Worst case Fault currents from Generator Decrement Curve

[Wfg42438]

Hello Everyone,

I was reading this article on generator decrement curves created by cummins and its was quite interesting.

On Figure 3 of the article, they show that in there example for a terminal fault at around 0.1 seconds both the line to line and L-G fault currents are larger than the 3Ph fault current.

Loading…

www.stamford-avk.com

I have seen articles on the L-G being largest but was surprised to see the LL current also being larger than the 3Ph current.

I wanted to ask in practice how is this accounted for and how common is this case?

I am thinking in practice your limiting factor is the LG fault current since the generator will have a maximum time it can sustain for under a LG fault which for the example in the article is much smaller than for the LL case.

Is the idea to simply use protection that operates prior to the time where this phenomenon occurs?

if anyone can shed some light on this it be much appreciated.

 

[dpc]

I didn't read the article, but since it's about generator decrement curves, I'm assuming these are transient values, calculated for a fault at the generator terminals. The decrement curves generally assume no change to generator excitation, but in the real world there are often fault current support systems (current boost) that will impact the resulting fault current. For longer duration faults, the problem becomes the lack of fault current since the synchronous reactance be greater than 100%, meaning the fault current is less than the rated current. The type of excitation also has an impact.

Larger medium-voltage generators are almost always grounded through some type of impedance to limit the SLG current for external faults. In addition, large generators are normally connected directly to the delta winding of a unit transformer to reduce the chances for SLG fault that is seen by the generator. For solidly-grounded machines, the generator can be braced to withstand the forces associated with full magnitude SLG faults (if specified this way).

LL faults are much less common than SLG. For very large machines, isolated phase bus duct is used to connect the generator to its unit transformer. This greatly reduces the risks of LL faults.

For generators breakers connected directly to the generator terminals, there are specific ratings and designs intended to allow the breaker to clear these faults at the very high X/R conditions that exist at the generator.

 

[Wfg42438]

I didn't read the article, but since it's about generator decrement curves, I'm assuming these are transient values, calculated for a fault at the generator terminals. The decrement curves generally assume no change to generator excitation, but in the real world there are often fault current support systems (current boost) that will impact the resulting fault current. For longer duration faults, the problem becomes the lack of fault current since the synchronous reactance be greater than 100%, meaning the fault current is less than the rated current. The type of excitation also has an impact.

Larger medium-voltage generators are almost always grounded through some type of impedance to limit the SLG current for external faults. In addition, large generators are normally connected directly to the delta winding of a unit transformer to reduce the chances for SLG fault that is seen by the generator. For solidly-grounded machines, the generator can be braced to withstand the forces associated with full magnitude SLG faults (if specified this way).

LL faults are much less common than SLG. For very large machines, isolated phase bus duct is used to connect the generator to its unit transformer. This greatly reduces the risks of LL faults.

For generators breakers connected directly to the generator terminals, there are specific ratings and designs intended to allow the breaker to clear these faults at the very high X/R conditions that exist at the generator.

Thank you for the detailed response.

If you are willing to quickly look at Figure 3 of the article in the link from cummins

there is a time after the transient region where the L-L and L-G currents are larger than 3 Ph for the generator in the sample

Any comments on why after some time the LL fault current goes from being less than 3Ph to greater than the 3Ph in Figure 3?

From what i understand this is based on the AVR and excitation system which will be designed to sustain under fault condition for some amount of time

in this case if you look at the tables they mention that for LL fault the generator can sustain for up to 5 seconds

for LG up to 2 seconds and for 3Ph faults up to 10 seconds

So in practice for this case is the LG case the limiting factor as it will have the highest magnitude of current and a shorter allowed duration?

Would the action to take simply be to make sure the generator protection can adequately handle the LG fault current and clear the fault much faster than 2 seconds?

Also in practice would you only take this to account for faults very close to generator and once you go at least one transformation away or more will these cases of LL & LG currents be less of an issue?

 

[Gr8blu]

Short circuit current decays with time. The variation depends on two types of decrement - AC and DC. The DC decrement occurs during the first few cycles of the short circuit (i.e. in the subtransient region). The AC decrement is spread over a much longer period, covering the subtransient, transient, and steady-state regions. Each region is a result of the respective impedance (X''d, X'd, Xd) within that time increment. At the terminals of a generator, the variation is quite pronounced.

When a short-circuit occurs in an electrical system supplied by generators, the short-circuit current presents an initial peak and then begins to decay rapidly since it does not have enough energy to sustain the initial value permanently. In most cases the permanent short circuit current may be less than the generator full load rating. This phenomenon occurs when the synchronous reactance is greater than 1 per unit.

The instantaneous short-circuit current value depends on the generator loading at the time of the fault and the elapsed time since the fault. From the unloaded to the rated load condition, there are families of decrement curves.

How is each component calculated for . . 3-ph fault . . . . . L-L fault . . . . . . . . . . . . . . . . L-N fault?
Steady state current . . . . . . . . . . . . . . . . . I = E / Xd . . . . . . I = 1.73 * E / (Xd + X2) . . . I = 3 * E / (Xd + X2 + X0)
Transient current . . . . . . . . . . . . . . . . . . . . I = E / Xd' . . . . . I = 1.73 * E / (Xd' + X2) . . . I = 3 * E / (Xd' + X2 + X0)
Subtransient current . . . . . . . . . . . . . . . . . I - E / Xd'' . . . . . I = 1.73 * E / (Xd'' + X2) . . I = 3 * E / (Xd'' + X2 + X0)

Here is the first kicker - the generator manufacturer doesn't know all the "system" details, so the only values that can be calculated are for the fault at the generator terminals.
And here is the second - the value of "E" changes between steady state, transient, and subtransient calculations depending on the prevailing conditions at the time of the fault (steady state excitation, prior non-zero load, or no load).

 

[davidbeach]

There are also zero and negative sequence impedances that start out no more than the subtransient impedance and then don't go up during the fault. The sequences being used by the fault matters.

 

[dpc]

If you're really interested in this topic, I'd recommend getting the book Analysis of Faulted Power Systems by Paul Anderson, et al. The old Westinghouse T&D book has good information as well, iirc.

As Gr8blu points out, the fault current decays with time. So for a 5 second fault, the fault current will be drastically lower at the end of five seconds, assuming it lasts that long. (five seconds would be an eternity for a fault on a properly protected generator).

Also be extremely careful about making assumptions regarding the excitation response that was used to develop the decrement curves. I would recommend getting specific decrement curves, excitation response and generator constants from the manufacturer for any generator.