The MFL Pipeline Inspection Process, Step by Step

Magnetic flux leakage surveys look simple from outside — a tool goes in, data comes out. Inside, the process has six stages, and the quality of the final report is decided in the first two. This walkthrough follows the process for the MFL tool fleet and the inspection service built on it.

Stage 1: Line data review

Before any hardware moves, the line is audited on paper: diameters (the fleet covers 159–1,422 mm), wall thicknesses, bend radii against the 1.5D minimum, deformation limits (15–20% OD depending on tool), valve and tee types, launcher and receiver condition, and operating envelope against the tool's 0–10 MPa and -10 to 80 degrees C ratings. Variable-diameter lines are flagged for the multi-calibre adaptive tools. Most failed runs are visible in this paperwork and simply not read.

Stage 2: Cleaning and gauging

Debris is the enemy of data. Cleaning pigs run first, then a gauging/caliper run proves the bore and measures deformation before the expensive sensor tool is committed. Skipping the gauge run to save a day is how tools get stuck and surveys get repeated.

Stage 3: The MFL run

The magnetiser tool is launched and carried by product flow, magnetising the wall as it travels and recording leakage fields at anomalies. Published continuous working time is more than 60 hours, which sets the maximum section length per run. Speed control matters: sensor quality degrades outside the tool's velocity band, so flow management during the run is part of the service, not an operational detail.

Stage 4: Data analysis

Analysts convert leakage signals into a feature list: metal loss (internal and external corrosion, mechanical damage), cracks and weld anomalies, and pipeline features — girth welds, bends, lengths, valves, tees. On the service side, multi-sensor fusion and adaptive filtering give a signal recognition accuracy of +/-5 cm, and features are referenced to the nearest circumferential weld within +/-0.1 m and +/-5 degrees circumferentially. Automated report generation cuts the documentation time by a stated 90% — but the classification judgement stays human. The method's physics, including its blind spots at field-parallel cracks and pinholes, is covered in the MFL testing guide.

Stage 5: Verification digs

No ILI report is believed until it is dug. A sample of reported features is excavated and measured directly; the comparison calibrates the whole report's confidence. Choose dig sites across severity classes, not just the worst calls — a programme that only verifies its scariest defects learns nothing about its quiet ones.

Stage 6: Reporting into the integrity register

The final report feeds the operator's integrity management: defect growth assessment against previous runs, repair scheduling, and re-inspection intervals. Operators running mixed networks pair MFL with the eddy current machine for small-bore lines — the programme guide shows the structure. Run well, the process repeats on a cycle measured in years; run badly, it is an expensive way to move a magnet.

Tool selection for the run

The run plan names the tool against the line's data: single-diameter lines take the standard tool in their size (20+ specifications span 159–1,422 mm); variable-diameter lines take the multi-calibre adaptive configuration; lines with a theft history add the tapping-detection configuration; small-bore laterals are scheduled for the eddy current machine instead. Deformation ability is checked against the gauging run's measurements (15–20% OD across the fleet), and any line that fails the gauge does not get the MFL tool — it gets a cleaning and remediation plan.

What a good report contains

• A feature list with coordinates — every anomaly referenced to the nearest girth weld (+/-0.1 m) and clock position (+/-5 degrees), sized and classified.
• A confidence statement — detection thresholds and the method's known blind spots (field-parallel cracks, pinholes) stated, not buried.
• Dig recommendations — a verification sample across severity classes, with the excavation data sheets the field crews need.
• Comparison with prior runs — growth rates per feature, because a defect's history is its prognosis.
• The raw data — archived so the next run, possibly by another vendor, can be compared honestly.

Re-inspection intervals

The output of one survey sets the date of the next: growth rates from verified digs convert into remaining-life calculations, and the interval follows the fastest credible growth on the line, not a calendar habit. Programmes that run the full process — the programme guide structures it — spend their inspection budget where the pipe is actually thinning.

Crew and HSE on site

An inspection run is a pressure operation. Launching and receiving involve opening pressure vessels on live lines, handling tools that may carry product residue, and — on crude lines — managing what the cleaning pigs bring out. The HSE plan covers pressure isolation and bleed verification, gas testing at the receiver, tool handling and decontamination, and waste disposal for debris and product samples. Crew composition is lean (run supervisor, tool technicians, data engineer, operator's line crew) but every role is named in the plan; ILI accidents are rare precisely because the industry treats these steps as ceremony.

Data ownership and archiving

Agree in the contract who owns the raw run data, in what format it is delivered, and how long the vendor retains it. Raw data is the asset that makes every future run more valuable — growth comparison across runs is how integrity engineers turn pictures into prognoses. An operator who holds only PDF reports holds conclusions without evidence; an operator who holds the raw data can re-analyse with better algorithms in five years and compare across vendors. This clause costs nothing at signing and a great deal later.

Common failure modes of a survey programme

Four recur across the industry. Running the sensor tool without a gauging pass, and learning about a dent from a stuck tool. Letting speed drift outside the tool's band and discovering, in analysis, that the data thins exactly where the line was dirtiest. Verifying only the scariest digs and calibrating nothing. And archiving reports without raw data, which we have already warned against. Every one of these is a process choice, not a technology limit — the six stages at the top of this post exist to make the choices deliberately. Programmes also fail quietly by never re-baselining after repairs: a line whose repairs are not fed back into the integrity register will be re-dug at the same locations by the next survey, at full price, forever.