Developing and implementing new processes to enhance the performance, reliability, and safety of aerospace hardware is an ongoing objective for both government and industry programs. TWI in Cambridge, U.K., invented friction stir welding (FSW) (Ref. 1) in the early 1990s and Lockheed Martin began its development activities in 1995. FSW development continued at Marshall Space Flight Center (MSFC) through 2001 for various NASA applications, including man-rated flight hardware.
Friction Stir Welding
Fig.1 - The friction stir welding process.
Friction stir welding is accomplished with both monolithic and multiple-piece pin tools rotating at several hundred RPM and traversing a square butt joint of the same design configuration used for fusion welding. A plunge load is imparted through a spindle, driven by a FSW machine, and reacted against a backside anvil. Frictional heating under the pin tool and around the pin tip generate sufficient heat to locally plasticize the aluminum alloys to be welded. Tool rotation during the FSW process imparts a material flow in three dimensions to the plasticized weldment, causing complete mixing of the alloys. Consolidation of the weldment occurs via an extruding/forging action under the pin tool shoulder as the pin tool is traversed down the length of the weld. A schematic representation of the FSW process is shown in Fig. 1.
Friction stir welding enjoys a number of advantages over fusion welding processes, including the elimination of welding consumables such as gas, filler metal, and electrodes. As a joining process based on frictional heating due to mechanical work, FSW has only three primary weld variables to control. These are plunge force, rotation speed, and weld travel speed.
The 2XXX series aluminum alloys have long been the workhorse of aerospace programs for high-strength, lightweight applications. New materials, such as Al2195 aluminum-lithium alloy, provided significant base material improvements over the predecessor, Al2219. Improved strength at both room and cryogenic temperatures were significant benefits of the new alloy; however, weldability was sometimes a challenge, which prompted efforts to improve the process and ultimately led to the development and implementation of friction stir welding. Al2195 alloy has proven to be highly receptive to the FSW process, overcoming some of the production difficulties experienced in early development and implementation of Al2195 with conventional fusion weld processes (Ref. 2).
Inspection of Friction Stir Welds
Attendant with the new friction stir weld process are new inspection requirements for both visual and NDE techniques. FSW enjoys freedom from most fusion weld process defects; however, the demands of flight safety require proof testing as well as full NDE of man-rated hardware.
Fig.2 - FSW microstructure.
Existing processes such as radiographic and penetrant inspection will remain for FSW inspection; however, those methods will be supplemented by new, automated NDE. Long term, automated NDE will replace part of conventional NDE and ultimately achieve a productivity enhancement for inspection.
Understanding the potential discontinuities for the FSW process requires an understanding of the metallurgy. Figure 2 provides a cross-sectional view of a completed FSW that allows observation of the metallurgical structure associated with a FSW of AL2195.
The FSW nugget is formed as frictionally heated metal flows around the pin tool and is consolidated under the shoulder. Discontinuities observed during FSW development present a challenge requiring a blend of several complementary NDE methods to provide adequate inspection. The discontinuities observed during FSW development range from surface defects, such as excess flash, to lack of fill under the FSW tool shoulder to internal porosity and incomplete joint penetration (IJP).
In every case, the FSW discontinuity was linked to one or more FSW process conditions or parameters directly related as causative factors of the defect.
To assess and select appropriate NDE techniques, a logic diagram was generated to integrate candidate NDE techniques, testing, and development for NDE, procedures and documentation, process validation, and the requirements of fracture control. Factors assessed in evaluating NDE techniques included critical initial discontinuity size, potential discontinuities detected by a given method, the capability of candidate NDE techniques, and their maturity for production use. This assessment has explored a wide variety of NDE methods encompassing visual, several liquid penetrant techniques, ultrasonic inspections of differing types, radiography, and eddy current. One of the newest NDE technologies assessed was MWM® conductivity, a technique that maps surface conductivity in the area of the weldment.
Visual Inspection
Perhaps the most straightforward and simplest inspection technique - visual inspection - is an excellent means of inspecting for surface features including excess flash, galling, shoulder voids, and even weld misalignment. Figure 3 shows an example of a shoulder void.
Workmanship standards were constructed to illustrate acceptable and unacceptable weld face and root surface conditions of this kind. These defects are visible to the naked eye and are attributed to out-of-family welding parameters such as excessive travel speed (in./min), excessive rotational speed (RPM), inadequate plunge force loads, and improper joint tracking.
Fig.3 - Shoulder void in FSW.
The principal unacceptable root side condition is IJP. Early on in the friction stir welding program, IJP was considered to be the most critical type of defect. As a result, most NDE testing was conducted with this discontinuity type.
Visual examination of the root side of the weld demonstrated IJP discontinuities were detectable when inspected in the postetched condition. Etching is a postweld chemical treatment performed most often to prepare mechanically worked surfaces prior to penetrant inspection. In this case, the etching process clearly delineates the weld nugget dynamically recrystallized zone (DXZ) and its surrounding heat-affected zone (HAZ), making the lack of FSW nugget a distinct feature visible to the trained eye. The reason for the successful detection rate is that it is easy to discern the DXZ from the surrounding base metal and HAZ in the postetched condition. Therefore, visual inspection is a reliable technique to confirm suspected IJP conditions. Figure 4 is a 3X magnification view of an IJP defect on the root side of a FSW panel after etching.
Fig.4 - IJP in FSW after postweld etch.
The metallurgical characteristics of an IJP discontinuity are the determining aspects of the discontinuity and relate directly to the ability of ultrasonics and penetrant inspection techniques to detect IJP. These characteristics are, likewise, directly linked to the weld process itself. Primary factors affecting IJP during welding include heat input or material flow and, most importantly, the depth of the FSW pin tool.
Figure 5 illustrates the metallurgical features, which include the total depth of IJP, the depth of plastically deformed material, and the tight bond at the IJP interface. The most significant of these with regard to NDE is the degree of "tightness" of the "kissing bond" created at the IJP interface. Conventional NDE techniques rely heavily on a physical separation - void or air gap - as the means to provide a response from such a defect. The less significant this separation, the more problematic its detection.
Penetrant Inspection
Penetrant inspection via P135E and P6F4 was performed on FSW test panels in the as-welded, single-etched, and double-etched condition. In addition, penetrant inspections were performed with and without developer and with varying penetrant dwell times. Penetrant inspection of the FSW test panels in the as-welded condition was determined to be an unacceptable method due to poor detection and the excessive background noise produced by the surface, which interfered with the inspection.
Fig.5 - Metallurgical cross section of IJP discontinuity.
Inspection of FSW in the etched condition via P135E and P6F4 consistently and successfully detected root-side IJP discontinuities. However, because the sensitivity level of detection for each penetrant solution is different, the results were dissimilar. P135E successfully detected IJP discontinuities that were greater than or equal to 0.064 in. deep, and P6F4 successfully detected IJP discontinuities that were greater than or equal to 0.050 in. deep. Double etching, via caustic etchant solution prior to the application of penetrant, enhanced the detection of IJP in comparison to single etching.
The difference between single etching and double etching is single etching removed 0.0002 to 0.0004 in. of metal and double etching removed 0.0004 to 0.0006 in. Test results demonstrated etching that removed a minimum of 0.0004 in. of metal prior to the application of penetrant improved the detectability of IJP.
Due to the outcome of test results, it was decided penetrant inspection should include the removal of 0.0004 to 0.0006 in. of metal via caustic etch solution prior to the application of penetrant solution. In addition, extended penetrant dwell times and the use of developer were evaluated; the results yielded no improvement in the detection of IJP discontinuities.
Ultrasonic Inspection
Automated Inspection Systems, RD/Tech, Lockheed Martin, and Marshall Space Flight Center (MSFC) NDE engineers and technicians performed ultrasonic inspection (UT) on FSW test panels. Conventional UT and multi-element probes were evaluated, as were L-wave and shear-wave techniques and multiple-angle transducers. The results initially demonstrated the technique(s) could detect IJP discontinuities at 15 to 20% of the material thickness or greater.
However, changes in FSW tooling directly affected the IJP discontinuity metallurgical characteristics, making the discontinuity more tightly closed and thus more difficult to detect. This affect of improving the weld process without sufficient regard for its effects on other parts of the manufacturing process, including inspection, became a recurring theme in pursuing automated NDE. Ultimately, improvements to RD/Tech's phased-array UT inspection technique resulted in detection at 25 to 30% of thickness and greater.
Fig.6 - Phased-array scan of IJP discontinuity.
The response for phased array provides multiple views of the friction stir weld at one time, allowing position location information as well as through-thickness data to be portrayed for detected discontinuities. This is accomplished through the use of a 32-element probe electronically rastering the UT beam across the weld as the probe is automatically scanned down the length of the weld. The result for an IJP discontinuity is depicted in Fig. 6.
The top portion of Fig. 6 provides a C-scan image of the weld with the weld and the discontinuity running from left to right. The lower portion of the figure is a longitudinal side view showing the material thickness and location of the discontinuity at the bottom of the image, which is the root side of the weld. Note detection is discontinuous at some points, which again relates to the metallurgical nature of the IJP discontinuity.
Radiographic Inspection
Radiographic inspection was performed via film and digital methods on FSW test panels. Test results demonstrated 90% probability/95% confidence in this method's ability to detect IJP discontinuities greater than or equal to 30% of the material thickness. However, dissimilar alloy welds posed a challenge to film radiography; results showed it was difficult to discern an IJP discontinuity.
Fig.7 - Al2219 to Al2195 dissimilar alloy FSW.
The reason for this is twofold. First, the joining of dissimilar alloys of aluminum yields a weldment that is a commingling of the two alloys, which vary in chemical composition by several percentage points of copper and lithium. The difference in copper greatly affects transmission of the X-ray, requiring an interpreter to "train" his eyes to accurately interpret the film radiograph. Figure 7 provides a view of the metallurgical difference evident in a dissimilar-alloy weld of Al2219 to Al2195. The lighter-etched portion is Al2219 and the wavy boundary where the two alloys intermix is reflected in radiographs of these welds.
The second reason for harder detectability in dissimilar-alloys FSW is the tendency for the IJP discontinuity to be more tightly bonded in this alloy combination (Al 2219 to Al2195). Several in-depth studies of the metallurgy of the FSW has proven the relationship, mentioned earlier, with the characteristics of the IJP and its NDE detectability.
Eddy Current and Conductivity Inspection
Conventional eddy current inspection (EC) was performed on FSW test panels by means of a 1-MHz pencil probe and a 300-kHz differential rotating probe. Initial eddy current results demonstrated reliable detection by both MSFC and Lockheed Martin techniques for Al2195/Al2195 friction stir welds containing at least 0.065 in., or deeper, IJP. The extreme difference in EC across dissimilar-alloy welds yielded an EC response from virtually all panels, making discrimination of IJP versus no IJP panels unreliable. These promising results changed as alterations were made to improve the FSW process by changing the FSW tooling.
Fig.8 - Jentek MWM multi-element sensor.
Reliable detection during automated NDE is critical to the integrity of aerospace applications. To assess the latest technology other than conventional EC, Lockheed Martin first contracted with Jentek Sensors, Inc., of Watertown, Mass., to develop its technology for FSW inspection.
This new approach to EC-type inspection is based on conductivity, first explored under LMCO IRAD activity (Ref. 3). Jentek Sensors, Inc., was asked to perform various tasks from 1998 through 2001 relative to process monitoring and postweld inspection with its inspection systems.
The promising results of the company's MWM® conductivity methods resulted in a contract to complete technique development and a custom sensor design specific for FSW applications. This work has been completed and provides a risk mitigation complementing current plans for radiographic, penetrant, and ultrasonic inspection techniques for production NDE of friction stir welds.
The multi-element MWM sensor (Fig. 8) has demonstrated detection of 0.050 in. and deeper IJP in Al2195-to-Al2195, as well as in dissimilar alloy Al2219-to-Al2195 FSWs (Ref. 4).
MWM Sensor System
The MWM system consists of a PC or laptop computer, Gridstation software, instrumentation module, and MWM probe and sensors.
Fig.9 - Full penetration FSW conductivity map.
The conductivity probe provides automated scanning; however, it can also be used in manual mode. Like the multi-element UT probes, the sensor is comprised of some 37 elements. The MWM array employs approximately 30 elements in the primary area of the weldment, with the remaining elements spaced approximately 3 in. apart to track the edges of the weld land. Individual element spacing and arrangement were customized to achieve optimum sensitivity for discontinuity detection.
Absolute electrical conductivity is a physical property of these aluminum alloys measured by the MWM array. Conductivity has long been used to inspect for heat-treat condition in aluminum alloy knowing its relationship to changes in alloy composition and metallurgy. Its application for FSW inspection actually maps conductivity on the root side of the weld with a precision more than an order of magnitude better than other conductivity applications. Data is then processed and displayed as a conductivity map at the weld root surface. A C-scan image and profile image for a good weld are shown in Fig. 9.
The C-scan view presents the inspection data as a top-down view of the friction stir weld. The weld in Fig. 9 extends from left to right. The circular region on the right edge of the image is the terminus of the weld and the yellow region indicates the FSW weld nugget (DXZ) exhibiting full weld penetration through the joint thickness.
Fig.10 - FSW with IJP conductivity map.
The lower portion of the image in Fig. 9 is a cross-sectional view of the inspection data. FSW DXZ is indicated in the middle of this profile view, while higher conductivity values, on either side of the DXZ, indicate changing conductivity in the heat-affected zone. Blue- to aqua-colored zones map the HAZ on either side of the DXZ.
Incomplete joint penetration (the failure of the FSW to fully penetrate the joint thickness) has significantly different conductivity patterns, as illustrated in Fig. 10. This FSW specimen contained 0.045-in.-deep IJP and exhibits minimal DXZ as well as several planar discontinuity indications.
Comparison of the profile in Fig. 9 with that of Fig. 10 reveals differences in conductivity values and their position are observed as changes to the shape of the profile. The presence of planar discontinuities is also noted as severe reductions (drop out) in the conductivity profile noted in Fig. 10.
Dissimilar-alloy friction stir welds yield quite different patterns of conductivity via the MWM array technique due to the large differences in base metal conductivity. Al2219-T8 exhibits a typical conductivity of 34% IACS, while Al2195-T8 is 20. The profile in Fig. 11 shows the high conductivity Al2219 to the left of the profile decreasing rapidly as the conductivity drops into the DXZ area. The DXZ is bounded on either side by slight peaks in conductivity indicating the HAZ.
The specimen for this example contained IJP that was 0.057 in. deep. The key to developing criteria for detection of IJP via this technique lies in differences affecting the shape of the conductivity map, including a sharp slope change (rate of decrease) in conductivity from the Al2219 side of the FSW and a reduction to the extent of the weld DXZ.
Summary
NASA and Lockheed Martin are pursuing implementation of friction stir welding and automated NDE as part of a larger program to improve performance, safety, and producibility for welded aerospace hardware. Friction stir welding is being implemented to take advantage of its high strengths and toughness and its near defect-free welds in the 2XXX aluminum and aluminum-lithium alloys used for numerous aerospace applications.
Fig.11 - Friction stir welding IJP conductivity map for dissimilar alloy.
Significant productivity gains are anticipated due to transitioning from conventional manual NDE inspection techniques to automated production NDE. Existing NDE methods, including liquid penetrant and radiography, will continue as automated phased-array ultrasonics is implemented and will subsequently be used to replace manual NDE.
To assure risk mitigation for conventional NDE inspection techniques, a new technology utilizing MWM conductivity mapping technique with a custom 37-element array sensor specific has been accomplished.
Acknowledgments
Acknowledgment and thanks are given to both NASA and Lockheed Martin personnel at the NASA Marshall Space Flight Center (MSFC) and to the staff at Automated Inspection Systems (AIS), Matec Instruments, and Jentek Sensors, Inc. for their efforts in this endeavor.
References
1. Thomas, W. M. et al. Friction Stir Butt Welding, International Patent Appl. No. PCT/GB92/02203 and GB Patent Appl. No. 9125978.8, Dec. 1991, U.S. Patent No. 5,460,317.
2. Arbegast, W., and Hartley, P. 1998. Friction stir weld technology development at Lockheed Martin Michoud Space Systems - an overview. June, AEROMAT.
3. Goldfine, N., Arbegast, W., et al. 2001. Friction stir weld LOP defect detection using new high-resolution MWM array and eddy current sensors. June, AEROMAT.
4. Kinchen, D.G., Goldfine, N., et al. 2002. Friction stir weld inspection through conductivity imaging using shaped field MWM® arrays. April, Trends In Welding Research.
DAVID G. KINCHEN (david.kinchen@maf.nasa.gov) is with Department 4630 M/S 4310 and ESMA ALDAHIR with Department 3700 P/A 3741 of Lockheed Martin Michoud Space Systems, New Orleans, La.
Inspection Trends
Front Page
About AWS | Contact AWS | Advertise with AWS
