volume : 500 lt

Return Rate : 5,000 LPH

Skimmer Rate : 5,000 LPH

Tank Rate : 5,000 LPH

Initial DOC : 100 mg/l

Skimmer Efficiency : 30%

Skimmer Exit : 10 mg/l

Time Step : 0.30 min

Now the thing with those system values are that they are scaleable. What this means isthat say you have a tank that is only 100lt. Well then if you skimmer then moves 500LPH, then the graphs still apply, i.e. five times the tank volume per hour is cycledthrough the skimmer. Time for some graphs .....

Tank and Sump Simulations

This is a graph of the four different base systems simulated with the tank and sump. As you can see from this, the end that you have the return go to makes a realdifference to how fast the DOC is removed i.e. the pink and yellow lines. And withthe same skimmer rate, it is much better to first send the water through the skimmerrather than into the sump i.e. the yellow and blue lines. So, the take home messagefrom that graph is you should have your return on the opposite end of the tank, and takewater from the overflow and put it through the skimmer.

Tank Only Simulations

Once again, the return at the opposite end of the tank makes a difference. Onething you might be asking now, is why that the ideal, fully mixed, systems are worse offthan those that are partially mixed. The reason for this is that the amount of DOCthat is removed is determined by the concentration. In the fully mixed system theDOC available to the skimmer is reduced faster than that is a partially mixed system. As a result it is then able to remove it from the system faster.

Skimmer Efficiency

The more efficient the protein skimmer is, the faster the DOC concentration goes down. Not that surprising that is it. ;-)

One thing you should note though, is that again the degree of improvement decreases for agiven increase in efficiency.

So there will be a point reached where you will no longer get a substantial improvement, i.e.spending more on a better skimmer would be better off spent on some other things that canbe done to make it work better.

Tank Circulation Rate

This would be expected to make a bit of a difference, but in actual fact it does notunless the return is at the same end of the tank. When the return is at the oppositeend, it fractionally decreases the amount you can remove as you are mixing things in thesystem faster. So the rate your water is moving around that tank does not make anydifference, unless you have it returned to the same end. This is applicable to people thathave hang on the back type skimmers.

Return Rate

It actually turns out that it is best to the the return rate around the same value asthe skimmer rate. I am not sure why this is the case yet, have to look into it more,as it contradicts another result that you get when you increase the skimmer rate. You can see this by the fact that the yellow line, highest return rate, is abovethe line for the return rate half of this.

Skimmer Rate

With the skimmer taking the feed from the sump, running the skimmer faster and fasterimproves the speed at which DOC is removed. Note that as you get higher and higherskimmer rates, then it will start to slow down the amount of improvement you get for agiven rise in flowrate. So there is a limit here where if you go to a bigger pumpthen you are just wasting money and power to get only a little more efficiency. Thisis the one that goes against what is in the above graph with the Return Rate.

I have some ideas on why this is so, but I have to look into it more before I can reallysay anything about it.

Tank and Sump Relative Sizes

If you keep the system total volume the same, but change the ratio of the tank to thesump volume, what happens? Well, basically not much at all with this particular basesystem. It really doesn't matter too much.

Skimmer Exit DOC

No surprises here, the lower the DOC from the skimmer, then the better off you are. And finally the last graph I have to show at the moment...

Tank Fed versus Sump Fed and Double Skimmer Rate

Another arrangement that people tend to use for their skimmer is to have it runningfaster than the return. Therefore water is recycling around in the sump. Thisgraph shows that if you just have the skimmer operating at the same rate, then you arebetter off to have the water coming directly from the tank to the skimmer. If youthen run your skimmer at a higher rate, then you are better off. So lets say youhave a certain sized pump running your skimmer. If it is close to the rate that yourreturn runs at, then you are better off feeding water directly to the skimmer. If onthe other hand, you have the choice between the skimmer fed from the tank and the skimmeroperating at double the rate then go for the latter. That is all I have to say aboutthe results, I will just go through something about the simulator if you happen to want todownload it and try it out.

Downloading the Simulator

It is available in two file formats, Excel 97 ( skim97.zip ) and Excel 95 ( skim95.zip ). The final file size is around the 3M mark for both of these, and the compressedversion is about 1.2M. So you need to have some file compression/decompressionprogram to be able to use them. Something like WinZip or PKZip would be fine.

If you are going to play around with the simulator, then it consists of the followingsheets, so that you know your way around. I have attempted to set it out nice and simply,so anyone can really have a play with it and get some interesting numbers out.

Sheet 1 : Parameters and Results of the 7 Systems

Sheet 2 : Graph of Comparison of the Three Different Skimmer Locations

Sheet 3 : Graph of the Four Tank and Sump Simulations

Sheet 4 : Graph of the Three Tank Only Simulations

Sheet 5 : Sample Results This is where you can copy some results from differentsimulations so that you can make direct comparisons.

Sheet 6 : Graph of Sample Results

Sheet 7 : Tank and Sump System Calculations

Sheet 8 : Tank Only Simulations

The last two are where all the calculations occur. You can have a look at some otherinteresting occurrences happening there, such as how the concentration in the differentcompartments change with time.

OK, that is all I have to say at the moment, time for any questions you might have.

How long did it take you to setup and perform the calcs. for each model?

Each one it took about 30min to get it how I wanted it. I started with the easiest,then got more completed with each step.

And then used each one to build up to the next. Then I like to make things pretty,such as those lovely colours ;-)

If I don't have a sump what does it mean contrasting to what you are saying, DBW?What is applicable?

If you don't have a sump, then what you need to do is either have the return to theother end of the tank i.e. not be able to be mixed quickly with water it is taking in... or have a lot of circulation in the tank if you must have it returning near to theintake. That will make it perform better. What you are aiming here is for there tobe as higher concentration of DOC entering the skimmer intake. And both of those steps gotowards that goal.

Tank circulation rate does not seem to effect system export efficiency, but it isstill important, isn't it?

No it doesn't effect the rate at which the DOC is removed and reduced in the systemfor the case that I showed the graph for.

But I am not saying that is not important, it is important for other reasons than skimmerefficiency and how fast it can remove DOCs. It will be important though if you havethe skimmer return at the same end that its intake is. Then you increase the DOCconcentration in the intake by mixing the more pure water straight from the skimmer withthat in the rest of the tank

I noticed on most graphs that once the doc reaches 20 the graph bottoms out... doesthat mean that the cleaner the water gets the harder it is to remove the doc?

Yes, that is what it means. And there are various reasons for that. The way a skimmeroperates, or the way I think it does and how I think it can be simulated, is as follows:

Lets say 100 mg/l of DOC goes into the skimmer. It then has the ability to removedown to say 10 mg/l i.e. that will be the exit concentration. It will not be able toremove all of it, that is just not possible. But on top of that you have to includehow efficient the skimmer is. In these simulations I have said 30%, which is a number thatsomeone came up with in a study recently. I believe it might have been Richard Harker. So, with that 100 mg/l water, it can get it down to 10 mg/l, but it is only 30%efficient. What does that mean? Well, it can only get 30% of the way to the lowest exitconcentration. In this case, that means that exit concentration will be 73 mg/l..... 30% of (100-10) is 27 ... so it can remove 27 mg/l in passing that water through it.

So each time the water passes through, it gets closer to the final value, but not quitethere. A good practical way to look at this is stand at the end of a room. Now move 1/2way to the front of the room. Then repeat that again, more 1/2 to the front of theroom. You are now 1/4 way to the front. Repeat that hundreds of times. Will you reach theother wall? No you don't, you get closer and closer, but never make it. It if verysimilar to that.

What do you think would be a good range of values for skimmer efficiency?

That is not an easy one to answer. There really needs to be some testing done on that.The only one I have found or been told about said that the skimmers were 30% efficient. Ireally can't say much more than that as I don't have a feel for how they perform etc.

Would it be possible to bound the thought experiment though? Say that a smallskimmer is X efficient and a Downdraft skimmer is Y efficient?

Well, with this simulator it is actually a little simpler than that I think. Theefficiency is linked to the bombardment rate. You can read up that in Escobal's book, hecovers it in detail. So you can then get a relative measure of how efficient eachtype of skimmers. Then you can feed in a certain flowrate through it, i.e. a size ofskimmer, and the efficiency will be similar. What I mean here is a small counter currentskimmer should be just as efficient or the same efficiency as a large one. That isall assuming that they are desgined in the same manner. You can then say that may bea downdraft is twice as efficient as a air driven skimmer, or something like that. Theproblem is we don't really know, or have a measure of how different they are. That's how Isee it anyway.

All graphs have shown that after about 200 minutes most systems are equal no matterwhat you do.... in the long run does it matter how the system is set-up since most removalcomes within the first 100 minutes?

Ahh, yes glad that someone asked that ;-) The reason that it doesn't make anydifference is because in the end that point that is being aimed for is the same. In thecase of most of the graphs shown, it was a minimum DOC concentration of 10 mg/l

Increasing the skimmer rate, return rate, etc. changes the speed at which it gets there,but does not move the end point.

Now, the thing that does make it important is when you start to add some generation of DOCin the system. This is where it stands out. I have only played with it for a littlebit, but what comes out is the faster that the DOC levels are reduced the better off youare. If you have a regular input of DOC, then speed is what counts. As an example,say the skimmer is 15% efficient, and compare that with one that is 45% efficient. Now, starting with 100 mg/l in both cases, and add 1 mg/l to the tank every 1minute. What you find is the 45% efficient case levels off within 200 min or so to around40 mg/L. But with only 15% efficiency, then it is roughly double, around the 80 mg/lmark.

So according to one graph, the optimal skimmer is one that removes a lot of waste atup to 10,000 lt/hr (obviously). Do you believe such a skimmer exists?

That is actually the spot where I can't say anything. I don't actually know that muchabout commercially available skimmers etc. The thing is that I don't have much exposure tothem here in Australia. Not many people most have to be imported.

What I have written here, and the principles behind them is not that hard. Anyone with abit of mathematics, or even just common sense could easily make something very similar. That is about all I can think to say about that question.

One thing you should also note, is the numbers that I put in there were just ones toillustrate relative points. Increase this by that amount this happens.

Thanks for the great talk, Dallas!