- Email: [email protected]
strain response of friction stir-welded titanium butt joints using moiré interferometry
ARTICLE IN PRESS Optics and Lasers in Engineering 48 (2010) 385–392
Contents lists available at ScienceDirect
Optics and Lasers in Engineering journal homepage: www.elsevier.com/locate/optlaseng
Elastic–plastic stress/strain response of friction stir-welded titanium butt joints using moire´ interferometry M. Ramulu a,, P. Labossiere b, T. Greenwell a a b
Department of Mechanical Engineering, University of Washington, Box 352600, Seattle, WA 98195, USA University of Manitoba, Winnipeg, Canada
a r t i c l e in fo
abstract
Article history: Received 28 July 2009 Received in revised form 3 October 2009 Accepted 5 October 2009 Available online 20 November 2009
An experimental investigation is conducted to examine, evaluate, and characterize the fundamental elastic–plastic stress/strain response of friction stir-welded butt joints in thin-sheet, fine grain Ti–6Al– 4V titanium alloy under normal tensile loading using the full-field optical strain analysis technique of moire´ interferometry. It was found that the overall strength of friction stir-welded Ti–6Al–4V is comparable to the accepted values for pure mill-annealed Ti–6Al–4V and the overall strain performance of friction stir-welded Ti–6Al–4V is roughly half that of the accepted values for pure mill-annealed Ti–6Al–4V. In addition, friction stir-welded Ti–6Al–4V demonstrates a consistent pattern of strain localization between the onset of yielding and ultimate failure. The resolvability of such strain localizations and their impact on far-field behavior can only be realized with full-field measurements such as those obtained using the moire´ interferometry technique. Published by Elsevier Ltd.
Keywords: Friction stir welding Titanium Moire´ interferometry Full-field measurements
1. Introduction Titanium (Ti) is currently an extremely important material to the aerospace industry. Its strength to weight ratio, operating temperature range and resistance to corrosion offer many advantages over aluminum alloys. Unfortunately joining titanium is very difficult to do with traditional fusion welding techniques and riveting can lead to poor tolerances. Titanium alloys on the other hand, are currently welded by a number of fusion welding processes, but some problems associated with fusion welding, such as formation of a brittle cast structure, large distortion and residual stress, still remain. Friction stir welding (FSW) of titanium alloys has emerged as one of the enabling technologies, allowing the fabrication of very large single piece components [1–3]. This technique avoids problems associated with the traditional fusion welding of titanium alloys since FSW is a solid-state joining process. Much has been accomplished in characterizing FSW joints in 2XXX and 7XXX series aluminum alloys for aerospace use, but little published research exists regarding FSW joints in titanium alloys. This is due to the more challenging issues associated with FSW of titanium alloys, such as the need for high performance, high temperature, wear resistant tool materials and well-controlled processing conditions due to the sensitivity of FSW to temperature variations. These added challenges limit the application of FSW titanium [4]; however, preliminary research indicates FSW of titanium alloys creates a robust weld with strength equal to or greater than the base metal and capable of further post-processing [2–6]. Corresponding author.
E-mail address: [email protected] (M. Ramulu). 0143-8166/$ - see front matter Published by Elsevier Ltd. doi:10.1016/j.optlaseng.2009.10.003
Only recently FSW has been applied successfully in joining of titanium [5–14], which has sparked much interest in the Aerospace industry as this process could play a major role in next generation manufacturing techniques. Recently reported research on FSW of titanium and titanium alloys have already shown some important knowledge on the microstructure and the properties in the welds [5,6,9–12], but effect of microstructure on mechanical properties of the weld has not been fully understood. Our Industry—University collaborative preliminary research has verified that FSW of titanium alloys creates a robust weld with good strength and allows further post-processing [10,11]. Review of currently available research finds global stress response and microstructural characterization of friction stir welded Ti–6Al–4V available, but no data is available, which describes the full-field stress–strain response of the weld-base metal interface. Therefore the purpose of this paper is to examine, evaluate, and characterize the fundamental elastic–plastic stress/strain response of friction stir-welded butt joints in thin-sheet, fine grain Ti–6Al–4V titanium alloy under normal tensile loading using the full-field optical strain analysis technique moire´ interferometry. The objective of the present study is to help clarify the relationship between microstructure and mechanical properties of friction stir welded Ti–6Al–4V alloy. 2. Experimental set-up and procedure 2.1. Materials The material used in the friction stir welds considered in this investigation is fine grain Grade 5 (Ti–6Al–4V) titanium alloy
ARTICLE IN PRESS 386
M. Ramulu et al. / Optics and Lasers in Engineering 48 (2010) 385–392
sheets. The specific material used was provided by the Boeing Company. The original material base sample from which all research test specimens were generated was made from two 1.8 mm thick 101.6 mm wide 700 mm long sheets of identical alloy butt welded together along their respective long edges. As with the supplied material used, the friction stir-welding of the base sample was performed by the Boeing Company using its own FSW machine. Typical weld parameters used are shown in Table 1. While the specifics of the tool geometry are proprietary, the tool had the same conical pin and narrow shoulder typical of tungsten FSW tools intended for use on titanium. Fig. 1 shows a photograph of the friction stir welded sheets showing close-up views of the tool plunge and the tool extraction locations. Fig. 2
Table 1 Titanium friction stir welding parameters. Spindle speed Tool travel speed Forge load Tool plunge Depth Tool tilt Tool material Tool pin length Tool pin diameter Tool shoulder diameter
550 rpm 120 mm/min 15.6 kN 1.6 mm 31 from direction of travel Tungsten Pin and Anvil 1.4 mm 8.6 mm 15.9 mm
shows a photograph of the cross-section of a friction stir weld where the weld zone and depth of penetration can be clearly seen. 2.2. Specimen geometry This research used uniquely designed tensile test specimens made from friction stir-welded Ti–6Al–4V sheet. The specimens were designed such that the weld runs transverse to the direction of loading (long axis of the specimen) and through the center of the specimen gage section. A photograph of a tensile test specimen is shown in Fig. 3. The gage length is 36 mm and the cross-section is 2.0 mm wide. The dimensions of the specimen were not held in accord with ASTM E8-04 standards for tensile test specimens for two reasons: first, the equipment necessary to conduct the moire´ interferometry experiments would not accommodate a large load frame, so the specimen was required to fail at loads below 4448 N and second, the resolvable field of view of the camera and lenses used in the moire´ interferometry apparatus was on the order of 3.0 mm square. 2.3. Specimen preparation Specimen preparation consisted of a set of common processing steps for the specimens for the global measurements followed by unique processing steps for the specimens for the full-field strain
Fig. 1. Photograph of friction stir welded titanium sheet showing close-up views of the tool plunge (left top) and tool extraction (right top).
Fig. 2. Micrograph of friction stir weld cross-section.
Fig. 3. Tensile specimen with a 36 by 2 mm gage section. Also visible is the grating applied to gage section of the specimen where the weld traverses.
ARTICLE IN PRESS M. Ramulu et al. / Optics and Lasers in Engineering 48 (2010) 385–392
measurement processes. All research test specimens were first cut from the base FSW sheet by a high-pressure abrasive waterjet cutter. Once cut to shape, the burrs and slag formed on the upper weld surface by the FSW process and the remnant weld bead on the underside of the FSW surface were removed from each specimen using a grinding wheel. Each specimen was then polished to achieve a mechanically flat surface by adhering three to six specimens at a time to the finished face of an aluminum block and then hand sanding against a mechanically flat ceramic backing surface using silicon carbide sandpaper with a sequence of grit ratings from 60 to 120 grit until the friction stir-weld was no longer visible and the specimen had a uniform thickness. This level of initial processing left each specimen with typical final gage dimensions of 2.0 mm wide by 1.6 mm thick. Specimens processed to this extent were used for global stress–strain tensile tests; fullfield strain tests required further unique post-processing
´ interferometry specimens 2.4. Preparation of moire Preparation of moire´ interferometry specimens required further polishing to achieve as uniformly flat and as smooth a surface finish as possible for the grating application to minimize fringe distortion [15]. To this end, hand polishing continued from the initial common preparation through successively finer grit silicon carbide sandpaper from 150 to 2000 grit. Once polished, each moire´ specimen was drilled with a 2.0 mm diameter pin-hole in each tab to secure the specimens into the custom-made pinsecured load frame grips used during moire´ testing. The final and most important step in moire´ specimen preparation was the application of the specimen surface grating. This research used previously fabricated evaporated-aluminum dual-field gratings with symmetrical perpendicular patterns of 300 lines/mm. The grating was epoxy replicated to the gage surface of the specimen including the weld area as can be seen in Fig. 2. Subsequently, the grating surface was marked at the weld boundary to provide fiducials for imaging the weld area of interest.
387
3. Experimental set-up and analysis The following describes the experimental set-ups and the analysis methodology for the experiments performed. All testing was performed on specimens prepared using the same original welded titanium sheets. 3.1. Global strain measurement set-up In establishing the mechanical behavior of the friction stirwelded titanium alloy, it was first desired to establish the global stress–strain character of the subject welded material. To determine a baseline for the global stress–strain behavior of the subject material, four initially prepared specimens were dedicated to global stress–strain tensile tests. The global stress–strain test apparatus consisted of an Instron 5585H Floor Model Test System load frame with Instron 2716 series mechanical wedge style grips. The test parameters used in this testing were an extensioncontrolled increase rate 0.03 mm/s to a load-controlled maximum of 4448 N. Load and extensometer data were monitored and recorded throughout all tests performed. 3.2. Moire´ interferometry set-up The moire´ interferometry apparatus used in this research consisted of a load frame to induce strain in the specimen; a moire´ interferometer to generate moire´ fringes on the strained specimen; a camera and computer to capture the moire´ fringe imagery; and a load cell, data acquisition device, and a dedicated computer to measure applied load. The load frame used was an Instron model TM-M-L table-top model. Load frame control was performed by manual load frame adjustment based on desired load increments. This research used a custom-built, four-beam, dual-field, fiber optic, piezoelectric phase-shifted moire´ interferometer. Fig. 4 shows a photograph of the moire´ set-up. The laser, power supplies and phase shifter are not visible in the photograph. The moire´ interferometer used a
Fig. 4. Photograph of the moire´ interferometry test set-up.
ARTICLE IN PRESS M. Ramulu et al. / Optics and Lasers in Engineering 48 (2010) 385–392
620 nm wavelength helium–neon laser passed through a phaseshifter and split into four separate beams and transmitted through fiber optic cables into four separate fully adjustable collimator projector lenses set at angles to produce separate v-field and u-field reference gratings of 600 lines/mm The four-beam, dualfield apparatus used in this research simplified sequential capture of v-field and u-field imagery by allowing quick switching between v- and u-fields by blocking the horizontally aligned u-field or the vertically aligned v-field projector lenses, respectively. Moire´ imagery data collected by a Sony 1280 960 pixel 10-bit resolution XCD-SX910 monochromatic CCD camera using a 10x manual zoom lens with fixed focal length. A 451 first surface optically flat turning mirror was used to direct specimen and moire´ fringe imagery into the camera. Image data was captured on a computer by means of an IEEE 1394 interface and phase shifting was performed using custom software written by Perry [16]. Two laptops were used: one dedicated to the phase shifting and image collection software; and the other dedicated to the custom written Labview interface for farfield measurements using a custom data collection routine. This program captured load response data from the load cell in the Instron TM-M-L load frame channeled through a National Instruments DAQPad model 6015 USB multi-function data acquisition system.
3.3. Analysis procedure The methods of analysis used in this research differed significantly in the type of data collected and the available tools for analysis. The individual and unique methods of analysis for global stress–strain and moire´ interferometry are discussed in the following. Global specimen data for each test was collected by the Instron BlueHill 2 software and global stress and strain were calculated from the load and extensometer data. The first step in the moire´ data analysis was the conversion of raw moire´ imagery into high-fidelity gray scale wrapped moire´ fringe images. In order to achieve high-fidelity moire´ fringe images, each image capture process involved collecting five separated phase-shifted images and wrapping the images to improve resolution and reduce noise from surface imperfections, vibrations, etc. The approach of phase shifting (or fringe shifting) is well established and known for greatly improving moire´ data quality [15]. The result here is a 10-bit gray scale (1024 shades of gray) refinement of wrapped fringe images. Fig. 5 shows a sequence of images illustrating how the raw image and phase shift mask are combined to achieve a high-quality wrapped moire´ fringe image. With the known reference grating frequency and the specimen grating frequency, each single fringe, either in Fig. 5(a) or (c), represents 1.667 mm of relative displacement; however, change in the gray scale of the wrapped fringe image of Fig. 5(c) also represents finer resolution of displacement in that each relative change in the gray scale also represents relative displacement. Strain was calculated at areas of interest, which included the weld
nugget (WN) beginning 1.0 mm inside the weld border and moving in toward the center, the thermo-mechanically affected zone (TMAZ) spanning from the weld border to the WN, the heat affected zone (HAZ) spanning from the weld boundary to 1.0 mm outside of the weld border, and the parent material (PM) beginning at the other side of the HAZ and moving out from the center.
4. Results 4.1. Global stress–strain measurements Fig. 6 shows representative global stress–strain data taken from four specimens dedicated to this task. The yield and ultimate stresses show good accord with typically accepted values for titanium, though the failure strain values of 2.2–5.5% are significantly less than the Ti–6Al–4V expected 10–14% elongation at break. Comparison of the global data obtained in this research with previous research conducted at the University of Washington by Edwards [11] on similarly friction stir-welded Ti–6Al–4V material shows correlation of yield stress values within 9%. Experimental percent elongation from Mr. Edwards’ research, however, resulted in values from 4.4% to 12.1% versus the 2.2% to 6.2% elongation obtained here. Given that only twelve specimens were evaluated for this research effort and that their values for yield and ultimate strength were in accord with accepted and previous empirical data, the relatively low values for percent elongation may be indicative of the effects of the weld on the material properties and reflect the unique behavior of the particular samples of Ti–6Al–4V used. The difference from expected elongation may also be due to the presence of the FSW spanning a larger section of the gage section being spanned by the extensometer. 4.2. Moire´ strain measurements Nine specimens were prepared for data collection using moire´ interferometry for this research effort. It should be noted that although nine specimens were set aside specifically for moire´ testing, a number of specimens required multiple moire´-specific preparation due to failed initial attempts to apply a high quality surface grating. Of the nine successfully prepared moire´ specimens, two were used in setting up and verifying the experiment apparatus and were not used for data collection. The first of these set-up specimens experienced repeated loading/unloading cycles, which might have altered its strain behavior under load. Six of the 1200 1000 Stress (MPa)
388
800 600 400
TAG GS 1 TAG GS 2 TAG GS 3 TAG GS 4
200 0 0
Fig. 5. (a) Raw fringe patter image, (b) phase shift mask and (c) wrapped fringe image.
0.01
0.02 0.03 0.04 Strain (mm/mm)
0.05
0.06
Fig. 6. Representative global stress–strain behavior of the friction stir weld specimens.
ARTICLE IN PRESS M. Ramulu et al. / Optics and Lasers in Engineering 48 (2010) 385–392
remaining seven specimens produced sufficient data to hold the last specimen as reserve material for further testing if needed. The moire´ experiments followed a flexible approach to explore the strain behavior in elastic and plastic loading at the surface of a friction stir weld for variations at each of the different weld interface regions and to explore the strain behavior in both the axial (v-field) and transverse (u-field) directions for variations in transverse strain behavior. Local strains were determined by differentiating the full field displacement data and interrogating the results at key locations. The results of all six successful tests show distinct areas of elevated localized strain near the weld boundary developing near the yield strength and continuing through plastic loading. In two cases, testing continued until necking developed, which showed weld boundary strain localization at a maximum until overtaken as the maximum by the necking region associated with the point of ultimate failure. A summary of the strain data from these specimens is presented in Table 2. Fig. 7 presents a down-sampled sequence of typical moire´ specimen data during loading and unloading in the vicinity of the weld boundary. The wrapped moire´ fringes in the image progression indicate relative changes of displacements that correspond to changes in strain on the specimen surface. Fig. 7(a) shows a typical null image without any discernable fringes. Fig. 7(b) shows a typical elastic fringe pattern with consistent fringe spacing across the entire length of the specimen gage. Fig. 7(c) shows the advent of strain localization in the area of
389
tighter fringe spacing as the applied stress exceeds the yield stress. This localization continues to increase to a maximum shown in Fig. 7(d), which corresponds to the maximum applied stress before the specimen was unloaded. Fig. 7(f) shows the residual plastic strain in the fully unloaded specimen. Note the change in fringe spacing from the maximum applied stress image; the area of maximum localization, corresponding to the weld boundary, remains highly strained while the areas above and below, corresponding to the parent material and weld nugget, relax significantly more.
5. Discussion The foremost purpose of this research was to evaluate the stress–strain response of friction stir-welded Ti–6Al–4V in elastic and plastic loading conditions using global and full-field strain measurement techniques in order to demonstrate the utility in determining variations in strain behavior due to the presence of the weld. Table 3 summarizes the global specimen failure data and shows a distinct difference in the total strain depending on the location of the ultimate failure with respect to the weld boundary. The global specimen tests represent the lowest fidelity measurement technique used in this research and only present an average of the total strain across the gage section including the entire weld region and some of the parent material. When
Table 2 Moire´ specimen failure and maximum strain summary. Specimen
Calculated maximum strain Max stress (MPa)
TAGMS1 TAGMS2 TAGMS3 TAGMS4 TAGMS5 TAGMS6
1007 961 1007 866 1009 1009
Failure location (mm) N/A N/A N/A N/A 4.2R 1.0R
PM N/A N/A N/A 0.0094 0.0167 0.0123
HAZ 0.0173 0.0115 0.0173 0.0257 0.0200 0.0167
TMAZ a
0.0203 0.0153 0.0194a 0.0317a 0.0233a 0.0283a
WN
Total
0.0116 0.0213a 0.0110 0.0095 0.0233 0.0110
0.0172 0.0198 0.0163 0.0124 0.0344 0.0129
R= retreating side of weld. a
First region of visible strain localization.
Fig. 7. Sequence of wrapped moire´ fringe images during loading and unloading: (a) Load step 1:null, (b) load step 4:650 MPa, (c) load step 6:941 MPa, (d) load step 8:1007 MPa, (e) load step 11:610 MPa and (f) Load step 1 MPa.
ARTICLE IN PRESS 390
M. Ramulu et al. / Optics and Lasers in Engineering 48 (2010) 385–392
evaluated with the specific failure behavior of each specimen, however, the global specimen tests demonstrated that the stress– strain behavior differs substantially between specimens, which failed in or near the weld region versus those which failed well away from the weld boundary in the parent material. Noteworthy in the stress–strain behavior shown in Fig. 6 is indicative of a ductile failure and represents the specimens, which failed near the weld interface, defined in this research as within 5 mm of the visible weld boundary. These stress–strain curves demonstrated a gradual decline in stress magnitude and higher elongation in general before reaching ultimate failure. The second type of curve shape we observed (not shown here) was indicative of a less ductile failure and represents the specimens, which failed well into the parent material, arbitrarily defined in this research as more than 0.197 in. (5 mm) away from the weld boundary. Though this change in global strain behavior does not provide sufficient insight into material behavior to support specific conclusions, it does indicate changes in material properties based on proximity to the friction stir-weld. These results suggest that the region near the weld interface, which has been affected by the FSW process, is more ductile than the parent material. Typically, moire´ specimens were used to evaluate specific narrow areas of interest around the weld region. In a few instances, to provide a notional understanding of the strain manifestations at the different weld regions, multiple panned images were captured along the gage length to create a composite image of strain localization behavior across larger portions of the gage, as shown in Fig. 8. Fig. 8 shows a panoramic sequence of wrapped moire´ fringe images for both fields in the vicinity of the
retreating weld boundary. The discontinuities in the field occur from image to image in the sequence due to vibrations and are inherent in phase shifting. The approximate locations of the heat and thermo-mechanically affected zones are shown for reference. Strain localization is clearly apparent in the vicinity of the approximate HAZ and TMAZ with less strain in the PM and even less in the WN. Upon further loading well into the plastic region and subsequent unloading, the residual strains show significantly different manifestations in the various regions. Fig. 9 shows another composition of wrapped moire´ fringe images for both fields in the vicinity of the retreating weld boundary upon unloading. The PM has very little residual strain compared to the residual strain localized in the vicinity of the HAZ and the TMAZ. This information is instrumental in understanding the behavior of FSW joints and cannot be resolved with the global stress–strain measurements. Comparison of the results with the stress–strain curves derived from moire´ interferometry data shown in Fig. 10 shows agreement
Table 3 Global specimen failure and strain summary. Specimen
Maximum stress (MPa)
Failure from weld (mm)
Extensometer indicated strain
TAG TAG TAG TAG
1007 961 1007 866
0.0 6.3 8.0 1.3
0.0550 0.0225 0.0299 0.0530
GS GS GS GS
1 2 3 4
A R A R
A= advancing side of weld; R= retreating side of weld.
Fig. 9. Panoramic sequence of wrapped moire´ fringe images for both fields in the vicinity of the retreating weld boundary upon unloading. The approximate locations of the heat and thermo-mechanically affected zones are shown for reference.
Fig. 8. Panoramic sequence of wrapped Moire´ fringe images for both fields in the vicinity of the retreating weld boundary. The approximate locations of the heat and thermo-mechanically affected zones are shown for reference.
ARTICLE IN PRESS
Stress (MPa)
M. Ramulu et al. / Optics and Lasers in Engineering 48 (2010) 385–392
1200
6. Conclusions
1000
Experimental and numerical results of this research support the following conclusions:
800
Overall strength of friction stir-welded Ti–6Al–4V is compar-
600
able to the accepted values for mill-annealed Ti–6Al–4V.
Total Parent Material Weld Boundary Weld Nugget
400 200
Overall strain performance of friction stir-welded Ti–6Al–4V is
0 0
0.005
0.01 Strain (mm/mm)
0.015
0.02
Fig. 10. TAG MS 5 weld region stress–strain plot showing PM and WN values.
0.03
Applied Stress [MPa]
Strain [mm/mm]
391
0.02
1010 1010 1009 986 892 716 512 234
0.01
roughly half that of the accepted values for pure mill-annealed Ti–6Al–4V. Friction stir-welded Ti–6Al–4V demonstrates a consistent pattern of strain localization between the onset of yielding and ultimate failure. Following yielding, strain localizations develop at the weld boundaries and dominate the strain response during the initial stages of plateau loading. During plateau loading, strain localizations migrate away from the weld boundary and slightly into the parent material, where ultimate failure typically occurs. Even under plastic loading to near ultimate failure, the entire central portion of a friction stir-weld (the weld nugget) does not manifest significant strain when compared to the weld boundaries and parent material, nor does significant residual strain remain in the weld nugget after unloading as compared to the weld boundaries and parent material. Failure in friction stir-welded Ti–6Al–4V will likely occur at or just outside of the weld boundary with no apparent preference for advancing or retreating side of the weld. Moire´ interferometry provides a good means of characterizing specific areas of localization and quantifying atypical localized strain behavior.
0 Parent Material
Weld Boundary
Weld Nugget
Fig. 11. TAG MS 5. Strain versus approximate location with increasing applied stress.
in the onset of yielding in the WB versus the WN regions. PM stress–strain behavior data from the moire´ measurements was concentrated on the known areas of initial localization near the weld boundaries. Typically, the narrow frame region of the moire´ images allowed capture of the weld boundary and one other region without panning for composite images. Composite images were only captured at plateau loading or fully unloaded, which left limited data that captured both WN and PM behavior for a single specimen in the range of yielding. Though the stress–strain plots shown in Figs. 6 and 10 do not show the same amount of comparative data, when considered with typical moire´ fringe distribution around the weld region, a qualitative pattern of behavior is clearly illustrated: the WN and PM regions exhibit similar strain behavior until ultimate failure begins and the WB consistently exhibits lower yielding onset and higher localized strain. Fig. 11 shows a representative evolution of strain localization in the WB with increasing applied stress. The results were obtained by extracting wrapped moire´ data from the centerline of a sequence of images down the middle of the specimen at various loads, differentiating the relative pixel intensities to obtain the strain values and averaging to minimize the inherent noise in the results. It is apparent from the resulting curves that localization does not initiate at one discrete location in the WB, but rather over a short range. Subsequently, this range narrows and further extensive localization occurs typically to the side of the PM. This representation gives a clear view of the nonuniform nature of the localization.
Acknowledgements The authors are grateful to the Boeing Company for supplying the welded sheet of Titanium and specifically we extend our thanks to Dr. Daniel Sanders and Mr. Paul Edwards for their support and encouragement. In addition the author, M. Ramulu is very grateful to Professor A.P. Reynolds of University of South Carolina for his help in our earlier research. References [1] Mishra RS, Ma ZY. Friction stir welding and processing. Materials Science and Engineering R 2005;50:1–78. [2] Trapp T, Helder E, Subramanian PR. FSW of titanium alloys for aircraft engine components. TMS Annual Meeting, Friction Stir Welding and Processing II 2003:173–8. [3] Jata KV, Subramanian PR, Reynolds AP, Trapp T, Helder E. Friction stir welding of titanium alloys for aerospace applications: microstructure and mechanical behavior. In: Proceedings of the international offshore and polar engineering conference; 2004. p. 22–7. [4] Ramirez AJ, Juhas MC. Microstructural evolution in Ti–6Al–4V friction stir welds. Materials Science Forum V 2003;426–432:2999–3004. [5] Russell MJ, Freeman R. FSW for titanium. Metalworking production; May 2007, p. 30. [6] Lienert TJ. Microstructure and mechanical properties of friction stir welded titanium alloys. Friction stir welding and processing, ASM international, Chapter 7, p. 123–54. [7] Lee WB, Lee CY, Chang WS, Yeon YM, Jung SB. Microstructural investigation of friction stir welded pure titanium. Materials Letters 2005;59(26): 3315–3318. [8] Reynolds AP, Hood E, Tang W. Texture in friction stir welds of Timetal 21S. Scripta Materialla 2005;52(6):491–4. [9] Sanders D, Grant G, Ramulu M, Klock-McCook E, Leon L, Booker G, Foutch D, Reynolds T, Tang W. Superplastic forming of friction stir welded Ti 6-4 sheet. Presented at the ASM AeroMat Conference; 2005. [10] Klock-McCook EJ. Characterization of friction stir welded and superplasticly formed friction stir welded joints of titanium. Masters Thesis, University of Washington; 2005.
ARTICLE IN PRESS 392
M. Ramulu et al. / Optics and Lasers in Engineering 48 (2010) 385–392
[11] Edwards PD. Experimental and numerical characterization of friction stir welded and superplastically formed—friction stir welded titanium. Masters Thesis, University of Washington; 2006. [12] Zhang Y, Sato YS, Kokawa H, Park SHC, Hirano S. Microstructural characteristics and mechanical properties of Ti–6Al–4V friction stir welds. Materials Science and Engineering A 2008;485:448–55. [13] Sanders D, Ramulu M, Edwards P. Superplastic forming of friction stir welds in titanium alloy 6Al–4V: preliminary results. Materialwissenschaft und Werkstofftechnik 2008;39(4):353–7.
[14] Sanders D, Ramulu M, Klock-McCook E, Edwards P, Reynolds A, Trapp T. Characterization of superplastically formed friction stir weld in titanium 6Al–4V: preliminary results. ASM Journal of Materials Engineering and Performance 2008;17(2):187–92. [15] Post D, Han B, Ifju P. High sensitivity moire´: experimental analysis for mechanics and materials. New York: Springer-Verlag; 1994. [16] Kenneth E. Perry. EchoBio LLC., Bainbridge Island, Washington.
Contents lists available at ScienceDirect
Optics and Lasers in Engineering journal homepage: www.elsevier.com/locate/optlaseng
Elastic–plastic stress/strain response of friction stir-welded titanium butt joints using moire´ interferometry M. Ramulu a,, P. Labossiere b, T. Greenwell a a b
Department of Mechanical Engineering, University of Washington, Box 352600, Seattle, WA 98195, USA University of Manitoba, Winnipeg, Canada
a r t i c l e in fo
abstract
Article history: Received 28 July 2009 Received in revised form 3 October 2009 Accepted 5 October 2009 Available online 20 November 2009
An experimental investigation is conducted to examine, evaluate, and characterize the fundamental elastic–plastic stress/strain response of friction stir-welded butt joints in thin-sheet, fine grain Ti–6Al– 4V titanium alloy under normal tensile loading using the full-field optical strain analysis technique of moire´ interferometry. It was found that the overall strength of friction stir-welded Ti–6Al–4V is comparable to the accepted values for pure mill-annealed Ti–6Al–4V and the overall strain performance of friction stir-welded Ti–6Al–4V is roughly half that of the accepted values for pure mill-annealed Ti–6Al–4V. In addition, friction stir-welded Ti–6Al–4V demonstrates a consistent pattern of strain localization between the onset of yielding and ultimate failure. The resolvability of such strain localizations and their impact on far-field behavior can only be realized with full-field measurements such as those obtained using the moire´ interferometry technique. Published by Elsevier Ltd.
Keywords: Friction stir welding Titanium Moire´ interferometry Full-field measurements
1. Introduction Titanium (Ti) is currently an extremely important material to the aerospace industry. Its strength to weight ratio, operating temperature range and resistance to corrosion offer many advantages over aluminum alloys. Unfortunately joining titanium is very difficult to do with traditional fusion welding techniques and riveting can lead to poor tolerances. Titanium alloys on the other hand, are currently welded by a number of fusion welding processes, but some problems associated with fusion welding, such as formation of a brittle cast structure, large distortion and residual stress, still remain. Friction stir welding (FSW) of titanium alloys has emerged as one of the enabling technologies, allowing the fabrication of very large single piece components [1–3]. This technique avoids problems associated with the traditional fusion welding of titanium alloys since FSW is a solid-state joining process. Much has been accomplished in characterizing FSW joints in 2XXX and 7XXX series aluminum alloys for aerospace use, but little published research exists regarding FSW joints in titanium alloys. This is due to the more challenging issues associated with FSW of titanium alloys, such as the need for high performance, high temperature, wear resistant tool materials and well-controlled processing conditions due to the sensitivity of FSW to temperature variations. These added challenges limit the application of FSW titanium [4]; however, preliminary research indicates FSW of titanium alloys creates a robust weld with strength equal to or greater than the base metal and capable of further post-processing [2–6]. Corresponding author.
E-mail address: [email protected] (M. Ramulu). 0143-8166/$ - see front matter Published by Elsevier Ltd. doi:10.1016/j.optlaseng.2009.10.003
Only recently FSW has been applied successfully in joining of titanium [5–14], which has sparked much interest in the Aerospace industry as this process could play a major role in next generation manufacturing techniques. Recently reported research on FSW of titanium and titanium alloys have already shown some important knowledge on the microstructure and the properties in the welds [5,6,9–12], but effect of microstructure on mechanical properties of the weld has not been fully understood. Our Industry—University collaborative preliminary research has verified that FSW of titanium alloys creates a robust weld with good strength and allows further post-processing [10,11]. Review of currently available research finds global stress response and microstructural characterization of friction stir welded Ti–6Al–4V available, but no data is available, which describes the full-field stress–strain response of the weld-base metal interface. Therefore the purpose of this paper is to examine, evaluate, and characterize the fundamental elastic–plastic stress/strain response of friction stir-welded butt joints in thin-sheet, fine grain Ti–6Al–4V titanium alloy under normal tensile loading using the full-field optical strain analysis technique moire´ interferometry. The objective of the present study is to help clarify the relationship between microstructure and mechanical properties of friction stir welded Ti–6Al–4V alloy. 2. Experimental set-up and procedure 2.1. Materials The material used in the friction stir welds considered in this investigation is fine grain Grade 5 (Ti–6Al–4V) titanium alloy
ARTICLE IN PRESS 386
M. Ramulu et al. / Optics and Lasers in Engineering 48 (2010) 385–392
sheets. The specific material used was provided by the Boeing Company. The original material base sample from which all research test specimens were generated was made from two 1.8 mm thick 101.6 mm wide 700 mm long sheets of identical alloy butt welded together along their respective long edges. As with the supplied material used, the friction stir-welding of the base sample was performed by the Boeing Company using its own FSW machine. Typical weld parameters used are shown in Table 1. While the specifics of the tool geometry are proprietary, the tool had the same conical pin and narrow shoulder typical of tungsten FSW tools intended for use on titanium. Fig. 1 shows a photograph of the friction stir welded sheets showing close-up views of the tool plunge and the tool extraction locations. Fig. 2
Table 1 Titanium friction stir welding parameters. Spindle speed Tool travel speed Forge load Tool plunge Depth Tool tilt Tool material Tool pin length Tool pin diameter Tool shoulder diameter
550 rpm 120 mm/min 15.6 kN 1.6 mm 31 from direction of travel Tungsten Pin and Anvil 1.4 mm 8.6 mm 15.9 mm
shows a photograph of the cross-section of a friction stir weld where the weld zone and depth of penetration can be clearly seen. 2.2. Specimen geometry This research used uniquely designed tensile test specimens made from friction stir-welded Ti–6Al–4V sheet. The specimens were designed such that the weld runs transverse to the direction of loading (long axis of the specimen) and through the center of the specimen gage section. A photograph of a tensile test specimen is shown in Fig. 3. The gage length is 36 mm and the cross-section is 2.0 mm wide. The dimensions of the specimen were not held in accord with ASTM E8-04 standards for tensile test specimens for two reasons: first, the equipment necessary to conduct the moire´ interferometry experiments would not accommodate a large load frame, so the specimen was required to fail at loads below 4448 N and second, the resolvable field of view of the camera and lenses used in the moire´ interferometry apparatus was on the order of 3.0 mm square. 2.3. Specimen preparation Specimen preparation consisted of a set of common processing steps for the specimens for the global measurements followed by unique processing steps for the specimens for the full-field strain
Fig. 1. Photograph of friction stir welded titanium sheet showing close-up views of the tool plunge (left top) and tool extraction (right top).
Fig. 2. Micrograph of friction stir weld cross-section.
Fig. 3. Tensile specimen with a 36 by 2 mm gage section. Also visible is the grating applied to gage section of the specimen where the weld traverses.
ARTICLE IN PRESS M. Ramulu et al. / Optics and Lasers in Engineering 48 (2010) 385–392
measurement processes. All research test specimens were first cut from the base FSW sheet by a high-pressure abrasive waterjet cutter. Once cut to shape, the burrs and slag formed on the upper weld surface by the FSW process and the remnant weld bead on the underside of the FSW surface were removed from each specimen using a grinding wheel. Each specimen was then polished to achieve a mechanically flat surface by adhering three to six specimens at a time to the finished face of an aluminum block and then hand sanding against a mechanically flat ceramic backing surface using silicon carbide sandpaper with a sequence of grit ratings from 60 to 120 grit until the friction stir-weld was no longer visible and the specimen had a uniform thickness. This level of initial processing left each specimen with typical final gage dimensions of 2.0 mm wide by 1.6 mm thick. Specimens processed to this extent were used for global stress–strain tensile tests; fullfield strain tests required further unique post-processing
´ interferometry specimens 2.4. Preparation of moire Preparation of moire´ interferometry specimens required further polishing to achieve as uniformly flat and as smooth a surface finish as possible for the grating application to minimize fringe distortion [15]. To this end, hand polishing continued from the initial common preparation through successively finer grit silicon carbide sandpaper from 150 to 2000 grit. Once polished, each moire´ specimen was drilled with a 2.0 mm diameter pin-hole in each tab to secure the specimens into the custom-made pinsecured load frame grips used during moire´ testing. The final and most important step in moire´ specimen preparation was the application of the specimen surface grating. This research used previously fabricated evaporated-aluminum dual-field gratings with symmetrical perpendicular patterns of 300 lines/mm. The grating was epoxy replicated to the gage surface of the specimen including the weld area as can be seen in Fig. 2. Subsequently, the grating surface was marked at the weld boundary to provide fiducials for imaging the weld area of interest.
387
3. Experimental set-up and analysis The following describes the experimental set-ups and the analysis methodology for the experiments performed. All testing was performed on specimens prepared using the same original welded titanium sheets. 3.1. Global strain measurement set-up In establishing the mechanical behavior of the friction stirwelded titanium alloy, it was first desired to establish the global stress–strain character of the subject welded material. To determine a baseline for the global stress–strain behavior of the subject material, four initially prepared specimens were dedicated to global stress–strain tensile tests. The global stress–strain test apparatus consisted of an Instron 5585H Floor Model Test System load frame with Instron 2716 series mechanical wedge style grips. The test parameters used in this testing were an extensioncontrolled increase rate 0.03 mm/s to a load-controlled maximum of 4448 N. Load and extensometer data were monitored and recorded throughout all tests performed. 3.2. Moire´ interferometry set-up The moire´ interferometry apparatus used in this research consisted of a load frame to induce strain in the specimen; a moire´ interferometer to generate moire´ fringes on the strained specimen; a camera and computer to capture the moire´ fringe imagery; and a load cell, data acquisition device, and a dedicated computer to measure applied load. The load frame used was an Instron model TM-M-L table-top model. Load frame control was performed by manual load frame adjustment based on desired load increments. This research used a custom-built, four-beam, dual-field, fiber optic, piezoelectric phase-shifted moire´ interferometer. Fig. 4 shows a photograph of the moire´ set-up. The laser, power supplies and phase shifter are not visible in the photograph. The moire´ interferometer used a
Fig. 4. Photograph of the moire´ interferometry test set-up.
ARTICLE IN PRESS M. Ramulu et al. / Optics and Lasers in Engineering 48 (2010) 385–392
620 nm wavelength helium–neon laser passed through a phaseshifter and split into four separate beams and transmitted through fiber optic cables into four separate fully adjustable collimator projector lenses set at angles to produce separate v-field and u-field reference gratings of 600 lines/mm The four-beam, dualfield apparatus used in this research simplified sequential capture of v-field and u-field imagery by allowing quick switching between v- and u-fields by blocking the horizontally aligned u-field or the vertically aligned v-field projector lenses, respectively. Moire´ imagery data collected by a Sony 1280 960 pixel 10-bit resolution XCD-SX910 monochromatic CCD camera using a 10x manual zoom lens with fixed focal length. A 451 first surface optically flat turning mirror was used to direct specimen and moire´ fringe imagery into the camera. Image data was captured on a computer by means of an IEEE 1394 interface and phase shifting was performed using custom software written by Perry [16]. Two laptops were used: one dedicated to the phase shifting and image collection software; and the other dedicated to the custom written Labview interface for farfield measurements using a custom data collection routine. This program captured load response data from the load cell in the Instron TM-M-L load frame channeled through a National Instruments DAQPad model 6015 USB multi-function data acquisition system.
3.3. Analysis procedure The methods of analysis used in this research differed significantly in the type of data collected and the available tools for analysis. The individual and unique methods of analysis for global stress–strain and moire´ interferometry are discussed in the following. Global specimen data for each test was collected by the Instron BlueHill 2 software and global stress and strain were calculated from the load and extensometer data. The first step in the moire´ data analysis was the conversion of raw moire´ imagery into high-fidelity gray scale wrapped moire´ fringe images. In order to achieve high-fidelity moire´ fringe images, each image capture process involved collecting five separated phase-shifted images and wrapping the images to improve resolution and reduce noise from surface imperfections, vibrations, etc. The approach of phase shifting (or fringe shifting) is well established and known for greatly improving moire´ data quality [15]. The result here is a 10-bit gray scale (1024 shades of gray) refinement of wrapped fringe images. Fig. 5 shows a sequence of images illustrating how the raw image and phase shift mask are combined to achieve a high-quality wrapped moire´ fringe image. With the known reference grating frequency and the specimen grating frequency, each single fringe, either in Fig. 5(a) or (c), represents 1.667 mm of relative displacement; however, change in the gray scale of the wrapped fringe image of Fig. 5(c) also represents finer resolution of displacement in that each relative change in the gray scale also represents relative displacement. Strain was calculated at areas of interest, which included the weld
nugget (WN) beginning 1.0 mm inside the weld border and moving in toward the center, the thermo-mechanically affected zone (TMAZ) spanning from the weld border to the WN, the heat affected zone (HAZ) spanning from the weld boundary to 1.0 mm outside of the weld border, and the parent material (PM) beginning at the other side of the HAZ and moving out from the center.
4. Results 4.1. Global stress–strain measurements Fig. 6 shows representative global stress–strain data taken from four specimens dedicated to this task. The yield and ultimate stresses show good accord with typically accepted values for titanium, though the failure strain values of 2.2–5.5% are significantly less than the Ti–6Al–4V expected 10–14% elongation at break. Comparison of the global data obtained in this research with previous research conducted at the University of Washington by Edwards [11] on similarly friction stir-welded Ti–6Al–4V material shows correlation of yield stress values within 9%. Experimental percent elongation from Mr. Edwards’ research, however, resulted in values from 4.4% to 12.1% versus the 2.2% to 6.2% elongation obtained here. Given that only twelve specimens were evaluated for this research effort and that their values for yield and ultimate strength were in accord with accepted and previous empirical data, the relatively low values for percent elongation may be indicative of the effects of the weld on the material properties and reflect the unique behavior of the particular samples of Ti–6Al–4V used. The difference from expected elongation may also be due to the presence of the FSW spanning a larger section of the gage section being spanned by the extensometer. 4.2. Moire´ strain measurements Nine specimens were prepared for data collection using moire´ interferometry for this research effort. It should be noted that although nine specimens were set aside specifically for moire´ testing, a number of specimens required multiple moire´-specific preparation due to failed initial attempts to apply a high quality surface grating. Of the nine successfully prepared moire´ specimens, two were used in setting up and verifying the experiment apparatus and were not used for data collection. The first of these set-up specimens experienced repeated loading/unloading cycles, which might have altered its strain behavior under load. Six of the 1200 1000 Stress (MPa)
388
800 600 400
TAG GS 1 TAG GS 2 TAG GS 3 TAG GS 4
200 0 0
Fig. 5. (a) Raw fringe patter image, (b) phase shift mask and (c) wrapped fringe image.
0.01
0.02 0.03 0.04 Strain (mm/mm)
0.05
0.06
Fig. 6. Representative global stress–strain behavior of the friction stir weld specimens.
ARTICLE IN PRESS M. Ramulu et al. / Optics and Lasers in Engineering 48 (2010) 385–392
remaining seven specimens produced sufficient data to hold the last specimen as reserve material for further testing if needed. The moire´ experiments followed a flexible approach to explore the strain behavior in elastic and plastic loading at the surface of a friction stir weld for variations at each of the different weld interface regions and to explore the strain behavior in both the axial (v-field) and transverse (u-field) directions for variations in transverse strain behavior. Local strains were determined by differentiating the full field displacement data and interrogating the results at key locations. The results of all six successful tests show distinct areas of elevated localized strain near the weld boundary developing near the yield strength and continuing through plastic loading. In two cases, testing continued until necking developed, which showed weld boundary strain localization at a maximum until overtaken as the maximum by the necking region associated with the point of ultimate failure. A summary of the strain data from these specimens is presented in Table 2. Fig. 7 presents a down-sampled sequence of typical moire´ specimen data during loading and unloading in the vicinity of the weld boundary. The wrapped moire´ fringes in the image progression indicate relative changes of displacements that correspond to changes in strain on the specimen surface. Fig. 7(a) shows a typical null image without any discernable fringes. Fig. 7(b) shows a typical elastic fringe pattern with consistent fringe spacing across the entire length of the specimen gage. Fig. 7(c) shows the advent of strain localization in the area of
389
tighter fringe spacing as the applied stress exceeds the yield stress. This localization continues to increase to a maximum shown in Fig. 7(d), which corresponds to the maximum applied stress before the specimen was unloaded. Fig. 7(f) shows the residual plastic strain in the fully unloaded specimen. Note the change in fringe spacing from the maximum applied stress image; the area of maximum localization, corresponding to the weld boundary, remains highly strained while the areas above and below, corresponding to the parent material and weld nugget, relax significantly more.
5. Discussion The foremost purpose of this research was to evaluate the stress–strain response of friction stir-welded Ti–6Al–4V in elastic and plastic loading conditions using global and full-field strain measurement techniques in order to demonstrate the utility in determining variations in strain behavior due to the presence of the weld. Table 3 summarizes the global specimen failure data and shows a distinct difference in the total strain depending on the location of the ultimate failure with respect to the weld boundary. The global specimen tests represent the lowest fidelity measurement technique used in this research and only present an average of the total strain across the gage section including the entire weld region and some of the parent material. When
Table 2 Moire´ specimen failure and maximum strain summary. Specimen
Calculated maximum strain Max stress (MPa)
TAGMS1 TAGMS2 TAGMS3 TAGMS4 TAGMS5 TAGMS6
1007 961 1007 866 1009 1009
Failure location (mm) N/A N/A N/A N/A 4.2R 1.0R
PM N/A N/A N/A 0.0094 0.0167 0.0123
HAZ 0.0173 0.0115 0.0173 0.0257 0.0200 0.0167
TMAZ a
0.0203 0.0153 0.0194a 0.0317a 0.0233a 0.0283a
WN
Total
0.0116 0.0213a 0.0110 0.0095 0.0233 0.0110
0.0172 0.0198 0.0163 0.0124 0.0344 0.0129
R= retreating side of weld. a
First region of visible strain localization.
Fig. 7. Sequence of wrapped moire´ fringe images during loading and unloading: (a) Load step 1:null, (b) load step 4:650 MPa, (c) load step 6:941 MPa, (d) load step 8:1007 MPa, (e) load step 11:610 MPa and (f) Load step 1 MPa.
ARTICLE IN PRESS 390
M. Ramulu et al. / Optics and Lasers in Engineering 48 (2010) 385–392
evaluated with the specific failure behavior of each specimen, however, the global specimen tests demonstrated that the stress– strain behavior differs substantially between specimens, which failed in or near the weld region versus those which failed well away from the weld boundary in the parent material. Noteworthy in the stress–strain behavior shown in Fig. 6 is indicative of a ductile failure and represents the specimens, which failed near the weld interface, defined in this research as within 5 mm of the visible weld boundary. These stress–strain curves demonstrated a gradual decline in stress magnitude and higher elongation in general before reaching ultimate failure. The second type of curve shape we observed (not shown here) was indicative of a less ductile failure and represents the specimens, which failed well into the parent material, arbitrarily defined in this research as more than 0.197 in. (5 mm) away from the weld boundary. Though this change in global strain behavior does not provide sufficient insight into material behavior to support specific conclusions, it does indicate changes in material properties based on proximity to the friction stir-weld. These results suggest that the region near the weld interface, which has been affected by the FSW process, is more ductile than the parent material. Typically, moire´ specimens were used to evaluate specific narrow areas of interest around the weld region. In a few instances, to provide a notional understanding of the strain manifestations at the different weld regions, multiple panned images were captured along the gage length to create a composite image of strain localization behavior across larger portions of the gage, as shown in Fig. 8. Fig. 8 shows a panoramic sequence of wrapped moire´ fringe images for both fields in the vicinity of the
retreating weld boundary. The discontinuities in the field occur from image to image in the sequence due to vibrations and are inherent in phase shifting. The approximate locations of the heat and thermo-mechanically affected zones are shown for reference. Strain localization is clearly apparent in the vicinity of the approximate HAZ and TMAZ with less strain in the PM and even less in the WN. Upon further loading well into the plastic region and subsequent unloading, the residual strains show significantly different manifestations in the various regions. Fig. 9 shows another composition of wrapped moire´ fringe images for both fields in the vicinity of the retreating weld boundary upon unloading. The PM has very little residual strain compared to the residual strain localized in the vicinity of the HAZ and the TMAZ. This information is instrumental in understanding the behavior of FSW joints and cannot be resolved with the global stress–strain measurements. Comparison of the results with the stress–strain curves derived from moire´ interferometry data shown in Fig. 10 shows agreement
Table 3 Global specimen failure and strain summary. Specimen
Maximum stress (MPa)
Failure from weld (mm)
Extensometer indicated strain
TAG TAG TAG TAG
1007 961 1007 866
0.0 6.3 8.0 1.3
0.0550 0.0225 0.0299 0.0530
GS GS GS GS
1 2 3 4
A R A R
A= advancing side of weld; R= retreating side of weld.
Fig. 9. Panoramic sequence of wrapped moire´ fringe images for both fields in the vicinity of the retreating weld boundary upon unloading. The approximate locations of the heat and thermo-mechanically affected zones are shown for reference.
Fig. 8. Panoramic sequence of wrapped Moire´ fringe images for both fields in the vicinity of the retreating weld boundary. The approximate locations of the heat and thermo-mechanically affected zones are shown for reference.
ARTICLE IN PRESS
Stress (MPa)
M. Ramulu et al. / Optics and Lasers in Engineering 48 (2010) 385–392
1200
6. Conclusions
1000
Experimental and numerical results of this research support the following conclusions:
800
Overall strength of friction stir-welded Ti–6Al–4V is compar-
600
able to the accepted values for mill-annealed Ti–6Al–4V.
Total Parent Material Weld Boundary Weld Nugget
400 200
Overall strain performance of friction stir-welded Ti–6Al–4V is
0 0
0.005
0.01 Strain (mm/mm)
0.015
0.02
Fig. 10. TAG MS 5 weld region stress–strain plot showing PM and WN values.
0.03
Applied Stress [MPa]
Strain [mm/mm]
391
0.02
1010 1010 1009 986 892 716 512 234
0.01
roughly half that of the accepted values for pure mill-annealed Ti–6Al–4V. Friction stir-welded Ti–6Al–4V demonstrates a consistent pattern of strain localization between the onset of yielding and ultimate failure. Following yielding, strain localizations develop at the weld boundaries and dominate the strain response during the initial stages of plateau loading. During plateau loading, strain localizations migrate away from the weld boundary and slightly into the parent material, where ultimate failure typically occurs. Even under plastic loading to near ultimate failure, the entire central portion of a friction stir-weld (the weld nugget) does not manifest significant strain when compared to the weld boundaries and parent material, nor does significant residual strain remain in the weld nugget after unloading as compared to the weld boundaries and parent material. Failure in friction stir-welded Ti–6Al–4V will likely occur at or just outside of the weld boundary with no apparent preference for advancing or retreating side of the weld. Moire´ interferometry provides a good means of characterizing specific areas of localization and quantifying atypical localized strain behavior.
0 Parent Material
Weld Boundary
Weld Nugget
Fig. 11. TAG MS 5. Strain versus approximate location with increasing applied stress.
in the onset of yielding in the WB versus the WN regions. PM stress–strain behavior data from the moire´ measurements was concentrated on the known areas of initial localization near the weld boundaries. Typically, the narrow frame region of the moire´ images allowed capture of the weld boundary and one other region without panning for composite images. Composite images were only captured at plateau loading or fully unloaded, which left limited data that captured both WN and PM behavior for a single specimen in the range of yielding. Though the stress–strain plots shown in Figs. 6 and 10 do not show the same amount of comparative data, when considered with typical moire´ fringe distribution around the weld region, a qualitative pattern of behavior is clearly illustrated: the WN and PM regions exhibit similar strain behavior until ultimate failure begins and the WB consistently exhibits lower yielding onset and higher localized strain. Fig. 11 shows a representative evolution of strain localization in the WB with increasing applied stress. The results were obtained by extracting wrapped moire´ data from the centerline of a sequence of images down the middle of the specimen at various loads, differentiating the relative pixel intensities to obtain the strain values and averaging to minimize the inherent noise in the results. It is apparent from the resulting curves that localization does not initiate at one discrete location in the WB, but rather over a short range. Subsequently, this range narrows and further extensive localization occurs typically to the side of the PM. This representation gives a clear view of the nonuniform nature of the localization.
Acknowledgements The authors are grateful to the Boeing Company for supplying the welded sheet of Titanium and specifically we extend our thanks to Dr. Daniel Sanders and Mr. Paul Edwards for their support and encouragement. In addition the author, M. Ramulu is very grateful to Professor A.P. Reynolds of University of South Carolina for his help in our earlier research. References [1] Mishra RS, Ma ZY. Friction stir welding and processing. Materials Science and Engineering R 2005;50:1–78. [2] Trapp T, Helder E, Subramanian PR. FSW of titanium alloys for aircraft engine components. TMS Annual Meeting, Friction Stir Welding and Processing II 2003:173–8. [3] Jata KV, Subramanian PR, Reynolds AP, Trapp T, Helder E. Friction stir welding of titanium alloys for aerospace applications: microstructure and mechanical behavior. In: Proceedings of the international offshore and polar engineering conference; 2004. p. 22–7. [4] Ramirez AJ, Juhas MC. Microstructural evolution in Ti–6Al–4V friction stir welds. Materials Science Forum V 2003;426–432:2999–3004. [5] Russell MJ, Freeman R. FSW for titanium. Metalworking production; May 2007, p. 30. [6] Lienert TJ. Microstructure and mechanical properties of friction stir welded titanium alloys. Friction stir welding and processing, ASM international, Chapter 7, p. 123–54. [7] Lee WB, Lee CY, Chang WS, Yeon YM, Jung SB. Microstructural investigation of friction stir welded pure titanium. Materials Letters 2005;59(26): 3315–3318. [8] Reynolds AP, Hood E, Tang W. Texture in friction stir welds of Timetal 21S. Scripta Materialla 2005;52(6):491–4. [9] Sanders D, Grant G, Ramulu M, Klock-McCook E, Leon L, Booker G, Foutch D, Reynolds T, Tang W. Superplastic forming of friction stir welded Ti 6-4 sheet. Presented at the ASM AeroMat Conference; 2005. [10] Klock-McCook EJ. Characterization of friction stir welded and superplasticly formed friction stir welded joints of titanium. Masters Thesis, University of Washington; 2005.
ARTICLE IN PRESS 392
M. Ramulu et al. / Optics and Lasers in Engineering 48 (2010) 385–392
[11] Edwards PD. Experimental and numerical characterization of friction stir welded and superplastically formed—friction stir welded titanium. Masters Thesis, University of Washington; 2006. [12] Zhang Y, Sato YS, Kokawa H, Park SHC, Hirano S. Microstructural characteristics and mechanical properties of Ti–6Al–4V friction stir welds. Materials Science and Engineering A 2008;485:448–55. [13] Sanders D, Ramulu M, Edwards P. Superplastic forming of friction stir welds in titanium alloy 6Al–4V: preliminary results. Materialwissenschaft und Werkstofftechnik 2008;39(4):353–7.
[14] Sanders D, Ramulu M, Klock-McCook E, Edwards P, Reynolds A, Trapp T. Characterization of superplastically formed friction stir weld in titanium 6Al–4V: preliminary results. ASM Journal of Materials Engineering and Performance 2008;17(2):187–92. [15] Post D, Han B, Ifju P. High sensitivity moire´: experimental analysis for mechanics and materials. New York: Springer-Verlag; 1994. [16] Kenneth E. Perry. EchoBio LLC., Bainbridge Island, Washington.