Continue to Site

Eng-Tips is the largest engineering community on the Internet

Intelligent Work Forums for Engineering Professionals

  • Congratulations SDETERS on being selected by the Eng-Tips community for having the most helpful posts in the forums last week. Way to Go!

Distributed Generation

Status
Not open for further replies.

PRSharma

Electrical
Nov 4, 2007
12
Distributed Generation: What should be the preferred interface transformer's winding configuration (utility side Delta, Star solid grd, or Star effectively grd. a) preferred by utility; b) preferred by generating facility. However I am interested in utility's preference particularly when utility's distribution is 3 p 4 w.
I understand that this is highly debateable issue and I am familiar with over-voltage and ground fault coordination issues.

 
Replies continue below

Recommended for you

Thanks Slavag
I understand that it is not simple. However, I am involved in connection impact assessments of such DG. Distribution feeders have re-closers (1 and 3 phase), regulators (1 or 3 phase), fuses, and switches. There is always a trade off between grounded or ungrounded interface transformer winding connection. Unfortunately, most of the North American utilities do not specify their favoured connection in their technical requirements and IEEE 1547 is also silent on critical issues. (IEEE 1547 revision is under way)

Regards,
PRS
 
PRSharma,welcom to the club. not only North American.
On this moment, we work on such application.
2 DG ( sum 16MW) connected to distribution network 22kV (petersen coil system) with 3ph autorecloser. 3ph autorecloser it's big problem ( 300ms recloser time).
We used df/dt function for the decoupling. Additional option Voltage vector protection and PPS protection (popular in several Eroupean countries).
Those DG connected with y/D (22kV side) xfr. I go recommend to customer (plant) add zig-zag with grounding switch. Same problem:grounding fault in the island mode.

We also used directional grounding relay.

Think about 67N, on this way, you can separate internal and external grounding fault.

What is your concept about DG operation? In case of some problem possible off DG w/o problem?

Please send SLD of plant, try think together.

What is a problem with OV? your EQ not for phase to phase voltages?
Regards.
Slava

 
Thanks Jghrist.
From my point of view, it's only small part of subject.
I still say:
1. Y/d connection it's best solution.
2. In case of island mode : close grounding switch.
May be, it's good for our distribution, I don't know.
Why not Delta on the HV side. In case of island mode, we need send transfer signal to all relays:
change setting to 67N for ungrounded system.
Why not Yn connection: It's influence on the ground current
in the network, and request some approvals from utilities.

And I still say: Very important point in such application it's detect " loss of main situation".
Regards.
Slava
 
Thanks jghirst for the link. Truely speaking all articles which I have come across on the subject till date are just of information level and there is no conclusion in any of them. I understand that there is always a trade off for interface transformer winding connection, because both connection types have merits and de-merits. Depending upon the system conditions one has to select the best suitable connection for particular location.

Slavag there is no best connection. However I have serious concerns if HV winding is star and the neutral is solidly grounded. This screws up the ground cordination to an extent that fuse saving schemes becomes garbage and possiblity of unattended ground fault increases. If interface transformer is star then neutral should be grounded through reactor (sometimes people also use resistors). Appropiate sizing of these reactors or resistors is the biggest challange because for higher impeadance though the ground faults would be reduced but the over voltages would exceed the thresholds.
Neither rate of change of frequeny (df/dt) or vector shift is good for anti-island protection if the cummulative generating capacity is greater than 50% of the minimum load of the feeder or the section of feeder down stream of interrupting device and up-stream of DG. Both of the above schemes, and infact all passive local detection schemes have non detection zones. Active local detection schemes(generally employed by invertor based interface)are much better, but they also have a narrow non detection zone and some destorted voltage/currents may remain unchecked.
 
If the utility line is grounded and serving line-neutral(ground) loads and there is any possibility of the utility disconnecting from the circuit during a ground fault before the generator is disconnected, it is imperative that the generator installation be able to source ground current. If the generator installation (generator, transformer(s), etc.) can not source sufficient ground current there will be a 1.73pu voltage on the two unfaulted phases. This overvoltage will cause problems for all phase-neutral loads connected to these phases. If the generator remains connected for any time at all you will also have a number of lightning arrestor failures. Fixing the protection to deal with the additional ground source if vastly easier than cleaning up the mess. Generator owner pays for utility protection upgrades but utility would be liable for damage to customer equipment from overvoltages. Easy decision.
 
Easy technical decision, but practicaly...
Our utilities "kill" me, if I say to them "upgrade your protection". Answer will: "it's your problem".
Rules of our utilities are very simple:
1.Your generator must disconnet first in all cases.
2.Your xfr. don't change any things in our grounding system.
Regards.
Slava
 
I'm on the utility side of the fence now, and there is no way we will upgrade our protection at our cost, the generator owner pays all, but we will gladly make all necessary upgrades. We can't pay for it as that would come out of the pocket of all of our rate payers for the benefit of just one customer. There is no way I would require a generator to be off first, that simply would mean that I have to delay tripping to give him time to get off and I won't do that. If the generator is large enough I will require transfer trip so that any time I open my breaker I trip the generator off the line, doesn't matter why or when I open my breaker. If the generator is less than 33% of the minimum circuit load I won't require transfer trip but will require the generator to get off the line in compliance with IEEE-1547.

If the generator provides a ground source that will cause problems with my ground overcurrent, the generator owner will pay to have my ground overcurrent changed to directional ground overcurrent. No problem, just open your check book. Because we do fast reclose on our distribution feeders (that have reclosing) I will require a VT and voltage relay at the substation on every circuit with synchronous generators connected, no matter how small, and the cost of that becomes a barrier against really small synchronous machines. If somebody wants to spend more on upgrading our substation than on the rest of their project I'd be glad to accept a 50kW synchronous machine, but I doubt I'll see much below 500kW. Nothing outright prohibited, if not behind network protectors, but everything has its price. If your utility won't work with you then it is their loss.
 
Thanks David.
Right desition!
May be with time our utilities understand it!
Our biggest problem, that our utilities are gov. company.
PRSharma, I think David closed all of your problems.
Regards.
Slava
 
Hi Davidbeach
Sorry for the delayed response. YOU have touched the original question. There won't be any delay in tripping utility breaker. Usually the typical tripping time without DG ( across North America) should more or less look like: 4 ms (relay time) + 50 ms (electromechanical aux relay) + 50 ms (3 cycle utility breaker) = 104 ms (tripping time)
Now, when high speed transfer trip (T/T) is applied, and DG CB is tripped before utility CB, the tripping time would look like: 8 ms (relay time) + 20 ms (T/T signal communication time) + 50 ms (3 cycle DG CB) = 78 ms (tripping time). With new numeric relays (GE Multilin F60 or Schweitzer make feeder relay) the need of 50 ms electromechanical aux relay is eliminated. Utility keeps tirpping time same and DG is asked to have fast tripping time. Thus, there is no delay in tripping time. Since beigning I wanted your comment on above tripping sequences.

Next, IEEE 1547 advocates 33% minimum circuit load criteria for T/T is bit conservative, and there is nothing solid to support it. Infact, on basis of simmulations and field tests we are even going upto 50%. Furthermore, IEEE 1547 anti-islanding criteria of 2 sec, is in real sense, restricted to utilities reclosing time. Now when we deal with reclosing times between 500 ms to 1 sec, how DG can remain in complaince with IEEE 1547. How far you guys go to verify whether DG is in complaince with IEEE. My personal belief is that even DG with invertor based interface are not able to comply with IEEE 1547.1 testing requiremnet (which is theoriticaly correct but in practical field is pretty stringent).

Ground fault coordination: Directioning of relay does not solve the protection desensitizing issues. A DG connected close to the station for feeder end faults definitely desentisizes feeder relay.

Three or single VT at every cicuit with DG at station solves the out of step reclosing but realiability is certainly sacrificed whenever it is called to operate. For DG with dedicated feeder voltage supervision before reclosing is OK.

Thus, I do not think that topic can closed at this point and would like to continue sharing comments and practices from everyone.
 
You're right, directional ground won't solve the desensitizing problem but it will solve the problem of the DG circuit tripping for a ground fault on an adjacent feeder. What I outlined above is the essence of what we require. Transfer trip is implemented using Mirrored Bits, we use the 1/3 criteria from IEEE 1547 for deciding when to require transfer trip, although we will, under certain circumstances, allow feeder load monitoring to determine when to allow/block parallel operation. We've got over 40MW of DG on the system and what we require all seems to work fine.
 
Hi.
What is a desensitizing problem?
I heard this first time, may be possible use other term.

PRSharma, next point, what do you mean 4ms, 8ms relay time?

David, what is your practic for the trip/AR sequency/coordination?
For example what we use in utilities:
1. First AR time is 300ms.
2. Second is 75sec.
3. if AR ready, operated overcurrent ( 50 and 50N)with 40ms time delay, but only one time. if AR is not success, we'll block those stages and only IDMT function continue operation.
4. if AR not ready, in operation only IDMT functions.
5. for two first second ( after CB closed position) we add SOTF function ( also blocked in case of AR running situation).
Regards.
Slava
 
Thanks Davidbeach and Slavag to be in loop:

We are currently facing extremely high level of penetration of DG in sub transmission and distribution system (though majority of them are yet at cost estimate stage but around 250 MW is already connected).

Davidbeach: Yes we also use Freewave radios based on mirrored bits where the line of site permits. However, leased line with NSD570 is pretty suitable for T/T (forget the price, proponent pays it any how). Recently we are also trying to use ethernet based MDS radios in combination with GE Multilin make relays to reduce the price. What are your thoughts or practices regarding implementation of T/T for pure induction machines or invertor based interfaces? Do you apply the same 1/3 criteria? What about Transformer's reverse powerflow capabilities or ULTC compatibility. How do you deal with voltage regulators? I have started believing that these DG are not Plug and Play but are more Plug and Pray type applications.

Slavag:
300 ms is indeed very fast AR, because if you have five or eight cycle breaker then there is hardly any dead time for breaker left after arc quenching. I believe you are dealing with urban/city or industrial areas. We have mix of almost every type of loading pattern and used 400 ms to 800 ms AR in transmission system. For rural sectors this fast AR is not worth and usually the AR is around 1 sec to 2 sec. Second and third shot timings are pretty routine. 4 ms and 8 ms are typical primarly relay operating times. The other word widely used for desensitizing is blinding of
protection. Basically it means that, in-spite of increased total faults currents due to conection of DG, the source contribution reduces to an extent that there is not enough current for utility relays to sense untill the DG trips.
 
Hi PRSharma.
Yes, my area is urban/city/industrial, our rural sectors are same.
Thanks, terms "blinding" is clear for me.

Sorry for my previos post " you need some expert", I see you are expert.

4ms and 8ms, total relay operating time I assume is 20-30ms,
4ms is may be function time. But it's not important.

It's very intresting case, we don't have these problems, but only on this time, we are only on begining of IPP on the distribution networks.

What do you think about NPS voltages and current protection?
I think, combination of these functions can solve problem of desensitizing.

Xfr's reverse powerflow, actually it's not electrical problem, in any case possible add reverse power relay or may be underpower relay for Xfr protection. In our co-generation application we use both of them.

Voltage regulator, I think, it's depend on size of generator and distance between generator and utilities source.

Regards.
Slava
 
Slavag
Thanks for the complement. I am learning DGs and still there is long way to go.
Traditional O/C;E/F protection are proven practices for typical radial distribution system, but are operational only when there is enough current to exceed the settings. I am not aware of Negative Phase Sequence protection being used for protection of distribution or transmission lines, however, it is common feature in generator protection. Though I am working on NPS for local detection based protection for anti-islanding in DGs, but due to lot of associated uncertainities and parameters involved, it is very difficult to differentiate between disturbances or fault conditions.

X'mer reverse power flow is understandable not to be a problem since, X'mer being a static device except ULTC. But just think of this senario, when cummulative generation from DG substantially increases than load, the active power through X'mer reverses but the reactive power is still supplied by X'mer. Which means that through X'mer, active power flows in reverse direction and reactive power folws in forward direction. The x'mer is operated at very low power factors.

Votage regulators have different issue. Think of a regulator located upstream of DGs, due to cummulative generation the voltage at the load side of regulator is high so the regulator taps are either at neutral or lowest position. Suddenly the due to some reason (gusty wind drawing wind turbines to cut speed)DG ceases to generate, then the voltage will drop, now voltage would be regulated by regulator but not before inital 25 to 30 seconds delay and 2 to 3 seconds for each subsequent tap change operation. Thus for this period the other customers may be subjected to voltages below standard limits.
 
Hello PRSharma.
Senario with forward direction of active power and reverse direction of reactive power is common practic for big generator. It's mode "absorbe of reactive power". But of course GSU xfr w/o OLTC.
Voltage regulators issue, it's "issue".
Actually, I see this as, utilities issue ( who pay for this it's other thing). It must be central regulation of reactive power flow with option of generators AVR regulation directly from utilities center.
Next way, in this cases, install several capcitor banks in the critical points with auto/manual control.
Regards.
Slava
 
Status
Not open for further replies.

Part and Inventory Search

Sponsor