Asset management, CCTV, Condition assessment, Stormwater, Wastewater, Water

Veolia partners with VAPAR for CCTV defect recognition solution

Inspection and condition assessments of network infrastructure are critical for water utilities and city councils to ensure structural integrity and functionality of sewer pipes and stormwater drains. These assessments are necessary to identify the pipes requiring rehabilitation before they deteriorate past the point of renewal.

For this reason, water utilities and city councils across Australia spend millions of dollars every year to maintain their sewer and stormwater assets. The traditional CCTV assessment methodology of network assets presents challenges for utilities and operators, such as:

  • the time required to visually review CCTV inspection videos and identify defects.
  • the operator experience and subjectivity.
  • the tracking, geo-locating and organising large amounts of CCTV data.

In this context, VAPAR CEO Amanda Siqueira and CTO Michelle Aguilar founded the company in 2018 to provide smart solutions that automate the condition assessment of network assets. Ms Siqueira and Ms Aguilar developed an algorithm able to auto-code CCTV videos using artificial intelligence (Figure 1).

“The technology provides an accurate, cost effective and quick alternative to laborious video inspections I worked with during my intern days,” says Ms Siqueira.

Veolia, as a long-term provider of network services for municipal water entities, is always looking for innovative digital technologies that can improve quality of service while reducing labour-intensive tasks. A partnership with VAPAR was therefore the logical step to address the challenges operators face every day.

“Fully committed to digital transformation, Veolia is always looking at different ways of using data to improve outcomes for both our clients and Veolia’s operations,” says Veolia General Manager Water Victoria Jean-Michel Seillier.

This led Veolia to conduct a trial in Victoria in February 2020 to assess the accuracy of VAPAR software and provide a market-leading case study for the application of AI technology to pipeline inspections.

The trial’s positive results point to exciting developments for Veolia, VAPAR and Australian water utilities and councils.

The trial

This trial was based on 198 videos, representing 3.6 km of pipes selected from various inspection campaigns performed by Veolia Network Services (VNS) in Victoria, and covering a wide range of diameters and materials. The condition of these pipes had been assessed by operators during these campaigns.

In parallel, VAPAR used its algorithm to detect defects from all selected videos. Two assessments were subsequently available for the same video: the ‘Operator Assessment’ and the ‘VAPAR Analysis’. Veolia then mandated an independent seasoned expert to carefully review these 198 videos to provide a ‘Reference’, used as a source of truth.

The Operator Assessment and the VAPAR Analysis were then compared to the Reference.

To facilitate this comparison, the various defect types were classified across six categories: cracks; roots; obstructions; joint defects; connection defects; and other. The performance of the assessments was then estimated for each of these categories.

The following two indicators were calculated to assess the performance of both the Operator Assessment and the VAPAR analysis against the ‘Reference’:

  • Precision: the precision is an indicator of the number of ‘false alarms’. For example, a precision of 80 per cent means that 20 per cent of defects listed in an assessment are actually not defects (i.e. false alarms).
  • Recall: the recall is an indicator of the number of missed defects. A recall of 70 per cent for example means that 30 per cent of the actual defects were missed in the assessment.

Results

The outcome of this analysis revealed the VAPAR algorithm performs relatively well, missing only 13 per cent of the defects, compared to 37 per cent for the Operator. This is mostly explained by the capacity of the VAPAR algorithm to detect micro-defects that are typically not reported by busy operators.

Figure 2. Recall (missed defects) and precision (false alarms) for each category, comparing VAPAR and the operator.

The algorithm was found particularly performant at identifying roots, cracks and joint displacements. As shown in Figure 2, the VAPAR algorithm is particularly oversensitive on displaced joints: approximately 50 per cent of the defects reported by VAPAR were not clearly visible in the video frames.

Figure 2. Recall (missed defects) and precision (false alarms) for each category, comparing VAPAR and the operator.

What is particularly important in condition assessment of network assets are the service and structural grades of the pipes. These grades, between 1 (good conditions) and 5 (critical conditions), are typically used by water utilities and city councils to direct maintenance and rehabilitation programs.

The most valuable result of this trial was that the VAPAR algorithm significantly improves the accuracy of the grading. Indeed, Figure 3 shows that VAPAR assessment for structural grades is correct for 80 per cent of the pipes, compared to only 48 per cent for the Operator Assessment.

Figure 3. Overall comparison of VAPAR and the Operator for defects detection, pipeline structural grades and service grades.

Similar results were obtained for the service grades, with 76 per cent of correct assessment in VAPAR analysis against only 52 per cent in the Operator Assessment. This improved accuracy will allow for optimisation of the maintenance and rehabilitation programs of network assets.

In the search for innovative technical solutions that can contribute to improving operational performance, Veolia was pleased to perform this successful trial with VAPAR. The VAPAR algorithm was found to improve the existing methodology to assess the condition of sewer pipes and stormwater drains.

In particular, the capacity of detecting micro-defects can significantly refine the condition assessment of these critical network assets. Despite a slight over-sensitivity, this trial demonstrated the VAPAR algorithm is more accurate than operators to grade network assets.

As the structural and service grades are typically the metrics used to direct the maintenance and rehabilitation programs in water utilities and city councils, this improved assessment can help better optimise the investment in this critical public infrastructure.

For this reason, Veolia launched a partnership with VAPAR and is currently building a range of offers to support councils and water utilities to maximise the return on their investments in sewer and stormwater assets.

This article was featured in the September 2020 edition of Trenchless Australasia. To view the magazine on your PC, Mac, tablet or mobile device, click here.

For more information visit the Veolia website.

If you have news you would like featured in Trenchless Australasia contact Assistant Editor Sophie Venz at svenz@gs-press.com.au

Send this to a friend