- Email: [email protected]
strain measurements in composite laminates with a fibre-optic Raman probe
Composites: Part A 30 (1999) 1187–1195
Surface and bulk stress/strain measurements in composite laminates with a fibre-optic Raman probe B.P. Arjyal a, P.A. Tarantili b, A.G. Andreopoulos b, C. Galiotis a,c,* a
Materials Department, Queen Mary & Westfield College, Mile End Road, London E1 4NS, UK Department of Chemical Engineering, National Technical University of Athens, 9 Iroon Polytechniou Str., 157 80 Zografou, Athens, Greece c Institute of Chemical Engineering and High Temperature Chemical Processes, Foundation of Research & Technology—Hellas, P.O. Box 1414, GR-265 00, Patras, Greece b
Received 4 October 1998; accepted 8 March 1999
Abstract Fibre stress/strain measurements in unidirectional, as well as, multidirectional aramid/epoxy composites have been conducted with the use of a laser Raman microprobe. The composite was incrementally loaded in tension while Raman measurements were taken. Fibre-optic probes sandwiched between adjacent laminae were employed for channelling the laser excitation light to a specified location within the bulk of the composite. The direction of the fibre-optic was either perpendicular or parallel to the reinforcing fibres. For comparison purposes, the same fibre-optic probe was used to scan the surface of the laminates. The perpendicular configuration was found to reduce the tensile strength of the as-received composite coupon by 10% whereas the parallel second configuration had no effect. In the unidirectional coupons the stress or strain in the principal fibre direction could be measured prior to loading and at every increment of applied tensile load up to fracture. The take-up of fibre strain for both bulk and surface set of measurements was identical with that obtained from the attached electrical resistance strain gauges. In the case of multidirectional coupons the stress or strain in the principal direction could be measured within successive plies situated at angles u to the loading direction. The results for the 08 plies were in good agreement with those obtained by conventional laminate analysis whereas small deviations from linearity were observed in the angle plies. The proposed methodology paves the way for simultaneous in-service stress/strain measurements on the reinforcing fibres situated on the surface or within the bulk of a composite laminate. q 1999 Elsevier Science Ltd. All rights reserved. Keywords: Stress; A. Laminates
Nomenclature
af am T E11, E22
e 1, e 2, e 3 s 11, s 22, s 33 n 12 G12, G23 Ef, E m Vf, V m Gm ASTM
longitudinal coefficient of thermal expansion for the fibre coefficient of thermal expansion of the matrix temperature elastic moduli in the principal material directions principal normal strains principal normal stresses Poisson’s ratio shear moduli in the 1–2 and 2–3 planes fibre and matrix Young’s moduli volume fractions of fibre and matrix shear modulus of matrix American society for testing and materials
* Corresponding author. E-mail address: [email protected] (C. Galiotis)
CCD LRS PDA ReRaM
charge coupled device laser Raman spectroscopy polydiacetylene remote Raman microprobe
1. Introduction Over the last two decades, lot of work has been put into developing structural materials that can incorporate functions of biological systems such as sensing, actuation and control. Advanced fibrous composites already possess many desirable characteristics for a whole range of structural applications. By adding the biological features of sensing, actuation and control, they can take a step further and given certain ‘true life’ features and, ‘intelligence’ [1]. Needless to say that such a development would substantially lessen concerns over the introduction of composite materials into aerospace, automotive and civil industries, as well as
1359-835X/99/$ - see front matter q 1999 Elsevier Science Ltd. All rights reserved. PII: S1359-835 X( 99)00 025-1
1188
B.P. Arjyal et al. / Composites: Part A 30 (1999) 1187–1195
Table 1 Elastic parameters for the fibre, the matrix and the Hexcel unidirectional 914k-49 composite E11 (GPa)
80.3 a
E22 (GPa)
8.0
G12 a (GPa)
2.1
n 12
0.35
Ef (GPa) # 0.5%
. 0.5%
110
125
Em (GPa)
a f (10 26 K 21)
a m (10 26 K 21)
3.5
6.5
60
Supplied by manufacturer.
significantly reduce the cost [2]. Already, there is a great deal of activity in developing optical fibres as strain and temperature sensors, shape memory alloys (SMA) as actuators, and piezoelectric materials as both strain sensors and actuators [3]. Neural networks are looked at as possible selflearning control systems for composite ‘smart’ structures [4]. For a structural composite, monitoring of strain represents one of the most important sensing endeavours. The standard method to measure strain in laminates is still by the use of intrusive electrical resistance strain gauges attached to the surface of the structure, and then by determining the ply-by-ply strains using laminate theory [5]. The technique of incorporating fibre-optics in composite materials for non-destructive evaluation and/or strain measuring purpose has progressed rapidly since the late 1970s. Fibreoptics sensors are being used for ‘smart’ structures as they have numerous advantages, the main one being their ability to serve the dual role of sensor and pathway for the signal. Since the optical fibre is a dielectric fibre it is quite compatible with the composite material and avoids creating electrical pathways within the structure. They are also of comparable size to a single ply of unidirectional laminate and have a relatively high modulus. These features enable the fibre-optic to integrate into the structure without affecting the structural properties particularly when it is placed parallel to the ply direction, whereas there is a slight decrease of strength in the transverse direction [6]. An optical sensor may be defined as a device in which an optical signal is modulated in response to a measured field. An optical signal is characterised by its wavelength, phase, intensity, and polarisation. In an optical sensor any one, or a combination of these parameters, may be modulated in response to external influence, i.e. strain, temperature, pressure [2,7,8]. However, independent concurrent measurements of all the external parameters are not easy to perform [9]. In composites, intensiometric (microbend) [6], interferometric (Mach-Zehnder, Michelson, FabryPerot, Bragg grating) [3,8,10,11], and polarimetric (high and low birefringence) [3], fibre-optic sensors have been evaluated primarily for strain measurements and detection of damage generated by impacts, manufacturing flaws, excessive loading or fatigue. Over the last few years, a Raman spectroscopic technique for non-destructive stress/strain measurements in composites has emerged [12]. The physical principles of this stress/strain sensing method has been described elsewhere
[13]. The main fundamental difference between the Raman sensor and all other existing sensors and techniques is that the measurements are conducted at the reinforcing fibre itself, at a resolution which is governed by the wavelength of the exciting laser light (typically 0.5–1 mm). Measurements at the surface or near-surface of an optically transparent or semi-transparent fibre composite can be performed non-destructively by just focussing at a single fibre or at a whole array of fibres with an appropriate microscope objective. Scanning of a whole composite component can also be done in service by using specially designed remote probes [14]. Thus the Raman sensor is advantageous over conventional electrical resistance strain gauges and fibre-optics as it provides higher resolution tailored to the testing requirements, it is non-intrusive, and is not limited to a specific sampling area. For composites, its usefulness stems from the fact that it is the only existing method to-date that yields the fibre stress in one of the composite principal directions (fibre axis). Calibration is usually conducted by stressing single fibres in air and by measuring independently the Raman wavenumber shift with stress and/or strain. It is worth adding that the Raman measurements correspond to physical changes of the measurand itself brought about by an applied stress field. On the contrary, electrical resistance or light modulation measurements are performed on attached or embedded external devices, that are assumed to be strained equally with the surrounding composite material. The laser Raman technique has been used extensively in the past to monitor the composite micromechanics in single fibre geometries and in full composites [12], to measure the residual thermal stresses [15] and to map the stress concentration in composites containing circular notches [16]. All this work was performed on either single fibres embedded into resins or on the near-surface plane of composites. The first attempt to obtain strain measurements from the bulk of the composite involved (a) the attachment of a polydiacetylene (PDA) fibre on the tip of fibre-optic and (b) the incorporation of the assembly into a carbon fibre composite [17]. This configuration was only partially successful due to the effect that the size of the PDA fibre had upon the results obtained [18]. In this work, an attempt is made to obtain high quality Raman data obtained directly from the reinforcing fibres in the bulk of the composite without the need for attaching a PDA crystal to the tip of the fibre-optic. To channel the laser light to the bulk of the composite fibreoptic cables are embedded into the laminae in unidirectional
B.P. Arjyal et al. / Composites: Part A 30 (1999) 1187–1195
1189
and multi-directional Kevlarw/epoxy laminated composite coupons. The work on multi-directional composites opens up the possibility of using this technique to conduct ply-byply point stress measurements in composites in service. Finally, the effect that an embedded fibre-optic cable has upon the mechanical properties of the composite is assessed.
2. Experimental 2.1. Materials/specimen preparation
Fig. 1. (a) Photograph of ‘bulk perpendicular’ standard Kevlarw49/epoxy composite tensile coupon. (b) Schematic of ‘bulk-perpendicular’ configuration. The items are not shown to scale.
Fig. 2. (a) Photograph of ‘bulk parallel’ standard composite Kevlarw49/ epoxy tensile coupon. (b) Schematic of ‘bulk-parallel’ configuration. The items are not shown to scale.
Eight-ply laminates were produced from unidirectional Kevlar 49w/epoxy resin pre-impregnated tapes supplied by Hexcel Composites (type 914k-49-54.8%, Table 1). To produce the bulk-perpendicular configuration the prepreg tapes were cut to strips of 203 mm in length and 12.7 mm in width and were then laid on top of each other with a cleaved fibre-optic embedded in the centre (Fig. 1). The multimode fibre-optic cables employed in this work were supplied by Newport Co. (model F-MLD) and had a nominal core diameter of 100 mm and a numerical aperture of 0.29. To produce the bulk-parallel configuration the cleaved fibre-optic cable is wrapped in a strand of Kevlarw fibres and placed in the centre of laminates in a direction parallel to the axis of the reinforcing fibres (Fig. 2). To produce the multidirectional [02, 245, 145]s configuration the laminates were cut to dimension and cleaved fibre-optic cables were placed in-between the two 1458 plies at the very centre of the laminate and at a direction perpendicular to the axis of the reinforcing fibres. Simultaneous measurements on the 08 ply were conducted by either interrogating the fibres at the surface of the laminate or by embedding a second fibre-optic between the 08 plies at a direction perpendicular to the reinforcing fibres. In all cases the pre-impregnated tapes were laid on the top of each other and cured in an autoclave for 1 h at 1758C under a pressure of 420 kN m 22, according to the manufacturer’s instructions. To avoid fracture of the cleaved fibreoptic during lay-up in the autoclave by resin bleeding onto the fibre, care was taken to seal the non-embedded part of the fibre-optic within sleeves made of peel-ply. The reinforcing fibres located near the surface of the laminate can be interrogated directly with the Raman microprobe. However, for consistency with the bulk measurements as above, the reinforcing fibres were interrogated through a cleaved fibre-optic bonded to a small squared PMMA block (8 mm). As shown in Fig. 3, the tip of the fibre-optic was then attached on to the surface of the laminate with a rubber-based adhesive. For volume fraction measurements, a typical test coupon was cut perpendicular to the reinforcing fibres, embedded into resin and polished. After polishing, the volume fraction of the reinforcing was determined using an Optomaxw V image analyser. Measurements in the bulk and on the surface were taken from 10 different composite samples
1190
B.P. Arjyal et al. / Composites: Part A 30 (1999) 1187–1195
Fig. 5. Experimental set-up for the mechanical testing of the tensile coupons. The bright spot at the middle of the specimen indicates the position in the bulk of the composite, which is analysed with the embedded fibre-optic.
Fig. 3. (a) Photograph of standard composite Kevlarw49/epoxy tensile coupon employed for surface measurements. (b) Schematic of ‘surface’ configuration. The items are not shown to scale.
and average volume fractions of 60 1 2% were obtained after autoclave curing. In Fig. 4(a) and (b), micrographs of sections of the composites for the bulk-perpendicular and bulk-parallel configurations, respectively, is shown. As can be seen, the fibre-optic that runs perpendicular to the ply direction creates a resin wedge, the long axis of which is approximately six times the fibre-optic diameter (Fig. 4(a)). On the contrary, no resin pockets are created in the bulkparallel configuration (Fig. 4(b)). 2.2. Mechanical testing The composites were tested in tension following the ASTM D3039-76 standard procedure [19]. Prior to testing, the ends of the tensile coupons were sand blasted and endtabbed with standard, 2.4 mm thick, glass-reinforced-plastic tabs. Strain gauges of gauge resistance of 350 ^ 1.0 V and of gauge factor of 2.03 were attached to the middle of the gauge section for each coupon. A total of 15 specimens were tested in tension; five of those specimens contained no embedded fibre-optics, whereas five incorporated bulkperpendicular and five bulk-parallel fibre-optic cables. All specimens were loaded up to fracture on a 20 kN screwdriven Hounsfield mechanical tester at a strain rate of approximately 0.002 min 21. 2.3. Laser Raman experiments
Fig. 4. (a) Micrograph depicting section through the fibre-optic and resin wedge, which run perpendicular to the ply direction. (b) Micrograph showing section through the fibre-optic and the reinforcing fibres for ‘parallel’ configuration.
A remote laser Raman microprobe (ReRaM) [14] was employed for all Raman measurements described here. The ReRaM incorporates flexible fibre-optics for the delivery of the laser light and for the collection of the inelastically (Raman) scattered light. The use of flexible fibre-optic cables permits operation of the microprobe in horizontal, vertical and multi-angle positions. Such an arrangement allows the analysis of specimens of any size and shape and under a variety of different environments such as
B.P. Arjyal et al. / Composites: Part A 30 (1999) 1187–1195
1191
Fig. 6. (a) Raman wavenumber shift as a function of tensile strain for Kevlarw49 fibres. The solid line is a least-squares-fit to the experimental data. (b) Raman wavenumber shift as a function of tensile stress for Kevlarw49 fibres in air. The solid line is a least-squares-fit to the experimental data.
mechanical testing labs, factory floors, service depots, chemically hostile chambers, etc. [14]. In Fig. 5, the experimental set-up employed for Raman spectra acquisition during mechanical loading of the composites is shown. For the stress/strain experiments, the input laser light of an argon-ion laser excited at 514.5 nm, was directed to the objective of the ReRaM and then was focused onto the free tip of the embedded fibre-optic. For moderate levels of laser power the 514.5 nm wavelength was not found to cause any damage to the reinforcing fibre. The Raman sensor was calibrated by loading single Kevlarw fibres in air and by plotting the shift of Raman frequency as a function of axial stress or strain. The light emitted by the fibre-optic was nonpolarised and, therefore, the calibration of the stress sensor in air was also carried through using non-polarised laser light. Both relationships shown in Fig. 6(a) and (b), can be considered linear, within experimental error although, in general, the Raman frequency vs. strain relationship can be best fitted with cubic spline polynomial curves [20]. The slope of the least-squares-fitted straight lines represent
the sensitivity of the stress or strain sensors and can be used to convert Raman frequency into values of fibre stress or strain in composites. The slope of the least-squares-fitted straight lines for stress was found to be 23.65 cm 21 GPa 21 (Fig. 6(a)), whereas the corresponding value for strain was 24.13 cm 21 % 21 (Fig. 6(b)). The 1808 scattered light from the bulk or the surface of the composite was collected via the same embedded fibreoptic and, by means of the ReRaM, was directed to the single monochromator for analysis and Raman light detection. The laser power was maintained constant throughout each test. In the case of the bulk–perpendicular experiments, 3 mW of incident laser radiation was directed to the embedded fibre-optic and measurements were taken at 10 s exposure intervals during the mechanical loading of the composite. For the bulk-parallel experiments, a laser power of 7 mW was used and measurements were taken every 20 s. At each Raman measurement, both the strain in the laminate—measured by means of the attached strain gauge— and the corresponding applied load, were independently recorded.
1192
B.P. Arjyal et al. / Composites: Part A 30 (1999) 1187–1195
Fig. 7. Fibre strain measured by the remote Raman microprobe using an embedded fibre-optic in a direction perpendicular to the reinforcing fibres of the composite vs. the strain in the composite measured by means of an attached strain gauge. The solid line is a least-squares-fit to the experimental data.
3. Results The fibre strain measured by the ReRaM is compared with composite strain, measured by means of an attached strain gauge, in Figs. 7–9. On average, 100 individual measurements of fibre strain vis-a`-vis strain gauge values were taken during tensile testing of the specimen. In all three cases (Figs. 7–9) the fibre strain measured using ReRaM increases in tandem with the strain gauge measurements. The slope of the least-squares-fitted straight lines is approximately 1 for all three configurations (Table 2). Another important observation made from Figs. 7–9 is that prior to loading the tensile coupons, the strain in the reinforcing Kevlarw49 fibres is found to be compressive. Residual compressive strain values were measured by taking the average of 10 Raman measurements prior to stressing the laminate. For bulk-perpendicular and bulkparallel configurations, fibre compressive strains of magnitude 400 and 300 me, respectively, have been obtained. The corresponding compressive strain for the surface measurements is 400 me. In Table 3, the effect of the presence of an embedded fibre-optic upon the integrity of the composite is investigated. As can be seen the presence of a fibre-optic cable
Fig. 9. Fibre strain measured by the remote Raman microprobe near the surface of the laminate vs. the strain in the composite measured by means of an attached strain gauge. The solid line is a least-squares-fit to the experimental data.
perpendicular to the reinforcing fibres reduces the tensile strength by approximately 10%, whereas the presence of a fibre parallel to the reinforcing fibres has no practical effect upon the tensile strength of the composite. Careful monitoring of the fractured process for the perpendicular configuration revealed failure in two out of total five specimens initiated at the point where the fibre-optic is embedded (Fig. 10). However, the post-mortem examination in all five specimens shows a typical ‘broom-like’ fracture for all specimens. In the case of the bulk-parallel configuration (Fig. 11), failure initiation was random as in the case of dummy coupons that contained no fibre-optics. In Fig. 12 shows the measured stress by ReRaM on fibres embedded perpendicular to 0 and to 1458 plies as a function of the applied stress over the whole [02, 245, 145]s composite coupon. As expected, the axial fibre stress in the 08 plies increase steeply with the applied stress, whereas the principal fibre stress in the 1458 ply is only a fraction of the applied composite stress (Fig. 12).
4. Discussion The results obtained from unidirectional composites (Figs. 7–9) show clearly that the fibre strain values obtained with the technique of Raman spectroscopy are in agreement with theoretical predictions, as well as, conventional strain gauge measurements. The superiority of the Raman sensor Table 2 Results of the gradients of fibre strain measured using ReRaM versus composite strain measured by means of an attached strain gauge
Fig. 8. Fibre strain measured by the remote Raman microprobe using an embedded fibre-optic in a direction parallel to the reinforcing fibres of the composite vs. the strain in the composite measured by means of an attached strain gauge. The solid line is a least-squares-fit to the experimental data.
Configuration
Slope of the least-squaresfitted straight lines for strain measurements
Standard error of the slope
Surface measurements Bulk perpendicular Bulk parallel
1.00 1.02 1.03
0.01 0.02 0.01
B.P. Arjyal et al. / Composites: Part A 30 (1999) 1187–1195
1193
Table 3 Ultimate tensile strength data for: (I) as-received coupons; (II) bulk perpendicular coupons; and (III) bulk parallel coupons No. of specimen
Ultimate tensile strength of bulk ' coupons (GPa)
Ultimate tensile strength of bulk k coupons (GPa)
Ultimate tensile strength of asreceived coupons
1 2 3 4 5 Mean (GPa) Standard deviation (GPa)
1.18 1.25 1.32 1.23 1.19 1.23 0.06
1.33 1.38 1.41 1.35 1.42 1.38 0.04
1.43 1.39 1.35 1.41 1.27 1.37 0.06
over other existing sensors is its ability to provide independently values of fibre strain, and stress if required, from composite sample volumes as small as 1 mm 3. In addition, this is the only technique that can directly measure stress in composites as most of the currently available non-destructive methods can only provide strain measurements [1]. As has also been demonstrated by Arjyal et al. [21,22] the Kevlarw49 sensor can also be incorporated in composites that contain fibres such as glass, whose Raman signal is considered too weak for accurate measurements. The observed compressive stress on fibres of the laminate prior to loading is due to the mismatch of the thermal expansion coefficients of the fibre and matrix and comes about as a
Fig. 10. Photograph of failed ‘bulk perpendicular’ standard composite Kevlarw49/epoxy tensile coupons.
result of curing and post-curing the composites at elevated temperatures. These high compressive stresses/strains are balanced by the corresponding tensile stresses/strains in the matrix and are responsible for matrix microcracking and/or premature matrix cracking/splitting during tensile loading of composites [13]. Nairn [23] and Nairn and Zoller [24] derived analytical treatments for the determination of the magnitude of residual thermal stresses, on the micromechanical level, in graphite and/or Kevlarw unidirectional composites:
sRf 2
ZT0 T
Da dT; 1=Ef 1 Vf =Em Vm
1
Da is the difference between the longitudinal thermal expansion coefficient of the matrix and the thermal expansion coefficient of the fibre, T and To are the ambient temperature set at 208C and the temperature for the onset of stress build-up (,1758C), respectively, Ef and Em are the longitudinal Young’s moduli of the fibre and the matrix respectively, and Vf and Vm are the fibre and matrix volume fractions of the composite respectively. All the relevant values are given in Table 1. By substituting the corresponding material properties (Table 1) in Eq. (1) the longitudinal thermal stresses in the embedded fibres is predicted to be 19 MPa. The obtained residual thermal stress in the fibre direction in the bulk range from 30 to 45 MPa for the parallel and perpendicular configurations, respectively. The corresponding value for the surface measurements is of the same order of magnitude (40 MPa) and, therefore, the presence of the fibre-optic does not seem to have any significant effect upon the measured values of fibre residual stresses. Since the estimated standard deviation of the mean values is of the order of ^25 MPa, one can safely assume that, within experimental error, there is a broad agreement of the Raman data with the analytical predictions. Numerous articles in the literature [25–27] have described the effect of optical fibres upon the mechanical properties of the structures that have been incorporated into. All these studies led to the conclusion that the integrity of the composites with embedded fibre-optics is only slightly weakened if the maximum fibre diameter is no more than
1194
B.P. Arjyal et al. / Composites: Part A 30 (1999) 1187–1195
Fig. 11. Photograph of failed ‘bulk parallel’ standard composite Kevlar849/ epoxy tensile coupons.
the ply thickness (,125 mm) and if their respective volume fraction is not too high. The work carried out by Jensen et al. [26] indicated that the effect of embedded optical fibre orientation on the laminate tensile mechanical properties is of the same order of magnitude with the experimental scatter, which originates from the material inhomogeneity and brittleness. However, the same study pointed out that the optical fibres embedded perpendicular to either loading direction and/or the reinforcing fibres induce the largest degradation (,10%) in the mechanical properties (e.g. tensile strength and stiffness) of composite laminates. As the angle between the optical fibre and adjacent ply
Fig. 12. Fibre stress in the 0 and 1458 plies, measured by the remote Raman microprobe using embedded fibre-optics in a direction perpendicular to the fibres. The geometry of the coupon is [02, 245, 145]s.
Fig. 13. (a) Fibre stress measured by the remote Raman microprobe in the 08 ply vs. the calculated fibre stress obtained by LAP. The geometry of the coupon is [02, 245, 145]s. The solid line is a least-squares-fit to the experimental data. (b) Fibre stress measured by the remote Raman microprobe on 1458 ply vs. the calculated fibre stress using LAP. The geometry of the coupon is [02, 245, 145]s. The solid line is a least-squares-fit to the experimental data.
increases, the span of the resin-rich zone also increases [27]. The resin pocket acts like a void running through the laminate and can trigger the initiation of interlaminar failure under mechanical load. As has been demonstrated here, the Raman stress/strain sensor can be employed for unidirectional, as well as, multidirectional composites. The stress measurements using ReRaM, on the fibres in the 0 and 1458 plies have been compared against the calculated fibre stress using the Laminate Analysis Programme [28], which employs standard laminate theory to calculate stress on each individual ply. The program requires knowledge of E11, E12, G12, n 12 ply thickness, ply orientation and curing temperature to calculate the theoretical stress along the fibre direction in any given ply. Fibre stress measurements using ReRaM are compared with the calculated fibre stress using LAP for the 145 ply and 08 ply respectively in Fig. 13(a) and (b). The slope of the least-squares-fitted straight lines for measurements in the 0 and 1458 plies are 1.05 1 0.01 and 0.97 1 0.02, respectively. The ReRaM fibre stress measurements in the 0 ply and 1458 ply, prior to loading the tensile coupons, show compressive stresses in the reinforcing Kevlar849 fibres of magnitude 34 and 126 MPa, respectively. The high compressive stresses in the principal
B.P. Arjyal et al. / Composites: Part A 30 (1999) 1187–1195
(1458) direction in an angle ply are generated during cooling of the laminate from the curing temperature (1758C) down to room temperature [29]. A closer inspection of the data points revealed that the relationship is not quite linear. Useful information on the deformation mechanisms in these geometries can be extracted from the relationship between the stress in the principal fibre direction within an angle ply to that of the applied stress in the 08 direction (Fig. 13). Conventional laminate analysis does not account for nonlinear deformations in the principal direction of an angle ply, possibly due to matrix yielding or other non-linear effects, as observed in the 1458 ply examined here. 5. Conclusions A new stress/strain sensor for localised measurements in polymer based composites has been built and tested. Measurements can be conducted on the near-surface of a unidirectional laminate and also in the bulk. The latter measurements are performed by employing the fibre-optic cables to transport the exciting laser radiation to the embedded reinforcing fibres in the composite. The two configurations tested for bulk measurements (parallel and perpendicular) yielded similar results, however, the presence of the fibre-optic perpendicular to the reinforcing fibres reduces slightly the tensile strength of the composite. The take-up of fibre strain for both the bulk and surface set of measurements was identical with that obtained from the attached electrical resistance strain gauges. The residual fibre stress in the as-manufactured composites was also measured and was found comparable to analytical predictions. An attempt was also made to measure the fibre stresses in the principal direction in angle plies of a multidirectional composites. The results for the 08 plies were in good agreement with those obtained by conventional laminate analysis whereas small deviations from linearity were observed in the angle plies. The proposed methodology paves the way for simultaneous in-service stress/strain measurements on the reinforcing fibres situated on the surface or within the bulk of a composite laminate.
Acknowledgements The experimental work was conducted in the Materials Department of Queen Mary & Westfield College (QMW), University of London. Mr V. Chohan and Dr A. Paipetis are thanked for assisting with the acquisition of the experimental data. The British Council is thanked for
1195
providing Dr Tarantili with a visiting scholarship to QMW. References [1] Rogers CA. In: Gardiner P, Kelly A, Bunsell AR, editors. Smart composites, ECCM-6, Cambridge: Woodhead Publications, 1993. pp. 3. [2] Measures RM. Prog Aerospace Sci 1989;26:289. [3] Friend MC. Interdisciplinary Sci Rev 1996;21(3):195. [4] Boller C. In: Gobin PF, Tatibouet J, editors. Proceedings of the Third International Conference on Intelligent Materials/Third European Conference on Smart Structures and Materials, Philadelphia: SPIE, 1996. pp. 16. [5] Tsai SW, Hahn HT. Introduction to composite materials. Lancaster, PA: Technomic, 1980. [6] Claus RO, Bennett K, Jackson B. Review of progress in quantitative NDE, Williamsburg, VA, Vol. 5, 1985. p. 1149. [7] Medlock RS. J Opt Sensors 1986;1:43. [8] Culshaw B, Dakin J. Optical fibre sensors, 1. Norwood, MA: Artech House, 1988 Vol. 2 (1989). [9] Grattan KTV, Meggitt BT. Optical fiber sensor technology. London: Chapman and Hall, 1995. [10] Jackson DA, Jones JDC. Fibre optic sensors. Opt Acta 1986;33:1469. [11] De Paula P, Moore EL. Fibre optic sensors overview. Fibre optic and laser sensors III, 566. Philadelphia: SPIE, 1985. pp. 2. [12] Galiotis C. Micromechanics of reinforcement using laser Raman spectroscopy. In: Summerscales J, editor. Microstructural characterisation of fibre-reinforced composites, Cambridge: Woodhead Publishing, 1998. pp. 224. [13] Schadler L, Galiotis C. A review of the fundamentals and applications of LRS microprobe strain measurements. Int Mater Rev 1995;40(3):116. [14] Paipetis A, Vlattas C, Galiotis C. J Raman Spectrosc 1996;27:519. [15] Filiou CD, Galiotis C, Batchelder DN. Composites 1992;28(1):28. [16] Filiou CD, Galiotis C. Composite Science and Technology. Submitted for publication. [17] Underwood FM, Sharpe D, Batchelder DN. In: Bunsell AR, Lamicq P, Massiah A, editors. Developments in the science and technology of composite materials, ECCM3, London: Elsevier, 1989. pp. 759. [18] Underwood FM. PhD thesis, University of London, 1990. [19] American Society for Testing and Materials. Standard test method for tensile properties of fibre–resin composites. Designation D3039-76, ASTM, Philadelphia, 118 (Re-approved 1989). [20] Vlattas C, Galiotis C. Polymer 1994;35(11):2335. [21] Arjyal BP, Galiotis C, Ogin SL, Whattingham RD. J Mater Sci 1998;33(11):2745. [22] Arjyal BP, Galiotis C, Ogin SL. Whattingham. Composites Part A 1998;29(11):1363. [23] Nairn JA. Polym Composites 1985;6:123. [24] Nairn JA, Zoller P. J Mater Sci 1985;20:355. [25] Roberts SSJ, Davidson R. Conference proceedings. SPIE, Vol. 1588, 1991. p. 326. [26] Jensen DW, Pascual J, August JA. Smart Mater Struct 1992;1:24. [27] Dasgupta A, Wan Y, Sirkis JS. Smart Mater Struct 1992;1:101. [28] Laminate Analysis Program (LAP), DOS version. Center of Composite Materials, Imperial College, London. [29] Hull D, Clyne TW. An introduction to composite materials. Cambridge: Cambridge University Press, 1996.
Surface and bulk stress/strain measurements in composite laminates with a fibre-optic Raman probe B.P. Arjyal a, P.A. Tarantili b, A.G. Andreopoulos b, C. Galiotis a,c,* a
Materials Department, Queen Mary & Westfield College, Mile End Road, London E1 4NS, UK Department of Chemical Engineering, National Technical University of Athens, 9 Iroon Polytechniou Str., 157 80 Zografou, Athens, Greece c Institute of Chemical Engineering and High Temperature Chemical Processes, Foundation of Research & Technology—Hellas, P.O. Box 1414, GR-265 00, Patras, Greece b
Received 4 October 1998; accepted 8 March 1999
Abstract Fibre stress/strain measurements in unidirectional, as well as, multidirectional aramid/epoxy composites have been conducted with the use of a laser Raman microprobe. The composite was incrementally loaded in tension while Raman measurements were taken. Fibre-optic probes sandwiched between adjacent laminae were employed for channelling the laser excitation light to a specified location within the bulk of the composite. The direction of the fibre-optic was either perpendicular or parallel to the reinforcing fibres. For comparison purposes, the same fibre-optic probe was used to scan the surface of the laminates. The perpendicular configuration was found to reduce the tensile strength of the as-received composite coupon by 10% whereas the parallel second configuration had no effect. In the unidirectional coupons the stress or strain in the principal fibre direction could be measured prior to loading and at every increment of applied tensile load up to fracture. The take-up of fibre strain for both bulk and surface set of measurements was identical with that obtained from the attached electrical resistance strain gauges. In the case of multidirectional coupons the stress or strain in the principal direction could be measured within successive plies situated at angles u to the loading direction. The results for the 08 plies were in good agreement with those obtained by conventional laminate analysis whereas small deviations from linearity were observed in the angle plies. The proposed methodology paves the way for simultaneous in-service stress/strain measurements on the reinforcing fibres situated on the surface or within the bulk of a composite laminate. q 1999 Elsevier Science Ltd. All rights reserved. Keywords: Stress; A. Laminates
Nomenclature
af am T E11, E22
e 1, e 2, e 3 s 11, s 22, s 33 n 12 G12, G23 Ef, E m Vf, V m Gm ASTM
longitudinal coefficient of thermal expansion for the fibre coefficient of thermal expansion of the matrix temperature elastic moduli in the principal material directions principal normal strains principal normal stresses Poisson’s ratio shear moduli in the 1–2 and 2–3 planes fibre and matrix Young’s moduli volume fractions of fibre and matrix shear modulus of matrix American society for testing and materials
* Corresponding author. E-mail address: [email protected] (C. Galiotis)
CCD LRS PDA ReRaM
charge coupled device laser Raman spectroscopy polydiacetylene remote Raman microprobe
1. Introduction Over the last two decades, lot of work has been put into developing structural materials that can incorporate functions of biological systems such as sensing, actuation and control. Advanced fibrous composites already possess many desirable characteristics for a whole range of structural applications. By adding the biological features of sensing, actuation and control, they can take a step further and given certain ‘true life’ features and, ‘intelligence’ [1]. Needless to say that such a development would substantially lessen concerns over the introduction of composite materials into aerospace, automotive and civil industries, as well as
1359-835X/99/$ - see front matter q 1999 Elsevier Science Ltd. All rights reserved. PII: S1359-835 X( 99)00 025-1
1188
B.P. Arjyal et al. / Composites: Part A 30 (1999) 1187–1195
Table 1 Elastic parameters for the fibre, the matrix and the Hexcel unidirectional 914k-49 composite E11 (GPa)
80.3 a
E22 (GPa)
8.0
G12 a (GPa)
2.1
n 12
0.35
Ef (GPa) # 0.5%
. 0.5%
110
125
Em (GPa)
a f (10 26 K 21)
a m (10 26 K 21)
3.5
6.5
60
Supplied by manufacturer.
significantly reduce the cost [2]. Already, there is a great deal of activity in developing optical fibres as strain and temperature sensors, shape memory alloys (SMA) as actuators, and piezoelectric materials as both strain sensors and actuators [3]. Neural networks are looked at as possible selflearning control systems for composite ‘smart’ structures [4]. For a structural composite, monitoring of strain represents one of the most important sensing endeavours. The standard method to measure strain in laminates is still by the use of intrusive electrical resistance strain gauges attached to the surface of the structure, and then by determining the ply-by-ply strains using laminate theory [5]. The technique of incorporating fibre-optics in composite materials for non-destructive evaluation and/or strain measuring purpose has progressed rapidly since the late 1970s. Fibreoptics sensors are being used for ‘smart’ structures as they have numerous advantages, the main one being their ability to serve the dual role of sensor and pathway for the signal. Since the optical fibre is a dielectric fibre it is quite compatible with the composite material and avoids creating electrical pathways within the structure. They are also of comparable size to a single ply of unidirectional laminate and have a relatively high modulus. These features enable the fibre-optic to integrate into the structure without affecting the structural properties particularly when it is placed parallel to the ply direction, whereas there is a slight decrease of strength in the transverse direction [6]. An optical sensor may be defined as a device in which an optical signal is modulated in response to a measured field. An optical signal is characterised by its wavelength, phase, intensity, and polarisation. In an optical sensor any one, or a combination of these parameters, may be modulated in response to external influence, i.e. strain, temperature, pressure [2,7,8]. However, independent concurrent measurements of all the external parameters are not easy to perform [9]. In composites, intensiometric (microbend) [6], interferometric (Mach-Zehnder, Michelson, FabryPerot, Bragg grating) [3,8,10,11], and polarimetric (high and low birefringence) [3], fibre-optic sensors have been evaluated primarily for strain measurements and detection of damage generated by impacts, manufacturing flaws, excessive loading or fatigue. Over the last few years, a Raman spectroscopic technique for non-destructive stress/strain measurements in composites has emerged [12]. The physical principles of this stress/strain sensing method has been described elsewhere
[13]. The main fundamental difference between the Raman sensor and all other existing sensors and techniques is that the measurements are conducted at the reinforcing fibre itself, at a resolution which is governed by the wavelength of the exciting laser light (typically 0.5–1 mm). Measurements at the surface or near-surface of an optically transparent or semi-transparent fibre composite can be performed non-destructively by just focussing at a single fibre or at a whole array of fibres with an appropriate microscope objective. Scanning of a whole composite component can also be done in service by using specially designed remote probes [14]. Thus the Raman sensor is advantageous over conventional electrical resistance strain gauges and fibre-optics as it provides higher resolution tailored to the testing requirements, it is non-intrusive, and is not limited to a specific sampling area. For composites, its usefulness stems from the fact that it is the only existing method to-date that yields the fibre stress in one of the composite principal directions (fibre axis). Calibration is usually conducted by stressing single fibres in air and by measuring independently the Raman wavenumber shift with stress and/or strain. It is worth adding that the Raman measurements correspond to physical changes of the measurand itself brought about by an applied stress field. On the contrary, electrical resistance or light modulation measurements are performed on attached or embedded external devices, that are assumed to be strained equally with the surrounding composite material. The laser Raman technique has been used extensively in the past to monitor the composite micromechanics in single fibre geometries and in full composites [12], to measure the residual thermal stresses [15] and to map the stress concentration in composites containing circular notches [16]. All this work was performed on either single fibres embedded into resins or on the near-surface plane of composites. The first attempt to obtain strain measurements from the bulk of the composite involved (a) the attachment of a polydiacetylene (PDA) fibre on the tip of fibre-optic and (b) the incorporation of the assembly into a carbon fibre composite [17]. This configuration was only partially successful due to the effect that the size of the PDA fibre had upon the results obtained [18]. In this work, an attempt is made to obtain high quality Raman data obtained directly from the reinforcing fibres in the bulk of the composite without the need for attaching a PDA crystal to the tip of the fibre-optic. To channel the laser light to the bulk of the composite fibreoptic cables are embedded into the laminae in unidirectional
B.P. Arjyal et al. / Composites: Part A 30 (1999) 1187–1195
1189
and multi-directional Kevlarw/epoxy laminated composite coupons. The work on multi-directional composites opens up the possibility of using this technique to conduct ply-byply point stress measurements in composites in service. Finally, the effect that an embedded fibre-optic cable has upon the mechanical properties of the composite is assessed.
2. Experimental 2.1. Materials/specimen preparation
Fig. 1. (a) Photograph of ‘bulk perpendicular’ standard Kevlarw49/epoxy composite tensile coupon. (b) Schematic of ‘bulk-perpendicular’ configuration. The items are not shown to scale.
Fig. 2. (a) Photograph of ‘bulk parallel’ standard composite Kevlarw49/ epoxy tensile coupon. (b) Schematic of ‘bulk-parallel’ configuration. The items are not shown to scale.
Eight-ply laminates were produced from unidirectional Kevlar 49w/epoxy resin pre-impregnated tapes supplied by Hexcel Composites (type 914k-49-54.8%, Table 1). To produce the bulk-perpendicular configuration the prepreg tapes were cut to strips of 203 mm in length and 12.7 mm in width and were then laid on top of each other with a cleaved fibre-optic embedded in the centre (Fig. 1). The multimode fibre-optic cables employed in this work were supplied by Newport Co. (model F-MLD) and had a nominal core diameter of 100 mm and a numerical aperture of 0.29. To produce the bulk-parallel configuration the cleaved fibre-optic cable is wrapped in a strand of Kevlarw fibres and placed in the centre of laminates in a direction parallel to the axis of the reinforcing fibres (Fig. 2). To produce the multidirectional [02, 245, 145]s configuration the laminates were cut to dimension and cleaved fibre-optic cables were placed in-between the two 1458 plies at the very centre of the laminate and at a direction perpendicular to the axis of the reinforcing fibres. Simultaneous measurements on the 08 ply were conducted by either interrogating the fibres at the surface of the laminate or by embedding a second fibre-optic between the 08 plies at a direction perpendicular to the reinforcing fibres. In all cases the pre-impregnated tapes were laid on the top of each other and cured in an autoclave for 1 h at 1758C under a pressure of 420 kN m 22, according to the manufacturer’s instructions. To avoid fracture of the cleaved fibreoptic during lay-up in the autoclave by resin bleeding onto the fibre, care was taken to seal the non-embedded part of the fibre-optic within sleeves made of peel-ply. The reinforcing fibres located near the surface of the laminate can be interrogated directly with the Raman microprobe. However, for consistency with the bulk measurements as above, the reinforcing fibres were interrogated through a cleaved fibre-optic bonded to a small squared PMMA block (8 mm). As shown in Fig. 3, the tip of the fibre-optic was then attached on to the surface of the laminate with a rubber-based adhesive. For volume fraction measurements, a typical test coupon was cut perpendicular to the reinforcing fibres, embedded into resin and polished. After polishing, the volume fraction of the reinforcing was determined using an Optomaxw V image analyser. Measurements in the bulk and on the surface were taken from 10 different composite samples
1190
B.P. Arjyal et al. / Composites: Part A 30 (1999) 1187–1195
Fig. 5. Experimental set-up for the mechanical testing of the tensile coupons. The bright spot at the middle of the specimen indicates the position in the bulk of the composite, which is analysed with the embedded fibre-optic.
Fig. 3. (a) Photograph of standard composite Kevlarw49/epoxy tensile coupon employed for surface measurements. (b) Schematic of ‘surface’ configuration. The items are not shown to scale.
and average volume fractions of 60 1 2% were obtained after autoclave curing. In Fig. 4(a) and (b), micrographs of sections of the composites for the bulk-perpendicular and bulk-parallel configurations, respectively, is shown. As can be seen, the fibre-optic that runs perpendicular to the ply direction creates a resin wedge, the long axis of which is approximately six times the fibre-optic diameter (Fig. 4(a)). On the contrary, no resin pockets are created in the bulkparallel configuration (Fig. 4(b)). 2.2. Mechanical testing The composites were tested in tension following the ASTM D3039-76 standard procedure [19]. Prior to testing, the ends of the tensile coupons were sand blasted and endtabbed with standard, 2.4 mm thick, glass-reinforced-plastic tabs. Strain gauges of gauge resistance of 350 ^ 1.0 V and of gauge factor of 2.03 were attached to the middle of the gauge section for each coupon. A total of 15 specimens were tested in tension; five of those specimens contained no embedded fibre-optics, whereas five incorporated bulkperpendicular and five bulk-parallel fibre-optic cables. All specimens were loaded up to fracture on a 20 kN screwdriven Hounsfield mechanical tester at a strain rate of approximately 0.002 min 21. 2.3. Laser Raman experiments
Fig. 4. (a) Micrograph depicting section through the fibre-optic and resin wedge, which run perpendicular to the ply direction. (b) Micrograph showing section through the fibre-optic and the reinforcing fibres for ‘parallel’ configuration.
A remote laser Raman microprobe (ReRaM) [14] was employed for all Raman measurements described here. The ReRaM incorporates flexible fibre-optics for the delivery of the laser light and for the collection of the inelastically (Raman) scattered light. The use of flexible fibre-optic cables permits operation of the microprobe in horizontal, vertical and multi-angle positions. Such an arrangement allows the analysis of specimens of any size and shape and under a variety of different environments such as
B.P. Arjyal et al. / Composites: Part A 30 (1999) 1187–1195
1191
Fig. 6. (a) Raman wavenumber shift as a function of tensile strain for Kevlarw49 fibres. The solid line is a least-squares-fit to the experimental data. (b) Raman wavenumber shift as a function of tensile stress for Kevlarw49 fibres in air. The solid line is a least-squares-fit to the experimental data.
mechanical testing labs, factory floors, service depots, chemically hostile chambers, etc. [14]. In Fig. 5, the experimental set-up employed for Raman spectra acquisition during mechanical loading of the composites is shown. For the stress/strain experiments, the input laser light of an argon-ion laser excited at 514.5 nm, was directed to the objective of the ReRaM and then was focused onto the free tip of the embedded fibre-optic. For moderate levels of laser power the 514.5 nm wavelength was not found to cause any damage to the reinforcing fibre. The Raman sensor was calibrated by loading single Kevlarw fibres in air and by plotting the shift of Raman frequency as a function of axial stress or strain. The light emitted by the fibre-optic was nonpolarised and, therefore, the calibration of the stress sensor in air was also carried through using non-polarised laser light. Both relationships shown in Fig. 6(a) and (b), can be considered linear, within experimental error although, in general, the Raman frequency vs. strain relationship can be best fitted with cubic spline polynomial curves [20]. The slope of the least-squares-fitted straight lines represent
the sensitivity of the stress or strain sensors and can be used to convert Raman frequency into values of fibre stress or strain in composites. The slope of the least-squares-fitted straight lines for stress was found to be 23.65 cm 21 GPa 21 (Fig. 6(a)), whereas the corresponding value for strain was 24.13 cm 21 % 21 (Fig. 6(b)). The 1808 scattered light from the bulk or the surface of the composite was collected via the same embedded fibreoptic and, by means of the ReRaM, was directed to the single monochromator for analysis and Raman light detection. The laser power was maintained constant throughout each test. In the case of the bulk–perpendicular experiments, 3 mW of incident laser radiation was directed to the embedded fibre-optic and measurements were taken at 10 s exposure intervals during the mechanical loading of the composite. For the bulk-parallel experiments, a laser power of 7 mW was used and measurements were taken every 20 s. At each Raman measurement, both the strain in the laminate—measured by means of the attached strain gauge— and the corresponding applied load, were independently recorded.
1192
B.P. Arjyal et al. / Composites: Part A 30 (1999) 1187–1195
Fig. 7. Fibre strain measured by the remote Raman microprobe using an embedded fibre-optic in a direction perpendicular to the reinforcing fibres of the composite vs. the strain in the composite measured by means of an attached strain gauge. The solid line is a least-squares-fit to the experimental data.
3. Results The fibre strain measured by the ReRaM is compared with composite strain, measured by means of an attached strain gauge, in Figs. 7–9. On average, 100 individual measurements of fibre strain vis-a`-vis strain gauge values were taken during tensile testing of the specimen. In all three cases (Figs. 7–9) the fibre strain measured using ReRaM increases in tandem with the strain gauge measurements. The slope of the least-squares-fitted straight lines is approximately 1 for all three configurations (Table 2). Another important observation made from Figs. 7–9 is that prior to loading the tensile coupons, the strain in the reinforcing Kevlarw49 fibres is found to be compressive. Residual compressive strain values were measured by taking the average of 10 Raman measurements prior to stressing the laminate. For bulk-perpendicular and bulkparallel configurations, fibre compressive strains of magnitude 400 and 300 me, respectively, have been obtained. The corresponding compressive strain for the surface measurements is 400 me. In Table 3, the effect of the presence of an embedded fibre-optic upon the integrity of the composite is investigated. As can be seen the presence of a fibre-optic cable
Fig. 9. Fibre strain measured by the remote Raman microprobe near the surface of the laminate vs. the strain in the composite measured by means of an attached strain gauge. The solid line is a least-squares-fit to the experimental data.
perpendicular to the reinforcing fibres reduces the tensile strength by approximately 10%, whereas the presence of a fibre parallel to the reinforcing fibres has no practical effect upon the tensile strength of the composite. Careful monitoring of the fractured process for the perpendicular configuration revealed failure in two out of total five specimens initiated at the point where the fibre-optic is embedded (Fig. 10). However, the post-mortem examination in all five specimens shows a typical ‘broom-like’ fracture for all specimens. In the case of the bulk-parallel configuration (Fig. 11), failure initiation was random as in the case of dummy coupons that contained no fibre-optics. In Fig. 12 shows the measured stress by ReRaM on fibres embedded perpendicular to 0 and to 1458 plies as a function of the applied stress over the whole [02, 245, 145]s composite coupon. As expected, the axial fibre stress in the 08 plies increase steeply with the applied stress, whereas the principal fibre stress in the 1458 ply is only a fraction of the applied composite stress (Fig. 12).
4. Discussion The results obtained from unidirectional composites (Figs. 7–9) show clearly that the fibre strain values obtained with the technique of Raman spectroscopy are in agreement with theoretical predictions, as well as, conventional strain gauge measurements. The superiority of the Raman sensor Table 2 Results of the gradients of fibre strain measured using ReRaM versus composite strain measured by means of an attached strain gauge
Fig. 8. Fibre strain measured by the remote Raman microprobe using an embedded fibre-optic in a direction parallel to the reinforcing fibres of the composite vs. the strain in the composite measured by means of an attached strain gauge. The solid line is a least-squares-fit to the experimental data.
Configuration
Slope of the least-squaresfitted straight lines for strain measurements
Standard error of the slope
Surface measurements Bulk perpendicular Bulk parallel
1.00 1.02 1.03
0.01 0.02 0.01
B.P. Arjyal et al. / Composites: Part A 30 (1999) 1187–1195
1193
Table 3 Ultimate tensile strength data for: (I) as-received coupons; (II) bulk perpendicular coupons; and (III) bulk parallel coupons No. of specimen
Ultimate tensile strength of bulk ' coupons (GPa)
Ultimate tensile strength of bulk k coupons (GPa)
Ultimate tensile strength of asreceived coupons
1 2 3 4 5 Mean (GPa) Standard deviation (GPa)
1.18 1.25 1.32 1.23 1.19 1.23 0.06
1.33 1.38 1.41 1.35 1.42 1.38 0.04
1.43 1.39 1.35 1.41 1.27 1.37 0.06
over other existing sensors is its ability to provide independently values of fibre strain, and stress if required, from composite sample volumes as small as 1 mm 3. In addition, this is the only technique that can directly measure stress in composites as most of the currently available non-destructive methods can only provide strain measurements [1]. As has also been demonstrated by Arjyal et al. [21,22] the Kevlarw49 sensor can also be incorporated in composites that contain fibres such as glass, whose Raman signal is considered too weak for accurate measurements. The observed compressive stress on fibres of the laminate prior to loading is due to the mismatch of the thermal expansion coefficients of the fibre and matrix and comes about as a
Fig. 10. Photograph of failed ‘bulk perpendicular’ standard composite Kevlarw49/epoxy tensile coupons.
result of curing and post-curing the composites at elevated temperatures. These high compressive stresses/strains are balanced by the corresponding tensile stresses/strains in the matrix and are responsible for matrix microcracking and/or premature matrix cracking/splitting during tensile loading of composites [13]. Nairn [23] and Nairn and Zoller [24] derived analytical treatments for the determination of the magnitude of residual thermal stresses, on the micromechanical level, in graphite and/or Kevlarw unidirectional composites:
sRf 2
ZT0 T
Da dT; 1=Ef 1 Vf =Em Vm
1
Da is the difference between the longitudinal thermal expansion coefficient of the matrix and the thermal expansion coefficient of the fibre, T and To are the ambient temperature set at 208C and the temperature for the onset of stress build-up (,1758C), respectively, Ef and Em are the longitudinal Young’s moduli of the fibre and the matrix respectively, and Vf and Vm are the fibre and matrix volume fractions of the composite respectively. All the relevant values are given in Table 1. By substituting the corresponding material properties (Table 1) in Eq. (1) the longitudinal thermal stresses in the embedded fibres is predicted to be 19 MPa. The obtained residual thermal stress in the fibre direction in the bulk range from 30 to 45 MPa for the parallel and perpendicular configurations, respectively. The corresponding value for the surface measurements is of the same order of magnitude (40 MPa) and, therefore, the presence of the fibre-optic does not seem to have any significant effect upon the measured values of fibre residual stresses. Since the estimated standard deviation of the mean values is of the order of ^25 MPa, one can safely assume that, within experimental error, there is a broad agreement of the Raman data with the analytical predictions. Numerous articles in the literature [25–27] have described the effect of optical fibres upon the mechanical properties of the structures that have been incorporated into. All these studies led to the conclusion that the integrity of the composites with embedded fibre-optics is only slightly weakened if the maximum fibre diameter is no more than
1194
B.P. Arjyal et al. / Composites: Part A 30 (1999) 1187–1195
Fig. 11. Photograph of failed ‘bulk parallel’ standard composite Kevlar849/ epoxy tensile coupons.
the ply thickness (,125 mm) and if their respective volume fraction is not too high. The work carried out by Jensen et al. [26] indicated that the effect of embedded optical fibre orientation on the laminate tensile mechanical properties is of the same order of magnitude with the experimental scatter, which originates from the material inhomogeneity and brittleness. However, the same study pointed out that the optical fibres embedded perpendicular to either loading direction and/or the reinforcing fibres induce the largest degradation (,10%) in the mechanical properties (e.g. tensile strength and stiffness) of composite laminates. As the angle between the optical fibre and adjacent ply
Fig. 12. Fibre stress in the 0 and 1458 plies, measured by the remote Raman microprobe using embedded fibre-optics in a direction perpendicular to the fibres. The geometry of the coupon is [02, 245, 145]s.
Fig. 13. (a) Fibre stress measured by the remote Raman microprobe in the 08 ply vs. the calculated fibre stress obtained by LAP. The geometry of the coupon is [02, 245, 145]s. The solid line is a least-squares-fit to the experimental data. (b) Fibre stress measured by the remote Raman microprobe on 1458 ply vs. the calculated fibre stress using LAP. The geometry of the coupon is [02, 245, 145]s. The solid line is a least-squares-fit to the experimental data.
increases, the span of the resin-rich zone also increases [27]. The resin pocket acts like a void running through the laminate and can trigger the initiation of interlaminar failure under mechanical load. As has been demonstrated here, the Raman stress/strain sensor can be employed for unidirectional, as well as, multidirectional composites. The stress measurements using ReRaM, on the fibres in the 0 and 1458 plies have been compared against the calculated fibre stress using the Laminate Analysis Programme [28], which employs standard laminate theory to calculate stress on each individual ply. The program requires knowledge of E11, E12, G12, n 12 ply thickness, ply orientation and curing temperature to calculate the theoretical stress along the fibre direction in any given ply. Fibre stress measurements using ReRaM are compared with the calculated fibre stress using LAP for the 145 ply and 08 ply respectively in Fig. 13(a) and (b). The slope of the least-squares-fitted straight lines for measurements in the 0 and 1458 plies are 1.05 1 0.01 and 0.97 1 0.02, respectively. The ReRaM fibre stress measurements in the 0 ply and 1458 ply, prior to loading the tensile coupons, show compressive stresses in the reinforcing Kevlar849 fibres of magnitude 34 and 126 MPa, respectively. The high compressive stresses in the principal
B.P. Arjyal et al. / Composites: Part A 30 (1999) 1187–1195
(1458) direction in an angle ply are generated during cooling of the laminate from the curing temperature (1758C) down to room temperature [29]. A closer inspection of the data points revealed that the relationship is not quite linear. Useful information on the deformation mechanisms in these geometries can be extracted from the relationship between the stress in the principal fibre direction within an angle ply to that of the applied stress in the 08 direction (Fig. 13). Conventional laminate analysis does not account for nonlinear deformations in the principal direction of an angle ply, possibly due to matrix yielding or other non-linear effects, as observed in the 1458 ply examined here. 5. Conclusions A new stress/strain sensor for localised measurements in polymer based composites has been built and tested. Measurements can be conducted on the near-surface of a unidirectional laminate and also in the bulk. The latter measurements are performed by employing the fibre-optic cables to transport the exciting laser radiation to the embedded reinforcing fibres in the composite. The two configurations tested for bulk measurements (parallel and perpendicular) yielded similar results, however, the presence of the fibre-optic perpendicular to the reinforcing fibres reduces slightly the tensile strength of the composite. The take-up of fibre strain for both the bulk and surface set of measurements was identical with that obtained from the attached electrical resistance strain gauges. The residual fibre stress in the as-manufactured composites was also measured and was found comparable to analytical predictions. An attempt was also made to measure the fibre stresses in the principal direction in angle plies of a multidirectional composites. The results for the 08 plies were in good agreement with those obtained by conventional laminate analysis whereas small deviations from linearity were observed in the angle plies. The proposed methodology paves the way for simultaneous in-service stress/strain measurements on the reinforcing fibres situated on the surface or within the bulk of a composite laminate.
Acknowledgements The experimental work was conducted in the Materials Department of Queen Mary & Westfield College (QMW), University of London. Mr V. Chohan and Dr A. Paipetis are thanked for assisting with the acquisition of the experimental data. The British Council is thanked for
1195
providing Dr Tarantili with a visiting scholarship to QMW. References [1] Rogers CA. In: Gardiner P, Kelly A, Bunsell AR, editors. Smart composites, ECCM-6, Cambridge: Woodhead Publications, 1993. pp. 3. [2] Measures RM. Prog Aerospace Sci 1989;26:289. [3] Friend MC. Interdisciplinary Sci Rev 1996;21(3):195. [4] Boller C. In: Gobin PF, Tatibouet J, editors. Proceedings of the Third International Conference on Intelligent Materials/Third European Conference on Smart Structures and Materials, Philadelphia: SPIE, 1996. pp. 16. [5] Tsai SW, Hahn HT. Introduction to composite materials. Lancaster, PA: Technomic, 1980. [6] Claus RO, Bennett K, Jackson B. Review of progress in quantitative NDE, Williamsburg, VA, Vol. 5, 1985. p. 1149. [7] Medlock RS. J Opt Sensors 1986;1:43. [8] Culshaw B, Dakin J. Optical fibre sensors, 1. Norwood, MA: Artech House, 1988 Vol. 2 (1989). [9] Grattan KTV, Meggitt BT. Optical fiber sensor technology. London: Chapman and Hall, 1995. [10] Jackson DA, Jones JDC. Fibre optic sensors. Opt Acta 1986;33:1469. [11] De Paula P, Moore EL. Fibre optic sensors overview. Fibre optic and laser sensors III, 566. Philadelphia: SPIE, 1985. pp. 2. [12] Galiotis C. Micromechanics of reinforcement using laser Raman spectroscopy. In: Summerscales J, editor. Microstructural characterisation of fibre-reinforced composites, Cambridge: Woodhead Publishing, 1998. pp. 224. [13] Schadler L, Galiotis C. A review of the fundamentals and applications of LRS microprobe strain measurements. Int Mater Rev 1995;40(3):116. [14] Paipetis A, Vlattas C, Galiotis C. J Raman Spectrosc 1996;27:519. [15] Filiou CD, Galiotis C, Batchelder DN. Composites 1992;28(1):28. [16] Filiou CD, Galiotis C. Composite Science and Technology. Submitted for publication. [17] Underwood FM, Sharpe D, Batchelder DN. In: Bunsell AR, Lamicq P, Massiah A, editors. Developments in the science and technology of composite materials, ECCM3, London: Elsevier, 1989. pp. 759. [18] Underwood FM. PhD thesis, University of London, 1990. [19] American Society for Testing and Materials. Standard test method for tensile properties of fibre–resin composites. Designation D3039-76, ASTM, Philadelphia, 118 (Re-approved 1989). [20] Vlattas C, Galiotis C. Polymer 1994;35(11):2335. [21] Arjyal BP, Galiotis C, Ogin SL, Whattingham RD. J Mater Sci 1998;33(11):2745. [22] Arjyal BP, Galiotis C, Ogin SL. Whattingham. Composites Part A 1998;29(11):1363. [23] Nairn JA. Polym Composites 1985;6:123. [24] Nairn JA, Zoller P. J Mater Sci 1985;20:355. [25] Roberts SSJ, Davidson R. Conference proceedings. SPIE, Vol. 1588, 1991. p. 326. [26] Jensen DW, Pascual J, August JA. Smart Mater Struct 1992;1:24. [27] Dasgupta A, Wan Y, Sirkis JS. Smart Mater Struct 1992;1:101. [28] Laminate Analysis Program (LAP), DOS version. Center of Composite Materials, Imperial College, London. [29] Hull D, Clyne TW. An introduction to composite materials. Cambridge: Cambridge University Press, 1996.