Jump to content
Engineer Boards
Sign in to follow this  
Omer

differential protection on 3 phas TXR

Recommended Posts

Omer    0

This post is to discuss about the differential protection on 3 phases transformer. specifically, star delta connection. I think it is a rich topic in the concepts and might well appear on the PE

How to connect CTs, 30 degrees phase shift, instantaneous currents, zero sequence currents, triple harmonic currents, etc..

you are all welcome to give your thoughts for the benefit of all.

the below figure is for your convenience

Capture.GIF

Share this post


Link to post
Share on other sites
TNPE    15

To compensate for the 30 degree phase shift, the CTs on either side should be connected in the opposite configuration of that side of the XFMR.  For example, a D-Y XFMR would be CT'd in Y-D for a differential protection scheme.  However, relays exist today that can compensate and do the work for you, regardless of winding configuration of XFMR or CTs.  

Share this post


Link to post
Share on other sites
rg1    4

If we ignore digital age relaying systems which can do anything and everything you need and go back to analog relays then

1. The CTs on Delta side should be star connected and star side should be delta connected as a rule rather than convention. This is helpful  not only for differential but in other protections too.

Its benefits are

a. Eliminates the game of 30 degree in star delta. (two wrongs makes it right).

b. Zero sequence and triplen harmonic (triplens behave similar to zero sequence) currents of star connected power system gets a circulating path in delta connected CTs; so these villains are prevented to go to relay side and make some mess there.

c. Take care of selection of CT ratios when you are connecting in star on one side and delta on other . If say 1:1 Xmer is deferential protected  with star connected CTs of 1:1 ratio then delta connected CTs should be 1:1/sqrt3. This sqrt3 will make corrections to 1:1 in a delta connection. ( So many wrongs making corrections?)

d. The differential protection is not absolute i.e the tripping is not based on absolute value of differential current. It is biased differential current protection. so the setting is say 10, 20, 30% biased on through current. Meaning is- if through current is 100A the differential current should be 30 A (at 30%) for the relay to give a trip command and  at 1000A through current it will need 300A. This is done to eliminate the ratio error or mismatch in the ratio of CTs. This is achieved by restraining coil (R in the diagram) in the relay which sees through current, so the setting of relay is Diff current/ through current or say Id/((I1+I2)/2). O in the diagram is operating coil which sees only differential current (I1-I2). 

e. The magnetizing current and so is inrush magnetising current contain second harmonic currents. The magnisisng current is only present in primary (feeder) side of the Xmer and so they form part of the differences of the currents and may act to trip the relay. Tripping protection from these demons is provided by providing a second harmonic restraint feature into the relay.

f. The little mismatch in CT ratios is permitted in the Xmer differential protection but not in diff protection of Motors, Bus bars, Generators etc. Why- I do not remember now? 

I welcome comments on the items I mentioned. 

Share this post


Link to post
Share on other sites
TNPE    15
23 hours ago, rg1 said:

If we ignore digital age relaying systems which can do anything and everything you need and go back to analog relays then

1. The CTs on Delta side should be star connected and star side should be delta connected as a rule rather than convention. This is helpful  not only for differential but in other protections too.

Its benefits are

a. Eliminates the game of 30 degree in star delta. (two wrongs makes it right).

b. Zero sequence and triplen harmonic (triplens behave similar to zero sequence) currents of star connected power system gets a circulating path in delta connected CTs; so these villains are prevented to go to relay side and make some mess there.

c. Take care of selection of CT ratios when you are connecting in star on one side and delta on other . If say 1:1 Xmer is deferential protected  with star connected CTs of 1:1 ratio then delta connected CTs should be 1:1/sqrt3. This sqrt3 will make corrections to 1:1 in a delta connection. ( So many wrongs making corrections?)

d. The differential protection is not absolute i.e the tripping is not based on absolute value of differential current. It is biased differential current protection. so the setting is say 10, 20, 30% biased on through current. Meaning is- if through current is 100A the differential current should be 30 A (at 30%) for the relay to give a trip command and  at 1000A through current it will need 300A. This is done to eliminate the ratio error or mismatch in the ratio of CTs. This is achieved by restraining coil (R in the diagram) in the relay which sees through current, so the setting of relay is Diff current/ through current or say Id/((I1+I2)/2). O in the diagram is operating coil which sees only differential current (I1-I2). 

e. The magnetizing current and so is inrush magnetising current contain second harmonic currents. The magnisisng current is only present in primary (feeder) side of the Xmer and so they form part of the differences of the currents and may act to trip the relay. Tripping protection from these demons is provided by providing a second harmonic restraint feature into the relay.

f. The little mismatch in CT ratios is permitted in the Xmer differential protection but not in diff protection of Motors, Bus bars, Generators etc. Why- I do not remember now? 

I welcome comments on the items I mentioned. 

To elaborate and expand further on e). above, most all relays should have a differential bypass to address your concerns with inrush current.  Generally speaking, when a XFMR is being placed online, it endures an extensive array of tests to validate it's integrity for safe operation (at least it should.. if not, you need to question the engineer overseeing this).  Hence, it is practically 100% safe, at this point, to disable the differential protection for energization.  Once energized, enable the appropriate protection schemes.  I have been around situations where you may isolate a section with limited protection, maybe backing up on the grid to the transmission/generator sections, but that's just the way it is in some scenarios.  Safe practices and measures can be put in place to mitigate this, though (i.e. may still be utilizing over-current protection, one-shot enabled, ground trip block, etc.).  I'm aware you don't work in the electric utility industry, but you want to eliminate every blink possible.  This doesn't mean you put personnel at risk, just that work around these scenarios in the best, safest way possible.

Share this post


Link to post
Share on other sites
TNPE    15

A differential scheme is looking for 0 amps (minus whatever tolerance levels may be in place).  Essentially, the burden is comparing input and output current based on the turns ratio (i.e. it should register ~0 amps, if not, it is recognizing an internal fault and opens up).  I may be misunderstanding you, but there should not be a tolerance as high as 300 amps, with a 1000 amp system, when utilizing a differential scheme.  Depending on XFMR design and voltage ratings, very, very low voltages could be present in an internal fault, where insulation breakdown can (and often) occur in adjacent windings (i.e.   consecutive windings on primary or secondary).

Share this post


Link to post
Share on other sites
rg1    4
1 hour ago, TNPE said:

A differential scheme is looking for 0 amps (minus whatever tolerance levels may be in place).  Essentially, the burden is comparing input and output current based on the turns ratio (i.e. it should register ~0 amps, if not, it is recognizing an internal fault and opens up).  I may be misunderstanding you, but there should not be a tolerance as high as 300 amps, with a 1000 amp system, when utilizing a differential scheme.  Depending on XFMR design and voltage ratings, very, very low voltages could be present in an internal fault, where insulation breakdown can (and often) occur in adjacent windings (i.e.   consecutive windings on primary or secondary).

I appreciate you pointed this out. This was an example to take the reader to maths;  to make him understand the percent bias;  not the exact  values,  but yes percent bias is there in the settings of a differential protection. In case there is a mismatch of CTs it is used. I have  myself decided upto 15% setting at some places.  Exact pick up like over current (for difference of currents)  will  give you spurious tripping. This is supported by the fact that there is R coil- restrain coil sensing through current. Imagine a mismatch of 10% in CTs and then do the maths, you are going get it.

 

Share this post


Link to post
Share on other sites
rg1    4
2 hours ago, TNPE said:

To elaborate and expand further on e). above, most all relays should have a differential bypass to address your concerns with inrush current.  Generally speaking, when a XFMR is being placed online, it endures an extensive array of tests to validate it's integrity for safe operation (at least it should.. if not, you need to question the engineer overseeing this).  Hence, it is practically 100% safe, at this point, to disable the differential protection for energization.  Once energized, enable the appropriate protection schemes.  I have been around situations where you may isolate a section with limited protection, maybe backing up on the grid to the transmission/generator sections, but that's just the way it is in some scenarios.  Safe practices and measures can be put in place to mitigate this, though (i.e. may still be utilizing over-current protection, one-shot enabled, ground trip block, etc.).  I'm aware you don't work in the electric utility industry, but you want to eliminate every blink possible.  This doesn't mean you put personnel at risk, just that work around these scenarios in the best, safest way possible.

You are right, this feature of second harmonic restrain can be achieved by many means, time delay (disabling while inrush) is one of them, the other is kick fuse ( i do not know how it functions, must be bypass as you mention), putting a filter for second harmonics and then using it for restraining purpose is also used. But yes, one thing is true that we have to tackle second harmonics.

Share this post


Link to post
Share on other sites
rg1    4
53 minutes ago, rg1 said:

I appreciate you pointed this out. This was an example to take the reader to maths;  to make him understand the percent bias;  not the exact  values,  but yes percent bias is there in the settings of a differential protection. In case there is a mismatch of CTs it is used. I have  myself decided upto 15% setting at some places.  Exact pick up like over current (for difference of currents)  will  give you spurious tripping. This is supported by the fact that there is R coil- restrain coil sensing through current. Imagine a mismatch of 10% in CTs and then do the maths, you are going get it.

 

If you have tailor made Xmers and CTs, percent Bias may not be of much use but, practical situations warrant this. Secondly this also eliminates mismatch due to unequal saturation conditions of CTs, difference in their ages may also cause this mismatch. ( Say I had to replace one of the CTs due to failure and this latest addition may be 20 years younger than the older lot,  whose B-H characteristics might have drifted a little. Now you will require bias for this purpose. I have not come across a text book or manufacturers manual which describes diff protection without percent bias. I will request you share, if you have. 

Share this post


Link to post
Share on other sites
TNPE    15
16 minutes ago, rg1 said:

You are right, this feature of second harmonic restrain can be achieved by many means, time delay (disabling while inrush) is one of them, the other is kick fuse ( i do not know how it functions, must be bypass as you mention), putting a filter for second harmonics and then using it for restraining purpose is also used. But yes, one thing is true that we have to tackle second harmonics.

It is usually in the form of a toggle switch on the panel in the relay cabinet. You're essentially removing the differential scheme by breaking the differential protection circuit.  As for your other post, I don't readily have any documentation that says you should design your scheme with "x" tolerance.  But let's be honest, XFMRs (CTs included) are well-designed from all applicable engineering standpoints, and should accomplish and fill the needs for metering, relaying, protection, etc. with precision.  I say this with regards to an engineer designing a suitable scheme/configuration.  But yes, it is easily doable if done properly.  Personally, I'm all about precision.  Blame it on the engineer in me. I much rather prefer a well-designed (with all considerations given to CT ratios, saturation, wiring configurations, available fault currents, types of faults available, etc.) scheme as opposed to giving excessive tolerances to avoid unwanted trips/operations.  Especially when we're talking about a XFMR, it is the lifeblood of a power system, as well as the most expensive (minus generation), so it is of utmost importance that it is protected properly.

Share this post


Link to post
Share on other sites
TNPE    15

Also, differential schemes are EXTREMELY fast!!!!  Much faster than a generic over-current scheme.  But it should be.  We're talking about protecting a piece of equipment that could be valued in a range from ~$500K to several million.  Differential schemes are akin to a hot line tag for over-current devices.  Operation can occur within the first half cycle to 3 cycles.

Share this post


Link to post
Share on other sites
cos90    3

@rg1 you have done a good job explaining the details of transformer differential protection

@TNPE I am curious what your plan is for the second energization of your transformers if you don't enable differential protection during the first energization?? especially since the next time you energize will be in less than ideal conditions.

my comments: 

every source I can find for power says do not consider "even" harmonics but there is a second harmonic prominent during transformer energization. why does everyone say it then?

removing the zero sequence components with delta connected cts is a negative. if you have a ground bank inside of the transformer cts on the delta side and there  is an external ground fault you will have a differential trip. im not smart enough to realize this on my own but instead just read it in the ge t60 manual.

 

Share this post


Link to post
Share on other sites
cos90    3

oh i forgot to mention a major source of CT error as mentioned by rg1: variation from the load tap changer!! 

also if you are not tnpe or rg1 this discussion is way out of scope for the pe exam

Share this post


Link to post
Share on other sites
TNPE    15

@cos90  

Yes, you're correct in saying the level of this conversation is safely out of range for what the PE will test for.  

Second energization?  Hopefully you don't lose your XFMR often.  I'm speaking to initial energization.  That said, should you lose a XFMR, and you have no indication of internal damage, I would disable differential protection and let the OC carry it, just as you would during initial energization. But be aware of XFMR behavior and noises.  If it starts emulating a washing machine, kill it.  Without an advanced scheme or alternate profile, energization could be extremely difficult, even impossible (the inrush would trigger the differential and prevent energization).  On the other side, if you suspect internal damage or have shrapnel and oil spewed everywhere, it's safe to say energizing ain't happening soon.  At this point, I hope you have means to backfeed or access to a spare XFMR or a mobile sub.

Share this post


Link to post
Share on other sites
rg1    4
2 hours ago, cos90 said:

@rg1 you have done a good job explaining the details of transformer differential protection Thanks

@TNPE I am curious what your plan is for the second energization of your transformers if you don't enable differential protection during the first energization?? especially since the next time you energize will be in less than ideal conditions. I think the disabling of the diff protection is only for energisation ( everytime you energise you do it). This is one of the technique to bypass the effects of second harmonics and inrush diff current(Inrush is only in feeding end)

my comments: 

every source I can find for power says do not consider "even" harmonics but there is a second harmonic prominent during transformer energization. why does everyone say it then? Good thought. Why I missed it? 

removing the zero sequence components with delta connected cts is a negative. This is not removing of zero sequence from Power ckt. This removal is only in relay ckt. if you have a ground bank inside of the transformer cts on the delta side and there  is an external ground fault you will have a differential trip.  Can you post the exact language or let me know page no. I would like to see that. .  im not smart enough to realize this on my own but instead just read it in the ge t60 manual.

 

 

Share this post


Link to post
Share on other sites
cos90    3
9 minutes ago, rg1 said:

Can you post the exact language or let me know page no. I would like to see that.

DownloadFile.aspx?prod=t60&type=3&file=1

page 372 

5.5.4.6 Phase and zero sequence compensation 

Share this post


Link to post
Share on other sites
cos90    3
11 minutes ago, TNPE said:

@cos90  

Yes, you're correct in saying the level of this conversation is safely out of range for what the PE will test for.  

Second energization?  Hopefully you don't lose your XFMR often.  I'm speaking to initial energization.  That said, should you lose a XFMR, and you have no indication of internal damage, I would disable differential protection and let the OC carry it, just as you would during initial energization. But be aware of XFMR behavior and noises.  If it starts emulating a washing machine, kill it.  Without an advanced scheme or alternate profile, energization could be extremely difficult, even impossible (the inrush would trigger the differential and prevent energization).  On the other side, if you suspect internal damage or have shrapnel and oil spewed everywhere, it's safe to say energizing ain't happening soon.  At this point, I hope you have means to backfeed or access to a spare XFMR or a mobile sub.

what if you lose feed from the utility and they close back in before you have someone at the station to turn the differential off? lockout differential on inrush ? genuinely curious 

 

Share this post


Link to post
Share on other sites
rg1    4
8 minutes ago, cos90 said:

what if you lose feed from the utility and they close back in before you have someone at the station to turn the differential off? lockout differential on inrush ? genuinely curious 

 

I had this question in my mind. In my working I have never seen this feature. I accept it as a theory only. And I do not know when technology provides you all time solution why you have to have a manual mode for bypassing purpose. ( I remember this inherent restrain feature in analogue English electric relays long ago)

Share this post


Link to post
Share on other sites
TNPE    15

That's a good question.  If you did lose feed from the G&T, or whoever your up line supplier is, it's customary that they be in contact with their downline utilities to coordinate events such as this and for safety measures (i.e. all clear).  With today's technology, you should also be able to handle this remotely via SCADA.  

Share this post


Link to post
Share on other sites
rg1    4
23 hours ago, cos90 said:

DownloadFile.aspx?prod=t60&type=3&file=1

page 372 

5.5.4.6 Phase and zero sequence compensation 

I saw this. could not find anything like you mentioned. I think the theory of diff protection, I have covered the whole thing in my earlier post. We can achieve the same thing by many means. The manual t60 details how it is achieved in their relay. These new relays are very flexible in things like dynamically changing the relay characteristics etc. I had a chance to install and test 8 such relays of another make in last 2/3 years. So I have used all three types of relays- Analogue ( those old ones, disc types), then came electronic and now microprocessor based type. These are really wonderful. We did not have such big manuals for analogues and electronic types.  If one can pay attention the theory of GFCI, Differential and Restricted Earth Fault protections, is all same. That is---If incoming power/Current is equal to outgoing power/current the equipment is healthy, if not, there is some leakage within the equipment. 

Share this post


Link to post
Share on other sites
cos90    3
20 hours ago, rg1 said:

I saw this. could not find anything like you mentioned. I think the theory of diff protection, I have covered the whole thing in my earlier post. We can achieve the same thing by many means. The manual t60 details how it is achieved in their relay. These new relays are very flexible in things like dynamically changing the relay characteristics etc. I had a chance to install and test 8 such relays of another make in last 2/3 years. So I have used all three types of relays- Analogue ( those old ones, disc types), then came electronic and now microprocessor based type. These are really wonderful. We did not have such big manuals for analogues and electronic types.  If one can pay attention the theory of GFCI, Differential and Restricted Earth Fault protections, is all same. That is---If incoming power/Current is equal to outgoing power/current the equipment is healthy, if not, there is some leakage within the equipment. 

Here is the text:

In our example, the transformer has the Δ-Y connection. Traditionally, CTs on the Wye connected transformer winding (winding 2) are connected in a delta arrangement, which compensates for the phase angle lag introduced in the Delta connected winding (winding 1), so that line currents from both windings can be compared at the relay. The Delta connection of CTs, however, inherently has the effect of removing the zero sequence components of the phase currents. If there is a grounding bank on the Delta winding of the power transformer within the zone of protection, a ground fault results in differential (zero sequence) current and false trips. In such a case, it is necessary to insert a zero sequence current trap with the Wye connected CTs on the Delta winding of the transformer.

This is a very special case. If there weren't so many special cases we wouldn't need engineers. :)

Share this post


Link to post
Share on other sites
rg1    4
On 8/15/2017 at 1:09 PM, cos90 said:

Here is the text:

In our example, the transformer has the Δ-Y connection. Traditionally, CTs on the Wye connected transformer winding (winding 2) are connected in a delta arrangement, which compensates for the phase angle lag introduced in the Delta connected winding (winding 1), so that line currents from both windings can be compared at the relay. The Delta connection of CTs, however, inherently has the effect of removing the zero sequence components of the phase currents. If there is a grounding bank on the Delta winding of the power transformer within the zone of protection, (IMHO this situation can be imagined same as star connected wdg with ground, on main Delta wdg side too- So the situation becomes like this - Both sides, primary and secondary of the Xmer are star with grounded neutral- and the CTs on one side are star connected (because main wdg is delta) and other side it is delta connected. In such a case the zero sequence currents on star connected CTs will flow through the Operating coil of the relay.  In this case you will have diff relay tripping on through L-G faults because they give rise to zero sequence currents in the star connected CTs. To avoid this the traditional method is to connect both side CTs in delta connection, attached is the diagram for connection. This is called marketing strategy of T60,  nothing else- as if till digital relays were not there, we were without electricity) a ground fault results in differential (zero sequence) current and false trips. In such a case, it is necessary to insert a zero sequence current trap with the Wye connected CTs on the Delta winding of the transformer.

This is a very special case. If there weren't so many special cases we wouldn't need engineers. :)

I hope the explanation makes sense. I have never dealt with grounding winding of a transformer because I have never worked in a ungrounded system, I mean it is really funny to have undergrounded system and then ground it by a separate wdg. However the concept is no rocket science, I must have understood it rightly. Discussion welcome. 

Capture17.JPG

Share this post


Link to post
Share on other sites
rg1    4
17 hours ago, rg1 said:

I hope the explanation makes sense. I have never dealt with grounding winding of a transformer because I have never worked in a ungrounded system, I mean it is really funny to have undergrounded system and then ground it by a separate wdg. However the concept is no rocket science, I must have understood it rightly. Discussion welcome. 

Capture17.JPG

When you have delta winding and a star connected neutral grounded winding on same side of transformer, you have two parallel paths for Zero sequence currents, one is to circulate with in delta and the other in the neutral. How much will go where will depend on the  resistances of the paths. I have taken everything going  into neutral (extreme case) just for better appreciation.

Share this post


Link to post
Share on other sites
TNPE    15

Per the original drawing, no zero sequence current flows to the relay coils.  Zero sequence circulates in delta CTs and no zero sequence present on wye CTs due to the absence of a neutral.  X2 bushings are bonded together and floating.  Am I missing something you're trying to say?

The primary of the XFMR winding is wye grounded, thus zero sequence currents will flow between that point and it's upline source.  This is a rather odd connection.  I'm used to the whole configuration being flipped.

Share this post


Link to post
Share on other sites
rg1    4
3 hours ago, TNPE said:

Per the original drawing, no zero sequence current flows to the relay coils.  Zero sequence circulates in delta CTs and no zero sequence present on wye CTs due to the absence of a neutral.  X2 bushings are bonded together and floating.  Am I missing something you're trying to say?

The primary of the XFMR winding is wye grounded, thus zero sequence currents will flow between that point and it's upline source.  This is a rather odd connection.  I'm used to the whole configuration being flipped.

The discussion started with Cos90 brought  a scenario from t60 manual which states like this

In our example, the transformer has the Δ-Y connection. Traditionally, CTs on the Wye connected transformer winding (winding 2) are connected in a delta arrangement, which compensates for the phase angle lag introduced in the Delta connected winding (winding 1), so that line currents from both windings can be compared at the relay. The Delta connection of CTs, however, inherently has the effect of removing the zero sequence components of the phase currents. If there is a grounding bank on the Delta winding of the power transformer within the zone of protection, a ground fault results in differential (zero sequence) current and false trips. In such a case, it is necessary to insert a zero sequence current trap with the Wye connected CTs on the Delta winding of the transformer." 

So this is a situation in which there are three windings on Xmer, Delta, Star, and one more star on delta side to provide ground on delta side. The relay manufacturer (t60, page 372) while pointing out to better characteristics of his relay states that there will be spurious tripping in case of the conventional methods of relaying. I tried to analyse  why there be spurious tripping in the event of through L-G   fault and how it could be avoided even in conventional methods. Now if the problem statement is clear you can go through my explanation whether it is true.

Share this post


Link to post
Share on other sites
TNPE    15

So, this sounds like a zigzag has been installed near the delta so that a ground reference can be made for the secondary side.  In this case, yes, I can see where you would have to protect in a seemingly unconventional manner to offset the introduction of zero sequence from the zigzag.

That said, can anyone explain to me where a XFMR arrangement like the one presented in the OP would be used, or even practical, in a utility environment?  Personally, from my experience and the applications I'm accustomed to, I would never use that arrangement at the sub-transmission or distribution levels.

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now

Sign in to follow this  

×