Summary
This case study focuses on monitoring a long-distance crude oil pipeline to detect leaks in real time. Traditional methods like SCADA and periodic inspections failed to identify small leaks early, leading to environmental damage and product loss. OctaSense implemented an advanced AI-driven system combining acoustic sensing, pressure monitoring, and thermal detection to accurately identify leak signatures. The solution enabled rapid detection, precise location identification, and immediate response, significantly reducing spill volume and environmental risk.
Background & Context
The pipeline traverses three national parks, two river crossings, and a mountainous section with 1,200 m elevation change. Manual patrol is conducted bi-monthly. The previous system — a basic SCADA mass-balance algorithm — had a minimum detectable leak size of 5% throughput and a detection lag of 6–24 hours. Three significant spill events had occurred in 4 years, resulting in $22M in environmental remediation costs.
Leak Physics:
Hydrocarbon leaks generate three detectable physical signatures:
1. Negative Pressure Wave (NPW): A leak event initiates a pressure drop that propagates in both directions from the leak point at the speed of sound in the fluid (~1,200 m/s). Cross-correlating arrival times at two sensor stations localises the leak to within ±150 m.
2. Sustained Pressure Imbalance: A steady-state leak creates a persistent differential between the expected and measured pressure profile — detectable by the hydraulic gradient deviation algorithm.
3. Acoustic Emission: Fluid escaping through a defect generates broadband acoustic energy (2–20 kHz) propagating along the pipe wall and through the soil.
Sensor Deployment
Key Outcomes & Results
Minimum Detectable Leak
0.3% throughput — 16× improvement over SCADA mass-balance
Alert-to-Location Time
Average 5.8 minutes from event to crew dispatch with GPS coordinates
False Positive Rate
1.2 per week — reduced to 0.3/week after model fine-tuning in month 4
Third-Party Interference
47 unauthorised excavation events detected in 24 months
Environmental Savings
Zero reportable spill events in 30 months since deployment
5.8 min
Alert-to-locate time
0.3 %
Min detectable leak
47
Intrusions detected
680 km
Pipeline covered
Deployment Snapshot
Location
East Africa
Operating Pressure
82 bar
Product
Crude oil
Sensor Count
680 km DAS + 104 discrete