A 6-node triangular solid-shell element for linear and nonlinear analysis

Yeongbin Ko, Phill-Seung Lee *

Department of Mechanical Engineering, Korea Advanced Institute of Science and

Technology, 291 Daehak-ro, Yuseong-gu, Daejeon, 34141, Republic of Korea

Abstract
In this paper, we present an effective 6-node triangular solid-shell element (MITC-S6), with
particular attention on shear locking and thickness locking. To alleviate shear locking, the
assumed transverse strain field of the MITC3+ shell element is used while modifying the
bending enhancement mechanism. Thickness locking is treated using the assumed and
enhanced strain methods for thickness strain. Two independent enhancements of strains are
applied: The in-plane and transverse shear strain fields are enhanced using the strain fields
obtained from a bubble interpolation function for in-plane translations and the thickness
strain field is enhanced for linear variation in the thickness direction. The general three
dimensional material law is employed. The proposed element passes all the basic tests
including zero-energy mode, patch, and isotropy tests. Excellent performance is observed in
various linear and nonlinear benchmark tests, wherein its performance is compared to that of
existing 6-node triangular and 8-node quadrilateral solid-shell elements.

Keywords: shell structures; solid-shell finite elements; triangular element; 6-node element;
shear locking; thickness locking

* Corresponding author: Associate Professor, [email protected]

This article has been accepted for publication and undergone full peer review but has not
been through the copyediting, typesetting, pagination and proofreading process which may
lead to differences between this version and the Version of Record. Please cite this article as
doi: 10.1002/nme.5498

This article is protected by copyright. All rights reserved.

1. Introduction

Solid-shell elements with unique aspects in the finite element analysis of shell structures have
become popular in recent decades. In addition to conventional shell theories, solid-shell
elements can represent extended physics with the inclusion of stretch in the thickness
direction. Solid-shell elements have only translational degrees of freedom (DOFs), leading to
the following advantages: the elements can easily simulate the thickness change of shell
structures, merge well with solid finite elements and avoid complex rotation updates in
geometric nonlinear analyses [1-13]. The general three-dimensional (3D) material law can be
directly employed, which is beneficial in material nonlinear analyses [3-7].

An ideal triangular solid-shell finite element should meet the following requirements: The
solid-shell element should satisfy the basic tests such as patch [1-5,7-11,14-18], isotropy
[9,15-18] and zero energy mode tests [2,6,9-11,14-18]. The element should properly express
the physics of the shell structures using the general three-dimensional material law [4-6,14].
Finally, uniformly optimal convergence of the solution should be seen in various bending-
dominated or mixed shell problems [15-18]. It is extremely difficult to develop an ideal
triangular solid-shell element. In the development of effective solid-shell elements, the major
obstacles are shear locking, membrane locking and thickness locking.

Alleviation of shear locking in structural finite elements has been accomplished using the
assumed strain (Assumed Natural Strain) and MITC (Mixed Interpolation of Tensorial
Components) methods. The concepts of the assumed strain method were introduced for 4-
node quadrilateral plate elements [19,20] and the method was then successfully applied to
shell finite elements [14]. It was also useful to alleviate shear locking of 8-node quadrilateral
solid-shell elements [3-8,12,15]. However, finding effective shear locking treatments for low-
order triangular plate or shell finite elements progressed quite lately, leading to subsequent
developments of 3-node shell elements (MITC3 and MITC3+ shell elements) [15,16]. Unlike
shear locking, membrane locking occurs only when the element geometry is curved [18].
Therefore, membrane locking is not treated in 3-node triangular shell and 6-node triangular
solid-shell elements due to their flat geometries.

This article is protected by copyright. All rights reserved.

Thickness locking in solid-shell elements is categorized as Poisson thickness locking and
curvature thickness (or trapezoidal) locking. Poisson thickness locking predominantly
deteriorates the solution accuracy of solid-shell elements whenever Poisson‟s ratio is not
equal to zero. Linear variation of thickness strain [21] or quadratic variation of transverse
displacement [22,23] was studied to alleviate the locking, and consequently the enhanced
assumed strain (EAS) method, originally proposed in Ref. [24], was widely employed for the
treatment of Poisson thickness locking [2-8,12]. Curvature thickness locking occurs when
mid-surface normal directions are naturally distorted in solid-shell element models. In
alleviating such locking, the assumed strain (ANS) methods was also useful for quadrilateral
shell elements with thickness stretching [25] and subsequently for solid-shell elements [1-
8,12,13]. At this point, it is worthy to note that alleviation of locking is also a very important
issue in isogeometric analysis of shell structures [26,27].

In this paper, we propose a new 6-node solid-shell element, named MITC-S6, for general use
in linear and nonlinear analyses. In order to alleviate shear locking, the assumed transverse
shear strain field of the MITC3+ shell element [16] is adopted. However, the bending
mechanism is differently enhanced with two internal translations interpolated using a bubble
function [16,28]. To avoid Poisson thickness locking, the thickness stretch mechanism is
enhanced by enabling linear variation in thickness strain [23]. Finally, by utilizing the
assumed constant thickness strain sampled from the nodal points, curvature thickness locking
is treated. In the formulation, the general 3D material law is employed.

The basic aspects of the proposed element can be summarized as follows: The element has
six nodes, each carrying only three translations as external DOFs. The rotational DOFs or
update of director vectors are unnecessary. The element passes basic tests, and adopts three
internal DOFs due to the EAS method used to reduce locking. The constitutive model is fully
general, and can be defined in any material axis, as with solid elements. The predictive
capability of the present element is particularly good for irregular meshes.

In the following sections, the formulation of the new MITC-S6 solid-shell element is
presented. The performance of the element is demonstrated through basic tests, classical
benchmark tests and detailed convergence studies. Geometric nonlinear analyses are also

This article is protected by copyright. All rights reserved.

presented to show the robustness of the element in the simulation of large displacements.

2. New 6-node triangular solid-shell element

In this section, we present the total Lagrangian formulation of the new triangular solid-shell
element for a geometric nonlinear analysis. In the formulation, the super- and subscript t
denotes “time” for a general analysis, and in static solutions “time” simply denotes load steps
and configurations.

2.1 Geometry and displacement interpolations

The geometry of the proposed element in the configuration at time t shown in Figure 1 is
interpolated by
3 3
1 1
t
x(r , s,  )  (1   ) hi t xbot
i  (1   )  hi t xtop
i , (1)
2 i 1 2 i 1

where h1  r , h2  s , h3  1  r  s and the superscripts “bot” and “top” respectively

denotes the bottom and top surfaces of the solid-shell element.

The incremental displacement vector u from the configuration at time t to the

configuration at time t  Δt is
u(r , s, )t  t x(r , s, )t x(r , s,  ) , (2)
and hence
3 3
1 1
u  (1   ) hiubot
i  (1   )  hiutop
i . (3)
2 i 1 2 i 1

2.2. Green-Lagrange strain

In this section, the Green-Lagrange strain of the proposed solid-shell element is formulated
for the nonlinear analysis. The super- and subscript 0 ( t  0 ) denotes the initial reference

This article is protected by copyright. All rights reserved.

configuration, see Figure 2 for the three configurations considered.

The covariant components of the Green-Lagrange strain are given by

1 0x t t x 0
t
ε ij  ( t g i t g j  0 g i 0 g j ) with 0
gi  , gi   g i  u ,i , (4)
ri ri
0
2

u
where u ,i  , ut x0 x , r1  r , r2  s , r3  ξ .
ri

The incremental covariant strain components are

1 u
ε ij  t Δ0t ε ij  0t ε ij  (u ,i t g j  t g i  u , j  u ,i  u , j ) with u ,i  . (5)
ri
0
2
which can be decomposed as
1 1
0 ε ij 0 eij  0 ηij with 0 eij  (u ,i t g j  t g i  u , j ) , 0 ij  (u ,i  u , j ) , (6)
2 2
where e
0 ij and 0ij are the linear and nonlinear parts, respectively.

Substituting Equation (3) into Equation (6), the linear and nonlinear parts of the covariant in-
plane strains are directly calculated as follows:
1 1
e  (u ,i t g j  t g i  u , j )  B ij U e , 0ij  (u ,i  u , j )  12 U e N i , U e for i, j  1, 2 ,
T
0 ij (7)
2 2
in which B ij is the linear strain-displacement relation matrix for the strain component ij ,

U e is the incremental nodal displacement vector including u bot

i and u top
i , and N ij is the

matrix representing the relation between the nonlinear strain components and the nodal
displacements.

For the transverse shear strains and the corresponding variations, we adopt the assumed strain
field of the MITC3+ triangular shell element [16]. The linear parts of the covariant transverse
shear strains e
0 23 and e
0 13 are substituted by

2 ( A) 1 ( A) 1 ( C ) 1
0 23e~  ( 0 e23  0 e13 )  ( 0 e13  0 e23
(C )
)  c~(1  3r )  B 23U e , (8a)
3 2 3 3
2 ( B ) 1 ( B ) 1 (C ) 1
0 13e~  (0 e13  0 e23 )  ( 0 e13  0 e23
(C )
)  c~(3s  1)  B 23U e , (8b)
3 2 3 3

This article is protected by copyright. All rights reserved.

with
c~0 e13
(F )
0 e13
( D)
0 e23
(F )
 0 e23
(E)
, (8c)

in which the tying points are shown in Figure 3. As the distance d in Figure 3 varies from
1 / 6 to 0 , the three tying positions (D) , (E ) , and (F ) move from the centers of the
edges to the barycenter, resulting in a smaller in-plane twisting stiffness [16,17]. In this study,
we use the fixed value d  1 / 100,000 and thus d is not a problem-dependent value.

The same assumed strain field with the same tying positions is used for the nonlinear part of
the covariant transverse shear strains:

0 ~i 3  12 U e T N i 3 U e for i  1, 2 . (9)

To reduce curvature thickness locking, the assumed strain method is also used. The linear and
nonlinear parts of the thickness strain are assumed as

~ 1
0 33e (r , s)  (0 e33 (0,0) 0 e33 (1,0) 0 e33 (0,1))  B33U e , (10a)
3
1
0 ~33 (r , s)  ( 0 33 (0,0) 0 33 (1,0) 0 33 (0,1))  12 U e T N 33U e . (10b)
3
Note that, unlike the linear strain field in Ref. [1], the constant assumed strain field is
employed for thickness strain in this element.

2.3. Strain enhancements

We here derive the enhanced strains to improve the behaviors of the 6-node triangular solid-
shell element. The enhanced strains are used for the linear part of the covariant in-plane,
transverse shear and thickness strains.

In order to construct the enhanced strains, we consider the following interpolations defined
using the bubble-type functions:
1
u in-plane  hb ξ (αV1  βV2 ) , (11a)
2

This article is protected by copyright. All rights reserved.

 1
u thickness   hq γV3 , (11b)
2
where hb  27rs(1  r  s) and hq  1  ξ 2 are the two-dimensional bubble and one-

dimensional quadratic functions, V1 , V2 and V3 are vectors denoting the direction of

translations, and  ,  and γ are internal variables.

The interpolation in Equation (11a) is used to effectively enhance the bending mechanism of
the triangular elements, see Refs. [10,16,28,29]. In the plate and shell elements [16,29], the
variables  and  are defined as rotations about V1 and V2 . However, in this 6-node
solid-shell element, the variables  and  represent the in-plane translation of shell

surfaces in the directions of V1 and V2 . As long as the two vectors can fully define the in-
plane translations, i.e. they are not coincident to each other, the same numerical results are
obtained. Thus, it is not necessary to update V1 and V2 in the nonlinear solution procedure.
In this study, we simply use V1  i x and V2  i y .

The interpolation in Equation (11b) has been used in three-dimensional shell elements
[6,22,23] to introduce the quadratic thickness stretch, which is useful to avoid Poisson
thickness locking. For this purpose, we take the vector V3 identical to the third

contravariant base vector V3 t g 3 , where the contravariant base vectors t g i satisfy the

relation t g i t g j  δij for the Kronecker delta δij .

Using the bubble interpolation in Equation (11a), we first derive the enhanced in-plane
covariant strain as follows:
1
eenh
0 ij  (u in,i -planet g j  t g i  u in,j -plane )  G ij Λ e for i, j  1, 2 , (12)
2
u in -plane
where u in,i -plane  , G ij is the enhanced strain-displacement relation matrix, and Λ e
ri
is the vector of internal variables. This strain field is also effective for improving the bending
performance of the 6-node triangular solid-shell element.

This article is protected by copyright. All rights reserved.

The transverse shear covariant strains due to the bubble interpolation in Equation (11a) are
calculated by
1
bub
0 i3e  (u in,i -planet g 3  t g i  u in,3-plane ) , (13)
2
and the enhanced transverse shear strains are obtained using the following assumed field:
2 bub( A) 1 bub( A) 1 bub(C )
enh
0 23e  (0 e23  0 e13 )  (0 e13  0 e23
bub( C )
)  G 23Λ e , (14a)
3 2 3
2 bub( B ) 1 bub( B ) 1 bub(C )
enh
0 13e  (0 e13  0 e23 )  ( 0 e13  0 e23
bub( C )
)  G13Λ e , (14b)
3 2 3
where the tying points ( A , B and C ) used are shown in Figure 3. Note that this assumed
strain does not include the terms linear in r and s from Equation (8), resulting in simple
matrices G 23 and G13 (see Appendix A).

In order to alleviate Poisson thickness locking, the following enhanced strain is employed for
the thickness normal component:
enh
0 33e  t g 3  u ,thickness
3    G 33Λ e , (15)

u thickness
where u ,thickness  . In Equation (15), a single internal variable was used, as in Ref.
r3
3

[2], with the physical meaning of thickness normal translation being quadratic in ξ .

Note that the interpolation in Equation (11a) is used only to enhance strain components „11‟,
„22‟, „12‟, „23‟ and „13‟, while the thickness strain (component „33‟) is enhanced by Equation
(11b). That is, two strain enhancements are independently applied to each other. This is an
important characteristic for the present 6-node solid-shell element to behave well in various
shell problems.

We finally obtain the linear part of the incremental covariant strains for the 6-node MITC
solid-shell element:

0 e ij  0 eij  0 eijenh  Bij U e  G ij Λ e for i, j  1, 2 , (16a)

0 i3e  0 e~i 3  0 eienh

3  Bi 3U e  G i 3 Λ e for i  1, 2 , (16b)

0 e 33 0 e~ 33  0 e33
enh
 B33U e  G 33Λ e , (16c)

This article is protected by copyright. All rights reserved.

where Λ e  [α β γ]T is the vector of internal variables for the element. Finally, the

resulting form of the enhanced strain matrices ( G ij ) is given in Appendix A.

2.3. Incremental equilibrium equations

The general nonlinear response is calculated using incremental equilibrium equations. We

here derive the tangent stiffness matrix and internal force vector of a single solid-shell
element. The incremental equilibrium equations for an entire shell finite element model can
be easily constructed using the direct stiffness procedure [30].

Linearizing the principle of virtual work in the configuration at time t  t about the known
configuration at time t , the following total Lagrangian formulation is given [30].

 C ijkl 0 eij 0ekl d 0V  0 S ij 0ij d 0V t t   0 S ij 0eij d 0V ,

t t
0 0 0 0 (17)
V V V

t  t
where 0V is the volume of the solid-shell element at time 0 ,  is the external virtual
t
work due to the applied surface tractions and body forces, and 0 S ij denotes the

contravariant second Piola-Kirchhoff stress and 0 C ijkl is the corresponding general 3D

material law tensor [6,21,31].

Substituting Equations (7)-(10) and (16) into Equation (17), the total Lagrangian formulation
is discretized as follows:
δU e
T
0
V
B ij
T
0 C ijkl B kl d 0V  0 N ij 0t S ij d 0V U e  δU e
V
 T
 0
V
B ij
T
0 
C ijkl G kl d 0V Λ e

 δΛ e
T
 0
V
G ij 0 C ijkl B kl d 0V Ue  δΛ e
T
 T
0
V
T
G ij 0 C ijkl G kl d 0V Λ e 
 
T t Δt
 δUe R e  δUe S ij d 0V  δΛ e
T T t T T t
0
Bij 0 0
G ij 0 S ij d 0V , (18)
V V

t  t
where R e is the external load vector at time t  t .

We then obtain the following linearized equilibrium equation in the element level:

This article is protected by copyright. All rights reserved.

 ˆ
 tK t
Γ e  U e   t Δt R e   t Fˆ e 
t T      ,
e
(19)
 Γe
t
Α e  Λ e   0   t H e 

with
ˆ  B T C ijkl B d 0V  N t S ij d 0V , t Γ  B T C ijkl G d 0V ,
0 ij 0 0 ij 0 0 ij 0
t
K e kl e kl
V V V

Α e  0 G ij C ijkl G kl d 0V , t Fˆ e  0 Bij 0t S ij d 0V , t H e  0 G ij 0t S ij d 0V .
t T T T
0
V V V

The internal variables related to the strain enhancements can be easily condensed out in the
element level, and we finally obtain the condensed equilibrium equation [3,4,6,10]:
t
K e U e t Δt R e t Fe with t ˆ t Γ t Α 1 t Γ T , t F t Fˆ t Γ t Α 1 t H
K e t K (20)
e e e e e e e e e

and the contravariant second Piola-Kirchhoff stress is updated by

t Δt
0 Sij 0t Sij  0 Sij (21)
with
1 T
0 S ij  0 Cijkl (B kl U e  G kl Λ e  U e N kl U e ) , (22)
2
Λe t Αe1 (t He t Γe Ue ) .
T
(23)

In the incremental solution procedure, the geometry is updated using Equation (2). For the
evaluation of the element stiffness matrix and internal nodal force vector, we use 3-point
Gauss integration in the r - s plane and the 2-point Gauss integration in the  -direction.

3. Remarks on previous studies

We here briefly comment on two studies on previous 6-node solid-shell finite elements [1,2].

First, we remark on the formulation of stress-strain (constitutive) laws used for 6-node solid-
shell finite elements. Sze et al. [1] developed a 6-node solid-shell element based on the
hybrid stress method and employed a modified laminated constitutive law instead of the
general three-dimensional material law. The constitutive equations regarding membrane,
bending, transverse shear and thickness strains are decoupled and separately defined. Flores

This article is protected by copyright. All rights reserved.

[2] can employ the general three-dimensional material law, but still require transformation of
strains onto a specific material axis. However, in the present element, the general full three-
dimensional constitutive law is employed on the material axes independent of the orientation
of the solid-shell element. Special treatments on neither specific components of the stiffness
matrix nor the material law are necessary.

Second, we remark on the convergence tests that have been performed. Sze et al. [1] tested
the convergence of displacements on several linear benchmark problems focusing on coarse-
meshes. In the study by Flores [2], the convergence behavior was not thoroughly tested
especially for thin shell structures. In our study, we aim to show the convergence
performance of the present solid-shell element from coarse to fine meshes over shell
problems with various curvatures encompassing a practical range of shell thickness (ratio of
thickness to overall dimension, t / L  1 / 100 ~ 1 / 10000 ).

4. Basic numerical tests

In this section, basic numerical tests are conducted for the proposed solid-shell element. The
isotropy, patch and zero energy mode tests are considered.

Spatially isotropic behavior is an important requirement for the triangular elements. The
element behavior should not depend on the sequence of node numbering, i.e. the element
orientation [9,15-18]. The proposed solid-shell element passes this test.

We perform three patch tests: the membrane, bending and transverse shearing patch tests, see
Refs. [1-5,7-11,14-18] for the patch tests. The geometry of the mesh is shown in Figure 4(a).
The loading and boundary conditions for the membrane, bending and transverse shearing
patch tests are shown in Figures 4(b) to 4(d), respectively, in which the boundary conditions
for the shaded regions are specified. The patch of elements is additionally subjected to the
minimum number of constrains to prevent rigid body motions. If the correct constant stress
fields are calculated, the patch tests are passed. The proposed element passes all the patch
tests. Also, the element expresses constant normal stress in the constant compression test [1]

This article is protected by copyright. All rights reserved.

performed using a single element, see Figure 5.

In the zero energy mode tests, the number of zero eigenvalues of the stiffness matrix of a
single unsupported element is counted [2,6,9-11,14-18]. For the present element, only the six
zero eigenvalues corresponding to the six correct rigid body modes are obtained.

5. Classical benchmark tests

We here solve widely-used linear problems to benchmark the proposed solid-shell elements.
The problems considered are the fully clamped square plate problem, the Scordelis-Lo roof
problem, twisted beam problems and hyperboloid shell problems [1-7,9-11,13-22,32-39]. As
in the literature, convergence is measured using a representative displacement in a specific
location of the shell structure.

The solutions of the MITC-S6 solid-shell element are compared with those of the 6-node
solid-shell element by Sze et al. [1], the 6-node solid-shell element, SC6R, used in the
commercial software ABAQUS [40] and the 8-node solid-shell element in the literature,
denoted as MITC-S8 [3-7,12], see Appendix B for its formulation.

5.1. Fully clamped square plate problem

The plate bending problem [1,3-5,9,10,13,15-19,21,32] is shown in Figure 6. The square

plate with dimensions of 2L  2L and thickness t is subjected to uniform pressure q  1.0 .
We consider three different thicknesses, t / L  1 / 100 , 1 / 1000 and 1 / 10000 with L  100 .

The material properties used are E  1.0  104 and   0.3 .

Due to symmetry, only a quarter of the plate corresponding to the shaded region ABCD in
Figure 6 is modeled with the mesh patterns shown in Figure 7. The boundary conditions are
v  0 along the edge AB, u  0 along the edge AD and u  v  w  0 along the edges CD
and BC. We use N  N meshes with N  2 , 4 , 8 , 16 and 32 . For the case of

This article is protected by copyright. All rights reserved.

t / L  1 / 100 , the reference vertical deflection is wref  1.384 . As the plate thickness

becomes smaller, the deflection value varies with (t / L) 3 .

Table I presents the vertical deflection at the plate center, w A , normalized by the reference

values. Figure 8 shows convergences of the normalized displacement, wA / wref according to

N . It is observed that the MITC-S6 element shows good convergences regardless of t / L

and outperforms the 6-node solid-shell developed by Sze, compared in the cross mesh. In the
regular mesh patterns (types 1 and 2) considered here, the predictive capability of the MITC-
S6 triangular solid-shell element is comparable to that of the MITC-S8 quadrilateral solid-
shell element. Note that, in plate bending problems, most triangular elements perform much
better in a cross mesh than in regular meshes (types 1 and 2).

Figure 9 presents stress- xx at the plate center, σ Axx , normalized by the reference stress of

σ ref
xx
obtained using 72×72 uniform mesh of MITC9 shell elements [38,39]. Stresses are

evaluated at the top surface for the type 1 regular mesh used. Unlike the SC6R element, the
normalized effective stress ( σ Axx / σ ref
xx
) of the MITC-S6 element converges uniformly toward

the reference value regardless of the thickness of the plate.

5.2. Scordelis-Lo roof problem

The Scordelis-Lo roof problem [1,2,5,11,14,20,22,33-35] in Figure 10 is considered. A roof

structure of length L  50 , radius R  25 and thickness t  0.25 is supported by rigid end
diaphragms along its curved edges and is subjected to a self-weight with density ρ  360 and

gravity constant g  1.0 . Its material properties are E  4.32  108 and ν  0.0 .

We discretize only one quarter of the structure corresponding to the shaded region ABCD in
Figure 10 with the mesh patterns shown in Figure 7. The boundary conditions are v  0
along the edge AB, u  0 along the edge DA and u  w  0 along the edge CD. We use
N  N meshes with N  2 , 4 , 6 , 8 , 10 , 12 , 14 , 16 and 32 .

This article is protected by copyright. All rights reserved.

Table II shows the normalized vertical deflections at the center of free edge (point B ), wD ,
normalized by the reference value of wref  0.3024 [33]. Figure 11 shows the convergence of

the normalized vertical deflections ( wB / wref ) according to the mesh division N for the

type 1 regular mesh of the 6-node solid-shell elements. The proposed solid-shell element
outperforms the SC6R element.

Figure 12 presents effective stresses at the top surface along the edge AB ( σ eff
AB ) with

reference to the stresses obtained by 72×72 uniform mesh of MITC9 shell elements [38,39].
The stresses are sampled at the mid-points of the element edges. For the 6-node solid-shell
element, the type 1 regular mesh is used. Convergence of the effective stress ( σ eff
AB ) of the

MITC-S6 element is similar to that of the SC6R element.

5.3. Twisted beam problem

The twisted beam problem [5,7,9,11,34,36] shown in Figure 13 often has been considered to
test membrane locking. A cantilever beam of length L  12 , width w  1.1 , and twist of 90 
is loaded by concentrated forces at the center of free tip, point A. The material properties are
E  2.9  107 and ν  0.22 . We consider two different thicknesses, t  0.32 and t  0.0032 .

The whole beam is modeled with N  6 N meshes with N  2 , 4 , 8 , 16 and 32 . Two

loading cases are considered: in-plane ( P ) and out-of-plane ( F ) loadings. For t  0.32 ,
P  F  1.0 is used and the reference deflections at point A are vref  0.5424  102 and

wref  0.1754  102 for in-plane and out-of-plane loadings, respectively [34]. For t  0.0032 ,

P  F  1.0 10 6 is used, and the reference deflections at point A are vref  0.5256  102

and wref  0.1294  102 for in-plane and out-of-plane loadings, respectively [34].

Table III presents the in-plane and out-of plane deflections at point A, v A and w A ,

This article is protected by copyright. All rights reserved.

normalized by the reference values of each. Figure 14 shows the convergence of the
normalized vertical deflections ( v A / v ref and w A / wref ) according to N . All the solid-shell

elements considered here show excellent performance. We additionally present in Figure 15

the deformed shapes of the beam where the maximum displacements are normalized by 30%
of the beam length.

6. Detailed convergence studies

In this section, we perform in-depth studies on the solution convergence of solid-shell

elements for plate and hyperboloid shell structures with various values of t / L . Unlike the
previous section where we solved the benchmark problems using only regular meshes of
solid-shell elements, the behaviors of the elements adapting to irregular meshes are presented.

Solution convergences are measured using the following normalized strain energy error E h ,

Eref  Eh
Eh  , (24)
Eref

in which E h and Eref denote strain energies stored in the entire structure obtained from the

finite element and reference solutions, respectively. Optimal convergence for low-order
elements considered in this study is obtained if Eh  ch 2 , where c is a constant

independent of the shell thickness and h is the element size.

This article is protected by copyright. All rights reserved.

In order to investigate how much thickness locking deteriorates the solution accuracy, we also
implemented an intermediate 6-node solid-shell element: MITC-S6 without thickness locking
treatment (MITC-S6i). The solutions obtained using the MITC-S6 and MITC-S6i solid-shell
elements are compared with those of the MITC-S8 solid-shell elements (see Appendix B).

6.1. Fully clamped square plate problem

To further study the solution convergence of the solid-shell elements with various values of
t / L , we again solve the same plate bending problem considered in Section 5.1 with different
geometric and material constants [15-18,32]. The plate of dimension 2L  2L and thickness
t is subjected to uniform pressure q  1.0 . We consider three ratios of thickness to
dimension: t / L  1/ 100 , t / L  1/ 1000 and t / L  1/ 10000 with L  1.0 . The material

properties used are E  1.7472 107 and ν  0.3 .

To avoid difference between regular meshes (type 1 and type 2), the whole plate shown in
Figure 6 is modeled with the regular mesh pattern shown in Figure 7. The boundary
conditions are u  v  w  0 along the edges A´B´, B´C´, C´D´ and D´A´. We use N  N
meshes with N  4 , 8 , 16 and 32 .

In order to study the performance of the solid-shell elements in irregular meshes, the mesh
patterns shown in Figures 16(a) and (b) are also considered for the triangular and
quadrilateral elements, respectively, where each edge is discretized in the following ratio:
L1 : L2 : L3 : …… L N = 1 : 2 : 3 : …… N [15-18]. The similar distorted mesh patterns have

been used for testing triangular elements, see Refs. [9,10].

The reference solutions are obtained using a 72×72 uniform mesh of MITC9 shell elements
[38,39]. Figure 17 shows the convergence curves of the solid-shell elements when the regular
and irregular mesh patterns are used. The element size in the convergence curves is
h  L / N . The performance of the MITC-S6 solid-shell element is again comparable to that
of the MITC-S8 solid-shell element, and is particularly better when the irregular mesh pattern

This article is protected by copyright. All rights reserved.

is used. This is a very interesting result and additional investigations are required.

On the other hand, the convergence of the MITC-S6i solid-shell element severely deteriorates
and the errors do not diminish well with the mesh refinement. This is due to the Poisson
thickness locking, and the optimal convergence behavior is successfully recovered through
the EAS technique.

6.2. Hyperboloid shell problems

We solve the hyperboloid shell problem [15-18,37-39] shown in Figure 18(a). The shell
structure has length L  1 , radius R  1 , and thickness t . The mid-surface of the shell
surface is given by
x2  z 2  1  y 2 ; y [ 1, 1] , (25)
and a smoothly varying pressure in Figure 18(b) is applied over the entire structure:
p( )  p0 cos(2 ) with p0  1.0 . (26)

A bending-dominated behavior is obtained with free ends and a membrane-dominated

behavior is given with clamped ends. We again consider three ratios of thickness to
dimension: t / L  1/ 100 , t / L  1/ 1000 and t / L  1/ 10000 . The material properties used are

E  2.0 1011 and ν  1/ 3 .

Due to symmetry, only one-eighth of the structure corresponding to the shaded region
A´B´C´D´ in Figure 18(a) is modeled for the analysis. For the membrane-dominated case, the
clamped boundary condition is imposed: w  0 along B´C´, u  0 along D´A´, and v  0
along D´C´ and u  v  w  0 along A´B´. For the bending-dominated case, the free
boundary condition is imposed: w  0 along B´C´, u  0 along D´A´, and v  0 along
D´C´. For both cases, we use N  N meshes with N  4 , 8 , 16 and 32 .

For the membrane-dominated case, the regular mesh graded in a boundary layer of width

6 t shown in Figure 18(c) is considered, see Refs. [37-39]. For the bending-dominated case,

This article is protected by copyright. All rights reserved.

the regular mesh is graded in a boundary layer of width 0.5 t [37,39]. We also perform
convergence studies with the irregular meshes shown in Figures 16(a) and (b).

Figure 19 shows the convergence curves of the solid-shell elements for the regular and
irregular mesh patterns when the clamped boundary condition is imposed. A 72×72 uniform
mesh of MITC9 shell elements is used to obtain the reference strain energy. The element size
is h  L / N . The performance of the MITC-S6 solid-shell element is similar to that of the
MITC-S8 solid-shell element. As expected, it is observed that thickness locking is inherently
not present in this membrane-dominated problem.

Figure 20 shows the convergence curves of the solid-shell elements when the free boundary
condition is used. In both regular and irregular meshes, the MITC-S6 solid-shell element
presents excellent convergence behavior, while the convergence of the MITC-S8 solid-shell
element severely deteriorates for the case of irregular mesh. This is due to membrane locking,
which is investigated in detail in Ref. [18]. In addition, thickness locking manifests even
more severely for this problem, notably from both curvature and Poisson thickness locking.
The combined use of the AS and EAS techniques for thickness strain indeed facilitates
solution convergence in this doubly-curved shell problem.

7. Geometric nonlinear analysis

In this section, we present the performance of the MITC-S6 solid-shell element in the
numerical examples involving large displacements and large rotations through three shell
problems: a cantilever beam under shear tip force [8,41-43] and a pinched cylindrical shell
[36,43,44].

This article is protected by copyright. All rights reserved.

7.1. Cantilever beam under shear tip force

A cantilever beam structure of length L  10 , width b  1 and thickness t  0.1 shown in

Figure 21 is considered [8,41-43]. The cantilever beam is subjected to a shear force P at its

free tip. The material properties are E  1.2  10 6 and ν  0 . The mesh used is shown in
Figure 21. The analysis was performed until the maximum shear force P  4.0 is reached.

Figure 22 shows the load-displacement curves: non-dimensional applied load PL2 / EI (with
I  bt 3 /12 ) versus non-dimensional tip displacements utip / L and wtip / L . The load-

displacement curves are in very good agreement with those obtained by Sze et al. [43].
Deformed shapes for load levels P  1.0 , 2.0 and 4.0 are shown in Figure 23.

7.2. Pinched cylindrical shell

The pinched cylindrical shell problem [36,43,44] is considered here, as shown in Figure 24.
The cylinder of length L  3.048 , radius R  1.016 and thickness t  0.03 is clamped at
one end and is subjected to a pair of concentrated forces F at the free end. The material
properties used are E  2.0685  10 7 and ν  0.3 . Due to symmetry, only one quarter of the
cylinder is modeled using 14×14 meshes of the proposed 6-node solid-shell elements. As in
Ref. [36], the analysis was performed neglecting the possible contact within the shell
structure up to the maximum force F  2.0 103 .

The radial displacements at points A and B of the shell are shown in Figure 25. The overall
load-displacement curves are in good agreement with Refs. [36,43,44]. The deformed shape
at F  700 is shown in Figure 26.

This article is protected by copyright. All rights reserved.

8. Concluding remarks

In this study, we developed a new 6-node solid-shell element by extending the previous 3-
node MITC3+ shell element. We adopted the assumed transverse shear strain field of the
MITC3+ shell element onto the new MITC-S6 solid-shell element. The concept of enhancing
the bending mechanism using the cubic bubble function is similar to the MITC3+ shell
element, but in this solid shell element the mechanism is applied for the in-plane translation
of the shell element through the EAS method. To avoid Poisson thickness locking, the
thickness stretch mechanism is enhanced to include quadratic variation in the thickness
direction. The assumed constant thickness strain is adopted to avoid curvature thickness
locking.

The new MITC-S6 solid-shell element satisfies basic tests and achieves uniformly optimal
convergence on various shell problems considered in this study while providing a simple
formulation, as in the MITC3+ shell element. The MITC-S6 solid-shell element also
maintains various advantages in computational aspects of nonlinear simulations: No rotation
updates are necessary and the general three-dimensional material law is directly used, which
facilitates further application of the present solid-shell element to the analysis of elastoplastic
behavior of shell structures when thickness change is important.

Acknowledgments

This work was supported by the Basic Science Research Program through the National
Research Foundation of Korea (NRF) funded by the Ministry of Education, Science and
Technology (No. 2014R1A1A1A05007219).

Appendix A: The enhanced strain matrices

In this section, we give the enhanced strain matrices G ij in Equation (16), which are useful

for practical implementation of the proposed solid-shell element.

This article is protected by copyright. All rights reserved.

G11 G 22 G 33 G12 G 23 G13 
T

T
 hb ,1ξV1t g1 hb , 2ξV1t g 2 0 1
h ξV1t g1  12 hb ,1ξV1t g 2
2 b, 2
1
V1t g 2 1
V1t g1 
 
4 4

 hb ,1ξV2 t g1 hb , 2ξV2 t g 2 0 1

h ξV2 t g1  12 hb ,1ξV2 t g 2
2 b,2
1
4
V2 t g 2 1
4
V2 t g1  , (A.1)
 0 0 ξ 0 0 0 

hb
in which hb ,i  for i  1, 2 .
ri

Appendix B: The 8-node quadrilateral solid-shell element (MITC-S8)

In this section, we briefly present the linear formulation of the 8-node solid-shell element,
denoted as MITC-S8, which is used for comparison in this study. Shear locking and thickness
locking are treated using the AS and EAS methods.

For the 8-node solid-shell element, the geometry and displacement interpolations are given
by (Figure 27)
4  4 
 1  1 
x(r , s, ξ )  (1  ξ ) hi x bot
i  (1  ξ )  hi x top
i , (B.1)
2 i 1 2 i 1

4  4 
 1  1 
u(r , s, ξ )  (1  ξ ) hi u bot
i  (1  ξ ) hi u top
i . (B.2)
2 i 1 2 i 1

 1  1  1  1
with h1  (1  r )(1  s) , h2  (1  r )(1  s) , h3  (1  r )(1  s) , h4  (1  r )(1  s) .
4 4 4 4

The linear part of the covariant strain is calculated by

 1    
eij (r , s, ξ)  (u ,i  g j  g i  u , j ) (B.3)
2
 
 u  x
with u ,i  , gi  , r1  r , r2  s , r3  ξ . (B.4)
ri ri

The following assumed transverse shear strains are employed to treat shear locking, see Refs.
[3-7,12]

This article is protected by copyright. All rights reserved.

~ 1 (J ) 1 (I )
e23 (r , s, ξ )  (1  r )e23  (1  r )e23 , (B.5)
2 2
1  1 
e~13 (r , s, ξ )  (1  s)e13( H )  (1  s)e13(G ) , (B.6)
2 2
where the tying points are shown in Figure 28.

In order to reduce thickness locking, the following assumed strains is used for thickness
strain, see Refs. [3-6,12,22],
   
e~33 (r , s, ξ )  h1e33 (1,1)  h2 e33 (1,1)  h3e33 (1,1)  h4 e33 (1,1) . (B.7)

The enhanced assumed thickness strain is also used as follows:

 γ1 
γ 
j0 2
e33 (r , s, ξ )  t33[ξ rξ sξ rsξ ] 2  ,
enh
(B.8)
j  γ3 
 
γ4 
   
with t33  g 3 (r , s, ξ )  g 3 (0,0,0) , g i (r , s, ξ )  g j (r , s, ξ )  δij , (B.9)
   T
in which j and j0 are the determinants of the Jacobian matrix g1 g 2 g 3  at (r , s, ξ)

and at (0,0,0) , respectively, and γ i are internal variables, see Refs. [3,4,12]. We note that,
unlike the MITC-S6 triangular solid-shell element, four internal variables are necessary to
satisfy the basic tests.

We note that this element is equivalent to the 8-node solid-shell element proposed by Klinkel
et al. [3,4]. In the numerical examples, the 2  2  2 Gauss integration is employed.

References

[1] Sze KY, Chan WK. A six-node pentagonal assumed natural strain solid–shell element.
Finite elements in Analysis and Design 2001;37(8):639-55.

[2] Flores FG. Development of a non-linear triangular prism solid-shell element using ANS

This article is protected by copyright. All rights reserved.

and EAS techniques. Computer Methods in Applied Mechanics and Engineering
2013;266:81-97.

[3] Klinkel S, Gruttmann F, Wagner W. A robust non-linear solid shell element based on a
mixed variational formulation. Computer Methods in Applied Mechanics and Engineering
2006;195(1):179-201.

[4] Klinkel S, Gruttmann F, Wagner W. A continuum based three-dimensional shell element

for laminated structures. Computers & Structures 1999;71(1):43-62.

[5] Schwarze M, Reese S. A reduced integration solid‐shell finite element based on the EAS
and the ANS concept–Geometrically linear problems. International Journal for Numerical
Methods in Engineering 2009;80(10): 1322-55.

[6] Hauptmann R, Schweizerhof K. A systematic development of „solid-shell‟ element

formulations for linear and non-linear analyses employing only displacement degrees of
freedom. International Journal for Numerical Methods in Engineering 1998;42:49–69.

[7] Sze KY, Yao LQ. A hybrid stress ANS solid‐shell element and its generalization for smart
structure modelling. Part I–solid‐shell element formulation. International Journal for
Numerical Methods in Engineering 2000;48(4):545-64.

[8] Sze KY, Chan WK, Pian THH. An eight‐node hybrid‐stress solid‐shell element for
geometric non‐linear analysis of elastic shells. International Journal for Numerical Methods
in Engineering 2002;55(7):853-78.

[9] Kim JH, Kim YH, Lee SW. An assumed strain formulation of efficient solid triangular
element for general shell analysis. International Journal for Numerical Methods in
Engineering 2000;47(8):1481-97.

[10] Hong WI, Kim JH, Kim YH, Lee SW. An assumed strain triangular curved solid shell
element formulation for analysis of plates and shells undergoing finite rotations. International

This article is protected by copyright. All rights reserved.

Journal for Numerical Methods in Engineering 2001;52(7):747-61.

[11] Sze KY, Zhu D. A quadratic assumed natural strain curved triangular shell element.
Computer Methods in Applied Mechanics and Engineering 1999;174(1):57-71.

[12] Harnau M, Schweizerhof K. About linear and quadratic „solid-shell‟ elements at large
deformations. Computers & Structures 2002;80:805-17.

[13] Kim CH, Sze KY, Kim YH. Curved quadratic triangular degenerated‐and solid‐shell
elements for geometric non‐linear analysis. International Journal for Numerical Methods in
Engineering 2003;57(14):2077-97.

[14] Dvorkin EN, Bathe KJ. A continuum mechanics based four-node shell element for
general nonlinear analysis. Engineering Computations 1984;1(1):77-88.

[15] Lee PS, Bathe KJ. Development of MITC isotropic triangular shell finite elements.
Computers & Structures 2004;82(11):945-62.

[16] Lee Y, Lee PS, Bathe KJ. The MITC3+ shell element and its performance. Computers &
Structures 2014;138:12-23.

[17] Lee Y, Yoon K, Lee PS. Improving the MITC3 shell finite element by using the
Hellinger-Reissner principle. Computers & Structures 2012;110-111:93-106.

[18] Ko Y, Lee PS, Bathe KJ. The MITC4+ shell element and its performance. Computers &
Structures 2016;169:57-68.

[19] Hughes TJR, Tezduyar T. Finite elements based upon Mindlin plate theory with
particular reference to the four-node bilinear isoparametric element. Journal of Applied
Mechanics 1981;48(3):587-96.

[20] Macneal RH. Derivation of element stiffness matrices by assumed strain distributions.

This article is protected by copyright. All rights reserved.

Nuclear Engineering and Design 1982;70(1):3-12.

[21] Büchter N, Ramm E, Roehl D. Three‐dimensional extension of non‐linear shell

formulation based on the enhanced assumed strain concept. International Journal for
Numerical Methods in Engineering 1994;37(15):2551-68.

[22] Bischoff M, Ramm E. Shear deformable shell elements for large strains and rotations.
International Journal for Numerical Methods in Engineering 1997;40:4427–49.

[23] Parisch H. A continuum‐based shell theory for non‐linear applications. International

Journal for Numerical Methods in Engineering 1995;38(11):1855-83.

[24] Simo JC, Rifai MS. A class of mixed assumed strain methods and the method of
incompatible modes. International Journal for Numerical Methods in Engineering
1990;29(8):1595-638.

[25] Betsch P, Stein E. An assumed strain approach avoiding artificial thickness straining for
a non‐linear 4‐node shell element. Communications in Numerical Methods in Engineering
1995;11(11):899-909.

[26] Echter R, Oesterle B, Bischoff M. A hierarchic family of isogeometric shell finite

elements. Computer Methods in Applied Mechanics and Engineering 2013;254:170-80.

[27] Hosseini S, Remmers JJ, Verhoosel CV, De Borst R. An isogeometric continuum shell
element for non-linear analysis. Computer Methods in Applied Mechanics and Engineering
2014;271:1-22.

[28] Fricker AJ. An improved three‐noded triangular element for plate bending. International
Journal for Numerical Methods in Engineering 1985;21(1):105-14.

[29] Auricchio F, Taylor RL. A generalized elastoplastic plate theory and its algorithmic
implementation. International Journal for Numerical Methods in Engineering

This article is protected by copyright. All rights reserved.

1994;37(15):2583-608.

[30] Bathe KJ. Finite element procedures. New York:Prentice Hall;1996.

[31] Lee PS, Bathe KJ. Insight into finite element shell discretizations by use of the “basic
shell mathematical model”. Computers & Structures 2005;83(1):69-90.

[32] Lee PS, Bathe KJ. The quadratic MITC plate and MITC shell elements in plate bending.
Advances in Engineering Software 2010;41(5):712–28.

[33] Belytschko T, Stolarski H, Liu WK, Carpenter N, Ong JS. Stress projection for
membrane and shear locking in shell finite elements. Computer Methods in Applied
Mechanics and Engineering 1985;51(1):221-58.

[34] Belytschko T, Wong BL, Stolarski H. Assumed strain stabilization procedure for the
9‐node Lagrange shell element. International Journal for Numerical Methods in Engineering
1989;28(2):385-414.

[35] Lee PS, Bathe KJ. On the asymptotic behavior of shell structures and the evaluation in
finite element solutions. Computers & Structures 2002;80(3):235-55.

[36] Jung WY, Han SC. An 8-node shell element for nonlinear analysis of shells using the
refined combination of membrane and shear interpolation functions. Mathematical Problems
in Engineering 2013.

[37] Chapelle D, Bathe KJ. The finite element analysis of shells – fundamentals. Berlin:
Springer; 2003.; 2nd edition, 2011.

[38] Bathe KJ, Lee PS, Hiller JF. Towards improving the MITC9 shell element. Computers &
Structures 2003;81(8):477-89.

[39] Hiller JF, Bathe KJ. Measuring convergence of mixed finite element discretizations: an

This article is protected by copyright. All rights reserved.

application to shell structures. Computers & Structures 2003;81(8):639-54.

[40] Systèmes D. ABAQUS 6.12 Theory manual. Dassault Systèmes Simulia Corp.,
Providence, Rhode Island, 2012.

[41] Park HC, Cho C, Lee SW. An efficient assumed strain element model with six DOF per
node for geometrically non‐linear shells. International Journal for Numerical Methods in
Engineering 1995;38(24):4101-22.

[42] Jeon HM, Lee Y, Lee PS, Bathe KJ. The MITC3+ shell element in geometric nonlinear
analysis. Computers & Structures 2015;146:91-104.

[43] Sze KY, Liu XH, Lo SH. Popular benchmark problems for geometric nonlinear analysis
of shells. Finite elements in Analysis and Design 2004;40(11):1551-69.

[44] Arciniega RA, Reddy JN. Tensor-based finite element formulation for geometrically
nonlinear analysis of shell structures. Computer Methods in Applied Mechanics and
Engineering 2007;196(4):1048-73.

This article is protected by copyright. All rights reserved.

Table I. Normalized vertical deflections at the plate center (point A ) for the clamped plate
problem. The mesh patterns are shown in Figure 7.
Type 1 Type 2 Regular
Cross mesh
t/L N regular mesh regular mesh mesh

MITC-S6 SC6R MITC-S6 SC6R MITC-S6 SC6R Sze et al. MITC-S8

2 0.743 0.029 0.993 0.043 0.933 0.773 0.781 0.957

4 0.925 0.554 1.000 0.668 0.981 0.937 0.946 0.988
1/100 8 0.981 0.941 1.001 0.970 0.995 0.985 0.991 0.997
16 0.995 0.989 1.000 0.996 0.999 0.997 - 0.999
32 0.998 0.998 1.000 0.999 1.000 1.000 - 1.000
2 0.741 0.001 0.991 0.001 0.931 0.771 0.778 0.956
4 0.924 0.032 0.998 0.034 0.979 0.933 0.943 0.987
1/1000 8 0.979 0.557 0.999 0.484 0.993 0.982 0.987 0.995
16 0.993 0.941 0.999 0.956 0.997 0.994 - 0.998
32 0.997 0.994 0.998 0.996 0.998 0.997 - 0.998
2 0.741 0.000 0.990 0.000 0.931 0.770 0.778 0.956
4 0.923 0.006 0.998 0.006 0.979 0.933 0.943 0.987
1/10000 8 0.979 0.091 0.999 0.092 0.993 0.982 0.987 0.995
16 0.993 0.527 0.999 0.589 0.997 0.994 - 0.998
32 0.997 0.922 0.998 0.934 0.998 0.997 - 0.998

This article is protected by copyright. All rights reserved.

Table II. Normalized vertical deflections at the center of free edge (point B ) of the
Scordelis-Lo roof problem. The mesh patterns are shown in Figure 7.
Type 1 Type 2 Regular
Cross mesh
t/L N regular mesh regular mesh mesh

MITC-S6 SC6R MITC-S6 SC6R MITC-S6 SC6R Sze et al. MITC-S8

2 1.074 0.747 0.746 0.573 1.045 0.964 - 1.192

4 0.740 0.625 0.670 0.590 0.872 0.829 0.892 0.943
6 0.819 0.735 0.786 0.704 0.923 0.890 0.946 0.960
8 0.878 0.811 0.858 0.803 0.952 0.928 0.975 0.973
0.005 10 0.914 0.863 0.900 0.858 0.967 0.951 0.992 0.980
12 0.936 0.898 0.926 0.894 0.977 0.965 1.003 0.984
14 0.951 0.921 0.943 0.919 0.984 0.974 1.014 0.987
16 0.961 0.938 0.955 0.937 0.988 0.980 1.014 0.989
32 0.987 0.986 0.985 0.985 1.010 0.997 - 0.994

This article is protected by copyright. All rights reserved.

Table III. Normalized deflections at the free end of the twisted beam in the direction of
applied load.
In-plane load Out-of-plane load
t/L N
MITC-S6 SC6R MITC-S8 MITC-S6 SC6R MITC-S8

2 0.981 0.996 0.990 0.832 0.825 0.918

4 0.990 0.996 0.995 0.922 0.907 0.973
0.0267
8 0.995 0.999 0.997 0.974 0.967 0.991
(t  0.32)
16 0.997 1.001 0.998 0.991 0.991 0.996
32 0.998 1.001 0.998 0.996 0.999 0.997
2 0.998 0.994 0.979 0.990 0.995 0.974
4 0.988 0.985 0.991 0.990 0.992 0.992
0.000267
8 0.993 0.995 0.997 0.996 0.998 0.997
(t  0.0032)
16 0.996 0.999 0.998 0.998 1.001 0.999
32 0.998 1.000 0.998 0.999 1.001 0.999

This article is protected by copyright. All rights reserved.

Figure 1. A 6-node triangular solid-shell element.

This article is protected by copyright. All rights reserved.

Figure 2. Nonlinear kinematics of the 6-node triangular solid-shell element. The covariant
base vectors at r  s  ξ  0 in the initial ( 0 ), previous ( t ) and current ( t  Δt )
configurations are depicted.

This article is protected by copyright. All rights reserved.

Figure 3. Tying positions for the transverse shear strains of the MITC-S6 solid-shell element.

This article is protected by copyright. All rights reserved.

Figure 4. Patch tests. (a) The mesh geometry. The loading and boundary conditions used for
the (b) membrane, (c) bending, and (d) transverse shearing patch tests.

This article is protected by copyright. All rights reserved.

Figure 5. An element used for the constant compression test.

This article is protected by copyright. All rights reserved.

Figure 6. Fully clamped square plate problem.

This article is protected by copyright. All rights reserved.

 Figure 7. Mesh patterns used for the benchmark tests: Regular meshes of (a) type 1 and (b)
type 2 for triangular elements, (c) cross mesh for triangular elements and (d) regular mesh for
quadrilateral elements.

This article is protected by copyright. All rights reserved.

Figure 8. Convergence of the normalized vertical deflections for the fully clamped square
plate problem with (a) t / L =1/100, (b) t / L =1/1000 and (c) t / L =1/10000. The solid and
dotted lines denote type 1 regular and cross mesh in Figure 7, respectively. The horizontal
lines represent the exact vertical deflection.

This article is protected by copyright. All rights reserved.

Figure 9. Convergence of the normalized stress- xx for the fully clamped square plate
problem. The horizontal lines represent the reference stress.

This article is protected by copyright. All rights reserved.

Figure 10. Scordelis-Lo roof problem.

This article is protected by copyright. All rights reserved.

 Figure 11. Convergence of the normalized vertical deflections for the Scordelis-Lo roof
problem with the type 1 regular mesh shown in Figure 7. The horizontal lines represent the
reference vertical deflection.

This article is protected by copyright. All rights reserved.

Figure 12. Convergence of the effective stress for the Scordelis-Lo roof problem.

This article is protected by copyright. All rights reserved.

Figure 13. Twisted beam problem ( 2  12 mesh). (a) Out-of-plane load case. (b) In-plane
load case.

This article is protected by copyright. All rights reserved.

Figure 14. Convergence of the normalized deflections for the twisted beam problem with (a)
in-plane and (b) out-of-plane load cases. The horizontal lines represent the reference
deflection.

This article is protected by copyright. All rights reserved.

Figure 15. Deformed shapes for the twisted beam problem ( 8 48 mesh, t  0.0032 ) for (a)
out-of-plane load and (b) in-plane load cases.

This article is protected by copyright. All rights reserved.

Figure 16. Irregular mesh patterns used ( N  4 ) for (a) triangular and (b) quadrilateral
elements.

This article is protected by copyright. All rights reserved.

Figure 17. Convergence curves for the fully clamped square plate problem with the (a)
regular and (b) irregular mesh patterns shown in Figure 16. The bold lines represent the
optimal convergence rate.

This article is protected by copyright. All rights reserved.

Figure 18. Hyperboloid shell problem. (a) Problem description. (b) The distributed loading.
(c) Graded mesh used for the membrane-dominated case (8×8 mesh, t / L  1 / 1000 ).

This article is protected by copyright. All rights reserved.

Figure 19. Convergence curves for the clamped hyperboloid shell problem with the (a)
regular and (b) irregular mesh patterns shown in Figure 16. The bold lines represent the
optimal convergence rate.

This article is protected by copyright. All rights reserved.

Figure 20. Convergence curves for the free hyperboloid shell problem with the (a) regular and
(b) irregular mesh pattern shown in Figure 16. The bold lines represent the optimal
convergence rate.

This article is protected by copyright. All rights reserved.

Figure 21. Cantilever beam under shear tip force.

This article is protected by copyright. All rights reserved.

Figure 22. Load-displacement curve for the cantilever beam under shear tip force.

This article is protected by copyright. All rights reserved.

Figure 23. Deformed shapes at shear tip forces P  1.0 , 2.0 or 4.0 for the cantilever beam
problem.

This article is protected by copyright. All rights reserved.

Figure 24. Pinched cylindrical shell.

This article is protected by copyright. All rights reserved.

Figure 25. Load-displacement curves for the pinched cylindrical shell.

This article is protected by copyright. All rights reserved.

Figure 26. Deformed shape at F  700 for the pinched cylindrical shell.

This article is protected by copyright. All rights reserved.

Figure 27. An 8-node quadrilateral solid-shell element.

This article is protected by copyright. All rights reserved.

Figure 28. Tying positions for the transverse shear strains of the MITC-S8 solid-shell element.

This article is protected by copyright. All rights reserved.