Many types of critical equipment are subject to nondestructive testing methods, which are specified as part of a program to reduce the risk of unexpected structural failure.

For example, in power plants, the steam lines may be subject to ultrasonic wall-thickness testing. The latest methods are even capable of getting information about both the remaining steel thickness and the thickness of the oxide layer built up on the inaccessible inside of the line. Magnetic particle inspection or dye penetrant may be used to make cracks on visible surfaces easier to see. And nondestructive metallography by replica inspection may be used to determine whether the mechanical properties of the steel are still in an acceptable range or whether long-time exposure to high temperatures has caused excessive degradation.

In metal-forming operations or plastic molding operations, tie rods, hydraulic cylinders, giant screws or other highly stressed components may be subjected to ultrasonic inspection for the presence of transverse cracks, for example. This is usually done primarily after a painful failure of a component that caused lost production time and costly repair or replacement issues.

Years ago, I was asked to look at a broken fragment from a large-diameter screw from a large stamping press. The company had drilled a hole in the center of the fracture face in order to get a specimen to check the composition. They were disappointed in the answer: The material met the spec! Now what? The crack was clearly a fatigue crack and had initiated from two locations approximately 180 degrees apart. Detailed study of the press history revealed that the crack likely took almost three years to propagate to final separation. So, this left time for some sort of nondestructive inspection of the multiple remaining screws in service in five or six similar presses all over the country. Ultrasonic testing was selected as the method of choice.

Lo and behold, one of the screws tested was found to have a crack! The screw, reported to cost about $40,000, was “sacrificed,” even though the crack was MUCH shorter than the one that caused the separation event. The client gave me a stub within which the “crack” was found and asked me to examine the crack and determine the cause. So, instead of a 16-foot-long piece, I had a stub about 10 inches long. I took the stub to an NDT house and watched the technician re-inspect the part. There was an “indication” in the approximate center of the cut length.

For those of you who are not familiar with NDT methods, the work “indication” has always seemed to me to fit into the following rough analogy: Indication is to Discontinuity as Insurance Coverage is to Insurance Company Payment.

What does that mean? Well, if we buy an insurance policy and they tell us X, Y or Z condition or service is “COVERED,” that’s great. But what we REALLY should be asking is whether they will pay for the covered service! Likewise, the technician finds an “INDICATION” when using some nondestructive test, so sometimes people jump to the conclusion that there is an unacceptable discontinuity, such as a crack. This may or may not be the case.

After much work on that screw stub, including MULTIPLE nondestructive method inspections followed by destructive metallography, NO CRACK WAS FOUND. Instead, there was a small colony of stringer inclusions slightly subsurface to the thread root. This colony of stringers would probably NEVER have been the main factor in a crack initiation event.

In other catastrophic events, indications from inspections were not even noticed by the inspector, and equipment was allowed to operate that was presumed in good condition and failed unexpectedly. The presence of dye-penetrant residue on a crack surface that is being evaluated after a catastrophic event generally leads to much discussion about why the inspector did not see the discontinuity if it was big enough to collect the penetrant!

These situations, and some other similar situations that I have encountered over the years, led me to meditate on what factors lead us down the path of either assuming a component is bad because it has an indication or assuming it is good because it does not. In the end, we have to realize that INDICATIONS are INDICATIONS OF SOMETHING, not necessarily a crack or void. It may be a surface lap, a burr, a chemical in homogeneity, a deposit on the surface or many other features that we can’t determine nondestructively. Likewise, just because one method reveals NO indication, it does not prove that the component is intact.