Page 196 - Fundamentals of Computational Geoscience Numerical Methods and Algorithms
P. 196

8.2  Some Numerical Simulation Issues Associated with the Particle Simulation Method  187

                           ∂u s         ∂v s         1     ∂u s  ∂v s
                      ε x =   ,    ε y =   ,    γ xy =      +      ,     (8.25)
                           ∂x           ∂y           2   ∂y   ∂x

            where σ x and σ y are normal stresses of the solid matrix in the x and y directions; ε x
            and ε y are the normal strains of the solid matrix in relation to σ x and σ y ; τ xy and γ xy
            are the shear stress and shear strain of the solid matrix; u s and v s are the horizontal
            and vertical displacements of the solid matrix; E and G are the elastic and shear
            modulus respectively; v is the Poisson ratio of the solid matrix; f x and f y are the
            body forces in the x and y directions;.
              Note that Eqs. (8.20) and (8.21) represent the equilibrium equations, whereas
            Eqs. (8.22), (8.23), (8.24) and (8.25) are the constitutive equations and strain-
            displacement relationship equations, respectively.
              By using either the variational principle or the Galerkin method (Zienkiewicz
            1977, Lewis and Schrefler 1998), the above-mentioned equations can be expressed
            in the finite element form:

                                         e  e     e
                                      [K] {u} ={F} ,                     (8.26)
                    e
                                                             e
                                                     e
            where [K] is the stiffness matrix of an element; {u} and {F} are the displacement
            and force vectors of the element.
              In the finite element method, the quasi-static equilibrium problem is solved in
            a global (i.e. system) manner. This means that the matrices and vectors of all the
            elements in the system need to be assembled together to result in the following
            global equation:

                                       [K]{u}={F},                       (8.27)

            where [K] is the global stiffness matrix of the system; {u} and {F} are the global
            displacement and force vectors of the system.
              Similarly, in the distinct element method, the quasi-static equilibrium equation
            of a particle is of the following form:
                                                   p
                                        p
                                            p
                                     [K] {u} ={F} ,                      (8.28)
                                                     p
                    p
                                                             p
            where [K] is the stiffness matrix of a particle; {u} and {F} are the relative dis-
            placement and force vectors of the particle.
              In order to reduce significantly the requirement for computer storage and mem-
            ory, the distinct element method solves the quasi-static equilibrium equation at
            the particle level, rather than at the system level. This requires that a quasi-static
            problem be turned into a fictitious dynamic problem so that the explicit dynamic
            relaxation method can be used to obtain the quasi-static solution from solving the
            following fictitious dynamic equation:

                                                        p
                                    p
                                       p
                                              p
                                                 p
                                [M] {¨ u} + [K] {u} ={F} ,               (8.29)
   191   192   193   194   195   196   197   198   199   200   201