Is it possible to use anisotropic materials in Karamba3d? I know orthotropic is possible, to my understanding this is a material that has different material properties in two different directions. I want to calculate 3D printed concrete, having different properties in X, Y and Z direction. Is it possible to create a material like this in Karamba3D?
at the moment (Version 1.3.3 & 2.3.0 WIP) isotropic and orthotropic material behaviors are the only options in Karamba3D. In addition, to accurately model the material in X,Y and Z direction it would be necessary to apply volumetric finite elements - which are also not there yet.
A work around could be to use a dense volumetric mesh where equivalent truss elements form the cell edges. The question is however, whether such a simulation fulfills your requirements regarding accuracy with respect to displacements and ultimate limit state.
I don´t know the 3D printing technology you´re using, I´m only used to Fused Filament Fabrication (Fused filament fabrication - Wikipedia) 3D printing technology, si I don´t know how well would this apply for concrete 3D printing.
In FFF, the filament layers are usually printed in orthogonal directions alternatively in the XY plane. This means that if the first layer is printed in +45º direction, the second one will do in -45º direction, and so on. This arrangement of layers, is not quasi-isotropic, but if your slicer allowed you to use a quasi-isotropic layout (Carbon Fiber 101: What do Isotropic, Quasi-Isotropic, and Anisotropic Mean? | DragonPlate) (for example, like a 0º/-45º/+45º/90º laminate) it would be considered isotropic and facilitate modeling, at least in the XY plane.
But this would strongly depend on the relative dimensions of the printed parts. It would only approximate an isotropic material for relatively thin shells. If you are printing bulky objects like a house… it will definitively not work.
By the way, @karamba3d, is the implementation of a solid finite element in Karamba´s development road map?
Hi Vigardo, thank you for the quick reply.
We use a similar printing technique; Simply put we use a robot arm to move a nozzle around and use a pump to steadily pump the wet concrete trough (it’s not pourable, but slightly stiff concrete). We only print the shell, most often one wall thick (sometimes two) and we often print two or four bottom layers where we use a concentric infill pattern. The nozzle is 3/4" in diameter, which results in a layer width of about 30mm and a layerheight of 10mm. The shapes are often organic and not larger than 1.5m by 1.5m, and 1 meter tall.
Therefore talking about rotated layers is not relevant I think (like I said, we only print the shell) and I’m not sure if this is even possible in finite elements softwares like Karamba, because of the organic and unpredictable shapes we make and the material having a lot of different directional properties.
I was thinking that maybe every layer can be seen as an orthotropic material, with the first direction being tangential to the printpath and the second direction perpendicular to that. I might be wrong however, I don’t understand orthotropic materials fully yet. Also, I’m not sure how to get the layers to connect to each other and this approximation would only work with single wall thicknesses.
I don’t think a dense mesh is what I need, because of the way we print. Like I mentioned in the message to Vigardo, we only print the shell of solids, with a filled in bottom. A dense mesh would have to be fitted inside the layerwidth of 30mm’s and follow the contour of the product, because the material properties follow this contour as well. I think this will not be accurate enough, but if you think this might work I’ll give it a try!
If your layers are thin in one direction (e.g. Y), then you can use as direction 1 the X axis and as direction 2 the Z axis. This would be similar to printing relatively thin walls.
This way, the orthotropic FE model should work… at least for displacements prediction, and as long as you provide reasonable E1 and E2 stiffnesses.
I really don´t have any idea about the material strenght failure hypothesis that applies here.
If the thin shell approximation des not apply, you will have severe inaccuracies because some crossterms of the anisotropic stiffness matrix can´t be neglected. A solid material model (anisotropic or not) should be used instead in such cases.
I’m having trouble understanding the way orthotropic material is calculated. As I understand, FEM takes the mesh edges and uses them as beams, to which simple mechanical principles are applied. In the case of an orthotropic material, does the system use the direction of the beam as one direction and every radial direction, perpendicular to the beam, as another? Or does it take the beam as one direction (X) and perpendicular in the Z axis as a second direction, ignoring the third (Y) axis? Or is it something else?
I need to know the way the material properties are used, as to give the system the right data. The walls are 30mm’s thick, so I’m also not sure if I should use an infinitely small wall thickness for the orthotropic calculation, to give it as little error as possible, or to use the 30mm thick wall. Understanding how this is calculated will help me to make the right choice here as well.
in the Finite Element Method volumetric- and shell-elements are not discretized as beams at the mesh edges. The relationships between the nodes result from integrating the weak form of the partial differential equations that govern the problem over the element volumes.
In Karamba3D the direction of orthtropy can be arbitrarily set (see here).
If the wall thickness is small as compared to the other dimensions you can use shell elements for the continuity regions where the flow of forces is not disturbed by concentrated external loads or boundary conditions.