Special Issue Article

Advances in Mechanical Engineering

2017, Vol. 9(8) 1–12
Ó The Author(s) 2017
Visualized simulation and design DOI: 10.1177/1687814017714729
journals.sagepub.com/home/ade
method of mechanical system
dynamics based on transfer matrix
method for multibody systems

Xiaoting Rui1, Junjie Gu1, Jianshu Zhang1, Qinbo Zhou1 and Haigen Yang2

Abstract
The transfer matrix method for multibody systems is a new method with very high computational speed developed in
recent 20 years for studying multibody system dynamics. By combining transfer matrix method for multibody systems,
computer graphics, and open-source software, this article puts forward an approach and software MSTMMSim for visua-
lized simulation and design of mechanical system dynamics. The approach includes the following procedures in sequence:
design of functional model, design of three-dimensional solid model, design of dynamics model, automated formation of
dynamics equations, and software MSTMMSim for visualized simulation and design of mechanical system dynamics. The
proposed method and software provide a platform to realize the simulation and design of complex mechanical systems
with the following characteristics: (1) automatic deduction of the overall transfer equation, (2) high computational speed,
and (3) high visualization and programming of dynamics simulation and design process. The proposed method and soft-
ware are verified by the practical example of simulation and design of a tank system dynamics using this platform.

Keywords
Multibody system dynamics, transfer matrix method for multibody systems, visualized simulation and design, pre-proces-
sor, transfer matrix method for multibody system solver, post-processor, software

Date received: 31 August 2016; accepted: 16 May 2017

Academic Editor: Chuanzeng Zhang

Introduction multibody dynamics methods, there are two common

characteristics. First, it is necessary to establish the glo-
The dynamics design of mechanical systems has become bal dynamics equation of system. Second, the system
the digital virtual pattern to promote properties and matrix is rather high for complex mechanical system,
enhance reliability of systems, thus resulting in a shor- thus resulting in a heavy computation burden.
tened development cycle and less cost.1 Various mechan- Optimization of dynamics of complex mechanical
ical systems, such as vehicles, ships, aircrafts, spacecraft,
modern weapon systems, numerically controlled 1
Institute of Launch Dynamics, Nanjing University of Science and
machine tools, and robots, may be modeled as multi-
Technology, Nanjing, P.R. China
body systems comprising rigid and/or deformable bod- 2
Engineering Research Center of Wideband Wireless Communication
ies connected with hinges in various fashion.1–6 The Technology, Ministry of Education, Nanjing University of Posts and
software of multibody system dynamics, such as UG, Telecommunications, Nanjing, P.R. China
Pro/E, CATIA, ADAMS, and ANSYS, has been widely
Corresponding author:
used nowadays for the simulation and design of Xiaoting Rui, Institute of Launch Dynamics, Nanjing University of Science
mechanical system dynamics. Now for almost all and Technology, 200 Xiaolingwei, 210094 Nanjing, P.R. China.
dynamics solvers of software based on ordinary Email: ruixt@163.net

Creative Commons CC-BY: This article is distributed under the terms of the Creative Commons Attribution 4.0 License
(http://www.creativecommons.org/licenses/by/4.0/) which permits any use, reproduction and distribution of the work without
further permission provided the original work is attributed as specified on the SAGE and Open Access pages (https://us.sagepub.com/en-us/nam/
open-access-at-sage).
2 Advances in Mechanical Engineering

systems will take long time if using these kinds of soft- method and MSTMMSim. It lowers the demand for
ware. Consequently, there is an urgent demand for high theoretical knowledge of technicians who use the simu-
speed of dynamics computation to realize the simulation lation platform and provides a stylized pattern to simu-
and design of complex mechanical system dynamics. late and design the dynamics of a mechanical system
The transfer matrix method for multibody systems with high efficiency.
(MSTMM) is a new method of multibody system
dynamics with high computational speed developed in
the past 20 years. It has been widely used in the analysis Structure of MSTMMSim
and design of complex multibody systems because of
Basic structure of MSTMMSim
its following features: no need for global dynamics
equations of system, very low order of system matrix, In this article, MSTMM is used as the core of solver
high computational speed and programming, and auto- for dynamics computation. Combining OSS including
matic deduction of the overall transfer equation.6–8 OCC and Qt,9,10 a new software for simulation and
With the rapid development of computer graphics, var- design of multibody system dynamics, named
ious open-source software (OSS), such as Open MSTMMSim similar to Yang et al.,11 is developed. It
CASCADE (OCC) and Coin3d, provides great conve- realizes the following functions: direct calling of vari-
nience for developing simulation software of multibody ous formats of three-dimensional (3D) model file, auto-
system dynamics, especially the pre-processor and post- matic generation of the dynamics parameters of a
processor. In order to simulate and design the mechani- multibody system, visualized modeling, rapid computa-
cal system dynamics efficiently, this article puts for- tion, data processing, and animation displaying.
ward an approach and software MSTMMSim based The basic structure of MSTMMSim is shown in
on MSTMM for visualized simulation and design of Figure 1. The fundamental parts of MSTMMSim con-
mechanical system dynamics by combing computer sist of three main modules: pre-processor, MSTMM
graphics technology and OSS. The practical example of solver, and post-processor. The graphical user interface
a tank system dynamics using the proposed method (GUI) of MSTMMSim is developed primarily using
and software is shown in this article to verify the C++ language and Qt framework.10 The OCC

Figure 1. Basic structure of MSTMMSim.

Rui et al. 3

Table 1. Modules of MSTMMSim.

Basic modules Sub-modules

Pre-processor Parametric geometry solid modeling based on OCC

CAD file import and export
Structure parameter automatic generation
MSD parameter automatic generation
MSD model and topology structure definition file automatic generation
MSTMM solver Transfer matrix library
Import file parser
Automatic generation of transfer equation and transfer matrix of element, overall transfer equation,
overall transfer matrix, and boundary conditions of the system
Numerical computation
Post-processor Data storage of computational results
Animation output
Curve and chart output

MSTMM: transfer matrix method for multibody systems; CAD: computer-aided design; MSD: multibody system dynamics; OCC: Open CASCADE.

geometric modeling kernel library is integrated into processing, and storage of 3D solid even if the research-
MSTMMSim. Users can import various types of ers have no specific knowledge on computer graphics.9
computer-aided design (CAD) solid models built by
other software. In MSTMMSim, the application pro- Pre-processor module of MSTMMSim
gram interface (API) of OCC can be called to achieve
The pre-processor module of MSTMMSim is devel-
basic geometry modeling. Complex geometric model
oped using OCC geometric modeling kernel. The main
can also be established rapidly through the Boolean
function of the pre-processor is to set up the models of
operations, that is, union, difference, and intersection.9
the mechanical system including functional model, 3D
The topology structure and parameters of a multibody
solid model, and dynamics model. Using pre-processor
system are automatically generated in MSTMMSim module, users can not only design the mechanical sys-
according to the connecting relations and material tem but also import CAD files with standard format,
properties defined by the user. MSTMMSim solver such as IGES, STEP, and DWG/DXF, generated by
reads the defined parameters, computes system other CAD software. As for homogeneous bodies, each
dynamics, and stores the resulting data. Then, the body has a base coordinate system whose position and
results are returned to the user in any chosen form such orientation with respect to the global coordinate system
as animations, curves, numerical values, pictures, and are used to represent the absolute position and orienta-
tabulations. The modules of MSTMMSim and their tion. The connection points can be defined by establish-
sub-modules are given in Table 1. ing the body-fixed coordinate systems on the body
whose position and orientation are represented in the
base coordinate system as shown in Figure 2. After the
Visualization of mechanical system dynamics
solid geometric figures of the mechanical system are
The visualization is a technique utilizing computer gra- generated, by setting the parameters of material prop-
phics and image processing to convert data into graph erties, the dynamics parameters such as mass, position
or image displayed on the screen, and to carry on inter- of mass center, and moment of inertia can be computed
active processing. The visualization of mechanical sys- automatically and represented in the base coordinate
tem dynamics includes the visualization of dynamics system by calling the functions getMass(),
modeling, simulation process, and computational getCenterOfMass(), and getMatrixOfInertia() in the
results, among which the visualization of the 3D solid GProp_Gprops base class of OCC.9 These positions,
models plays a significant role. The essence of visuali- orientations, and mass parameters are stored in an
zation of the 3D solid is to transform discrete 3D solid XML-based format file. The length, width, and height
data into discrete two-dimensional image data dis- of the box shown in Figure 2 are 15, 10, and 10 mm.
played on the screen in accordance with the rules of The base coordinate system is identified by position
projection of a body. coordinates (0.01, 0.01, 0.005) of its origin with respect
Currently, various OSS, such as OCC and Coin3d, to the global coordinate system and is rotated by 15°
provides a convenient solution for the visualization of about the Z-axis. The density of the body is set to be
3D solid. This makes it much easier for developers of 7900 kg/m3. Then, the parameters stored in the XML-
software to realize the modeling, displaying, interactive based file are formatted as follows:
4 Advances in Mechanical Engineering

Figure 2. Definition of connection point.

\RigidBody Name = ‘‘Box’’ ID = ‘‘1’’ Label = are used to obtain the dynamics parameters of the whole
‘‘Box’’. component. The parameters are stored in the format of
\AbsolutePosition.0.01, 0.01, 0.005\/ ABAQUS inp document. Figures 3 and 4 show the finite
AbsolutePosition. element model of a beam and connection points on it,
\AbsoluteOrientation.0, 0, 0.2617994\/ respectively. After the parameters of the bodies are set,
AbsoluteOrientation. they are saved in the XML document and the inp docu-
\Marker ID = ‘‘1’’. ment, which will be invoked and computed by the solver.
\Position.0, 0, 0\/Position.
\Orientation.0, 0, 0\/Orientation.
.. MSTMM solver module of MSTMMSim
\/Marker. MSTMM solver module is developed based on
\Marker ID = ‘‘2’’. MSTMM using C + + language, where the transfer
\Position.0.015, 0.01, 0.01\/Position. equations and transfer matrices of each element are
\Orientation.0, 0, 0\/Orientation. generated automatically based on the corresponding
.. dynamics model and its topology figure by calling the
\/Marker. transfer matrix library.6–8 The overall transfer equation
\MassCenter. and overall transfer matrix of the system are generated
\Position.0.0075, 0.005, 0.005\/Position. automatically by following the automatic deduction
\Orientation.0, 0, 0\/Orientation. theorem of overall transfer equation of multibody
.. system.7 Furthermore, the computation of system
\/MassCenter. dynamics can be implemented using the algorithm of
\Density.7900\/Density. MSTMM.8 The automatic deduction theorem of over-
\Mass.0.01185\/Mass. all transfer equation is based on tree systems. As for
\JI.7.9e207, 1.28375e206, 1.28375e206, multibody systems which contain closed loops, the
24.44375e207, 24.44375e207, 22.9625e207\/JI. technique of auto selecting the cut hinges is established
.. by introducing the correlation matrix and the dynamic
\/RigidBody. connectivity matrix that depict the connecting state of
As for the inhomogeneous bodies, as well as beams elements. For a multibody system containing nb body
and shells, the finite element scheme is utilized where elements and nh hinge elements, bodies are numbered
each finite element is regarded as homogeneity. The as 1, 2, . . . , nb , while hinges are numbered as
parameters of the body are first specified in each ele- nb + 1, nb + 2, . . . , nb + nh . The correlation matrix of
ment, and then the strategies of finite element method the bodies and hinges is defined as
Rui et al. 5

Figure 3. Finite element model of the beam.

dynamic construction process starting from an empty

system in the following:

1. Set k = 0; treat the system as an empty system

with no bodies and hinges, so that the connec-
tivity matrix is Sk = Onb 3 nb .
2. Set k : = k + 1 according to the ascending
order of the first line in J.
Figure 4. Connection point on the beam. 3. The system will be augmented by adding hinge

nb + k and its connected bodies B+ k and Bk .
Herein, a set Ak , which consists of bodies in
2 3 connected status corresponding to the introduc-
nb + 1 nb + 2    nb + k  nb + nh
J = 4 B+1 B+
2  B+
k  B+
nh
5 tion of hinge nb + k, is defined. Apparently, B+ k

B1 
B2  B
k  
Bnh and B k are connected according to equation (1);

also, B+k (or Bk ) may be already connected with
ð1Þ 
other bodies in a set form fB+ k, m g (or fBk, n g)
where the first line denotes the sequence number of in the present system. Sk is changed

hinges, and the second and third lines indicate the with Si, j = 1 (i, j 2 Ak ), where Ak = fB+ k , Bk g[
+  + 
sequence numbers of bodies connected to the hinge fBk, m g [ fBk, n g, and fBk, m g (or fBk, n g) is deter-

listed in the first line. If a hinge is directly linked to a mined by the column that B+ k (or Bk ) lies in
boundary, the corresponding number in second or Sk1 .
third line will be ‘‘0.’’ The matrix J can determine the 4. If Sk = Sk1 , it indicates that introducing hinge
topology of the system completely. Another matrix S nb + k results in a closed loop since it will not
with order (nb 3 nb ) called dynamic connectivity matrix change the connected status. By recording the
is introduced in the following to represent the connec- sequence number nb + k of this hinge, the closed
tion status of bodies. If there exists at least one path, loop is found. Furthermore, the closed loop can
which is made up of bodies and hinges between body be eliminated by cutting the one of the connec-
Bi and Bj , then Bi and Bj are called ‘‘connected.’’ The tion points.
5. Terminate if k = nh ; otherwise, go to step 2.
elements in S are defined as

1 Bi and Bj are conneted or i = j It should be noted that in MSTMM, boundaries will
Si, j = ð2Þ not cause closed loops, so if B+ 
0 Bi and Bj are not conneted k = 0 or Bk = 0, steps 3
and 4 should be skipped, and one should continue with
For example, if S2, 3 = 1, body 2 and body 3 are step 5. Then, a general multibody system will be con-
called ‘‘connected’’ with other bodies or hinges in verted into a derived tree multibody system by eliminat-
between. The system topology can be regarded as a ing the closed loops.
6 Advances in Mechanical Engineering

‘‘input = 0’’ (the input end of the system), one can

search the ‘‘output’’ of the element and consequently
find out its adjacent element connected to its output
end. The above procedure can be repeated until the ele-
ment whose ‘‘output = 0,’’ indicating the output end of
the system is reached. The sub-matrix of the main
transfer equation7 corresponding to the input end of
the system can be obtained by multiplying the transfer
matrix of the elements in the search path. Analogously,
starting from the element whose ‘‘input = 0,’’ one can
search its adjacent element connected to its output end.
This procedure can be repeated until the element with
multiple input ends is found. Then, the sub-matrix of
the geometric equation can be obtained by multiplying
the transfer matrix of the elements in the search path
and pre-multiplying the geometric coefficient matrix.
By adding the sub-matrices of the main transfer equa-
tion and the geometric equation to the block in the
overall transfer matrix corresponding to the boundary
state vector, the overall transfer matrix of the system
Figure 5. Data structure of a tree system. can be obtained, and the automatic deduction of the
overall transfer equation is realized in the computer.

Herein, object-oriented programming method is used

Post-processor module of MSTMMSim
to build the base class of the element, where the body
element class and the hinge element class are derived. The post-processor module is developed using OCC,
Furthermore, element objects are defined using these Qt, and Qwt, where the computational result datum
element classes. The data structure of a tree system from MSTMM solver is stored for each interested
stored in a computer is shown in Figure 5, where the body, including their positions, velocities, accelerations,
‘‘ID’’ represents the sequence number of the element; angles, angular velocities, angular accelerations, inter-
the ‘‘input’’ (or ‘‘output’’) represents the sequence num- nal torques, and internal forces. The data outputs are
ber of an element or boundary connected to its input achieved by means of animations, curves, numerical
end (or output end); the ‘‘data’’ stores all the para- values, pictures, and tabulations.
meters of the element; and the ‘‘TM’’ denotes the trans-
fer matrix of the element. According to the definition of Basic principle and steps of simulation and
the topology of the dynamics model of a multibody sys-
design of MSTMM
tem in the context of MSTMM, it can be found that
Basic principle of simulation and design of
1. Each element has only one ‘‘output’’ and multi- mechanical system dynamics
ple ‘‘inputs’’; The design of mechanical system dynamics is an itera-
2. The element whose ‘‘output’’ equals to zero is tive process starting from a functional model of the
the root of the system, and the connection point mechanical system. It repeatedly predicts, evaluates the
on the element corresponding to ‘‘output’’ is the current overall design scheme, and compares with the
output end of the system; previous one to acquire a design scheme that better
3. The connection point on the element corre- satisfies the performance requirement of a mechanical
sponding to ‘‘input = 0’’ is the input end of the system. It includes the following procedures in
system. sequence: design of functional model, design of 3D
solid model, design of dynamics model, design of the
The automatic deduction of the overall transfer mathematic model, simulation and optimization of the
equation of a multibody system has already been multibody system dynamics, virtual test, formation of
studied in Rui et al.7 The realization of such automa- the scheme, and experimental verification. By compar-
tism in a computer lies in finding the transfer paths ing with the experimental results, the final design
from tip elements to root element, and to bifurcated scheme can be formed if it satisfies the performance
elements. Herein, we illustrate such process according requirement; otherwise, the design parameters will be
to Figure 5 as follows. Starting from the element whose repeatedly optimized until the requirement is satisfied.
Rui et al. 7

Figure 6. Flow chart of the overall design process of mechanical system dynamics.

Figure 6 shows the flow chart of the overall design pro-

cess of a mechanical system dynamics. The proposed
method in this article has been applied to the dynamics
designs of many practical complex mechanical systems.

Functional model design of a mechanical system

The design of the functional model is to figure out the
Figure 7. Functional model of tank.
shape and dimensions of major parts or subsystems
that have relative motion among a mechanical system.
The functional model of mechanical system is gener-
ated through the 3D modeling module embedded in
MSTMMSim. This process is mainly based on a
designer’s imagination or existing drawings. Upon
designing a mechanical system, the designer will first
identify the basic function of the system according to
practical requirement or specifications, which provides
basic dimensions for the follow-up detailed 3D solid
models. As shown in Figure 7, the functional model of Figure 8. 3D solid model of tank.
a tank is designed using MSTMMSim. The tank should
have the function including marching, turning around, physical structural parameters for dynamics computa-
firing, and adjusting of the pitch and yaw angles. tion such as the mass, moment of inertia, and position
According to these functional requirements, designer of mass center based on its functional model. The 3D
designs the approximate shape and dimension of the solid model can be built directly by the 3D modeling
tank, the number of road wheels, and figures out the module embedded in MSTMMSim or imported from
shape of the caterpillar band, turret, elevating part, other commercial 3D modeling software such as UG
recoil part, and so forth. and Pro/E. In MSTMMSim, the 3D solid model of a
mechanical system can be generated by calling the
modeling module of OCC and displayed by the OCC
3D solid model design of a mechanical system Viewer module, where the parametric modeling tech-
The design of the 3D solid model is to precisely depict nique is utilized. Figure 8 shows the 3D solid model of
the 3D entity of a mechanical system and to design the the tank established using MSTMMSim according to
8 Advances in Mechanical Engineering

position, rotation angle, velocity, angular velocity,

acceleration, angular acceleration, internal forces, and
torques between two bodies. Typical hinges include
smooth ball-and-socket hinges, pin hinge, translational
hinges, elastic hinges, and dampers.6–8
Dynamics model of the multibody system and its
topology figure of the tank are automatically gener-
Figure 9. Dynamics model of tank. ated, as shown in Figures 9 and 10, respectively. The
topology figure describes the transfer relationship
among the state vectors of adjacent elements and the
its functional model. Via human–computer interaction, transfer directions in a system, where a circle ‘‘o’’
the assembling relations of recoil part and elevating denotes a body element and an arrow ‘‘!’’ denotes a
part, elevating part and revolving part, revolving part hinge element and the transfer direction of state vec-
and hull, hull and suspension, caterpillar bands and tors. The bodies and hinges are numbered uniformly.
road wheels, and caterpillar bands and the ground are In Figure 10, the recoil part of the tank (containing
defined. gun barrel, gun breech, muzzle brake, etc.) is numbered
1, the gun cradle is numbered 3, the turret is numbered
5, the tank hull is numbered 7, the balance elbows are
Dynamics model design and topology figure of a
numbered 32–37 and 38–43, the drive sprockets are
mechanical system numbered 56 and 68, the road wheels are numbered
The design of the dynamics model is to establish the 57–62 and 69–74, the support wheels are numbered 63–
dynamics model and its topology figure of multibody 66 and 75–78, while the idlers are numbered 67 and 79.
system automatically according to the 3D solid model There are totally 164 creeper treads with 82 creeper
of a mechanical system. By defining the coordinates of treads lying on each side, respectively. The creeper
connection points of body elements and calling the treads on the right side of the tank in its forward
defined hinge elements from the transfer matrix library, motion are numbered 80, 82, 84, ., 242, respectively,
the dynamics model will be automatically generated via while the left 244, 246, 248, ., 406. The track link pins
the 3D solid model. By exporting the data of the multi- are considered as hinges. The hinges corresponding to
body system in the form of Graphviz dot language and the track link pins on the right side of the tank are
calling the Graphviz software, the topology figure of numbered 81, 83, 85, ., 243, respectively, while the left
multibody system will be automatically generated 245, 247, 249, ., 407. Through human–computer
according to the symbol conventions of MSTMM and interaction, the dynamics properties of each element in
automatic deduction theorem of overall transfer equa- the tank can be set. The mass, moment of inertia, posi-
tion of multibody system.7 A multibody system is com- tion of mass center, and so on, can be automatically
posed of bodies and hinges, where bodies include generated by the MSTMMSim. Meanwhile, users
lumped masses, rigid bodies, elastic bodies, and so should also set the dynamics parameters of each hinge
forth, while hinges indicate the general relationship of element including stiffness and damping. Furthermore,

Figure 10. Topology figure of tank dynamics model.

Rui et al. 9

the connection parameters of caterpillar band and road multiplication of transfer matrices of elements in the
wheels, and caterpillar band and the ground, should transfer path from the tip to root. The order of system
also be set. matrix in MSTMM is very low and equals to the order
of transfer matrix of a rigid body, thus resulting in very
high computational speed. Comparisons on computa-
Basic idea of MSTMM tional precision and speed of MSTMM with other
The basic idea of MSTMM is to decompose a multi- dynamics methods can be found in Rui et al.6–8
body system into elements containing bodies and hinges
whose dynamics properties can be readily expressed in
matrix form. These matrices of elements are considered Automated formation of dynamics equations
as building blocks that provide the dynamics properties Seen from equations (5) and (6), the overall transfer
of the entire system when assembling them together equation can be deduced automatically. The state vec-
according to the topology of the dynamics model of tors involved in an overall transfer equation are the col-
multibody system. Especially, bodies and hinges are umn matrix comprising the state vectors of all
considered the equivalent position in transfer equations boundary ends of the system. The overall transfer
and transfer matrices. matrix U1n of a chain system is the successive multipli-
For a chain multibody system undergoing spatial cation of transfer matrices of every element of the sys-
motion, the state vector of a connection point between tem along the transfer path from the tip to the root. In
any two elements is defined as fact, one can deduce automatically the overall transfer
equation for a general system according to the topol-
z= ogy figure of dynamics model and the automatic
 T
_x
€x €y €z V _y
V _z
V mx my mz qx qy qz 1 deduction theorem of overall transfer equation of mul-
tibody system applying MSTMM.7 Particularly, the
ð3Þ
transfer equations and the library for most elements
where r€ = ½ €x €y €z T , _ =½V
V _x V _y V_ z T , are also explicitly set and can be found in Rui et al.6–8
T T The transfer equation and transfer matrix of the
m = ½ mx my mz  , and q = ½ qx qy qz  are the
column matrices of absolute accelerations, absolute tank element will be formed automatically by calling
angular acceleration, internal torques, and internal the transfer matrix library installed in the MSTMMSim
forces in the right-handed Cartesian inertial coordinate shown in Rui et al.6–8 The overall transfer equation of
system, respectively. the tank like equation (7) will be formed automatically
For an element numbered j, it can be proved that its using automatic deduction theorem of overall transfer
transfer equation keeps the form8 equation of multibody system7 as follows

zj, O = Uj zj, I ð4Þ Uall zall = 0 ð7Þ

where zj, O and zj, I are the state vectors at the outboard where
end and inboard end of the element, respectively; Uj is 2 3
--------------------

the element’s transfer matrix. I T156 T157    T179

6------------------------------------7
According to kinematics and Newton’s third law, 6 O G9, 56 G9, 57    G9, 79 7
and the specific sign conventions,7 the state vectors of 6 7
6 7
Uall = 6 O G10, 56 G10, 57    G10, 79 7 ð8Þ
element j’s output end coincide with that of k’s input 6 . .. .. .. .. 7
6 . 7
end, where k is the outboard element of j. This allows 4 . . . . . 5
to substitute input states zk, I by output states zj, O and O G31, 56 G31, 57    G31, 79
the associated transfer equation (4), that is,
zk, O = Uk zk, I [Uk zj, O = Uk Uj zj, I , which finally leads to zall = ½ zT1, 0 zT56, 0 zT57, 0  zT58, 0 T ð9Þ
products of element transfer matrices. Consequently, 8
the overall transfer equation and overall transfer > T156 = U1 U2 U3 U4 U6 U7 U8 U56
>
< T157 = U1 U2 U3 U4 U6 U7 U9 U32 U44 U57
matrix of chain system take the form
.. ð10Þ
>
> .
:
zn, 0 = U1n z1, 0 ð5Þ T179 = U1 U2 U3 U4 U6 U7 U31 U79
U1n = Un Un1    U2 U1 ð6Þ 8
>
> G9, 56 = H9, 8 U56
>
<G =H U U U
where zn, 0 and z1, 0 are the boundary state vectors of the 9, 57 9, 9 32 44 57
.. ð11Þ
system, and U1n is the overall transfer matrix. >
> .
>
:
Seen from equation (6), for a chain system, the G9, 79 = H9, 31 U79
overall transfer matrix equals to the successive
10 Advances in Mechanical Engineering

Figure 11. Interface of animation display of simulation of tank dynamics.

Figure 12. Interface of animation display of simulation of the tank dynamics.

Rui et al. 11

Figure 14. Comparison on time cost by the new version of

MSTMM and Lagrange method.

Under the above optimal scheme, Figure 12 shows

the visualized interface of the tank system dynamics
during marching, and Figure 13 shows the interface of
the time history of velocity and acceleration of the mass
Figure 13. Interface of simulation results of velocity and center of the hull of the tank, respectively.
acceleration of the tank in the vertical direction: (a) velocity and Furthermore, we consider a typical system to
(b) acceleration. demonstrate the computational speed of the proposed
method and the software. A planar pendulum system is
8 formed by N (0\N  1000) homogeneous rods with
>
> G10, 56 = H10, 8 U56 length 1 m and mass 1 kg, which are connected by
>G
< 10, 57 = H10, 9 U32 U44 U57 smooth hinges. The CPU used is an i7-3610QM proces-
.. ð12Þ sor with maximum frequency of 3.3 GHz. Both the new
>
> .
>
: version of MSTMM and Lagrange equations are com-
G10, 79 = H10, 31 U79
8 puted using a single thread. According to Figure 14, the
>
> G31, 56 = H31, 8 U56 time consumption of the new version of MSTMM is
>
<G =H U U U
31, 57 31, 9 32 44 57 much less than that of the Lagrange method. Especially
.. ð13Þ for 1000 bodies, the time cost of the new version of
>
> .
>
: MSTMM is 52 s, while that of the Lagrange method is
G31, 79 = H10, 31 U79
39,224 s, which demonstrates the high computational
speed of the new version of MSTMM.
Simulation and optimization
The MSTMMSim will call the MSTMM solver to com- Conclusion
pute and optimize the tank system dynamics. An
unconstrained optimization problem for tank multi- Concentrating on the speedy simulation of multibody
body system dynamics is set up as follows: the root system dynamics, and by combing computer graphics
mean square (RMS) value of the vertical acceleration technology and OSS, this article puts forward the new
of the mass center of hull is selected as objective func- method and software MSTMMSim for visualized simu-
tion f(x), while the damping coefficient of the torsional lation and design of mechanical system dynamics based
stabilizer is chosen as a single design variable x with on MSTMM. A tank and a pendulum system are taken
bounds [0, 3 3 104] Nm/rad. The Brent’s optimization as two practical examples to verify that the simulation
method which combines the reliability of the golden- and design of complex mechanical system dynamics
section search with the speed of parabolic interpolation own the following features if using MSTMMSim: very
is adopted. As shown in Figure 11, the minimum of the high computational speed, and very high visualization
objective function f(x) is found with 25 iterations, and programming. It provides powerful tool for the
where the optimal value of damping coefficient of the simulation and design of complex mechanical system
torsional stabilizer is 4344 Nm/rad. dynamics.
12 Advances in Mechanical Engineering

Acknowledgements 2. Wittenburg J. Dynamics of systems of rigid bodies. Stutt-

The authors own special thanks to Prof. Dieter Bestle from gart: B.G. Teubner, 1977.
Brandenburgische Technische Universität, Cottbus, 3. Schiehlen W. Multibody systems handbook. Berlin:
Germany, who offered important suggestions for this paper Springer-Verlag, 1990.
in his academic visit to Nanjing University of Science and 4. Shabana AA. Dynamics of multibody systems. Cam-
Technology. bridge: Cambridge University Press, 2009.
5. Bauchau OA. Flexible multibody dynamics. Dordrecht:
Springer Science & Business Media, 2011.
Declaration of conflicting interests 6. Rui XT, Yun LF, Lu YQ, et al. Transfer matrix method
The author(s) declared no potential conflicts of interest with for multibody system and its application. Beijing, China:
respect to the research, authorship, and/or publication of this Science Press, 2008.
article. 7. Rui XT, Zhang JS and Zhou QB. Automatic deduction
theorem of overall transfer equation of multibody sys-
tem. Adv Mech Eng 2014; 2014: 378047-1–378047-12.
Funding 8. Rui XT, Bestle D, Zhang JS, et al. A new version of
The author(s) disclosed receipt of the following financial sup- transfer matrix method for multibody systems. Multibody
port for the research, authorship, and/or publication of this Syst Dyn 2016; 38: 137–156.
article: This research was supported by the Research Fund 9. Object libraries application framework user guide, http://
for the Doctoral Program of High Education of China (no. www.OpenCASCADE.org
20133219110037) and the Basic Project for Defense Base of 10. Qt software reference documentation, http://www.qt.io
China (no. JCKY 2013620133008). 11. Yang HG, Rui XT, Zhan ZH, et al. Virtual design soft-
ware for mechanical system dynamics using transfer
matrix method of multibody system and its application.
References Adv Mech Eng 2015; 7: 1–24.
1. Rui XT, Wang GP, Gu JJ, et al. Study on system design
method of firing precision for multiple rocket launcher. J
Mech Eng 2015; 52: 164–177.