Study of induced seismicity for reservoir characterization

Abstract

The main goal of the thesis is to characterize the attributes of conventional and unconventional reservoirs through passive seismicity. The dissertation is comprised of the development and applications of three new methods, each of which focuses on a different aspect of fractures/faults and the resulting seismicity. In general, the thesis work discusses reservoir characterization from two aspects: 1) understanding fractures and faults in reservoirs as seismic sources with induced seismicity, and then inferring other properties of the reservoirs, such as stress regime and velocity structure (Chapters 2, 3, 4); 2) understanding the fractures in reservoirs as seismic scatterers (Chapter 5). First, I introduce a new method to determine the source mechanisms of the induced earthquakes by incorporating high frequency waveform matching, first P-arrival polarities and average S/P amplitude ratios. The method is applied to 40 induced earthquakes from an oil/gas field in Oman monitored by a sparse near-surface seismic network and a deep borehole seismic network. The majority of the events have a strike direction parallel with the major NE-SW faults in the region, and some events trend parallel with the NW-SE conjugate faults. The results are consistent with the in-situ well breakout measurements and the current knowledge of the stress direction of this region. The source mechanisms of the studied events together with the hypocenter distribution indicate that the microearthquakes are caused by the reactivation of preexisting faults. Then I introduce a new method to locate microseismic events induced by hydraulic fracturing with simultaneous anisotropic velocity inversion using differential arrival times and differential back azimuths. We derive analytical sensitivities for the elastic moduli (Cij) and layer thickness L for the anisotropic velocity inversion. The method is then applied to a microseismic dataset monitoring a Middle Bakken completion in the Beaver Lodge area of North Dakota. Our results show: 1) moderate-to-strong anisotropy exists in all studied sedimentary layers, especially in both the Upper Bakken and Lower Bakken shale formations, where the Thomsen parameters (E and y) can be over 40%; 2) all events selected for high signal-to-noise ratio and used for the joint velocity inversion are located in the Bakken and overlying Lodgepole formations, i.e., no strong events are located in the Three Forks formation below the Bakken; 3) more than half of the strong events are in two clusters at about 100 and 150 meters above the Middle Bakken. Re-occurrence of strong, closely clustered events suggests activation of natural fractures or faults in the Lodgepole formation. Finally, I introduce a new hybrid method to model the shear (SH) wave scattering from arbitrarily shaped fractures embedded in a heterogeneous medium by coupling the boundary element method (BEM) and the finite difference method (FDM) in the frequency domain. The hybrid method can calculate scattering from arbitrarily shaped fractures very rapidly, thus Monte Carlo simulations for characterizing the statistics of fracture attributes can be performed efficiently. The advantages of the hybrid method are demonstrated by modeling waves scattered from tilted fractures embedded in complex media. Interesting behaviors of the scattered waves, such as frequency shift with the scattering order and coherent pattern of scattered waves through strong heterogeneities, are observed. This method can be used to analyze and interpret the scattered coda waves in the microseismic observations, e.g., the reverberating multiples in the Bakken microseismic data which cannot be explained by the determined layered anisotropic velocity model alone.