Page 432 - Applied Numerical Methods Using MATLAB
P. 432

FINITE ELEMENT METHOD (FEM) FOR SOLVING PDE  421
                 For each subregion s = 1,... ,N s , this solution can be written as

                             N n            N n

                   φ s (x, y) =  c n φ n,s (x, y) =  c n (p n,s (1) + p n,s (2)x + p n,s (3)y)
                            n=1             n=1
                                                                         (9.4.6)
              4. Set the values of the boundary node coefficients in c 1 to the boundary
                 values according to the boundary condition.
              5. Determine the values of the interior node coefficients in c 2 by solving the
                 system of equations
                                            A 2 c 2 = d                  (9.4.7)

                 where


                       N s                T                T
                            ∂      ∂          ∂      ∂
                                                                       T
                  A 1 =       ϕ 2,s  ϕ 1,s  +  ϕ 2,s  ϕ 1,s  − g(x s ,y s )ϕ 2,s ϕ 1,s   S s
                            ∂x     ∂x        ∂y     ∂y
                       s=1
                                                                         (9.4.8)
                                  ϕ 1,s = [ φ 1,s  φ 2,s  ·  φ N b ,s ] T
                                 ∂
                                  ϕ 1,s = [ p 1,s (2)p 2,s (2) ·  p N b ,s (2) ] T
                                ∂x
                                 ∂
                                  ϕ 1,s = [ p 1,s (3)p 2,s (3) ·  p Nb,s (3) ] T
                                ∂y

                       N s                T                T
                            ∂      ∂          ∂      ∂
                                                                       T
                  A 2 =       ϕ 2,s  ϕ 2,s  +  ϕ 2,s  ϕ 2,s  − g(x s ,y s )ϕ 2,s ϕ 2,s   S s
                            ∂x     ∂x        ∂y     ∂y
                       s=1
                                                                         (9.4.9)
                                ϕ 2,s = [ φ Nb+1,s  φ Nb+2,s  ·  φ Nn,s ] T
                              ∂
                                ϕ 2,s = [ p Nb+1,s (2)φ Nb+2,s (2)  · φ Nn,s (2) ] T
                              ∂x
                              ∂                                  T
                                ϕ 2,s = [ p Nb+1,s (3)φ Nb+2,s (3)  · φ Nn,s (3) ]
                              ∂y
                                            N s

                                 d =−A 1 c 1 −  f(x s ,y s )ϕ 2,s  S    (9.4.10)
                                            s=1
                 (x s ,y s ): the centroid (gravity center) of the sth subregion S s
              The FEM is based on the variational principle that a solution to Eq. (9.4.1)
            can be obtained by minimizing the functional


                                         2              2

                                ∂              ∂
                     I =          u(x, y)  +     u(x, y)
                           R    ∂x            ∂y

                                         2
                                − g(x, y)u (x, y) + 2f (x, y)u(x, y) dx dy  (9.4.11)
   427   428   429   430   431   432   433   434   435   436   437