You are on page 1of 16
Composites: Par B 33 2002) 505-819 Part B: engineering wi clever comioeatfeompostesd Free vibration analysis of composite sandwich plates based on Reddy’s higher-order theory AK. Nayak", S.S.J. Moy”, R.A. Shenoi** "School of Engineering Sciences, Unversity of Souhampton, Highcld,Southarypton SO1? IBY, UK “Department of Chil and Environment! Engineering, Univers of Souhanpton, Mighell, Southampton SOI7 181, UK Received 23 July 2001: revised 17 June 2002; aeepted 19 June 2002 Abstract ‘Two new C° assumed strain finite element formulations of Reddy's higher-order theory are used to determine the natural frequencies of isouropic, orthotropic, and layered anisotropic composite and sandwich plates. The material properties typical of glas ibe polyester resins forthe skin and HEREX C70 PVC (polyvinyl chloride) foam materials forthe core are used to show the parametic effets of plate aspect ratio, lengthto-thickness ratio, degre of rthotropy, numberof layers and lamination scheme on the natural frequencies. A consistent mass ‘matrix is adopted in the resent formulation. The results presented in this investigation could be useful fora beter understanding of the behaviour of sandwich laminates under free vibration conditions and potentially beneficial for designers of sandwich structures. © 2002 Elsevier Science Lid. All rights reserved Keyword: Free vibration 1. Introduction ‘The use of sendwich structures in marine transport, civil construction and aerospace applications is growing rapidly because of advantageous features such as high strength to weight ratio and low maintenance cost. With the increased interest and use of sandwich structures in many challenging situations over the years, itis therefore very important 10 study its behaviour in various circumstances, Sandwich plate and shell behaviour ina static and strength context has been well documented in the texts of| Plantema [1], Allen [2], Zenkert [3] and Vinson [4]. The same does not hold true for dynamic behaviour, which is in sharp contrast to the analyses of laminated plates. The + (aie) + (Sie) where the stress resultants N,, Mj, P,, Q; and Rj are defined by (Ms MisPi Pypeadde 0-126 sn wa (uk) f asCh,2 ne ” (Ona) [eth ve The inertia [= 1,2,3,4,5,7) are defined by a Cntototattd= [hat ete 0) Note that by neglecting terms associated with , and in Eq. (8), the equations of motion for FSDT can be AK: Naya eta. Compesties: Par B33 (2002) 505-519 50 obtained (see Ref. [36]). The various stress resultants are aiven by N) pal w) qe M |i (D1 (FI |} a? ay Py Liem unde, teb-Lon willed « where N=(MN2No)", (ene) Mah MMe, =U GET PH=(PP2Pi, P= (dd Ae)™ 2=@,0., f=" RRR) Ka OS AT An Ar Ai [l=] An An Aas Are Aas Aes [B1, (DI, (£1, (F1 and [41] can be expressed in a similar wf =| “LA Ae [D'] and [F*] can be expressed in a similar manner. The ‘elements of matzices of plate stiffnesses ((A), (81, (D], (E], (F1, (HN, [4°], (D"] and [F*)) are given by Ay Bip Dy Egy FH Pia Oahnee 2 .2)de G=1,2,6 aa : Dn FD =f VaQhZchle I= 59) (3) Using Eqs. (8), (11) and (12) we get the principle of virtual work equation in the following form [, fuser! seman! sere eatin! + Ba DIR”) + BF] + Bx [E]e” + BLE + Bx THC + Be [A]! + Be TD! ]e! + 8 (Det anTimeaaar+ f [gow aa at =f f[ooan-+ni amy FL By + Vv + ib V+ HBF) + (FE) (G.di0+ dv +is86, +84) + (Ge) He. as, + dan+dow,) iy Hck ats bows (Se )(d.86,+8,865) sar a 24, Assumed transverse shear strain rate interpolation One basic problem inherent in the use of standard interpolation of the strains for the transverse shear components is that the element locks when itis thin. The reason for this locking is that the element, when loaded in pure bending, will exhibit spurious transverse shear energy. In order to overcome the shear locking, Dvorkin and Bathe (22) and Huang and Hinton (23] proposed assumed interpolations for the shear strain to develop 4 node and 9 rhode assumed strain plate elements, respectively. In this paper, the above formulations are used to derive new 4 node snd 9 node Reddy type elements. The substitute shear strain fields are choven as follows. For the 4 node element: & => YPeomel and oxerines, where Ou) = 40/2, Q)= 0-9/2, PY =1 @= én) For the 9 node element: => SP@qmed as = ¥ Yor imey of where OG) = 1 +2)/2, Q@) = (1-2), Qs(2) = Xe ~ 1/2, PA 1+2/a)/2, as) P= (1-2/2 én 310 AK Nayak eal. / Composites: Pant B 33 (2002) 505-519 a "1 ete wb) . UG 2 2 @ (b) p> Sampling potion ser intron in he 4 node assumed suain element ef.) in which a= 0.577, eff and €% are the mn unknown substitute shear strain parameters associated with two sets of mx n sampling points (&, 3) and (6,7) (see Figs. 2 and 3 for details). In order to eliminate locking, the following equations are obtained Es B= Es) Ey) = eG) C= Ay Itis possible to write 6 «-{2} as) where € and é; are obtained from €4, and €3, given by Eqs. (16) and (17), respectively, by tensor transformation, ‘The tvansformation of the strain tensor in curvilinear ‘coordinates may be written as et aes oot Beg Beg t Lm JS ln) (19) A= Lyng) (17) ay ‘where it is assumed that egg isthe strain tensor in the (&,) coordinate system and €y i the strain tensor in the (x,)) system. For further details see Ref. (22] for 4 node element and Ref. (23] for 9 node element. For implementation purpose, e* in Eq, (14) is replaced by & where é isthe substitute shear strains to remove spurious zero energy modes as mentioned earlier in the paper. n : | bos eb eal Jan : Gy ot ykan * 2 3 GP @2 G2) G1) 2 3 @ o eee a sein mest @) 4, 0 3. Einite element formulation ‘The general procedure for obtaining fnite element formulations trom the principle of virtual work is well known and hence only a bref overview ofthe same is given, The finite element equations are obtained by discretiing the plane region & into « number of isoparametric elements Each element e has n nodes, where each node i(i = 1,..., 1) is identified with seven degrees of freedom Uj, — sy Wo Me Bo. For simplicity, we assume the ste’ interpolation for all seven. variables in each element, ic m= SN, % Sn, = SN, w= SNM, 4,= 306 20 Where Nj, i= 1,m, a¥e the interpolation functions. Knowing the generalized displacement vector (Ui (N}(8}.0) at all points within the element ¢, the generalized mid-surface strains at any point given by Eq. (4) can be expressed in terms of nodal displacements as follows 8 = [Bele OS BI) = BIB, BOR BIg = (BI ho, where (80), (BQ), (Bt), (82), (BE) are generated strain isplacement matrices. For arbitrary values of virtual displacements, Eq. (14) finally leads to the following assembled equations for free vibration analysis tanta) +1114} = (0) en Here the unknown vector (4) is generated by the assemblage of clement degrees of freedom {d)!, Iya. total degrees of freedom in the region R. The assembled stifiness matrix, mass matrix for free vibration analysis are wi-¥f, (ees aywe 1+ (Bey (BIB) (BEIT LEMB]+ 1851" (BILBE1+ 0880" (DICE) HBR TFB BET EUBE+ (Bid FB +B B+ BA B+ BE DE FBT DUB + BT FIBA 2) AK. Naya eta. / Composites: Part B 3 (2002) 505-519 su wt xf, eves 23) ‘where [Mj] is the mass matrix containing inertia terms which can be derived from the right-hand side of the expression in Eq, (14). ‘The 3X3 and 2X2 Gauss—Legendre rules (ie. full integration scheme) have been employed to integrate bending, membrane, shear and inertia terms in the energy ‘expressions for the 9 node element and the 4 node element, respectively. The resulting eigenvalue problem in Eq. (21), has been solved using Amoldi’s process. Details regarding Amoldi’s process can be found in the works of Lehouca, Sorensen and Yang [37]. 4, Results and discussions ‘The performance characteristics of the new Reddy type clements in predicting the natural frequencies of composite sandwich plates are evaluated. The results obtained for some benchmark problems are presented, starting with the isotropic case, then the orthotropic case and finally laminated composite and sandwich plates. Some of the ‘examples sre compared with those already reported by other ‘authors. The following boundary conditions have been used. ‘Simply supported (S) (For cross ply): up = wy = Ye = 0.aty=0,b; vy = wo = Vy = 4 = Oat r= 0,4; (8) (or angle ply): vp Y= 4, = 0 at y= 0,0; ty 4.1. Free vibration analysis of a simply supported square plate (isotropic) ‘To examine the numerical accuracy of the new Reddy type elements, natural frequencies are first calculated for a thick isotropic square plate with a Iength-to-thiekness ratio of 10, An 88 finite element mesh for the whole plate is used in this example. The results are compared with those from 3D elasticity (exact) solution [38], higher-order theory {39] and classical plate theory (CPT) in Table 1. The results from the present formulation are found to be quite close to Table 1 CCompasison of non-dimensional frequencies 11= hJp/G of « square simply supported ple wich »= 0.30, a/h = 10 Modes Exact “SDT Present’ Present CPT. a 0.0981 ous 0.083085 a2 om 02253 ommz 0360 22) Osa 03463 oss a3 aise 04299 0416904629 es 05221 05368 0528105951 aa 0msts 06540 045599 0.1668 oy 06862 0081688008090 ea) ee the exact results. The analytical results of the higher-order theory [39] presented in Table 1 appear to be better than the finite element results with both 4 and 9 node elements. Tt should, however, be noted that the finite element method is ‘an approximate method and the present results are obtained ‘with an 8 x8 mesh in a full plate Further refinement of the ‘mesh could yield improved results. It is seen that the results ‘obtained from the present elements are very close to those of | the 3D elasticity solution, The CPT overestimates the frequencies and the effect of transverse shear deformation increases with increasing mode numbers. 4.2. Free vibration analysis ofa thin square plate (isotropic) ‘with simply supported and clamped boundary conditions As mentioned i the literature review section, when shear deformable plate bending elements are used to analyse thin plates, overstff results are obtained, which is popularly known as ‘locking’ phenomenon, This has been investigated by Hinton and Bicanic [40] in free vibration situations. We hhave taken this problem as testcase to investigate locking in our new 9 node and 4 node elements, The results for free vibration of a simply supported and clamped square plate With increasing length-to-thickness ratio are presented in Fig. 4. The results are normalized with the non-dimensional frequency O= OQJphat/Dn*, where Dis Eh'/12(1 ~ 2 ‘An 8X68 finite eloment mesh for the whole plate is again used in this example. As seen from the results the present elements do not lock in thin plate situations 4.3. Free vibration analysis of a simply supported square plate (orthotropic) ‘The numerical results fora simply supported orthotopic square plate are presented in Table 2 as test case for further validation, The following orthotropic meterial properties, typical of aragonite crystals have been used [39] Ej 143.52GPa, E,= 75.38 GPa, Gi» = 42.03 GPa, Gy, = 25.56 GPa, Gzs = 42.65 GPa, riz = 0.44, 21 = 0.23. The value of pis arbitrary because ofthe non-dimensionalization used (set to unity here). The elastic constant Cy used in “Table 2 has the value of 159.85 GPa. The results from the 88 mesh over the fll plate are compared with the exact solutions from 3D elasticity theory [41], higher-order theory [G9] and classical plate theory [41]. For lower modes, the results from the 9 node element are close to the exact elasticity solution. The results from the 4 node element are also quite close to the 3D elasticity solution. As seen from the previous example, CPT overestimates the frequency. ‘The results could be improved by refining the mesh 4.4. Free vibration analysis ofa simply supported eross ply (0/90/80/0) square plate In the fourth example, one set of dimensionless material properties (typical of high modulus graphite epoxy) is used siz AK Mayak etal. / Composites: Pat B 33 (2002) 505-S19 "Normalisd non-Limensional fundamental frequency ofa simply Supported square pate 1.044 normalised frequency 2 40 a0 80 Lenath fo thickness ratio == Present @ Se Present 4 rr a rT ee) @ "Normalised non-dimensional fundamental frequency of a clamped square pate rermalised froquency os" a a aT) =e Present Se Present 4 120 140~=~«t60~=~=C«BO~C«2O ‘Length fo thickness ratio Fig. 4. Nor-dimensionl normalized fundamental fequeney = A/a! /Dr of (2) amply supported square plat () clamped square pat vs ncresing ath to. to study the dynamic behaviour of laminated composite plates. Orthotropic layers with the following engineering constants [39] are used. E/E, = 40.0, Gas/E, = 0.50, G2/ Ey = Grs/Ez = 0.6, v2 = 0.25. The normalized smal- lest fundamental frequencies [39] as a function of plate side to thickness ratio are tabulated in Table 3. The non- No.oflayers Source 6i/Bs , 0 OM 3 Buact 26474 3.2641 38241 41089 43006 Present4 26283 32608 37053 39469 41109 Pretnt 9 26258 32515 34978 3.0413 41053 CLT 29198 1261 54043 6.4336 73196 5 xact_ 2.6587 34089 39792 43140 45374 dieu 33782 39351 426% 4aase Prevent 9 2.6367 33727 39790 42576 44789 CPT 29198 4.1264 5.4043 6.4336 73196 ° Exact 2.6640 34422 4.0547 44210 4.6679 Prevent 4 26412 33782 39351 42638 44852 Present 2.6367 33727 39290 42576 44789 CLT 29188 4.1254 54068 6.4336 73196 dimensionless fundamental frequency have been examined. The following engineering constants have been used: E/E, = vatiable, Gya/E, = 050, Gya/E, = Gys/Ey 0.6, vq = 0.25. The results are given in Table 4. Again resulls from the formulations in this paper are compared ‘with 3D elasticity theory [42] and classical lamination plate theory (CLPT). The present results are in good agreement with the 3D elasticity solution which are obtained by using a 6X6 mesh in a quarter plate. The error in the CLPT. predictions is mainly attributed to the neglect of shear deformation, Further mesh refinement could yield stil better accuracy. 4.6, Free vibration analysis of a three layer sandwich plate with a soft core In the sixth example, @ problem has been selected for Which an experimental result is available. The problem deals with a three layer rectangular simply supported sandwich plate. The following geometrical and material properties have been used [43 For face layers: Fy =F;=68.9GPa, 2 =0.30, p= 2.77 10° N sim For cote layer: Giz = 0.134 GPa, Gzs = 0.052 GPa, 0.122% 10° N/m‘. Geometrical properties: hg = 4.06% 10-* m, he, 183m, b 0.0064 m. 1.22 m, ‘The results from the present formulation along with HSDT (44), analytical [43] and experimental results [43] are tabulated in Table 5. The results from a 6 x6 mesh in the full plate are quite close to the analytical solution, although the results could be improved by increasing the mesh size, sia AK: Maya era. /Composts: Part B 88 (2002) 505-519 ‘Teble S Natural fequencies of tre layer simply suppred ecanguar plate Modes Analytical Experimental HSDT Present Present 9 ' 2 = none na 2 4“ 45 4 4510 45.0 3 n e nm 740 71.44 4 0 ® so 8619 81.26 5 or 2 39636 92.56 6 ms 19 2813588 12830 7 19 133 m9 ast na73 se 132 1311643615040 9165 169 im we 17108 0m 7 ~ 97m 7980 ‘Tabe 6 Dimensionless aatual frequencies = Me! /hjpeTE for « square sandwich plat with a/h= 10 and h/t = 038 Modes ap0/acee/900 857 — 451450 = 45145) — 45 HSDT Prescot Present HSDT Present 4 Preset 9 1 1528 15341504638 16431609 2 6 IS 8102968117 9D 3 MOL 319 920 HS 3117S 4386 A094 3776 a0.00 278386, 47. Free vibration analysis of seven layer composite sandwich plate using PVC foam in the core ‘The aim of this example is to compare the accuracy of the present formulation with an analytical solution [8] when applied to FRP (fibre reinforced plastics) sandwich plates made with PVC (polyvinyl chloride) foam core, The dimensionless natural frequencies are calculated for two square sandwich plates, with (0/90/0/core/0/90/0) and the (45) ~ 45/45/core/ — 45/45/ ~ 45) lay-ups, with the skins ‘made of glass polyester resins and the core of HEREX €70.130 PVC foam. The material properties of ui lirectional glass fibre in a polyester resin matrix are [8]; 24.51 GPa, E = 7.77 GPa, Gi 1.34 GPa, p, = 1800 kg/m’, 1, rial properties of the range of HEREX C70 foam core products are [8]; HEREX C70.40: £, = 22.69 MPa, 4 MPa, p. = 40 kel’, y. = 0.32. HEREX C70.55) 22 MPa, p. = 55 kg/m’, v, = 0.32. HEREX 70.75: E, = 56.95 MPa, G.=30MPa, 2 32. HEREX C7090: E, = 73.11 MPa, 90 kgfm?, 2 = 0.32. HEREX C70,130: 1, Ge=50MPa, p. = 130 kg/m’, 0.32. A.6% 6 mesh in the full plate is used, and the results are compared with analytical results [8] in Table 6, The Noncimensional fundamental frequency r 8 Noniimensional fundamental frequency 03 ry os 08 oF oe 09 ho © Fig. 5 tect of coe thickness on fundamental frequency of 0/0eoreAVBOM0 (8) heavy cone () light cove AK. Nayak eral / Composites: Part B 33 (2002) 505-519 results from the present formulation are quite close to the analytical results, Based on the foregoing convergence and comparison studies, it can be concluded that the present elements give very accurate vibration results for the test problems ‘considered. Since the 9 node element gives excellent results ‘even for a 2X 2 mesh as seen in problem 4, this element ‘used to generate new sets of vibration frequencies of composite sandwich plates with laminated facings and it, will also be used to investigate the effects of various design parameters on the frequencies. 48. Effect of plate thickness and material properties Simply supported square sandwich plates with PVC foam core and cross ply layers for the facings (O/90/0/ core/0/900) have been used to investigate the effect of core thickness and the total plate thickness on the frequency parameter (2a? /hy/p./E.). The length to thickness (a/h) ratio is taken to be 5, 10, 20, 50 and 100. Two different materials for the core and the same material of glass fibre for skin are selected. As discussed in the previous example, the light core material selected is, HEREX C70.40 and the heavy core material is HEREX €10.130. By considering these two material types, the sis vibration behaviour of sandwich plates with a heavy core against their counterparts with a lighter core may be examined. Fig. 5(a) and (b) shows the effects of varying core thickness (hh) and total length-to-thickness ratio on the fundamental frequency of vibration. A 6% 6 mesh over the full plate has been adopted to generate results. The vibration behaviour is similar for both kinds of sandvvich plates with slight variation in magnitudes of frequency. As shown in the figures, the fundamental frequency parameter for thick plates (a/h = 5, 10) will first decrease and later Increase as the cote thickness ratio (i/A) increases. This js due to the fact that for thick plates. as the core thickness increases, the plate becomes less stiff leading to a lowering of the frequency value but at the same time the plate is also getting lighter which causes the frequency value to increase. Thus, there is @ trade off between plate rigidities and mass effects on the natural frequencies for this ease However, for relatively thin plates (a/h = 20, 50, 100), the frequency parameter inereases monotonically with inereas- ing core thickness ratio because the effect of decreasing ftexural rigidity is overshadowed by the effect of decreasing mass density. As seen in the figures in case of thin plates (alk = 50 and 100), the frequency parameter variation differs slightly with inereasing core thickness. ‘There is not much gain in going from a/h = 50 to 100. "Nondimensional fundamental frequency Nondimensionalfundemental frequency 08 or 05 oh Fis 6. Eect of cor thickness on fundamental frequency of 45I-45/4SIeore! ~ 45/4 ~ 45 (2) avy core (light core sis c F 1 s sc er F s © ¢ © ® © F eee Js a é @ © Fig. 7. Dierea boundary condiors (a) SSSS, (b) CCC, (e) CFE, (ECR, @) SCS. The effect of core thickness on the fundamental frequency of angle ply laminates (45/ ~ 45/4Sicore! ~ 45/ 45/ ~ 45) is investigated in Fig. 6(a) and (6). The same observation as that of cross ply laminates holds good. However, there is a slight increase in fundamental frequency parameters in angle ply laminates when com- pared to cross ply laminates even with same materials. This demonstrates the significance of laminate orientations in a ‘composite sandwich plates. AK. Nayak eal / Composite: Part B 33 (2002) S0S-519 4.9, Effect of boundary conditions To study the effect of boundary conditions on the vibration frequencies, five combinations of free, simply supported and clamped edge boundary conditions, desig- nated as SSSS, CCCC, CFFE, CFCE, CSCS have been ‘considered as shown in Fig. 7. The leters F, § and C denote free, simply supported and clamped edges, respectively, The square sandwich plates analysed here have length-to- thickness ratio of 5. The PVC foam core of the plate is sandwiched by two symmetric cross ply facings (0/90/0/ core/0/90/0). The heavy and light core materials mentioned in the previous example are used again. The influence of boundary conditions on the fundamental frequency is shown in Fig. 8(a) and (b), It can be seen that the fundamental frequency increases with greater geometric constraint imposed at the four edges in the following sequence (CFFF, CFCF, SSSS, CSCS, CCC). Among these edge conditions the CCC plate has the highest fundamental frequency which is about 5-12 times higher than its CFFF ‘counterpart depending on the plate thickness ratios. It can also be observed that as f./h increases, the frequency parameter decreases up to h/t = 0.70 for all plates, and then increases. For the light core, it decreases monotoni- cally. The fundamental frequency of SSS boundary ‘condition is higher than that of CFFF condition. It appears Effct of core thickness on fundamantalrequency /0D/O/eoreK/90/0 rT 3 a 8 8 i | | "Noneimensional fundamental requency 2 Fig. 8. Bier of core thickness on diferent houndry conditions S888, CCCC, CFFF, CECE, CSCS, AXE Nayak et al. / Composites: Part B 33 2002) 505-519 su 100 “E> Manufacturer date sof |“ Line of eet at 60) 40} ‘Shear modulus in MPa 20| 0 40 o 20 70 a0SSCSSCSSSSCD Flexural modulus in MP2 a 280 g TE Manatectirer + Line of best ft Densty in Kai gis 8 ee 0 «0 60 0, 0 0SCADSC*~«sRSSC Flexural modulus in MPa Fig. 9, Star modulus and density as a function of fal modus for he range of HEREX C70 prodvets Elect of Boon dmensionossfequoncy Nondmeraionalrequancy Ecin Mpa Fig. 10, Elect of Bon dimensionless fequeney. sis AX. Nayak ah. / Composites: Prt B 33 (2002) 505-519 Tobe? Effect of pate spect ratio (fb), length o thickness rio on (lh), on menscnless natural frequencies 1 = Maia. ora square sandwich pate with A jh=08t ah Modes ‘af, bUvtcor/90 al 45) — A514Sleontts— 45/48 as 10 1s 20 os 10 1s 20 s 1 729 9a 1236 sss 733 1262 13.80 2 943 1356 74 1995 30 1799 20.18 3 1020 1396 2041 2041 68 2M 232 10 1 naa 1508 21.08 2807 1108 1614 2204 2495 2 1505 28.10 3286 am 1 an 3350 3876 3 2041 2020 fos wos 2 m2 829 son 20 1 ass 23 27 4165 1x68 212 sno aa38 2 24 4170 si 028 2161 363 sam 6159 3 29115 4438 nt S168 2219 4509 ‘nas 33 30 t 1460 2049 aun 4129 1436 aaa 3590 sit 2 2050 4136 5065 ey Bal saat ers B03 Al iss 5203 8573 11016 3602 5235 eR 1381 0 1 1490 2099 2290 4991 Me 2418 sre sis6 2 201 2999 5390 n00 2417 5336, 63.96 86.46 3 sad 5558 95.26 2364 317% 5590) 9791 a7 20 1 soe 210 3350 5129 uaz mss 3350 56.38 2 21.26 5137 6621 sos mse 55108 ne 098 3 sans ara 100.19 13191 sas sre 324 13809 that for thick plates, the simply supported constraint is more effective than having CFFF, CECF. 4.10. Influence of material properties on the natural frequencies ‘This part concerns with the study of the influence of a change in core modulus on the natural frequencies with (ish = 0.88). Fot a given range of products, such as the HEREX C70 foam, the density and shear moduli vary linearly as a function of the flexural modulus as shown in Fig. 9(a) and (b). Using the characteristics of the HEREX ‘Table 8 : Dimensionless natural frequencies =u? /pcTE, for 9 square sandwich No. oftayers Moves adicore00 5 1 385 1983 1503 1756 1998 2 651 1843 2843 2985 3253 a 1562 26402878 31503288 4 589 2885 3755 3994 4214 5 Wm 3098 4082 4082 4872 avv00/corens00 7 1 36 1352 1504 1758 2001 2 643 1521 2810 3023 3223 3 1s07 26832920 la 3534 4 1539 28323776 3996 27 5 94330354082 4082486 70, the density and shear moduli of the core material for any material for any value of E, arc obtained by interpolation. Knowing these parameters, itis now possible to obtain the data required for the analysis. Fig. 10 shows that an increase in the value of E results in an increase in the natural frequency of mode (1,1) for ab = 1, afh = 5 and ‘hclh = 0.88. Similar observations have been made with other boundary conditions. 4.11. Tabulated results for composite sandwich plates As there is a dearth of vibration results for composite pplates with laminated facings, the first frequencies of sandwich plates composed of a PVC core and (0/90/0) cross ply facing are presented in Tables 7 and 8, In particular, Table 7 gives the results for SSSS plates with 0.5, 1.0, 1.5, 2.0 and length to thickness 5, 10, 20, 30, 40, 50, Table 8 gives the natural frequencies of CCC, CFFF, CFCF, CSCS with an aspect ratio a/b = 1.0 and width to thickness ratio = 10, These tabulated frequency parameters should be useful as bench- ‘mark solutions for researchers who are developing numeri- cal techniques and software for solving sandwich plate vibration problems. 5. Concluding remarks In this paper, two new Reddy type elements with the assumed strain approach have been developed. AX. Nayak etal / Composites: Part B 33 (2002) 505-519 519 (1) Based on the Reddy's higher-order theory, these clements have been implemented with a very simple ‘and understandable mathematical framework and are easily programmed. (2) To control spurious zero energy modes, the assumed strain method as suggested by earlier investigators has been implemented in the present formulation (3) In terms of the variational principle, the resulting equations are similar to those derived from FSDT. except that higher-order terms are added to replace the shear correction factors. (4) The applicability of the Reddy type elements has been demonstrated over a wide range of free vibration problems, with varying material combinations, geo- metric features, and boundary conditions. References [1] Phatema FI, Sandwich construction: the bending and bucling of sandvich beams, plates and shells, New York: Wiley; 1966 [2] Allen HG. Analysis and design of totaal sandwich panels. Oxford: Pergamoa Press 1959, Zenkest D. An ntodution to sandwich construction. West Midlands, UK: EMAS Publication, 1995, ‘Vinson JR. The behaviour of sandwich sities of fotopis and composite materials, USA: Techno Publishing Company: 1999. Bose P, Reddy IN. Analysis of composite plates using varius plate thoores, Par I, Formulation and analytical solaions. Swuct Engng Mech 1998:6(6,583-612 Bose P, Redy IN. Analysis of composite plates using various pate theories, Pat 2 Finite element model and numerical results, Stuck Engng Mech 1998;(7): 727-46. (71 Yang PC, Nomis CH, Stavsky Y, Blac wave propagation ia heterogeneous plates nt] Solids Strut 1966:2:665-84 {8} Meunier M, Shenei RA. Free vibration analysis of composite sandwich plas. Proc ImechE, Pant C.J Mech Engng Sei 1999; 21340) 715-21. (9) Viachoutsis §. Stear conection factors fr plates and shel. Int J Namer Meth Engng 1992:33(7:1537-52 [0] Reddy IN. A simple higher-order theory for laminatst composite pints. App! Mech Trans ASME 1984 51(¢).745~52 (11) Reddy JN. A refined nonlinear teary of plates with transverse seat ‘deformation It J Solids Stet 19842009-10):881-96 12} Lewinski T. A not on recent development inthe theory of elastic plats wih moderate thickness Rorpeawy la 1986:34(4:53 (03) Robwer K. Application of higher-order theories to the beading analysis of layered composite plates. at J Solids Suet 1992:29(1): 15-19. (14) Noor AK, Burton WS, Bet CW. Computational models for sandwich panels ad shes. Ap! Mech Rey 199649031559, (15) Yang ITY, Suiza S Masad A, Kapaa RK. A survey of recent shell finite element Int Numer Meth Engng 2000/7(13):101 27 116} Khatua TP, Cheung YE. Triangular element for muki-ayer sandwich plates. J Engng Mech, ASCE 197298:1225-28 [071 Khatua TP, Choung YK. Bending and vibration of mali-ayer sandwich beams and plates Int J Numer Meth Engng 1973:6- 1-24 [18} Zieakiewiex OC, Taylor RL. Too IM. Reduced integration technique ingeneral alysis of plates ad sels, IJ Numer Meth Engng 197 3275-90 [U9] Hughes THR, Cohen M, Haroun M, Reduced an selective iteration techniques inthe inte element analysis of pecs. Nul Engng Des 1978:46:203-22, 0 ia 15) 6 (20) Hughes THR, Teaduyar TE. Finite elements based upon Minin plat ‘heory with pacieulr reference tothe four node bilinear soparametc clement. J Appl Mech, Trans ASME 1981;¢8:581-96, (21) MacNeal RH. simple quadelateral shell element, Comput Stet 1978 :175-83. (22) Dvorkin EN, Bato KJ A coztinuum mechanies based four nod sell clement for general ontnear analysis. Engng Compt 19841:77—. (23) Huang HC, Hinton E. A nine node Lagrangian Mindia plate element with hanced shear interpolation. Engng Comput 1984136879 [24] Simo JC, Hoghes TIR. On the vesitional foundations of assumed strain metho. J Appl Mech, Trans ASME 1986:5:51—4 [25] Hinon F, Huang HC. A family of quadslatral Mindlin plate

You might also like