Dear All

Recently Bearing Inspection and Dry Gas Seal replacement activity has been carried out on a Motor driven Centrifugal Compressor during a planned shutdown. The Compressor was operating smoothly with vibration values at all four probes well within 20 microns prior to shutdown.

The Operating Speed of Compressor is 15712 rpm. High Vibration Alarm is at 50 microns and Trip at 65 microns. Calculated 1st critical of Compressor is 5000-5700 rpm. The Compressor is equipped with Bently Nevada NSV type proximity probes and proximitors. The Shaft Diameter at Probe Target Area is 3 inches.

During Start up, Vibration values on both probes at Non Drive End reached 120-130 microns at around 4000 rpm causing the compressor to Trip.

Bode Plot for both Non Drive End probes showed nearly flat Vibration Vs RPM curve during startup and coast down. In fact during coast down from around 3000 to below 100 RPM, the vibration values at both Non Drive End probes remained nearly constant at 80-85 microns on one probe and 90-95 microns on other.

Waterfall plot revealed that vibration was almost entirely at 1X component during startup and coast down.

Further it was found that even with Compressor standstill, vibration values at both probes at Non Drive End was 12 microns. Vibration Values at all other Probes on Compressor Train in standstill condition is 1 micron. Historical vibration Data was checked and it was found that previously with Compressor standstill vibration was used to be 1-2 microns.

NDE Probes were connected to DE Proximitors and it was found that vibration values in Compressor Stand Still condition swapped 

Further it was found that with Compressor Stand Still, whenever one probe is disconnected from Proximitor or removed from Compressor Body and connected to Proximitor, the vibration on other probe reduces to 1 micron.

With Compressor Standstill, all Compressor Bearings RTDs were disconnected and Probes and Proximitors were replaced with like to like but problem persisted.

With Compressor Standstill, one probe was replaced with 8 mm Dia probe without changing Proximitor (keeping NSV Proximitor), vibration values at both probes reduced to 1 micron with Compressor Stand Still.

 

.

 

 

FSX

Original Post

There is a minimum distance between transducer tips.  Keep in mind that a conical (more or less) field comes out of the probe tip. The transducer manual gives guidance on the distance and I have attached the applicable page showing mounting requirements. Observe that the minimum shaft diameter for 5 or 8 mm radial transducers is 51 mm (2.0 inches). If your shaft diameter is less than that then cross talk between the radial transducers is likely. If substantially more than 51 mm then cross talk between the radial transducers is far less likely.

Do you also have a axial position probe that could be a contributing factor?

Attachments

Hi John

Thanks for useful information provided. The Shaft Diameter at radial probe location is 3 inches which I believe is far more than what is given in Bently Document for NSV probes. Shaft Material is AISI 4130

We didn't disconnect Axial Probes to see the effect of these. I have attached the Probe Mounting Drawing for your review.

What we are not understanding is that this problem was not there before as I mentioned that vibration Data during previous two trip events for this Compressor in July 2018 and December 2017 indicate that at 0 RPM vibration on these two probes was 1 micron.

Also can this cross talk be contributing factor towards high 1X that we observed during startup attempt as explained in previous message ? I mean if some AC signal is generated due to cross talk  in standstill can this cause high 1X observed during startup.

Can shaft be a contributing factor

 

Attachments

Photos (2)

Recently Bearing Inspection and Dry Gas Seal replacement activity has been carried out on a Motor driven Centrifugal Compressor during a planned shutdown. The Compressor was operating smoothly with vibration values at all four probes well within 20 microns prior to shutdown. Did this require lifting a top cap or removing bearing support or removing probes? Assume motor/gearbox/compressor? Was the seal replacement done on the NDE of the compressor that has high vibration?

Even with the 12 microns (0.472440945 mils) at rest, this value alone will not change the 1X values enough to keep machine from tripping, your measured values are twice your alarm values. 120 microns=4.7 mils

Seems odd the vibration values at 4000 are that high when natural frequency is 5000-5700, change in stiffness/bearing clearance?

I assume that the vibration levels are monitored continuously when unit is started, is there a time delay for the alarms to let the machine pass thru the critical speed or the machine is designed not to exceed 50 microns on startup?

Dave

Hi Dave

Many Thanks for sharing the views. Below is answer to your queries

The Dry Gas Seals at both Ends (Drive End & Non Drive End) replaced.

Removal & Installation of Dry Gas Seal requires removal of Vibration Probes (Radial as well as Axial), both halves of Bearing Shell/Carrier as well as both halves of Bearing Housing. (The Bearing Housing is located in Compressor End Wall with a guide)

Its Motor, Gear Box and Compressor.

Kindly note that a second start attempt was made with one Probe, Extension Cable and Proximitor replaced with 8mm 9meter set. No AC signal observed with Compressor standstill.

Upon restart, Vibration Values at both probes at NDE reached to 130-140 microns at around 4000-4200 rpm causing Compressor to Trip.

 Review of Data Plots i.e. Bode Plot, Water Fall Plot and Spectrum indicated that 1X component remained below 25 microns during second start attempt.

Further there is no clear dominant frequency in Water Fall and Spectrum Plots. In fact we have observed a scattered spectrum (with peaks up to more than 15 times running speed)

The DC Gap (Shaft Average Centre Line) was also abnormal indicating a shaft average center line moving down during ramp up as speed increased and it was found moving up during coast down as speed reduced. The Shaft Average Center Line movement was normal at Drive End showing shaft moving up nearly vertically as Compressor ramped up and Vice Versa during Compressor Coast down

During Coast down vibration values around 80 microns observed at low speed (below 400 rpm) with 1X less than 10 microns in that range

Currently alarm and trip for this Compressor are at 50 microns and 65 microns respectively. We have verified from system 1 that there is no trip multiplier during startup.

Appreciate if you please look into the above findings and let us have your inputs. Kindly note that non availability of Compressor is causing production loss and therefore you are requested to have a look into this on priority and advise the possible way forward.

Due to certain limitations with access to server, I am still unable to have a copy of vibration data plots but I will try my best to share these as soon as I get

 

We have been able to start the compressor by disconnecting one of the two probes at Non Drive End and making it NA in Bently Nevada and System 1. 7m NSV probe has been disconnected and made NA whereas 9m  8mm probe (along with extension cable and proximitor) has been kept in service. Maximum vibration during startup on single NDE probe reached up to 80 microns and in steady state it is 28-30 microns

As per original configuration, Compressor would trip if 1 out of 4 probes reaches trip and any other reaches alarm.  After disconnecting one probe we have changed voting logic to trip the compressor in case there is alarm on any of the remaining 3 probes.

Another observation is that with NSV probe disconnected and 8 mm probe in service, Keypahsor at Compressor NDE has started giving erratic reading

If understand your recent post you got the machine running?

As per original configuration, Compressor would trip if 1 out of 4 probes reaches trip and any other reaches alarm.  After disconnecting one probe we have changed voting logic to trip the compressor in case there is alarm on any of the remaining 3 probes. Is there a time delay, like wait 3 seconds then have the alarms become active?

If shaft is moving up and down on startup / coastdown, oil whirl?, assembly problem? I assume you are using prox probes only for data collection, no portable data collector with casing readings?

Microns are the units to cause trip, but voltage is the source of the signal, if the the voltage from the prox driver is erratic then your reading will be erratic. Have you monitored the prox driver output during the 80 micron excursion or anytime at all

Any electrical work done in proximity of probes or wiring going to the probes, vfd recently added, possible noise being induced by nearby cables, pinched cable, pin hole from welding slag....

The probe pair shown in the 3944 RPM plot is showing a scratch or possibly a situation of “bleed” from the keyphasor (Kph).  Is that negative going spike apparent on other plots of probe pairs that are involved in the trip.  The actual dynamic amplitude without the negative spike is likely down around 4 um.

I have seen instances where cables in a cable trap are extremely close to the Kph signal cable and this causes the Kph to bleed into other signals.

 

Dear John/Dave

Thanks for your responses. 

Proximity Probes were removed, sensitivity check done and were reinstalled

No VFD added or electrical work done.

This Compressor was taken for maintenance first time since commissioning and it was found that all the probes installed on Compressor are NSV probes whereas configuration in Bently was for 8 mm probe. At the time of installation all the probes on Compressor (Radial, Axial and Key Phasor) were set at 8 volts (mid of linearity range for NSV probes) instead of 10 volts. However Rack Configuration was kept unchanged i.e. Rack is still configured for 8 mm probe.

John Appreciate if you please further elaborate this "Keyphasor Bleed" When we started the Compressor with one probe disconnected, the keyphasor has gone erratic. Form your point I think may be keypahsor signal had been interfering with probes

 

The rack being configured for an 8mm probe vs. the Nsv probe won’t matter.  The important parameter is the scale factor which is 200 mV/mil or its metric equivalent.

I made a type in my previous reply saying “cable trap” when I should have said cable trap.  What I meant to not have a radial probe cable in close proximity to a Kph cable.  I don’t recall the BN exact recommendation but you generally don’t want to bundle the cables together.  Having said that, I see an angular displacement between the spikes on the time domain signals.  Where is the physical Kph notch relative to this probe pair?  There is a minimum spacing, again see the PDF I supplied earlier. 

 

By the way, your one plot has a machine speed of 3944 RPM while you earlier mentioned the compressor speed is over 15000 RPM.  Is this part of a transient speed start or stop, or steady state data?

Just for some clarification, can you get a screenshot of the Sys1 machine diagram and add it to the thread?

 

Hi John

These plots are transient data plots captured during startup attempt. The Compressor normal speed is 15800 rpm but during start attempt, it tripped at 4000-4200 rpm.

I am unable to provide screen shot of System 1. However Instrument Layout Dwg is attached which will enable you to understand the layout of machine and location of probes.

I have attached Rotor Dwg as well as Probe Mounting Dwg. From the values on these Drawings, the axial distance between Radial probe and Key Phasor is at least 125mm

Attachments

Photos (3)

Hi john

We are already working on acquiring BN services for this issue.

Concentrating on Wiring or Monitor Condition I would like to have your expert opinion on following points

The two probes at Compressor Drive End as well as Axial probes are also connected to same BN card. I believe issue with BN card seems unlikely ?

Throughout the problem, with Compressor standstill, the DC Gap voltage reading on System 1 of both Non Drive End probes (which are causing trip) remain stable. We didn't observe any fluctuation. Any wiring problem either in wiring from Monitor till proximitor or from proximitor to probe should have caused fluctuation in DC gap of the probes ? Or is it possible that there may be a wiring problem affecting only AC signal generated by the probes ?

I believe the point you mentioned about bleeding KPh signal can cause fluctuation in probes signal only when shaft is turning ?

FSX posted:

Hi john

We are already working on acquiring BN services for this issue.

Concentrating on Wiring or Monitor Condition I would like to have your expert opinion on following points

The two probes at Compressor Drive End as well as Axial probes are also connected to same BN card. I believe issue with BN card seems unlikely ?

Throughout the problem, with Compressor standstill, the DC Gap voltage reading on System 1 of both Non Drive End probes (which are causing trip) remain stable. We didn't observe any fluctuation. Any wiring problem either in wiring from Monitor till proximitor or from proximitor to probe should have caused fluctuation in DC gap of the probes ? Or is it possible that there may be a wiring problem affecting only AC signal generated by the probes ?

I believe the point you mentioned about bleeding KPh signal can cause fluctuation in probes signal only when shaft is turning ?

Taking your questions one by one:

The two probes at Compressor Drive End as well as Axial probes are also connected to same BN card. I believe issue with BN card seems unlikely ?

I don't believe this will matter subject to how the channels are configured.  In that respect I don't have a copy of the rack configuration software to do a test configuration.

Throughout the problem, with Compressor standstill, the DC Gap voltage reading on System 1 of both Non Drive End probes (which are causing trip) remain stable. We didn't observe any fluctuation. Any wiring problem either in wiring from Monitor till proximitor or from proximitor to probe should have caused fluctuation in DC gap of the probes ? Or is it possible that there may be a wiring problem affecting only AC signal generated by the probes ?

If the compressor is at a standstill (as you state) the gap voltages should essentially be stable.  There might be some small variation which is one of the reasons I suggested making sure your lube oil pumps are off.  I have seen some variation with pumps on, and perhaps more so in tilt pad bearings although I've never done a study to prove that to myself.

With respect to a wiring problem causing a fluctuation in DC Gap, it depends on how you define "wiring problem."  Problems like a bad proximitor, bad cable, loose connections, moisture in connections, etc. may cause gap voltage fluctuations.  Something like a ground loop would be more likely to cause an issue in the AC voltage component.  In that respect usually with the machine stopped, a spectrum will show line frequency and 2X line frequency.   

I believe the point you mentioned about bleeding KPh signal can cause fluctuation in probes signal only when shaft is turning?

This generally is the case.  Looking at the output side of the proximitor, if a signal cable and the Keyphasor (Kph) cable lie along side each other (in a cable tray for instance), the Kph spike may show itself on the signal cable.  But if the machine is shut down, the Kph spike would not be present so it would not show.  Having said that I have seen cables from many machine share a cable tray and sometimes even encountered a spike from an unrelated machine.  However, usually this was due to poor installation, hence the reason it is a good idea to follow the recommendations of BN when installing the cabling.

Now, a question for you...do you get that spike on the waveforms when the machine is shut down?

 

Add Reply

Post

HACKATHON in association with IMC-2019

×
×
×
×