GFCI again

Then I suggest you leave "gud enuf" alone. You might want to "add on" some kind of device to clamp the "kick" voltage. Amen! Without "grounds" that isn't a return path to cause the GFCI to trip.
You can easily set up a very dangerous situation where two separate faults result in a potentially leathal voltage to appear between to metal objects (or metal and water) in your setup. Were I you, I wouldn't even "experiment" with lifting ground connections. That's the only one I can think of.
I have used GFCIs on all kinds of loads including refrigerators and washing machines (dish and clothes.)
When it tripped on the fridge I determined that some of the water generated in the defrost cycle wasn't going in to the right place. In the dishwash, I found than I had a leak which caused a bit of water to spray over the connections to the heater.

Leakage (a short) between the neutral and the ground wire will also cause the GFI to trip.

I realize that the setup I describe could be hazardous. I would not use it permanently. It would only be used to confirm rather that the inductive kick forces current to flow to ground when tripping occurs. For a permanent setup, I would keep the capacitors and a protective ground.
I have had pumps with sufficient leakage to ground so that I could not use a regular GFCI. In such a case, I would have to use a grounded pump without GFCI.
Bill

Why would breaking an inductive load cause leakage between neutral and ground?
Is it possible that a spike between hot and neutral causes a GFIC to trip? I don't see how, but there are various implementations of GFCI and some might behave peculiarly.
Bill

You should get a 12V DC system and use recreational watercraft bilge pumps, which are made to be submersible. This way, you solve many problems at once. No need to worry any longer about excessive voltages, and no need to use GFCI or worry about it again. For your more than a bit dangerous SC Line fed setup, I would simply plug it into a power strip so that the turn off pulses do not flip the GFCI
Maybe the relay is not closing the contacts in unison, or it is closing the fault return line late, if it switches that line at all. Look at solid state AC timer switches if you are really trying to nail down this issue.
Still, I would go with the 12VDC fed marine bilge pumps. That is the right path for this task. Hell, you might be able to mount the pump in a large vat and feed all the hydro units from one pump.
It may be the timer as well. Can you simulate the on/off cycle without the timer in place to see if the event id tied to the pumps or the timer(s)?
It is very unlikely that there would be a spike high enough to pierce wire insulation. Flux collapse spikes have essentially ZERO actual work capacity, so it gets clamped to a virtual zero pretty easily, and that insulation requires a longer term event to get breached.
The timer? Also just as an experiment, put an earth grounded lead into the tank and do the test again. Then, try the same thing with the fault return line.

Not "SC" "AC" in the above.
Anyway, you could also take the protection out of a power strip and install it inside your timer. Either side (input or output) should do.
I suspect the timer is the root cause.

Bill, an inductive kick can generate broadband RF. The RF can then activate the trip function in the GFI circuit. Kind of a false alarm in a way.
The kind of GFI used for hot tubs seems like it would be best for this application.

Interesting though. If so, that would explain why a capacitor helps. I would also expect that removing the ground should not greatly affect the action of rf on the circuitry.
Bill

It clamps the HF transients.
When testing the PARD of a power supply, it is often common to place a 10uF EL cap plus a 0.01uF ceramic across the output being tested.

You may have the solution there. Spikes that don't cause any problems with insulation or performance of electromechanical devices have been known to kill electronic devices. While your GFIC should be able to ride through these spikes, (past history of electronics in what was once treated as 117V rms pure sine wave in electronic applications in power circuits is full of failures) if there is some leakage below the trip level, a spike may be sufficient to bring this over the critical level. Your capacitor may be satisfactory as a permanent solution but doesn't answer your question. Have you tried replacing the GFIC device?- it could be borderline. I've had these trip for no apparent reason and then, when reset, go on happily after that (and another that repeatedly tripped (again no apparent reason) until it was replaced and sent to its grave. So, all I can suggest is keep the capacitor but if it trips again under similar conditions- replace the device. ----- Don Kelly cross out to reply

I have had the problems on two separate GFCI units installed by the same electrician at different times. It always happens on the breaking of the circuit.
I also had a pump with obvious leakage to ground. It got so bad, that I could not use it on a GFCI at all. The leakage was low enough so that it would work well with just the protective ground. Because of the salt water it was moving and the consequent hazard, I decide to use that pump at all.
I decided to find the problem. To my chagrin, when I opened the housing, I found that the motor was immersed in oil. Oil was everywhere. Being careful not to slip on the floor, I disposed of everything without ever finding out what caused the electrical leakage. My wife was not happy.
I will post separately on one experience I had with line interference on electronics.
Bill

This is a digression from a previous post.
At one time I was working on a viable voltage regulated supply for tungsten halide lamps. I had to fix someone else's design by improving its regulation. It used thyristors triggered by unijunction transistors (UJT). It was a cheap way to go compared to the magnetic cards designed during thyratron days and sold by Sprague Electric.
When I was all done, I noticed some strange fluctuation in the lamp output. It turned out that the turn-on of the thyristors produced enough noise to trigger the UJT at odd times. This was especially true if the UJT circuit was near to triggering.
When there was only one of these units going, there was no problem. Two or more could end up reacting with each other. I had built a gizmo with a thyristor to put spikes on the line. The quick fix was to make an LC line filter that would keep line spikes from getting into and out of the unit. I was testing this filter while another unit was on final check before delivery. The unit to be delivered failed from these spikes. The one with the filter did just fine.
It turned out that a customer gave me the clue to solve the problem. He suggested that I add a transistor that completely discharged the time base capacitor every half cycle. That worked.
Bill

Stop using PVC for wet installations.
THHN or oil resistant... stuff like that.

I am not sure that I understand what you are trying to say. The actual wiring in the two cases is to standard outlet boxes with regular three-prong receptacles (hot, neutral and ground). The only wiring going into the water is the electric cords for the submersible pumps. The cords to the air pumps would only get wet by accident.
Bill

You said something earlier about thinking that it was tripping due to leaking via the insulation. I was referring to better wire, that's all.

If you can, try changing the motor switch to a solid state relay.
Have you tried MOV's across the line?
I had a problem with a certain type of power strip. This one had a neon indicator from hot to ground (to indicate proper wiring) it was enough to trip out the protection built in to an inverter.

A thyristor switch could be handy. It could allow lagging inductive current to continue to flow until zero current causes the thyristor to become nonconducting. But what I am using is a cheap Harbor Freight timer that allows any time on.off schedule in 15 minute combinations.
Bill

Are you the expert, Archie?
Do the people here know of your list of 75 nyms? How many did you use in this thread?

THE DIMBULB SCORECARD
Abbey Somebody AM AnimalMagic Archimedes' Lever AtTheEndofMyRope AwlSome Auger BarnCat Bart! BigBalls BillyPilgrim 10 BlindBaby Bungalow Bill Capt. Cave Man CellShocked ChairmanOfTheBored Chieftain of the Carpet Crawlers
Copacetic Corbomite Carrie DarkMatter DarkSucker 20 Do I really need to say? Dorothy with the Red Shoes on Dr. Heywood R. Floyd DrParnassus FatBytestard FigureItOut FunkyPunk FieldEffectTrollsistor FunkyPunk FieldEffectTrollsistor George Orr GoldIntermetallicEmbrittlement
30 Hattori Hanzo Herbert John \Jackie\" Gleason" HiggsField IAmTheSlime ItsASecretDummy Jupiter Jaq Kai LargeMarge life imitates life Lil Red Riding In The Hood
40 lurch MadManMoon MakeNoAttemptToAdjustYourSet snipped-for-privacy@thebarattheendoftheuniverse.org
MeowSayTongue Mr.Eko Mr. Haney MrTallyman My Name Is Tsu How Do You Do Mycelium 50 Mycelium Neanderthal Nunya OutsideObserver Perenis ? snipped-for-privacy@hereforlongtime.org? Pieyed Piper Phat Bytestard RoyLFuchs scorpius
SkyPilot 60 SomeKindOfWonderful
Son of a Sea Cook SoothSayer Spurious Response StickThatInYourPipeAndSmokeIt Sum Ting Wong Sum Ting Wong
SuspendedInGaffa The Great Attractor
TheJoker 70 The Keeper of the Key to The Locks
The Last Mimsy The Loner The_Giant_Rat_of_Sumatra TheGlimmerMan TheKraken TheQuickBrownFox TralfamadoranJetPilot snipped-for-privacy@thebarattheendoftheuniverse.org
UltimatePatriot 80 UpGrade UpYerNose ValleyGirl VioletaPachydermata WallyWallWhackr WarmUnderbelly 100WattDarkSucker 87