You are on page 1of 7

Engineering Analysis with Boundary Elements 27 (2003) 603–609

www.elsevier.com/locate/enganabound

Plate bending analysis using the classical or the Reissner –Mindlin models
L. Palermo Jr.
Faculty of Civil Engineering, University of Campinas, C.P.6021, Campinas 13084-970, Brazil
Received 12 December 2001; revised 8 September 2002; accepted 12 November 2002

Abstract
Plates can be solved with the classical or the Reissner – Mindlin plate model using the same computer code with an appropriate treatment of
the direct boundary element formulation. A field decomposition is employed to establish a connection between these plate models and to
obtain a boundary element formulation for the classical model from that used for the Reissner – Mindlin one. The classical model is related to
the irrotational component of the field related to the rotations of the plate. The use of this field component in both the fundamental solution
and the direct boundary integral equation of the Reissner – Mindlin model carries the classical model approach. Furthermore, the well-known
fundamental solution derived from the classical model can be employed as the irrotational field of the fundamental solution of the Reissner –
Mindlin model. In this way, the fundamental solution used by Danson to perform classical analysis was used as the irrotational component
and the obtained results were compared with those obtained with the well-known Weeën’s formulation.
q 2003 Elsevier Science Ltd. All rights reserved.
Keywords: Reissner–Mindlin model; Plate bending; Classical model; Fundamental solution

1. Introduction contributions from Tottenham [4], Danson [5], Bézine [6]


and Stern [7]. Mukherjee [8] and Paiva [9] used the classical
The classical plate bending model is written in terms of theory with an additional degree of freedom for the
the out-of-plane displacement and its derivatives or an tangential boundary rotation in a boundary integral
irrotational field representing the behavior of the plate [1]. formulation and showed that many difficulties associated
The integral equation contains four boundary parameters with the treatment with two-boundary parameters of this
and it is necessary to satisfy two-boundary conditions to theory could be overcome. Weeën [10] introduced a BEM
obtain a single solution. A concentrated reaction must be formulation applied to Reissner’s model; Vilmann [11] and
placed at each corner as an additional parameter in the De Barcellos [12] used similar formulations to treat
boundary value problem when the boundary of the plate is a Mindlin’s model. The boundary integral equation (BIE)
non-smooth line. The inaccuracy of the classical model used to perform those analyses contained six boundary
turns out to be significant in the edge zone of a plate and parameters and no additional unknown had to be introduced
around holes that have a diameter not larger than the when polygonal plates were considered. Katsikadelis [13]
thickness of the plate. In order to overcome the above- presented a BEM formulation for Reissner’s model, which
mentioned features due to the one-displacement dependence was expressed in terms of two potentials, one biharmonic
of the classical model, Reissner [2] and Mindlin [3] and one Bessel potential. The plates analyzed with that
developed similar models considering the shear deformation strategy needed three integral equations plus three finite
effect. The curvatures have not been directly related to the difference equations. The fundamental solutions for the
out-of-plane displacement derivatives in the constitutive Reissner – Mindlin models and the relations with the
equations and three boundary conditions should be satisfied classical model have been studied by Westphal et al. [14,
in the boundary value problem rather than the two of the 15]. It was shown that a general solution for the Reissner –
classical model. Mindlin model includes the solution for the classical model
The plate analysis with the classical model using the and free coefficients could be introduced in the general
boundary element method (BEM) received important formulation.
0955-7997/03/$ - see front matter q 2003 Elsevier Science Ltd. All rights reserved.
PII: S 0 9 5 5 - 7 9 9 7 ( 0 2 ) 0 0 1 3 9 - X
604 L. Palermo Jr. / Engineering Analysis with Boundary Elements 27 (2003) 603–609

2. Plate equations related to effect of the assumed s33 distribution along the
thickness and not with the shear deformation. Mindlin [3]
The expressions will be presented with the Latin indices has justified the absence of this term because the linearly
in the range {1 – 3} and Greek indices in the range {1, 2}. A weighted average effect of s33 was neglected rather than s33
plate of uniform thickness is referred to midline coordinates itself with reference to Reissner’s model. It must be
xa and thickness coordinate x3. The transverse shears Qa ; remembered that the average expressions for the displace-
the bending and twisting moments Mab ; all per unit length, ments have been introduced in the second version of the
have similar definitions in both theories with reference to Reissner model [16] and the constitutive relations were not
stresses. In the classical theory, the constitutive relations changed (Eqs. (4) and (5)). It has carried to effective values
written in terms of deflections are of s33 equal to ^ 0.5q at x3 ¼ ^h=2; respectively, or the
same values assumed at the surfaces in the classical or the
Mab ¼ 2D½ð1 2 nÞw;ab þ ndab w;gg ; ð1Þ
Mindlin’s model. However, the values of s33 should be q at
Qa ¼ 2Dw;gga ð2Þ x3 ¼ h=2 and zero at x3 ¼ 2h=2 according to the real
loading condition of the plate, which the distributed load q is
D ¼ Eh3 ½12ð1 2 n2 Þ21 ; ð3Þ a traction on the upper surface in the x3 direction and all
others tractions on both surfaces are null. Further, Lévy,
D is the flexural rigidity, E is the elasticity modulus and n is
Boussinesq and Timoshenko [17] made interesting com-
Poisson’s ratio.
ments on the stress distribution and the constitutive
Reissner conceived his theory from an assumed stress
equations assumed by Reissner, including the s33 distri-
distribution. He introduced generalized displacement
bution along the thickness.
expressions with a weighted average across the thickness
[16] to get the deflection w of the plate in the middle surface
and quantities ba representing components equivalent but
3. Integral equations
not identical to the components of changes of slope. The
constitutive equations written in terms of displacements
The well known Weeën’s formulation [10] for Reissner’s
ðba ; wÞ and the distributed load q on the plate surface are
  theory is given by
12n 2n
Mab ¼D ba ; b þ bb ; a þ b d þ þ
2 1 2 n g;g ab 1
Cij uj þ Tijp uj dG ¼ Uijp tj dG
n 2
þq 2 d ; ð4Þ
l ð1 2 nÞ ab G G
ðð  
n
12n 2 þ Ui3p 2 Uipa;a q dV;
Qa ¼ D l ðba þ w;a Þ; ð5Þ V l2 ð1 2 nÞ
2 ð8Þ
l is a constant related to the shear effect and is equal to
pffiffiffi
10=h: Cij is the Kronecker delta for a smooth boundary, ua is ba ;
Mindlin’s model [3] was conceived from an assumed u3 is w, ta is the product Mab nb ; t3 is the product Qa na ; dG
strain distribution. The ‘equivalent changes of slope’ ba of and dV denote boundary and domain differentials, respect-
Reissner’s model were replaced by ca which were obtained ively. Uij represents the rotation ðj ¼ 1;2Þ or the deflection
without average expressions due to the assumed strain ðj ¼ 3Þ due to a unit couple ði ¼ 1; 2Þ or a unit point load
hypothesis. The integrations that determined shears, bend- ði ¼ 3Þ:
ing and twisting moments were performed but the   
p 1 2 1 1
coefficients of the integrals containing ga3 were replaced Uab ¼ ½AðzÞ 2 BðzÞra rb  2 dab lnðzÞ 2
by constants whose magnitude were obtained from wave 2pD 1 2 n 2 2

propagation analyses. The constitutive relations for Mind- 1
2 ra rb ;
lin’s model are: 2
 
12n 2n  
Mab ¼ D ca;b þ cb;a þ cg;g dab ; ð6Þ 1 1
2 12n Uap 3 ¼ lnðzÞ 2 rra ¼ 2U3pa ;
4pD 2
12n 2
Qa ¼ D l ðca þ w;a Þ: ð7Þ
2 
p 1 2 1 2
Mindlin showed that l is related to the Poisson ratio and the U33 ¼2 lnðzÞ 2 z ½lnðzÞ 2 1 :
2pDl2 1 2 n 4
vibration mode including the transverse shear deformation
and the rotatory inertia effects. In the present study, l will be With z equal to lr and
assumed as p=h independently of the Poisson ratio.  
It should be noted in Eq. (6), in comparison with Eq. (4), 1 1
AðzÞ ¼ K0 ðzÞ þ K1 ðzÞ 2 ;
that the distributed load effect is not included. This term is z z
L. Palermo Jr. / Engineering Analysis with Boundary Elements 27 (2003) 603–609 605

  The solution used by Bézine – Stern is obtained from Eq.


2 1
BðzÞ ¼ K0 ðzÞ þ K1 ðzÞ 2 ; (14) when k is set equal to zero or l is equal to e (the base of
z z
natural logarithms) and that used by Danson when k is equal
Kn is a modified Bessel function of order n. to 0.5 or l is e 0.5. The fundamental solution for a unit
It is necessary to remember that the integrand of the couple in the classical analysis is obtained from a derivative
domain integral in Eq. (8) contains the additional factor of the fundamental solution for a unit point load, as
from Reissner’s theory related to the distributed load Timoshenko showed in Ref. [17].
(linearly weighted average effect of s33) that is null in When Eqs. (10) and (11) are compared, a connection
Mindlin’s model. This term should be disregard to analyze between the classical and Mindlin models can be noted. The
plates using Mindlin’s model: boundary integral Eq. (11) could be obtained from Eq. (10)
ðð  n
 ðð when an irrotational approach for rotations is considered or
Ui3p 2 Uipa;a 2
q dV ) Ui3p q dV: ð9Þ when un and us are made equal to 2w;n and 2w;s ;
V ð1 2 nÞl V
respectively. The other end of the connection can be
The boundary integral equation for Mindlin’s model with established when the field decomposition [1] is taken into
the boundary parameters written with reference to normal- account in the fundamental solution. Thus, a vector
tangential (ns) directions is given by containing the rotations of the plate can be written as [1]
þ    
1 ›f ›f ›H ›H
Cij uj þ ðMni un þ Mns
i
us þ Qin wÞdG ð c1 ; c2 Þ ¼ ; þ ;2 ; ð15Þ
2 ›x 1 ›x 2 ›x 2 ›x 1
G
þ ðð f and H represent functions whose components are
¼ ðMn uin þ Mns uis þ Qn wi ÞdG þ wi q dV; ð10Þ independent of x3 [1].
V The obtained functions wp, f and H for a unit couple Fa
G
are given by [1]
uj is related to the fundamental solution used and
1 › 2
corresponds to the deflection for the solution due to a unit wp ¼ ½r ðlnðlrÞ 2 1Þ ¼ 2f; ð16Þ
point load ði ¼ 3Þ or the rotation for the solution due to a unit 8pD ›xa

couple (i ¼ 1 or 2). un and us are cn and cs, respectively. 1 ›
H¼ F10 ðlnðlrÞ þ K0 ðlrÞÞ
The integral equation for the classical model with two pDð1 2 nÞl 2 ›x2
unknown boundary parameters can be written as 

þ 2 F20 ðlnðlrÞ þ K0 ðlrÞÞ ; ð17Þ
1 X a ›x 1
Cab ub þ ðVna w 2 Mna w;n ÞdG þ Rci wci
2 Nc Fa0 can be understood as the result from domain integration
G
on a circle containing the Dirac-delta force in the direction a
þ X ðð (it is equal to 1) and it was introduced to distinguish each
¼ ðVn wa 2 Mn wa;n ÞdG þ Rci waci þ w a q dV ;
Nc V couple effect, see Ref. [1].
G
The functions f and H are obtained from derivatives of a
ð11Þ biharmonic function and a combination of a harmonic
Cab is the Kronecker delta for a smooth boundary, ub is the function with the solution of a Bessel’s equation, respect-
deflection when the solution for a unit point load ða ¼ 1Þ is ively. They are according to Reissner’s studies on plate
considered or the rotation for the unit couple solution ða ¼ differential equations with shear deformation effect that are
2Þ: The equivalent shear Vn is introduced in the integral d’Alembert – Euler (Cauchy – Riemann) equations. Thus,
equation and the corner reaction Rc should be considered this paper presents the fundamental solution composed
when the boundary line has corners. The corner reaction Rc from a scalar and a vector potential (two potentials) that are
at corner i is the difference between twisting moments at the according to Reissner’s studies presented in Ref. [2].
corner neighborhood on the forward side Mns F
and the The obtained functions wp and f for a unit point load are
backward side Mns : B given by [1]:
 
F B 1 1 2 lnðlrÞ
Rci ¼ ðMns 2 Mns Þ; ð12Þ wp ¼ r ðlnðlrÞ 2 1Þ 2 ; ð18Þ
pD 8 ð1 2 nÞl2
›Mns
Vn ¼ Qn þ ¼ 2D½w;gga na þ ð1 2 nÞw;nss : ð13Þ 1 2
›s f¼2 r ðlnðlrÞ 2 1Þ: ð19Þ
8pD
A biharmonic function is used as the fundamental solution
The function f is related to the irrotational field and it can
for classical analyses and a general expression for this
be shown that the solution of H is null due to the radial
solution is given by:
symmetry of the unit point load problem on an infinite plate.
1 2 1 2 The solutions obtained by Weeën can be found if the
wp ¼ r ½lnðrÞ 2 k ¼ r ½lnðlrÞ 2 1: ð14Þ
8pD 8pD results shown in Eqs. (16) – (19) are introduced in Eq. (15).
606 L. Palermo Jr. / Engineering Analysis with Boundary Elements 27 (2003) 603–609

Further, the fundamental solutions used to perform analyses the Reissner –Mindlin’s theory. When expression (20) is
with the classical model could be employed as functions for introduced in the relations for the transverse shear (7), they
the irrotational field of the fundamental solution of the can be written as
Reissner –Mindlin model [1]. The fundamental solution will  
1 2 n 2 ›H ›H
be written next with reference to normal-tangential (ns) ðQ1 ; Q2 Þ ¼ D l ;2 ;
2 ›x 2 ›x1
directions to be used in Eq. (10). It should be noted on the
complete expressions written with reference to normal- and
tangential (ns) directions that they are the same presented in      
›H ›H ›H ›H ›H ›H
the literature but the corrections introduced by the shear ;2 ¼ n þ s ;2 n þ s :
›x 2 ›x 1 ›n 2 ›s 2 ›n 1 ›s 1
deformation effect are clearly identified when the field
decomposition is used. The transverse shears Qa can be written with reference to ns
directions:
3.1. Solution for a unit point load  
1 2 n 2 ›H ›H
ðQn ; Qs Þ ¼ D l ;2 : ð21Þ
The bending and twisting moments have similar 2 ›s ›n
expressions to those used for the classical model because The relation obtained in Eq. (21) can be used in Eq. (20), or:
the solenoidal field is null ðH ¼ 0Þ and their expressions  
›w ›w 2
written with reference to ns directions are given by: ð cn ; c s Þ ¼ 2 ; 2 þ ðQn ; Qs Þ: ð22Þ
"  2 #
›n ›s Dð1 2 nÞl2
p 1 ð1 2 nÞ ›r
Mn ¼ 2 ð1 þ nÞlnðlrÞ 2 þ ð1 2 nÞ ; Expression (22) gives a physical meaning for the correction
4p 2 ›n introduced by the Reissner –Mindlin’s model. The final
p ð1 2 nÞ ›r ›r expression for the transverse shear Qn due to a unit couple in
Mns ¼2 : the a direction is given by:
4p ›n ›s
    
The transverse shear defined in Eq. (7) can be written with 1 1 ›r 2 K1
Qn ¼ 2 r 2 na þ l K0 þ na
reference to the normal direction (n) and the following 2p r 2 ›n a z
  
expression is obtained ðQn ¼ Qa na Þ : K ›r
2 K0 þ 2 1 r : ð23Þ
1 ›r z ›n a
Qpn ¼ 2 :
2pr ›n The first term in expression (23) is equal to the transverse
The rotations cn and cs (or un and us, respectively) are given shear due to a couple obtained from the classical model. The
by: second term or the term in square brackets is the correction
  due to Reissner – Mindlin’s hypotheses.
1 1 ›r The bending and twisting moments are given by
cpn ¼ 2 r lnðlrÞ 2 ;
4pD 2 ›n " #
  ›2 w 2 ›Qn
1 1 ›r Mn ¼ D ð1 2 nÞ 2 2 7 w þ 2 2
; ð24Þ
cps ¼ 2 r lnðlrÞ 2 : ›s l ›n
4pD 2 ›s
 
The main difference when comparing with the classical ›2 w 1 ›Q n ›Qs
Mns ¼ 2Dð1 2 nÞ þ 2 þ ; ð25Þ
model appeared only in Eq. (18) for the deflection using ›n›s l ›s ›n
Reissner –Mindlin’s hypotheses in which the first term Eqs. (24) and (25) contain the classical relation (Eq. (1))
corresponds to the deflection obtained from the classical plus a correction introduced by the shear deformation effect
model and the second term is the shear deformation to remind an expression presented by Reissner [2].
effect. The deflection, rotations and efforts were written with
reference to boundary directions ns at field points and for a
3.2. Solution for a unit couple unit couple in the a direction at the source point. They can
be written with reference to a couple in the n direction using
Eq. (16) gives the deflection due to a unit couple in a a similar procedure as it is shown in the following example:
similar form to that obtained under the classical hypotheses.
The rotations are related to the deflection derivatives using wn ¼ wa na ; ð26Þ
Eq. (15) because the deflection w is 2 f in this solution. Eq. n
Mn ¼ Mna na : ð27Þ
(15) written in terms of ns and w is given by:
    Boundary directions at the source point are represented by n
›w ›w ›H ›H n
ð cn ; cs Þ ¼ 2 ;2 þ ;2 : ð20Þ and s in Eqs. (26) and (27). Mn and wn are the deflection and
›n ›s ›s ›n
the bending moment due to a couple in the direction n;
The notation presented in expression (20) shows respectively. Mna and wa are the deflection and the bending
the explicit connection between the classical and moment due to a couple in the direction a, respectively.
L. Palermo Jr. / Engineering Analysis with Boundary Elements 27 (2003) 603–609 607

The use of the field decomposition has shown that responses with the classical model would be obtained
Reissner –Mindlin’s model contains the classical model as with an irrotational approach of the plate behavior and
the irrotational field component. The boundary element the terms related to the shear correction would be
formulation allows the classical model to be checked with disregarded.
reference to Reissner – Mindlin model in the same computer
code if the present expressions are considered. Unless
mostly studies with the classical model employ the 4. Numerical example
boundary element formulation with two unknown boundary
parameters, the formulation for the classical model with an A square plate was analyzed with the classical and
additional boundary parameter could be more advantageous Reissner – Mindlin models. Linear shape functions were
to allow the switching between the plate models. The next used for the approximation of the displacements and the
equation is similar to that used by Mukherjee [8] and Paiva efforts on the elements in the numerical implementation. All
[9] to perform classical analyses with an additional nodal parameters were positioned at the ends of the
boundary parameter: elements and the collocation points were placed on the
þ boundary to perform analyses using the classical model or
1 the Reissner – Mindlin models. When discontinuous bound-
Cij uj þ ðQin w 2 Mni w;n 2 Mns
i
w;s ÞdG
2 ary elements were used, the collocation points were shifted
G
þ ðð to inside the element at a distance equal to a quarter of the
¼ ðQn wi 2 Mn wi;n ÞdG þ w i q dV : ð28Þ element length. Analytical expressions were employed
V
G when the integration was performed on the boundary
element containing the source point and Gauss –Legendre
The following features can be noted on the relations scheme with 10 points when the integration was performed
between Eqs. (10) and (28): on the boundary elements not containing the source point.
The results were not significantly affected if continuous or
† The integral Eq. (28) can be obtained from integral Eq. discontinuous elements were used or if the location of the
(10) whether cn and cs are set equal to 2w;n and 2w;s collocation points in discontinuous elements was changed.
and a soft restraint condition is used (Mns is released on Two meshes with double nodes placed at the corners were
the boundary); used: 32 elements (36 nodes) and 128 elements (132 nodes).
† The fundamental solutions for the classical analysis can The refined mesh (128 elements) was used in order to
be obtained from those used for Reissner –Mindlin’s compare results with those obtained with 64 quadratic
model if the shear deformation effect in expressions for elements in Ref. [18]. Eqs. (11) and (28) were used to
the displacements and efforts are disregarded; perform the classical model with the parameter used by
† Eq. (28) can be understood as a weak formulation of plate Danson. The responses from Mindlin’s model were labeled
models with shear deformation effect. Thus, the by Weeën and Danson when the value of l in the irrotational
Table 1
Values at the center of a simply supported square plate

Thickness w M11
Units
0.08 0.8 0.08 0.8
( £ 1022 m) ( £ 1025 m) (N m/m) (N m/m)

Classical Ref. [17] 0.7094 – 0.4905 –


BEM-Eq. (11) 0.7109 – 0.4908 –
0.7100a 0.4905a
BEM-Eq. (28) 0.7107 – 0.4906 –
0.7099a – 0.4904a –
Hard Mindlin/Danson 0.7079 0.8601 0.4886 0.4909
0.7111a 0.8590a 0.4903a 0.4905a
Mindlin/Weeën 0.7078 0.8599 0.4885 0.4908
0.7111a 0.8590a 0.4903a 0.4905a
Reissner/Weeën 0.7076 0.8358 0.4885 0.4969
0.7108a 0.8350a 0.4903a 0.4966a
Reissner [18] 0.7110 – 0.4904 –
Reissner [19] 0.7097 0.8351 – –
Soft Mindlin/Danson 0.7232 0.9688 0.4964 0.5479
Mindlin/Weeën 0.7230 0.9686 0.4963 0.5477
Reissner/Weeën 0.7226 0.9391 0.4963 0.5511
a
Values obtained with 128 elements.
608 L. Palermo Jr. / Engineering Analysis with Boundary Elements 27 (2003) 603–609

Table 2
Values at the center of a clamped square plate

Thickness w M11
Units
0.04 0.4 0.04 0.4
( £ 1021 m) ( £ 1023 m) (N m/m) (N m/m)

Classical Ref. [17]a 0.1761 – 0.2365 –


BEM-Eq. (11) 0.1769 – 0.2346 –
BEM-Eq. (28) 0.1769 – 0.2346 –
Hard Mindlin/Danson 0.1773 0.2103 0.2346 0.2373
0.1772b 0.2106b 0.2346b 0.2375b
Mindlin/Weeën 0.1773 0.2103 0.2346 0.2373
0.1772b 0.2106b 0.2346b 0.2375b
Reissner/Weeën 0.1773 0.2098 0.2347 0.2416
0.1772b 0.2101b 0.2346b 0.2418b
Reissner [19] 0.1772 0.2103 0.2346 0.2420
Soft Mindlin/Danson 0.1773 0.2113 0.2346 0.2380
Mindlin/Weeën 0.1773 0.2113 0.2346 0.2380
Reissner/Weeën 0.1773 0.2109 0.2347 0.2424
a
Values obtained from Table 35 on page 202 of Ref. [17].
b
Values obtained with 128 elements.

part was equal to p=h and e 0.5, respectively. A uniform load could be represented in the Reissner –Mindlin theory by
equal to 0.64 N m22 was applied on the plate surface, the a restraint on the twisting moment Mns at the boundary.
Young modulus was 2 MPa, the Poisson ration was 0.3 and On the other hand, the efforts distribution on the
the side of the plate had a length equal to 4 m. Table 1 boundary using the soft condition was closer to the
presents the results for a square plate simply supported in all classical approach than those obtained with the hard
sides with two thickness, h ¼ 0.08 and 0.8 m. The results condition. It is not an inconsistency but a consequence of
shown in Table 2 were obtained for a clamped square plate the classical hypothesis on twisting moments, which
with two thickness, h ¼ 0.04 and 0.4 m. The responses were enables the sum of the effect of variation of twisting
obtained at the center of the plate and two kinds of boundary moments with the vertical shear forces applied at
conditions related to the twisting moments were used for the same edge and introduces corner reactions for the
Reissner – Mindlin models: ‘soft’ when the twisting equilibrium.
moments were released on the boundary and ‘hard’ when
they were restrained. The results with the soft restraint
condition were presented with 32 elements because the Acknowledgements
differences on the response obtained with 128 elements
were similar to those shown for the hard restraint condition. The author is grateful to FAPESP for financial support of
The features presented in Tables 1 and 2 can be these studies on plates.
summarized as:

† The differences between results obtained from Reissner References


and Mindlin models were less than 0.1% on the boundary
[1] Palermo Jr, L. A study about fundamental solutions in plates. Brebbia
nodes and at the internal points for thin plates. The
CA, Power H, editors. 22nd Conference on the Boundary Element
linearly weighted average effect of s33, which has Method (BEM 22); 2000. p. 373–84.
appeared as an additional term in Eq. (4), introduced the [2] Reissner E. The effect of transverse shear deformation on the bending
main differences in the final values of the bending of elastic plates. J Appl Mech 1945;A69–A76.
moments for thick plates. [3] Mindlin RD. Influence of rotatory inertia and shear on flexural
motions of isotropic elastic plates. J Appl Mech 1951;March:
† The results with l equal to e 0.5 for the irrotational field
31– 8.
did not present significant differences from those [4] Tottenham H. The boundary elements for plates and shells.
obtained with l equal to p=h: Other boundary conditions Development in BEM, vol. 1. Barking: Applied Science Publishers;
were studied and the differences were less than 1%. 1979. 173–205.
† The results at internal points with the ‘hard condition’ [5] Danson DJ. Analysis of plate bending problems by direct boundary
element method, Southampton, UK, MSc Dissertation, University of
were closer to the classical approach than those obtained
Southampton, Southampton, UK; 1979.
with the ‘soft condition’, even for thick plates. This [6] Bézine G. Boundary integral formulation for plate flexure with
shows that the irrotational approach of the rotation field arbitrary boundary conditions. Mech Res Commun 1978;5:
at the internal points introduced by the classical analysis 197 –206.
L. Palermo Jr. / Engineering Analysis with Boundary Elements 27 (2003) 603–609 609

[7] Stern MA. A general boundary integral formulation for the numerical [13] Katsikadelis JT, Yotis AJ. A new boundary element solution of thick
solution of plate bending problems. Int J Solids Struct 1979;15: plates modeled by Reissner’s theory. Engng Anal Bound Elem 1993;
769–82. 12:65–74.
[8] Guo-Shu S, Mukherjee S. Boundary element method: analysis of [14] Westphal Jr T, De Barcellos CS. Reissner/Mindlin’s plate models and
bending of elastic plates of arbitrary shape with general boundary the boundary element method. Brebbia CA, editor. International
conditions. Engng Anal Bound Elem 1986;3:36–44. Conference on Boundary Element Technology; 1992. p. 589–604.
[9] Paiva JB, Neto LO. An alternative boundary element formulation for [15] Westphal Jr T, Andrä H, Schnack E. Some fundamental solutions for
plate bending analysis. International Conference on Boundary the Kirchoff, Reissner and Mindlin plates and a unified BEM
Element Technology X, Southampton: CMP; 1995. p. 1–8. formulation. Engng Anal Bound Elem 2001;25:129–39.
[10] Weeën F. Application of the direct boundary element method to [16] Reissner E. On bending of elastic plates. Q Appl Math 1947;5:55– 68.
Reissner’s plate model. Bound Elem Meth Engng 1982. p. 487– 99. [17] Timoshenko SP, Woinowsky-Krieger S. Theory of plates and shells,
[11] Vilmann O, Dasgupta G. Fundamental solutions of Mindlin plates 2nd ed. New York: McGraw-Hill; 1959.
with variable thickness for stochastic boundary elements. Engng Anal [18] Rashed YF, Aliabadi MH, Brebbia CA. Hypersingular boundary
Bound Elem 1992;9:47–59. element formulation for Reissner plates. Int J Solids Struct 1998;
[12] De Barcellos CS, Silva LHM. A boundary element formulation for the 35(18):2229 –49.
Mindlin’s plate model. In: Brebbia CA, Venturini WS, editors. [19] Long SY, Brebbia CA, Telles JCF. Boundary element bending
Boundary element techniques: applications in stress analysis and heat analysis of moderately thick plates. Engng Anal Bound Elem 1988;5:
transfer. Southampton: CMP; 1989. p. 122 –30. 64–73.

You might also like