in random response analysis [12].

Karpel presented a procedure for finite-element models for TB structures using superelement
the dynamics analysis of complex structures based on fictitious-mass modeling; the dynamics analysis segment expands on the modal
CMS, which is in excellent agreement with that performed for the full analysis and frequency response analysis processes, and further
model [13]. Zhang applied the superelement approach to the provides a comparison of the feasibilities of superelement and
modeling and dynamics simulation of bodies in white (BIW) based traditional modeling methods in solving vehicle dynamics problems
on the design requirements regarding automotive comfort [14]. with the application of the method in the simulation of a full vehicle
Lagache introduced an orthogonal method into the CMS of dynamic under accelerating conditions.
responses in coupled acoustic vibrations, in order to increase the
accuracy of mode superposition [15]. Legay studied the coupled
acoustic vibrations of vehicle interior noise using a combination of Theory of Superelement Modeling-Based
CMS and an expanded finite element approach [16]. Dynamic Analysis
The superelement modeling method proposed by this study is
In this paper, we build on the aforementioned studies to propose a developed based on the theory of component mode synthesis. During
superelement modeling-based CMS approach for the simulation and the computation of dynamics problems, a typical superelement
analysis of vehicle structure dynamics. This approach was used to modeling approach consists of three stages, superelement
perform simulated body dynamics analysis for a certain multi- condensation, solution and data restoration of residual structure, and
purpose vehicle (MPV) model. The paper expands on the data restoration of superelements, as shown in Figure 2.
superelement modeling of a finite-element trimmed body (TB) model.
The feasibility of superelement modeling in vehicle dynamics
analysis is examined through application of the method to full-vehicle
analysis under accelerating conditions, and comparing the accuracy
and computational times of superelement and traditional modeling
methods in the dynamics analysis (mode and frequency response
analysis) of body structures. With this method, the efficiency of
dynamics computation can be greatly increased without reducing the
accuracy of computation.

Objective and Approach

Figure 2. Process of superelement modeling-based dynamics analysis

From a theoretical perspective, the method can be further divided into

the following computational steps. And the terms used in the
following equations have been defined in the table of Definitions/
Abbreviations section.

Step 1: Subsystem Division

The initial system is divided into several subsystems along interfaces
depending on the purpose of analysis, which are deemed
superelements. Here the set of a superelement’s DOF consist of the
B-set and the O-set. Thus the [Mff] and [Kff] can be represented as:

Figure 1. Technical approach

This study aims to improve the computational efficiency of vehicle (1)

dynamics simulation, particularly to reduce the computational time of
TB dynamics analysis for NVH characteristics, through the use of the
proposed superelement modeling method to build finite-element TB (2)
models for dynamics simulation and analysis. Its approach is as
shown in Figure 1, i.e. to examine the advantages of superelement
modelingbased dynamics simulation and analysis from the angles of Step 2: Extracting Constraint Modal Parameters
vehicle modeling, dynamics analysis and engineering applications. The superelement modal parameters of boundary nodes under unit
The vehicle modeling segment examines the construction of displacement constraints are solved using the following matrix
formula:
 3. The oil tank, radiator, trim, etc. are simulated using concentrated
mass simulation of non-structural members;
(3) 4. Window weather strips, door locks and door weather strips are
simulated using three-directional spring-dampener elements
Step 3: Extracting the Main Modal Parameters under (CBUSH);
Fixed Boundary Conditions 5. The main material properties are shown in Table 1.
The main modal parameters under fixed boundary conditions are
Table 1. Main material properties of vehicle body
extracted by solving the following matrix formula:

(4)

Step 4: Solving the General Mass and Stiffness Matrix

The main mode and constraint mode are combined into a general
coordinate transformation matrix, represented by the following:

(5)
Calibration of the Finite-Element Model
The BIW’s modal frequencies and shapes directly affect the vehicle’s
The matrix is used to transform the mass and rigidity matrices, dynamics performance, and reliable BIW modal parameters are
obtaining the general mass matrix [Maa] and the general stiffness necessary for the vehicle body’s finite-element model[17]. For this
matrix [Kaa], represented as follows: reason, we verified the accuracy of the finite-element model by
calibrating the BIW model’s free modal frequencies and shapes.
Figures 3 and 4 respectively shows the BIW’s finite-element model
(6) (including the front windshield) and suspension status in free modal
testing. Table 2 shows the simulation and test results of BIW free
modal parameters within the 20∼80Hz range. As indicated by the
(7) table, the simulation results have good consistency with the test
results, with generally consistent modal shape, and modal frequency
Step 5: General Assembly and Solution error less than 3%. Therefore the finite-element model is sufficiently
The superelement and residual structure matrices are assembled to obtain accurate for the purpose of the superelement model.
the condensed matrix formula of the full structure, and the system’s
parameters are acquired by solving the dynamics balance formula.

Construction of Superelement Model of Vehicle

Body Structure
In the superelement modeling approach, the vehicle body’s
superelement model can be obtained through modifying a finite-
element model constructed using the traditional modeling method. In
order to ensure its accuracy, the model is calibrated based on a BIW
mode test. A superelement body model has been constructed for a
certain MPV body as follows.
Figure 3. Finite-element BIW model (with front windshield)

Construction of Traditional Finite-Element Model

The traditional finite-element model of a certain MPV body (including
BIW and closure system) contains approximately 1,680,000 elements.
The model was built using the finite-element preprocessing software
Hypermesh based on the automotive industry’s computer-aided
engineering (CAE) simulation and analysis standards and relevant
research, and adheres to the following principles:

1. All stamping parts are simulated as 4-node quadrilateral shell

elements;
2. Point welds are simulated using area contact models (ACM),
and weld joints are simulated using rigid elements (RBE2);
Figure 4. Suspension status of BIW modal test
Table 2. BIW modal parameters: simulation vs. test values Table 3. Information of superelement model

Construction of Body Superelement Model

The superelement model is constructed based on the calibrated
finite-element model. This is done by dividing the full body
system into subsystems, creating the superelement models of Dynamics Simulation and Analysis of Vehicle Body
residual structures and various subsystems, according to the The superelement model obtained using the aforementioned method
following principles: can be used to perform dynamics simulation and analysis, with the
computational efficiency and accuracy compared to that of simulation
1. The number of interface nodes between superelements is and analysis based on the traditional modeling method. To represent a
reduced as much as possible to achieve the optimal reduction of typical analysis, we focused on the modal analysis and frequency
model DOF; response analysis within the 20∼100Hz range, commonly used in
2. The division of subsystems should ensure the integrity of the NVH studies. All analyses were conducted using the same Dell
component structures, which is beneficial for the management, T7600 workstation, and the simulation and analysis processes were
analysis and optimization of the models; run without redundant concurrent programs.
3. When using nodes of rigid element (RBE2) as boundaries, the
independent node of RBE2 should be deemed the boundary node.
Modal Analysis of Vehicle Body Structure
Table 4. Free modal information of vehicle body structure
Based on the principles and the characteristics of the MPV body, the
subsystems are divided as shown in Figure 5. The finite-element
model is converted into a superelement model, with the closure
system’s left front door, right front door, left rear door, right rear
door, back door, engine cover, and BIW and attachments are each
considered an independent subsystem, and converted into a
superelement; the undefined structures are considered the residual
structures, consisting primarily of information on boundary nodes
and scalar points. Thus the superelement model is established, with
information on each subsystem superelement (ID number and size)
shown in Table 3.

Free modal analyses were performed on the vehicle body using the
traditional finite-element model and superelement model, and the
accuracy and computational efficiency of the two approaches were
compared. In terms of accuracy, Table 4 shows the resulting modal
frequency of the superelement model, which displays errors within
the 1% range compared to traditional models, caused primarily by
high-order modal truncation; the modal shape corresponding to each
modal frequency order are generally consistent, without any
information loss. In terms of efficiency, the superelement model has
Figure 5. Division of subsystems an extraction time for modal information of 9323 seconds, while the
traditional method has an extraction time of 10225 seconds, which is
an 8.8% conservation of computational time. Therefore using
superelement modeling for the analysis can result in a significant time
reduction of about 8.8% without decreasing accuracy.
Frequency Response Analysis of Vehicle Body Structure the results of the superelement model analysis are generally
Frequency response analyses were performed based on the traditional consistent with the traditional model, indicating that the method can
finite-element model and superelement model. To represent a typical obtain comparable accuracy in frequency response analysis of vehicle
analysis as per the actual vehicle situation of vibration overload in bodies. Compared to the test, the peak frequency points and the
front row floor, we performed simulation and analysis of the full general trends of simulation display good approximation of the
vehicle under accelerating conditions, using the vibration frequency floor’s vibration responses. The points of main problem frequencies
response function of the interior front row floor as the target of in particular display errors within 3%, as shown in Table 5. Some
observation for a comparison with a full-vehicle road test. The differences between the analysis and the test are shown in the
locations of monitored points in the simulation and the road test are amplitudes, which is mostly caused by the simplification of upwind
shown in Figure 6. In order to simulate the accelerating conditions, resistance, tire-road surface excitations, and higher-order excitations
the system’s boundary conditions were set as follows: the DOF of the to the engine. However, since the peak frequency points are the focus
mounting’s active side were constrained, and the translational DOF of of frequency response analysis, the deviations in amplitudes are
the mounting’s passive side were released.; 2nd-order accelerating acceptable. Therefore, using superelement modeling for the analysis
excitations (as the excitations of a four-cylinder engine are primarily of vehicle frequency responses can achieve consistency in peak
of 2nd-order) in X, Y and Z directions corresponding to each frequency points and general trends between simulation and testing
accelerating condition (see also the full-vehicle coordinate system in for interior front row floors.
Figure 5) were applied to the engine mounting’s passive side,
consisting of amplitude, frequency and phase components, as the
system’s load excitations. As an example, Figure 7 shows the
amplitude and phase of 2nd-order excitation applied to the rear
mounting’s passive side under the 3rd gear wide open throttle (3G
WOT) condition.

Figure 6. Locations of monitored points in interior front row floor for

vibration response

Figure 8. Acceleration response of front row floor point under different

acceleration conditions: (a) 3rd gear, (b) 4th gear.

Table 5. Main problem frequency points: simulation vs. test

Figure 7. Amplitude and phase of 2nd-order accelerating excitation at rear

mounting’s passive side under 3G WOT condition

The simulation results and the existing results of road test for
vibration responses under the 3rd and 4th gear accelerating conditions
are shown in Figure 8. As can be seen, the main problem vibration
peak frequencies to the 3rd and 4th gear accelerating conditions are
34Hz, 53Hz and 42Hz, 49Hz, respectively, and the peak frequency
points in vibration response curve, which impacting and increasing Through further analysis by means of the transfer path analysis,
the vehicle interior sound pressure level with same frequency, were modal contribution factor analysis and operation deformation
generally defined by means of the objective and subjective NVH test; analysis, it was found that the main problem frequencies are related
to the modal frequencies of the closure system, and optimization of analysis computation for vehicle bodies. For the purpose of structural
the closure system structure was needed to avoid the current problem optimization, which involves repeated computations, the
frequencies. However, even tiny structural optimizations to the computational time can be reduced by 87.7% for frequency response
closure system would require us to spend more development times to analyses of optimizations.
repeat the simulation and analysis under accelerating conditions. This
The superelement modeling-based dynamic simulation and analysis
issue allows the superelement model’s advantage in efficiency to
for vehicle body structure has good engineering value. Using the
shine. In theory, the superelement model’s computational time in
method, consistency between simulation and testing can be achieved
dynamics analysis is the sum of the time for subsystem condensation
on peak frequency points and general trends for the vibration
and the time for assembly and solution. During the initial
frequency response functions of interior front row floors under
computation, the model reduces the number of equations to be solved
accelerating conditions.
through condensation, i.e. the dimensions of the solutions are
reduced, which reduces the computational time; during the repeated
computations, with the changes made to the structure, only the References
modified superelements need to be condensed again, while the 1. Afaneh, A., Abdelhamid, M., and Qatu, M., "Engineering
condensation results from the first computation can be directly Challenges with Vehicle Noise and Vibration in Product
utilized, thus further reducing the computational time. Based on the Development," SAE Technical Paper 2007-01-2434, 2007,
aforementioned theoretical advantage of superelement method, while doi:10.4271/2007-01-2434.
the various schemes of modified closure system structure were 2. Albert, A. and Schwarz, A., "Automated Identification of
carried out by the applicaiton of traditional and superelement method, NVHPhenomena in Vehicles," SAE Technical Paper 2011-01-
respertively, for obtaining the corresponding vibration frequency 1656, 2011, doi:10.4271/2011-01-1656.
response function, the development times with the usage of 3. Gladwell G.M.L., "Branch mode analysis of vibrating systems,"
superelement method could be saved with the increasing of scheme Journal of Sound & Vibration. 1:41-59, 1964, doi:10.1016/0022-
number. Thus, a comparison of efficiency between superelement 460X(64)90006-9.
modeling and traditional modeling shows as Table 6. As can be seen, 4. Hurty, W. C.,"Dynamic analysis of structural systems
the superelement modeling can conserve 31.7% of the computational using component modes," AIAA Journal. 3:678-685, 1965,
time during the first simulation, and for repeated computation during doi:10.2514/3.2947.
each optimization, 87.7% of time can be conserved. If we consider
5. Craig, Roy. "A new substructure system identification method"
the repeated computational time to be a fixed value, the efficiency
36th Structures, Structural Dynamics and Materials Conference.
advantage of the superelement model becomes even greater as the
doi: 10.2514/6.1995-1298.
number of optimizations increases.
6. Craig, Roy. "Coupling of substructures for dynamic analyses
Table 6. Computational time: superelement model vs. traditional model -An overview" 41st Structures, Structural Dynamics, and
Materials Conference and Exhibit. doi:10.2514/6.2000-1573.
7. Agrawal, O. P.and Kumar, R.,"A superelement model for
analysis of multi-body system dynamics," Computers
& Structures. 32:1085-1091, 1989, doi:10.1016/0045-
7949(89)90410-0.
8. Butland, A.and Avitabile, P.,"A reduced order, test verified
component mode synthesis approach for system modeling
applications," Mechanical Systems and Signal Processing.
24:904-921 ,2010,doi:10.1016/j.ymssp.2009.08.008.
9. Lu, K., Liu, Y., Zhang, W.and Qiu, H.et al.," Free-Interface
Component Mode Synthesis Method with Link Substructure
as Super-Element," Procedia Engineering. 16:685-694 ,2011,
Summary/Conclusions doi:10.1016/j.proeng.2011.08.1142.
In this paper, we proposed a method of dynamics simulation and 10. Ding, Z., Li, L.and Hu, Y.,"A free interface component mode
analysis based on superelement modeling to increase the efficiency of synthesis method for viscoelastically damped systems,"Journal
dynamics simulation for vehicle body structure. The method is used of Sound and Vibration. 365:199-215 ,2016, doi:10.1016/j.
to perform simulation and analysis on a certain MPV model, and the jsv.2015.11.040.
results of modal analysis and frequency response analysis are
11. Unruh, J. F.,"Finite element subvolume technique for structural-
compared to the traditional model to examine the method’s
borne interior noise prediction," Journal of Aircraft. 17(6):1021-
advantages in accuracy and efficiency. The results show that:
1026 ,1986, doi:10.2514/3.57921.
In the modal analysis, the modal frequency of the superelement 12. SUNG, S. H.and NEFSKE, D. J., "Component mode synthesis
model displayed <1% error in main problem points. In terms of of a vehicle structural-acoustic system model,"AIAA Journal.
efficiency, a significant time reduction of about 8.8% was obtained 24(6):1021-1026 ,1986, doi:10.2514/3.9379
without decreasing accuracy.
13. Karpel, M., Moulin, B.and Feldgun, V., "Component mode
In the frequency response analysis, the points of main problem synthesis of a vehicle system model using the fictitious mass
frequencies display errors within 3% using superelement modeling. method,"Journal of Vibration and Acoustics. 129(1):73-83
This method can also effectively increase the efficiency of dynamic ,2006, doi:10.1115/1.2202156.
14. Zhang, J.,"Dynamic analysis and optimization of car's body- Acknowledgments
inwhite based on superelement approach," Journal of Southeast
This material is based upon work supported by the National Natural
University, 20(3):324-327,2004, doi: 10.3969/j.issn.1003-
Science Foundation of China (Grant No. 51575410). Any opinions,
7985.2004.03.012.
findings, and conclusions or recommendations expressed in this
15. Lagache, J., Assaf, S.and Schulte, C.,"Finite element synthesis publication are those of the author(s) and do not necessarily reflect
of structural or acoustic receptances in view of practical design the views of the National Science Foundation. This work also
applications,"Journal of Sound and Vibration. 310:313-351 supported by the Fundamental Research Funds for the Central
,2008, doi:10.1016/j.jsv.2007.07.051. Universities (WUT:2014-VII-004).
16. Legay, A., "The extended finite element method combined
with a modal synthesis approach for vibro-acoustic problems,"
International Journal for Numerical Methods in Engineering. Definitions/Abbreviations
101:329-350 ,2014, doi:10.1002/nme.4798. B-set - DOF set of superelement’s boundary nodes
17. Liu, B., Zhan, Z., Zhao, X., Chen, H. et al., "A Research on O-set - DOF set of superelement’s internal nodes
the Body-in-White (BIW) Weight Reduction at the Conceptual [Mff] - Superelement’s mass matrix
Design Phase," SAE Technical Paper 2014-01-0743, 2014,
[Kff] - Superelement’s stiffness matrix
doi:10.4271/2014-01-0743.
[Moo] - Mass matrix of internal nodes
Contact Information [Koo] - Stiffness matrix of internal nodes
Author: Chihua Lu [Mbb] - Mass matrix of boundary nodes
Mailing address: School of Automotive Engineering, Wuhan [Kbb] - Stiffness matrix of boundary nodes
University of Technology
[Mbo] - Coupled mass matrices of boundary nodes
No.122 Luoshi Road, Wuhan, Hubei Province,430070, P.R. China
[email protected] [Mob] - Coupled mass matrices of internal nodes
Tel: 086-18507103566 [Kbo] - Coupled rigidity matrices of boundary nodes

Co-author: Wenxin Yang [Kob] - Coupled rigidity matrices of internal nodes

Mailing address: School of Automotive Engineering, Wuhan {ϕob} - Characteristic vector of an internal node under unit
University of Technology displacement
No.122 Luoshi Road, Wuhan, Hubei Province,430070, P.R. China {lbb} - Unit matrix
[email protected]
{Pb} - Cohesive force of a superelement interface
Tel: 086-15071309070
ωk - Characteristic value
Corresponding-author: Hao Zheng
{ϕoo} - Vector of a superelement’s fixed boundary condition
Mailing address: School of Automotive Engineering, Wuhan
University of Technology {ϕG} - General coordinate transformation matrix
No.122 Luoshi Road, Wuhan, Hubei Province,430070, P.R. China [Maa] - General mass matrix
[email protected] [Kaa] - General stiffness matrix
Tel: 086-13554236675
BIW - Body in white
Co-author: Jingqiang Liang MPV - Multi-purpose vehicle
Mailing address: SAIC GM Wuling Automobile Co.,Ltd.
SE model - Super-element model
No.18 Hexi Road, Liuzhou, Guangxi Province,545000, P.R. China
[email protected] TB - Trimmed body
Tel: 086-18077208096 Superelement method - Finite element modeling method with using
superelement
Co-author: Guang Fu
Mailing address: SAIC GM Wuling Automobile Co.,Ltd. Traditional method - Finite element modeling method without using
No.18 Hexi Road, Liuzhou, Guangxi Province,545000, P.R. China superelement
guang [email protected] closure system - System constituted by all body closures
Tel: 086-15677021076

The Engineering Meetings Board has approved this paper for publication It has successfully completed SAE’s peer review process under the supervision of the session organizer The process
requires a minimum of three (3) reviews by industry experts

All rights reserved No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, electronic, mechanical, photocopying, recording, or
otherwise, without the prior written permission of SAE International

Positions and opinions advanced in this paper are those of the author(s) and not necessarily those of SAE International The author is solely responsible for the content of the paper

ISSN 0148-7191

http://papers sae org/2016-01-8050