2250HP Motor Start with 2500KVA Transformer 2

[ProjEngr]

We have a 2250 HP 4160V 3P 60Hz 1.15SF 270FLA Induction motor on a large compressor. There are surge protectors and capacitors (GE 1.5uF, 3 Pole, 4160V) on each phase at the motor. The motor dropped out one day in the middle of unit operation - no alarms on the Multilin, no obvious signs of any issues except that the starter cubicle was hot inside. So we replaced the starter with a new Eaton Ampgard MVC and Multilin 369. The transformer is a 13.8kV/4.16kV Delta-Delta, 2500KVA, that feeds only this motor. When trying to start, the motor would start to turn then everything would drop out. We get an under voltage alarm on the Multilin. The control power voltage dropped to 89V and the incoming line voltage dropped to 3225V before tripping. We tested the motor (high pot & surge), we tested all the feeders (megger), we tested the transformer oil (no signs of degradation - no dissolved gases) we tested the caps. We found one of the capacitors to be bad, so we changed all three (it's really one assembly with (3) caps in it). We checked the Schweitzer relay in the substation and found no issues with voltage drop or amp draw on the 13.8kV side. To get it started, we supplied 110v control power to the starter from a nearby UPS. This move helped keep the starter coil pulled in and ultimately allowed restart of the motor. This motor has been in service for 15 years with no issues - it's generally started twice a year. We once installed a soft start, but the machine started blowing out couplings - this was attributed to having critical speeds (machine natural frequency) that were between 0 and operating RPM - so we went back to an across the line type starter. So I have a few questions for those who would like to take a stab...and hopefully I have supplied enough information!

To me, all signs point to the transformer and it's voltage drop during inrush.
1. Can a transformer degrade in performance without showing signs of this in the oil?
2. Is a Delta-Delta the proper transformer for this application?
3. Does this type of setup need the starter capacitors? Should their sizing be reviewed?
4. Should I get a transformer with a different inductance? It's a 5.6% IZ @ 85C.

Your thoughts are appreciated!!

 

[waross]

The old starter would hold in at a little lower voltage than the new one. If the motor has been in service for 15 years and neither the motor nor the transformer show signs of damage or degradation there is probably nothing wrong.
The transformer is acting as a reactance starter, similar to a primary resistance starter but with reactance instead of resistance.
I would continue to use the UPS twice a year.
I understand the desire to find the cause of the original trip to save a future unexpected failure, but it may be too late now. I suspect the original starter had a gremlin.

Bill
--------------------
"Why not the best?"
Jimmy Carter

 

[dpc]

This is a relatively large motor. If it dropped out, you really need to try to figure out what caused it. I assume the motor was checked and the contactor was tested.

Most likely, the new starter is dropping out during start because of the voltage drop in the control circuit. But that is just a guess. With a 2250 hp motor and a 2500 kVA xfmr there will be significant voltage drop during starting if starting across-the-line. The original design most likely had a reduced-voltage starter of some type for this reason.

It sounds like you could really benefit from having a EE on site to try to get to the bottom of the original problem.

David Castor

http://www.cvoes.com

 

[burnt2x]

Thinking backwards, could it be that somethings changed on the supply side of things (13.8 kV side)? Assuming your down-the-line equipment are okay now (caps replaced as well as the starter), were there additional loads on the 13.8 kV side? I am looking at he problem based on you post telling "the incoming line voltage dropped to 3225V". Can you give us a picture of how you source your power from?

 

[waross]

The bad cap may have been the original cause of the trip. The voltage dip may be reasonable for that motor starter/combination.
The starting issue now is most likely due to a less voltage tolerant starter.

Bill
--------------------
"Why not the best?"
Jimmy Carter

 

[LionelHutz]

I'm betting the old starter did not have the Multilin, and if it did have a Multilin then the settings were different. It's also quite possible the old starter had a different control scheme and a latching contactor to hold-in during the starting voltage dip.

 

[jraef]

That (latching contactor) is exactly what I was thinking too.

We once installed a soft start, but the machine started blowing out couplings - this was attributed to having critical speeds (machine natural frequency) that were between 0 and operating RPM - so we went back to an across the line type starter.

I'm also surprised at the assessment that the soft starter somehow damaged the couplings. Generally the use of a soft starter PRESERVES couplings. If it ran at a critical speed long enough to cause coupling damage, that tells me someone didn't know how to set up the soft starter correctly. Either that, or the power supply was too weak even for the soft starter and it got stuck in current limit because of the same votlage drop issue you were having with the Across-the-Line starter.

And if you failed to remove the caps from the motor when you tried the soft staryter, that would explain the damage to the caps, maybe the soft starter failure too.

"Dear future generations: Please accept our apologies. We were rolling drunk on petroleum."
— Kilgore Trout (via Kurt Vonnegut)
For the best use of Eng-Tips, please click here -> faq731-376

 

[burnt2x]

Nice sleuthing there, jraef! A star for you.

 

[jraef]

I commissioned a 2000HP 4160V pump with a soft starter once without opening the connection box first to check everything, because I trusted the electrician. It didn't occur to me that he had not read the installation manual that warned him 5 times not to have capacitors connected to the motor when powered from a soft starter (either that or he had no idea the huge devices inside the connection box were capacitors). The caps lasted about 3 seconds and the SCRs in the soft starter shorted on the very next attempt after removing the dead caps.

It was a very expensive lesson that I only had to learn once...

"Dear future generations: Please accept our apologies. We were rolling drunk on petroleum."
— Kilgore Trout (via Kurt Vonnegut)
For the best use of Eng-Tips, please click here -> faq731-376

 

[ProjEngr]

Thanks for all the great input - very interesting stuff (as you may be able to tell, I'm mechanical...)

A few clarifications (and I apologize for being so lengthy - just trying to give you guys all the details...) The "soft start" we installed was actually a Toshiba VFD. We had a reputable electrical company specify and design the system. They are the ones that specifically put the caps on the motor, which was also Toshiba. I checked the starter schematics and verified that the old motor did not have caps. We didn't change the horsepower when we switched the motor - which we still keep as a spare. I think we had a 15 second ramp up time. We blew out (2) couplings before we figured it out. The couplings were VERY good couplings from a well respected company. The natural frequencies combined with the sheer mass of the system (this compressor has a gear box, which ramps the compressor speed to about 10,000rpm) exerted tremendous pressure on the coupling. I plan to research the comment about start capacitors on a soft start though...

The old multilin on the original starter was a 239. We had the RTD's connected to a separate temperature monitoring device, built into the VFD. We replaced this with a 339, and now have the RTD's connected to it. Not sure about the 239, but I found out this week that you have to tell the 339 to alarm for most any event. Funny it wouldn't come from the factory pre-configured to alarm for everything. I'm not sure if the 239 has the same functionality, but I know we didn't even have PT's on the old starter feed wiring. I think this is why we had no indication of any voltage problems during the initial shutdowns. The guys that put it in probably knew better and left them off!

I don't believe the old contactor was a latching type. It was a standard vaccum breaker. I'll look more into that too.

I think it's plausible that the bad cap could have been responsible for the mid-run shutdown. We've also always thought the starter had a gremlin. I guess something that I should have mentioned was location. We are a Gulf Coast refinery. We have been experiencing very hot weather. I pulled the trends from the (2) mid-run shutdowns we've had on this motor and they happen to correlate to the hottest days of the year. Seeing this, I took a look at the cooling coils on the transformer and found the fans not running - the transformer temperature gauge showed 75C, which seemed hot. I asked our technicians to test the fans and thermostat - the fan wasn't even wired up!! We setup some external cooling on this transformer and now have it operating at a nice 55C. Not having a tremendous amount of experience with transformers, could this temperature issue be the root cause?

I've started to shop for a new transformer. I'm looking to get a 3000-3500KVA (~1.5 KVA/HP) I'm checking all the feeders to ensure proper sizing, etc. We've always standardized on CU/CU transformers and have gotten excellent life and reliability from them. What do you guys think of AL/AL?

 

[Skogsgurra]

I have looked through your input and the comments given above.

I agree that the capacitors are a no-no. Are you sure they are connected on the motor side? And that they are capacitors and not snubbers (RC units)?

Also, I do not think that the capacitors caused the coupling breakdown. That was most likely a resonant situation, as you say.

You said nothing about the VFD. If it is a Toshiba T300MVi, then the output is an eight-level PWM and that is not as bad as a pure (three-level) PWM since voltage steps are only around 2000 V and not taking the around 7000 V step directly. Still, I see no explanation for the capacitors. And I would never even think about installing them and I see no recommendation in the Toshiba literature either.

A DOL started motor usually (always) draws lots of reactive power. PF capacitors (large ones) help. You could also improve the situation by adding reactance to the motor (coiled cables often help and can be used to find a suitable uH value) to reduce starting current. If you are lucky, you can leave them in circuit when running but most likely you need to short-circuit them once you are up in speed.





Gunnar Englund

http://www.gke.org

--------------------------------------
Half full - Half empty? I don't mind. It's what in it that counts.

 

[waross]

After 15 years I would not be in a hurry to expend capital on a new transformer.
From the voltage drops that you give for power and control, it looks as if you may be running your control voltage at 115 Volts. I would try putting the control voltage up to 120 Volts, and setting the low volt alarm a little lower or adding a short delay to the under voltage trip.
One unexplained trip in 15 years that was coincident with a failed capacitor may not be indicative of serious issues.

Bill
--------------------
"Why not the best?"
Jimmy Carter

 

[jraef]

Gunnar,
Caps on the motor side when run off of a VFD may have been part of a filter package. ProEng being an ME may not have been able to notice the subtle difference. Caps still look like caps even if there is other "stuff" with them.

ProEng,
Notwithstanding the good point made by waross, do you have an EE looking after this? Seems not and it also seems like you are in need of one, especially one NOT associated with the vendors. You have issues here that appear to be a lot of expensive hardware thrown at a problem without enough investigation, a classic sign of a "vendor engineering" solution.

Skogs (Gunnar) is pretty good at this sort of thing, he'll fly anywhere for the price of a plane ticket, a few good meals and a sampling of local ale (plus maybe a modest fee?)



"Dear future generations: Please accept our apologies. We were rolling drunk on petroleum."
— Kilgore Trout (via Kurt Vonnegut)
For the best use of Eng-Tips, please click here -> faq731-376

 

[Skogsgurra]

Not so modest any more. Jeff. David (zdas4) told me to be less modest. And that works pretty well. But beer sampling still interesting.

Gunnar Englund

http://www.gke.org

--------------------------------------
Half full - Half empty? I don't mind. It's what in it that counts.

 

[waross]

jraef makes a lot of sense in his post. Gunnar is still too modest.
Swedish class self appreciation but world class abilities. (Or maybe it's the same thing) grin

Bill
--------------------
"Why not the best?"
Jimmy Carter

 

[electricpete]

the transformer temperature gauge showed 75C, which seemed hot. I asked our technicians to test the fans and thermostat - the fan wasn't even wired up!! We setup some external cooling on this transformer and now have it operating at a nice 55C. Not having a tremendous amount of experience with transformers, could this temperature issue be the root cause?

The temperatures you mentioned don't seem particularly hot to me. Is it an "oil temperature" or "winding hotspot" temperature? What is the temperature rise rating (55 or 65C?).

Also if transformer develops a defect, I'm inclined to think you would more likely see a transformer trip on differential rather than motor starter trip. And as you said, no dissolved gases.

2. Is a Delta-Delta the proper transformer for this application?

I don’t know the answer. Completely ungrounded on lv side? Does the soft starter introduce any ground reference ? We have ungrounded systems, but only at 480vac level. It raises another question about the caps: installed phase to ground? That would raise even more questions both about the presence of the caps and the implications of their removal (by the way I hope you didn’t gloss over jraef’s comment that caps don’t belong fed from soft starter).

I'm also surprised at the assessment that the soft starter somehow damaged the couplings. Generally the use of a soft starter PRESERVES couplings. If it ran at a critical speed long enough to cause coupling damage, that tells me someone didn't know how to set up the soft starter correctly. Either that, or the power supply was too weak even for the soft starter and it got stuck in current limit because of the same votlage drop issue you were having with the Across-the-Line starter.

Regarding the potential for soft starting to cause coupling failure due to torsional oscillations exciting a torsional resonance, I think it is possible.

Torque oscillation occurs during DOL start of induction motor is a little mysterious to me, not much discussed outside of academic literature, was discussed in the following thread
thread237-283114
Note in my post today, I have attached an updated powerpoint related to the questions of that thread which I put together just after that post ended, but never attached before.

The torque oscillations during DOL start have the following characteristics.
1 - the frequency is s*LF (starts at LF and decreases to almost 0 at full speed). (See slide 27 in stationary ref frame and slide 36 in rotor ref frame). For better example, see attachment to post 13 Oct 10 10:01 in other thread linked above.
2 - the torque oscillation magnitude is strongest upon startup and decays with open circuit time constant of Lm+L2 / [R2/s]. See slide 9 and 10. Also compare slides 9 to 27 and we see torque decays away much slower in time for locked rotor start.
3 – The magnitude can exceed breakdown torque. (slide 27)
Note the parameters of example motor which was simulated are shown on slide 8

Based on the above, it seems there are are 3 factors which might act in direction to make soft start torque oscillations thru torsional resonance potentially more damaging than DOL:
1 - (based on 1 above) - The frequency of oscillation s*LF will increase slower during reduced voltage start than DOL start, so spend more time in the resonant region.
2 - (based on 2 above) - The torque oscillations die away much slower with respect to time during reduced voltage start than they do in DOL start due to less damping of the rotor currents (larger s and smaller R2/s). One could argue this is irrelevant because what matters is how fast the torque oscillations die away with respect to speed. That is absolutely true. But the important thing that #2 tells us is we should not assume the torque oscillations have decayed away further by the time we reach resonant speed simply because it takes longer time to get there..... the reduced voltage start oscillations last longer in time than the DOL start oscillations.
3 – The soft starter currents are richer in harmonic content and therefore possibly in torque content. This would mean more excitation frequencies and more potential to excite torsional resonance.

On the other side of the coin:
1 - The magnitude of the s*LF torque oscillations during reduced voltage start would be less than DOL start, presumably factor V^2 although I havne’t checked by simulation,.
2 - Also, it’s worth considering that higher frequency oscillations are more attenuated by the inertia of the motor before even getting to the coupling than are lower frequency oscillations, which tends to diminish the significance of oscillations at s*LF when s near 1.0.

All in all we have a lot of factors. I don’t think one could say a given installation will have more severe torsional conditions during start as a result of installation of soft starter, but I definitely don’t think one can say we will have less severe torsional conditions. If coupling failures begin upon installation of soft starters, then excitation of a torsional resonance by motor torque oscillations (which have different characteristics during soft start) seems like a credible explanation to me.

=====================================
(2B)+(2B)' ?

 

[electricpete]

By the way, I don't particularly have reservations about the transformer but maybe I'm missing something. If you do have reservations about the transformer and want to gather more info, electrical tests (power factor, ratio tests, resistance tests etc) are something to consider.

=====================================
(2B)+(2B)' ?

 

[Skogsgurra]

It takes time to build amplitude at resonant frequencies. A DOL does not dwell long enough at a possible resonant frequency so no dangerous amplitudes can build up.

A soft starter with a 15 seconds ramp will dwell much longer in each resonance band and it is therefore possible to have damage with a VFD.

I think that a VFD with 'frequency jump' could work (it usually does) but I also think that a DOL start with increased impedance (reactors) will work also.

Gunnar Englund

http://www.gke.org

--------------------------------------
Half full - Half empty? I don't mind. It's what in it that counts.

 

[electricpete]

fwiw I agree 100% with the dwell time comment (roughly the same thing I said before). It is well known in the lateral vibration world where turbines and vfd's are moved quickly through resonance for this reason. In the lateral vibration case, the unbalance type exictation frequency varies directly with speed. The applicability to the present scenario is not well known until we suggest the source of the oscilation and its speed dependence. Are you envisioning s*LF frequency torsional oscillations as predicted by Krause?

=====================================
(2B)+(2B)' ?

 

[electricpete]

If you’all will permit me to go a little off track, I think it is interesting to step back and look at 3 different frequencies of electromagnetic torque oscillation that are predicted in motors in various circumstances:

#1 = 2*LF (slip independent). Occurs in single-phase motors and 3-phase motors with current unbalance. This is a key torsional resonant frequency to be avoided on turbine generators.

#2 = 2*s*LF = #poles*<slip speed> = “pole pass frequency”. This occurs when there is an asymmetry present on the rotor (dynamic eccentricity varying at 1x, or rotor bar defect). Also is occurs on salient pole sync machines due to reluctance variation within a pole.

#3 = 1*s*LF. This is (imo) the least well known. Large torque oscillations during DOL start predicted at this frequency by Kraus' model

We are familiar with the fact that current creates vibration of various kinds at double the frequency of the current or flux. The first two follow that pattern, but the last doesn’t.
#1 = 2*LF oscillation arises from 1*LF varying stator current and flux.
#2 = 2*s*LF oscillation arises from s*LF varying rotor current and flux.
#3 = s*LF oscillation arises from s*LF varying rotor current and flux combining with dc component of rotor current or flux.

The first two involve a sinusoid squared, which gives double frequency. The last involves product of terms of two different frequencies (s*LF frequency and 0 frequency), which explains why there is no doubling. There is some more discussion of the combination of s*LF and dc (0 frequency) in the attachment above.

=====================================
(2B)+(2B)' ?