← BACK TO WORK

OPM Pipeline Corrosion Detection

A proposed quantum-sensing framework that uses optically pumped magnetometers (OPMs) to detect hidden corrosion in South Africa's buried steel water pipes before it turns into a leak.

STATUS

Research & System Design, Emerging Technology

ROLE

Researcher & System Designer

TIMELINE

2026

Quantum Sensing   ·   OPM Magnetometry   ·   Signal Processing   ·   Feature Extraction   ·   Machine Learning   ·   System Design

§01Purpose / Problem

South Africa's municipal non-revenue water rose from 37% in 2014 to about 47% in 2026, and part of that loss comes from ageing underground pipes. Corrosion weakens buried steel pipes long before a leak or burst shows at the surface, and current inspection and failure-prediction methods struggle to catch it early. Rand Water already uses the nanotesla-scale Magnetic Memory Method to find corrosion zones. The open question is whether a far more sensitive sensor could find them sooner.

§02What I Designed

I researched OPM quantum magnetometry, which uses laser light and rubidium-87 atoms to measure magnetic fields at femtotesla scale, and traced how it grew out of conventional and magnetic pipeline inspection. I then designed a six-layer framework that turns raw OPM readings taken above ground into a corrosion risk map, and compared it with the Magnetic Memory Method on sensitivity, stand-off distance, cost and IT integration. The framework is built to support existing condition monitoring, not replace it.

§03The Framework

OPM sensors above ground measure the magnetic field around a buried steel pipe. Corrosion changes the pipe's magnetic behaviour, so the framework looks for that change, filters out everything else, and flags the section for engineers to inspect.

OPM-based water-pipeline corrosion detection setup An OPM sensor above ground measures the magnetic field of a buried steel water pipe 0.1 to 1 metre below. Corrosion on the pipe changes its magnetic field, which shows up as a peak in the measured field gradient directly above the corroded section. MEASURED FIELD GRADIENT ANOMALY POSITION ALONG PIPE → GROUND LEVEL OPM SENSOR Rb-87 vapour cell STAND-OFF 0.1 – 1 m BURIED STEEL WATER PIPE CORROSION SHIFTS THE FIELD magnetic signal from the pipe corrosion area
Sensing setup: the anomaly in the measured field gradient lines up with the corroded section of pipe.

INPUTS

  • Magnetic field measurements from OPM sensors
  • Sensor position and orientation
  • Pipe depth and location
  • Pipe material and dimensions
  • Background environmental magnetic readings
  • Known corrosion conditions, for validation

Simulated measurements can stand in during prototyping if OPM hardware isn't available.

  1. 01

    QUANTUM SENSING

    OPM Sensing

    One or more OPM sensors outside the pipe measure its magnetic field. A multi-sensor array adds spatial detail by comparing measurement points.

  2. 02

    QUANTUM SENSING

    Data Acquisition

    Records every field measurement together with the sensor's position and the measurement conditions.

  3. 03

    SIGNAL PROCESSING

    Noise Reduction & Preprocessing

    Separates the pipe's signal from environmental magnetic interference such as power lines, vehicles and the Earth's field.

    Baseline correction   ·   Filtering   ·   Sensor calibration   ·   Magnetic gradient calculation

  4. 04

    SIGNAL PROCESSING

    Feature Extraction

    Turns the cleaned signal into measurable features that can be compared across pipe conditions.

    Field magnitude   ·   Field gradient   ·   Spatial variation   ·   Signal amplitude   ·   Anomaly location

  5. 05

    DETECTION & DECISION SUPPORT

    Corrosion Detection

    Decides whether the features point to corrosion, possibly with a machine-learning model. Final classes are set once experimental data exists.

    Normal   ·   Potential deterioration   ·   Probable corrosion   ·   Higher-risk corrosion

  6. 06

    DETECTION & DECISION SUPPORT

    Visualisation & Risk Output

    Presents the results so maintenance teams can act on them.

    Pipe and sensor location   ·   Anomaly measurement   ·   Estimated condition   ·   Areas for further inspection

OUTPUTS

  • Processed magnetic field data
  • Magnetic anomaly maps
  • Detected corrosion signatures
  • Corrosion condition class
  • Confidence values
  • Locations needing further inspection

§04Evaluation Plan

Measured on steel pipe samples with known corrosion against samples without it, under controlled conditions. Targets are starting points to be tested, not assumptions.

  • ≥ 80% DETECTION SENSITIVITY Corroded pipes correctly detected: TP ÷ (TP + FN).
  • ≥ 80% SPECIFICITY Healthy pipes correctly left alone: TN ÷ (TN + FP).
  • ≥ 80% CLASSIFICATION ACCURACY Test cases classified correctly, given enough labelled data.
  • < 20% FALSE-POSITIVE RATE Healthy pipes wrongly flagged, which means unnecessary excavation.
  • 0.1–1 m DETECTION DISTANCE Sensor-to-pipe stand-off tested to find the maximum effective range.
  • 0.3 · 0.6 · 1.0 m BURIAL DEPTH Simulated depths, to see how depth weakens the signal.
  • SNR SIGNAL-TO-NOISE RATIO Measured at each distance and noise level. The detection threshold is tested, not assumed.
  • Repeat REPEATABILITY Repeated scans of the same pipe under the same conditions must agree.

§05Outcome / Impact

A complete research proposal: literature survey, problem statement, framework design, an evaluation plan with measurable targets, and a comparison with the conventional alternative. It also sets out ethical safeguards. Anomaly maps of critical infrastructure must be access-controlled, rollout must not leave under-resourced municipalities behind, and engineers, not the model, make the final maintenance call.

§06What I Learned

Sensor sensitivity alone doesn't solve the problem. A femtotesla sensor still has to separate a weak pipe signal from the Earth's field and urban noise, so most of the value sits in the software around it: signal processing, feature extraction, honest error rates and human oversight. It also taught me to separate a technology's potential from its practical readiness.