Does number of integration points or number of DOF drive Abaqus Explicit solution time?

[jball1]

When I run a dynamic model with implicit time integration, I typically assume that solution time and number of DOF are correlated. The number of DOF is correlated to the size of the stiffness matrix, and factoring the stiffness matrix drives the solution time (NLGEOM=ON).

I am less familiar with Abaqus-Explicit. When I run a dynamic analysis in Abaqus-Explicit, is solution time more correlated with integration points or DOF?

I ask because I am building a large model in Abaqus-Explicit. The model includes a large composite shell. I will be representing the composite with continuum shell elements, which I have never used before. The continuum shell elements will be light on DOF but heavy on integration points.

Thank you for the help!

 

[FEA way]

Yes, the number of integration points influences the computational expense of element calculations in Explicit. That's why this solver mainly uses reduced integration for first-order elements.

 

[jball1]

Thanks FEA way. Do you have a sense for which of the two (# of DOF and # of integration points) have more of an influence on solution time for an explicit analysis?

 

[FEA way]

It would be best to run some simple tests (even using a single element) with different element orders and numbers of integration points. This way you will get the most specific answer. Even better if you run tests using simplified versions of your target model (the same type of elements and analysis settings, just with a very coarse mesh).

 

[jball1]

Doing that now... thanks.

 

[bkal]

I would also look into the stable timestep, a good starting point is a section on Explicit dynamic analysis of the manual.