Blast Design Optimization Using Artificial Neural Network Approach and Langaford – Kihlstrom Model for Selected Rock Types in Nigeria

Blasting is an important process in mining that affects loading, hauling, and crushing. A well-planned blast design helps to achieve good fragmentation, reduce energy use, and lower costs. This study develops an optimized blast design model by combining the Langaford-Kihlstrom method with an Artificial Neural Network (ANN) to improve blasting efficiency for selected rock types in Nigeria. The research examined the geomechanical and structural properties of limestone, granite, and marble from Lafarge Ewekoro Mines, Mercy Quarry, and Jakura Marble Obajana Mines. Parameters such as Uniaxial Compressive Strength (UCS), Rock Quality Designation (RQD), and Blastability Index (BI) were determined through laboratory and field tests. A comparison of ANN predictions with the Langaford-Kihlstrom model and quarry data showed that ANN provided more accurate estimates of explosive requirements and reduced the overestimation problems of empirical models. For limestone at Ewekoro Mines, ANN predicted explosive usage of 1,873.737 kg for Blast 1, closely matching the quarry data (1,900 kg), while the Langaford model overestimated at 2,316.5 kg. In granite blasting at Mercy Quarry, ANN estimated 1,628.187 kg for Blast 1, compared to quarry data (1,900 kg) and the Langaford model (2,316.5 kg). Similarly, for marble at Jakura Marble, ANN predicted 7,123.353 kg for Blast 1, closely matching the quarry data (7,025 kg), while the Langaford model significantly overestimated at 20,606.3 kg. The ANN model also reduced costs. For granite blasting, ANN lowered explosive costs by 39.7% compared to quarry estimates and 48.2% compared to the Langaford model. ANN optimization resulted in a 16.6% overall reduction in total explosive costs across all rock types. ANN predictions also improved fragmentation, producing sizes that reduced the need for secondary blasting. Performance evaluation showed that the ANN model achieved an R² score of 92.3% and an RMSE of 6.03, demonstrating its accuracy in blast design. The findings highlight the usefulness of ANN in improving blast planning, reducing costs, and improving efficiency in mining operations.