Electrolysis Alkalization Experiments

From coral
Jump to: navigation, search

Applying a given current to a cathode with electrolysis should yield a predictable amount of hydroxide ions and resulting alkalization of the water. These experiments test how well these predictions work in practice.

4/13/2019

This is a simple experiment to do a basic test of the ability for electrolysis to alkalize water. A tub was filled with seawater from a lagoon, then divided into two compartments connected by a gap at the bottom. Electrodes were placed in each compartment, then used to perform electrolysis and alkalize one compartment while acidifying the other. The cathode compartment's pH was measured afterwards. The amount of current needed to achieve a certain level of alkalization was more than predicted, by a factor of 45% to 120%. It's not clear if this discrepancy is a problem with the theory, the equipment used, or the fundamental crudeness of the experimental setup.

Equipment:

  • HI98194 meter (measures pH/Salinity/Temperature among other things)
  • TekPower TP1540E power supply
  • 31 gallon rubbermaid tub w/divider
  • 2x MMO anodes

Initial setup:

  • Meter calibrated via quick cal HI9828-0 solution

Setup when replacing water:

  • Fill rubbermaid with 60 l of water
  • Place electrodes at opposite ends of rubbermaid
  • Connect electrodes to power supply

For each measurement:

  • Place divider approximately in middle. Divider has gap at bottom of ~1cm
  • Measure initial pH
  • Apply specified current for specified duration
  • While holding divider in place, stir cathode compartment with pH meter for several seconds
  • Measure pH in approximate center of the cathode compartment
  • Remove divider
  • If reusing water for the next experiment, stir contents, wait several minutes, then stir contents again

Data:

Cathode Compartment Size (l) Fresh water? Salinity (S) Temp (°C) Initial pH Current (A) Time (s) Final pH Actual A s Predicted A s Percentage over
30 yes 36.28 30.28 8.13 4 180 8.30 720 396 82%
30 no 36.28 30.28 8.14 4 90 8.25 360 249 45%
30 no 36.28 30.28 8.17 3.8 60 8.23 228 137 66%
30 no 36.28 30.28 8.19 4 120 8.30 480 267 80%
30 yes 36.25 29.83 8.13 4 120 8.26 480 294 63%
30 no 36.25 29.83 8.15 3.85 60 8.20 231 110 110%
15 yes 36.3 29.95 8.16 3.3 60 8.24 198 90 120%
30 yes 36.25 29.94 8.16 4 120 8.26 480 230 108%

6/15/2019

Setup: 2x 15.2cm x 15.2cm x 1mm titanium plates

One side of each plate covered with electrical tape

Per Abdel-Aal et al 2010, at 6 V hydrogen production for their cell is roughly 20 ml/h. At a rate for another voltage of 85 ml/h from 120 mA cm^-2, this gives 20 / 85 * 120 = 28 mA cm^-2. With a 15cm x 15cm cell the current should be about 0.028 * 15 * 15 = 6.3 A.

The resistance of 12 AWG wire is 0.162 ohms / 100' (sailboat electrics simplified pp 47) => 5.4e-3 ohm/m

With approximately two meters of wiring on either side of the cathode/anode => 4 m total, or 5.4e-3 * 4 = 2.2e-2 ohms of resistance. At 6.3 A this is a voltage drop of 6.3 A * 2.2e-2 ohm = 1.4e-2 V, which is small enough to ignore. We're not accounting for voltage drop as current travels within the two plates.

Cut 2x 2m lengths of 12 AWG wire, attach to plates with terminal hardware, cover with electrical tape (covers 1cm x 2cm of exposed plate surface)

Cut four 2cm x 2cm x .64cm separators out of plastic, tape to corners of plates and then tape plates together. There is a .64cm separation, with each plate having 15.2cm x 15.2cm - 4 * 2cmx2cm - 2cmx1cm area exposed == 15.2 * 15.2 - 4 * 2 * 2 - 2 * 1 = 213 cm^2. At 6V there should be roughly 0.028 * 213 = 6.0 A of current.

Hook the plates to a power supply set at 6V, fill a bucket with about 12 liters of seawater and put the plates into it.

Trial #1: Current is initally a little over 1 A, and in the next 25 minutes it increases to 13.7 A, and is still climbing as the trial ends. The rate of increase seems to slow down as time goes on, but measurements were not made. The bucket's contents turn cloudy and eventually develop a film of bubbles at the surface.

Trial #2: Current is initially 0.2 A. After 5 minutes it is 1.3 A. After 10 minutes it is 2.2 A. After 15 minutes it is 3.3 A. After 20 minutes it is 4.0 A. After 25 minutes it is 4.9 A, and the trial ends.

Trial #3: Current is initially 0.2 A. After 5 minutes it is 1.9 A. After 10 minutes it is 2.8 A, and the trial ends. The attitude of the plates in the bucket has not been controlled (they are at various angles off of vertical) and may be affecting measurements. For remaining trials, the plates are oriented vertically by pinning them against the wall of the bucket with a diving weight.

# of Minutes 5 10 15 20
Trial #4 2.3 3.4 4.2 5.2
Trial #5 2.8 4.2 5.2 6.0
Trial #6 3.2 4.7 5.7 6.7
Trial #7 3.5 5.0 6.0 6.9

Analysis: That in all trials the current gradually increased over time is rather mysterious, and needs further investigation.

Update (6/22/2019): The behavior of the current seems related to the titanium electrodes eroding over time.

6/17/2019

Setup: reusing titanium plates from last experiment, whose edges on the anode have been eaten away by 1-2mm.

Add a small propeller attached to a stand which is hooked up to an adjustable voltage power supply. Place both in a large bin and add 75 liters of water, enough to fully cover propeller and plates. Plates are held in a vertical position, a couple inches off the floor of the bin, several inches in front of the propeller and oriented parallel to the flow of water from the propeller. A 5m 2x 12 AWG extension cable is added for connecting electrodes to power supply, giving a total of 7m of 12 AWG between each electrode and the power supply.

Trial #1: Set propeller's power supply to 1.25 V. Propeller pulls about 0.9 A and agitates water gently. Set plate power supply to 6 V. Current is initially 0.1 A. After 5 minutes it is 0.5 A. Bubbles from the plates are being driven away from the propeller, and it is definitely agitating the water in between the plates. After 10 minutes current is 1.6 A. After 15 minutes it is 2.2 A. After 20 minutes it is 2.8 A. After 25 minutes it is 3.3 A. After 31:30 minutes (not paying attention) it is 3.9 A. After 35 minutes it is 4.1 A. Bin contents are starting to get cloudy. After 40 minutes current is 4.5 A. After 45 minutes it is 4.9 A. After 50 minutes it is 5.2 A. After 55 minutes it is 5.6 A. After 60 minutes it is 5.9 A. After 65 minutes it is 6.1 A. After 70 minutes it is 6.5 A. After 75 minutes it is 6.7 A. The bin is getting pretty cloudy, and 60 liters of fresh seawater are added to it, overflowing the bin a little. The water is still somewhat cloudy. After 80 minutes the current is 7.0 A. After 85 minutes the current is 7.4 A. Water is removed, added, removed, then added again to cycle much of it out. The water is clearer but still a little cloudy. After 90 minutes the current is 7.5 A. After 95 minutes it is 7.7 A. After 100 minutes it is 7.8 A. After 105 minutes it is 8.0 A. The water in the bin is cycled as before several more times. After 110 minutes current is 8.3 A. After 115 minutes it is 8.3 A. After 120 minutes it is 8.6 A. The plates are removed from the water and the current drops to 0 A (there is no short circuit that is adding current).

Trial #2: Reset experimental setup as before, but set plate power supply to 12 V. After 2 minutes current is 34.1 A. After 3 minutes current is 38.1 A, near the 40 A limit of the power supply. Reduce voltage to 9 V. After 4 minutes current is 26.0 A. After 5 minutes current is 26.3 A. Reduce voltage to 6 V. After 6 minutes current is 11.6 A. After 7 minutes current is 11.2 A. After 9 minutes it is 11.2 A. Reduce voltage to 5 V. After 10 minutes current is 6.0 A. After 11 minutes current is 5.5 A. After 12 minutes it is 5.7 A. After 13 minutes it is 5.7 A. Reduce voltage to 4 V. After 14 minutes current is 2.0 A. After 15 minutes it is 1.9 A. After 16 minutes it is 1.8 A. Reduce voltage to 3.5 V. After 17 minutes current is 0.4 A. After 18 minutes it is 0.4 A. Reduce voltage, with 0.0 A at 3.1 V (a more precise amp meter registers 0.06 A).

Trial #3: Reset experimental setup as before, but set plate power supply to 10 V. After 2 minutes current is 8.4 A. Increase voltage to 11.5 V. After 4 minutes current is 18.4 A. Increase voltage to 12 V. After 5 minutes current is 22.0 A. Reduce voltage to 6 V. After 6 minutes current is 6.3 A. After 7 minutes current is 6.1 A. Increase voltage to 12 V. After 8 minutes current is 24.5 A. Increase voltage to 13 V. After 9 minutes current is 22.7 A. Current started getting very erratic, and it was discovered that the terminal hardware attached to the anode had corroded and was no longer providing a good electrical connection.

Analysis: Trial #1 and #2 had an interesting contrast. Current increased only very slowly in #1 when maintaining a constant voltage, whereas #2 stepped up the voltage early on and then down to the target voltage later, reaching a higher current much quicker at that target voltage. Trial #2 and variations will need to be repeated after the plates have been repaired.

Update (6/22/2019): Stepping up voltage as in trials #2 and #3 isn't necessary with MMO electrodes.

6/18/2019

Setup: reusing titanium plates from last two experiments. The terminal hardware on the anode was replaced, using lanocote instead of electrical tape for sealing.

Trial #1: Set plate power supply to 12 V. After 1 minute current is 26.3 A. After 2 minutes current is 30.2 A. Reduce voltage to 6 V. After 3 minutes current is 8.9 A. After 4 minutes it is 8.9 A. After 5 minutes it is 9.0 A. After 10 minutes it is 9.4 A. After 15 minutes it is 9.7 A.

Trial #2: Set plate power supply to 11.5 V, then 12 V 45 seconds later. After 2 minutes current is 19.1 A. After 3 minutes it is 23.6 A. After 4 minutes it is 25.8 A. After 6 minutes it is 29.5 A. After 7 minutes it is 30.7 A. Reduce voltage to 6 V. After 8 minutes current is 8.5 A. After 9 minutes it is 8.5 A. After 11 minutes it is 8.3 A. After 13 minutes it is 8.5 A. After 15 minutes it is 8.6 A. After 20 minutes it is 8.9 A.

Trial #3: Set plate power supply to 12 V. After 1 minute current is 9.5 A. After 3 minutes it is 21.5 A. After 5 minutes it is 25.5 A. Increase voltage to 13 V. After 6 minutes current is 30.8 A. After 7 minutes it is 32.1 A. Reduce voltage to 9 V. After 8 minutes current is 18.9 A. Reduce voltage to 6 V. After 9 minutes current is 8.2 A. After 10 minutes it is 8.3 A. After 15 minutes it is 8.5 A. After 20 minutes it is 8.8 A.

Trial #4: Set plate power supply to 12 V. After 1 minute current is 12.0 A. After 2 minutes current is 19.2 A. After 5 minutes it is 26.7 A. Increase voltage to 13 V. After 6 minutes current is 31.7 A. After 7 minutes current is 33 A. Reduce voltage to 9 V. After 8 minutes current is 19.3 A. Reduce voltage to 6 V. After 9 minutes current is 8.4 A. After 10 minutes it is 8.5 A. After 15 minutes it is 8.6 A. After 20 minutes it is 9.0 A. After 25 minutes it is 9.3 A. After 30 minutes it is 9.6 A. After 35 minutes it is 9.9 A. After 40 minutes it is 10.0 A. After 45 minutes it is 10.3 A. After 50 minutes it is 10.5 A. After 55 minutes it is 10.8 A. After 60 minutes it is 11.0 A. After 65 minutes it is 11.0 A. After 70 minutes it is 11.2 A. After 75 minutes it is 11.3 A. After 80 minutes it is 11.5 A. After 85 minutes it is 11.7 A. After 90 minutes it is 11.8 A.

Analysis: Stepping up the voltage early on allows for higher currents at the target voltage without having to wait a while, and provides pretty consistent results across trials. As shown in trial #4, it can still take a while after stepping down before the current stabilizes.

Update (6/22/2019): Stepping up voltage as in trials #2 and #3 isn't necessary with MMO electrodes.

The experimental setup was changed so that the two plates are separated by 10cm instead of 0.64cm. The calculations below suggest this should increase the circuit resistance (and decrease the current) by roughly a factor of 3:

seawater conductivity is approximately 3.5 S/m 3.5 S/m => 3.5 ohm^-1 m^-1 3.5 ohm^-1 m^-1 * 0.0213 m^2 = 0.07455 m ohm^-1 1 / 0.07455 m ohm^-1 = 13.4 ohm/m for the plates used in this experiment

at 0.64 cm separation this gives 0.64 / 100 * 13.4 = 0.086 ohm at 10cm separation this gives 10 / 100 * 13.4 = 1.34 ohm

6 V / 8.9 A = 0.67 ohms total circuit resistance at end of trial #2 0.67 - 0.086 = 0.58 ohms for the wires and electrode reactions

With a 10cm separation, this suggests a 0.58 + 1.34 = 1.92 ohm circuit resistance, or 6 V / 1.92 ohm = 3.1 A of current 1.92 / 0.67 = 2.86 times as much resistance as with a 0.64cm separation

Trial #1: Set plate power supply to 12 V. After two minutes current is 7.7 A. After 4 minutes it is 9.0 A. After 7 minutes it is 10.2 A. Reduce voltage to 6 V. After 9 minutes current is 3.0 A. After 12 minutes it is 3.0 A. After 15 minutes it is 3.0 A. Plates have started to lean over (forming a rhombus instead of a rectangle) but this will only bring them closer together. After 20 minutes current is 3.0 A.

Analysis: separating plates by a larger distance gives results consistent with theory. Electrodes must be as close to each other as possible for the circuit to behave efficiently.

6/22/2019

Some concerns arose after the previous experiments. The titanium anode has been eaten away from pretty significantly, the water got pretty cloudy when running trials, and there was no noticeable chlorine smell after the experiment, as happened during the experiment conducted back in April. It seems like titanium oxides are being produced at the anode instead of chlorine and oxygen gas. We need an anode material that doesn't dissolve while running, and it seems good to do some more tests using different electrode materials, to see what differences in the results can be found.

So, using a mesh mixed metal oxide (MMO) electrode, whose length is 25.5cm, width is 23.59mm, and is missing about 42% of its area (see below), the total surface area on both sides is 25.5 * 2.359 * 2 * 0.58 = 69.8 cm^2. Adding 42.2 cm^2 for the exposed area in the cross section of the anode gives 69.8 + 42.2 = 112 cm^2, which is pretty approximate.

Measuring the size of the cutout portions of one segment of the MMO anode:

5.87mm x 23.59mm = 138.5 mm^2 cutout 2.5mm x 2.6mm / 2 = 3.25 mm^2 cutout 3.2mm x 3.2mm / 2 = 5.12 mm^2 3x cutout 8.12mm x 3.2mm - 2.89mm x 3.2mm = 16.74 * 3 = 50.2 mm^2 => (50.2 + 5.12 + 3.25) / 138.5 = 42% missing

Measuring the exposed surface area in the cross section of the MMO anode:

1.60mm thickness 4x 9mm + 2x 8mm + 2x 7mm = 66mm linear exposed length per segment 66mm * 1.6 mm = 105.6 mm^2 40x segments * 105.6 = 4224 mm^2 / 100 = 42.2 cm^2

After connecting the electrodes with the same 0.64cm separation as before, there is 1.5 + 8.4 + 8.3 = 18.2 cm of electrode length exposed, giving a surface area of roughly 18.2 / 25.5 * 112 = 80 cm^2

Trial #1: Set plate power supply to 12 V. After two minutes current is 23.5 A. Noticeable but subtle chlorine smell when sniffing near the water surface. After 3 minutes current is 23.4 A. Reduce voltage to 9 V. After 4 minutes current is 15.7 A. After 5 minutes it is 15.7 A. Reduce voltage to 6 V. After 7 minutes current is 8.1 A. After 8 minutes it is 8.1 A. After 10 minutes it is 8.1 A. The masking tape holding up the electrodes has failed and the electrodes are lying flat on the bottom of the bin. After 15 minutes the current is 8.1 A. There is no noticeable discoloration or cloudiness in the water.

Some of the lanocote was worn away from the terminal hardware on the anode (apparently from rough handling), and more was gooped on. The anode was leaned up against the wall of the bin to keep it somewhat vertical.

Trial #2: Set plate power supply to 6 V. After 15 seconds current is 8.1 A. After 1 minute it is 8.1 A. After 5 minutes it is 8.0 A. Reduce voltage to 5 V. After 6 minutes current is 5.7 A. After 7 minutes it is 5.7 A. Reduce voltage to 4 V. After 8 minutes current is 3.4 A. After 9 minutes it is 3.4 A. Reduce voltage to 3.5 V. After 10 minutes current is 2.3 A. After 11 minutes it is 2.3 A. Reduce voltage to 3 V. After 12 minutes current is 1.2 A. After 13 minutes it is 1.2 A. Reduce voltage to 2.5 V. After 14 minutes current is 0.3 A. After 15 minutes it is 0.3 A. Reduce voltage to 2 V. Current is 0 A.

Trial #3: Set up the electrodes in a bucket of seawater, measured at 8.07 pH, 30.68 PSU (even after recalibrating, surprising, some rain water mixed in at the surface?), 27.04 C. For each of the following voltages, set the power supply to that voltage, wait for current to stabilize (10-15 seconds), then measure current. Measured data for amperage at each voltage is shown below, along with the calculated circuit resistance (V / A = Ohm).

Voltage Amps Ohms
2.0 0 Infinite
2.2 0.1 22
2.4 0.3 8
2.6 0.5 5.2
2.8 0.8 3.5
3.0 1.2 2.5
3.2 1.6 2
3.4 2.1 1.62
3.6 2.5 1.44
3.8 3.0 1.27
4.0 3.4 1.18
4.2 3.9 1.08
4.4 4.4 1
4.6 4.9 0.939
4.8 5.4 0.889
5.0 5.9 0.847
5.2 6.3 0.825
5.4 6.8 0.794
5.6 7.3 0.767
5.8 7.8 0.744
6.0 8.3 0.723
6.2 8.9 0.697
6.4 9.3 0.688
6.6 9.7 0.68
6.8 10.3 0.66
7.0 10.8 0.648
7.2 11.4 0.632
7.4 11.8 0.627
7.6 12.3 0.618
7.8 12.8 0.609
8.0 13.3 0.602
8.2 13.7 0.599
8.4 14.3 0.587
8.6 14.8 0.581
8.8 15.3 0.575
9.0 15.7 0.573

Bucket is slightly cloudy with precipitates. After stirring, water is measured at 8.35 pH, 30.57 PSU, 27.46 C. There is no noticeable erosion on the anode.

Analysis: Using MMO electrodes instead of titanium electrodes gave behavior much closer to that expected. It does seem that with titanium a large part of the anode reaction is consuming the titanium itself, leading to slow changes in amperage over time and cloudy water. The grade of the titanium is unknown; both the titanium and MMO electrodes were purchased off eBay for not a lot of money.

As shown above, circuit resistance steadily decreases as voltage increases. Resistance in wires and seawater should not vary with voltage, though I'm not 100% sure; reductions in resistance should be due to increased efficiency of the chemical reactions occurring at the electrodes.

Calculating resistance of circuit components:

at 6 V total circuit resistance is 0.723 ohm

seawater conductivity is approximately 3.5 S/m 2.359 * 18.2 = 42.9 cm^2 area of seawater separating electrodes = 4.29e-3 m^2 3.5 S/m = 3.5 ohm^-1 m^-1 3.5 ohm^-1 m^-1 * 4.29e-3 m^2 = 0.015 m ohm^-1 1 / 0.015 m ohm^-1 = 66.7 ohm/m for the electrodes

at 0.64 cm separation, seawater resistance is 0.0064 * 66.7 = 0.427 ohm

electrodes are connected to power supply with 7m of 12 AWG wire on each side: 2m connected to the electrodes with terminal hardware, then 5m of 12 AWG extending to the power supply, with terminal hardware on either side.

The resistance of 12 AWG wire is 0.162 ohms / 100' (sailboat electrics simplified pp 47) => 0.162 / 30 = 5.4e-3 ohm/m. With 14m of wire, total resistance from wire is 0.0756 ohm

Terminal hardware is in good shape and shouldn't contribute appreciably to resistance. Remaining resistance is 0.723 - 0.427 - 0.0756 = 0.22 ohm

So, resistance of circuit components is roughly:

0.427 ohm: seawater between electrodes 0.0756 ohm: wires connecting to power supply 0.22 ohm: reactions occurring at electrodes

Only 0.22 / 0.723 = 30% of circuit resistance is doing useful work

6/23/2019

Setup: in order to further analyze the effect of circuit resistance on electrolysis efficiency, try to reduce the resistance from seawater and wires. First, setup the MMO electrodes with a separator of 1.2 mm instead of 6.4 mm. With the tape required to hold things in place, the separation between the electrodes is more like 1.5-2mm, and at such close distances the distance traveled by ions as they migrate to/from the far sides of the electrodes matters more.

Trial #1: Set up the electrodes in a bucket of seawater. Set power supply to 6 V. Current settles on 10.0 A.

Trial #2: Leave the electrodes in the same bucket, but replace the 5m extension cable with two 5cm pigtails (for adapting the terminal hardware). Set power supply to 6 V. Current settles on 11.6 A.

Trial #3: Repeat trial #1. Current settles on 9.9 A.

Trial #4: Repeat trial #2. Current settles on 12.7 A.

Trial #5: Repeat trial #2 in a fresh bucket of seawater. Current settles on 12.2 A.

Trial #6: Repeat trial #2 in a fresh bucket of seawater. Current settles on 12.2 A.

Analysis: Minimizing circuit resistance is clearly crucially important for efficient electrolysis.

Circuit resistance with the electrodes close together is 6 / 10 = 0.6 ohm, a reduction of 0.723 - 0.6 = 0.123 ohm from yesterday's experiment. The difference is a combination of reduced seawater resistance and reduced resistance for the electrode reactions: a lower voltage drop across the seawater means a higher potential at the electrodes and less resistance for the reactions. As mentioned above, with the electrodes so close together it's hard to estimate the average distance that ions have to travel through seawater to complete the reaction, and thus the average resistance of the seawater.

Circuit resistance with the electrodes close together and shorter wires is 6 / 12.2 = 0.49 ohm, a reduction of 0.6 - 0.49 = 0.11 ohm from the same setup with longer wires. The extra 10m of wiring should have 5.4e-3 * 10 = 0.054 ohm of resistance. The remaining 0.11 - 0.054 = 0.056 ohm of reduced resistance seems due to the higher potential at the electrodes and resulting more efficient reaction, as shown below.

6 V @ 10.0 A, 0.6 ohm total

wire: 14m * 5.4e-3 ohm/m * 10 A = 0.756 V reactions + seawater: 6 - 0.756 = 5.244 V

6 V @ 12.2 A, 0.49 ohm total

wire: 4m * 5.4e-3 ohm/m * 12.2 A = 0.264 V reactions + seawater: 6 - 0.264 = 5.736 V

The shorter wire gives an extra 5.736 - 5.244 = 0.492 V available for the reaction. We can estimate the effect this will have on the reaction's resistance using the data from the voltage sweep on 6/22. At 6 V the circuit resistance is 0.723 ohm. Averaging the circuit resistance at 5.4 V and 5.6 V gives (0.794 + 0.767) / 2 = 0.781 ohm, which has an extra 0.781 - 0.723 = 0.058 ohm of resistance compared to the circuit at 6 V. This 0.058 ohm accounts for the remaining reduced resistance when using the shorter wires.

Reducing circuit resistance therefore has a compounding effect: the lower resistance increases the voltage available for the reactions, which lowers the resistance of those reactions. Conversely, increasing circuit resistance will also increase the reaction resistance.

6/25/2019

A new experimental design was constructed. A divider was built from 1/4" plastic for separating the bin into two equal sized compartments. Two holes of size 10.5 cm x 2.359 cm were cut in the divider, with recessed areas to the side allowing for installing different membranes between the two compartments. Plastic clamps were installed on both sides of the divider for pinning the MMO electrodes in place over the holes in the divider. Right angle brackets were used to secure the divider in place in the bin, and the edges of the divider sealed with a polyurethane sealant (sikaflex 291).

Setup: Install membranes in the holes which are solid plastic, and seal the edges with lanocote. Install electrodes and connect to power supply using extension cable (14m total 12 AWG wiring). Water was added to both compartments to fill them above the level of the electrodes. Some water was observed leaking from one compartment to the other at the base (there is not a perfect seal between the sealant and bin walls), but no deformation was observed of the bin walls as water was added, i.e. the seal should still be reasonably good.

Trial #1: Set power supply to 6 V. Current stabilizes at 1.1 A.

Remove electrodes, remove solid plastic membranes, then reinstall electrodes.

Trial #2: Set power supply to 6 V. Current stabilizes at 7.0 A.

Analysis: the solid plastic membranes increase circuit resistance from 6 / 7 = 0.86 ohm to 6 / 1.1 = 5.45 ohm. In a previous experiment we calculated seawater resistance for these electrodes as 66.7 ohm/m, so 5.45 ohm roughly means that ions have to travel on average 5.45 / 66.7 = 0.082 m. Given the geometry of the divider this seems to mean that ions are mostly traveling around the edge of the solid plastic membrane, which was supposed to be sealed but maybe wasn't sealed very well.

Reinstall the solid plastic membrane, and put a thick seal of lanocote around its edge facing the cathode. When filling the compartments, water easily flows from one to the other along the base (the seal has completely failed).

Trial #3: Set power supply to 6 V. Current stabilizes at 1.1 A.

Analysis: more work is needed to make sure there is a good seal between the compartments.

6/30/2019

More angle brackets were added to attach the the base of the divider to the floor of the bin, and the base of the divider was resealed. The experiment was setup as before, with solid plastic membranes thickly sealed with lanocote.

Trial #1: Set power supply to 6 V. Current stabilizes at 0.7 A.

Remove electrodes, remove solid plastic membranes, then reinstall electrodes.

Trial #2: Set power supply to 6 V. Current stabilizes at 5.8 A.

Some more water was added to the bin (it was just above the height of the electrodes).

Trial #3: Set power supply to 6 V. Current stabilizes at 5.9 A.

The electrodes were removed and attached together with a 0.64 cm gap, then placed in the bin, as in the setup on 6/22.

Trial #4: Set power supply to 6 V. Current stabilizes at 6.3 A.

The terminal hardware on the anode was found to be loose and the terminal was exhibiting corrosion (despite being coated with lanocote). New terminal hardware was installed on the anode and the electrodes reassembled as for trial #4.

Trial #5: Set power supply to 6 V. Current stabilizes at 7.0 A.

The electrodes were separated and attached on either side of the bin divider, with no plastic membrane.

Trial #6: Set power supply to 6 V. Current sabilizes at 6.6 A.

Analysis: care needs to be taken when comparing current rates between trials done at different times. Resistance can come in to play pretty easily with this setup and can vary quite a bit due to corrosion. Later on we'll need to figure out a way to manage corrosion and get dependable connections between the power supply and electrodes, but for now we'll just be careful with comparisons.

Regardless, using the divider with no membrane produces a drop in the current between the electrodes (and thereby an increase in the circuit resistance) of only about 6% (Trial #3 vs. #4 and #5 vs. #6 have differences of 5.9 / 6.3 = 0.937 and 6.6 / 7.0 = 0.943, respectively).

The electrodes were removed, visible lanocote residue removed from around the divider, and the electrodes replaced.

Trial #7: Set power supply to 6 V. Current stabilizes at 6.7 A.

Visible lanocote residue was wiped off the solid plastic membrane and it was installed with no new lanocote sealing its edges.

Trial #8: Set power supply to 6 V. Current stabilizes at 1.5 A.

Holes were drilled in the membrane at intervals of 1 cm. The membrane thickness is 1.21 mm, the drill bit size is 0.93 mm, with 2 * 3 * 10 = 60 holes in total. The membrane was reinstalled.

Trial #9: Set power supply to 6 V. Current stabilizes at 1.6 A.

Analysis: Per the calculations below, in retrospect it isn't surprising that the current increased so little.

no membrane: 6 / 6.7 = 0.896 ohm solid membrane: 6 / 1.5 = 4 ohm membrane with holes: 6 / 1.6 = 3.75 ohm

Each hole is pi * (0.93/2)^2 = 0.68 mm^2 Total surface area is 0.68 * 60 = 40.8 mm^2 * 1e-6 = 4.08e-5 m^2 3.5 ohm^-1 m^-1 * 4.08e-5 m^2 = 1.4e-4 m ohm^-1 1 / 1.4e-4 m ohm^-1 = 7010 ohm/m 7010 * 1.2e-3 m = 8.4 ohm

there are two pathways past the membrane with holes: around the edges or through the holes. combined: 3.75 - 0.896 = 2.85 ohm

around the edges: 4 - 0.896 = 3.1 ohm

1/2.85 = 1/3.1 + 1/N 1/N = 1/2.85 - 1/3.1 N = 1/(1/2.85 - 1/3.1) = 35.3 ohm

If N actually was 8.4 ohm, the resistance across the membrane would be 1/(1/3.1 + 1/8.4) = 2.26 ohm, the total circuit resistance would be 2.26 + 0.896 = 3.16 ohm, and the current would be 6 / 3.16 = 1.9 A.

Say we want the current to drop by 50% compared to no membrane. Then the current is 6.7 / 2 = 3.35 A, the resistance is 6 / 3.35 = 1.79 ohm, and the resistance across the membrane is 1.79 - 0.896 = 0.894 ohm. Note that these calculations are all pretty approximate in assuming the resistance of the rest of the circuit is constant when the membrane is removed or changes. The membrane's state will affect the paths through the seawater which ions take as they migrate, such that the seawater resistance varies between setups.

1/0.894 = 1/3.1 + 1/N 1/N = 1/0.894 - 1/3.1 N = 1/(1/0.894 - 1/3.1) = 1.26 ohm

More holes were drilled in the membrane, at intervals of 0.5 cm, giving 2 * 5 * 21 = 210 holes in total. The holes were very slightly enlarged with a pick. The membrane and electrodes were reinstalled as before.

Trial #10: Set power supply to 6 V. Current stabilizes at 2.3 A.

Analysis: repeating the calculations from earlier, to get the approximate resistance of the membrane with more holes:

circuit resistance: 6 / 2.3 = 2.61 ohm approximate combined resistance of membrane edges + holes: 2.61 - 0.896 = 1.71 ohm resistance of the membrane with more holes:

1/1.71 = 1/3.1 + 1/N 1/N = 1/1.71 - 1/3.1 N = 1/(1/1.71 - 1/3.1) = 3.81 ohm

the theoretical resistance is 8.4 / (210 / 60) = 2.4 ohm

At least these are in somewhat better agreement...

Suppose we enlarge holes with a 2.32 mm drill bit to double the surface area. Current area is 210 * 0.68 mm^2 = 142.8 mm^2 Each larger hole is pi * (2.32 / 2)^2 = 4.23 mm^2

N * 4.23 + (210 - N) * 0.68 = 142.8 * 2 N * 4.23 + 210 * 0.68 - N * 0.68 = 142.8 * 2 N * (4.23 - 0.68) = 142.8 * 2 - 210 * 0.68 N = (142.8 * 2 - 210 * 0.68) / (4.23 - 0.68) = 40.2

For the sake of symmetry, 60 of the 210 holes were drilled out using the 2.32 mm bit, using the original 1 cm spacing. The membrane and electrodes were reinstalled as before.

Trial #11: Set power supply to 6 V. Current stabilizes at 2.8 A.

Some bubbles were observed on the cathode, which were brushed off.

Trial #12: Set power supply to 6 V. Current stabilizes at 2.9 A.

Analysis: repeating the calculations from earlier:

circuit resistance: 6 / 2.9 = 2.07 ohm approximate combined resistance of membrane edges + holes: 2.07 - 0.896 = 1.17 ohm resistance of the membrane with some big holes:

1/1.17 = 1/3.1 + 1/N 1/N = 1/1.17 - 1/3.1 N = 1/(1/1.17 - 1/3.1) = 1.89 ohm

theoretical resistance:

surface area is 60 * 4.23 + 150 * 0.68 = 355.8 mm^2 8.4 * 40.8 / 355.8 = 0.96 ohm

Still pretty far off...

The electrodes were reinstalled without a membrane.

Trial #13: Set power supply to 6 V. Current stabilizes at 6.9 A.

The remaining 150 holes in the membrane were drilled out with the 2.32 mm bit. The membrane and electrodes were reinstalled as before.

Trial #14: Set power supply to 6 V. Current stabilizes at 3.8 A.

Analysis: repeating the calculations from earlier:

circuit resistance: 6 / 3.8 = 1.58 ohm approximate combined resistance of membrane edges + holes: 1.58 - 0.896 = 0.68 ohm resistance of the membrane with all big holes:

1/0.68 = 1/3.1 + 1/N 1/N = 1/0.68 - 1/3.1 N = 1/(1/0.68 - 1/3.1) = 0.87 ohm

theoretical resistance:

surface area is 210 * 4.23 = 888.3 mm^2 8.4 * 40.8 / 888.3 = 0.39 ohm

7/1/2019

The bin was set up with the membrane from the last experiment (210 2.32 mm holes) and electrodes installed as before. Propellers were placed in both sides of the divider to circulate the water in the two sides. A pH meter was placed in the same side as the cathode. 60 liters of water was added, and the propellers turned on at 1.2 V, pulling about 1.9 A. The water in both compartments is visibly circulating.

Trial #1: The pH reading is initially 8.09, and the temperature is 26.37 C. The salinity reading is 27.44 but I don't believe it, so using 35 for doing calculations. Calculated values for raising the pH to 8.2, 8.3, and 8.4 are 211, 437, and 696 A s, respectively. Set power supply to 8 V (well, initially 6 V, then adjust to 8 V within 15 seconds or so). Current settles at 5.7 A after a minute or so.

Time (min)  pH     Current (A)
2                 8.17  5.7
3                 8.23  5.6
4                 8.28  5.4
5                 8.32  5.4
6                 8.36  5.4
7                 8.40  5.3
8                 8.43  5.3

At 8 minutes the current is shut off. After 9 minutes pH is 8.44, after peaking at 8.45. After 11 minutes pH is 8.41. After 13 minutes pH is 8.38. After 17 minutes pH is 8.33.

Empty the bin, set things up as before, and refill the bin with 60 liters of water.

Trial #2: The pH reading is initially 8.11, and the temperature is 26.21 C. Set power supply to 8 V.

Time (min)  pH     Current (A)
2                 8.22  5.8
3                 8.27  5.6
4                 8.31  5.5
5                 8.36  5.5
6                 8.39  5.5
7                 8.42  5.4
8                 8.45  5.4

Empty the bin, set things up as before, and refill the bin with 60 liters of water.

Trial #3: The pH reading is initially 8.13, and the temperature is 26.15 C. Set power supply to 8 V.

Time (min)  pH     Current (A)
2                 8.23  5.9
3                 8.29  5.8
4                 8.34  5.7
5                 8.38  5.5
6                 8.42  5.6
7                 8.46  5.4
8                 8.49  5.5

Analysis: A script was written (see electrolysis.js in the coral repo) to compare the amp-seconds required to raise the pH of one compartment with the theoretical requirement. Each 3 minute time period was used and the ratio of the actual amp-seconds to the theoretical amp-seconds was computed. For trial #1:

2 -> 5 Ratio 2.975933889612167
3 -> 6 Ratio 3.166967340782892
4 -> 7 Ratio 3.135960842801048
5 -> 8 Ratio 3.265488440469809

For trial #2:

2 -> 5 Ratio 3.0514246251970123
3 -> 6 Ratio 3.3074338573935274
4 -> 7 Ratio 3.3799263000126794
5 -> 8 Ratio 3.922611869702964

For trial #3:

2 -> 5 Ratio 2.815908724245556
3 -> 6 Ratio 3.0342021274530455
4 -> 7 Ratio 3.017967520017932
5 -> 8 Ratio 3.12351785681507

Raising the pH requires significantly more amp-seconds than the optimum predicted by theory. There are several possible causes here which will need to be investigated in future experiments. Fortunately, the trials here gave somewhat consistent results and it should be possible to do this investigation without having to deal with a lot of experimental noise.

7/5/2019

Before each trial the bin was set up in an identical fashion to the last experiment, filling with fresh saltwater.

Trial #1: The pH is initially 8.09, and the temperature is 25.56 C. Set power supply to 8 V.

Time (min)        pH    Current (A)
2                 8.19  5.2
3                 8.25  5.0
4                 8.29  5.0
5                 8.33  4.8
6                 8.36  4.8
7                 8.39  4.6
8                 8.41  4.7

2 -> 5 Ratio 2.855869072409772
3 -> 6 Ratio 3.3390398870039357
4 -> 7 Ratio 3.4265908909514033
5 -> 8 Ratio 4.070202337805971

Trial #2: The pH is initially 8.12, and the temperature is 25.61 C. Set power supply to 8 V.

Time (min)        pH    Current (A)
2                 8.23  5.1
3                 8.27  5.0
4                 8.31  5.1
5                 8.35  5.0
6                 8.38  5.0
7                 8.41  4.9
8                 8.44  5.0

2 -> 5 Ratio 3.219353012017835
3 -> 6 Ratio 3.3096618991027986
4 -> 7 Ratio 3.4704760316255747
5 -> 8 Ratio 3.6820183796649686

Trial #3: The pH is initially 8.12, and the temperature is 25.59 C. Set power supply to 8 V.

Time (min)        pH    Current (A)
2                 8.24  5.2
3                 8.28  5.1
4                 8.32  5.1
5                 8.34  5.1
6                 8.36  5.1
7                 8.37  5.0
8                 8.39  5.0

2 -> 5 Ratio 3.9403657335509203
3 -> 6 Ratio 4.675053623031953
4 -> 7 Ratio 7.155533643185846
5 -> 8 Ratio 6.964310750011725

It's curious why this trial was so ineffective compared to others. There isn't an obvious cause. After it ended the pH meter was swished around in the cathodic compartment and was still around 8.39. Perhaps the propellers were oriented such that more convection than normal occurred between the compartments? The propellers have to be set up from scratch for each trial, though the position and orientation are roughly the same.

Trial #4: The pH is initially 8.12, and the temperature is 25.59 C. Set power supply to 8 V.

Time (min)        pH    Current (A)
2                 8.27  5.1
3                 8.31  4.9
4                 8.35  5.1
5                 8.38  5.0
6                 8.40  4.9
7                 8.42  5.0
8                 8.44  4.9

2 -> 5 Ratio 3.3433797244041306
3 -> 6 Ratio 3.8053514788294147
4 -> 7 Ratio 4.845035699082789
5 -> 8 Ratio 5.364501695579108

Still not very consistent...

The membrane was removed, its holes smeared with lanocote to block them, then reinstalled with some more lanocote to seal the visible passages between compartments. The intent here is not to provide a perfect seal, but to block convection as much as easily possible between the compartments.

Trial #5: The pH is initially 8.14, and the temperature is 26.06 C. Set power supply to 11 V.

Time (min)        pH    Current (A)
2                 8.20  3.7
3                 8.20  3.7

The water was stirred with the pH meter and while pH readings were pretty inconsistent, they seemed to indicate a pH of around 8.12, i.e. virtually unchanged. Bubbling was observed on screws in the anodic compartment which screwed into the cathodic compartment. If current travels through these screws then the end in the anodic compartment acts as a cathode (generating hydrogen) and the end in the cathodic compartment acts as an anode (generating chlorine). The net effect is that these reactions act in opposition to the others in their compartment, with the net result being no change in alkalization for current traveling through these screws. Additionally, the catholyte was pretty discolored, and the anolyte less so. It's possible this discoloration is due to the lanocote.

7/18/2019

The membrane between the compartments was redesigned to use a thin plastic film, approximately 0.1 mm thick. Screws between the compartments were sealed with a polyurethane sealant to prevent the screws from acting as electrodes. 44 holes were poked into the membranes between the electrodes (2x2x11). The experiment was set up as before.

Trial #1: The pH is initially 8.09, and the temperature is 26.2 C. Set power supply to 8 V.

Time (min)        pH    Current (A)
2                 8.14  1.3
3                 8.15  1.2
4                 8.15  1.3
5                 8.15  1.3

Some stirring was done with the pH meter and some inconsistent readings were produced, and the trial was abandoned.

Trial #2: The pH is initially 8.12, and the temperature is 26.07 C. Set power supply to 15 V.

Time (min)        pH    Current (A)
0                 8.12  3.1
2                 8.23  3.1
3                 8.27  3.0
4                 8.29  3.0
5                 8.31  2.9
6                 8.32  2.9
7                 8.34  3.0
8                 8.36  2.9
9                 8.37  2.9

0 -> 2 Ratio 1.6904316818110865
2 -> 3 Ratio 2.043428291936218
3 -> 4 Ratio 3.848039218813171
4 -> 5 Ratio 3.6772713466675295
5 -> 6 Ratio 7.078865867273463
6 -> 7 Ratio 3.5262018340310313
7 -> 8 Ratio 3.4312082057748574
8 -> 9 Ratio 6.611768299035464

It's interesting how in this and many other trials, efficiency compared with theory drop as the pH difference between the compartments increases. It seems that the two compartments are neutralizing each other to a greater degree than expected. This could be because the neutralizing effects of ion migration and diffusion are greater than expected, or because there is convection between the two compartments, which we are assuming is nil.

It seems good to test membranes that don't allow convection at all between the compartments. The electrodes were removed and the anode was placed in a paper coffee filter, which covered about 1/3 of it. The electrodes were fastened together at a 0.64 cm as done previously, then partially submerged in seawater, such that the top of the coffee filter was entirely out of the water. At 6 V the current is 1.2-1.4 A. The electrodes were reassembled without the coffee filter and placed in the water as before. Current is 1.2-1.3 A. The filter material seems suitable for largely eliminating convection while still allowing migration and diffusion.

7/21/2019

A new experimental design was conceived and constructed. The electrode and membrane system were retained, but the separator was detached from the bin (so that it is a plate to attach things to rather than having an intrinsic function), and housings for the electrodes were constructed, with hose barb outlets, and inner volume of roughly 13" x 3.75" x 1.75". These outlets were connected to pumps. After filling the bin with water and putting the assembly into it, the pumps can be turned on to pull water from the bin, through the housings, and then pump it either into a bucket or into the sea. There are several reasons for this new design:

- It should be much easier to manage variations between experiments with this setup; the previous design gave inconsistent results and made it difficult to control the flow of water past the electrodes.

- Measurements are easier to make (the catholyte outlet should have a constant pH which is easy to measure), and it is easier to reset things between trials.

- A number of issues point towards using a flow-through design like this in actual installations. This design serves better as a prototype for such an installation and makes it easy to extrapolate results.

The experimental methodology for trials is as follows:

- Place the apparatus in the bin, hooked up to the pumps and power supply. For initial trials, the separator is horizontal, with the cathode over the anode.

- Fill the bin with around 80 liters of water (it doesn't matter much, there just needs to be enough that the apparatus doesn't dry out). Reusing water already in the bin after a previous trial is fine, since it is unadulterated seawater.

- Measure the bin's temperature and pH.

- Turn the pumps on.

- Turn on the electrode power supply and adjust to the desired constant current.

- Let the pumps run for 15-30 seconds to settle.

- Direct the flow from the catholyte pump into the bucket, measuring the amount of time it takes for the bucket to fill with the specified amount of water.

- As the bucket fills, measure its pH. Keep track of the approximate power supply voltage, which can change over time.

- After the bucket fills and the time measured, remove the catholyte flow, turn off the electrode power supply, then run the pumps for several seconds before turning them off, to flush most of the catholyte and anolyte out of the bin.

The theoretical amp-second requirement to alkalize the bucket is determined from the bucket's volume and the pH of the bin and bucket. The actual amp-seconds used are the electrode power supply's current multiplied by the time taken to fill the bucket. As before, the difference between the theoretical and actual amp-seconds under different experimental conditions is the subject of interest.

For the trials today, the electrodes are positioned 0.64 cm apart, and the membrane used is a double-thickness layer of coffee filter material (i.e. a coffee filter folded flat and cut to the right size).

Some calibration of the pump power supply was performed. At 4.5 V the catholyte output was about 5 liters per minute, and at 8 V the catholyte output was about 9 liters per minute. At 8 V, the anolyte output at the same voltage (different pump) was slightly higher at about 10 liters per minute. The pump power supply will be at 8 V for initial trials. At this voltage the pumps pull about 4.5 amps total, so power usage can be higher than the electrolysis circuit itself. Ultimately we'll want more efficient pumps (e.g. the propellers used earlier were much more energy efficient at moving water around), but these are nice for providing a consistent and easily adjustable rate of flow.

Trial#    startPH    Temp   Liters   Amps     Volts   Seconds     endPH    Actual / Theoretical
1         8.11       25.75  18       3.1      6.1     130         8.36     1.231

Some confusion arose over where the 18 liter point is in the bucket (a little more than 18 liters was put in the bucket it seems), so a 15 liter point was marked on the bucket and will be used for subsequent trials.

Trial#    startPH    Temp   Liters   Amps     Volts   Seconds     endPH    Actual / Theoretical
2         8.11       25.73  15       3.1      5.8     105         8.35     1.253

The bin was emptied and refilled, as the electrolysis had been left running after the pumps were turned off, and the bin's pH had gotten out of whack.

Trial#    startPH    Temp   Liters   Amps     Volts   Seconds     endPH    Actual / Theoretical
3         8.11       25.69  15       3.1      6.0     103         8.34     1.293
4         8.11       25.73  15       3.1      6.1     102         8.34     1.280
5         8.11       25.74  15       3.1      6.3     105         8.34     1.317
6         8.11       25.75  15       3.1      6.2     102         8.34     1.279    

It's time to change some of the experimental parameters. Let's see what happens to efficiency when amperage changes.

Trial#    startPH   Temp    Liters   Amps     Volts   Seconds     endPH    Actual / Theoretical
7         8.11      25.77   15       2.0      5.1     103         8.25     1.468
8         8.11      25.75   15       2.0      5.2     105         8.25     1.497
9         8.11      25.78   15       2.0      5.2     105         8.25     1.496
10        8.11      25.81   15       4.5*     8.3     103         8.43     1.258*
11        8.11      25.81   15       4.5*     8.0     104         8.42     1.321*

Trials #10 and #11 (asterisked above) had a lower amperage later on due to power supply issues --- a lower amperage was noticed as the power supply was turned off. No such amperage anomalies were observed beforehand. The problem is that the power supply switched over to constant voltage from constant amperage as the voltage went up, and the power supply was adjusted to avoid this problem in the future. These two trials should be disregarded.

Trial#    startPH   Temp    Liters   Amps     Volts    Seconds     endPH   Actual / Theoretical
12        8.11      25.83   15       4.5      9.1-10+  102         8.44    1.199
13        8.11      25.83   15       4.5      8.1-10+  103         8.44    1.210
14        8.11      25.84   15       4.5      9.0-10+  102         8.44    1.198

Some bubbling was observed on the bronze plumbing fitting exiting the cathode housing. As noted before, this is a sign that at this high voltage the fitting is acting as a cathode on the outside and an anode on the inside of the housing, reducing the efficiency of the reaction and adding chlorine to the catholyte. Care is needed when assembling these electrolysis cells! Other metal fasteners connecting the electrode housings to each other or to the bin water were sealed, but these fittings were overlooked.

Anyways, let's see what happens when amperage is constant but the rate of water flow changes. The pump power supply was changed from 8 V, first to 6 V for a lower flow rate, then 9 V and 10 V for a higher flow rate.

Trial#    startPH   Temp    Liters   Amps     Volts    Seconds     endPH    Actual / Theoretical
15        8.11      25.86   15       3.0      6.5-7    102         8.34     1.235
16        8.11      25.88   15       3.1      7.8      103         8.34     1.289
17        8.11      25.91   15       3.1      6.4-7.1  141         8.40     1.336
18        8.11      25.95   15       3.1      6.6-7.6  140         8.40     1.325
19        8.11      25.96   15       3.1      7.1-8.2  142         8.41     1.290
20        8.12      25.97   15       3.1      7.0+     93          8.32     1.346
21        8.12      25.93   15       3.1      6.7+     84          8.30     1.373
22        8.12      25.94   15       3.1      6.8-7.2  84          8.30     1.373

Summary: All in all the new design is a huge improvement over previous designs, regarding the consistency of experimental results, the much closer match between experimental and theoretical results, and the general usability and convenience of the setup. There are more things to investigate here, after fixing the design flaw noted above. In particular, voltages creeped up over time for runs at similar amperages. The cathode developed a thin film of (presumably) calcium carbonate on it, which may be adding resistance to the circuit. Alternatively, the terminal hardware on the anode may have corroded again --- it's hard to inspect the electrodes without disassembling and resealing them.

7/22/2019

Some more trials would be nice to study the relationship between the water flow and the alkalization efficiency, by separately adjusting the water flow for the catholyte and anolyte (previously they had the same power supply and were adjusted together). The exposed plumbing fittings on the electrode housings were also sealed to prevent unwanted anodic reactions from occurring in the cathode housing. Catholyte/anolyte flows are specified as voltages. The flow rates at these voltages were separately measured and are given below.

Catholyte?      Voltage   Liters  Seconds  LPM
No              3         5       98       3.1
No              5.5       15      141      6.4
No              7.9       15      92       9.8
No              11        15      65       13.8
Yes             3.5       5       83       3.6
Yes             5.5       15      156      5.8
Yes             7.95      15      105      8.7
Yes             11.5      15      71       12.7

Trial#  Cath V   An V        startPH   Temp    Liters   Amps     Volts       Seconds     endPH    Actual / Theoretical
1       7.95     7.9         8.08      26.00   15       3.1      5.5         105         8.32     1.303
2       7.95     7.9         8.10      25.98   15       3.1      5.7         102         8.32     1.362
3       7.95     7.9         8.10      25.98   15       3.1      5.8         103         8.32     1.375
4       7.95     7.9         8.10      25.98   15       3.1      5.8         103         8.32     1.375
5       7.95     5.5         8.11      25.99   15       3.1      6.1         104         8.32     1.444
6       7.95     5.5         8.09      26.03   15       3.1      5.7         104         8.31     1.408
7       7.95     5.5         8.10      26.15   15       3.1      6.0         104         8.31     1.461
8       7.95     3           8.10      26.14   15       3.1      5.7         104         8.31     1.461
9       7.95     3           8.10      26.14   15       3.1      6.8         104         8.31     1.461
10      7.95     11          8.10      26.18   15       3.1      5.9         104         8.31     1.460
11      7.95     11          8.10      26.17   15       3.1      5.7         102         8.31     1.432
12      5.5      11          8.11      26.17   15       3.1      5.3         149         8.40     1.405
13      5.5      11          8.11      26.16   15       3.1      ?           149         8.41     1.348
14      11.5     11          8.12      26.23   15       3.1      5.5         71          8.26     1.530
15      11.5     11          8.12      26.23   15       3.1      5.4         72          8.26     1.552
16      3.5      3           8.12      26.30   10       3.1      5.9         167         8.55     1.418
17      3.5      3           8.12      26.30   10       3.1      5.6         163         8.56     1.344

Analysis: Varying the water flow through the cathode and anode housings did not have much of an effect on the design's efficiency. This is kind of nice, as the flow can be adjusted according to the applied current to get the desired level of alkalization, without affecting efficiency.

One thing to note here is the pretty significant experimental margin of error. The pH meter only has a resolution of two decimal places, and the voltage and current meters one decimal place. Movement in any of these less than the resolution has a pretty significant effect on the measured efficiency. For example, changing the starting pH in trial #14 from 8.12 to 8.11 improves the measured efficiency from 1.530 to 1.439. Measuring volumes with marks on a bucket is also pretty imprecise. Ultimately, the exact efficiency numbers here should be taken with a grain of salt, but these measurements are accurate enough to point towards ways of improving the design.

7/24/2019

Some refinements were made to the design. The brackets holding the electrodes in place get in the way of the flow of water in the housings and can potentially create turbulence and eddies in the regions where the reactions take place, potentially affecting the system's efficiency. These brackets were removed, and new brackets were constructed which were installed on the inside of the housing and hold the electrodes in place using some sealant, in a way that the flow of water should not be obstructed. The new brackets will also allow the distance between the electrodes and the membrane to be adjusted. The membrane holder was also changed to allow a larger membrane surface area and to avoid sharp edges which might induce turbulence. The terminal hardware on the anode was somewhat corroded and was replaced, and the terminal hardware on both electrodes was sealed with a polyurethane sealant.

Trial#    startPH   Temp    Liters   Amps     Volts       Seconds     endPH    Actual / Theoretical
1         8.09      26.03   15       3.2      5.4         110         8.30     1.623
2         8.10      26.02   15       3.2      5.5         109         8.30     1.677
3         8.10      26.02   15       3.2      5.5         109         8.30     1.677

Efficiency is a good deal worse than was the case with the last design. Back to the drawing board... On the plus side, experimental results continue to be very consistent, and show that the cell design has a significant impact on efficiency. Some more work is needed to find a cell that works well. One possibility here is that the bulk of the water flow is occurring away from the electrodes.

The polarity of the electrodes was reversed, and 3.2 A of current was applied for 5 minutes while the pumps weren't running, with the voltage rising from 5.5 to 5.8. The housing was then left for 15 minutes before being disassembled. Afterwards, most of the accreted material had sloughed off the cathode: many flakes of the material came out of the cathode housing when it was removed from the water, and accreted material was visible only at the tips of the electrode mesh.

7/26/2019

More refinements were made to the design. Foam and plastic was added to the interior of the housing to create a channel in which the electrodes are roughly in the middle. The electrodes are slightly further apart, but are still only separated by roughly 1 cm.

Trial#    startPH    Temp   Liters   Amps     Volts       Seconds     endPH    Actual / Theoretical
1         8.14       28.46  15       3.2      5.8         109         8.28     2.245
2         8.15       28.42  15       3.1      5.8         110         8.30     2.001
3         8.15       28.39  15       3.1      5.8         109         8.32     1.732
4         8.15       28.36  15       3.1      5.8         109         8.32     1.733
5         8.15       28.36  10       3.1      5.8         150         8.51     1.479
6         8.16       28.39  10       3.1      5.8         150         8.52     1.460
7         8.16       28.55  10       1.9      4.5         150         8.29     2.899
8         8.16       28.20  10       1.9      4.6         148         8.31     2.460
9         8.16       28.04  10       3.1      5.9         149         8.50     1.565
10        8.16       27.94  5        3.1      6.0         118         8.70     1.386
11        8.16       27.88  5        3.1      6.0         117         8.73     1.282
12        8.16       27.85  5        3.1      6.1         117         8.72     1.313

Analysis: Unlike the original design tested on 7/22, this design's efficiency is very sensitive to the water flow between the compartments. One possibility is that at higher flows there is turbulence which leads to ions in the catholyte and anolyte neutralizing each other more readily. More trials at different water and current flow rates are needed to better examine the characteristics of this design.

8/16/2019

The old design was reused, except that the 5m extension cable was replaced with short pigtail connectors, and the electrode compartments were arranged to be side-by-side instead of one-over-the-other, with a tilt so that generated gas came out the inlet. The goal is to do some more trials at low water flow rates, to get some idea of how the system will work when scaled to a larger size. Because of the low water flows, the system was left running between adjacent trials, letting it sit running for a few minutes after adjusting water flow or voltage.

Trial#   startPH    Temp   Liters    Amps    Volts       Seconds      endPH    Actual / Theoretical
1        8.07       27.22  1         4.1     6.0         36           8.96     1.596
2        8.09       27.20  1         4.1     6.0         39           9.15     1.364
3        8.09       27.19  1         4.1     6.0         38           9.12     1.377
4        8.09       27.18  1         4.1     6.0         42           9.24     1.335
5*       8.09       27.18  2         4.1     6.0         42           8.78     1.287
6        8.09       27.18  1         4.1     6.0         50           9.46     1.322
7        8.10       27.20  1         4.1     6.0         61           9.70     1.408
8        8.11       27.21  1         4.1     6.0         70           9.73     1.599
9*       8.11       27.21  2         4.1     6.0         70           9.06     1.382
10       8.11       27.23  1         4.1     6.0         65           9.75     1.471
11       8.11       27.24  1         3.0     5.0         44           9.05     1.288
12       8.11       27.25  1         3.0     5.0         44           9.00     1.380
13*      8.11       27.25  2         3.0     5.0         44           8.63     1.406
14       8.11       27.35  1         3.0     5.0         44           9.03     1.322
15       8.10       27.39  1         3.0     5.0         71           9.47     1.369
16       8.10       27.39  1         3.0     5.0         71           9.48     1.360
17       8.11       27.39  1         1.9     4.0         63           8.84     1.620
18       8.11       27.39  1         1.9     4.0         62           8.83     1.624
19       8.10       27.41  1         1.9     4.0         64           8.84     1.632
20*      8.10       27.41  2         1.9     4.0         64           8.54     1.628
21       8.10       27.43  1         2.4     4.5         74           9.17     1.487
22       8.11       27.55  1         2.4     4.5         76           9.21     1.469
23       8.10       27.62  1         2.4     4.5         75           9.20     1.457
24       8.10       27.64  1         2.4     4.5         49           8.92     1.374
25       8.11       27.61  1         2.4     4.5         48           8.88     1.449
26       8.11       27.57  1         2.4     4.5         45           8.86     1.407
27       8.11       27.55  1         1.9     4.0         51           8.79     1.439
28       8.10       27.56  1         1.9     4.0         53           8.73     1.674
29       8.11       27.53  1         1.9     4.0         51           8.72     1.664
30       8.10       27.56  1         5.6     7.0         64           9.81     1.923
31       8.12       27.64  1         5.6     7.0         51           9.80     1.549
32       8.12       27.69  1         5.6     7.0         62           9.80     1.883
33       8.12       27.74  1         5.6     7.0         50           9.79     1.525
34       8.12       27.79  1         5.6     7.0         44           9.69     1.404
35       8.12       27.77  1         5.6     7.0         41           9.62     1.356
36       8.12       27.77  1         7.0     8.0         41           9.76     1.583
37       8.13       27.75  1         7.0     8.0         52           9.83     1.957
38       8.13       27.76  1         7.0     8.0         69           9.80     2.623
39       8.14       27.79  1         7.0     8.0         27           9.42     1.278
40       8.14       27.73  1         7.0     8.0         26           9.36     1.292

* These trials were produced by mixing the previous trial's result with 1 liter of seawater and measuring the resulting pH.

Other than some oddness towards the end where inconsistent water flows produced some odd results, these trials are pretty consistent with previous experiments and are a good source of data for estimating the behavior of a larger installation. A few observations:

- At lower voltages the alkalization effectiveness decreases somewhat, which is rather mysterious.

- The trials which mixed together seawater with the water alkalized by the previous trial had about the same efficiency. While the theoretical calculations underestimate the amount of hydroxide required to alkalize a given amount of water, they do predict the results of mixing water together well, so we can calculate how much alkalization is required to reach a target pH by alkalizing to a higher pH and then diluting with seawater.