AI in Sewers
The importance of the pipe network beneath out feet
Our cities and suburbs are supported by a vast underground network of water, wastewater and storm water infrastructure. This network of gravity pipes, pumps and filtration systems play a very important role in the quality of our life, eliminating disease, safeguarding the environment, and protecting communities.
However, parts of this aging infrastructure are nearing the end of its useful life and now (more than ever) requires closer attention. Without attention, this situation is not sustainable.
Most of our water and wastewater infrastructure were installed during the 19th century and municipalities are facing the challenge of broad-scale infrastructure replacements or repairs costing hundreds of millions of dollars.
Adding to all this is the changing climate factor, meaning systems that were designed 30 or so years ago may not be sufficient to support everchanging environment around it.
To extend the life of infrastructure, reliance on smart city technology capabilities is critical. By creating visibility into buried assets to understand the conditions of underground infrastructure, utilities can compare current performance with expectations, and predict when and where problems may arise. This also leads the way to prioritisation of maintenance work, decreasing downtime of the assets, resulting in reduced interruptions.
With sensors and actuators becoming more cost effective, an array of technologies is becoming available for the pipe industry. For pressure pipes or pipes transporting materials under high pressure, static sensors are being used to help monitor the health of the asset. In sewer and stormwater applications, inspection by video still is widely adopted with assessment being carried out visually by an expert.
With operational technology (OT) and information technology (IT) coming together, data that were once only available in isolated networks is now available via the world wide web. What this means is CCTV operators are no longer needing to download inspection videos to a hard-drive in order to assess the condition of the pipe in the office, instead they can upload the video file over the cloud.
AI at your service
With more data being available and accessible, a path has been paved for advanced technology such as artificial intelligence or AI. These smart algorithms feed on data, in-fact, the more data that is available, the quicker and more accurate an AI system can become.
Like other technologies, AI is tool to better understand a problem so to make data driven decisions. One of the areas where AI is helping the pipe industry is in the field of video processing. The traditional means of CCTV condition assessment presents several challenges including time taken to review the videos and identify defects, the operator subjectivity and field conditions making visual inspections difficult.
The AI models are pre-trained to detect certain anomalies, in this case pipe features and defects. The inspection video is then ingested and inferenced against the trained model. The result is the identification of the type and importance of anomalies.
Integrating the above-mentioned technologies, the VAPAR.Solutions platform leverages cloud computing and its AI engine to automatically assess inspection videos that users upload. The platform is accessed via any web browser where videos can be uploaded, analysed, manually audited by an expert (if required), with a report generated and stored, eliminating the need for hard drives to back up the video data and corresponding reports.
With this approach, both asset owners and CCTV contractors are reducing the time taken for assessments, standardising the process to remove any subjectivity and utilising AI to deep dive into the data to get better outcomes.
In 2020, VAPAR worked with asset owners in Victoria, Australia, where the results showed that the solution outperformed the same inspection carried out manually. The AI algorithm missed fewer defects and was more accurate in grading the pipes. To date, VAPAR has processed over 3 million images, which means the AI has only become quicker and more accurate.
With the need for the pipe network needing special attention, technology is adding another lens to take a closer look. It’s empowering engineers, operators, and decision-makers to make data driven decisions more cost effectively and proficiently.