I am trying to solve set of PDE's by using Gelarkin Finite element method for gaseous phase adsorption of VOCs on activated carbon for drawing breakthrough curves. I have used a quadratic polynomial as a basis function. I have taken 5 elements (i.e. 11 nodes). I have a made a program in MATLAB. When I run the program it gives absurd results like it says matrix is close to singular and badly scaled. Now my question is whether 5 elements are sufficient or should I increase the no. of elements? and how to get rid of this error.
Hi, you will find numerous links to existing software implementing this approach here:
http://www.swmath.org/?term=Galerkin%20method
Here are some helpfull links:
FEM using Matlab in practice - http://vefur.simula.no/~hpl/INF5620/books/Larson_Bengzon.pdf
Application sample - http://file.scirp.org/pdf/JSEMAT_2016042714593445.pdf
I hope they will be usefull to you.
My first would be to check the boundary conditions. Also if you have programmed the element from scratch (i.e. you have programmed also the numerical integration for the in plane discretization) check the shape functions assignment to the corresponding degrees of freedom.
- Badges
- Science topic