From Fedem FAQ

Contents 1 Welcome to the Fedem FAQ 1.1 Q: Why will not some parts reduce in my model ? 1.2 Q: Why does not the beam part reduce ? 1.3 Q: Why does my model diverge / does not solve ? 1.4 Q: What do I do when my model is diverging / does not solve ? 1.5 Q: How do I see what is going on during Initial Equilibrium calculation ?

[edit] Welcome to the Fedem FAQ

Here You will find answers to frequently asked questions about the Fedem Software, and how to use it. This is still in an early experimental stage, but we hope to develop it into something useful soon.

[edit] Q: Why will not some parts reduce in my model ?

A: Errors in reduction process may have several causes. Check that the mass is correct in the *.res file in the result file browser. Also check the material properties directly in the *.nas file. If there is something wrong with the mesh input, common problems are coincident nodes/elements, loose nodes or disconnected mesh.

[edit] Q: Why does not the beam part reduce ?

A: If none of the above suggestions helps, typically for beam parts is that they contain a less number of nodes than solid parts. If triads are placed on most nodes, there is not enough DoFs for the part to calculate all component modes. Try to set the component modes to zero in the "reduction options" for the part.

[edit] Q: Why does my model diverge / does not solve ?

A: This is really a frequently asked question, but it is impossible to answer in a FAQ, because the answers is normally completely different from model to model. The question that can be answered, however is:

[edit] Q: What do I do when my model is diverging / does not solve ?

A: This is a somewhat incomplete list of steps to go through to "debug" or pinpoint problems in your Fedem model:

1. Read the error messages, and try to understand them.
2. Then read them again.
   Did the solver start, or did the pre-solver test report some errors ?
   1. The solver did not start:
      Try to fix the modeling errors reported by the Pre Solver Test.
      This could be triads that you have forgot to attach, you are trying to solve a model without any degrees of freedom, some link does no have any triads etc.
   2. The solver started:
      Open the Result File Browser and have a look at the fedem_solver.res file.
      When did the solver fail ?
      1. Before the anything really starts.
         This seldom happens. You would probably want to contact support on this one. Could be a bug.
      2. Just as the Equilibrium Iterations are about to start.
         Singularity in the Equilibrium model is by far the most common reason for it to fail. That means that some DOF(s) in your model do have a zero or very low stiffness compared to the rest of the mechanism. The algorithm used for the equilibrium solution depends on all dofs having some kind of stiffness. That means that parts that is free to move some way or another (in the exact modelling position) needs to be constrained in some way during the equilibrium calculations.
         This is normally done either by adding additional boundary conditions to triads, or by adding some small fake stiffnesses to certain joint DOFs.
      3. During the Equilibrium Iterations
         This happens often due to model errors. It can be quite challenging to get the initial configuration correct, when it it different from the modeling position.
         It might also fail due to the path the model is moved toward equilibrium. It might take a path that is impossible or result in a model that is numerically instable.
         1. Have a look at the iterations that Fedem actually does. (see question below on how to create an animation of this)
            It will often give a clue to where the model errors are, or how to tweak the model or the additional boundary conditions to avoid problems.
         2. If the model is moving too much each step, reduce the Iteration step size limit.
            (Dynamics Solver Options, Initial Equilibrium tab).
      4. Just as the dynamic solution is about to start.
      5. During the iterations for the first timestep
      6. At one of the first timesteps
      7. At a time step somewhat into the analysis.
3. Check the Links (Parts). Do this by opening the Result File Browser and look at the ".res" file for each reduced link.
   1. Look at the eigenvalues of the reduced links.
      Make sure that the 6 first eigenvalues are small compared to the ones from nr. 7 and up.
   2. Check the masses and inertia for each link reported from the Reducer.
      Make sure they are sane. This is to rule out unit conversion trouble, or other FE-model conversion related issues.
   3. Look for warnings about negative pivots.
      They are indicating that your FE model is not numerically robust, and are a potential threat to a stable and robust dynamics solution.
      Fedem is particularly picky on the quality of shell meshes, so:
      Shell + negative pivots + Dynamics Solver Trouble = Probably need to remesh.

[edit] Q: How do I see what is going on during Initial Equilibrium calculation ?

A: There are some hidden options in Fedem that will make the Dynamics solver output the results for each iteration. In that way, you can post process the progress of the initial Equilibrium solution in the same way as for the normal dynamics. That means curves and animation. The option to use, that has to be entered into the Dynamics Solver field in the Additional Solver Options is :

"-saveIterStart -1 -savestart -1 -saveIterStop 1e-9"

To actually animate these results, you need to create an animation that starts from time = -1 to 0.