Sign up Calendar Latest Topics Donate
 
 
 


Reply
  Author   Comment  
electricpete

Sr. Member
Registered:
Posts: 528
Reply with quote  #1 

I have in the PAST posted a dynamic absorber case study regarding an 1800rpm rod drive motor generator where the generator had a resonance just below running speed:
 https://www.machineryanalysis.org/post/show_single_post?pid=1295307026&postcount=22&forum=473361

This NEW post is not about that machine in the case study, but about an identical sister machine (same design motor, generator, and support structure) except the sister machine does not have the dynamic absorber installed.

I noticed the directionality increased while that sister machine was unloaded.  I included in the tabulation below and the graph attached the vibrations for position 3 (generator inboard) and position 4 (gen outboard) H and V readings for 1X and OA (overall).

What is noteable to me is that under no-load condition (the 2nd reading), the horizontal 1X vibrations significantly increases but the vertical 1X stays the same (or even decreases).

Like everything, I have a theory about it.   But I thought it would be interesting to ask what you guys think might be the explanation.

Posn    Load    NoLoad Load

3H1X  0.194   0.354   0.215

3HOA  0.202   0.355   0.22

4H1X  0.209   0.349   0.236

4HOA  0.216   0.354   0.241

3V1X  0.0243 0.0102 0.0192

3VOA  0.0426 0.0483 0.0382

4V1X  0.0745 0.0651 0.0651

4VOA  0.0789 0.0711 0.0708

 

 
Attached Files
pdf MG22_DirectionalityNoLoad.pdf (44.88 KB, 15 views)

ivibr8

Sr. Member
Registered:
Posts: 133
Reply with quote  #2 
Pete
Is this machine on sleeve bearings?

I ask because I like to start simple. If vibrations change (in this case load vs. no load), either force changes or the stiffness changes  OK, there are other factors but again I start simple

If these are supported on sleeve bearings, I can see how a change in rotor position (e.g. magnetic center) could ?perhaps? affect stiffness/damping.
Don't know the construction but usually the vertical has higher K values compared to horizontal

Regards
Jim
OLi

Sr. Member
Registered:
Posts: 1,519
Reply with quote  #3 
Is  this one of those that have a interesting flywheel attachment?
__________________
Good Vibrations since early 1950's, first patented vibrometer 1956 in the US.
http://www.vtab.se
Walt Strong

Sr. Member
Registered:
Posts: 716
Reply with quote  #4 
The sister machine was reportedly operating near resonance at 1X shaft speed, and you reduced vibrations with the TMD. It is reasonable to expect that this machine also has a natural frequency in the horizontal direction that is close to shaft speed, although probably not identical to the sister machine. The sister machine had vertical cracks in the concrete pad foundation, and anchor bolts were not evident to restrain the steel skid/base. If this is the case on this machine, then nonlinear vibration behavior could be expected. Drive torque (load) could alter the support stiffness at the top of the concrete foundation. In addition to the static torque change a dynamic torque at 1X shaft speed may change as a result of shaft misalignment and poor gear coupling condition (tooth wear or lubrication).

I would consider the following actions:
Conduct and ODS test with and without load to detect motion across skid frame and foundation.
Measure airborne ultrasound near shaft coupling with and without load.
Measure shaft alignment and inspect coupling condition.

Walt
electricpete

Sr. Member
Registered:
Posts: 528
Reply with quote  #5 

Quote:
Oli wrote: Is  this one of those that have a interesting flywheel attachment?

Yup, that’s the one, Oli. 

Quote:
Walt wrote:The sister machine was reportedly operating near resonance at 1X shaft speed, and you reduced vibrations with the TMD. It is reasonable to expect that this machine also has a natural frequency in the horizontal direction that is close to shaft speed, although probably not identical to the sister machine. The sister machine had vertical cracks in the concrete pad foundation, and anchor bolts were not evident to restrain the steel skid/base. If this is the case on this machine, then nonlinear vibration behavior could be expected. Drive torque (load) could alter the support stiffness at the top of the concrete foundation.

Bingo. Yup, that’s exactly what I was thinking, Walt.  It doesn’t seem like an increase in excitation (based on the verticals), so it seems like it must be tuning of a resonance. What could possibly tune the resonance between loaded and unloaded… a stiffening effect of the support under load which moves resonance above running speed while loaded and drops it back down near resonance when unloaded.  I can’t think of any other explanations.   (I'm not really suspecting the coupling since the vertical is not increasing when the horizontal increases) 

The stiffening under static load is never something I’ve had an opportunity to observe directly before, but I’ve read about it in other contexts (couplings, sliding bearings come to mind).

We might try a bump test loaded and unloaded (assuming we can manage that).   That would be mostly as a curiosity to explain what we’ve seen.   In the event we decide to use the dynamic absorber (we already have it designed/approved based on the first machine), this information might suggest some care in tuning the dynamic absorber that might not otherwise apply if the resonant frequency was constant.

Quote:
Jim wrote:Is this machine on sleeve bearings?

I ask because I like to start simple. If vibrations change (in this case load vs. no load), either force changes or the stiffness changes  OK, there are other factors but again I start simple

If these are supported on sleeve bearings, I can see how a change in rotor position (e.g. magnetic center) could ?perhaps? affect stiffness/damping.
Don't know the construction but usually the vertical has higher K values compared to horizontal

No Jim, this particular machine has only rolling bearings.   But I understand why you ask, it's a very similar thought process as described above. I wouldn't be as surprised by these readings if we had sleeve bearings... the stiffening phenomenon is more familiar for sleeve bearings than for machine support stiffness. 

electricpete

Sr. Member
Registered:
Posts: 528
Reply with quote  #6 
I think the explanation is fairly straightforward, but I'm curious if anyone else has ever actually seen a similar case where resonant frequency apparently shifted with machine loading level, or even between off and on (assuming the change couldn't be explained by other factors like sleeve bearing characteristics, fluid mass draining out of a vertical pump when secured etc).
Curran919

Avatar / Picture

Sr. Member / Supporter
Registered:
Posts: 476
Reply with quote  #7 
Quote:
Originally Posted by electricpete
this information might suggest some care in tuning the dynamic absorber that might not otherwise apply if the resonant frequency was constant.


In theory (linear), the resonant frequency has no bearing on the tuning of the elimination frequency of the system. If the machine system changes, that is okay. Only the absorber system and its attachment point matters when determining the elimination frequency (antimode).

In practice, absorbers can be over-tuned or under-tuned to maximize width of the elimination frequency, depending if the base system is subcritical (positive separation margin) or supercritical (negative separation margin). If you are just tuning to the lowest response, it doesn't really matter.
spciesla

Sr. Member
Registered:
Posts: 92
Reply with quote  #8 
Assuming that the vibration is primarily 1X, then phase measurements would be very telling in regards to resonant behavior at the different load conditions.
electricpete

Sr. Member
Registered:
Posts: 528
Reply with quote  #9 

Quote:
Steve wrote Assuming that the vibration is primarily 1X, then phase measurements would be very telling in regards to resonant behavior at the different load conditions.

Thanks Steve, I didn’t think about that.  I think we’ll try to get bump test in loaded (easy in the current condition) and unloaded or shutdown (can’t get until later) conditions.  To me the bump test gives more relevant info than the phase. If we have problems performing / interpreting a running bump test than we’ll think about phase.


Quote:
Curran wrote:In theory (linear), the resonant frequency has no bearing on the tuning of the elimination frequency of the system. If the machine system changes, that is okay. Only the absorber system and its attachment point matters when determining the elimination frequency (antimode).

In practice, absorbers can be over-tuned or under-tuned to maximize width of the elimination frequency, depending if the base system is subcritical (positive separation margin) or supercritical (negative separation margin). If you are just tuning to the lowest response, it doesn't really matter.

Thanks Curran.   I mentioned care in tuning as a placeholder to remind myself to think some more about the implications of this unique situation.  And I'm glad you chimed in to help me think through that.   I agree with your analysis.  I think the minimum amount of care required (assuming we do first tuning while unloaded or secured) is simply to recheck the tuning while loaded (which is how the machine runs most of the time).   If we’re being extra thorough, it might be worthwhile to bump / monitor near the base of the absorber in the final loaded condition and look for the zero at the tuned frequency and the spacing to running speed of the two peaks on each side.  Because as you point out the zero is less centered between peaks the more the machine-alone resonant frequency is different than the tuned resonant frequency.  We know the normal loaded condition already has machine-alone resonant frequency somewhat different than tuned frequency since that normal loaded condition is the less-directional condition prior to installing the absorber.  And machine stiffness might also drift if it is affected by cracks in the foundation.   I expect we will have some loaded running  bump test results available long before we install the absorber... when we get those it might be interesting to speculate about effect of possible loosening of absorber mount and/or machine stiffness. Right now I don't have enough info to even begin thinking about that,  but at least we designed a high mass into the absorber to give us some margin to accommodate those drifts if they occur. 

Another thought process I have is that final tuning accounts for some unknowns in the design process, such as the simplifications we make that turn the 3-d continuous potentially-nonlinear system into a 1-d discrete 2dof linear system.  One example is we assume the machine moves purely horizontal with no rocking.  I don’t really know the effect of these simplifications but it is intuitively satisfying to know we have verified the tuning in the final condition (loaded).  Perhaps some of those simplifications result in the real system acting slightly different than predicted our 2dof model such that maybe change in support stiffness does slightly alter the real system's tuned zero frequency and nearby peaks.   I don’t know how likely that is, but rather than trying to analyse it, we’ll simply rechecking the tuning in the final loaded condition to improve our confidence in the final tuning.

In the end, I know the above discussion is probably making things way more complicated than needed.  The primary strategy to address all of that is simply to recheck tuning in the final loaded condition, and to continue to monitor vibration over time and re-evaluate if/when we see an increase.

Previous Topic | Next Topic
Print
Reply

Quick Navigation:

Easily create a Forum Website with Website Toolbox.