You are on page 1of 7

feb /mar 08

the magazine for maintenance & reliability professionals

Walter Reed Army Medical Center


Rebuilding A Maintenance Program

Asset Management’s Upbringing • Practical Advice for Reliability Engineers


Putting Reliability Into Ice Cream • Ultrasound: The Power of Two
si u 08

Uptime is a registered trademark of Reliabilityweb.


com. The following article is used with permission from
In ch 20
de re

Uptime Magazine. Copyright© 2008 by Reliabilityweb.


ro -
B M

com. All rights reserved.


C
R

www.uptimemagazine.com
Vibration Training

Do you really understand


vibration analysis? Do you want to?
Before you turn the page...

Setting up a program, taking measurements, and viewing spectra is not


difficult. But doing it correctly is difficult. Making late calls, false calls, or
missing the fault altogether is not good for your confidence or your career.
We help you to unleash the analyst within.

Mobius offers a better way to learn


In this day and age you can do better than attend a three day course filled
with boring slides... Mobius training begins the moment you register. You
are e-mailed a link to our on-line training center. You can view interactive,
narrated slides, read the manual, and take some quizzes. You can do that
after the course too; for refresher training. During the course you will view
slides filled with cool animations and super-cool simulations. The simulators
turn complex theories into simple, visual procedures that are easy to
understand and remember. And if you understand you can solve anything.

If you don’t have the time or budget to come to our course, buy
iLearnVibration and learn at your own pace, in your own place.

Please visit our Web site, look at the simulators, check the course schedule,
and then make the call that will help you evolve into the analyst that you
know you can be: 877 550 3400.

www.ilearninteractive.com
Offices in Melbourne, Seattle, and Knoxville. Training centers in 22 countries.
Toll free: 877 550 3400 e-mail: info@ilearninteractive.com
vibrationupload All Cracked Up
Examining the Vibrational Effects of a Cracked Shaft
by Dr. Zhuang Li & Dr. Suri Ganeriwala

A
shaft crack is one of the most common defects in a rotor system and early detection of this failure mode is a
very serious matter. In this study, shaft cracks were simulated and analyzed using SpectraQuest’s rotor Ma-
chinery Fault Simulator™ (MFS). A series of experiments were conducted to observe the behavioral changes
of the cracked shaft in critical speed, 1X and 2X frequency responses. The experimental results were found
to be consistent with the theoretical prediction of the shaft crack.

A shaft crack is a slowly growing fracture of the rotor. If Maximum Tensile Stress
undetected in an operating machine, as a crack grows,
the reduced cross section of the rotor will not be able
to withstand the dynamic loads applied to it. When this
happens, the rotor will fail in a fast brittle fracture mode. Transverse Crack
The sudden failure releases a large amount of energy
that is stored in the rotating system, and the rotor will Torsion Crack
fly apart. This kind of failure may cause serious injury or
even death to anyone unfortunately standing near the
machine at that moment. Obviously, shaft crack detec-
tion is a very serious matter, and machines that are sus- Maximum Tensile Stress
pected of having a crack must be treated with the utmost
caution. Figure 1 - Transverse and Torsion Cracks

Shaft bending stiffness is related to the shaft cross-sec-


Cracks are initiated in the shaft in regions of high local
tion area. As a crack propagates across the shaft, the re-
stress. Shafts are subjected to large-scale stresses due to
maining cross section becomes smaller, and the bending
bending, torsion, static radial loads, constrained thermal
stiffness of the shaft decreases. This will reduce the reso-
bows, thermal shock, and residual stresses from heat
nance frequency as well as the critical speed of the rotor
treatment, welding and machine operations. All of these
system. The reduction in shaft stiffness also causes the
stresses combine to produce a local stress field that
rotor to bow more in response to a static or dynamic
changes periodically. In a small, local region where
load, such as a rotating unbalance. The bow is likely to
stresses exceed the maximum that the material can with-
change over time. As a result, it will change the effective
stand, a crack will form in the material.
location and magnitude of the heavy spot, which will
consequently change the 1X rotor response.
If the cyclic stresses are sufficiently high, the leading
edge of the crack will slowly propagate so that the plane
If a rotor with a crack has a steady, unidirectional radial
of the crack is perpendicular to the orientation of the
load, then a strong 2X response may appear when the ro-
tensile stress field. The orientation of this stress field is
tor is turning at half of any resonance speed. As a breath-
determined by the type of stress (bending or torsional)
ing crack involves both the closing and opening in one
and by geometric factors. If the rotor is subjected only
revolution, the rotor will respond at the 2X frequency. If
to simple bending stresses, then the stress field will be
a resonance exists at twice running speed, then the 2X
oriented along the long axis of the rotor, and the crack
vibration will be amplified.
will propagate directly into and across the rotor section,
forming a transverse crack. The pure torsional stress
In this study, a series of tests were carried out on a Spec-
will produce a tensile stress field that is oriented at 45°
traQuest MFS with cracked shafts to observe their be-
relative to the shaft axis. A crack in this stress field will
havioral changes, including the critical speed, 1X and 2X
propagate into the rotor and tend to form a spiral on
frequency responses, compared with an intact shaft.
the shaft surface. Figure 1 shows these two types of
cracks. In most rotor systems, the stress field contains a
mixture of bending and torsional stress. Bending stress, Experimental Setup
however, is usually the dominant component, thus the
crack will usually propagate into the rotor as, basically, a Three experiments were designed to approach the objec-
transverse crack. tive of this study. They were:

78 february/march 2008
with a simple 90 degree Time Waveform
4E+0
0.150” deep notch. A small
2E+0
tapped hole in the center
allowed a small filler piece 0E+0

g
to be clamped in the notch -2E+0
to vary the stiffness change -4E+0
due to the notch. Tests 2.4 5 10 15 20 25 30 35 40 45 50 55 60 65 68
were run with the filler Time (s)

piece removed, partially Figure 4 - Time waveform of the start-up and cost-down
and fully tightened. test for “all bolts tight” condition.
Figure 2 - SpectraQuest’s Machinery Fault
Simulator (MFS) Experimental Results Spectrogram
247.2
and Analysis -21.2

A. Cracked shaft simulated using Flange All experimental data have been 220

collected and analyzed by using 200


B. Notched shaft with V-shape crack 1.6”
SpectraQuest’s VibraQuest soft-
from the inboard bearing housing 180 -1.1E+2
ware package. Transient, Wa-

Frequency (Hz)
C. Notched shaft with V-shape crack 1” from terfall and Analysis tools of this 160

the inboard bearing housing software package were mainly 140


used in this study.
The changes of critical speeds for the intact 120

and cracked shafts were studied in experi- Change of Critical Speeds 100
ments A and B. The changes of 1X and 2X
frequency responses for the intact and cracked The critical speed in start-up 80

shafts were investigated in experiments B and tests can be identified by using 60


C. The running speeds were chosen to be time-frequency spectrogram,
40.9
2000 rpm and 4000 rpm. time waveform, and a waterfall 40.9
plot of the acceleration signals,
The tests were conducted on the rotor MFS shown in Figs. 4 through 6 re- Figure 5 - Intensity plot of the stat-up in the flange-
which is illustrated in Figure 2. Four acceler- spectively. In Fig. 5 the abscissa simulated crack test.
ometers were mounted on the inboard and is time (second). The red spot
outboard bearing housings in the vertical and where a cursor points repre-
horizontal direction, respectively. The setup sents the first critical. The
of experiment A is shown in Figure 3. cursor values indicates that
the critical occurs at 21.7 sec-
Figure 3 shows the flange-simulated cracked
ond and 65.62 Hz. For Experi-
shaft. It consists of two separable shafts
ment A, the critical speeds for
joined at the mating flanges. The four (4)
different cracked conditions
bolts compress Belleville washers which can
on the shaft are listed in Table
be loosened or tightened in a pattern to cre-
1, where the ‘all bolts tight
ate an unsymmetrical time varying stiffness
condition’ simulates no crack
and simulate the opening and closing of a
on the shaft, and ‘three-loose’
transverse crack. The large disc next to the
means the most seriously
flanges provides gravity loading and a shaft
cracked condition.
bending moment. Tests were run with one,
two and three bolts loosened. The critical speeds for differ- Figure 6 - Waterfall plot of the start-up and coast-down in
ent cracked conditions on the flange-simulated crack test
The setup for experiments B and C use a shaft
the shaft in Experiment B are
listed in Table 2. Intact means no crack on
the shaft and “No-filler” represents the most

Conditions Critical Speed (Hz) Conditions Critical Speed (Hz)


All Bolts Tight 65.62 Intact 65.00
One Bolt Loose 65.00 Filler Fully Tightened 62.50
Two Bolts Loose 64.37 Filler Partially Tightened 62.50
Three Bolts Loose 64.37 No Filler 60.00
Table 1 - Change of the critical speed Table 2 - Change of the critical speed
Figure 3 - Setup for experiment A in Experiment A in Experiment B

www.uptimemagazine.com 79
C ha nge of the c ritic a l frequenc y in
ex periment A a nd B
Change of the Critical Frequency
in Experiment A and B
68
68
the c ritic a66
l 66
frequenc y in
periment A a nd B
64
64
Experiment A
62
62
Experiment B
60
60
Experiment A
58
58
Experiment B
Critical
56 56 frequency (Hz)
Crack Condition (from intact to worse)
C r a c k c o n d itio n (fr o m
z)
Figure in
7 -taChanges
c t to woof rcritical
s e c r aspeeds
c k ) as
n d itio n (fr o m the crack conditions change
wo r s e c r a c k )

serious crack condition. Filler fully tightened


and Filler partially tightened (more serious
than Filler fully tightened) stand for the middle
cracked conditions.

Both tables have demonstrated that as the


crack grows, the critical speed decreases due
to the reduced stiffness. The overall trend of
these results matches the theory of shaft crack
quite well. Figure 7 clearly shows this trend
for the changes of critical speeds in different Figure 8. Frequency analysis of 1X and 2X frequency responses
crack conditions.
1X and 2X Frequency Response

SDT 170
Using the analysis functions in the VibraQuest
software package, as shown in Fig. 8, it is easy
to obtain the 1X and 2X frequency response
for each steady-state test in Experiments B

HEAR MORE
and C. Because the critical speeds of the rotor
systems were around 60 Hz to 65 Hz, when
the running speed is 2000 rpm (33.33 Hz) in
the tests, 2X will be close to the resonance
frequency. So both the 1X and 2X compo-
nents need to be analyzed. When the running
speed is 4000 rpm (66.67 Hz), 1X is close to
the resonance frequency and 2X will be far
away from it. Therefore, only the 1X frequen-
cy response has been analyzed.

Tables 3 through 6 show the results of 1X


and 2X frequency response for all cases of Ex-
periments B and C. Comparing these results
shows that normally the 1X and 2X frequency
responses of the cracked shaft are larger than
that of the intact shaft. This means the crack
The factory floor is a noisy place where problems go unheard. in the shaft has changed and amplified the
Leaky pipes, dry bearings, failed steam traps, and faulty electrical 1X and 2X vibration responses. This result is
panels are potential failures that lurk quietly, waiting for the most consistent with the theoretical consequence
inopportune time to shut you down. Don’t be caught not listening. about the shaft crack which has been men-
Get your SDT170 Today... and HEAR MORE. tioned before in the introduction.
SDT 170 Summary
Ultrasonic Detector A shaft crack is a slowly growing fatigue frac-
HEAR MORE Call Now: 800-667-5325 ture of the rotor. Damage of a crack failure
www.sdtnorthamerica.com can cause serious injury and major downtime.

80 february/march 2008
Experiment B, 2000 rpm 1X (g) 2X (g) Experiment B, 4000 rpm 1X (g) two B.S. degrees in precision instrument en-
Intact Shaft 9.578e-3 1.753e-2 Intact Shaft 8.732e-2 gineering and industrial engineering and the
Cracked Shaft (No filler) 1.746e-2 2.125e-2 Cracked Shaft (No filler) 2.756e-1
M.S. degree in precision instruments all from
Tianjin University in 1998 and 2001, respec-
Table 3 - 1X and 2X frequency responses for Table 4 - 1X frequency responses for tively. He obtained the Ph.D. degree in me-
2000 rpm in Experiment B 4000 rpm in Experiment B chanical engineering from Auburn University
in 2005. His research interests include acous-
Experiment C, 2000 rpm 1X (dg) 2X (g) Experiment c, 4000 rpm 1X (g) tics, vibration, and signal processing. He has
Intact Shaft 9.578e-3 1.753e-2 Intact Shaft 8.732e-2
published more than twenty journal papers
and book chapters. Dr. Li is also serving as the
Cracked Shaft (No filler) 2.729e-2 1.471e-2 Cracked Shaft (No filler) 1.714e-1
associate editor of the International Journal of
Table 5 - 1X and 2X frequency responses for Table 6 - 1X frequency responses for Acoustics and Vibration.
2000 rpm in Experiment C 4000 rpm in Experiment C
Dr. Suri Ganeriwala is founder/president of
Therefore, detection of a shaft crack is of vital package. A series of experiments were con- Spectra Quest, Inc. He has over twenty-five
importance. Shaft crack reduced the bending ducted to observe the behavioral changes years of industrial and academic experience in
stiffness of the shaft due to the reduced avail- of the cracked shaft in critical speed, 1X and machinery vibration diagnostics and control,
able cross-section area, which will change the 2X frequency responses. The results show signal processing, and viscoelastic materials
critical speed of the rotor system, and its 1X that the critical speed decreased as the crack characterization. Suri has developed a unique
and 2X frequency response when the system is increased, and the 1X and 2X frequency re- method of instruction using the Spectra Quest
operated at one-half of a resonance frequency. sponse for cracked shaft increased compared Machinery Fault Simulator (MFS), which is his
All of these symptoms provide the diagnosis with the intact shaft. Those results are consis- brainchild from concept to completion. He
for shaft cracks. tent with the theoretical consequence of the has authored over thirty papers and articles in
shaft crack. journals, magazines, and books. He obtained
In this study, shaft cracks were simulated and a Ph.D. in Mechanical Engineering from the
analyzed using SpectraQuest rotor Machinery Dr. Zhuang Li is a research and development University of Texas at Austin.
Fault Simulator and the VibraQuest software engineer at Spectra Quest Inc. He received

www.uptimemagazine.com 81
Like What You’ve Read?
Get your FREE subscription to

uptime t h e m a g a z i n e f o r Pd M & C B M p r o f e s s i o n a l s

Now!

In each issue, Uptime Magazine delivers high quality articles on Predictive Maintenance and
Condition Based Monitoring.

You receive articles covering:

Information Technology
Infrared Thermography
Lubrication/Oil Analysis
Maintenance Management
Motor Diagnostics
Alignment/Balancing
Reliability
Ultrasound
Vibration
Plus a feature article on relevant topics to help you increase your knowledge about the in-
dustry and take your company’s predictive maintenance to a higher level.

Subscribe to Uptime Magazine now! Click anywhere on this page and you will be directed to the
uptime website. It will take only a minute or two of your time to receive your FREE subscription.

You might also like