Complicated geological structures make it difficult to analyze the stability of rock slopes, such as faults, weak intercalated layers or joint fissures. Based on 3D geological modeling and surface block identifying methods, an integrated methodology framework was proposed and realized to analyze the stability of surface blocks in rock slopes. The surface blocks cut by geological structures, fissures or free faces could be identified subjected to the four principles of closure, completeness, uniqueness and validity. The factor of safety(FOS)of single key block was calculated by the limit equilibrium method. If there were two or more connected blocks, they were defined as a block-group. The FOS of a block-group was computed by the Sarma method. The proposed approach was applied to an actual rock slope of a hydropower project, and some possible instable blocks were demonstrated and analyzed visually. The obtained results on the key blocks or block-groups provide essential information for determining potential instable region of rock slopes and designing effective support scheme in advance.
Due to the complex nature of multi-source geological data, it is difficult to rebuild every geological structure through a single 3D modeling method. The multi-source data interpretation method put forward in this analysis is based on a database-driven pattern and focuses on the discrete and irregular features of geological data. The geological data from a variety of sources covering a range of accuracy, resolution, quantity and quality are classified and integrated according to their reliability and consistency for 3D modeling. The new interpolation-approximation fitting construction algorithm of geological surfaces with the non-uniform rational B-spline(NURBS) technique is then presented. The NURBS technique can retain the balance among the requirements for accuracy, surface continuity and data storage of geological structures. Finally, four alternative 3D modeling approaches are demonstrated with reference to some examples, which are selected according to the data quantity and accuracy specification. The proposed approaches offer flexible modeling patterns for different practical engineering demands.