Rotating
Machinery Stability Test Rig
Student:
C. Hunter Cloud
Advisors:
Lloyd E. Barrett, Eric H. Maslen
Project
Start Date: January 1999
Executive
Summary
The continuing high costs incurred by turbomachinery
users due to rotordynamic instability problems provide the main
motivation for this research effort. Unlike most research in this area
which is focused on the destabilizing components within turbomachinery,
this project centers on characterizing and measuring industrial machines’
stability level and margin, i.e., their stability robustness. The
research objectives include evaluating and experimentally verifying
techniques for determining a rotor system’s stability characteristics
through forced response testing. An experimental test rig that simulates
the vibrational behavior of many types of industrial turbomachinery will
be employed. Results are available to all ROMAC members. A variety of
companies are providing support and funding, the primary one being
ExxonMobil Research and Engineering.
Motivations
One of the most prevalent problems with newly
purchased or re-rated turbomachinery is rotor lateral instability, a
condition of rotor operation in which various elements in the rotor
system induce self-excited vibrations. Such vibration can impact project
schedule, limit the machine’s operating range which results in reduced
unit production rates, as well as cause severe failure, threatening
personnel and environmental safety.
In order to avoid such problems, turbomachinery users
are often forced to incur the costs of expensive testing at full speed,
power, and gas density at the machine manufacturer’s facility in order
to simulate the actual operating forces that are expected in the field.
These tests are designed to assess, to some degree, the stability of the
machines. However, these complex tests often do not provide definitive
answers regarding the machine’s stability because of the differences
between shop and field operating conditions and machine installation.
Nor do they provide a measurement of the machine’s stability level and
stability sensitivity versus that predicted through rotordynamic
modeling.
With these current cost and testing limitations, most
research efforts are focused on improving the rotordynamic analysis and
prediction of stability. Because of the documented high modeling
uncertainties associated with components which destabilize turbomachines
(typically components with tight clearances where fluids flow and
rotate, namely, labyrinth seals, bearings, oil seals, wear rings, etc.),
a stability sensitivity analysis is typically performed as shown. In
this type of analysis, the destabilizing forces are varied on the
rotor/bearing system in order to determine the stability threshold.
Combined with the modeling predictions of the components’
destabilizing forces, this prediction allows a stability margin or
"safety factor" to be estimated. However, the stability
threshold which determines this safety factor is also highly uncertain
because, to date, no experimental studies have been performed to verify
such threshold predictions and the accuracy of the sensitivity curve
characteristics.
With little documented experimental correlation
verifying the predictions of stability robustness, and no techniques
available for measuring this robustness on actual machines, two serious
technological "gaps" remain to be examined, in addition to the
inaccuracies of modeling destabilizing components. This project hopes to
help avoid instabilities by focusing on solutions to the first two of
these gaps.
Objectives
The primary objectives of this research project are the following:
- Evaluate methods for measuring stability levels and margins for
industrial turbomachinery.
- Verify these methods experimentally using a test rig with magnetic
actuators which provide measured and controllable forces.
- Develop new techniques for assessing stability margins from forced
response measurements.
An additional objective is to experimentally examine
how non-linearities within a rotor/bearing system affect stability
levels and margins. By examining this area, a greater understanding of
the influence of bearing design parameters and unbalance levels will be
achieved which will provide additional design options.
Measuring stability level is basically a system
identification problem where the parameter of interest is the damping
ratio (or log decrement) associated with each natural frequency/mode.
The accuracy and determination of this measurement depends on the forced
excitation technique (impulse, sine sweep, etc.) used as well as the
damping estimation technique (amplification factor, phase slope, RFP,
etc.) applied to the test data. This project focuses on what techniques
are most practical and accurate for industrial turbomachines.
Experimental Test Rig
The test rig is designed to simulate the dynamic
behavior of an industrial between-bearing turbomachine. Driven by a
variable frequency drive motor, the rotor can reach a maximum speed of
10,000 rpm and is supported by 2.75" diameter fluid film journal
bearings. The bearing housings can accommodate any journal bearing
capable of being tested in the Fluid Film Bearing Test Rig.
Each 8 pole heteropolar magnetic actuator or
"shaker" has 1,400 pounds static force capability and 500
pounds of dynamic force capability at 10,000 cpm.
As part of the effort to understand the influence of
bearing design on stability characteristics, several tilting pad bearing
designs are being tested. A 5 pad, load between pad bearing with rocker
back, center offset pivots and 0.3 preload will serve as the base design
for the testing. Several variations of this base design are being
studied, particularly, with respect to preload, pivot offset, and pivot
design. The figure below shows the stability sensitivity of this base
design as well as that for some of the other bearings in the test
program.
Testing using this experimental rig will help to shed
light on issues surrounding the frequency dependent nature of tilting
pad bearings’ stiffness and damping properties and their influence on
stability characteristics. This influence is illustrated below where
synchronously reduced coefficients, those used by many in industry,
predict much higher stability level and threshold versus the full pad
coefficients, which incorporate frequency dependency.
