Jtam.dvi JOURNAL OF THEORETICAL AND APPLIED MECHANICS 47, 4, pp. 737-750, Warsaw 2009 MICROMECHANICAL MODEL OF AUXETIC CELLULAR MATERIALS Małgorzata Janus-Michalska Cracow University of Technology, Institute of Structural Mechanics, Cracow, Poland e-mail: mjm@limba.wil.pk.edu.pl An effective anisotropic continuum formulation for auxetic cellularmaterials is the objective of this paper. A skeleton ismodelled as a plane beam elastic structure with stiff joints. The skeleton topology, forming concave polygons, is responsible for negative Poisson’s ratio effect. The essential macroscopic features ofmechanical behaviour are inferred from the deformation response of a representative volume element using the framework ofmicromechanical analysis. The strain energy of a unit cell is calculated by adding the tensile, shearing and bending strain energy of individual members. The equivalent continuum is based on averaging this energy, thus formulating the basis for computing the anisotropic stiffness matrix. The structural mechanics methodology and ANSYS finite element code are applied to solve the beam model of the skeleton.Graphical representationof certainmaterial constants such as Young’s modulus, Poisson’s ratio, shear modulus and generalized bulk modulus is given. The results of included parametric study may be used for proper choice of geometric and material data of the skeleton for a given structural application of the anisotropic continuum. Key words: auxetic cellular materials, anisotropy, effective model, elasticity 1. Introduction Auxetic materials are of interest because of enhanced material properties re- lated to negative Poisson’s ratio. Such materials are called also dilatational materials because they exhibit substantial volumetric changes when loaded. First auxetic cellular structures were created as 2D silicone rubber or alu- minum honeycombs and were extensively investigated by Lakes (1991-1993), Lakes et al. (1988), Lakes andWitt (2002). Materials with negative Poisson’s ratio aremore resilient than non-auxeticmaterials, and the linear strain-stress relationship can reach up to 0.40 compressive ultimate strain compared to 0.05 strain for conventional cellular materials. Auxetic materials also exhibit lower 738 M. Janus-Michalska stiffness and greater resistance to indentations than ordinary materials. This propertymay beuseful in certain applications such asmatresses andwrestling mats. Auxetic materials better redistribute strain under external loads. Ne- gative Poisson’s ratiomay significantly influence stress distribution in contact problems, reduce stress concentrations, substantially affect the Saint Venant effect and yield double curvature effect in 3D bending problems. This no- vel behaviour is important in design considerations. An increasing number of materials and processing routines are developed and tailored to specific appli- cations. Auxetic behaviour can be described by classical theory of elasticity and it does not depend on scale (Lakes, 1991b). Deformation takes place on the micro or macro level. This means that one may consider auxetic matrials as well as auxetic structures. Physical origin of mechanical behaviour of cellular materials is based on structural considerations. Structural analysis of a skeleton on the micro level explains macroscopic behaviour of such structured bodies. The overall effecti- ve properties are determined by considerations using transition betweenmicro andmacro scales of observation. This corresponds to effectivemodel construc- tion (Hori andNemet-Naser, 1999;Nemat-Naser andHori, 1999). The effective properties are then used to determine the response of structural elements on the macro scale and emerge naturally as a consequence of micro-macro rela- tionswithoutdependingon specificphysicalmeasurements on themacroscopic level (Nemat-Naser andHori, 1999). Thismethod, typical formicromechanics, has been applied to aluminum foams Janus-Michalska and Pęcherski (2003), and a group of 3D cellular structures forming positive Poisson’s ratio cellular materials Janus-Michalska (2005). The existing solutions concerning stiffness matrices for auxetic cellularmaterials aremainly based on the homogenization approach (Kumar andMcDowell, 2004). Themost recent solution for auxetic structures is the missing ribs method by Smith et al. (2000). The approach based onmicromechanicalmodelling presented in this paper is advantageous because the dependence of effective properties onmicrostruc- tural parameters is easily readable here.Thus structural topology and skeleton material may be properly chosen in order to obtain the required properties of the cellular material. 2. Micromechanical analysis A micromechanical framework similar to that proposed in Janus-Michalska and Pęcherski (2003), Janus-Michalska (2005) has been developed to deter- mine equivalent continuum properties. The internal structure of a material is Micromechanical model of auxetic cellular materials 739 assumed to be periodical one, and is the basis for studying overall mechanical properties of a cellular auxetic material. 2.1. Representative unit cell A theoretical investigation is conducted for a material exhibiting two the dimensional reentrant hexagonal microstructure as shown in Fig.1a. Fig. 1. A cellular structure and a representative unit cell Formulation on the micro level begins by identifying the unit cell of the spatially periodic array.The idea adopted here is based on the fact that the es- sential feature of arbitraryuniformdeformation of thewhole beamstructure is affinityofmidpointdisplacements (WarenandKraynik, 1988; Janus-Michalska andPęcherski, 2003). It results in the same structural response of parts of the skeleton consisting of the node and halves of beams. The representative unit cell defined as the smallest part of the skeleton with repetitive response was successively applied to foams (Janus-Michalska and Pęcherski, 2003), and 3D periodic cellular structures (Janus-Michalska, 2005). The detailed description of unit cell construction is given in Janus-Michalska (2005). The cell corresponding to the reentrant hexagonal structure is given in Fig.1b. The skeleton of the cell is modelled as an elastic beam structure with a stiff joint (vertex 0). The strut midpoints are described by the position vectors b 0 1 = ( 0, h 2 ) b 0 2 = (L 2 sinγ, L 2 cosγ ) b 0 3 = ( − L 2 sinγ, L 2 cosγ ) where: |b01|= h/2, |b02|=L/2, |b03|=L/2. Beam lengths and γ angle should satisfy the following conditions, which are necessary to obtain the representa- tive volume element (RVE) shown by the dotted line in Fig.1a: Lcosγ ¬h/2 and cosγ¬L/h. 740 M. Janus-Michalska The thickness of RVE is assumed to be H =1 in the direction perpendi- cular to the (x,y) plane. The volume of the unit cell is V = [(h−Lcosγ)Lsinγ]H Three surfaces of RVE perpendicular to struts i, where i = 1,2,3 are considered for further calculations, and they are computed as follows A1 = (L−hcosγ sinγ ) H A2 =A3 = (h−Lcosγ sinγ ) H The part of the skeleton included in RVE consists of beams of rectangular cross-sections As = tH, where t denotes beam thickness. The skeleton vo- lume is thus determined as: Vs = (L+ h/2)tH. The cellular material may be characterised by its relative density: ρ∗ = ρ/ρs, where ρ and ρs denote the cellular material and skeletonmaterial densities, respectively. The relative density may be expressed by relation ρ∗ = ( L+ h 2 ) t (h−Lcosγ)Lsinγ The structure is specified by four geometric structural parameters: L, h, t, γ. Three material parameters: Rse – yield stress, Es – Young’s modulus, νs – Poisson’s ratio must be given to describe the skeleton material. On the basis of the unit cell model, effective properties of cellular material can be derived. 2.2. Uniform strains for linear elasticity Uniformplain deformation of a solid with repetitivemicrostructure results in displacement affinity (Janus-Michalska, 2005). Periodic skeleton structure requires that the individual beams deformantisymmetrically about theirmid- points, so the resultant forces at eachmidpoint reduce to axial and transversal force (Janus-Michalska, 2005) (resultant bendingmoment at themidpoint va- nishes). Inmicromechanics, strains are defined as volumetric averages of themicro field variables (Nemat-Naser and Hori, 1999) and read as follows ε= 〈εs〉V = 1 V ∑ Ai sym(ni⊗ui) dS (2.1) where: 〈·〉V stands for the volumetric average in the skeleton s taken over V , ni is the outer unit normal on the boundary ni = b 0 1/|b0i |, ui – themidpoint displacement on the surface Ai. Micromechanical model of auxetic cellular materials 741 Three independent uniform unit deformations Kε̃, described by the follo- wing strain vector components (Kelvin’s notation in 6-D space) are considered 1ε̃= εx =1 2ε̃= εy =1 3ε̃= √ 2εxy =1 (2.2) This results in the following midpoint displacements: — for uniaxial extensions ∆i( 1ε̃)= 1ε̃(b0i ·ex)ex ∆i(2ε̃)= 2ε̃(b0i ·ey)ey i=1,2,3 (2.3) — for pure shear ∆i( 3ε̃)= 3ε̃[(b0i ·ex)ey+(b0i ·ey)ex i=1,2,3 (2.4) where ex, ey are unit vectors parallel to the x, y axes. 2.3. Mechanical model of cellular skeleton structure Atypicalmicrostructure skeleton consists of thickbeams,whichmaybede- scribedby theTimoshenko beammodel. For structures consisting of long slen- der beams, thus yielding low relative cellular material density, the Bernoulli- Euler beam model is sufficient. Structural mechanics methods are used to analyse the skeleton structure. Resultant forces, i.e. normal force, tangent for- ce and bendingmoment for the sets of midpoint displacements given by Eqs. (2.3) and (2.4), related to uniform deformation, are obtained by making use of the ANSYS FEM code. The skeleton part within the RVE is therefore as shown as in Fig.2. Fig. 2. A part of the skeleton with the stiff node 0 and a local beam system of coordinates The following notation is used: ξi – local coordinate axis for strut i, ξi(0)= 0, 0¬ ξi ¬ li, li =h/2 for i=1, li =L/2 for i=2,3 742 M. Janus-Michalska ni – versor normal to the i-th beam cross-section τi – versor tangent to the i-th beam cross-section. Theaxial and transversal force functions are constant along thebeamaxes, while the bendingmoment changes linearily as follows KF̃ni(ξi)= KF̃ni KF̃τi(ξi)= KF̃τi KM̃i(ξi)= KF̃τi(li− ξi) K =1,2,3 i=1,2,3 (2.5) Due to linearity of the problem, superposition is possible, and thence for an arbitrary strain state ε = (1ε,2ε,3ε), the forces are a linear combination of solutions computed for unit strains and read as follows Fni(ε)= 3∑ K=1 KεKF̃ni Fτi(ε)= 3∑ K=1 KεKF̃τi (2.6) 2.4. Equivalent continuum based on averaging of strain potential The approach adopted here is based on equivalence of the strain potential for the discrete structure and the strain potential of an effective continuum. It refers to averaging the strain energy density (Nemat-Naser andHori, 1999) as written below ΦE = 〈sΦE〉V = 1 V ∫ Vs (sΦE) dVs (2.7) The strain potential of the beam skeletonmay be obtained using the following formula (Piechnik, 2003) U = ∫ Vs (sΦE) dVs = 3∑ i=1 ( li∫ 0 (Fni) 2 dξi 2EsAs +µ li∫ 0 (Fτi) 2 dξi 2GsAs + li∫ 0 [Fτi(li− ξi)]2 dξi 2EsJ ) (2.8) where Es,Gs – Young’s and shear modulus for the skeleton material As,J – beam cross-sectional area andmoment of inertia µ – energy cross-sectional coefficient (for rectangular cross- -section µ=1.2). Due to linearity of the stress-strain relationship, the strain energy density function is represented by the following quadratic form ΦE = 1 2 ε :S : ε (2.9) Micromechanical model of auxetic cellular materials 743 For simplicity, it is convenient to express strains, stresses and stiffness ma- trix in terms of 6D space (Kelvin notation), where the plane stress tensor is represented by a vector and the stiffness tensor representation is a 3× 3 matrix. Introducing relation (2.6) to the expression of strain potential and diffe- rentiating with respect to strain components as follows SIJ = 1 V ∫ Vs ∂2(sΦE) ∂(Iε)∂(Jε) (2.10) one obtains the following formula for stiffness matrix components SIJ = 1 V 3∑ i=1 [ li EsA IF̃in JF̃in+ ( µli GsA + l3i 3EsJ ) IF̃iτ JF̃iτ ] I,J =1,2,3 (2.11) It is worth to emphasise that the energy based definition of S in general does not lead to the same quantities as those obtained through the stress-strain relations based on averaging of stresses in the skeleton over volume of the representative cell (Janus-Michalska, 2005; Nemat-Naser and Hori, 1999). The stiffness matrix for the considered type of symmetry is as follows S= S11 S12 0 S12 S22 0 0 0 S33 (2.12) Its zero components result from the distribution of resultant forces KF̃ni, KF̃τi in the skeleton.Those forces are symmetrical for K =1,2andantisymmetrical for K =3. The constitutive relation for the equivalent continuum is determined by the following linear relation σ=S : ε (2.13) 2.5. Anisotropic linear properties Anisotropic properties are represented by a stiffnes matrix or by a direc- tional distribution of stiffness moduli. For plane structures, the direction is represented by the vector n shown in Fig.3, defined through the angle θ measured from the axis x of the cell coordinate system to the direction n. 744 M. Janus-Michalska Fig. 3. Directions n,mwith respect to the cell coordinate system The following stiffness moduli are considered: Young’s modulus E(n), Poisson’s ratio ν(n,m), shear modulus G(n,m), generalized bulk modu- lus K(n). They can be obtained using the following definitions 1 E(n) = (n⊗n) ·C · (n⊗n) −ν(n,m) E(n) = (n⊗n) ·C · (m⊗m) 1 2G(n,m) = (n⊗m) ·C · (n⊗m ) 1 3K(n) = I ·C · (n⊗n) (2.14) where n – versor specifying the tensile direction in a tension test m – versor perpendicular to the direction of tension n C=S−1 – compliance tensor I – unit tensor n⊗n,n⊗m – diadic tensors. InKelvin’s notation, the diadics and tensors have the representation given below n⊗n= ( cos2θ,sin2θ, 1√ 2 cosθsinθ ) n⊗m= ( −cosθsinθ,cosθ sinθ, 1√ 2 (cos2θ− sin2θ) ) (2.15) I=(1,1,0) C= C11 C12 0 C12 C22 0 0 0 C33 Micromechanical model of auxetic cellular materials 745 The anisotropy parameters defined as follows J1 = S11 S22 J2 = S12+2S33 S11 are introduced tomeasure the degree of anisotropy (for isotropy J1 = J2 =1). 3. Examples 3.1. Comparison with results available in literature Theeffective elastic stiffnessmatrix S calculated according to theproposed method is compared with the matrix S∗ available in literature (Overaker et al., 1998) for the following geometrical data: L = 3.152mm, h = 3.152mm, t = 0.15mm, γ = 70◦, thus yielding the material of relative density ρ/ρs = 0.1155. Skeleton material data are: Es =10GPa, νs =0.3, sRe =100MPa. The computed values S∗= 73.112 −18.49 0 0 4.804 0 0 0 0.038 ·kPa S= 73.122 −18.489 0 0 4.803 0 0 0 0.038 ·kPa show that application of both methods leads to similar results. 3.2. Structural modelling and obtained material properties Geometric parameters of the skeleton structure are chosen to satisfy the assumptions of theTimoshenko beam theory (beams are not overly thick) and to avoid buckling (beams are not overly slender). Calculations areperformed for thematerial dataas in theprevious example and geometric parameters given in Table 1. The obtained effective stiffness moduli and anisotropy coefficients for the given structures are put together in Table 2. The dependence of stiffness moduli on the angle θ yields plots presented inFigs.4-7. Theplots are given for angles 0◦ ¬ θ¬ 180◦ (for 180◦ ¬ θ¬ 360◦ the plots are the same). The analysis shows that cell shape (ratio L/h and angle γ) determines the type of anisotropy and plots shapes. Comparison of these plots leads to conclusion that for materials of identical cell shapes (structure (2) and (6)), which result in similar anisotropy parameters, the curves E(θ), G(θ), K(θ) 746 M. Janus-Michalska Table 1.Geometric parameters Structure L h γ t min { L t , h t } max { L t , h t } type [mm] [mm] [–] [mm] (1) 1.50 1.50 80◦ 0.15 10.0 10.0 (2) 1.50 2.00 60◦ 0.15 10.0 13.33 (3) 1.50 3.00 80◦ 0.15 10.0 20.0 (4) 3.00 1.40 80◦ 0.15 9.333 20.0 (5) 1.50 3.00 60◦ 0.15 20.0 20.0 (6) 3.00 4.00 60◦ 0.15 20.0 26.67 Table 2. Stiffness moduli and anisotropy coefficients Structure EX EY νXY νYX max ( G K ) J1 J2 type [kPa] [kPa] (1) 46.183 1.297 −5.058 −0.142 5.52 35.556 −0.1317 (2) 6.270 1.941 −1.718 −0.532 28.9 3.230 −0.5192 (3) 20.900 2.808 −2.289 −0.307 10.1 7.442 −0.2996 (4) 20.071 0.0586 −17.708 −0.0527 14.4 342.07 −0.0495 (5) 3.6713 3.4074 −0.9823 −0.9119 108.3 1.0774 −0.8970 (6) 0.7834 0.2402 −1.786 −0.5472 34.2 3.1616 −0.5448 Fig. 4. Young’s modulus in dependence of the angle θ for different structures are also similar and their maximum values decrease approximately with the third power of dimension of the proportion ratio. This phenomenon may be explained by domination of bending over tension in skeleton beams, resulting in such a dependence (Janus-Michalska, 2005).Materials with slender skeleton beams are very compliant. Micromechanical model of auxetic cellular materials 747 Fig. 5. Poisson’s ratio in dependence of the angle θ for different structures For anisotropic materials the bounds of negative Poisson’s ratio are wider than for isotropic materials, and theoreticaly may reach infinity (for isotropic materials in two-dimensional problems the acceptablePoisson’s ratio is limited by −1 ¬ ν ¬ 1). The existence of directions with high auxetic behaviour in cellular materials is bound with high anisotropy (both parameters influence the plot shape). Fig. 6. Generalized bulk modulus in dependence of the angle θ for different structures Thequantity 1/[3K(θ)] represents the relative change of volumeper tensile unit stress in the direction given by θ. The bulk modulus is sensitive only to the first anisotropy parameter J1. Contrary to the bulk modulus K, shear modulus G is only sensitive to the second anisotropy parameter J2. 748 M. Janus-Michalska Fig. 7. Shear modulus in dependence of the angle θ for different structures Due to lack of typical symmetry, the distribution of stiffness moduli de- pends only on the choice of geometric andmaterial parameters, which results in various anisotropy parameters and values of stiffness moduli. 4. Conclusions An effective model based on the framework of micromechanical analysis is built and applied for an auxetic cellular material to predict elastic properties on the macroscale. The same framework can be applied for arbitrary cellular material with open cells of a regular skeleton structure. It is especially useful for cells representing complicated shapes of the skeleton (multiple nodes, lack of symmetry). Various structural topologies of cellular materials and skeleton materials result in different macroscopic properties, which can be tailored to special demands.Theproper choice ofmicrostructural geometrical parameters determines expected elastic properties. The sensitivity analysis of structural parameters for the auxetic cellularmaterial allows for formulation ofmodelling indications. References 1. GibsonL.J.,AshbyM.F., 1997,Cellular Solids, 2nd edition,CambridgeUni- versity Press 2. Hori M., Nemat-Nasser S., 1999, Onmicromechanics theories for determi- ningmicro-macro relations in heterogeneous solids,Mechanics ofMaterials,31, 667-682 Micromechanical model of auxetic cellular materials 749 3. Janus-Michalska M., 2005, Effective models describing elastic behaviour of cellular materials,Archives of Metallurgy and Materials, 50, 3, 596-608 4. Janus-Michalska M., Pęcherski R.B., 2003, Macroscopic properties of open-cell foams based onmicromechanicalmodelling,Technische Mechanik, 23, 2/4 5. Kordzikowski P., Janus-Michalska M., Pęcherski R.B., 2005, Speci- fication of energy-based criterion of elastic limit states for cellular materials, Archives of Metallurgy and Materials, 50, 3, 621-634 6. KumarR.S.,McDowellD.L., 2004,Generalizedcontinuummodelingof 2-D periodic cellular solids, Int. Journal of Solids and Structures, 41, 7399-7422 7. Lakes R.S., 1991a, Deformation mechanisms of negative Poisson’s ratio ma- terials: structural aspects, Journal of Material Science, 26, 2287-2292 8. Lakes R.S., 1991b, Experimental micromechanics methods for conventional and negativePoisson’s ratio cellular solids asCosserat continua,J. Engineering Materials and Technology, 113, 148-155 9. Lakes R.S., 1992, No contractile obligations,Nature, 358, 713-714 10. Lakes R.S., 1993a, Advances in negative Poisson’s ratio materials,Advanced Materials, 5, 293-296 11. Lakes R.S., 1993b, Design considerations for negative Poisson’s ratio mate- rials,ASME Journal of Mechanical Design, 115, 696-700 12. Lakes R.S., Friss E.A., Park J.B., 1988, Negative Poissons ratio for poly- meric andmetallic materials, Journal of Material Science, 23, 4406-4414 13. Lakes R.S., Witt R., 2002, Making and characterizing negative Poisson’s ratio materials, Int. Journal of Mechanical Engineering Education, 30, 50-58 14. Mehrabadi M.M., Cowin S.C., 1990, Eigentensors of linear anisotropic ela- stic materials,Quarterly Journal of Mech. Appl. Math., 43, 15-41 15. Nemat-Naser S., Hori M., 1999,Micromechanics, 2nd edition, Elsevier 16. Ostrowska-Maciejewska J., Rychlewski J., 2001, Generalized pro- per states for anisotropic elastic materials, Archives of Mechanics, 53, 4/5, 501-518 17. Overaker D.W., Cuitino A.M., Langrana N.A., 1998, Elastoplastic mi- cromechanical modeling of two-dimensional irregular convex and nonconvex (Re-entrant) hexagonal foams,Transactions of ASME, 65, 748-757 18. Piechnik S., 2003,Wytrzymałość materiałów, manual for students, Kraków 19. Smith C.W., Grima J.N., Evans K.E., 2000, A novel mechanism for gene- rating auxetic behaviour in reticulated foams: missing rib foam model, Acta Materialia, 48, 4349-4356 20. Warren W.E., Kraynik A.M., 1988, The linear elastic properties of open- cell foams, Journal of Applied Mechanics, 55, 341-346 750 M. Janus-Michalska Model mikromechaniczny materiałów komórkowych o ujemnym współczynniku Poissona Streszczenie Celem pracy jest sformułowanie efektywnego anizotropowego continuum spręży- stego dlamateriałówkomórkowycho ujemnymwspółczynnikuPoissona. Szkieletma- teriału jest modelowany przez płaską strukturę belkową połączoną w sztywnych wę- złach tworzącą układwielokątówwklęsłych.Kątywklęsłe w strukturzemateriału od- powiadają za efekt ujemnego współczynnika Poissona. Poprzez zastosowaniemodelu mikromechanicznego istotne cechy mechaniczne materiału komórkowego są wypro- wadzone z wyników analizy komórki reprezentatywnej. Potencjał sprężysty szkieletu komórki jest wyznaczony jako suma energii w belkach tworzących szkielet od ich rozciągania, ścinania i zginania. Efektywne continuum jest oparte na uśrednianiu po- tencjału sprężystego, co jest podstawą konstruowania macierzy sztywności. Metoda analizy strukturalnej przeprowadzona za pomocą programuMES-ANSYS jest stoso- wanadlamodelu belkowego szkieletu. Jakowynik tej analizyprzedstawionograficznie rozkładymodułuYounga, współczynnika Poissona,modułu na ścinanie i uogólnione- go współczynnika ściśliwości objętościowej. Studium parametryczne umożliwia prze- śledzenie wpływu parametrów geometrycznych struktury i charakterystykmateriału szkieletu na własności kontinuum zastępczego jako materiału o zastosowaniu struk- turalnym. Manuscript received November 13, 2008; accepted for print April 6, 2009