Test is complete. Attached are some pictures. Notes and questions below.
MOTOR DATA: 13.2kv, 3500hp, 324 rpm (torque equivalent of 30,000hp 2-pole), 50,000 pounds assembled.
CALCULATIONS RELATED TO TUB FILL HEIGHT: It is deisred to fill the tub to the right height so that level would be above connections, but not overflow the tank. Tub is 11.5' diameter and 80" tall. Stator weight unassembled is: 20,000 lbm. Estimated stator volume based on 500 lbm/ft^3 desntity of steel would be 40 ft^3. Since more dense copper is also present, it will be a little less volume. Left only 8" between water level and top of the tank prior to inserting stator. That doesn't look like much, but it equates to something like 50 ft^3 which was enough. Result was acceptable (did not overflow and covered the connections).
WATER TYPE: Pure demin water (non-conducting) was used and a non-ionic (covalent) wetting agent was added to obtain desired surface tension 31 dyn/cm specified in NEMA MG-1. The wetting agent supposedly helps the water get into smaller cracks than it otherwise could. NEMA MG-1 does not say much about the conductivity of the water to be used.
ELECTRICAL TESTS BEFORE PUTTING STATOR IN WATER:
1 - Megger at 10kvdc. 3.8 Gigaohms at 1 minute, 25 gigaohm at 10 minute
2 - Surge test – Sat I forgot the voltage.
3 - AC Hi-pot to 27,500vac for 1 minute for each phase to other two grounded. Sat. Amps 1250. Smell of ozone and sound resembling bath shower. Compute impedance of 22,000 ohms (capacitive)
4 - Winding resistance – balanced within 0.1%
5 - PD 7600vac. Individual phase to others/ground: 260milliamps. Compute impedance of 27,500 ohms (capacitive)
ELECTRICAL TESTS WITH STATOR IN WATER
1 - Performed IR at 1kvdc(*). 4.9Gigaohm at 1 minute. 33 gigaohm at 10 minutes
2 – Performed ac test at 12.5kvac. 2200 millamps (all 3 phases). Compute impedance 5680 ohms. Could not obtain target 15kvac due to ac hi-pot trip setting 2500 milliamps
3 - Repeat IR at 1kvdc(*). Results Gigaohms. (slightly better than before).
* I agree with Edison that 500vdc or 1kvdc is only a drop in the bucket for a 13.2kv motor. However the NEMA MG-1 standard says only 500vdc. There is no precedent or standard to apply higher dc voltage to a motor underwater that we know of. Based on discussions between motor shop and myself, we decided to go 10 1kvdc, but don't want to press our luck any higher.
PLAN: Motor will be dried in over for 12 hours, then repeat megger/PI at 10kvdc, followed by dc step voltage test to 28kvdc
1 – What do you think is explanation for small variability in computed impedance between AC hi-pot (27,000 ohms) and pd test (22,000 ohms).
2 – More importantly, there was a larger variation unexplained in the submerged ac test. I expected the capacitive impedance to decrease by some factor less than 3 (maybe 2?) when we went from single-phase-to-ground tests (before putting into water) to the all-phases-together ac test (under water). After all the single-phase tests include not only ground contribution, but also phase-to-phase contribution. But what we actually saw was impedance approx factor of 4 lower. Could the presence of the water bringing a ground-plane closer to the endwinding explain this? (it makes sense it would have some effect although I don't know how much since water is demineralized and non-conducting)
3 - Does the sequence of testing make sense to you? I am thinking about requesting the 27,500vac/2E+1 factory ac hi-pot after drying following the water test next time. That would give us even higher confidence that as-left condition of winding is very robust. Shop says factory ac test before submergence test is standard procedure they have always done. The standards appear silent on this aspect.
4 – Silly question. What role does water play in revealing a failure in the endwinding if the water has low conductivity (demin water). To complete the circuit to ground, the current still has to travel to the core somehow. Does dielectric breakdown of water play any role? Or more likely, even though it is pure water compared to tapwater, it still has significant enough conductivity that will facilitate enough current through water to trip the test set with pinhole leak with no water dielectric breakdown required? I'm not sure but interested in opinions/thoughts on what role the water plays in detecting endwinding weaknesses. Do you think different conductivity of the water (like tapwater plus wetting agent) would give significantly different test results?