Using Percentiles and Distributions to Evaluate Drilling Performance
Percentiles locate observations within a defined population. Metric direction, analytical unit, weighting, cohort, spread, and sample size determine their meaning.
Read article →Technical resources
Drilling engineering, explained from the data up.
Practical articles about drilling data architecture, statistics, engineering models, real-time systems, directional work, and applied AI.
The Average Trap: Comparing Drilling Performance With Distributions
Averages can hide repeatability, multimodal behavior, and rare operational delays. Distributions reveal what is typical, variable, and worth investigating.
Read article →Depth-Based and Time-Based Analysis of Real-Time Drilling Data
Time preserves operational sequence; depth preserves location. Converting between them changes eligibility, weighting, aggregation, and resolution.
Read article →Choosing the Right Sampling Rate for Real-Time Drilling Analytics
Choose sampling rate from the shortest behavior that must remain visible. Acquisition, analysis, aggregation, and display can operate at different rates.
Read article →Why Rig-State Classification Is Foundational to Real-Time Drilling Analytics
Rig state gives raw channels operational meaning and routes each interval to the correct KPI, model, event detector, or alert. Classification errors propagate everywhere.
Read article →The Hidden Data-Engineering Work Behind Real-Time Drilling Analytics
Reliable real-time analytics starts with channel identity, units, clocks, quality, wellbore and BHA context, lineage, and recovery from delayed data.
Read article →Two Clocks, One Well: Modeling Time- and Depth-Indexed Data
Time and depth answer different drilling questions. Reliable alignment requires explicit grain, eligibility, direction, tolerance, survey authority, and provenance.
Read article →Why Drilling Data Is Not Just Time-Series Data
Drilling data needs time, depth, wellbore, rig-state, equipment, unit, and provenance context. Without that structure, valid measurements can still produce the wrong engineering answer.
Read article →