Micro and macro analysis of sisal fibre composites hollow core sandwich panels

Composites: Part B 43 (2012) 2738–2745

Contents lists available at SciVerse ScienceDirect

Composites: Part B journal homepage: www.elsevier.com/locate/compositesb

Micro and macro analysis of sisal fibre composites hollow core sandwich panels S. Rao, K. Jayaraman, D. Bhattacharyya ⇑ Centre for Advanced Composite Materials, Department of Mechanical Engineering, The University of Auckland, Auckland Mail Centre 1142, New Zealand

a r t i c l e

i n f o

Article history: Received 13 May 2011 Received in revised form 7 August 2011 Accepted 8 August 2011 Available online 18 May 2012 Keywords: A. Polymer–matrix composites (PMCs) B. Stress relaxation C. Micro-mechanics C. Finite element analysis (FEA)

a b s t r a c t In the view of the growing environmental concerns, hollow cores from recyclable natural fibre composites were manufactured to reduce the undesirable impact on the environment. To evaluate the feasibility of using short sisal fibres as reinforcements in the composites, existing micromechanical models have been used to predict properties starting from the intrinsic properties of its constituents. The stress relaxation behaviour of the composites was examined experimentally by performing tensile stress relaxation tests and to understand the process, it was modelled using variations of Maxwell’s model. A steady-state finite element analysis in the linear range was performed in ANSYS environment to examine flexural properties of the panels, and the shear strength of the hollow cores was experimentally determined by subjecting them to flexural loads in a four-point bending scheme. The micromechanics models indicated that the fibres had failed to provide effective reinforcements with their existing lengths, acting as fillers rather than reinforcements. The stress relaxation models indicated that the formed part needs to be cooled to room temperature within the die under suitable forming loads to avoid local deformations due to warping. The mid-span deflections of the sandwich panels predicted by the FE model agree well with the experimental results, the analysis predicted facing buckling as a mode of failure when wood veneers facings of modulus 4.5 GPa and thickness 1.7 mm were used. The specific shear strengths of the reinforced core are more than twice than those of the unreinforced polypropylene cores, increasing the scope of such panels as structural members in various engineering facets. Ó 2012 Published by Elsevier Ltd.

1. Introduction In the current age of global environment awareness and the strict legislations that are being passed by the concerned governments, bio-composites have received significant attention to help mitigate the perceived pollution caused by their synthetic counterparts. The relatively low cost of the recyclable thermoplastic polymers, renewability of the fibres and the increasingly favourable mechanical properties exhibited by the natural fibre reinforced thermoplastics have attracted many industries to utilise them in various facets of engineering in order to reduce synthetic material consumption and the energy footprint [1,2]. At the same time as minimising the undesirable impact on the environment, the composite material must meet the needs of the end application. Many a times, due to their relatively poor mechanical properties, the natural fibre composites are limited to non-structural applications. Hence, to extend its functionality, a relatively new concept of manufacturing and evaluating hollow cores reinforced with short natural fibres composites has been attempted in this study. As the cores were to be used in structural panels, it is necessary that they possess suitable mechanical properties, which in turn ⇑ Corresponding author. E-mail address: [email protected] (D. Bhattacharyya). 1359-8368/$ - see front matter Ó 2012 Published by Elsevier Ltd. http://dx.doi.org/10.1016/j.compositesb.2012.04.033

depend on the properties of the composites they have been fabricated from. Many a times, the manufacturing processes of such composites limit the possibility of exploiting their properties to the fullest, for example, the process of extrusion in a counter rotating twin screw extruder may limit the length of the reinforcing fibres being used, and compression and injection moulding process between flat platens may restrict the fibre orientation of the short fibres to being randomly oriented. In essence, the mechanical properties of short fibre reinforced polymer composites are directly or indirectly influenced by the manufacturing processes limiting the choice for material selection. This could be determined experimentally or can be estimated using mathematical models based on the properties of the individual components of the composite and their arrangement. The advantage of mathematical modelling is it reduces the cost and time consumed in experimentation by providing approximate solutions in terms of best possible combination of the materials. Therefore, by applying micromechanics models effectively selection of raw materials is possible [3–8]. For example, Kalaprasad et al. [9] have used models based on rule of mixtures (ROM) [3] and shear-lag [4] to extract the elastic modulus and tensile strength of sisal fibre composites and concluded the discrepancy in agreement to be due to fibre–matrix interfacial bonding condition, indicating a need for a coupling agent to improve the interfacial bonding state. In many cases, the difference

S. Rao et al. / Composites: Part B 43 (2012) 2738–2745

2739

Table 1 Micromechanics models used evaluate extruded sisal-PP composite sheets. Equation

Name

Model

Nomenclature

1 [7] 2 [3]

Voigt’s model Reuss’s model

Ec = EfVf + EmVm; Tc = TfVf + TmVm

3 [9]

Hirsh’s model

Ec ¼ xðEf V f þ Em V m Þ þ ð1  xÞ ðEf V mfþEmm V f Þ

E E

E E

T c ¼ xðT f V f þ T m V m Þ þ ð1 

4 [4]

Halpin–Tsai model

Ec ¼ Em

g¼ Cox’s shear-lag model



1þngV f 1gV f



; Tc ¼ Tm



1þngV f 1gV f



f =Em Þ1 g ¼ ðE ðEf =Em Þþn ;

L

2ðTL Þ;

D – Fibre diameter

g – Shear-lag parameter

ð2Þ

0

Nairn modified shear-lag model

2

p Pf = p for square packing and 2p/ 3 for hexagonal packing

112 2Em

Ef ð1þtm Þ ln

6 [6]

c – composite m – matrix f – fibre x – interfacial adhesion parameter n – shape fitting parameter L – Fibre length

ðT f =T m Þ1 ðT f =T m Þþn

g ¼ 1r @

V – volume fraction

T T xÞ ðT f V mfþTmm V f Þ

n¼2 D   gL tanh ð Þ Ec ¼ Ef 1  gL 2 V f þ Em V m ; ð2Þ   tanhðgLÞ T c ¼ T f 1  gL 2 V f þ T m V m n¼

5 [5]

E – elastic modulus T – strength

T T

Ec ¼ V m MffþVm f Mm ; T c ¼ V m T ffþVmf T m

 A Pf Vf

2

g ¼ 4r2 E2m Ef 4

3312 Vm 1 4Gf þ2Gm

h

Ef V f þEm V m 1 Vm

ln



1 V f þv



1V2m

i

þrD1s

v – correction factor

55 Gm – shear modulus of the fibre Gf – shear modulus of matrix Ds – adhesion parameter R – radius of the fibre centres

in the polarity of the fibre and matrix, presence of moisture in the fibres, uneven dispersion in the matrix and their physical variability make it challenging for researchers to develop a general model for composites with natural fibres as reinforcements. Andersons et al. [10] have tackled the length and dispersion problem of flax fibre reinforced thermoplastics by incorporating correction factors in the ROM equations whereas Facca et al. [11] have extended the models by incorporating correction factors which take moisture content into consideration, providing opportunity to use the extended models in case of variable fibre distributions and moisture content, some commonly used micromechanics models in predicting mechanical properties of the composites are listed in Table 1. Most of the models are based on certain assumptions [8] and considering them valid for the material considered, it may be possible to point out the areas of deficiency in the composite materials such as fibre–matrix adhesion, fibre lengths which could be modified prior manufacturing, in effect reducing wastage and cost incurred during experimentation. Thermoplastic composites are often subjected to secondary operations such as thermoforming to manufacture articles of desired shape and performance. In this study, the composite sheets were thermoformed into hexagonal and sinusoidal profiles to serve as cores in structural sandwich panels. There are several conventional methods of manufacturing hollow cores which are well established and frequently used [12] but their application to thermoplastic composites is relatively new. Many patents regarding new manufacturing techniques for thermoplastics have been filed in the recent years [13–15] but none mention incorporating short natural fibres in their cell walls. Therefore, to add value to natural fibre composites, hollow cores in the study were manufactured from short natural fibre composites with the intention of improving the mechanical performance of their unreinforced thermoplastic counterparts. The extruded natural fibre composites were thermoformed between matched dies or roll formed through matching rolls to obtain corrugated sheets for the cores. During such secondary operations, after processing and subsequent cooling of composites

from a relatively high forming temperature to service temperature, residual stresses frequently arise due to the mismatch of shrinkage of the matrix and the fibres of the composite. These residual stresses affect the matrix dominated properties such as moisture absorption by accelerating it and causing the matrix to swell, which in turn would change the stress state in the composites, affecting the fibre–matrix interfacial adhesion [16]. The most common effect of residual stresses in composite products is the dimensional non-conformance of the formed part with the die geometry due to spring-in or spring-back effect of the angled parts. Due to the time-dependent nature of the polymer matrix, the resulting residual stresses become time dependent too, hence when a certain strain is applied and held constant, the composite material shows an increase in stress initially, which gradually decreases with time due to the molecular relaxation process that takes place within the polymer [17]. This stress-time information of the specimen could be valuable in practical applications such as thermoforming processes where it could be applied to reduce the shape conformance problems of the formed part either by incorporating the results towards the die design or by altering the cure kinetics. The honeycomb cores in this study were manufactured via thermoforming between matched dies or matching rolls, which were then ultrasonically bonded to form honeycomb cores for sandwich panels. As sandwich panels behave similar to I-beams, the shear properties of the cores contribute substantially toward the overall performance of the panel. Therefore, four-point flexural test was simulated in ANSYS finite element environment using to examine the shear properties of the cores. The cell walls were modelled as a repetitive unit cell which could be translated in the x and y directions to form a complete core. The faceplates were assembled in the transverse direction (out-of-plane, along z axis) to form a honeycomb core sandwich panel and SHELL181 elements were used to model individual cell walls and the faceplate. The supports were placed at a distance L of 180 mm apart and equally distributed point loads as force of P/2 were applied to nodes at a distance L/4 from the supports. As the beam was simply supported, the

2740

S. Rao et al. / Composites: Part B 43 (2012) 2738–2745

Table 2 Properties of sisal fibres and PP matrix used for this study.

3

qf (kg/m ) Vf Ef (GPa)

rf (MPa) tf Gf (MPa)

Sisal fibre

PP

1300 0.13 0.23 5.25 578 0.17 2.4

900 0.87 0.77 2.6 30 0.42 0.41

Matrix property

qm (kg/m3)

4

Vm Em (GPa) rm (MPa)

tm Gm (MPa)

nodes at one of the support were constrained for translation in the x, y and z direction (Ux, Uy, Uz, constrained and free to rotate Rotx, Roty and Rotz) and the other end was constrained for translation in the z direction. The length, thickness and the angle between the horizontal members and the inclined members of the cell wall were measured under an optical microscope to be 9.8 mm, 1.3–1.5 mm and 58–60°, respectively. Therefore, for modelling, cell wall thickness of 1.4 mm, angle of 60° and length of 9.8 mm were considered. As the cores were manufactured from composite sheets which exhibited negligible orthotropy (along x–y plane), the composite material and facings were assumed to be isotropic, other material properties are listed in Table 2.

Modulus (E 1) [GPa]

Fibre property

5

Voigt's model Hirsch's model at x = 0.5 Experimental Reuss's model Halpin-Tsai Shear lag model Modified shear lag model

3

2

1

0 0.05

0.1

0.15

0.2

0.25

Volume fraction (V f) Fig. 1. Micromechanics models for modulus of sisal-PP composites compared with experimental values.

2. Materials and manufacturing Sisal rope (supplied by E.C Attwood Ltd., Auckland) was chopped (fibre lengths varying between 1.5 and 7 mm) using rotary blade pelletiser and Radiata pine wood fibres in the form of saw dust were used as received. The average diameter and density of sisal fibres were 0.17 ± 0.02 mm and 1300 ± 200 kg/m3, respectively and the density of the wood fibres was 450 kg/m3. Polypropylene (Moplen HP 555G, MFI 1.3 g/10 min) was chosen as the matrix, Licocene PP MA 6452 TP copolymer was used as a coupling agent/lubricant and talc (Talc HTP 30) obtained from Unimin NZ Ltd. was used as the nucleating agent. The chopped sisal fibres and wood fibres were dry blended with talc (10% content), copolymer (1%) and PP (59%) which was extruded through a 35 mm diameter (TC35 Cincinnati Milacron) counter rotating conical twin screw extruder with a rectangular. A full factorial design of experiments based on Taguchi method was used to manufacture composites and the details of the analysis can be found in [18]. Rolls of extruded composites were obtained by passing the extruded composite sheets between calendering rolls to reduce the thickness from 2.5 mm to 1.5 mm for sisalPP composites and 0.7 mm for wood fibre-PP composites. The composite sheets were thermoformed into half-hexagonal and sinusoidal profiles between matched dies, maintained at a temperature of 155 °C using a forming rate of 500 mm/min. The profiled sheets were assembled and bonded using ultrasonic methods to form cores for sandwich panels, which were then bonded to face sheets using LoctiteÒ 401 after treating the bonding surfaces with LoctiteÒ 770. The micromechanics models listed in Table 1 were used to predict the elastic properties of sisal-PP composites. The average fibre length and diameter was 4 mm and 0.17 mm, respectively and fibre weight fractions (wf) 0.15 and 0.30 were considered in this study. The mechanical properties of the fibre and the matrix are listed in Table 2. The Poisson’s ratio of the sisal fibre has been adopted from literature [19] and the shear modulus of the sisal fibre and PP was calculated considering it to be isotropic.

3822 [20]. Sisal fibres of 5 mm were tested in a universal testing machine, Instron 5567, using 200 N load cell (serial number C 74513) at a constant crosshead speed of 0.05 mm/min. All the fibres were conditioned at the laboratory temperature, 21 °C and humidity 55% for 24 h before testing. The ultimate tensile stress of the fibre was taken to be the strength of the fibre and the Young’s modulus was measures as a slope of the chord 0.05% and 0.25% strain after applying the machine compliance correction. The mechanical properties of the composites were determined as per the standard ASTM D 638 [21] in a universal testing machine, Instron 5567 between flat grips at a crosshead speed of 5 mm/min. The tensile chord modulus of the tested specimens was measured as a slope of the chord between 0.05% and 0.25% strain, and the ultimate tensile stress was considered to be the tensile strength of the sheet. Tensile stress relaxation tests were performed at 25, 50 and 80 °C inside an Instron environmental chamber (model 3119-006) mounted on Instron 5567 universal testing machine. All the specimens were milled on a CNC milling machine conforming to the specifications of standard ASTM D 638 [21]. The specimens were conditioned at their respective temperatures before testing. A stress that was 30% of the tensile strength of each specimen at room temperature was applied and the strain was held constant for a period of 60 min. The decay in stress and the normalised stress under constant strain were plotted against time to study the relaxation behaviour of the composites. Four-point bending with two point loads equidistant from the supports was used and the flexural test on flat sandwich construction was conducted according to ASTM C 393 [22] to determine the flexural stiffness of the construction and shear strength of the core and validate the FE model.

4. Results and discussion 4.1. Micromechanics analysis

3. Mechanical tests of fibres, matrix and their composites The mechanical performance of the sisal fibres has been estimated by performing static tensile tests as per standard ASTM D

As expected, a steady increase in the tensile modulus with the increase in fibre volume fraction was displayed by the theoretical values. At lower fibre volume fractions, it can be seen from Fig. 1

2741

S. Rao et al. / Composites: Part B 43 (2012) 2738–2745

that the experimental tensile modulus values appear to be within the upper and lower bound (Voigt and Reuss) models, respectively. As most of the models are based on the rule of mixtures concept, they all seem to predict well at lower fibre volume fractions, but at higher volume fractions, most of the models under-predict the tensile moduli of the composites. However, Hirsch’s model which is a combination of Voigt and Reuss models, coupled with the correction factor x agrees well with the experimental tensile modulus at lower volume fractions. It is to be noted that Hirsch’s model reduces to Reuss’s model when the parameter x is 0 and to Voigt’s model when the parameter is 1. It is well known that the stress transfer to the fibres depends mainly on the fibre–matrix interfacial strength which in turn would influence the critical fibre length, the controlling parameter x in Hirsch’s equation may be considered as that depending on the fibre length or the fibre–matrix interfacial stresses, which in this case is 0.5 or higher, indicating a poor fibre– matrix adhesion. Similar results were obtained using shear-lag equation in Eq. (5) in Table. 1 where the tensile strength of the composite was being consistently over predicted. The difference in the experimental results and the predicted results by the shear-lag theory may be due to the discrepancy in the considered assumptions and real case. As noticed, many fibres in the composites appeared to be well dispersed within the matrix in a more or less staggered manner which undermines the assumption that the fibre–matrix are arranged as concentric cylinders and moreover due the polarity differences between the fibre and the matrix the stresses may undermine complete stress transfer to the fibre from the matrix as the theory assumes perfect fibre–matrix bonding for efficient stress transfer. As the fibres appeared well packed within the matrix, the over prediction by the models can be attributed mainly towards the poor fibre–matrix interfacial adhesion. In the modified shear-lag equation, Nairn [6] enabled the original shear-lag equation to be expressed in a generalised form with the inclusion of shear stresses as shape functions and the interfacial adhesion parameter Ds. It was seen that the predicted tensile strength values agreed well with the experimental values (Fig. 2) when the factor Ds = 1.1, and by recalling that the value of Ds = 1

implies a perfect fibre–matrix interfacial adhesion; it suggests that the low tensile strengths in the composites was mainly due to poor fibre–matrix interfacial adhesion. A listing of percentage errors in predicted and experimental values of strength and modulus can be found in Table 3. When Halpin–Tsai equations were used, the predicted result agreed well with the experimental results when the shape factor n assumed the value of 0.25, resulting in a fibre aspect ratio of 0.125 (originally 24), which similar to that of a particulate. For the fibre strength and diameter considered in this study and an interfacial shear strength 2.5 MPa [23], the critical fibre length lc = 19.7 mm. Therefore, for the fibres of tensile strength used in this study, to experience a maximum stress, they will have to be 5 times more than the length of the fibres used in the composites in this study. As the average fibre lengths are much lower than the critical fibre length, the fibre stress may have never reached the ultimate fibre strength, failing prematurely either in matrix or at the interface. The average fibre strength considered in this analysis was 578 MPa which may not be true for all the reinforcing fibres, as the values vary from 300–700 MPa [23,24]. Therefore, the maximum fibre stress that can be achieved for the length of 4 mm considered in this study will be (r = 2sl/d) 118 MPa, which is again 5 lesser than the ultimate fibre strength. However, to achieve the ultimate fibre strength for a length of 4 mm, the fibre–matrix interfacial strength should be 12.3 MPa. To check for the fibre–matrix interface, the fracture surface of the composite was viewed under a SEM in an environmental chamber. In Fig. 3, it can be seen that most of the fibres have been pulled out leaving pockets and a residue of the matrix along the fibre surface and its tip which is characteristic of a weak fibre–matrix interface. The micromechanical models indicate that the fibre strength is not being completely utilised, as a result of poor fibre–matrix interfacial adhesion and the length of the fibre. As fibre shortening is inherent to the process of extrusion, it would be more practical to increase the fibre–matrix adhesion. Noting that the experimental results of sisal-PP composites that were used for micromechanics models contained Licocene PP MA 6452 (a copolymer added to

175

150

175 Voigt's model Experimental Reuss's model Halpin-Tsai Shear lag model Modified shear lag

150

125

Stress (σ1) [MPa]

Stress (σ1) [MPa]

125

100

75

100

75

50

50

25

25

0 0.05

Voigt's model Experimental Reuss's model Halpin-Tsai Shear lag model Modified shear lag model

0.1

0.15

0.2

0.25

0 0.05

0.1

0.15

0.2

Volume fraction (Vf)

Volume fraction (Vf)

(a)

(b)

0.25

Fig. 2. Micromechanics models for strength of sisal-PP composites compared with experimental values (a) without correction (b) with correction.

2742

S. Rao et al. / Composites: Part B 43 (2012) 2738–2745

Table 3 Percentage errors in the predicted and experimental strength and modulus values. Models

Predicted values

Error (%)

r1 (MPa)

E1 (GPa) Weight fractions Experimental values Voigt’s model Reuss’s model Halpin–Tsai Cox’s shear-lag Nairn-modified shear-lag

0.15 2.9 2.9 2.7 2.9 2.8 2.8

0.30 4.1 3.3 3.0 3.3 3.1 3.1

0.15 30 93.2 33.7 34.5 56.8 32.1

r1 (MPa)

E1 (GPa) 0.30 41 161.7 38.9 40.9 92.9 41.5

0.15 – 0 6.8 0 3.4 3.4

0.30 – 19.5 26.8 19.5 24.4 24.4

0.15 – 210 12.3 15.0 89.3 7

0.30 – 294 5 0.2 127 1.2

Fig. 3. ESEM images of fracture surfaces of tensile test specimens.

improve the fibre–matrix interfacial adhesion), indicates a failure of the copolymer to improve the interfacial strength in the composite material. Obtaining an interfacial strength close to the yield stress of the matrix in case of natural fibre composites may prove to be a difficult task, in which case it would be more prudent to use inexpensive fibres of lower strengths (e.g. woodfibres) in extrusion of composite sheets such that the current interfacial strength would be adequate to utilise the total fibre strength. Therefore woodfibres obtained from local sawmills were used as reinforcement instead of sisal fibres, and the properties obtained was more or less similar to those of the sisal-PP composites, Table 4. 4.2. Stress relaxation The experimental stress relaxation data of the sisal-PP composites with the results predicted by the relaxation model is shown in Fig. 4 and the values of the integration constants are shown in Table 5. The PP and sisal-PP specimens tested at 25, 50 and 80 °C was fitted with a Maxwell element with a spring in parallel and two Maxwell’s elements with a spring in parallel, respectively. The constants for the best fit were obtained using solver function in MS Excel. It was observed that both the models provided relatively good fit onto the experimental data. The relaxation in this model is Table 4 Properties of wood-PP composites. Property

Value

Sheet density Tensile strength

qs (kg/m3) r1 (MPa) r2 (MPa)

Tensile modulus

E1 (GPa) E2 (GPa)

Poisson’s ratio

m12

1000 39.3 26.2 3.65 2.70 0.40

a combination of the Hookean springs and Newtonian dashpots (Maxwell elements), meaning that the total stress is initially experienced by the elastic springs that relax with the extension of the dashpot which is dependent on the negative exponential time function. Hence, including another Maxwell element would increase the viscous damping of the whole system; however, as no significant change in the fit was observed by the addition of a Maxwell element to the previous model, it suggests that the system is dominated by the elastic spring, meaning that the strain recovery in this material is high. This is true at lower temperatures where the r0 values are high, but as the temperature increases, there is a significant decrease in r0 and a simultaneous increase in the relaxation time constant s (Table 5), meaning that at higher temperatures permanent deformation due to the translation of the polymer chain segments is eminent but the relaxation rate in case of fibre reinforced composites is controlled by the reinforcing fibres because, at higher temperatures, the polymer matrix is at a softened state and the bonding between the fibre and matrix is expected to be weaker and especially in the composites considered in this study, the short fibres behave like polymer rich areas which fail to share the imposed load, increasing the stress relaxation of the neat polymer. The results for the PP specimens were fitted with a Maxwell element and a spring in parallel, and those for the sisal-PP composite specimens were fitted with two Maxwell elements and a spring, all in parallel (inserts in Fig. 4). Least squares method was used to obtain the best fit, and from Table 5, the sum of the residual squares reveals an excellent fit with the experimental data. It can be observed from Table 5, that the values of the constant r0 decreases monotonically with the increase in temperature, it has to be noted that the value of r0 is that of the parallel spring alone and without it, the model would relax to the value of zero in due time. Interestingly, this decrease in r0 was accompanied by a steady increase in the constants r1, r2, and s2 of the Maxwell elements

2743

S. Rao et al. / Composites: Part B 43 (2012) 2738–2745

1

1

Experimental

Experimental

0.9

0.9

Maxwell model

Maxwell model

0.8 25

0.8

25

0.7 0.6

σ/σ 0

σ /σ 0

0.7

0.6

80

50

0.5 0.4

80

0.3

0.5

0.2

0.4 0.1

0.3

0

25

50

75

100

125

150

175

0

0

40

80

120

160

Time [min]

Time [min]

(a)

(b)

200

240

280

Fig. 4. Time-stress history of experimental stress relaxation data fitted with the results of a relaxation model (a) sisal-PP fitted with a Zener model with a Maxwell element in parallel (b) two extreme temperatures of PP fitted with Zener model.

Table 5 Values of all the constants used in the stress relaxation experiment. Material and temp (°C)

r0

r1

r2

s1

s2

Sum of square errors

PP 25 PP 80 Sisal-PP 25 Sisal-PP 50 Sisal-PP 80

0.76 0.54 0.64 0.45 0.33

0.21 0.24 0.20 0.31 0.52

– – 0.14 0.24 0.36

2.48 15.58 0.70 0.70 0.35

– – 15.44 17.56 19.02

0.05 0.19 0.01 0.01 0.01

which indicated that the viscous damping of the material increased with the increase in temperature. This viscous damping remains as a permanent deformation and the elastic deformation can be recovered after the release of the load. The addition of a Maxwell element to the original model used for the PP indicates that the reinforcement of the fibres in the PP has increased the stress relaxation in the material. After the initial application of load, the model predicts an initial increase in stress, which gradually decreases and reaches a constant relaxation after 190 min at 25 °C, after 225 min at 50 °C, and after 265 min at 80 °C. The sisal-PP honeycomb cores were manufactured using matched-die thermoforming and from the results of the stress relaxation tests, the part was allowed to cool in the die. The

Profiled sheets were then cut and bonded using ultrasonic bonding methods to maintain complete recyclability and the details of the process optimisation is reported in [25]. The honeycomb cores were assembled to form sandwich panels and the shear properties of the core were obtained using FEM and flexural tests.

4.3. Finite element analysis of four-point bending The static FE analysis predicts a maximum mid-span displacement of 3.1 mm, shown in Fig. 5 and facing bending stress, Fig. 6 which was measured on the top facing between the loading points as 44.5 MPa in the middle of the panel and 45.3 MPa closer to the loading zone, which is because of stress concentrations. The experimental results reveal excellent correlation of mid-span deflection (3.8 ± 0.4 mm) and facing bending stress (44.1 ± 1.2 MPa) at 4.4 ± 0.2 kN with the FE prediction. The presence of slight variation (0.3–0.7 mm) in deflection was because of the variations in the core modulus exhibited during experimental testing of the sisal-PP composite sheets (2.17 ± 0.3 GPa), which was considered as a constant value of 2.0 GPa FE analysis. The stresses, deflection and other associated parameters of the experimental and FE predicted is shown in Table 6. The shear stress in the core from the FE analysis was estimated using the relation from [26]:

Fig. 5. Deflection contours of sandwich panel consisting of 3-ply wood veneer facings and sisal-PP honeycomb core, under 4-point flexural loading.

2744

S. Rao et al. / Composites: Part B 43 (2012) 2738–2745

Fig. 6. Facing bending stress of a sandwich panel consisting of 3-ply wood veneer facing and sisal-PP honeycomb core subjected to 4-point bending.

Table 6 Comparison of experimental and FE predicted results.

D (mm) r (MPa)

Material

s (MPa) G (MPa)

3-ply facings

Exp 3.8 ± 0.4 44.1 ± 1.2 1.4 ± 0.2 FE

1

s0xz pffiffiffi s; 3

t l

3.1

44.5

1.0

36.20 152.86

EI (N mm2) 62.97 155

Table 7 Flexural properties of sisal-PP honeycomb with 3 and 6-ply wood veneer facings – Experimental.

U (kN) 166  106 150  106

ð7Þ

where sxz is the shear stress in the core, ss is the shear stress in the cell wall material, t is the thickness of the cell wall and l is the length of the cell wall. The discrepancy in the FE and experimental shear values is due to the assumption of uniform cell diameter in the FE model which may not be true in experimental specimens as the corrugated profiles may have relaxed causing spring-back or -forward causing variations in the measured sandwich width. A mid span deflection of 3.8 ± 0.4 which is 10–38% higher than that of the FE analysis results can be seen in Table 6. These discrepancies may be because the FE model assumes a perfect bond between the core and the faceplates, which may not be true in the real case, as it was observed during experimentation that the failure of the sandwich panel was mainly due to the failure of the adhesive between the core and the faceplates. Hence, the presence of flaws and the discrepancies in bond strength within the core and between the core and face plates (even before testing) may have contributed towards the increased mid-span deflections during experimentation. Moreover, the total deflection of the sandwich panel is due to the combination of bending of the panel and shear of the core, and when total deflection increases with the same span length, the shear contribution towards the total deflection will increase. In sandwich panels it is assumed that the facings carry all the bending stress and the core carry all the shear stresses. In the experimental values, the contribution from the bending of the panel towards total bending is 57% and the rest 43% is due to the shear of the core, in the FE model the contribution of bending towards total deflection is 80% and that of shear is 20%, meaning that cores being stiffer than the experimental cores may carry some of the bending load too. From simple beam theory, with larger loading span and high span length to sandwich depth ratio (>20), the contribution of shear towards the total bending becomes minimum and is usually neglected. However, in this case, the span length to depth ratio is 8, meaning that the shear contribution towards total deflection will prevail. In the experiment, the core shear stresses evaluated was an underestimate as the cores failed with the facings yielding in compression rather than failing in shear. Hence, the 3-ply wood veneer was replaced with 6-ply wood veneer facings to increase the stiffness to withstand buckling due to compressive stresses at the top

Panel property

Facing (6-ply)

Facing (3-ply)

Cell diameter (mm) Core density (kg/m3) Core thickness (mm) Sandwich width (mm) Maximum deflection d (mm) Panel shear rigidity U (N)

12 160 30 (±0.15) 60 3.45 (±0.5) 116.97 (±11.69)  103 1.45 (±0.10) 271.86  106

12 160 25 (±0.25) 61 4.90 (±1.04) 28.74  103 (±7)  103 1.31 (±0.17) 176.92  106

30.42 (±2.16) 64.2 (±21.52)

31.93 (±4.16) 16.40 (±4.16)

Core shear strength s (MPa) Panel bending stiffness D (N mm2) Facing bending stress rf (MPa) Core shear modulus Gc (MPa)

Table 8 Specific shear strength of honeycombs. Material, cell size (mm)

Density (kg/m3)

s/q(Nm/kg)

Polypropylene core, 6 Aluminium (5056), 10 Nomex (phenolic), 10 Sisal-PP hexagonal core, 12 Sisal-PP sinusoidal cores, 12

49 (±12.0) 37 (±3.0) 48 (±3.0) 151 (±7.6) 145 (±3.2)

8  103 32  103 27  103 10  103 12  103

faces. Though the compressive failure in the faceplate was eliminated by increasing their stiffness the increase in shear strength of the core was marginal (Table 7) and the failure in all the hexagonal core panels tested was initiated due to de-bonding at the interface. Although the experimental shear strength values were underestimates, the specific shear strength even at such lower values exceeded that of the unreinforced ones by at least two fold, Table 8. 5. Conclusion The micromechanical models indicate that the fibre strength is not being completely utilised, as a result of poor fibre–matrix interfacial adhesion and the length of the fibre. Therefore inexpensive fibres could be used as reinforcement instead of sisal fibres, and the properties obtained by replacing sisal fibres with sawdust was more or less similar to those of the sisal-PP composites. The stress relaxation experiments suggested a need to cool the formed part to room temperature within the die under suitable forming loads to avoid dimensional instability due to warping. In this study, it was seen that the stresses at 80 °C and 50 °C relax dramatically; meaning that, if the formed material is extracted from the die at 80 °C or 50 °C during the cool-down time, there is a possibility of part dimensional instability due to warping of the material because the stresses in the material are still relaxing at those

S. Rao et al. / Composites: Part B 43 (2012) 2738–2745

temperatures. Hence, for better shape conformance with the die, it is essential to have the (originally) flat sheet in its deformed state under suitable forming load until the material reaches room temperature. The stress relaxation experiments also provide valuable information about the component behaviour at the application stage of the composite product. If the ‘frozen-in’ residual stresses in the formed component were to start relaxing at service temperatures, the material would tend to lose its formed shape which in essence would affect the structural integrity of the component. Higher mid-span deflection was observed in the experimental specimens compared to the FE deflections which may be due to the assumption of uniform honeycomb cell diameters throughout the core. Although sisal fibre reinforced composites are linear in the elastic region, the structure as a whole behaves non-linearly. As the failure in the sandwich panels was predominantly due to faceplate delamination and some local buckling under the loading points, a non-linear analysis in the view to investigate the failure may be essential. The shear strength of the core obtained during flexural testing was an underestimate as the failure in the sandwich panels was predominantly due to faceplate de-bonding from the core at their interface. However, the underestimated shear strength at these levels was more than twice than those of the unreinforced cores, generating potential to be used in structural panels. Acknowledgements The authors would like to thank BioPolymer network and Foundation for Research Science and Technology (FRST), New Zealand for their financial support in this research. References [1] Holbery J, Houston D. automotive applications. J [2] Pritchard G. Plants move 2007;9(4):40–3. [3] Piggott MR. Load bearing Press; 1980.

Natural-fiber-reinforced polymer composites in Miner Met Mater Soc 2006;58(11):80–6. up the reinforcement agenda. Plast Addit Compd fibre composites. Willowdale, Ontario: Pergamon

2745

[4] Halpin JC, Kardos JL. The Halpin–Tsai equations: a review. Polym Eng Sci 1976;16(5):344–52. [5] Cox HL. The elasticity and strength of paper and other fibrous materials. Br J Appl Phys 1952;3(3):72–9. [6] Nairn JA. On the use of shear-lag methods for analysis of stress transfer in unidirectional composites. Mech Mater 1997;26(2):63–80. [7] Jones RM. Mechanics of Composite Materials. Philadelphia, PA: Taylor & Francis, Inc.; 1999. [8] Mallick PK. Fibre-reinforced composites materials: manufacturing and design. New York: Marcel Dekker, Inc.; 1998. [9] Kalaprasad G, Joseph K, Thomas S, Pavithran C. Theoretical modelling of tensile properties of short sisal fibre-reinforced low-density polyethylene composites. J Mater Sci 1997;32:4261–7. [10] Andersons J, Sparnins E, Joffe R, Wallström L. Strength distribution of elementary flax fibres. Compos Sci Technol 2005;65(3–4):693–702. [11] Facca AG, Kortschot MT, Yan N. Predicting the elastic modulus of natural fibre reinforced thermoplastics. Compos A Appl Sci Manuf 2006;37(10):1660–71. [12] Bitzer T. Honeycomb technology: materials, design, manufacturing, applications and testing. UK: Chapman & Hall; 1997. [13] Court C, Castorina L. Continuous process for forming structure suitable for use as a core member, WO 99/41061, United States; 1999. [14] Lhommeau RE, OrgeSavigny S, Ducury R. Apparatus for upwardly extruding and cooling a thermoplastic resin multiple tube structure, 4478563, France; 1984. [15] Gerhard KH, Baumbach R. Methods and device for joining sections of thermoplastic continuous web material, 6907912, USA; 2005. [16] Barnes JA, Byerly GE. The formation of residual-stresses in laminated thermoplastic composites. Compos Sci Technol 1994;51(4):479–94. [17] Findley WN, Lai JS, Onaran K. Creep relaxation of nonlinear viscoelastic materials amsterdam. North-Holland: Publishing Company; 1976. [18] Rao S, Bhattacharyya D, Jayaraman K, Fernyhough A. Manufacturing and recycling of sisal-polypropylene composites. Polym Polym Compos 2009;17:467–79. [19] Ntenga R, Be´akou A, Ate´ba JA, Ohandja LA. Estimation of the elastic anisotropy of sisal fibres by an inverse method. J Mater Sci 2008;43:6206–13. [20] ASTM D 3822 standard test method for tensile properties of single textile fibers; 2007. [21] ASTM D 638 standard test method for tensile properties of plastics; 2007. [22] ASTM C 393 standard test method for core shear properties of sandwich constructions by beam flexure; 2006. [23] Yuan X, Jayaraman K, Bhattacharyya D. Plasma treatment of sisal fibres and its effects on tensile strength and interfacial bonding. J Adhes Sci Technol 2002;16(6):703–27. [24] Li Y, Mai Y-W, Ye L. Sisal fibre and its composites: a review of recent developments. Compos Sci Technol 2000;60(11):2037–55. [25] Rao S, Coombs B, Ultrasonic bonding of thermoplastic honeycomb cores: A statistical approach in parameter selection. In: Processing and fabrication of advanced materials (PFAM-19), Auckland, New Zealand; 2011. [26] Banerjee S, Battley M, Bhattacharyya D. Shear strength optimisation of reinforced honeycomb materials. Mech Adv Mater Struct 2010;17(7):542–52.