Fig. 5-12 Finite Element mesh used for the simulations
This experiment has also been simulated
with the finite elements technique. The mesh definition and the calculations
have been performed by Dr. Laurent Humbert (LMAF, EPFL). We present in Fig. 5-12
the defined mesh, where only one eighth of the sample has been considered due
to the symmetry properties of the sample (adding limits conditions). The mesh
density is increased near and inside the fiber region and near the notch
region.
The linear behavior of the sample to
axial stress loading allows defining a normalized strain distribution f(z) at
the fiber core location
|
(5-10) |
This normalized strain function is
presented in Fig. 5-13 (top). From equation (5-9), we have the
experimental axial strain distribution ez,b(z) for z Î [-1.4 , 9.1]. In this range, the minimal strain ez,b(z = 0)
can be used with the normalized value at the origin
f(z = 0) = 1.0476 to calculate the finite element
simulation axial strain distributions for the different loading cases. These
simulations are presented in Fig. 5-13 (bottom).
Fig. 5-13 Normalized strain distribution along the fiber (top) and axial
strain distributions for the four loading cases (bottom)
The comparison with the experimental
axial strain distributions is shown in Fig. 5-14. An overall agreement is
observed but the position scale between experimental and calculated strains
does not match well. This effect is not explained yet and further
investigations on the finite element simulations are currently conducted.
Fig. 5-14 Left : axial strain distributions for different axial stress
loading forces and right : difference with the axial strain value at
z = L; the discrete points represent the experimental results
from the OLCR measurements and the lines represent the calculated values obtained
with the finite element method
|