transients on ungrounded delta

[s22j26]

Hi,
I think I have some of the same issues described in thread238-227276. I hope you can help.
I'm working to develop a device to monitor and control an oilwell pumpjack and its motor. My device takes its power from the 480V/3-phase supply that runs the motor. Lately I've had a rash of catastropic failures (just to my box, not the entire well, thankfully) on one particular oil-field. I assume it's the result of transients on the line. Unfortunately I haven't been able to capture any transients yet.
Most pumpjack motors get their 3-phase power from a 12kV-to-480V wye-to-ungrounded-delta transformer configuration. I realize that ungrounded-delta is not recommended, but that's what I have.
I would like to design a transient suppressor for the input to my box without yet knowing what kind of transients I'm dealing with. I was going to just put an MOV from each phase to ground, but I've read that line-to-ground MOVs are not recommended for an ungrounded-delta. I have heard that this is because an ungrounded-delta can "float" to a high voltage and trigger the MOV.
- Is this really possible? How high a voltage can it float to?
- If this floated voltage is just the result of electrostatic accumulation, can I "bleed off" or "pull down" this charge? If so, how?
- If I can't put MOVs line-to-ground, should I settle for line-to-line?

As always, thanks for your help.

 

[waross]

Three small dry type transformers may be used in a wye:delta configuration to develop a symmetrical ground point. If the wye point is grounded through a resistor, the transformers need only be rated for the resistor current. The transformers and resistor must be rated for line to line voltage (480V).
The delta winding floats with no external connections and the voltage is unimportant.
This will limit the voltage rise to 480 V to ground.
In a floating delta, the secondary may be linked to the primary winding by either insulation leakage or capacitive linkage.
The actual voltage may vary from none to some fraction of the primary voltage.
Either a delta primary connection or a wye connection with a floating wye point will be a worse case for floating high voltage on the secondary. If you have a wye:delta transformer configuration the recommended primary connection is a floating primary wye point.
Copper thieves have been electrocuted by the resulting voltage on the tanks of large transformers when they cut the case grounding conductor to steal the copper.


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

 

[s22j26]

Bill,
Thanks for your reply. I'm a bit out of my element, but does this implementation have an effect on the floating delta, or just provide me a safe vantage point from which to observe it?
I would still need surge suppression on the lines of the floating delta, and I'm not clear on what I can use if the lines are floating.
Sorry for my ignorance. Thanks for your help.

 

[waross]

In regards to the wye:delta connection, the delta secondary must balance. If the delta vector diagram does not close on itself, there will be a circulating current that will increase until the delta does close on itself.
This results in the primary wye point being held at the electrical neutral position. The primary neutral/wye point may then be impedance grounded or solidly grounded.
The same effect as a zig-zag but cheaper and easier to understand for some of us.

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

 

[busbar]

This doesn't directly answer your question, but here is a ~50-year-old account of the perils on LV ungrounded systems—as opposed to high-resistance grounded systems. This all-time classic description is detailed in a chapter on system grounding in Beeman’s Industrial Power Systems Handbook.

Chapter 6 — System Grounding

About midafternoon one day in a West Coast manufacturing plant, normal operations became suddenly disrupted. The first evidence of trouble came in the form of a motor failure on the 480-volt system, then another, and still another in close succession. An inspection of switchboard voltmeters (measuring line-to-line volts) and ammeters indicated no unusual conditions. System equipment continued to fail. A test voltmeter was rigged up having a full-scale calibration of 1200 volts. Upon connecting it phase-to-ground, the pointer went off scale. A phase-to-ground potential on a 480-volt system of more than 1200 volts existed!

At once the incoming service transformers were suspected of internal breakdown between high- and low-voltage windings. As the last of these transformers was isolated and individually tested, it became evident that they were not at fault. System equipment continued to fail, and the situation was desperate.

A frantic group went into a huddle and decided that the only way out was to trip the main incoming service breaker, which would deenergize the entire system. At this point one of the workmen noticed a small wisp of smoke coming from a motor-starting autotransformer and, upon approaching, could hear a buzzing noise inside. This circuit was switched clear of the system, and the overvoltages disappeared. During the two-hour period that this arcing fault existed, between 40 and 50 motor windings had failed.

Finally it was found that the autotransformer enclosing case had been hashed in and was practically in contact with the coil. The spot where arcing had taken place was evident although not badly burned. An attempt was made to show the plant engineer what had been the trouble. A solid connection was made between the frame and the burned spot on the coil. Much to the bewilderment of the operating men and according to the expectations of the plant engineer, no more than the 73 per cent increase in the voltage to ground on the other two phases occurred. The main ingredient of the overvoltage (discontinuous conduction) had been omitted.

This is an actual case of severe prolonged experience of overvoltage of repetitive restrike origin on a 480-volt ungrounded system. This story is spectacular because of the magnitude of the disturbance and consequential damage. Similar occurrences of lesser extent are not uncommon, however, and there is evidence that they are more frequent than realized. It is a characteristic of ungrounded systems that they are subject to relatively severe transient overvoltages. This trouble can be avoided by proper grounding of the system, and other important benefits are also obtained.

Pp. 337-338 L. J. Carpenter and L G. Levoy, Jr.
Industrial Power Systems Handbook
DONALD BEEMAN, Editor
Manager, Industrial Power Engineering / Industrial Engineering Section, General Electric
McGRAW-HILL BOOK COMPANY, INC. 1955

 

[electricpete]

Also there's this:

http://www.postglover.com/Literature/GT210-08_Zigzag_Trans.pdf

The Zigzag should be connected to the system on the line side of the main breaker, as close as possible to the power transformer secondary terminals.

It sounds like you are not in a position to install anything close to the power transformer secondary, but are limited to your particular load..., and might be tricky to managing the protection when grounding the customer's system through a load. Not to mention cost of installing a system grounding device at an individual load. I guess if the zig-zag transformer is pursued it would have to be through the customer, and directly downstream of the transformer.

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

 

[electricpete]

Just to step back a bit, the failure have been only to your device/electronics, and not to the motor?

If yes, I'll throw out an idea for consideration (maybe others can shoot it down). How about isolating your entire electronics box from all ground? Then it would not see line to ground voltages. i.e. if you have a metal box enclosure, insulate it from grounded parts of the machine. Those ground overvoltages may well still be seen by other parts of the system including the motor, but if the electronics is the most sensitive part of the system, then maybe it's the only part that needs to be protected and isolated. Of course there may be some personnel safety concerns about leaving your box ungrounded which would have to be very carefully studied.

Also going back again to the idea of connecting resistances raised by WilliamMcCormik... I raised concern about losses, but that's probably not a huge issue because it doesn't take much ground current to have profound effect on ungrounded system. However the bigger problem seems to be considering all the effects including detection and protection for faults at this and other locations of the modified system.

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

 

[electricpete]

To address safety concerns I guess you could go with box within a box. The outer box is grounded normally and inner box is floating... insulation lives between the boxes.

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

 

[electricpete]

Also the spacing between boxes selected large enough to limit capacitive coupling.

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

 

[s22j26]

Thanks for everyone's comments/suggestions.
Yes, the damage is only to my box and not to the motor. Currently the electronics in my box are grounded. I have considered a floating neutral for the electronics, contained inside an grounded enclosure for safety. I'll give it some more thought.
In the meantime is there any benefit to connecting phase-to-phase MOVs, as I can't go phase-to-ground while the neutral is floating?

 

[waross]

If the problem is just your electronics, install a suitable isolating transformer and ground the secondary. You can protect the secondary with MOVs. The fairly high impedance of the small transformer will help the MOVs clamp the voltage. They will only have to handle the control transformer fault current, not the current of the motor supply transformers when hit with an overvoltage transient.

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

 

[electricpete]

Perhaps it would be helpful to know more about the circuit. Are there power electronic “switches” that directly control the main power? Or instead do the electronics just provides an input to mechanical contactor operation?

A simple diagram might help. Also is there any info about specifically which components failed?

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

 

[PHovnanian]

A bit different than electricpete's recommendation: Keep the electronics grounded, but provide suitable isolation on its power supply (via a suitably rated isolation transformer) and I/O interfaces to the system (by transformer again or opto-isolators).

We'd have to see the details of the system to provide better advice. But keep in mind that, if the 480V system is seeing transients sufficient to blow sensitive electronics, other insulation (motors, etc.) are also being stressed. Just not to the point of failure (yet). A solution applicable to the entire secondary system might be a wise investment. Your box is just the canary in the coal mine.

 

[s22j26]

Actually the majority of the damage was in the area of the control circuitry which drives the gate of the solid-state switches (SCR/thyristor).
We had an interesting development yesterday in which we saw similar damage as a result of over-currenting the switching device during the motor startup surge. I'm pretty sure that we didn't have this overcurrent in the earlier failures, but if not, then it makes me suspect the ratings of the switches. We selected them based on their ability to handle full-load-amps x 6 for about 1 second.
I'll keep you posted.

 

[electricpete]

So you do have power SCR’s associated with the main power. In that case isolation transformer doesn’t seem like much of an option to me (unless you can transmit main power through your isolation transformer… which would require a big transformer). The electronic switch scenario is the one that made me think you want to float all of your electronics…. as I previously described “box within a box.” Or maybe just an insulated floating metal panel within your outer grounded box…. All components mounted to the insulated floating panel with no connection to the outer grounded box. Just my thoughts fwiw... interested anyone else wants to comment whether that makes sense or is out in left field.

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

 

[PHovnanian]

Damage to the gate drive circuitry and overcurrent damage to the switches themselves may be unrelated. Or they could both be due to voltage transients. Its possible that a grounded controller driving ungrounded SCRs could result in spurious SCR triggering if the cathode voltage gets pulled down by a system transient. If this occurs at the wrong time (w.r.t. the three phase supply) that unwanted SCR conduction may result in a short circuit (line to line) through another SCR.

SCR gates also have a certain amount of parasitic capacitance. If the cathode is pulled one way or another by a voltage transient, the dV/dt can produce a current spike in the gate circuit.

Driving a set of SCR/Triacs from a controller with a different ground reference is a bad idea. If the controller must be grounded (for safety reasons), the gate drive outputs must be isolated from the floating SCRs by some means. Opto isolators are one way.