Page 280 - Numerical Analysis and Modelling in Geomechanics
P. 280

ENRICO PRIOLO 261








            and C is the elastic stiffness matrix, given by










            Using  this  notation,  the  generalized  Hooke’s  law  is  written  as  σ=Cε,  and  the
            components of the strain vector ε are related to the displacement field by ε=Du.
              In the SH case (equation (9.1)), Ξ=u , and
                                           y
                                                                         (9.8)


                                                                         (9.9)



                                                                        (9.10)


            Finally, in the case of the pressure waves equation (9.3), the bilinear forms are
            similar to those of the SH case (9.8–9.10), with the following changes: p→u , 1 /
                                                                         y
              2
            (ρc )→ ρ and 1/ρ→µ.
              Equation (9.5) can easily account for a more general constitutive equation, just
            by changing matrix C coefficients. See, for instance, Seriani and Priolo (1995) for
            the extension to the case of a transversely isotropic medium.
              In the following two sections, the author first introduces the discretisation of
            the physical domain, and then that of the wave equation. For the sake of brevity,
            the latter is done only for the P-SV case (equations 9.4–9.7), which is the most
            general. A detailed treatment of other cases can be found in: Priolo and Seriani
            (1991)  for  the  1-D  acoustic  equation;  Seriani  et  al.  (1991,  1994)  for  the  2-D
            acoustic equation with straight and curved elements, respectively; Seriani et al.
            (1992) for the 2-D elastic equation with curved elements.

                             Discretisation of the physical domain

            The  first  step  of  the  discretisation  process  is  to  build  up  the  approximating
            functional  spaces.  To  do  this,  the  spatial  domain  is  decomposed  into
   275   276   277   278   279   280   281   282   283   284   285