UBC Theses and Dissertations
Computer modeling of rock fall : improvement and calibration of the PIERRE 2 stochastic rock fall simulation program Mitchell, Andrew David
Rock falls are a common hazard affecting people and infrastructure below natural cliffs and in man-made environments such as roads, railways, quarries and open pit mines. Computer models have become standard in assessing the hazards posed by rock falls, and there are a wide variety of models currently available. The main difference between these models is the way in which they represent the impact process. New models utilizing rigorous solutions for rigid body impacts have become more common, however, the complexity of these models makes them difficult to calibrate and difficult to apply in practice. The PIERRE 2 model, presented here, returns to a simplified lumped-mass model assuming collinear impact conditions. The natural variability of rock falls is represented using the following features: • The slope angle is varied by a stochastic roughness angle to approximate the effects of non-collinear impacts. • The conservation of momentum during an impact is determined using hyperbolic relationships so that high energy impacts will have greater losses than low energy impacts. • A stochastic shape factor is used to vary the sphere dimensions at impact to approximate the rotational effects of real, non-spherical boulders. Impact mechanics theory was used as the basis for these features; however their validity is demonstrated through 2D and 3D model calibration. The model was calibrated using data from experiments with detailed kinematic information on two natural slopes, Vaujany, France and Ehime, Japan, and excavated slopes at three Austrian quarries. The results from the calibrations were then applied to a natural slope, Tornado Mountain, and an excavated slope at Nicolum quarry, both in British Columbia. The results of this calibration and validation program showed the model can reproduce runout and kinematic behaviours of rocks based on relatively limited site characterization data. Select results were also compared to design guidelines. Using reliability engineering principles, the probability of the design parameters derived from the model being exceeded was found to be acceptably low. The model is also capable of reproducing observed relationships between impact incidence angle and the normal restitution coefficient reported by others.
Item Citations and Data
Attribution-NonCommercial-NoDerivs 2.5 Canada