Probabilistic bit selection using bit performance data, bootstrap sampling, and Monte Carlo method

Purpose This paper aims to propose a probabilistic framework for selecting the right drill bit in the presence of uncertainties related to various input parameters. For this purpose, a deterministic bit selection method based on dull bit grading (DBG) was adopted to extract the bit dullness data. Then, a limit state function was proposed for the bit selection according to the deterministic model concerned. Defining the input parameters as random variables, the problem was established as a reliability model. Next, an algorithm was suggested via establishing a historical data set and then applied to generate artificial data using bootstrap sampling. The generated artificial data were subsequently loaded into a Monte Carlo algorithm, wherein the exceedance probability of the performance index was calculated. Afterward, the resulting probability was used as the decision criterion for the bit selection. The results revealed that the exceedance probability drastically dropped as the performance index elevated, reaching sub-5% values at performance indices higher than 0.65, which indicated the unlikeliness of the bits achieving such performance levels. However, out of the four bits examined in this work, i.e. 517, M23, M223 and M1665, the M223 exhibited higher exceedance probabilities, making it a more suitable bit for drilling. Design/methodology/approach This study introduces a probabilistic framework for drill bit selection, incorporating uncertainties in input parameters. Initially, DBG is used to assess bit dullness. A limit state function based on the deterministic bit performance model is developed, and the input parameters are treated as random variables. Bootstrap sampling generates artificial data sets, which are analyzed using the Monte Carlo method to calculate exceedance probabilities of bit performance indices. These probabilities serve as the decision criterion for bit selection. The method integrates reliability analysis and historical data to propose a novel and robust approach for optimizing drill bit selection. Findings This study demonstrates that bit selection probabilities decrease significantly as the performance index increases. Among the four examined bits (517, M423, M223, M1665), the M223 bit consistently exhibited the highest exceedance probabilities across a range of performance thresholds, marking it as the most reliable for the evaluated geotechnical conditions. The probabilistic approach provides a nuanced understanding of bit performance, accommodating uncertainties and offering a superior alternative to deterministic methods. Additionally, convergence analysis confirmed the stability and accuracy of the Monte Carlo simulations, ensuring that the reliability-based findings are robust and practical for field implementation. Originality/value This research introduces a novel integration of probabilistic modeling and reliability analysis into the drill bit selection process, bridging a significant gap in deterministic methodologies. By leveraging bootstrap sampling and Monte Carlo simulations, the proposed framework offers a more accurate and uncertainty-aware decision-making tool. The study’s findings, including exceedance probability curves and comparative performance indices, provide actionable insights for field applications. The adaptability of the proposed framework to different data sets and grading systems further enhances its value, making it a versatile solution for optimizing drill bit selection in diverse drilling environments.