Sie sind auf Seite 1von 8

Example 10 - Bending

Summary
The bending of a straight cantilever beam is studied. The example used is a famous bending test for shell
elements. The analytical solution enables the comparison with the quality of the numerical results.
Carefully watch the influence from the shell formulation. In addition, the results for the different time step
scale factors are compared.
Title
Bending
Number
10.1

Brief Description
Pure bending test with different 3- and 4-nodes shell formulations.
Keywords

Q4 and T3 meshes

QEPH, Belytshcko&Tsay, BATOZ, DKT shells

Mesh, hourglass, imposed velocity, quasi-static analysis, bending test

RADIOSS Options

Imposed velocity (/IMPVEL)

Rigid bodies (/RBODY)

Compared to / Validation Method

Analytical solution

Input File
BATOZ:
<install_directory>/demos/hwsolvers/radioss/10_Bending/BATOZ/.../ROLLING*
QEPH: <install_directory>/demos/hwsolvers/radioss/10_Bending/QEPH/.../ROLLING*
BT (type1):
<install_directory>/demos/hwsolvers/radioss/10_Bending/BT/BT_type1/.../ROLLING
*
BT (type3):
<install_directory>/demos/hwsolvers/radioss/10_Bending/BT/BT_type3/.../ROLLING
*
BT (type4):
<install_directory>/demos/hwsolvers/radioss/10_Bending/BT/BT_type4/.../ROLLING
*
DKT18:
<install_directory>/demos/hwsolvers/radioss/10_Bending/DKT18/.../ROLLING*
RADIOSS Version
51e
Technical / Theoretical Level
Beginner benchmark

Overview
Physical Problem Description
The purpose of this example is to study a pure bending problem. A cantilever beam with an end moment
is studied. The moment variation is modeled by introducing a constant imposed velocity on the free end.
Units: mm, ms, g, N, MPa
Several kinds of element formulation are used.
The material used follows a linear elastic law (/MAT/LAW1) and has the following characteristics:

Initial density: 0.01 g/mm 3

Reference density: .01 g/mm 3

Young modulus: 1000 MPa

Poisson ratio: 0

Fig 1: Geometry of the problem.

Analysis, Assumptions and Modeling Description


Modeling Methodology
Three beams are modeled using quadrilateral shells and one beam with T3 shells. A rigid body is defined
at the end of each beam for applying the bending moment.
The four models are integrated into one input file. The shell element formulations are:

Q4 mesh with the Belytshcko&Tsay formulation (I shell =1, hourglass control type 1, 2, 3)

Q4 mesh with the QEPH formulation (Ishell =24)

Q4 mesh with the QBAT formulation (Ishell =12)

T3 mesh with the DKT18 formulation (Ishell =12)

RADIOSS Options Used


At one extremity of the beam, all d.o.f. are blocked. A rotational velocity is imposed on the master node of
the rigid body placed on the other side.
This velocity follows a linear function: Y=1

Fig 2: Beam meshes.

Simulation Results and Conclusions


Numerical Results Compared to Analytical Solutions
As shown in Fig 1, rotation around X and displacement with regard to Y of the free end are studied.
The analytical solution of the Timoshenko beam subjected to a tip moment reads:
(a)
which yields the end moment for a complete loop rotation 2

KN-mm

Fig 3: Bending of beam

The following tables summarize the results obtained for the different formulations. From an analytical point
of view, the beam deformed under pure bending must satisfy the conditions of the constant curvature
which implies that for = 2 , the beam should form a closed ring. However, depending on the finite
element used, a small error can be observed, as shown in the following tables. This is mainly due to beam
vibration during deformation as it is highly flexible. Good results are obtained by the QBAT, QEPH and
DKT18 elements, respectively. This is mainly due to the good estimation of the curvature in the
formulation of these elements. The BT family of under-integrated shell elements is less accurate. With the
type 3 hourglass formulation, the model remains stable until = 6rad. However, the moment-rotation
curves do not correspond to the expected response.
To reduce the overall computation error, smaller explicit time steps are used by reducing the scale factor
in /DT. The results reported in the end table show that a reduction in the time step enables to reduce the
error accumulation, even though the divergence problems for BT elements cannot be avoided.
The following parameters are chosen for drawing curves and displaying animations:
BATOZ

QEPH

BT

DKT

Scale factor

0.6

0.9

0.9

0.2

Imposed velocity
rot.

0.005 rad/ms

0.005 rad/ms

0.005 rad/ms

0.005 rad/ms

The following curves show the evolution previously shown (rotation and nodal displacement by moment):

Fig 4: Moment versus rotation around X.

Fig 5: Moment versus displacement along Z.

Fig 6: Moment versus rotation around X.

BATOZ
Sf=0.
9

Sf
=0.8

QEPH
Sf
=0.6

Sf
=0.9

Sf
=0.8

BT
Type 1
Sf
=0.9

DKT

Type 3

Sf Sf =0.9 Sf =0.1
=0.1

Type 4
Sf
=0.9

Sf
=0.1

Sf
=0.3

Sf
=0.2

Sf =0.1

CPU
42.6
(normalize 2.18 2.43 3.14 1.23 1.34
4
7.07
2.62 108.60 1.03 7.17 5.44 8.21 16.21
d)
9530 10721 14295 9357 10527 8153 73080 25150 1824124 8133 73066 42004 62945 125867
# cycles
5
9
3
7
4
8
8
1
8
2
6
1
0
8
Error
=2
(%)

0%

0%

0%

0%

0%

99%

99%

-78%

-53%

99.9 99.9% 15%


%

6.91

6.89

4.36

4.53

6.06

5.98

4.38

4.51

6.37

396

395

250

260

347

343

251

258

365

err =20%

8.9%

1.5%

(rad)
degree
Dz

= 2

(mm)

Mx

= 2
+5
(x10 kN-mm)

-500.5 -500.5
-500.5
-525.8
500.5
500.
491.
518.33
5
2
3

-506.0

-433.8 -476.5 -496.5 -499.4


529.8

-4.04 -4.05

-2.38

-0.07 -0.02

-4.06 -4.01 -4.01 -0.21 -0.11

-3.13

-3.09

-3.02

-3.08

Conclusion
A description summary of the different tests is provided below:

QBAT element:
This formulation gives a 2 -revolution of the beam with no energy error. However, a 20% error is
attained for = 384.
Note that the decrease of the scale factor enables obtaining better results.
QEPH element:
This formulation seems to be the best one to treat the problem. It enables a 2
beam to be obtained. The error remains null until = 400.

-revolution of the

BT formulation:
This formulation does not provide satisfactory results and is not adapted to this simulation,
whatever the anti-hourglass formulation. This is mainly due to using a flat plate formulation and the
fact that the element is under-integrated. The type 3 hourglass formulation seems to be better than
others.
For DKT formulation:
The bending is simulated correctly. However, the element is costly and the CPU time is much
longer.

Das könnte Ihnen auch gefallen