CUSTOMER CASE STUDY
Detection of anomalies and hazardous
events affecting gas pipelines
- Minimized false alerts to the minimum.
- Reliable AI-powered solution for the online detection of excavator attacks and 3rd party intervention.
- No digging necessary, no downtime.
- Can be installed and removed easily.
- Minimizes maintenance cost.
- Prevent environmental hazards.
- Minimizes personnel health risk.
- Maximize return of damage claims to third parties.
False Alert Reduction
Sensewaves worked together with a world leading gas company to provide an AI-driven, cost-effective detection system for excavator attacks and hazardous events that threaten pipelines’ integrity.
The excavator and other type of attacks is a major problem for gas pipeline companies. It takes a lot of time and effort to detect such attacks and complete repairs. This results in considerable costs, serious safety hazard, downtime of the pipeline and potential damage to the environment. Traditional maintenance and operational approaches through regular inspections, flow and pressure monitoring, helicopter surveillance and forensics are also considered ineffective and costly, as one needs to engage a high level of resources just to detect such problems and usually with a considerable delay.
The company has started initiatives for the online, remote monitoring of the pipelines to enable predictive maintenance and early detection of hazardous events. They have formalised a framework for sensory equipment installation and data acquisition that would allow detection of such events while being cost-effective and not causing any downtime.
allow detection of such events in a way that is reliable, cost efficient and, in the same time, inobtrusive, in the sense that the pipeline would not need to be shut down (the last being the case of installing of any sensors in the interior of the pipe).
The reason for deploying Adaptix was to use AI that can cope with the analysis of fast, noisy and diverse data from the pipelines and produce smart alerts of high quality that ensure a) an excellent success rate in identifying a attack event and b) a fair minimization of false alerts, compared with other machine learning methods already tested within the system.
The ultimate goal of the pilot was to create a reliable, end-to-end solution for the detection of excavator attacks and 3rd party intervention, which can be applied and removed ad hoc and within the shortest time frame, implying zero digging or downtime. The key for such a solution was to benefit from existing pipeline data, such as scada or cathodic protection, and maximize the exploitation of the data through AI.
Sensewaves has used Adaptix.AI together with the Adaptix pipeline dedicated software package applied on CP and pipeline SCADA data.
Among the tools that are deployed:
- Smart tagging: a tool allowing to tag special events and learn offline and incrementally to recognize similar and quasi-similar events in the future.
- Smart alerts: based on self-learning, the system automatically identifies abnormal situations that have been not encountered before and may constitute a potential threat.
TEST BEFORE DEPLOYMENT
In the PoC that was organised, a number of datasets was provided together with numerous excavator attacks data that have been simulated across a number of different pipelines and under diverse weather conditions.
COMPARISON OF ADAPTIX PERFORMANCE WITH OTHER AI/MACHINE LEARNING METHODS
Adaptix was equally reliable if not superior to other methods, keeping false negatives at a minimum.
- FALSE ALERT REDUCTION:
Adaptix reduced false alerts by 45% to 90x.
- SPEED OF DETECTION:
Adaptix response was in sub-second time.
TECHNOLOGY STACK, INTEGRATION AND DEPLOYMENT
Next step is to deploy an Adaptix dedicated instance in the cloud deployment and make available Adaptix functionalities through REST API. A pilot will run in 12 months and will lead to full scale deployment for 250 pipeline segments (around 5000 km of pipeline). The purpose is to train site managers and focus on integrating Adaptix to the pipeline operator software and familiarize pipeline integrity engineers with its workflow.