Troubleshooting

When the Radioss stops with the one of following error messages:

If this error message is printed, two cases are possible.
STIFFNESS MATRIX IS NOT DEFINITE
  1. The stiffness matrix [ K ] = [ K0 ] + [ Kg ] (geometrical) is not definite. This can happen when some of the DOF values in stiffness matrix are zero (for example: one of the rotational stiffness of a spring TYPE8 is defined as zero), or elastic modulus for a material is too small (often related to inconsistencies in the units used in the model).
  2. The model still has rigid body modes even after boundary conditions have been applied. This is a more common source of this error. Examples include two parts that are connected only by contact interfaces or a 6 degrees of freedom spring connect to a 3 degrees of freedom solid element.

    Simulations fail due to the implicit resolution (matrix inverse), even with an iterative (PCG) solver.

    Radioss implicit will automatically fix these free degrees of freedom in the first case (refer to Radioss Engine Input for more details).

    To check the rigid body modes in the second case, employ one of the following methods by running:
    • A model check (with /IMPL/CHECK in the Engine Input file). The results of this check will be listed in the *n.out file.
    • A Linear Quasi-static Analysis (a direct solver is preferred) with a sufficiently large time step. The free part can be found by reviewing the animation files.
Some possible workarounds for eliminating such rigid body modes are:
  • Add artificial springs with (small stiffness) to connect the free parts.
  • Create slight initial penetration in the contacts, if interfaces are defined between the free parts.
  • Run a Quasi-static Analysis.
The error message below occurs when no loading has been defined in the model or when loading is zero during the first step in a nonlinear analysis.
IMPLICIT STOPPED DUE TO LOADING DATA
The run terminates with the following Radioss message:
**RADIOSS IMPLICIT STOPPED DUE TO LOADING DATA** 
ISTOP =       0
A few suggested workarounds are:
  1. Check loading in the model.
  2. Increase the initial time step.
  3. Create a slight contact penetration, in cases where loading is due to contact.
If the error message shown below is printed:
TIMESTEP REDUCED DUE TO DE-ACTIVATION
The following warning could be printed in nonlinear simulations with contact:
**TIMESTEP WILL BE REDUCED TO AVOID DE-ACTIVATION IN INTERFACE**: 14
In such a case, Radioss will try to restart the iteration with reduced time step and eventually the run might fail, if the time step drops below the prescribed minimum. This is usually seen in models with interface TYPE11 (edge to edge contact) with small gap.
** ERROR : RADIOSS IMPLICIT STOPPED DUE TO MODELING DATA **
     ISTOP=      -1

This message is generally following different modeling error messages: "STIFFNESS MATRIX IS NOT DEFINITE", "...TOO LARGE ENERGY VALUE".

Often, it was about the constraint conditions or too large time step.
**ERROR: RADIOSS IMPLICIT STOPPED DUE TO TIMESTEP LIMIT**
    ISTOP =      -2

When the time step is smaller than the input minimum one (see /IMPL/DT/STOP), Radioss will stop the run with this message.

The reason is generally due to the convergence issue (DT_min), if DT_min was already small enough or not. There is no general workaround for convergence issue, as outlined below:
  1. Adding more printouts in out files (/PRINT/n, for example, n=1, /IMPL/PRINT/NONLIN/n, for example, n=1), and analyzing the reason of failing.
  2. Check the model, constraint conditions, materials, contact interface defines and so on.
  3. If the fail was due to instability, try /IMPL/DT/3 (Riks method) for snap-thru type simulations or /IMPL/QSTAT for general instability case.
  4. Switching other convergence criteria (refer to Itol in /IMPL/NONLIN) or losing the tolerance, if possible.
  5. Try /IMPL/NONL/SMDISP to see if the issue came from geometric nonlinear.
IMPLICIT STOPPED DUE TO SOLVER ERROR
The run terminates with the following Radioss message:
"**ERROR: RADIOSS IMPLICIT STOPPED DUE TO SOLVER ERROR** 
     ISTOP =      -4"
Two suggested workarounds are:
  1. Check which solve method is used in /IMPL/SOLVER. Has the direct solver (BCS) been used in SMP run? If it is, then use another solve method.
  2. Another frequent case is when the model is relatively large and the available free disk is not large enough. In that case, check the working space free volume (in ./ if TMPDIR is defined). If the issue came from memory management, you could also force Radioss to use out-of-core resolution (/IMPL/SBCS/OUTCORE).

If this error is still there with the above workarounds, contact your local Altair Simulation support team.

If BCS or mix solver (using BCS) was used, BCS error IDs will printout on the screen of running shell, report them to support team.