Global Equations with Sparse Solver

[epn]

Dear All,

I have recently started using global equations to apply a constraint on the average of a field over the whole finite-element domain, which works for iterative solver. However, when I have a global equations, sparse solver (dire,spar) does not work, meaning it does not converge. If I would like to fix this problem, where in the source code should I take a look at? I am using FEAP8.6.

Is it possible to partition the problem into two parts and solve the global constraints in a staggered manner?

Thanks for any input in advance.

[Prof. S. Govindjee]

How have your implemented you global constraint?

[epn]

I apply the constraint using a Lagrange multiplier as discussed in the programmers manual in Sec. 5.11 ("Global constraints on elements"). I have a user element where I add the terms associated with the constraint in the suitable locations of the element residual vector and tangent matrix. I access the Lagrange multiplier from ulg(*) array.

[Prof. R.L. Taylor]

Adding the global constraint should be possible, but will involve quite a few changes. I will look a bit to see what might be the best option.

[Prof. R.L. Taylor]

This requires more changes than I am able to do at the present time.

[epn]

Thank you for the respsone.