Вы находитесь на странице: 1из 9

Engineeringwith Computers(1993) 9:27-35

9 1993 Springer-VerlagLondonLimited

Engineering
C~nputers

Form Finding of Shells by Structural Optimization


K.-U. Bletzinger and E. Ramm
Universityof Stuttgart,Institutfiir Baustatik,D-7000 Stuttgart 80, Germany

Abstract. Shell structures are known to be extremely


parameter sensitive; even small changes of the initial design,
e.g., to the shape of the shell, may drastically change the
internal stress state. The ideal case for concrete shells is a
pure membrane stress state in compression for all loading
conditions. Since in many realistic situations the solution for
an 'optimal' shape is not obvious, the need for form finding
methods is evident. This paper presents computational methods
of structural optimization as a general tool for the form finding
of shells. The procedure as a synthesis of design modelling,
structural analysis and mathematical optimization is discussed
with special emphasis on the modelling stage. Several examples
show the power of the approach and the similarities to
experimental solutions.

changes of certain parameters, e.g., the reduction of


the buckling load due to only small initial imperfections. Since realistic situations of design, such as single
loads, support conditions, free edges, or shape
incompatibilities, make it very difficult to fulfil the
basic membrane-oriented design rules, a modification
of the original design could substantially improve the
structural behaviour. Additionally, the specific properties of the chosen material have to be considered.
It is obvious that the shape of a concrete shell should
be designed to avoid tension to the greatest possible
extent. The ideal situation would be a pure membrane
state in compression, hopefully avoiding any buckling.
The problem is to find shapes of that kind.

Keywords. Computer aided geometrical design; Form


finding; Shells; Structural optimization
1.2. Traditional Form Finding Methods

1. Introduction
1.1. Shells as Structures of Optimal Behaviour
Shells are the most efficient structural alternative for
a number of extreme situations such as structures of
long spans, minimum mass, or high resistance [13. At
the same time, they appear very light and graceful,
and meet aesthetical demands in a natural manner.
There is no doubt that shells are the epitome of
structural elegance. The extraordinary behavior of
shell structures is caused by the 'double arch effect'
which, in contrast to one-dimensional curved structures, allows them to carry several different load
constellations by a pure membrane action. That means
that shells designed to act as membranes are already
optimal structures. In some cases they may also show
typical characteristics of 'over-optimized' specializations with high sensitivity with respect to small
Correspondence and offprint requests to: K.-U. Bletzinger,Univer-

sitfit Stuttgart, Institut f/Jr Baustatik, Pfaffenwaldring7, Postfach


801140, D-7000 Stuttgart 80, Germany.

It can be recognized that most shells are of regular


shapes which are often analytically defined. This
reflects interactions of the usual design practice and
classical shell theory which gives closed solutions for
analytical defined geometries only. Typical design
practice is to experiment with standard geometries like
spheres, cylinders, toil, cones and HP-surfaces. To
adjust the final shape to the prescribed plan, segments
are cut out or different shell types are put together,
neglecting the basic conditions of membrane theory
with respect to both equilibrium and compatibility.
This leads to high bending stresses or large displacements, which are usually avoided by additional
stiffening elements. Typical examples are the cylindrical roof shell with edge beams and diaphragm walls
at the ends, or the spherical shell over a polygonal
plan with heavy beams at the free edges.
An experimental principle long used to find the
optimal shape of structures in compression is the
hanging model and its inverse. The inverted catenary
was used, for example, in 1748 by Giovanni Poleni to
compare the shape with Michaelangelo's design of St.
Peter's in Rome, and extensively by Gaudi at the
beginning of this century for many structures around

28

Barcelona (chapel in Colonia Gfiell, Sagrada Familia,


etc.). Extended to two dimensions, the hanging fabric
can be used to find shapes of shells over almost
arbitrary plans. In many cases, heavy edge beams can
be avoided by this method, to yield pleasing, naturally
shaped shells with free edges. This principle was used
very successfully and economically in many practical
applications by H. Isler [-2, 3].
The basis of the hanging model experiment is that
one characteristic load case is used to generate the
final shape by large deflections of a given membrane.
So far, the method can easily be simulated by modern
analysis codes which are able to consider geometrically
non-linear structural behaviour. The procedure is
powerful [1] but has some drawbacks: for example, it
is not possible to find a compromise when different
load cases are dominant, or to consider criteria for the
genesis of shapes not based on elastic deformations.

2. Structural Optimization - A General


Tool of Shape Design
Computational methods which offer a more general
approach to shell design than the principle of inverting
hanging models are the procedures of structural
optimization. Their foundation is a strict separation
of structural geometry, mechanical behaviour and the
design objective which is responsible for the generation
procedure. The idea of form finding by these methods
is therefore very much related to the obvious
engineering approach:

K.-U. Bletzinger a n d E. R a m m

which are not necessarily related to mechanical


behaviour, and loads such as body forces and support
conditions which change with every modification of
shape and which can only be solved with difficulty by
experiments, and sometimes not even then.
The bulk of investigations is still devoted towards
optimization of cross-sections. In shape optimization,
most work addresses two-dimensional systems [4].
Shell problems are usually restricted to the axisymmetric case; relatively little has been described for
general shells so far [5]. Usually, as in this paper,
non-linear structural response is not taken into
account because of the exponential increase in the
complexity of the problem. Considerable research is
currently underway to include these phenomena.
The methods of structural optimization have
reached a remarkable level [6], and they have been
used as design tools to improve structural quality in
many industrial applications, especially in aircraft [7]
and automotive as well as in other mechanical
industries. Their potential for alternative design in civil
engineering has not yet been fully exploited. This lack
of practical acceptance may be explained by some
frustrating experiences in the 1960s and early 1970s.
But progress in computational methods and hardware
since those times is also considerable, and this has
rekindled interest in structural optimization.
A typical problem of structural optimization is
characterized by an objective f(x) and constraints 9(x)
and h(x) which are non-linear functions of the
optimization variables x. It can be stated as:

f(x)
subject to: hj(x) = 0;
minimize:

(1) choose a shape;


(2) evaluate the structural behaviour according to the
given load cases and support conditions, by finite
element methods, for example;
(3) check stresses, displacements, buckling load and
other safety requirements;
(4) compare the quality of the design with the chosen
optimality criteria;
(5) if necessary, propose a new and better design by
means of sensitivity analysis, and repeat the
process.
This procedure is absolutely different from the
hanging model and other related principles where the
generating rule itself (i.e., mechanical reactions to
given loads) is already the criterion for optimality.
Nevertheless, the same results can be achieved if
equivalent objectives and load conditions are chosen.
Because of their general formulation, methods of
structural optimization can tackle problems with
many load conditions, arbitrary design objectives

gj(x) ~

0;

x L~<x~<xu;

1,...,

me

j -= m e -t- l . . . .

(1)
, m

x~R"

where xL and x U are lower and upper bounds on the


variables, respectively.
It is now possible to consider, for typical objective
functions which are used in the sense of the design
rules above discussed, how to improve the structural
behaviour substantially. For example, strain energy fE
is an appropriate objective to achieve shapes which
act in a membrane state of stresses, i.e., tension and
compression but no bending:
fE = ~

ere d V

(2)

To generate shapes which are able to carry loads

Form Finding of Shells

29

mainly in compression, stress levelling fs can be used

E8]:
fs = fv (~ - a")2 dV

(3)

where cr, is a prescribed goal of stress. There are other


objective functions, such as construction cost, weight,
or natural frequencies, which are interesting in the
shell structure design and often dominate the two
objectives of natural significance already mentioned
above. Applying the methods of multicriterion optimization, several, even conflicting, objectives can be
considered simultaneously to obtain an 'optimal'
compromise of structural design. This is done, for
example, when strain energies of more than one load
case have to be minimized (see the example in
subsection 4.2).
Inequality constraints g(x) are taken into account
to impose safety and reliability requirements. Typical
constraints of this type are stress and displacement
limits. If the stiffness of a structure is to be maximized
for a prescribed structural mass, an equality constraint
h(x) is introduced. Otherwise, unrealistically massy
solutions could be obtained, as is usually the case
if external loads dominate the self-weight of the
structure.
Form finding implies optimization of geometry.
Characteristic optimization variables are therefore
geometric parameters defining the structural shape.
The number of variables can be reduced dramatically
without loss of generality if CAGD concepts are used
[-9, 10]. By these methods, shapes of free form shells
can be described by the coordinates of a few so-called
'design-nodes' which can be chosen as variables. In
addition, the thickness variation can be optimized
where discrete thicknesses at design-nodes are taken
as variables. The use of these methods in modelling
and modifying surfaces will be described in more detail
in the next section.
Form finding of shells results in non-linear optimization problems which exhibit many different properties of mathematical optimization. Depending on the
objective (strain energy, weight, etc.), the constraints
(equality, non-equality) and their combinations, the
optimization problems vary from totally unconstrained (often stress levelling as objective) via little
constrained but with equality constraints (strain
energy minimization with fixed structural mass) to
highly constrained problems like weight minimization,
which tends to reduce mass until the limit of material
resistance is reached. Therefore, only robust and
sophisticated methods can be recommended, such as
SQP techniques (sequential quadratic programming)

[11], which are able to handle all these problems, or


approximation methods such as the method of moving
asymptotes [-12] (MMA), which are superior in special
cases like weight minimization. Both methods have
been successfully used in shape optimal design of free
form shells [-8, 9, 13].
As already mentioned, structural optimization is
understood to be a synthesis of various individual
disciplines [10]: (a) design modelling (CAGD),
(b) structural analysis (e.g., FEM), (c) behavioursensitivity analysis, (d) mathematical programming
and (e) interactive computer graphics as an important
additional aid. Expertise in all these fields is necessary
to get satisfactory results. The authors developed their
own program system CARAT (Computer Aided
Research Analysis Tool) [14, 15] which provides
the designing engineer with coordinated program
modules. Design modeller, FE-analysis and optimization algorithms are coordinated from the root of
development and are integrated in a general design
procedure based on a unique in-core database which
allows fast data exchanges without any loss of
accuracy.

3. Modelling and Modifying Structural


Shapes
Design modelling, as an important part of structural
optimization, is the backbone of the whole procedure.
The general methods of Computer Aided Geometrical
Design (CAGD) [16, 17] are the basis of modern
pre-processors to design structural geometries in two
and three dimensions. Shapes are approximated
piecewise by 'design patches'. Within each design path,
the resulting shape r is parametrized in terms of shape
functions ~i, patch parameters u, v, w, and design
nodes rdi which describe the location of the patch in
space:

r(u, v, w) = ~ ~i(u, v, w)r~i

(4)

i=1

Many different shape functions are available, e.g.


Lagrangian interpolation, Coons' transfinite interpolation, B6zier and B-spline approximations. Depending on their formulation, a huge variety of shapes
can be described without severe restrictions on the
manifold solutions. In shape design of free form shells,
one-dimensional cubic B4zier and B-splines, and
two-dimensional bi-cubic B6zier patches (Fig. 1)
appear to be superior to others. This is because only
the corner nodes of those design patches interpolate
the resulting shape. The remaining inner nodes only
approximate the shape, which yields no differences in

30

K.-U. Bletzingerand E. Ramm

continuity

d e s i g n nodes

a) four B6zier

patchesdefininga plate

continuity
patches

roo
Fig. 1. B6zierpatch.

b) shift designnodes
the results compared with equivalent Lagrange
interpolation schemes, but allows the construction of
continuity conditions between adjacent design patches
if composite surfaces are to be defined. This is
demonstrated by an example which also reflects the
interactive capabilities of CARAT. A plate is defined
by four 16-noded B6zier patches as shown in Fig. 2.
These elements are connected in such a way that the
shape generated is continuous in slopes across the
common edges of adjacent patches. To obtain this,
corresponding design nodes of the involved patches
have to remain on a common line during all
subsequent shape modifications. The same rule holds
for the second dimension which leads to linear
dependencies between, at most, nine design nodes.
These topological relations are formulated in superimposed 'continuity patches' [9]. They are generated
automatically and preserved during manual user
interactions and shape optimization (Fig. 2(a), (b), (c)).
Figure 2 shows different types of continuity patches
depending on whether they are connecting two or four
design patches, or if they are defined at an isolated
corner. In all cases, four nodes are independent and
control the locations of the remaining nodes, leading
to a reduction of geometrical degree of freedom which
is very welcome in structural optimization to stabilize
the procedure.
The idea of continuity patches is very helpful in
interactive design of free form shells which can serve
as initial shapes for subsequent optimization runs or
as valuable interactive pre-processor tools for input
preparation of complex shapes. Figure 3 shows the

c) generatedcontinousshape
Fig. 2. Interactive surface modification; continuity patches connecting four B6zierpatches.
plan of a free form shell described by a total of 16
B6zier patches, and the generated shape modelled by
8-noded isoparametric shell elements. The generated
model of a sea urchin shell is shown in Fig. 4, and
was the subject of a biomechanical study together with
biologists [18]. The stiffening effects of the wrinkles
were the main objective of this investigation. The shape
was interactively adjusted to measured data using
B-splines which were linearly blended in cylindrical
coordinates (Coons' interpolation).
Another important fact which must be considered
when 'membrane' shapes are to be determined is the
generation and modification of corresponding support
conditions. CARAT supplies rules to generate support
conditions which are tangential and normal to surfaces
or edges, respectively, and remain so during the whole
form finding process.

Form Finding of Shells

31

Fig. 3. Free form shell using


16 B~zier patches.

ground plan

generated shape

4. Examples
4.1. Bi-parabolie Roof Shell

Fig. 4. Finite element model of a sea urchin shell.

This example is used to demonstrate the effects of


different objective functions and the variety of shapes
which can be generated by using only two variables.
The structural situation is shown in Fig. 5. A shell of
rectangular plan (b = 6 m , l = 12m) and uniform
constant thickness (t = 0.05m) is supported by
diaphragms at the smaller edges. The shape is
generated by four B6zier patches. The design nodes
are linked (a) to preserve double symmetry and (b) to
describe a bi-parabolic surface which can be controlled
by two vertical coordinates as indicated. In the initial
design, both coordinates are set to sl/2 = 3 m ,
describing a cylindrical shell. The structure is loaded

linked design node


parabolic shape functions
variable

~---'--b = 6m-----~ "~t ~-----~1


Fig. 5. Parabolic roof shell:
problem statement.

diaphragm

= 12mr

material properties: E = 30,000 MPa, v = 0.2 (concrete)

32

K,-U. Bletzinger and E. Ramm

optimizer: SQP
intial values:
Sl = 3m
s2 = 3m

a) initialshape

1200
Z~
~

optimal values:
s2* = 3.12m

1080
960
840

e~

720
600

0 1 2 3 4 5
iterations

360,

b) optimal

shape, minimization of strain energy

optimal values:
Sl* 0.90m
s2* = 1-97m
=

340 i

280 k

~ 260, ~ .
240:
012345678
iterations
36 i
34

c) optimal shape, stress leveling


t

optimal values:
sl* = 1.64m
s2* = 1.34m

~ 32

.~

30

28
26
24

0123456
iterations

d) optimal shape, weight minimi~tion

Fig. 6. Parabolic roof shell; initial design and optimal


solutions.

Form Finding of Shells

33

oround olan:

load:

load cases:

snow:

~!:!i!i~:.i!i!i~:.~!~!iiii~i:i~iiii!~i!iii!:i:~:~:!!i!i~:i!i!ii!i~i~i!!i!!%i|P

p=5~
nl 2

I::iii]i:i~iii~i~i~iii!i]!iii:ii!i!:~i:i]

and

'~

~iiiiiiiiiiiiiiiii~,,

lOm

[i!!i!!!i!i!!ii!i!i!i~-i!i!:ii!ii~iiii!il]

dead load

,i::::::::::::::::::::::::
:i i i i ili~i :i~:~i:i:i i ~il,

Ii:i:!:i:i:!:~:i:!:~:!:!:i:i:!:]:i:]

1,9-----------I~,~-------I~
lOm

t o = 8.1cm

,~

lOm =,...

lOm

t* = 8.0cm

lOm =,

Fig. 7. Free form shell subjected


to two load cases.

initial shape

by uniform vertical load p = 5 kN/m 2 (snow). Support


conditions are fixed hinges. Because of the symmetry
of loads and the structure, only one-quarter of the
shell has to be analyzed. This was done by 72 8-noded
isoparametric shell elements which are 2 x 2 reduced
integrated. The material properties of concrete are
chosen as E = 3.0 x 107 kN/m 2, v = 0.2.
In a first optimization run, strain energy was chosen
as the objective function without a stress constraint,
assuming the structure is sufficiently reinforced to
resist high tension forces. The resulting shape (Fig.
6(b)) is an anticlastic surface (HP), very similar to a
minimal surface which acts almost like a membrane
in tension and compression. Since the structural
thickness is fixed, the result is alternatively restricted
by an upper bound (6 m) on variable s 1.
To get a more suitable design for concrete, the
objective 'stress levelling' was used to reduce tension
stresses in the lower fibres of the structure, which are
caused by interactions of normal forces and bending
moments. A goal stress of o a ~--- - - 1 0 0 kN/m 2 was
prescribed. The optimal structure (Fig. 6(c)) is a
synclastic shape (EP) where the area of tension in the
lower fibres is reduced to a minimum. Tension cannot

optimal shape

be avoided totally because of the simple shape function


and the rectangular plan of the structure. It is
remarkable that the diaphragm - although possible does not disappear. If it vanishes, the resulting shape
has a horizontal tangent plane at the corner, leading
to negative curvature and increased bending.
By using 'weight' as ~he objective function, any
shape between the 'minimal surface' and a plate can
be determined, which is forced by the additional
constraints on stresses and displacements. Figure 6(d)
shows a result obtained with constraints on v. Mises
effective stresses are not allowed to exceed an
arbitrarily chosen value of a m = 400 kN/m 2.

4.2. Non-regular Shell with Two Loading Cases


A concrete shell of quadratic plan and uniform thickness t ( E = 3.4 x 107kN/m 2, v = 0 . 2 ,
7 = 25 kN/m 3) has been investigated with respect to
strain energy minimization (Fig. 7). The total material
volume is kept constant during shape optimization;
no further constraints are introduced. Two combinations of dead load and uniform live load p = 5 kN/m 2,
which acts either on the entire shell or on one-half of

34

K.-U. Bletzinger and E. R a m m

side views:

A-A

B-B

lore

~r. i'
I,dl

7.3m

I~
r.,,i

18rn

r~

continuity patches
IS
'11

sl

'" ....... " Ss l

~ :

" fixed hegiht

~~l
~ls|

$23

~'22 -

~-4

v l

elan with 23 desien variables s,:


$1

10m

141

"
El

25.3m

variable height

linked height

-$21 -$20
Fig. 8. Geometric model of free
form shell.

Is
it, are considered as different load cases. The sum of
the non-weighted individual strain energies defines a
compromise objective. The initial geometry with a
thickness of t = 8.1 cm was evaluated in a preliminary
design process. The shape is defined by four Brzier
patches. Design variables are linked and continuity
patches are introduced to preserve symmetry and
continuity of the structure. The optimal shape exhibits
a maximal stiffness. Since a pure membrane state is
otherwise not possible, the shell needs a boundary
stiffened by a distinct negative curvature.

4.3. Tennis Hall


The shell in Fig. 8 is related to the reinforced concrete
tennis hall designed by H. Isler [3]. The material data
are E = 3.0 x 107 kN/m 2, v = 0.2, ~ = 25 kN/m 3. One
load case of dead load plus uniform live load,
25 kN/m 2, is considered. One-quarter of the shell is
idealized by 4 Brzier elements linked by continuity
patches and by 23 vertical nodal coordinates S~ as
design variables (Fig. 8). A total of 126 reduced
integrated 8-node shell elements are used for a linear
structural analysis. Strain energy is minimized. It can
be recognized that the shape near the free edges is
sensitive and might even result in a sharp local
curvature representing a kind of edge beam. If this is
not tolerated by prescribed geometrical constraints,

the optimal shape again shows a clear negative


curvature near the free boundary (Fig. 9) as was
demonstrated by Isler in hanging model experiments
and the related beautiful shell structures.

5. C o n c l u s i o n s
Shape-sensitive structures like shells require high
quality design, analysis and manufacturing. Therefore,
the main objective is a membrane-oriented design,
avoiding as far as possible bending, and also buckling
phenomena.
The present paper presents the methods of structural
optimization as general computational tools to find
the shape of shells subjected to different load cases
and certain boundary conditions. The key to the
approach is a flexible design modeller which allows
the generation and modification of even complex
shapes by only a few design parameters. Different
design objectives can be applied. It was shown that
useful information on optimal shells of preferable
stress state can be achieved by the objectives of
'minimal strain energy' and 'stress levelling'. Although
the static analysis is assumed to be linear in this
investigation, the entire optimization procedure is
highly non-linear, demanding sophisticated algorithms and experienced personnel. Together with
progress in computational sciences and hardware,

35

Form Finding of Shells

[3

Fig. 9. (a) Initial and (b)


optimal shape of free form shell.

structural optimization can become a valuable design


aid for shell structures, which will reduce the planning
time and the experimental expense.

9.

10.

Acknowledgements
This work is part of the research project SFB 230'Natural Structures
Light Weight Structures in Architecture and Nature' supported
by the German Research Foundation (DFG) at the University of
Stuttgart. The support is gratefully acknowledged. The authors also
would like to thank their former research associate and colleague
Stefan Kimmich.
-

References
1. Ramm, E.; Mehlhorn, G. (1991) On shape finding methods and
ultimate load analyses of reinforced concrete shells, Engineering
Structures, 13, 178-198
2. Isler, H. (1990) Elegante Modelle - die moderne Form des
Schalenbaus, Deutsche Banzeitung, db 7, 62-65
3. Ramm, E.; Schunck, E. (1986) Heinz Ister - Sehaten, Kr~mer,
Stuttgart
4. Haftka, R.T.; Grandhi, R.V. (1986) Structural shape optimization - a survey, Comp. Meth. Appl. Mech. Engng, 57, 91-106
5. Kruzelecki, J.; Zyczkowski, M. (1985) Optimal structural design
of shells - a survey, SM Archives, 10, 101-170
6. Schmit, L.A. (1981) Structural synthesis - its genesis and
development, AIAA Journal, 19, 1249-1263
7. Petiau, C. (1991) Structural optimization of aircraft, Thin
Walled Structures, 11, 43-64
8. Kimmich, S. (1990) Strukturoptimierung und Sensibilitfits-

11.

12.

13.

14.

15.

16.

17.
18.

analyse mit finiten Elementen, PhD Dissertation, Institut f/Jr


Baustatik, Universit/it Stuttgart
Bletzinger, K.-U. (1990) Formoptimierung yon Fl/iehentragwerken, PhD Dissertation, Institut f/Jr Baustatik, Universit/it
Stuttgart
Braibant, V.; Fleury, C. (1986) Shape optimal design and CAD
oriented formulation, Engineering with Computers, 1,193-204
Schittkowski, K. (1981) The nonlinear programming method
of Wilson, Han and Powell with an augmented Lagrangian type
line search function, Numerische Mathematik, 38, 83-114
Svanberg, K. (1987) The method of moving asymptotes - a new
method for structural optimization, Int. J. Num. Meth. Engng,
24, 359-373
Ramm, E.; Bletzinger, K.-U.; Kimmich, S. (1991) Strategies in
shape optimization of free form shells, Nonlinear Computational Mechanics - State of the Art (Wriggers, P. and Wagner,
W., Editors), Springer-Verlag, Berlin, 163-192
Btetzinger, K.-U.; Kimmieh, S.; Ramm, E. (1991) Efficient
modeling in shape optimal design, Computing Systems in
Engineering, 2, 483-495
Kimmich, S.; Ramm, E. (1989) Structural optimization and
analysis with program system CARAT, Proc. GAMM Seminar, October 5-7, 1988, Siegen, on Discretization Methods
and Structural Optimization - Procedures and Applications
(Eschenauer, H.A. and Thierauf, G., Editors), Lecture Notes in
Engineering, Springer-Verlag, Berlin, 186-193
Brhm, W.; Farin, G.; Kahmann, J. (1984) A survey of curve
and surface methods in CAGD, Comp. Aided Geom. Des., 1,
1-60
Faux, I.D.; Pratt, M.J. (1979) Computational Geometry for
Design and Manufacture, Ellis Horwood, Chichester
Philippi, U.; Nachtigall, W. (1991) Constructional morphology
of sea urchin tests, to be presented at the 2nd International
Symposium on Natural Structures - Principles, Strategies, and
Models in Architecture and Nature, SFB 230, Stuttgart, October
1-4

Вам также может понравиться