What Each NDT Method Can Miss on a Weld

No test sees everything. Nondestructive testing (NDT) checks a weld without damaging it, and each method has blind spots built into its physics. A clean report means the method found nothing it could reject, under the conditions it was used in. It does not mean the weld has no flaws. A senior inspector chooses a method by asking what could be wrong with this weld and which test can actually see that. Getting this wrong is expensive, because a test that cannot see the flaw gives you confidence you have not earned.

NDT is performed by qualified technicians working to a written procedure. Your job as an inspector is to pick the right question, read the results, and know the limits. Reading about a method does not qualify you to run it or to sign for it.

What You Will Be Able to Do

• Sort flaws by shape and location, and match them to a method.
• Explain what PT, MT, UT and RT can each miss, and why.
• Choose a method for a given material, thickness and joint.
• Combine methods to cover each other's blind spots.
• Review an NDT report for gaps.

Ask the Question First

Before you pick a method, answer these questions:

1. What flaw are you worried about? Cracks, lack of fusion, porosity or slag?
2. Where would it be? At the surface, just under it, or deep inside?
3. Which way would it lie? Along the weld, across it, or along a bevel?
4. What is the material? Is it magnetic? Is it coarse-grained?
5. What is the joint? How thick is it, and can you reach both sides?
6. What does the code or contract require? That may settle the method and the extent.

Two words make the choice clearer:

• Volumetric flaws have some bulk, like pores and slag.
• Planar flaws are thin and flat, like cracks and lack of fusion.

Planar flaws are usually the most dangerous, since their sharp edges can grow under load. They are also the easiest to miss, because they are thin.

The Methods Compared

Penetrant Testing

In penetrant testing (PT), a liquid is drawn into open flaws. The surface is cleaned, and a developer pulls the liquid back out so that it shows.

The whole method depends on the flaw being open to the surface and empty. So it can miss:

• Anything under the surface. Even a flaw just below the surface will not show.
• Cracks filled with oil, water or dirt. The penetrant has no room to enter.
• Cracks smeared shut. Grinding, wire brushing or blasting can push metal over a crack opening. Clean with the method the procedure allows, and be careful with mechanical cleaning just before PT.
• Shallow flaws that were over-washed. Too much cleaning pulls the penetrant out.

Dwell time is how long the penetrant stays on before you remove it. It comes from the procedure. Maker and federal guidance commonly gives 10 minutes or more, and cold metal needs longer because the liquid thickens. One maker reports that at about 45 F, dwell should be roughly double. Testing below the procedure's temperature range is not valid, and any frost or moisture has to be removed and the part dried first.

PT works on stainless, aluminum and other metals that MT cannot test. Rough weld surfaces can hold penetrant and give false indications, so a good surface matters.

Magnetic Particle Testing

In magnetic particle testing (MT), a magnetic field runs through the steel. A crack disturbs that field, and fine iron particles collect over the flaw.

Direction matters most. Flaws show best when they lie at 90 degrees to the magnetic field. TWI, a welding research institute, says flaws up to about 45 degrees from that still show. A crack running along the field can stay invisible. So you test each area twice, with the second pass about 90 degrees from the first, to catch cracks in every direction.

MT can miss:

• Cracks lined up with the field. This happens when you test in only one direction.
• Flaws deeper than near-surface. MT is not a test for the middle of a thick weld.
• Everything in nonmagnetic metal. Most stainless steels and all aluminum cannot be tested this way.
• Weak-field flaws. A worn yoke or a wide leg spacing gives a poor field. Check the yoke with a lift test before use. An AC yoke is commonly required to lift 10 pounds, and the DC weight depends on the standard you are working to.

MT beats PT for tight cracks in carbon steel, since it does not need the crack to be open and empty. Sharp geometry such as a weld toe can give false indications, so a trained eye has to read the result.

Ultrasonic Testing

Ultrasonic testing (UT) sends high-frequency sound into the steel. Flaws reflect it back. Angle beam probes send sound in at an angle, to reach the weld from the plate beside it. The beam can also bounce off the back wall to reach parts of the weld from the other direction.

UT is better than radiography for finding planar flaws such as cracks and lack of fusion, and it can give depth. It needs access to one side only. But it can miss:

• Flaws that do not face the beam. A flat flaw returns the most sound when the beam hits it square on. Sound hitting it at a shallow angle goes elsewhere.
• Parts of the weld the scan did not reach. TWI's example is lack of sidewall fusion. A beam at one distance from the weld may catch one fusion line and miss the other, so the probe has to scan from more than one distance.
• Flaws in coarse-grained metal. Grain boundaries scatter sound. Austenitic stainless and nickel alloy welds are hard to test for this reason, and lower frequency probes may be used, usually within about 2 to 5 MHz.
• Flaws in thin sections. Echoes crowd together. The UT procedure in your code states the thickness range it covers.
• Flaws near the probe. The area right under the surface can be hidden in the first signals.
• Anything an operator overlooks. Manual UT depends on the person. It has a limited record, and TWI gives its accuracy as about 2 mm for a careful, competent operator.

The scanning surface matters. Loose scale, spatter and loose paint have to be removed, and the probe needs couplant, a gel or liquid that lets sound pass into the steel.

Advanced forms of UT include phased array (PAUT), which steers the beam electronically, and time of flight diffraction (TOFD), which uses the signals from a flaw's tips. Both can give a stored record. Whether either is allowed depends on the code and contract, and they need specially trained people.

Radiographic Testing

Radiographic testing (RT) sends X-rays or gamma rays through the weld onto film or a digital detector. Thin or empty areas let more radiation through and show up on the image. It is excellent for porosity and slag, and it gives a lasting record that someone else can review.

Its weakness is planar flaws. The radiation has to travel along the flaw to show it. So RT can miss:

• Tight cracks. A crack that is nearly closed leaves almost no image.
• Planar flaws at an angle to the beam. Lack of fusion along a bevel may not show unless the beam runs along that bevel. Detection falls as the angle grows.
• Small cracks that get misread. TWI notes that small cracks can look like less serious thread-like flaws.
• Flaw depth. The image is flat, so it does not tell you how deep a flaw lies.

RT also needs access to both sides, which rules it out on many T-joints and corner joints, and fillet welds are a poor fit for it. An image quality indicator (IQI) on the image shows that the technique was sensitive enough. It proves the technique, not the weld.

Warning: Radiation cannot be seen or felt. Never enter a marked radiography area, and follow the barriers and signals. Federal rules in 10 CFR 34.41 require a second qualified person to be present at field sites, watching the operation and ready to stop anyone entering. Radioactive sources are licensed by the Nuclear Regulatory Commission or by your state, and state rules also cover X-ray machines, so the rules vary by location.

Matching the Method to the Job

CJP means complete joint penetration. These are general fits. The code or contract can name a specific method, and that overrides your preference.

Worked Example: Covering the Blind Spots

A 1-1/2 inch double-bevel CJP butt weld in carbon steel sits on a cyclically loaded crane girder. The worry is cracking and lack of fusion, and you can reach both sides.

• RT alone? It would show porosity and slag well, but bevel lack of fusion and tight cracks are the flaws you care most about, and those are its weak spots.
• UT? It is suited to planar flaws in this thickness. It may not catch surface-breaking cracks well.
• MT? It is strong at the surface and near it, and the steel is magnetic. It will not see the middle of the weld.

A good plan, if the specification allows it, is VT first, then MT for surface and near-surface cracks in two directions, then UT for the full volume. The methods cover each other's blind spots. Inspect after the required hold time if there is one, and test before anything gets painted.

Reading an NDT Report Critically

Do not stop at "accept." Check that the report answers these questions:

1. Procedure. Which written procedure and which code edition were used?
2. Person. Who did the work, and at what level? Personnel are commonly qualified under ASNT SNT-TC-1A or CP-189.
3. Calibration. Was the equipment checked against the right reference?
4. Coverage. Which welds, which lengths, and which areas could not be reached?
5. Timing. Was the test done after any required delay or heat treatment, and after any repair?
6. Indications. Are the location, size and type recorded?
7. Result. Which clause or table was the result judged against?

Levels matter. A Level I technician generally works under direction. A Level II technician commonly sets up the test and interprets the results. A Level III technician develops and approves procedures. Your code or contract says which level may do what.

Common Mistakes

• Treating a clean report as proof of no flaws. It only covers what that method can see.
• Using RT to look for lack of fusion on a bevel. It may not show unless the beam lines up.
• Testing MT in one direction only. Cracks along the field can stay hidden.
• Grinding hard just before PT. You can smear the crack shut.
• Choosing MT on austenitic stainless. It is not magnetic.
• Ignoring the coverage notes. "Not tested" is a finding too.

Check Your Understanding

A carbon steel fillet has a suspected crack along its toe. A tech ran MT with the yoke legs placed on either side of the weld, and also along it. Why both?

Flaws show best at 90 degrees to the field. Testing in two directions about 90 degrees apart covers cracks lying either way.

A radiograph of a bevel joint is clean, but a later bend test shows lack of fusion along the bevel. How did RT miss it?

Lack of fusion is planar, and it only shows when the beam runs along it. If the shot was not lined up with that bevel, it can leave little or no image.

Which method would you choose for a surface crack check on an aluminum weld, and why not MT?

PT, because aluminum is not magnetic and MT will not work on it. Clean the surface first without smearing it.

A UT report on a thick weld lists coverage from one side only, with a note that the other side was blocked. What should you do?

Treat the blocked area as untested. Ask whether another scan angle or method can cover it, and raise it to whoever owns the inspection plan.

Key Takeaways

• A clean report covers only what that method can see under those conditions.
• Choose by flaw, location, direction, material, joint and code requirement.
• PT needs an open, clean flaw. MT needs magnetic steel and two directions.
• UT is the stronger test for planar flaws, but it depends on angle, access and operator skill.
• RT is the stronger test for volumetric flaws, and it can miss tight or badly angled planar flaws.
• Combine methods to cover blind spots, and read the coverage notes on every report.