Setting active equations,set 'ix' for the isw=1

[Shilmin]

Dear Sir and Madam,

I am writing to ask about the Setting active equations in 5.4.1 of Programmer Manual. My question is: can I change the ix(ndf,*) during the analysis? For example, in step n, ix(1,3) = 0, and in step n+1, ix(1,3) changes to 1.

Thank you! Wish you have a good day!

Shilmin

[Prof. R.L. Taylor]

No, you cannot reset the dof this way.  You may be able to do it using partitions in which each partitiion has the dof's you want to be active.

[Shilmin]

Dear Prof. R.L. Taylor

Thank you for your timely response.

I want to do some work on the Extended Finite Element Method, there are a lot of dof's which are inactive during the analysis, and only some different parts of dof's which are active at the different analysis steps. Can you give me some suggestions? Thank you!

Shilmin

[Prof. R.L. Taylor]

You will have to write a user module (umacr*.f) to modify the equation structure as you proceed.

Assume you are doing a 2-d problem which normally hass 2-dof's/node (u, v) but you want to have additional dof's for the XFEM.  You set the maximum number of dof's on the control record that you need.  Feap will automatically delete the unused dof's (> 2) during set up.   When you do a solution step and decide what dof's to activate you need to reset the id(ndf,numnp,2) array for the "2" at the dof/node you want active (set it to zero).  Then call pnewpr() and a new profile will be created in which the added dof is considered. 

[Shilmin]

Dear Prof. Taylor

Thank you very much for your advice!

Acccording to my understanding, when computing tangent matrix and residual vector in 'elem**.f', we just need to compute the terms for the active dof's and keep the other terms being zero for the inactive dof's. Then reset the 'id' array for the '2' at a solution step which will keep the inactive dof's out. This process looks like changing the boundary conditions during the analysis.

Thank you for your patient.

Sincerely,
Min