Summary of the Aluminium Honeycomb Simulation
The video shows a pressure load applied to the top Carbon Fibre Reinforced Polymer (CFRP) skin layer on a curved aluminium honeycomb structure. The plot of velocity magnitude (z-component) allows us to visualize the propagation of the wave through a non-defective component.
The microstructure of composite structures significantly increases the complexity of simulating wave propagation: analytical and semi-analytical approaches no longer appropriate for most applications and transient FEA problems can become very large. OnScale is capable of running multi-billion element simulations, allowing scope to capture the necessary detail in the composite microstructure.
OnScale offers a variety of anisotropic material models that allow accurate simulation of CFRP behavior:
- Transversely isotropic materials: useful for simulating yarn bundles, and chopped strand matting
- Orthotropic materials: useful for simulating woven layers
- Fully anisotropic: useful for layers with complex weaves, using full stiffness tensor
Structures can be represented using a combination of continuum and shell elements, allowing great flexibility over the level of detail included in the simulation.
Technical details of the OnScale simulation
The modeled curved component material specifications:
- Hexcel IM7/8552 CFRP, 8 ply, [45/90/-45/0]s, 1 mm thick
- 20 mm thick with 20 mm cells, 0.5 mm wall thickness
The excitation signal was a 20 kHz pulse applied as a pressure load on the top CFRP skin. The model was simulated in quarter symmetry with the outer surfaces at X-max and Y-max fixed.
In NDT where bulk wave propagation dominates it is clear that individual yarn bundles will scatter sound, and can have an impact on the inspection. At lower frequencies where surface waves dominate, it may be appropriate to approximate the structure as a series of layers or shell elements. Using shell elements also avoids the issue of drastically decreasing the simulation timestep when modeling thin structures which will increase runtime.
OnScale also allows users to create custom meshes to accurately represent complex shapes and curved geometries in simulation.
Computational Resources required for the OnScale simulation
- Elements: The model used 537,600 skewed elements to setup the grid space and 122,080 shell elements
- Processor: 2 x 8 core Intel Xeon E5 Workstation
- RAM: 276 MB
- Time required: 14 minutes, 22 seconds
The simulation produces a number of useful outputs, including:
- Video of wave propagating through component to understand where reflections may be coming from
- Analysis of various metrics over time such as velocities, displacements, stresses and strains.