A new method of replacement optimization of soft rock mass is proposed and applied to a large cavern group. The method combines FEM with Genetic Algorithms (GAs). At the beginning, a set of' initial tentative replacement schemes is randomly generated. The volume of damage zone is calculated for each tentative replacement scheme to evaluate applicability of the scheme. A group of new schemes are then generated by operation of GAs. The process continues until the best solution is found. Compared with the other existing methods, this new method has a faster speed, and is easier to be carried out on personal computer, and can obtain optimum solution in global space. This new method is applied to the analysis of the stability of a cavern group in hydroelectric project. An optimum scheme of replacement with the reasonable height, width and depth of each sort rock layer around the main chamber is determined through evolution of 10 generations with 10 individuals. Some reasonable suggestions are also given for safe construction.
Currently, the scale of the underground structure becomes larger and larger. The layout and the geological condition also become more and more complex. For example, the cavern group of the underground chamber of the building projects such as Longtan, Xiaowan and Shuibuya, have an all-time scale and a great technological difficulty in construction (Xiao 2000). On dealing with the excavation of the cavern group and the analysis of the stability, the traditional method is to simulate the progress of the construction of excavation, and calculate several possible schemes by 2-D or 3-D elasto-plastic finite element method (FEM). After the analysis of deformation, the distribution of plastic zone and the state of stress, the scheme with a preferable stability of the surrounding rock after excavation was selected (Zhang 1999, Xu 1999). But to the optimization of excavation process for the large-scale underground cavern group, especially to the optimization and the corresponding analysis of the stability in soft rock, it obviously takes so much time and is not so easy to carry out on personal computer. Taking the method of dynamic programming to do the calculation and optimization of the large-scale complex project brings a first transition of method. With the method the excavation scheme is determined by optimization stage by stage(Zhu1995). Because of optimization in stage, this method can't obtain a global optimum solution. Therefore, a new method of replacement optimization of soft rock mass is proposed in this paper. The method is applied to the analysis on the stability of a large-scale cavern group in soft rock mass.
Genetic Algorithm (GA) is a global optimization method, and is especially fit for optimization of multi-extremum. It imitates the rule of evolution in nature(i.e. selecting advantage and survival of the fittest) and the concepts of reproduction, crossover, and mutation are introduced into the algorithm. Through constructing a set of initial tentative solution population.
