Generator Grounding

[joan271273]

I am working at a data center project were we have 4-2MW (480V) gen sets tied to a paralelling gear, these are currently solidly grounded and will be energized in a couple of weeks. Now our customer is stating that they are thinking on 1) grounding these generators via reactors or 2) making the system grounded at the Unit Substations via the reactors. My concern is that lead times for these reactors will basically put the project on the following situation:

a) The data center will become active with IT equipment.
b) We would have to do the grouding changes later with the project on line.

I would like some opinions, will the implementation of this grounding system with a facility that is operational be a source of disruptions. Were should the system be grounded at gen sets or Unit S/S.

Thanks for any help.

 

[EEJaime]

I am assuming you are writing about grounding the neutral of a 480Y/277v system for each of the generators. And do you mean neutral grounding RESISTORS?

There are many factors which need to be considered when determining which approach to take, but if your load profile allows, you should be able to re-work the grounding on each of the gen-sets one at a time so that you are not fully off line at any given time. Trying to accomplish this at the substation will require it be off-line which, if this switchgear is in fact feeding mission critical IT equipment, would be an expensive and disruptive process. I would suggest preparing as much work as possible while awaiting the delivery of the resistor assemblies. Each generator can be taken off line while the grounding change is implemented with the other unit on-line to provide at least partial emergency power availability. Remember that the neutral grounding resistors will heat up very rapidly and dissipate the heat to their surrounding environment gradually, but they will get very hot. Care should be excersized in locating them. Outdoors would be ideal, very often they are located on the generator cableing enclosure.

Hope this helps.

 

[dpc]

I'm sure you're aware that if the system uses any type of impedance grounding, you cannot have any single-phase (277V) loads.

Reactance grounding would be **extremely** rare for a 480V system (in the US anyway). High-resistance grounding would be more common. Reactance grounding might be used to limit generator line-to-ground fault current to three-phase fault current values, but I've never seen it used on a 480V system.

480V systems run inherent risk of arcing faults anyway, so if neutral is impedance-grounded, it is almost always high-resistance grounded. Low-resistance grounding is considered poor design practice and dangerous for 480V systems.

 

[robertb]

Per IEEE Std 142-1991, Chapter 1.6.3 (Multiple Power Sourcs) the system ground may be accomplished...
1) Each source grounded, with or without impedance
2) Each system neutral connected to a common neutral bus, which is grounded,with or without impedance.

Different grounding arrangements are shown in Westinghouse T&D Book, Chapter 19 ("Grounding of Power Ssystem Neutrals"), page 656.

Per the code discussion in IAEI magazine, May/June 2003, page 49: "Once paralleled the generators become one system, single-point grounding of the paralleling switchgear is acceptable".

See IEEE Std 446 for reference.

See IEEE IA #6, 1981, page 553 "Grounding of generators connected to industrial plant distribution buses" for information on fault and circulating currents.

 

[peebee]

"Where should the system be grounded at gen sets or Unit S/S."

If you are in the US. . .

Your utility service entrance is required to be grounded at the utility transformer, at the utility service disconnect, at both locations, or anywhere in between per NEC 250.30(A)(1).

If this is a 4-wire system (277-v), with 3-pole paralleling, (gens are NOT separately derived), then the only place NEC would permit you to ground the system is at the utility service entrance.

But I assume you were smart enough to stay away from a 4-wire system for a data center application. . . .

On a 3-wire system, NEC would permit you to ground both the gens & the utility service entrance (gens and utility would be considered separately derived). The utility service entrance would still be required to be grounded somewhere between the utility transformer and the service disconnect per NEC, no question there. BUT -- NEC would also permit you to ground the generators. Keep in mind, such a system gets much trickier to analyze. . . . If it were me, I'd try HARD to stick with a single-point grounded system unless you're dealing with VERY long distances between the utility and the generators.

 

[alehman]

What is the reason for changing the grounding? Grounding reactors are commonly used to reduce the ground fault current to manageable levels.

I agree with peebee's recommendation to stay with a single-point grounded system.

If you use a common grounding reactor, you must have a neutral disconnect for each generator. If you don't, maintenance personnel will be in danger if electrical service on any of the generators is performed while any of the others are operating.

Be sure to think about how you will test the system once the revisions are completed. Of course you will need to re-do the short circuit calculations and overcurrent coordination.

 

[peebee]

I need to qualify a statement I made:

"On a 3-wire system, NEC would permit you to ground both the gens & the utility service entrance (gens and utility would be considered separately derived)."

If you are paralleling full-time, then this statement is false. The systems would NOT be considered separately derived, and once again you would be forced into single-point grounding at the utility service entrance.
------------------
Also, note that if you do select single-point grounding at the utility service entrance, you might still elect to provide neutral-ground switches at the generators so that in the event of a catastrophic utility equipment failure, or replacement, you would be able to easily set up the generators as a separately derived system. If you did this, though, I'd strongly suggest posting some big instruction plaques indicating that the n-g switches should NEVER be closed EXCEPT in the event of the removal of the utility n-g bond.

 

[jbartos]

Suggestion to joan271273 (Electrical) May 19, 2004 marked ///\\I am working at a data center project were we have 4-2MW (480V)
///Please, clarify 480V system. Is it 480V 3ph 3wire neutral solidly grounded or 4-wire with a neutral solidly grounded?\\

 

[joan271273]

The system is made with 3phase-3w, I will check the project drawings to be sure that there is no neutral. The system voltage is 480V. The grounding system right now is a solid ground system. The system is been changed to a neutral reactor grounded system, Not H-R system. I believe the reason is to bring down the SC values. However I am concerned about changing the grounding scheme a data center operational. Remember the data center cannot be shut down so the change will happen with IT equipment operational. What are the exposures and how to mitigate them is my concern.

 

[busbar]

It is probably a good idea to study what changes may be necessary in protective-device settings with addition of a reactor.

Typically, short-circuit characteristics are greater with the 'utility feed' compared to that of gensets as a power source. If generator-stator protection during ground faults is of primary concern, a scheme link winding-differential relaying may be a desirable alternative.

 

[rbulsara]

Inserting impedance ( reactive or resistive) in the neutral does not help bring SCC down! As it affects only L to G fault currents. Phase to Phase and 3-phase bolted SCC remains the same as before. You still need to provide the equipment and breaker of full SCCR and Interrupting rating!

Imepedance grounding only permits you to continue operation (for a while) in case of a L-G fault, the most common of all. Most important benefit (and goal) is to limit the damage to the generator system. You need to clear this before it develpes in to more catastophic another L-G fault, resulting i L-L fault. This again is only true when you are on the generator. You have to ground the utiltiy service solidly.

Also as suggested by other posts, for a 480V, 3W system High resistor grounding is the most common in the USA. A 30 ohm resitor will limit the GF to less than 10 amps of L-G fault current! (277V/30 ohm) It does not get simpler or cheaper than this, I don't understand all other options being considered.

You can add resistor to one Generator at a time withoug affecting the data center, as long as remaining generators can support the load. Also pick a bright sunny day to do this!

I don't understand the need or reasoning for reactor grounding the unit substations. Where in the system are this unit substations, btw?

 

[alehman]

Clarification to rbulsara's statement "Inserting impedance ( reactive or resistive) in the neutral does not help bring SCC down!"

The zero sequence impedance (Z0) of most generators is much less than the positive sequence impedance (Z1). Therefore the L-G fault current of most generators is significantly higher than the L-L or L-L-L value. Adding neutral grounding impedance reduces the L-G fault current. Typically the object being to obtain Isc(L-G) <= Isc(L-L-L).

With multiple LV generators, L-G fault current can exceed switchgear and generator ratings. Also, many medium voltage generators require grounding impedance to prevent damage to the coils in case of a L-G fault.

These would be considered fault current limiting or low impedance grounds. High-impedance grounding will also correct this problem, but is used for other reasons as well.

 

[jbartos]

Suggestion to joan271273 (Electrical) May 23, 2004 marked ///\\The system is made with 3phase-3w, I will check the project drawings to be sure that there is no neutral.
///Yes, this is important.\\ The system voltage is 480V. The grounding system right now is a solid ground system. The system is been changed to a neutral reactor grounded system, Not H-R system. I believe the reason is to bring down the SC values.
///Yes, this is sometimes done.\\ However I am concerned about changing the grounding scheme a data center operational. Remember the data center cannot be shut down so the change will happen with IT equipment operational.
///In order to keep the data center operational in case of single phase to ground fault, high-resistance system grounding is recommended. The best choice will be to allow the ground fault for an indefinite period of time to be able to locate it and remove it. This will require 173% cable and transformer insulation levels. Alternately, 133% insulation level might be considered if the fault will be possible to remove within one hour. A pulse generator in the neutral is recommended to be able to pinpoint the ground fault location faster.\\

 

[alehman]

Comment to the previous post: there is considerable disagreement in the business as to the benefits of a high-resistance ground toward improving reliability in data center environments. Reliability is only improved in so far as maintenance personnel identify and correct the "first fault" before a second occurs. It is generally preferrable to design a system from the outset with redundant paths such that a single circuit breaker trip can be tolerated.

Also, one must be sure to confirm that all equipment (e.g. UPS's) can withstand and operate correctly with the required 173% overvoltage.

 

[jbartos]

Comment on the previous posting: I agree that it is better to have a fast solid state transfer switch to provide and uninterruptible power supply to the data center; however, the original posting does not address this option.
The 173% level insulation class does not imply 173% overvoltage, it merely reflects a thickness of insulation that can withstand higher overvoltage and the ground fault for an indeterminate amount of time.

 

[rbulsara]

To alehman's comment:

I agree that L-G faults could be higher than L-L fault currents. But the difference is usually not that drastic. The reduction of overall SCC is not as drastic as the reduction in L-G fault currents using a N resistor.

For example, one may have 95kA of L-L SCC and 104kA of L-G SCC. By inserting impedance in N, one may get the 104A L-G SCC down to, say 10Amps, but the the equipment still needs to be rated for 95kAIC. In the same example, however, it may make a difference between using a 100kAIC equipment vs. 200kAIC rated equipment. (I chose the example on purpose, where a N ressitor CAN make a difference in choosing the rating of the equipment)

 

[alehman]

rbulsara: My experience in some cases has been there is a much larger difference between L-L and L-G fault values than in your example. It depends highly on the winding configuration of the generator (Onan generators are much worse than Caterpillar in this regard). Where it is a problem, a small amount of neutral reactace will reduce the L-L value to the same or less than L-G. It's not a common occurance, but it can save a lot of money on circuit breakers. I mention it because I recently came across this situation.

 

[joan271273]

Thanks for all the replies.

A) One Line:

1) The one line is made up as follows: The Utility serves a 12 kV gear , this gear connects to two (2) Unit substations (12kV-480V).
2) These Unit S/S then serve all of the project loads. But each load feeder run through an ATS that is also connected to a paralelling gear.
3) This paralelling gear is the one that connects to the gen sets.

B) System Data

1) I received the following SC info: The calculated maximum fault current with 3 generators is 60,617 amps SYM (3 Phases) and 13,723 amps SYM SLG. The calculated parameter is based on the following per unit sequence:
(+) X" =0.2072 X/R=23.7500
(-) X" 0.1764 X/R= 23.7500
(0) X" 0.0592 X/R= 23.7500

Neutral Impedance R=0 Ohms; X= 0.05 Ohms

The generators are to be grounded with 0.05 Ohm reactances, 15000 Amps/10 secs.

I hope this helps.

Can someone clarify the 173% insulation issue!!!!. All of the 12kV wiring is 133% , as this is a data center how can I check if the UPS can operate in this scenario or the IT equipment per se.

Thanks

 

[busbar]

On MV-cable insulation, IEEE Std 141-1993 [Red Book] §7.2.5 There are three levels of conductor insulation for medium-voltage cables: 100, 133, and 173% levels. The solidly grounded system permits the use of 100% insulation level. When the fault on the other system will raise the system voltage above normal during the time of the fault, 133% insulation level should be specie if the fault is cleared within one hour. When the fault will remain on the system for an indefinite time, 173% voltage level insulation should be used.

 

[peebee]

Assuming you do have a 3-wire system (you said you were going to confirm this, but you haven't done that yet), and assuming your ATS's do not permit continuous paralleling (most don't, but you should check that too):

Your generators are a separately derived system. You can ground them without concern to the utility system. The best place to ground them is probably at the paralleling switchgear, assuming that they are not very far away from the paralleling gear.

As the ATS's do not permit continuous paralleling of the gens with the utility, the analysis of the ground return paths is greatly simplified. Treat the generators and the utility substations separately. Look at the ground impedance of the gens, figure out how much fault current you would see on a fault, and set the relaying appropriately. Perform a similar (but separate) analysis for the utility substations.

Hope that helps.