Improvement of Microseismic Location Based on an Anisotropic Velocity Model
Dai Feng,Guo Liang,Xu Nu-Wen,Fan Yi-Lin,Xu Jian,Jiang Peng
DOI: https://doi.org/10.6038/cjg20160914
2016-01-01
Abstract:As a kind of three-dimensional detection of seismic events, microseismic (MS) monitoring techniques have been widely used in the world for many years to assess the stability of surrounding rock. The location of MS events is foundation of MS monitoring. Resulting from anisotropy and inhomogeneity of regional rock masses, the velocity of stress waves may be different in different directions and areas. So, it is difficult to figure out a reasonably accurate location. Aimed at the regional rock masses with stratified velocities, a simplified and stable stratified velocity location model (SV) is proposed.The SV model is proposed based on the anisotropic velocity model, which simplifies the propagation path of stress waves into a straight line, which consists of horizontal and vertical velocities, that is, one for underground sensors and the other for surface sensors. Taking into account that the vertical stress waves pass through some rock formations at different velocities, the model is constructed part by part according to the values of each interval velocity. The first thing is to derive the nonlinear equation of stress wave travel time by spatial analytic geometry's operation and analysis. The target function of SV is established by using L2 norm of the time residual between the measured value and calculated value of the travel time difference of each two sensors. Secondly, based on the established target function, the SV is proposed, which consists of parameters-preparation module, interval velocities inversion module and seismic location module. In the parameters-preparation module, the regional rock mass is divided into some wave velocity layers according to engineering geological data. And, using spatial analytic geometry's operation and analysis, the analytical expressions for layer interfaces are derived. In the interval velocities inversion module, the target function to solve velocities is figured out according to the form of the target function of SV. Then we collect some blast events or rockburst with known coordinate values. Using the genetic algorithm (GA), we can get optimization solutions of velocities. In the seismic location module, using the determined velocities, the target functions to locate MS events are figured out according to the form of the target function of SV as well. It should be emphasized that during the optimal solving, for every wave-velocity layer, the concrete forms of target function and searching regions are different from each other. Using GA, the optimization solutions of a MS event for each wave-velocity can be worked out, and the minimum of optimization solutions is considered to be the location of one MS event.In order to verify the effectiveness and reliability of SV, we applied it to a slope engineering on the left bank of the Baihetan hydropower station, where a MS monitoring system (ESG, from Canada) has been installed. Firstly, the slope was divided into 11 wave-velocity layers, which included 4 kinds of velocities, according to the sonic wave testing given by engineering geological investigation data. And, the analytical expressions for layer interfaces were worked out. Secondly, we collected 4 blast events that occurred at 610 m-elevation irrigation and drainage galleries. Using GA, we got optimization solutions of velocities, which are 4500 m . s(-1), 4900 m . s(-1), 3750 m . s(-1), 5000 m . s(-1), respectively, through minimizing the target function. And then, 11 target functions to locate MS events for each wave-velocity were figured out by using the determined velocities and parameters of layer interfaces. The MS events were collected from 1 to 18, April, 2015. The locations of these MS events were determined by using the simplified single-velocity model, of which the velocity is 4600 m . s(-1) (approximate mean value of 4 velocities above). Finally, those MS events, which were scattered around two dislocation bands (LS331, LS337), were relocated by using the SV. The relocation results show that the maximum, minimum and average deviations from dislocation bands decrease by 57. 17%, 36. 51% and 57. 35%, respectively. It is clear that SV is more convenient and reliable than the single velocity location model aimed at those regional rock masses with stratified velocities.The improved location model (SV), which is based on the anisotropic velocity model, takes into account the impact from the different velocities of rock formations with different lithological characters. Compared with the single velocity model, the accuracy of MS events location aimed at the regional rock masses with stratified velocities is improved obviously. However, like all MS events location models, it requires an accurate velocity-distribution of regional rock masses. Further study on improving the accuracy of MS events location can be achieved through considering velocity partitions inside each rock formation and finding a stable location model, which is more suited to the calculation of the actual stress wave traveltimes.