3. What is Fatigue?
In materials science, fatigue is the
progressive and localized structural
damage that occurs when a material is
subjected to cyclic loading.
The nominal maximum stress values are
less than the ultimate tensile stress limit,
and may be below the yield stress limit of
the material.
4. Introduction to Imaging
Defect detection technologies have a wide range
of variances (cost, data, etc.) and an even bigger
range of applications.
Reliability (precision, accuracy)
Versatility (applications, speed, sensitivity)
Cost (system + training + cost of runtime)
5. Imaging Technologies: Digital Camera
Digital color imagers allow the capture of objects
and scenes visible by the naked eye (400-700nm)
(color, shape, size, texture), with little to no
distortion.
Remotely, an inspector can point out other obvious
features to make reasonable predictions about the
image content of a part.
6. Imaging Technologies: Thermal Imager
Thermal systems are most commonly used for
night vision, or mold inspection due to their acute
ability to detect infrared wavelengths (750nm-
1mm) (heat), invisible to the human eye.
Infrared radiation primarily acts to set molecules
into vibration. This allows the easy detection of
hot spots, which are usually problem areas.
7. Imaging Technologies: Hyperspectral Imager
Hyperspectral imaging systems can take images
over a wide range of wavelengths (400-1000nm)
and fuse them together with spatial data.
Relative brightness for each pixel of data at each
wavelength is recorded, creating a spectral profile
or “finger print”
Varied by elevation, texture, as well as changes in
material properties.
8. Motivation
Since little specifics are known about fatigue, we
will analyze aluminum castings with three vision
systems to find defects.
Fatigue is the most common form of mechanical
failure, fatigue loading will be used to validate
break points
9. Goals
To create a repeatable process for
accurately analyzing parts and predicting
life remaining
To determine the most effective of the
three vision technologies to
nondestructively detect defects.
10. Application
This Projects simulates the defect
included on an as-cast part.
This is applicable to industry where little
machinating is needed from forging a part
to delivery
In theory, this process could be used
analyze major components of mechanical
assemblies (crank shafts, turbine blades,
hydrolic systems, aircraft wings)
11. Calculations
s=(M*c)/I, M=(s*I)/c
C=D/2
I=p(D^4)/64
D=.355’’
Assuming s=100MPa
M=F*DL
N=10^5 cycles at 180rpm
About 9 hours of cycling
Fatigue-Life Method
13. Experiments: Sample preparation & Data
Acquisition: Digital Color Image Acquisition
Lay each casting horizontally on the countertop
Turn off overhead light and flash to reduce bright
spots
Keep the camera distance from the castings constant
for each measurement (casting should be in focus, at
about 10 feet)
Take images of both sides of each casting
14. Experiments: Sample preparation &
Data Acquisition: Hyperspectral Image
Acquisition
Lay a casting on the auto-moving stage and adjust
vertical stage till casting in focus and in camera view
Set system to reflection mode and set the stage to scan in
330 micron step size increments
Set scan length to 22 mm and the integration time to 50
ms
15. Experiments: Sample preparation & Data
Acquisition: Thermal Image Acquisition
Put castings in an oven set at 200-215
o
F
After the castings reach temperature equilibrium, remove
castings from the oven two at a time quickly to avoid excessive
heat loss
Put hot castings on a black foam board to reduce noise caused
by the hot castings’ exposure to the environment, increasing
background temperature
Take measurements of the both sides of each casting, with
camera lens three feet away from the castings
16. Data Analysis Methods: Digital
Visually Inspect the part for signs of fatigue, and
obvious defects
17. Data Analysis Methods: Thermal
temperature range of interest was adjusted using a software
tool to a specific range to reduce noise in the images.
Hot spots are easily identifiable as these hot spots represent
brighter areas in the image.
Specific points were analyzed and picked as the hottest and
most substantial hot spots on the samples as points of
defects.
18. Data Analysis Methods: Hyper-spectral
These figures demonstrates the spectral profiles
obtained from the region where the Moment
applied will be at a Maximum.
These spectral profiles have similar shape, but
significant variations in reflectance values are
most likely due to changing surface finish.
19. Discussion (Validation In Progress)
Due to the Destructive nature of
Fatigue testing and the small number
of custom samples, defects have not
been completely validated.
Would like to extract more data before
breaking.
Current Hyperspectral capabilities do
not allow for integration of round
surfaces
20. Predictions
I predict that overloading will become the
greatest influence.
X=e*cos(w*t)
X’=-e*w*sin(w*t)
X’’=-e*w^2*cos(w*t)
Added force of X’’*M(mass of weight)
M*g vs. e*w^2 (1mm of deflection could
triple load)
21. References
M. Sharifzadeh, S. Alirezaee, R.
Amirfattahi, and S. Sadri, “Detection of
steel defect using the image processing
algorithms,” Proc. of the 12th IEEE
International Multitopic Conference, 2008,
p. 125-127.
R. C. Juvinall, K. M. Marshek,
Fundamentals of Machine Component
Design, third edition
22. Acknowledgements
This material is based upon work
supported by the National Science
Foundation und Grant No. 0960000
Would like to thank LSAMP for
continued support