Today I tried winding a rather poor 75-turns air coil (improvised from an insulated 2 mm-thick copper wire), and with a ferromagnetic bracket I attempted to reproduce the basic process used in Alexander Parkhomov's "Woodpecker" (see references further below), but here instead at 12V DC and about 40A peak input from the usual Corsair HX520 switching DC power supply that I used recently. Due to the relatively large current involved and the speed at which the electrodes got separated, the counter EMF must have had a voltage greater than 12V (see Lenz's law), but how exactly it's hard to quantify.
A messy photo of the quick setup:
It seems to operate just like a welding machine. I've made a short video:
The coil got warm, but most of the heat probably was generated towards the end of the test, after the electrodes welded together and about 40A continuous were passed through the coil. Using metallic electrodes there's always a risk of this occurring.
I haven't run it with water yet, and I don't plan to until I can make a better crafted version that got rid of the bugs.