Abstract
Hydraulic fracturing is carried out in most shale gas fields to enhance reservoir permeability. Thousands of microseismic events are observed with geophone arrays during fracking stages. Microseismic hypocenter distributions are essential information to delineate stimulated reservoir volume (SRV), and there are some useful automatic processing tools to get hypocenter locations. However, recorded seismograms contain noise and complex phases that cause ambiguity in the data processing. In addition, the velocity model selected has a great influence on the accuracy of the hypocenter information.
In this research, to enhance the accuracy of microseismic event hypocenter locations, we picked P and S phases using an array seismogram volume (ASV). The ASV consists of shot gathers and receiver gathers of selected events and receivers. The events should be limited to those that have enough waveform similarity. Once we pick an onset time of a single trace in the ASV, the picked seed trace is compared with neighboring traces using cross correlation; most of the onset times in the ASV are picked automatically. Because the process requires picking of single traces, we call this procedure semi-auto picking. This semi-auto picking tool can reduce the time needed for manual picking and the process is efficient and provides high-quality results.
Shale sediment are frequently characterized by anisotropic parameters. The sediments in Barnett, our study area, also show anisotropy. Therefore, an accurate velocity model is required to achieve precise microseismic event analysis. A tilted layer orthorhombic velocity model was adopted in this study. P-wave sonic and gamma ray logs were available for a reference well located close to the study area. The 3D seismic survey interpretation provided a tilted layer angle. The principal horizontal stress direction was provided by a previous study in Barnett. Using this geoscience information as constraints, a number of optimized parameters could be reduced. Seven Thomsen parameters and the Vp/Vs ratio of the defined layers were optimized using perforation shots.
A grid search location approach was applied to locate 932 events during a single fracking stage. To obtain final locations, the following objective functions were combined: (1) P, Sh, and Sv travel time misfits; (2) travel time differences between receivers; (3) travel time differences between different wave phases; (4) travel time differences between microseismic events and master events (perforation shots); (5) P phase polarization; and (6) P phase polarization differences between microseismic and master events. Spatio-temporal behavior of the located event cloud was investigated using R-T plot analysis. The SRV was observed to grow with the injection volume in this stage and the effective fracture thickness was almost 10 mm after treatment.
