Real-time drilling monitoring: what to watch, and when to call

A live drilling screen shows dozens of channels, and most of the time nothing on it needs a decision. The skill is knowing which few channels matter in each operation, what normal looks like for them, and which changes mean stop and call. This article works through 21 hours of a 6⅛″ lateral: one stand and its connection read channel by channel, a pit gain the alarms caught or missed, two alarm rule sets compared, and a break in the feed.

What to watch, by operation

The channels that matter change with what the rig is doing. A torque spike while drilling is information; the same spike with the bit off bottom and the pumps off is noise. Read every channel against the rig's state first.

Two rules follow. Each warning sign is a change against a baseline, so the baseline must be known: the model's hook loads, the circulating pressure at this rate, the volume the hole should take. And each sign has a first action that needs no meeting. A pit gain means a flow check. Overpull past the margin means stop pulling.

A stand and a connection, read

The example is the 6⅛″ lateral of a horizontal oil producer, set in typical North Oman conditions: a carbonate reservoir at 1,500 m TVD, 1.12 SG drill-in fluid and 4″ drill pipe. The 21-hour window starts circulating at the 7″ liner shoe at 1,760 m, drills twelve 29 m stands to 2,108 m, circulates bottoms-up, makes a wiper trip, waits 75 minutes, and circulates again. Thirteen channels arrive every 5 seconds.

While drilling, the hook reads 76.5 klbf: the rotating weight of 87.5 klbf less the 11 klbf on the bit. When the stand is down the driller stops rotating and picks up. For 35 seconds the hook reads 96.1 klbf, the model's pick-up weight, then hangs still at the pumps-off rotating weight before the slips go in. In slips the hook carries only the block, 22 klbf, for 2 minutes 20 seconds. On the way back down the string slacks off at 78.5 klbf, then washes the last metre to bottom rotating at 87.9.

That connection carries three readings worth keeping. The pick-up and slack-off weights go on the torque and drag model's broomstick, stand after stand; a pick-up that drifts above the model is the earliest sign of a cuttings bed or a tight zone (see torque and drag, worked through on a lateral). The off-bottom torque, 1.4 to 1.8 kft·lbf here while washing down, is the same check for rotation. And the pit with the pumps off shows how much the annulus gives back, which is the baseline for every flow check.

Trips need the same arithmetic on the trip tank. Each 29 m stand of 4″ drill pipe with a 3.34″ bore displaces (4² − 3.34²) / 1,029.4 = 0.00471 bbl/ft of steel, 0.448 bbl a stand. Pull ten stands dry and the hole must take 4.5 bbl from the trip tank to stay full. If it takes 3 bbl, the other 1.5 bbl came from the formation, and the call is a flow check before the next stand. Fill the tank sheet stand by stand, compare it with the calculated line, and treat a hole that takes less than its steel as a kick until shown otherwise. The trip sheet article works this through for a whole trip.

Standpipe pressure tells its own story through the connection: 2,900 psi on bottom, 2,750 psi as the bit lifts, zero with the pumps off, and back to 2,900 when the bit takes weight. A return to a different on-bottom pressure after a connection is worth a note on the log.

A hand check of the pit

Pit volume is the channel that cannot wait, and the one most often misread, because it is never flat while drilling. Work out what it should do.

1. Hole volume. A 6⅛″ hole holds 6.125² / 1,029.4 = 0.0364 bbl/ft, or 0.120 bbl/m. Each 29 m stand takes 3.5 bbl from the pit.
2. Rate while drilling. Stand 11 drills at 21 m/h, so the pit should fall by 21 × 0.120 = 2.5 bbl an hour, 0.042 bbl a minute.
3. The alarm limit. The fixed rule re-sets at each connection to the volume then plus 3 bbl. After the connection at 2,050 m the pit reads 385.1 bbl, so the limit is 388.0. By 2,060 m, where the transfer starts, the hole has taken 10 × 0.120 = 1.2 bbl more.
4. Time to alarm. The transfer must make up 3 + 1.2 = 4.2 bbl, at 0.5 − 0.042 = 0.458 bbl a minute net: 9.2 minutes.

DrillIQ's Alerts evaluator, run every 15 seconds on the example's feed, raises the alarm at 9.0 minutes with 4.5 bbl in. The small difference is the sensor's ±0.3 bbl ripple.

Pit volume through the transfer

The transfer starts at minute 0 and runs for 10 minutes. The fixed rule catches it at minute 9 and, because the pit stays 5 bbl up, keeps re-alarming every 5 minutes, its cooldown, until the condition clears. It fires once more at the next connection, when the usual 2 bbl drain-back lands on top of the transfer. After that connection the limit is re-set on the higher volume and the rule goes quiet.

The automatic limit never comes close. It is the 99th percentile of the channel's last 5,000 samples, about 7 hours, and the pit has been falling all that time as the hole took mud. At the start of the transfer the limit sits at 401.1 bbl, 16 bbl above the pit. On this channel it raised ten alarms in the window: one at start-up, and nine for the pit being lower than its own past, which is what a pit does while drilling.

A fixed limit that is never re-set fails the other way. Set at 423.1 bbl at the start of the shift, it ended the window 40 bbl above the pit and never fired. Either way, a pit alarm that does not follow the hole volume drilled is not watching for a gain.

In this example the gain was a transfer the derrickman forgot to announce. The rig only learns that after the flow check. The right response to the alarm is the same either way: pick up, stop the pumps, flow-check, and call the toolpusher and the company representative, then look for the cause.

Alarm rules that earn their place

The two rule sets watched the same four channels: standpipe pressure, hook load, torque and pit. Set A left the Log Viewer's channel alarms on their automatic limits. Set B is four rules an engineer sets for this well from its plan: SPP above 3,150 psi for a pack-off, hook load above 115 klbf for overpull, torque above 6.5 kft·lbf, and the pit rule re-set at each connection.

Set A raised 104 alarms in 21 hours, about five an hour, every one at critical severity, and none for the transfer. SPP raised 46, torque 31, hook load 17 and the pit 10. Sixty-two came mid-stand, while drilling, on ordinary sensor scatter; 27 at or just after a connection, where the pumps and rotary stop; the rest before and after the drilling. A limit at the 99th percentile is, by definition, crossed by 1 % of the samples. Set B raised five, all for the transfer, and was silent otherwise. Nothing else happened in this window: no pack-off, no overpull, no torque event. Silence was the right answer.

ANSI/ISA-18.2 frames the test well: an alarm needs an abnormal condition, a defined response and time to act. A limit that fires on normal drilling teaches the crew to acknowledge without looking. Build each rule from the well's own numbers, check it against a day of recorded data before it goes live, and remove any rule that nobody acts on.

The same thinking applies to severity. Keep critical for the conditions where the first action is immediate (a pit gain, overpull past the margin, a loss of returns) and use warning for trends that need a look. For how statistical bands, rules and physics models compare at catching a developing pack-off, see predictive drilling analytics.

Data freshness: a silent feed looks calm

During bottoms-up the example's feed stopped for 6 minutes. The alarm evaluator kept running every 15 seconds and read the same pit value, 383.05 bbl, 24 times. A gain in those 6 minutes would have shown nothing. Only the age of the newest sample, which reached 360 seconds, said the screen was blind.

That is why DrillIQ's Rig Watch marks every reading fresh up to 8 seconds after it was measured, stale up to 30 seconds and dead after that. When a reading goes dead while the rig is working, treat it as an instrument failure on the rig floor: tell the driller, put a person on the pits and the flow-line, and call the data provider. How late a healthy feed runs, and where the delay comes from, is worked through in WITSML explained.

When to call

A monitoring screen supports the driller; it does not replace the driller's own flow checks and weight readings. Agree the calls before the well starts, write them on the rig's standing instructions, and make the first action something the driller does without asking.

A checklist for the rig team

• Read every channel against the rig's state: drilling, connection, tripping, static.
• Know the baseline for each warning sign: model hook loads, circulating pressure, hole volume per stand.
• Re-set the pit alarm at every connection, a margin above the drain-back the connections show.
• Set alarm limits from the well's own numbers: plan, model and hole volume.
• Keep critical for alarms whose first action is immediate; check a rule set on a recorded day first.
• Watch the age of the newest sample; a stale feed hides gains and alarms alike.
• Write the first action and the person to call for each sign before the well starts.

How DrillIQ handles it

DrillIQ's Rig Watch shows each channel as a gauge or a metric card, with caution and alarm zones from the channel's range and a fresh, stale or dead marker on every reading; tanks, gauges and cards are set per rig. The Log Viewer draws any channel on time or depth tracks, live as samples arrive, with saved layouts, shared templates, annotations and per-channel warning and critical limits, automatic or set by hand. Alerts evaluates rules on any live channel every 15 seconds, with severity, cooldown, a cap per hour, automatic resolve once the condition has stayed clear, and escalation by email after a set delay; alarms are acknowledged, snoozed or resolved on screen and go out by app, email, SMS or webhook. The platform reads the rig's data and never sends commands to the rig. Everything above was worked out with the same engines.

Book a demo → See all capabilities

References

1. ANSI/ISA-18.2-2016. Management of Alarm Systems for the Process Industries. International Society of Automation, 2016.
2. EEMUA Publication 191. Alarm Systems: A Guide to Design, Management and Procurement, 3rd ed., 2013.
3. Grace, R.D. Blowout and Well Control Handbook, 2nd ed. Gulf Professional Publishing, 2017.
4. Johancsik, C.A., Friesen, D.B. and Dawson, R. Torque and Drag in Directional Wells: Prediction and Measurement. SPE 11380, 1984.

About this example. The well in this article is a representative scenario set in typical North Oman conditions and worked through with DrillIQ's engines. It is not a client's well, and its figures are not field results.