PipeInsights™ : The Use of Machine Learning & Artificial Intelligence in Gravity Sewer Defect Recognition
The inspection of gravity sewers is labour intensive and can be subjective. Sewer asset owners are continually managing budget constraints while trying to maintain aging systems to prescribed levels of service. Canada’s aging sewer infrastructure requires a more focused understanding of its condition. This will require increasing the frequency and efficiency of condition assessment surveys. The use of Artificial Intelligence/Machine Learning (AI/ML) technology will be a disruptor in the sewer assessment industry. In anticipation of this, AECOM is focused on improving methods and workflows to enhance the speed and accuracy of the sewer assessment process and develop the most cost-effective capital improvement programs.
There are more than 292,432 km of storm and sanitary sewer pipelines across Canada (Statscan. 2018). These pipelines are subject to deterioration due to external influences including type of flow, soil condition, surcharging, and external loads. With the increasing frequency and severity of storm events due to climate change, these factors can exacerbate the deterioration process. Recent North American Infrastructure Report Cards suggested that the existing sewer infrastructure is Grade D+ or poor/at risk condition. (ASCE 2021; Canadian Infrastructure Report Card 2019).
To meet these challenges, municipalities and utility owners allocate annual budgets to monitor the condition of sewer assets and develop engineering solutions to rehabilitate or replace failing infrastructure. The inspection of sewers is the first step towards proactive maintenance and a sustainable asset management program.
The most commonly used inspection technique in sewers is Closed Circuit Television (CCTV). Based on the video interpretations of the operator, defects are coded and graded to rate the overall pipeline. In general, each inspection is reviewed three or four times before inspection results are applied to a Capital Improvement Plan (CIP). This process is labour intensive given that much of the time could be spent assessing defect-free pipe.
AECOM developed SewerLogic as an AI/ML driven tool to automatically identify and code defects, recommend rehab strategies, and improve the CIP development process
The AECOM team won this prestigious event against other AECOO competitors. SewerLogic uses AI and vision detection for:
SewerLogic also uses geospatial data exchange standards that allow organizations to leverage their enterprise GIS data. Connected to a GIS interface, users can quickly locate infrastructure assets and navigate the full history of inspections on that asset — both in the office and in the field.
SewerLogic is a SaaS-based offering that performs fully-automated defect recognition and is designed to utilize non-working hours to process sewer inspections through the use of AI. The typical workflow involves uploading the sewer video, inspection databases and GIS data via the SewerLogic interface. Once uploaded, SewerLogic’s AI begins assessing the inspection – validating defects provided by the contractor, detecting uncoded defects, generating validation statistics and creating both AI- developed defect databases and AI-created inspection videos complete with AI- generated bounding boxes for each defect.
SewerLogic has been trained on 93,000 images with over 500,000 labelled defects to date. When developing SewerLogic, AECOM’s infrastructure engineers prioritized training of the Automated Defect Recognition (ADR) software based on the most common and impactful defects from our current infrastructure rehab programs focusing on the most common or critical defects first. SewerLogic is trained to detect/validate defects at the group/family level and is trained on over 110 individual NASSCO PACP modifiers representing 95% of the PACP defects typically seen on infrastructure rehab programs.
SewerLogic can be applied to all pipe sizes and has been successfully tested with remotely piloted aircraft systems (RPAS or drones) on very large pipes. AECOM recognizes, however, that the true value of ADR algorithms exists in the rapid processing of smaller (<1200mm), common-diameter pipes.
SewerLogic helps asset owners build informed CIPs for resilient and sustainable sewer infrastructure. The solution was designed to produce consistent evaluations and accurate interpretations while reducing review time and allowing asset owners to prioritize work and effectively manage their networks. This optimized tool can suggest sustainable construction methods including trenchless technology methods that maximize available funding, and reduce construction footprints, social disruptions, and carbon dioxide (CO2) emissions.
The strategic investment in infrastructure pays dividends in overall economic activity and GDP growth. A 2020 GI Hub study that found the economic multiplier for public investment (including infrastructure) is 1.5 times greater than the initial investment in two to five years — much higher than other forms of public spending. This is a compelling argument for infrastructure as a post-COVID-19 economic stimulus.
The GI Hub is also encouraging government and industry to look beyond infrastructure’s impact on economic growth and consider it as a means to achieve transformative outcomes. Specifically, for the application of sewer systems, the accurate identification of defects via CCTV yields opportunities for trenchless rehabilitation. AECOM’s analysis on various programs, including the City of Winnipeg and the Detroit Water and Sewerage Department, estimates the benefit/cost ratio of accurate condition assessment to be >30 when compared to the cost of open-cut replacement. This means that the cost of CCTV inspections, coupled with accurate defect evaluation to identify trenchless rehab opportunities saves vast amounts of money and allows sewer rehab programs to be expanded, compounding increased economic prosperity and the health of our cities and environment.
At its core, SewerLogic was created as a tool to more cost-effectively deliver infrastructure programs, and address the immediate needs of aging sewer systems while improving levels of service to the public. In short – to improve the lives of people through improved infrastructure investment. AECOM has been an industry leader in the application of new technologies in the assessment of gravity sewers. SewerLogic saves engineers and project teams time and allows asset owners to quickly process large volumes of data. AECOM plans to scale SewerLogic to work on defect coding systems in Europe, the Middle East and Australia.