Jtam-A4.dvi JOURNAL OF THEORETICAL AND APPLIED MECHANICS 56, 3, pp. 829-839, Warsaw 2018 DOI: 10.15632/jtam-pl.56.3.829 MODELLING OF FGM PLATES Grzegorz Jemielita Warsaw University of Life Sciences, Faculty of Civil and Environmental Engineering, Warsaw, Poland e-mail: g.jemielita@il.pw.edu.pl Zofia Kozyra Warsaw University of Technology, Faculty of Civil Engineering, Warsaw, Poland e-mail: z.kozyra@il.pw.edu.pl The paper presents formulation of the problem of layered plates composed of two various isotropic materials.We assume that the first materialM1 is characterized by the following parameters: Young’s modulusE1 and Poisson’s ratio ν1, whereas the second one byE2 and ν2, respectively. Let us consider twomodelling cases for functionally gradedmaterial (FGM) plates.These cases are related to anappropriatedistribution of thematerialwithin two-layer and three-layer systems. Our objective is to compare the stiffness of both the two-layer and there-layer plates with the FGMplate containing various proportions between thematerial componentsM1 andM2. Keywords: modelling, layer plate, FGM plate, stiffness of the plate 1. Introduction The effect of a continuous change in the plate properties (through the thickness) can be obtained in variousways. The overall properties of FMGs are unique and differ fromany of the individual material forming it (Mahamood et al., 2012). Usually, we assume isotropic plates with variation of two constituents: ceramic and metal (Mokhtar et al., 2009). Another possibility is to assume aluminium-alumina FGMplates (Rohit Saha andMaiti, 2012). Thus, the created plate material is inhomogeneous with both the composition andmaterial properties varying smoothly through thickness of the plate. The material properties along the thickness direction of the FGM plate vary in accordance to a power-law function, exponential function, sigmoid function, etc. Anothermodelling examples known in the literature are related to asymptotic and tolerance modelling (Woźniak, 1995; Nagórko, 1998, 2010; Wągrowska and Woźniak, 2015; Woźniak et al., 2016). Most material properties through the plate thickness are expressed by p(z)= (p1−p2)f(z)+p2 (1.1) where f(z)= (1 2 + z h )n p= ρ,E,ν (1.2) where h is the plate thickness, the subscripts 1 and 2 indicate the top z = h/2 and bottom z = −h/2 surfaces, E – Young’s modulus, ν – Poisson’s ratio, ρ is mass density, n – material 830 G. Jemielita, Z. Kozyra parameter (Reddy, 2000; Efraim, 2011; Kim and Reddy, 2013; Kumar et al., 2011). Mokhtar et al. (2009) use the following definition of the function p(z) p(z)=    f1(z)pc+[1+f1(z)]pm for 0¬ z¬ h 2 f2(z)pc+[1+f2(z)]pm for − h 2 ¬ z¬ 0 (1.3) where f1(z) =1− 1 2 ( 1− 2z h )n f2(z)= 1 2 ( 1+ 2z h )n (1.4) Delale and Erdogan (1983) and Rohit Saha and Maiti (2012) used an exponential function in order to describe the variation of Young’s modulus in the following form E(z) =Emexp ( z+ h 2 ) B= 1 h ln Ec Em − h 2 ¬ z¬ h 2 (1.5) In this work, we present a new FGM plate model formulated by using an appropriate mo- delling related to double-layer and three-layer plates. 2. Modelling of an FGM plate developed from a two-layer plate Let us assume a two-layer plate whose scheme is shown in Fig. 1. Fig. 1. A scheme of the two-layer plate Our aim is to construct a gradient plate developed from a two-layer plate as well as to compare the stiffnesses of both systems in the case of any quotient η=h1/h. 2.1. Plate geometry The function describing the properties of the plate is defined as follows p(z)= { p1 for 1−η¬ ξ¬ 1 p2 for 0¬ ξ¬ 1−η } = p2[1+εs(ξ)] p= ρ,E,ν (2.1) where s(ξ)= { 0 for 0¬ ξ¬ 1−η 1 for 1−η¬ ξ¬ 1 (2.2) where ρ denotes density, E – Young’s modulus, ν – Poisson’s ratio, ε= (p1−p2)/p2, ξ = z/h, η= h1/h, 0¬ η¬ 1, h= h1+h2. Visualization of the discontinuous function s(ξ) (defined via Eq. (2.2)) is presented in Fig. 2. We are looking for a continuous-density function ρ(ξ) satisfying the following conditions: — the law of mass conservation 1∫ 0 ρ(ξ) dξ= ρ1h1+ρ2h2 h = ρ2(1+ερη) ερ = ρ1−ρ2 ρ2 η= h1 h (2.3) Modelling of FGM plates 831 Fig. 2. Discontinuous function s(ξ), see Eq. (1.2) —positive-density condition ρ(ξ)­ 0 0¬ ξ¬ 1 (2.4) — boundary conditions ρ(0)= ρ2 ρ(1)= ρ1 (2.5) The function under consideration ρ(ξ) can be described as follows ρ(ξ)= ρ2[1+ερf(ξ)] (2.6) where the function f(ξ) satisfies the conditions as follows 1∫ 0 f(ξ) dξ= 1∫ 0 s(ξ) dξ= η f(0)= 0 f(1)= 1 f(ξ)­ 0 0¬ ξ¬ 1 (2.7) The functions characterizing elastic properties of the FGM material, i.e. Young’s modu- lusE(ξ) and Poisson’s ratio ν(ξ), should satisfy the following conditions E(ξ)­ 0 0¬ ξ¬ 1 E(0)=E2 E(1)=E1 ν(ξ)­ 0 0¬ ξ¬ 1 ν(0)= ν2 ν(1)= ν1 (2.8) The forms of analyzed functions E(ξ) and ν(ξ) differ. In the case of metallic alloys, one can assume ν1 = ν2 = ν because of small differences between Poisson’s ratios and the fact that the effect of Poisson’s ratio on the deformation is much less than that of Young’s modulus (Delale, Erdogan, 1983). However, analyzing the plates, it can be assumed that the function E(ξ)/[1−ν2(ξ)] is analogous to the form of the density function expressed by Eq. (2.6) E(ξ) 1−ν2(ξ) = E2 1−ν22 [1+εEf(ξ)] εE =α−1 α= E1(1−ν 2 2) E2(1−ν 2 1) (2.9) and the function f(ξ) satisfies conditions (2.7). Let us note that taking the function E(ξ)/[1−ν2(ξ)] along with conditions (2.8)1 leads to a simple interpretation of the static problem. Tensile strength of the FGM cross-section with Young’s modulus andPoisson’s ratio described byEqs. (2.9) is the same as for the cross-section shown in Fig. 1. Figure 3 shows the functions s and f satisfying conditions (2.7) for η=0.2. We anticipate the following form of f(ξ) f(ξ)= fr(ξ)= ξ r(a1ξ+a2) (2.10) where r=n or r=1/n, and n is an arbitrary natural number. 832 G. Jemielita, Z. Kozyra Fig. 3. The functions f and s The proposed function satisfies boundary condition (2.7)2, whereas condition (2.7)3 is satis- fied ∀r 6=0 if a2 =1−a1 ⇒ fr(ξ)= ξ r[a1(ξ−1)+1] (2.11) As a special case r=n=0, condition (2.7)3 will be satisfied with a1 =1. Condition (2.7)1 gives 1∫ 0 f(ξ) dξ= 1∫ 0 ξr[a1(ξ−1)+1] dξ= 2+r−a1 (1+r)(2+r) = η (2.12) what implies a1 =(2+r)[1−η(1+r)] (2.13) Taking condition (2.7)4 ξr[a1(ξ−1)+1]­ 0 (2.14) it follows that at n> 0, f(ξ)­ 0 for −r¬ a1 ¬ 1 1 r+2 ¬ η¬ 2 r+2 (2.15) It is obvious that for each η, we can find an appropriate natural number n for r = n or r=1/n. Thus, f(ξ) is written as follows f(ξ)= fr(ξ)= ξ r{1− (1− ξ)(2+r)[1−η(1+r)]} (2.16) As a special case of f0(ξ), we obtain η=1/2, f0(ξ)= ξ. Taking r=1 f1(ξ)= ξ[1− (1− ξ)3(1−2η)] for 1 3 ¬ η¬ 2 3 (2.17) For n­ 2, we have two different functions fnξ)= ξ n{1− (1− ξ)(2+n)[1−η(1+n)]} (2.18) or gn(ξ)= f1/n(ξ)= ξ 1 n { 1− (1− ξ) ( 2+ 1 n )[ 1−η ( 1+ 1 n )]} (2.19) Modelling of FGM plates 833 Fig. 4. Ranges of η for f n (ξ) and g n (ξ) satisfying conditions (2.7) For a given value η, one can find such a value n at which the functions fn(ξ) or gn(ξ) meet conditions (2.7). Figure 4 shows such ranges of η at which the functions fn(ξ) and gn(ξ) satisfy the conditions expressed via Eqs. (2.7). For η = 1/3, Fig. 4 depicts four possible values of n, i.e. n = 1,2,3,4. On the other hand, we have three functions to be shown in Fig. 5a. In the case of η=0.2, we obtain six functions satisfying conditions (2.7) for n= 3,4,5,6,7,8 (see graphs visualized in Fig. 5b). It is obvious that for any η it gives a collection of functions satisfying conditions (2.7). Fig. 5. Graphs of f n (ξ) for: (a) η=1/3 and n=1,2,3,4; (b) η=0.2 and n=3,4,5,6,7,8 Moving back to the graph illustrated in Fig. 4, it is obvious that for 1/2 ¬ η ¬ 1 we get an infinite number of functions satisfying conditions (2.7). Figure 6 shows variation of func- tions E(ξ)(1− ν22)/{E2[1− ν 2(ξ)]} = 1+ εEf(ξ), f(ξ) = fn(ξ) or g(ξ) = gn(ξ) for α = 2, η=1/5,1/6,1/3 along the FGM plate thickness. 2.2. Stiffness of the two-layer and FGM plates The stiffness of the two-layer plate is expressed by the formula Dw = E2h 3 12(1−ν2) [ η3α+(1−η)3+ 3(1−η)αη 1−η(1−α) ] (2.20) 834 G. Jemielita, Z. Kozyra Fig. 6. Cross section of the FGM two-component plate: (a) α=2, n=3, η=1/5, (b) α=2, n=1, η=1/3, (c) α=2, n=1, η=2/3 The stiffness of the FGM plate depends on the functions f or g. These functions depend, in turn, on the variable ξ, on the parameter η=h1/h and on the natural number n Ef(ξ,α,n,η) = E 1−ν2 (ξ,α,n,η) = E2 (1−ν22 [1+(α−1)fn(ξ)] (2.21) or Eg(ξ,α,n,η) =Ef ( ξ,α, 1 n ,η ) = E2 1−ν22 [1+(α−1)gn(ξ)] (2.22) The stiffness of the FGM plate material is expressed via formulas: — for 1 n+2 ¬ η¬ 2 n+2 Df(FGM)(α,n,η) = E2h 3 1−ν22 1−ef∫ −ef [1+(α−1)fn(ξ+ef)]ξ 2 dξ = 1 3 −ef +e 2 f +(α−1) ( ηe2f + 2+(1+n)(2+n)η (3+n)(4+n) − 2ef[1+(1+n)(2+n)η] (2+n)(3+n) )(2.23) — for n 1+2n ¬ η¬ 2n 1+2n Dg(FGM)(α,n,η) =Df(FGM) ( α, 1 n ,η ) = E2h 3 1−ν22 1−eg∫ −eg [1+(α−1)g(ξ+eg)]ξ 2 dξ= 1 3 −eg +e2g +(α−1) ( 2n2+η{(1−eg)[1−eg +n(3−7eg)]+2n 2(1−4eg +6e 2 g)} (1+3n)(1+4n) − 2n2eg (1+2n)(1+3n) ) (2.24) whereas ef(α,η,n) = 4+5n+n2+2α+2(1+n)(2+n)(α−1)η 2(2+n)(3+n)[1+(α−1)η] eg = ef ( α,η, 1 n ) = 1+5n+4n2+2n2α+2(1+n)(1+2n)(α−1)η 2(2n+1)(3n+1)[1+(α−1)η] (2.25) Modelling of FGM plates 835 In general, we have the following inequalities Df(FGM) ( α,n,η= 2 2+n ) 6=Df(FGM) ( α,n+1,η= 1 2+n ) Dg(FGM) ( α,n,η= 2n 2n+1 ) 6=Dg(FGM) ( α,n+1,η= n 2n+1 ) (2.26) but there always exists such η̂ meeting the following equality for each α DFGM(α,n, η̂)=DFGM(α,n+1, η̂) (2.27) For each α, the following equality is satisfied Df(FGM) ( α,n,η= 1 2+n ) =Df(FGM) ( α,n+1,η= 1 2+n ) (2.28) In the case of n/(1+2n)¬ η¬ 2n/(1+2n), there exist such values of η̂ satisfying the following equalities Dg(FGM)(α,n, η̂)=Dg(FGM)(α,n+1, η̂) (2.29) For example:Dg(FGM)(α=2,n=1, η̂)=Dg(FGM)(α=2,n=2, η̂), for η̂=0.424472. Figure 7 presents diagrams of the function r(η) describing the stiffness of the plate D = E2h 3r(η)/[12(1− ν22)] for α = 2 where Dw and DFGM denote the two-layer plate and the gradedmaterial plate, respectively. Fig. 7. A diagram of the function r(η) 3. Modelling of the gradient plate developed from the three-layer plate Let us consider a three-layer plate with a symmetrical arrangement of layers. In such a case, it is convenient to take the intermediate layer thickness equal to 2h2 (Fig. 8). 3.1. Determining the functions f n , g n Similarly, as in the case of two-layer plate, we assume that Eqs. (2.9) hold provided that the functions fn and gn have the following forms fn(ξ)= ξ 2n[1−a1(1− ξ 2)] gn(ξ)= (ξ 2) 1 n [1− b1(1− ξ 2)] (3.1) 836 G. Jemielita, Z. Kozyra Fig. 8. A scheme of the three-layer plate Satisfying the following conditions (see (2.7)) 1∫ 0 f(ξ) dξ= 1∫ 0 s(ξ) dξ= η= h1 h f(±1)= 1 f(0)= 0 f(ξ)­ 0 −1¬ ξ¬ 1 (3.2) we obtain the functions fn and gn of the form: — for 1 3+2n ¬ η¬ 3+4n (1+2n)(3+2n) fn(ξ)= ξ 2n ( 1− 1 2 (3+2n)[1−η(1+2n)](1− ξ2) ) (3.3) — for n 2+3n ¬ η¬ n(3n+4) (2+n)(2+3n) gn(ξ)= f1/n(ξ)= (ξ 2) 1 n ( 1− 1 2n2 (2+3n)[n(1−η)−2η](1−ξ2) ) (3.4) Figure 9 shows the ranges of η for the functions fn(ξ) and gn(ξ) satisfying conditions (2.7). Fig. 9. Ranges of η for f n (ξ) and g n (ξ) satisfying conditions (2.7) Figure 10a presents functions f1 = g1 within the range of validity 3/15¬ η¬ 7/15, whereas the functions g10 for 30/96¬ η¬ 85/96 are visualized in Fig. 10b. Similarly, as in Section 3, the functionE(ξ)/[1−ν2(ξ)] can be expressed as follows E(ξ) 1−ν2(ξ) = E2 1−ν22 [1+εEf(ξ)] (3.5) Modelling of FGM plates 837 Fig. 10. (a) Functions f1 = g1 for 3/15¬ η¬ 7/15, (b) functions g10 for 30/96¬ η¬ 85/96 where the following designations are adopted εE =α−1 α= E1(1−ν 2 2) E2(1−ν 2 1) f(ξ)= fn(ξ) or f(ξ)= gn(ξ) (3.6) Figure 11 presents variations of the function E(ξ)(1− ν22)/{E2[1− ν 2(ξ)]} = 1+ εEf(ξ), f(ξ)= fn(ξ) or g(ξ) = gn(ξ) for α=2, η=1/5,7/15,5/8 along the FGM plate thickness. Fig. 11. Cross section of the FGM two-component plate: (a) f1(α=2,η=1/5), (b) f1(α=2,η=7/15), (c) g2(α=2,η=5/8) 3.2. Stiffness of the three-layer FGM plate The stiffness of the three-layer plateDw can be described as follows Dw = 2E2h 3 3(1−ν22) [(1−η)3+αη(3−3η+η2)] (3.7) In the case of the FGM plate, the following formulas can be applied: — for 1 3+2n ¬ η¬ 3+4n (1+2n)(3+2n) Df(FGM)(α,n,η) = 2E2h 3 3(1−ν22) 3(3+2α)+4n(4+n)+3η(α−1)[3+4n(2+n)] (3+2n)(5+2n) (3.8) — for n 2+3n ¬ η¬ n(4+3n) (2+n)(2+3n) Dg(FGM)(α,n,η) = 2E2h 3 3(1−ν22) 4+16n+9n2+6αn2−3η(2+n)(2+3n)(α−1) (2+3n)(2+5n) (3.9) 838 G. Jemielita, Z. Kozyra The stiffnesses of the layered plate Dw and the FGM plate DFGM as a function of η for α=E1/E2 =2 are visualized in Fig. 12. Fig. 12. Layered plate stiffness D w and FGMplate stiffness DFGM as a function of η for α=E1/E2 =2 4. Conclusions Figure 7 shows the dependence of the stiffness of a two-layer plate and an FGM plate on the parameter η. Analyzing this graph, one concludes that taking two plates (the two-layer and FGM) with the same amount of the material and for certain values of the parameter η, the stiffness of the FGM plate is greater than the stiffness of the two-layer plate. It proves that we can construct an FGM plate, according to the procedure outlined in Section 2, with a higher stiffness of the two-layer plate andwith the same amount of thematerial for both plates. In the case of the three-layer plate and its corresponding FGM plate, the FGM plate stiffness is less than the stiffness of the three-layer plate for all values of the parameter η, except for η=0 and η=1, for which the stiffnesses are the same. References 1. Delale F., Erdogan F., 1983, The crack problem for a nonhomogeneous plane,ASME Journal of Applied Mechanics, 50, 609-615 2. Efraim E., 2011,Accurate formula for determination of natural frequencies of FGMplates basing on frequencies of isotropic plates,Procedia Engineering, 10, 242-247 3. Kim J., Reddy J.N., 2014, Analytical solutions for bending, vibration, and buckling of FGM plates using a couple stress-based third-order theory,Composite Structures, 103, 86-98 4. Kumar J.S., Reddy B.S., Reddy C.E., Kumar Reddy K.V., 2011, Geometrically non linear analysis of functionally gradedmaterial plates using higher order theory, International Journal of Engineering Science and Technology, 3, 1, 279-288 5. Mahamood R.M., Esther T. Akinlabi E.T., Shukla M., Pityana S., 2012, Functionally gradedmaterial: an overview,Proceedings of theWorld Congress on Engineering, WCE 2012, III, London, U.K. 6. Mokhtar B., Abedlouahed T, Abbas A.B., Abdelkader M., 2009, Buckling analysis of functionally graded plates with simply supported edges, Leonardo Journal of Sciences, 16, 21-32 Modelling of FGM plates 839 7. NagórkoW., 1998,Twomethods ofmodeling of periodic nonhomogeneous elastic plates, Journal of Theoretical and Applied Mechanics, 36, 291-303 8. NagórkoW., 2010,Asymptoticmodeling inelastodynamicsofFGM, [In:]MathematicalModelling andAnalysis inContinuumMechanics ofMicrostructuredMedia,WoźniakC. et al. (Edit.), Silesian University of Technology, Gliwice, 143-151 9. Reddy J.N., 2000, Analysis of functionally graded plates, International Journal for Numerical Methods in Engineering, 47, 663-684 10. Rohit Saha, Maiti P.R., 2012, Buckling of simply supported FGMplates under uniax ial load, Iternational Journal of Civil and Structural Engineering, 2, 4, 1035-1050 11. Wągrowska M., Woźniak C., 2015, A new 2D-model of the heat conduction in multilayered medium-thickness plates, Scientiarum Polonorum Acta Architectura, 14, 1, 37-44 12. Woźniak C., 1995,Microdynamics: continuummodeling the simple compositematerials, Journal of Theoretical and Applied Mechanics, 33, 267-289 13. Woźniak C., Wągrowska M., Szlachetka O., 2016, On the tolerance modelling of heat conduction in functionally graded laminated media, Journal of Applied Mechanics and Technical Physics, 56, 2, 274-281 Manuscript received August 22, 2016; accepted for print February 7, 2018