| The whole lifetime of a high-level radioactive waste repository spans usually over thousands of years,therefore the surrounding host rock will become the main barrier against the catastrophic radionuclide leakage and migration once the engineering barrier lost its normal function at any possible moment after backfilling of the repository.However,the existence of the small scale fractures in the near field(near the periphery of the repository)and the large scale faults in the far field(encircled by the candidate site border)comprises the main channels within which radionuclides could migrate through the engineered barrier and spread into the surrounding rock mass.In this situation the possibility of these harmful materials migrating into the biosphere via the fractures and faults greatly increases.It is well recognized that the near-field migration of radionuclides is controlled by the overall seepage characteristics of the stochastic discrete fracture system,while the far-field migration behavior is largely related to the flow properties of the faults.Prediction and evaluation of radionuclide migration are thus considerably dependent on the characterization of flow features of fracture system,as well as the acquisition of the hydraulic properties of faults.It is common practice that investigation upon the seepage property of a fracture system is primarily based on the regional fracture network.The small-scale fractures in the near field are generally distributed in a random manner,which may be properly described by stochastic theory.However,a huge gap between real situation and the simplified model could arise if the systematically distributed fractures or faults with apparent structural patterns are totally replaced by random distributed fractures,which may result in a decrease of model accuracy.To overcome this shortcoming,an enhanced model which could simultaneously account for the stochastic fractures as well as the deterministic structural pattern is proposed in this dissertation.By performing model correction based on the identified patterns of fractures/faults,this model is believed to be a more reasonable simplification of the reality.This thesis is devoted to the investigation and characterization of the small-scale fracture system and regional faults in Jiujing area,which is located in Gansu Beishan,one of the preselected areas for Chinese high-level radioactive waste repository.Completed work consists of detailed mapping of fractures in the study area,identification of structural pattern,acquisition of the geometrical and seepage parameters of fractures in large amount by digitalization,simulation of stochastic fracture network,model correction considering the hierarchical relationship,structural pattern and hydraulic connection among fractures of different scales,laboratory flow tests on the undisturbed fault zone samples,etc.These work aims at a better understanding of the seepage property with respect to both the stochastic fracture system and faults.The main results of this study are as follows:1 Data pre-processing is crucial for the improvement of modelling accuracy.Due to the sampling bias induced by the discrepancy between real frequency of fractures in rock mass and the apparent frequency exposed on outcrops whose orientation is sub-parallel to the mean orientation of fracture set,a frequency bias correction procedure is needed.Here the traditional Terzaghi bias correction method is adopted.After correction the sampling frequency is closer to the real one.In addition,fractures with dip angle less than 20 degrees are prone to be sampled with marked bias and should be considered first while conducting pre-processing.2 In order to investigate the merits and restrictions among various methods of homogeneity delineation,different algorithms were realized in program.Modifications with respect to the inconvenience in each algorithm were then carried out.The improved algorithms were applied to analyze the study area in Beishan.Results show:an ambiguous geologic background is associated with correlation coefficient method,which may lead to a less reliable delineation result.On the contrary,method proposed by Mahtab has a sound geological meaning,which,nevertheless,suffers from the restrictions imposed by too many factors and could result in a pronounced contingency.Method proposed by Miller has received wide-spread application due to its well-developed theory.However,some inconvenience is introduced by its complex algorithm,which may be improved further by algorithm modification.A preliminary analysis on the rock mass near borehole BS03 has shown that outcrop D9 and D11 has the most pronounced homogenous feature.It is thus suggested that these two outcrops should be combined into a single homogeneity.Miller's method(34 grids large area block method)is then applied to analyze Jiujing area.Results indicate that there are two homogeneities inside the study area.Subzone A and B belong to homogeneity ?,whereas C and D belong to homogeneity ?.Furthermore,two and three fracture sets are identified with respect to homogeneity ?and ?,respectively.3 Discrete fracture network(DFN)model for homogeneity ? is generated based on the mapped fracture data obtained by detailed investigation in Jujing area,Beishan.Model validation is performed using the sampling window test and statistical index test.Maximum value of difference ratio,variation coefficient,spherical deviation are 16.7%,26.82%,0.21,respectively,which indicates that the proposed DFN model meets the requirements imposed by statistics and is relatively reliable.Therefore,stability and seepage analysis based on the proposed model could be used to treat problems appeared in practical applications.4 DFN model modification based on the identified structural pattern has been conducted.After inserting those deterministic fractures,seepage simulation is then carried out.Compared with the DFN model,simulation result of this newly-proposed coupled model is more consistent with the in-situ packer test.An improvement in consistency is 1/3 more than the DFN model.Besides,the seepage results with respect to different test sections fluctuate between 10-6?10-4 m3/s,while in DFN model the flow rate stabilizes at 10-6m3/s,which means little fluctuation could be anticipated without considering the effects of deterministic fractures.It is therefore concluded that the DFN model only captures the random feature of fracture system,while the coupled model could account for the intrinsic structure and hydraulic connection of a specific fracture system.The deterministic fractures thus play a very important role in controlling regional seepage characteristics.In other words,these fractures should not be neglected when building up the regional DFN seepage model.5 Seepage simulation for the fracture system in Jiujing is carried out using a secondary developed program based on the commercial code.Results indicate that the equivalent hydraulic conductivity on three mutually perpendicular plane show little fluctuation.Values around 10-7m/s are obtained.The volumes of the representative element for the three directions are also similar in terms of size,namely 8?10m.6 Laboratory seepage tests on fault damage zone samples taken from Shiyuejing area were performed.In order to quantitatively describe the structure of the inner fractures which are contributed to the flow process,samples were first stained,and then sliced into pieces after tests.Statistical parameters of fractures were extracted from these sliced pieces.Results show that fracture set ?,?,and ? have a hydraulic conductivity of about 7.48×10-5cm/s,3.17×10-5cm/s,and 4.02×10-6cm/s,respectively.These data were then assigned to the seepage model.The size of REV in Shiyuejing fault zone is approximately 6?8m,and the magnitude of the equivalent hydraulic conductivity is about 10-5m/s.A great difference was obtained between the equivalent hydraulic conductivity along and perpendicular to the strike of the fault.On average the equivalent hydraulic conductivity along the strike is 30%larger than that normal to the fault,whereas the maximum ratio between the former and the latter may approach 3.It is concluded that fluid flow process mainly occurs along the strike of the fault zone. |
PDF Full Text Request