Papers by victor akinlosose
This study aims to enhance the blast design efficiency by incorporating Langefors-Kihlstrom model... more This study aims to enhance the blast design efficiency by incorporating Langefors-Kihlstrom model and artificial neural networks (ANN) while considering rock strength and structural properties of sedimentary limestone deposit at Dangote cement, Ibese mine site. Blast data were obtained from seven drilling and blasting operation at phase II of the mine site, important blast parameters recorded includes bench height, drilled hole diameter, round width, burden, spacing, specific charge, and hole depth among others. Rock strength properties including an estimated mean unconfined compressive strength (UCS) average value of 44 MPa obtained from the rock samples, a rock density of 2.2 g/cc, blastibility index (BI) of 30, and an average rock quality designation index (RQD) of 60% were obtained from joint spacing and volumetric joint data. ANN model was trained using blast datasets collected from the Langefors-Kihlstrom model, which incorporated the UCS, RQD, BI, and rock density as inputed weights to predict the outcomes of different blast designs. The trained datasets were then tested and compared to the actual blast results to evaluate the effectiveness of the blast design optimization. The results showed that the ANN and Langefors-Kihlstrom model enhanced the efficiency of the blast design at phase II saving an average of 16.6% of explosive cost when compared to existing data. The regression analysis between the existing data and Langefors-Kihlstrom data as shows that the model has a high R-squared value of 0.90 and a low RMSE of 9.19. This indicates that the blast model is able to explain a large portion of the variability in the data and is performing well. The ANN data, on the other hand, has a higher R-squared value of 0.97 and a lower RMSE of 6.06, which indicates that the ANN model is performing even better. Further research can be carried out using machinelearning algorithms such as random forest and support vector machine in predicting blast design that can be suitable for generic rock properties.
Uploads
Papers by victor akinlosose