The thorough and accurate measurement of rock mass deformation or convergence is often limited to the use of a small number of discrete point to point measurements. These measurements are often not representative of how a rock mass is responding to the excavation of material over a larger domain. Advances in LiDAR (Light Detection and Ranging) technology and data processing techniques have developed portable, and accurate devices for three-dimensional mapping of excavations in GPS (Global Positioning System) deprived environments. This technology has the ability to generate large, representative spatially continuous data sets showing excavation convergence. These data sets can be especially useful for the calibration and forecasting of excavation convergence in non-linear, three-dimensional, strain softening, discontinuum finite-element models (DFEM). This paper discusses the findings from an ongoing study into the use of this new measurement technique and the resulting data sets for the calibration of numerical models and the importance of incorporating sufficient structural resolution in the models to match the observed convergence with sufficient accuracy.
Numerical modelling techniques in rock mechanics have advanced significantly in recent times. There are no longer limitations requiring a rock mass to be assumed as 2-dimensional, linear-elastic and homogeneous. Non-linear, 3-dimensional, strain-softening, discontinuum Finite-Element simulations, with multiphysics coupling can now accurately represent a rock mass at many different length scales.
With these sophisticated numerical simulations, it is common now to be limited in the model construction phase by a lack of representative scale structural data, and representative observational data during the calibration phase. The inability to to incorporate realistic structures on the correct length scale due to a lack of data can be overcome through the use of discrete fracture networks where no actual data can be acquired. However, these kind of models must be calibrated in order to allow a realistic representation of the rock mass response.
Commonplace in-situ measured and observational datasets include stress measurements, seismicity records, damage mapping, extensometers or point closure measurements. Stress measurements and seismic records provide a valuable insight into the rock mass state and reaction, however the representative strain response of the rock mass has been very difficult to measure. Laboratory testing can provide some insight, however post-peak behavior is rarely considered, and size limitations restrict the upscaling to a rock mass representation.