Analysis options - 'system' and 'algorithm'

Forum for OpenSees users to post questions, comments, etc. on the use of the OpenSees interpreter, OpenSees.exe

Moderators: silvia, selimgunay, Moderators

Post Reply
sriarun
Posts: 3
Joined: Fri Dec 13, 2019 7:48 pm

Analysis options - 'system' and 'algorithm'

Post by sriarun »

I'm working on a complex Opensees model of a structural steel building that can model fatigue, buckling, fiber-based progressive yielding, etc. I would like to use mixed damping (a combination of modal damping for 10 modes and mass proportional damping for all modes as mentioned here https://www.researchgate.net/profile/Xi ... alysis.pdf). I understand that I cannot use the 'Diagonal' system since the off-diagonal terms cannot be ignored (damping matrix is full). I intend to use 'UmfPack'. However, the explicit transient integrator (ExplicitDifference) requires 'Linear' algorithm which can be problematic as this article (https://portwooddigital.com/2019/09/12/ ... l-damping/) explains. Essentially, when I use modalDampingQ, the solver will need some iterations to converge because I'm using a sparse matrix storage scheme (UmfPack) and therefore, BFGS or Krylov Newton would seem appropriate. However, explicit integrators don't let me use them. My reason for sticking with explicit integrators is that I'm trying to simulate behavior close to collapse (large nonlinearities) which implicit schemes cannot handle well. Any suggestions?
Post Reply