Continuous (automatic) water changes?

Continue water change is very popular in flesh water goldfish and discus farms because
1. flesh water is cheap there
2. It is the best way to remove excess organic waste

But in saltwater, we have protein skimmer and saltwater is expensive. So even though continue water change is good but it is not necessary.
 
Math is cool isn't it?!? ;)

Actually, you can take this a step, or many steps further with the following equations:

dx/dt = D(xnew - x)

Where dx/dt is the rate of change of any chemical or component in the tank. Xnew is the concentration of that component in the water you are pumping into the tank, and x is the concentration of the component in the water at any time t. D is called the dilution rate, defined as the rate of liquid pumped in divided by the size of the vessel (tank).

If you do a little calculus you can solve the differential equation and come up with the following:

ln [(xnew-x2)/(xnew-x1)] = - D (t2-t1)

Where x2 is the concentration at any time t2, and x1 is the concentration at any time t1.

The following is an example of how you could use the equation. Suppose you measure the magnesium level in your tank and find it low at 1100 ppm. You test your freshly preped Instant Ocean water and find it at 1350 ppm. You decide that you don't feel comfortable with magnesium that low in your tank and decide you want it at least up to 1300 ppm. You set your automatic water change system to 1 gallon per day, and your tank is 100 gallons. D = 0.01/day (I know reciprocal days are hard to comprehend, but don't give up yet!). So how long would it take to get back to 1300 ppm?

ln [(1350-1300)/(1350-1100)] = - 0.01 (t2-t1)

ln means natural log, BTW, you know that button on your calculator you haven't used since 10th grade!. The solution....161 days...wow, long time!

You can assume any rates you like for any of this. How about 10 gallons per day? It would then take 16 days.

All of this of course assumes that the system is well mixed, and that there is no consumption or production in the system (obviously not exactly accurate, but depends on the component). We can add additional terms for consumption and production, but the equations start to become more difficult to solve, and it helps to have accurate ideas of the rates of consumption, etc.

You could also do the calculation if you were worried about the level of some contaminant in your tank. Suppose you dosed with copper and wanted to know how long to get back to a low ?safe? level. Assume copper was at 0.2 ppm (therapeutic dose for ich) and that there was very little, or none in your make up water. How long to get to 0.002 (probably safe for most inverts)? At 10 gallons per day:

ln [(0-0.002)/(0-0.2)] = - 0.1 (t2-t1)

46 days. Of course copper tends to precipitate on things, particularly calcium carbonate sand, so it gets more complex, but anyhow, you get the idea.
 
greg, how often and at what rate to you change your water? do you use the formula above or do you guess at it like most of us?
 
I remember Greg saying he doesn't do water changes. Only when he sells a ton of frags will he have to add new salt water to replace.
 
Ed, no way do I pull out my calculator, though the math was fun for a kick above.

Wrassefan,

Basically that's correct for some tanks. Although I also use a fair amount of water when I'm fragging stuff out as well. For tanks where I don't really take much out I think a good plan in 10% or so per month. A good skimmer and some carbon can reduce the need for too many water changes, but you all knew that.
 
Hi Greg,

Thanks for the math tips. My original calculations were analytical as they were pretty simple, but when I did the ones with production I did it numerically. I couldn't figure a good way to deal with both the production and the discontinuities of large (not continuous) water changes. The pure side of me kind of feels that numerical methods are "cheating", but my math skills aren't what they used to be, and CPUs are really fast these days. I must say, the results with production surprised me, but I think they are basically correct - but then again, I'm often wrong... :)


Note: my first post was not really accurate, when I said "numerically" it didn't mean numerical methods - it was just the evaluation of the function, while the limit of the function was found using L'Hopital's Rule.
(ok - I know, things might be getting a little geeky here :cool: ).
 
Last edited:
Back
Top