Equipment Damage due to Power Disturbance 4

[MechProjectEng]

Hello all,

Hoping for some advice:

I work on a large industrial site, and a few weeks ago, we had a site wide power failure caused by a short circuit of one of our 11kV switchboards due to flooding.

The site is served by 2 incoming 11kV feeders, and then distributed by 6 seperate 11kV rings (Ring 1 to 6).

The fault occured on one of these rings (Ring 2). However, the fault caused the main incoming 11kV breaker to trip, causing all rings to de-energize.

Following the incident and subsequent re-energization of the facility, the following equipment damage was observed:

1. 30 No. 1000W Metal Halide Light fittings in one of our buildings burnt out (burnt out items are the capacitors, ignitors, bulbs, and 6" of the cable connected to the light fittings)
2. The power board of a 4kVA UPS burnt out.

We cannot see any technical reason for the damage other than a voltage transient/spike.

However, what is perculiar is that the damaged equipment above was connected to substations on Ring 6 (not Ring 2 where the fault occured), and there is no trace of any damage occuring closer to the source of the fault.

We are investigating the cause of the equipment damage as well as recommendations to avoid recurrance of this incident.

Your comments would be appreciated.

Thank you.

Here are some links to some photographs of the damage:

http://files.engineering.com/getfil...c892-4bbc-b54d-781c9f93abd5&file=IMG_2395.jpg

http://files.engineering.com/getfil...dbb1-439b-88ff-9bccd45bb5a7&file=IMG_2592.jpg

http://files.engineering.com/getfil...61c7-4026-9c1a-4b2e7c58a455&file=IMG_2599.jpg

http://files.engineering.com/getfil...966e-4793-9cb9-0a552cd82d86&file=IMG_2595.jpg

http://files.engineering.com/getfil...8dc5-4b08-9242-4d11299d1b9f&file=IMG_2590.jpg

 

[Skogsgurra]

I have had a similar situation where it was supposed that a transient overvoltage burned out devices. It turned out to be something entirely different. And I think that your fixtures are victims to the same failure mechanism.

Most modern halide fixtures have a constant wattage ballast. They try to keep power delivered to the lamp at a constant level and does that very effectively. That means that current drawn from the grid increases when voltage drops. If there is no undervoltage supervision and if the fixtures have no turn-off mechanism built in, then a low voltage - typical in a situation like the one you describe - will make current increase until something burns or explodes. See recordings below. They illustrate such a situation.

http://www.gke.org/presentationer/files/SICK PSU overcurrent when grid drops out 1.jpg

I think that is what happened to the MH fixtures.

The UPS sees a similar situation. It "wants" to deliver power to the load and tries as much as it can to do it. When input voltage gets low, it draws more current. And, just like the fixtures, something burns or explodes.

I am sure there are other devices that were about to die also. But you were lucky this time.

I think that a deeper investigation is needed to find out exactly what happened. Are there any recordings showing voltage on the different rings? That would be a good start.

I do not think that the damages were caused by transient overvoltages. There is too much power involved in the damages. Also, I do not quite see how massive overvoltage transients could occur on a loaded bus, after the breakers.

Gunnar Englund

http://www.gke.org

--------------------------------------
Half full - Half empty? I don't mind. It's what in it that counts.

 

[DRWeig]

Thank you Gunnar, for the illumination.

I had a similar experience with a warehouse full of metal halide hi-bays with CW ballasts. As you stated, the damage was too severe to have been a spike-up. I will rest easier.

Best to you,

Goober Dave

Haven't see the forum policies? Do so now: Forum Policies

 

[ScottyUK]

I am far from convinced that this was an under-voltage event, although it is certainly has the appearance of a prolonged high energy fault rather than a transient. A magnetic ballast is not a constant-power load (unlike a high frequency type, which is what Gunnar has accurately described) and an under-voltage event isn't likely to cause any damage. The UPS is a mess - I think I can see the remains of a blue disc varistor in the blackened area, suggesting that this may the area of the board where the raw power input connects via a filter. Destruction of the filter again could indicate a high voltage, high energy fault.

Is the ring 3-wire 3-phase? What is the protection for the ring itself? If relays do you have any disturbance records for the magnitude and duration of the fault? If fuses did all three phases clear or just one? Do you have any information relating to the earthing scheme? Is Ring 6 in the return path for HV fault current? Where was the ring open point? Are the individual substations classed as 'hot sites'? What condition are the earth electrodes in, especially on Ring 6?

 

[davidbeach]

Scotty - The constant wattage autotransformer ballast for Metal Halide lamps is not that unusual.

It would be very helpful to know the duration of the fault and the associated abnormal voltage.

 

[ScottyUK]

I have learned something there - I didn't realise that a magnetic ballast behaved as a constant wattage load. I'm aware of the negative resistance slope for a discharge lamp, but hadn't made the connection with how the ballast behaves.

Makes sense - thanks David.

 

[davidbeach]

Most magnetic ballasts don't, just the family of CW ballasts. And large MH lamps on magnetic ballasts are often on CW ballasts.

 

[zer0sequence]

Are we talking about an ungrounded power system? If so, then a transient over-voltage due to an intermittent ground fault would be a likely culprit.. need more information regarding the power system and fault that occurred.

 

[MechProjectEng]

Thank you all for your input.

It appears that there is a general consensus that our damage occured due to an under-voltage event.

The next query: How can we prevent this type of damage from occuring again?

Thanks again for your assistance.

Answerwing some queries:

-Skogsgurra:
*Are there any recordings showing voltage on the different rings? [highlight #EF2929]We are trying to retrieve the voltage logs for all substations at the time of the event from our monitoring system. However, this is proving harder than we expected. It is questionable if the information can be retrieved at this stage due to the configuration of the monitoring system and memory of the server.[/highlight]

-ScottyUK:
* Is the ring 3-wire 3-phase? [highlight #EF2929]Yes.[/highlight]
* What is the protection for the ring itself? [highlight #EF2929]The VCBs on the HV switchgear in Ring 2 triped, as well as the SF6 breaker at the main incoming HV supply to the facility.[/highlight]
* If relays do you have any disturbance records for the magnitude and duration of the fault? If fuses did all three phases clear or just one? [highlight #EF2929]The magnitude of the fault was 13kA. I don't have the time of the fault to hand. I will check and reply.[/highlight]
* Do you have any information relating to the earthing scheme? Is Ring 6 in the return path for HV fault current? [highlight #EF2929]No, Ring 6 is not in the return path. Ring 6 is a seperate HV ring to Ring 2 where the fault occured.[/highlight]
* Where was the ring open point? Are the individual substations classed as 'hot sites'? [highlight #EF2929]No, the individual substations where the damage occured are not considered as 'hot sites'.[/highlight]* What condition are the earth electrodes in, especially on Ring 6? [highlight #EF2929]The earth electrodes are in good condition.[/highlight]

-Zer0sequence:
Are we talking about an ungrounded power system? [highlight #EF2929]No, the system is a grounded system.[/highlight]

 

[Skogsgurra]

There are a few short term solutions to this.

The obvious one, but not the best, is to put individual overcurrent protection before each object.

A better solution, but one that will be blamed every time you get a nuisance trip, is to add delayed (1 second or so) undervoltage tripping to each ring. Then, the undervoltage is not allowed to act for any appreciable time and thermal damage will be avoided. Check if the fixtures have a lower voltage limit (I can imagine -20% or so) and set the trip voltage just above this limit.

A simple test in order to verify my theory about prolonged low voltage would be to subject one of the fixtures to a succesively lower and lower voltages and record voltage, current and temperature of key elements in the fixture. If current increases beyond what's good for the fixture - then the failure mechanism has been found and a delayed undervoltage trip should be applied.

Gunnar Englund

http://www.gke.org

--------------------------------------
Half full - Half empty? I don't mind. It's what in it that counts.

 

[MechProjectEng]

Thanks Skogsgurra.

Do you also have any theories as to why we experienced such major damage on ring 6, while we had no damage in ring 2 (where the fault occured)?

Thanks again.

 

[davidbeach]

Could be that in ring 2 the voltage went low enough that things just shut off rather than increasing their current draw. In ring 6 you could have had the right (wrong?) combination of voltage sag, duration, and load mix to get that damage; higher voltage and the current would be lower, lower voltage and the MH lamps would have gone out. Faster clearing would have reduced the level of damage.

 

[Skogsgurra]

Davidbeach may have given an explanation already.

One major cause could be that you have more motors running with heavy inertia loads - maybe even with phase compensation capacitors - or if you have synchronous motors that kept voltage up in ring 6 for a prolonged period.

You are really asking much of us. It would be good to have the one-line diagrams for the plant available. Voltage recordings would also be very good. Especially for ring 6.

Gunnar Englund

http://www.gke.org

--------------------------------------
Half full - Half empty? I don't mind. It's what in it that counts.

 

[MechProjectEng]

Thanks David/Skogsgurra,

I know I'm asking a lot! I have attached a simplified sketch which might assist:

http://files.engineering.com/getfil...-4184-ae78-ed8bca831d4a&file=201301291857.pdf

A few other observations which may assist:
- Undervoltage protection is provided on the LV (415V) switchgear. The settings are 0.5s @ 0.7PU, 2s @ 0.8PU and 5s @ 0.85PU.
- The allowable voltage range for the Metal Halide lamps is 228 - 252V
- The allowable voltage range for the UPS is 160 - 280V
- Ring 6 does not have many motors with high inertia loads. The load on Ring 6 at the time of the fault would have been mainly lighting, and a few small motors for building HVAC.
- However, Ring 6 is the most heavily loaded ring by far, the average load on ring 6 being more than twice that of any of the other rings.
- The time delay between the first trip on earth fault on Ring 2 and the trip of the LV switchgear feeding the lighting and UPS on Ring 6 was approx 1.5 second (according to the time the alarms came in on our monitoring system).

A few other queries we have:

- With reference to the photographs of the burnt out lighting units. Why would only the first 6" of the power lead burn like this? If there was a high current in the supply line, wouldn't the whole lead be damaged?
- If there was a current spike caused by an undervoltage condition, why didn't the overcurrent protection at the distribution boards feeding the lighting and the UPS trip?

Thanks

 

[MechProjectEng]

Hi All,

Can anyone help with these queries:

- With reference to the photographs of the burnt out lighting units in the original post. Why would only the first 6" of the flexible power lead burn like this? If there was a current spike caused by undervoltage, why wouldn't the whole lead be damaged?
- If there was a current spike caused by an undervoltage condition, why didn't the overcurrent protection at the distribution boards feeding the lighting and the UPS trip?

Thanks.