Drilling and blasting procedures are a critical part of mine planning activities and improvements in this stage can lead to better productivity downstream and lower costs. One potential improvement would be better understanding the characteristics of the rock for blast design purposes. The distribution of material properties within a rock mass is very unpredictable so to more accurately determine its characteristics a controlled drilling environment is needed. Many mines possess the capacity to record Measurement While Drilling (MWD) data but don't utilize it. This project investigates and analyzes MWD data from an anonymous iron ore mine. Machine learning was used to analyze the MWD data for the sake of improving blast optimization and productivity and has been used to successfully implement MWD data in other studies. Based on previous work, it has been demonstrated that the utilization of MWD data can assist with developing a better understanding of rock mass properties and other variables of importance during the drill, blast, and mine planning processes. This report investigates using MWD data to classify and predict lithology and utilize regression modeling to identify potential soft spots within blast patterns for blast optimization. The MWD data of six blast patterns from an anonymous mine underwent data processing and then were modeled. The lithology was able to be approximately classified with new information of potentially revealed bed boundaries and blast pattern soft spots. / Master of Science / In the mining industry, liberating ore from the ground is necessary to process the material and generate products. To accomplish this liberation objective a process of drilling and blasting is utilized. A pattern is designed, and holes are drilled that match the spacing and depth of the design. The blast holes are loaded with explosives and detonated to create fractured rock for the liberation of desired material. During the drilling process, drilling parameters are recorded called Measure While Drilling (MWD) data. Previous research has demonstrated that modeling techniques using MWD data can assist with developing a better understanding of rock mass properties and other variables of importance during the drill, blast, and mine planning processes. Utilizing MWD data and machine learning to improve blasting procedures by classifying and predicting bed assignment and potential soft spots in a blast hole will be investigated in this research. The MWD data comes from 6 blast patterns from an anonymous iron ore mine. After the data was processed and modeled the lithology was classified with a validation accuracy of approximately 78% and potential soft spots estimated.
| Identifer | oai:union.ndltd.org:VTETD/oai:vtechworks.lib.vt.edu:10919/117337 |
| Date | 10 January 2024 |
| Creators | Arnold, Joshua Ryan |
| Contributors | Mining Engineering, Westman, Erik Christian, Ripepi, Nino S., Pandey, Rohit |
| Publisher | Virginia Tech |
| Source Sets | Virginia Tech Theses and Dissertation |
| Language | English |
| Detected Language | English |
| Type | Thesis |
| Format | ETD, application/pdf, application/pdf |
| Rights | In Copyright, http://rightsstatements.org/vocab/InC/1.0/ |
Page generated in 0.002 seconds