Friction Stir Welding Dem...

Friction Stir Welding Demonstrated for Combat Vehicle Construction
Improved tools and process parameters were used to fabricate structures of 2519 aluminum armor for the U.S. Marine Corps' Advanced Amphibious Assault Vehicle

By Kevin J. Colligan, Paul J. Konkol, James J. Fisher, AND Joseph R. Pickens

The Advanced Amphibious Assault Vehicle (AAAV) is an armored personnel carrier under development by General Dynamics Amphibious Systems for the U.S. Marine Corps. Concurrent Technologies Corp. (CTC), through the National Center for Excellence in Metalworking Technology (NCEMT), recently completed the second phase of a project to evaluate the feasibility of friction stir welding (FSW) to enhance weld performance and reduce fabrication costs for the AAAV.
The U.S. Marine Corps' Advanced Amphibious Assault Vehicle. (Above)
In Phase I, single-pass butt-joint welds were made at 1.6 in./min travel speed in 1-in. 2519-T87 aluminum armor plate that exhibited dramatically superior strength and ductility than those made by gas metal arc welding (GMAW) (Ref. 1). Ballistic shock test panels made using FSW butt-joint welds passed the ballistic shock impact test at velocities 30% more than MIL-STD-1946A requirements. Conventional 2519 GMAW butt-joint welds have never passed this demanding test requirement. Similar results were reported by General Dynamics Land Systems and the Edison Welding Institute for two-pass FSW butt-joint welds in 1.25-in. 2519 plates (Ref. 2).

In Phase II, the objective was to fabricate and test an AAAV floor section (used to verify mine blast performance of the AAAV hull) that would incorporate both FSW joints and extruded 2519 T-stiffeners. Design and construction of the test section required consideration of many practical manufacturing issues, including development of procedures for joining the floor to the sidewalls by friction stir welded corner joints, FSW around the chine actuator mount, and FSW of the floor panel to the lower glacis.

Fig. 1 - Schematic of the friction stir welding process.
Improved tool designs were developed and process parameters were optimized to increase the travel speed during FSW while improving the mechanical properties attained in Phase I. This article describes FSW process parameter and welding tool optimization, fabrication and testing of the ballistic shock test panels, and construction of the AAAV floor test section.

Process Description
Friction stir welding is a solid-state joining process in which a wear-resistant, rotating, frustum (truncated cone) pin tool is plunged into the joint and traversed along the joint line - Fig. 1. The welding tool is comprised of a specially profiled pin, which plunges nearly through the workpiece, and a shoulder that rides on the surface of the plate. Heat generated by a combination of frictional heating and plastic deformation of the workpiece softens the material adjacent to the tool to a temperature approaching the solidus, where no generalized melting is observed. Once the pin is plunged into the joint, the tool traverses along the joint, "stirring" the interface and producing a solid-state weld. Because no discernable melting occurs, FSW can join materials that suffer from solidification-related defects, for example, nonfusion-weldable 2xxx and 7xxx aluminum alloys. The solid-state nature of FSW offers other advantages, including improved mechanical properties; elimination of welding fumes, porosity and spatter; low shrinkage; and reduced weld distortion. Furthermore, the process can be performed in a single pass and in all workpiece positions.

Fig. 2 - Improved FSW tool design.
Optimized Parameters
Improved tool designs and process parameters were developed to enable increased travel speeds while also improving strength and maintaining the excellent ductility attained in the Phase I weldments. The plate used for this work was nominally 1-in.-thick 2519-T87. The tool used in Phase I was a monolithic design made from H13 tool steel. The Phase II tools were derived from ones developed by The Welding Institute (TWI) for thick-section plate (Ref. 3). These two-piece tools (Fig. 2) featured pins made of MP159, which provides high strength at the temperatures reached during FSW of aluminum, and shoulders made of H13 tool steel. The Phase II welding tools used a frustum-shaped pin profile, each with three equally spaced flats machined into the profiled surface, and a scroll or spiral shoulder profile. The scroll shoulder design enables welding without tilting the welding tool relative to the workpiece, which facilitates welding around corners. Additionally, the pins were flat ended, which helps produce a better stir zone or weld nugget penetration to the back of the workpiece.
Fig. 3 - Effects of travel speed on transverse tensile strength in single-pass 1-in. 2519 FS weldments.
Three welding tools were studied in Phase II in order to evaluate differences in shoulder geometry and material. Each tool was used to produce welds at different travel speeds to determine the strength and ductility of the joint produced. Initially, welds were evaluated at travel speeds between 1.2 and 2 in./min. After reviewing the tensile test results, one tool design was tested at travel speeds between 2.5 and 4 in./min, using a higher spindle speed. After each trial, five transverse tensile test specimens and three metallographic specimens were cut from each weldment. The tensile specimens were 1 in. wide by 12 in. long by full plate thickness without a reduced gauge section. Selected weldments were also radiographically inspected in accordance with MIL-STD-2035A, Nondestructive Testing Acceptance Criteria.

The tensile results are summarized in Figs. 35, which show the progression in the transverse weld tensile strength, yield strength, and elongation, respectively, as a function of travel speed. The figures also show the minimum yield, tensile, and elongation values established for conventional gas metal arc welds in 2519 aluminum for the AAAV. In Phase II, the new tool designs allowed travel speeds up to 4 in./min with the following excellent combination of properties: 56.4 ksi tensile strength, 33.8 ksi yield strength, and 13.6% elongation. These values represent improvements of 48, 69, and 289% in tensile, yield, and elongation, respectively, compared to the minimum values established for GMAW. Although travel speeds were increased more than three times over those in Phase I, tensile strength increased 8 ksi while maintaining excellent elongation of nearly 14%. The increase in weld strength was attributed to the reduced heat input per unit length of weld, which resulted in less overaging of the HAZ, where the majority of the tensile fractures occurred. Transverse weld elongation increased or decreased with weld travel speed (Fig. 5), depending on tool geometry. It is interesting to note that tool design D produced welds that had increasing elongation with increasing travel speed. Another important consideration in judging a welding tool design is the forces required to produce welds. The high forces required to make friction stir welds mean the structure of the welding machine and the fixtures holding the work must be robust. Compared to the Phase I tool, the welding tools tested in Phase II reduced the axial force required to make a weld from approximately 16,000 to 9000 lb.

Fig. 4 - Effect of travel speed on transverse yield strength in single-pass 1-in. 2519 FS weldments.
It should be noted the tensile fractures were all located in the HAZ in welds produced at travel speeds of 3 in./min or less. All the welds produced at 3.5 or 4 in./min had tensile fractures in the stir zone, although tensile strengths were the highest observed and ductility was also high. It is possible that at speeds of 3 in./min and lower, the strength is dominated by the metallurgical response of the HAZ to overaging, while at speeds over 3 in./min, increasing HAZ strength shifts the fracture location to the next weakest location, which apparently is the stir zone. However, additional microscopic examination by transmission electron microscopy is warranted to further explain these results.

A transverse metallographic specimen of a 1-in.-thick 2519 weldment made at 4 in./min travel speed is shown in Fig. 6. The stir zone (SZ) was relatively narrow and straight sided and there was a very narrow thermo-mechanically affected zone (TMAZ) compared to that produced by more conventional welding tool designs. As is typical, the advancing side SZ boundary was sharp compared to the retreating side boundary.

Fig. 5 - Effect of travel speed on transverse elongation in single-pass 1-in. 2519 FS weldments.
Corner Weld Process Development
Construction of complex structural components such as those for the AAAV often requires various joint configurations, such as 90-deg corner joints. This type of weld requires a welding fixture to hold the plates in the perpendicular orientation during welding and to react to the applied and thermal-expansion forces produced in the welding process. Two types of joints were evaluated: butted corner and rabbeted corner. Both joint designs are shown in Fig. 7. A 1- to 2-in. joint is shown, which represents the joining of the 1-in. floor section to the 2-in. sidewalls on the AAAV floor test assembly; however, only 1-in. plates were used during weld process development.
Fig. 6 - Transverse metallographic section of FSW weldment .
The rabbet joint is similar to the traditional joint design used in conventional arc welding. For ballistic impact applications, however, it leaves an unwelded segment between the horizontal and vertical members. In arc welding, this is compensated for by depositing a fillet weld on the inside corner to transfer the loads. The butted corner joint requires less preparation but requires special tooling to support the horizontal leg by means of an anvil fitted to the inside corner.

Process development was conducted initially using prototype tooling to verify the tooling design. Initial trials were conducted using 1- X 6- X 30-in. plates. Figure 8 shows the plates in the prototype tooling during the initial plunge into the workpiece.

Fig. 7 - Corner joint designs.
Once the tooling concepts were verified and process parameters developed for 2519 aluminum, a larger fixture was made to accommodate 54-in.-long corner ballistic shock test weldments made between 1- and 2-in. plates using the butt corner joint design in Fig. 8. Two additional weldments were made using the rabbeted joint design. All welds were made at a travel speed of 4 in./min. The weldments were trimmed to 48 in. long and shipped to the U.S. Army's Aberdeen Proving Ground for ballistic shock testing.

Ballistic Shock Testing
In addition to the butted and rabbeted corner panels, flat butt-joint weld ballistic shock test panels were made in 1-in. 2519-T87 plates at a travel speed of 4 in./min. Ballistic shock impact testing was performed per MIL-STD-1946A (Ref. 4). The test evaluates the performance of welds under high strain rate loading by firing a 75-mm-diameter by 150-mm-long soft aluminum slug into the joint at a specified velocity. To pass, the length of any cracks produced in the joint must total less than 12 in. Results of the test, which is typically performed on flat and 90-deg corner joints, are summarized in Table 1.

The friction stir welded flat-panel butt-joint weld made in Phase II passed the ballistic shock test, as did the Phase I weldments. The two 90-deg corner butt-joint weld panels passed the test on both the 1- and 2-in. faces. Acceptable performance of the butt corner joints is significant because use of the butt corner presumably reduces production costs by eliminating the machining of the rabbet in the sidewall. Note that the butt corner joint design places the point of maximum shear loading on the center of the weld, but the superior deformation and fracture properties of the FSW stir zone, created by the fine-grained microstructure, enable it to pass the test.

Fig. 8 - One-in. butt corner joint during FSW in prototype fixture.
The rabbet corner weld panels failed the test due to excessive cracking. This weld joint design, which had a 0.675-in. rabbet, placed the HAZ of the weld at the corner of the intersection of the two plates, which is also the location of the maximum shear stress during ballistic shock testing, as shown in Fig. 9. The HAZ in heat-treatable aluminum alloys is typically softened due to overaging of the strengthening precipitates and recovery of any work hardening present in the base metal. The fracture path during ballistic shock testing of other alloys has been shown to initiate in this softened zone (Refs. 5, 6). It is likely a joint with a shallower or deeper rabbet would pass the test, as this would place the maximum shear plane away from the softened HAZ.

Fabrication of Floor Test Assembly
The floor mine blast test was designed to demonstrate the ability of the AAAV hull to withstand shock loading from a mine detonation beneath the vehicle. The test assembly represents the forward third of the AAAV and contains plates, stiffeners, and structural details representative of the actual vehicle.

Figure 10 shows a schematic of the floor test assembly. The 1-in.-thick floor plate and lower glacis (the lower, forward portion of the vehicle) and 2-in-thick sidewalls were built of 2519-T87 aluminum plate. The chine actuator mounts were fabricated by GMAW from 1- and 2-in. plate components. CTC personnel made all the welds at the Materials and Processes Welding Laboratory of The Boeing Co., Huntington Beach, Calif. <

Fig. 9 - Rabbet 90-deg corner geometry.
FSW is most easily performed on flat butt joints. Although it is possible to make friction stir welds between angled plates, it was decided this approach added risk without significant benefit to the project. The floor test article required the lower glacis to be oriented at 12 deg relative to the floor. To accommodate the requirements of friction stir welding, the floor was welded to the lower glacis by joining each separately to a newly designed transition piece machined from 2-in. plate and at a 12-deg angle. The two plates were prevented from lifting by a surface roller used on the welding machine. This roller rode just in front of the welding tool and pressed down on the plates with approximately 1000 lb force. Once joining the transition piece to the lower glacis was completed, the fixture was reconfigured for joining the lower glacis/transition piece assembly to the floor plate in a similar manner.

The general procedure to fabricate the weld that joined the floor to the sidewall and chine actuator mount was to plunge in at the top of the 12-deg ramp and weld toward the chine actuator mount. As the welding tool reached the chine actuator mount, it made a 90-deg turn at each corner until it returned to the floor/sidewall joint. The weld then proceeded along the floor/sidewall joint to the end of the panel, where the welding tool was driven onto a run-off tab and retracted.

An earlier practice section showed the floor tended to separate from the sidewall as the tool passed around the chine actuator mount, probably resulting from the forces of thermal expansion and weld shrinkage in the actuator mount segments of the weld. As a result, the complete-joint-penetration welds between the floor and sidewall/chine actuator were preceded by 0.2-in.-penetration FS tack welds along the floor to sidewall segments. This completely eliminated any relative movement between the floor and sidewall.

Fig. 10 - Floor test assembly.
The last welds to be made were the two 14-in.-long segments between the lower glacis and the two sidewalls. To fabricate these welds, the welding machine spindle assembly was tilted to a 12-deg orientation and the weld was made in load control by running down the 12-deg slope. As the weld carriage traversed along the X-axis, the weld spindle continually fed out to maintain the programmed vertical axis load on the weld tool.

To ensure complete weld penetration to the top of the ramp, it was necessary to begin the weld above the top of the ramp and weld through it. This was accomplished by plunging the welding tool on a small wedge that was GMA tack welded to the transition piece - Fig. 11. This wedge was manually removed after FSW was completed.

After completion, the floor test weldment was sent to General Dynamics Land Systems for subsequent attachment of floor stiffeners and preparation for mine blast testing, then it went to the U.S. Army Aberdeen Test Center for a simulated mine blast test. Both the FS butt-joint and corner welds passed the test. The successful welding and testing of the mine blast floor test assembly demonstrated that relatively large structures made from thick-section aluminum could be constructed using friction stir welding.

Fig. 11 - Starting tab for welding lower glacis to sidewall.
Over the duration of this project, new FSW tool materials and process parameters were developed to increase the travel speed during FSW of 1-in. 2519 aluminum armor from 1.2 to 4 in./min with good weld soundness and minimal distortion. Along with the greater than threefold increase in travel speed, the transverse weldment tensile strength was increased to a reproducible 56 ksi while maintaining high weld ductility of 14%. Furthermore, the new tool designs reduced the axial welding tool loads required for welding from approximately 16,000 to 9000 lb. Flat and 90-deg corner ballistic shock test panels were fabricated by FSW and were successfully shock tested. Flat butt-joint welds made by GMAW in 2519 have never been able to pass the ballistic shock test. The technology was then used to fabricate and successfully test a floor mine blast test assembly, representing the forward one-third of the AAAV tub section, in 2519-T87 aluminum. Thus, the viability of this new FSW technology has been demonstrated for fabricating thick-section aluminum armor components for combat vehicles such as the AAAV.

This work was conducted by the National Center for Excellence in Metalworking Technology, operated by Concurrent Technologies Corp. under Contract No. N00014-00-C-0544, to the U.S. Navy as part of the U.S. Navy Manufacturing Technology Program. The authors appreciate the suggestions of Dr. Bill Herman, Dave Gerty, and Glenn Campbell of General Dynamics and the support of Scott Story and Robert Cross of the AAAV Program Office.

The authors acknowledge Doug Waldron of The Boeing Co. for the use of the friction stir welding equipment and the invaluable assistance of Boeing employees Keith McTernan, Scott Forrest, and Herman Villoria. Finally, the authors appreciate the contributions of Bill Majer, John Gover, and Bruce Williams of CTC.

References 1. Pickens, J. R., Colligan, K., and Fisher, J. J. 2000. Aluminum 2519 material evaluation - Interim report - Friction stir welding of aluminum 2519 plate for the advanced amphibious assault vehicle. NCEMT TR No. 00-54. Concurrent Technologies Corp.
2. Campbell, G., and Stotler, T. 1999. Friction stir welding of armor grade aluminum plate. Welding Journal 78(12): 4547.
3. Dawes, C. J., and Thomas, W. M. 1999. Development of improved tool designs for friction stir welding of aluminium. Proceedings of the 1st International Symposium on Friction Stir Welding. Thousand Oaks, Calif. Cambridge, U.K.: The Welding Institute.
3. MIL-STD-1946A (MR). 1989. Welding of Aluminum Alloy Armor.
4. Konkol, P. J., Colligan, K., and Nickodemus, G. H. 2001. Ballistic shock performance of friction stir weldments in aluminum alloys 5083-H131 and 2195-T8P4. NCEMT TR No. 01-149, Concurrent Technologies Corp.
5. Colligan, K., Nickodemus, G. H., Konkol, P. J., and Solich, J. 2001. Mechanical and ballistic shock performance of 90-deg corner friction stir welded aluminum alloy 5083 and 2195. NCEMT TR No. 01-059, Concurrent Technologies Corp.

KEVIN J. COLLIGAN, PAUL J. KONKOL, JAMES J. FISHER, and JOSEPH R. PICKENS are with Concurrent Technologies Corp. Colligan is Principal Engineer, Harvest, Ala.; Konkol ( is Principal Engineer, Pittsburgh, Pa.; Fisher is Mechanical Engineer, Johnstown, Pa.; and Pickens is Chief Scientist, Gleneig, Md.