Vertical Pump Resonance

Problem
Gus Altimany, Air Products and Chemicals
Patrick Smith, Air Products and Chemicals
Biographies
• Patrick J. Smith is a Principle Engineering Associate - Machinery in
the Operational Excellence Technical Team at Air Products &
Chemicals. He is also an AP Fellow. He is based in Allentown, PA and
has over 35 years of rotating machinery experience.
• Gus Altimany is a Machinery Engineering Manager at Air Products &
Chemicals. He is based in Allentown, PA and has 25 years of rotating
machinery experience.

TURBOMACHINERY & PUMP SYMPOSIA

Pump Overview
• 17- stage vertical cryogenic pump
• Driven by 800 HP variable speed motor
• Pump and motor hard coupled
• Pump operated at a preset speed between 3300
RPM (55 Hz) and 3600 RPM (60 Hz) depending
on process duty requirements
• Motor fitted with DE casing vibration sensor
• Motor fitted with DE and NDE bearing
temperature probes
• Pump mounted in a pit
Pump on test stand in shop

TURBOMACHINERY & PUMP SYMPOSIA

Problem Statement
• During commissioning pump experienced periods of high vibration
• Magnitude of vibration affected by speed, bearing temperature,
ambient conditions, and ice build up
• See next slide for plots

TURBOMACHINERY & PUMP SYMPOSIA

Problem Statement

TURBOMACHINERY & PUMP SYMPOSIA

 Initial Findings
• Vibration all at 1X running speed
• Bump check showed a clear resonance at around 3540
RPM (59 Hz)
• Coast down vibration plot showed possible resonances
at running speed and at around 3300 RPM (55 Hz)
• Five identical pumps at site showed the same behavior
• High vibration was not observed during shop testing
• Pumps were an old design with many references.
However, the new application was a higher power and
used larger, heavier motors.

TURBOMACHINERY & PUMP SYMPOSIA

Initial Analysis
• Rotordynamics reviewed – concluded there were no problems
• Suspected issue with structural resonance. In a vertical pump, the motor is a
large cantilevered mass. The natural frequency associated with a cantilever
mode is sometimes referred to as a reed critical frequency due to the mode
shape being similar to that of a musical reed. If the pump motor combination
reed critical frequency occurs in the speed range of the pump, high vibrations can
occur.
• Motor vibration provides the exciting force.
• Natural frequency is function of stiffness and mass. To change the natural
frequency, the stiffness and/or the mass can be changed.
• Lowering the excitation force can lower the vibration response, but does not
change the natural frequency.

TURBOMACHINERY & PUMP SYMPOSIA

Initial Field Fixes
• Several attempts were made to stiffen the motor/pump
laterally by adding temporary supports. However, this
resulted in no change in the vibration behavior.
• In parallel the motor was field balanced in an effort to
reduce motor excitation force. This resulted in slightly
lower vibrations, but still not acceptable.
• 896 kg of weight was hung from nozzle head, resulting
in low, stable vibration. Note the following:
• Weight of nozzle head = 600 kg
• Weight of motor = 2405 kg
• Weight of smaller motors used in the past < 1500 kg

TURBOMACHINERY & PUMP SYMPOSIA

Path Forward
• The hanging weights were considered a temporary solution.
• Similar vertical pumps were being installed at many other sites.
These pumps used newer, high head/stage hydraulics. This resulted
in significantly less stages, but this only impacted the cold end. The
warm end and motor sizes were similar to the pumps with the
vibration problem.
• An engineering solution was desired that could address the existing
pumps and the pumps that were yet to be installed.

TURBOMACHINERY & PUMP SYMPOSIA

Finite Element Modal Analysis – Simulations
• Components model included nozzle
head, barrel w/mounting flange,
soleplate, bowls, suction bell and
motor
• Further models included simulation of:
➢Nozzle head ice build-up
➢Hanging weights (steel curtains)
➢Horizontal & vertical supports
➢Nozzle head windows
➢Vertical rods support

TURBOMACHINERY & PUMP SYMPOSIA

Finite Element Modal Analysis - Results
• Identified various modes of system Natural Frequencies

natural frequencies: Main focus

were torsional & second bending
• Field System Natural Frequency
(SNF) Tests Revealed Close
Proximity to Design Speed of 59HZ
→ Excitation Leading to High
Vibrations
• Industry Preferred Separation
Margin away from SNF is ±20%
(10% Acceptable)
Mode 10 – Mode 11 –
53.8613 Hz 61.1212 Hz

TURBOMACHINERY & PUMP SYMPOSIA

Attempted Modifications & Effect
Modification What was done Effect Result
Nozzle Head Removed/Cut Reduction of Bending mode up
Windows material from nozzle torsional mode closer to running
head speed
Nozzle Head Added angular rods No change No change
Tie Rods in both tension and
compression
Support Rods Added 4-vertical Vibration amplitude Still within 5%
rods under motor reduction for short margin to system
corners period but trend up natural frequency
w/minor excitation
Steel Curtains Added various Vibration amplitude Tried progressive
hanging weights reduction=F(mass) weights

TURBOMACHINERY & PUMP SYMPOSIA

 Curtains Rods Structure

TURBOMACHINERY & PUMP SYMPOSIA

Adopted Modifications
Modification What was done Effect Result
Engineered Steel Built structure Radial (bending) Torsional mode
Structure around nozzle head reinforcement of remained close to
w/adjustable braces motor DE/nozzle running speed
(horizontal/vertical) mounting flange

Steel Curtains Added hanging Vibration amplitude Optimized on

weights to nozzle reduction below 1000Kg after head
head ribs 10% margin stress analysis

Engineered Hanged detuning Vibration amplitude Required frequent

Vibration Absorber panels (2x head @ reduction @ specific tuning / adjustment
90°) from head frequency

TURBOMACHINERY & PUMP SYMPOSIA

Engineered Support Structure
• Design & fabricate SS structure
around nozzle head. Tried 18 different
variations to find a structure that
meets the design requirements,
considered piping attachment
stiffness, weights, and conducted
sensitivity analysis/checks →
complete integrated comprehensive
analysis

Vertical Posts bolted to

plates welded to soleplate
and located above concrete
– transferring dynamic
loading to concrete
TURBOMACHINERY & PUMP SYMPOSIA
Engineered Curtains Vibration Absorber
Attach detuning
engineered
vibration
absorbers to
nozzle head –
these are not
damping devices
that change
system natural
frequency but
detuning devices
that absorb the
specific HZ
vibration
amplitude

TURBOMACHINERY & PUMP SYMPOSIA

Location of Field Data Vibration Measurements

TURBOMACHINERY & PUMP SYMPOSIA

Field Study – System Dynamic Analysis
• Field data measurements and analysis
• Developed FEA models based on collected data and operating
deflection shape (ODS) measurements
• Higher damping of the system natural frequency mode typically
indicates that this mode will not lead to high vibrations
• Proposed plan was to evaluate simple modifications to the support
structure and measure the effect on the various modes
➢Support structure with horizontal and vertical braces around the nozzle head
➢ Loosen bolts and remove braces → Measure impact progressively
• Alternative plan was to engineer vibration absorber – 2xhead for
directional detuning

TURBOMACHINERY & PUMP SYMPOSIA

Final Solution
• The solution was to leave the hanging weights installed on the
pumps. This was the simplest solution and provided the most
consistent results of all the attempted modifications.
• There have been no structural resonance issues to date.

TURBOMACHINERY & PUMP SYMPOSIA

 Conclusions
• High vibration was due to a resonance problem, but the cause
was different depending on the mounting method of the
pump k

• If the pump was mounted on a steel structure, the

m
structure behaved more like a simple mass and spring
system. In this case the flexibility of the steel structure Mass and Spring
largely determined the structural natural frequency. This is
why this wasn’t seen during shop testing.
• If the pump was mounted on concrete in a pit, the m
structure behaved more like a cantilever spring and the
stiffness and mass of the nozzle head largely determined L I
the structural natural frequency. This is why stiffening the
mounting had little to no effect
Cantilever Spring

TURBOMACHINERY & PUMP SYMPOSIA

Conclusions
• The hanging weights resolution could have been replaced with a
heavier weight nozzle head assembly if the FEA were conducted
accurately during early design phase

TURBOMACHINERY & PUMP SYMPOSIA

Lessons Learned
• When reviewing a pump for a new installation, it is important to
understand there are risks when stepping outside of a suppliers
experience. The pumps discussed in this case study were typically
driven by much smaller (power and weight) motors. The size was a
step out.
• When installing large motors on vertical pumps, consider performing
a system finite-element analysis (FEA) during the design phase to
ensure there are no structural resonances within the operating speed
range. The system should include the pump assembly, supporting
structure and piping connections to the nozzle head.

TURBOMACHINERY & PUMP SYMPOSIA