Energy issues
In a recently completed research project, the University of Texas at Austin conducted two-year surveys of earthquake activity in the Fort Worth basin of Texas, the Eagle Ford shale of Texas, the Bakken shale/Williston basin of North Dakota and Montana, and the Haynesville shale of Texas and Louisiana. The studies were conducted under the Research Project to Secure Energy for America (RPSEA), with funding from the U.S. Department of Energy’s National Energy Technology Laboratory.
In the Fort Worth basin, researchers identified 67 earthquakes occurring during the 2009-2011 survey period, of which only eight were reported by the U.S. Geological Survey. Of these 67 earthquakes, all the reliably located events occurred in eight clusters, and each of these clusters was situated within 3.2 km of one or more high-volume injection disposal wells, i.e., wells having maximum monthly injection rates exceeding 150,000 bbl/month. This suggests that higher-volume injection wells in the Fort Worth basin did trigger earthquakes; however, there were numerous such high-volume wells that did not have earthquakes nearby. Thus, it is unclear why earthquakes occurred near some high-volume wells and not others.
In the Eagle Ford, researchers identified 62 earthquakes occurring during the 2009-2011 survey period, of which only four were reported by the U.S. Geological Survey. Of these 62 earthquakes, the majority (47 events) occurred following increases of fluid extraction (not injection) at nearby wells. The most reasonable conclusion is that most earthquakes identified in the Eagle Ford are triggered by petroleum operations—mostly by production/extraction of petroleum and water. But the triggering mechanism is clearly different than in the Fort Worth basin, where earthquakes are associated with injection wells.
In the Bakken shale/Williston basin, the project identified only nine earthquakes occurring during the 2008–2011 study period. Of these, only three were near injection wells. Thus, injection wells possibly triggered earthquakes, but were far rarer in the Bakken shale/Williston basin region than in the Fort Worth basin or the Eagle Ford regions of Texas.
In the Haynesville shale of Texas and Louisiana, researchers identified 39 earthquakes during the February 2010 through 2012 study period. They identified a cluster of activity near Bienville Parish, La., that had not been reported previously. A total of seven earthquakes between magnitudes 1.2 and 2.7 occurred over just a few months (August-October 2011). The smaller-magnitude earthquakes, without a preceding larger event, for the Bienville Parish activity are most consistent with classification as a seismic swarm. Data are available concerning monthly injection volumes, injection intervals, etc., for wells in Texas, but not for Louisiana.
The Texas data indicate that, within a few km of two high-volume injection wells in Texas, a series of earthquakes occurred that included the magnitude 4.8 earthquakes south of Timpson, Texas, on May 17, 2012. With collaborators from Stephen F. Austin University and the U.S. Geological Survey, researchers operated temporary seismograph stations and located numerous aftershocks following the 2012 earthquake. The locations and focal depths of these earthquakes suggest that they occurred at, and deeper than, the depths of injection along a mapped basement fault. The evidence strongly suggests that the sequence was triggered by injection.
A surprise finding. When researchers conceived of the project, they anticipated that the relationship between seismicity and injection would be similar in the four geographic regions studied and, thus, it would be reasonable to use statistical tools to assess salient features of the relationship. However, to their surprise, the relationship was hugely different in each of the four geographic regions: injection disposal triggered nearby earthquakes in the Fort Worth basin; fluid extraction triggered earthquakes in the Eagle Ford; earthquakes were virtually nonexistent in the Bakken; and two earthquake sequences occurred within the Haynesville, including a magnitude 4.8 triggered event. Otherwise, there was little apparent triggered activity.
Thus, the project did not pursue a statistical assessment of the relationship. However, the observation that the injection/seismicity relationship is significantly different in different geographic regions is important, and has implications for managing injection waste disposal operations. For example, it implies that before crafting regulations or implementing hazard-reduction actions in any particular locality, surveys should be undertaken to assess the relationship between injection and seismicity within that area.
To that end, later sections of this report concern developing geomechanical models to better understand the subsurface conditions, where injection/extraction leads to seismic activity. The most detailed, focused modeling effort was applied to the subsurface surrounding the epicenter of the magnitude 4.8, Timpson, Texas, earthquake of May 17, 2012. The modeling indicated that the injection in two wells near the epicenter changed Coulomb stresses sufficiently to cause failure along a mapped (but previously unnoticed) fault.
A practical implication of this work is that for proposed injection disposal operations, it indicates that geomechanical modeling to evaluate Coulomb stress levels is a reasonable tool for assessing the geographic extent and location of possible earthquakes. This is true, especially if there is information about rock properties and subsurface faulting near a proposed well.
Additional information can be found at http://www.rpsea.org/media/files/project/5f6a4f5e/11122-27-FR-Relationships_Seismicity_Fluid_Injection_Strategies_Disposal_Shale_Hydrocarbon_Plays-04-03-15_P.pdf. 
