Crack Paths 2009

0

0.05

Dlsilzlacomam iiaii;

Figure 5: Global behavior: experimental (bold), numerical answer (grey) and mean

(circles)

The correlation between the experimental result and the meancurve is quite good as the

experimental result is contained in the set of the numerical answers and is very close to

the meananswer. Themacroscopic model provides not only a global answer but also

somelocal information on cracks opening and distribution. In fig.6 a typical crack

pattern and crack opening curve are presented. It is interesting to observe that not only a

main macro crack is represented but also the multi-cracking character of the global

failure is represented. Finally, a numerical traction test on a cube proves the 3D

capabilities of the approach. In fig.7, a typical crack pattern of a traction test performed

on a cube is presented.

Dlsplfl(cemerlm)

-S0

-5u

—40

1.31.",

Figure 6: Typical crack patterns and crack opening (numerical -grey-, mean answer -

circles- and experimental -bold-)

1 2 1 “ I n

Int-0|

int-m

1“

m

amt-m

I: m I

\

1.95501 "2 4 1

.1

\

In

If

i \

:'

‘

ZJIIIEIH

\

m

.

2.11m

\‘

1.1"?“

X‘ I

ll

\

IJIIBII

a.‘

I

I

I

%

E

I

III

|

.I

Z

I.‘

I

l l 1

1.:

:nn‘I

Figure 7: Traction test on a cube: cracks opgning and load/displacement global ansizver

885