Switch from ground overcurrent to ground distance 1

[HamburgerHelper]

For transmission, has anyone completely replaced their ground overcurrent schemes with ground distance protection. How difficult was it? Were there extra studies needed? What scenarios were looked at to validate the settings so they didn't underreach or overreach due to mutual coupling or infeed? I would like to if possible replace our ground overcurrent elements with ground distance due to our models never being 100% correct and not being forced to recheck coordination.

 

[cranky108]

Some people have made the switch, and had problems.

Why aren't your models 100% correct?

Why do you think ground distance is the solution of having bad models?

Actually using ground overcurrent is more of a solution to having bad models, and rechecking your coordination is becoming a requirement in some parts of the world.
If your models are incorrect have you looked at line differentials?

 

[HamburgerHelper]

The models have issues that protection is being asked review and catch. Sometimes things like transformers are not modeled with the correct connection codes and it doesn't get caught right away.

Ground distance has less of an issue with source changes. If someone screws up a transformer model, you have more ground current and maybe your ground instant. now overreaches. With ground distance, I can tolerate more wrongness in the rest of the system model before a misoperation happens.

On a project while I was in planning, I compared real word data to our power flow models. Everything was surprisingly within 10% but that only validates positive and negative sequence information. The zero sequence and mutual coupling data never gets compared to the real world. Planning won't catch transformer mismodeling looking at their power flows.

 

[Mbrooke]

This is your friend

https://www.omicronenergy.com/en/products/grounding-system/ground-impedance-test/

https://www.youtube.com/watch?v=nsn-Tw_wzyg

As for switching to ground distance I think its necessary- at least from my limited understanding of transmission line protection. In fact when I first learned utilities were using ground over current for transmission lines I was surprised and to be honest I still don't understand why it was ever done. My math says its just not possible.


FWIW around here in the past 30 years everything outside of line differential has used ground distance.


*EDITED out error in choosing correct terminology to describe what I had in mind*

 

[Mbrooke]

Also, if you need a pointer regarding mutual coupling. I am certainly not say it will work for you, but more of a pointer or general guideline to consider.

http://www.eng-tips.com/viewthread.cfm?qid=424649

 

[HamburgerHelper]

Mbrooke,

I think you stated ground distance instead of ground overcurrent in your post.

 

[Mbrooke]

Oh! Yes, yes I did! My mistake, let me correct that. I meant ground over current but got tripped up...


With that said I can understand the use of ground distance, but ground over current not so much. Maybe your system has much fewer generation dispatch variants.

 

[davidbeach]

My take - it is much easier to contain the misoperations with ground distance than it is with directional ground overcurrent.

The immunity to source impedance changes is huge. To set an underreaching instantaneous ground element using ground overcurrent I have to know every zero sequence impedance in the surrounding network; any of them, all of them, can impact the magnitude of the zero sequence current. The instantaneous ground distance element is only impacted by the errors in calculating the zero sequence impedance of that line. With ground overcurrent it takes just one goofus and it's all over; the most common errors in zero sequence modeling tend to overstate the actual zero sequence impedance resulting in a ground instantaneous element that reaches, real life, much further than the model predicts. Ground overcurrent, with its strong dependence on source impedance is also more adversely affected when infeeds change.

We're certainly not to the point of complete elimination of ground overcurrent, but are well on our way. All of the ground over-trips that come to mind since we started converting from ground overcurrent to ground distance have been ground overcurrents, not ground distance elements. It's true that all of the ground distance settings are newer than almost any of the ground overcurrent settings and thereby benefit from other improvements in our processes; but if the susceptibility to misoperation were comparable we'd have had a bunch of ground distance misops that just haven't happened.

Eagerly awaiting the day we eliminate ground overcurrent from the transmission system.

 

[Mbrooke]

The immunity to source impedance changes is huge.

Oh yes, 100% agreed! Source impedance variants are a given being impossible to avoid.

To set an underreaching instantaneous ground element using ground overcurrent I have to know every zero sequence impedance in the surrounding network; any of them, all of them, can impact the magnitude of the zero sequence current.

And even if you did know every single zero sequence impedance value for a network (line, cable, bus, transformer, ect) it will change the most when protective relaying, coordination and operational security is needed the most. Selective coordination does not stop after N-1-1 or N-2. An example of that would be hurricanes, ice storms and repeated lightning strikes. There are storms in history that have put overall networks in N-6 (and much higher) with multiple generation outages mixed in. Granted most networks would need load shedding if this occurred during peak periods; but since most storms that cause numerous transmission outages also cause numerous over head distribution outages its not uncommon for loading on the transmission system to end up being low enough to allow successful operation of metropolitan areas with underground distribution networks unaffected by wind and lightning. In such a case the last thing you would want is miss operation after an N-2 contingency when thermal, voltage and transient stability limits are still very well within acceptable values. To me ground over current should only be a last resort backup for when differential (if present) and Quad/MHO elements fail.

 

[davidbeach]

Differential may fail due to loss of comms, but if your Quad/Mho elements fail it means the relay has failed and the ground overcurrent won't do any better. I've never found a hole in the element performance; it could still happen, but at this point if the relay specs say the relay will detect some given set of voltages and/or currents I've not found any counter examples (allowing for the specified tolerances).

A while back we had a heck of storm and at N-2 (N-1-1 in planner speak) we were getting ground overcurrent over trips, but no problems with ground distance. In fairness, the worst affected areas were also the heaviest concentrations of ground overcurrent; I doubt there was any correlation there though. Never the less, I know that one of them would have been prevented with ground distance; no idea one way or the other for the others yet.

 

[Mbrooke]

Differential may fail due to loss of comms, but if your Quad/Mho elements fail it means the relay has failed and the ground overcurrent won't do any better. I've never found a hole in the element performance; it could still happen, but at this point if the relay specs say the relay will detect some given set of voltages and/or currents I've not found any counter examples (allowing for the specified tolerances).

I understand where you are coming from, but if you loose your 3 phase voltage inputs to the relay you also loose distance and directional protection. In that case the relay reverts to simple none directional over current (instantaneous and time delay). Of course I think I may have miss spoken, in that if the ground over current uses directional elements, then you loose ground over current in the directional sense when the VT circuit opens. My error- but again- anything to me that involves any type of over current in transmission lines I think of as a last ditch effort.

A while back we had a heck of storm and at N-2 (N-1-1 in planner speak) we were getting ground overcurrent over trips, but no problems with ground distance. In fairness, the worst affected areas were also the heaviest concentrations of ground overcurrent; I doubt there was any correlation there though. Never the less, I know that one of them would have been prevented with ground distance; no idea one way or the other for the others yet.

The truth is that most systems outside of peak load can handle far more than N-2. And even if you exceed N-2 under peak load, load shedding will (should) prepare the system for handling more contingencies beyond N-2. By ignoring selectivity beyond N-2, you are basically saying any negative resulting impact- up to complete voltage collapse- from any two elements being out is ok. Yes you have reclosing, but if an N-3 turns into an N-5 simply from miss-operation, the phase angles can become great enough to disable or prevent successful automatic reclosing.

Not trying to bash the OP here, just my opinion based on the protection philosophy I hold.

 

[davidbeach]

When a relay is in an LOP condition we do turn on non-directional overcurrents. At that point over tripping is a risk to be taken simply make sure faults get clear. LOP also sends an alarm and I don't remember seeing those outside of testing.

At N-1 or N-2 there's no problem carrying load, and the distance elements continue to behave generally as expected. But at N-2 there's not much hope for many of the overcurrents. If you set them based on two sources out they'll way over reach during system normal and if you set them for two downstream in-feeds out they'll way under reach during system normal. We should get good results for the vast majority of N-1 cases but N-2 calls for too many compromises, particularly for overcurrents. Distance elements, phase or ground, are much more tolerant of N-2 and we don't see N-2 very often under planned conditions.

 

[Mbrooke]

X2, well said and I agree.

 

[HamburgerHelper]

Davidbeach,

How do you go about setting the K-value on the ground distance relay? I am kind of confused as to how to set that. Is it better than just pulling back the reach and leaving K alone and giving yourself more margin? What methodology do you use to determine how to set that value? I can play with the K value in ASPEN to adjust the reach but I am unsure of what I am getting and losing by changing it. I know I am making the relay more or less sensitive to zero sequence current but if someone asked "well, why not just set K=-1 and marginalize I0 in the impedance equation?" I don't know what I would say. If you set it too low or negative, do you have to worry about it picking up on phase faults?

This is kind of separate but when you check coordination, do you assume the fault will have a certain amount of impedance? We don't assume any fault impedance in our studies and I am not sure that is best. It mitigates the effects of one side being a very strong source has on the weak side. What is a realistic value for transmission level fault impedance?

 

[stevenal]

ASPEN calculates the k value for you. Hit the line impedance button in the lower left corner of your distance relay box, and provides the formula as well. Of course you need to have the line modeled correctly.

 

[HamburgerHelper]

Stevenal,

I believe this is ASPEN just calculating K = (Zo-Z1)/(3Z1). That doesn't help me in determining what I should be setting K to mitigate things like mutual coupling.

 

[davidbeach]

The k value that ASPEN will calculate, based on the entered line impedance will almost always be good enough. You don't ever have to do anything else, and for many years I never particularly considered using any other value.

What I've found, though, is that if you're working with lines that have particularly difficult conditions due to mutual coupling that calculating the k values you can get settings that are more sensible. On a line with no mutual coupling, and k set based on the line impedance, your ground and phase reaches are numerically identical - the whole point of k. Where mutual coupling is present you can't do that; either the reaches are numerically different or the coverage of the line is different. I like to have both zone 1 reaches the same. So, you can calculate the k0M1 and k0A1 values such that the reaches match at some point on the line, typically 85% in my cases. Then set k0M and k0A based on the line impedance or some other specific fault case; the remote bus for instance. This means that zone 1 and zone 2 (and higher) see different apparent impedances for the same fault.

I like the reach selection process using the fault based k values but you can trial and error your way into perfectly good reach settings using an impedance based value for k. Given the nature of the situations where a fault based k value is appropriate, you still need to check a bunch of cases to make sure that there are not any reach anomalies. My favorite reach anomaly is the pair of lines where a ground fault at the second bus out produces a lower apparent impedance than a ground fault at the first bus. Those are set with an impedance base k value. Next time they need to be touched I'm going to try the fault based selection of different k values and see if I can make the ground reaches much more like the phase reaches.

 

[cranky108]

If you don't look at fault impedance in ground distance relays, then you may not operate for impedance faults, like tree limbs, dry soil, etc.
That's why ground over current works so well, is you normally don't have to put much into thinking about ground impedance at the point of contact.

Fault studies with no fault impedance it good for a maximum value, but fault impedance is real and must be looked at, other wise you have a conductor on the ground, and no breakers operating.
That can happen anyway, but you want to reduce that as much as you can.

I have seen a live fault test, and a 7 kV feeder dropped in a sand box will not operate a 15T fuse, but will still make glass.

 

[davidbeach]

Fault resistance is a piece of cake to deal with; you run a fault case with the resistance your setting criteria call for and set the relay based on that. We expect to clear a 50 ohm fault on any line; but we do allow it clear sequentially if simultaneous clearing results in excessively long reaches. Ground quads are great for adding resistive coverage without extending the reactance reach.

 

[HamburgerHelper]

50 ohm seems like a lot for a fault? What is the normal range for fault impedance? The only thing that I have come across that might give me some idea is the arc impedance in ohms being close to - Z = 440*feet/amps.