Induced seismicity in Eastern Ohio has been ongoing for the past several years, much of which is attributed directly to hydraulic fracturing (HF) wells. To help parse which factors influence the generation of earthquakes in the region we gathered public well data for the HF wells in Harrison, Monroe, and Noble county. This dataset allowed us to investigate the relationships between the number of wells on a pad, injected volume, depth of the well, and when different well laterals were in operation on the same pad. To establish a 95% confidence interval in our results we used odds ratios to compare seismic and aseismic wells, and logistic regression to plot the results of comparing chance of seismicity vs. depth/injection volume/number of laterals per pad. There is a negative trend with injected volume, however the higher the volume injected the more uncertain the probability of seismicity. This is similar to the chance of seismicity when compared with the number of laterals on a well pad, however there is significantly more uncertainty than the relationship between volume injected and seismicity. In comparison the depth of the HF wells had a much stronger positive trend where the deeper the well, the higher the likelihood. We categorized the type of well operations into zipper, sequential, and isolated wells. After parsing which well laterals on which well pads generated which behavior we analyzed the resulting binary data with odds ratios to determine which type of operation had a higher likelihood of causing an induced earthquake. Well depth had the strongest relationship of these factors, in addition odds ratios comparisons indicate that zipper or sequential methods have a much higher likelihood of inducing seismicity. Our results are consistent with previous work on the topic and region, suggesting that operational procedures and number of laterals per pad, along with depth are major controlling factors in inducing seismicity in HF wells.
Author: Seth Young
Faculty Advisor: Dr. Michael Brudzinski, Geology Department
You must be logged in to post a comment.