STOCHASTIC ESTIMATION OF FRACTURE SIZE THROUGH SIMULATED SAMPLING

Engineering fractured rock masses requires accurate geometrical descri ption of the fractures. A critical parameter is fracture size which st rongly affects the degree of interconnection between fractures, and co nsequently permeability, mass transport and incipient instability. Exi sting methods for estimating size from borehole data or from planar ex posures that are smaller than many of the fractures are too unconstrai ned. A new approach using stochastic forward modeling makes it possibl e to estimate the size/frequency distribution of fracture sizes for se verely truncated data. Its predictive accuracy is demonstrated through simulation. This approach is illustrated using severely truncated fie ld data gathered from an underground cavern. The size distributions fo r each of three fracture sets, as well as the major conductive fractur es are estimated.