Dealing with possible defects in welds which are not examined

by Nicolas Kohout
(France)

This is not really a story but more of a questioning about what I think is recurring when dealing with welded equipment inspections and risk management.

I exclude aerospace construction because it's a field in which no defect at all is tolerated (in the extent of what it is possible to examine of course).
If you consider a pressure vessel, you have to follow strict design principles which are first dependent on where you plan to install it (local regulations) and then there's a bit of a choice from the customer and/or the manufacturer (which will chose an appropriate design code regarding the vessel, the local regulations, and so on...), additionally the manufacturer will often have to follow customer specifications.

Now, considering regulations, code and specifications require for example 20% radiographic examination on longitudinal welds and you find some repeated defects in the welds (let's be specific and say lack of fusion). If several repairs are done respecting all standards and specifications, and that you are able to clear all the defects, what posture should you have considering risk management of potential presence of other defects in remaining welds.
Codes require some control extension when you find defects but not always 100% examination so you could say you're good to go, assemble, test and install your vessel on site. On the other hand, you can reasonably believe that defects are remaining. But it's only a believing, big defects, small ones, of the same kind or not, on which length and so on, all of that you cannot know until you examine remaining welds, but the code does not require it. How do you decide to go on with manufacturing or perform additional examinations to clear your technical doubts? Even considering potential defects in some case (depending mostly on fluid, temperature and pressure cycles) it's highly improbable that the vessel will explode but you cannot be sure... The risk is big (explosion) but the probability of it to happen is uncertain...

What would be your position? Of course, depending on the field I guess it would not be the same, but what would be a good decision making attitude that you could use in many situations?

Comments for Dealing with possible defects in welds which are not examined

Average Rating starstarstarstarstar

Click here to add your own comments

Feb 22, 2017
Rating
starstarstarstarstar
Think of Partial Inspection from a Statistical Point of View
by: Niekie

Nicolas, you ask a very good question.

When it comes to partial inspection, such as spot radiography of pressure vessel welds, then we need to understand that we are using the inspection as a statistical sampling tool.

It is understood that there could be defects that are missed, but the argument is that if the sample inspection is representative of the rest of the welding, and this sample inspection is acceptable, then the probability that there are unacceptable defects is rather low.

For pressure equipment, different amounts of inspection will result in different design factors being used by the designer. For instance, when using 100% radiography, the design factor may be 1. This assumes that the weld has 100% of the strength of the base metal. When using spot radiography, the design factor may go to 0.85. This assumes that the weld can only carry 85% of the stress of the base metal.

Most codes will cover the rules of what to do if the spot radiograph exposes unacceptable defects. In the case of ASME VIII Div 1, this is dealt with in section UW-52. This requires that 2 additional spot radiographs need to be taken for each of those that failed. If any of the additional radiographs show unacceptable defects, then the whole weld "increment" represented by those radiographs is unacceptable, and needs to be removed and re-welded, or 100% radiography needs to be performed. Obviously, any unacceptable defects found need to be repaired.

So, if you have complied with all code requirements, and you have not ended up doing 100% inspection, but there were none-the-less a number of unacceptable defects found, and repairs performed, there could obviously be legitimate concerns regarding the amount of defects in the uninspected welds. My way of thinking about this is as follows:

1) All design codes have "factors of safety" to cover the unknown and unknowable. As mentioned in the article, these can easily be around 3.
2) The designers will assign the level of NDT, based on the risk associated with the vessel. For high risk applications, the level of RT or UT will be 100% of highly stressed welds.
3) When partial examinations are performed, additional safety factors are introduced, to compensate for "unfound" defects.
4) Ductile materials are actually quite tolerant of defects, especially if they are relatively thin. (Say less than 20mm for pressure equipment.)

If you are the inspector, and you have a suspicion that the fabricator is trying to "pull the wool over your eyes" by hiding poor quality welds, then there is obviously a trust problem. Best you assign the areas where you want the NDT performed, then this should not happen. Also make sure that each welder's work is included in the spot NDT "sample".

Click here to add your own comments

Join in and write your own page! It's easy to do. How? Simply click here to return to Sensors in Welding.