Modeling for the Improvement of Dissimilar Weld Inspection

Ultrasonic inspection of dissimilar welds is still a challenge due to the different acoustic impedances across the interfaces between parent metals, welded, butter or cladded regions. Typically parent metals are fine grained carbon/ferritic/20MnMoNi55 steel, which show isotropic character. Whereas weld/butter/cladded metals are INCONEL/ austenitic steel, which exhibit anisotropic character, attributed to their columnar grained texture resulting from the high thermal inputs during welding process. The ultrasound wave propagation through such anisotropic materials is not straightforward due to the skewing of energy flow direction with respect to the direction of phase velocity. Therefore the analysis of the experimental data necessitates many years of experience. A simple theoretical model based on ray-tracing concepts can help to optimize the ultrasonic experimental procedures for the detection of transverse defects (defects with their orientation perpendicular to the welding direction) at the inner surface of the pipe in the circumferential weld. In this presentation a comparison between theoretical and experimental results will be described Introduction: Weld joining dissimilar materials with different chemical and mechanical properties is a challenge for nondestructive testing due to the variation of physical properties in the examination region. If the viewpoint is limited only to the application of ultrasonic testing the change in the acoustical impedance along the sound path is the most important influence. Fig 1 shows an example of a dissimilar weld at a nozzle. In nuclear power plants such welds are employed among other things as austenitic and/or inconel joints between austenitic pipes (material: 1.4550 or 1.4551) and the ferritic cladded nozzle of a reactor pressure vessels. The motivation for the present investigations is due to the fact that such welds are sensitive to stress corrosion cracking. In some nuclear power stations abroad, cracks were detected on the pipe inner surface of the dissimilar weld especially if inconel material was in contact with the cooling water. These cracks due to their orientation perpendicular to the weld direction (weld di rection circumferential) are designated as transverse cracks or axial oriented cracks. A typical microstructure of such materials combination is illustrated in Fig. 1 [1-3]. Clearly visible in the micrograph are the interface areas between the different materials where a sound interaction must be assumed. In some earlier publications by this author the examination of dissimilar welds are described where longitudinal cracks are in the focus of interest. Due to recent evidence on such newer defects on dissimilar welds or inconel welds we have focussed our activities on the detection of transverse cracks [4]. Therefore, in the present contribution only such oriented cracks are under examination. Optimisation of Inspection Metho The difference between the inspection of longitudinal cracks as written in some papers [510] and those of transverse cracks is mainly given by the surface condition of the weld region [1114]. By the inspection of longitudinal cracks starting from the interior surface, the probe position is at a offset distance from the weld e.g. a proper coupling of the probe is guaranteed. Otherwise the examination for transverse cracks regards the coupling direct on the weld cover but it is well known that in this region the surFig 1: Dissimilar weld – joints and interface structures face has a waviness due to the shrinkage between the weld and the base metal. Good coupling conditions for the ultrasonic transducers are found only in the vicinity of the weld and thus on the base metal surface. In the following the basic steps for the optimisation of the ultrasonic examination method regarding the above mentioned conditions is described. For the detection of transverse cracks in dissimilar welds different probe arrangements and types are applicable under the utilisation of longitudinal and shear waves. Conventional probes as well as phased array probes for pulse echo or transmitter receiver operation are, in general, applicable. Due to the fact that such welds are often used to weld pipes together the probes must be adapted to the curvature of the pipe. It is understandable that for the development and optimisation of a new technique on a number of test specimens the experimental investigation time in addition to expenditure increases. In order to reduce the experimental time a complimentary theoretical model based on the ray tracing principle can be used to confirm the influence of some inspection parameters on sound propagation in the anisotropic region of dissimilar weld. Test specimen for the Experiments A number of test specimens with transverse cracks in dissimilar welds are available for the optimisation of an ultrasonic technique. Fig. 2: Specimen with dissimilar weld (Tk-E nozzle) One such test piece is shown in Fig. 2 with an inconel 182 weld. The buttering has a thickness of approximately 10 mm. and the interface between butter and carbon steel is perpendicularly orientated to the inner surface, similar to the micrograph in Fig. 1. The defect situation is pictured in the schematic drawing of Fig. 2. As shown in the figure the spark eroded notches have a depth extension between 2 mm and 8 mm and the position is in the weld and butter as well as in the austenitic or ferritic base metal. The extension in length is10, 20 and 26mm. Modeling of the inspection method The theoretical model used in the present section is based on the ray tracing method. The principles of the method are described in [15/16]. Other investigations using theoretical models for the optimisation of inspection techniques and for the interactions of the sound field with the structure of dissimilar welds are described in [17-23]. The following assumptions were made for the calculations using the method described in [15/16]. • The grain texture is simulated by an empirical formula, which is not based on the principles of solidification mechanics • A layback angle of 10° is assumed in the welding direction • Elastic constants for austenite and Inconel 182 are taken from the literature and must be therefore not exactly in accordance with those of the testblock • Multi reflection on the grain boundaries and sound attenuation are not considered • Planar waves are further bases of model Fig 3: “V” arrangement of two transducers Fig. 4: Coordinate system (top view) Due to the fact of waviness at the surface in the welded region the surface is hand grinded and a V-arrangement of the ultrasonic probes as demonstrated in Fig. 3 were chosen. The advantage of such an arrangement is clearly visible because both of the probes are coupled on the base metal of the austenite or the carbon steel. With the definition of he azimuth angle Φ as the angle between the weld middle line and the main beam (fig. 4) the optimisation of the probe coupling position for the detection of transverse cracks will be explained. Fig 5 shows the transmitter coupled on the austenite base metal at an x-position of 20 mm (for the definition of the coordinate system see Fig. 4). Fig. 5: Ultrasound propagation in an Inconel dissimilar weld The angle of incidence for the longitudinal wave in that particular case was 40°, the azimuth angle 30° and the wall thickness of the test specimen 47 mm (OD 478 mm). The example in Fig. 5 shows that the sound field (nine incidence beams were calculated) impacts the assumed defect area. Also clearly visible is the divergence sound field reflected at the defect but the sound field direction (x and y direction) is not distinguishable from the figure. This information can be estimated as in Figure 6 where the top view is presented. Viewing in the direction of the reflected sound field the receiver must be at a position in opposite of the transmitter coupled on the carbon steel surface. To make sure that the receiver probe has the right “receiving” angle the side view presented in Fig. 7 is very helpful. Fig. 6: Ultrasound propagation in an Inconel dissimilar weld (top view) Fig. 7: Ultrasound propagation in an Inconel dissimilar weld (side view) With the help of the theoretical model the probe arrangement can be optimised and the experimental investigation times can be reduced. The results of the calculation gives the angle of incidence for the transmitter probe, the position of this probe as well as the receiver probe on the surface, the azimuth angle and the “receiving” angle of the receiver probe. The reliability of the calculated results because are strongly dependend on the input of the mechanical parameters into the model. In table 1 the parameters used for the optimisation of ultrasonic methods for the detection of transverse defects in dissimilar welds are listed. In the next chapter the comparison between experimental and theoretical results are described. Comparison of the results between experiments and model The testblock Tk-E nozzle was used for the comparison between the theoretical and experimental results. This specimen (see Fig. 2) has a wall thickness of 35 mm, an outer diameter in the weld region of 220 mm and four transversal defects with depth extensions of 2, 5 and 8 mm at the inner surface. The length extensions of the 2 mm deep notches are 20 mm and 26 mm whereas the location of the longer notch is between the butter and the 6 CrNi 18 11 Inconel 182 [10 N/mm] C11 2,4110 2,78 C12 0,96920 1,15 C13 1,3803 1,3889 C44 2,4012 2,5376 C66 1,1229 1,06 103Kg/m3 ρ 7,82 8,61 Table 1: Used parameter Fig. 8: Asymmetric probe arrangement austenitic base metal and the shorter notch between the weld centre line and the butter. An asymmetrical “V” form arrangement of two transducers with 35° angle of incidence is shown in a drawing in Fig. 8. An asymmetrical arrangement means that the transducers have different positions at the circumference (ydirection). In Fig. 8 the deference between the two transducers is 10.5 mm. One of the experimental results is presented in Fig. 9. There are four clear indications visible in the time displacement image (TD); two of the indications –not marked – are from the disconnections (the specimen was made from two cylindrical half shells) of the test specimen. The other indications were generated from the 2, 5 and 8mm deep notches. With the help of the transducer positions on the coupling surface the physical effective angle of incidence can be estimated to 39°. While the signal to noise ratio (SNR) of the 2 mm deep notch was 3 dB (to small for a mechanized inspection of the weld), so for the 5mm and the 8 mm deep notches a SNR of 13 dB and 17 dB was meas ured. The scanning direction for this measured example was counterclock48 46 44