International Journal of Automotive Technology, Vol. 12, No. 3, pp.

375−381 (2011) Copyright © 2011 KSAE

DOI 10.1007/s12239−011−0044−6 1229−9138/2011/058−08

STRUCTURAL DESIGN OF AN OUTER TIE ROD

FOR A PASSENGER CAR
J. K. KIM , Y. J. KIM , W. H. YANG , Y. C. PARK and K.-H. LEE
1) 2) 3) 4) 4)*

Graduate School, Department of Mechanical Engineering, Dong-A University, Busan 604-714, Korea
1)

Technical Center, Central Corporation, 54 Seongsan-dong, Changwon-si, Gyeongnam 641-315, Korea

2)

Die Engineering R & D Team, Hyundai Motor Company, 700 Yangjeong-dong, Buk-gu, Ulsan 683-791, Korea
3)

Department of Mechanical Engineering, Dong-A University, Busan 604-714, Korea

4)

(Received 30 November 2009; Revised 1 November 2010)

ABSTRACT−This study proposes a structural design method for an outer tie rod installed in a passenger car. The weight of
the outer tie rod is optimized by using the aluminum alloy Al6082M, which is developed as a steel-substitute material, and
applying structural optimization techniques. The high strength aluminum with improved mechanical properties was developed
to reduce the weight of the outer tie rod. The newly developed aluminum alloy Al6082M is applied as the material of the outer
tie rod. The static strength due to inertia force, durability and buckling performances are considered in the structural design
of the outer tie rod. At the proto design stage of a new outer tie rod, it is cost-effective to utilize FE (finite element) analysis
to predict each of these performances. In addition, the current trend in the structural design of automobile parts is to use
optimization techniques to reduce the weights of the parts. First, for an arbitrary base design, the static strength, the life cycle
and the buckling load are calculated to check whether the design satisfies its criteria. Then, the critical performance is selected
so as to include its loading condition only in the optimization process. In this study, the metamodel based optimization process
using kriging is adopted to obtain the minimum weight satisfying the critical design requirement. Then, the feasibility of the
determined optimum shape is investigated against the other performances. Finally, the optimum design of outer tie rod is
modified by considering forging efficiency. The performances of the final design are investigated through simulation and
experiment.
KEY WORDS : OTR (Outer tie rod), Structural design, Kriging, Buckling

1. INTRODUCTION under reasonable assumptions and produce accurate

results. Second, the optimization of car parts is mostly
Light-weight automobile parts have received much shape optimization. The finite element modeling of most
attention recently in light of the regulations on hazardous car parts uses a shell or solid element. The use of these
gases and the development of environment-friendly elements makes the setting of a shape variable difficult and
automobiles. Lightweight automobile parts can be realized distorts the finite elements during the iteration process of
by approaches focusing on material development and on the optimization so that the optimization process does not
design and manufacturing technology (Kim et al., 2009; progress. Third, in the case of the optimization application,
Song et al., 2009). These approaches have already been it is difficult to consider the technological performance
used in large-scale projects such as ULSAB (Ultralight related to manufacturing such as manufacturability. In this
Steel Auto Body) programme, ULSAC (Ultralight Steel study, optimization using a metamodel is applied to solve
Auto Body Closure) programme, ULSAS (Ultralight Steel the second problem.
Auto Suspension) (Automotive Group, 2006) programme, The outer tie rod (OTR) of the present study is the part
USAMP, USCAR and others since the end of the 1990s, and mounted on a passenger car, which is produced at the A
their importance has already been confirmed to carmakers company. The parts manufacturer producing the OTR is
and parts manufacturers. currently manufacturing the OTR as a forging part using
Structural optimization is a design technology and has steel. This steel product weighs 665 g. However, the newly
often been applied to car design problems. However, there developed aluminum is being used to replace steel to make
is a restriction in the application of this technique. First, a lightweight OTR. At the same time, the criteria of static
because structural optimization is based on finite element strength, durability and buckling of the aluminum OTR are
analysis, the finite element analysis must be performed investigated to suggest a lightweight OTR satisfying all
these criteria of static strength, durability and buckling.
*Corresponding author. e-mail: [email protected] Although there have been studies on the structural

375
376 J. K. KIM et al.

optimization of the control arm and knuckle, there has been

no study on the structural optimization of an OTR because
an OTR weighs less than a control arm or a knuckle.
However, the OTR is also becoming a target of lightweight
optimization with the recent trend of producing lightweight
parts.
The optimization technique applied in the present
research is metamodel-based optimization using kriging
(Guinta and Watson, 1998; Lee and Kang, 2007; Song et al.,
2009; Song and Lee, 2009; Kim et al., 2009). This type of
optimization solves the problem of the shape optimization
module in the existing commercial software (MSC, 2004),
where the distortion of the finite elements complicates the
optimization process. For convenient application of the Figure 2. FE model and load condition for static analysis.
metamodel-based optimization, the critical performance
among static strength due to inertial force, durability and
buckling is selected for OTR design. This performance is Table 1. Material properties of Al6082M.
set as the constraint in the optimization of the weight
minimization. Other performances are investigated for the Material Al6082M
optimum design calculated. If the criterion for each Yield stress (MPa) 340
performance is satisfied, an optimum solution has been Tensile strength (MPa) 380
found. At this time, the shape design variables are defined in
the CAD model and are programmed to automatically Young’s modulus (MPa) 72,000
perform the finite element analysis with a number of sample Density (kg/mm ) 3
2.71×10 -6

points. The performances of static strength, durability and

buckling load were calculated by MSC.Nastran (MSC,
2004), MSC.Fatigue (MSC, 2007) and Abaqus (Dassault, present research applied the inertial force conditions and
2008), respectively. criteria used by the A company. This study replaced steel
with the newly developed aluminum material to realize a
2. STRUCTURAL ANALYSIS OF OTR lightweight OTR. The material properties are given in
Table 1. The basic design of the aluminum OTR includes a
2.1. FE Analysis for Static Strength circular section of 10.0 mm in radius.
The OTR is a part in the steering system that changes the Its finite element model and boundary conditions are
direction of the car according to the movement of the given in Figure 2. The initial weight of the base model made
steering wheel. The steering power transferred from the of aluminum is 226.9 g. On the contrary, the weight of the
steering wheel to the steering gear, the pitman arm, the steel OTR is much heavier than that of the aluminum OTR.
relay rod and the inner tie rod is finally transferred to the The static strength originating from the inertial force is
wheels through the knuckle after passing through the outer calculated by the inertia relief analysis. The inertia force was
tie rod. The OTR is installed as shown in Figure 1. In obtained from ADAMS, which is software for multibody
Figure 1, x is the longitudinal direction of the car body dynamic analysis. The four loading conditions for its inertia
coordinate system, y is the width direction, and z is the relief analysis are related to the loadings of static G.V.W.
height direction. (gross vehicle weight), braking, ultimate vertical impact and
The inertial force is the general load received by a cornering. The maximum von-Mises stresses for the four
vehicle while it is being driven. Car makers or chassis parts loading conditions are shown in Table 2.
manufacturers consider load conditions and criteria in their Tetrahedron elements with three degrees of freedom
strength assessment of a vehicle due to inertia force. The were used for the FE analysis, and the size of the element
was decided by convergence analysis. Among the four
loading conditions, the maximum stress appears in the 4 th

loading condition, as shown in Figure 3. Its value is 69.7

MPa, which satisfies the criterion of static strength. The
load represented in Figure 2 is related to the 4 loading th

Table 2. Static strength due to inertia force.

Loading condition 1st 2nd 3rd 4th
Figure 1. Position of outer tie rod. Stress(MPa) 12.9 40.0 51.0 69.7
 STRUCTURAL DESIGN OF AN OUTER TIE ROD FOR A PASSENGER CAR 377

Table 4. Load cases for fatigue analysis.

Load case Fx(N) Fy(N) Fz(N)
End A -50 281 -12
Case 1
End B 50 -281 12
End A 493 -2749 117
Case 2
End B -493 2749 -117
End A 104 -578 25
Case 3
End B -104 578 -25
End A -135 753 -32
Figure 3. Static analysis result of base model (von-Mises Case 4
End B 135 -753 32
stress).

condition.
2.2. FE Analysis for Durability
This study applied a quasi-static load to the base model for
durability analysis. The quasi-static durability analysis is a
method for predicting the fatigue life by applying a load
history. In addition, the strain-life prediction method was
used for predicting the fatigue life. Because the static
structural analysis was carried out in the linear elastic
region, it could not consider the plastic region. Accordingly,
among the methods supported by the commercial software, Figure 4. Durability analysis result of the base model.
the Neuber Rule was selected. The SWT (Smith-Watson-
Topper) approach was used to consider the average stress
effects. adopted. The durability analysis conditions and assessment
The three loading conditions and their corresponding criteria are shown in Table 3. The parameters C1, C2 and C3
criteria are summarized in Table 3. Each loading condition in Table 3 show the life criteria for the respective loading
is composed of one load case or two load cases. The load conditions. The result of the 1st loading condition in Table 3
cases are specified by A company, and the number is four. is shown in Figure 4. The durability analysis results using
The force in each load case is represented as a function of MSC.Fatigue show values close to infinite life for all
GVW. For this research, four load cases are shown in Table criteria. The base model also satisfies the durability criteria.
4. The forces were obtained from ADAMS. End A and End
B in Table 4 represent the ends of the OTR in Figure 2. The 2.3. FE Analysis for Buckling
analysis procedures are the same as the ones for static Buckling due to the coupling position and shape
analysis mentioned in Sec. 2.1. For the quasi-static characteristics of the OTR is a serious issue in OTR
durability analysis, the equivalent load replacing the true structure design. The boundary and loading conditions for
load history, which is utilized in the A company, was buckling analysis are shown in Figure 5. It is necessary to
investigate buckling strength with respect to the
Table 3. Loading conditions and specifications for compressive load in the longitudinal direction. A buckling
durability assessment. analysis was carried out with the following procedures. The
Load time OTR is combined with an ITR (inner tie rod) of steel. A
Loading Static histories Assessment displacement load δ is applied on point A in the D direction
condition analysis (Cycle) by fixing all degrees of freedom of point B, which is a
Signal position combined with ITR. At this time, the buckling load,
case 1 Constant which is the reaction force generated at point A, should be
1st C1 produced by more than R0, which is as the criterion. The
case 2 Sine(-1~+1) section shape of OTR should be such that the diameter of
case 1 Constant the OTR is within the constrained length so that the OTR
2nd C2 does not interfere with other parts. The buckling analysis of
case 3 Sine(-1~+1)
the base model made of the newly developed aluminum was
3rd case 4 Sine(-1~+1) C3 conducted by using Abaqus. The analysis result of the base
378 J. K. KIM et al.

Figure 7. Definition of the design variables.

The lower bounds and the upper bounds of design

variables r1, r2 and r3 were set to proper values to make the
section smaller. The formulation for the shape optimization
Figure 5. Boundary and loading conditions for buckling. of the OTR is represented as:
Minimize W(r) (1)
Subject to Rmax ≥ R0 (2)
rL ≤ r ≤ .
rU (3)
where W and Rmax are the weight and buckling load,
respectively, and L=[14.0 2.0 8.0]Tmm and U=[17.0 5.0
r r

10.0]Tmm. However, the classical structural optimization

makes the setting of a shape variable difficult and distorts
the finite elements in the optimization process. In addition,
a number of non-linear analyses are performed in an
optimization process to solve equations (1)~(3), and the
iterations may be not terminated. Accordingly, it is more
Figure 6. Displacement vs. reaction force curve. realistic for optimization to be carried out by replacing the
reaction force appearing in equations (1)~(3) with the
metamodel. There are methods such as a response surface
model is indicated in Figure 6. The buckling load of the model, kriging, a neural network and others that generate a
base model is 12,430 N, which is less than R0, and thus, it surrogate model. The present research utilizes the kriging
does not satisfy the criterion of the A company. metamodel suitable for the prediction of a highly nonlinear
From the result of the investigation on the static strength function. When an approximate model is used, the
performance, durability performance and buckling of the objective and constraint functions in equations (1)~(2) can
base model, it was found that the OTR base model satisfied be replaced as follows:
the static strength performance and durability performance
but not the buckling performance. Accordingly, this study Minimize w(ˆr ) (4)
included a structural optimization of the OTR by consider-
ing only the buckling performance. Subject to Rmax
ˆ
≥ R0 (5)
3. OPTIMIZATION CONSIDERING BUCKLING where ^ represents the estimator of the response. That is, the
LOAD approximate optimum can be obtained by solving equations
(3)~(5). Because true responses such as the weight and
3.1. Optimization Formulation buckling load are substituted into the metamodels, the
The base model of the OTR has a circular section of 20 mm responses are mathematically expressed as functions of
in diameter. However, the circular section is disadvantageous design variables, facilitating the shape optimization.
for making the OTR lightweight. Accordingly, it was made
into an ellipse shape by using the design variables of r1 and 3.2. Kriging Theory
r3, as shown in Figure 7. The length of the minor axis could Kriging is an interpolation method named after an
be decided within 20 mm. On the contrary, the design engineer, D. G. Krige, who developed the technique while
variable r2 controls the geometric shape limit. The definition trying to increase the accuracy in predicting ore reserves. In
of the design variables leads to the maximum moment of the kriging model (Guinta and Watson, 1998; Fang et al.,
inertia about the y axis in Figure 7. 2006; Lee and Kang, 2007; Song et al., 2009; Kim et al.,
 STRUCTURAL DESIGN OF AN OUTER TIE ROD FOR A PASSENGER CAR 379

2009), the approximation model y(ˆr) for a response (r) is y

represented as
y (ˆr ) = β̂ + cT( r )C–1( y – β̂ i ) (6)
where β is a known function, c is the correlation vector, C
is the correlation matrix, y is the observed data, and i is the
unit vector, respectively. In this study, (r) can be replaced y

by and max, respectively.

W R

The correlation matrix and the correlation vector are

defined with equations (7) and (8) (Guinta and Watson,
1998; Fang ., 2006; Lee and Kang, 2007; Song
et al ., et al

2009; Kim ., 2009), respectively.

et al

n
Figure 8. Design process.
C( r , r ) = Exp
j k
–
∑θ r
i=1
i
j k2
i– i r
( = 1,.., s), ( = 1,.., s) (7) was obtained by solving equations (3)~(5). As a final step,
j n k n
for the determined optimum, the remaining performances
c(r)=[C(r,r(1)),C(r,r(2))…C(r,r(ns))]T (8) were calculated.

where is the number of design variables, θi is the -th 3.3. Optimization Results
n

correlation parameter, and s is the number of observed

i
This study solved the optimization problem represented by
data. To determine a kriging model expressed as equation
n
equations (3)~(5) by using an in-house program (Song et

(6), the parameters θ should be defined. The unknown al., 2009). The optimizer of the in-house program is the
correlation parameters, θi are calculated from GRG method, which is an optimization algorithm built in
EXCEL. All the parameters in the optimization process
( σˆ2 ) + ln C were set to their default values. The optimum parameters
Maximize – -n---s--•---ln
------------------------------ (9) were investigated by changing more than 10 initial values,
2

where σˆ2 is the estimated variance of a metamodel and θ i but the same optimum values were obtained. For the
should be greater than 0. To solve equation (9), the GRG kriging models of weight and reaction force, the optimum
(Generalized Reduced Gradient) algorithm built in EXCEL parameters are summarized in Table 5. For the DOE, the
was utilized. maximum entropy method was used as the sampling
The validity of an approximated kriging model can be method, and the number of sampling points was set as 50.
evaluated by several indexes. The root mean squared error The sample points and their responses are summarized in
(RMSE) and the error percentage are defined as Table 6. The FE analysis on each row was carried out, and
the metamodels were generated by setting the weight and
RMSE = ---1- (y – yˆ )
reaction force as response values.
n∑
n

(10)
t
2

t
i =1
i i
The validation of two kriging models measured by
Table 5. Optimum parameters of θ and β.
yˆ – y × 100
Average % error = n---1- ∑ -----------
n

(11)
t

Responses β* θ1* θ2* θ3*

i i

t
=1
y i

253.4 0.130 0.074 0.093

i

Ŵ
where t is the number of sample points for validation,
n R̂max 14547.1 0.580 0.838 1.058
which was set to 10 in this study.
The overall design process carried out in this research is Table 6. DOE using maximum entropy method.
represented in Figure 8. First, a FE model of OTR was
constructed. Second, for a base design, static analysis based No. r1 2r 3 r max(N)
R W(g)
on the inertia force, durability analysis and bucking analysis 1 14.21 3.53 8.82 12179.7 241.0
were performed to select the critical performance. Third, the 2 15.95 2.87 8.86 15519.6 271.8
DOE (Design of Experiments) using the maximum entropy
method was carried out. In this step, the shape design 3 16.49 3.17 8.14 13057.4 259.1
variables were defined in a CAD model and programmed to .. .. .. .. .. ..
automatically perform FE analysis with the number of . . . . . .
sample points. Thus, the responses of interest were replaced 50 16.01 4.13 8.06 12097.4 242.4
by the kriging models. Fourth, an approximate optimum
380 J. K. KIM et al.

Table 7. Validation of Kriging models. investigated. The optimum shape satisfied the buckling
Weight (g) Buckling load (N) criterion and the remaining two criteria.
No.
W Ŵ Rmax R̂max 4. REDESIGN CONSIDERING
1 274.5 274.5 15184.2 15861.7 MANUFACTURABILITY
2 253.3 253.3 14554.0 14821.4 An outer tie rod of aluminum was produced using a forging
3 244.3 244.3 12024.1 12119.5 process. However, the optimum design suggested in Sec. 3
4 274.3 274.3 14068.6 14711.1 did not consider its manufacturability. Given the forging
efficiency, the section of the suggested optimum design
5 264.4 264.4 14307.3 14935.2 should be modified. The appearance is somewhat different,
6 300.1 300.1 18057.7 19757.3 but its overall size is similar.
7 277.7 277.7 15895.4 16324.0 The weight of the final design was increased by 1.3 g
compared to that of the suggested optimum design. The
8 266.2 266.2 16072.1 16623.6 buckling analysis result of the final design is shown in
9 239.7 239.7 11723.2 11898.9
10 232.2 232.2 12111.8 12169.8
RMSE 0.031 689.73
% Error 0.003% 3.35%
Table 8. Optimum design and its validation.
r1* r2* r3* max (N) W (g)
* *
R

Predicted
optimum 15.42 3.56 10.00 18141 W’
Analysis 18181 W’’
% error 0.2% 0.003% Figure 10. Analysis result of buckling of the final design.

Figure 11. Experiment equipment for the buckling test.

Figure 9. Buckling analysis result of the optimum design.

equations (10) and (11) is summarized in Table 7. Each

metamodel has a relatively small error compared with the
result. Table 8 shows that the optimum calculated through
the metamodel is comparable to the estimated values and
the analysis values of weight and reaction force. The
estimated value and the true value were almost the same in
the optimum design.
The buckling analysis result of the OTR with the optimal
section shape is shown in Figure 9. The buckling load
exceeded the target reaction force R0. The remaining two
criteria for the calculated optimum design were also Figure 12. Results of buckling experiment.
 STRUCTURAL DESIGN OF AN OUTER TIE ROD FOR A PASSENGER CAR 381

Figure 10. Its reaction force is greater than R0. To verify the Project for Regional Innovation.
final design, the buckling experiment was carried out for
three specimens. The equipment for the experiment is REFERENCES

shown in Figure 11. The loading and boundary conditions

are the same as those for the simulation. Automotive Group of International Iron and Steel Institute.
The result of the buckling experiment is shown in Figure (2006). http://www.ulsab.org.
12. The range of reaction force is between 18,500 N and Dassault Systèmes Simulia Software Corporation (2008).
19,000 N. ABAQUS 6.8.1 Documentation. RI. USA.
Fang, K. T., Li, R. and Sudjianto, A. (2006). Design and
5. CONCLUSION Modeling for Computer Experiments. Chapman & Hall/
CRC. London.
This paper proposed a structural optimization technique for Guinta, A. and Watson, L. (1998). A comparison of
OTR, which is one of the parts of the steering system of a approximation modeling techniques: polynomial versus
passenger car. The optimal shape of the OTR satisfying the interpolating models. Proc. 7th AIAA/USAF/NASA/ISSMO
criteria of static strength based on inertial force, durability Symp. Multidisciplinary Analysis and Optimization, St.
and buckling strength was determined through the Louis, Mo, AIAA, 2, 392−440 (AIAA-98-4758).
suggested design process. The conclusions of the present Kim, J. K., Kim, S. G., Park, H. S., Lee, S. M., Hwang, S.
research can be drawn as follows: C., Seo, S. M., Park, Y. C. and Lee, K. H. (2009).
(1) This study succeeded in the development of a Structural optimization of an outer tie rod for passenger
lightweight OTR by replacing the current OTR, whose car. Proc. 15th Pacific Automotive Engineering Conf., 2,
initial model was made of steel, with an Al6082M APAC15−372.
OTR, whose model is based on a newly developed Lee, K. H. and Kang, D. H. (2007). Structural optimization
aluminum material, and applying the optimization of an automotive door using the kriging interpolation
technique. The suggested optimum design has 57.2% method. J. Automobile Engineering, Proc. Institution of
lighter weight than the initial steel model. The OTR Mechanical Engineers, Part D 221, 12, 1525−1534.
design was most affected by the buckling load criterion MSC Software Corporation (2004). MSC.NASTRAN 2004
rather than the static strength criterion or the durability Design Sensitivity and Optimization User’s Guide. Santa
criterion. Once the buckling load criterion was satisfied, Ana. California.
the optimized result also satisfied the remaining criteria. MSC. Software Corporation (2007). MSC. Fatigue BASIC
(2) Although the optimization technique using the kriging User’s Guide. Santa Ana. California.
metamodel utilized in this research cannot provide a Song, B. C. and Lee, K. H. (2009). Structural optimization
mathematical Kuhn-Tucker point, it can be understood of a circumferential disk brake with consideration of
that it does provide a realistic optimum. thermoelastic instability. Int. J. Automotive Technology
(3) The optimum design model was redesigned considering 10, 3, 321−328.
forging efficiency. The final design considering forging Song, B. C., Park, Y. C., Kang, S. W. and Lee, K. H.
manufacturability satisfied all performance criteria. (2009). Structural optimization of an upper control arm
considering the strength. J. Automobile Engineering,
ACKNOWLEDGEMENT−This study was funded by the Proc. Institution of Mechanical Engineers, Part D 223,
Ministry of Education, Science Technology and Korea Industrial 6, 727−735.
Technology Foundation through the Human Resource Training