Convergence behavior of the MITC3+ triangular shell element

*Youngyu Lee1) and Phill-Seung Lee2)

1)
Agency for Defense Development, Daejeon 305-600, Korea
2)
Department of Mechanical Engineering, KAIST, Daejeon 305-701, Korea
1)
[email protected]

ABSTRACT

In this study, we present the convergence behavior of the MITC3+ triangular shell
element, which has been recently developed (Lee et. al, 2014, Jeon et. al, 2015, Lee et.
al, 2015). A cubic bubble function for the rotations is used to enrich the bending
displacements and the corresponding degrees of freedom can be statically condensed
out on the element level. The MITC (Mixed Interpolation of Tensorial Components)
method is employed to alleviate shear locking. The MITC3+ shell element passes all
the basic tests (the patch, zero energy mode and isotropy tests). In convergence
studies, excellent convergence behaviors are shown. In uniform meshes, the
convergence behavior of the MITC3+ shell element is as good as that of the 4-node
MITC4 shell element, which has been widely used in finite element analysis of shells. In
distorted meshes, the MITC3+ shell element shows much better convergence behavior
than the MITC4 shell element, particularly, in bending-dominated shell problems.

1. INTRODUCTION

The finite element method has been dominantly applied to analyze shell structures.
It is important to use reliable and effective shell finite elements. There is still required to
improve predictive capability of shell elements (Bathe, 1996, Chapelle and Bathe,
2011). One of important issues is to develop the effective 3-node triangular shell
element, which is very useful for automatic mesh generations of complex shell
structures (Lee and Bathe, 2004, Lee et. al, 2007, Lee et. al, 2012).
Recently, a new 3-node triangular shell finite element (MITC3+) was developed
(Lee et. al, 2014). A cubic bubble function is used for the interpolation of the rotations
to enrich the bending displacement fields. The corresponding rotation degrees of
freedom can be statically condensed out on the element level. A new assumed
transverse shear strain field was established with a new tying scheme to reduce shear
1)
Senior Researcher
2)
Associate Professor
locking while satisfying the consistency and ellipticity conditions. The MITC3+ shell
element passes the three basic tests (the patch, zero energy mode and isotropy tests)
and shows excellent convergence behaviors in various shell problems even when
highly distorted meshes are used (Lee et. al, 2014).
In this presentation, we show the convergence behavior of the MITC3+ triangular
shell element. The results are compared with the 4-node MITC shell element (MITC4),
which has been widely used in various commercial software.
In the following sections, we review the formulation of the MITC3+ triangular shell
finite element and its performance is briefly presented. We then present conclusions.

2. The MITC3+ triangular shell finite element

The geometry interpolation of the MITC3+ shell element, shown in Fig. 1, is given
by (Lee et. al, 2014)

t Bubble node
s

1
4
r
Vn2

r
V22 z, w
r r
r
V12 2
iz
iy
y, v
r r
ix
x, u
Fig. 1. Geometry of the MITC3+ shell element with an additional bubble node.

 3
 t 4   1   
x (r , s, t )   hi (r , s) xi   ai f i (r , s) Vni with a 4Vn4  (a1Vn1  a 2Vn2  a3Vn3 ) , (1)
i 1 2 i 1 3

in which hi (r , s ) is the two-dimensional interpolation function of the standard


isoparametric procedure corresponding to node i , xi is the position vector of node i
 
in the global Cartesian coordinate system, and a i and Vni denote the shell thickness
and the director vector at the node, f i (r , s) are two-dimensional interpolation functions
that include the cubic bubble function f 4 corresponding to the internal node 4

1 1 1
f1  h1  f4 , f 2  h2  f4 , f 3  h3  f4 , f 4  27 rs (1  r  s ) . (2)
3 3 3

From Eq. (1), we obtain the displacement interpolation

 3
 t 4  
u (r , s, t )   hi (r , s)ui   ai f i (r , s) (V2i i  V1i  i ) , (3)
i 1 2 i 1

in which ui is the nodal displacement vector in the global Cartesian coordinate system,
  
V1i and V2i are unit vectors orthogonal to Vni and to each other, and  i and  i are
  
the rotations of the director vector Vni about V1i and V2i , respectively, at node i (  4
and  4 are the rotation degrees of freedom at the bubble node).

s s

1
1
: Tying points : Tying points
2/ 3 ( A)
(E ) (F )

(C ) (B )
1/ 6
(D)

0 1/ 6 2/3 1 r 0 1 r
2d d

(a) (b)
Fig. 2. Tying positions (A)-(F) used for the assumed transverse shear strain field of the
MITC3+ shell finite element.

The geometry of the MITC3+ shell element is flat because the bubble node with
only rotation degrees of freedom is positioned on the flat surface defined by the three
corner nodes of the element. Hence, only the transverse shear strain components are
assumed using the MITC method. The assumed transverse shear strain field is given
by (Lee et. al, 2014)
 2 ( B ) 1 ( B ) 1 (C ) 1
eˆrtMITC 3  (ert  e st )  (ert  e st(C ) )  cˆ(3s  1) ,
3 2 3 3
2 1 1 1
eˆ stMITC 3  (e st( A)  ert( A) )  (ert( C )  e st( C ) )  cˆ(1  3r ) , (4)
3 2 3 3

in which cˆ  ert( F )  ert( D)  est( F )  est( E ) and the 6 tying points (A)-(F) with the tying distance
d are shown in Fig. 2. A fixed value d  1 / 10,000 is used in Ref. (Lee et. al, 2014).
The partly clamped hyperbolic paraboloid shell problem shown in Fig. 3 is
considered (Lee and Bathe, 2002). The surface is defined as

Z  X 2  Y 2 ; ( X , Y )  [ L / 2; L / 2)]2 , (5)

Fixed

L/ 4 Sym.

z 0 C A - L/ 2

- L/ 4

- L/ 2
Free
0
Free y
0
D
x
L/ 2
L/ 2

Fig. 3 Partly clamped hyperbolic paraboloid shell problem ( L  1.0 , E  2  1011 and
  0.3 ).

L4 L3 L2 L1 L1 L2 L3 L4
B A

C D
Fig. 4 Distorted mesh pattern for N  8 .
 MITC3+ MITC4 MITC3+ MITC4
0 0 0 0

-0.6 -0.6 -0.6 -0.6

-1.2 -1.2 -1.2 -1.2

log E h -1.8 log E h -1.8 log E h -1.8 log E h -1.8

t/L = 1/100 t/L = 1/100 t/L = 1/100 t/L = 1/100

-2.4 t/L = 1/1000 -2.4 -2.4 -2.4
t/L = 1/1000 t/L = 1/1000 t/L = 1/1000
t/L = 1/10000 t/L = 1/10000 t/L = 1/10000 t/L = 1/10000

-3 -3 -3 -3

-3.6 -3.6 -3.6 -3.6

-2.0 -1.6 -1.2 -0.8 -2.0 -1.6 -1.2 -0.8 -2.0 -1.6 -1.2 -0.8 -2.0 -1.6 -1.2 -0.8
log h log h log h log h
(a) (b)

Fig. 5 Convergence curves for the partly clamped hyperbolic paraboloid shell problem.
The bold line represents the optimal convergence rate. (a) Uniform mesh. (b) Distorted
mesh.

and clamped along the side X   L / 2 and loaded by its self-weight. This shell
problem is classified as a bending dominated behavior. Due to symmetry, only one half
model is considered. Both the uniform mesh shown in Fig. 3 and the distorted mesh
shown in Fig. 4 are considered. The s-norm is used for the convergence studies (Hiller
and Bathe, 2003). Fig. 5 presents the convergence curves of the MITC3+ and MITC4
shell elements. Even though the distorted mesh is used, the MITC3+ shell element still
shows good results unlike the MITC4 shell element.

3. CONCLUSIONS

We presented the performance of the MITC3+ triangular shell element recently

developed in convergence studies. The results are compared with the MITC4
quadrilateral shell element, which has been widely used in finite element analysis of
shells. The MITC3+ shell element showed the excellent behavior even when distorted
meshes are used. In uniform meshes, the performance of the MITC3+ shell element is
as good as that of the 4-node MITC4 shell element. In distorted meshes, the MITC3+
shell element presented much better performance than the MITC4 shell element in the
bending-dominated shell problems. We expect that, due to its superior performance,
the MITC3+ triangular shell element will be widely used in finite element analysis of
shells.
REFERENCES

Bathe K.J. (1996), “Finite element procedures”, Prentice Hall, New York.
Chapelle D., Bathe K.J. (2011), “The finite element analysis of shells – fundamentals,
Second Edition”, Springer-Verlag, Berlin.
Hiller J.F., Bathe K.J. (2003) “Measuring convergence of mixed finite element
discretizations: an application to shell structures”, Computers and Structures, 81,
639-654.
Jeon, H.M., Lee Y., Lee P.S. and Bathe K.J. (2015), “The MITC3+ shell element in
geometric nonlinear analysis”, Computers and Structures, 146, 91-104.
Lee Y., Jeon, H.M. and Lee, P.S. and Bathe K.J. (2015), “The modal behavior of the
MITC3+ triangular shell element”, Computers and Structures, 153, 148-164.
Lee Y., Lee, P.S. and Bathe, K.J. (2014), “The MITC3+ shell finite element and its
performance”, Computers and Structures, 138, 12-23.
Lee Y., Yoon K., Lee P.S. (2012) “Improving the MITC3 shell finite element by using the
Hellinger-Reissner principle”, Computers and Structures, 110-111, 93-106.
Lee P.S., Noh H.C., Bathe K.J. (2007), “Insight into 3-node triangular shell finite
elements : the effects of element isotropy and mesh patterns”, Computers and
Structures, 85, 404-418.
Lee, P.S., Bathe, K.J. (2004), “Development of MITC isotropic triangular shell finite
elements”, Computers and Structures, 82, 945-962.
Lee, P.S., Bathe, K.J. (2002), “On the asymptotic behavior of shell structures and the
evaluation in finite element solutions”, Computers and Structures, 80, 235-255.