Keep Simulink simulation stable

I would like to prevent my Simulink model from crashing due to dynamic instabilities. My coworker mentioned some measures that have stabilized his models:

• Replace switches, which cause abrupt changes, with Saturation blocks, though I'm not clear on implementation
• Avoid differentiation. For example, simulate position and integrate to get acceleration, rather than simulate acceleration and differentiate to get position.

What other measures can help stabilize a Simulink model? If there is documentation that covers this, please let me know - haven't found any.

An overview of keeping a simulation stable and accurate is at http://www.mathworks.com/help/toolbox/simulink/ug/f11-8983.html

Answer 1

The points that you describe are for stability with respect to system dynamics - not software stability (ie. crashes). In general, you should never see a crash with Simulink unless you have written C-MEX S-functions that are causing a SegV. If you see a crash due to Simulink library blocks, that should be reported as a bug to MathWorks Tech Support.

Then I will have to check whether the system dynamics are going unstable (there are all kinds of complicated restraints that might be incorrectly implemented). Would love to see a list of 'best practices for dynamical stability of Simulink models' if you have run across it...

K E: I was only trying to point out that your title "Keeps Simulink from crashing" is inaccurate, because you are talking about system stability, not software stability. I'll let you know if I see anything more than the documentation link that you have already posted for system stability.

You may also want to edit your subject to make it more relevant. Other experienced users may ignore your question thinking that you are asking about a software crash.

I changed the title as you suggest. Thanks for pointing me to dynamic rather than numerical stability as the source of my problem.