Surge Caps and Lighting Arrestors on 5kV motor 2

[rockman7892]

I have an installation of a new 350hp 4.16kV motor in which the motor termination box contining the surge capacitors and lighting arrestors is too large to fit in the location next to the motor where it was intended.

Because of this several people here have proposed some solutions.

Solution 1 was to eliminate the Surge Caps and LA's all together and therefore eliminate the large terminatation box thus allowing a smaller termination box to be used. I dont feel as if it is a good idea to get rid of these components but dont have enough knowledge to defend my thought. Can someone help me understand why they are needed and point me to some info to back my thought.

Solution 2 was to keep the Surge Caps and LA's in this large box, however mount it elsewhere away from the motor where there is room. We would then come from this box to a much smaller box next to the motor itself to pick up the motor leads. My qustion and concern is, what is the effect of locating the surge caps and LA's away from the motor itself. If this is allowable, what is a safe or acceptable distance or location?

 

[electricpete]

All machines are built to withstand such surges [closing] - even if it results in a transient being up to twice the peak voltage of the applied voltage.

What is the basis for this statement?


=====================================
Eng-tips forums: The best place on the web for engineering discussions.

 

[electricpete]

Sorry, I did not intend to ignore your prior post. That there are capacitors located remotely and it is different than what the IEEE says supports the common theme I have promoted throughout this thread: surge protection is a complex subject. Practices vary. There may be thumbrules that work for some people in some situations but these are not universally agree. It is difficult to make generalizations.

=====================================
Eng-tips forums: The best place on the web for engineering discussions.

 

[electricpete]

The attached analysis represents my simple understanding of why it would be preferred to place the cap at the motor vs the breaker.

It assumes a fairly simple model with the breaker represented as the voltage source, and lumped impedances for the capacitor, the upstream inductance, and the downstream inductance.

Solving simple lumped circuit equations to determine the transfer function Vmotor/Vsource, under some assumptions of high frequency, we find that the motor votlage increases as we increase the downstream inductance and decreases as we increase the upstream inductance.

I would be glad to hear any comments on the assumptions or the conclusion. I am by no means claiming this is a perfect analysis (as I said before I think surge analysis can be much more complicated than I understand... and I am also a little leery of relying heavily on lumped elements to predict wave phenomena).

=====================================
Eng-tips forums: The best place on the web for engineering discussions.

 

[electricpete]

Whoops - I see an error in the sign of my Vc term in KCL. Let me correct it and try again.

=====================================
Eng-tips forums: The best place on the web for engineering discussions.

 

[electricpete]

Here is revised version. The resulting expression for Vmotor is a little simpler - no longer depends on downstream inductance from this revised analysis.

Vmotor is still minimized by maximizing the upstream inductance (between the cap and the breaker). Suggests that cap at the breaker would not be a good idea.

Once again I'll be glad to hear any comments.

=====================================
Eng-tips forums: The best place on the web for engineering discussions.

 

[electricpete]

OK - another possible error is my voltage divider to determine Vc. It is correct if we have no current drawn from the terminal labeled Vm. Probably not a good model. Will try again maybe tomorrow.

=====================================
Eng-tips forums: The best place on the web for engineering discussions.

 

[electricpete]

Attached is yet another revision. I have added the motor impedance into the model. I have assumed the motor behavior is dominated by the capacitance to ground at the high frequencies of interest.

With this model, the result is again very similar to predicted from my first model:
* the motor votlage increases as we increase the downstream inductance
* the motor voltage increases as we increase the upstream inductance



=====================================
Eng-tips forums: The best place on the web for engineering discussions.

 

[electricpete]

typographical correction:
"* the motor votlage increases as we increase the downstream inductance
* the motor voltage decreases as we increase the upstream inductance"

=====================================
Eng-tips forums: The best place on the web for engineering discussions.

 

[electricpete]

Attached I have cleaned up the algebra just a little to make it slightly easier to follow.

(no changes to the model or conclusions)

=====================================
Eng-tips forums: The best place on the web for engineering discussions.

 

[electricpete]

The hard to follow math file can probably be replaced by a simpler discussion:

Let's say I have to build a filter to protect the motor from surges. I am given a single parallel capacitor and a single series inductor. Should I put the capacitor in front of or behind the inductor?

The answer depends in part on how you model the breaker. If you model it as anything resembling a voltage source, then putting the capacitor behind the inductor (cap next to motor) makes a better filter and will protect the motor better.

=====================================
Eng-tips forums: The best place on the web for engineering discussions.

 

[Skogsgurra]

Pete.

You have now moved the question from that of permissible practical changes to an inductive kick-back protection scheme to that of the topology of a general low-pass filter.

In my opinion, you have arrived at an answer that is valid for an LC filter. An answer that we, and all engineers, knew was right from the beginning. It is, after all, one of the classical examples we all met in engineering school - in the earlier grades.

The main problem with your math is that you start with the wrong assumption, that of the surge coming from a voltage source with zero internal impedance. In fact, the source is a current source that is short-circuited up to t=0 and then applied to the cable, caps/arresters and motor.

I would like to see a new analysis starting from those correct assumptions and then see if moving the protective elements along the cable really makes such a catastrophic difference as you seem to think.

I do not have the time right now, but you seem to have it. I will check in later, shall we say in 15 hours? to see what you have found.



Gunnar Englund

http://www.gke.org

--------------------------------------
100 % recycled posting: Electrons, ideas, finger-tips have been used over and over again...

 

[electricpete]

Sticking with your approach:

If we represent the source as current source Is with parallel Zs, we can replace it with voltage source Vs = Is * Zs and impedance Zs in series with Vs.

It looks just like the previous model, except instead of ZLup, we have Zseries = ZLup + Zs.

It suggests an easy way to determine the solution without repeating the analysis.

Define a new parameter Leq =Zseries/(j*w). = (ZLup + Zs.)/(jw) = Lup + Zs/jw.

Now we can see the effect on the solution for any source impedance simply by substituting Leq instead of Lup in the solution.

So let's examine the possible values of Zs

Case 1 – ASSUME that Zs is pure inductive. Zs = Ls*jw.
Leq = Lup + Zs/jw = Lup + Ls*jw/jw = Lup + Ls.
Now substitute into the Vm solution instead of Lup: Leq = Lup + Ls.
The conclusions remain the same.

Case 2 – ASSUME that Zs is pure capacitive. Zs=1/<jwCs>
Leq = Lup + Zs/jw = Lup + 1/(jwCs)/jw = Lup – 1/(Cs*w^2)
Now substitute into the Vm solution instead of Lup: Leq = Lup – 1/(Cs*w^2)
Making the usual high frequency assumptions, Leq>0, the conclusion remains the same.

Case 3 – ASSUME that Zs is pure resistive. Zs = Rs.
Leq = Lup + Zs/jw = Lup + R/jw
Now substitute into the Vm solution instead of Lup: Leq = Lup + R/jw
Note the numerators has not been affected, so the numerator conclusion stays the same (increasing Ldown increases Vmotor).
The effect on the denominator is to add an imaginary component where there was none before. The magnitude of the denominator is sqrt(Real^2+Imag^2). If we look at changes to Lup it will affect only the real part and the results are again the same.

So, even with your suggested changes to the model, the results are the same.

With all that said, I don't think current source is the right approach. Look at any textbook showing a surge traveling down a line and you'll see it modeled as a voltage source on the front end of the line and a voltage output at the end of the line. The voltage spike travels down the line from one end to the other. There is current flow associated with that voltage spike, but it is primarily through the distributed capacitance to ground of the cable, not end-to-end through the cable.



=====================================
Eng-tips forums: The best place on the web for engineering discussions.

 

[electricpete]

Take a look at attached excerpt from "The Vacuum Interrunter, Theory, Design, and Application" by Paul G. Skide ISBN 978-0-8493-9091-3

It is "TABLE 5.6 - The Effect of Surge Suppression when Switching Motors with Vacuum Circuit Breaker Compared to an SF6 or Air Circuit Breaker"

These results are based on typical values. Some things to note from the table:

1 – In the absence of surge protection, the switching-on transient seen at the motor is the same regardless of whether we use vac interuppter or air breakers. The switching off transient (still in absence of surge protection) is only slightly worse with vacuum interrupter (4 – 5 p.u. in .2-.5 usec) than with air breakers (4 – 4.5 p.u. in .2-.5 usec). It would suggest that presence or absence of a vac interrupter may not be appropriate for use as the sole criteria for determining if surge protection is required. As I have said before, it may very well be a practical and useful thumbrule and as such is a valuable contribution. But it is not necessarily the end of the story. We can exceed motor specification levels even without vacuum interrupter **

2 – As others have said, the switching on transient is not as severe as the switching off transient. And I will point out that I have never said otherwise. What I did say was that switching off is not the sole purpose of surge protection as others had implied. I believe that my point is supported by this chart which shows not a huge difference between the unprotected switch-on (4 pu in 0.2-0.5usec) and an unprotected switch-off (4-4.5 or 4-5 pu 0.2-0.5 usec). Again it supports the conclusion that the switching on transient is also important (switching on transient can exceed motor speicification). **


3 – For the protection scheme including MOX and R/C, better protection is afforded when the protection is placed at the motor than at the panel. The preferred location at the motor is consistent with the excerpt from the IEEE Guide. (although I can't say I have ever heard of R / C surge protection on a motor).

4 – The author analysed several other protection schemes at the motor, but didn't even bother to analysi other protection schemes located at the panel. I can't speak for the reason for this omission, but it is certainly possible that he doesn't envision that protection at the panel is a common approach.

** On points 1 (vac interupter vs air breaker) and point 2 (switching off vs switching on), we were faced with a question of whether a change from at most 4 to 5 pu in 0.2-0.5 usec is significant. A very relevant number to bring into this analysis is the motor specification for surge withstand. NEMAMG-1 and IEEE522 specifies that new motors be tested to 2.5 pu or 3.5 p.u. with the same 0.2-0.5usec rise time. So, at 4 pu, the closing transients on air breakers are still beyond what we tested the motor to when new (at the most stringent test level of 3.5). This particular fact does not support the notion that we can ignore the closing transient and does not support the notion that we can ignore the any transient associated with air breakers. This fact does lead us again to the question: what is the basis for the claim that ALL motors are designed to withstand the closing transient ?

=====================================
Eng-tips forums: The best place on the web for engineering discussions.

 

[electricpete]

Sorry, my last paragraph was in error. The 3.5 p.u. test increases to 3.5 pu in a period 0.1 usec to 0.2 usec per IEEE522, so the p.u. values are proabaly not directly comparable in terms of the associated dv/dt. To be honest, I have no idea how we can allow such a large variation in the rise time numbers quoted by both sources since 100% variation ion rise time causes 100% variation in dv/dt.

At any rate, ignore the very last paragraph of my last post.

=====================================
Eng-tips forums: The best place on the web for engineering discussions.

 

[electricpete]

A transition to the simple world of electricpete - what is the importance of dv/dt vs Vpeak?

Going back to my simple model: the wave travels through the winding. If it is traveling fast enough, then the voltage difference is seen between adjacent turns.

So let's say applied voltage step increaes 0 to Vmax in Tstep

Case 1 - time to travel through 1 loop is Tloop = 2*Tstep.
The voltage between adjacent turns is Vmax/2.

Case 2 - time to travel through 1 loop is Tloop = 1*Tstep.
The voltage between adjacent turns is Vmax.

Case 3 - time to travel through 1 loop is Tloop = 0.5*Tstep.
The voltage between adjacent turns is still Vmax.

And of course as we decrease Tloop further, we can never get a difference more than Vmax.

So there is a threshhold effect of rise time. At relatively slow rise times, the voltage between turns depends on the rise time. At faster rise times, we hit the maximum, and the stress is no longer dependent on the rise time.

I am not sure where the 0.1-0.2use and 0.2-0.5sec would be in relation to the threshhold. I guess I could use typical coil dimensions, make some assumptios about mu and epsilon to determine speed of the wave, and try it out. A project for another day.

=====================================
Eng-tips forums: The best place on the web for engineering discussions.

 

[electricpete]

Correction in bold:
And of course as we decrease Tstep further, we can never get a difference more than Vmax.

=====================================
Eng-tips forums: The best place on the web for engineering discussions.

 

[waross]

As we are converned with inductie surges when breaking a circuit, am I oversimplifying to consider a simple DC inductive/resistive circuit? When voltage is applied to an induction, the current increases exponentially depending on the induction and the resistance of the circuit. The voltage never exceeds applied voltage. When the current is interrupted, the voltage is determined by the inductance and the resistance of the circuit. As a vacuum contactor interrupts the current very rapidly and presents an extremely high resistance, the "inductive kick" on de-energization may be many times the applied voltage. The addition of a shunt capacitor provides a path for the current and presents an impedance to a steep wave front that decreases as the wave front becomes steeper.
When considering inductive kick and interuption of current is it valid to ignore the AC and consider the current as fast decaying DC? Am I on the right track here, Gunnar?

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

 

[electricpete]

That simple model does not seem to match up too closely with what we see in my attachment 23 Aug 08 17:32. Unprotected motor with air breaker sees 4 pu closing , 4-4.5 pu opening (rise time 0.2-0.5used). Maybe the data is flawed? The author doesn't explain a lot where the chart came from. Anyone else wants to present a reference... please do.

Returning to my quest to compare the p.u. values in the attachment to the surge test levels.

I mentioned two levels: 2.5 and 3.5 for testing. Actually it is 2.0 or 3.5

NEMA MG-1 specified that the 2.0 is "standard" and the 3.5 is "special"

So, let's just use the standard motor at 2.0 instead of the 3.5 special.

If the thresshold rise time is somewhere up around 5 usec, than rise time is not relevant, clearly the 2.0 that we specify the motor tested to is less than the 4.0. The 4.0 pu stresses seen during closing with air breaker are higher than the test.

But where is the real threshhold time? I would venture to say it must be above 2.0usec. Otherwise why on earth would IEEE522 specify a wide range 0.1-0.2usec (only makes sense if you are below the threshhold where rise time is not important).

But even if we are absurdly generous and use the other assumption that the threshhold rise time is somewhere down below 0.1usec (in which case the IEEE limit makes no sense whatsoever), then the turn stress is inversely proportional to the rise time. So this 2 pu in 0.1-0.2usec gives a voltage stress equivalent to to 4 pu in 0.2-0.4 usec.

Our brand new standard new motor is tested to withstand a stress of 4 pu in 0.2-0.4 usec. And we are going to subject it 4pu in 0.2-0.5us during switching on with an air circuit breaker. And 4-4.5 during switching off with an air circuit breaker. We have exceeded the test stress during switching off and gone right up to the limit when switching on. That doesnt' give me a lot of confidence, considering the normal test strategy (for example hi-pot) is supposed to test beyond the operating envelope when the motor is brand new. so that we have some margin to operate reliably in the field for many years, including allowance for degradation from aging. And let's remember the unrealistically generous assumption of threshhold below 0.1usec.

All things considered, I would say this analysis does support the assertions of the last paragraph of my 23 Aug 08 17:32 post qualitatively (although the quantitative conclusion is different). i.e. The typical levels of stress created even by closing (vs opening) of air circuit breaker (vs vacuum contactors) in an unprotected environment (ie 4.0 pu in 0.2-0.5usec) IS at or above the stress level created during testing of new motors at the NEMA MG-1 "standard" level 2.0 pu.

By the way, my real world experience is that most motors do fine without surge arresters (all of the 4kv motors at our plant). But again, I am just cautioning against generalizations.

=====================================
Eng-tips forums: The best place on the web for engineering discussions.

 

[electricpete]

1 -

http://www.nepsi.com/msp.htm

Due to the wavelength and travel time of voltage surges, the MSP [Motor Surge Protector] is most effective when placed as close as practical to the motor terminals with the ground leads being as short as possible. This will limit the surge voltage seen by the motor to the discharge voltage of the arrester. This should be done for all critical and large medium voltage motors. Where there are many small motors or explosion proof motors in hazardous locations, a single MSP at the motor control center is recommended.

This particular manufacturer of motor surge protective devices clearly indicates that locating the surge protector as close as possible to the motor gives the best protection and therefore is preferred for critical and large motors. They mention some special situations when the surge protector would be located at the MCC... for multiple smaller motors protected by one surge protector (not clear to me how this would work) or explosion proof motors in hazardous locations.
2 -

http://books.google.com/books?id=V1...+‘ON’"&sig=ACfU3U2znF1siLAvlTIb0JMkMktws7pXgg

(ii) Surges generated during a switching ‘ON’ operation
Field data collected from various sources have revealed failure of a motor’s windings, even during an energizing process. It has been shown that a motor may be subject to an overvoltage of 3-5 p.u. with a front time ‘t1’ as low as 1 us or even less (signifying the steepness of the TRV) during switching ON.

The above link goes on to explain the reasons. It has to do in part with timing of the closure of the contacts on the three phases.

=====================================
Eng-tips forums: The best place on the web for engineering discussions.

 

[edison123]

pete

As a rewinder, I agree with you on many motors failing during switching on rather than on switching off or during running. Based on my clients' feedbacks, I would roughly categorize the percentage failures as 80% switching on, 18% during running and only 2% on switching off.

And I have never seen a surge protector located away from the motor or generator.

And I appreciate all your time and efforts in this thread. Wish I could give another lps.