RESIDUAL STRESSES AND IN-SITU MEASUREMENT OF PHASE TRANSFORMATION IN LOW TRANSFORMATION TEMPERATURE (LTT) WELDING MATERIALS

Transcription

1 755 RESIDUAL STRESSES AND IN-SITU MEASUREMENT OF PHASE TRANSFORMATION IN LOW TRANSFORMATION TEMPERATURE (LTT) WELDING MATERIALS ABSTRACT Thomas Kannengiesser 1, Arne Kromm 1, Michael Rethmeier 1, Jens Gibmeier 2* and Christoph Genzel 3 1 Federal Institute for Materials Research and Testing (BAM), Unter den Eichen 87, Berlin, Germany 2 University of Karlsruhe, Institute for Materials Science and Engineering I Kaiserstrasse 12, Karlsruhe, Germany 3 Hahn-Meitner-Intitute (HMI), c/o Bessy, Albert-Einstein-Straße 15, Berlin, Germany Thomas.Kannengiesser@bam.de The investigations carried out in this study using innovative high strength filler materials with specifically lowered M s /M f -temperatures have demonstrated that different transformation temperatures may significantly affect the welding residual stresses. For this kind of filler materials for the first time high energy synchrotron radiation was applied in order to characterise the effect of lowered martensite start temperatures on the welding residual stresses in-situ by means of energy dispersive diffraction. This way likewise the phase transformation temperatures could be determined. Using synchrotron white beam diffraction the phase-specific residual stresses in the martensitic and the austenitic phase were analysed within the same experiment. Low transformation temperature welding material with varying nickel content between 8 % and 12 % was investigated in this study. Butt welds using the high strength base material S690 applying manual metal arc welding were produced in two passes. Residual stress analysis was carried out in longitudinal as well as transverse direction to the weld line. The results clearly indicate that the residual stress distributions show in general a decrease in magnitude caused by phase transformation of the weld metal. Particularly in the transition region from the weld metal to the heat affected zone relatively high compressive residual stresses up to -350 MPa were determined in the martensitic phase. The stress magnitude and distribution in the austenitic phase is strongly influenced by the nickel content and the amount of associated retained austenite. INTRODUCTION Safety of welded high strength steel structures demands for the precise knowledge of the welding residual stress distributions, which profoundly influence the crack resistance and the service load. The fact that the filler material, apart from constructional design and purposeful heat control, may contribute significantly to residual stress reduction with regard to the development of high strength filler material with correspondingly lowered martensitic transformation temperature (Low Transformation Temperature (LTT) filler material) is reflected e.g. in [1]-[4]. First investigations into the effect of this filler material on the welding residual stresses were recently carried * formerly at Hahn-Meitner-Institute Berlin

2 This document was presented at the Denver X-ray Conference (DXC) on Applications of X-ray Analysis. Sponsored by the International Centre for Diffraction Data (ICDD). This document is provided by ICDD in cooperation with the authors and presenters of the DXC for the express purpose of educating the scientific community. All copyrights for the document are retained by ICDD. Usage is restricted for the purposes of education and scientific research. DXC Website ICDD Website -

3 756 out by [5]-[9]. However, no precise quantifications of residual stress distributions in welded butt joints are available up to the present. The use of high strength filler materials with correspondingly lowered martensitic transformation temperature may finally have positive effects also on the cold cracking resistance, as for example with the help of M f -temperatures below room temperature with respective proportions of retained austenite. EXPERIMENTAL MATERIAL Investigations in this work focused on three selected LTT filler metals with variation of the nickel content between 8 % and 12 %. Table 1 shows the nominal composition of LTT filler material and their M s - temperatures calculated according to [10]. With respect to the intended application for high strength joints, the high strength fine grained structural steel S690Q was used as base material. Table 1: Nominal composition of filler metal and M s - temperatures LTT filler Alloying elements in Wt.% Calculated M s - No. C Ni Cr Mn Si temperatures (after [10]) WELDING PARAMETERS Specimens for residual stress analysis were welded using the parameters shown in Table 2. After finishing the root run, this run was cooled down to room temperature in order to assure complete transformation into martensite. Subsequently, the final run was welded and cooled down under the same conditions. The welds were produced according to Fig. 1 and Fig. 2. Welding process Welding current Welding voltage Travel speed Heat input Table 2: Welding parameters MMAW 87 A 26 V 2 mm/s 1 kj/mm 60 final pass root pass 1.6 Fig. 1: Weld coupon geometry Fig. 2: Joint geometry

4 757 IN-SITU PHASE ANALYSIS The energy dispersive synchrotron diffraction experiments were carried out at the materials science beamline EDDI of the Hahn-Meitner-Institute (HMI) at BESSY, Berlin (Germany) [11]. Here a white beam with usable photon energies in the range between 20 and 150 kev is provided by a 7T-multipole-wiggler. Further details about the experiment layout can be found in [11]. By energy dispersive diffraction always diffraction spectra showing diffracted intensities vs. the available energy range are recorded within one experiment under a fixed diffraction angle, allowing for simultaneous determination of stress and phase composition information for all crystalline phases of a material contributing to the diffraction spectra. In the present study heat cycles were realised in a first step by means of a carbon domed heating stage type DHS 1100 (Anton Paar) using Nitrogen as shielding gas. The heating stage is optimised for the integration in conventional diffractometer systems, whereas the thin-walled carbon dome is almost completely transmissible by the high energy synchrotron radiation. Diffraction analyses were carried out in-situ using a discrete time stepping of 5 s to record the diffraction spectra. The experiment allows for the determination of the martensite start temperature during the cooling cycle for penny shape specimens with a diameter of 10 mm and a thickness of 0.3 mm. The samples were prepared from weld metal produced by multi-layer welding. The thermal cycles applied for heating and cooling are summarised in Table 3. Table 4 shows the parameters used for synchrotron analysis. Neglecting the beam divergence and attenuation due to the absorption of the irradiated material the measuring volume can be determined to approximately mm³. In addition, thermal analysis was carried out for all investigated LTT materials using the SS-DTA technique [12]. The in-situ diffraction studies have to be understood as feasibility studies for in-situ diffraction analysis during real welding using the innovative LTT-filler materials investigated here, since applied heating and cooling cycles are slower compared to weld regions. Table 3: Parameters for heating and cooling cycles Cycle Start temperature Finish temperaure Heating rate 1 ambient 1100 C 500 K/min C ambient 500 K/min Table 4: Parameters for in-situ phase analysis Primary beam cross section 1x1 mm² Absorber 8 cm graphite Optics at the secondary side Double slit system (equatorial axial) 0.03 x 5 mm² Diffraction angle 2θ = 6 Measuring mode Reflexion Measuring time 5 s / spectrum Calibration Tungsten powder RESIDUAL STRESS ANALYSIS Residual stress analysis of the weld coupons was carried out for the ferritic (martensitic) as well as for the austenitic phase, which coexists in the weld zone next to the martensitic phase. Measuring parameters are shown in Table 5. Neglecting again the beam divergence and attenuation due

5 758 to the absorption of the irradiated material the measuring volume can be determined to approximately mm³. Fig. 3 illustrates the measuring range and the measuring directions schematically. The well known sin²ψ -method for X-ray stress analysis [13] was used for determination of residual stresses in longitudinal as well as transverse direction to the weld line. Weld primary beam 10 mm σlong, Φ=0 σ trans, Φ= 90 Fig. 3: Measuring range and directions on specimen Table 5: Parameters for strain analysis Beam cross section (primary beam) 1 x 1 mm² Absorber 2 cm graphite Optics at the secondary side Double slit system (equatorial axial) 0.03 x 5 mm² Diffraction angle 2Θ = 14 Measuring mode Symmetrical Ψ - mode (reflexion), ψ = 0 80 Measuring time / spectrum 100 s Evaluated diffraction lines ferrite: 110, 200, 211, 220, 310, 321, 330/411 austenite: 111, 200, 220, 311, 222, 400, 331, 420 Diffraction elastic constants (DEC) s 1 (hkl) and ½s 2 (hkl) were calculated from the single crystal constants using the Eshelby/Kröner-model Calibration Tungsten powder, measured under similar conditions RESULTS AND DISCUSSION IN-SITU PHASE ANALYSIS Considering the diffraction patterns during cooling of LTT materials, it is remarkable that retransformation into martensite starts at different temperatures for respective diffraction lines as indicated by the 2D-density-plots in Fig. 4. It is obvious that martensite diffraction lines at lower energies, i.e. the 110α diffraction line at a photon energy of 58 kev, appear earlier than corresponding diffraction lines occurring at higher photon energies. An explanation for that phenomenon could be the effect of a small temperature gradient that might occur within the 0.3 mm thick samples which are heated from the bottom side by means of a resistance wire. Diffraction lines occurring at lower photon energies represent surface near penetration depths, whereas diffraction lines occurring at higher photon energies more likely represent the material volume for higher distances to the surface. Due to the slight temperature gradient martensite transformation obviously starts earlier compared to core regions. When transformation continues, further diffraction lines arise at higher photon energies. At a distinct temperature, no further qualitative change in the diffraction spectra can be observed during cooling down to ambient temperature apart from

6 Copyright JCPDS-International Centre for Diffraction Data 2009 ISSN the shifts of the individual diffraction lines. The phase transformation obviously decays at ambient temperature with some amount of retained austenite, indicated by austenite diffraction lines. Fig. 4: Density plots of diffraction spectra during thermal cycle of LTT materials showing austenite to martensite transformation. (a) LTT filler No. 1 with 8 % Ni, (b) LTT filler No. 2 with 10 % Ni, (c) LTT filler No. 3 with 12 % Ni The Ms - temperatures determined from in-situ diffraction analysis compared to thermal analyses (SS-DTA) are summarised in Fig. 5. Fig. 5: Comparison of Ms - temperatures determined by thermal and synchrotron analysis The Ms - temperatures show a linear decrease with increasing nickel content. Results prove that the experimental findings are in good agreement. The measurement uncertainties for the analyses are below 6 C and result from limited time resolution. RESIDUAL STRESS ANALYSIS The residual stresses determined in the following for each case represent phase specific stresses which are the sum of I. order stresses (so called macrostresses ) and the mean values of the homogeneous microstresses of II. order according to σ α = σ I + σ II α. (1) From stress analysis by means of energy dispersive diffraction the diffraction lines can be assigned to different information depth. I principal residual stress depth distribution can be determined that way (see e.g. [11]). For the stress analysis of the welded joints only robust average

7 760 values of the residual stress measures will be presented here. The average stresses for the individual phases are calculated taken into account all measured interference lines of the corresponding phase whilst the residual stress values determined for each {hkl}-lattice plane family are weighted by their multiplicity factor. Since the diffraction information for diffraction lines occurring at higher photon energies stem from depths of up to about 150 µm, a plane stress condition cannot be assumed furthermore. The phase specific residual stresses specified in the following always represent the difference between the surface parallel stress components σ i 11 or σ i 22 and the apparent normal stress component σ i 33, whereas i is assigned to the phase α - martensite or γ - austenite (average values from all measured interference lines of the corresponding phase). The σ 33 stress components can be assumed to be small in case of the present weld geometry. The assessment of potential stress gradients is possible by evaluation of different diffraction lines attributed to different penetration depths (see Genzel et al. [11]). In the following diagrams the phase specific residual stresses of the martensitic and the austenitic phase across the weld are represented for the three investigated LTT welding materials. For visualisation of the results the reasonable assumption was made that the stress distributions are symmetric with respect to the weld centre, hence the residual stress distributions are complemented by mirroring at the centre although measurements were carried out only on one side of the weld. In longitudinal direction, the martensitic phase (Fig. 6) of all three welding materials shows a distribution as is typical for transformation-affected welds. Therefore, transformation-associated compressive residual stresses were found at a distance of 3 mm from the weld centre, which change into low tensile residual stresses at the weld centre that can be assigned to quenching effects. The quenching occurred between the surface and the core regions with an obvious temporally delayed shift of the transformation. This shift might give rise to the development of tensile residual stresses within the surface as compensation for compressive residual stresses being developed in the core. In transverse direction, this phenomenon is even more pronounced, as illustrated by the distributions presented in Fig. 7. The residual stresses of the martensitic phase are levelled more pronounced in transverse direction to the weld than in welding direction in particular in the transition area to the HAZ. In principle, the residual stress distributions in the martensitic phase are qualitatively similar with respect to the chemical composition and, thus, of the M s - temperature. Differences arise particularly in transverse direction at the height of the compressive residual stresses at the boundary to the HAZ and at the tensile residual stress maximum in the weld centre. Here the peak value occurs as expected for the LTT-filler material with 8 % Ni showing the highest M s -temperature. Finally it can be summarised that for the LTT-filler materials investigated here the residual stress values are lowered remarkably by the phase transformation as stated in literature [1]-[4]. Furthermore considerable amounts of compressive residual stresses are induced during welding. However, these effects are not homogenous across the weld. Compressive residual stresses are merely observed in the weld boundary regions rather than in the weld centre. On the other hand the austenitic phase shows more strongly deviating residual stress distributions (Fig. 8-Fig. 9). Whilst a LTT weld with 10 % Ni exhibits a nearly homogeneous compressive residual stress distribution, welds with 8 % and 12 % Ni, respectively, show more drastically varying residual stress distributions as indicated by the strongly scattering residual stress data in Fig. 8. In contrast to the martensitic phase, this effect is more pronounced in longitudinal direction. The lower the M s - temperature and the higher the associated contents of residual austenite, the higher the compressive residual stresses are in longitudinal as well as in transverse direction. However, in order to confirm this statement quantitative analysis of the corresponding amount of retained austenite will be carried out in forthcoming investigations.

8 Copyright JCPDS-International Centre for Diffraction Data 2009 ISSN Fig. 6: Phase specific residual stresses longifig. 7: Phase specific residual stresses transtudinal to welding direction in martensitic phase verse to welding direction in martensitic phase (mirrored at weld centre) (mirrored at weld centre) Fig. 8: Phase specific residual stresses longitudinal to welding direction in austenitic phase (mirrored at weld centre) Fig. 9: Phase specific residual stresses transverse to welding direction in austenitic phase (mirrored at weld centre) After due consideration it appears apparent that the effect of the phase transformation is more pronounced for the austenitic phase at lower transformation temperature. This results furthermore in an increase of the content of retained austenite within the weld. Thus, with increasing nickel content the transformation-specific compressive residual stresses develop more evidently in the austenitic phase of the weld. However, all three LTT weld materials examined here appear to be suitable for generating beneficial compressive residual stresses in the weld. But, an intended utilisation of this finding for purposes of fatigue resistance improvement or cold cracking avoidance finally depends on the availability of local mechanical properties from the weld. This issue should be one focus of future studies on the applicability of LTT-filler materials. Specifically with a view to cold cracking, existing residual austenite contents as well as transformation temperature linked processes of hydrogen diffusion represent a complex interaction which needs to be clarified before this material may be permitted to be applied in high strength steel welding. CONCLUSIONS For the first time, residual stress distributions in Low Transformation Temperature (LTT) welding material could be assessed by means of energy dispersive diffraction using high energy synchrotron radiation. Additional diffraction studies for the determination of Ms - temperatures during a thermal cycle by means of a heating stage and by using heating and cooling rates of 761

9 K/min were successfully conducted as feasibility studies for in-situ diffraction analysis during real welding. By application of energy dispersive diffraction using high energy synchrotron radiation robust phase-specific residual stresses of the involved martensitic and austenitic phase can be analysed successfully taking into account a multitude of {hkl}-reflections of both phases. In relation to the chemical compositions and the respective M s - temperatures of the LTT fillers, characteristic residual stress distributions occur especially in the martensitic phase of the welds showing typical criteria of phase transformation influence. This applies longitudinal as well as transverse to the welding direction. The residual stress distribution in the martensitic phase is qualitatively equal for different nickel contents, but stress peak values vary considerably. The highest compressive residual stresses of up to -350 MPa were found at the boundary of the weld to the HAZ. At the same time, tensile residual stresses of about 300 MPa are induced at the weld centre. The residual stress distribution in the austenitic phase is strongly influenced by the M s - temperature and by its content. At lower M s - temperature, the residual stress values of this phase show higher amounts of compressive residual stresses up to about -300 MPa in longitudinal as well as transverse direction. Finally it can be maintained that the LTT materials investigated here are qualified for creating compressive residual stresses in the weld and in the HAZ by using the effect of martensite transformation at low temperatures. For evaluation of the residual stress state of LTT welds the coexisting martensitic and austenitic phases have to be taken into account. The macro residual stress state may also be influenced by the relative content of the two phases. REFERENCES [1] Ohta, A.; Watanabe, O.; Matsuoka, K.; Shiga, C.; Nishijima, S.; Maeda, Y.; Suzuki, N.; Kubo, T., Welding in the World, 1999, 43, 6, [2] Suzuki, N.; Ohta, A.; Maeda, Y., Welding International, 2004, 2, [3] Zenitani, S.; Hayakawa, N.; Yamamoto, J.; Hiraoka, K.; Shiga, C.; Morikage, Y.; Kubo, T.; Yasuda, K., 6th International Trends in Welding Research Conference Proceedings, April 2002, Pine Mountain, USA, ASM International, 2003, [4] Mikami, Y.; Mochizuki, M.; Toyoda, M., Mathematical Modelling of Weld Phenomena 8. Verlag der Technischen Universität Graz: Graz, 2007, [5] Badeshia, H.K.D.H.; Francis, J.A.; Stone, H.J.; Kundu, S.; Rogge, R.B.; Withers, P.J.; Karlsson, L., 10th International Aachen Welding Conference Proceedings, October, 2007, Aachen. Shaker Verlag: Aachen, 2007, [6] Kromm, A.; Kannengiesser, TH.; Gibmeier, J.; Genzel, CH.; van der Mee, V., IIW - Doc. No. IIW-II , 2007, [7] Mochizuki, M.; Matsushima, S.; Toyoda, M.; Morikage, Y.; Kubo, T., Welding International, 2005, 19, 10, [8] Shiga, C.; Mraz, L.; Bernasovsky, P.; Hiraoka, K.; Mikula, P.; Vrana, M., Welding in the World, 2007, 51, 11/12, [9] Francis, J.A.; Stone, H.J.; Kundu, S.; Rogge, R.B.; Bhadeshia, H.K.D.H.; Withers, P.J., Proceedings of PVP2007, ASME Pressure Vessels and Piping Division Conference, San Antonio, July , USA, PVP , ASME, 2007, 1-8. [10] Steven, W.; Haynes, A.G., Journal of the Iron and Steel Institute, 1956, August, [11] Genzel, CH.; Denks, I.A.; Gibmeier, J.; Klaus, M.; Wagener, G., Nucl. Instrum. Methods in Phys. Research, 2007, A 578, [12] Alexandrov, B.T.; Lippold, J.C., IIW - Doc. No. IIW-IX , 2004, [13] Macherauch, E.; Müller, P.; Z. angew. Physik, 1961, 13,