Specifications often say “100 percent NDT” or “NDT as required” without saying which method, on which welds, at which stage. The result is that the fabricator picks the cheapest method that satisfies the wording, and the buyer discovers later that the chosen method could not have found the flaws that matter. Good NDT method selection avoids that gap by starting from the flaw, not from the equipment.
This article offers a practical way to match the method to the weld: what you are worried about, what the material and thickness allow, how the joint can be reached, what record you need and what site conditions will permit. It also flags a change in the governing standard that many projects have not yet noticed, which affects which edition your contract should cite.
- Start with the flaw you fear
- Five methods at a glance
- Material, thickness and access
- Records, speed and cost
- Why the standard changed in 2025
- Method standards and acceptance standards
- Extent, sampling and timing
- Frequently asked questions
Start With the Flaw You Fear
The right question is not “which NDT method is best?” but “which flaws would hurt us, and which method can find them?” A pipeline weld in a high-pressure line worries about cracks and lack of fusion. A structural weld in a lightly stressed frame worries more about gross defects and dimensions. A stainless steel vessel worries about surface cracks and lack of fusion at the root.
Flaws split into two broad families. Surface flaws, such as cracks open to the surface, are found by visual, penetrant and magnetic methods. Internal flaws, such as buried lack of fusion, slag and porosity, need volumetric methods, which in practice means ultrasonic and radiographic testing.
Planar flaws, such as cracks and lack of fusion, are generally more damaging than rounded ones. Methods differ in how well they detect each. Ultrasonic testing tends to be strong on planar flaws, while radiography is strong on volumetric ones and weaker on planar flaws that are not aligned with the beam. Starting from the flaw therefore narrows the field before any other factor is considered.
Five Methods at a Glance
| Method | Finds | Main limits |
|---|---|---|
| Visual (VT) | Surface flaws, dimensions, profile | Surface only, depends on inspector and lighting |
| Penetrant (PT) | Surface-breaking flaws on non-porous materials | Needs clean surface, surface-open flaws only |
| Magnetic particle (MT) | Surface and slightly subsurface flaws | Ferromagnetic materials only |
| Ultrasonic (UT) | Internal flaws, especially planar, with depth | Needs surface access, skilled operator, thickness limits |
| Radiographic (RT) | Internal volumetric flaws, reviewable image | Radiation safety, planar flaws can be missed, access to both sides |
The table is a starting point and not a substitute for the code. Real projects often combine methods, such as VT with MT on a carbon steel weld, or VT with PT on stainless steel, and then add UT or RT for volumetric assurance. A weld that passes a surface test has not been shown to be free of internal flaws.
Each of these methods has its own detailed limits, which are set out in the relevant standards. The point of the comparison is to show that no method covers everything, and that a sound plan combines two or three of them.
Material, Thickness and Access
Material decides what is possible. Magnetic particle testing works only on ferromagnetic materials, so austenitic stainless steel and aluminium need penetrant testing for surface checks. Ultrasonic testing standards for welds often assume ferritic steel and a minimum thickness, so thin sections and other metals need special procedures or other methods.
Thickness affects both UT and RT. Very thick sections require high-energy sources for radiography, while thin sections may fall outside the range of conventional manual ultrasonic standards. If your equipment sits at either end of the thickness range, check the scope of the standard before assuming that the method applies.
Access completes the picture. Radiography usually needs access to both sides of the joint, one for the source and one for the detector. Ultrasonic testing needs a clear scanning surface next to the weld. Penetrant and magnetic testing need the surface to be reachable and clean. A joint in a tight corner may rule out a method that looks perfect on paper.
Records, Speed and Cost
The need for a permanent record steers the choice. Radiography produces an image that can be reviewed by someone else, and advanced ultrasonic techniques can store data for later review, whereas conventional manual UT and the surface methods depend mainly on the written report and any photographs. If the owner requires independent review of results years later, the method should be chosen accordingly.
Speed and disruption matter as well. Visual and magnetic testing give immediate results. Radiography requires exclusion zones and often night scheduling, which can disrupt other work. Penetrant testing needs dwell and development times. The method that is quickest on a data sheet may not be quickest on your site.
Cost comes last, not first. A cheaper method that cannot detect the flaws of concern is false economy, while a more expensive method applied everywhere may be unnecessary. The best plans use inexpensive methods broadly and expensive ones selectively, guided by risk.
Why the Standard Changed in 2025
ISO 17635 is the standard that provides general rules for non-destructive testing of welds in metallic materials, including how to choose methods and evaluate results. According to ISO, the 2016 edition was withdrawn on 2 April 2025 and replaced by ISO 17635:2025.
That matters for contracts. Specifications written before 2025 may still cite the 2016 edition, and newer ones may cite the 2025 edition or an undated reference. The edition determines which rules apply, so confirm with the buyer and the fabricator which one governs, and record the decision in the quality plan.
The older edition explained that acceptance levels are not a direct translation of the quality levels in other weld standards, but relate to the overall quality of a production batch of welds. Whether the new edition keeps the same framing is something to confirm in its text, not to assume. In general, it is safer to name the edition explicitly than to trust a vague reference.
Method Standards and Acceptance Standards
A recurring source of confusion is the difference between how to test and what to reject. Method standards describe the technique, such as the setup and the coverage. Acceptance standards define the limits, such as the size or type of indication that is allowed. They are often separate documents, and the contract must name both.
For example, standards exist for acceptance levels for penetrant testing and for magnetic particle testing as distinct documents, separate from the method standards. A plan that cites only the method leaves the inspector to decide acceptance, and different inspectors will decide differently.
Quality levels are a third element. A weld quality level defines how good the weld must be overall, and the NDT acceptance level should be chosen to be consistent with it. A mismatch, such as a stringent quality level paired with a lenient method or coverage, can leave the project with a false sense of security.
Extent, Sampling and Timing
Extent describes how many welds are tested and how much of each. It may range from a sample of production welds to every weld, and the right amount depends on the code, the joint category, the service and the fabricator’s track record. The contract should state the extent per method and say how it increases if rejects are found.
Timing matters too. Surface methods are normally applied after visual inspection and before volumetric ones, and some materials call for a waiting period after welding before final NDT. Testing too early can miss delayed cracks, while testing too late can leave no time for repair before shipment.
Tie the NDT plan to the inspection test plan so that each method has a place in the production sequence and a hold or witness point where needed. G.I Quality Services lists visual inspection, ultrasonic, magnetic particle, dye penetrant and radiography testing among its services, and any provider can help to map methods to welds, but the decision on what you need should be made against your own risks.
Frequently Asked Questions
How do I choose the right NDT method for a weld?
Start with the flaws that matter for the service, then check what the material, thickness and access allow. Add the need for a permanent record, site conditions and cost, and combine methods where one cannot cover everything.
What is the difference between surface and volumetric NDT?
Surface methods, such as visual, penetrant and magnetic testing, find flaws at or near the surface. Volumetric methods, such as ultrasonic and radiographic testing, examine the inside of the weld.
Is one NDT method enough for a critical weld?
Usually not. A surface method does not show internal flaws, and each volumetric method has blind spots. Critical welds are commonly checked with a combination chosen to cover different flaw types.
Which edition of ISO 17635 should I cite?
According to ISO, the 2016 edition was withdrawn in April 2025 and replaced by ISO 17635:2025. Confirm with your contract parties which edition governs, and state it explicitly in the quality plan.
Why do specifications need an acceptance standard as well as a method standard?
The method standard tells the inspector how to test, while the acceptance standard says what indications are allowed. Without both, the decision to accept or reject is left to individual judgment.
How much NDT is required on a project?
It depends on the code, the joint category, the service and the contract. Specify the extent per method and how it increases if rejects are found, rather than leaving it to general wording.