Thermoelastic Forum Vol.1, No.2, October
1992
First Words
The R&D Side
Tech Tips
Events
UW Corner
Thank you for the encouraging response to our first newsletter. We hope
that we can continue to bring you newsletters that keep you up-to-date with
the latest in thermoelastic techniques, equipment and software.
The November SEM conference in Chicago is coming up! In our "Events"
column we have provided information on sessions that we think you will find
interesting. Please take a minute to read the session and paper titles.
There will be no written proceedings from this conference, so if you are
not attending and you have a particular interest in one of the topics you'll
have to get in direct contact with one of the authors. Stress Photonics
can provide you with their phone numbers and addresses.
The "UW-Corner" deals with thermoelastic stress analysis of randomly
loaded structures. This is a topic which has received significant study
at Stress Photonics as well as at the University of Wisconsin. If you are
interested in using your SPATE system for non-sinusoidal loading contact
us for a chat and we'll do our best to help you.
Again, we are glad to see that so many of you found the newsletter valuable.
If you have any suggestions as to how this newsletter can better serve you
please describe your ideas on the comment card provided and return them
to us. We would be happy to incorporate your suggestions into upcoming editions
of the Thermoelastic Forum.
By Jon Lesniak
I have for some time been concerned about image distortions caused by the
SPATE scan mirrors. Recently, I looked at this problem in more detail.
There are three common image distortions affecting SPATE data sets:
Figure 1a shows a combination of these effects.
Scan Mirrors
As can been seen by looking into the SPATE camera, there are two scan mirrors.
The rectangular one on the bottom is the y-scan mirror and the oval one
above it is the x-scan mirror. The distance between their axes of rotations
is about 2.0 inches.
The mirrors convert the analog steps from the computer DACs to proportional
angular steps. The x-y scan position is related to these angular steps by
where d is camera distance, ax and ay are angle/DAC step, and Nx and
Ny are the number of DAC steps.
Mirror Separation Distortion
As a result of the difference in the true x and y distance, each y DAC step
yields a smaller y distance on the specimen than each x DAC step. If, for
example, you were to mark off a 3 in. x 3 in. square on your specimen, you
would find that it takes more DAC steps in the y direction than in the x
direction. The resulting image would be displayed as a tall rectangle. The
amount of y expansion in our camera is plotted in Fig. 2. Note that the
angular steps of our mirror motors appear to have been set for distortion
free imaging at 60 inches.
Unfortunately, the y expansion is largest at a camera distance of 10 in.;
the working distance often used to compare SPATE to theory.
Small Angle Approximation Distortion
The SPATE system approximates TAN (Theta) as Theta. This can have a significant
effect on larger scans. For example, for a scan extending ± 25°
a 7% dimensional error occurs at the edges of the scan.
Camera Angle Distortion
The details of camera angle distortion can be slightly more complex than
one would first expect. But simply, if the camera is viewing the surface
at an angle b, then there is a compression of the image of 1-cos(b) in the
direction of the angle.
Image Correction Software
It is not difficult to correct any of the aforementioned image distortions.
The scan shown in Fig. 1a was taken at 18 in. from the specimen with a 30°
camera angle. A camera angle of 30° results in a 13% compression of
the x dimension. At 18 in. the scan mirror separation causes a 6% expansion
of the y dimension.
Figure 1b is the result of a new analysis routine that I have created to
correct these image distortions. The routine is a simple spatial transformation
which interpolates data for the new coordinates from the original data set.
Notice that the image is not affected in any way other than the spatial
correction. The data can still be considered "raw data".
In 1992, the clock battery inside many HP310 and 320 computers started
to run out. As a result, the data file creation date in the upper left corner
of SPATE 9000 outputs may be incorrect. This is a particularly frustrating
problem with two-computer systems because the Data Acquisition and Control
(DAC) computer does not have a keyboard or a video monitor; therefore, setting
the time is tricky. This tip applies to two-computer systems but can easily
be applied to one- computer systems.
Replacing the Battery
Replacing the battery is as easy as changing the battery in a calculator.
You will need a BR2325 Lithium battery. A Radio Shack catalog # 23-168 ($1.98)
will work fine. To replace the battery follow these steps:
Setting the Time and Date
Now that you have the new battery in, you need to set the time and date
to current values. Once this is done the computer will keep fairly good
time for the next few years. To set the time and date you need to get the
DAC computer to execute a SET TIMEDATE command. You do this by making a
change to the AUTOST program that both computers run when they first boot
after power up. Here are the steps;
330 CONTROL Hpib,3; Dac_cpu_adrs ! address to 22
340 DISP "LOADing DAC"
350 MASS STORAGE IS ":,700,0,0"
355 SET TIMEDATE DATE ("30 Sep 1992")+TIME ("08:15:00")
360 LOAD "DAC" ! THE routine loaded here has
If you are an experienced programmer this should be a sufficient set of
instructions. However, if you are a novice, call or email us and we'll FAX
you the step-by-step instructions.
SEM will program jointly with the American Society for Nondestructive
Testing (ASNT) at the Quality Testing Show and 1992 Fall Conference to be
held at the Sheraton Chicago Cityfront Center.
The Quality Testing Show attracts NDT professionals from all over the globe
to share the latest in NDT techniques and applications. The Quality Testing
Show typically attracts more than 2,000 people.
The theme of ASNT's Fall Conference is, "Quality Through NDT: A Partner
for Success," and will offer approximately 125 presentations. Sessions
pertaining to thermal methods are detailed below.
For further information about the program or exhibit, please contact the
SEM office, 7 School Street, Bethel, CT 06801; (203) 790-4472.
Technical Session 2. Thermographic Measurement for Structural Assessment
Tue., Nov. 17th, 10am-Noon
IR Thermography of Fast Transient Phenomena
Steve Shepard, U.S. Army Tank-Automotive Command (USA)
Advanced Thermographic NDE via Dynamic Pattern Projector
J. Lesniak, Stress Photonics, Inc. (USA)
Thermoelastic Stress Characterization in Aircraft Structures
K.E. Cramer, NASA Langley Research Center and C.S. Welch, College of
William and Mary (USA)
Thermal Characterization of Structural Material Using Time-resolved Techniques
J.C. Murphy, J.W. Spicer, W.D. Kerns and L.C. Aamodt, The Johns Hopkins
University (USA)
Thermoelastic Quantification of Distributed Damage in a Composite Material
B.J. Mahoney, Ford Motor Co. (USA)
Technical Session 4. Nondestructive Testing Applied to Composite
Materials Wed., Nov., 18th 9am-Noon
Low Frequency Ultrasonic Scans of Flaws in Composites
D.K. Hsu, Iowa State University (USA)
Review of Progress in Quantitative NDE for Fiber-reinforced Composites
L.J. House and F.B. Stulen, Battelle Memorial Institute (USA)
Nondestructive Detection of Compressive Fatigue Damage in Thick Composites
R.E. Green and C. Byrne, The Johns Hopkins University (USA)
Damage Accumulation in Thick Composites During Compressive Fatigue
C.Byrne and R.E. Green, The Johns Hopkins University (USA)
Use of Piezoelectric Polymer Sensors for Sensing Stresses and Defects in
Adhesives and Composites
G.L. Anderson, Thiokol Corporation; D.A. Dillard, J. Mommaerts, J. Duke
and B. Tang, Virginia Polytechnic Institute and State University (USA)
Studies of Impact Damage in Structural Composites by Thermal Wave Imaging
R. Thomas, Wayne State University (USA)
Tutorial: Thermoelastic Stress Analysis Tue., Nov., 17th 1:30pm-5pm
Speakers: Bradley R. Boyce and Jon R. Lesniak, Stress Photonics; and
Jamal Dajani, Ometron
The tutorial is planned to run two and one-half to three hours and will
cover the fundamentals of thermoelastic theory, instrumentation, and application.
Application examples with detailed explanation of instrumentation, and data
analysis will be used to illustrate the capabilities and shortcomings of
the technology. Some advanced applications of Thermographic Stress Analysis
(TSA) will also be discussed briefly.
Publications: "A Method of Thermographic Stress Separation,"
Boyle, J.T., Hamilton, R., Applied Solid Mechanics IV Conf. Leicester, U.K.,1991.
By J.P. Miles, B.J. Mahoney, B.I. Sandor
Department of Engineering Mechanics, UW-Madison
Traditional Thermoelastic Stress Analysis (TSA) uses a sine-wave reference
signal as an input to the SPATE lock-in amplifier. Harwood and Cummings1
have demonstrated the use of the Fourier technique under random excitation
whereby a frequency response function between a reference input and the
TSA output is determined. This approach is useful when the system exhibits
frequency-dependent response. However, the method suffers from a lack of
user knowledge of the frequency domain technique, and an increased total
data acquisition time. A different signal processing technique has been
developed for the TSA system which accommodates a component subjected to
variable amplitude, random loading.
Using a strain gage placed locally in the area of interest as a reference,
the technique exploits the linearity between the known strain and the resulting
TSA signal. A ratio between the total power of the TSA signal and the total
power of the reference signal is calculated. Under non-modal behavior, this
ratio is equivalent to the square of the frequency response function and
is constant at all frequencies.
An aluminum flat plate with a hole in the center was used to test the random
signal analysis method. The specimen was loaded in a closed-loop servohydraulic
testing machine and subjected to a
10-40 Hz tension-compression random load. The reference signal was obtained
from a strain gage placed at the far-field stress location. Figure 1 shows
the results of a line scan passing through the center of the hole. For comparison,
the results of a traditional sinusoidal line scan are also shown and are
in excellent agreement with the line scan produced by the random signal
analysis method.
1Harwood, N. and Cummings, W. M., "The Theoretical Basis of the Use
of Random Excitation Signals for Thermoelastic Stress Analysis", SPIE,
713, 32-43,1987
