A B S T R A C THydraulic fracturing, a powerful completion technique used to enhance oil or gas production from impermeable strata, may trigger unintended earthquake activity. The primary basis for assessment of triggered and natural seismic hazard is the classic Gutenberg-Richter (G-R) relation, which expresses scale-independent behaviour of earthquake magnitudes. Here we use a stochastic approach to simulate and test magnitude-distance trends expressed by microseismic catalogues derived from three hydraulic fracture monitoring programmes in North America. We show that a widely observed rapid fall-off in large-magnitude events, almost universally quantified using the G-R b value, may in our case be an artefact of the strongly laminated character of the stimulated oil and gas reservoirs. We also show that, for the three reservoirs considered, mechanical bed thickness can be approximated by a lognormal distribution. For a stratabound fracture network, this leads asymptotically to a Gaussian decay for induced magnitudes. We show that the stratabound model provides a more significant correspondence with our observations. If applicable in general, this result has important implications for determining the energy balance of hydraulic fracture systems (i.e. radiated seismic energy versus injected energy) as well as hazard assessments based on the probability of occurrence of anomalous seismic events.

TX 75083-3836, U.S.A., fax 01-972-952-9435. AbstractMicroseismic mapping is extensively utilized in the Barnett Shale, to map hydraulic fracture complexity associated with interactions of the stimulation with pre-existing fractures. Previous studies have indicated a fair correlation between the well performance and extent of the seismically active volume. However, in addition to this measure of the extent of the stimulated fracture network, the characteristics of this fracture network is also expected to impact the well performance. In particular, the fracture spacing is believed to be important factor controlling the potential gas flow. In this paper, we utilize the density of the total seismic moment release (a robust measure of the microseism strength) as an indication of the seismic deformation that may correlate to the fracture density. The study uses a set of microseismic maps of hydraulic fracture stimulations, including cases where the stimulated reservoir volume measured by the extent of the seismically active region poorly correlated with the well performance. Incorporating the seismic moment density to assess the fracture density with the network extent, an improved correlation with the well performance was observed.

In this paper we compare the input and output energy during a hydraulic fracture treatment to evaluate microseismic efficiencythe fraction of total input energy that is radiated as high-frequency seismic waves. We use three approaches to compute energy. "Injection energy" is the total energy input into the system, which can be accurately quantified using the product of treatment pressure, injection rate and duration. "Fracture energy" is an estimate of the work done during the deformation process. This parameter has considerably greater uncertainty and is computed using the product of the in situ treatment pressure, total effective area of the fracture surface (estimated using the microseismic event distribution) and the effective opening width of the fractures (based on an educated guess). Finally, "radiated seismic energy" is computed by summing the energy for all recorded high-frequency microseisms, after converting the reported moment magnitudes into energy based on scaling relations developed in earthquake seismology. This estimate is highly uncertain for various reasons, including epistemic uncertainty, which reflects lack of knowledge, and aleatory uncertainty, which reflects uncertainty that is inherent to the observations. Epistemic uncertainty stems primarily from the extent of extrapolation required for use of moment-energy relations, which are calibrated for large earthquakes and here are applied to events that are more than 10 orders of magnitude smaller. Aleatory uncertainty is related to noise, missing data and amplitude uncertainty due to radiation patterns. We attempted to correct the seismic energy for missing data by estimating the Gutenberg-Richter b value to estimate the contribution for all events greater than magnitude-3 up to the maximum recorded magnitude. To our knowledge, such a correction has not been previously attempted, but it is very significant for determining the seismic energy. Injection, fracture and radiated seismic energy were calculated for 10 stimulated stages. Fracture energy was found to be 12-41% of total injected energy, whereas the ratio of radiated energy to fracture energy ratio was consistently less than 1% for all stages. A current viewpoint is that the fracture energy typically represents 20-80% of the total injection energy, and that the shear mechanisms are only a small part of the total fracturing process (N. Warpinski, pers. comm., 2011). This viewpoint is adopted here as a working hypothesis. Our results confirm previous studies and show that the radiated high-frequency seismic energy constitutes a small fraction of the total energy in the system. Theory and/or Method Data from four different wells in an unconventional shale gas play that were hydraulically fractured are considered in this study. Microseismic maps for 10 stages were produced and used to estimate the fracture area and total high-frequency seismic energy. The following sections describe the energy calculations and the parameters extracted from each stage in order to determine these values.

Microseismic mapping is extensively used in the Barnett Shale to map hydraulic fracture complexity associated with interactions of the stimulation with pre-existing fractures (fracs). Previous studies have indicated a fair correlation between the well performance and extent of the seismically active volume. However, in addition to this measure of the extent of the stimulated fracture network, the characteristics of this fracture network is also expected to impact the well performance. In particular, the fracture spacing is believed to be an important factor controlling the potential gas flow. In this paper, we use the density of the total seismic moment release (a robust measure of the microseism strength) as an indication of the seismic deformation that may correlate to the fracture density. The study uses a set of microseismic maps of hydraulic fracture stimulations, including cases in which the stimulated reservoir volume measured by the extent of the seismically active region poorly correlated with the well performance. Incorporating the seismic moment density to assess the fracture density with the network extent, an improved correlation with the well performance was observed.

Along with horizontal drilling and hydraulic fracturing technology, monitoring-induced microseismic activity during hydraulic fracturing has played a significant role in the economic development of today's prolific tight-gas and shale resource plays. Many microseismic fracture monitoring papers in the current literature discuss the equipment, how to acquire and process microseismic data, and how to determine an appropriate velocity model. Unfortunately, for an operating company's completion engineers and geophysicists charged with interpreting these data and integrating them with other data sets, only a few papers discuss the impact of artifacts and location uncertainties on the interpretation. While microseismic data have been very useful in the economic development of resource plays, there are a number of significant pitfalls and interpretational issues that, if not understood, will potentially lead to erroneous interpretations. This is especially important for any real-time decisions during a hydraulic fracture treatment.