CO2 Loss in Large Aquariums

(an experiment testing CO2 loss as a function of powerhead, trickle filter, and turbulence)

by George and Karla Booth [bibl. citation]
If you don't want to read this whole thing, jump ahead to their conclusion.
We are in the process of setting up a new 90 gallon plant tank based on concepts presented in Dupla's "The Optimum Aquarium". Before we finalized the set up, we had the chance to conduct some experiments in the bare tank. We ran a "CO2 loss" test recently and wish to share the results with the AGA. We verified that the major culprit in CO2 loss is surface turbulence (no surprise here). However, counter to our expectations, trickle filters DO NOT necessarily cause rapid CO2 loss.

The test tank was a "90 gallon" glass tank (48"x18"x24" tall, 79 gallons of water). It was set up with Dupla heating coils and a Dupla "DuplaTherm" temperature monitor/controller to maintain the water temperature at 79.0 +/- 0.1 degrees F. A Dupla Reactor "S" was used TO inject CO2 via the sump of a trickle filter.

CO2 concentrations and pH were measured with LaMotte test kits. Note that the LaMotte CO2 test kit has a resolution of 1 ppm (mg/l) and an error of about +/- 2 ppm. The LaMotte pH test kit has a resolution of 0.1 pH units and uses an "octet color comparator". The pH error IS ABOUT 0.05 pH units based on comparisons with a Sandpoint II pH/ORP controller and comparisons with a pH/KH/CO2 table.

An AquaClear 802 powerhead was used to circulate water in the tank. It was placed near the bottom in three tests to provide a gentle circulation current with little surface turbulence. A 0.3 ft/sec surface current was noted, giving a smooth surface pattern that looked like "heat waves" rising off a highway in the summer. In a forth test, the powerhead was placed at the surface and was adjusted to give vigorous ripples without splashing.

The trickle filter used was an Amiracle "100 gallon" unit with a bio-media capacity of 3.99 gallons. The media space is 16.125" long x 7.625" wide x 7.5" high. The media used was 238 Dupla BioKascade bio-balls, with the internal slats arranged roughly horizontally to allow the water to move through the media in a gentle, cascading manner.

The filtered water is circulated by a Quiet One pump controlled by ball valves to provide a 400 gallon per hour flow, turning over the tank five times per hour. The trickle filter has two water returns. One is directed across the bottom 1/3 of the tank, providing a flow at what will eventually be the top of the gravel. The other return utilizes a Magnum 330 water return fitting. To provide surface turbulence, a Magnum diffuser was used to direct the return flow across the surface, producing ripples equivalent to the powerhead when placed at the surface. For tests without turbulence, the diffuser was removed, allowing the water to be directed towards the bottom of the tank.

The tank is bare except for the equipment mentioned - no gravel, no livestock, no plants. Lighting is room ambient. The top is open.

Before we turned on any equipment, we filled the tank with tap water and adjusted the water hardness by adding 3 tablespoons of calcium carbonate (CaCO3) and 1 tablespoon of sodium bicarbonate (NaHCO3) to achieve a GH of 3.5-4 degrees and KH of 7 degrees as measured by a Tetra test kit. Note that GH has no bearing on the CO2 measurements, but a KH of over 3.5 is needed to ensure accurate CO2 test kit readings. A KH of 7 was selected to keep pH readings in the range of the LaMotte test kit.

After letting the water equilibrate for one day we measured dissolved CO2 at 2-3 ppm. We then set up a large powerhead to circulate the water (Project RS-500, ~500 gph) and let it run for a day. The CO2 remained about 2-3 ppm. At the end of most of the tests, CO2 again measured about 2-3 ppm, indicating that this was the equilibrium value for the experimental conditions (note that the altitude was 5000 feet above sea level).

After the initial tests, the heating coils, trickle filter and CO2 injection were set up. The large powerhead was replaced with the AquaClear 802. For the first test ("trickle, turbulence"), the trickle filter was run with the Magnum diffuser producing surface turbulence and with the powerhead running at the bottom. For the second test ("powerhead, turbulence"), the filter was turned off and the powerhead was placed near the surface. For the third test ("trickle, quiet"), the trickle filter was run without the diffuser and with the powerhead running at the bottom. For the last test ("powerhead, quiet"), just the powerhead was used, running at the bottom. In all tests, the trickle filter and reactor were used to get the CO2 level up to the point were the test started. At that time, the CO2 was turned off and the reactor allowed to clear of residual CO2 before actually starting the test.

Due to some difficulty in getting the CO2 to the same starting point for each of the tests (actually, a lack of patience on our part), the CO2 readings were normalized for the table below. To normalize the readings, the raw data was plotted with the CO2 concentration on a log scale. A best-fit line was drawn by eye through the data points. The numbers in the table below were then read from the plotted lines at hourly intervals. Just CO2 data in the range of 10-33 ppm is shown, since we consider that range to be the most relevant for planted tanks. The raw data is shown at the end of this note.


CO2 concentration (ppm)

               trickle,   powerhead,   trickle,   powerhead, 
Time (hrs)    turbulence  turbulence    quiet      
quiet -------------------------------------------------------------
   0             33          33           33        33
   1             21          24           28        28.5
   2             13.5        17           24        25
   3              -          12.5         20        21.5
   4              -           -           18        19
   5              -           -           14.5      16
   6              -           -           12.5      14
   7              -           -           10.5      12
   8              -           -            -        10.5 

At KH=7, the following table relates CO2 to pH:

   CO2 (ppm)  pH
   -------------
   42        6.7
   33        6.8
   26        6.9
   21        7.0
   17        7.1
   13        7.2
   10.5      7.3
    8        7.4 

Conclusion

What surprised us was the fact the trickle filter itself was not responsible for much CO2 loss (compare the last two tests). It should be noted that air was not pumped into the media chamber during the tests. We suspect that any out-gassing of CO2 by the media will quickly produce a concentration of CO2 in the chamber such that it reaches equilibrium with the CO2 in the water.

In our other trickle-filtered tank, we have noted very high CO2 loss (we go through a 10 pound tank in 6-7 weeks). We now suspect that the loss is caused by the Ehiem canister filter spray-bar return. We plan to run further experiments on that tank to verify this conjecture.

Although some authorities recommend pumping air into the media chamber of a trickle filter, we have found no evidence of a need for this. Thriving plants will provide plenty of oxygen for the aerobic bacteria colonies during the day and we have noticed no problems at night when the plants are at rest. We ran a long term test on another tank using a Sandpoint II pH/ORP controller and found no difference in ORP with the air pump on or off. We also noted that less CO2 was used with the air pump off (longer intervals between CO2 bottle refills). >From the table, it would appear that a 10 pound CO2 tank will last about 5 months when we finally get the new tank set up. We will have a KH of about 5 and will regulate the pH to be 6.8 +/- 0.5. This is a CO2 concentration swing of 5 mg/l (27 mg/l to 22 mg/l) times 300 liters and should occur within 1.25 hours for a usage of 29 grams per day. Of course, the usage by the plants will increase this by some amount, but that's another experiment!


CO2 test raw data

CO2 concentration in ppm and measured pH ()

 Clock          trickle,   powerhead,   trickle,   powerhead,
 Time          turbulence  turbulence    quiet       quiet
-------------------------------------------------------------
 6:00 pm          -           -           23 (7.0)  47 (6.65)
 7:00 pm          -           -           20        37 (6.75)
 7:30 pm          -          33 (6.8)      -         -  
 8:00 pm         27 (6.9)    27 (6.9)     17        31 (6.8)
 8:30 pm         20 (7.0)    24 (7.0)     -          -  
 9:00 pm         17 (7.1)    19 (7.0)     14        27 (6.9)
10:00 pm         11 (7.3)    14 (7.2)     10        23 (7.0)
11:00 pm          8 (7.4)    11 (7.3)     10        21 (7.0)
12:00 am          -           -            9        17 (7.1)
 3:00 am          -           -            -        15     
 8:00 am          2           2            -         8 (7.4)  
12:00 pm          -           -            -         5        

---------------------------------------------------------------------- George Booth "Nothing in the world is more dangerous booth@hplvec.lvld.hp.com than sincere ignorance and conscientious Freshwater Plant Tank Technology stupidity" - Martin Luther King, Jr. ---------------------------------------------------------------------

--



To Bruce Hallman's plant page.