Increasing drilling efficiency: rock drillability assessment with improved reliability methods

doi:10.1007/s00500-025-10982-5

2026-01-21

PRE

Hadi Fattahi *

Fateme Jiryaee

DOI：10.1007/s00500-025-10982-5

原文链接

原文求助

摘要

En 中文

The assessment of a successful and cost-effective drilling operation hinges on the evaluation of the Drilling Rate Index (DRI). Moreover, the DRI’s value, subject to fluctuations driven by various rock parameters, remains beyond control, introducing an element of unpredictability. Consequently, deterministic methods prove inadequate for the analysis of rock drillability. Nonetheless, the reliability analysis, which incorporates the uncertainty surrounding these parameters while addressing the problem, offers the advantage of providing assessments that better align with real-world conditions. In this study, rock drillability was appraised by employing various reliability methods, including the First Order Reliability Method (FORM), Monte Carlo Simulation (MCS), and the Second Order Reliability Method (SORM). Additionally, Genetic Algorithm (GA) was employed in the optimization processes associated with these methods. A critical aspect within reliability methods concerns the definition of the Limit State Function (LSF), which, in this context, is articulated using the Group Method of Data Handling (GMDH) neural network approach, yielding an explicit function. In this approach, the random variables include Brazilian Tensile Strength (BTS), Uniaxial Compressive Strength (UCS), axial point load strength index (Is(50)↓), diametral point load strength index (Is(50)→), and Shore Scleroscope Hardness (SSH), while the DRI serves as the output parameter. To evaluate the efficacy of the proposed reliability methods, field datasets from previously published literature were employed. Lastly in this paper, according to the US Army Corps of Engineers, the probability of failure (Pf) calculated using the FORM, SORM, and MCS methods is low.

Keyword:

Rock drillability

Reliability analysis methods

GMDH neural network

Genetic algorithm