I trying to optimize the material on a shell by combining two material (higher and lower performance ones) on a single shell. I have segmented the mesh in two meshes based on principal stress and assigned a lower performance material in the areas with lower principal stress values. While verifying the results, I also tested assigning the same material to both meshes to see if the results are the same as when assigning the material to a model with a single mesh. We assume that the materials are perfectly bound to one another and there are no cold joints. In this sense the meshes are used to define the areas of different materials but they should be considered to work as one model.
Here, it the calculations were different. For example, when assigning the higher performance material to a single mesh the utilization domain is -7.9 to 49 but when I assign the same material to the model with two meshes the utilization is -7.9 to 81. The domain is different in tension however other calculations (mass and displacement) are the same.
Thank you in advance for the support! I have been trying to figure out if Karamba can be use for multi-material shell simulations and your advise or support on this will be very much appreciated!
From what I can see, you’re working with two types of steel that have different strengths but share the same Young’s Modulus in your calculations. That’s why the stress distribution itself must be identical in both the strong–strong and strong–weak material cases – the only difference lies in the stress-to-strength ratios.
The variation you’re noticing in the tensile stress/strength ratio might come from the stress peaks at the point supports, which act like infinitely sharp needles. To reduce these singularities, you could try modeling the supports in a more realistic way, for example by embedding short beam elements (like small crosses) into the shell.
I am actually working with two types of custom concrete, which have both different strengths and different Young’s Modulus. The general idea is to test concrete panels for lifting conditions, assuming four lifting anchors close to the corners and a perpendicular load (representing self-load).
What I noticed is that the difference comes if I am using the ShellView component, directly linking the model from the Analyze component. However, when I pass the model data from Analyze to ModelView and then to the ShellView component, the utilization data seems to work consistently.
