Marc
When linear FEA just doesn’t cut it
The world is nonlinear. True, a large subset of models can be approximated as linear to great effect. For those models that cannot be approximated as linear however, we obviously must run a nonlinear analysis.
For structures, nonlinearity can come from only three major sources, each with subcategories:
- Geometric (Large displacement, Large strain, Follower loads)
- Material (Plasticity, Hyperelasticity, Failure, etc. [Any other nonlinear material effects])
- Contact (Touching contact, Friction)
This list contains all the possible nonlinear structural effects. All of them. If an FEA solver supports all of these, then it can perform nonlinear simulation, right? Right?
Well, kind of since it is a True Lie. While no other nonlinearities exist, there are additional functionalities needed to run some of the models that require nonlinear analysis.
Here are examples of functionality implemented in Marc to perform the analysis:
For the post-failure response of a structure, you need the ability to model the removal of failed regions during the analysis. Whether this is by removing failed elements during the analysis or removing the stiffness of a layer in a composite material applied to an element, without it you can only predict when failure starts, not how it will progress. Both are available in Marc in addition to other methods to model post-fracture behaviour.
Some materials undergo a significant deformation during the nonlinear process. Typical of this would be rubber materials or modelling the forging process. In these cases, the elements will become distorted to such an extent that the element accuracy degrades too much for useful results. In this case, remeshing is needed. Marc allows remeshing while solving to overcome these challenges.
Rubber models such as Ogden or Mooney-Rivlin use a polynomial approach to model the behaviour of the material. Finding these material coefficients requires curve fitting of the data sourced from a series of measurements (uniaxial, equibiaxial, biaxial, etc.). Mentat, the pre-processor for Marc can perform the curve fitting based on the measured data and provides visual and numeric feedback on the accuracy of the fit.
Concrete structures typically contain rebars. Meshing these in detail would result in impossibly large models. To make matters worse, changing the rebar layout requires a complete re-meshing operation. March has special rebar elements where you simply need to specify the location, amount and direction of the rebar layers which are added into the parent material (the concrete in this example). Changing the amount of rebars to use simply requires changing a number in the rebar material property. Adding a layer of rebars? Edit the rebar material, add the layer, and then specify the additional rebars' location, amount and orientation for the new layer.
Ever looked at the cross-section of a tire? There are multiple different types of rubber and often regions with a steel mesh, nylon cloth and steal beads. The same rebar element method allows modelling a tire with a lot less elements than detailed modelling would require, with very little impact on the accuracy. An additional method is available in Marc where elements can be “inserted” into others without requiring the meshes for the rebars and the rubber to match. This allows moving these inserted layers without having to recreate the complete mesh.
More functionality in Marc is available such as adding or removing materials during the analysis. This can be used to model additive manufacturing processes, welding, machining, or modelling the process of building a dam wall.
But there are other interesting interactions in nonlinear analysis as well. What would happen to a structure that can contact another while applying heating to one of them? The contact state between the components determines how the heat flows between them, while the temperature distribution combined with thermal expansion changes the contact state. This requires simultaneous solution of the thermal and structural physics. This is available in Marc.
What if the heat input was caused by electrical current flowing? The current flowing impacts the thermal solution, the temperatures impact the structural solution and the structural response impacts the flow of electricity and heat. This can also be performed in Marc.
Temperature
Voltage
Density
There are a lot more interesting additions available in Marc to allow complex nonlinear problems to be solved.
By this point it should be obvious that supporting all the nonlinear effects that can occur in structures is the bare minimum required to solve some nonlinear structural problems. Solving generic nonlinear problems often requires additional capabilities that complement the nonlinear capabilities.
This is where Marc shines: If it can be solved using FEA at all, chances are that it can be solved with Marc.
