cheking hot spots in a stator core 1

[uhpo]

Hi everybody, hope this new year would be full of new knowments and sharing of them.... i have a stator core i would like to cheek searching for hot spots on the core, firstly i would present our friend, this is an motor 1500 rpm 850 kw and 660 V, and cause an ausence on manteinance, the bearing on drive end broke, and the rotor touch for long time the stator core ending this process on a burned stator, now i would like to cheek the stator core before rewinding, the winding is on wire with wounded coils.....
My questions are this:
I think about pass through four wires with current through the inner side of stator core, on 4 points on 90 degrees, and put some current, but my problem is that i don´t have magnetic information, just the current per phase, so the first question is how could i know how much current i must circulate onto the circuit.
The idea is to cheek the temperature on the core with a thermographic camera, in order to cheek the hot spots, but again a new question, in example if i see a average temperature on 40º C and some points with 60ºC, could we think this core is useful or not, what are the limits?
Do you know if there is some literature that speaks about this or some related standard.

thanks for your atention

 

[bartb102]

We had a stator core tested on a 750KW 120RPM generator, core is 10ft diameter. For the test a machine that could produce 2000A AC into a short circuit was brought in. Stator winding had been removed (damaged by a fire) so core was empty. Lead from tester was dropped into one slot, return lead around the outside (single pass through the frame). Current was turned on, it took less than 10 minutes for the hot spots (bad core) to bcome 10 to 20C hotter than rest of the core. Very visible on the thermal camera. We had a spare core that had no hot spots when tested.

Doesn't matter how much current you use, as long as it is enough.

 

[uhpo]

thanks barb,
i agree with you, but what i´m not sure is how much diference could be admisible, i mean, if we have a diference on 6 º C i think it´s not significative, cause on normal situation, this core would be at 50º C on the average of the core...but on the hot spot, you would have 56ºC , this is my thinking but i´m not sure i´m right
do you use only one turn onto the slot, i read that it´s needed to use 4 turn separates 90º mechanical..really i feel a little confused.. not about how to do the test, but how to know if the core must be changed.

thanks in advance

 

[bartb102]

If a spot increases 6C above it's surrounding steel with just the test excitation it is a bad spot, will continue to get hotter as time goes on. With the 2000A drive it took 10 minutes to get the bad spots hot. It only took 5 minutes to start seeing the difference.

Use judgement for what the ambient air may be doing.

Yes, machine we used was a single turn, but at 2000A the cable had to be put in two adjacent slots due to physical size.

I'm looking for a picture of the test, can't find one.

 

[electricpete]

I think about pass through four wires with current through the inner side of stator core.....first question is how could i know how much current i must circulate onto the circuit.

EASA Suggests a target backiron flux density of 85,000 lines per square inch (roughly 1.3T) for a core test.

For “typical” motor core iron, this flux density of 85,000 is achieved with an excitation H = 9 amp-turns per inch (this comes from a commonly-referenced curve originally published in IEEE 432).

The "per inch" part refers to per inch of MPL = mean path length of flux. Let D be the diameter to the midpoint of the backiron (halfway betwen bottom of slot and outside of core). Then MPL = Pi * D

Amp-Turns = H*MPL = 9 * Pi * D ~ 28 * D

For 4 series turns, you would target Amps = 7*D

There is a lot more that can be said. How much voltage does it take to produce this current? It can be estimated of course but if you have variable voltage source you may not care.

If you know actual flux density from winding configuration, you may want a different target flux density to represent actual flux density or some multiple of it (1.05 often mentioned). But sharpening your pencil may not be productive because you are stil relying on the 9 Amp-turns per inch based on typical iron which may not be representative. So if you want to get even more precise you can use a search coil ... from which we can calculate flux density more directly.

Some good references are any EASA document on core testing, or IEEE432, or Jim Oliver's EPRI paper.

If more questions, let us know

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

 

[electricpete]

It may be obvious, but maybe I should clarify: since I have expressed H as amp-turns per inch, then D must also be expressed in inches to use the formulas above.

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

 

[electricpete]

i see a average temperature on 40º C and some points with 60ºC, could we think this core is useful or not, what are the limits?

The standard limit is a 10C rise from core hottest spot to core average (or core minimum, depending on reference). In your case 60-40=20 would exceed that limit. The limit must go along with a test procedure that gives sufficient flux density (discussed above) and duration. Duration typically at least 20 minutes.


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

 

[uhpo]

Thanks for all this information, i would study your formulas, and try to undertand... on other hand, the number of a 10º C rise form hottest spot to core average is a good reference, could you indicate in which standard i could finde, maybe on the named above IEEE 432.
THANKS and for sure i would have some more questions but first i want to study it.

 

[uhpo]

ok,i was estudying all this documents, EASA and IEEE, and they was of interest, but now i try to apply to my motor..... really i´m interested on doing this test before take away the burned winding, cause if i must change the stator core, i would prefer to buy a new stator... for this i´m thinking....only is an idea, maybe an crazy idea, that i could open one group of coils far away from the area where i see the posible damage, and in this coils inject the nominal current or close, and see how the core is heating, so if there is some problem on the core i would see heating some areas...let me know what do you think about this maybe crazy idea

 

[electricpete]

I think in theory it might work, but the problems for me are:
It varies from standard practice.
It will not be easy/straightforward to calculate required current)*).
Results will not be compare-able against standard tests.

* I don't think full load current works for a variety of reasons including the following:
- If you use single phase vs three phase, that's a difference.
- You have to consider not only reluctance of the iron (as in loop test) but also reluctance of the airgap. So calculations are not so easy, and more amp-turns will be required.
- Even if you did successfuly re-create operating flux density, it is somewhat a different pattern than the loop test flux density. And while the loop test flux density makes a rough attempt to recreate operating flux density, the pattern has some differences which will be tough to reconcile. And for better or worse, the standards are based off of loop test flux density patterns.

By the way, unrelated comment: you have probably seen that core loss test (watts per pound) is often used to evaluate core condition in conjunction with hot spot tests. One frequent spec is no more than 5% increase in watts-per-pound from before burnout to after burnout. The absolute limit (rather than change) varies from 2 watts-per-pound to 7 watts-per-pound depending on that steel. (Very generally higher end of the range for cores that are old, or cheap, or small motors which tend to use cheaper materials... very generaly lower end of the range for cores that are new, quality materials such as those often used on very large motors)


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

 

[uhpo]

Hi electricpete.... i understand your point, but in anyway i wont try to fullfit any standard, just i want to be sure i have a good stator core.... but i didn´t see why it´s worse to use in example 4 coils of one phase to check the heating of the core, really i think we don´´t need to have the nominal flux, cause really we need to see the diference between diferents areas of the stator core....
maybe i´m a little square head, i really i appreciate your information....but really i see dificult to calculate the current to use and the number of turns, and really... i thought this system is more easy... i have a doubt about the other coils of the winding, i asumms that must connect them on short circuit..how do you see it....

thanks a lot for your time, ideas and experience

 

[electricpete]

but really i see dificult to calculate the current to use and the number of turns

I see it the opposite. It is fairly easy to calculate required number of turns and realtionship to back-iron flux density for wires looped around core, using simple toroidal model.

If you energize the windings, it’s very tough to know what you have.

You mentioned full load currents, but those include load current component it has no direct relationship to flux density.

OK, so instead you may inclined to focus on operating no-load current which is related to magnetizing current which is related to airgap flux density, which can be related to backiron flux density. BUT, operating conditions have rotor installed, you can't do a test with rotor installed so operating no-load amp-turns for given flux density tell you very little about flux turns to get comparable flux density with rotor removed. The exciting amp-turns required for given flux density will be much higher with rotor removed for test than with rotor installed.

If we use winding, we don’t have ability to change number of turns. If we use cables looped around core, we do.

If we use winding, you may be inclined to energize only one phase (especially since you mentioned damaged coils). If you think about the flux pattern, the stator core teeth inside the span of coils in the energized phase will link a smaller flux than the other teeth. In contrast for loop test, all teeth see the same.

Finally and most importantly to me, we need some frame of reference to compare the data. Using standard test methods provides that. Inventing a new test will give you data, but what it means will be tougher to understand.

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

 

[uhpo]

hi, finally i do a loop test, and finally i find some hot spots, so now it´s time for the question, the hot spots are due to the touch between stator and rotor, so.. it seems it´s surface, i try to clean it with a grind, but only takes more time to heat it up....
does somebody have ideas about how to do it???

 

[ScottyUK]

On large generator cores techniques many techniques have been tried with limited success. The key is to avoid any metallic connection between laminations, so very careful grinding may achieve this, or fine grinding with a final acid etch to remove any smearing. Obviously it is very important to neutralise any acid once the repair is complete if the latter is used.

A partial re-stack of the core is likely to yield the best repair as the individual laminations can be repaired and possibly re-distributed within the core.


----------------------------------

If we learn from our mistakes I'm getting a great education!

 

[electricpete]

Chuck Yung from EASA has published a number of copyrighted IEEE papers about core repair using acid solution.

I googled and found this:

http://www.docstoc.com/docs/51966924/An-AlternAtIve-method-to-lAbor-IntensIve-core-rep

I don't know what docstoc is and I'm not sure if I would download, but you can scroll to see the entire paper. And it's not copyrighted. This particular one came from Iris Rotating Machinery Conference 2007. IRIS used to make these all freely accessible (papers from each IRMC filed by year). Now after beingtaken over by Qualitrol their website is re-organized. There are still some papers but I couldn't find this one.

There are other methods as well. Spreading, cracking etc.

If it doesn't clear, it is possible the damage has spread further in and cannot be repaired without at least partial restack.



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