Contact and linear analysis
Contact and linear analysis
(OP)
Hello,
I made an analysis with 3D elements and I defined contact(to create a welded contact between tubes. I defined my material properties as linear isotropic. I checked all the results(eq stress, eq strain, rotation,displacement) in order to validate the linear elasticity field, and it's ok.
But since I used contact (MPC algorithm with Always bonded condition) which have a non linear behaviour(I guess), can I validate my model and these results in linear elasticity?
Thank you for you help,
Regards,
Mickaël
I made an analysis with 3D elements and I defined contact(to create a welded contact between tubes. I defined my material properties as linear isotropic. I checked all the results(eq stress, eq strain, rotation,displacement) in order to validate the linear elasticity field, and it's ok.
But since I used contact (MPC algorithm with Always bonded condition) which have a non linear behaviour(I guess), can I validate my model and these results in linear elasticity?
Thank you for you help,
Regards,
Mickaël





RE: Contact and linear analysis
yes. The non-linearity here comes from the existence of the "Contact": probably you have used as default to "update contact stiffness at each equilibrium iteration" or "at each step", which causes the solver to enter the non-linear mode (and which is not helpful for a bonded contact). You'd better check that you have "update contact stiffness -> never": in this case, you remain in the linear field by definition (the contact stiffness is evaluated in the initial condition and then never updated, just like the other structural K terms).
Regards
RE: Contact and linear analysis
I try to "never update contact stiffness", but I didn't find it in the contact element type option, there are several option for "update contact stiffness"(K10) but not "never".
Thank you,
Regards
Mickaël
RE: Contact and linear analysis
sorry, I forgot that for MPC, Keyopt(10) is disregarded.
For a well-conditioned problem, however, the Contact in this case won't exhibit non-linear behaviour in a linear problem, so your "all-linear" hypothesis is preserved, even if the program shifts to non-linear solver techniques for internal numerical reasons.
Regards
RE: Contact and linear analysis
Mickaël