Figure 1

The problem had been solved. The test method had been developed. The equipment had been installed. The test had passed its manufacturing feasibility test and had been working successfully for months. Then a failure of catastrophic proportions was noticed. What happened? What should be done?

To be more specific, steel rod was being procured to be cut into blanks to use in the extrusion forming of chemical shock absorber tubes at a supplier to an automobile manufacturer. The rod was known to have seams introduced in its own manufacturing process. To eliminate most of the seamed material, it was determined that the rod stock should be turned down by a certain amount before the extrusion. The extrusion operation involved a plunger to pierce a hole down the centerline of each blank almost to the end of the blank. The plunger operation expanded and lengthened the blank. If the seam were still present, the seam would turn into a crack because of the expansion.

A crack would be catastrophic to the survivability of the chemical shock absorber tube. A chemical shock absorber is the shock absorber mounted behind the 8 km/h (5 mph) bumper of an automobile (see Figure 1). Two behind each bumper absorb all the energy in the first 8 km/h (5 mph) of impact. A crack would produce a rupture of the cylinder during the crash, negating the 8 km/h (5 mph) safety factor. Government regulations would be involved, as well as lawsuits, etc.

---

Bingo! It was obvious!

---

Because some blanks came from segments of the long rods with seams too deep to be totally removed by the lathe turning, the parts made from these blanks had to be sorted out and discarded. The cracks were tightly shut by the extrusion operation and could not be detected visually. An NDT test was required. A reputable vendor of NDT developed an eddy current test to find the cracks. The eddy current probe was a pair of D shaped coils mounted back to back in a cylindrical housing and connected as a differential sensor. One coil was energized with the input current, and the other coil picked up a signal if something changed between the two coils. The coils were positioned near the cylindrical parts with the narrow space between the Ds parallel to the axis of the cylinders and hence parallel to the sought after cracks. The cylinders were spun on their axes, and the probes were traversed from one end of the cylinder to the other. The whole test mechanism was mounted on a carousel on which the parts were undergoing several processes. As a cylinder spun on its axis, the double D coils would have a spike output when a crack passed under the space between the Ds.

This test was shown to have technical feasibility in the vendor's laboratory and then shown to have manufacturing feasibility by detecting all the cracks from among many blanks in a blind test of cracked and sound parts. The test equipment was mounted on the carousel in the parts vendor's plan and used successfully for months. No cracks were missed. All anomalies were found.

Then the roof fell in. A failure of a chemical shock absorber tube occurred. We in the NDT development group at the automobile manufacturer were informed of the failure immediately because the eddy current NDT equipment apparently did not catch the fault.

The failure occurred in the paint shop in an automobile assembly plant. The failure was within a long tank in the rustproofing primer process. The liquid in the tank into which an entire car body was dipped was run at an elevated temperature. At that point, the body in white was fully assembled with all its metal body parts except for the chrome bumpers. The chemical shock absorber tubes and bumper reinforcing bars were installed. (Of course there were no glass, engines, tires, axles, batteries, upholstery, etc.)

The failure was as follows: One chemical shock absorber tube with an undetected crack ruptured because of the heat of the surrounding liquid in the tank. The silicone gel shock absorbing material filling the tube expanded and blew the tube open. The silicone gel spread throughout the thousands of gallons of liquid in the tank, contaminating the liquid, the giant tank, and all the bodies in white which passed through it until it was discovered at a later painting stage that the paint was not sticking. Production had to be decontaminated. The solution (enough to float a small yacht) had to be junked, and the tank had to be decontaminated. Production could not be resumed until the NDT instrument was fixed.

Despite the fact that the instrument was not of our design, my NDT group was called in as the resident experts to troubleshoot the problem. A team traveled to the parts vendor's plant to investigate. We found that the eddy current instrument was working, the double D probe was working, and the system was capable of detecting the anomalies in the standard anomalous parts kept to calibrate it.

So what had happened? Had an operator set the sensitivity too low one day? Had there been an intermittent (perhaps one time) malfunction? Had one part with an anomaly too small for the probability of detection gotten through? Had an operator or a janitor inadvertently put a rejected part into a bin of good parts?

We looked closer at the probe. It was encapsulated in a perfectly smooth circular cylinder casing. This casing was then held in the structure of the carousel by a setscrew in the wall of a perfectly smooth cylindrical tube.

Bingo! It was obvious! The setscrew had become loose, and the probe had rotated 90 degrees because of the twist of the wires or some other force. In that orientation, the probe could detect nothing. Someone had returned it to the correct orientation before our visit so the test was working when we arrived. But the damage had been done to the tank and the car bodies.

The solution to the problem was obvious, too, as was the blame. There should have been either a flat or a spline on the sidewall of the cylindrical probe to define the direction of the Ds. The mounting tube should have accepted the flat or the spline only in the correct orientation. The mounting of the probe should have been unique with the backs of the Ds parallel to the length of the cracks. The lack of the flat or the spline permitted the probe to rotate 360 degrees and ruin the test at a random, unknown time.

We ordered the fix, which was to machine a flat on the probe OD so that the setscrew would hold the probe in the proper orientation. After the fix, no more failures ever occurred.

A thorough failure mode and effect analysis on the part of the NDT vendor would have indicated that the flat or the spline should have been specified in the original design. As in the old poem, "for the sake of a nail the shoe was lost, for the sake of the shoe the horse was lost, for the sake of the horse the rider was lost, for the sake of the rider the message was lost, for the sake of the message the battle was lost …"

*  Quality Systems Concepts, Inc., 379 Diem Woods Dr., New Holland, PA 17557-8800; (717) 355-9809; fax (717) 355-9812; papadakis@desupernet.net.

Copyright © 2001 by the American Society for Nondestructive Testing, Inc. All rights reserved.

[ Materials Evaluation ]

IRRSP, NDT Handbook, The NDT Technician and www.asnt.org are trademarks of the American Society for Nondestructive Testing, Inc. ACCP, ASNT, Level III Study Guide, Materials Evaluation, Nondestructive Testing Handbook, Research in Nondestructive Evaluation and RNDE are registered trademarks of the American Society for Nondestructive Testing, Inc. ASNT exists to create a safer world by promoting the profession and technologies of nondestructive testing.