Who’s watching when you aren’t?
AI-powered machine vision – enhancing remote monitoring and real-time pipeline leak detection in metals mining
Written by Gurjeet Bansal, Corporate Communications Manager, IntelliView Technologies Inc.
Environmental accountability and safety are longstanding priorities in the business of metals extraction and processing, where effluents such as cyanide for cyanidation in gold mining, sulfuric acid, processed water, and slurry form a crucial part of operations.
Pipelines transport these hazardous fluids to and from various mining facilities based on the stage of the process, including plants, concentrators, tailings ponds, and leach pads. While tailings safety and dam failure are top global concerns, especially in high-incident countries such as China, Canada, US, Mexico, and Brazil, according to the 2017 ‘Mine Tailings Storage: Safety Is No Accident’ report by United Nations Environment Programme, slurry and water pipelines are also susceptible to leakage. Mechanical failure, worker oversight, vandalism, and environmental factors, such as corrosion, landslides, and ground movement, could compromise pipeline integrity.
An undetected pipeline leak has the potential to endanger workers, damage local ecosystems beyond rehabilitation, contaminate drinking water supply, and leave community livelihoods in ruins. The financial liability of serious incidents can amount to millions of dollars from operational interruption, restoration work, and fines. It can also cost a company its reputation and social license to operate. As with the oil and gas industry, such incidents threaten the sustainability of the mining industry, which in the foreseeable future is expected to grow owing to the increasing demand for gold, copper, silver, and other metals.
For an industry widely cognizant of a spill’s detrimental impacts, being proactive in adopting new risk-mitigating and safety-enhancing approaches has become less an option and more an expectation – a standard.
A future with zero catastrophic incidents, zero failures, and zero pollution has been deemed achievable by industry and regulatory experts, and therefore should be endeavored. Technology has a major role to play to meet this challenge, but successful products must overcome hurdles to adoption. These include price-performance-benefit ratio, coverage sufficiency, level of technology disruptiveness, infrastructure interoperability, and ease of use, just to name a few.
Gaps in coverage and adequacy
Industry standard pipeline leak detection techniques come with their own merits and limitations. While statistical volume/flow measurement and pressure monitors, for instance, are designed to provide 24/7 monitoring and near real-time detection, their detection accuracy, typically one to five per cent as observed in oil and gas environments, can be degraded by transients introduced at pump stations and or fluctuations in the flow rate due to production variations. Acoustic or fiber optic sensors can be run along the length of a pipeline but cannot be passed through valves and pumps, where leaks could also occur. Manned patrols by foot, car or drones provide visual, close-up inspections but typically not real-time detection. Additionally, the high cost of operation, reliance on favorable weather, and manpower make it unsuitable for continuous monitoring, and therefore reserved for periodic and post-event inspections to overlap with non-visual approaches.
The inadequacies of these technologies promote opportunities for small to medium leaks to go undetected long enough to become a serious problem. It also encourages innovation to help meet tougher industry standards.
Enhancing safety and efficiency with leak detection automation and remote validation
The mining industry is continuously seeking ways to increase automation, efficiency, productivity, and the safety of their operations. Cameras have been around for decades for their role in security and safety. Over time, they have been supplemented with motion detection software to become smarter. However, these devices still lack adequately advanced intelligence to allow complete autonomy and to have wider industrial applicability.
Innovation by Canadian software company IntelliView Technologies takes video and image analysis to new heights through the coalescence of leak analyzing video analytic artificial intelligence with dual sensor (thermal and optical) techniques. The IntelliView Vision System (IVS) is an edge-based internet of things (IOT) machine vision platform developed in 2014 to cost-effectively address the leak detection needs of a leading oil and gas midstream operator. Currently installed at numerous NA oil and gas facilities, the IVS is a product of over 300,000 hours of field testing and refinements. Patented and proprietary software contained in the Dual Camera Analytic Module (DCAM™) or the IntelliView HVR of the IVS process thermal and video feeds in real time. Data analysis goes beyond pixel temperature and motion detection by taking into account behavioral characteristics of small leaks, spills, sprays, and pooling.
The process of detecting, qualifying and reporting a leak transpires in seconds. Unlike with cloud computing where input is processed offsite, any potential processing or reporting delay is minimized. An exclusively cloud-based infrastructure can be restrictive as well as expensive. Transporting good quality video files for processing is not recommended for remote sites given their limited bandwidth capabilities and the high data cost. Often, in such situations, only photographic images are sent, which may not be sufficient for thorough and accurate analysis in the cloud.
The IntelliView system sends notification, with snapshot and video link, only for qualified events. This minimizes data overload and nuisance alerts that would otherwise take up valuable man hours. By also making live and recorded video available online through various internet-enabled devices, operators are able to remotely review and validate events at the soonest possible time, consequently enabling prompt action that’s needed to mitigate downtime, productivity loss, safety concerns, and negative environmental impacts.
Given the high toxicity level of liquids and chemicals used in mining, having prior visual field intelligence of an incident’s immediate environment better informs responders of the potential dangers and the safety risks associated with their reparation and remediation efforts. Users also benefit from improved operational efficiency and long-term value brought about by reduced windshield time, fewer scheduled site visits, asset security, and monitoring of personnel and contractor attendance.False alerts are a common challenge long faced by providers of smart camera systems and automated solutions. IntelliView’s in-house optimization, comprehensive array of proprietary detection fine tuning tools and environment filtering algorithms reduce false positives well below industry average. Because thermal sensors create images by capturing heat signature of objects, ambient lighting and unfavorable weather conditions, such as heavy snow, rain and fog, have little effect on performance.
Wider coverage and applicability in mining
The IntelliView Vision System offers a multitude of capabilities to meet the unique and complex needs of industrial settings. It is designed to handle applications of any scale, single system or multi-system/multi-site, and to integrate with IntelliView hardware, third-party devices, and client infrastructure. The flexibility in coverage also extends to the software side. A single system can operate both as a leak detection solution and a security solution, or any other combination from IntelliView’s line of solution offerings, bringing additional cost savings.
One of the advantages of an IoT architecture is the ability to distribute processing power at any point of the network. This allows site-specific requirements and restrictions to be accommodated, including the on-site or off-site supplementation of deep learning artificial intelligence-based classification – which can detect and filter objects such as cars, persons, and wildlife – to provide an additional layer of qualification and further reduce false alerts.Currently, the IntelliView Vision System is implemented for leak detection and security monitoring at a prime gold mining facility. By making available the combined potentials of imaging and video analytic technologies, IntelliView has brought to the fore new applications in mining. The system delivers value in processes where parameters such as movement, color or temperature are closely monitored.
To learn more about how IntelliView can protect assets, life and the environment of your mining operations, visit www.intelliviewtech.com.
About this Author
Gurjeet Bansal is the Corporate Communications Manager for the Canadian software company IntelliView Technologies Inc., a leader in AI-driven machine vision systems that help to improve operation efficiency and mitigate financial, safety, and environmental risks through the automation of real-time leak detection and industrial surveillance. Gurjeet has more than 15 years of combined experience in the field of marketing communications within media, agency, and corporate environments. Her educational background consists of a master’s degree in integrated marketing communications from the University of Westminster (England), a master’s degree in event management from the University of Technology Sydney (Australia), and a bachelor’s degree in journalism from the University of Santo Tomas (Philippines).