I am working on the application of a Local maximum-entropy approximation based meshfree method to the fluid problems.

As a first step, I chose the Convection-Diffusion problem to study and decided to validate my results with the classical SUPG FEM.

In order to first validate the SUPG FEM MATLAB code, a benchmark problem that has an analytical solution is taken and performed the convergence test. In SUPG FEM as I am increasing the number of nodes from 10 to 100 the error is gradually increased as follows

10 2.22E-05

20 6.30E-05

30 1.16E-04

40 1.79E-04

50 2.50E-04

60 3.28E-04

70 4.14E-04

80 5.05E-04

90 6.03E-04

100 7.06E-04

What are the possible mistakes that took place in my code?

Considered only homogeneous mesh for this work and used vpa (Variable precision arithmetic) function where ever necessary to ensure double-precision.