I have a dynamic model that undergoes repeated snap-through behavior under the influence of a sliding contact moving at a fixed velocity. I'm using Implicit Dynamics and MODERATE_DISSIPATION

I run the model through several snap-through events and output the results at fixed intervals of time. I want to sample the forces and other quantities at a uniform rate in time. In between snap-through events, the solver can take large timesteps and the time-step size rises to the maximum time interval.

I am doing it this way because I don't care about the short-timescale details of the snap-through on timescales less than the desired output timestep. The solver uses much smaller increments to successfully resolve the dynamics near snap through, but I don't want to write all that data to the .odb and I don't need to know what happens during the numerous increments of 1e-4 to 1e-5 second stepping that is needed there.

Sometimes everything is fine, and the model runs for a long time. But it always terminates before the end of the job like this:

 4   185   1     2     3     5  3.16       0.856      0.006250 

   4   186   1     0     3     3  3.16       0.862      0.006250 

   4   187   1     0     3     3  3.17       0.869      0.006250 

   4   188   1     0     3     3  3.18       0.875      0.006250 

   4   189   1     1     2     3  3.18       0.881      0.006250 

   4   190   1     1     3     4  3.19       0.887      0.006250 

   4   191   1U    4     1     5  3.19       0.887      1.065e-015

 THE ANALYSIS HAS NOT BEEN COMPLETED

There are several (five to ten, maybe?) steps taken at the largest output interval (usually it's twice this value, 0.0125s), then this happens:

TIME INCREMENT REDUCED TO   1.065E-15 TO MEET NEXT OUTPUT TIME POINT

 ITERATION SUMMARY FOR THE INCREMENT:   4 TOTAL ITERATIONS, OF WHICH

   1 ARE SEVERE DISCONTINUITY ITERATIONS AND  3 ARE EQUILIBRIUM ITERATIONS.

 TIME INCREMENT COMPLETED  6.250E-03,  FRACTION OF STEP COMPLETED  7.744E-02

 STEP TIME COMPLETED       0.887    ,  TOTAL TIME COMPLETED         3.19   

  INCREMENT   191 STARTS. ATTEMPT NUMBER  1, TIME INCREMENT  1.065E-15

The simulation stops with  ***ERROR: TIME INCREMENT REQUIRED IS LESS THAN THE MINIMUM SPECIFIED

It says in the manual that this is possible with fixed time points, that the required increment may be smaller than the minimum, and I find I can avoid the problem by setting 

myModel.FieldOutputRequest(name='F-Output-Slide',createStepName='sliding',timeInterval=OUTPUT_TIME_INTERVAL,timeMarks=OFF)

(something like "output at approximate times" in CAE), but then the time increment goes substantially beyond my desired maximum, sometimes double.

Is there a way to get more uniform timed outputs without the risk of crashing the solver or changing the solution technique?

Thanks,

Dan