We utilize several lines of evidence to argue that slow slip on pre-existing fractures and faults is an important deformation mechanism contributing to the effectiveness of slick-water hydraulic fracturing for stimulating production in extremely low permeability shale gas reservoirs. First, we carried out rate and state friction experiments in the laboratory using shale samples from three different formations with a large range of clay content. These experiements indicated that slip on faults in shales comprised of less than about 30% clay is expected to propagate unstably, thus generating conventional microseismic events. In contrast, in formations containing more than about 30% clay are expected to slip slowly. Second, we illustrate through modeling that slip induced by high fluid pressure on faults that are poorly oriented for slip in the current stress field is expected to be slow, principally because slip cannot occur faster than fluid pressure propagates along the fault plane. Because slow fault slip does not generate high frequency seismic waves, conventional microseismic monitoring does not routinely detect what appears to be a critical process during stimulation. Thus, microseismic events are expected to give only a generalized picture of where pressurization is occurring in a shale gas reservoir during stimulation which helps explain why microseismicity does not appear to correlate with relative productivity. We review observations of long-period-longduration seismic events that appear to be generated by slow slip on mis-oriented fault planes during stimulation of the Barnett shale. Prediction of how pre-existing faults and fractures shear in response to hydraulic stimulation can help optimize field operations and improve recovery.
We investigate a series of long period and long duration (LPLD) seismic events observed during hydraulic fracturing in a gas shale reservoir. These unusual events, 10–100 seconds in duration, are observed most clearly in the frequency band of 10–80 Hz and are remarkably similar in appearance to tectonic tremor sequences first observed in subduction zones. These complex but coherent wave trains have finite move-outs obtained from cross-correlation. The move-out direction of the events confirms that they originate in the reservoir from the area where the fracturing is going on. Clear P- and S-wave arrivals cannot be resolved within the LPLD episodes, but in some cases, small micro-earthquakes occur in the sequences. Whether these micro-earthquakes are causal or coincidental is not known. It has also been observed that in three contiguous frac-stages, all LPLD events appear to come from two distinct places along one of two fracture planes. Interestingly, the stages which have the largest number of LPLD events also the have the highest observed pumping pressures during fracturing and the highest density of natural fractures. One possible explanation of these LPLD events is that low effective normal stress (resulting from high pore fluid pressure during hydraulic fracturing), causes slow slip on the pre-existing fault planes. In the absence of elevated pressure, slip would not be expected on these planes as they are poorly-oriented to the stress field. Slip on these fault planes is occurring because the fluid pressure is close to the magnitude of the least principal stress. We observe a few events after pumping is stopped indicating that once triggered, these planes continue to slip, perhaps due to trapped pore pressure within the fault planes.