Smart Glass Epoxy Laminates with Embedded Ti-based Shape Memory Alloy

Transcription

1 Materials Transactions, Vol. 45, No. 7 (2004) pp to 2421 #2004 The Japan Institute of Metals Smart Glass Epoxy Laminates with Embedded Ti-based Shape Memory Alloy Yung K. Choi* 1 and Michelle Salvia* 2 LTDS UMR CNRS 5513, Ecole Centrale de Lyon, ECULLY Cedex, France Growing attention has been given in the last years to the development of smart materials and structures. Among them, composite materials play evidently a leading role. In fact, it is relatively easy, taking in account their processing techniques, to embed in the structural material itself some sensors, processors and actuators at a mesoscopic scale. Shape memory alloys (SMA) are particularly fitted as actuators because they can be easily drawn into thin wires and incorporated in polymeric and metallic matrices or in classical fibre epoxy laminate structure. To test the influence of the materials and processing conditions on the actuation properties of adaptive hybrid composites four sets of asymmetric composite systems based on a glass epoxy laminate with embedded wires of a shape memory Ti-Ni-Cu alloy were processed. The SMA wires in One Way Shape Memory Effect (OWSME) or (Two Way Shape Memory Effect TWSME) conditions were incorporated as far as possible away from the neutral plan. These asymmetric hybrid laminate beams were tested in clamp-free conditions. With the actuators heated by Joule effect undergoing a reversible martensite to austenite transformation, the reversible bending was induced due to the recovery strain in relation with the shape memory effect. The most important deflection of the composite was obtained for the material, processed with embedded wires in TWSME conditions. Nevertheless, for samples just prestrained for the OWSME, a self-training effect occurred in relation to the reverse polarised austenite to martensite transformation, during cooling after actuation. In order to follow the phase transitions in the embedded SMA wires, resistance measurements have been performed during actuation. Describing the macroscopic behaviour in the frame of the unidirectional approach de Liang and Rogers and using metallurgical parameters defined from a Clausius-Clapeyron diagram, a description of the temperaturedeflection curves can be obtained. Nevertheless some parameters have to be mastered in order to process real structural parts. (Received November 6, 2003; Accepted April 30, 2004) Keywords: shape memory alloy, composite, epoxy, electrical resistance, modelling 1. Introduction In Titanium based metallic alloys the shape memory effect is related to the displacive, diffusionless and reversible transformation between a low temperature martensitic phase and the high temperature austenitic phase. When such an alloy undergoes a permanent deformation in the martensitic state this deformation can be wholly recovered by heating above the martensite to austenite transformation temperature. That is the so-called one way shape memory effect (OWSME). If the shape memory was prevented to occur by means of clamping devices or by embedding into polymer or composite material, an important recovery stress appears. Recently, much progress and many potential applications appeared in the field of composite materials with embedded shape memory alloys: improvement in the level of the stressstrain curve, 1) stress concentration reduction around a crack area, 2,3) stability of composite shells and plates, 4,5) increased bandwidth for structural control, 6) tracking tabs for advanced rotor systems, 7) vortex leveraging, 8) adaptive wing, 9,10) biomimetic active hydrofoil, 11) damage suppression. 12,13) Actuation applications require, in principle, the use of SMA exhibiting the two way shape memory effect (TWSME). Such effect or training was achieved by suitable temperature cycling between the martensite and austenite phases under constant stress or strain. In these conditions, a reversible phase transformation with reversible shape memory or recovery stress can be initiated only by heating and cooling. The main objective of this work is to investigate the * 1 Graduate Student, Ecole Centrale de LYON. Present address: Shinshu University, Faculty of Engineering Department of Mechanical Systems Engineering, Nagano , Japan * 2 Contact author, Corresponding michelle.salvia@ec-lyon.fr influence of the processing conditions on hybrid composite actuation ability. So, glass-epoxy composites with embedded Ti-Ni wires have been processed. Either the structural state, or the mechanical conditions of the SMA are varied during the cure of the surrounding composite, in order to relate the actuation properties to these different features. Moreover, in order to increase the efficiency of these hybrid composites as actuators, asymmetric laminates have been processed. Describing the macroscopic behaviour in the frame of the unidirectional approach de Liang and Rogers and using metallurgical parameters defined from a Clausius-Clapeyron diagram, a description of the temperature-deflection curves can be obtained. 2. Materials The SMA wires produced by WEG Industries (Belgium) ( ¼ 120 mm) were Ti-Ni-Cu alloys with 5 at% of Copper in order to avoid the R phase. The wires were strained up to 8% in martensitic phase after annealing during one hour at 425 C. The four characteristic transformation temperatures (M f, M s, A s, A f ) were measured as a function of the applied tensile stress using hysteresis curves determined by electrical resistance measurements. The slope c A, of the martensite! austenite transformation domain boundaries in the (stress), T (Temperature) plane, is 6 MPa/ C. Four composite systems, which vary in structural state and embedding conditions of the SMA alloys, have been processed. These four systems (named A to D) were made from Vicotex XE12 prepreg tape (E-glass/Epoxy resin) from Hexel-Composites. The process was achieved by prepreg lay up and vacuum bag moulding. All systems were cured for 2 hours at 120 C. In order to process asymmetric laminates, the SMA wires were incorporated during lay-up, between the first and the second ply, as far as possible away from the

2 2418 Y. K. Choi and M. Salvia TiNiCu tension E-glass/Epoxy resin prepreg tape tension (a) vacuum (b) Siliconized paper Polyamide vacuum film mold vacuum sealant tape 2 hours at 120 C NiTiCu wire 215 µm (c) GFRP (d) Fig. 1 (a) Prepreg lay up; (b) Schematic drawing of the process; (c) Hybrid composite beam (part); (d) Cross section of the hybrid composite (part). neutral plan (Fig. 1). For the hybrid composites A and C only a small tensile stress (20 MPa), enough to secure the wires alignment, was applied during processing. In the cases of B and D, a stress of 500 MPa was maintained during the different stages of the processing in order to prevent premature actuation at the cure temperature. For composites C and D the wires were trained in TWSME conditions: after stretching up to 8%, the Ti-Ni-Cu wires were subjected to nine cooling and heating cycles within the temperature range: T 1 (below M f ) T 2 (above A f ). Table 1 gives the main characteristics of hybrid composites constituents and defines the four hybrid composites studied. 3. Experimental Results 3.1 Interface characterisation The actuator is obviously efficient only if the strain associated with the martensite-austenite transition is well transmitted to the structural material. The alloy-matrix interface shear strength was studied by means of the pullout test. This test consists in performing a tension test on a SMA wire fractionally embedded in the composite. 14) The wire is subjected to a tension F at a constant strain rate until the fibre is pulled out of the composite. These tests provide the force F d at which the debonding matrix SMA-wire occurs. In all the cases the failure takes place at the interface and the interface shear strength is estimated in the frame of the average stress modelling: Table 1 Definition and characteristics of the four systems. Hybrid composite System A B C D Training No No Yes Yes Stress during curing (MPa) Fibre E Glass volume content (%) 64 Ti-Ni volume content (%) 0.6 Host composite Longitudinal modulus, E L (GPa) 52 T at the beginning of the sharp decrease of the modulus ( C)* 125 T ( C) (defined as the max of the loss factor versus temperature variation)* 160 NiTiCu alloy M s ** M f ** A s ** A f ** 30 C 44 C 52 C 65 C *Torsion Dynamic Spectrometry: Frequency 1 Hz; Temperature rate equal to 10 /min -. ** DSC measurements after annealing at 425 C during 1 h. i ¼ F d =2rL where 2r is the wire diameter and L the embedded length. These tests were performed at 23 C for the processing conditions of the four types of hybrid composites (A, B, C ð1þ

3 Smart Glass Epoxy Laminates with Embedded Ti-based Shape Memory Alloy 2419 Table 2 Shear strength at the alloy-matrix interface. System A B C D E i (MPa) Current Aluminium Induction sensor Deflection SMA wire electrical resistance and D). Moreover, in the case D, the test was also performed at 120 C in order to estimate the shear strength near the glass transition of the host composite (Ref (E)). The shear strength i are shown in Table 2. These values of i are high enough to transmit the recovery stress to the composite structure (critical length l c ¼ 250 mm) and very close of that obtained by Jannalagadda et al. 15) in the case of a sandblasted wire. Therefore it should be thought these values could be associated to a superficial roughness due to the development of well oriented superficial martensitic variants. Liquid nitrogen for cooling TiNiCu Hybrid composite Clamping (a) Temperature Acquisition card PC 3.2 Actuators evaluation The hybrid laminate beams (5 1:5 105 mm 3 ) were tested in clamp-free conditions (Fig. 2). The SMA wires, were connected in series and electrically activated by means of resistive heating. Since the actuators were off the beam neutral plane, bending was induced during actuation, due mainly to the recovery strain, or shape memory effect. The tip deflections, f, were used to evaluate actuation effect. Measurements of the deflection were performed by means of an induction sensor, with an accuracy of 5 mm (Fig. 2b). The superficial temperature of the beam on the side close to SMA wires was measured by a thin thermocouple (Fig. 2). Moreover the electrical resistance was monitored during the test. The first bending deflection of different systems (A, B, C and D) according to the temperature is given in Fig. 3. In all the cases the NiTiCu after cutting off the heating current the sample returned by cooling to its original shape thanks to the properties of the host laminate composite These results show that:. the most largest deflection appears on the sample containing the SMA wires trained for the TWSME and before embedding and submitted to a tensile stress of about 500 MPa during process. a self-induced TWSME can be seen on samples, with SMA wires only prestrained up until 8% before embedding Moreover, the electrical resistance of the Ti-Ni wires during the actuation was recorded in order to estimate A s and A f. It is important to note that in this part of the work the Ti- Ni wires are not only used as actuators but also as sensor. Taking in account the fact that the resistivity of the Ti-Ni alloy varies with the temperature, the applied stress and the volume fraction of the different phases, the results could be complex. In fact, for all the processing conditions and using the classical technique for the determination of the characteristic temperatures, all the samples exhibit the same transition temperature values with A s 45 C and A f 70 C. 4. Modelling In order to model the macroscopic behaviour of the hybrid composite, the sample of thickness h and width b is replaced Fig. 2 Deflection, f/mm 0,7 0,6 0,5 0,4 0,3 0,2 0,1 Fig. 3 Thermocouple Clamping Induction sensor Hybrid composite beam SMA wire (b) DC power source Actuation test device: (a) principle; (b) photograph of the device Temperature, T/ C Bending deflection, f of the different systems versus temperature. by a bi-material with a glass/epoxy beam of thickness h 1 and a superficial layer of shape memory alloy of thickness h 2 with h 1 þ h 2 ¼ h (h 2 ¼ n 2 =4b), where n 4) is the number of SMA wires, and their diameter (120 mm)). The bending of this bimaterial plate occurs with a temperature change, due mainly to the shape memory effect of the SMA layer. As a D A B C

4 2420 Y. K. Choi and M. Salvia first approximation, and taking into account the relative thermal capacities, the temperature of the host composite beam, is supposed constant. A quantitative analysis can be performed 16) using the classical method of a bimaterial device 17) and assuming the continuity of the strain at the SMA/composite interface. Several one-dimensional constitutive models are able to simulate experimental results obtained on wires of shape memory alloys. The model considered here is one originally formulated by Tanaka, 18) then modified by Liang and Rogers 19) and Brinson: 20) 0 ¼ E W ð" " 0 ÞþðT T 0 Þþð 0 Þ ð2þ where is the thermoelastic tensor, the transformation tensor, classically given by ¼ " m E W where " m is the recovery strain limit, E W the modulus of the alloy, the martensite volume fraction and T 0, 0, " 0 and 0 are the initial values of temperature, stress, strain and martensite volume fraction, respectively. In bimaterial structures the mechanical requirements of strain compatibility and stress equilibrium between the two components produce coupling between the SMA constitutive variables. For such a spring-sma system, the loading is proportional to the recovery strain, and the stress-strain relation in the SMA layer is given by: 0 E Lh 1 ð" " 0 Þ ð2 bisþ 4h 2 where E L is the longitudinal modulus of the host composite. Moreover in order to describe experimental results, a kinetic law giving the evolution of the martensite volume fraction, as a function of the temperature and stress has to be chosen. A simple expression has been proposed by Armstrong: 21) ¼ 0 1 ðt A s =c A Þ ð3þ ða f A s Þ Then, with the initial conditions 0 ¼ " 0 ¼ 0, 0 ¼ 1, neglecting the thermal term in comparison with the shape memory effect and using the simple kinetic law proposed by Armstrong the relationship 2) can be written as follows: ðtþ ¼ " m E W ð ðt A s =c A Þ=ðA f A s ÞÞ ð4þ ð1=ð1þ4e W h 2 =E L h 1 ÞÞ where c A is the slope of the martensite-austenite transformation domain boundaries in the, T plane. The tip deflection f as a function of temperature is then given by: 5) f ðtþ 3L 2 =E L h 2 1 h 2 ðtþ ð5þ 5. Discussion Figure 4 gives the experimental and theoretical deflection values, versus temperature in the case of the sample D, for which the SMA wires were trained for TWSME before embedding (" m ¼ 8%; ða f A s Þ¼25 C, n ¼ 4). The modulus of the alloy E W varies with the volume fraction of martensite and the determination of ðtþ requires an iterative process. Figure 4 shows also the evolution of the electrical resistance of the embedded wires and its derivative. Fig. 4 Experimental and theoretical deflect ions and electrical resistance versus temperature for material D. Taking in account the relative variations of the electrical resistance the results can be divided in three temperature domains:. domain I: the alloy is in the martensitic phase and the deflection of the composite is negligible.. domain II: the shortening of the SMA.alloy results in a bending but two zones (a) and (b) can be observed: zone (a): the experimental deflection is small (with a slope 0.1 of the theoretical one), zone (b): near the maximum value of the phase transition rate it becomes possible to compare experimental and modelled deflection, but the experimental slope is always smaller than the theoretical one and decreases with the increasing temperatures. So, it seems possible to conclude that the phase transition rate is one essential parameter in the knowledge of the actuation phenomenon. This behaviour is evidently due to the viscoelastic properties of the host polymer based composite, probably at the local scale of the interface between the alloy and the organic matrix. It is clear now that the purely elastic model used in the previous approach has to be modified in this hypothesis.. domain III: it is not from now possible to compare experimental and theoretical curves. Nevertheless the experimental deflection of the hybrid composite increases slowly till 130 C although the classical interpretation of the resistance curves allows to estimate the A f temperature around 70 C for the embedded wires. In fact the rate of the transition martensite-austenite decreases with the increasing temperatures and the end of this phenomenon (real A f ) is probably at a markedly higher temperature. 6. Conclusion In conclusion, adaptive systems based on hybrid composite materials with embedded shape memory alloys wires are now

5 Smart Glass Epoxy Laminates with Embedded Ti-based Shape Memory Alloy 2421 emerging as a technical reality. In order to reach the best possible properties, the SMA wires have to be trained in TWSME conditions before embedding in the laminates. Maintaining the alloy at a level of stress avoiding the martensite-austenite transition during the cure seems also useful. Nevertheless it appears possible to obtain almost the same results with embedded wires just cold-worked of 6-8% in the martensitic phase before embedding. Thanks to this remarks the elaboration of these adaptive hybrids laminates will be markedly simple. Moreover a modelling developed in the frame of a purely elastic model is obviously not able to describe the global behaviour of these hybrid systems. So it seems necessary to include in this approach the viscoelastic properties of the host polymer based composite. 22) Finally the use of the embedded actuators wires as sensors by the interpretation of the variation of the electrical resistance versus the temperature is complex and a lot of experimental work has to be done also in this field. REFERENCES 1) H. Izui, K. Hamada, K. Ogawa, M. Taya and K. Inoue: Materials for smart systems II (Materials Research Society Symposium Proceedings 1997) pp ) Y. Furuya: Proceedings of the International Symposium on Microsystems, Intelligent materials and robots, ed. by J. Tani and M. Esashi, (Tohoku University, Sendai, 1995) pp ) W. Zheng, W. Jiansheng, X. Ke and T. Baoqui: Progress in Advanced Materials and Mechanics, ed. by W. Tzuchiang, T. W. Chou, (Peking University Press, Beijing, 1996) pp ) V. Birman: International Journal of Mechanical Sciences 39 (1997) ) M.-T. Ho, R. R. Chen, L.-C. Chu: Proceedings of Smart Structures and Materials, Vol 3041, ed. by M. E. Regelbrugge, (SPIE, San Diego, 1997) pp ) A. V. Srinivasan and D. M. McFarland, H. A. Canistraro and E. K. Begg: Journal of Intelligent Material Systems and Structures 8 (1997) ) V. Giurgiutiu, C. A. Rogers and J. Zuidervaart: Proceeding of Smart Structures and Materials, Vol. 3041, ed. by M. E. Regelbrugge, (SPIE, San Diego, 1997) pp ) T. R. Quackenbush, A. J. Bilanin, P. F. Batcho, R. M. McKillip Jr and B. F. Carpenter: Proceeding of Smart Structures and Materials, Vol 3044, ed. by J. M. Sater, (SPIE, San Diego, 1997) pp ) V. Weissberg, L. Shikhmanter and A. K. Green: Tenth International Conference on Adaptive Structures and Technologies, ed. by R. Ohayon & M. Bernadou (Technomic, Paris, 1999) pp ) D. C. Lagoudas, J. C. Strelec, J. Yen and A. Khan: Proceeding of Smart Structures and Materials, Vol 3984, ed. by V.V. Varadan (SPIE, Newport Beach, 2000) pp ) L. J. Garner, L. N. Wilson, D. C. Lagoudas and O. K. Rediniotis: Smart Materials and Structures 9 (2000) ) T. Ogisu, N. Andou, J. Takaki, T. Okabe, N. Takeda: Proceedings of the 11th ICAST, ed. by Y. Matsuzaki et al., (Technomic, Nagoya, 2000), pp ) Y. Xu, K. Otsuka, H. Yoshida, H. Nagai, R. Oishi, H. Orikawa, T. Kishi: Intermetallics 10 (2002) ) F. Mezzanotti and M. Salvia: Journal of Applied Mechanics and Engineering V (2000) ) K. Jonnalagadda, N. R. Sottos, M. A. Quidwai and D. C. Lagoudas: Smart Materials and Structure 9 (2000) ) Y. K. Choi, M. Salvia: Smart Structures and Materials: Active Materials, Behaviour and Mechanics, Vol 4333, ed. by C. S. Lynch, (SPIE, Newport, Beach, 2001) pp ) S. Timoshenko: Résistance des matériaux; Théorie élémentaire et problèmes, ed. by Dunod, (Dunod, Paris, 1968) pp ) K. Tanaka: Res. Mechanica 18 (1986) ) C. Liang and C. A. Rogers: Journal of Intelligent Materials Systems and Structures 8 (1997) ) L. C. Brinson: Journal of Intelligent Materials Systems and Structures 7 (1996) ) W. D. Armstrong: Journal of Intelligent Materials Systems and Structures 7 (1996) ) S. S. Sun, G. Sun and Q. H. Li: Computational Materials Science 21 (2001)