'anbul 2004
NS AND
the images
targets and
ermined by
how global
> Ta). Only
ctors focus
I direction)
x this the
n overview
another iterative
optimization
of length
d material
real crack
k can be
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B5. Istanbul 2004
Measurement: Crack probability
$ s fy |
rd f
a 1
re CTS Bs 7e Ha
£ 7 e 7 Jo. 24 a.’ |
/ |
ee sii: |
ty === = |
N |
\ c |
m ^M
fu fuf n Be) Hs [nik
=
Figure 9: Calculation of crack probability from measured
displacement and crack width at the material
The crack probabilities are the basis for the following
calculation of crack structures.
IIl. A controlling module uses the crack probabilities and
controls the process of extraction of cracks. It decides about the
sequence and use of analysis methods and defines defaults. The
controlling is manual, but as an alternative it research is done
on automation with an expert system. This uses the knowledge
about optimisation with the decision of analysis methods in the
program.
IV. The first revision of data aims at the intensification of the
contrast between cracked and not cracked areas.
These are displayed by the crack probabilities, which are
between 0% (not cracked) and 100% (cracked). The use of
intensification methods extracts cracked areas. For the
intensification the following methods are used:
e Rules from mechanics (i.e.: dispersion length)
e Filter methods from image processing
e Crack history over the load steps
e Iterative correction by finite element analysis
After the local calculation of crack probability and
intensification, crack ways, which belong together, must be
tracked.
V. The algorithm for tracking of crack structures follows a
contour tracking method by WAHL (1987). It starts at an
initialisation edge and searches the way of highest crack
probability. There it marks located edges and a surrounding
area of dispersion length. Is the crack probability lower than a
limit the tracking is stopped and a new initialisation edge for a
new crack is searched. The way of crack is weighted in the
direct and indirect neighbourhood with the crack probability.
In addition at special specimen, i.e. tension and shear tests
cracks do not change the direction large scale, which constrains
the choice of crack ways.
VI. The graphic output is done by the finite element program
ORFEUS (CHUDOBA & KONRAD 2003) or by the CAD- System
MicroStation. The display of crack pattern allows cracked
edges on a grid, but also difference displacements. The result of
the crack extraction can be controlled by optical comparison
(figure 10).
2
AX
Figure 10: Displacement pattern and extracted crack pattern
(displayed by ORFEUS)
4.1 Detailed investigations
As an additional module a routine for the calculation of crack
width and crack edge displacement exists. For this the
following method by GÖRTZ (2004) was used (figure 11).
© p
Figure 11: Calculation of crack opening vector V, , crack width
w and crack edge displacement v from the
: P
displacement vectors V, to V, of one element
A broken element, which consists of four neighboured targets,
is separated by the crack into two parts (figure 11, crack in the
example between the targets 1, 3, 4 and 2). The displacement of
p P ; -
the targets (V, to V, ) are summarized in both parts to a mean
value (AB):
B - v, (1)
The difference (V, = Ä = B) is the relative displacement of
the crack. Because of the difference between the angle O and
the real crack angle B, the relative displacement can be
converted by using trigonometric functions in crack width w
and crack edge displacement v:
w =|V,|-cos(# - ©) ; v - Vol sin(£- ©) (2)
The calculation of crack width and crack edge displacement can
also be done using the difference displacements between two
crack-divided targets (figure 12).
Figure 12: Displacement at two edges
Crack width w and crack edge displacement v can be derived
from the formulas (1) (2).
W = [7 =~ Y -cos( B — arctan E
dpl y dpl y (3)
Vx dpi = sin f - arca ( 9-5 =X )
dol yvy dpl y
