Selection and peer-review under responsibility of the Indira Gandhi Centre for Atomic Research.doi: 10.1016/j.proeng.2013.03.231
6th International Conference on Cree
Evaluation of Low Cycle Weld Joints for Hig
Vani Shankar∗, K. Mari
Mechanical Metallurgy Division, Indira G
Abstract
Grade 91 steel is a heat treatable steel and hence consisting of a heterogeneous microstructure exhibitpresence of soft zone in the heat affected zone (HAparameters such as temperature and strain amplituapplication, strain localization occurs in the soft intesurface creep cavity formation in the soft region and lower fatigue life of the weld joint at high temperaattributed to the presence of surface oxides.
Modified 9Cr-1Mo steel is a plain 9Cr-1Mo extensively used for super heater tubings, headewith steam temperatures up to 866 K [1, 2].corrosion cracking compared to the austenitic tensile, creep and fatigue properties at elevagenerators (SG) are complex large structures tWelds are weak links in any structure and manweld-related [3-5]. During welding, the temperavarious extents depending upon the temperatutemperature is reached by heat affected zone (His a complex variation of microstructure with v
Fatigue Damage in Grade 91 Steegh Temperature Applications
iappan, R. Sandhya, M. D. Mathew
Gandhi Centre for Atomic Research, Kalpakkam-603102, India
its microstructure is very sensitive to temperature. The welts a lower fatigue life than that of the base metal. This is due
AZ). The failure location in the weld joint is very sensitive to ude and application of hold. At high temperatures and undercritical HAZ (ICHAZ) which results in the failure in the HAZtheir linkage cause enhanced crack propagation and that transla
atures. Occurrence of compression dwell sensitivity in the mat
or peer-review under responsibility of the Indira Gandhi Cen
steel in which Nb and V are optimized (Grade 91) and isers and pipings of conventional as well as nuclear power The alloy has better resistance to thermal fatigue andstainless steels. Also, the Grade 91 steel has better monated temperatures compared to plain 9Cr-1Mo steel. that are made in parts and integrated by welding technny failures in high temperature components are reportedature sensitive microstructure of Grade 91 steel gets modiure it is exposed to during the weld thermal cycles; h
HAZ) closest to the weld metal. Hence the narrow band ovaried mechanical properties. Many premature failures a
joints have been identified to occur in a narrow fine grained or inter critical region of the HAZ in stronger 9-12 Cr steels such as Grade 91 ferritic steel [3-6] and they are termed as type IV failures.
In sodium cooled fast reactors (SFRs), the components are often subjected to high temperature fatigue due to the temperature gradient induced cyclic thermal stresses as a result of start-ups, shutdowns and transients. In addition, steady state operation at elevated temperatures introduces creep, resulting in creep-fatigue interaction condition. In order to have more reliability in the performance of a component, it is essential to understand the deformation mechanisms operative under pure fatigue, pure creep and creep-fatigue interaction conditions. Creep experiments on weld joints of Grade 91 steel have shown that fracture usually occurs in the base metal in the short term tests at high stresses, whereas under long term creep conditions at low stresses Type IV failure mode appears [7-9]. Type IV failure results in significant reduction in rupture life as compared to the specimens without weld joint [6, 9]. Creep performance of the weld joints of Grade 91 steel is much better understood compared to the LCF and CFI behavior. Hence the aim of the present paper is to study low cycle fatigue and creep-fatigue interaction behavior (CFI), identify the underlying deformation mechanism and correlate with the failure mode occurring in Grade 91 steel weld joints.
2. Experimental
Chemical composition of the 30 mm thick rolled plate used for current study and the weld metal is given in Table 1. The base material was given a normalizing treatment at 1313 K for 1 hour followed by air cooling and subsequently tempering at 1033 K for 1 hour followed by air cooling, before machining the fatigue specimens. Microstructural characterization was carried out using optical microscope, scanning and transmission electron microscopes (TEM). Samples for the optical metallography were etched using Vilella’s reagent (1g of picric acid + 5 ml conc. HCl + 100 ml ethyl alcohol).
Table 1. Chemical composition (wt %) of base metal and weld metal.
Element C Cr Mo Ni V Nb N S P Cu Co
Base metal 0.11 9.3 0.99 0.14 0.25 0.1 0.068 0.008 0.020
For fabricating LCF specimens containing weld joint, the plates were joined along the rolling direction by shielded metal arc welding (SMAW) process using voltage and current approximately 20 V and 100 Amperes respectively. A double-V configuration, with an included angle of 70o, a root face of 2 mm, and a root gap of 3.15 mm was used. Multiple passes were employed to fill the groove. An inter-pass temperature of 423 K was maintained during welding. Matching filler wire was used for welding. The weld pads were examined by radiography for their soundness. Standard LCF specimens containing joint in the centre were machined from the plate bars. Bars of 110 mm length and 25 x 25 mm square cross section were cut from the welded plate. Post welding heat treatment (PWHT) was given to the bars containing weld joints at l033 K for 3 h followed by air cooling. Low cycle fatigue (LCF) and creep fatigue interaction (CFI) tests were conducted in air, under fully reversed, total axial strain control mode in accordance with ASTM specification E606 [10] in a closed loop servo hydraulic testing system equipped with a radiant heating furnace. The temperature variation along the gauge length of the specimen did not exceed ± 2 K. Continuous cycling tests were performed using a triangular waveform and hold time experiments were carried out using a trapezoidal waveform. LCF tests under continuous cycling condition were carried out at room temperature at strain amplitudes of ±0.6% and at 823 K and 873 K at strain amplitude ±0.25%, ±0.4%, ±0.6% and ±1%. CFI tests were conducted by introducing holds ranging from 1 min to 10 minute at peak tension or in peak compression under ±0.6% strain amplitude. All the tests were carried out at a constant strain rate of 3×10-3 s-1. Fractographic analysis was performed using SEM after which the longitudinal sections of the two separated parts were analyzed using an optical microscope equipped with an image analyzer. Microstructure near the fractured surface was studied for all the tested
specimens. Microhardness profile was taken alsurface and the two fusion lines were determined
3. Results and discussion
The normalized and tempered base materiaprecipitation of M23C6 at the boundaries and MXtowards base material, the microstructure is asstructure, tempered martensite structure with iinter-critical structure (Fig. 1(d)) and base matewidth of the HAZ is ~3 to 4 mm and the intercriline and close to BM. A gradient in the hardness(f)); a minimum in hardness value corresponds t
Fig. 1. Graded microstructure in Grade 91weld joint due toand the resulta
long the longitudinal section of the sample from the frad using an optical microscope.
al reveal a tempered martensitic microstructure with extX type precipitates within the matrix. Starting from welds follows: weld metal (Fig. 1(a)) with a tempered martslands of -ferrite (Fig. 1(b)), fine grained region (Fig.
erial (Fig. 1(e)) with a tempered martensitic structure. Thitical region is approximately 2.5 to 3 mm away from the s values due to gradient in the microstructure is observed to the ICHAZ microstructure.
o exposure to different temperatures during weld thermal cycles (a) throuant variation in microhardness (f).
The cyclic stress response of the base metal softening from first cycle onwards is observesoftening material. As the temperature is increaalso viewed that whereas at 823 K there is not mthe weld joint (Fig. 2 (b)), the disparity betweensection of the failed and separated samples testethe microstructure near the fractured surface coron failed samples show that whereas the majorsample tested at 823 K, the crack is confined tothe microhardness profile taken across the seconThe profiles are of the samples tested at 823 Khardness values. There is also a dip in the haFGHAZ/ICHAZ at 873 K. It is hence speculatehigher temperatures is related to the difference zones of the weld joint as compared to the basecompared to 823 K as observed in Fig. 2 (d) iannihilation and coarsening of carbides as the te3 (a) and (b) that there is a tendency to form cellfatigue deformation, the high number of dislocaand forth movement under cycling [11] and theycells and subgrains causing cyclic softening annihilated and coarsening of carbides occur; thi
(a)
(c)
Fig. 2. (a) Effect of temperature on the cyclic streof base metal and weld joint at 823 K, (c) at 873
and weld joint is compared in Fig. 2 (a). The continuoused in both base and weld joint and is typical of a cycased, there is an overall decrease in the cyclic stress valuemuch difference in the fatigue life values of the base me
n the life values increases at 873 K (Fig. 2 (c)). The longited at 823 K and 873 K under an optical microscope showrrespond mostly to intercritical heat affected zone. Replicar crack initiates in HAZ and propagates into weld metal o the HAZ for the sample tested at 873 K. Figure 2 (d) dnd fusion line which is close to the HAZ and that had not K and 873 K. It shows that higher the temperature lowerardness profile at a distance of ~ 2mm which correspoed that the disparity between the fatigue lives of BM and
in the microstructural changes occurring in the heterogee metal. The larger decrease in the hardness values at 87is due to enhanced recovery process in the form of disloemperature of straining is increased. This is also clear frols/subgrains occur at higher temperatures such as 873 K. Dations present in the martensitic lath structure undergoes y try to get rearranged into a lower energy configuration sin the material. During the process, many dislocatio
is process is highly sensitive to temperature.
(b)
(d)
ess response curves of base metal and weld joint, (b) fatigue life comparK and (d) effect of temperature on microhardness profile across fusion
cyclic clically es. It is tal and tudinal ws that a taken for the depicts failed.
Fig. 3. Effect of temperature on the resultant microstructure corresponding to the samples cycled at +0.6% strain amplitude up to failure at (a) ambient and (b) at 873 K.
The fractured surface of the samples tested at 873 K under continuous cycling (Fig. 4) depicts dark contour (marked by white dotted lines) on the periphery of the fractured surface (~0.5 mm below the sample surface). Magnified image of the contour shows severe localized plastic deformation. This indicates that the contour could be the soft intercritical region in the HAZ. It may be stated that during each cycle, the plastic strain accumulated in different regions of the sample is different depending upon the mechanical properties of individual microstructures. During high temperature experiment, the soft zone of HAZ accumulates larger plastic strain which is constrained by the surrounding material such as the comparatively stronger weld and base metal. Carbide counting and their size measurement performed at various locations (1 to 7) of the sample across the weld joint depicted in Fig. 5 (a) shows comparatively coarser M23C6 carbides at one location. This corresponds to the ICHAZ. This is in line with that reported under pure creep conditions [7-10]. Bright field image (Fig. 5 (b)) taken on a thin foil prepared from the intercritical region of HAZ also shows scanty coarsened carbides and dislocation rearrangement. SEM image of the intercritical HAZ illustrates cavities around the coarsened M23C6 carbides (Fig. 5 (c)) due to localized plastic deformation and multiaxial state of stress. Crack linkage along the periphery of the fractured surface (Fig. 4) is observed which could be the soft ICHAZ. Hence strain localization due to heterogeneity in microstructure and cavitation around coarsened carbides due to multiaxial stress occur in the weld joint that together leads to the difference in the cyclic stress response and life values.
Fig. 4. Fractograph of specimen tested at 873 K under continuous cycling shows contour in the outer extremity of the fractured surface (marked by white dotted lines) marks localized plastic deformation.
Fig. 5. (a) Mean carbide size at various locations of the weld joint specimen, (b) bright field image as a result of dislocation rearrangements and (c) SEM image of intercritical HAZ illustrating cavities around the coarsened M23C6 carbides due to localized
plastic deformation and multiaxial state of stress in the soft ICHAZ.
The effect of application of hold in peak strain in either tensile or compressive direction is to deteriorate the fatigue life. Hence application of hold in any direction decreases the fatigue life values; compressive hold is found to be more damaging than tensile hold (Fig. 6 (a)). Hardness profiles taken across the fusion line under various test conditions such as tension hold (1 min), compression hold (10 min) and untested conditions are depicted in Fig. 6 ((b). It is clear that application of hold in either direction result in an overall softening of the weld joint. Hence it is speculated that the annihilation of dislocation and carbide coarsening is much more pronounced during the extended periods of hold at elevated temperatures. This leads to lowering of the microhardness profile and the resultant lowering of fatigue life values as observed in Fig. 6 (b) and Fig. 6 (a) respectively. Under pure creep condition, the importance of multiaxiality of stress for creep cavitation around coarse precipitates in Cr-Mo steel weld joints have been brought out. On similar understanding, under creep-fatigue interaction condition, the vital role played by multiaxiality of stress for cavity nucleation around the coarsened carbides in soft ICHAZ cannot be ruled out. Vacancies generated during fatigue enhance creep climb and carbide coarsening that assist in cavity nucleation and hence enhanced creep deformation occur during elevated tests under continuous cycling and creep-fatigue interaction conditions.
(a) (b)
Fig. 6. Effect of application of various holds on (a) fatigue life and (b) comparison of hardness profiles taken across fusion line in untested condition and under application of hold.
The role played by surface oxides in crack initiation and propagation are important under compressive holds (Fig. 7). Multiple crack initiation sites and oxidation assisted crack initiation and propagation for sample tested under compression hold is observed. Also, the number of secondary cracks found under compression hold is much higher than found under tension hold (Fig. 7 (a). Brittle surface oxides formed on the sample surface during hold at high temperatures break upon unloading and further cycling in tensile direction after removal of compression hold (Fig. 7 (b)). Oxygen impregnates through the freshly exposed metal layer causing more damage in the material.
(a) (b)
Fig. 7. Large number of short secondary cracks perpendicular to loading direction under CH as compared to fewer unidirectional longer cracks under TH (a) is due to the role played by surface oxides during compression hold experiment (b).
A shift in the failure location from base metal to heat affected zone is observed as the temperature is increased from ambient to elevated temperatures. At room temperature, major crack is in the base metal, at 773 K and 823 K the crack initiate in HAZ and propagate in WM. At 873 K, the major crack is contained in HAZ and resemble type IV cracking due to creep effects. The different failure location is due to the different response of each microstructural zone present in the weld joint to the temperature, strain amplitude and strain hold combination. At elevated temperature and under creep-fatigue interaction, multiaxiality of stress for creep cavitation around coarsened carbides in the ICHAZ and their linkage cause the failure. At lower temperatures, such preferential coarsening of carbides in the ICHAZ is not so accelerated and hence the failure location is shifted to other locations. Further analysis to explain the shift in failure location with various test parameters is in progress.
4. Conclusions
The high dislocation density initially present in the martensitic lath structure undergoes a rearrangement into a lower energy configuration such as cells and subgrains causing cyclic softening in the material and this process is highly sensitive to temperature and application of hold. The lower cyclic stress response curve of the weld joint compared to the base metal is attributed to the soft zone present in the heat affected zone of the weld joint. Failure location in the weld joint is found to depend upon temperature and application of hold. Strain localization in the soft zone of the heat affected zone (HAZ), sub-surface creep cavity formation in this region and their linkage had caused enhanced crack propagation that translated into lower fatigue life of the weld joint at high temperatures. The compression dwell sensitivity was largely due to the deleterious effect of the surface oxides.
Acknowledgements
The authors are grateful to Sri S. C. Chetal, Director, IGCAR and Dr T. Jayakumar, Group Director Metallurgy and Materials Group, IGCAR, Kalpakkam for their support and encouragement during this research.
[1] J. M. Vitek, R. L. Klueh, Precipitation Reactions during the Heat Treatment of Ferritic Steels, Metall Mater Trans A 14 (1983) pp. 1047-1055.
[2] B. W. Jones, C. R. Hills, D. H. Polonis, Microstructural Evolution of Modified 9Cr-lMo Steel, ) Metall Trans A 22 (1991) 1049-1058. [3] I. A. Shibli, Parsons 2003, Engineering Issues in turbine machinery, power plant and renewables, Proc. Of the sixth international
Charles Parsons Turbine Conf., 16-18 Sept., 2003, Trinity College Dublin, Ireland, Eds. A. Strang, R.D.Conroy, W.M.Banks, M.Blackler, J. Leggett, G.M.McColvin, S.Simpson, M.Smith, F.Star and R.W.Vanstone, pp. 261-279
[4] C. Middleton and E. Metcalfe, Review of laboratory type IV cracking data in high chromium ferritic steels, Paper C386/027, IMechE Proceedings, London, UK., 1990.
[5] Shibli, Creep and fatigue crack growth in P91 weldments, Proceedings of the Swansea Creep Conference, Organized by the University of Swansea and EPRI, Swansea, UK, 2001.
[6] H. Cerjak and E. Letofsky (1998) Behaviour of advanced 9-12 Cr steels and its weldments in short and long term tests. Advanced Heat Resistance Steels for Power Generation. San Sebastian, Spain, 1998, pp. 611-621.
[7] J. M. Brear, A. Fairman, C. J. Middleton and L. Polding, Predicting the creep life and failure location of weldments. Key Eng. Mater., 171-174 (2000), 35-42.
[8] C. J. Middleton, J. M. Brear, R. Munson and R.Vishwanathan, Proc. 3rd Conf. on Advances in Materials Technology for Fossil Power Plant. Edited by R.Vishwanathan, W. T. Bakker and J. D. Parker, The Inst.Mater., London, 2001, pp. 69-78.
[9] E. Letofsky, H. Cerjak, I. Papst and P. Warbichler, Proc. 3rd Conf. on Advances in Materials Technology for Fossil Power Plant. Edited by R.Vishwanathan, W. T. Bakker and J. D. Parker, The Inst.Mater., London, 2001, pp. 133-142.
[10] ASTM E606-92, "Standard recommended practice for constant-amplitude low-cycle fatigue testing", Annual Book of ASTM Standards, 1994, 03.01, pp. 522-536
[11] D.C. Lord, L.F. Coffin Jr, Low Cycle Fatigue Hold Time Behavior of Cast Rene 80, Metall. Trans. 4 (1973) 1647–1654.
