Spurious Differential trip of GE Digital Generator Ptotection Relay

[Sackeyfio]

There are 3 identical generator step-up transformers T1,T2,T3,2 step-up transformers, 2 identical generator step-up transformers(these were added later) GSU1,GSU2 and 2 identical station service transformers SST1 and SST2.

Realized that when generator is in service only
GE Digital generator Protection relays for T1 and T2 trips spuriously on differential when either GSU1 or GSU2 transformers is energised on no-load from the HV busbar. Because GSU1 and GSU2 are normally kept energised even with generator off-line this problem of spurious tripping comes about when GSU1 and GSU2 have to be taken out for maintenance isolation and restored to service afterwards.

Observation Made so far:
1. GSU1 and GSU2 have lower impedances than the other transformers.
2. Current oscillographs taken from DGP relay after the trip show waveform distortion on system side current without any corresponding distortion on return-side current. This is likely to come from harmonics associated with the inrush.
3. Energizing T1,T2, SST1 and SST2 no load does not trip the differential.
4. DGP relay for T1 and T2 are ver 4.12400F 1995-1997 whilst DGP for T3 is ver 4.12500F 1997-1998. DGP for T3 does trip on differential when on line.
5. Most of the spurious trips occur on phase C
6. Relay has been checked for stability OK for phase C.
7. CT ratios checked OK.

Anybody has some experience with these DGP relays ver 4.12400F. What do you think could be the causes of these spurious trips? I will appreciate your contributions. I have contacted GE Multilin who asked me to send oscillographs to them which I did.

 

[dpc]

Is this a new problem? Have you been able to successfully load the transformer with these relays? Does trip occur every time you energize the transformer?

Offhand, it sounds like a problem with the harmonic restraint in the relay. I'm not familiar with the DGP, but in many digital relays, the harmonic restraint can be turned off. Some relays also allow the differential to be disabled on energization.

You should also review the minimum pickup for the differential element.

If you've had this problem since day one, it is probably a relay setting issue. If it is a new problem, see if the relay configuration has been changed.

 

[Sackeyfio]

The transformers GSU1 and GSU2 share the same HV busbar with transformers T1, T2,T3. It is the back-energisation of either GSU1 or GSU2 that cause generator differential of units feeding T1 or T2 to trip on generator differential.
We are not too sure whether it is new problem because those 2 transformers GSU1 and GSU2 are normally kept energized from the HV even when their generators are off-line.

Once those transformers are already energized, we can load the units in question without any problem.

Since we became aware of this problem, we have tried to avoid the energisation of GSU1 or GSU2 when generators feeding T1 or T2 are on-line. Infact the last three energisations have led to tripping of generator differential of units feeding T1 or T2. So I can it always trips.

I agree with you that it is a harmonic restraint problem but unfortunately, the DGP is for generator protection and not for transformer therefore it is not equipped with any harmonic restraint function. About disabling of the differential during energisation, I think it will not be feasible here as the transformers being energized is not associated with the generator that is tripping.

The minimum pickup setting for the differential element is 0.3A. The maximum settable pickup is 1.0A.
Find below current values from oscillopgraphs of trip values.

1. Ph C(system side)= -1.51A ;Ph C(Return side)= +1.23A
2. Ph C(system side)= +1.06A ;Ph C(Return side)= -1.07A

With these trip values even the maximum settable pickup of 1.0A will not prevent a trip.

One of the events had all three units for T1,T2 and T3 in service. The units for T1 and T2 tripped on phase C generator differential but unit for T3 did not trip on differetial. DGP for T3 unit is a later firmware version.
Do you think that the C-phase CTs for T1 and T2 units are probably generating too much harmonics or the older DGP relay for T1 and T2 units are susceptible to harmonics?

 

[jghrist]

Does the oscillography show Ph C system side of opposite polarity than Ph C return side before energizing GSU1/2? It looks like a CT wiring error if I am interpreting your current values correctly. Which side is the generator side, system or return?

I'm not familiar with the GE DGP, but usually, numeric relays are not responsive to harmonics and act on fundamental only because of digital filtering. Harmonics are used for restraint purposes. For this reason, I doubt that CTs generating harmonics are the problem. I could see the GSU1/2 differentials operating on inrush if there is no 2nd harmonic restraint, but this would not explain the tripping of T1 or T2.

 

[Sackeyfio]

The oscillography data shows system side and return side of same polarity normally but because of the waveform distortion of the system side current, there exist an instant when polarity of system side is opposite to that of return side. It is not a CT wiring error otherwise the generator would even trip during normal loading. The system side is the terminal side of the generator and the return side is the neutral side of the generator.

If usually, numeric relay filter out harmonics, then we the current waveform should be clean. The waveform on the C-ph system current is distorted with some DC offset. Does this mean that the DGP does not filter our harmonics?

 

[stevenal]

I'll disagree with the idea that this is a harmonic restraint issue. Harmonic restraint desensitizes the relay of the transformer being energized, since the harmonic rich inrush would otherwise appear as operating current. The issue here is energization of a downstream transformer, which should not appear as operating current, regardless of the harmonics present.

I agree with jghrist, check those C phase CT polarities. A CT polarity tester can check the entire path from primary to relay.

 

[dpc]

I have to apologize - I misread your initial post. I was thinking **transformer** differential, not generator.

I agree with stevenal that harmonic restraint is not really the issue.

However, if the generator differential relay can normally carry generator load current, I don't see how this can be a CT wiring or polarity issue - if the CTs were miswired, the relay would trip on any generator load.

If I understand the situation (always a risky assumption), you may have a sympathetic inrush condition on transformers T1 and T2 when energizing GSU1 or GSU2. This can create a situation where dc offset current flows in the energized bank, causing iron saturation, and it will look like offset inrush current. If your differential zone includes the generator and the transformer, it will appear as an internal transformer fault.

Not all digital relays provide dc offset filtering and not all provide harmonic filtering. I know the newer GE digital relays (Multilin) do have dc offset filters, but I don't know about the DGP. You might want to discuss with GE/Multilin folks.

As far as harmonic currents, the CTs should have good fidelity at these lower harmonics and their output should be good reflection of the primary current harmonics. DC offset is another matter - this will quickly saturate any CTs.

Not sure there is an easy solution other then possible relay upgrade.

 

[jghrist]

Sackeyfio,

The oscillography probably uses unfiltered data, whereas the relay operation would use filtered data. I agree with dpc; you should discuss this with GE.

Are the transformers included in the differential zone?

How do the oscillograph current values during energization compare to full load current?

dpc,

I thought sympathetic inrush occurred when the transformers were in parallel. This doesn't appear to be the case here, with only the high side connected together.

 

[dpc]

jghrist - you may be right. I thought sympathetic inrush was possible with only one side of the transformers connected together. It seems that the dc current would flow directly into the winding, regardless of how the other winding on the core is connected, or not connected.

I've never really had a problem with sympathetic inrush, so I'm not speaking from experience.

 

[alehman]

These areticles apply to the GE SR489 generator relay, but may be of use in diagnosing your problem. The basic conclusion is that it is important for your generator differential CT's to match as closely as possible. Slight differences can cause unequal performance going into saturation (such as during transformer inrush). My experience with the SR489 is that significant time delay on differential overcurrent seems to be required to prevent false trips when energizing large transformers (not a good fix).

http://geindustrial.com/products/support/489/get8403.pdf

http://geindustrial.com/products/support/489/get8402.pdf

 

[Sackeyfio]

The generator differential zone is just the neutral to terminal of the generator. It does not include the transformers T1 or T2 neither does it include GSU1 nor GSU2.

The oscillography current values during back-energisation of adjacent transformer is lower than full load current. There is no appreciable increase in current magnitude, it is just the distortion which is very significant.

The issue is that Generators feeding HV busbars through transformers T1 or T2 are reacting(tripping on generator differential) to the inrush associated withback-energising of adjacent unit step-up transformers GSU1 and GSU2.
The terminal side current waveform undergoes distortion and dc offset during the energisation.Because the neutral side current waveform does not undergo similar distortion there is always some operating current that is enough(very high) to trip. Why is the neutral side current waveform not under similar distortion?

We have contacted GE Multilin and have given them all the oscillographs but we are yet to hear from them.

dpc,

Relay upgrade is something we are considering but that maybe more expensive. Maybe GE Multilin can upgrade the relay firmware if not already done to address this issue. we have later version DGP on our steam turbine generator which has not so far experienced this spurious trip even though it had been on-line a couple of times during the back-energisation of GSU1 or GSU2.

 

[dpc]

I think the DGP is an older GE (pre-Multilin) design. I'm not sure of their level of support for this device.

alehman comments on the SR489 are interesting. I've used a lot of these and have not seen this problem. Multilin has told me that the SR489 has a dc offset filter and harmonics filter in software.

I'm not sure why the neutral side phase inputs don't show the same waveform as the line side inputs. This may indicate differences in the two sets of CTs.

 

[e5fornow]

I am not sure this will help but I think you should verify phasing. I do not know if you have looked at your other phases during these trips, but you might find phase C return current on phase B system current.

 

[buzzp]

The 489 has a technical advisory out for such a problem with nuisance trips. Check out their web site for serial numbers of affected devices.

 

[Sackeyfio]

buzzp,
The relay in question is not the 489 but the DGP.

e5fornow,
Phasing has been verified to correct already; all phase angles checked on load OK.

Folks,

The DGP relay manufacturer after looking at the trip current oscillographs indicated that CT matching could could be the problem and the DGP relay is fine.

Subsequently, we checked all the CT name plates and found out that the CTs on the generator terminal side are rated 50Hz and the CTs on the generator neutral side are rated 60Hz. We also plotted the CT magnetisation curve and found out that though the linear portion of the magnetisation curves are the same, they have different knee points and non-linear curves. The 60Hz CT has a higher saturation curve.
The GT1 generator terminal CTs have an averagely higher burden(actual external load) of 0.83ohms as compared to a burden of 0.54ohms for corresponding neutral CT. Likewise the GT2 terminal CTs have an averagely higher burden of 0.88ohms as compared to a burden of 0.59ohms for corresponding neutral CTs. However, the CTs rated burdens of B1.8 are OK for the indivual loads

When the CT manufacturer was contacted, they indicated that the 60Hz CT excitation curve is different from that of the 50Hz CT. And that the core cross-section for the 50Hz CT is 16% less.
For me this mis-match of CT excitation curves could be the major contributor to the problem of tripping on generator differential when an adjacent power transformer is back-energised. This new finding now explains why the 50Hz CT(terminal side) which has a Lower exc. curve, distorts the secondary current waveform whereas the 60HzCT(neutral side)does not comparatively.

We have recommended that all the six 60Hz CTs be put on one generator and the six 50 Hz CTs on another generator to satisfy the CT magnetisation curve matching. Secondly if the trips still occur, the CT external load will have to be matched as close as possible before trying to revise upwards the K-factor for the differential protection.

What do you think about these recommendatiions?
Will keep you posted after implementing them.

 

[dpc]

Did GE indicate if the DGP had internal filtering of dc offset current?

 

[stevenal]

If you are determined to salvage the non-nominal frequency CTs, I would ask the manufacturer to reclassify their accuracy at the new frequency. Then go back to your calculations and see if the reclassified CTs will work in the application. The trial and error approach proposed sounds like it could waste a lot of time.

 

[RRaghunath]

I agree with the findings of GE and yours, Sackeyfio.

You may also like to check the actual value of the stabilising resistor in the differential protection circuit as against the calculations.

This is assuming that the generator differential protection is high impedance type (you indicated a setting of 0.3A and no bias setting). I do not know the relay DGP much.

Once you sort out Ct mismatch issue, I hope you will not have the sort of spurious trips experienced.

 

[dangnm]

dpc,
GE Multilin indicated that the DGP has both hardware and software filtering for dc offset currents.

stevenal,
The CT manufacturer advised that the accuracy class error for the 60Hz CT being used on 50Hz system is negligible.So the CTs will still keep their accuracies.

The only issue now is the matching of the burdens. Using the CT burdens above for both units, there will be a current error of aprrox 13%. Since I am not too sure of the figure,Can you confirm this calculation for me folks. This percentage current error will be too much for the k-factor of 2% in the DGP. So I think the burdens should matched after swapping CT before doing a trip test.

rraghunath,
The DGP is a not a high impedance type relay.

Folks,
What is the best and easiest way of matching the CT burdens? I propose adding lower gauge wire say 1.5sqmm of about 15meters to each phase wire and star-point. Let me know what you think?

 

[alehman]

The burden compensation should take power factor into account. Is the majority of the burden from wiring or relays?

I think this solution is a patch at best. It's important that the CT's on both ends respond identically going into saturation as well as in their linear region. The accuracy class and burdens should match as well. It's normally recommended that CTs be from the same manufacturer and model. You said they have different curves. That would raise concern for me regardless of what they told you about performance on 50Hz vs. 60Hz.