The strength of layered siliciclastic fault rocks as a function of composition and structure

Normal faults are common in sedimentary basins and often associated with reservoirs in interbedded sands and clays. Fault rocks therefore also consist of some mixture of sand and clay. Outcrop studies have shown, that these fault rocks can occur as homogeneous mixtures, (multiple) parallel layers of sand and clay without intense grain-scale mixing, or complex structures with brittle clasts of one material embedded in a ductile sheared matrix of the other. Both, the composition and the structure of the fault rock affect its the overall frictional strength at any given position.

The strength of faults in sedimentary basins is crucial information when producing fluids from faulted reservoirs in critically stressed conditions. Increasing pore pressure during injection phases bears the risk of fault reactivation. To minimize the risk of reactivation while maximizing the recovery, our goal is to improve the prediction of fault friction. The predicted friction coefficient can then be used in dynamic reservoir models to calculate the maximum allowed pore pressure increase. 

From literature we compile the friction coefficients for various homogeneous sand-clay mixtures at different effective normal stresses, measured in laboratory tests. The resulting function shows a linear increase of the friction coefficient with increasing sand content, while normal stress only shows an effect for stresses larger than expected at reservoir conditions. We can now use this function to predict the friction coefficient for any given homogeneous sand-clay mixture.

However, fault rocks are often not homogeneous mixtures. To gain insights into natural fault rock compositions, we investigate field and sample data in 2D and 3D from outcrops in northwest Borneo/Malaysia. These show the complex structure of fault rocks on various scales for faults with displacements from cm to decameter range.

In exploration and production workflows, commonly algorithms such as the shale gouge ratio are applied to predict the average volume of clay (Vclay) in the fault rock, based on the amount of clay in the unfaulted rock and the displacement. The average Vclay is then loosely correlated to a friction coefficient, often proprietary to the used software packages. We propose that the structure of the fault rock, i.e. the distribution of clay and sand, affects the frictional properties estimated for the average Vclay.

We use discrete element numerical simulations to study the effect of complex fault rock structures on the fault friction coefficient. We reproduce natural structures from outcrop and sample data and calibrate the mechanical properties of the individual components in the model to fit the natural prototype. In direct-shear tests we then measure the friction coefficient of the entire modelled fault rock. Preliminary results show a discrepancy between the friction coefficient of a homogeneous sand-clay mixture and a more complex geometry with the same clay volume. This suggests errors in currently used approaches that are solely based on Vclay.