RESISTANCE WELDING OF THERMOPLASTIC COMPOSITE SKIN/STRINGER SPECIMENS

Martine Dubé1, Pascal Hubert1, Ali Yousefpour2, 1, Johanne Denault3, Matthew Wadham-Gagnon1

1McGill University, CREPEC, Department of Mechanical Engineering

817 Sherbrooke Street West, Montréal, Québec, H3A 2K6, Canada

2Aerospace Manufacturing Technology Centre, Institute for Aerospace Research National Research Council Canada

5145 Decelles Avenue, Montréal, Québec, H3T 2B2, Canada

3Industrial Materials Institute National Research Council Canada

75 de Mortagne, Boucherville, Québec, J4B 6Y4, Canada

ABSTRACT An investigation of resistance welding of thermoplastic composites skin/stringer specimens is presented. Skin/stringer configurations with square-ended and 20° taper-ended flanges were resistance-welded using a metal mesh heating element. The skin and stringer laminates were made of 16-ply APC-2/AS4 PEEK/carbon fiber composite. The objective of this work was to study the feasibility of resistance welding to assemble aerospace structures with non-uniform cross-sections such as a taper-ended skin/stringer configuration. Combinations of different clamping distances and power levels were used. The welded specimens were analyzed using short beam tests, ultrasonic inspections and optical and scanning electron microscopy. The mechanical performance of the skin/stringer configuration was evaluated by three-point bending tests. Comparison between resistance-welded tapered and square-ended specimens was performed and better performance was obtained with the taper-ended specimens. KEY WORDS: Welding/Thermal Joining, Polyetheretherketone (PEEK), Mechanical Properties

1 INTRODUCTION Joining plays an important role in manufacturing of composite structures in marine, automotive and aerospace industry (1). Mechanical fastening and adhesive bonding are widely being used to assemble metals or composite components (2). However, there are disadvantages associated with these methods such as stress concentration induced by drilling holes in mechanical fastening or extensive surface preparation during adhesive bonding. In the case of joining thermoplastic composite components, fusion bonding, or welding, is an alternative method that can eliminate the abovementioned disadvantages.

Fusion bonding occurs by heating the surfaces of the parts to be joined above the polymer glass transition temperature, or melting temperature, and then allowing the weld interface to cool down, under application of pressure. Various fusion bonding techniques are available for thermoplastic composites. They can be classified into three categories: friction welding (in which heat is generated by frictional work between the two surfaces to be welded), thermal welding (in which heat is directly applied to the surfaces to be joined) and electromagnetic welding (in which heat is induced by electrical current or electromagnetic field at the weld interface). As one type of the electromagnetic welding class, resistance welding has exhibited processing, performance and cost benefits. For the first time, this method has been used for welding thermoplastic composite components of J-nose leading edges for the Airbus 340-600 and Airbus 380 airplanes. Composite stiffened panels are now being used in aerospace industry and are replacing the metallic counterparts. These structures were investigated at NASA Langley Research Center for adhesively bonded thermoset composites (3-7). It was shown that the failure mode of a skin/stringer configuration (Figure 1 – a) is the same as the failure mode of a simplified configuration composed of a skin laminate reinforced by a flange (Figure 1 – b) (6). Benefiting from these results, the failure mode of a skin/stringer specimen can be studied using a simplified skin/stringer specimen that is easier and less expensive to manufacture. Henceforth, the simplified skin/stringer configuration is called skin/stringer specimen in the text.

Figure 1: Skin/stringer configuration (a) and simplified skin/stringer configuration (b)

In a previous study, Dubé et al. (8) investigated the processing and mechanical performance of resistance-welded thermoplastic composite skin/stringer specimens using square-ended flanges (i.e. θ = 90°). However, it is well known that using a taper-ended flange (θ ~ 20°) reduces the stress concentration at the flange tip (3-7). In this study, the effect of a taper-ended flange on processing and performance of resistance-welded skin/stringer specimens was investigated. The effects of input power level and clamping distance on the quality of the welds were explored and a comparison between the mechanical performance of the welded specimens using tapered and square-ended flanges was performed.

2 EXPERIMENTAL

2.1 Materials, Welding Set-Up and Specimen Geometry The material system used in this study was quasi-isotropic [±45/90/0]2S, 16-ply APC-2/AS4 composite. The APC-2/AS4 material was provided by CYTEC Engineered Materials Inc. and was compression-molded under standard PEEK molding condition, i.e., processing temperature of 390°C, residence time of 20 minutes, molding pressure of 0.7 MPa and cool down rate of approximately 7°C/minute. The heating element was a stainless steel mesh with wires diameter of 0.04 mm and open gap of 0.09 mm. The thickness of the heating element was 0.08 mm. Two neat PEEK polymer films (thickness of 0.127 mm) were placed on each side of the heating element to provide a resin rich region at the weld interface. Two flange geometries were considered: (a) square-ended flange (Figure 2 - a) and (b) taper-ended flange with 20° angle (Figure 2 - b).

Figure 2: Specimen geometries (a): square-ended flange and (b): taper-ended flange

The resistance welding set-up consisted of a DC power supply, a pressure system, a computer/data acquisition system and a resistance welding rig (8). The computer/data acquisition system was used to monitor the temperature, current and voltage in the welds. A control system was developed using Labview software to control the input power. The control system was designed such a way that when temperature at the weld interface reached a pre-determined temperature, the electrical current was turned off. Ceramic block insulators were used to apply pressure and prevent excessive heat losses in the environment. Electrical current was introduced to the heating element using copper electrical connectors. Welding was conducted under a

pressure of 1 MPa. Figure 3 presents the final weld stack used for the square-ended specimens. For the taper-ended specimens, the top ceramic block was modified such a way that pressure was applied on the oblique edges of the flange.

Figure 3: Weld stack for square-ended specimens For the square-ended skin/stringer specimens, two input power schemes were used. The first method consisted of applying a constant voltage and the second method consisted of applying a ramped voltage to the heating element. Two constant voltages (corresponding to power levels of 249 kW/m2 and 282kW/m2) and three ramped voltage rates (0.005 V/s, 0.01 V/s and 0.02 V/s) were used. In all cases, the input power was stopped and the joints were allowed to cool down when the weld interfaces reached a pre-determined temperature of 440°C. Talbot et al. (9) have shown that clamping distance, i.e. distance between the edge of the specimen and electrical connector (Figure 3), has an influence on the temperature distribution at the weld interface. When a large part of the heating element is exposed to air, the edges of the specimen tend to burn (local heating) before the temperature in the middle of the weld reaches the processing temperature. The problem of local overheating arises from sudden change of heat transfer mechanisms from convection and radiation to conduction at the edges of the weld (10). The areas of the heating element exposed to air have poor heat transfer properties due to free convection. As a result, the exposed areas reach a higher temperature faster than the areas of the heating element in contact with the weld interface. The sudden temperature rise of the exposed areas generates a sharp temperature gradient at the edges of the weld that leads to local overheating, possible polymer degradation and melt front propagation through the weld thickness. For this investigation, two clamping distances (1.5 mm and 2.0 mm) were used for the constant power levels. The effect of the clamping distance on the temperature uniformity in the weld was measured using two thermocouples positioned at the center and the edge of the weld

area, between the neat PEEK film and the flange laminate (Figure 4). For the ramped power level technique, clamping distances of 1.0 and 2.0 mm were used for the tapered and square-ended specimens, respectively. Table 1 presents the different welding conditions.

Table 1: Resistance welding conditions

Specimen Power level or ramp voltage rate

Clamping distance (mm)

Square-ended 249 kW/m2 1.5 Square-ended 249 kW/m2 2.0 Square-ended 282 kW/m2 1.5 Square-ended 282 kW/m2 2.0 Square-ended 0.005 V/s 2.0 Square-ended 0.01 V/s 2.0 Square-ended 0.02 V/s 2.0 Taper-ended 0.01 V/s 1.0

2.2 Mechanical Testing and Characterization Methods Short beam tests were performed to evaluate the Interlaminar Shear Strength (ILSS) of the welded samples according to ASTM D2344. The samples were 4.6 mm thick and 8.5 mm wide. The samples were cut from the welded area of the square-ended flanges. Five samples were tested for each welding condition. The fractured specimens were analyzed by optical microscopy. Ultrasonic inspection was also performed on one sample of each welding condition. The optimal welding condition was determined from the short beam tests, ultrasonic inspection and optical microscopy and was used to investigate the properties of the square-ended specimens under three-point bending tests. The same welding condition was also used to weld taper-ended specimens and investigate their properties under three-point bending tests. A support span of 80 mm was used with the central load applied on the backside of the skin laminate. Specimens were tested at 23°C and at relative humidity of 50% under a crosshead speed of 2 mm/min. Five replicated samples were welded and tested for each specimen geometry. Scanning electron microscopy was used to observe the fracture surfaces of the specimens.

Figure 4: Short beam specimens

3 RESULTS AND ANALYSIS

3.1 Weld Interface Temperature History The time-temperature curves obtained for the two clamping distances of the constant power level of 249 kW/m2 are shown in Figure 5. T1 and T2 are temperatures in the middle and at the edge of the weld, respectively. For a clamping distance of 1.5 mm, T1 was close to T2, showing a small temperature gradient between the center and the edge of the weld. For a clamping distance of 2.0 mm, a larger temperature gradient between the center and the edge of the weld was observed. This is due to local heating at the edges of the weld because of the larger part of heating element exposed to air (edge effect). A typical time-temperature curve for a ramped voltage of 0.01 V/s and clamping distance of 2.0 mm is shown in Figure 6. The welding time for those samples was about 210 seconds, which was 150 seconds longer than those welded under the constant power method. However, the time spent above the melting temperature (343°C) is decreased from 34 seconds to 20 seconds, which is likely to reduce the size of any heat-affected-zone or polymer degradation.

Figure 5: Time-temperature curves for square-ended samples welded under a constant input power level of 249 kW/m2 and clamping distances of (a): 1.5 mm and (b): 2.0 mm

Figure 6: Time-temperature curve for a square-ended sample welded under the ramped

voltage technique of 0.01 V/s (clamping distance = 2.0 mm)

3.2 Interlaminar Shear Strength Results The short beam results for all welding conditions of the square-ended specimens are presented in Figure 7. Three welding conditions showed similar interlaminar shear strengths and standard deviations: the ramped voltage at a rate of 0.01 V/s and the constant power levels of 249 kW/m2 and 282 kW/m2 with a clamping distance of 1.5 mm. The small standard deviations obtained indicated a uniform weld quality over the weld area. Since the samples for each welding condition were cut at different places over the weld area (Figure 4), it was possible to relate the short beam strength of each sample to the actual welding temperature that was measured at those positions. The resulting graph is shown in Figure 8. A significant increase in ILSS was observed as the temperature increased from 385°C to 440°C. Below 415°C, optical microscopy showed that the samples failed at the weld interface, between the polymer film and the skin or flange’s first ply (Figure 9 – a). This failure mode led to lower ILSS and was observed for some samples welded at 0.005 V/s and 282 kW/m2 with a clamping distance of 2.0 mm. Between 440°C and 450°C, some scatter was observed in the data and the samples failed in the skin or flange laminates, at the interface between the 90° plies and the 0° or 45° plies (Figure 9 – b). At temperatures higher than 450°C, a decrease in ILSS was observed, probably due to polymer degradation.

Figure 7: Short beam results

Figure 8: ILSS as a function of temperature

Figure 9: Short beam failure mode (middle of short beam samples) (a): crack at the weld interface and (b): cracks at the 90° ply interfaces

3.3 Ultrasonic Inspection and Microscopy The effect of temperature on the quality of the welds was investigated using ultrasonic inspection. Figure 10 shows the C-Scan of the square-ended specimens welded under the three ramped power levels and two constant power levels. No significant differences were observed between the specimens but it was evident that all the defects or porosities were located near the edges of the flanges. Figure 11 presents the B-Scan of the specimens. It was seen that the porosities at the edges of the flanges were located in the skin laminate. Approximately the same amount of porosities was detected for the different power levels except for the constant power level of 249 kW/m2 where less porosities were observed. Away from the edges the flanges, weld interfaces seemed free from any defects. These observations were confirmed by optical microscopy. Figure 12 shows the weld interfaces at the edges of the flanges for power levels of 249 kW/m2 and 282 kW/m2. No significant defects were observed at the edge of the samples welded under 249 kW/m2 (Figure 12 – a). For a power level of 282 kW/m2 voids and defects were observed at the weld interface and in the skin and flange laminates (Figure 12 – b). These observations are in good agreement with C-Scan and B-Scan images.

Figure 10: C-Scan of square-ended specimens for different power levels

Figure 11: B-Scan of square-ended specimens for different power levels

Figure 12: Weld interfaces at the edges of the flanges for constant power levels of (a): 249 kW/m2 and (b): 282 kW/m2

3.4 Three-Point Bending Results A ramped power level of 0.01 V/s was used to weld tapered and square-ended specimens. Unintentionally, two processing temperatures were used for the square-ended specimens. For the higher temperature (460°C), distortion of the laminates was observed with a lot of polymer squeeze out. For the lower temperature (440°C), laminates remained intact. Three-point bending tests were performed on specimens welded under both processing temperatures and on taper-ended specimens (for which the processing temperature was kept at 440°C). Surprisingly, the square-ended specimens that experienced distortion showed a better performance than the specimens that remained intact (Figure 13). Cracks in the three-point bending tests always started at the flange tip and then propagated in the weld and skin laminate. The temperature reached at the flange tip was therefore a crucial parameter in this configuration. For the over-processed specimens, the temperature at the flange tip was around the optimal temperature of 440°C and the temperature in the center was 460°C. Therefore, cracks at the flange tip formed at a higher load. For the lower temperature, one can see in Figure 6 that the temperature at the flange tip was around 405°C when the temperature in the middle of the specimen was 440°C. This low temperature at the flange tip could explain why cracks formed at a lower load in the test. These results are in good agreement with the temperature – ILSS relationship that was presented (Figure 8).

Figure 13: Three-point bending results for (a): square-ended specimens (460°C), (b):

square-ended specimens (440°C) and (c): taper-ended specimens (440°C) Even though the mechanical performance of the over-processed specimens was better, distortion of the specimens was unacceptable and the high temperature involved in the center of the weld would probably cause polymer degradation and reduce the toughness of the welds. Scanning electron microscopy was used to observe the fracture surfaces of both specimens, at the center of the flanges (Figure 14). As expected, good fiber/matrix adhesion was observed at the optimum processing temperature (440°C), while a weaker adhesion was seen at the higher processing temperature of 460°C. The lower matrix ductility observed at the higher processing temperature was also an indication of polymer degradation. It was shown that PEEK undergoes chemical bonding during degradation, which leads to brittle fractures (11, 12).

Figure 14: Fracture surfaces for (a): processing temperature of 440°C and (b): processing

temperature of 460°C

Comparison between tapered and square-ended specimens was performed. A maximum load increase of 13% was reported for the taper-ended specimens (Figure 13). This better performance was attributed to the reduction of the stress concentration factor at the flange tip. However, it is believed that this performance could be further improved by enlarging the clamping distance. The temperature at the edges of the flange was only 400°C when the temperature in the middle was 440°C. Enlarging the clamping distance would allow the temperature at the edges to increase and approach the optimal temperature of 440°C. Observation of the fracture surfaces of the tapered and square-ended flanges confirmed the lower temperature at the edges. In both cases, cracks initiated at the flange tip, between the weld interface and the 45° layer of the skin and propagated towards the center of the flange, in the skin laminate (Figure 15 – a). In the case of the over-processed square-ended specimens, cracks initiated at the flange tip, between the 45° and -45° layers of the skin laminate and propagated in the skin (Figure 15 – b).

Figure 15: Cracks initiation and propagation in three-point bending test for (a): taper-ended specimens (440°C) and (b): over-processed square-ended specimens (460°C)

4 SUMMARY AND CONCLUSIONS In this study, APC-2/AS4 composite skin/stringer specimens were resistance-welded using a stainless steel mesh heating element. Different constant and ramped power levels were used to introduce power to the heating element and the quality of the welds was investigated using short beam tests, optical and scanning electron microscopy and ultrasonic inspection. A comparative study was performed between tapered and square-ended specimens. It was shown that better mechanical performance is obtained with taper-ended specimens. However, the same failure mode was observed in both cases. Cracks initiated at the flange tip between the weld interface and the first ply of the skin laminate and then propagated in the skin laminate towards the center of the flange.

5 ACKNOWLEDGEMENTS This work was supported by funding from the Natural Sciences and Engineering Research Council of Canada. The authors also greatly acknowledge Mr. Christian Néron from Industrial Materials Institute for his help concerning the ultrasonic inspection and Mr. David Leach from Cytec Engineered Materials Inc. for providing the materials used in this study.

6 REFERENCES 1. A. Yousefpour, M. Hojjati and J.-P. Immarigeon, Journal of Thermoplastic Composite Materials, 17, (4), 303 (2004). 2. M. J. van Wijngaarden, SAMPE Fall Technical Conference, (2005). 3. R. Krueger, et al., Journal of Composite Materials, 34, (15), 1263 (2000). 4. M. K. Cvitkovich, et al. Proceedings of the 13th Annual Technical Conference on Composite Materials, 1998. 5. M. K. Cvitkovich, T. K. O'Brien and P. J. Minguet, Proceedings of the 7th Symposium on Composites: Fatigue and Fracture, 97 (1998). 6. P. J. Minguet and T. K. O'Brien, Composite Materials: Testing and Design, Proceedings of the 12th Symposium, 12, 105 (1996). 7. P. J. Minguet and T. K. O'Brien, Tenth International Conference on Composite Materials. I. Fatigue and Fracture, 245 (1995). 8. M. Dubé, P. Hubert and A. Yousefpour, SAMPE Fall Technical Conference, (2005). 9. E. Talbot, et al., Annual Technical Conference (ANTEC) of the Society of Plastics Engineers, (2005). 10. E. Eveno, and J. W. J. Gillespie, Journal of Thermoplastic Composite Materials, 1, 322 (1988). 11. T. Vu-Khanh and J. Denault, Journal of Thermoplastic Composite Materials, 4, (4), 363 (1991). 12. J. Denault and T. Vu-Khanh, Polymer Composites, 13, (5), 361 (1992).