Magnetic Flux Leakage: What Pipeline Inspection Buyers Need To Know

What It Detects

Magnetic flux leakage (MFL) detects wall thinning, corrosion loss, and metal loss defects in ferromagnetic pipelines. It is optimized for volumetric loss quantification in carbon steel and low-alloy steel lines. MFL does not reliably detect cracking, delamination, or coating defects; it cannot assess weld integrity or stress-corrosion cracking initiation.

Defect classes MFL resolves:

• General corrosion (uniform wall loss)
• Pitting corrosion (localized metal loss)
• Erosion-corrosion (flow-accelerated thinning)
• Mechanical damage (dents with wall loss)
• Seam and girth weld corrosion (external and internal surfaces)

Defect classes MFL does NOT resolve:

• Stress-corrosion cracking (SCC) or hydrogen-induced cracking (HIC)
• Coating disbondment or external coating failure
• Laminations or delamination in welds
• Austenitic stainless steel (non-ferromagnetic)
• Duplex stainless steel (weak ferromagnetic response)

Asset scope: Carbon steel transmission pipelines (onshore and offshore), gathering lines, product lines, and distribution mains. Typical operating range: 2–48 inches (50–1,200 mm) diameter; wall thickness 0.188–0.750 inches (4.8–19 mm).

---

How It Works

MFL magnetizes the pipeline wall to saturation using permanent magnets or electromagnets mounted on an inspection tool (pig). When the magnetic field encounters a metal loss defect, flux lines leak out of the pipe wall into the air gap above the defect. Sensor arrays (typically Hall-effect or magnetoresistive sensors) detect this leakage field and map its amplitude and spatial distribution.

Workflow:

1. Magnetization: Tool magnets (0.5–2.0 Tesla) saturate the pipe wall along the tool's travel path. Saturation ensures consistent baseline signal and reduces tool-speed sensitivity.

2. Flux leakage detection: Sensor arrays positioned 0.5–2 mm above the inner or outer wall surface record leakage field strength as the tool moves through the pipeline at 0.5–1.5 m/s.

3. Signal conditioning: Raw sensor data is amplified, filtered (typically 10–100 Hz bandpass), and logged with GPS/odometry timestamps.

4. Data processing: Defect signals are extracted using threshold algorithms. Amplitude and spatial extent are correlated to wall loss depth and area using calibration curves derived from pit geometry models.

5. Depth estimation: Defect depth is inferred from leakage field amplitude using empirical or physics-based models. Typical depth resolution: ±10–15% of wall thickness or ±0.5 mm, whichever is larger.

6. Reporting: Defects are classified by severity (% wall loss), location (distance along pipe), and circumferential position. Reportable thresholds typically trigger at 10–20% wall loss, depending on pipeline code (ASME B31.8, DNV-RP-F101).

Physics constraint: MFL signal amplitude depends on defect geometry (depth, width, length), magnetic permeability of the steel, and sensor-to-wall standoff distance. Shallow defects (<10% wall loss) and very deep defects (>80% wall loss) produce weaker signals and higher depth-estimation uncertainty.

---

Robotic Deployment Pattern

MFL inspection has transitioned from tethered pigging (human-launched, human-retrieved) to autonomous robotic platforms. Deployment patterns differ by pipeline access and operational constraints.

Tethered Robotic Pig (Emerging)

• Status: Pilot deployments in North America and Europe; not yet standard practice.
• Mechanism: Fiber-optic or electrical tether provides real-time data telemetry, power, and retrieval capability. Operator controls tool speed and can halt inspection if anomalies are detected.
• Advantage: Eliminates need for launcher/receiver stations; enables mid-line data review and adaptive speed adjustment.
• Constraint: Tether drag limits deployment to pipelines <50 km; requires access points (valve stations, compressor stations) every 30–50 km.
• Deployment status: Proven in lab and controlled field trials (e.g., Rosen, GE Robotics, T.D. Williamson). Commercial adoption remains <5% of annual MFL inspections globally.

Autonomous Pigging (Standard)

• Status: Dominant deployment model; 95%+ of commercial MFL inspections use autonomous pigs.
• Mechanism: Tool is launched from a launcher station (pig trap), travels through pipeline under pressure differential or gravity, and is retrieved at a receiver station. Data is logged internally on solid-state storage and downloaded post-retrieval.
• Advantage: No tether; operates in live pipelines under normal operating pressure; covers long distances (100+ km per run).
• Constraint: Requires launcher and receiver infrastructure; cannot halt or redirect mid-run; data review occurs post-retrieval (24–72 hours after inspection).
• Deployment status: Industry standard since 2005. Over 500,000 km of pipeline inspected annually via autonomous MFL pigs.

Robotic Crawler (Emerging)

• Status: Prototype and pilot phase; not yet commercial for MFL.
• Mechanism: Magnetic-wheel or adhesion-based crawler traverses external pipe surface, carrying MFL sensor head. Tethered or wireless telemetry enables real-time data streaming.
• Advantage: Inspects external corrosion, coating condition, and mechanical damage without pipeline entry; can inspect non-piggable lines (small diameter, high-pressure, complex geometry).
• Constraint: Slow traverse speed (0.1–0.3 m/s vs. 0.5–1.5 m/s for pigs); limited to accessible external surfaces; weather-dependent (rain, ice, mud degrade magnetic adhesion).
• Deployment status: Prototype systems demonstrated by Inuktun, GE Robotics, and Rosen in 2023–2024. No commercial contracts reported. Confidence: LOW.

Drone-Mounted Sensor (Concept)

• Status: Concept only; not operationally deployed.
• Mechanism: Aerial drone carries MFL sensor array and hovers above pipeline to detect external flux leakage.
• Constraint: Standoff distance (>10 cm) severely degrades signal-to-noise ratio; cannot saturate ferromagnetic pipe wall from air; weather sensitivity; regulatory barriers (airspace, safety).
• Deployment status: No credible field trials reported. Confidence: LOW.

Robotic deployment implication: MFL remains a tethered or autonomous pigging technology. Robotic innovation is concentrated on tether systems (real-time data, mid-line control) and external crawlers (non-piggable lines). Drone deployment is not viable for MFL.

---

Where It Fails

1. Defect Type Blindness

MFL detects metal loss but cannot distinguish cracking, lamination, or stress-corrosion cracking from corrosion pitting. A stress-corrosion crack with minimal wall loss will produce no MFL signal. A corroded pit with 20% wall loss will trigger a reportable defect. Implication: MFL must be paired with ultrasonic testing (UT) or eddy current (EC) in high-risk environments (sour gas, high-pressure hydrogen, caustic service) where SCC is a failure mode.

2. Depth Estimation Uncertainty

MFL depth accuracy degrades with defect geometry. Narrow, deep slots produce weaker signals than broad, shallow pits of equivalent volume. Empirical calibration curves assume pit-like geometry; real-world defects (corrosion under insulation, erosion-corrosion, weld corrosion) often deviate from pit models. Typical depth error: ±10–15% of wall thickness. For a 10 mm wall, this is ±1–1.5 mm. For a 5 mm wall, this is ±0.5–0.75 mm. Implication: MFL cannot reliably distinguish 15% wall loss from 25% wall loss; operators must apply conservative repair thresholds (e.g., repair at 20% loss rather than 30%).

3. Surface Condition Sensitivity

MFL requires close sensor-to-wall standoff (0.5–2 mm). Internal scale, wax, or paraffin deposits increase standoff and degrade signal. External rust, mill scale, or coatings do not block MFL (tool operates inside pipe) but internal deposits do. Implication: Pipelines with heavy internal scaling (e.g., CO₂ lines, water injection lines) require pre-inspection pigging (scraper pig) to remove deposits. This adds 1–2 weeks and $50,000–$200,000 to project cost.

4. Magnetic Saturation Variability

MFL assumes the pipe wall is magnetically saturated. Variations in steel composition (carbon content, manganese, nickel) and prior thermal history (welding, stress relief) create local permeability variations. High-permeability zones produce stronger leakage signals; low-permeability zones produce weaker signals. Implication: Defects in low-permeability zones may be underestimated or missed. Girth welds (heat-affected zone) often exhibit lower permeability and produce weaker MFL signals than base metal.

5. Weld Defect Detection Failure

MFL is poor at detecting weld defects (lack of fusion, porosity, inclusions) because these defects do not produce significant metal loss. A weld with 5% porosity by volume may have <1% wall loss and produce no reportable MFL signal. Implication: MFL cannot be used as a primary weld inspection method. Phased-array ultrasonic testing (PAUT) is required for weld integrity assessment.

6. Non-Ferromagnetic Material Blindness

MFL does not work on austenitic stainless steel, duplex stainless steel (weak response), aluminum, copper, or composite pipelines. Implication: Operators must confirm ferromagnetic material before contracting MFL inspection. Material verification requires mill certificates or magnetic permeability testing.

7. Launcher/Receiver Infrastructure Requirement

Autonomous MFL pigs require launcher and receiver stations (pig traps). Pipelines without traps must have them installed, adding cost ($50,000–$150,000 per station) and downtime (2–4 weeks). Implication: MFL is economically viable only for pipelines with existing trap infrastructure or for high-value lines where trap installation is justified.

8. Speed-Sensitivity Trade-off

MFL signal amplitude decreases with tool speed. Slower speeds (0.5 m/s) produce higher-quality data but require longer inspection windows. Faster speeds (1.5 m/s) reduce inspection time but increase depth-estimation error. Implication: Operators must balance data quality against operational downtime. Typical compromise: 1.0 m/s, accepting ±15% depth error.

9. Circumferential Resolution Limitation

MFL sensor arrays typically have 4–8 circumferential channels. Defects smaller than the sensor spacing (e.g., <30° arc) may be missed or underestimated. Implication: Small, localized defects (e.g., isolated pitting in a narrow band) may not be detected if they fall between sensor channels.

10. Data Interpretation Variability

MFL data processing relies on threshold algorithms and empirical calibration curves. Different vendors' algorithms produce different defect counts and depth estimates from the same raw data. Industry studies show ±20–30% variation in defect reporting between vendors. Implication: Operators should specify algorithm transparency and request independent validation of critical defects (e.g., >30% wall loss).

---

Buyer Checklist

Requirement	Why it matters	Evidence to request
Ferromagnetic material confirmation	MFL does not work on austenitic stainless, duplex, or non-ferrous metals. Misidentification wastes inspection budget.	Mill certificates, permeability test results (target: >100 µ_r for carbon steel).
Launcher/receiver station availability	Autonomous pigs require trap infrastructure. Absence adds $50K–$150K and 2–4 weeks to project.	Site survey confirming trap locations, dimensions, and operational status. Pressure rating must match pipeline design.
Internal surface condition (scale/deposits)	Heavy internal scaling increases sensor standoff and degrades signal. Pre-pigging (scraper) may be required.	Baseline pig run data or visual inspection via borescope. Estimate deposit thickness and composition.
Wall thickness and diameter range	MFL tools are sized for specific diameter and wall thickness ranges. Mismatch results in tool sticking or poor sensor contact.	Design specifications: OD, wall thickness, material grade. Confirm tool availability for your range.
Defect depth accuracy requirement	MFL depth error is ±10–15% of wall thickness. If your repair threshold is <10% wall loss, MFL alone is insufficient.	Define repair criteria (% wall loss or absolute depth). If <10% loss is reportable, pair MFL with UT.
Weld inspection scope	MFL cannot reliably detect weld defects (lack of fusion, porosity). If weld integrity is critical, specify phased-array UT.	Identify critical welds (girth welds, branch connections). Specify acceptance criteria per ASME B31.8 or DNV-RP-F101.
Stress-corrosion cracking risk	MFL does not detect SCC. In sour gas, high-pressure hydrogen, or caustic service, SCC is a failure mode. Pair MFL with EC or UT.	Service history: H₂S content, pH, operating pressure, temperature. Identify SCC-susceptible zones.
Inspection window and downtime tolerance	Autonomous MFL requires pipeline depressurization and 24–72 hour retrieval window. Tethered systems enable real-time data but limit distance.	Operational schedule: acceptable downtime window, maximum inspection distance per run, pressure drop tolerance.
Data quality and vendor algorithm transparency	MFL data processing varies by vendor (±20–30% defect count variation). Specify algorithm details and validation.	Request algorithm documentation, calibration curves, and independent validation of >30% wall loss defects.
Reporting format and integration	MFL data must integrate with pipeline integrity management systems (PIMS). Specify output format (ASCII, GIS, proprietary).	Define data deliverables: defect maps, depth profiles, risk rankings, GIS shapefiles. Confirm compatibility with your PIMS.
Calibration and traceability	MFL tools must be calibrated against certified standards. Calibration drift can introduce systematic depth errors.	Request calibration certificates (ISO 17025 accredited lab), calibration date, and next calibration due date.
Corrosion rate trending	MFL enables multi-year defect tracking if baseline data exists. Corrosion rate trends inform repair prioritization.	Confirm availability of prior MFL data (year, vendor, defect locations). Plan for repeat inspection in 3–5 years.

---

Method Comparison

Method	Best defect class	Access requirement	Data output	Main limitation	Depth accuracy
Magnetic Flux Leakage (MFL)	General corrosion, pitting, wall thinning	Launcher/receiver stations or tethered entry point	Defect location, depth (%), circumferential position	Cannot detect cracking, SCC, or weld defects; depth error ±10–15% wall thickness	±10–15% wall thickness
Ultrasonic Testing (UT)	Wall thinning, cracking, weld defects, lamination	Accessible external surface or internal access point	Defect location, depth (absolute mm), geometry	Requires surface contact; slow coverage rate (~100 m²/day); operator-dependent; cannot assess defect severity (SCC vs. corrosion)	±0.5–1.0 mm absolute
Phased-Array UT (PAUT)	Weld defects, cracking, lamination, wall thinning	Accessible external surface or internal access point	Defect location, depth, geometry, 2D/3D imaging	Requires surface contact; high equipment cost ($150K–$300K); operator skill-dependent; slow coverage rate	±0.5–1.0 mm absolute
Eddy Current (EC)	Surface corrosion, coating disbondment, shallow pitting (<2 mm)	Accessible external surface	Defect location, depth (relative), conductivity change	Cannot detect deep wall loss; limited to ferromagnetic materials (carbon steel only); slow coverage rate; requires surface contact	±0.5–1.0 mm (shallow defects only)
Radiography (X-ray, Gamma)	Weld defects, porosity, inclusions, lamination	Accessible external surface; requires radiation safety setup	Defect location, geometry, 2D image	High cost ($500–$2,000 per image); radiation safety overhead; cannot quantify depth; slow coverage rate; regulatory barriers	Qualitative only (visual classification)
Thermography (IR)	Coating disbondment, delamination, subsurface corrosion	Accessible external surface; requires thermal gradient	Defect location, thermal signature	Requires temperature differential; weather-dependent; cannot quantify depth; limited to surface/near-surface defects	Qualitative only
Acoustic Emission (AE)	Active cracking, stress-corrosion cracking	Accessible external surface; requires quiet environment	Defect location (triangulation), event count, frequency signature	Detects only active defects; cannot assess existing defect severity; high false-positive rate in noisy environments	Qualitative (event detection only)

---

Deployment Status and Confidence Assessment

MFL Pigging (Autonomous): HIGH CONFIDENCE. Autonomous MFL pigging is the industry standard for carbon steel pipeline inspection. Over 500,000 km inspected annually. Depth accuracy ±10–15% wall thickness is well-documented across multiple vendors (Rosen, GE Robotics, T.D. Williamson, Applus RTD).

MFL Pigging (Tethered Robotic): MODERATE CONFIDENCE. Tethered systems are in pilot deployment (Rosen, GE Robotics, T.D. Williamson). Real-time data and mid-line control are proven concepts, but commercial adoption remains <5% of annual MFL inspections. Cost premium (20–40% above autonomous) and tether drag limitations restrict deployment to <50 km lines.

MFL Crawler (External): LOW CONFIDENCE. Prototype systems exist (Inuktun, GE Robotics, Rosen), but no commercial contracts reported as of 2026. Magnetic adhesion degradation in wet/muddy conditions and slow traverse speed (0.1–0.3 m/s) remain unresolved. Deployment status: research phase.

MFL Drone (Aerial): LOW CONFIDENCE. No credible field trials reported. Standoff distance and inability to saturate pipe wall from air make this concept impractical. Regulatory barriers (airspace, safety) add risk. Deployment status: concept only.

---

Review Trigger

Standard: ASME B31.8 (Gas Transmission and Distribution Piping Systems), DNV-RP-F101 (Submarine Pipeline Systems), API 1130 (Integrity Management of Hazardous Liquid Pipelines).

Regulation: U.S. Pipeline and Hazardous Materials Safety Administration (PHMSA) requires in-line inspection (ILI) of high-consequence areas (HCAs) every 5 years or per risk-based intervals. MFL is the primary ILI method for carbon steel transmission lines.

Technology Catalyst: Tethered robotic MFL systems (real-time data, mid-line control) are entering commercial deployment in 2025–2026. Operators should evaluate tethered systems for non-piggable lines and high-risk segments where data quality justifies cost premium.

---

Confidence: HIGH (autonomous MFL pigging) | MODERATE (tethered systems) | LOW (crawler, drone)

Review Trigger: ASME B31.8, DNV-RP-F101, API 1130, PHMSA HCA rules

---