Table of Contents ECNDT '98
Session: Automative Industry
Nondestructive Characterization of State and Properties of Aluminium StructuresE. Schneider
Fraunhofer Institut Zerstörungsfreie Prüfverfahren (IZFP), Saarbrücken
|TABLE OF CONTENTS|
The contribution gives a short overview of IZFP investigations towards the process integrated characterization of microstructure and evaluation of mechanical properties of Al structures. The propagation velocities of ultrasonic waves allow the evaluation of the elastic constants of the alloy and good correlations are found between the acousto-elastic constant of different alloys and their yield strengths. The tensile strength is evaluated using an eddy current technique. Using the times-of-flight of different ultrasonic waves, the residual stress states in plates and welded parts are evaluated. The texture in Al-sheets is determined and the earing parameter is characterized using measurements of ultrasonic times-of-flight as function of the ultrasonic propagation direction. Ultrasonic backscattering technique is applied to visualize the distribution of ceramic reinforcements in automotive pistons and conrods made of Al-based metal matrix composites. The volume content on reinforcement is evaluated from the ultra-sonic velocities.
It is common practice to evaluate the elastic constants Young`s- and shear moduli using the material density and the propagation velocities of ultrasonic longitudinal and shear waves. The evaluated constants are generally found in a good agreement with those determined by tensile testing. Ultrasonic velocity meaurements can be successfully used to detect areas of a compo-nent with microstructural variations caused e.g. by pores, damages or plastic deformations. Since the velocities can be determined with an accuracy better than 0.1%, those changes of the microstructure which are influencing the elastic behaviour in that order of magnitude can be detected and characterized nondestructively (Schneider et al 1992; Dobmann et al 1995).
The above mentioned elastic constants describe the elastic behaviour in first approximation. A more precise description of the real elastic properties of a material is possible using third order elastic constants. Acousto-elastic constants finally are combinations of the elastic constants Young`s, shear moduli and the third order constants. As expected, these acousto-elastic constants are more sensitive to elastic properties than the moduli. In experimental investiga-tions good correlations are found between the acousto-elastic constants and the percentages on solid solution phase of three work hardening and two heat treatable Al-alloys. Furthermore, linear correlations are found between the acousto-elastic constants and the yield strengths of the mentioned alloys as to be seen in Figure 1 (Schneider et al 1985).
Fig. 1 The acoustoelastic constant AEC33 of a longitudinal wave propagating perpendicular to the load for work hardening and precipitation hardening Al-Alloys with different content on solid solution phase (left) and with different yield strengths (right)
Similar investigations dealt with the characterization of the elastic properties of Al-based metal matrix composites with different types and volume percentages of reinforcements (Lee et al 1989). The influence of fatigue damages on the third order elastic constants and on the acousto-elastic constants is shown by Wahnschaffe (1994). Besides the elastic constants also their temperature dependencies are found to be suitable quantities to characterize micro-structural states (Mohrbacher 1991).
A two frequency eddy current device has been developed for the in process evaluation of the tensile strenght of Al air bag components and for the simultaneous detection of surface defects (Becker et al 1996). In Figure 2 the nondestructively evaluated strength values are shown in comparison with the tensile test results.
|Fig. 2: Nondestructively evaluated strength values of Al air bag components in comparision with the tensile test results|
|Fig. 3: The principal stress acting perpendicular to the weld seam|
In Figure 4 the profiles of the principal stress acting parallel to the weld along two traces perpendicular to the weld seam is displayed. Comparisons with the results of an established technique demonstrate the reliable applicability of ultrasonic techniques (Arenz 1996, Schneider and Arenz 1996).
|Fig. 4: Change of residual stress parallel (l ) to the weld seam along two traces across the weld in an aluminium sheet|
Both, the residual stress state and the texture of sheets are changed by the welding process. And it is obvious that a smooth and homogeneous change of both properties along a trace perpendicular to the weld is better with respect to the dynamic behaviour of the welded part as inhomogeneous changes with strong gradients. Stress and texture influence the propagation velocities and the direction dependence of ultrasonic waves. A measure for this elastic aniso-tropy is the relative change of the times-of-flight of a linearely polarized shear wave. Measure-ments are made at different orientations of the shear wave vibration with respect to the length direction of the weld. The largest difference of the times-of-flight is found if the vibration is parallel and perpendicular to the weld. It is found by Mourik (1997) that the relative change of the time-of-flight of the two shear waves polarized along the principal directions correlates with the welding energy. The difference of the times-of flight as well as the lateral expansion of the area around the weld with significant time-of-flight differences increase with the welding energy as shown in the Figures 5 a and b.
|Fig. 5a: The relative difference of the times-of-flight of a shear wave polarized along the principal directions of welded Al sheets||Fig. 5b: Lateral expansion of the areas around the welds with significant values of the relative time-of-flight difference|
Regarding the aluminium beverage can production, the economical significance of the right texture becomes obvious: The two predominant crystallographic textures that exist in the aluminium sheet need to be balanced. One kind of texture arises during annealing of the alloy after the strip is hot-rolled from the ingots. It causes four ears to appear every 90 degrees. The second kind results from cold rolling the sheet producing four ears also. Proper control of annealing and rolling can lead to a combination of the two textures such that ears caused by one fill the valleys caused by the other process. The result is eight very low ears and a reduc-tion of the amount of aluminum needed. Up to now, the texture and/or the earing behavior is checked by destructive testing. Samples of the production are usually cut from the head and the tail of the strip. Slight changes of the processing parameters can cause changes of the texture in the head and tail part of a roll and hence, earing values are evaluated which are not representative for the quality of the strip.
The interdependences between the crystallographic texture and the propagation velocities of ultrasonic waves are known and have been widely used and published as are the inter-dependences between the texture and the earing behavior. The ultrasonic SH-wave was found to be the most suitable ultrasonic mode to characterize the texture. The major advantage of using a SH-wave is the possibility to keep a distance between the ultrasonic sensor and the sur-face of the strip in order to prevent any surface damage. The distance between the ultrasonic transmitter and receiver is usually kept constant. Hence, the ultrasonic time-of-flight is the only measuring quantity. Using our prototype system, the reproducibility of the time-of-flight measurement is better than 0.3 . Measurements were performed at the positions 50%, 75% and 95% of the width of the strip in order to check on the homogeneity of texture. The absolute time-of-flight data are taken at stepwise changes of the ultrasonic propagation direction with respect to the rolling direction. As example, the relative change of the ultrasonic time-of-flight with the propagation direction of the SH-wave is displayed in Figure 6. The difference of the extreme values of the relative time-of-flight data is found to be in a very satisfying correlation with the earing parameters as shown in Figure 7 (Schneider et al 1994).
Fig. 6: Relative change of the ultrasonic time-of-flight versus the angle between the sound propagation and the rolling direction of a hot rolled aluminium sheet
Fig 7: Correlation between the ultrasonic quantity and the earing parameter of hot rolled aluminium sheets
The earing parameters were evaluated by the manufacturer of the strips using the established technique (Schneider et.al.1994) A similar ultrasonic system for on line texture analysis and characterization of deep drawability of ferritic steel strips is already in industrial use (Borsutzki et.al.1993).
The density of MMC components can be calculated from the densities of the two constituents and their volume or weight percentage using the law of mixture. The Young's- and shear moduli do not follow the law of mixture. Calculations using the Voigt or Reuss model result in the extreme values; the average of both (approximately the Hill model) usually fits with experimental results. The change of the elastic constants with the volume percentage of short fibers are calculated for the above mentioned MMC. Using the calculated dependencies of the Young's and shear moduli on the volume percentage of short fiber reinforcement and the corresponding dependency of the density, the ultrasonic longitudinal and shear wave velocities in MMC`s are calculated as function of the volume percentage of Al2O3-fibers. The relative changes of the longitudinal and shear wave velocities with the volume percentages are found to be different: 1% change of volume percentage causes about 4.3 change of the longitudinal velocity and a change of about 7.8 of the shear wave velocity. Applying linearly polarized shear waves, the change of the volume percentage on Al2O3 short fibers in piston crowns has been evaluated. The results also show whether the pistons have a planar isotropy or not (Schneider et al 1994).