Page 201 - Fundamentals of Computational Geoscience Numerical Methods and Algorithms
P. 201

192        8  Spontaneous Crack Generation Problems in Large-Scale Geological Systems

            occurrence of any unphysical damage/crack in the particle model, whereas a frozen
            period can be comprised of thousands of time steps. In the loading period, a veloc-
            ity increment is applied to the loading boundary of the system, while in the frozen
            period the loading boundary is fixed to allow the system to reach a quasi-static equi-
            librium state after a large number of time steps. It needs to be pointed out that a pair
            of load and displacement (or stress and strain) is correctly recorded at the end of a
            frozen period.
              Based on the above-mentioned conceptual understanding, the theoretical expres-
            sion of the proposed loading algorithm can be deduced as follows. An applied force
            in the quasi-static system can be divided into M loading increments.

                                            M

                                       P =    ΔP i ,                     (8.31)
                                           i=1

            where P is the applied force; ΔP i is the ith loading force increment and M is the
            total number of loading steps.
              For each loading force increment, ΔP i , it is possible to find a solution, ΔS i ,
            which satisfies the following condition:


                                   lim (ΔS ij − ΔS ij−1 ) = 0,           (8.32)
                                  j→N i

            where ΔS ij is the solution at the jth time step due to the ith loading force incre-
            ment; ΔS ij−1 is the solution at the (j-1)th time step due to the ith loading force
            increment; N i is the total number of time steps for the particle system to reach a
            quasi-static equilibrium state after the application of the ith loading force increment
            to the system.
              From the numerical analysis point of view, Equation (8.32) can be straightfor-
            wardly replaced by the following equation:

                             &            &
                             & ΔS ij − ΔS ij−1  &
                         max  &           &  <δ        (at j = N i ),    (8.33)
                             &            &
                                  ΔS ij
            where δ is the tolerance of the solution accuracy.
              It needs to be pointed out that the value of N i can be determined using the con-
            vergent condition expressed by Eq. (8.33), so that different values of N i may be
            obtained in the numerical simulation. It is the convergent condition that approxi-
            mately warrants the solution of the quasi-static nature, if the value of δ is not strictly
            equal to zero.
              Thus, the total solution S corresponding to the applied force P can be expressed as

                                            M

                                        S =    ΔS i .                    (8.34)
                                            i=1
   196   197   198   199   200   201   202   203   204   205   206