HI, I'm still looking for help with this problem. it would be much appreciated. thanks
On Monday, January 15, 2018 at 11:39:16 PM UTC, Jane Lee wrote: > > Hi Wolfgang, happy new year and hope you had a good break. > > I'm back working on this and I just don't understand what I am doing > wrong. > > Firstly, I decided not to use the normal vector for now. Since the normal > vector is 2D, i wasn't sure how to implement the rest so that it is a > 'double' since my g (neumann condition vector) has 3 components when the > normal vector will only have 2? > > I went back to the beginning - step-22 tutorial and started from absolute > basics. Testing with 1D, i have started with exact solution u=(0,-y^2)^T > and p = y^3, ie purely vertical. Doing a manufactured solution type, i get > the right hand side vector f = (0, 4+3z^2, 2z)^T. with my solution, on a > rectangular domain, i have no tangential stresses anywhere, no flux on > sides (unit normal being (/pm 1,0)), and plus and minus (0, z^3 + 4z) at > the top. > > I have added the following changes to step-22: > > std::set<types::boundary_id> no_normal_flux_boundaries; > > no_normal_flux_boundaries.insert (0); > > VectorTools::compute_no_normal_flux_constraints (dof_handler, 0, > > > no_normal_flux_boundaries, > > constraints); > > for the no flux condition. > > > tangential stresses are automatically satisfied, and on the top (and > similarly for the bottom), i add the following with necessary > accompaniments: > > local_rhs(i) -= > (topstress_values[q_point](component_i)* > > > fe_face_values.shape_value(i,q_point) > * > > fe_face_values.JxW(q_point)); > > Note that I am solving this DIRECT, so my local_matrix only has three > terms (the last is omitted which is used in the schur complement). Could > this be causing problems? > > For ease of understanding and hopefully for swift closing of this issue, > i've attached the full code. When you run it, the pressure seems to be > fine, but the velocity is playing up. the error for the pressure is > converging properly too, but not for the velocity. > > Dirichlet boundary conditions work so i think it is indeed how I am > implementing stress conditions. > > Any help would be very very much appreciated. > -- The deal.II project is located at http://www.dealii.org/ For mailing list/forum options, see https://groups.google.com/d/forum/dealii?hl=en --- You received this message because you are subscribed to the Google Groups "deal.II User Group" group. To unsubscribe from this group and stop receiving emails from it, send an email to dealii+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
