Hi Deepak, You've chosen a difficult example for us to evaluate, because your implementation of the plane stress condition is rather hard to follow. Can you provide a reference from which you're implementing this?
At the very least your implementation of "add_point_source_to_rhs" looks suspicious to me. The ordering of the vertices has, in general, nothing to do with the ordering of the DoFs. You should use the vertex_dof_index <https://www.dealii.org/developer/doxygen/deal.II/classDoFAccessor.html#a5560151b5407e4851d5c1009c7753764> function to retrieve the DoF associated with a vertex. There are a number of posts <https://groups.google.com/forum/#!searchin/dealii/vertex_dof_index%7Csort:relevance> that discuss the use of this function. J-P On Thursday, October 6, 2016 at 7:39:28 PM UTC+2, Deepak Gupta wrote: > > Hi Wolfgang, > > Yes it is ConstraintMatrix::distribute_local_to_global. > > I have finally come up with a small example that clearly shows the issue I > am facing. Attached is the code and 3 outputs. I have a rectangle domain of > 4X8 elements. A point load (0, 1) is applied at the second vertex from the > top at the right boundary and the left boundary is fixed. The > y-displacement will be maximum at the load point. Now I look at 3 cases: > > 1) Using *globally biquadratic elements* -> displacement value is *420.2 *(as > in biquadratic.png) > *.*2) Using* globally bicubic elements* -> displacement value is *426.8* > (as in bicubic.png). > 3) Using *biquadratic for all elements*, except that the *element with > index 6 is bicubic*, displacement value is *61.45 *(as in mix.png). > > I would expect the result of *3) *to be between *1) *and *2) *ideally. I > have tested it for finer meshes, but such issues exists. Could someone help > me in locating the problem. Thanks in advance. > > Best regards > Deepak > > > > On Thu, Oct 6, 2016 at 7:16 PM, Wolfgang Bangerth <> wrote: > >> >> Deepak, >> >> By the way, I am constructing the cell matrices using B'DB, where B is >>> the strain-displacement matrix and D is the constitutive relation. Then >>> I add it to the system matrix using distribute_local_to_gobal function. >>> Since B and D are the same for global and adaptive case, I do not expect >>> any difference in the cell matrices. The system_rhs is correct since I >>> have only a point load and I can check it by printing the vector. Thus, >>> the possible issue can be only in the assembly (for the local refinement >>> one). My problem is vector-values. In case this information can lead to >>> any other suggestion from someone, it would be great. >>> >> >> I don't have any experience with this approach, but at the end of the >> day, you get a local matrix A_K = B_K' D_K B_K for each cell K, and that >> should be all that matters. I assume you mean >> ConstraintMatrix::distribute_local_to_gobal, not >> cell->distribute_local_to_gobal, right? >> >> >> Meanwhile I am going to extend the simple example for a basic version of >>> the elastic problem I am trying to solve. Hope then I can figure out the >>> error. >>> >> >> Good plan. >> >> Best >> W. >> >> -- >> ------------------------------------------------------------------------ >> Wolfgang Bangerth email: [email protected] >> www: http://www.math.colostate.edu/~bangerth/ >> >> -- >> 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 [email protected]. >> For more options, visit https://groups.google.com/d/optout. >> > > -- 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 [email protected]. For more options, visit https://groups.google.com/d/optout.
