Page 72 - Fundamentals of The Finite Element Method for Heat and Fluid Flow
P. 72

64
                           The quantity (dx/dζ) is called the Jacobian of the coordinate transformation and is
                        denoted by [J]. For a one-dimensional coordinate, transformation [J] is calculated using
                                                  dx   dN i    dN j  THE FINITE ELEMENT METHOD
                                                                       dN k
                                            [J] =    =     x i +  x j +    x k             (3.108)
                                                  dζ    dζ     dζ       dζ
                        Example 3.2.4 Derive the shape function derivatives for a one-dimensional quadratic ele-
                        ment that has nodal coordinates x i = 2, x j = 4 and x k = 6.
                           The Jacobian matrix is written as
                                                dx
                                          [J] =
                                                dζ
                                                dN i    dN j    dN k
                                              =    x i +   x j +    x k
                                                dζ      dζ       dζ

                                                   1                   1
                                              = − + ζ 2 + (−2ζ)4 +       + ζ 6
                                                   2                   2
                                              = 2 + 8ζ − 8ζ = 2                            (3.109)
                        thus,
                                                                1
                                                           −1
                                                        [J]  =                             (3.110)
                                                                2
                           The shape function derivatives are written as follows:
                                                     
                                      dN i         dN i                     
                                                                 1          −1    ζ
                                                     
                                                     
                                                  dζ                        
                                                                                +
                                     dx                    − + ζ            
                                                                            
                                                            2         4    2  
                                      dN j      −1  dN j     1
                                           = [J]          =      −2ζ    =     −ζ           (3.111)
                                                               1
                                                                          
                                     dx           dζ     2           1      
                                                                            ζ 
                                                                  + ζ
                                                                            
                                                                        +   
                                                  dN k 
                                                              2           4   2
                                     dN k 
                                                     
                                      dx             dζ
                           For two-dimensional cases, we may express x and y as functions of ζ and η,thatis,
                                                x = x(ζ, η) and y = y(ζ, η)                (3.112)
                           Since we deal with Cartesian derivatives for the calculation of the stiffness matrix, we
                        transform the derivatives of the shape functions using the chain rule as follows,
                                                ∂N i       ∂N i ∂x  ∂N i ∂y
                                                   (x, y) =      +
                                                ∂ζ         ∂x ∂ζ    ∂y ∂ζ
                                                ∂N i       ∂N i ∂x  ∂N i ∂y
                                                   (x, y) =      +                         (3.113)
                                                ∂η         ∂x ∂η    ∂y ∂η
                        which can be written as
                                                                        
                                                    ∂x ∂y
                                                   
                                           ∂N i             ∂N i       ∂N i 
                                                                         
                                            ∂ζ      ∂ζ ∂ζ    ∂x          ∂x
                                                                         
                                                 =                = [J]                  (3.114)
                                                    ∂x ∂y 
                                         ∂N i              ∂N i       ∂N i 
                                                                         
                                            ∂η      ∂η ∂η      ∂y          ∂y
                                                                         
   67   68   69   70   71   72   73   74   75   76   77