We experienced the simultaneous failures of two 4kV fan motors on a high resistance grounded system. They were on the same switchgear bus, a 6000 HP fan (located outside) and a 600 HP fan (inside) – both less than 300 feet from the bus. The 6000 HP fan has surge protection at its terminals – the only other surge protection is on high side of main transformer. Our control system shows the loss of the 6000 HP fan one second before the 600 HP fan, however, the points have a 0.5 second scan time and it is too close to call if that is real or not. The 6000 HP fan went on instantaneous at 8920 A (740 FLA) and one end coil was visibly blown out, and the 600 HP fan on time overcurrent at 1160 A (76 FLA). No abnormal weather or operating conditions. A similar situation occurred previously wherer we got a 2-fer in failures. Any ideas?
Tek-Tips is the largest IT community on the Internet today!
Members share and learn making Tek-Tips Forums the best source of peer-reviewed technical information on the Internet!
-
Congratulations cowski on being selected by the Eng-Tips community for having the most helpful posts in the forums last week. Way to Go!
Simultaneous 4kV Motor Failures 2
- Thread starter joepower
- Start date
electricpete
Electrical
There are a large number of possibilities. Easy to speculate, tough to prove
#1 - An abnormal voltage (surge or steady state) was initiating event and caused both motors to fail.
#2 - Failure of first motor (likely the 6000hp motor) caused voltage transient which caused 2nd motor to fail.
#3 - Ungrounded system may have contributed to item #1 or #2. There have been lots of threads on intermittent arcing faults on ungrounded system.
#4 - Failure of 6000hp was the initating event and the 600hp tripped due to the high current in 600hp motor as a result of the 6000hp motor fault not being cleared rapidly? An induction motor acts like a generator during a fault.
Maybe some monitoring of system transient and steady state voltages to ground would shed some light if there are occasional destructive conditions in the power system.
#1 - An abnormal voltage (surge or steady state) was initiating event and caused both motors to fail.
#2 - Failure of first motor (likely the 6000hp motor) caused voltage transient which caused 2nd motor to fail.
#3 - Ungrounded system may have contributed to item #1 or #2. There have been lots of threads on intermittent arcing faults on ungrounded system.
#4 - Failure of 6000hp was the initating event and the 600hp tripped due to the high current in 600hp motor as a result of the 6000hp motor fault not being cleared rapidly? An induction motor acts like a generator during a fault.
Maybe some monitoring of system transient and steady state voltages to ground would shed some light if there are occasional destructive conditions in the power system.
Are the short-circuit numbers {8920/1160A} ø-ø or ø-g values? They seem large for ground faults on a typical high-resistance grounded system. In this configuration, when a phase-to-ground fault occurs, voltage on the unfaulted phases raises to phase-to-phase values, but normally this is allowed for in the system insulation levels. If the resistive ground-fault current does not equal or exceed the system capacitive charging current, some oscillatory/transient overvoltage may occur. Have equipment and cables been added after the grounding resistor was sized for then-present conditions?
electricpete
Electrical
Whoops. Delete my #3. I was thinking ungrounded, not high-resistance grounded.
jbartos
Electrical
- Jan 15, 2000
- 6,440
Suggestion: Assuming that the high-resistance system grounding did not fail, then the shorts were phase to phase. The reasons can be:
1. Large surge beyond surge protector ratings
2. Harmonic distortion, i.e. high voltage harmonic content, especially, if there are ac motor variable speed drives
3. Voltage spikes in the power supply, (Are there any static switcher?).
4. Winding insulation failure due to insulation defects causing short(s).
1. Large surge beyond surge protector ratings
2. Harmonic distortion, i.e. high voltage harmonic content, especially, if there are ac motor variable speed drives
3. Voltage spikes in the power supply, (Are there any static switcher?).
4. Winding insulation failure due to insulation defects causing short(s).
-
1
- #8
I would tend to agree with Busbar. We had a similar instance on a 5 kv MCC where 2 motors failed within about 45 sec of each other. At that time all of our motors on the MCC ( about 12) had surge caps. We found that we exceeded to maximum charging current rule for HRG systems if we had 2 or more caps connected. In your case, it sounds as if you have one surge cap and the capacitance of the cables to consider.
We were able to show that our motor cable length was adequate enough to protect our motors (basically the cable is also a capacitor, albeit smaller than a surge cap). So we disconnected the surge caps.
We were able to show that our motor cable length was adequate enough to protect our motors (basically the cable is also a capacitor, albeit smaller than a surge cap). So we disconnected the surge caps.
electricpete
Electrical
I need some time to work thru thisw discussion of high-r grounding by busbar and Gord.
One item was mentioned a max expected fault current delievered from the source. But doesn't that exclude the capacitive current mentioned by Gord which in theory can be much higher limited only by fault resisstance as capacitance discharges thru the fault.
We have high R systems with many long high-C cabbles and many motor surge caps attached. I'll have to look at that scenario closer to see if it applies to us.
One item was mentioned a max expected fault current delievered from the source. But doesn't that exclude the capacitive current mentioned by Gord which in theory can be much higher limited only by fault resisstance as capacitance discharges thru the fault.
We have high R systems with many long high-C cabbles and many motor surge caps attached. I'll have to look at that scenario closer to see if it applies to us.
electricpete
Electrical
I did locate the calculation used to size our neutral grounding resistors. It does compare total estimated system cacitive reactance to ground to grounding resistance as described above.
I was a little surprised to see that they considered cable capacitance to ground (based on cable type and lenght) and motor capacitance to ground (based on horsepower and speed), but not surge capacitor. I'm guessing perhaps the surge caps are small contributor to the total?
I was a little surprised to see that they considered cable capacitance to ground (based on cable type and lenght) and motor capacitance to ground (based on horsepower and speed), but not surge capacitor. I'm guessing perhaps the surge caps are small contributor to the total?
- Thread starter
- #11
The motors were running under normal full load conditions when the event happened with non other switching. No major system changes have been made since the system was designed or the surge protection added. There are no harmnonic generators on the system such as VFDs. Both motors did fail, and the rapir shop just said that the 6000 HP motor was burned in at least three places, but I don't have the full report yet on it or the 600 HP motor. The maximum charging current rule is intriguing and I've also got to do some research there.
Unlike most power-factor capacitors in medium-voltage applications, the surge-protective capacitor wyepoint is intended to be connected to machine frames and station-ground bus. [Example at www.geindustrial.com/products/manuals/GEH-2730B.pdf] This could significantly increase charging current as effectively in parallel with the dielectric of shielded cables, transformers and rotating equipment—lowering zero-sequence capacitive reactance, with the prospect of increased phase-to-ground oscillatory/transient overvoltage.
-
1
- #13
Pete...in our installation, the surge capacitor current was a significant portion of our resistor let-thru current. We used 0.5 uF caps which result in a charging current of 1.35 A or about 50% of our resistor let-thru.
We had operated for about 20 years in the above configuration. Old-timers did indicate that we used to lose motors in groups ( within days or a few weks of each other). My previous posting was the first time that motors failed within sec's of each other.
So just because the charging current exceeds the resistor let-thru does not mean you imediately have a problem. The problem occurs when you have an arcing ground fault. I modelled our installation using PSpice and found that when charging current equalled resistor let-thru current, the resulting transient voltage rise was about 200% - this matched to textbooks I had read on HRG systems. As the charging current increased, transient voltage increased.
I think, in simplistic terms, a ground fault results in the ground capacitances getting charged up when the fault is applied. If the fault is removed (ie arcing), these capacitances try to discharge thru the neutral resistor. If the resistor is small, the ground capacitances can almost totally discharge befor the next application of the fault; if the resistor is to big, the energy from successive applications of the fault adds onto the capacitor voltage remaining.
We had operated for about 20 years in the above configuration. Old-timers did indicate that we used to lose motors in groups ( within days or a few weks of each other). My previous posting was the first time that motors failed within sec's of each other.
So just because the charging current exceeds the resistor let-thru does not mean you imediately have a problem. The problem occurs when you have an arcing ground fault. I modelled our installation using PSpice and found that when charging current equalled resistor let-thru current, the resulting transient voltage rise was about 200% - this matched to textbooks I had read on HRG systems. As the charging current increased, transient voltage increased.
I think, in simplistic terms, a ground fault results in the ground capacitances getting charged up when the fault is applied. If the fault is removed (ie arcing), these capacitances try to discharge thru the neutral resistor. If the resistor is small, the ground capacitances can almost totally discharge befor the next application of the fault; if the resistor is to big, the energy from successive applications of the fault adds onto the capacitor voltage remaining.
electricpete
Electrical
I agree with everything that has been said. I see in IEEE documents where the importance of keeping neutral resistance < Xc is stated to prevent damaing transients is stated.
I analysed the system consisting of balanced 3-phase resistance grounded source powering three identical capacitances connected to ground, with a time-varying fault resistance connected across phase as shown in:
Equation 11 gives a solution:
d(En(t))/dt = -1/3*En(t)/Rn/C - 1/3*(En(t)+Ea(t))/C/Rf(t)
where En(t) is neutral voltage referenced to ground, Ea(t) is A phase voltage referenced to ground, C is capacitance to ground, Rf(t) is time-varying fault resistance, Rn is neutral grounding resistance.
At times when Rf(t) is infinity, the first term gives a decaying exponential response which always decreases |En| over time.
When Rf becomes low, the 2nd term can act to increase |En(t)| over time, but ONLY when En and Ea are opposite sign AND |Ea| > |En|.
Therefore we can see the peak value that En can attain is the peak value of Ea.
If we have no capacitance and a solid ground short, then En(t)= - Ea(t).
in that case Vb(t) = En(t)+Eb(t) = -Ea(t)+Eb(t) will have a peak value of sqrt(3) times the nominal line-to-ground voltage, since there is 120 degree angle between Ea and Eb.
Now if we add the capacitance and remove the short when Ea hits a peak, the peak value of En(t) will decay slowly. At 60-degrees later it will be 180 degrees apart from Eb or Ec and will create a voltage approaching 2x nominal line-to-ground voltage. (exactly 2x if no decay occurs during that 60 degree time span).
Now the question…. how do we expect a machine to respond to line-to-ground voltage increasing by a factor of 2 above nominal?
IEEE432-92 and others specify that ac hi-pot tests for machines with service-aged insualtion be performed at a level of 125% to 150% of rated machine line-to-line voltage for one minute. Taking the lower limit of 125%, that corresponds to a 1.25*sqrt(3) ~ 215% of nominal line-to-ground voltage. If a machine were to fail at less than 200% of nominal line-to-ground voltage for duration of less than one minute, it seems to me that the insulation was already weak. What do you guys think?
It also makes me wonder whether voltage can increase higher than the factor of 2 predicted above if we change the model....
I think that if we modeled other motors connected to the system ot would have no effect… they continue to see balanced voltage applied to their terminals (even though line-to-ground voltage is changing).
But I do think that we can get higher voltages and surges if we add a series inductance into the supply circuit representing transformer and cable impedances. That can give some oscillatory behavior in reponse to a step-change in resistance. If I get a chance I will try to model that.
Although we have spent a lot of time in discussion of this particular aspect (partly due to my comments) I would recommend to the original poster to keep an open mind to a wide range of possibilities for his problem.
I analysed the system consisting of balanced 3-phase resistance grounded source powering three identical capacitances connected to ground, with a time-varying fault resistance connected across phase as shown in:
Equation 11 gives a solution:
d(En(t))/dt = -1/3*En(t)/Rn/C - 1/3*(En(t)+Ea(t))/C/Rf(t)
where En(t) is neutral voltage referenced to ground, Ea(t) is A phase voltage referenced to ground, C is capacitance to ground, Rf(t) is time-varying fault resistance, Rn is neutral grounding resistance.
At times when Rf(t) is infinity, the first term gives a decaying exponential response which always decreases |En| over time.
When Rf becomes low, the 2nd term can act to increase |En(t)| over time, but ONLY when En and Ea are opposite sign AND |Ea| > |En|.
Therefore we can see the peak value that En can attain is the peak value of Ea.
If we have no capacitance and a solid ground short, then En(t)= - Ea(t).
in that case Vb(t) = En(t)+Eb(t) = -Ea(t)+Eb(t) will have a peak value of sqrt(3) times the nominal line-to-ground voltage, since there is 120 degree angle between Ea and Eb.
Now if we add the capacitance and remove the short when Ea hits a peak, the peak value of En(t) will decay slowly. At 60-degrees later it will be 180 degrees apart from Eb or Ec and will create a voltage approaching 2x nominal line-to-ground voltage. (exactly 2x if no decay occurs during that 60 degree time span).
Now the question…. how do we expect a machine to respond to line-to-ground voltage increasing by a factor of 2 above nominal?
IEEE432-92 and others specify that ac hi-pot tests for machines with service-aged insualtion be performed at a level of 125% to 150% of rated machine line-to-line voltage for one minute. Taking the lower limit of 125%, that corresponds to a 1.25*sqrt(3) ~ 215% of nominal line-to-ground voltage. If a machine were to fail at less than 200% of nominal line-to-ground voltage for duration of less than one minute, it seems to me that the insulation was already weak. What do you guys think?
It also makes me wonder whether voltage can increase higher than the factor of 2 predicted above if we change the model....
I think that if we modeled other motors connected to the system ot would have no effect… they continue to see balanced voltage applied to their terminals (even though line-to-ground voltage is changing).
But I do think that we can get higher voltages and surges if we add a series inductance into the supply circuit representing transformer and cable impedances. That can give some oscillatory behavior in reponse to a step-change in resistance. If I get a chance I will try to model that.
Although we have spent a lot of time in discussion of this particular aspect (partly due to my comments) I would recommend to the original poster to keep an open mind to a wide range of possibilities for his problem.
RRaghunath
Electrical
I see the high resistance grounding as problematic on two counts.
One that it cannot and should not be applied where there are capacitors (including surge suppressors) in the system or / and where the equipment is spread out in the plant involving lot of cabling (capaciatance of cables). The thumb rule is that the total capacitive currents in the system shall not exceed 10A.
The second pertains to detection of earth faults or more precisely identifying the faulty feeder to isolate. It is generally not possible considering the low fault current magnitude and the earth fault in the system is identified by measurement of residual voltage and annunciated. It also happens in many plants that annunciations are not given much attention (rightly so though, some times) and as a result, the feeder continues to be in service with an earth fault in the motor or the connecting cable.
Under above conditions, the voltage of the healthy phases is raised, putting more stress on the insulation. No surprise, that a fault in another phase occurs sooner than later and both the feeders (fault in one phase in one feeder and another fault in the second feeder being one of the possible cases) trip which means the client is saddled with two simultaneous failures, like the one in the post.
I believe that high resistance grounding or isolated systems demand high level of education and awareness and education among plant operation and maintenance personnel to reap the benefits of the system (such as continity in production even with a ground fault). Further, proper design, good quality of equipment and installtion too is a prerequisite. Raghunath
One that it cannot and should not be applied where there are capacitors (including surge suppressors) in the system or / and where the equipment is spread out in the plant involving lot of cabling (capaciatance of cables). The thumb rule is that the total capacitive currents in the system shall not exceed 10A.
The second pertains to detection of earth faults or more precisely identifying the faulty feeder to isolate. It is generally not possible considering the low fault current magnitude and the earth fault in the system is identified by measurement of residual voltage and annunciated. It also happens in many plants that annunciations are not given much attention (rightly so though, some times) and as a result, the feeder continues to be in service with an earth fault in the motor or the connecting cable.
Under above conditions, the voltage of the healthy phases is raised, putting more stress on the insulation. No surprise, that a fault in another phase occurs sooner than later and both the feeders (fault in one phase in one feeder and another fault in the second feeder being one of the possible cases) trip which means the client is saddled with two simultaneous failures, like the one in the post.
I believe that high resistance grounding or isolated systems demand high level of education and awareness and education among plant operation and maintenance personnel to reap the benefits of the system (such as continity in production even with a ground fault). Further, proper design, good quality of equipment and installtion too is a prerequisite. Raghunath
This has been an interesting discussion. Just a couple of points:
In the U.S., HRG is not generally used for medium-voltage distribution system, mainly because it is not possible to selectively trip on ground faults. I believe I have a paper by Louie Powell or other GE engineer regarding renewed interest in HRG on medium voltage systems to limit damage to motor stator iron on ground faults.
My experience with HRG is primarily on large generators with unit transformers. Selectivity is not an issue in this case. Surge capacitors are universally used in combination with surge arresters at the generator terminals. The capacitance is factored into the resistor sizing calculations.
I don't see an inherent problem with use of surge capacitors on an HRG system, provided the capacitance is taken into account. Use of PF correction caps is probably not a great idea. Another drawback to HRG systems.
I'll see if I can dig up that paper on Monday.
In the U.S., HRG is not generally used for medium-voltage distribution system, mainly because it is not possible to selectively trip on ground faults. I believe I have a paper by Louie Powell or other GE engineer regarding renewed interest in HRG on medium voltage systems to limit damage to motor stator iron on ground faults.
My experience with HRG is primarily on large generators with unit transformers. Selectivity is not an issue in this case. Surge capacitors are universally used in combination with surge arresters at the generator terminals. The capacitance is factored into the resistor sizing calculations.
I don't see an inherent problem with use of surge capacitors on an HRG system, provided the capacitance is taken into account. Use of PF correction caps is probably not a great idea. Another drawback to HRG systems.
I'll see if I can dig up that paper on Monday.
faulty, you raise some good points about high-resistance grounding limitations. The facility owner/staff must understand that the first ground fault allows process continuity, but that, if not acted upon, the likelihood of two faults increases, and will disrupt more equipment and cause progressively greater expense and disruption.
If immediate efforts are not dedicated to eliminate the first fault, then the method loses its operational advantage.
In the case of expecting a ground alarm not to be acted upon, then yes, definitely it is not the best choice, and a first ground fault should operate an overcurrent device, as in the case of low-resistnce or solid grounding. That capability and limitation is ~50 years old, but is not a cure-all.
Also, high-resistance grounding should be applied to the load serving a manageable section of the electrical system, and if too large (high zero-sequence capacitive reactance) then it is also a poor choice.
If the client operation does not understand the tradeoffs, then it is probably counterproductive to implement this grounding method.
dpc, I may be wrong, but my understanding is that surge capacitors are intended to be phase-to-ground connected, whereas PF-correction capacitors typically are not.
electricpete
Electrical
Maybe I'm mistaken but I think the comments about education, responding to ground alarms etc apply to ungrounded, not high-R grounded.
We have 13.2kv system with 20ohm neutral grounding resistor. I think it falls in the category of high-R grounded.
I know we don't have any ground alarms on our high-resistance grounded systems, only ground fault trips.
Individual motor loads are protected with residually connected ct's or differential-connected ct's. The transformer supplying the bus is protected by ground current sensed in neutral (higher lelel). I think this provides plenty of sensitivity for selective tripping.
We have 13.2kv system with 20ohm neutral grounding resistor. I think it falls in the category of high-R grounded.
I know we don't have any ground alarms on our high-resistance grounded systems, only ground fault trips.
Individual motor loads are protected with residually connected ct's or differential-connected ct's. The transformer supplying the bus is protected by ground current sensed in neutral (higher lelel). I think this provides plenty of sensitivity for selective tripping.
electricpete
Electrical
I'm not sure about that resistance value. Maybe 400 ohms and 20 amps. I'll look it up on Monday.
Pete...the theoretical voltage rise due to an arcing fault can approach that of an ungrounded system (ie 6X). I think you will see this when you add the effect of inductance and continue the analysis for several cycles.
In your analysis you say you remove the short when the voltage was at a peak. Shouldn't you apply the short when the voltage exceeds a level - insulation breakdown depending on a vols/mil being exceeded?
When I did the PSpice analysis of my situation, I got a voltage peak of about 4-5X rated. The frequency of the oscillation was also quite high so the motor was shown to be overstressed when compared to the surge voltage withstand envelope (this is the curve relating motor insulation capabity vs microseconds). This is a much snaller scale than the hi-pot test and reflects the fact that transients take a finite time to propogate through the winding.
In your analysis you say you remove the short when the voltage was at a peak. Shouldn't you apply the short when the voltage exceeds a level - insulation breakdown depending on a vols/mil being exceeded?
When I did the PSpice analysis of my situation, I got a voltage peak of about 4-5X rated. The frequency of the oscillation was also quite high so the motor was shown to be overstressed when compared to the surge voltage withstand envelope (this is the curve relating motor insulation capabity vs microseconds). This is a much snaller scale than the hi-pot test and reflects the fact that transients take a finite time to propogate through the winding.
- Status
Similar threads
- Question
- Question