## General Mechanical

#### Sub step in non linear analysis

• Cri1994
Subscriber
Hi everyone How can I know the number of necessary substeps to have a good results?
I’m doing a non linear transient analysis.
Can I calculate it ?
• jj77
Subscriber

Better to define dt (substeps divides the total step time by the amount of substeps) - so use define by Time instead.

Typically if you are looking at dynamics say of a building (say earthquake) then typically one can take dt~1/(10*freq_max). freq_max is the largest frequency of interest.

If you are looking at impact then take the dt~1/100*(impact duration) so it is dictated by the impact event.

Also if it is a shock problem the dt is described by the shock duration.

More info can be found here (an in the help manual - look for transient analysis)

I

• Cri1994
Subscriber
I’m studying an impact. The impact duration is 0,026 s. So the time step should be 1/100*(0,026). Is it right?
• jj77
Subscriber

That is correct. Basically we want to get enough time points to capture the impact event. Even if this is close to some resonance of the structure we can still capture frequencies of about 384 Hz (based on the formula above for dt). Because in impacts it is the relative relation of the time duration to the period of the structures that will give rise to amplification see here:

https://pdhonline.com/courses/s164/s164content.pdf

back to FEA. have in mind though that if it is NL then the solver might need to cut down on the time to get convergence thus we can have a minimum time that is 1/10 (or smaller say 1/100) of the time step you propose so the min step will be 1/1000*0.026).