Paper presented at the 8th ECNDT, Barcelona, June 2002

Abstract

The influence of surface roughness in the amplitude of the ultrasonic signal was evaluated. We used the technique of pulse-echo by direct contact in order to determine the presence of flaw indication.

We studied steel AISI/SAE 4340 samples with roughness of 0,5m, 2.0m, 8.0m and 12m. We also, drilled holes of several diameters, witch act as reference reflectors at frequencies of 2.25, 5.0, 7.5 and 10 MHz. Best indications of reflectors were obtained at the frequencies of 5.0 and 7.5 MHz in steel samples.

It can be seen, from test result, that the attenuation of the signal depends strongly on the surface roughness. Spectral analysis of signals shows that attenuation increases as the surface roughness.

Introduction

The roughness of the surface used to introduce an acoustic beam in a sample material is an important fact to consider in an ultrasonic testing by direct contact. A high surface roughness may reduce the energy transmitted to the sample by the ultrasonic beam, therefore reducing the amplitude of the received signal (1). As a result, the capacity of detection of discontinuity such as cavities, cracks, etc., is reduced.

Actually, the industry produce elements and parts of several degrees of surface finish. The surface condition will depend on the fabrication process. It also may happen that the surface condition of an element is affected during service. The material to be evaluated in this work, steel AISI/SAE 4340, is frequently used in elements and parts that will be under high level of dynamic stress, such as shaft, lever axis, torsion bars, cast for plastic injection, etc. For this reason, we explore the behavior of roughness range until 12 m, a usual range of surface condition in the applications previously stated.

If the surface roughness is less than the wavelength of the sample beam, noise will be generated and the transmitted beam amplitude will be reduced, but the shape of the beam will be kept. In addition, if the variation in the surface is larger than the wavelength of the sample beam the surface condition will distort the beam. These ideas establish that any variation of the surface condition of a sample specimen will generate changes in the ultrasonic beam used in the inspection. Therefore, it is very important to evaluate the changes in the beam due to a series of roughness that may approach that generated in a component in service. In the literature published about this phenomenon some interesting points are highlighted, let us say:

• Effect of surface roughness in the ultrasonic contact test (3,8).
• Attenuation of transmitted and reflected waves on samples of rough surface (4,5,7).
• Measurements of surface roughness using the ultrasonic reflection coefficients (5,6).

Thavasimuthu et al. (3) showed an evaluation of surface roughness in the range between 6 to 50m, and observed that as the surface roughness increases the smaller discontinuity gave a bigger reflectivity at low frequencies, and also larger discontinuity gave lower reflectivity at higher frequencies. In addition, discontinuities of different size showed the same amplitude of the reflected signal. Accordingly, they recommend performing test at other frequencies in order to do an estimation of the right size of the discontinuity.

In our study, we make samples of steel AISI/SAE 4340. This material is of special strength to traction, with god wear and impact resistance. The range of roughness in the samples, obtained by machining processes, is approximately from 0.5m to 12m. Our results allow us to evaluate, throughout the spectral analysis technique, the effect of small roughness in the transmitted beam as well as the effect of different coupling fluids in the technique of ultrasonic testing.

The Experiment

The objective of this work is to study the effect of the surface roughness of steel samples from the behavior of the signal obtained from an ultrasonic contact test.

In this research we prepare the samples with different roughness in the front face and holes drilled of several sizes that acts as reference reflectors.

After the preparation of samples they were inspected by the known method of Contact Pulse Echo. The method consists of the transmission of ultrasonic wave pulses and the later detection of the echoes reflected from a discontinuity or the back wall of the test sample.

The digitalization of the signal was made using an oscilloscope, and the data processing of the signal with MATLAB 5.3. This processing includes the determination of the frequency spectrum of the ultrasonic wave after it has propagated through the test sample.

Samples

Samples were made of steel AISI/SAE 4340 with dimensions 300 x 120 x 20 mm, (11,81 x 4,72 x 0,78 in) following the design of Thavasimuthu, et all. (3)

The roughness of the front surface, of each sample, was achieved by mechanical brushing, while the back wall roughness of the sample was kept constant (0.5m). The values of the roughness of the samples are shown in Table 1.

Sample	SR	R1	R2	R3
Roughness (mm )	0.5	2	8	12
Table 1:

In one side of the sample holes were drilled with diameters of 2, 3, 4 y 5 mm, (0.08, 0.12, 0.16, y 0.2 in.) and a constant depth of 60 mm (2,36 in.). These drills act as reference reflectors. The centers of these drills were separated 60 mm (2,36 in.) each other, as depicted in the Figure Nº. 1

Fig 1: Schematic representation of test specimen (3).

Ultrasonic Inspection

The ultrasonic test was done with an Ultrasonic Flaw Detector KRAUTKRAMER model USN52L, with probes of 2.5, 5.0, 7.5, and 10 MHz and with two different couplings fluids (glycerin, oil SAE 50).

In order to calibrate the system for the sample SR of 20 mm depth, it was chosen an echo in the far field of the sonic field of the probe; this is done to eliminate interference effect. A death weight of 320 gr. was placed on top of the probe to avoid variation of amplitude due to changes in the pressure applied to the probe. We performed five measurements in each sample: one direct to the back wall of the test sample (20 mm depth) and one over each of the reflectors in the test sample.

Data Processing

To obtain the signals we use an oscilloscope Tektronix model TDS3010 of 100 MHz and a sampling speed of 1 Gs/s. Once the signal was digitalized the Fast Fourier Transform (FFT) was performed, allowing us to do the interpretation of the signal in the time and frequency domains. An schematic representation of the experimental set up is shown in the Figure Nº 2.

Fig 2: Schematic representation of experimental setup.

Spectral Analysis

Spectral analysis was used on the echoes obtained from the technique of pulse-echo (9,10,11). This permits us to do an analysis of the pulse transmitted through the sample in the frequency domain. The spectral analysis technique has become a conventional method of characterization of materials (2).

When an ultrasonic pulse propagates throughout the material the wave interacts with the microstructure (grains, inclusions, porosities, and micro cracks). Through these interactions, the size and shape of the microstructure or discontinuity have been quantitatively related to the changes in the spectral components and wavelengths of the incident beam (1). Therefore, allowing us to study the geometrical properties of the test sample and its surface condition in both time and frequency domain.

When determining the wavelength values associated to the each one of the frequencies used we find the following results (Table 2):

F ( MHz)	2.25	5	7.5	10
l(m)	2609	1174	782	590
N(mm)	8.64	19.20	28.80	38.40
Table 2:

According to the results, the roughness values range (0.5-12 m ) in this research is a lot less than the wavelength of the test beam. Due to this fact, the attenuation shown in the signals is not produced only by roughness type (1,2). In the literature studied, we could find that the majority of the evaluated roughness covers great values and because of this, the roughness effect is more significant. Another important aspect, is the fact that the possibility to observe certain reflector sizes is not only determined by the superficial roughness and the probes frequency to be used but by the discontinuity geometry and by incident beam dispersion.

In general as observed in figures 3 and 4, the spectrum produced from the reflection of the back echo are not simple, showing different relative maximum. As the frequency of the probe is incremented, as well as the surface roughness the maximums tend to overlap.

Fig 3: Frequency spectrum of the back wall echo using glycerin as couplant.

Fig 4: Frequency spectrum of the back wall echo using oil as couplant.

Fig 5: Frequency spectrum of the reflector at 7.5 MHz using glycerin as couplant.

Fig 6: Frequency spectrum of the reflector at 7.5 MHz using oil SAE 50 as couplant.

It is very important to observe that for each frequency there is a shift of the maximums towards the low frequency regions. This shift can be explained by influence of the viscosity effect of the couplant on the system.

In the figures 3 and 4, the back echo attenuation increases with roughness. It also increases as the probe frequency increases. A shift toward the low frequency peaks can also be observed.

Comparing figures 3 and 4, it can be observed that using glycerin as couplant the signal is attenuated a lower degree, and this effect is more visible at high frequencies.

The figures 5 and 6 show the effect of roughness in each of the reflectors using glycerin as couplant and SAE-50 oil. They show a shift of frequency spectrum to the left much higher at high frequencies that to lower frequencies and a signal attenuation that increase with the roughness of the front face.

Due to the discontinuity geometry, the beam dispersion is not the same for each of the reflectors. This fact complicates the quantitative analysis of the energy reflected and the reflectivity comparison considering the discontinuity dimension, the frequency and the roughness.

As a result of the facts previously explained, we are preparing reflectors shaped as flat bottom holes which axis are parallel to the incident beam direction. The evaluation of them will lead us to a quantitative analysis of the discontinuities size at a roughness and an established frequency.

Conclusions

• This work show that at the frequencies studied the glycerin is the most appropriate couplant in this range of roughness (0.5-12 m).
• The couplant viscosity generates loses of the signal amplitude due to viscosity effects.
• A clear possibility is that the shift of the peaks is related to the couplant viscosity.
• From our results we can conclude that the determination of the defect size depends on the defect geometry as well as on the front face roughness.

References

1. J. Krautkämer . Ultrasonic Testing of Materials. Springer-Verlag. New York, 1.969.
2. ASM International ASM Handbook, Ultrasonic Inspection, Methods of Nondestructive Evaluation, 231-277.
3. M. Thavasimuthu, C. Rajagopalan, T.JayaKumar T and R. Baldu, “Effect Of Font Surface Roughness On Ultrasonic Contact Testing A Few Practical Observations”. Material Evaluation, April 1998, 1302-1309.
4. P. Nagy and L. Adler,. “ Surface Roughness Induced Attenuation Of Reflected And Transmitted Ultrasonic Waves”. Acoustical Society of America, 1987, pp.193-197.
5. D. Nicoletti and B. Sorli , “ Measurement Of Changes In Surface Roughness Using Ultrasonic Reflection Coefficient”. Review of Progree in Quantitative Nondestructive Evaluation, Vol. 14, 1995, pp.1853-1860..
6. B. Drink , P. Cawley and R. Dwyer-Joyce, “A Study Of The Transmission Of Ultrasound Across Real Rough Solid-Solid Interfaces”. Review of Progress in Quantitative Nondestructive, Evaluation .Vol. 14, 1995, pp. 1829-1836.
7. M. Bilgen and J. Rose. “Effects Of One Dimensional Random Rough Surfaces On Ultrasonic Backscatter Utility Of Phase Screen And Fresnel Approximations”. Acoustical Society of America., 1994, pp.2849-2857.
8. B. Drink and P. Cawley , “ The Effect Of Dust And Surface Roughness On The Efficiency Of Solid Coupled Wheel Probes”. Review of Progress in Quantitative Nondestructive, Evaluation .Vol 15, 1996, pp. 955-962.
9. J. Szilard, Ultrasonic Testing, Non Conventional Testing Techniques. John Wiley & Sons. New York, 1982..
10. J. Bressler, T. Barghese ,N. Bilgutay, “ Spectral Correlation In Ultrasonic Pulse Echo Signal Processing”. IEEE Transaction On Ultrasonic, Ferroelectrics And Frequency Control. Vol. 40, No 4. July 1.993.
11. S.E Kruger, J.M.Rebello and P.C. Camargo. “Hydrogen Damage Detection By Ultrasonic Spectral Analysis”. NDT & International, 1999, pp. 275-281.

© NDT.net - info@ndt.net

|Top|