Thermoelastic Forum Vol.1, No.4, September
1993
First Words
R&D Side
Tech Tips
Events
New Products
Tid Bits
Considering the length of time since our last issue, Stress Photonics
has many new ideas, products, and events to relate to you. One item is a
new level of cooperation between Stress Photonics and Ometron. SP and Ometron
will have some joint product development efforts: Ometron will be selling
SP products outside North America, and SP will assist with sales and service
in North America.
In this issue: 1) Make sure you don't miss reading about three exciting
new products: Data Maximizer Software, V-Amp and the SPATE 4000 camera head.
2) The Events column summarizes important technical papers presented at
the June Society for Experimental Mechanics conference in Dearborn and the
July conference on Aging Aircraft jointly sponsored by SEM and SPIE. 3)
As always, Tech Tips provides a helpful little hint for your spate work.
4) For your more intellectual moments, the R&D Side provides the technical
detail behind Stress Photonics' new V-Amp product. 5) Our new column, Tid
Bits, describes a hint from a SPATE user. If you would like to contribute
an article or even just a "tid bit," send it to the Thermoelastic
Forum at Stress Photonics.
By Jon Lesniak
In many situations it would be extremely time efficient and informative
to be able to perform TSA scans while a component is undergoing its routine
road-load simulation or other multiaxial testing. In these situations, SPATE
scanning via single-frequency correlation is inefficient because the load
energy is distributed over a range of frequencies.
This article describes an extension to the single-input variable-amplitude
approach developed by Stress Photonics under NASA funding. In that project,
we showed that non-periodic signals can be analyzed with a simple digital-correlation
algorithm. Essentially, the algorithm least square fits the SPATE signal
to the sampled reference (fig. 1). Of course, the reference must be free
of phase changes, DC offsets, and other distortions. A load cell signal
usually provides an acceptable reference.
For a multiple input system like the John Deere biaxial test fixture depicted
in fig. 2, the stress at any given point will be a linear combination of
all load inputs described by a multiple regression equation of the form
where F(t) and G(t) are the time histories of the load inputs and A(x,y)
and B(x,y) are the regression coefficients taking the form of distinct images.
Each image describes the stress pattern influenced by each respective input.
By using the multiple regression equation the simultaneous stress images
can be determined by the simple relation
where K1, K2 are the running sums
A variable amplitude scan takes slightly more time than a normal SPATE
scan, although not nearly as long as some earlier FFT techniques.
As with the single input case, the reference input must be free of phase
changes, DC offsets and distortions. A load cell signal is found to be adequate
in most cases. In order to separate the influences of multiple inputs, the
load shapes must be distinctly different.
Because heat transfer will attenuate DC and low frequency temperature changes,
low frequencies are removed from both the reference and the SPATE signals
by matched high pass filters set at a 1-5 Hz cutoff. Low pass filters are
used (usually set at 100 Hz) to minimize high frequency noise.
The software allows the imaging to be based on any linear combination of
the two load inputs. For example, in the case of John Deere biaxial tests,
it is preferable to compile separate images representing the stresses due
to bending and torsion. To do this, the software samples the load cell references
and combines them to form the desired references.
The running sums are calculated substituting R1 for F(t) and R2 for G(t).
Even if the images are collected using F(t) and G(t) directly, the torsion
and bending images can be calculated through image addition and subtraction.
Indicators are used to assess the correlation between the fit and the SPATE
signal, and the ability to separate between the two references.
For example, the formula below determines whether the two inputs are sufficiently
independent.
This definition is similar to the vectorial definition of cosine. Like
the cosine function, a value of 1 implies the reference functions are similar,
and a value of 0 implies that they are distinct. In the event of insufficient
loading or inseparable references, the sample period is repeated.
Figure 3 is a TSA scan of a John Deere biaxial test specimen. This image
shows the stresses caused by bending. This sample was undergoing a fatigue
test loaded with a SAE Log Skidder load history. The SPATE scan was collected
without interfering with the test in progress.
We would like to thank Paul Swanson and the John Deere Technical Center
for their cooperation in the collection of these early data sets.
By Brad Boyce
Within the SPATE 9000 software there is a short routine that calculates
the proper focus ring setting for the working distance entered by the user.
When focus data is generated for a scan unit, a formula of the form
(where d is the working distance) is fit to the data. The parameters
A, B, C, and D are recorded and entered into the scan software for that
particular head. If a new focus curve is run for a head or if a new head
is matched up with a particular set of scan software, then the parameters
A, B, C, and D need to be changed in the software.
To enter new parameters follow these steps:
1. Load your SPATE 9000 software system and start it.
2. Press the Stop key and issue the commands
3. On your screen should be the explicit assignment of the parameters
A, B, C, and D. Change them to the values listed on the recent metric focus
curve.
4. We suggest that you enter a comment line near Focus_a that dates the
entered parameters and indicates the serial number of the scan unit for
which they are valid.
5. Make the changes permanent by saving the SPP program with the command
6. Now, to activate the modified SPATE 9000 software just type
If you have any difficulties or need help please feel free to call or fax Stress Photonics at: ph(608) 224-1230 fax (608) 224-1233
By Brad Boyce
Nine papers of particular interest were presented this past June at the
Spring SEM conference in Dearborn, Michigan.
"Infrared Staring Arrays and Digital Signal Processing"
by Ryall and Wong
This paper may have raised the most interest at the conference.
By using an array detector and digital signal processing, the Australian
based Aeronautics Research Laboratory was able to make quality 486 x 510
TSA images in as little as 10 minutes. Predictions are that soon array detectors
will collect images even faster.
"Thermoelastic Stress Analysis of Structures Under Random Loading"
by Miles, Sandor, and Mahoney
"Thermoelastic Stress Measurements on a Bridge Loaded by Traffic"
by Boyce, Zuraski, and Miles
Both papers deal with the same topic covered in this month's R & D Side.
"Quantifying Distributed Damage in Composites Under Random Loading
Using Thermoelastic Stress Analysis" by Mahoney, Miles, and Sandor
"The Use of Thermoelasticity to Evaluate Stress Redistribution and
Notch Sensitivity in Ceramic Matrix Composites" by Mackin
These two papers describe how TSA can be used to quantify and track damage
in composites. The papers give some hope that in the future one may be able
to assess the level of damage in a composite component or structure with
TSA.
"Progress in Thermoelastic Evaluation of Mixed-mode Stress Intensity
Factors" by Stanley and Dulieu-Smith
This paper explores the exciting capability of TSA to be used readily to
make stress intensity factor measurements, a feature that, until now, has
seen far too little attention.
"Thermoelastic Data Improvements" by Lesniak
As an extension of information presented in earlier R&D Side articles,
this paper describes many techniques that are now available in the Data
Maximizer products by Stress Photonics.
"Recent Advances in Determining Individual Stresses from Thermoelastically
Measured Isopachics", by Rauch and Rowlands
Rauch and Rowlands put forward a number of interesting ideas in this paper
that have been evolving over several years of work.
Stress Photonics has been exploring the idea of using differential thermography
in non-destructive inspection applications. Jon Lesniak has developed the
ideas for using differential thermography for NDE and presented them for
comment at two NDE conferences. These ideas have received very favorable
responses and work is now under way at Stress Photonics to develop this
technology further. We refer to the new technology as Force Diffusion
Thermography (FDT). Jon's most recent presentation of FDT was at a July
aging aircraft NDE session jointly sponsored by SPIE and SEM.
FDT projects a dynamic pattern of heat onto the surface of the structure.
The structure absorbs this energy and conducts heat from hot to cool regions.
Flaws will impede the flow of heat, causing disturbances in the local thermal
pattern. Patterns resembling moving parallel lines have been found to find
cracks as well as disbonds and delaminations.
Dynamic patterns allow the use of the same differential IR cameras used
for thermographic stress analysis, and the continuous illumination means
low wattage projectors can be used in place of the high wattage flash lamps
used by other thermal methods.
If you would like to learn more about FDT, watch for your next issue of
the Thermoelastic Forum and/or use the reader response card to request "Force
Diffusion Thermography" by Lesniak and Boyce.
Stress Photonics now offers an economical product that makes it possible to use SPATE when constant-amplitude loading is not available. Many labs perform variable-amplitude fatigue/durability testing and would like to be able to use their SPATE system to obtain stress patterns from these tests. Stress Photonics has designed V-Amp, a hardware/software package that makes it practical to use the SPATE 9000 to collect stress patterns from single and multiple input variable amplitude (V-Amp) tests. Sampling times are typically 0.1 to 0.5 points/sec for single input systems and twice that for dual input systems. The V-Amp system can measure stresses caused by loadings between 2 and 100 Hz. For more technical detail on V-Amp, refer to the R&D Side column. Prices vary throughout the world, so contact Stress Photonics or the nearest Ometron representative for pricing.
Now available in the U.S. are the Image Manipulator and Smart Smoother
options of the Data Maximizer Software package for SPATE 9000. The Image
Manipulator option removes spatial distortions and provides mirror image
correction, image rotation, and subscan extraction. These features combine
to make a powerful package that improves SPATE outputs. Image Manipulator
orients the outputs to more natural or expected views and assists the comparison
of SPATE data with other stress analysis techniques. As an added bonus,
the Image Manipulator improves operation of the Scan Geometry and Screen
Annotator portions of the SPATE 9000 software.
The Smart Smoother software uses Fourier transform techniques to identify
and reduce noise from SPATE images. Optical, thermodynamic, and stress gradient
criteria all indicate the maximum possible spatial gradient in a scan. Smart
Smoother takes advantage of this information to remove gradients above the
threshold of possibility. The main advantage of the Smart Smoother is that
it eliminates the guess work from choosing the proper amount of smoothing
and does not blur important detail.
You can get more information on the Data Maximizer software packages by
contacting Stress Photonics, your nearest Ometron representative, or by
sending in the reader response card.
By Brad Boyce
Most SPATE operators find plotted focus curves difficult to read. Our experience
is that mistakes are regularly made when reading these curves. One SPATE
8000 user found a solution to this problem. He fit a curve to his focus
data, programmed the curve into an inexpensive hand-held calculator, and
stuck the calculator to the side of the camera head. Now the operator can
quickly enter the working distance and, voilà!, the correct focus
ring setting appears.
If you want help fitting a curve to your focus data, contact Stress Photonics.
We're happy to be of service.
