Central Alberta Leak Detection 2023-2024

Fiber Optic DAS Implementation - Alberta Energy Corridor

Challenge

A major transmission pipeline operator managing 2,400 kilometers of NPS 36 crude oil pipeline across central Alberta faced increasing regulatory pressure to enhance leak detection capabilities. Traditional computational pipeline monitoring (CPM) provided adequate detection for significant leaks but lacked sensitivity for small releases. The pipeline traversed environmentally sensitive areas including river crossings, wetlands, and proximity to populated communities requiring rapid detection and response.

Existing aerial patrol programs conducted quarterly flights but could not provide continuous monitoring. The operator sought a technology solution providing:

  • Continuous monitoring along entire pipeline length
  • Detection of leaks below CPM sensitivity threshold (< 1% flow rate)
  • Rapid alerting (under 2 minutes) enabling immediate response
  • Integration with existing SCADA and emergency response systems
  • Minimal operational disruption during installation

Solution Implementation

Following comprehensive technology evaluation, the operator selected distributed acoustic sensing (DAS) based on fiber optic cables. The implementation proceeded in phases:

Phase 1: Pilot Installation (150 km segment, 4 months)

Fiber optic cables were installed along a representative pipeline segment including multiple terrain types, two major river crossings, and sections through agricultural and forested areas. Cables were attached to pipeline coating using specialized clamps at 2-meter intervals. Interrogation units at segment endpoints transmitted laser pulses analyzing backscattered light for acoustic signatures.

Phase 2: Algorithm Training (6 months)

Machine learning algorithms were trained using six months of operational data to establish baseline acoustic signatures for normal operations. The training dataset included pump starts/stops, pressure transients, pigging operations, and environmental conditions (weather, wildlife, third-party construction). Controlled release tests validated leak detection sensitivity and location accuracy.

Phase 3: System Integration (3 months)

DAS monitoring platform integrated with existing SCADA system and emergency response protocols. Operators received training on system interpretation, alarm response procedures, and verification techniques. Mobile applications provided field personnel with real-time DAS data and alert notifications.

Phase 4: Full Deployment (18 months)

Following successful pilot validation, fiber optic installation expanded across remaining pipeline segments. Installation progressed at approximately 15 kilometers per week using specialized crews minimizing disruption to pipeline operations.

Results and Outcomes

The DAS system achieved operational status in March 2024. Performance over the first operating year demonstrated significant improvements:

90 sec
Average Detection Time

Compared to 4.5 hours with previous methods

3
Small Leaks Detected

Averaging 45 L/hr, below CPM threshold

67%
Reduction in False Alarms

Decreased unnecessary field investigations

±50m
Location Accuracy

Enabled rapid crew deployment to incident sites

Incident Example: August 2024 Detection

On August 17, 2024, the DAS system detected acoustic anomalies at kilometer marker 1,247.3 at 03:42 local time. Control room operators received automated alerts within 45 seconds. SCADA data showed no significant pressure changes indicating a small leak below CPM detection threshold. Emergency response crews deployed to the location within 35 minutes, discovering a 3mm crack in a girth weld releasing approximately 60 liters per hour. The pipeline segment was isolated within 90 minutes of initial detection. Total product release was estimated at 180 liters. Without DAS detection, the leak would likely have continued undetected until the next quarterly aerial patrol scheduled three weeks later, potentially releasing 70,000+ liters.

Lessons Learned

  • Algorithm Training Critical: Initial false alarm rates were elevated until ML algorithms completed training on site-specific conditions. Operators should plan for 6-9 month training period with expected higher false positive rates.
  • Operator Training Essential: Effective response requires operators understanding DAS acoustic signatures and verification procedures. Comprehensive training programs prevent both missed detections and unnecessary field mobilizations.
  • Integration Complexity: SCADA integration required careful planning to ensure DAS alerts appropriately prioritized within existing alarm management workflows without causing operator overload.
  • Environmental Factors: Heavy equipment operation near pipelines, severe weather events, and wildlife activity can generate acoustic signatures requiring algorithm refinement to maintain low false alarm rates.
Southern Manitoba Predictive Maintenance 2022-2025

Predictive Maintenance Program - Manitoba Distribution Network

Challenge

A regional natural gas distribution utility serving 75,000 customers across southern Manitoba operated 3,200 kilometers of distribution pipeline installed between 1975 and 2010. The aging infrastructure faced increasing maintenance costs, particularly during harsh winter months when frozen ground and extreme temperatures stressed pipeline systems.

The utility's time-based preventive maintenance program resulted in:

  • Escalating maintenance costs averaging 12% annual increase over five years
  • Unplanned service disruptions affecting customer satisfaction and regulatory metrics
  • Inefficient resource allocation with maintenance crews addressing scheduled work while urgent issues arose elsewhere
  • Limited visibility into asset condition between inspection intervals
  • Difficulty justifying capital replacement programs to regulatory commission

Solution Implementation

The utility partnered with a technology provider to implement comprehensive predictive maintenance program combining inline inspection data, cathodic protection monitoring, SCADA analytics, and environmental factors.

Data Integration Platform

Cloud-based data warehouse consolidated information from diverse sources including SCADA systems (pressure, flow, temperature across 450 monitoring points), inline inspection results from MFL and UT tools, cathodic protection surveys, soil analysis data, weather stations, and historical maintenance records spanning 30 years. Data cleansing and normalization processes addressed inconsistencies in legacy records.

Predictive Models Development

Analytics team developed multiple predictive models addressing different asset classes and failure modes:

  • Corrosion Progression Model: Combined ILI metal loss measurements with soil conditions, CP effectiveness, and pipe age to forecast corrosion rates and predict when wall thickness would reach minimum acceptable levels.
  • Coating Degradation Model: Analyzed CP current requirements and coating resistance measurements to estimate coating condition and forecast protection system maintenance needs.
  • Mechanical Damage Risk Model: Incorporated construction activity databases, frost depth measurements, and ground movement sensors to identify segments at risk from external forces.
  • Valve Failure Prediction: Analyzed valve cycling frequency, operating conditions, and historical failure data to predict optimal inspection and replacement timing.

Decision Support Dashboard

Interactive dashboard provided maintenance planners with asset health scores, risk rankings, and optimized maintenance schedules. Work order generation system automatically created maintenance tasks with technical specifications, parts requirements, and crew scheduling recommendations.

Results and Outcomes

Three-year program implementation (2022-2025) delivered measurable improvements across multiple performance metrics:

38%
Maintenance Cost Reduction

From CAD $8.7M to $5.4M annually

54%
Fewer Emergency Responses

Winter emergency calls decreased from 187 to 86

127
High-Risk Segments Identified

Requiring priority intervention

92%
Customer Satisfaction

Up from 79% before program

Financial Impact

Total program implementation costs including software licensing, data integration, model development, and training amounted to CAD $2.1M. Annual operating costs (software, data management, analyst time) total $350K. Three-year cumulative savings of $9.9M yielded return on investment of 373% with payback period of 8 months.

Lessons Learned

  • Data Quality Foundation: Six months spent cleaning and normalizing historical data proved essential. Poor data quality delays model development and undermines prediction accuracy.
  • Change Management: Transitioning from time-based to condition-based maintenance required cultural change. Comprehensive communication and training programs addressed field crew skepticism about algorithm recommendations.
  • Regulatory Engagement: Early engagement with provincial regulator demonstrated program rigor, securing approval for revised maintenance intervals and supporting rate applications.
  • Phased Implementation: Starting with pilot assets and expanding after demonstrating success built organizational confidence and allowed refinement of processes before enterprise rollout.
  • Continuous Improvement: Establishing feedback loops capturing maintenance outcomes and incorporating results into model refinement ensures ongoing accuracy improvement.

Explore More Solutions

Learn about the technologies and approaches behind successful implementations