Open Access
Liu, Wenpeng
Graduate Program:
Mining Engineering
Doctor of Philosophy
Document Type:
Date of Defense:
November 17, 2017
Committee Members:
  • Derek Elsworth, Dissertation Advisor/Co-Advisor
  • Derek Elsworth, Committee Chair/Co-Chair
  • Shimin Liu, Committee Member
  • Jeremy M. Gernand, Committee Member
  • Asok Ray, Outside Member
  • Jamal Rostami, Special Member
  • Joint Detection
  • Rock Strength Estimation
  • Drilling Parameters
  • Roof Bolter
  • Pattern Recognition Algorithms
  • Wavelet Analysis
  • Ground Support Optimiztion
  • Ground Control
  • Composite Parameters
  • Mining Health and Safety
  • Roof Bolts
Accurate understanding of geological features, including locations of joints, cracks, bed separations, and rock strength, allows for optimization of ground support measures and mitigation of ground instabilities in underground structures. The concept of using operational parameter data, collected by monitoring the work cycle of a roof bolting unit drilling into roof and ribs, to predict geological features of interest has been proposed in the past and studied in the last few decades. Some smart drilling systems have been developed to implement this concept, but despite their limited success on joint detection and/or rock classification, they fail to identify hairline joints, (aperture less than 3.175 mm) and discriminate between rocks with similar strengths. This research aimed to advance the existing smart roof bolting systems to enhance their capabilities to sense geological features of interest along boreholes. To achieve this objective, full-scale laboratory tests were conducted involving a set of concrete blocks with various strength properties and small joints to simulate drilling from rocks of various strength properties into another. Some pattern recognition algorithms were developed to detect pre-designed joints. To improve capabilities of the existing algorithms, several composite parameters were introduced to provide collaborative decisions for locating the joints. Moreover, wavelet analysis was also employed to improve pattern recognition algorithms and therefore to enhance their capabilities for joint detection. A set of additional holes were also drilled into a block that included joints at four different angles (15o, 30o, 45, and 60o) relative to the direction of drilling. The area between the joints were filled with grout having various strength. Also, a sample composed of blocks of various rocks were cast in grout to represent variation of rock strata while drilling. The rocks used in this composite block included soft shale, sandstone, limestone and shale with strength ranging from 3 to 130 MPa. These tests allowed examination of the capabilities to identify angled joints, while generating data for the programs for estimating rock strength. The result of the analysis of the drilling parameters proved that joints with smaller aperture (less than `3 mm, 1/8th inch) could be successfully detected at high rates, reaching 94% by using feed pressure. The algorithms have also resulted in generating various amounts of false alarms, but the improved algorithms have been able to reduce the false alarms down to 14 in a set of 156 drill holes tested. The use of composite parameter RP/FP/PR and the same algorithms could increase the detection rate to 97%, with false alarms reduced to 9. Use of wavelet and other noise filtering systems could also improve the detection rates and reduce false alarms compared to the straight use of single drilling parameters but could not substantially increase the detection rates. Therefore, it was concluded that the use of composite parameters was sufficient for the data set that is currently available. The same was true for detection of angled joints, but the available data in this setting was only on a few drill holes. As for the estimation of rock strengths by monitoring drilling parameters, data from drilling into the composite sample showed very good correlation between Field Penetration Index (FPI), which is calculated from feed pressure and drilling rate, and rock strength values. This is especially true when a Wear Index (WI), based drilling distance on a given bit, was used to adjust the calculated values of FPI and account for the wear on the drill bit. Correlation coefficient for statistical analysis of rock strength data from drilling parameters in the limited full-scale drilling tests were around R2 = 92%.