Case Histories: Applying vibration and related test and analysis methods to solve problems

1. Case Histories: Applying vibration and related test and analysis methods to solve problems by Eric J. Olson – Vice President Mechanical Solutions, Inc. ejo@mechsol.com 973-326-9920 ext. 125 www.mechsol.com Whippany NJ January 22, 2014

2. Morton Effect – Oscillating Vibration Phenomenon on a Double Suction Feed Water Pump Excessive Vibration Amplitude at the Inboard Bearing of a Reactor Feedwater Pump

3. Case Study 1: Morton Effect – Oscillating Vibration Phenomenon on a Double Suction Feed Water Pump 3

4. Background • February 2011 one of the single-stage double-suction pumps in feed water service, driven by a steam turbine, indicated a step change in 1x vibration accompanied by a significant change in performance. The 1x vibration level increased from 1.8 to 6.5 mils pk-pk. Simultaneously, flow rate decreased from 22400 to 21600 gpm. This issue was a result of a broken impeller vane causing imbalance and rubbing. • April 2012, after a major outage, the same pump began increasing shaft vibration at the inboard and outboard end. The pump was completely refurbished during the outage. • The vibration amplitude was not constant. The overall amplitude was fluctuating with a period between 10 to 60 seconds. • The overall vibration was mostly at 1x rpm.

5. Background

6. Vibration Testing – Prox probe outboard bearing housing Vibration amplitude was not constant (prox probes). The overall amplitude was fluctuating with a period between 10 to 60 seconds

7. Vibration Testing – Prox probe inboard bearing housing

8. Vibration Testing RPM fluctuations 4845rpm to 4888 rpm (350 second sample) Average Speed

9. Vibration Testing 0 to 344 seconds

10. Vibration Testing The steady average mechanical imbalance vector combined with a secondary non-constant phase imbalance vector (0 to 344 seconds)

11. Vibration Testing • Morton Effect: the circles are the result of a static imbalance vector (i.e. the steady average mechanical imbalance) combined with a secondary non-constant phase imbalance vector where the secondary vector cycles its phase from 0 – 360 degrees. The secondary (cycling) imbalance vector is produced from a thermal distortion of the shaft in which a localized high temperature area of the shaft travels the full circumference with every vibration swell. The imbalance vectors behaving in this way is what causes the vibration level cyclic trends such as those experienced by the pump. • This phenomenon is typically caused by either a light rub or a “near rub” (boundary lubrication at the “pinch point”) in a synchronously whirling shaft leading to a “hot spot” on the “heavy side” of the shaft that leads to thermal bow of the shaft, changing the “heavy side” location. Such a condition can occur in either a bearing or a seal, but in the case of a bearing the vibration cycles are typically longer (on the order of 5 to 10 minutes) than observed in the present case.

12. Vibration Testing • Morton Effect is dependent on the thermal state, structural internal clearances, and fluid dynamics within the seal cavities of the pump. However, Morton Effect may be affected by changes in pump speed, seal water injection temperature, and flow rate. • Plant personnel tried different ways to eliminate this phenomenon and detected a strong influence on the condensate seal water temperature (seasonal) or by changing the seal water injection flow. However, the results were not always consistent and sometime without any effect.

13. Vibration Testing Morton Effect not detected after slight reduction of seal water injection flow and increment of drain temperature. June 2012

14. Vibration Testing End of Morton Effect after reducing seal water injection flow and drain temperature. August 2012

15. Vibration Testing Start of Morton Effect after slight increment of seal water injection flow and increment of drain temperature. August 2012

16. Vibration Testing No effect on Morton Effect after seal water injection flow increment and reduction of drain temperature. August 2012

17. Operating Deflection Shape (ODS)Testing • Key Symptoms: • Erratic vibration at 1x but not classic imbalance. • Sensitive to seal drain temperature. • What was the root cause? • Used an ODS test to help answer ODS Animation @1x rpm (72.3 Hz)

18. Root Cause, Conclusions, and Solution • After a thorough investigation between site personnel and MSI, it was concluded that the root cause of the Morton Effect was due to a casing distortion that was altering the internal clearances between the shaft and the long seal sleeve. • The torque applied to the hold-down bolts to the pedestals at the outboard and inboard end of the pump was approximately 4,400 ft-lb. The required torque, based on the OEM manual, is 1,800 ft-lb at the inboard end and 550 ft-lb at the outboard end. This would allow the casing to grow axially while it reaches the thermal equilibrium.

19. Case Study 2: Excessive Vibration Amplitude at the Inboard Bearing of a Reactor Feedwater Pump 19

20. Background • In August 2013, the Bravo single-stage double-suction pump in feed water service, driven by a steam turbine, indicated elevated vibration amplitude (over 0.8 in/s RMS @ 1x rpm) when the pump was operating at its maximum speed of 3930 rpm (100% load, 65.5Hz). • The plant decided to operate the pump at 90% load (3510 rpm, 58.5HZ) to maintain the 1x rpm vibration near 0.60 in/s RMS. • Others were recommending that the pump needed to be shut down. • Questions: • Can the pump continue to operate? • If needed is there a fix that can be implemented with the pump operating? • If needed, what is a long term fix?

21. How Natural Frequencies Affect Vibration Typical FFT gs of vibration per pound of force *FRF plot from impact test Forces causing the problem. Will be amplified by a natural frequency *FRF is a Frequency Response Function. Used FFT on log scale to determine resonance was a likely issue.

22. Modal Impact Testing – FRF plot before rebuild Log scale – vibration acceleration (gs) per pound of force 0.000282 gs of vibration per pound of force at 57.5 Hz Frequency Frequency Response Function (FRF) Plot Resulting from Modal Test Cold Non-Operating Condition Before Rebuild Measured at West End Pump Foot Mid-Span in the Horizontal Direction (Horizontal Casing Mode at ~57.3 Hz, 90% of load)

23. Operating Deflection Shape – before rebuild 90% load OBB 1x rpm OB and IB probes 1.8 and 5.4 mils pk-pk ODS Animation @1x rpm (58.6 Hz) Baseline Conditions

24. Conclusions at this point - 1 • The high vibration root cause was predicted to be a soft or sprung foot along most or all of the foot length along with pump’s west side (no issue on east side). • The looseness and/ or excessive flexibility at the foot-to-pedestal-top-pad west side locations enabled the casing to swing in an elongated circular arc, moving on the order of 5 plus mils p-p, primarily in the horizontal direction, with largest motion at the bearing housings. • Although the pump or driver were not in immediate jeopardy, the casing vibration was strong enough that the bearing cavity walls are repetitively striking the shaft, resulting in a very brief impact and rub within the bearing, approximately once per revolution over a small arc (maybe 30 degrees or so) which will lead to premature bearing wear. • The large casing motion can happen when welds of any gussets and cross-members internal to the pedestal develop fatigue cracks and/ or gradually corrode and break free. If there was not a soft foot, this situation could be tolerated with no problem until the next planned outage, but the soft foot allows a strong (on a relative basis) contortion of the pedestal, facilitating greater casing vibration.

25. Conclusions at this point - 2 • Fix the flexible foot and the extra vibration caused by this condition will be minimized to acceptable levels. • The reason for the sudden appearance of the soft foot is unknown. It is possible that action by piping loads during load swings contributed to this condition. A fatigue crack loosening the entire top plate of the pedestal or the foot gussets on the west side is another possibility. • Recommendations: • The foot bolts should be tightened. This was done immediately to limit further increase in vibration and gradual damage to the pump IB bearing.   • Tightening of the foot bolts will shift the casing natural frequency from 57.3 HZ (about 90% of running speed) upwards, and the natural frequency may still be in running speed range (max rpm 3930rpm, 65.5Hz). This may not necessarily be a problem depending on damping and system stiffness.

26. Modal Impact Testing - FRF after rebuild/foot bolt torquing Log scale – vibration acceleration (gs) per pound of force 0.00057 gs of vibration per pound of force at 67.5 Hz Frequency Response Function (FRF) Plot Resulting from Modal Test Cold Non-Operating Condition After Rebuild Measured at West End Pump Foot Mid-Span in the Horizontal Direction (Horizontal Casing Mode at ~67.5 Hz, 100%)

27. Operating Deflection Shape – after rebuild at 100% where the NF is now located Turbine Bearing 1x rpm OB and IB probes 2.1 and 0.7 mils pk-pk ODS Animation @1x rpm (65.5 Hz) After Modifications

28. Results – after second round of tests - 1 The pump foot west side “soft foot” condition appears to be corrected upon examination of the ODS 1x animations. This was largely due to the installation of the dowel pins on the pump feet. As a result, the pump structural natural frequency (which involves significant horizontal relative motion) had changed from ~57.3 to 67.5 Hz due to the general increase of the system stiffness which also allows the pedestal better restrain the pump casing motion. The overall vibration levels as measured on the pump casing are generally low, on the order of ~0.1 in/s rms. Seismic spectra also indicate that the pump is not significantly exciting the 67.5 Hz structural natural frequency at the 65.5 Hz (3930 rpm) running speed. ODS animations at the 1x running speed confirms that the casing motion is significantly low relative to the rotor vertical indications and the IB turbine bearing.

29. Results - after second round of tests -2 Vibration levels at 1x rpm on the OB and IB probes changed respectively from 1.8 and 5.4 mils pk-pk before the rework to 2.1 and 0.7 mils pk-pk after the rework. Furthermore, an examination of the proximity probe spectrum after the rework for the IB probe does not indicate any serious detrimental condition involving the coupling such as severe misalignment or imbalance.

30. Final Recommendations The root cause of the soft or sprung foot on the pump’s west side should be investigated during the next outage opportunity to avoid a chronic problem. Plant personnel should check for pedestal weld cracks and any fatigue damage to supports / cross members in the pedestal structure. Such damage may contribute to pedestal flexibility and would certainly be beneficial to correct if present.