Common errors with manual pipe inspection coding
None of us is infallible. We all make mistakes. The fact that we are human means there is always a margin for error. In whatever we do.
If you are a commercial airline pilot, the room for error is just not there. Lives are at stake if you make a fundamental error. So, there are many aspects that come into play to ensure that risk is mitigated.
Planes, of course, are highly automated these days; autopilot can navigate and land the plane and arguably is less likely to make an error than a human. But the human is essential. You cannot rely on computers to do everything, and the human must be there in case of an emergency. The recent history of Boeing’s 737 Max is an example of what happens when this does not happen. This is where the best training comes in. Not just one-off training, but regular training and review. Then, of course, are the checklists and manuals. Specific processes to go through in every event.
Although planes are now highly automated, human input is still an essential element.
So how does this relate to the coding of pipes? Well, unlike planes, CCTV camera surveys are not yet fully automated. The reporting system still heavily relies on human input to generate a report. Of course, the information is only as good as what is put into the system by humans.
With pipe coding systems (and there are several of them around the world, including MSCC5 and PACP), there are a set of rules to follow in each instance. Fairly straightforward, right? – Well, yes, if everyone is trained the same way and sees the same thing. But we are humans, and we not only make mistakes but also see things from different perspectives.
What one might see may be different to another, and this becomes apparent when you look through vast amounts of reports and footage.
So, what are some common misconceptions, errors, and mistakes humans make when interpreting CCTV recordings?
Cracks vs Fractures
Without doubt the most common. What is the difference, and how can you tell? A crack is just a broken line visible on the pipe wall, be it circumferential or longitudinal. It will not be open or moved apart; the pipe is still in place. A fracture you can clearly see where the pipe is open on the pipe wall and shows a visible ‘black line’, pieces of the pipe are still in place.
Can you tell the difference between a crack and a fracture or cracks <1mm or >1mm” in Australia?
Pipe Material or Material Change
Many times, the pipe material is coded wrong or missed completely. Often where there is a material change midline it is missed, and many coders can’t tell the difference between certain materials. Common types that get mixed up are pitch fibre and cast iron as they are both dark and can be hard to see. The difference between Vitrified clay and PVC and concrete and asbestos cement.
Getting the code wrong here could lead to the wrong repair specification or inappropriate recommendations. With pitch fibre, this could cause a longer-term issue if the presence of the material is not identified at an early stage in the survey process and will affect the choice of no dig repairs or excavations.
Broken vs Hole
Another one we see a lot is the difference between a broken pipe and a hole in the pipe. People’s definitions are mixed up here and without the correct training and reference materials these are often miscoded. A broken pipe shows pieces of pipe that have moved from their original place but are still there, whereas a hole is a visible hole in the fabric of the pipe with the pieces missing.
The broken pipe may need to be removed, whereas a hole has already gone. It may be necessary to recover the evidence. If this is coded wrongly then further work required may not be correctly identified, leading to further blockages.
Junction or Connection
Lots of coders will confuse junctions with connections. The definition of a junction is one that is performed or purpose-made and built into the sewer. A connection, however, has been added afterwards, usually with a boss or just smashed into the pipe, causing a defect.
Junctions are meant to be there. If they are wrongly coded as connections, a recommendation may be made to remove and replace them when it’s not needed. Leading to costly, unnecessary excavations.
Deposits
Attached or settled, fine or course, grease, or encrustation. All areas where easy mistakes can happen. Encrustation is often coded as grease and vice versa.
So, what is the difference between grease and encrustation? The easiest way to remember is that grease is essential ‘man-made’ whereas encrustation is usually mineral deposits caused by nature. These are usually attached to the pipe.
Deposits can cause flow issues, blockages and reduction in hydraulic capacity in the pipe and these foreign materials can stick to the inner pipe wall and become relatively permanent.
Settled deposits can be fine (DE S) like sand or silt or coarse (DE R) like gravel or rubble. There are plenty of descriptor codes in this category which often leads to confusion and if in doubt the code (DS Z) is a catch-all for all deposits that can’t be classified by the other codes.
Getting the type of deposit right is important because it will build a case as to what the issue with the pipe is. The evidence is important, especially if the pipe is being abused with grease and fat. If it is coded wrongly, the correct action may not be taken where needed.
Clock references
There are often errors in clock references. Either when inputting a ‘to and from’ or a specific point (and determining when its required). Also, the actual clock reference itself. One person’s 7 o’clock is another’s 8.
Getting the position wrong could affect where patches are installed or how it affects the serviceability of the pipe. A hole at 11 o’clock is not so damaging as a hole at 7 o’clock if the pipe is running at half bore.
A junction at 3 o’clock or is that 4 o’clock?
Percentages
Used for several codes, percentages, like clock references, are open to human interpretation. Used in deformation, cross-sectional loss, and water levels these are often under or overestimated and not accurate.
The extent of the intrusion or root mass or the water level is important to determine remedial action and the timescale. If cross-section loss is estimated at 50% when it’s only 20% this could overplay the urgency.
Summary
So, can AI and automation, like the functionality provided by VAPAR, help with these errors?
Of course, the answer is yes! Setting out and agreeing on the exact parameters which define a code can allow the software to take out the subjectivity. Although, like a human, an AI will not be 100% accurate, 100% of the time. This is where the provision of a targeted human check of the AI outputs allows the accuracy to exceed using just the AI or Human capability. This approach is no different to how an airline pilot would use manuals and checklists on a plane to ensure the automated systems are working as intended.
Of course, we are not dealing with hundreds of lives on a plane here, and these types of errors will not necessarily cause a catastrophic consequence. However, the pipe inspection coding errors in extreme cases can lead to the internal flooding of someone’s property or the pollution of a nearby water course. A Combined Intelligence model, like that used by VAPAR, effectively merges the AI and human inputs to minimise mistakes, improve both accuracy and consistency, and generate efficiencies by grading pipes appropriately to be used in the asset management of piped networks.
This article is Co-Authored by Nathan & Anthony
Learn more about Sewer Network
VAPAR automates sewer and stormwater pipe condition assessment for councils, utilities and CCTV contractors. Learn how we help improve the monitoring and maintenance of the underground pipes using AI.