These material parameters can be derived for anyconfiguration of sandwich construction (see, forexample, [2, 9).
The type of nodal degrees of freedom
Simple finite elements, possessing only the basicgeometrical nodal degrees of freedom (e.g. mid-planetranslations and section rotations which are easilytransformed) are suitable for the analysis of three-dimensional systems composed of panels meetingat an angle, e.g. folded plates, and panelizedconstruction. Finite element models with additionaldegrees of freedom such as curvatures, higher orderderivatives of displacements or shear strains aretherefore more suitable for the analysis of flat platesor shells. Also, these elements can be very accurate,particularly when interlaminar shear stresses are ofspecial interest.
Since interlaminar shear stresses are continuousacross layers, they should not be evaluated using anindividual layer’s elastic constants, but rather byother means, such as equilibrium consideration, orby using higher order assumed-stress elements. Adifferent problem, involving ‘concentrated’ forces,where the local transverse shear stresses are discon-tinuous, can also be tackled by more sophisticatedelements.An additional complication arises in the case ofsandwich construction with unsymmetrical layers,of which half-sandwich is the simplest type, wherebending and in-plane stretching are coupled(although such coupling tends to reduce the plateeffective stiffness, other considerations may prevail).In finite element analysis, this coupling must beincluded at the formulation stage, since a simplesuperposition of the bending and stretchingbehaviours is invalid for this case.In general, finite element solutions for transverseshear stresses and edge reactions are less accuratethan those for deflections and moments; this isparticularly true in the displacement formulation ofsandwich and Mindlin plate elements.The specification of boundary conditions for manyrefined elements can be complicated because of thenumerous degrees of freedom assigned across theplate thickness. For sandwich construction with rigidedge inserts or stiffeners, compatibility requiresspecial consideration. It is unfortunate that develop-ers of finite element models rarely address thesepractical matters.
Assumptions in sandwich plate formulations
The variety of assumptions used in sandwich plateformulation can be confusing. Obviously, the moresophisticated formulations make fewer assumptionsat the cost of more complexity and analysis-effort.A paper by Cook  investigates the effects ofsome of these assumptions for three-layer sandwichplates of isotropic facings. The various degrees ofapproximation are classified below to facilitate laterreference.1.
These are common tomost formulations: (a) the transverse deflection isconstant across the thickness; and (b) perfect bondingbetween layers (the effects of ‘elastic interlayer slips’on sandwich beam-columns were discussed by Fazio
et a1.).2. Materialproperties.
Materials for the facing andcore can be: (a) isotropic; (b) orthotropic; or (c)anisotropic.3.
Stress distribution in facing section.
(a) Thinfacing theory: the face is treated as a solid membrane,i.e. constant a,, by,
(b) Thick facing theory: theface is treated as a classical thin plate to allow forstretching and local bending, i.e. linear distribution ofQ,, aY, rxY.c) All five
components o,, au,
are defined, and are often allowed to varyacross the layer thickness.4.
Stress distribution in the core section.
(a) Flexiblecore: constant transverse shear stresses
(b)Stiff core: similar to 3(c) above.5.
The approximations sofar are concerned with individual layer behaviour.Since the layers are bonded together, certain quan-tities such as interlaminar shear stresses and displace-ments are continuous across layers. This, combinedwith the need to enforce strain compatibility withinthe layers (as required in the displacement finiteelement formulation), necessitates an assumed kin-ematic deformation mode for the plate section. (a)Linear variation (Fig. 1): the normal remains straightacross the entire cross-section of the plate, but notnecessarily perpendicular to the middle plane. (b)Piece-wise linear variation with partial continuity(Fig. 2). (c) Piece-wise linear variation with fullcontinuity (Fig. 3). (d) Quadratic or cubic variationto allow for warping of the cross-section.The discontinuity of the inplane displacements atthe layer interface shown in Fig. 2 arises from theneglect of transverse shear deformations in the stifflayers. In addition, if the small displacement u* in thestiff layers is considered then we have the so-calledthick facing theory [i.e. assumption 3(b)], otherwisethe theory is for thin facings [i.e. assumption 3(a)].This discussion also illustrates the interdependencybetween the various assumptions.
Linear displacement variation.