by Quincy Howard and Steve Enzukewich*

 

While heat treatment does not typically affect the velocity of acoustic waves, it definitely can effect the acoustic attenuation, as seen in this month's NDT Solution. The lesson learned is that reference standards with appropriate heat treatment are necessary for some alloy steels to match the complete acoustic response of the material. G.P. Singh Associate Contributing Editor

 

Figures 1-3
Figures 4-6

Introduction
Reference standards are used in ultrasonic testing to calibrate the instrument (screen range, gain setting, etc.) prior to performing an inspection. In the aerospace industry, it is common practice to use reference standards that are specific to the inspection to be performed, as opposed to a general standard like the IIW block. These reference standards are designed to have the same material and geometric properties as the part to be inspected, and are commonly manufactured by an airline operator or a local machine shop. Since the heat treatment condition of the material does not greatly effect the sound velocity it is usually not necessary to specify a heat treatment condition.

 

Problem
Several instances of excessive ultrasonic sound attenuation in materials used to manufacture reference standards have come to our notice at Boeing and through reports from airline customers. The instances have all been alloy steel materials (4330, 4340, etc.) that were being tested at 10 MHz. Figures 1 and 2 are two instrument screen displays of a 10 MHz longitudinal beam transmitted in two apparently identical 76 mm (3 in.) long alloy steel samples. Figure 1 is the expected result with no signals between the initial pulse and the backwall signal at 80 percent of screen width. In Figure 2, however, it is not possible to distinguish which signal is the back surface reflection because of all the noise signals.

Sound attenuation as severe as in Figure 2 makes instrument calibration impossible. Even less severe sound attenuation is a concern as it can cause very high instrument gain settings which in turn can lead to false calls during the inspection. Alloy steel airplane parts, which are fully heat treated, do not have these noisy characteristics. A reference standard with these high noise characteristics, therefore, does not truly represent the characteristics of the parts to be inspected.

---

Another way to limit the amount of sound attenuation in reference standards is to match them acoustically to the part to be inspected.

---

 

Possible Solutions
One solution to the situation above is to require that alloy steel reference standards be heat treated to the "on-airplane" condition. A full heat treatment would match the reference standard's metallurgical characteristics with the part to be inspected and thus prevent any excessive sound attenuation. This treatment consists of two high temperature heating steps (austenitizing and tempering) separated by a rapid cooling (quenching) step. Full heat treatment, however, is costly and many small machine shops do not have the facilities necessary for proper heat treatment of alloy steels. Another difficulty with a full heat treatment is the possibility of geometric changes caused by the quenching step.

Another way to limit the amount of sound attenuation in reference standards is to match them acoustically to the part to be inspected. Two simpler heat treatments exist for alloy steels: normalizing and subcritical annealing. Although these two heat treatments will not match the exact microstructure of a fully heat treated part, they can produce a material that has similar ultrasonic results.

These two heat treatments are used together to prepare alloy steels for machining. Normalizing is used to homogenize the part material (i.e., to create a uniform microstructure) before it is annealed. Normalizing takes a material at some unknown microstructural state and brings it to a known starting condition. The annealing step is done after normalizing to soften the material prior to machining. It is called a subcritical anneal when it is performed at temperatures below the point at which austenite forms for the particular steel alloy. These two heat treatments offer an advantage over a full heat treatment because they involve only one heat cycle, and the cooling is accomplished in air as opposed to rapid quenching.

The important question is whether either (or both) of these heat treatments will produce a microstructure that is similar to a fully heat treated part in its ultrasonic characteristics. In the past, we have used normalizing of some reference standards in instances where excessive sound attenuation difficulties have occurred, but a direct comparison to a fully heat treated part (or subcritical annealed part) has not been performed.

 

Experiment
An experiment was designed to compare these three different heat treatments. We exposed 24 samples of "noisy" alloy steel material to one of three different heat treatments (a full heat treatment, normalizing, or a subcritical anneal). The samples were taken from three different developmental reference standards. The heat treatments were performed according to Boeing process specifications in a vacuum furnace to prevent scaling of the surface. Some of the fully heat treated samples were tempered to the 1,030-1,170 MPa (150-170 ksi) range and the others to the 1,520-1,650 MPa (220-240 ksi) range to get a cross section of fully heat treated values.

The relative sound attenuation was measured before and after the heat treatments by comparing the amount of instrument gain needed to get an 80 percent of full screen height backwall signal. In many of the samples it was not possible to distinguish the backwall signal in the initial state because of the excessive noise signals. In these cases the gain settings were estimated based on the known screen range and the expected location of the backwall signal.

The same ultrasonic flaw detector, transducers, and instrument settings were used for all of the readings. Gain settings on all of the samples were taken by the same two individuals and averaged. Gain settings for each sample were taken at both 10 MHz and 5 MHz. A 3.2 mm (0.125 in.) diameter element, longitudinal wave transducer was used for the 10 MHz readings. This transducer was selected because it is used in one of the inspections where difficulties with sound attenuation were first noticed. A 6.25 mm (0.25 in.) diameter element, longitudinal wave transducer was used for the 5 MHz readings.

Three of the samples were also sectioned and examined metallurgically to determine what caused the excessive sound attenuation and what the heat treatments did to reduce it. These samples were all taken from the same original piece so that microstructure comparisons would be of the same original material.

 

Results
Prior to any heat treatment, all of the samples at 10 MHz showed excessive sound attenuation. In all cases it was very difficult to determine with certainty which signal was the back surface reflection. Figure 3 is an example of a typical screen display. It took an average of 88 dB of instrument gain to get an 80 percent of full screen height signal. At 5 MHz, however, it took an average of only 63 dB to get the same signal. The 5 MHz signals were very "clean" with no excess noise signals present.

As expected, the fully heat treated samples showed the most improvement from the original samples. Figures 3 and 4 show the same sample at 10 MHz before and after the full heat treatment respectively. The full heat treatment removed all of the excess noise as can be seen in Figure 4 and made distinguishing the backwall signal very straightforward. The improvement in the samples at 10 MHz ranged from 32.5-40 dB after the full heat treatment was performed. There was very little change in the sound attenuation of the samples when tested at 5 MHz. Because of the relatively low dB starting point at 5 MHz, the samples improved only 4.3-8 dB after the full heat treatment.

The samples that were normalized showed improvements very similar to the fully heat treated samples both at 10 MHz and 5 MHz. The samples improved 24.9-33.2 dB at 10 MHz after normalizing. Figures 5 and 6 show one of the samples at 10 MHz before and after the normalizing, respectively. The screen display shown in Figure 6 is very similar to that for the fully heat treated sample shown in Figure 4.

The subcritical annealed samples showed little or no improvement at either 10 MHz or 5 MHz. In the case of the 5 MHz readings, more gain was needed after the subcritical anneal to get the same 80 percent screen height signal. The relatively low temperatures of the subcritical annealing process did not make a significant enough microstructural change to affect the sound attenuation readings.

Table 1 summarizes the sound attenuation results for all of the samples at both frequencies. The greatest improvement can be seen in the 10 MHz readings; little change can be seen at 5 MHz. On average it took 8 dB more at 10 MHz to get the same signal in the normalized samples than it did in the fully heat treated samples. In both cases the signals were very clean, with no noise signals between the initial pulse and the backwall signal.

Three samples, representing the untreated condition, the fully heat treated condition, and the normalized condition were cross-sectioned, etched, and examined under 250 magnification. The untreated sample showed a microstructure typical of an as-cast condition (no subsequent mechanical or thermal treatments) which is very inhomogenous in both grain shape and size. There were several very large areas of pearlite (a mixture of ferrite and plate-like carbide) present in the microstructure. At high ultrasonic test frequencies (and therefore short wave lengths), these areas, if they are sufficiently large, can become reflectors of the sound. This explains why the noise signals were present at 10 MHz and not present at 5 MHz. The presence of these relatively large pearlite areas and the general inhomogenous condition of the material at the grain boundaries both could have contributed to the high sound attenuation seen in these samples.

The fully heat treated sample showed a very fine and homogenous grain structure. The uniformity of the grain structure and the very small grain size would account for the relative ease in which the sound traveled through the samples. The normalized sample showed a very homogenous grain structure as well, but with grains many times larger than the fully heat treated sample. Even though the grains were larger than those of the heat treated sample, they were made up of ferrite (pure iron) and not the plate-like pearlite. A quick test was made at 15 MHz to see if the larger grain size of the normalized samples had any sound attenuation effect at a higher frequency. There was no noticeable difference between the normalized and fully heat treated samples which indicated the larger grain size of the normalized samples did not affect the sound transmission.

 

Table 1 Comparison of experimental results

	Samples at 10 MHz	Samples at 5 MHz
Full heat treatment	Average of 37 dB improvement	Average of 6 dB improvement
Normalized	Average of 29 dB improvement	Average of 4 dB improvement after normalizing
Subcritical anneal	Little to no effect	Slightly negative effect

 

Conclusions
Heat treating an alloy steel ultrasonic reference standard to the "on airplane" condition is the best way to match the standard to the part. Reference standards in this condition best represent the part to be inspected, both from a metallurgical standpoint and an ultrasonic standpoint.

The tests described in this paper showed that normalizing alloy steel reference standards is a simple and relatively inexpensive way to match the acoustic characteristics of alloy steel reference standards to those of a fully heat treated part. Even though a normalized microstructure is not the same as a fully heat treated microstructure, normalizing will remove unwanted noise signals and requires only one short heating cycle followed by cooling in air. Although gain settings appear to be slightly higher with normalized reference standards compared to fully heat treated ones, the gain settings will not be high enough to cause excessive noise signals or false calls.

The tests showed that a subcritical anneal does not improve noise signals in alloy steel reference standards. The temperature for subcritical annealing is too low to change the microstructure enough to remove unwanted noise signals.

The tests also showed that the excessive sound attenuation in these samples was very dependent on the instrument/transducer frequency. In the longitudinal mode, normalizing was not necessary when inspecting the samples at 5 MHz. (Because of the slower sound velocity of shear and surfaces waves, and the subsequent shorter wave lengths, sound attenuation may be a difficulty when inspecting at 5 MHz in these modes.) Other alloy steel materials may have other microstructural conditions where 5 MHz longitudinal wave inspections encounter excessive sound attenuation.

 

Acknowledgments
The authors thank Brian Pearce for the "noisy" test piece materials, Norm Munsey for the heat treating, and Luther Gammon for the metallography.

 

* The Boeing Company, (425) 342-3164; fax (425) 342-0665; e-mail quincy.howard@boeing.com.

Copyright © 1997 by the American Society for Nondestructive Testing, Inc. All rights reserved.

[ Materials Evaluation ]