Professional Documents
Culture Documents
3, June 2014
for simulating multi-body dynamics [3]. Similar to ADAMS, III. DEVELOPMENT OF KERNEL OF DYNAMIC ANALYSIS FOR
ODE derives equations of motion for multi-body systems MULTI-BODY SYSTEM
using augmented formulation. ODE can treat only rigid The kernel of dynamic analysis is developed for the
bodies, however, not flexible bodies. Moreover, ODE cannot multi-body system. In this section, the recursive formulation
handle hydrostatic and hydrodynamic forces. applied for the construction of kinematic equation of
RecurDyn [4] is a three-dimensional simulation software multi-body system is explained.
that combines dynamic response analysis and finite element
analysis tools for multi-body systems. It is 2 to 20 times faster A. Forward and Inverse Dynamics
than other dynamic solutions because of its advanced fully The dynamics of a rigid-body system are described by its
recursive formulation. Various joints and external forces can equation of motion, which specifies the relationship between
also be applied to the multi-body systems, but RecurDyn the forces that act on the system and the accelerations they
cannot handle hydrostatic and hydrodynamic forces. produce. The main concern in this section is the algorithms
This study presents the development of a dynamics kernel for the following two particular calculations.
for the dynamic analysis of offshore structures such as Forward dynamics: The calculation of the acceleration
semi-submersible drilling rig or drill ship. The equations of response of a given rigid-body system to a given applied
motion for multi-body systems were derived using recursive force
formulation. Moreover, the external force calculation module Inverse dynamics: The calculation of the force that must
can generate hydrostatic force by considering the nonlinear be applied to a given rigid-body system to produce a
effects and the linearized hydrodynamic force as external given acceleration response
forces. Forward dynamics is used mainly in simulation. Inverse
It is possible to interface these general programs with dynamics will be explained first, however, since it is easier
hydrodynamics using user-subroutine. For example, than forward dynamics in the manner of the explanation of
Jonkman developed a module called HydroDyn for the recursive formulation [6].
calculating hydrodynamic force in time domain and interface
B. Inverse Dynamics of Recursive Formulation
it with other program [5]. For calculating hydrodynamic
force in time domain, Jonkman transforms the analysis The equations of motion for each body of the multi-body
results from WAMIT, which is a commercial program for system based on recursive formulation can be summarized as
calculating hydrodynamic force in frequency domain, into following.
time domain using Cummins equation. In other words, the
analysis results from WAMT are required to use v i i Xi 1 v i 1 Si qi (1)
Hydrodyn. In this research, to make the developed kernel
independent from any other programs, 3D Rankine panel a i i Xi 1 a i 1 Si qi S i qi v i Si qi (2)
method is applied for direct calculation of hydrodynamic
force in time domain. To interface 3D Rankine panel method fiB I i a i v i * I i v i (3)
with the existing programs of dynamic response analysis for
multi-body, input of hydrodynamic force is not enough. To
fi fiB i X*i 1 fi 1 fiext (4)
increase numerical stability, it is required to modify the mass
and inertia of the floating body using added mass. Therefore
it is not easy to interface with other commercial programs. i SiT fi - (5)
Moreover, as mentioned above, to make the developed kernel
independent from any other programs, the functions for In here, v i is velocity vector(6 elements) of body , a i is
analyzing dynamic response of multi-body and calculating acceleration vector(6 elements) of body , qi is generalized
hydrostatic and dynamic force are integrated together. Table
coordinate(joint value), S i is velocity transformation matrix,
I shows the features of the different dynamics kernels that
I i is mass and mass moment of inertia of body , fi is
were compared in this study. B
TABLE I: COMPARISON OF THE FEATURES OF THE DEVELOPED DYNAMICS resultant force exerted on body , fiext is external force
KERNEL IN THIS STUDY WITH COMMERCIAL DYNAMICS KERNEL
This study ADAMS ODE RecurDyn exerted on body , fi is force across the joint , and i is
Multi-body Recursive Augmented Augmented Recursive
formulation formulation formulation formulation formulation force generated by joint (generalized force). In inverse
Various dynamics, since the positions qi , velocities qi , and
joints
accelerations qi of generalized coordinates are given, the
Flexible
X X
body velocities v i and accelerations a i of each body can be
Hydrostatic
computed. Furthermore, the forces fi and the generalized
force
Linearized
hydrodyna
forces i , which should be exerted on each link, can be also
mic force computed in a recursive fashion [6], [7].
*(: Supported; : Can be only interfaced by the developer of the dynamics For example, the equations of motion for the three-link
kernel) multi-body system can be formulated as shown in Fig. 2.
279
International Journal of Computer Theory and Engineering, Vol. 6, No. 3, June 2014
f B I a v * I v
1 1 1 1 1 1 (1-c)
fiB I i a i v i * I i v i fiext (8)
f1 f1B 1 X*2 f2 f1ext (1-d)
fi fiB i X*i 1 fi 1
S T f
1 1 1
(1-e) q3
Equations of motion for link 2
(9)
v 2 X1 v 1 S 2 q2
i
(2-a) 3
a 2 2 X1 a 1 S 2 q2 S 2 q2 v 2 S 2 q2 (2-b) 4) Because fiext is given, v i * I i v i fiext can also be
f2B I 2 a 2 v 2 * I 2 v 2 (2-c)
f f B 2 X* f f ext (2-d)
q2 determined without considering the other equations.
2 2 3 3 2
pi v i * I i v i fiext .. (10)
(3-b) B ase
xn Fine: 1 , 2 , 3
E
f B I a v * I v
3 3 3 3 3 3 (3-c)
f3 f X f4 f
3
B 3 *
4
ext
3 (3-d)
3 S3T f3 (3-e)
By substituting (6), (8) ~ (10) into (2) ~ (5), (11) ~ (14) are
Fig. 2. Example of an inverse dynamics problem: three-link multi-body
system. derived.
a i i Xi 1 a i 1 Si qi ci (11)
1) The velocity of link 1, v 1 , can be determined using
(1-a), since the velocity of the base, v 0 , is zero. Then the fiB I i a i pi (12)
velocities of the other bodies can be determined using
(2-a) and (3-a). fi fiB i X*i 1 fi 1 (13)
2) The acceleration of link 1, a1 , can be determined using
(1-b), since the acceleration of the base, a 0 , is zero. i SiT fi (14)
Then the accelerations of the other bodies can be (11) ~ (14) can be rearranged into (15) ~ (18), and they are
determined using (2-b) and (3-b). the equations of motion for the forward dynamics of the
3) Because the velocities and acceleration are determined, multi-body system based on recursive formulation.
the resultant forces can be determined using (1-c), (2-c),
and (3-c). a i i Xi 1 a i 1 Si qi ci (15)
4) Because link 4 does not exist, the force, f4 , exerted on
S I X
1
qi STi I iASi i
T
i
A
i
i
a ci p iA
i 1 i 1
(16)
link 4 by link 3 is zero. Therefore, the force, f3 , exerted
1
on link 3 by link 2 can be determined using (3-d). Then I iA I i i X*i 1 I iA1 i 1 Xi i X*i 1 I iA1 Si 1 STi 1I iA1Si 1 STi 1I iA1 i 1 Xi (17)
determined by suppressing the constraint forces from Based on the equations of motion for the forward and
forces f , f , and f . 1 2 3
inverse dynamics, the kernel of dynamic analysis for
multi-body system is developed.
C. Forward Dynamics of Recursive Formulation
For the simple explanation of inverse dynamics, (1) ~ (5)
will be expressed in a more compact form through the IV. VERIFICATION OF THE DEVELOPED KERNEL
following steps. In this section, the developed kernel is verified before the
1) Because qi is given and velocities v i can be dynamic response analysis of the multi-body system is
determined in the recursive fashion using only (1), v i performed by comparing it with other commercial kernels for
will be considered as given. two test models.
2) When the velocities v i are considered a given, the last
C .O .G . of B ody 1: 1kg
term of (2), S q v S q , can be determined
i i i i i
without considering the other equations. Therefore, it Revolute Joint2
x Joint3
Revolute Joint3
3) As shown in (7), when the new notation fiB ' denotes qq1 B od y3
z 1
fiB fiext , (3) and (4) are rewritten as (8) and (9).
Fig. 3. Multi-body system composed of three bodies, three revolute joints,
one cylindrical joint and one closed loop.
280
International Journal of Computer Theory and Engineering, Vol. 6, No. 3, June 2014
The first test model is a multi-body system composed of of motion can be used for dynamic response analysis after the
three bodies, three revolute joints, one cylindrical joint, and input of external force. The external forces considered in this
one closed loop, as shown in Fig. 3. simulation are hydrostatic force with non-linear effect,
Body 1 is attached to the base by revolute joint 1, and body linearized hydrodynamic force, gravitational force, and wind
2 is connected to body 1 by revolute joint 2. Body 3 is force.
attached to body 2 with revolute joint 3, and moves
A. Hydrostatic Force
perpendicular to the x axis due to cylindrical joint 1.
The hydrostatic force applied to floating platform is
Position Velocity Acceleration calculated while considering its instantaneous position and is
expressed as seen in (19) [9].
Calculation
result from
reference e
f Hydrostatic SW g[0 ;0 ; dV ; yBuoyancy dV ; xBuoyancy dV ;0 ]T (19)
V V V
[rad/s] [rad/s] [rad/s2]
15.0
Result 3.5
3.0
4.0
10.0
calculated 2.5 2.0
5.0
using
2.0
1.5
q1
q2
0.0 0.0
q1_dot
-5.0
q1_twodot
q2_twodot
In here, SW is density of the sea water, g is gravitational
developed 1.0
0.5
-2.0
q2_dot
-10.0
kernel 0.0
-4.0 -15.0
acceleration, V is submerged volume of the floating body,
0.0 0.5 1.0 t[sec] t[sec] 0 0.5 1
t[sec]
0 0.5 1
Fig. 4. Results of the simulation of the position, velocity, and acceleration of and xBuoyancy , yBuoyancy are represent the coordinates of
generalized coordinates q1 and q2 compared with those yielded by the
reference. center of buoyancy, respectively.
calculated 3.50E-01
3.00E-01
1.00E+02
5.00E+01
1.20E+01
1.00E+01
8.00E+02
using
2.50E-01 6.00E+02
0.00E+00 8.00E+00
2.00E-01 0 0.005 0.01 0.015 0.02 0.025 0.03 0.035
-5.00E+01 6.00E+00 4.00E+02
developed
1.50E-01
1.00E-01
5.00E-02
-1.00E+02
-1.50E+02
4.00E+00
2.00E+00
2.00E+02
wave.
kernel 0.00E+00
0 0.005 0.01 0.015 0.02 0.025 0.03 0.035
-2.00E+02
0.00E+00
0.00E+00
0 0.005 0.01 0.015 0.02 0.025 0.03 0.035
0 0.005 0.01 0.015 0.02 0.025 0.03 0.035
281
International Journal of Computer Theory and Engineering, Vol. 6, No. 3, June 2014
potential, and represents radiation coefficient. Fig. 7. The lug arrangement for lunching process of the ship using two
Problems defined as above can be solved as follows. floating cranes.
G d G d
d d n G n dS SF d n G n dS
S Body
(29)
(3) (4)
(3) (4)
(5) (5)
1 1
G ( x, x)
4 x x
(30)
(5)
Fig. 8. The
(5)
result of the simulation for the lunching process of the ship using
In here, x is position vector of 3D Rankine source, x' is two floating crane.
position vector of panels, Sbody is body surface, and SF is free
surface. Fig. 9 shows the result of simulation that is the tension of
To increase the numerical stability of this procedure, when two wire ropes numbered H3, H4 among total of 8 wire
ropes. The tension changes according to the stage of the
we calculate acceleration of the ship at time (t+dt), a(t dt ) ,
launching procedure, and we could confirm that the dynamic
m add a(t ) is considered as the additional external force loads are 10 ~ 15% larger than the static loads.
with the hydrodynamic force at time (t+1), and madd is added
to real mass of the floating body. This is the reason why it is
not very simple to apply the 3D Rankine panel method to the
commercial program for multi-body response analysis. In
here, madd is added mass and it is also calculated using 3D
Rankine panel method.
282
International Journal of Computer Theory and Engineering, Vol. 6, No. 3, June 2014
(to n )
simulation and measured data does not exceed 10%. 180
160
This 10% difference is mainly caused by uncertainty of 140
120
synchronization between the cranes, and we are planning to 100
Tens
consider this uncertainty as a factor of multi-crane 80
60
simulation. 40
20
0
0 200 400 600 800 1000 1200 1400 1600
Time
H o istin g H o istin g B lo ck B lo ck
-U p -U p Tu rn -o ver Tu rn -o ver
VII. SIMULATION FOR BLOCK TURNOVER USING TWO starts en d s starts en d s
H o istin g H o istin g
Tran sp o rtatio n o f Tran sp o rtatio n o f
GOLIATH CRANES th e b lo ck starts th e b lo ck en d s -D o w n -D o w n
starts en d s
Fig. 1 (a) shows the block lifting using 2 Goliath cranes. Fig. 12. Tension of the wire rope calculated from the simulation for the block
The weight of the block is 830ton which exceeds the turn-over process.
capability of Goliath crane 600ton. Therefore, to move and
turnover the block into the dock, two Goliath cranes is being
used, and this process is being simulated.
Fig. 13. Maximum static and dynamic tension of the wire rope calculated
from the simulation.
Fig. 14. the results of structural analysis using the maximum static and
dynamic loads.
283
International Journal of Computer Theory and Engineering, Vol. 6, No. 3, June 2014
to the structural analysis for evaluate the dynamic effects on [7] J. Y. S. Luh, On-line computational scheme for mechanical
manipulators, Journal of Dynamic Systems, Measurement, and
the block. Control, vol. 102, pp. 69-76, 1980.
In further research, this simulation program will be applied [8] E. J. Haug, Intermediate Dynamics, Prentice-hall, 1992.
to other various simulations in order to enhance its liability. [9] K. Y. Lee, J. H. Cha, and K. P. Park, "Dynamic response of a floating
crane in waves by considering the nonlinear effect of hydrostatic
Moreover, other several functions, for example, the
force," Ship Technology Research, vol. 57, no. 1, pp. 62-71, 2010.
equalizer, the module for calculating contact force between [10] D. C. Kring, Time domain ship motions by a three-dimensional
the wire rope and body, etc., will be developed to make this rankine panel method, MIT, Ph.D. Thesis, 1994.
program can consider more field environment.
Namkug Ku is a research engineer of Education and
Research Center for Creative Offshore Plant Engineers at
ACKNOWLEDGMENT Seoul National University, Korea. He holds a B.S., M.S.,
and Ph.D. in naval architecture and ocean engineering
This study was partially supported by the Industrial from Seoul National University. His main area of research
Strategic Technology Development Program (10035331, interests include ship and offshore plant design,
Simulation-based Manufacturing Technology for Ships and multi-body dynamics, top-side process engineering of
offshore plant, and so on.
Offshore Plants) funded by the Ministry of Knowledge
Economy (MKE) of the Republic of Korea and by the Brain Sol Ha is a research engineer of Engineering Research
Korea 21 plus program of the Marine Technology Education Institute at Seoul National University, Korea. He holds a
and Research Center, the Research Institute of Marine B.S., M.S., and Ph.D. in naval architecture and ocean
engineering from Seoul National University. His main
Systems Engineering, and the Engineering Research Institute area of research interests include ship and offshore plant
of Seoul National University. design, methodology of modeling and simulation, risk and
reliability analysis, GPU-accelerated parallel processing,
fluid analysis based on lattice Boltzmann method, and so
REFERENCES on.
[1] N. Orlandea, M. A. Chace, and D. A. Calahan, A sparsity-oriented
approach to the dynamic analysis and design of mechanical Myung-Il Roh is a professor of naval architecture and
systems-part 1 and 2, Journal of Engineering for Industry, ocean engineering at Seoul National University, Korea.
Transactions of the ASME, vol. 99, no. 3, pp. 773-779, 1977. He holds a B.S., M.S., and Ph.D. in naval architecture and
[2] W. Schiehlen, Multibody Systems Handbook, Springer, pp. 361-402, ocean engineering from Seoul National University. His
1990. main area of teaching and research interests include ship
[3] R. Smith, Open dynamics Engine v0.5 User Guide, pp. 15-20, 2006. and offshore plant design, simulation-based design and
[4] FunctionBay, Inc., RecurDyn V7R5 Release Notes, 2011. production, optimization, CAD, CAM, CAE, CAPP, and
[5] J. M. Jonkman, Dynamics of offshore floating wind turbines - model so on.
development and verification, Wind Energy, 2009.
[6] R. Featherstone, Rigid body Dynamics, Springer, 2008.
284