FAULT GOUGE POROSITY FROM 2D SIMULATIONS OF FRACTAL PARTICLE SIZE DISTRIBUTION

Two dimensional computer simulations were carried out to determine porosity in various regions of simulated Westerly granite gouge deformed at room temperature and 25 MPa normal stress to 44, 79, and 387mm of shear displacement in rotary shear. The input data was approximated by a power simulation of the actual gouge particle size distribution(PSD). The algorithm takes as input, values due to a distribution of circle equivalents of N particles of size S in a descending order as tangent circle solutions pack the circles such that least number of same size particles touch. The 2D porosity was calculated by summing up the void space areas in the simulated texture. The algorithm was without limitation on size and number of particles and produced maximum packing density under the described conditions. For comparison, porosities of the same gouge samples were also estimated by image processing method on telescopic SEM images. Gouge particle outline overlays were made manually on images of the deformed and undeformed gouge samples. The overlays were used for thresholding, PSD, fractal dimension, and particle shape (the number of straight sides per particle)measurements.

Close to zones of shear localization, the porosity values determined by both methods showed 2 orders of magnitude decrease from undeformed porosity of about 35%. However, the difference between porosity values determined by the two methods decreased greatly with increasing amount of shear displacement. The shape factor decearsed while fractal dimension increased with decreasing porosity.

The results suggest that particles tend to become more equant in shape in regions of extreme comminution along Y and R-shears in the gouge. The sharp decrease in local porosity results in packing density increases. Based on brittle-ductile transition data for porous rock material this porosity reduction could cause local embrittlment and shear fracture of the gouge. The microstructural evidence is consistent with the above conclusion.