A Non-parametric Discrete Fracture Network Model

Abstract A discrete fracture network (DFN) model based on non-parametric kernel density estimators (KDE) and directional-linear statistics is developed. The provides a characterization of the with distributions orientation size jointly. solution to Bertrand paradox used for calculation disk sizes from trace lengths, latter calculated intersection disks highwall faces by triangulation. Poisson point process applied generation model, fractures assumed be flat circular in shape, number per unit volume (P 30 ) adjusted match experimental length area 21 ). Length censoring traces due surface dimension considered calculations including semi-bounded traces, i.e., censored one their ends. Orientation biases are corrected weighting function random sampling. truncation effect whereby no shorter than some cut-off recorded, addressed randomized optimization algorithm. joint orientation-size distribution developed tested maps discontinuities measured photogrammetric models twelve quarry benches, outstanding results. Computational advantages over traditional parametric addressed.