Professional Documents
Culture Documents
Real-time-monitoring-of-phase-transformation-and-s_2014_Journal-of-Materials
Real-time-monitoring-of-phase-transformation-and-s_2014_Journal-of-Materials
a r t i c l e i n f o a b s t r a c t
Article history: For a newly developed 10% Cr and 10% Ni low transformation temperature (LTT) weld filler material, the
Received 24 January 2014 local phase transformation kinetics and the strain evolution during gas tungsten arc welding (GTAW)
Received in revised form 6 June 2014 under real welding conditions was studied. An experimental set-up and a measuring and evaluation
Accepted 7 June 2014
strategy are presented to gain a real time insight into the welding process. The experiments were carried
Available online 16 June 2014
out at the beam line ID15@ESRF using a two detector EDXRD (energy dispersive X-ray diffraction) set-up
and high energy synchrotron X-rays. The time-resolved diffraction analysis during welding was carried
Keywords:
out locally throughout the weld in longitudinal as well as in transverse direction to the weld line to
In situ synchrotron X-ray diffraction
Low transformation temperature
examine the interdependence of the strain state and the transformation kinetics. This comprehension is
Welding crucial for the optimization of the weld process, and thus for the tailoring of the resulting residual stress
states, which is one of the main issues for the application of LTT alloys. Using the herein proposed approach
EDXRD diffraction pattern can be monitored during real welding with a counting rate of 5 Hz. By means
of the time resolved diffraction data the local transformation temperatures and times were determined
and the local phasespecific strain evolutions are discussed with respect to the transformation rates and
the time-delayed phase transformations.
© 2014 Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.jmatprotec.2014.06.008
0924-0136/© 2014 Elsevier B.V. All rights reserved.
2740 J. Gibmeier et al. / Journal of Materials Processing Technology 214 (2014) 2739–2747
Table 1
Chemical composition in wt% of the LTT-alloy and the base material S690Q as well as selected material properties with indication of the references from which some of the
data are taken from. The martensite start temperature was calculated according to Steven and Haynes (1956).
C Ni Cr Mn Si P S Mo Nb V B Fe
temperature for the chosen LTT-alloy. Irrespective of this item, 7.6 mm in beam direction is available. Strain measurements were
austenite observed at room temperature is considered as retained performed in transmission such that the strain was determined in
austenite in the following. the weld longitudinal (Ge-det 0) and weld transverse (Ge-det 1)
direction simultaneously within one exposure. With the described
∗ = 3◦ away
set-up, the strain is actually measured in a direction QL/T
2.2. Thermal exposure for in situ welding
from the weld longitudinal (QL ) and transverse (QT ) directions:
The thermal cycle for the in situ diffraction experiment was however, the difference in strain between the measured and the
applied by online gas tungsten arc welding (GTAW). A special actual in-plane directions is negligible.
designed online welding device, which was first presented by To provide fast data analysis a special readout configuration of
Kannengiesser and Kromm (2009) as applied for comparable stud- the multi-channel (MCA) electronics developed by the ESRF BLISS
ies, is depicted in Fig. 2a) with two parallel translation axes, was group was provided. This high speed software/hardware architec-
used with the welding torch mounted on one axis and the weld ture consists of a commercial MCA acquisition card (XIA LLC) which
sample on the second one. can be read out in ∼20 ms and a multi-purpose digital I/O NIM
This set-up allows translating the welding torch along the static module built around a programmable logic device and a micro-
sample to monitor the material behavior at a fixed weld line posi- controller (=OPIOM; developed by the Instrumentation Services &
tion. For simulation of the welding process, a S690Q base plate Development Division at ESRF). The Opium NIM module was pro-
(100 mm × 80 mm × 6 mm) with a layer of LTT material welded in grammed to synchronize the acquisition of the MCA data and the
multiple beads onto the long edge of the base metal and subse- operation of the welding unit as firstly reported by Altenkirch et al.
quently machined to 100 mm × 10 mm × 5 mm was used. By this (2012). In order to observe the phase kinetics and strain evolution
means, the welding process monitored during the in situ diffrac- during welding throughout the sample, diffraction spectra were
tion experiments is a re-fusion of the pre-deposited weld made collected every 200 ms (5 Hz) at 10 lateral distances (0.5–5 mm
from the LTT-filler material, which leads to similar results of phase distance from the top of the weld line in 0.5 mm steps by shif-
transformation kinetics and strain evolution in case of solidification ting the sample along the transverse direction). For each run the
without constraint as for the we prior welding process. A schematic weld tool was traversed along the pre-welded LTT-material. Data
illustration of the plate is shown in Fig. 2b). GTAW was performed were collected until the sample reached ambient temperatures. In
at a feed rate of 80 mm/min over a 65 mm long weld path with a between runs a 5 min delay was included to ensure the sample
welding current of 80 A and argon as shielding gas. The tungsten was at the same starting temperature for each measurement. The
electrode was positioned approximately 1 mm above the surface. welding cycle start was triggered electronically by the diffraction
Surface temperature measurements as a function of time and dis- measurement.
tance to the weld pool were carried out using an infrared camera
during offline welding under identical conditions. The temperature
data were confirmed using thermocouples. 2.4. Data analysis
Fig. 2. Illustration of the welding device used for online welding (a) and a schematic illustration of the test samples (b). 10 layers off the LTT alloy were pre-welded onto the
long edge of a 6 mm thick base plate (S690Q) and subsequently machined into shape.
well as the hkl plane specific Young’s modulus Ehkl a {h k l} plane NIST (Chantler, 1995) the phase fraction of retained austenite was
specific weighting factor m∗hkl was calculated using: calculated according to the procedure given in the textbook of
Macherauch (1992):
FT · mhkl · Ehkl
m∗hkl = (2)
E
which was presented by Daymond (2004) for the analysis of mul- 100 vol%
v = (4)
tiple diffraction peaks of one phase, where E is the macroscopic m m
˛i ·I˛i i ·Ii
phase specific and temperature dependent Young’s modulus. The 1+ R˛i
/ m˛i / Ri
/ mi
variation of Ehkl and E as a function of temperature was calculated i=hkl i i=hkl i
based on data taken from Davis (1996) for the ˛ - and -phase,
respectively. The unit cell parameter a for each phase was calcu-
lated using: Assuming 1 = v + v˛ the phase fraction of the ˛ -phase can be
a · m∗hkl
hkl hkl
determined.
a= ∗
(3) For the calculation of the lattice strain the stress free unit cell
hkl
mhkl
parameter a0 has to be known as a function of the temperature for
This determination of the unit cell parameter is valid for the both the ˛- and the -phase. For the -phase it was assumed that
austenite and for the martensite phase. Due to the high amount of the first solidified grains are stress free at high temperatures. The
nickel and chromium in the LTT alloy a cubic body centered mar- stress free unit cell parameter a0 of the -phase was determined at
tensite lattice structure exist that corresponds to the findings of the solidification temperature and afterwards calculated for lower
Reed (1962) for studies on a comparable alloy. temperatures via the coefficient of thermal expansion (CTE). The
The phase fraction of retained austenite was calculated accord- temperature measurement, which is the basis of this approach only
ing to the procedure defined in Laine (1978) who proposed an account to 2% in maximum, hence, the effect will be neglected in
evaluation procedure to determine the volume fractions of ferrite in the following. The stress free unit cell parameter of the ˛ -phase
austenitic steels using EDXRD and Faninger and Hartmann (1972) was determined at room temperature using the comb-method in
who discussed the physical bases of quantitative X-ray phase anal- V-groove weld joints as reported by Gibmeier et al. (2014) for the
ysis. Using the integrated peak intensity Ihkl of the ˛ - and -phase identical welding samples to a0,˛ = 2.871 Å. The CTE was measured
peaks and the multiplicity factor m∗hkl as well as the temperature earlier by Altenkirch et al. (2011) to ˛˛ = 11.39 K\10−6 for the ˛ -
dependent R-factor Rhkl (calculation is based on data taken from phase and to ˛ = 28.39 K\10−6 for the -phase.
Fig. 3. Schematic illustration of the EDXRD diffraction experiment set-up for monitoring the phase transformation kinetics and the strain evolution using an online welding
rig, which allows for automatically traversing the GTAW weld torch along the pre-welded layer of LTT material. As a result of the set-up, the scattering vectors are slightly
rotated by out of the in-plane directions.
J. Gibmeier et al. / Journal of Materials Processing Technology 214 (2014) 2739–2747 2743
Fig. 5. Time and temperature evolution of the martensite (˛ ) and austenite () phase fraction in the LTT test weld at 3 mm from the weld center line in longitudinal direction.
not reached. Even though the martensite finish temperature is weld cycle (b) as a function of distance from the weld surface. The
expected above room temperature, the ongoing increase in ˛ - Ac3 -temperature is relatively constant at around 700 ◦ C in the com-
phase fraction may be attributed to local chemical variations plete welding zone and also in the transition zone to the base metal.
causing for a martensite formation at even lower temperatures The comparison of the Ms -temperature as a function of lateral dis-
(Fig. 7). tance from the weld surface reveals only a slight increase of the
Fig. 6 displays the Ms - and Ac3 -temperatures (a) as well as the value for Ms up to the depth of 4 mm from about 110 ◦ C to 150 ◦ C.
starting time of the martensite transformation after start of the Beyond that distance Ms increases strongly. As soon as the transi-
tion zone to the base metal is reached the effect of the increased
Ni-content on the transformation temperatures vanishes and the
Ms -temperatures increase to values which are more typical for low
alloyed steel alloys. Hence, the strong effect might be due to local
demixing of the filler material. Moreover, upon cooling and mar-
tensite formation a complex local stress field may arise, which can
lead to further deformation induced martensite formation. Fig. 6b
clearly shows that the martensite transformation starts at the bot-
tom of the weld at the transition to the base material. Since in this
region the local heat input is smaller and the heat conductance
through the base material is higher than in the regions closer to
the weld surface, the martensite start temperature is reached ear-
lier with a higher distance to the weld surface. The larger delay
between the transformation start at 3.5 and 4 mm is due to the
increase of the Ms temperature (Fig. 5a) that results from the mixing
of the LTT filler material with the S690 base material that contains
less nickel. Thus, the martensite start temperature is reached even
earlier in distances higher than 4 mm. An additional influence can
also be assigned to the locally varying (residual) stress states. The
Fig. 6. (a) Martensite start Ms and Ac3 temperature determined in longitudinal direc-
tion (b) begin of the martensite transformation as a function of the distance from Fig. 7. Amount of retained austenite after cooling down to room temperature as a
the weld line. function of the distance from the weld line. Results averaged from the two detectors.
J. Gibmeier et al. / Journal of Materials Processing Technology 214 (2014) 2739–2747 2745
Fig. 9. Strain (elastic plus thermal expansion strain) development in the ˛ -phase in longitudinal direction at different distances z from the weld top surface (left) as a function
of time and (right) as the function of the temperature.
weld. Apart of the amount of retained austenite, which varies with (DFG) is acknowledged for financial support through project GI
the location in the weld, the phasespecific strains in martensite do 376/4-1 and RE 1648/2-1.
not show a significant dependency on the local austenite content.
However, the variation in retained austenite will certainly affect the
resulting macrostress state and must be considered for calculation References
of macro (residual) stresses.
Altenkirch, J., Steuwer, A., Peel, M., Richards, D.G., Withers, P.J., 2008. The effect of
tensioning and sectioning on residual stresses in aluminum AA7749 friction stir
4. Conclusions welds. Mater. Sci. Eng. A 488, 16–24.
Altenkirch, J., Gibmeier, J., Kostov, V., Kromm, A., Kannengiesser, T., Doyle, S., 2011.
A method was presented that allows real time monitoring of Time- and temperature-resolved synchrotron X-ray diffraction: observation of
phase transformation and strain evolution in novel low temperature transfor-
the local phase transformation and strain evolution during welding mation weld filler materials. J. Strain Anal. Eng. Des. 46, 563–579.
under real welding conditions using time resolved energy disper- Altenkirch, J., Gibmeier, J., Buslaps, T., Honkimäki, V., 2012. EDXRD setup for real
sive synchrotron X-ray diffraction (EDXRD). time observation of a gas tungsten arc (GTA) welding process. Mater. Sci. Forum
706–709, 1655–1660.
Bhadeshia, H.K.D.H., 2004. Developments in martensitic and bainitic steels: role of
• Measurements at a rate of 5 Hz were carried out successfully the shape deformation. Mater. Sci. Eng. A 378, 34–39.
Chantler, C.T., 1995. Theoretical form factor, attenuation, and scattering tabulation
at ID15A@ESRF, which give a very good resolution in terms of
for Z = 1–92 from E = 1–10 eV to E = 0.4–1.0 MeV. J. Phys. Chem. Ref. Data 24,
resolved temperatures. 71–643.
• Local information about the start and the end of phase transfor- Dai, H., Francis, J.A., Stone, H.J., Bhadeshia, H.K.D.H., Withers, P.J., 2008. Character-
mations was extracted together with the transformation rate. izing phase transformations and their effects on ferritic weld residual stresses
with X-rays and neutrons. Metall. Mater. Trans. A 39, 3070–3078.
Davis, J.R., 1996. ASM Specialty Handbook: Carbon and Alloy Steels. ASM Interna-
We focused on phase transformation kinetics in a LTT weld filler tional, Materials Park, Ohio, USA.
Daymond, M.R., 2004. The determination of a continuum mechanics equivalent
material with 10%Cr and 10%Ni. The results determined for the LTT- elastic strain from the analysis of multiple diffraction peaks. J. Appl. Phys. 96,
weld can be summarized as follows: 4263–4272.
Faninger, G., Hartmann, U., 1972. Physikalische Grundlagen der quantitativen
röntgenographischen Phasenanalyse (RPA). HTM Härterei-Techn. Mitt. 27,
• the delayed martensitic transformation successfully counteracts 233–244.
the thermal strain resulting in a compressive stress state at room Gibmeier, J., Obelode, E., Altenkirch, J., Kromm, A., Kannengiesser, T., 2014. Residual
temperatures. stress in steel fusion welds joined using low transformation temperature (LTT)
filler material. Mater. Sci. Forum 768–769, 620–627.
• not only the local chemical constitution is important for Kannengiesser, T., Kromm, A., 2009. Formation of welding residual stresses
the austenite-martensite transformation, but also the local in low transformation temperature (LTT) materials. Soldagem Inspeção 14,
stress/strain field has a significant influence on the transforma- 74–81.
Kannengiesser, T., Kromm, A., Rethmeier, M., Gibmeier, J., Genzel, Ch., 2009.
tion strain and the martensite start temperature. Residual stresses and in-situ measurement of phase transformation in low
• the time-resolved and local resolved results are essential for the transformation temperature (LTT) welding materials. Adv. X-ray Anal. 52,
understanding of the interdependence of transformation induced 755–762.
Kromm, A., Kannengiesser, T., 2009. In-situ-phase analysis using synchrotron radi-
strains and materials straining that results from local constraints ation of low transformation temperature (LTT) welding material. Soldagem
due to local differing cooling rates and time-delayed cooling Inspeção 14, 82–88.
down. Kromm, A., 2011. Umwandlungsverhalten und Eigenspannungen beim Schweißen
neuartiger LTT-Zusatzwerkstoffe. BAM-Dissertationsreihe. Band 72.
Laine, E.S.U., 1978. A high-speed determination of the volume fraction of ferrite in
The data monitored in real-time during welding are valuable austenitic stainless steel by EDXRD. J. Phys. F: Met. Phys. 8, 1343–1348.
for a proper validation of simulations of LTT welds and hence for Macherauch, E., 1992. Praktikum in Werkstoffkunde: Skriptum für Inge-
nieure, Metall-Werkstoffkundler, Werkstoffwissenschaftler, Eisenhüttenleute,
an effective improvement of the existing simulation models. Fertigungs- und Umformtechniker. 10. verbesserte Auflage. Vieweg + Teubner
Verlag.
Moat, R.J., Stone, H.J., Shirzadi, A.A., Francis, J.A., Kundu, S., Mark, A.F., Bhadeshia,
Acknowledgements H.k.d.h., Karlsson, L., Withers, P.J., 2011. Design of weld fillers for mitigation of
residual stresses in ferritic and austenitic steel welds. Sci. Technol. Weld. Joining
The authors would like to thank the European Synchrotron Radi- 16, 279–284.
Ohta, A., Matsuoka, K., Nguyen, N.T., Maeda, Y., Suzuki, N., 1999a. Fatigue strength
ation Facility (ESRF) in Grenoble for granting beam time through improvement by using newly developed low transformation temperature weld-
proposal number MA847. Further, German Research Foundation ing material. Weld. World 46, 38–42.
J. Gibmeier et al. / Journal of Materials Processing Technology 214 (2014) 2739–2747 2747
Ohta, A., Suzuki, N., Maeda, Y., Hiraoka, K., Nakamura, T., 1999b. Superior fatigue Stone, H.J., Bhadeshia, H.K.D.H., Withers, P.J., 2008. In situ monitoring of weld
crack growth in newly developed weld metal. Int. J. Fatigue 21, 113–118. transformations to control weld residual stresses. Mater. Sci. Forum 571–572,
Ohta, A., Matsuoka, K., Nguyen, N.T., Maeda, Y., Suzuki, N., 2003. Fatigue strength 393–398.
improvement of lap joints of thin steel plate using low-transformation- Williams, J.C., Starke Jr., E.A., 2003. Progress in structural materials for aerospace
temperature welding wire. Weld. Res. 04, 78–83. systems. Acta Mater. 51, 5775–5799.
Patel, J.R., Cohen, M., 1953. Criterion for the action of applied stress in the martensitic Wang, W., Huo, L., Zhang, Y., Wang, D., Jing, H., 2002. New developed welding elec-
transformation. Acta Metall. 1, 531–538. trode for improving the fatigue strength of welded joints. J. Mater. Sci. Technol.
Reed, R., 1962. The spontaneous martensitic transformations in 18% Cr, 8% Ni steels. 18, 527–531.
Acta Metall. 10, 865–877. Withers, P.J., 2007. Residual stress and its role in failure. Rep. Progress Phys. 70,
Steven, W., Haynes, A.G., 1956. The temperature of formation of martensite and 2211–2264.
bainite in low alloy steels. J. Iron Steel Inst. 183, 349–359.