How Much Radiography Is Enough? Weld Porosity, Inspection Coverage, and Confidence in Strategic API 650 Potable-Water Storage Tanks for Emergency Preparedness
This case study of two API 650 potable-water tanks reveals that weld porosity defects cluster by production batch rather than by welder or joint type, suggesting that current random radiographic sampling may miss critical flaws and that inspection protocols should consider batch-triggered extensions and enhanced environmental documentation to ensure structural confidence.
Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). This is an AI-generated explanation of the paper below. It is not written or endorsed by the author. For technical accuracy, refer to the original paper. Read full disclaimer
In many parts of the world, the safety of drinking water depends on massive steel tanks that sit on the ground, holding reserves for days or even weeks. These structures are built by welding thick metal plates together, a process that relies on human skill and the right conditions. To ensure these tanks will not fail, engineers use a standard set of checks. They look at the welds with their eyes, they test for leaks by creating a vacuum or using oil, and they fill the tank with water to see if it holds. However, the most critical check for hidden flaws is a technique called radiography. This is essentially an X-ray of the metal weld, which can reveal tiny pockets of gas trapped inside the joint that the other tests cannot see. While these tanks are vital for emergency preparedness—acting as a buffer if a desalination plant stops working or a disaster strikes—there has been little real-world data on how well the standard sampling method works. Specifically, no one knew if taking a few random X-ray pictures of the welds was enough to catch defects that might be hiding in clusters, or if those defects were spread out randomly.
A researcher named Ayman Ramadan set out to answer this question by examining the complete inspection records of two nearly identical water tanks built at the same time. Both tanks were made of steel plates ranging from 10 to 20 millimeters thick and were constructed with five layers of vertical seams. The plan was to take 122 radiographic images across both tanks, a sample size designed to represent the whole structure. In the end, 67 of these images were actually taken. The results revealed a pattern that challenges how we think about inspection. Out of the 67 spots checked, six were rejected because they contained porosity, which is a cluster of tiny gas bubbles trapped inside the weld metal. All six of these defective spots were found in just one of the two tanks. Even more telling, every single one of these six defects appeared in a specific group of welds made on the same day, in the same batch, and in the upper sections of the tank where the plates were 10 millimeters thick.
The distribution of these defects offered a crucial clue. The six rejected spots belonged to four different welders, and the same welders had successfully completed the identical joints on the other tank without issue. This ruled out the idea that a single worker was the problem. Instead, the defects seemed to be linked to the conditions of that specific day or the batch of materials used. The researchers noted that the other inspection methods used on the project—visual checks, vacuum tests, and oil penetration tests—had all passed without finding any issues. This is not surprising because those methods can only detect flaws that break the surface or create a leak path. The gas pockets found by the X-rays were trapped deep inside the metal, invisible to the surface tests. Furthermore, the hydrostatic test, where the tank was filled with water to its design level, did not stress the upper sections of the tank where these defects were found, meaning the water test would never have revealed them.
The study then looked at the odds of catching such defects with a standard sampling plan. If defects are scattered randomly, a sample that covers about a quarter of the welds might catch a single bad spot only one time out of four. However, the data showed that defects tended to cluster together in batches. When the researchers modeled this scenario, they found that if a cluster of bad welds existed, a random sample had a much better chance of hitting at least one of them. Once a single defect is found in a cluster, the standard practice is to expand the inspection to check all the welds made during that same period. In this case, that approach would have likely uncovered all the hidden defects in that batch. Without this expansion, a random sample would have missed the majority of the problems.
The findings suggest that while the current rules for inspecting water tanks are a good starting point, they might need to be adjusted to account for how defects actually form. The study proposes that when a defect is found, inspectors should look not just at the specific welder, but also at the time and materials used, expanding the check to include all welds made in that same batch. The researcher also emphasized the importance of recording details like the weather and the specific batch of welding rods used, as these factors might explain why a whole group of welds fails at once. For tanks that serve as strategic reserves for emergency water supplies, the confidence that comes from a thorough inspection is as important as the water itself. This research does not prove that all tanks are unsafe, but it does show that relying on a small, random sample without looking for patterns of batch-related failure could leave dangerous flaws undetected. The work highlights that in the world of critical infrastructure, understanding how and why defects group together is just as important as finding them.
Drowning in papers in your field?
Get daily digests of the most novel papers matching your research keywords — with technical summaries, in your language.