Anda di halaman 1dari 8

Finite Element Analysis of

Compression of Thin, High


Modulus, Cylindrical Shells with
Low-Modulus Core
Robert S. Joseph

Design Engineering Analysis Corporation, McMurray, PA

ABSTRACT
Long, cylindrical shells, of high modulus polymer with low
modulus elastomeric core, rest horizontally on the rigid
bottom of a groove with rigid side walls. At both sides, gaps
ranging from zero to approximately the dimension of the
shell thickness are allowed. Shell and core are assumed to
obey Hookes law. A uniformly distributed axial downward
acting load is applied to the top boundary. The system is
modelled using the ANSYS finite element program, Revision 5.0. The applied vertical load serves as the
independent variable. Dependent variables include the top
shell boundary reactions (loads and total deformation),
reaction at the side of the shell (load), and maximum von
Mises stresses and strains. Results can be reported numerically and graphically. The analytical model is
described briefly and its application is illustrated by three
examples. Purpose of this work is to provide parametric
trend data for estimating mechanical response of
AMPLIFLEX connector elements in reference 1.

1. INTRODUCTION
The behavior of certain elements of the AMPLIFLEX
connector was to be studied by the following model.l Cylindrical shells consisting of polyimide foil, an organic
polymer with relatively high modulus of elasticity, enclose a
core of low modulus silicone rubber. The shells are assumed to be of infinite length, and their cross sections can
be circular, oval or polygonal. They rest in a horizontal
groove with rigid bottom and side walls as shown schematically in Figure 1. Between the sides of the shells and the

side walls of the groove a gap of finite width can exist. At


the top, uniformly distributed parallel to the long axis of
the shell, a load is applied in a vertical, downward direction. The response to this load, in particular deformations
at the top and reactive loads at the top and the sides of the
shells, are of interest.
To avoid time consuming experimental studies requiring
preparation of parts with different shapes and dimensions,
the problem was to be modelled mathematically. Numerical analysis of mechanical systems has served design
engineers in finding optimal solutions for a long time. Usually, the system under consideration is described by a set of
higher order, nonlinear, partial differential equations and
boundary conditions specific to the system. Exact, closed
solutions of these problems are generally not possible.
Approximations were and still are developed by simplifying, sometimes drastically, the original mathematical
formulations. For a given system the degree of success of
this approach depends largely on the ingenuity of the analyst. If closed, exactor approximate solutions are not
required, the original problem can be rewritten in form of
difference equations. Using digital computers and observing the pertinent, system specific precautions, the rewritten
problem can then be solved with reasonable effort by conventional methods.2,3,4,5 For many of todays applications
even these approaches are unsatisfactory.
Difficulties encountered with these earlier conventional
procedures led to the development of the finite element
method (FEM). An early, fundamental discussion of its

Copyright 2004 by Tyco Electronics Corporation. All rights reserved.

16

R.S. Joseph

AMP Journal of Technology Vol. 4 June, 1995

and in Reference 1, the material nonlinearities (viscoelasticity, viscoplasticity, and hyperelasticity with the MooneyRivlin strain energy function) are available in ANSYS
should it become necessary to include these
approximations.

2. THE SYSTEM
Figure 1 shows the cross section of one of the examples
used in the study. Their symmetry and the assumption of
infinite length of the cylinders simplify the procedure
greatly. Three cases termed B0, Bl, and C where selected.
They represent combinations of different geometries and
boundary conditions:
B0 shell with circular cross section, rigid support at bottom, rigid support at both sides, load applied at top.

Figure 1. Cross section and schematic of support of one of


the examples analyzed. Infinite length of the cylinder was
assumed. Quantities are measured in conventional U.S.
units. Subscript o indicates outside dimensions of the shell, c
of the core.
h o = height of the shell,
wo = width of the shell,
hc = height of the core,
wc = width of the core,
ro = 0.5 wo = radius of curvature at top and bottom of the
outside,
rc = 0.5 wc = radius of curvature at top and bottom of the
inside,
t = 0.5 (ho h c ) = 0.5 (Wo wc) = shell thickness,
g = physical gap between sidewalls of shell and rigid support,
P = applied external load in lbs/in.

B1 shell with circular cross section, rigid support at bottom, gap between side walls of shell and support at both
sides, load applied at top.
C shell with oval cross section, rigid support at bottom,
rigid support at both sides, load applied at top.
Table 1 gives dimensions of the elements of each of the
examples, Table 2 the material constants for shell and core.
Justification for use of these constants and the linear materials model are given in reference 1. The effect of a finite
gap width between the side walls of the supporting structure and the shell is shown for a shell with circular cross
section.

Table 1. Dimensions used in the examples. Infinite length of


the cylinder was assumed. Definitions of the parameters are
given in Figure 1.

application to solving a number of non-trivial, specific engineering problems is presented for instance by Girault and
Raviart. 6 The most recent edition of Eshbachs Handbook
of Engineering Fundamentals contains a concise summary
of FEM, supported by selected examples and a brief bibliography. One of the most widely used and accepted FEM
codes in the world today is ANSYS8, introduced nearly 25
years ago by Swanson Analysis Systems, Inc.
Revision 5.0 of the ANSYS program provides extensive
nonlinear capabilities including geometric nonlinearities,
element nonlinearities, and material nonlinearities which
are required to solve contact problems of this type. In the
study described herein, the geometric nonlinearities (large
strain and large deflection effects) and element nonlinearities (contact surface elements with sliding and compression
capabilities are employed. Although not used in this study

AMP Journal of Technology Vol. 4 June, 1995

Table 2. Material constants used in the model. Shell and


core are assumed to obey Hookes law. Applied external
loads were 0.2, 1.0, 2.0, 4.0, 6.0 lb/in.

R.S. Joseph

17

3. THE FINITE ELEMENT ANALYSIS


Revision 5.0 of ANSYS is used to model and perform the
analysis of the long cylindrical shells discussed herein. A
one-half axial symmetry model of each geometry is developed using 2-D solid plane strain elements and contact
surfaces. Since the model exhibits reflective symmetry
along the length and the loading is symmetric, a one-half
symmetry model is only required for the solution. However,
for graphical presentation in section 4., the model results
are reflected so that the full model can be used to view the
displaced shape and the stress/strain contours. The ANSYS elements used to model the system described in

Table 3. ANSYS elements used to model the system described in section 2.

Figure 3. Finite element mesh for model C. The model exhibits reflective symmetry relative to the vertical, central
plane through its axis.

section 2. are listed in Table 3. The finite element meshes


for model B 0 with circular cross section and model C with
oval cross section are shown in Figures 2 and 3, respectively. The shell is modelled with one layer of 2-D
isoparametric elements (PLANE 42) with extra displacement shapes, which allow the elements to move more
flexibly. Friction between shells and the rigid supports is
assumed to be zero. For the purpose of the exploratory
study in reference 1, the modelling approximations regarding material properties, mesh sizes, friction and plain
strain end conditions are satisfactory.

Figure 2. Finite element mesh for model Bo. The model


exhibits reflective symmetry relative to the vertical, central
plane through its axis.

18

R.S. Joseph

The ANSYS program uses a frontal solver to solve the set


of simultaneous equations generated by the FEM. Since
geometric (large strain and large deflection) and element
(gaps) nonlinearities are included in the model, the program uses Newton-Raphson equilibrium iterations to
achieve convergence to a specified tolerance of 0.1%. The
solution results are saved on the results file and then they
can be conveniently reviewed (scanned, sorted, tabulated,
plotted) in the POST1 general postprocessor. A flow chart
illustrating the basic ANSYS concepts used in this analysis
is shown in Figure 4.

AMP Journal of Technology Vol. 4 June, 1995

4. THE FEM RESULTS


Tables 3 to 5 give summaries of the FEM results of particular interest for the three selected models. For a global view
they can also be represented graphically. Such graphs are
of importance if undesirable distribution of local stresses or
strains are to be identified. Figures 5 to 7 show the dis-

placements throughout the cross sections for the three


models. Figure 8 illustrates the von Mises strain distribution in shell and core for model C. In addition to these
more or less arbitrarily selected graphs, others can be generated from the ANSYS POST1 general postprocessor.

Table 4a. Summary of computed results for case Bo: Circular


cross section, no gap between shell and side walls of groove.
P is the load applied at the top of the shell. a) Reactions at
top boundary of shell; P/2 = total nodal contact force at top
boundary for 1/2 symmetry model = sum of the terms in the
column; = vertical displacement of top of shell.

Table 5a. Summary of computed results for case B1: Circular


cross section, gap of 1 mil between shell and side walls of
groove. P is the load applied at the top of the shell. a) Reactions at top boundary of shell; P/2 = total nodal contact
force at top boundary for 1/2 symmetry model = sum of the
terms in the column; = vertical displacement of top of
shell.

Table 4b. Reactions at side boundary of shell; Pside = total


normal load at the side boundary.

Table 5b. Reactions at side boundary of shell; Pside = total


normal load at the side boundary.

Table 4c. Maximum von Mises stress and strain;


Mises elastic stress; = von Mises elastic strain.

Table 5c. Maximum von Mises stress and strain;


Mises elastic stress; = von Mises elastic strain.

AMP Journal of Technology Vol. 4 June, 1995

= von

= von

R.S. Joseph

19

Figure 4. Flow chart illustrating ANSYS basic concepts.

Table 6a. Summary of computed results for case C: Oval


cross section, no gap between shell and side walls of groove.
P is the load applied at the top of the shell. a) Reactions at
top boundary of shell; P/2 = total nodal contact force at top
boundary for 1/2 symmetry model = sum of the terms in the
column; = vertical displacement of top of shell.

Table 6b. Reactions at side boundary of shell; Pside = total


normal load at the side boundary.

Table 6c. Maximum von Mises stress and strain;


Mises elastic stress; = von Mises elastic strain.

20

R.S. Joseph

= von

AMP Journal of Technology Vol. 4 June, 1995

Figure 5. Displacement plot for model B0, a) for applied load P = 0.2 lb/in, b) for applied load P = 6.0 lb/in.

Figure 6. Displacement plot for model B1, a) for applied load P = 0.2 lb/in, b) for applied load P = 6.0 lb/in.

Figure 7. Displacement plot for model C,


a) for applied load P = 0.2 lb/in,
b) for applied load P = 6.0 lb/in,
c) enlargement of upper portion of Figure 7b.

22

R.S. Joseph

AMP Journal of Technology Vol. 4 June, 1995

Figure 8. Plots of von Mises strain for model C at applied load P = 6.0lb/in, a) for the shell, b) for the core.

5. REFERENCES
1. E. W. Deeg, Mechanics of AMPLIFLEX Connector
Elements, AMP J. of Technol. 4 (1994), pp 24 to 40.
2. E. G. Keller and R. E. Doherty, Mathematics of Modern
Engineering, Volume I, (Wiley, New York, 1936), 163188.
3. H. T. Davis, Introduction to Nonlinear Differential and
Integral Equations, (Dover, New York, 1962), 467-488.
4. R. W. Hamming, Numerical Methods for Scientists and
Engineers, 2nd edition, (McGraw-Hill, New York, 1973).
5. M. E. Goldstein and W. H. Braun, Advanced Methods
for the Solution of Differential Equations, (NASA,
Washington, D. C., 1973), 320-345.
6. V. Girault and P.-A. Raviart, Finite Element Approximation of the Navier-Stokes Equations, (Springer, Berlin,
1979), 58-86.
7. J. N. Reddy in Eshbachs Handbook of Engineering Fundamentals, 4th ed. edited by B. D. Tapley, (Wiley, New
York, 1990), 2.145-2.168,2.191.

Robert S. Joseph is President and co-founder of Design


Engineering Analysis Corporation (DEAC), a professional
engineering consulting firm based in McMurray, PA.
Mr. Joseph earned his B.S. in Engineering Mechanics from
Pennsylvania State University in 1966 and his M.S. in Civil
Engineering from the University of Pittsburgh in 1971. He
started his professional career at Westinghouse Astronuclear Laboratory where he was employed for seven
years. There he was responsible for static, dynamic, and
stability analysis of various metallic and graphite components of the NERVA nuclear rocket engine.
During the past 21 years Mr. Joseph has worked as a consultant to both, industry and government agencies. He has
been extensively involved in the application of finite element analysis methods to solve a wide variety of complex
engineering problems involving static, dynamic, inelastic,
large deflection, and heat transfer analyses in many diverse
industries. He has published several technical papers dealing with structural dynamics using finite element methods
and has taught short courses on Section VIII, Division 1, of
the ASME code. Mr. Joseph is a Registered Professional
Engineer in Pennsylvania and a member of the American
Society of Mechanical Engineers.

8. ANSYS Users Manual for Revision 5.0, vol. I to IV.


Developed by Swanson Analysis Systems, Inc.,
Houston, PA.

AMP Journal of Technology Vol. 4 June, 1995

R.S. Joseph

23