h. samareh

Toe and boulder due to blasts in the open pit mines reduce the production efficiency and increase extraction costs. In this study, these phenomena were reduced to correct the blast parameters of the ore blocks in the Sirjan Golgohar-2 Mine. For this purpose, the values of seven effective parameters including, an average of depth holes, burden, sub-drilling, powder factor, spacing, rock quality designation, and blastability index were collected for 19 blasting blocks in the studied mine. In this research, the values of Geo-mechanical properties of rock mass were obtained by the photogrammetric method and discontinuity set extractor software. Subsequently, the experimental models were created to predict the volume of boulder and toe relative to the volume of blasting block by nonlinear multiple regression. The predicted ability related to each of the created models by statistical indicators was investigated, and it was determined that the polynomial model to product boulder and the exponential model for toe are more accurate with 94.43 and 98.13 coefficients of determination respectively. Then, the minimization process of these phenomena was performed to access optimal values of controllable parameters and their coefficients in each of the created two models simultaneously by the combinational algorithm of Particle Swarm Optimization-Genetic algorithm. Finally, to evaluate the predicted ability of two optimized models, four blasts based on the optimized information were performed on the mine. The results showed that the models predicted the volume of boulder and toe relative to the block volume with the Root Mean Square Error 0.47 and 0.08 respectively.

Drilling a certain number of blast holes per hour plays an important role in reaching the required annual mine production. Since, the proper use of the availability time of machine to drill the maximum number of blast holes is crucial. The drilling rate of blast holes is affected by various factors such as rock properties and system parameters. Thus, recognizing the effectiveness of these factors on the penetration rate of bit (PR), not only machine production can be increased but also drilling and blasting costs can be reduced in the mine. In this study to predict the PR in the selected mine, firstly, parameters of 91 holes related to 28 blasting block and in 9 various extracting benches were collected. Secondly, the sensitivity rate related to each of the independent parameters on the PR was studied using Cosine Amplitude Method (CAM). Finally, three models including non-linear multivariate regression (NLMR), artificial neural network (ANN), and fuzzy logic were produced to predict the PR. These models were validated using 12 series of data tests. It was shown that with a coefficient of determination of 0.68 and mean absolute percentage error (MAPE) of 12.15, the ANN model could predict the PR with a slightly higher precision compared to NLMR.