Resolving High Vibration On a

Vertical Pump
Mustafa Shalabi Sami Al-Mubarak
Sankar Ganesh Ahmad F Bn Nawawi
Baker Hughes, a GE Company Saudi International Petrochemical Company
 Presenter/Author bios
Sankar Ganesh – Technical Leader, MENAT Mustafa Shalabi – Lead MD Engineer
Sankar is the Technical Leader for GE Bently Mustafa works as Lead Machinery Diagnostics
Nevada Machinery Diagnostics Services in the Engineer in Bently Nevada General Electric
MENAT region. company, Saudi Arabia since December, 2012.
He received a Bachelor of Mechanical He is a certified vibration analyst CAT III by
Engineering from Bharathidasan University, Mobius Institute. Mustafa graduated in 2006 with
India, in 1993. B.Sc. degree in Mechanical Power Engineering
He has over 20 years experience in vibration from faculty of engineering, Alexandria university,
field and 9 years with GE Bently Nevada, Egypt. He has been working in vibration
including rotating equipment balancing, diagnostics of rotating equipment for over 10
vibration analysis, diagnostics and root cause years. He started his career in 2007 as a condition
analysis. monitoring engineer for rotating equipment.
He has published case studies in METS and In addition to his work as machinery diagnostics
Turbomachinery symposium. engineer, Mustafa is working as a Machinery
Diagnostics training instructor in Bently Nevada.
 Abstract
There is a single stage vertical pump installed in Acetic Acid area in a petrochemical plant. The unit is vertical high
pressure (HP) reactor feed pump (High speed single stage, 230 kW, 12338 rpm & 3.444 gear ratio) driven by induction
motor through a flexible shim pack coupling.

High vibration levels were observed on the motor, pump and structure for more than 2 years with 1X dominant
vibration (1X motor dominant frequency). Assuming that the issue is completely structural as this motor was running
well in the past, there were some structural modifications have been carried out at site with inadequate engineering
calculations. The modifications were welding a total mass of 100 kg onto the structure and the installation of braces
to the structure which did not yield any positive results. A systematic structural measurement along with the
Operating Deflection Shape carried out at site with the help of experts did not reveal a significant issue with the
structure itself except for the structural natural frequency being slightly closer to the running frequency of the motor.
The timely right decision by the experts to carry out an onsite balancing on the motor, which was intended to remove
the excitation force, yielded good results which exposed a mistake in the balancing activities carried out by a local
vendor at work shop.

This case study is designed to outline how the high vibration issue was successfully diagnosed using various tests on
site including the operating deflection shape, the root cause for the high vibration and finally how it was mitigated.
 Machine Details
▪ The unit is vertical HP
reactor feed pump
▪ Single stage, 230 kW,
12338 rpm & 3.444 gear
ratio
▪ Driven by induction
motor (260 kW, 3585
rpm, ball bearings)
▪ Flexible shim pack
coupling
 Problem Statement
▪ High vibration levels on the motor, pump and structure for more
than 2 years (11 mm/sec Peak at steady state)
1X motor dominant vibration component.
▪ Bump test showed natural frequency at 43Hz
▪ Many trials were done by the End User to reduce the high
vibration:
• Total mass of 100Kg were added to the structure
• Additional supporting braces
▪ Vibration levels remained as high as the original measurements
 Data Analysis– Coupled Run
▪ High overall vibration levels with
1X motor dominant frequency
▪ Transient Start-Up / Shut-Down
data showed that the unit is
running well above the 2nd critical
speed
▪ The highest amplitude while
passing the 2nd critical frequency
(70 mm/sec Peak)
Data Analysis– Coupled Run “cont’d”
1X Polar-MOB
Data Analysis– Coupled Run “cont’d”
1X Polar-MIB
 Data Analysis– Solo Run
▪ High vibration at same locations
as in coupled run
▪ High residual unbalance response
on the motor while passing 2nd
critical frequency
▪ Bode & polar plots confirmed the
residual unbalance response
 Data Analysis-Structural measurement
Coupled Run Solo Run
mm/sec Peak
Overall 1Xmotor Overall 1Xmotor
Str#1 12.9 11.2 11.5 11.2
Str#2 8.5 6.2 6.7 6.2
Str#3 1.3 0.2 0.4 0.2
Str#4 12.5 11.2 11.6 11.2
Str#5 9.1 6.7 7.3 6.7
Str#6 1.7 0.5 0.7 0.5
Str#14 11.2 10.5 10 9.9
Str#15 3.1 0.9 1.5 0.9
 Data Analysis-Structural measurement
Coupled Run Solo Run
mm/sec Peak
Overall 1Xmotor Overall 1Xmotor
Str#7 13.7 12.4 12.6 12.4
Str#8 9.8 7.3 7.8 7.3
Str#9 1.4 0.7 1 0.7
Str#10 12.7 11.5 11.9 11.5
Str#11 10.2 6.8 7.3 6.8
Str#12 1.6 0.7 0.9 0.7
Str#13 10.1 10.1 10.4 10.3
Str#16 4.1 0.9 2.4 1.2
Data Analysis-Impact Test
Data Analysis-Impact Test “Cont’d”

Point#5 48.5Hz
55Hz

Point#5 Showed Natural Frequency Close to The Running Speed

Data Analysis-Coupled Run ODS (Operating Speed 59Hz)

Circular Motion

Solo Run also showed the similar circular motion

 Action Taken
▪ Balancing shot was done by adding total mass of 178 g on the
motor coupling hub (2 bolts were attached to the hub)
▪ Significant decrease in vibration levels on the motor and
structure.
▪ The required final balance shot was 210 g as per vector
calculations (at this balance plane; hub).
▪ There is no access to add final balance weights on the hub.
▪ It was decided to check the motor cooling fan as a balance
plane in order to examine motor response to weights at NDE
side.
 Action Taken-Cont’d
Weights Added on the Coupling Hub Red: After
Blue: Before
Action Taken-Cont’d
 Action Taken-Cont’d
Coupled Run Red: After
Blue: Before
 Action Taken-Cont’d
Coupled Run Red: After
Blue: Before
 Post Analysis- ODS after balancing at 59Hz)

Twisting Motion
 Conclusion & Recommendations
▪ The extremely high vibration levels on the motor, pump and
structure -mainly due to motor rotor residual unbalance.
▪ ODS showed circular motion due to unbalance response.
▪ In-situ balancing reduced vibration levels to acceptable values.
▪ After balancing, the twisting motion remained due to the
natural frequency of 55 Hz.
▪ Structure design to be reviewed with the OEM to shift the
natural frequency above 20% operating speed.
 Lessons Learned
▪ Extensive measurement and detailed analysis in the field
helped to identify the root cause quickly - substantially
reduces the duration of unplanned downtime.
▪ Workshop balancing to be done competently to avoid
unnecessary balancing exercise
▪ A verification of vibration levels at workshop would have
avoided downtime and time for troubleshooting exercise in the
field.
Thank You
Appendix
Problem Statement
Spectrum- 1X
Problem Statement
Bump Test
Action Plan
Data Analysis– Solo Run “cont’d”
1X Polar-MOB
Data Analysis– Solo Run “cont’d”
1X Polar-MIB
Data Analysis-Structural measurement
 Data Analysis-Impact Test “Cont’d”
Different Locations

Other Points Showed Natural Frequency BUT Away from The Forcing Frequencies
Data Analysis-Solo Run ODS (Operating Speed 59Hz)

Circular Motion
 Conclusion
▪ ODS showed circular motion due to unbalance response.

▪ Extremely high vibration levels were observed on the motor,

pump and structure with 1Xmotor dominant frequency are
mainly due to motor rotor residual unbalance.
 Action Taken-Cont’d
Weights Added on the Coupling Hub Red: After
Blue: Before

OB-Solo Run
 Action Taken-Cont’d
Total of 186 g of Old Wrong Weights Removed from the Fan Red: After
Blue: Before

OB-Solo Run
 Action Taken-Cont’d
Total of 186 g of Old Wrong Weights Removed from the Fan Red: After
Blue: Before

IB-Solo Run
 Post Analysis- ODS after balancing at 55Hz)

Twisting Motion
 Conclusion & Recommendations
▪ ODS showed circular motion due to unbalance response in
addition to slight structural twisting motion.
▪ After balancing, the twisting motion remained due to the
natural frequency of 55 Hz as indicated from the impact test
data.
▪ Structure design to be reviewed with the OEM to shift the
structure resonance frequency to at least 20% above or below
the operating speed frequency.