Arcing at tank flange joint of Transformers

[prc]

When large transformers are energised a momentary arcing flash can be noticed across the tank flange (curb) joints at one or two locations.Normally cover and tank are bonded by earthing strips at two locations,but still the arc can be noticed.The earthing through the connecting bolts will not be effective due to paint film etc. The reason for this arcing seems the sudden charging of the ground capacitance between the winding to earthed tank brings up tank potential momentarily causing an arc to cover which will be at less potential.Any contributions?

 

[waross]

I wonder if the earthing straps were installed without cleaning the paint? What is the length of the arc path?

Bill
--------------------
"Why not the best?"
Jimmy Carter

 

[prc]

Earthing strips ( 2nos diagonally opposite positions) are OK. It is on stainless steel pad so that contact is perfect.Arcing is momentary and only at one place between cover and lower tanks

 

[electricpete]

It seems that any scenario some intermittent or poor contact between upper cover and lower tank.

Your scenario seems reasonable.

I could postulate another scenario. Let's say circulating current upon energization is interrupted by vibration-induced intermittent contact a few milliseconds later. Interrupting circulating current can certainly cause arcing. I have seen circulating currrent in scaffolds under the belly of the generator cause arcing when the poles are moved.


=====================================
(2B)+(2B)' ?

 

[electricpete]

I'm not sure if it is productive to do capacitive analysis to estimate voltage accross the gap using simplistic uniform field (parallel plate) model. But let's try and see where it ends up and you can think about it or ignore as you see fit.

Symbols
d = total separation distance
da = air distance between cover and tank
di = oil insulation distance
d = da+di= total separation distance
e0 = dielectric constant of vaccum approximately equal to that of air
ei = dielectric constant of insulation
eir = relative dielectric constant of insulation – epsi/eps0 ~ 2

Vapp = applied voltage (accross distance d)

Vair = portion of voltage drop occuring accross air
Eair = Vair/da = field in air
Ei = Vi/di = field in insulation
Xa = capacitive reactance of air
Xi = capacitive reactance of air

Approach 1: circuit analysis.
Vair/Vapp=*Xa/[Xa+Xi]
Use Xinsul~di/Epsi; Xair~da/Epsa
Vair/Vapp = da/ea / [da/ea + di/ei] = 1 / [1 + ea/da*di/ei] = 1 / [1 + (di/da)/eir ]
let eir = 2
Vair/Vapp = 1 / [1 + 0.5* di/da]
Vair = Vapp/ [1 + 0.5* di/da]

Let's say distance from 345kv VLL conductor to upper tank is 1 meter and gap between cover and tank is 0.1mm.
Then di / da = 10000
Vair = VLG/ [1 + 5000] ~ 0.0002 * Vapp ~ 0.0002 * 200kv = 40 volts.

With the numbers in this example, seems like arcing would not be likely from first look. But
1 - the assumed distance numbers might not be wrong
2 - uniform field assumption ignores possible voltage stress concentrations in the area of the small gap
3 - travelling wave behavior causes higher source voltages?


=====================================
(2B)+(2B)' ?

 

[electricpete]

1 meter - I was thinking about top of shell form tank clearance around bushing connections. But down below between windings and tank could be much smaller. You probably know better than me numbers to choose.

=====================================
(2B)+(2B)' ?

 

[prc]

pete, the issue is like this. I am referring only to core type transformers. The arcing is not continuous. It occurs only during the energisation and momentary. I have seen this only in large transformers. You have to be very careful to notice it.The arc jumps from the upper to lower tank or say lower tank to cover across the curb. The curb/flange will be almost touching each other with compressed elastomer ring between them.The bolts will be connecting the two curbs.Since paint film will not give effective contact, earth pads are provided on curbs of tank and cover at two diagonally opposite positions for connecting the earth leads, there by achieving the electrical connections between tank and cover.

At energisation, there will be no circulating current on tank as the unit will be with secondary disconnected.There will be inrush current ( of say 5-8 times full load current ) in HV winding. My thinking is the capacitance between winding and tank gets energised bringing up tank voltage momentarily. Since the length of tank is high say 5-10 metres long, the voltage differece between tank and cover cause momentary flash equalisng the voltage.

 

[waross]

I would be investigating the integrity of the grounding straps. There may be hidden corrosion. I agree that it is probably a capacitive charge but I would expect the grounding straps to prevent it from building up to a potential that will cause an arc.

Bill
--------------------
"Why not the best?"
Jimmy Carter

 

[FPelec]

Hello prc,
I have often noticed the phenomenon you describe in large transformers in our grid, mainly on large 3-limbs autotransformers.
Also in our case the tank cover is connected to the tank by means of 4 large (125 mm2) copper strips.
In my opinion this effect could be not only caused by the capacitance coupling between tank, tank cover and windings, but also by the stray fluxes exiting the core and jumping into the tank cover: i have never seen this happening when the transformer is energized using point on wave controller; on the contrary, when performing short circuit withstand test on large units in KEMA lab (using pre-set short circuit so that the point on wave controller is instead used for maximizing the current) this phenomenon always happens .
In my opinion the transient cause high-frequency overvoltages, so that the impedance of the grounding path makes them ineffective.
I also think that this phenomenon should not be dangerous for the transformers

Si duri puer ingeni videtur,
preconem facias vel architectum.

 

[prc]

FPelec, I agree with your views. It means this phenonenon is not due to capacitance effect but due to magnetising inrush current.This current will cause leakage flux from core and winding thrown on the tank. This may be the reason that this is not seen always but some times only.The magnitude of inrush current will depend on the point on voltage wave where energisation occurs.

I have never seen this causing any problem. But recently in case of a very large transformer this flashing caused burning of paint near to a bolt connecting cover to tank.Will connecting tank cover to ground by separate earthing lead will reduce this? ( instead of connecting cover to lower tank at curb)

You are right, in KEMA also this is noticed always.

 

[FPelec]

Hello prc,
I have never tried or seen the solution you proposed, so I'm not sure that it could be effective in avoiding arching: the inductance of the grounding path would be higher and thus I would expect also higher induced voltages.
If the arching happens always in the same place, maybe it could be better to add in that zone further grounding connections between tank and cover.
Generally, the tank and the cover are painted separately and then joined by bolts.
If the phenomenon is localized on a single bolt, maybe it could be sufficient to remove the paint on the bolt, the tank and the cover, in order to ensure a good conductivity, and then paint again after tightening the bolt.


Si duri puer ingeni videtur,
preconem facias vel architectum.