×
INTELLIGENT WORK FORUMS
FOR ENGINEERING PROFESSIONALS

Log In

Come Join Us!

Are you an
Engineering professional?
Join Eng-Tips Forums!
  • Talk With Other Members
  • Be Notified Of Responses
    To Your Posts
  • Keyword Search
  • One-Click Access To Your
    Favorite Forums
  • Automated Signatures
    On Your Posts
  • Best Of All, It's Free!
  • Students Click Here

*Eng-Tips's functionality depends on members receiving e-mail. By joining you are opting in to receive e-mail.

Posting Guidelines

Promoting, selling, recruiting, coursework and thesis posting is forbidden.

Students Click Here

Jobs

Chiller Waveform Question
2

Chiller Waveform Question

Chiller Waveform Question

(OP)
I obtained voltage and current waveforms from a centrifugal chiller of about 200 HP.  It was running at about 70% full load.  The voltage waveform looks fine. RMS voltage for A phase is 476.4V and the peak is 664V.  The current waveform for each phase is flat topped at the positive and negative peaks.    The A phase RMS is 161.5 amps.  The peak  is 230 amps.  The duration of the positive and negative flat peaks is about 2.9 miliseconds each.  Total harmonic distortion is 1.53% for voltage (A phase) and 7.21% for current (A phase).    Here is the question:  What characteristic of the motor produces the flat topped current waveform?

RE: Chiller Waveform Question

The problem is one's current sensor has gone into satuation.

RE: Chiller Waveform Question

(OP)
No chance--I used two different types of sensors to ensure there was no clamp problem.  One type of clamp was rated for 1000 amps.  The other was rated for 2000 amps.  The load was only about 150 amps.  I do not believe saturation could occur here.  Also, I read the current waveforms on a CSI 2120 spectrum analyzer and a Hioki Power Analyzer.  Both provided consistent results.

RE: Chiller Waveform Question

2.9ms? Is that a problem for you? It could be a lot of things. Is it only on Phase A or is it present on all 3 phases?

Eng-Tips: Help for your job, not for your homework  Read FAQ731-376

RE: Chiller Waveform Question

(OP)
Jraef--
Except for the flat current waveform, I have no evidence of a problem.  The vibration data I took on the machine looks great.  The three phase current and voltage data is well balanced. The flat topping does occur on all three phases. I suspect there  is no problem here but I am curious why the flat topping occurs in the current waveform.

RE: Chiller Waveform Question

Is it possible you are using a digital instrument near the bottom of it's range where the step due to bit size is significant (maybe 1% or more of your peak value). The software draws a straight line between samples (linear interpolation) so where the slope is high you don't see it (no stairstepping due to linear interpolation), but where the slope is lowest (at the peaks), the current does not change enough to toggle the last bit.

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

RE: Chiller Waveform Question

(OP)
Pete
As I mentioned, I have taken this test on two instruments using two different clamps. The CSI 2120 spectrum analyzer autoranges to prvide the best signal resolution.  Here it is seeing an RMS voltage of .15  This is certainly nowhere near the bottom of its range.  The second instrument offers a 500 amp range for a 1mv/amp clamp.  A 150 amp load will be accurately resolved.

RE: Chiller Waveform Question

Is this chiller cooling the motor with the refrigerant?  If so  I could imagine the V/f to be slightly too high so the motor is just barely saturating but with large cooling available not showing up otherwise.

Just a guess though.

Give the maker a call and ask them.

Keith Cress
Flamin Systems, Inc.- <http://www.flaminsystems.com>

RE: Chiller Waveform Question

(OP)
itsmoked

The chiller is hermetic; it does cool the motor with refrigerent.  I do a lot of chiller testing and see this issue somewhat regularly, although not usually to the extent exhibited by this machine.  Could you possibly explain in a little more detail how this would produce the appearance of saturation.  Thanks for your insights.

RE: Chiller Waveform Question

Help am I figuring this right? I get a flat spot of about 60 electrical degrees. That's about one third of a peak.
However an RMS of 161.5 times root 2 = 228.4
Compared to an indicated 230 peak volts, that is an error of 0.7%
Your RMS:Peak ratio is consistent with an almost perfect wave form. Try measuring the current with an old amprobe with a D'arsonval meter movement. If an old amprobe indicates any where near 161.5 amps, assume a good waveform and check the settings to see how you have caused two different instruments to clip the displays the same way. The old D'arsonval meter movements responded to average current or voltage rather than RMS. The scale was adjusted to take into account the ratio between the average value of a sin wave and the RMS value of a sin wave. (1.1 as I remember). This was called the form factor (from wave form). The indicated value only equalled the RMS value with a sign wave. It didn't take much wave-form distortion to throw the reading off.
respectfully

RE: Chiller Waveform Question

Although THD is lower on the current, your crest factor is more out of whack on the voltage than on the current:

CF(V)=664/476 = 1.39
CF(I)=230/161.5=1.42
For a sinusoid we expect crest factor sqrt(2)=1.41 so the current is closer to sinusoid.

In fact, one would have thought a current sinusoid with top chopped off would have crest factor < 1.41.  The deviation from sinusoidal is maybe a little more complex than just chopping off the top... I'm imagining the slope of this waveform as it crosses 0 must be less than we expect for a perfect sinusoid, lingering at lower values bringing the rms down. The slope coming off of that flat peak higher rate of change than a sinusoid to get down closer to 0 faster.  I don't exactly know what to make of that but just trying to reconcile the high crest factor with the chopped-off-peak description.

Here's a thought (might be way off base).  Even thought the voltage THD is lower than current THD, the problem might still be associated with the voltage (power supply) and somehow magnified in the current? (I would think this is more common in capacitive loads).  I have heard that during power quality surveys, it is not uncommon for a small Vthd to cause a larger Ithd (although I would tend to associate this with capacitive loads which amplify the higher frequency harmonics).

I assume you would have told us if vfd.  I really can't put my finger on what about the power system would cause it so just throwing out some off-the-wall stuff...   Anything weird in the power system?  Attached capacitors?  Unusual upstream transformer configuration? Nearby electronic power supplies? Or maybe not.

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

RE: Chiller Waveform Question

Hey waross - I wasn't ignoring you, just typing at the same time (I must be a lower typer than you).

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

RE: Chiller Waveform Question

Hey electricpete
I'm not a very fast typist. I think I started a lot sooner.
I was thinking along the lines of generators with non sinusoidal voltages depending on the distribution of the windings. Motors act as induction generators. I thought that this may be an explanation until I checked the peak ratio.
Remember, 2.9 mili-seconds is almost one third of a half cycle, and with a perfect wave-form ratio?

Did I ever tell you about the time I taugh a theory lesson right out of the text book on the ratio of line to line voltage to line to neutral voltages to a class? Then I sent them out to the shop to connect a bunch of small dry type transformers in Wye-Wye and verify the lesson. I got in trouble more than once that way. It wasn't long before the quicker students were all over me, because the voltage ratios were obviously too far off to be explained by acceptable experimental errors. We ended up putting a scope on the transformer outputs. It's so long ago that I don't remember which was which. That is the wave form from line to line versus the wave form from line to neutral. I clearly remember one wave form being normal and the other waveform having a flat spot on the zero line.  The distorted wave-form was a fairly good looking half wave and then a few electrical degrees of zero, and then a good looking negative half wave followed by a few electrical degrees of zero. The disparity in the voltage ratios as indicated by the D'arsonval meters was readily apparent.
I later ran into the same effect on a pumping station start-up, amidst a very puzzeled group of engineers.
The effect disappears as soon as a load is applied.
respectfully

RE: Chiller Waveform Question

Sorry testtech this isn't saturation.. I just looked again at your original post.  In saturation the current wouldn't flat top it would spike!

A saturated supply transformer would result in your motor's voltage flat topping.

I'm still thinking probe.

How about this.
The probe sees a large flux change during all of the current cycle except at just the peaks where essentially over a short period there is NO flux variation.  The output (a voltage) flattens out briefly during this point in the waveform.  Essentially DC very briefly.  (Not to be mistaken with saturation)

You could check this if there is any resistance in the power circuit.  Wouldn't need much.  The contactor's contacts for one phase would be plenty.  Watch the voltage drop across the contacts.  This will give the current without any magnetics in the picture.  (Note: it might get ugly if the contactor opens while you're looking at the 0.5 Volt scale)

Keith Cress
Flamin Systems, Inc.- <http://www.flaminsystems.com>

RE: Chiller Waveform Question

(OP)
HI All:

Thanks for your thoughts.  

Electicpete-I wonder if you are not right about the small distorion in the voltage being the culprit.  I have lots of saved chiller voltage/current signatures. I will go through some tonight and see if this makes sense.  I will let you know.

RE: Chiller Waveform Question

Hello testtech
I have a few questions.
Are there any capacitors in the system and if so are you looking at just the motor current or the combined motor-capacitor current?
Do I understand correctly that you have a flat line across the top of your wave-form for 2.9 Milliseconds. Almost a third of the width of the half cycle?
Can you describe the voltage wave-form please?
Thanks.
respectfully

RE: Chiller Waveform Question

(OP)
Hi Waross:

There are no capacitors.  If there were a way to post images, I could attach the waveform.  However, if you cut off the positive and negative peak and drew a straight line to connect the shoulders, you'd have this waveform.

I reviewed lots of data that I have here with chiller waveforms.  There is no obvious relationship between voltage waveform crestfactor and THD and current crestfactor and THD.  Certainly, I found quite a few that had the same voltage crestfactor or lower than the case here but had  normal looking or just slightly distorted current waveforms. The same applies to THD.  

So, at this point, I cannot explain the flat topping and high THD for this machine.



RE: Chiller Waveform Question

The problem might be one purely of instrumentation.  I wonder whether the preamp of the analyzer is being overloaded, similar to when a mic overloads a preamp on a tape deck or mixing board.  Does the amp probe give you the options for the ratio of V / A, such as my Fluke 80i-110s does:  100 mV/A  and 10 mV/A?  If it does, does the waveform look more "complete", less clipped, when using lower settings?  If it does, you are clipping the inputs of your analyzer.

BK

RE: Chiller Waveform Question

(OP)
Here is a picture of the current spectrum and waveform from the CSI 2120.

http://i1.tinypic.com/s1u4ip.jpg

Frequency is shown here in CPM (sorry, I neglected to set unit to Hertz).  Thus, 3600 CPM is 60 Hz, as you all know.

RE: Chiller Waveform Question

What are the dominant frequencies of your current distortion?  

RE: Chiller Waveform Question

There is a fair bit of low order harmonic in the current waveform. Hard to tell - looks like third or fifth: my guess would be fifth. Can your instrument give individual harmonic magnitudes as well as THD? I think you will find that one harmonic dominates the spectrum. You can almost pick out the peak and trough of the harmonic with a little imagination.

----------------------------------
  I don't suffer from insanity. I enjoy it...

RE: Chiller Waveform Question

(OP)
DPC and ScottyUK-
60 Hz- 154 A
180 Hz- 3.5 A
300 Hz- 9.4 A
420 Hz-4.5 A
660  Hz 1.9 A
2454 Hz- 1.7 A
All other harmonics are fractional amps.

RE: Chiller Waveform Question

Hi testtech
Well your pictures convinced me that it is not a clipping problem. The cut-off is not straight enough for clipping. Clipping is an instrumentation problem and does not cause harmonics.
I think you have enough harmonics to explain the waveform. The question is Why the harmonics?

Please note friends, I'm over my head here and my following comments are questions rather than a definitive answer based on first hand experience.
I notice a particularly strong 41 st. harmonic if I have calculated it correctly. The harmonic to the left of the 150 kCPM line. I make it 147.6 kCPM  or 2460 Hz., if I have calculated correctly.
I'm wondering, how many bars does the rotor have? 42 bars and a running speed of 3514 rpm may be generating 41st harmonics if a bar had a high resistance weld, or an inclusion in the casting.
If I'm completely out to lunch please be gentle with me.
respectfully

RE: Chiller Waveform Question

(OP)
waross

I am attaching a vibration spectrum.  The vibration amplitudes are extremely low, so I have not seen them as evidence of an electrical or mechanical problem.  However, the key frequencies are all harmonics of 120 Hz.  Unfortunately, I did not acquire vibration frequencies high enough to capture the rotor bar pass frequency.  However, the rotor bar test showed showed 2x pole pass side bands at about 50 dB down, so I think the high resistance weld is less likely.  Could an inclusion in the casting cause the symptoms we are seeing?  A rotor issue might make sense, since all three phases are impacted.

http://i1.tinypic.com/s25iqc.jpg

RE: Chiller Waveform Question

Hello testtech
With a cast squirrel cage, an inclusion in the casting will give one bar a higher resistance. If this is the case the unsymetrical bar may be generating the 2454? Hz. frequency. I estimated 2460, which is a multiple of 60 Hz. If the 2454Hz. is an accurate frequency and is NOT a multiple of 60 Hz. it may be caused by a bad bar sweeping the windings at the rotation frequency rather than the synchronous frequency.
Does any one have any idea how many bars this motor may have?
I'm guessing about 40 more or less. That would correlate with the 2400+ frequency.
respectfully.

RE: Chiller Waveform Question

I'm guessing that the current waveform is due to harmonics.  The high 5th and 7th is typical of 6-pulse rectifier distortion.  You probably have similar distribution of harmonics on the incoming voltage.  The current waveform looks suspiciously like 5th harmonics as Scotty suggested.

RE: Chiller Waveform Question

Hello testech.
What is the speed of the motor? Is this a dual speed motor?
I have seen a number of chillers using two speed motors so that they can slow the compressor to get energy savings. If the motor is a 4 pole/ 6 pole motor usin PAM techniques for generating the pole changing, you can get very strange current waveforms.

If the motor is delta connected and the motor is over fluxed, you will get current peaks coinciding with the crest of the voltage waveform across the windings. This will occur at 30 degrees before the phase - Neutral crest and 30 degrees after the phase - neutral crest. i.e. at 60 degrees and 120 degrees. This could explain the current waveform and certainly the high fifth harmonic. Is the motor delta connected?? I see that the line voltage is 476 volts. If the motor is wound for say 440 volts or even 460 volts, you could be saturating the iron on the voltage peaks, especially if the motor is cooled by the refridgerant. Where the motor is liquid cooled, it is common to increase the flux in the iron (or reduce the amount of iron for the same flux)

Best regards,

Mark Empson
http://www.lmphotonics.com

RE: Chiller Waveform Question

(OP)
HI All

I don't know how many rotor bars.

Voltage THD is 1.55%.  60 Hz-476V
3rd-.84
5th-6.97
7th-.76
11th- 1.63
13th-1.13
41st-.43

Motor is 2 pole, <3600 RPM. Should be single speed. It is star-delta connected.

Marke--If the motor is overfluxed, does this have any implication for reliability, especially when operated for long periods at full load?  I would guess over heating is not an issue here.  This client is trying to push off a 10 year tear-down for another year.  Except for this issue, I see no reason not to wait another year.

RE: Chiller Waveform Question

Could the 2.454kHz be a frequency imposed on the network from other sources such as the switching frequency of a VFD?

RE: Chiller Waveform Question

I have been thinking about my last post and have a correction. The frequency of a bar problem would not be related to the number of bars, but may be related to the number of coils in each winding. Probably not.
There are to many millions of 460 volt motors running on 480 volts for this to be an issue. This is standard industry practice and within the allowable motor tolerance.

dpc makes a very good point. Can you do a test on the voltage without the chiller on-line/ The chiller may be the victim of power line distortion rather than the cause.

sed2developer has a good point.
We had a discrepancy on the 41st harmonic. Can you verify the exact frequency? 2460 is 41st harmonic, 2454 may be externally generated.
respectfully

RE: Chiller Waveform Question

(OP)
waross

This is the 41st harmonic. It seems to me that a rotor bar issue that produced magnetic field imbalances would show up as side bands or been seen through demodulation processing, but not as a harmonic of 60 Hz.  Further, it is unclear how high frequency current distortion that is not present to a significant extent on the voltage (where it is .43V)  can produce the flat topped current waveform.  

At this point, the the concept described by Marke may be most relevant to the flat topping.  I hope he or someone else can weigh in and elaborate a bit.

As a digression, the most frequent source of flat topping I see is in electrical transformers that are running at or above full load.  The core saturates and secondary voltage and current show flat peaks.  This should not be happening here since the chiller is running at about 70% load.

RE: Chiller Waveform Question

Hello testech

You have confirmed that the motor is connected in delta.
When I look at the current waveform, it is apparent that the whole waveform is distorted, not just the flattened top.

The current flowing in phase 1 is made up of the sum of two currents, the current flowing through a motor winding betwteen phases 1 and 3, and the current flowing through a motor winding between phases one and 2. These two currents are displaced by 60 degrees from each other. If you are able, measure the current waveform flowing in one winding. This current will be less than the line current, but I expect that this current will show the peak on the crest of the waveform.
This flux saturation of the iron will result in an increase in the heat dissipated in the iron, but provided that there is adequate cooling, there is no issue. In an air cooled motor, the onset of saturation will cause a rapid rise in frame temperature.
It is common for very small air cooled motors to be operated in the region of the onset of saturation. This increases the heat, but due to the ratio for volume to surface area, the effective cooling is much higher and no problem results except where these motors are operated above their design voltage. Submersible pumps also operate with a much higher flux density than air colled motors because they are liquid cooled.

I look forward to seeing the winding current waveforms.

Best regards

Mark Empson
http://www.lmphotonics.com

RE: Chiller Waveform Question

(OP)
Marke:

I posted two images of currents.  One is 3 phases in a single plot.  It occurs just above the one you downloaded.  Let me know what you think.

RE: Chiller Waveform Question

Hello testtech

The current waveforms are the line curents and are what I would expect to see for a delta connected motor that is a little overfluxed. you can see a peak occuring at 60 degrees and another at 120 degrees coincicing with where the peak would be in the current flow in the individual windings.

Best regards,

Mark Empson
http://www.lmphotonics.com

RE: Chiller Waveform Question

(OP)
Thanks everyone.  

RE: Chiller Waveform Question

I think Mark figured it out and testtech waded through all the possibilities to find that right answer.

That's a great case study.

I think the the explanation is a little more complex than was mentioned.

If you were looking at an individual leg current of an UNLOADED motor, we would expect to see spike from saturation right at the "90 degree" sin wave peak as described above.  This is because the saturation occurs at highest flux, highest flux occurs at excitation current, and total current is equal to excitation current in the unloaded motor.

But if you are looking at a loaded motor such as posted, total current is not the same phase as excitation current. Total current leads excitation current by an angle (90-arccos(p.f.)).

A minor correction but still probably similar result.

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

RE: Chiller Waveform Question

Just to be clear, 90-arccos(p.f) is close to 90 degrees for a fully loaded motor.  For this motor p.f. angle looks like maybe 20 degrees so the spike in the leg current from saturation would occur approximately 70 degrees after the peak in the leg current.   

Then combining leg currents to phase currents.  The peak of the phase as mentioned differs 30 degrees from the peak of each of the peaks of the two associated legs.

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

RE: Chiller Waveform Question

It looks to me that this curent distortion is produed by the low order harmonics, 3rd, 5th and 7th. the 3rd harmonic is mainly produced by winding distribution and saturation of the magnetic circuit; the 5th and 7th mainly due to uneven leakage of flux due to a reduced and uneven airgap. It does not matter if the motor winding is wye or delta connected although normally hermetic motors are designed for wye start, delta run. The more you increase the operating line voltage the more distorted current will result.

RE: Chiller Waveform Question

If you look at it only in the frequency domain, it doesn't matter whether wye or delta.  However if you look at it in the time domain, you expect to see the current spike from s of saturation at a given point in the timewaveform (at the current peak if motor is unloaded and we are looking at leg current).  Where in the time waveform it shows up depends on the connection.

I do not think that winding distribution would result in 3rd time harmonic of current in the absence of saturation.

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

RE: Chiller Waveform Question

Well, I tried to recreate the waveforms of this case from the harmonics with some assumptions about power factor using a math program Maple.

http://home.houston.rr.com/electricpete/thetapfrun2.PDF

If you scroll through the graphs, you see the the first graph (MagnetizingCurrentsLeg) shows magnetizing current which was created with phase relationships to put the "spike" associated with saturation at the peak of the fundamental magnetizing current, which is where I think it should go.

The second graph "TotalLegCurrents" shows the sum of the above magnetizing currents plus a load current (2 graphs = 2 different legs).  More detials is provided in the third graph titled Aphase_exciting_load_total_legcurrents which shows for one LEG (I shouldn't have labled it phase), the magnetizing and load components and their sum.

The power factor determines the relationship between the magnetizing component and the load component (the phase difference and the magnitude ratio).  Power factor angle in this file (run2) was assumed to be 10 degrees.

The last plot (PhaseCurrent) is determined as the difference of two leg currents (as we expect in delta). This plot has an appearance somewhat similar to what was posted by testtech.

Now in another run (run1)
http://home.houston.rr.com/electricpete/thetapfrun1.PDF
I changed the powerfactor to 35 degrees.  

The results are kind of intriguing.... the LEG currents very closely resemble testtechs results for this configuration.  The phase current does not.

There is one error I made in both files. Looking at the phase relationship between load component and fundamental exciting component, the load component should lead the exciting component by 90-thetapf.  It looks like the opposite. I don't think this affects the final shapes (flip it mirror image around the peak).  I'm a little bit confused why my equations didn't produce the phase relationship I was looking for:

Iam(t):=A1m*cos(1*w*t)+A3*cos(3*w*t)+A5*cos(5*w*t)+A7*cos(7*w*t);

Iaload(t):=A1L*cos(w*(t+Pi/2-thetapf));

Both A1m and A1L are positive.
The fundamental Iaload should lead by Pi/2-thetapf according to the equations but according to the graph it looks to lag. Any ideas where I went wrong?




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

RE: Chiller Waveform Question

Actually the magnetizing current in the first example (run2) doesn't have a pretty shape to look at to show relationship of fundamental to the "spike".  The fundamental is too small and the waveform doesn't look realistic to me (I expect one spike at the peak of the fundamental).

The magnetizing current in the second example (run 1) has the shape I think it should have.  The second example (run 1) also produces a waveform which much more closely looks like testtech's.  

The second example (run 1) would make sense if we were looking at a wye motor.  But not if star-delta meaning star-start and delta run with the phase (not leg) currents measured while running.  Can you clarify the configuration again and did you measure phase or leg currents?

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

RE: Chiller Waveform Question

"The second example (run 1) also produces a waveform which much more closely looks like testtech's."

(the waveforms I'm referring to are labeled as leg currents, not the final phase current).

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

Red Flag This Post

Please let us know here why this post is inappropriate. Reasons such as off-topic, duplicates, flames, illegal, vulgar, or students posting their homework.

Red Flag Submitted

Thank you for helping keep Eng-Tips Forums free from inappropriate posts.
The Eng-Tips staff will check this out and take appropriate action.

Reply To This Thread

Posting in the Eng-Tips forums is a member-only feature.

Click Here to join Eng-Tips and talk with other members!


Resources