Effects of Clearance on Damping in Fluid Film Bearing

Spectra Quest, Inc.

8201 Hermitage Road
Richmond, VA 23228
(804) 261-3300
www.spectraquest.com

June, 2007

ABSTRACT
Clearance of a fluid film bearing is a critical parameter affecting the eccentric ratio as well as
dynamic characteristics of the rotor system. Four shaft/rotor configurations have been tested on
Spectra Quest’s Rotor Dynamics Simulator. For each test configuration, three bearing sets with
different clearances were used to investigate the effects of clearance on damping in the fluid
film bearing. It is observed that an increase in clearance will increase the modal damping
because of the higher shear deformation. The experimental results also show that the critical
speed and damping in the coast-down process are lower than those when the machine starts up.

1. INSTRUCTION
In Wikipedia, the bearing is defined as a device to permit constrained relative motion between
two parts, typically rotation or linear movement.1 Fluid bearings use a thin layer fluid to
support the bearing load so that there is no metal-to-metal contact when the shaft rotates, which
is the able to reduce wear. The portion of the rotating shaft at the bearing is often called the
journal. The working mechanism of fluid film bearing was discovered more then 100 years
ago. “Without relative motion or a converging clearance, no pressure or load capacity will be
developed. It is the pressure in the lubricant
film that carries the external load and bearing
separates the solid surfaces.” 2 Figure 1
illustrates a typical hydrodynamic bearing V
where the fluid is pumped in through an
orifice. When the shaft spins with an angular
velocity of ω, the fluid is dragged into a
convergent clearance. Due to gravity,
eccentricity, manufacturing imperfection, ω
h1
misalignment, unbalance, and other factors,
the rotor cannot always be perfectly centered.
Therefore, the fluid forms a wedge (h1 is
larger than h0 in Fig. 1) which generates h0 journal
pressure to support the applied load. By
assuming velocity continuity, the fluid
velocity on the surface of the journal is the
same as the velocity of the journal (V) at the
Figure 1. Schematic of a hydrodynamic
contact point, and is zero on the surface of
bearing and the fluid velocity distribution.
the bearing. The fluid velocity distribution

1
through the thickness is normally not linear, as shown in Fig.1. This is presumably caused by
fluid loss due to end leakage. So the fluid flows not only in the radial direction following the
rotating of the shaft, but also in the axial direction. As a result, the average velocity of the fluid
is slightly less than V/2.

As the fluid friction is proportional to the viscosity as well as the velocity gradient, an increase
in the running speed will increase the heat generation. On the other hand, lubricant’s viscosity
is very sensitive to temperature. In the meanwhile, the dynamic parameters, namely stiffness
and damping, are strongly affected by the fluid viscosity. Consequently, when a machine is
speeding up, an increase in temperature will decrease the bearing stiffness as well as damping,
which will further influence the dynamic behavior of the rotor system.

Bearing clearance is also an important parameter in rotor dynamics. For a cylindrical bearing,
the diametral clearance is defined as the difference between the bearing and shaft diameters.
The eccentric ratio, which is inversely proportional to the clearance, has a significant effect on
the dynamic characteristics of a rotor system. In addition, a change in the clearance will change
the fluid film and the shear stress, which will further affect the energy dissipation caused by the
shear deformation. Therefore, the clearance has quite complicated effects on the performance
on a rotor system. In this study experiments have been carried out to investigate the effects of
clearance on the bearing damping.

2. EXPERIMENTAL SETUP
The tests were conducted on the SpectraQuest’s Rotor Dynamics Simulator (RDS) which is
illustrated in Fig. 2. The RDS system is composed of two major subsystems, the simulator and
the lubrication oil circulation assembly.

speed display safety cover

controller
oil supply

motor

tachometer

proximity probe pressure gauge

Figure 2. Rotor Dynamics Simulator.

2
For the simulator, the motor drives the rotor shaft through a helical flexible coupling. The rotor
shaft is supported by fluid film journal bearings. The length of the rotor shaft between the two
bearings is 28 inches. The simulator is equipped with several shaft-bearing configurations of
different diameters, such as 1/2-, 5/8-, and 1-in. For each configuration, bearings with different
clearance selections are provided to perform in-depth study. The lubrication oil used is mineral
oil with ISO 13 viscosity at room temperature.

Proximity probes are mounted on the journal bearing housings to measure the relative
displacement of the shaft with respect to the bearing housing. Two proximity probes are
mounted on each bearing housing with 90° apart so their signals can be used to calculate shaft
orbit graphs. An optical tachometer is mounted near the motor output shaft. The corresponding
signal conditioner displays the running speed on a LED display, and sends the tachometer
signal to the data acquisition hardware.

The modular design of the RDS allows one to easily reconfigure the test rig and to perform
different kinds of experiments of rotor dynamics training and research, such as resonances, oil
whirl and oil whip. 3-5

The following configurations were used in this study.

(1) 1/2’’ shaft, one disk mounted at shaft center,
(2) 1/2’’ shaft, three disk evenly distributed,
(3) 5/8’’ shaft, one disk mounted at the shaft center,
(4) 5/8’’ shaft, one loader mounted at the shaft center.

For each configuration, three bearing sets with different clearances were tested with a startup
followed by a coast-down test. All the bearing numbers and their bore diameters listed in
Tables I and II. The ramp time for both the startup and coast-down is 30 seconds, in order to
excite the critical speed in a sufficient time. Figure 3 illustrates the three bearing sets for the
1/2'' shaft.

15-4

15-5

15-6

Figure 3. ½’’ Bearings with different clearance. Figure 4. SpectraPad data acquisition hardware.
3
The data acquisition and analysis were conducted using SpectraQuest’s VibratQuest system
along with SpectraPad, a multi-channel data acquisition system, as shown in Fig. 4. All the
channels are synchronized. Each input channel has its own 24-bit sigma-delta analog-to-digital
converter. The sampling rate can be up to 102.4 kSamples/second for each channel. As the
RDS is able to run up to 6000 RPM, the sampling rate was set to be 5120 Hz which is sufficient
to analyze up to the 20th order. The VibraQuest system provides steady state, hammer test, data
streaming, and on-line order tracking, four types of data acquisition mode. Since each test
lasted more than one minute in this study, the data streaming mode was used.

3. DATA ANALYSIS AND EXPERIMENTAL RESULTS

For each test, the modal damping was estimated in the Bode Plot Panel of the VibraQuest
system. The analysis procedure is as follows:
a. calculate speed profile using the tachometer signal,
b. calculate the magnitude and phase of the first order (the fundamental
order associated to the rotating speed) and show the Bode plot where
the magnitude is plotted in the dB scale,
c. smooth the Bode plot using the wavelet denoising method,
d. calculate damping ratio using the half-power point method for a
selected vibration mode.

All the calculations are performed automatically when the cursor is moved close to the peak
under investigation. Then the algorithm will search the true peak as well as the 3 dB down
points on the left and right of the resonance, and calculate the damping factor using the half-
power point method. The algorithm used also involves the wavelet denoising method and curve
fitting to smooth the amplitude curve and increase the accuracy of the calculation. The so-
called “half-power point method” is named because 10log (1/2) = -3 dB. That means for a
resonance peak in the Bode plot, the two frequencies at which the magnitude is 3 dB less than
that of the resonance peak have half energy compared to the resonance frequency. The damping
ratio ζ is then calculated as
1 ω − ω1
ζ = = 2
2Q ωr
where Q is the quality factor, ωr is the resonance frequency, and ω1 and ω2 are the half-power
frequencies on the right and left sides of the resonance frequency.

Figure 5 and 6 show the Bode plots of 1/2'' shaft with one disk configuration during the startup
and coast-down processes, respectively. Note the magnitudes are plotted in the dB scale for the
damping calculation. Once the red curser is moved close to a peak corresponding to a
resonance mode, the software will automatically catch the true peak, representing a critical
speed. It then calculates the associated modal damping. In both figures it can be seen that the
amplitude peak corresponds to a phase change, which means the peak is the first critical speed.

4
 Figure 5. Bode plot of the startup test of configuration #1 (1/2’’ shaft, one disk).

Figure 6. Bode plot of the coast-down test of configuration #1 (1/2’’ shaft, one disk).

The analysis was repeated for all the tests and results are shown in Tables I and II, as well as in
Figs. 7 and 8.

5
 Table I. Critical speeds and damping ratios of 1/2'' shaft tests
Loading Bore diameter Running Critical speed damping
Bearing
condition (inch) condition (rpm) ratio (%)
start up 1442.41 4.033
15-4 0.5025
coast down 1353.64 3.572
start up 1453.73 4.599
1 disk 15-5 0.5040
coast down 1355.12 4.071
start up 1442.17 5.26
15-6 0.5060
coast down 1342.49 4.739
start up 1049.41 3.784
15-4 0.5025
coast down 963.035 5.55
start up 1036.6 4.33
3 disks 15-5 0.5040
coast down 948.075 3.892
start up 1043.95 4.251
15-6 0.5060
coast down 901.228 3.596

Table II. Table I. Critical speeds and damping ratios of 5/8'' shaft tests

Loading Bore diameter Running Critical speed damping

Bearing
condition (inch) condition (rpm) ratio (%)
start up 1383.16 4.674
16-4 0.6275
coast down 1301.11 3.709
start up 1388.92 6.935
1 loader 16-5 0.6290
coast down 1303.68 3.935
start up 1380.62 6.667
16-6 0.6310
coast down 1300.88 4.39
start up 2213.68 2.622
16-4 0.6275
coast down 2132.78 2.476
start up 2237.92 4.056
1 disk 16-5 0.6290
coast down 2104.01 3.011
start up 2117.15 10.699
16-6 0.6310
coast down 2123.6 10.453

6
 1/2'' sahft with one disk
5.5

startup
Damping ratio (%) 5
coast down

4.5
(a)
4

3.5

3
15-4 15-5 15-6

Bearing number

1/2'' shaft with three disks

6

5.5
Damping ration (%)

startup
5 coast down

(b) 4.5

3.5

3
15-4 15-5 15-6
Bearing number
Figure 7. Damping ratios in terms of bearing clearance for 1/2’’ shaft tests.

7
 5/8'' shaft with one disk
12
start up
Damping ratio (%) 10
coast down

8
(a)
6

2
16-4 16-5 16-6
Bearing number

5/8'' shaft with one loader

7.5
7
start up
Damping ratio (%)

6.5
coast down
6
5.5
(b)
5
4.5
4
3.5
3
16-4 16-5 16-6
Bearing number

Figure 8. Damping ratios in terms of bearing clearance for 5/8’’ shaft tests.

4. DISCUSSIONS
First, it is observed that damping factor has an increasing trend as the clearance increases, as
shown in all the plots in Figs. 7 and 8, although some exemption exists. For a fluid film, the
energy is dissipated mainly due to the shear stress, which is inversely proportional to the film
thickness. Therefore, in order to study the effect of clearance on damping, we need to study the
relationship between clearance and film thickness. The film thickness to clearance ratio is a
function of the Sommerfeld number. But the Sommerfeld number also involves the clearance,
so the relationship between film thickness and clearance is quite complicated. More discussion
on this issue will be provided in another tech note. In some running conditions, an increase in
clearance will decrease the film thickness which will increase the shear stress in the fluid film.
Therefore the modal damping is improved. This explains our experimental results.

8
However, as we have seen, the effect of clearance on damping is a complicated problem. Some
researches reported the exactly opposite prediction. Ref. 6 provides an equation for the radial
damping of a hydrodynamic damping. If we calculate the derivative of the damping expression
with respective to the diametral clearance, it can be found that the radial damping is a
monotonically decreasing function of clearance size.

It is worth noticing that high damping is not the only criterion for noise and vibration control.
The overall effects of many factors such as mass, stiffness, and so on have to be considered as
well. In addition, some vibration problems benefit from, and others are hurt by high damping.
In fact for cylindrical bearings, high damping may be quite harmful, causing the oil whirl and
whip phenomena.5 In order to avoid complicated theoretical derivation, the bad effect of high
damping can be simply explained as follows. In the rotating coordinate system, the rotor
behavior is determined by not only the principal stiffness, but also, mathematically speaking,
the cross-coupled terms. These cross coupled terms, which are proportional to damping,
generate a tangential force which is in the same direction of rotation. Therefore, instead of
reducing the rotor’s whirling motion, such a damping-related tangential force is promoting the
shaft vibration. This will cause instability and may lead to serious consequences.

A careful study of Tables I and II shows that the critical speeds and damping ratios in the
startup and the coast down processes are different. For all the test conditions, the critical speed
and damping ratio during the coast down process is slightly less than that in the startup process.
As explained in the instruction section, oil viscosity is very sensitive to temperature. Less
viscosity will decrease the stiffness and damping coefficient of the fluid film. It is believed that
during the coast down process, the oil temperature is higher than during startup due to
increasing bearing temperature in the test.

5. FUTURE WORK
As seen in the last section, oil temperature is an important parameter in rotor dynamics.
Unfortunately, temperature was not measured in the experiments. We will use thermal couple
to measure the oil temperature in the future test in order to quantify the effect of temperature.

Cylindrical bearings are the most possible pattern to cause oil whirl and whip. Spectra Quest
Inc. is designing other anti-whirl bearings, such as tilting pad bearing, in order to reduce the
fluid circumferential flow velocity.

REFERENCES
1
http://en.wikipedia.org/wiki/Bearing_%28mechanical%29
2
M. He, C. H. Cloud, J. M. Byrne, “Fundamentals of Fluid Film Journal Bearing Operation and
Modeling”, 34th Turbomachinery Symposium (2005)
3
http://www.spectraquest.com/downloads/MFS_RDS.pdf
4
“Dynamic Behaviors of Rotor with Fluid Film Journal & Rolling Element Bearings”,
SpectraQuest Tech Note, November 2005

9
5
“Interesting Rotor Dynamics Observations on Oil Whirl and Whip”, SpectraQuest Tech Note,
April 2006
6
D. Bently, P. Goldman, T. Eldridge, “The Death of Whirl – What the SFCB Can Do for the
Stability of Rotating Machinery”, Orbit, First Quarter 2001 pp. 10-13.
http://www.bpb-co.com/articles/printableFiles/01deathofwhirl_print.pdf

Note: All the SpectraQuest tech notes are available online at

http://www.spectraquest.com/tech/index.html

10