I am solving the following mixed boundary elasticity problem with the finite element method on the unit circle, using a triangle mesh with 3 nodes on each element:

$ \frac {\partial}{\partial x_j}\sigma_{ij}(u(x_1,x_2))=0 $, $(x_1,x_2) \in \Omega$

$ u_1(x_1,x_2)=u_1^{an}(x_1,x_2)=\frac{1-\nu}{2G(1+\nu)}\sigma_0x_1$ on $\Gamma$

$\sigma_{2j}(u(x_1,x_2))n_j(x_1,x_2)=t_2^{an}(x_1,x_2)$ on $\Gamma$


$ t_2^{an}(x_1,x_2)=\sigma_{21}^{an}(x_1,x_2)n_1+\sigma_{22}^{an}(x1,x2)n_2(x_1,x_2) $ is the 2nd component of the traction vector $ t^{an}=(t_1^{an}, t_2^{an}) $

$ \sigma_i^{an}(x_1,x_2)=\sigma_0 \delta_{ij} $, $i,j=1,2$

$\sigma_0=1.5 *10^{10} N/m^2$ , $ \nu=0.34 $ , $G=3.35*10^{10}N/m^2$

With the following weak form:

$\int_\Omega \sum_{i.j}^2 \sigma_{ij}(u) \epsilon_{ij}(v)= \int_\Gamma v2*t2 $

I have already implemented it correctly for the pure Dirichlet problem. (so the global stiffness matrix is also good for this problem), but I do something wrong when implementing the Neumann boundary condition and I cannot figure out what. (I still get the right solution for u1 where I have the Dirichlet condition).

This is what I do when implementing the Neumann boundary condition:

I go over each triangular element:

-> I extract the coordinates of each node of the element: (x1,y1), (x2,y2), (x3, y3) (the elements are numbered in anticlockwise sense)

-> if (x1,y1) and (x2,y2) are on the boundary:

  • I calculate the outward unit normal: $$ n2=\frac{x_1-x_2}{\sqrt{(y_2-y_1)^2+(x_1-x_2)^2}} $$
  • I update the force vector F in the place corresponding to $C_e^12$ and $C_e^22$ ($v_2=\sum_{B=1}^3 N^BC_e^B2$) by adding : $$ \sigma_0*n_2*\sqrt{(x_2-x_1)^2+(y_2-y_1)^2}*0.5 $$

-> if (x1,y1) and (x3,y3) are on the boundary:

  • $ n2=\frac{x_3-x_1}{\sqrt{(y_1-y_3)^2+(x_3-x_1)^2}} $
  • update with $ + (-1)*\sigma_0*n_2*\sqrt{(x_3-x_1)^2+(y_3-y_1)^2}*0.5 $

-> if (x2,y2) and (x3,y3) are on the boundary:

  • $ n2=\frac{x_2-x_3}{\sqrt{(y_3-y_2)^2+(x_2-x_3)^2}} $
  • update with: $ +\sigma_0*n_2*\sqrt{(x_3-x_2)^2+(y_3-y_2)^2}*0.5 $

The methodology seems fine. The issue could be due to bugs in your code. Check your code carefully and see if there are any bugs.

  • $\begingroup$ I request people to explain why they downvoted my answer. $\endgroup$ – Chenna K Jul 27 '17 at 20:59

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.