Time-lapse Passive Seismic Velocity Tomography of Longwall Coal Mines: A Comparison of Methods
Luxbacher, Kramer Davis
MetadataShow full item record
Time-lapse passive seismic velocity tomography was conducted utilizing data from three underground longwall coal mines to produce a better understanding of the processes that lead to ground failure in mines, especially large, violent failures, such as bumps. Two of the datasets, US Western I and US Western II, were collected at bump-prone underground longwall coal mines in the Western United States using surface mounted receiver arrays, while the third data set was collected at an underground longwall coal mine in Australia utilizing an underground array. The Australian mine was experiencing problems with periodic caving and subsequent wind blasts, rather than bumps. Seismic velocity tomography allows for non-invasive imaging of a rock mass and inference of stress redistribution from the velocity images. These tomograms are unique as they are generated using source data that was collected remotely and the sources are mining-induced. Tomograms were generated using three inversion methods: simultaneous iterative reconstructive technique (SIRT), double difference least squares event relocation, and least squares event relocation. The three methods were compared and contrasted to determine if one is superior and if event relocation improves the image. Also, the tomograms were analyzed to determine if passive seismic velocity tomography is an appropriate technology for the study of stress in mines and assistance in forecasting of bumps. The tomograms were compared with known roof events, face advance, and fall locations at the mines to establish if expected stress features can be imaged with velocity. Finally, synthetic tomograms were generated using a starting velocity model that approximates the predicted â trueâ model for each mine to determine if the velocity images produced correlate with the theoretical stress state at the mine. Results indicate that high velocity zones correlating with high stress abutment regions can be imaged for the US Western I data set with all three inversion methods, but the SIRT method provided the best agreement when the synthetic tomogram was generated. Additionally, a low velocity zone that correlates with the gob is consistently imaged. These features also redistribute with face advance. The US Western II data set was not as densely sampled as the US Western I data set. A low velocity region was consistently present in the gob area and redistributed with face advance, but abutment stress features were not evident. Additionally an unexplained high velocity feature was evident on several of the tomograms. Synthetic tomography indicated that the double difference least squares event relocation method is most appropriate for this data set. Finally, the Moonee Colliery results, which were also not as densely sampled as US Western I were uncertain. While velocity anomalies were often present in the vicinity of a fall, the anomalies were not reliably high or low. Again, synthetic tomography indicated that the double difference damped least squares event relocation method was most appropriate for this data set. The tomograms presented indicate that source-receiver configuration and density and variable gridding are extremely important in the application of passive seismic velocity tomography to mines. The source-receiver configuration and density determine how well various areas of a model are constrained, and the variable gridding allows areas that are not well sampled to still be adequately constrained. As a result of this work several things can be drawn about requirements that must be met in order to utilize seismic velocity tomography for inference of stress in underground mines. First, typical longwall stress abutment patterns can be inferred from velocity images of underground coal mines. Second, synthetic tomography and analysis of this tomography, in addition to some knowledge of the general location and frequency of microseismic events, is necessary prior to designing receiver arrays for passive seismic velocity tomography. Suboptimal source-receiver configurations may be used for passive seismic velocity tomography, but there is a minimum threshold for the number of raypaths that must be met that is unique to each site. Finally, a good understanding of the mechanics of stress and failure at the site is necessary to interpret the tomograms.
- Doctoral Dissertations