Rig State Detection: Turning Realtime Data Into Operational Insight

Raw drilling data is valuable, but it is difficult to interpret at scale without context. Rig state detection classifies sensor streams into operational activities such as drilling, sliding, circulating, tripping, connection, reaming, and static time. Once the data is labeled, teams can measure where time is actually spent. This activity timeline becomes the foundation for performance analytics. It allows engineers to compare wells, sections, crews, rigs, and operational practices with a common language.

Manual activity coding can work for small reviews, but it becomes inconsistent across campaigns. Two engineers may classify the same interval differently, and important transitions can be missed when data volume is high. Automated rig state detection gives the team a repeatable baseline. Human review still matters. The strongest workflow combines automated classification with exception review, so engineers spend time on intervals that require judgment instead of labeling every minute from scratch.

Once activities are classified, the team can compare flat time, connection time, sliding efficiency, tripping speed, and invisible lost time across offset wells. This makes performance improvement more concrete. Instead of saying a section felt slow, the team can identify the exact activity that drove the variance. Rig state data also improves lessons learned. If a well had repeated circulation delays or long transition times, those patterns can be built into the next plan and tracked during execution.

Good rig state detection depends on consistent time-series data. Essential channels usually include depth, bit depth, hookload, block position, RPM, torque, flow rate, standpipe pressure, and weight on bit. Missing or noisy channels can still be handled, but the confidence score should reflect that uncertainty. For operational use, the model should expose both the assigned state and its confidence so engineers understand where the classification is strong and where review may be needed.