Documentos de Académico
Documentos de Profesional
Documentos de Cultura
DOI 10.1007/s00158-009-0473-2
RESEARCH PAPER
Received: 3 August 2009 / Revised: 10 December 2009 / Accepted: 10 December 2009 / Published online: 7 January 2010
c Springer-Verlag 2009
O. M. Querin (B)
School of Mechanical Engineering,
The University of Leeds, Leeds LS2 9JT, UK
e-mail: O.M.Querin@Leeds.ac.uk
M. Victoria P. Mart
Structures and Construction Department,
Technical University of Cartagena, Campus Muralla del Mar,
30202 Cartagena, Murcia, Spain
M. Victoria
e-mail: mariano.victoria@upct.es
P. Mart
e-mail: pascual.marti@upct.es
1 Introduction
Prager (Prager and Rozvany 1977) introduced the terms
truss-like continua for structures in which in some
regions members of infinitesimal cross-sectional area have
an infinitesimal spacing. Such structures are for example
Michell frames (Michell 1904). As Prager has declared,
while this type of structure is not practical, it is useful in
the evaluation of the efficiency of more practical designs
In the literature on topology optimization of truss-like
continua, relatively little attention has been paid to problems where the material behaves differently in tension and
compression. By behaviour we mean different stress limits or elastic modulus. There have been several approaches
using optimality criteria, ground structure, growth method,
SIMP, homogenization and heuristics to solve this problem.
The work of Michell (1904) has set the foundation for
research into the optimal layout of trusses, with materials
of the same or different strength in tension and compression. Michell proved that an optimum truss must follow
the orthogonal network of lines of maximum and minimum
strain in a constant magnitude strain field. This research was
further advanced by Hemp (1958, 1973) and Prager (1958).
Rozvany et al. (1993) showed that the optimal topologies
for plastic stress design and elastic compliance design are
the same, and the volume or weight of the latter is given
by the square of the volume of the former (multiplied by
a given constant), for a review see Rozvany et al. (1995),
p. 57.
Rozvany (1996, 1997) demonstrated that there was an
error in Michells (1904) derivation of optimality criteria,
and that the latter are only valid for a restricted class of
problems. Using three different methods, he derived the correct optimality criteria (with adjoint strains t = 1/t and
c = 1/c ), which confirmed those of Hemp (1973) and
26
demonstrated on a simple example that the new optimality criteria resulted in a much lower structural volume than
that of Michell. He also sketched the correct Michell cantilever with different permissible stresses in tension and
compression and gave a simple expression for the optimal
angle of members next to a line support (for a stress ratio of
1:3 these are 30 and 60 degrees).
Srithongchai and Dewhurst (2003), however expressed
appreciation for the early contribution of Prager (1958), in
which he established the optimality criteria that the strain
energy per unit volume, or per unit weight needed to be
constant in a structure with materials of different strength
in tension and compression. Their paper, like Rozvanys
compares Michell and what they called Prager structures.
Selyugin (2004) proposed a slight variation on the
adjoint strain optimality criteria of Hemp (1973) and
Rozvany (1996) for the optimization of bi-material structures with maximum stiffness. In it he proposed that the
layout should follow lines of maximum strains of magnitude k(t /t ) and k(c /c ), where k is a small positive
number but that the structural members should be designed
to all have equal strain energy per unit mass.
Dewhurst (2005) demonstrated that in order to achieve
an optimal layout of maximum stiffness and minimum volume/weight, the structural members should have a constant
strain with a fixed ratio between the tensile and compressive
members which is a function of their elastic moduli and densities. By achieving this, the structure would have a constant
strain energy per unit weight in all the members
Although these papers all dealt with structural layouts,
they formed the basis for the work of Nair (2005) and
Taggart et al. (2007) who extended this to continuum structures. They used a node based approach to carry out the
optimization, which is analogous to the Hard-Kill (Hinton
and Sienz 1995) or ESO (Xie and Steven 1993) methods,
and used the strain energy density as the optimality criteria.
The method used to incorporate the different stress limits in
tension and compression was based on modifying the modulus of elasticity of the material experiencing compressive
or biaxial stresses, whilst assuming an isotropic material
property, allowing an isotropic analysis of the structure.
Achtziger (1996) used the ground structure approach
coupled with a standard Linear programming (LP) solver to
derive the optimum truss topology with different stress limits in the tensile and compressive members. More recently
this has been extended to problems with up to 109 truss
elements in the ground structure by Gilbert et al. (2005),
which has resulted in a software application, Darwich et al.
(2008). Alternatively, Martinez et al. (2007) proposed a
growth method for the optimal design in a sequential manner of size, geometry and topology of plane trusses without
the need of a ground structure. This method is applicable
to single load case problems with stress and size constraints
Topology optimization of truss-like continua with different material properties in tension and compression
27
the compressive elements within the structure and satisfying the optimality criterion of a constant stress/strain ratio
in the structure. Three examples will be presented to show
the validity of the proposed approach.
2. The use of the same material for the tensile and compressive elements of the structure.
T
=
C
T E C
C E T
1
2
(1)
where:
Depending on the method of analysis, it may be more convenient to work with stresses rather than strain. Converting
(1) by dividing by the respective modulus of elasticity gives
the equivalent equation, which relates the tensile and compressive stresses in the optimal structure by a new factor k1 :
k
T , C
ET , EC
T , C
k1 =
T
C
=
T E T
C E C
1
2
(2)
where:
k1
T , C
EC =
ET
= ET k2
k12
(3)
28
4.
5.
6.
7.
8.
9.
At the initial stages of this research, two finite element analyses were carried out after every optimization step. The
first to determine the sign and orientation of the principal
stresses and calculate the orthotropic properties for each element, the second to determine the principal stresses to use
for the optimization. However; apart from the first iteration;
it was found that changes in the structural domain due to
the optimization step have a negligible effect in the changes
of the direction and sign of the principal stresses. Hence
only one finite element analysis between optimization steps
is required.
where:
M
F
L
M Optimal
VOptimal
F L 1T + 1C
where:
VOptimal
VOptimal Av
2F L
(6)
where:
4 Examples
M =
(5)
(4)
Fig. 1 Design domain for the Short Prager Cantilever
Topology optimization of truss-like continua with different material properties in tension and compression
Fig. 2 Optimal topology for the Short Prager Cantilever: a using the
method of Martinez et al. (2007), b this work
value of zero for getting solutions close to Michell structures. However, when this was tried there were problems
with the orthotropic elements so a Poissons ratio of v = 0.3
was used.
4.1 Short Prager Cantilever
This example given by Srithongchai and Dewhurst (2003)
represents a short cantilever of height W and width L with
a vertical force F acting along a distance L from the support
and a vertical distance D below the top of the support in
Fig. 1. Where the height to width ratio W/L = 1:0.5995.
In order to capture accurately the angles of the emerging
truss elements in the topology such that these are properly
aligned between the supports and load, a border around the
true design domain was placed here denoted by lengths B1
and B2 .
The dimensions used for this example are W = 1,000 mm
and L = 599.5 mm, represented by the dashed rectangle.
Fig. 3 Design domain for the
Prager Cantilever
29
B1
D
W
F
B2
B3
B4
30
Fig. 4 Optimal topology for the Prager Cantilever: a using the method
of Martinez et al. (2007), b this work
This example is the classic cantilever beam with a fixed circular boundary and an applied load F. The design domainis
a rectangle of length L + 1.5R and of height W = L e/42
(Hemp 1973), Fig. 5. As for the previous examples, a border denoted by lengths B1,2 was placed around the design
domain.
The dimensions used for this example are L = 1740 mm,
R = L/10 = 174 mm, 1.5R = 261 mm, W =
1121.94136 mm and, represented by the dashed rectangle. The lengths of the border were B1 = 36.42932 mm,
B2 = 34.8 mm. The entire design domain (solid outer
rectangle) was subdivided with a mesh of 206 351 elements. The distance to the vertical force F = 10 kN is D =
560.97068 mm with the thickness of the domain t = 1 mm.
The prescribed constant ratio between stresses was k1 = 2.
No theoretical solution was found in the literature for
this problem when the elements of the structure experiencing tension and compression have different permissible
B1
R
W
F
B1
1.5R
B2
Topology optimization of truss-like continua with different material properties in tension and compression
31
References
5 Conclusions
The method presented in this paper combines aspects of
the work of Plfi (2004), Dewhurst (2005) and Nair (2005)
in order to allow for the topology optimization of continuous structures with different material properties in tension
and compression. The method consists of a few simple
steps: identifying all negative (compressive) parts of a structure; modifying their modulus of elasticity; changing their
material properties to orthotropic; modifying the stresses
or strains; and satisfying the optimality criterion of a constant stress/strain ratio between the tensile and compressive
Achtziger W (1996) Truss topology optimization including bar properties different for tension and compression. Struct Optim 12(1):
6374
Alfieri L, Bassi D, Biondini F, Malerba PG (2007) Morphologic
evolutionary structural optimization. In: 7th world congress on
structural and multidisciplinary optimization, Seoul, Korea, paper
A0422
Cai K, Shi J (2008) A heuristic approach to solve stiffness design
of continuum structures with tension/compression-only materials.
In: Fourth international conference on natural computation, vol 1,
pp 131135
Darwich W, Gilbert M, Tyas A (2008) FORM: a practical layout optimization tool for civil and structural engineers. In: 8th world
congress on computational mechanics (WCCM8). 5th European
congress on computational methods in applied sciences and
engineering (ECCOMAS 2008), Venice, Italy
Desmorat B, Duvaut G (2003) Compliance optimization with nonlinear
elastic materials application to constitutive laws dissymmetric in
tension-compression. Eur J Mech A, Solids 22:179192
Dewhurst P (2005) A general optimality criterion for combined
strength and stiffness of dual-material-property structures. Int J
Mech Sci 47:293302
Duysinx P (1999) Topology optimization with different stress limit in
tension and compression. In: Proceedings (CD-Rom) of the third
world congress of structural and multidisciplinary optimization
(WCSMO3), Buffalo, NY, USA
Duysinx P, Bruyneel M (2000) Recent progress in preliminary design
of mechanical components with topology optimisation. Integrated
Design and Manufacturing in Mechanical Engineering: Proceedings of the Third IDMME Conference Held in Montreal. Kluwer,
Canada, pp 457464, (ISBN: 1-4020-0979-8)
Duysinx P, Van Miegroet L, Lemaire E, Brls O, Bruyneel M (2008)
Topology and generalized shape optimization: why stress constraints are so important? Int J Simul Multidisc Des Optim
2:253258
Geli S, Dolex G, Ishai O (1981) An effective stress/strain concept in the
mechanical characterization of the structural adhesive bonding.
Int J Adhes Adhes 1:135140
Gilbert M, Darwich W, Tyas A, Shepherd P (2005) Application
of large-scale layout optimization techniques in structural engineering practice. In: Proceedings of the 6th world congress of
structural and multidisciplinary optimization (WCSMO6), Rio de
Janeiro, Brazil, Paper 401
Graczykowski C, Lewinski T (2006a) Michell cantilevers constructed
within trapezoidal domains. Part I: geometry of Hencky nets.
Struct Multidisc Optim 32:347368
Graczykowski C, Lewinski T (2006b) Michell cantilevers constructed
within trapezoidal domains. Part II: virtual displacement fields.
Struct Multidisc Optim 32:463471
32
Graczykowski C, Lewinski T (2007a) Michell cantilevers constructed
within trapezoidal domains. Part III: force fields. Struct Multidisc
Optim 33:2746
Graczykowski C, Lewinski T (2007b) Michell cantilevers constructed
within trapezoidal domains. Part IV: complete exact solutions of
selected optimal designs and their approximations by trusses of
finite number of joints. Struct Multidisc Optim 33:113129
Guan H, Steven GP, Xie YM (1999) Evolutionary structural optimisation incorporating tension and compression materials. Adv Struct
Eng 2(4):273288
Guan H, Chen YJ, Loo YCH, Xie YM, Steven GP (2003) Bridge
topology optimization with stress, displacement and frecuency
constraints. Comput Struct 81:131145
Hemp WS (1958) Theory of the structural design. Report no 115,
College of Aeronautics, Cranfield
Hemp WS (1973) Optimum structures. Clarendon, Oxford
Hinton E, Sienz J (1995) Fully stressed topological design of structures
using an evolutionary approach. Eng Comput 12:229244
Lewinski T, Rozvany GIN (2007) Exact analytical solutions for some
popular benchmark problems in topology optimization II: threesided polygonal supports. Struct Multidisc Optim 33:337350
Lewinski T, Rozvany GIN (2008a) Exact analytical solutions for
some popular benchmark problems in topology optimization III:
L-shaped domains. Struct Multidisc Optim 35:165174
Lewinski T, Rozvany GIN (2008b) Analytical benchmarks for topology optimization IV: square-shaped line support. Struct Multidisc
Optim 36:143158
Martinez P, Marti P, Querin OM (2007) Growth method for size,
topology, and geometry optimization of truss structures. Struct
Multidisc Optim 33:1326
Michell AGM (1904) The limits of economy of material in framestructures. Philos Mag 8(47):589597
Nair AU (2005) Evolutionary numerical methods applied to minimum weight structural design and cardiac mechanics. PhD thesis,
Mechanical engineering and applied mechanics, University of
Rhode Island
Nowak M (2006) Structural optimization system based on trabecular
bone surface adaptation. Struct Multidisc Optim 32:241249
Plfi P (2004) Locally orthotropic femur model. J Comput Appl Mech
5(1):103115
Pereira JT, Fancello EA, Barcellos CS (2004) Topology optimization
of continuum structures with material failure constraints. Struct
Multidisc Optim 26:5066
Prager W (1958) A problem of optimal design. In: Proceedings of the
union of theoretical and applied mechanics, Warsaw