Smells like phantom voltage/current to me. Get a high value resistor at least 100kOhm, preferably 1MOhm to start with. Connect the chassis to ground via that resistor and take the same measurement again. The voltage you are seeing should collapse.

For reference, RCD (current leakage safety switches) pop at abour 30mA depending on the local rules. So you are orders of magnitude under that.

Cheap multimeters can show weird numbers when measuring high-impedance floating metalwork, especially around mains wiring and electrical noise. That does not prove it is safe, but the raw voltage reading may not mean what it first appears to mean.

The tingling I wouldn’t trust just because, tingling… BUT it can well be a thing with non grounded or poorly grounded gear. I get to tingling from my laptop even though it is connected through a power pack.

The big issue with metal chassis and single insulated housings is that without a ground if it does develop a fault (say something hot melts abd grounds out and r a motor winding cracks), then you are genuinely hot with nowhere for current to go. It really should be grounded somehow if it’s an old chassis that isn’t designed with multiple insulation.

At the very least. Get a safety protected plug (residual current device). They are cheap and effective. But they are you last line of protection not your first.

• not an engineer. YMMV. I don’t know anything and my advice probably faulty … etc. but definitely if you know what you’re doing and can do it safely: test voltage with a big resistor (should be an easy test), get safety switch.

It’s a nice idea but impractical. Energy = mass x gravity x height

Say 10 tonnes of water pumped into a massive 20m high tower… E =10000x9.8x20 Joules. ≈ 2 mega joules.

Theres 3600 joules to a watt hour so divide that and you get:

≈ 550 watt hours (0.55kWh) of storage (assuming perfect efficiency of the turbines). About the same a half a dozen big cordless tool battery packs.