Jtam-A4.dvi JOURNAL OF THEORETICAL AND APPLIED MECHANICS 52, 2, pp. 431-442, Warsaw 2014 NUMERICAL SIMULATION AND EXPERIMENTAL BENDING BEHAVIOUR OF MULTI-LAYER SANDWICH STRUCTURES Jamal Arbaoui Laboratoire de Physique des Milieux Denses, Metz, France; e-mail: jamal.arbaoui@ensta-bretagne.fr Yves Schmitt IUT de Thionville – Yutz, Yutz, France and P.A. Technologies, Thionville, France J-Luc Pierrot Laboratoire de Physique des Milieux Denses, Metz, France François-Xavier Royer Laboratoire de Physique des Milieux Denses, Metz, France and IUT de Thionville – Yutz, Yutz, France In this paper, an experimental investigation, an analytical analysis and a numericalmodel of a typical four-point bending test on a polypropylene honeycombmulti-layer sandwich panel areproposed.Thepolypropylenehoneycombcore ismodelledasa single solidandmulti-layer of equivalentmaterial properties. Analytical and numerical (finite element) homogenization approaches are used to compute the effective properties of the single honeycomb core and analytical homogenization of the multi-layer one. The results obtained by numerical simu- lation (finite element) of four-point bending are compared with the experimental results of a polypropylene honeycomb core/composite facing multi-layer sandwich structures. Keywords: multi-layer sandwich, polypropylene honeycomb,modelling, bending 1. Introduction Owing to theirmerits of a high strength/weight ratio, heat resistance, sound insulation and easy assembly, sandwich structures have beenwidely used in aerospace, automotive and construction industries (Yu and Cleghorn, 2005; Wang and Yang, 2000; Kim and Hwang, 2000). A typical sandwich panel is composed of three layers, in which two thin sheets (faces) of a stiff and strong material are separated by a thick core of low-density materials (Allen, 1961). Considering the very varied use of these materials in numerous fields, it is essential to know their mechanical properties in order to predict and calculate their behaviour in specific and diverse environments. One thus finds the faces possessing particular mechanical characteristics and the honeycomb core being able to have different specificmechanical properties. The assembly of these two parts is carried out by joining, welding or brazing with another material of different behaviour. The aim of this work concerning the research subject “Modelling of Composite Multi-layer Sandwichs” is to model the bending behaviour of sandwich structures. The main steps in this study are: i) the determination of elastic constants of the material by analytical and numerical homogenization, and ii) the comparison of the results obtained by numericalmodelling with the experimental data recorded for a polypropylene honeycomb core/composite facing multi-layer sandwich structure. 2. Sandwich material 2.1. Mechanical properties A typical sandwich panel consists of two thin faces with a thickness t, separated by a lightweight core of thickness hc, as illustrated in Fig. 1. The overall depth and width of the 432 J. Arbaoui et al. panel are h and b, respectively. The faces are typically bonded to the core to provide a load transfer mechanism between themain components of the sandwich panel. Fig. 1. A structure of a sandwich composite The flexural rigidity D of a sandwich beam is the sum of flexural rigidities of the faces and the core measured with respect to the centroidal axis of the entire section. It can be expressed as D= Efbt 3 6 + Efbtd 2 2 + Ecbh 3 c 12 =2Df +D0+Dc (2.1) where Ef and Ec are Young’s moduli of the face sheet and core, respectively, and d= t+hc. Df is the bending stiffness of the face sheet about its own neutral axis, D0 – stiffness of the face sheets associated with bending about the neutral axis of the entire sandwich, and Dc – stiffness of the core (Allen, 1961). Since the core is stiff in shear but generally soft, its Young’s modulus is much smaller than that of the face sheet. By assuming Ec ≪Ef and the face sheets are thin, the expression for D becomes D≈Ef b(h3−h3c) 12 (2.2) The shear stiffness Q is given by the following equation Q=Gc b(h− t)2 hc (2.3) The face stress is defined such that σf = P(L2−L1) 2tbd (2.4) In the core, the shear stress is given as τc = P 2bd (2.5) The elastic deflection wt for a sandwich beam at loading points (L2−L1)/2 is the sum of the flexural and shear deflections for a four-point bending (Fig. 2) wt =w1+w2 = P(L2−L1) 2(L2+2L1) 24D + P(L2−L1) 2S (2.6) Fig. 2. The four-point bending test Numerical simulation and experimental bending... 433 3. Materials and experimental method 3.1. Materials The sandwich panels used in this study consist of three main parts (Fig. 3): • Two face sheets of composite glass fibres (T800/M300)/polyester resin with the nominal face thickness of 1mm • A honeycomb polypropylene core • Intermediate layers of composite glass fibresM450/polyster resin with the nominal inter- mediate layers thickness of 0.05mm. The composite structures are pressed in only one pass. The mechanical properties of the basic materials are given in Tables 1 and 2. Table 1.Mechanical properties of a polypropylene honeycomb core Properties Polypropylene honeycomb core Density [kg/m3] 80 Shear modulus [MPa] 8 Shear strength [MPa] 0.5 Elastic modulus [MPa] 15 Table 2.Mechanical properties of polyester resin/glass fibresT800/M300 andM450 composites Properties T800/M300 composite M450 composite Young’s modulus [MPa] 9162 5500 Tensile strength [MPa] 321 200 Shear modulus [MPa] 2101 2115 Thickness [mm] 1 0.05 Poisson’s ratio 0.3 0.3 3.2. Experimental method The tests were carried out using a four-point bending testing fixture device shown in Fig. 4. The device, especially designed for such tests, was connected to a servo-hydraulic universal testing machine INSTRON 4302 controlled by an INSTRON electronic unit. These tests were performedwith respect to theNFT54-606 norm.To check the reproducibility of the results, five beams by composite type were tested. The crosshead displacement rate was 3mm/min. The sample dimensions are grouped in Table 3. Table 3. Specimen dimensions Specimen b [mm] h [mm] hc [mm] L1 [mm] L2 [mm] L [mm] Single core 35 22 20 120 300 440 Double core 35 22.05 20.05 120 300 440 Triple core 35 22.1 20.1 120 300 440 Quadruple core 35 22.15 20.15 120 300 440 434 J. Arbaoui et al. Fig. 3. Honeycombmulti-layer sandwich Fig. 4. Static four-point bending test 4. Experimental results Figure 5 depicts the load-displacement curve for multi-layer honeycomb composite structures solicited in four point bending. The bending behaviour is similar and can be described in three principal phases: the first phase is initial linear elastic behaviour followed by aphase of nonlinear one in which themaximum loading is reached. In the last phase, a reduction in the load applied is observed till the total rupture of the samples. The linear behaviour corresponds to the work of the skins in traction and compression, whereas the nonlinear behaviour mainly depends on the core properties under the effect of the shear stress. This figure shows also an increase of the mechanical properties of facing stress, core shear stress andbending stiffness by about 50, 51 and 36 percent, respectively, as the number of layers increases from single to quadruple layers. The assessed mechanical properties of these composite multi-layers sandwiches are given in Table 4. Table 4.Mechanical properties of the multilayer sandwich structures Specimen Load Facing stress Core shear stress Bending stiffness [N] [MPa] [MPa] [N·mm2] Single core 520 63 0.36 706 ·105 Double core 736 90 0.50 711 ·105 Triple core 931 113 0.63 714 ·105 Quadruple core 1065 129 0.72 718 ·105 Numerical simulation and experimental bending... 435 Fig. 5. Typical load/displacement curves for multilayer sandwiches 5. Effective properties of the single honeycomb core 5.1. Analytical homogenization approach Thedevelopment of constitutivematerialmodels for honeycombmaterials is complicated due tohighlyanisotropicpropertiesof thematerial.Computationally efficientmodellingmethodsand constitutive laws are required to reduce time and whilst being accurate enough to realistically represent the overall structural behaviour. The analytical expressions used to determine the effective elastic properties of the cellular hexagonal honeycomb core are based on the works of Gibson and Asby (1997), Masters and Evans (1996), Grédiac (1993), Shi and Tong (1995), Becker (1998), Xu and Qiao (2002), Meraghni et al. (1999). Appropriate expressions are given in Appendix A. Theelemental beamtheoryhasbeenadopted (Fig. 6) for each component inside theunit-cell’ to arrive at different expressions for effective properties employing the strain energy concept. The length of the diagonal and vertical struts including the angle as well as their thickness have been kept as variable. The presented analytical approach is simple and computes the effective properties in a fraction of the time that is required for FE analysis with a minimum change in the input file. The proper implementation of this method embedded in large quasi-static or dynamic simulations (where a part of the structure could be modelled with a detailed finite element mesh and the rest could be modelled with a single solid layer of equivalent material properties) would give high computational advantage, which is essential in large-scale modelling and simulation environment. Fig. 6. Deformationmode of a honeycomb structure according to Gibson and Ashby (1997) 436 J. Arbaoui et al. 5.2. Numerical homogenization approach Theaim is todetermine the elastic propertiesby anumerical homogenizationmethodapplied to aRepresentative Volume Element of the honeycomb to compare the results with those achie- ved analytically. TheRepresentative VolumeElement (RVE) (Fig. 7) consists in 40 cells meshed with plate finite elements with 4 nodes and 6 degrees of freedom per node. Every foil contains 12 elements: 4 according to height and 3 to length. To estimate the various elastic moduli, a displacement is imposed on the face of the RVE in a given direction while the opposite face is being fixed. Symmetries are taken into account by using the appropriate boundary conditions. Nine simulations are necessary to determine nine elastic constants of the single honeycomb. Fi- gure 8 presents an example of the tensile simulation along the direction i (i= x,y,z) which is used to determine the three elasticity moduli E1,E2 and E3 and then six Poisson’s ratios. The finite element results are depicted in Table 5. Fig. 7. Representative Volume Element “RVE In Fig. 7 lx, ly and lz are the lengths of the Representative Volume Element, with: lx = 64.7mm, ly = 36mm, lz = 10mm. The mechanical properties of the honeycomb are related to its geometrical characteristics which are (Fig. 6): c= a= 4.6188mm, e= 0.24mm, l=10mm, φ=30◦. Fig. 8. RVEwith the imposed displacement in the X, Y and Z directions The comparison between the results of Gibson’s analytical model and those obtained by the numerical simulation made it possible to better determine the values of the elasticity modu- lus. One can note that the variation of the results between the numerical simulation and the analytical model is approximately 7.8% for E1 and 5% for Poisson’s ratio, which is relatively Numerical simulation and experimental bending... 437 Table 5.Mechanical properties of a single polypropylene honeycomb core Polypropylene Finite element Gibson and Ashby honeycomb code (Ansys) (1997) E1 [MPa] 0.448 0.486 E2 [MPa] 0.545 0.486 E3 [MPa] 96.3 90 ν12 1.05 1 ν13 0.002 0.002 ν23 0.002 0.002 ν31 0.4 0.4 ν32 0.4 0.4 G12 [MPa] 0.1292 0.1214 G21 [MPa] 0.0664 G23 [MPa] 16.93 16.44 G32 [MPa] 1.2698 G13 [MPa] 17.11 G13min [MPa] 24.673 G13max [MPa] 27.415 G31 [MPa] 1 acceptable by takingGibson’smodel as the reference. In the case of traction in the Y direction, the variation of the results between the two models is approximately 12% and 25% for E2 and Poisson’s ratio, respectively. For displacement in the Z direction, the variation is approximately 7% for E3.On the other hand, the variation ofPoisson’s ratio is veryweak.Concerning the shear modulus G12, the variation of the results between Ansys and Gibson are approximately 6.4%. This relatively important error weakly affects the modulus, therefore it has little influence on the sandwich mechanical properties. The shear moduli G21,G31 and G32 are obtained only by the numerical simulation. Gibson does not give comparative values. 6. Effective properties of the multi-layer honeycomb core Themechanical characteristics of theM450 intermediate layer (isotropic compositematerial) are presented in Table 2. The analytical expressions used to assess the effective elastic properties of the multi-layer honeycomb core are determined by the following equation X = 1 h n ∑ i=1 Xihi (6.1) Themechanical properties of these composite multi-layer sandwiches are listed in Table 6. 7. Effective properties of the composite T800/M300 face The composites T800/M300 used in this study are made of fibers well-balanced and oriented along two perpendicular directions: one is called the warp and the other the weft direction. For an approximation of the elastic properties of the fabrics, one can consider them to consist of two plies of the unidirectionals crossing at 90◦ angles with each other. One can then use the following notation: n1 – number of warp yarns per meter, n2 – number of fill yarns per meter 438 J. Arbaoui et al. Table 6.Mechanical characteristics of the multi-layer core Multilayer cores Double core Triple core Quadruple core E1 [MPa] 14.2 27.84 41.42 E2 [MPa] 14.2 27.84 41.42 E3 [MPa] 103.49 116.9 130.27 ν12 1 1 1 ν13 0.002 0.003 0.002 ν23 0.002 0.003 0.002 G12 [MPa] 5.39 10.64 15.86 G23 [MPa] 10.54 26.84 32.02 G13 [MPa] 29.81 35 40.16 k= n1 n1+n2 = 1 2 and fiber modulus: Ef′ = 73000MPa, Gf = 30000MPa, νf = 0.25; resin modolus: Em =4000MPa, Gm =1400MPa, νm =0.4; fiber volume fraction Vf =28%. One can use the following relations to characterize the unidirectional ply: —modulus of elasticity along the direction of the fiber El =Ef′Vf +EmVm (7.1) —modulus of elasticity in the direction transverse to the fiber Et =Em ( 1 1−Vf + Em Ef Vf ) (7.2) — shear modulus Glt =Gm ( 1 1−Vf + Gm Gflt Vf ) (7.3) —Poisson’s ratio νlt = νfVf +νmVm (7.4) This two plies of the T800/M300 composite can be considered together. The fabric layer is replaced by a single orthotropic one. One can therefore obtain the mechanical characteri- stics of the T800/M300 which are determined by the formulas presented in Table 7 (Gay, 1997; Berthelot, 1997). Once the honeycomb core and T800/M300 composite are homogenized, the whole sandwich panel is likened to a beam constituting of three elastic layers: orthotro- pic/orthotropic/orthotropic that will be used in the numerical model described below. 8. Numerical simulation results Finite element calculations are also performed on CASTEM 2008. The honeycomb sandwich structure is modelled using 3D solid (eight nodes and six DOFs per node) elements (Mindlin, 1997). For symmetry reasons, only a quarter of the sandwich panel (Fig. 9) is considered in the present analysis. The applied boundary conditions are as follows: at the level of the support, the transversal displacement Uz is fixed to zero; at the symmetry level on face 1, the in-plane Numerical simulation and experimental bending... 439 Table 7.Three-dimensional elastic properties of composite T800/M300 E1 [MPa] E1 = kEl+(1−k)Et 9135 E2 [MPa] E2 =(1−k)El+kEt 9135 E3 [MPa] E3 =Ematrix 4000 ν12 ν12 = νlt k+(1−k) El Et 0.2 ν13 ν13 = Et 2G13 −1 0.3 ν23 ν23 = ν13 0.3 G12 [MPa] G12 =Glt =Gm ( 1 1−Vf+ Gm Gf Vf ) 1616 G23 [MPa] G23 =Gm+ GmVf Gm Gf −Gm + Km+2Gm 2Km+2Gm (1−Vf) 1769 G13 [MPa] G13 =Gm+ GmVf Gm Gf −Gm + Km+2Gm 2Km+2Gm (1−Vf) 1769 Avec Km = Em 2(1−2νm) Fig. 9. Modelling of four points on bending CASTEM 2008 displacement Ux and the rotations hy and hz are fixed to zero as well; then on face 2, the in-plane displacement Uy and the rotation hx and hz are likewise zero. The sandwich plate is composed of the composite face andmultilayer honeycomb corewhose dimensions are: length L = 440mm, width b = 35mm, core thickness hc = 20mm, face thickness t=1mm and intermediate layer thickness t′ =0.05mm. Prior to initiating the evaluation study, an analysis of mesh convergence is carried out to ensure the accuracy of the proposed finite element solution since it is considered in the present study as the reference. The convergence was achieved with 4200 elements: 20 elements following the x-axis, 15 elements in the thickness of the core, 3 elements in the thickness of each skin and 10 elements following the y-axis (Fig. 10). Figures 11a and 11b show the evolution of the load versus the displacement of tow sandwich materials with a single and double core, respectively. These figures compare the bending pro- perties obtained experimentally with that obtained by the 3D finite element model. The values 440 J. Arbaoui et al. Fig. 10. Finite-element mesh of the sandwich structure indicate a good prediction of the bending properties with amaximumdifference of 7% between the numerical prediction and experimental results. This gap is very reasonable by taking into account the systematic defects of the manufacturing process, in particular the air bubbles and uncertainties of the used devices. Fig. 11. Comparison of experimental and numerical bending results of the single (a) and double (b) honeycomb 9. Conclusion Experimental and numerical modelling investigations of a typical four-point bending test of single and multi-layer honeycomb sandwich structures have been performed. Analytical and numerical (FE) homogenization approaches have been used to compute the effective properties of the single andmulti-layer honeycomb core as well as the T800/M300 composite face. The correlation between the effective properties of the single honeycomb core obtained by the analytical and numerical modelling is in good agreement. Compared to the lower bound of Gibson, the shearmodulusG13 is low. This can be improved by coupling node cells of the single honeycomb core. The comparison between the experimental results and those obtained by the numerical si- mulation of the sandwich structures in honeycomb polypropylene shows a slight difference. This variation remains very reasonable by taking into account the systematic defects of themanufac- turing process, in particular the air bubbles and uncertainties of the used devices. Acknowledgments The authors would like to gratefully acknowledge P.A. Technologies for instrumenting the polypro- pylene honeycomb. This work was supported by the Departement de la Recherche et de l’Enseignement Suprieur of REGIONLORRAINE, France. Numerical simulation and experimental bending... 441 Appendix A Table 8.The effective elastic properties of honeycomb Linear elasticity The effective elastic properties of honeycomb Tensile modulus in theX1 direction E1 =Es (e l )3 cosφ (1+sinφ)sin2φ Tensile modulus in theX2 direction E2 =Es (e l )31+sinφ cos3φ Shear modulus in the (X1,X2) plane G12 =Es (e l )31+sinφ 3cosφ Poisson’s ratio in the (X1,X2) plane ν21 = (1+sinφ)sinφ cos2φ Poisson’s ratio in the (X1,X2) plane ν12 = cos2φ (1+sinφ)sinφ Tensile modulus in theX3 direction E3 =Es (e l ) e l +2 2(1+sinφ)cosφ Transverse shear G23 =Gs 1+2sin2θ 2cosφ(1+sinφ) e l Transverse shear G13 =Gs cosφ 1+sinφ e l Poisson’s ratio ν32 = ν31 = ν (solid material) ν13 = E1 E3 ν31 ν23 = E2 E3 ν32 References 1. AllenH.G., 1961,Analysis and Design of Structural Sandwich Panels, PergamonPress, London, U.K. 2. BeckerW., 1998,The in-plane stiffness of honeycombcore including the thickness effect,Archives of Applied Mechanics, 68, 334-341 3. Berthelot J.M., 1996, Composite Materials: Mechanical Behavior and Structural Analysis (in French), 4th edition 4. Gay D., 1997,Composite Materials (in French), Hermes edition 5. Gibson L.J., Ashby M.F., 1997,Cellular Solids: Structure and Properties, 2nd ed., Cambridge University Press, Cambridge, U.K. 6. GrédiacM., 1993,Afinite element studyof the transverse shear inhoneycombcores, International Journal of Solids and Structures, 30, 13, 1777-1788 7. Kim H.Y., Hwang W., 2002, Effect of debonding on natural frequencies and frequency response functions of honeycomb sandwich beams,Composite Structures, 55, 1, 51-62 8. Masters I.G., Evans K.E., 1996,Models for the elastic deformation of honeycombs,Composite Structures, 35, 403-442 442 J. Arbaoui et al. 9. Meraghni F., Desrumaux F., Benzeggagh M.L., 1999,Mechanical behaviour of cellular core for structural sandwich panels,Composites: Part A, 30, 767-779 10. Mindlin R.D., 1997, Influence of rotatory inertia and shear in flexural motions of elastic plates, Journal of Applied Mechanics, 18, 31-38 11. Shi G., Tong P., 1995, Equivalent transverse shear stiffness of honeycomb cores, International Journal of Solids and Structures, 32, 10, 1383-1393 12. Wang B., Yang M., 2000, Damping of honeycomb sandwich beams, Journal of Materials Pro- cessing Technology, 105, 1/2, 67-72 13. XuF.X., Qiao P., 2002,Homogenized elastic properties of honeycomb sandwichwith skin effect, International Journal of Solids and Structures, 39, 2153-2188 14. YuS.D.,CleghornW.L., 2005,Free flexural vibration analysis of symmetric honeycombpanels, Journal of Sound and Vibration, 284, 1/2, 189-204 Manuscript received June 10, 2013; accepted for print September 6, 2013