International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056

Volume: 02 Issue: 04 | July-2015 www.irjet.net p-ISSN: 2395-0072

TOPOLOGICAL OPTIMIZATION TECHNIQUES FOR LINEAR ISOTROPIC

STRUCTURES SUBJECTED TO STATIC AND SELF-WEIGHT LOADING
CONDITIONS
Naman Jain1, Rahul Joshi2, Rakesh Saxena3
1 M.Tech, Department of Mechanical Engineering, G. B. Pant University of Agriculture and Technology, Pantnagar,
Uttarakhand, India
2 M.Tech, Department of Mechanical Engineering, G. B. Pant University of Agriculture and Technology, Pantnagar,

Uttarakhand, India
3 Professor, Department of Mechanical Engineering, G. B. Pant University of Agriculture and Technology, Pantnagar,

Uttarakhand, India
-----------------------------------------------------------------------***------------------------------------------------------------------------

Abstract- This paper represents the optimal criteria Keywords- Optimality Criterion, SIMP, Topology
method for topological optimization of isotropic Optimization, Pseudo-densities and Compliance
material under different loads and boundary conditions minimization
with the objective to reduce mass of an existing
material and study the different shape obtained. 1 INTRODUCTION
Topological optimization mainly comprises of a
mathematical approach that optimizes the layout Topology optimization is a useful tool for a designer which
within a given design constraints, for a given set of generates the optimal conceptual shape of a mechanical
loads and boundary condition such that the structure. The structural shape is generated within a
performance matches with the prescribed set of predefined design space. In addition, the user defines
performance targets. Topological optimization solve the structural supports and loads. Without any further
problem of distributing a given amount of material in a decision and guidance of the user, the method will give the
design domain subjected to load and supports structural shape thus provides a first idea of an optimum
conditions, such that the compliance of the structure is geometry. A desired property of the structure is
minimized while the stiffness of structure is maximized. maximized by changing the shape of the given material.
For material distribution system solid isotropic with Usually this maximized property is stiffness. Another
penalization approach is used. In all the structures usage of topology optimization is minimizing the weight,
objective function is compliance, design variable is subjected to a given constraint (such as stress). Topology
pseudo density and state variables are the response of optimization method is a technique to find out optimal
structures that is deflection. Objected function is material distribution within predefined design domain. It
subjected to volume constraint and by minimize the can give the best conceptual design that can satisfy all
compliance stiffness of structures are maximize. design requirements. Topology optimization problem
Different numerical examples are taken to study the includes objective function, design domain and design
optimal criteria approach and validate the results constraints. Objective function represents the goal of the
obtained with SA-SIMP and BESO method. This paper optimization method which is to be minimized or
work represents topological optimization for static and maximized.
self-weight loading using finite element solver ANSYS.
APDL (ANSYS Parametric Design Language) has been With the exception of a few early landmark results [3 12],
employed for utilizing the topological optimization the historical development of the field of structural
capabilities of commonly used finite element solver optimization seems to have followed an opposite route to
ANSYS. 8 node 82 elements are used to model and mesh the actual structural design process [2 20]. Since its
the isotropic structures in ANSYS. inception, research in numerical optimal structural design
went from element stiffness design, through geometric and
shape optimization to topology optimization design. It is
also clear that the major impact on the structural
efficiency, in the sense of stiffness/volume or

© 2015, IRJET.NET- All Rights Reserved Page 52

 International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056
Volume: 02 Issue: 04 | July-2015 www.irjet.net p-ISSN: 2395-0072

stress/volume ratio, is determined at the conceptual stage each element, structure design can be improved to optimal
by the topology and shape of the structure. No amount of design.
fine-tuning of the cross-sections and thicknesses of the
elements will compensate for a conceptual error in the 2 SIMP METHOD
topology or the structural shape [13]. With the
development of high-speed computer, the topology The SIMP stands for Solid Isotropic Material with
optimization method using numerical approach has been Penalization method. It is also known as the power-law
growing quickly [1 5 16]. In the present work we will be approach, in which the material properties can be
studying the topology optimization of continuum expressed in terms of the design variable material density
structures with the help of Optimality criteria method using a simple “power-law” interpolation as an explicit
using ANSYS, also ANSYS use SIMP method for means to suppress intermediate values of the bulk density.
penalization of intermediate densities. The finite element This method has been presented by Bendsoe [3]. The
based continuum topology optimization as a generalized SIMP, material model where material properties are
shape optimization problem has experienced tremendous assumed constant within each finite element, discretizes
progress since the influential work of Bendsoe and Kikuchi the design domain with the design variables being the
[2]. They presented a homogenization based optimization element densities. At each point of the design domain, the
approach of topological optimization. They assumed that material properties are modeled as the relative material
the structure is formed by a set of non-homogenous density raised to some power times the material
elements which are composed of solid and void regions. properties of solid material. The common choice of design
They obtained optimal design under volume constraint parameterization is to take xi as the design variable by
through optimization process. In their method, the regions convention, xi = 1 at a point signifies a material region
with dense cells are defined as structural shape, and those while xi = 0 represents void. Each finite element (formed
with void cells are areas of unnecessary material. It has due to meshing in ANSYS) is given an additional property
also been demonstrated that the optimal material of pseudo-density, xi where 0≤xi≥1, which alters the
distribution can be considerably simplified by employing a stiffness properties of the material.
density dependent isotropic material. In both the
approaches, remeshing of the structural domain and the (1)
evaluation of shape density are avoided. This problem had
a discrete nature, since the material distribution consisted
ρi= Density of the ith element
of solid or void regions.
ρ0= Density of the base material
xi= Pseudo-density of the ith element

This Pseudo-density of each finite element serves as the

design variables for the topology optimization problem
and the intermediate values are penalized according to the
following scheme:

(2)

Here Ei is the material young modulus of the ith element

while E0 denotes the young modulus of the solid phase
Fig. 1: Design domain of typical topology optimization material. The stiffness of intermediate densities is
problem [10] penalized through the power law relation, so they are not
favoured. As a result, the final design consists primarily of
A scheme of design domain is shown in Figure 3.1, where solid and void regions.
Ft is the external force, Ω is the design domain, Ωs
denotes a solid domain and Ωv represents a sub-domain 3 MATERIAL AND METHOD
without material. Topology optimization methods are
based on FEM and sensitivity analysis. In FEM each finite 3.1 Optimal Criteria Approach
element is assigned a design variable which is the material
density of the element. By updating material density of Optimality criteria are necessary conditions for minimality
of the objective function and these can be derived by using
either variational methods or extremum principles of
© 2015, IRJET.NET- All Rights Reserved Page 53
 International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056
Volume: 02 Issue: 04 | July-2015 www.irjet.net p-ISSN: 2395-0072

mechanics. Optimality criteria (OC) method was constraint is zero. By using Kuhn-Tucker conditions, the
analytically formulated by Prager and co-workers in 1960. optimality conditions for the optimization problem can be
It was later developed numerically and become a widely stated as
accepted structural optimization method.OC methods can
be divided into two types. One type is rigorous
mathematical statements such as the Kuhn-Tucker
conditions. The other is algorithms used to resize the
structure for satisfying the optimality criterion. Different
optimization problems require different forms of
optimality criterion. In Kuhn-Tucker conditions, the
inequality constraints can be transformed into equality
constraints by adding slack variables. Here the
optimization in its most general form may be expressed as
follows
The optimal criteria method is one of the best-established
Minimize f(x) and widely accepted optimization techniques.
Such that hj(x) = 0 j=1,2, ….nj (3)
gk(x) ≤ 0 k=1,2,…nk 3.2 Numerical Examples

Where hj and gk are constraints, j and k are the number of Three numerical examples are taken to demonstrate the
equality of constraints and inequality constraints, validity and efficiency of the proposed approach. The
respectively specimens are taken from the work of Garicia-Lopez et al.
[8] and Huang and Xie [9]. All the models are under plane
The Lagrangian function of the optimization can be defined state of stress.
as
Model 1: Cantilevered beam under static loading
L(x,t, ) = f(x) +
A cantilever beam of thickness 1mm is considered in this
(5) case. The cantilever is under the state of plane stress and
supports a concentrated load of magnitude 1N at the
Where are Lagrangian multipliers
bottom right corner. The left edge is fixed as shown in
Figure 2. The meshing is done with 8-node quadrilateral
Differentiating the Lagrangian function (5) with respect to elements by giving element edge length one for each line.
x, t, λ, ζ we obtain The results were compared with combining simulated
annealing and SIMP approach [8]. Table 4 shows the final
compliance obtained with ANSYS (OC) and combining
simulated annealing and SIMP approach. Material
properties for Model 1 are shown in Table 1.

32 mm

20 mm
From equation 7 & 8

(10)
(11)
F
This implies that when an inequality constraint is not
active, the Lagrangian multiplier associated with the

© 2015, IRJET.NET- All Rights Reserved Page 54

 International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056
Volume: 02 Issue: 04 | July-2015 www.irjet.net p-ISSN: 2395-0072

Fig. 2: Geometry and boundary conditions for Model 1 Model 3: Messerschmitt Bolkow Bolhm beam under
self-weight

The beam is in the state of plane stress with a thickness of

1 mm. Here the classic MBB beam subjected to a
Table 1: Material properties, Load, Elements and Volume concentrated load and its self-weight is to be optimized.
usage fraction for Model 1 The dimensions and support conditions of the design
Young’s Poisson’s Load Elements Volume domain are shown in Figure 4. Due to the symmetry, only
modulus ratio Fraction half of the design domain is discretized with 100x50 8-
1 N/m² 0.3 1N 640(32*20) 0.4 node plane stress elements. The results are compared with
the results of X.Huang et al. [9] who utilized BESO method
Model 2: Messerschmitt Bolkow Bolhm beam under for topological optimization. The material volume
static loading constraint is set to be 40% of the whole design domain.
Material properties for model 2 are shown in Table 3.
The beam is in the state of plane stress with a thickness of
1 mm. The beam is optimized for minimum compliance. Table 3: Material Properties and Density Used for MBB
Due to symmetry of the model, only half of the model is Beam (Model 4)
considered with symmetry boundary conditions as it is Young’s modulus Poisson’s ratio Density Volume
symmetric about the vertical axis. The beam is supported fraction
by a roller support at the bottom right corner and 200 N/ mm2 0.3 78 kg/m3 0.4
symmetric boundary conditions are applied on the left
edge as shown in Figure 3. The meshing is done with 8 20 mm
nodes quadrilateral elements by giving element edge
length one for each line. The results are compared with
combining simulated annealing and SIMP approach [8].
Table 5 shows the final compliance obtained with ANSYS
5 mm
(OC) and combining simulated annealing and SIMP
approach. Material properties for model 2 are shown in
Table 2.

Table 2: Material properties, Load, Elements and Volume usage

fraction for Model 2 Fig. 4: Geometry and boundary conditions for Model 3

Young’s Poisson’s Load Elements Volume 4 RESULT AND DISCUSSIONS

Modulus ratio (ν) (N) Usage
(E) Fraction This section presents the detailed results of FE analysis
and optimization of the above structures. Final compliance
1 N/m² 0.3 1 1200(60*20) 0.5 and optimal shape of the models obtained with the help of
gradient based ANSYS based Optimality Criterion have
been compared with SA-SIMP and BESO method [8 9].
F 60 mm
Model 1: Cantilevered beam under static loading

The optimal shape of the cantilever beam has been

obtained through ANSYS (OC) as shown in Figure 5 (a).
20 mm
The shapes obtained through different methods are almost
same. The final value of compliance after topological
optimization is presented in Table 4 comprising of the
optimal compliance values ANSYS (OC) method give lowest
value. As it has been observed that, final compliance value
Fig. 3: Geometry and boundary conditions for Model 2 obtained through ANSYS is 2.084% lower than PS-RoA
method, 1.814% lower than RS-RoA method. From the
table it has been observed that number of iterations

© 2015, IRJET.NET- All Rights Reserved Page 55

 International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056
Volume: 02 Issue: 04 | July-2015 www.irjet.net p-ISSN: 2395-0072

required by ANSYS based OC is 39. In ANSYS (OC) method Fig. 5: Optimal shapes obtained using (a) ANSYS OC (b) PS-
convergence criteria is 0.0001 given Mesh density is same RoA (c) RS-RoA
in all the method. Above result show that ANSYS (OC) can
use for topological optimization and on comparison ANSYS
(OC) is more effective.

Table 4: Comparison between ANSYS OC, PS-RoA and RS-

RoA for Model 1
Method ANSYS OC PS–RoA RS–RoA
Compliance (Nmm) 52.224 53.3123 53.1714
Iterations 39 * *
*Not available

Fig. 6: Convergence of compliance values for cantilever

beam

Model 2: Messerschmitt Bolkow Bolhm beam under

static loading
(a)
The optimal shape of the MBB beam has been obtained
through ANSYS b(OC) as shown in Figure 7 (a). The shapes
obtained through different methods are almost same. The
final value of compliance after topological optimization is
presented in Table 5. On comparison of optimal
compliance value ANSYS (OC) method give lowest value.
As it has been observed that, final compliance values
obtained through ANSYS is 3.499% lower than PS-RoA
method, 3.44% lower than RS-RoA method. From the
Table 5 it has been observed that number of iterations
required by ANSYS (OC) is 32. In ANSYS (OC) method
convergence criteria is 0.0001 given. Mesh density is same
(b)
in all the method. From above result we conclude that
ANSYS can used for topological optimization and on
comparison ANSYS based OC method is more effective.

(c)
(a)

© 2015, IRJET.NET- All Rights Reserved Page 56

 International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056
Volume: 02 Issue: 04 | July-2015 www.irjet.net p-ISSN: 2395-0072

Model 3: Messerschmitt Bolkow Bolhm beam under

self-weight loading

The optimal shape of the cantilever beam has been

obtained through ANSYS (OC) as shown in figure 9 (a). The
shapes obtained through both methods are almost same.
The final value of compliance after topological
optimization is presented in Table 6 comprising of the
optimal compliance value ANSYS (OC) method give higher
(b) value. As it has been observed that, final compliance value
obtained through ANSYS is 5.88% higher than BESO
method. From the Table 6 it has been observed that
number of iterations required by ANSYS (OC) is 17 while
for BESO method is 76. Figure 10 show the convergence of
compliance values (OC). In ANSYS (OC) convergence
criteria is 0.0001 given. Mesh density is same in both the
method. From above result we conclude that ANSYS can
use for topological optimization and on comparison ANSYS
(OC) is more effective on the basis of number of iterations.
(c)
Fig. 7: Optimal shapes obtained using (a) ANSYS OC (b) PS-
RoA (c) RS-RoA

Table 4: Comparison between ANSYS OC, PS-RoA and RS-

RoA for Model 2
Method ANSYS OC PS–RoA RS–RoA
Compliance (Nmm) 183.345 189.7603 189.6530
Iterations 32 * *
*Not available (a)

(b)
Fig. 9: Optimal Shapes Obtained by (a) ANSYS (OC) and (b)
BESO Method

Table 6: Final compliance value for self weight for MMB

beam Model 3
Method Compliance Iterations
BESO 0.034 76
ANSYS based OC 0.036 17
Fig. 8: Convergence of compliance values by ANSYS for
MBB beam under static loading

© 2015, IRJET.NET- All Rights Reserved Page 57

 International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056
Volume: 02 Issue: 04 | July-2015 www.irjet.net p-ISSN: 2395-0072

[3] Bendsoe, M.P. 1989. Optimal shape design as a material

distribution problem, Structural Optimization 1: 193-
202
[4] Bendsøe, M.P., Kikuchi, N. and Diaz, N. 1993. Topology
and generalized layout optimization of elastic structures,
pages 159–205.
[5] Chapman, C. D. 1994. Structural topology optimization
via the genetic algorithm, Thesis, M. S. Massachusetts
Institute of Technology, America.
[6] Chiandussi, G. M. , Codegone and Ferrero, S. Topology
optimization with optimality criteria and transmissible
loads”, Computers and Mathematics with Applications
57 (2009) 772_788
[7] Diaz, A. and Sigmund, O. 1995. Checkerboard patterns in
layout optimization, Struct. Optim.. Vol: 10: 40-45
[8] Garicia-Lopez, N.P., Sanchez-Silva, M., Medaglia, A.L. and
Chateauneuf, A. 2011. A hybrid topology optimization
methodology combining simulated annealing and SIMP,
Fig. 10: Convergence of compliance values by ANSYS for Computers and Structures 89: 1512–1522
MBB beam under static loading [9] Huang, X and Xie Y.M. 2011. Evolutionary topology
optimization of continuum structures including design-
5 CONLUSIONS dependent self-weight loads, Finite Elements in Analysis
and Design 47:942-948
The results of Optimality Criteria method using ANSYS [10] Lee, E. 2011. A strain based topology
when compared with results obtained by SA-SIMP and optimization method. Thesis, SUJN, New Jersy
BESO method for the linear elastic isotropic structures [11] Michael, Thomas R. 2010. Shape and topology
taken for study are better. Compliance value obtained optimization of brackets using level set method”, An
through ANSYS for cantilever beam under static loading is Engineering project submitted to the graduate faculty
2.084% lower than PS-RoA method and 1.814% lower of Rensselaer Polytechnic Institute in partial fulfillment
than RS-RoA method and for MBB beam under static of the degree of Master of Engineering in Mechanical
loading is 3.499% lower than PS-RoA method and 3.44% Engineering. Rensselaer Polytechnic Institute Hartford,
lower than RS-RoA method also it takes lesser number of Connecticut
iterations to reach the optimal solution it is found that the [12] Michell, A.G.M. 1904. The limits of economy of
Optimality Criteria using ANSYS converges very fast in material in frame structures, Philosophical magazine
comparison to SA-SIMP method. The optimal topologies Series 6, 8(47):589-597.
obtained by both the methods are almost same. [13] Olhof, Niels, Bensoe, Martin P. and Rasmussen,
Compliance value obtained by ANSYS for MBB beam under John 1991. On CAD-integrated structural topology and
self-weight is 5.88% higher than BESO method where as design optimization, Computer methods in applied
number of iterations required by ANSYS is 17 while for mechanics and engineering 89:259-279
BESO method is 76. It is found that the Optimality Criteria [14] Rahmatalla, S. F. and Swan, C. C. 2004. A Q4/Q4
approach using ANSYS converges very fast in comparison continuum structural topology optimization
to BESO method. The optimal topologies obtained by both implementation, Struct. Multidisc. Optim. Springer-
the methods are almost same. Verlag, Vol 27: 130-135
[15] Rozvany, G. I. N. A critical review of established
REFERENCES methods of structural topology optimization, Struct
Multidisc Optim (2007)
[1] Allaire, G. , Jouve, F. and Toader, A. M. 2002. A level set [16] Sigmund, O. A 99 line topology optimization
method for shape optimization, C. R. Acad. Sci. Paris. code written in Matlab, Struct. Multidisc. Optim.
[2] Bendsøe, M. P. and Kikuchi, N. 1988. Generating optimal Springer-Verlag 2001, Vol 21: 120-127.
topologies in structural design using a homogenization [17] Sigmund, O. and Petersson, J. 1998. Numerical
method, Comput. Meth. Appl. Mech. Eng., vol: 71: 197- instabilities in topology optimization: A survey on
224. procedures dealing with checkerboards, mesh-

© 2015, IRJET.NET- All Rights Reserved Page 58

 International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056
Volume: 02 Issue: 04 | July-2015 www.irjet.net p-ISSN: 2395-0072

dependencies and local minima, Struct. Optim.. Vol 16:

68-75
[18] Swan, C. C. and Kosaka, I. 1997. Voigt-Reuss
topology optimization for structures with linear elastic
material behaviors, Int. J. Numer. Meth. In Eng. Vol: 40:
3033-3057
[19] Tcherniak, D. and Sigmund, O. A web-based
topology optimization program, Struct. Multidisc.
Optim. Springer-Verlag 2001, Vol 22: 179-187
[20] Zhou, M. and Rozvany, G.I.N. 1992. DCOC: an
optimality criteria method for large systems part
I:theory, Structure Optimization 5:12-25

BIOGRAPHIES

Naman Jain obtained his

bachelor’s degree in Mechanical
Engineering from College of
Engineering Roorkee. He is
currently doing M. Tech. in
Design and Production
Engineering from G. B. P. U. A &
T. His areas of interest are finite
element analysis, computer
aided mechanical design,
optimization and Numerical
Modeling and Simulation.

Rahul Joshi obtained his

bachelor’s degree (B. Tech.) in
Mechanical Engineering from
Uttarakhand Institute of
Technology. He is currently
doing M. Tech. in Design and
Production Engineering from G.
B. P. U. A & T.

© 2015, IRJET.NET- All Rights Reserved Page 59