how to calculate and limit fault currents in paralleled systems

[dseidel]

We need to design/manufacture a distribution system with a total capacity of 12000kVA of generated power.
This involves 6 off 2000kVA diesel generating sets paralleled onto a common busbar with each set having a 3200A air circuit breaker in line to feed the busbar.
Apart from the obvious; to break the busbars into manageable sections as suggested by us; what other magical means are available to handle the high prospective fault
currents on the busbars and to protect existing switchgear with " fault limiting inductors or other devices " as suggested by the consulting engineers office.

 

[bigamp]

dseidel,

Can you give a bit more of a description of your system. In particular:

operating voltage
number of feeds required from the "common busbar"
loads of each feed
nature of loads to be fed (existing MCC's maybe??)
If existing MCC's, their fault rating?
Have you calculated the prospective fault currents at the "common busbar", three-phase and phase to earth? what are they?

I assume you are looking at generating and distributing at LV, there must be a good reason for this. The sort of generating capacity you have suggests a smaller number of HV machines and an HV distribution system may have been better, with HV/LV transformers feeding the LV system so why not use an HV approach?

If you have to go LV, it is possible to obtain type tested 100kA LV switchboards, 100kA (150kA even) ACB's and 100kA MCCB's.

Your suggesation (splitting) is a very good one. Is it practical for your installation? Other possibilities include:

feed via fuses (may stuff up downstream discrimination)
feed via hi capacity current limiting MCCB's
feed via reactors (not sure if practical at LV)

Also, take into account the fault current attenuation that LV feeder cables provide.

Depending on the equipment you will be feeding, a combination of the above (with splitting) may well get you there.

Have you spoken to a switchboard or switchgear manufacturer?

Regards

 

[peterb]

In addition to the remedies suggested by Bigamp, there are current limiters available that are designed specifically for this scenario.
They are essentially fast acting fuses that are shunted in normal operation by a bridge. When the controls sense the inception of a fault current that will exceed the rated capacity of the protected switchgear, an explosive charge is fired that opens the shunt bridge. The current limiting fuse then interrupts the current before it reaches the first peak.
Sounds horrendous, but they do work. Check the ABB website for details.
I do prefer the passive current limiting reactor approach, if your system can accommodate the voltage drop. In fact, looking back at the above, I fully agree with Bigamp that it would be a far better idea to generate at medium voltage (say 2.4 kV, 3.3 kV or 4.16 kV depending on your location and standard voltages) and provide a single step down transformer to the existing LV switchgear. The transformer provides the current limiting reactance required.

 

[jwerthman]

I second the previous comments regarding paralleling these units at medium voltage, then stepping down at necessary locations to feed the existing low voltage systems. Based on experience, if this system is put together using current limiting reactors, you or they will regret it for years to come. These types of systems can be cobbled together, but they almost always end up being a problem in the future.

 

[gordonl]

A hybrid solution to the current limiting reactors is a current limiter as described peterb in parallel with a current limiting reactor. During normal operation current flows through the limiter with virtually no voltage drop, but on a fault the limiter opens up forcing the fault through the reactor, for normal discrimination of faults.

Once the fault is cleared the system operates through the reactor until the current limiter is replaced. You can also play with current limiters in startegic locations such as bus ties, main incomming, etc.

 

[gordonl]

By the way G&W makes current limiters as well.

http://www.gwelec.com/clip.htm

 

[busbar]

2000kVA/3200A looks like low voltage. I'd think something like Cutler-Hammer DSL's would cover a lot of territory. Seems like they would be perfectly adaptable to most any protective relaying or control scheme.

 

[Bung]

If you are feeling adventurous, speak to Australian Superconductors (you should be able to find their website via any good search engine). They make what is essentially a superconductor based magnetic amplifier - dc in a superconducting coil saturates an iron core, and lowers the inductance of an ac winding which is in series with your source. On detecting fault current, the dc is dropped, the core comes out of saturation, and bingo, the ac coil is choked. Could be done with conventional mag amp techniques, just a bit lossier. The superconductor is liquid nitrogen cooled, and the thermal lag on loss of refrigeration is days, not hours. They do have some commercial installations in service, I believe.

Bung

Just a suggestion!

 

[Etrix]

Hi..

Here's a question that's arisen from jwerthman's reply to this post:

What sort of problems do you anticipate on a system using current limiters? I am working on a project looking to install at least 3 current-limiters (possibly ABB Is-limiters) separating 6 off 40MW generators (11kV system). We are using it as gordonl described in his reply. Thank you!

Etrix

 

[alehman]

There is a fairly wide range of winding impedances available in low voltage genests. Caterpillar units typically are furnished with much lower subtransient reactance (higher fault current) than Onan/Cummins for example. This can be specified within limits. Availabiltiy depends on schedule. Take a look at your options.

Be sure to watch your line to ground fault values. They can be considerably higher than a 3-phase fault. Many low voltage generators have extremely low zero sequence reactances. This is largely a function of the winding pitch. I've been caught by this a time or two. You may need a neutral grounding reactor to reduce SLG fault values.

MCCB's (insulated case) work well to protect for high fault currents, but they don't coordinate well due to their instantaneous trip. MV is the way to go if you can.

 

[moosanbh]

thanks