Bit Selection Optimization Using Artificial Intelligence Systems

Abstract

Optimizing bit selection is one of the main challenges in drilling operations. Bit selection is based on the recorded performance of similar bits from offset wells. There are too many parameters intervening in drilling bit selection. Therefore, developing a logical relationship between them to assist in proper bit selection is extremely necessary and complicated though. In such a case, artificial neural networks (ANNs) have proven to be helpful in recognizing the complex relationship between variables. In this new approach, two models are developed with high proficiency using ANNs. The first model provides appropriate drilling bit selection based on the desired rate of penetration (ROP) to be obtained by applying specific drilling parameters. The second model uses proper drilling parameters obtained from an optimizing procedure to select the drilling bit that provides the maximum achievable rate of penetration. Genetic algorithms (GAs), as a class of optimizing methods for complex functions, are applied to help bit optimization and its related drilling parameters. With the given data sets, these new models predicted successfully the bit types and the optimum drilling parameters. The correlation coefficients for the predicted bit types and optimum drilling parameters in testing the obtained networks are 0.96 and 0.86, respectively. MATLAB software was used to perform ANN and GA solutions.