The practical challenges of managing and processing measuring-while-drilling (MWD) data in real life situations-Paper

Paper

Pdf Summary

The paper examines practical methods for processing Measuring-While-Drilling (MWD) data from fissure-grouting works in chalk formations at UK tunnel cross-passages. The study aimed to use drilling data to characterize ground conditions, improve drilling consistency, and ultimately forecast grout requirements and assess changes after primary grouting.<br /><br />The authors highlight key challenges: inconsistent units and sensor configurations, the lack of standardized rules for removing irrelevant data, and uncertainty about which parameters best represent a particular ground. They describe a preprocessing workflow using Python and spreadsheets. This involved standardizing units, accounting for equipment settings and borehole inclination, calculating derived parameters such as penetration per revolution and drilling indices, and filtering out non-drilling activities. An algorithm classified data as drilling, flushing, redrilling, or rod extraction, retaining only active drilling records for analysis.<br /><br />Exploratory analysis showed that guiding operators to maintain a relatively constant rotation speed improved relationships among drilling parameters, including advance speed, net force, and drilling energy. This supports the idea that reducing operational variability can make MWD data more useful for interpreting ground response.<br /><br />Attempts to correlate individual drilling measures, such as the Somerton Index, with grout volume and pressure produced weak results. The authors note that the interaction of multiple drilling variables may obscure simple one-to-one relationships. They conclude that accessible tools can provide valuable initial insights, while robust analysis requires careful data preparation and further investigation. Future work will apply multivariable correlation and clustering methods, prepare the full dataset on a larger data platform, and evaluate machine-learning approaches to predict grout quantities and improve ground-improvement planning.

Keywords

Measuring-While-Drilling

MWD data preprocessing

fissure grouting

chalk formations

tunnel cross-passages

drilling parameters

ground condition characterization

grout volume prediction

data standardization

machine learning