7.4 Phase 3 - Model development
125
be supported by using integrated parameter fitting algorithms, which run the FE-model
multiple times optimizing model parameters until convergence for predefined result figures is achieved. Computer aided engineering software packages often provide such capabilities (e.g. HyperStudy 1 ). However, in order to perform such curve fitting, the model
parameters to be adjusted have to be established, if not yet done. This can be done with
the help of a design of experiment (DOE) study, which reveals the impact of several
model parameters on the virtual test results. The model parameters that are typically
adjusted to calibrate the model are material or cohesive parameters (i.e. modulus,
strength or fracture toughness), while the corresponding material or cohesive model has
been selected based on the previous numerical studies.
In case of the investigated honeycomb core, there are four non-linear experimental
curves (from the four test setups, see Figure 32 to Figure 34), which have to be fitted
using a single set of material parameters for the cell wall material. This is a complicated
task for curve fitting algorithms. Therefore, curve fitting has been performed by manual
iteration based on a DOE study. The virtual test results of the implemented material
modelling approaches after calibration are illustrated in Figure 44, page 56 and
Figure 45, page 57. In addition, it has been shown that the calibrated model is capable
to reflect the visual damage patterns as in the experiments (Figure 46).
The third step and therefore the investigation concludes with a calibrated model of the
investigated structural test. The applied model parameters have been verified and they
are transferred into the Modelling Database for implementation in the following building
blocks or in future virtual tests.
7.4.2 Building blocks
The consideration of the building blocks is the starting point of the model development
phase. The developed approach suggests three building blocks that need to be addressed (see Figure 95). The constituents generally refer to the involved materials for
core, face sheets, adhesives and local reinforcements (i.e. inserts). Therefore, the primary objective of the constituent level is the implementation of suitable material models
for all materials that can be found in the reference test configuration. The structural elements level is primarily concerned with bilateral bonding of the constituents and the
effect of the bond on the combined mechanical properties. Therefore, the modelling of
bonded zones is of primary interest. In addition, it is common practice to derive the degraded properties of bonded face sheets on structural element level (see section 4.2 on
1 Altair® HyperStudy® is a multi-disciplinary, design exploration, study and optimization software. (Altair, 1820 E.
Big Beaver Rd., Troy, MI)
125
be supported by using integrated parameter fitting algorithms, which run the FE-model
multiple times optimizing model parameters until convergence for predefined result figures is achieved. Computer aided engineering software packages often provide such capabilities (e.g. HyperStudy 1 ). However, in order to perform such curve fitting, the model
parameters to be adjusted have to be established, if not yet done. This can be done with
the help of a design of experiment (DOE) study, which reveals the impact of several
model parameters on the virtual test results. The model parameters that are typically
adjusted to calibrate the model are material or cohesive parameters (i.e. modulus,
strength or fracture toughness), while the corresponding material or cohesive model has
been selected based on the previous numerical studies.
In case of the investigated honeycomb core, there are four non-linear experimental
curves (from the four test setups, see Figure 32 to Figure 34), which have to be fitted
using a single set of material parameters for the cell wall material. This is a complicated
task for curve fitting algorithms. Therefore, curve fitting has been performed by manual
iteration based on a DOE study. The virtual test results of the implemented material
modelling approaches after calibration are illustrated in Figure 44, page 56 and
Figure 45, page 57. In addition, it has been shown that the calibrated model is capable
to reflect the visual damage patterns as in the experiments (Figure 46).
The third step and therefore the investigation concludes with a calibrated model of the
investigated structural test. The applied model parameters have been verified and they
are transferred into the Modelling Database for implementation in the following building
blocks or in future virtual tests.
7.4.2 Building blocks
The consideration of the building blocks is the starting point of the model development
phase. The developed approach suggests three building blocks that need to be addressed (see Figure 95). The constituents generally refer to the involved materials for
core, face sheets, adhesives and local reinforcements (i.e. inserts). Therefore, the primary objective of the constituent level is the implementation of suitable material models
for all materials that can be found in the reference test configuration. The structural elements level is primarily concerned with bilateral bonding of the constituents and the
effect of the bond on the combined mechanical properties. Therefore, the modelling of
bonded zones is of primary interest. In addition, it is common practice to derive the degraded properties of bonded face sheets on structural element level (see section 4.2 on
1 Altair® HyperStudy® is a multi-disciplinary, design exploration, study and optimization software. (Altair, 1820 E.
Big Beaver Rd., Troy, MI)