Effect of Loading-Unloading Cycles on Impact-Damaged Jute/Glass Hybrid Laminates

Igor Maria De Rosa,1 Carlo Santulli,2 Fabrizio Sarasini,1 Marco Valente1 1 Department of Chemical Engineering, Materials, Environment Universita` di Roma- La Sapienza, 00184 Rome, Italy 2

Department of Electrical Engineering, Universita` di Roma- La Sapienza, 00184 Rome, Italy

The production of glass/plant fiber hybrid laminates is a possibility for obtaining semistructural materials with sufficient impact properties, and a better life cycle analysis (LCA) profile than fiberglass. The simplest and possibly the most effective configuration for the production of these hybrids would involve the use of a plant fiber reinforced laminate as the core between two glass fiber reinforced laminates. A main limitation to the use of composites including plant fibers is that their properties may be significantly affected by the presence of damage, so that even the application of a low stress level can result in laminate failure. In particular, it is suggested that when loading is repeatedly applied and removed, residual properties may vary in an unpredictable way. In this work, E-glass/jute hybrid reinforced laminates, impacted in a range of energies (10, 12.5, and 15 J), have been subjected to post-impact cyclic flexural tests with a step loading procedure. This would allow evaluating the effect of damage dissipation offered by the plant fiber reinforced core. The tests have also been monitored by acoustic emission (AE), which has confirmed the existence of severe limitations to the use of this hybrid material when impacted at energies close to penetration. POLYM. COMPOS., 00:000– 000, 2009. ª 2009 Society of Plastics Engineers

INTRODUCTION In recent years, the possibility to use plant fiber reinforced composites as semistructural materials has been widely investigated [1–4]. This possibility is mainly based on their reduced cost, coupled with a higher environmental friendliness, as suggested by LCA studies (e.g., [5]). However, there appear to be sound reasons for limiting the use of these materials in load-bearing applications such as environmental sensitivity, poor moisture resistance, unsatisfactory impact performance, problematic Correspondence to: Carlo Santulli; e-mail: [email protected] DOI 10.1002/pc.20789 Published online in Wiley InterScience (www.interscience.wiley.com). C 2009 Society of Plastics Engineers V

POLYMER COMPOSITES—-2009

damage detection, and limited knowledge of post-impact residual properties. In this context, glass/plant fiber hybrid laminates have been often investigated, for their ability to represent a suitable trade-off between environmental gain and mechanical properties [6]. In particular, a stacking sequence including a laminate reinforced with plant fibers sandwiched between two glass fiber reinforced laminates (hereinafter referred to as sandwich hybrid laminates) appears to be a relatively simple and suitable route for this purpose. Alternative processes have also been investigated, which show some limitations: for example, intermingling does not allow introducing a large volume of plant fibers [7] or commingling requires so far the use of polypropylene matrices [8]. A study on the falling weight impact properties of flax/E-glass fiber sandwich hybrids suggested that extensive damage present in the flax fiber reinforced core was able to dissipate a significant amount of energy, hence suggesting some structural and not just cosmetic function for the hybrid laminate [9]. Another main question would concern post-impact residual properties. It has been noted that the variation of residual properties of jute fiber reinforced laminates mainly depends on the number and extent of microscopic defects on the fibers, and so impact damage may allow pre-existing defects to degenerate. As a consequence, it is very difficult to correlate impact energy applied with the damage produced [10]. In this sense, it is important to clarify whether sandwich hybrid laminates show a more predictable behavior, as regards post-impact residual properties. This study follows the line traced from [9–11], with the aim of evaluating the differences in post-impact behavior between laminates, through a procedure consisting of flexural loading-unloading cycles carried out on specimens impacted in a range of energies. Impact energies were selected as sufficiently close to penetration energy for a plant fiber composite with a comparable fiber volume fraction, as measured, for instance, in Ref. 12.

FIG. 1. Experimental set-up of flexural tests monitored by acoustic emission.

This would allow understanding whether the plant fiber core of the hybrid laminate is able to withstand this level of impact loading; if confirmed, this would give some scope for using this composite structure in semistructural applications. Testing is assisted by AE, which has seldom been used on plant fiber composites [13]. MATERIALS AND TEST METHODS Jute/glass fiber hybrid laminates were manufactured using a RTM procedure, as a sandwich structure with jute/polyester layers as core and E-glass/polyester as skins. The stacking sequence was 7 Glass/4 Jute/7 Glass. Plain woven jute fabric (300 g/m2) and E-glass (Vetrotex VR38, 290 g/m2) were used as reinforcing fibers in this work. The resin was unsaturated polyester (1629 NT from Lonza). The overall fiber fraction of the composites (Vf) was 0.50 6 0.025 by volume, as determined from the preform basic weight (weight per unit area). From the plates were cut the four-point bend specimens having a length of 150 mm, a width of 30 mm, and a thickness of 5(6 0.2) mm. The specimens were impacted and then subjected to post-impact four-point bending tests. The impact point was located at the center of the specimens. The impact energy was changed varying the mass of the hemispherical drop-weight striker (ø ¼ 12.7 mm), thus having a constant velocity of 2.5 m/s. Impact tests were performed on an instrumented impact tower fitted with an anti-rebound device. Before being impacted, the specimens were clamped to ensure that they were not laterally displaced during the impact. Three different impact energies were considered: 10 J, 12.5 J, and 15 J. Flexural tests were carried out in accordance with ASTM D-790 (the configuration used was the quarter-point loading); however, the dimensions of the specimens were selected in order to place the acoustic emission sensors on the specimen surface. These tests were performed in a universal testing 2 POLYMER COMPOSITES—-2009

FIG. 2. Variation of flexural strength and modulus with impact energy for hybrid specimens (error bars represent mean 6 standard deviation).

machine (Zwick Roell Z010) with a support span length of 140 mm and a cross-head speed of 5 mm/min. Straingauges were used to measure strain at the mid-span. Five specimens were tested to measure the bending properties of the laminates; after this, five specimens were tested for each impact energy, including non-impacted specimens, which served as reference materials. To assess postimpact flexural properties of the hybrid laminates, tests were carried using a step loading procedure, similar to the one adopted in Ref. 10, on jute/polyester laminates. In particular, the cyclic procedure adopted, aimed at simulating service life, consisted of consecutive steps, each including 30 loading-unloading cycles, performed between 0 and rmax. The maximum stress value was increased in the ith step with respect to the (i–1)th step according to this rule: rmax-i ¼ rmax-(i-1) þ 10%rf, until final fracture occurred. In all cases, strain gauges were used to monitor the strain in composite laminates. Post-impact flexural tests were monitored by acoustic emission until final fracture occurred, using an AMSY-5 AE system by Vallen Systeme GmbH. The AE acquisition

FIG. 3. Flexural modulus of laminates vs. post-impact flexural load.

DOI 10.1002/pc

TABLE 1. Number of samples failed vs. level of flexural load (% flexural strength) per each impact energy. Impact energy (J)

Failed at 60% of flexural strength

Failed at 70% of flexural strength

Failed at 80% of flexural strength

Failed at 90% of flexural strength

0 10 12.5 15

– – – 3

– – 3 2

3 4 2 –

2 1 – –

settings used throughout this experimental work were as follows: threshold ¼ 40 dB, RT (Rearm Time) ¼ 0.4 ms, DDT (Duration Discrimination Time) ¼ 0.2 ms, and total gain ¼ 34 dB. The PZT AE sensors used were resonant at 150 kHz. Four sensors were used: two of them were placed on the surface of the specimens at both ends to allow linear localization, while the other two were used as guard sensors in order to discriminate between AE signals and noise and were positioned on the flexure fixture. Vacuum grease was used as a couplant and the sensors were also fastened using adhesive tape. The experimental set-up is shown in Fig. 1.

RESULTS From the results shown in Fig. 2, the main effect of impact appears to be a very significant reduction both of flexural strength and modulus, depending on impact energy. Cyclic flexural tests (Fig. 3) also indicate that a significant stiffness reduction is obtained for all the impacted laminates. In addition, while all of the nonimpacted and 10 J impacted laminates failed at least during the step at 80% of the ultimate strength, in some cases reaching also the 90% step, three out of five 12.5 J laminates failed at only 70%, and three out of five 15 J laminates reached only the 60% step: loading level leading to failure is reported for all the laminates in Table 1. This suggests a wide scattering in post-impact properties, which is likely to be due to degeneration of pre-existing defects in plant fiber reinforced laminates during cyclic loading. Acoustic emission analysis revealed that AE activity appears much earlier in the 15 J impacted laminates than elsewhere. Also, the number of localized AE events confirms that this impact energy produces an amount of damage largely exceeding that of the other two energy levels considered (Fig. 4). The analysis of acoustic emission during post-impact flexural tests was mainly focused on amplitude distributions. The spatial distribution of the events in the plots allowed discerning between two typical modes of emission: one due to flexural loading (Fig. 5a) and the other due to the prior presence of impact damage (Fig. 5b). Although the former results in the concentration of events in a straight region of the laminate, the latter is indicated by clustering of events in a quasi-circular region. DOI 10.1002/pc

In Fig. 6a–d typical amplitude distribution plots for laminates impacted at all energies are represented to offer a visual description of damage distribution and localization. To better clarify amplitude distribution in all the laminates, the results of an analysis similar to the one performed in Ref. 14 are also given in Table 2. The initial stages of loading, representing 10 and 20% of the flexural strength, show for all impact energies few and sparse acoustic emission events; this indicates that at this level of loading no further damage is produced. For higher stresses, the behavior of 15 J impacted specimens is markedly different, since a much larger number of acoustic emission events, spread in the whole area of the tested specimen, is produced at earlier stages (in particular 30 and 40%) than for laminates impacted at lower energies. This is likely to indicate that at this level of impact energy the core laminate is extensively damaged in the whole section and damage propagates also to the glass reinforced laminate on the nonimpacted side. As a whole, it can be noted that higher amplitude events (exceeding 75– 80 dB), often associated with extensive fiber breakage [15], are only detected in the last period of the final stage, where the failure of the laminate takes place. This behavior is also confirmed by the analysis of signal durations. As can be observed in Table 3, a significant number of long duration (>1 ms) AE events have been detected in the non impacted laminates. These could correspond to the opening of macro-cracks because of flexural loading, mainly in the less dampened glass fiber reinforced layers, as proposed in Ref. 16. Long duration

FIG. 4. Localized AE events vs. % of flexural strength.

POLYMER COMPOSITES—-2009 3

FIG. 5. (a) (left) AE amplitude distribution showing lining-up events (likely flexural damage) (nonimpacted laminate at 40% of ultimate flexural strength); (b) (right) AE amplitude distribution showing clustering of events (likely impact damage) (laminate impacted at 12.5 J at 20% of ultimate flexural strength).

FIG. 6. (a) Typical AE events amplitude distribution for non impacted laminates at all levels of flexural load; (b) Typical AE events amplitude distribution for laminates impacted at 10 J at all levels of flexural load; (c) Typical AE events amplitude distribution for laminates impacted at 12.5 J at all levels of flexural load; (d) Typical AE events amplitude distribution for laminates impacted at 15 J at all levels of flexural load.

FIG. 6. (Continued from the previous page)

events are also noted at high flexural stress for all impact energies, but in lesser number than for nonimpacted samples. This is likely to suggest that all levels of impact produce non-negligible delamination at the interface between core and skins, so that macro-cracks are present already in the laminate. As a consequence, the contribution of post-impact loading to damage development on impacted laminates becomes significant only at high flexural stress. As suggested in Ref. 17, a minor part of acoustic emission activity could also be due to jute fibers damage that causes their splitting from the strands. DOI 10.1002/pc

In fact, failure in all the hybrids started on the compressive side close to the interface between the glass skins and the jute core and was easily diffused in the laminate core, because of propagation of pre-existing defects. A limited number of these defects are in a sense inherent in the structure, because fibers like jute, which are extracted from the bast of a plant, have an irregular cellular structure with variable wall thickness and orientation, and therefore internal hollows have different geometries and size [18]. Furthermore, non optimal adhesion between jute fibers and matrix can be ascribed both to the hydroPOLYMER COMPOSITES—-2009 5

TABLE 2. Relative distribution (%) of AE events amplitudes in jute/glass hybrid composites. Impact energy (J)

Percentage of flexural strength (%)

0

10 20 30 40 50 60 70 80 10 20 30 40 50 60 70 80 10 20 30 40 50 60 70 80 10 20 30 40 50 60

10

12.5

15

Amplitude interval (dB) 40–49

64.28 17.24 17.02 16.34 28.70 33.94 8.76 25.00 12.50 32.35 26.19 24.36 33.04 29.64 14.28 27.27 20.75 13.64 11.89 24.82 20.10 44.76 16.67 38.74 35.83 35.59 42.42 36.79

50–59

60–69

75.00 21.43 45.98 46.81 47.03 53.39 50.80 35.65 50.00 25.00 37.50 26.47 42.86 61.54 40.00 39.55 28.57 63.64 67.92 73.86 82.16 59.49 64.02 44.83 61.11 58.56 54.72 59.89 50.00 46.17

25.00 7.14 29.88 24.47 28.22 11.73 10.93 46.19 50.00 25.00 37.50 14.70 28.57 12.82 17.39 21.55 28.57 9.09 11.33 11.36 4.86 10.95 13.23 8.87 22.22 1.80 7.82 3.87 6.83 13.61

phobic nature of resin used and to the variable diameter along different sections of each jute fiber. Once started, the failure could propagate along this weak interface, leading to extensive delamination and a relatively hindrance-free penetration of the hybrid laminate core (Fig. 7). This raises again concerns on the predictability of failure in laminates including plant fibers, which can be suggested to be even lower in the case of preimpacted laminates. This mode of failure is consistent with indications from previous works, that jute weave reinforced laminates break by fibers being torn off perpendicularly to the direction of loading [19]. This is likely to be because of the hardly effective load transfer from the matrix to the fibers, when using hydrophobic polymer resins, a problem which may be addressed using more complex warp-weft architecture [20]. To better explain the behavior of hybrid laminates, further analysis of acoustic emission data obtained for laminates impacted at high energy has been carried out. In particular, a further threshold on the amplitude has been set, to try to measure the fraction of total AE events characterized by high amplitude. This is a common practice in the analysis of composites behavior using AE [21]. In this way, R60, an amplitude-based variable, was defined as the ratio between the events exceeding 60 dB amplitude 6 POLYMER COMPOSITES—-2009

70–79

7.14 6.90 10.64 5.94 4.94 2.28 8.40

80–89

1.06 2.47 0.62 1.37 0.50

90–99

0.62 0.68 0.50

25.00 12.50 17.65 1.28 9.57 8.10 14.28

1.14 1.09 3.28 2.65 1.35

8.83 2.38

0.58 14.28

0.58

1.46 0.12

0.07

0.90 1.63 0.65 0.75 3.20

0.20

0.03

Loc. events 8 14 87 94 202 324 439 2373 2 4 8 34 42 78 115 1211 7 11 53 88 185 274 378 4283 18 111 307 930 2416 12637

and the total number of AE events at high impact energies. In Fig. 8, the values of R60 are plotted against the percentage of flexural strength. It should be noted that at 10 and 20% stages, the number of AE events is very low, sometimes being reduced to a few units, so that the related values of R60 are only reported for completeness, but should not be relied upon. In general, the study of R60 clarifies that only during the stage leading to failure (80% of maximum strength for impacts from 0 up to 12.5 J and 60% for 15 J), a larger number of high amplitude events are generated, as shown by the higher value of R60. In addition, for 15-J impacted laminates, the average number of AE events per fatigue cycle is considerably higher than for those impacted at 10 and 12.5 J (Table 4). The above results suggest that passing from 12.5 J to 15 J impacted laminates leads to an abrupt growth of damage, in spite of the small difference in impact energy between the two series, possibly because of the degeneration of the interface between skins and core. This, together with previous considerations on damage characterization, offers further evidence of the brittle mode of failure for laminates including plant fibers, a problem which can be aggravated further by the presence of low-velocity impact damage, so that it is difficult to predict the level of stress that will cause failure in the damaged laminate. DOI 10.1002/pc

DOI 10.1002/pc

POLYMER COMPOSITES—-2009 7

15

12.5

10

0

Impact energy (J)

10 20 30 40 50 60 70 80 10 20 30 40 50 60 70 80 10 20 30 40 50 60 70 80 10 20 30 40 50 60

Percentage of flexural strength (%)

100 100 73.56 84.04 80.20 90.74 93.62 53.14 100 50 100 73.53 88.09 94.87 80.87 70.19 57.14 90.91 94.34 95.45 96.75 87.23 93.39 91.36 83.33 76.58 89.90 90.43 90.48 76.20

0–199

26.47 9.53 5.13 18.26 17.50 28.58 9.09 5.66 4.55 3.25 12.77 6.61 7.28 16.67 23.42 10.10 9.57 9.44 12.78

50

26.44 13.83 16.35 7.41 5.02 34.93

200–399

0.08 4.25

0.72

0.87 4.87 14.28

2.38

5.80

2.13 0.99 0.62

400–599

2.41

0.28

2.39

1.68

600–799

1.54

0.12

1.15

0.31 0.45 1.09

800–999

0.89

0.02

0.58

0.23 1.09

0.49

1000–1199

0.61

0.07

0.66

0.55

0.31

1200–1399

Duration interval (ls)

0.38

0.02

0.41

0.29

1400–1599

TABLE 3. Relative distribution (%) of AE events durations in jute/glass hybrid composites.

0.20

0.41

0.21

0.49

1600–1799

0.20

0.25

0.23 0.25

0.99

1800–1999

0.54

0.12

1.57

0.49 0.62 0.45 0.97

[2000

TABLE 4. AE events per cycle averaged on all the laminates at all impact energy levels.

FIG. 7. 15 J impacted sample before loading (top) and at failure (bottom).

CONCLUSIONS As a conclusion, this work confirms concerns about low velocity impact resistance of hybrid laminates includ-

Impact energy (J)

Avg. AE events/cycle

0 10 12.5 15

71.53 43.33 51.39 91.3

ing plant fibers. Impact produces a significant level of damage, so that post-impact residual strength can be remarkably low, in particular when a cyclic loading is applied. The main critical factor which affects the strength appears to be the state of the core-skin interface; consequently, it would be worthy to limit damage propagation in jute fibers. This could be done, for instance, by limiting the number of pre-existing fiber defects or by alternative reinforcement architectures for jute fibers (e.g., weaving, braiding, or commingling). More specifically, this work would like to contribute to the creation of a database of post-impact residual properties for these materials. Further studies should concentrate on suggesting strategies to improve damage dissipation properties of plant fiber reinforced laminates. Acoustic emission can assist characterization, confirming properties degradation

FIG. 8. Average R60 ratio (AE events [ 60 dB/total AE events) on all the laminates at all levels of flexural load. 8 POLYMER COMPOSITES—-2009

DOI 10.1002/pc

resulting from the application of impact approaching penetration.

energies

REFERENCES 1. P. Wambua, B. Vangrimde, V. Lomov, and I. Verpoest, Compos. Struct., 77(2), 232 (2007). 2. T. Yuanjian, and D.H. Isaac, Compos. Sci. Technol., 67 (15–16), 3300 (2007). 3. A.K. Mohanty, M. Misra, and L.T. Drzal, Eds. Natural Fibers, Biopolymers, and Biocomposites, CRC Press, Boca Raton, FL (2005). 4. M.J. John and S. Thomas. Carb. Polym., 71(3), 343 (2008). 5. S.V. Joshi, L.T. Drzal, A.K. Mohanty, and S. Arora, Compos. A Appl. Sci. Manuf., 5(3), 371 (2004). 6. C. Santulli, J. Mater. Sci., 42(11), 3699 (2007). 7. C. Pavithran, P.S. Mukherjee, and M. Brahmakumar, J. Reinf. Plast. Comp., 10(1), 91 (1991). 8. H.D. Rozman, G.S. Tay, R.N. Kumar, A. Abubakar, H. Ismail, and I.Z.A. Mohd, Polym. Plast. Tech. Eng., 38(6), 997 (1999).

DOI 10.1002/pc

9. C. Santulli, M. Janssen, and G. Jeronimidis, J. Mater. Sci., 40(13), 3581 (2005). 10. C. Santulli, C. J. Mater. Sci., 41(4), 1255 (2006). 11. C. Santulli, NDT&E Int., 34(8), 531 (2001). 12. C. Santulli, Sci. Eng. Compos. Mater., 9(4), 177 (200). 13. T. Czigany, J. Compos. Mater., 38(9), 769 (2004). 14. J.S. Szˇabo and T. Czygany, Polym. Test., 22(6), 711 (2003). 15. M. Johnson and P. Gudmunson, Compos. Sci. Technol., 60(15), 2803 (2000). 16. S.C. Woo, D.J. Kim, and N.S. Choi, Key Eng. Mater., 306– 308(1–2), 19 (2006). 17. A.K. Sabeel, S. Vijayarangan, and C. Rajput, J. Reinf. Plast. Comp., 25(15), 1549 (2006). 18. H.J. Purz, H.P. Fink, and H. Graf, Das Papier, 52(6), 315 (1998). 19. C. Santulli and W.J. Cantwell, J. Mater. Sci. Lett., 20(5), 477 (2001). 20. L.H. de Carvalho, J.M.F. Cavalcante, and J.R.M. d’Almeida, Polymer Plast. Tech. Eng., 45(7), 791 (2006). 21. Z.Z. Xingmin and Y. Xiong, Compos. Sci. Technol., 66(3– 4), 444 (2006).

POLYMER COMPOSITES—-2009 9