Você está na página 1de 12

Struct Multidisc Optim

DOI 10.1007/s00158-015-1369-y

INDUSTRIAL APPLICATION

Structural optimization methods of nonlinear static analysis


with contact and its application to design lightweight gear
box of automatic transmission of vehicles
Takanori Ide1 Hiroyuki Kitajima2 Masaki Otomori1 Juan Pablo Leiva3
Brian C. Watson3

Received: 20 July 2015 / Revised: 19 October 2015 / Accepted: 26 October 2015


Springer-Verlag Berlin Heidelberg 2016

Abstract In this paper, we consider lightweight design


methods for gear box of automatic transmission of vehicles with contact constraints. Lightweight design is a fundamental requirement for protecting the environment and
improving fuel economy. In addition, durability is another
important requirement for safe driving. However, in the
design of automatic transmissions, these two requirements
are usually in a trade-off relationship and engineers spend
a long design study time. This paper deals with design
approaches using structural optimization method to design
lightweight structures and to minimize stress with contact constraints. Stress with contact constraints is solved
using the finite element method. Three different structural
optimization methods, topometry, topography and freeform
optimization, are applied for the design of a lightweight gear
box of an automatic transmission. The optimization results
show that the optimization methods successfully found the
lightweight gear box design and can be used at the early
stage of the design process of automatic transmissions.
Keywords Lightweight design Contact constraints
Automatic transmission

 Takanori Ide

i24824 ide@aisin-aw.co.jp
1

AISIN AW Co., LTD. Fujii-cho, Takane 10, Anjo, Aichi,


444-1192, Japan

AW ENGINEERING CO.,LTD. Sakurai-cho,


Kitaawarashimo 48-8, Anjo, Aichi, 444-1154, Japan

Vanderplaats Research & Development, Inc. 41700


Gardenbrook, Suite 115, Novi, MI 48375, USA

1 Introduction
The main purposes of this paper are to show:
1. a sensitivity analysis approach for nonlinear static analysis with contact constraints; and
2. the validity of structural optimization methods of nonlinear static analysis such as topometry, topography and
freeform for automatic transmission.
Our previous paper (Ide et al. 2014a) showed structural
optimization methods to design lightweight gear box with
low radiated noise for automatic transmission. We considered weakly coupled analysis of elastic and acoustic
problem.
Nonlinear analysis for contact is now well understood,
but structural optimization with contact conditions is not
as well developed or used due to the difficulty to implement it and the cost of the nonlinear sensitivity analysis.
Herskovits et al. (2000) proposed a bi-level method and
applied it to shape optimization problems. Yi et al. (2010),
utilizing the equivalent static load method coupled two commercial programs: ABAQUS, for the nonlinear analysis, and
NASTRAN, for the optimization; in their work they avoided
using expensive nonlinear sensitivities.
To find lightweight designs of automatic transmission
is one of the essential tasks to achieve fuel consumption
efficiency. Traditionally, engineers have relied on their intuition and experience to change their designs to improve
performance. This conventional approach typically requires
extensive studies and a long time to reach desired results.
To reduce design study time, we use structural optimization
methods. Schmit (1960) pioneered the numerical structural
optimization method, showing its effectiveness designing a
threebar truss. The literature provides plenty of material
concerning shape optimization (Zienkiewicz and Campbell

T. Ide et al.

1973; Imam 1982; Bennett and Botkin 1985; Kikuchi et al.


1986; Azegami et al. 1997; Allaire et al. 2002) (and their references) and topology optimization (Bendse and Kikuchi
1988; Wang et al. 2003; Allaire et al. 2004; Yamada et al.
2010; Sigmund and Maute 2013) (and their references).
Today, to solve industrial applications, some optimization methods are available in commercial computer software. In this paper, we apply different structural optimization methods such as topometry (Leiva 2004), topography
(Leiva 2003), and freeform optimization (Leiva 2010), to
find lightweight designs with contact constraints. Our procedure includes nonlinear static analysis using the finite
element method, sensitivity analysis and approximation
method.
The rest of this paper is organized as follows: In
Section 2, we discuss the basic formulation of general
optimization and commonly used structural optimization
types. In Section 3, a mathematical formulation for evaluating stress with contact constraints in numerical analysis
is presented. In Section 4, design considerations of vehicle automatic transmissions are discussed, and the results
of different types of optimization techniques are shown. We
will show the numerical result using topometry optimization. After that, we will summarize our previous work that
employs topography optimization (Kitajima et al. 2010).
Furthermore, we will show the numerical result using
freeform optimization. Then, we will compare the mass
change and the stress with contact constraints regarding
topometry, topography, and freeform optimization. Finally,
in Section 5, we present the conclusion of this work.

2 Structural optimization
In this section, we recall general optimization problem
and structural optimization methods (Leiva 2011; Ide et al.
2014a).
2.1 Optimization problem
The optimization problem can be stated as:
min f (x1 , x2 , , xn ) or max f (x1 , x2 , , xn )

(1)

subject to:
gj (x1 , x2 , , xn )  0; j = 1, 2, , m

(2)

ai  xi  bi ; i = 1, 2, , n,

(3)

where m and n are the number of constraint functions


and the number of design variables, respectively. f is the

objective function, gj are the constraints, xi are the design


variables and ai and bi are the side constraints associated
with the design variables (Vanderplaats 2007, 2011).
The objective function is a function of the design variables that needs to be either minimized or maximized. For
example, mass, maximum stress or contact pressure can be
used as objective functions.
The constraint functions are function of the design variables that need to be limited to satisfy prescribed desirable maximum or minimum values. Typical constraints are:
mass, stress, displacement and contact pressure.
Design variables are numerical quantities that determine
the configuration of the system and can be changed by
the optimizer during the optimization. In structural optimization, design variables are typically parameters that can
change, directly or indirectly, the dimension of elements,
grid locations, and/or material properties.
2.2 Structural optimization methods
Structural optimization is a kind of optimization used to
improve structures. In structural optimization, the responses
are obtained by solving the governing equation of the problem, using numerical methods such as the finite element
method while the design variables correspond to parameters
that describe the structure. The structural optimization types
can be broadly categorized into sizing, shape, and topology optimizations. A more detailed explanation of these
three types can be found in our previous paper (Ide et al.
2014a). Other types are described next, such as topometry
optimization, which can be considered as a special form of
sizing optimization, and topography and freeform optimizations, which can be considered as special forms of shape
optimization.
2.2.1 Topometry optimization: element-by-element sizing
optimization
Topometry optimization is a structural optimization type
used to design structural dimensions or properties of individual elements (e.g. thickness of shell). This method can
be used to find optimal thickness distributions on shell elements (Leiva 2004). Gambling and Wang (2013) used it
for designing composite structure. Ide et al. (2014b) used
it for designing wall thickness for gear box of automatic
transmission to reduce radiated noise.
2.2.2 Topography optimization
Topography optimization is a structural optimization type
used to design surface grids on shells or composite

Structural optimization of nonlinear static analysis with contact

elements. This method is recognized as a special form of


shape optimization. Shape optimization uses a technique to
superpose the predefined alternative shapes (Vanderplaats
1979; Imam 1982). In this type of optimization method, the
general shape Y is defined as follows:
Y = a 1 Y 1 + a2 Y 2 + + a n Y n

(4)

where ai is participation coefficients and Y i is basis vector


that defines the predefined shapes, which can be also given
in the form of Y i = Y 0 + dY i , where Y 0 is the original
shape and dY i can be called perturbation vector as it defines
a perturbed shape from original shape.
Topography optimization is typically used to find optimal
bead patterns and it uses automatically generated perturbation vectors (Leiva 2003). Topography optimization can also
be used to indirectly design meshes assembled with solid
elements by placing designable shell elements on surface
faces.
2.2.3 Freeform optimization
Freeform optimization is another type of shape optimization
that uses predefined shapes or perturbation vectors. In this
method, the program splits the given perturbation into multiple perturbations on a grid-by-grid basis. This split increases
the variability of the design space when compared with traditional shape optimization. The possible distortions of the
finite element mesh are avoided by automatically generated
distortion constraints and by the use of mesh smoothing.
Further details are available in Leiva (2010).
2.3 Use of Structural optimization
Form follows function is a design principle attributed to
famed architect Louis Sullivan and commonly used in modern architecture, industrial and automobile design. It has
been used for more than a century. The principle indicates
that the structure or the shape of a building or any object
should be primarily based upon its intended function or purpose. This principle also holds for automatic transmission
designs. However, designers are not free to just design, but
rather, they have to take into account what methods of fabrication exist. In addition, if they use optimization software,
they need to know which of them can be used. Therefore,
it is important to link the different structural optimization
types with the different manufacturing types. Table 1 shows
what types of optimization methods are appropriate for
some objective functions. The automatic transmission gearbox is manufactured using casting. The inside of a gearbox
is filled with automatic transmission fluid. We do not use
basic topology optimization to avoid the resulting holes that

Table 1 Types of structural optimization versus appropriate objective


functions and constraints for the design of automatic transmission

Topometry
Topography
Freeform

Stress

Contact

Lightweight

Yes
Yes
Yes

Yes
Yes
Yes

Yes
Yes

would allow the fluid to leak out. For their flexibility, we


use topometry, topography, and freeform optimization.
2.4 Techniques to lightweight design of automatic
transmission with contact constraints
Topometry, topography, and freeform optimization are used
in this work for preliminary studies of the design of our
automatic transmissions. Table 1 shows what types of optimization methods are appropriate to satisfy different design
requirements for the design of our automatic transmission.
Gear boxes, due to their complex shapes, are normally
modeled using solid finite elements. To use topometry optimization, we add a thin layer of shell elements to the
external surface. Then we design the extra layer by designing each shell element individually. This shows the locations
of the best places to reinforce. During the optimization, the
grid points of the solid model are not moved. Using topometry to design the extra layer is simple and quick but with this
method, it is not possible to reduce the mass. On the other
hand, in topography and freeform optimization, the grid
points can move during the optimization and it is possible
to reduce the mass.

3 Contact analysis and optimization


In this section, we consider nonlinear contact analysis and
its structural optimization.
3.1 Formulation of nonlinear static analysis
with contact constraints
There are two commonly used methods to solve contact
analysis problems. One is the Lagrange multiplier method
and the other one is the penalty method. In our work, we use
the penalty method. Using the penalty method, the governing equation of the static analysis with contact constraints
can be formulated as follows:
[K(u)]{ u} = {F }

(5)

T. Ide et al.

where [K(u)] is the stiffness matrix, {F } is the load vector,


and {u} is the static displacement.
The stiffness matrix in the above equation is a function
of the displacements because depending on their values,
springs that connect the contact surfaces are used or not used
to enforce such contact.
The normal contact condition can be written as follows:
GN (x) 0,

(6)

tN (x) 0,

(7)

GN (x)tN (x) = 0

(8)

where: x refers to any point on the surface of the structure;


GN is the gap between two surfaces of the structure that can
be in contact; and tN is the normal component of the surface
traction.
The inequality involving GN represents the kinematic
condition of impenetrability between the surfaces. The
inequality involving tN represents the requirement that
the normal contact traction be compressive. The equality equation involving GN and tN represent the requirement that only one of the two individual equations be
non-zero.
The system of equations above is solved in an iterative
way so that at the end of the process the surfaces that are
in contact are properly connected. The solution starts by
assuming all surfaces are connected, with this assumption
all springs have a non-zero stiffness value and the first set
of displacements is found. If the surfaces that were assumed
to be in contact are not in contact, the method corrects the
springs associated to those surfaces and zeros their stiffnesses. With this new information, a new iteration starts and
the assumption is checked again. If a pair of surfaces come
into contact, then the stiffness of their corresponding springs
are restored and if a pair of surfaces get separated, then the
stiffness of their corresponding springs is zero. The process
continues until no change in status is observed.
3.2 Sensitivity analysis
The sensitivity with contact conditions is evaluated using
the stiffness matrix and the displacements obtained from the
converged contact analysis. The equations to calculate the
sensitivities with contact are the same as the equations to
calculate the sensitivities of the standard static analysis. The
equations are shown next.
The sensitivity of a static response, R, (e.g. stress, displacement, strain energy) with respect to a design variable,

X, is determined by the chain rule of differentiation as


follows:
R
R u
dR
=
+
.
dx
x
u x

(9)

Using the governing global equilibrium equations


([K]{u} = {F }) the displacement sensitivities are
determined as:

 

F
K
u
u
(10)
= [K]1

x
x
x
where

 

K
F

u
x
x
are referred to as pseudo-loads.
Therefore, the response sensitivity becomes :

 

F
K
R
R
dR
u .
=
+
[K]1

dx
x
u
x
x

(11)

(12)

There are two methods to calculate sensitivities: the direct


method and the adjoint method and they are explained next.
The direct method first calculates the displacement senu
u
and uses that to calculate R
sitivity x
u x to form the
response sensitivity. This method requires performing a forward/back substitution (i.e. equivalent to solving a static
loadcase) for each design variable.
 T
and
The adjoint method first calculates [K]1T R
u
then dots that with the pseudo-load to form the second part
of the response derivative. Note that because K is symmetric, [K]1T = [K]1 . Therefore, this method requires one
forward/back substitution for each response. If the number
of retained responses is smaller than the number of design
variables, then the adjoint method should provide better
performance.
The pseudoloads are formed on an element-by-element
basis and are assembled into a global vector. Where possible
and efficient, an exact analytical process is used throughout
the sensitivity calculations. In other cases, a semi-analytic
technique is used, whereby the pseudo-load is calculated
by finite difference, but the remainders of the sensitivity
calculations are fully analytic.
3.3 Approximation problem
The structural optimization problem is solved using the
approximation concepts approach. In this approach, an
approximate analysis model is created using the sensitivities of the responses and optimized at each design cycle.

Structural optimization of nonlinear static analysis with contact

The design solution of the approximate optimization is then


used to update the finite element model, and a full system
analysis is performed to create the next approximate analysis model. The sequence of design cycles continues until
the approximate optimum design converges to the actual
optimum design. In the mid-seventies, Schmit et al. (1974,
1976) introduced approximation concepts for traditional
structural optimization. These concepts, in the eighties
and early nineties, were refined to improve the quality of
approximations (Vanderplaats and Salajegheh 1989; Canfield 1990). The approximate problem is solved using either
the BIGDOT (Vanderplaats 2000) or DOT (DOT Users
Manual 1999) optimizers. The purpose of using the approximation concepts approach is to reduce the number of design
cycles in order to reduce time. With these approximations a
good engineering answer can be typically found in 10 to 15
design cycles.

4 Numerical examples
As a demonstrative problem, we consider a front engine
and front wheel drive type (herein after, we call FF type)
automatic transmission. The interior parts of the automatic
transmission have little design freedom because of precise
machinery. As a design space, we consider the thickness of
the walls of the gearbox.
4.1 Finite element model
Our finite element model uses three-dimensional solid elements. Figure 1 shows the model of our automatic transmission (Kitajima et al. 2010). The model consists of 3,678,557

Fig. 1 Finite element model of FF type automatic transmission

elements (tetra and hexahedron) with 4,761,454 grids and


15,412,771 DOFs.
4.2 Notation
We define the notation that Kitajima et al. (2010) employed.
F (X) : Stress,

(13)

M(X) : Mass,

(14)

MT (X) : T arget mass,

(15)

where
X = {x1 , x2 , , xn } : Design variable.

(16)

Furthermore we define the maximum value of F (X) as


follows:

=
Max(F (X)).
def.

(17)

4.3 Optimization problem


Stress responses depend on the loading forces and we need
to minimize the peak values of the stresses over all of
the applied loading forces. However, the peak value of
the stresses can easily shift from one stress to another,
when changing the size or shape of the structure during the
optimization. To overcome this difficulty, we introduce an
artificial design variable called beta and add additional constraint equations using this beta. This method is called the
beta method (Taylor and Bendse 1984) and it is widely
used to solve the min-max problem (Vanderplaats 2007,
2011). The objective function is set to minimize beta and the
scaled maximum stress is constrained to be less than beta. If
beta is reduced, the maximum value of stress will be reduced
in order to satisfy the beta constraints.
With the beta method, two additional benefits can be
recognized as follows (Kitajima et al. 2010):
1. Even if the location of the maximum stress changes during optimization design cycles, the maximum response
and its sensitivities are stored throughout the process.
Therefore, continuous optimization can be carried out.
2. Due to the efficient constraint screening, only a limited
number of responses (those used in the beta constraints)
are retained.

T. Ide et al.
Fig. 2 Design variables; a
Housing, b Case

4.3.1 Topometry optimization


Our optimization problem for topometry optimization is
defined as follows:
Objective function
min
subject to
F (X)  .
Design variable
0.0005  xj  2.0mm (j = 1, 2, , n)
Figure 2a and b show the area where the design variables
will act, the areas correspond to surface shells generated
from the finite element model.
4.3.2 Topography optimization
Next, we consider topography optimization to reduce both
mass and stress. Here, the objective function is set to minimize stress. The stress is dealt using constraint functions
where the stress in topometry optimization. Then our optimization problem for topography optimization is defined as
follows:
Objective function
min
subject to
F (X)  ,
M(X)  MT (X).
Design variable
2.0  xj  2.0mm (j = 1, 2, , n)
The design area is the same as in topometry optimization.
But, in this case we design the location of the grids.
4.3.3 Freeform optimization
Finally, we consider freeform optimization. The mathematical formulation of optimization problem is the same as
topography optimization.

Objective function
min
subject to
F (X)  ,
M(X)  MT (X).
Design variable
2.0  xj  2.0mm (j = 1, 2, , n)
The design area is the same as topometry optimization
and topography optimization cases. In this case, like in
topography optimization, we design the location of selected
grids.
4.4 Numerical results
In this section, we present the numerical results obtained
using topometry, topography, and freeform optimizations,
applied to the gearbox design.
4.4.1 Topometry optimization
Topometry optimization was used to find the best places
which can reduce stress.
The initial thickness is set in this case to 0.0005mm. This
initial small value allows us to start with an initial design
practically equal to the nominal design.
Figure 3 shows the optimal thickness distribution of the
added shell elements. The red color indicates the upper
bound of the design variable thickness (0.8mm), while the
green color indicates the lower bound of the design variable thickness (0.0005mm). We note that although the lower
bound of design variable here is set to 0.0005mm, the lower
bound of color bar in the figure is set to 0.0mm to facilitate the comparison of thickness distribution with the other
results which we will discuss later. Designable areas of outer
surfaces are designated using 163,616 shell elements that
were generated to cover the solid elements in the surface.
Since each shell element is designed with a unique design
variable, 163,616 design variables are used in total on the
design.

Structural optimization of nonlinear static analysis with contact


Fig. 3 Topometry optimization
results

4.4.2 Topography optimization


Topography optimization was used to find the best places
that can simultaneously reduce mass and stress.
Figure 4 shows the optimal shape obtained using topography optimization. The red color region indicates the
upper bound of shape changes (the upper bounds of design
variables were set to 2.0mm) and the blue color region

Fig. 4 Topography
optimization results

indicates the lower bound of shape changes (the lower


bounds of design variables were set to -2.0mm). Designable areas of outer surfaces are designed using a thin
skin of shell elements that contain 78,392 grids on the
solid element surfaces. A corresponding 78,392 design
variables were automatically generated and used in this
design case. Details for a similar set up are described in
Kitajima et al. (2010).

T. Ide et al.
Fig. 5 Freeform optimization
results

4.4.3 Freeform optimization


Freeform optimization was used to find the best places that
can simultaneously reduce mass and stress.

Figure 5 shows the resultant optimal shape. As with


the topography optimization case described earlier the red
color region indicates the upper bound of shape changes
(the upper bounds of the design variables were set to

Fig. 6 Comparison of mass change (Vertical axis indicates the design cycle.)

Structural optimization of nonlinear static analysis with contact


Fig. 7 Baseline model von
Mises stress distribution

2.0mm) and the blue color region indicates the lower


bounds of shape changes (the lower bounds of design
variables were set to 2.0mm). Designable areas of the
outer surfaces are assigned; they contain 85,150 grids
on the solid element surfaces and 85,150 corresponding
design variables are automatically created and used in the
optimization.
4.4.4 Comparison of thickness distribution
Comparing the colored thickness distribution obtained using
topography and freeform optimization, we observe a similar
distribution. However, the topography optimization results
show larger and deeper blue areas that indicate more mass
reduction. The numerical results for the mass, as shown later
in Fig. 6, confirm this observation.
On the other hand, comparing the thickness distribution
obtained using topometry and the other two optimizations,
the distribution is quite different. This comes from the fact
that in this case topometry optimization has less design
freedom as it is only used for adding material which significantly affects the thickness distribution results.
4.4.5 Comparison of mass change
Figure 6 shows a comparison of mass change between the
baseline design and each of the topometry, topography, and
Fig. 8 Topometry optimization
von Mises stress distribution

freeform optimization designs. The three optimization cases


took different number of design cycles to finish: Topography took 24 design cycles, freeform took 35 and topometry took 40. It should be mentioned that the optimization
terminates if the changes in design variables are sufficiently small or the objective function is not significantly
improved. In addition, the maximum design cycle was set
here to 40. Topography and freeform optimization successfully reduced the total mass of our automatic transmission
gearbox model. While topography optimization reduced the
mass by 377.7g, freeform reduced it by 352.7g. On
the other hand, topometry optimization could not reduce
the mass because it was applied to an additional layer of
elements, resulting in that the mass being increased by
approximately +77.3g.
4.5 Comparison of von Mises stress
Figures 7, 8, 9 and 10 show a comparison of the von Mises
stress distribution for the baseline design, and the topometry, topography, and freeform final optimization designs.
Figure 11 shows close-up view of the hot-spot comparison of von Mises stress. In all three optimization cases, the
von Mises stress maximum peak value was reduced . The
reduction ratios of the maximum von Mises stress value
for topometry, topography and freeform optimizations are,
respectively, : 12.5 %, 16.6 % and 13.6 %.

T. Ide et al.
Fig. 9 Topography optimization
von Mises stress distribution

Fig. 10 Freeform optimization


von Mises stress distribution

Fig. 11 Close up view of the


comparison of hot spot of von
Mises stress distribution

Structural optimization of nonlinear static analysis with contact


Table 2 Types of structural optimization versus mass change and
maximum von Mises stress value of automatic transmission

Topometry
Topography
Freeform

Mass change

von Mises stress change

+77.3g
377.7g
352.7g

12.5 %
16.6 %
13.6 %

Table 2 shows comparison between the types of optimization methods, the mass changes, and the maximum von
Mises stress changes.

5 Conclusion
We proposed sensitivity analysis approach for nonlinear
static analysis with contact constraints. And the validity
of topometry, topography, and freeform optimization methods have been presented to design lightweight structure
and to minimize peak value of von Mises stress for gear
box of automatic transmission. Traditionally, most design
changes are coming from engineering judgment based on
experience and physical experimental results which takes
a long time to obtain the desired designs. Structural optimization methods, on the other hand, overcome this long
process. Topometry optimization was successfully applied
for design thickness reinforcement of automatic transmission to minimize the peak value of von Mises stress. Both
topography and freeform optimization achieved reducing
simultaneously mass and maximum peak value of von Mises
stress.

References
Allaire G, Jouve F, Toader AM (2002) A level-set method for shape
optimization. C R Math 334(12):11251130
Allaire G, Jouve F, Toader AM (2004) Structural optimization using
sensitivity analysis and a level-set method. J Comput Phys
194(1):363393
Azegami H, Kaizu S, Shimoda M, Katamine E (1997) Irregularity
of shape optimization problems and an improvement technique.
Comput Aided Optim Des Struct 5:309326
Bendse MP, Kikuchi N (1988) Generating optimal topologies in
structural design using a homogenization method. Comput Methods Appl Mech Eng 71(2):197224
Bennett JA, Botkin ME (1985) Structural shape optimization with
geometric description and adaptive mesh refinement. AIAA J
23(3):458464
Canfield RA (1990) High quality approximations of eigenvalues in
structural optimization of trusses. AIAA J 28(6):11161122
DOT Design Optimization Tools Users Manual, Version 5.0 (1999)
Vanderplaats Research & Development, Inc., Colorado Springs, CO

Gambling M, Wang S (2013) The development of TruPly, an efficient


composite optimization tool for Simulia Abaqus. In: Proceedings
of 10th world congress of structural and multidisciplinary optimization, Orlando, Florida, USA, 19-24 May, 5151
Herskovits J, Leontiev A, Dias G, Santos G (2000) Contact shape
optimization: a bilevel programming approach. Struct Multidisc
Optim 20(3):214222
Ide T, Otomori M, Leiva JP, Watson BC (2014a) Structural optimization methods and techniques to design light and efficient
automatic transmission with low radiated noise. Struct Multidisc
Optim 50(6):11371150
Ide T, Otomori M, Leiva JP, Watson BC (2014b) An efficient approach
to reduce radiated noise of automatic transmission of vehicles
using topometry optimization. Adv Comput Sci Eng 13(2):107
131
Imam MH (1982) Three-dimensional shape optimization. Int J Numer
Methods Eng 18(5):661673
Kikuchi N, Chung KY, Torigaki T, Taylor JE (1986) Adaptive finite
element methods for shape optimization of linearly elastic structures. Comput Methods Appl Mech Eng 57(1):6789
Kitajima H, Ide T, Leiva JP, Watson BC (2010) Structural optimization
with contact constraints applied to design automatic transmission
of vehicles. SAE Technical Paper 2014-01-0407
Leiva JP (2003) Methods for generation perturbation vectors for topography optimization for structures. In: Proceedings of 5th world
congress of structural and multidisciplinary optimization. Lido di
Jesolo-Venice, Italy, 19-23 May, A070
Leiva JP (2004) Topometry optimization: a new capability to perform
element by element sizing optimization of structures. In: Proceedings of 10th AIAA/ISSMO symposium on multidisciplinary
analysis and optimization Albany, New York, USA, 30 August-1
September, 2004-4595
Leiva JP (2010) Freeform optimization: a new capability to perform
grid by grid shape optimization of structures. In: Proceedings
of 6th China-Japan-Korea Joint Symposium on Optimization of
Structural and Mechanical Systems, Kyoto, Japan, 22-25 June,
J-13
Leiva JP (2011) Structural optimization methods and techniques
to design efficient car bodies. In: Proceedings of international
automotive body congress 2011, Troy, Michigan, USA, 9-10
November
Schmit LA (1960) Structural design by systematic synthesis. In: Proceedings of 2nd conference on electronic computation ASCE,
New York, USA, 8-9 September
Schmit LA, Farshi B (1974) Some approximation concepts for structural synthesis. AIAA J 12(5):692699
Schmit LA, Miura H (1976) Approximation concepts for efficient
structural synthesis, NASA CR-2552, March 1
Sigmund O, Maute K (2013) Topology optimization approach. Struct
Multidisc Optim 48(6):10311055
Taylor JE, Bendse MP (1984) An interpretation for min-max structural design problems including a method for relaxing constraints.
Int J Solids Struct 20(4):301314
Vanderplaats GN (1979) Approximation concepts for numerical airfoil
optimization. NASA Technical paper 1370
Vanderplaats GN (2000) Very large scale optimization. In: 8th
AIAA/USAF/NASA/ISSMO Symposium at Multidisciplinary
Analysis and Optimization, Long Beach, CA September 6-8
Vanderplaats GN (2007) Numerical optimization techniques for engineering design: with applications. Vanderplaats Research &
Development, Inc., Colorado
Vanderplaats GN (2011) Multidiscipline design optimization. Vanderplaats Research & Development, Inc., Colorado Springs

T. Ide et al.
Vanderplaats GN, Salajegheh E (1989) New approximation method
for stress constraints in structural synthesis. AIAA J 27(3):352
358
Wang MY, Wang X, Guo D (2003) A level set method for structural
topology optimization. Comput Methods Appl Mech Eng 192(1
2):227246
Yamada T, Izui K, Nishiwaki S, Takezawa A (2010) A topology
optimization method based on the level set method incorporating

a fictitious interface energy. Comput Methods Appl Mech Eng


199(4548):28762891
Yi SI, Lee HA, Park GJ (2010) Optimization of a structure with contact
conditions using equivalent loads. J Mech Sci Technol 25(3):773
782
Zienkiewicz OC, Campbell JS (1973) Shape optimization and sequential linear programming. In: Optimum structural design - theory
and applications. Wiley, pp 109126

Você também pode gostar