Time Increment Optimisation

[tvirkler1]

Hi all,

Still getting the hang of some of the finer intricacies of ABAQUS. I wanted to get your advice on job optimisation techniques. Occasionally I run into my job aborting because it the time increment needs to be less than the minimum I had set.

What are your techniques for:
a) Choosing a good time increment range to avoid this (and also not burn up the PC)
b) Increasing the required time increment

My models are relatively simple structural models looking at local failures of connection components if this provides any more context.

Cheers!

 

[FEA way]

I would recommend setting 0.1 or 0.01 (10% or 1%) as an initial increment size in most nonlinear cases. Then you can increase the maximum number of increments because the default value (100) is often too low. Reducing the maximum increment size is often important when using amplitudes or simulating transient phenomena (to capture them with a proper resolution - Abaqus may use too large increments and even skip amplitude values if convergence allows it). Finally, you may also decrease the minimum increment size but not too much. Use 1E-6 or 1E-7 if you want but not lower. It just doesn't make sense in most cases - if Abaqus has to reach so low increment sizes, it's very unlikely to converge anyway and you only delay this. So, despite the classic non-convergence error message mentioning too high minimum allowable increment size, it's only a consequence of some convergence issue and that issue should be investigated instead of just following the message directly.

Recently, a nice article about the basics of debugging non-convergence issues in Abaqus was posted on the Dassault Systemes blog. Look for "Debugging Abaqus Models" there.