Study Of Reservoir Induced Seismicity And Hydraulic Diffusivity

The Mead Lake is the first reservoir considered to be responsible for nearbyearthquakes reported in1945. Since then, about150reservoirs are believed to haveinduced seismicity. Reservoir-induced seismicity (RIS) has been drawn wideattention for a long time. In China, thirty four out of114reservoirs with dams over100m have induced remarkable seismicity. In order to solve the electric power supplyproblem, reservoirs have been built not only in eastern China with low level ofseismicity, but also in western China with high level of seismicity. After the2008Wenchuan MS8.0earthquake, whether the Zipingpu reservoir nearby the epicenterwas genetically related with this event has become a focused issue, but the questionremains to be controversial so far. This thesis attempts to explore the mechanism ofRIS taking the Zipingpu reservoir as an example.It is usually accepted that reservoir loading and pore pressure are two possiblegenetic factors for RIS. There have been many studies on the role of reservoir loadingin RIS, but more suggest that pore pressure should receive much attention. As it isdifficult to directly measure hydraulic diffusivities in the crust, the study on thisproblem seems to be in slow advance. The key problems are as the following:(1)How to develop a method to obtain the hydraulic diffusivities.(2) What role porepressure plays in reservoir-induced seismicity?(3) Typical patterns inspatial-temporal evolution of RIS.This thesis begins with analyzing the characters and main factors of RIS all overthe world. Then the Coulomb stress changes on actual fault by reservoir loading arestudied with different geometric factors. Finally, this study focuses on the role of porepressure in RIS and the pore pressure diffusivity.The major results of this work are summarized below.1. The Coulomb stress changes on the fault by reservoir loading are analyzed withdifferent fault geometries and different positions of the reservoir with respect to thefault. The reservoir plays different roles depending on its position on the hanging wallor the foot wall. Loading promotes fault slip in the former case and impedes faultmovement in the latter case. And then, a Coulomb stress change by the Zipingpureservoir before the Wenchan earthquake is obtained with the surface constraints and the pore pressure considered.2. The importance of hydraulic diffusivity distribution in the crust for studying RISis shown by analyzing Zipingpu reservoir impoundment and seismicity evolutionnearby. The fault geometric structure, rock type, fracture spacing and intensity mightinfluence hydraulic diffusivity distribution. Three possible models of hydraulicdiffusivity near the reservoir are obtained as a function of depth and orientations. Theresults indicate that hydraulic diffusivity is heterogeneous and anisotropic in the crust.3. The rate of change in water head increment is an influential factor for seismicityevolution. Comparison of the rate of change of water head increment and earthquakeswarms at Shenxigou and Shuimogou shows that it is the increase rate of water headinstead of the water head that played a role in the change of seismic activity. Theseismic activity began when the rate of change of water head increment began to rise.The seismic activity reached a high level when the rate of change of water headincrement was at the peak. The seismic activity turned to be low with the drop of rateof change of water head increment.4. The range of expansion coefficient of seismicity contains information about thecauses of earthquake. Shapiro S. A.(Shapiro,2005) found that spatial and temporaldistribution of seismicity induced by well injection included three processes: theinstant expansion phase, parabola expansion phase, and convergent phase. In thisstudy, the Shapiro’s method is applied to analyzing RIS near the Three Gorgesreservoir and Xiaolangdi reservoir, and then the expansion coefficients for isolatedearthquakes and aftershocks were estimated. It is found that these earthquakesequences of different types show similar three processes as seismicity was inducedby well injection. But the values of their expansion coefficients are significantlydifferent. The hydraulic diffusivities of well injection and reservoir impoundment arevery small. The expansion coefficients of aftershocks of some isolated earthquakesare also small, while that of large temblors is large.5. The range of expansion coefficient of seismicity contains information about thestress evolution on the fault. The expansion processes of range of seismicity afterZipingpu reservoir impoundment were analyzed in detail. The instant expansion phase,parabola expansion phase, and contraction phase also appeared with four times ofwater level rising. Then the expansion coefficient of range of seismicity is obtained.With the rising of water level, the expansion coefficient increased too. When the water level reached the first peak and the second peak, the expansion coefficient ofrange of seismicity suddenly increased up to158m2/s. The cause of the increment ofthe expansion coefficient is also discussed accordingly.This thesis suggests that the method of studying hydraulic diffusivity by analyzingthe temporal distribution of earthquakes of well injection can also be used in studyingRIS, as well as in estimation of the range of expansion coefficient of aftershocks.From the difference of expansion coefficients of different types of earthquakes,different mechanisms of earthquakes can be revealed. These results serve to helpanalyze the types of earthquakes. The change of the expansion coefficient of theZipingpu reservoir region is likely to reflect that the cause of seismicity before theWenchuan event changed from water-diffusion triggering to a phase dominated byfault-segment interaction and linking. This result may serve to help identify inidentifying the stress state of the related fault.