How to minimize the computational error of a particular measure which is solved through a subroutine?

The contact subroutine CNFSUB is giving slightly different result for the same set of parameters that is used in ADAMS VIEW contact module. What is the reason behind this difference?

0 Martin Kieltsch over 7 years ago

I'll put my eggs in the basket with the label "different handling in Solver prediction phase".

The theory behind that guess is that the build-in contact uses an analytical derivative whereas the usersub has to do it numerically?

At least the CNFSUB example doesn't have such section/variables whereas for example a fiesub can calculate partial derivatives.

I'd try tightening ERROR until the solutions converges better ?
Cancel
Vote Up 0 Vote Down

Sign in to reply

Verify Answer

Cancel
0 Jesper Slattengren over 7 years ago

And HMAX....
Cancel
Vote Up 0 Vote Down

Sign in to reply

Verify Answer

Cancel