International Review of Mechanical Engineering (I.RE.M.E.), Vol. xx, n.

Dynamic Modeling of a Generalized Stewart Platform by Bond Graph

Method Utilizing a Novel Spatial Visualization Technique
Đbrahim Yıldız1, Vasfi E. Ömürlü2, Ahmet Sağırlı3

Abstract – This paper represents dynamic modeling of generalized stewart platform manipulator
by Bond Graph method with a new spatial visualization method and the state-space representation
of the dynamic equations of the system. Dynamic model includes all the dynamics and gravity
effects, linear motor dynamics as well as the viscous friction at the joints. Following modeling of
actuation system and of main structure, unification of these two is accomplished. Linear DC
motors are utilized and are modeled as the actuation system. Since overall system consists of high
nonlinearity originated from geometric nonlinearity and gyroscopic forces, resultant derivative
causality problem caused by rigidly coupled inertia elements is addressed and consequential
system state-space equations are presented. A spatial visualization method is developed based on
classic vector Bond Graph method, due to the lack of representing overall kinematic bonds of the
stewart platform mechanism using classic vector bond-graph visualization. Copyright © 2008
Praise Worthy Prize S.r.l. - All rights reserved.

Keywords: Stewart Platform, Bond Graph, System Dynamics, Spatial Kinematics, State-Space

Nomenclature I. Introduction
{P} Upper platform coordinate system The dynamic models of parallel mechanisms are
more complicated than serial mechanisms because of
t& Upper platform linear velocities
their closed-form structure. Nevertheless, dynamic
ω Upper platform angular velocities
models are important for the control of parallel
L& Linear velocities of the linear actuators mechanisms. The stewart platform is a six-degree-of-
θ& Angular velocities of the linear actuators freedom mechanism with two plates mounted by six
rp Radius of upper platform linearly actuated legs and has first proposed as a flight
simulator in 1965 by D. Stewart as in [1]. Many
rB Radius of the lower platform applications related to physical simulation of spatial
λi Angle between upper connection points and motion has been put into realization. Present study
x axis of upper platform’s coordinate system mentions obtaining dynamic formulation of a
Λi Angle between lower connection points and generalized stewart platform using Bond Graph method
x axis of lower platform’s coordinate system with a spatial visualization technique.
•r
Velocity of upper connection points Bond-Graph method has been established as a new
S approach to model, analyze and control various
K1..3 Effects of angular velocity of upper platform dynamical systems by Prof. Henry Paynter in 1961, in
to the leg velocities which energy bonds are utilized to define the energy
K 4..6 Effects of linear velocity of upper platform flow between the physical elements of any system. Since
to the leg velocities
[~I ]
it is a graphical approach and resultant dynamic
Virtual inertial matrix equations are in state-space form, the method is straight
[G] Gyristor element forward and ideal for following control design problems.
r Many researchers have worked on Bond Graph
M Lωi Moments generated by actuators modeling of mechanical systems. Reference [2],[3]
r developed Bond Graph techniques for hydraulic,
FLti Forces generated by actuators
mechatronic and thermodynamic systems. Reference [4]
r
M θωi Reaction moment of upper platform acting studied a comprehensive research on Bond Graph
on legs techniques whose results are applicable to the robotics
r Reaction force of upper platform acting on
Fθti and spatially moving objects.
legs Reference [5] developed a Bond Graph technique
r
Mω Gyroscopical moments of upper platform obtaining Lagrange and Hamilton functions by modeling
systems with high degree of nonlinearity. The method is

Manuscript received January 2007, revised January 2007 Copyright © 2008 Praise Worthy Prize S.r.l. - All rights
reserved
 Đ.Yıldız, V.E.Ömürlü, A.Sağırlı

based on defining the dependent velocities of the system Rotation around xP , yP and zP axis; γ ,ϑ,υ
with general coordinates and local coordinates. respectively.
Reference [6] studied on a systematic modeling method
zP
of spatially moving mechanisms with Bond Graph P3 P2

technique. Reference [7] worked on modeling mp

yP
P4 λ
mechanical systems with vector Bond-Graph method. m 3 L3
Ib3
Ip
{P} εp L2
P1
rp
Reference [8] used Lagrange and Hamilton equations xP
m4
for solving the derivative causality problem on Ib 4 L4 P5 P6
m2 m L
Ib 2 I 1 1
zB b1
geometrically nonlinear systems. B3
θ3 L 6 θ2
B2
The inverse dynamics of the stewart platform yB
θ4 L5 m 5 m6 θ1
mechanism is solved using Newton-Euler method by Ib5 Λ Ib 6
B4 {B} εB B1
Reference [9], in which force-moment equalities are rB
θ6
θ5
xB
used to reach the transmitted forces through the joints
B5 B6
and the actuators, assuming symmetrical and thin legs
and frictionless joints. Reference [10] solved the inverse
dynamics of a system including all the dynamical and Fig. 1 A 3x3 Experimental Setup and the Physical Model of a
gravitational effects by using Newton-Euler approach. Generalized stewart platform mechanism
Reference [11] developed motion equations for a
system using Lagrangian equations which utilizes It is assumed that,
kinetic-potential energy equations to find the velocity • Center of gravity of upper platform is located at the
and acceleration characteristics of any dynamic system. origin of the mobile frame.
A similar equation system is developed as in [12] for • Center of gravity of actuators is located about a
control purposes. distance of the 2/3 of the total actuator length from
Reference [13] applied kinematic vector Bond Graph the lower connection point.
method to a telescopic rotary crane, obtained state space • External force and moment effects are neglected.
formulation of this mechanism and solved derivative • Bearing frictions and gravitational forces are present
causality problem by reducing the inertial elements to on the system.
the independent velocity ports using virtual inertia and a The parallel interaction between actuators causes
gyristor element. angular velocities on each leg.
Reference [14],[15] used linear graph technique to
generate the system equations of a stewart platform for
impedance control, yet, the details of the approach and II.2. Kinematics
the developed mathematical model were not provided. Position of upper and lower connection points can
In this study, Bond Graph modeling of a stewart be expressed with the following equations,
[ ]
platform mechanism is addressed including all r
dynamical and gravitational effects. The nonlinear
P
Pi = rp cos(λi ) rp sin( λi ) 0 T (1)
r
model includes linear motor dynamics and all joints B
Bi = [rb cos( Λ i ) rb sin( Λ i ) 0 ]T (2)
have viscous friction. The inverse kinematics equations
are used for modeling of the mechanism. Equation (1) denotes the position vector of upper
connection points with respect to mobile platform
frame {P}. Equation (2) represents the position vector of
II. Mathematical Modeling of a Stewart lower platform connection points with respect to base
Platform Mechanism frame {B}. Upper connection points can be expressed
with respect to the base frame by multiplying (1) with
the rotation matrix,
II.1. Assumptions and Presentation of the Stewart  c (υ )c (ϑ ) c (υ ) s (ϑ ) s (γ ) + s (υ )c (γ ) s (υ ) s (γ ) − c (γ ) s (ϑ )c(υ )
Platform Mechanism R =  − s (υ )c (ϑ ) c (υ )c (γ ) − s (υ ) s (ϑ ) s (γ ) s (υ ) s (ϑ )c (γ ) + c (υ ) s (γ ) (3)
 s (ϑ ) − c (ϑ ) s (γ ) c (ϑ )c (γ ) 
An experimental setup of 3x3 stewart platform and
where c=cos and s=sin are functions.
the physical model of a generalized stewart platform r r
qi = Rpi (4)
mechanism is shown in Fig. 1.
Modeled mechanism has six linear motors as From (3) and (4), leg vectors and leg lengths can be
actuators. Each actuator is attached to the upper expressed by the terms of upper platforms position
platform and the lower platform by spherical and variables.
r r r r
universal joints, respectively. Upper part of the stewart Si = qi + t − bi (5)
platform mechanism performs following motions: r
Li = S i (6)
Linear motion along x P , y P and z P axis; t x , t y , t z
Unit vectors along the legs can be expressed by
respectively. dividing leg vector by the leg length.

Copyright © 2008 Praise Worthy Prize S.r.l. - All rights reserved International Review of Mechanical Engineering, Vol.
xx, n. x
 Đ.Yıldız, V.E.Ömürlü, A.Sağırlı

r r
s =S/L (7) In (13) and (14), the angular velocity of upper
Linear velocity along the legs can be defined as, platform multiplies by the first, second and third
θ
r •r
L
• columns of K i and K i , same as the linear velocity of
L=sS (8)
upper platform multiplies by the fourth, fifth and sixth
where, L θ
•r columns of K i and K i .
r •
S =ω×q +t (9) Kinematic vectorial bond graph model of the
mechanism for one leg which is obtained from (13) and
And also, angular velocity of the actuators can be
(14) is given in Fig. 2. In complete model, each parallel
expressed as,
energy ports have 12 energy bonds and each leg has
•  • 
θ i =  si × S i  Li (10) different translational and rotational velocities. A new
  model is developed to understand the relationship
between the kinematic structure of the system and the
physical model. Fig. 2 shows two main velocity
III. Kinematic Bond Graph Model of a • •
variables of each leg, ( Li ,θ i ), which can be expressed
Stewart Platform Mechanism
with same energy bonds by using new kinematic
Translational and rotational velocities along the approach.
actuators can be expressed by the terms of upper Assuming two distinct energy ports “0” and “1”, two
platform velocities. vectorial effort variable are acting on these energy ports,
ω x  Fig. 3.
ω 
 y Left side of Fig.3 represents common vector bond-
ω z 
[ ]
•
Li = siz qiy − siy qiz six qiz − siz qix siy qix − six qiy six siy siz ⋅  •  (11) graph of two distinct energy bonds. In the right side of
 tx 
•  Fig.3, energy bonds and energy ports are associated by
ty  one energy bond and new complex vector bond-graph is
• 
 t z  established. The right side of Fig.3 represents two
ω x  distinct energy ports and two dependent energy ports.
ω 
 y “1.0” port shows one “1” and one “0” distinct energy
siy qiy + siz qiz − siy qix − siz qix 0 − siz siy  ω 
• 1   •z
θ iy = ⋅  − six qiy six qix + siz qiz − siz qiy siz 0 − six  ⋅
 tx  (12) ports. The energy associated with “a” flows from “0”
Li 
 − six qiz − siy qiz six qix + siy qiy − siy six 0   •  port to “1” port as in the energy related to the “b” flows
ty 
•  from the other “0” port to the other “1” port.
 t z 
[K ]
θi
[K ]
•ω ω ω MTF
Li
1..3

•ω
1..3

θi
MTF 1
• Li •
θi 0 0
Li

[K ]
θi
4..6
[ ] K 4L..i 6 •t
•t MTF 1 MTF Li
• •
θi Fig.4 Binding two distinct energy ports by one port
t t
Fig. 2 Kinematic vector bond-graph model for one leg of the stewart
platform mechanism
• • M
TF
• •
θ5 L5 θ 4 L4

• •
Fig.3 Binding and representing two distinct energy bonds in one • • M θ 3 L3
θ 6 L6
TF
diagram

Equations (11) and (12) can be shown as,

r
• ω
Li  • 
L i = K 1..3 K 4..6  r 
Li
14243  t 
[ ] (13)
•
θ1 L1
• • •
θ 2 L2
K Li
 

[142K ]⋅  r 
ω 
•r
θ i = K1θ..i 3 θi
4..6
•
(14) Fig.5 Complex kinematic vector bond-graph model of the stewart
4 43
4 t 
K θi
  platform mechanism

Copyright © 2008 Praise Worthy Prize S.r.l. - All rights reserved International Review of Mechanical Engineering, Vol.
xx, n. x
 Đ.Yıldız, V.E.Ömürlü, A.Sağırlı

If two distinct energy ports bounded each other by •r •

one energy port, this case can be shown as in Fig.4. In (15), θ i is 3 element matrix. Therefore, qb is a 24
Right side of Fig.4 illustrates developed complex element matrix. Independent velocities can be defined
vector bond-graph model and the one on the left side as,
represents using classical approach. • • • • 
T

Now, it is possible to show overall kinematic model q i = ω x ω y ω z t x t y t z  (16)
of the stewart platform mechanism by using complex  
kinematic vector Bond Graph method, Fig.5. Inertial velocities can be expressed as,
• • 
q
IV. Dynamic Bond Graph Model of a q I =  •b  (17)
 
Stewart Platform Mechanism  q i 
The relationship between independent and inertial
Attaching the physical elements of the system to the
velocities is TI matrix and can be defined as,
kinematic model, dynamic Bond-Graph model of
• •
stewart platform mechanism can be obtained. Vectorial q I = TI ⋅ q i (18)
dynamic Bond-Graph model of the mechanism for one
leg is shown in Fig. 6. All inertial elements of the system should be defined
IP with a single inertia matrix before obtaining virtual
[K ]
θi
[K ]
Li
1..3
inertial matrix and gyristor element. Inertial elements
1 ..3
Rθ •ω ω ω RL should be arranged using inertial velocity matrix. The
•ω
MTF 1 MTF
• θi •
• θi •
Li •
Li • virtual inertial matrix and the gyristor element can be
θi θi Li Li
1 0 0 1 expressed as,
[]
• •

I = [TI ]T ⋅ [I ] ⋅ [TI ]
θi [K ]
θi [K ]
Li
•t Li ~
•t
4 ..6 4..6
Li (19)
Iu θi MTF
•
1
•
MTF mu
t t and
•
[G] = [TI ]T ⋅ [I ]⋅ T I 
mP
(20)
Fig. 6 Vectorial dynamic Bond-Graph model of mechanism for one  
leg The derivative of relation matrix TI is needed to
calculate the gyristor element. It is useful way to
IV.1. Derivative Causality Solution derivate this matrix during simulation of complete
model of the system.
When two or more inertial elements are attached to After reducing the inertial elements to two ports,
“1” port, system state variables become dependent and bond graph of the system is shown in Fig. 7.
this causes derivative causality problem. This means,
independent velocities of these elements becomes EJS : MJR

dependent velocities when inertial elements attached and

[K ]
Mω ω

this causes derivative causality problem on the system. Rθ •ω

θi
1..3 ω ω [KMTF]
Li
1..3 RL
MTF
M θωi
1 •ω
• θi M Lωi FRi •
This problem results in an implicit differential equation M θi θi • Mθi ω ω
FLi
Li
• •
Li
θi [~I ] MI MG Li Li
and may cause numerical problems in solution. This 1 0 [G ] 0 1
FLi Fsi
[K ]
M θi M θi • •
FLi
[K ]
t t F
problem can be solved by transforming inertial elements θi
4..6
FI G Li •t
•t Fθti FLit 4..6 Li
of dependent velocities to the ports of independent θi
MTF
•
t
1
•
t
MTF

velocities. Therefore, inertial elements should be Fig. 7. Bond-Graph model of stewart platform mechanism after
reduced to the ports of independent velocity ports as a reducing the inertial elements
virtual inertia matrix and a gyristor element. Dependent ωx ωy
velocities, independent velocities and inertial velocities ωx Hz ωy
Mx 1 MJR 1 My
of the system should be defined before reducing the ωx ωy
system. Translational and rotational velocity of legs can ωx Ix I y ωy
be chosen as dependent velocities, therefore upper Hy Hx
MJ

platform’s translational and rotational velocities can be

MJ
R

Iz
chosen as independent velocities. All dependent and
ωz ωz
independent velocities can be chosen as inertial
velocities. Dependent velocities can be expressed as, 1
T
• • •r • •r • •r • •r • •r • •r 
q b =  L1 θ 1 L2 θ 2 L3 θ 3 L4 θ 4 L5 θ 5 L6 θ 6  (15) ωz M z
 

Fig. 8 Detailed view of one modulated gyrator element

Copyright © 2008 Praise Worthy Prize S.r.l. - All rights reserved International Review of Mechanical Engineering, Vol.
xx, n. x
 Đ.Yıldız, V.E.Ömürlü, A.Sağırlı

Fig. 7 shows the complete bond-graph model of The actuation of the system is based on linear DC
mechanical part of the system. The gyroscobic forces actuators which are chosen as legs of the stewart
acting on the platform is expressed by one modulated platform mechanism. Each linear motor is mounted to
gyrator element, Fig.8. The inertial elements which are mobile platform by spherical joints and is mounted to
shown in Fig. 8 are defined in the virtual inertial matrix. the stationary platform by universal joints. Complete
In Fig. 8, flow equations can be expressed as, bond-graph model of the mechanism for one leg is
M x = H zω y − H y ω z (21) shown in Fig. 9 after adding the actuation part.
The gravitational effect on legs and upper platform
M y = − H z ω x + H xω z (22) are modeled by effort sources, and the actuation inputs
M z = − H xω y + H yω x (23) of the legs are modeled by effort sources in Fig. 9. Right
side of the picture shows the linear motion and the left
Where, H x = I xω x , H y = I yω y , H z = I zω z .
side shows angular motion of the leg.

IV.2. Modeling the Actuation Mechanism (Linear DC

Motor)
EJS : MJR

[K ] [K ]
Mω ω
θi Li
Rθ 1..3 ω ω 1..3 RL La
•ω
MTF 1 MTF
θi M θωi M Li ω • ω
FRi
•
ia
M θi
•
ω ω Li Li VL
θi • M θi
FLi • •
θi Vexi
1 0 [~I ] MI MG
[G ] 0
Li
1
Li
GY
Kt Vsi
1 S e = Vexi
M θi • • FLi FLi Fsi ia ia

[KMTF ] [MTF
K ]
M θi t t F
FI
θi G Li •t VR ia
•t 4..6 Fθti FLti 4..6 Li
θi 1
2L i Sf = m u g. sin(θ iz )
Sf = m u g. cos(θ iz ) •
•
• Ra
3 t t
t Ft

Sf = m p g
Fig. 9. Bond-Graph model of a single leg of the stewart platform mechanism after adding the actuation part

equations. From the parallel junctions of the mechanical

V. SYSTEM STATE-SPACE EQUATIONS
part of the bond-graph model in Fig. 9, inner equations
State equations can be expressed as, are obtained as,
ωx  M I x 
r 6 r 6 r r r
ω  M I = ∑ M Lωi − ∑ M θωi − M G − M ω (26)
 
 y M I y  i =1 i =1
ω 
d  • z  ~ −1  M I z 
[]
r 6 r 6 r r
dt  t x 
= I 
FI x 
 FI = ∑ FLti − ∑ Fθt i
− FG − m p g (27)
 (24) i =1 i =1
•  F 
ty  Iy In equation (26) and (27), M G and FG terms are
 
•   FI z  effort variables of the gyrator element which are
 t z  
illustrated in Fig. 8. Open form of equation (26) can be
d
(ia ) = LaVLi , (i = 1..6) expressed as,

∑ (K11L FL )−
dt i 6
The constitutive law of the gyrator element can be M Ix = i
i
expressed as, i =1

 M Gx  ω x   θi  T  M i  
r (28)
ω  6  K11   θ x   ω 
M   θi i 
 Gy   y
ω z 
∑   K12   M θ y   − G1..6  •  − ( H z ω y − H yω z )
M G  i =1   θ i   i  t 
K
 z
 = [G ] •  (25)   13   M θ z  
 FGx  tx 
 FG   •

 y  t y 
 FGz  • 
 t z 
The right side of the equation (24) is created with the
terms of dependent variables which must be expressed
with the terms of state variables to find state-space
Copyright © 2008 Praise Worthy Prize S.r.l. - All rights reserved International Review of Mechanical Engineering, Vol.
xx, n. x
 Đ.Yıldız, V.E.Ömürlü, A.Sağırlı

∑ (K12L FL )−
6 • 2L
M Iy = i M θi z = Rθ θ iz − mu g i cos(θ iz ) =
i =1
i 3
r (36)
 θ i  T  M i  
ω
  2L
K r (29) Rθ K 3θ..i 6  r•  − mu g i cos(θ iz ), (i = 1..6)
6  21   θ x   ω 
 θi   t  3
∑  K 22   M θ y   − G2..6  •  − ( − H z ω x + H xω z )
i
Leg forces can also be given as,
i =1  θ i   i  t 
K r
 23   M θ z   ω
Li  • 
FLi = Fsi − RL K  r  − mu g . sin(θ iz ) (37)
 t 

∑ (K13L FL )−
6
M Iz = i Actuating force of the leg ( Fsi ) is also an unknown
i
i =1 variable in equation (37) and it can be expressed by
 θi  T  M i   using the bond-graph model in Fig. 8.
K
6   31   θ x  
r (30)
 θi  ω  VLi = Vexi − Vsi − VRi , (i = 1..6) (38)
∑   K 32   M θ y   − G3..6   − ( − H xω y + H y ω x )
i
•
i =1   θ i   i  t  where,
K r
  33   M θ z   • ω 
Vsi = L i K t = K L1  •  K t , Fsi = K t iai , VRi = iai Ra .
and open form of equation (27) also expressed as, t 
∑ (K14L FL )−
6
FIx = i
i
By writing the Vsi , Fsi , VRi in equation (38),
i =1 actuating forces can be expressed as,
r
 θ i  T  M i   Li ω 
K
6  14   θ x  
r (31) VLi = Vexi − K  •  K t − iai Ra , (i = 1..6)
 θi i  ω  (39)
∑  K15   M θ y   − G4..6  
• t 
i =1  θ i   i  t  From equations (26), (27), (38), M I x , M I y , M I z ,
K
 16   M θ z  
FI x , FI y , FI z , VLi terms can be expressed by the terms
∑( )−
6
FIy = Li
K15 FLi of state variables.
i =1 •
 θ i  T  M i   M I x = ω x Eωmx + ω y Eωmx + ω z Eωmx + t x Etmx +
K
6  24   θ x  
r (32) x y z x

ω 
( )
 θi i  • • 6 (40)
∑  K 25   M θ y   − G5..6  
•
t y Etmx + t z Etmx + ∑ ian Eimx + TLmx + Tθmx
i =1  θ i   i  t  y z
n =1
n n n
K
 26   M θ z   •
M I y = ω x Eω y + ω y Eω y + ω z Eω y + t x Et
m m m my
+
( )−
6
FIz = ∑ Li
K16 FLi x y z x

( )
i =1 • • 6 (41)
+ ∑ ian Ei
my my my m my
t y Et + t z Et + TL y + Tθ
 θ i  T  M i   y z n n n
 K 34   θ x   r (33) n =1
6
 θi i  ω 
∑  35θ   θ y   6..6  t  − m p g
K M − G •
M I z = ω x Eωmz + ω y Eωmz + ω z Eωmz + t x Etmz +
•
i =1 
K i   i    x y z x
 36   M θ z  
( )
• • 6 (42)
The G1..6 term in equation (28) shows the first row of t y Etm z + t z Etmz + ∑ ian Eimz + TLmz + Tθmz
y z n n n
n =1
gyrator matrix. Leg forces ( FLi ) and leg moments
•
( M θi ) are unknown variables of these equations. These FI x = ω x EωFx + ω y EωFx + ω z EωFx + t x EtFx +
x y z x

( )
unknown variables can be found by writing the • • 6 (43)
constitutive law for “0” energy ports. All unknown t y EtFx + t z EtFx + ∑ ian EiFx + TLFx + TθFx
y z n n n
n =1
variables should be expressed by the terms of state
•
variables. Leg moments can be given as,
FI y = ω x Eω y + ω y Eω y + ω z Eω y + t x Et y +
F F F F
r
•
θi  • 
ω x y z x

M θ x = Rθ θ ix = Rθ K1..6  r , (i = 1..6)
( )
i (44)
(34) • • 6
t y Et y + t z Et y + ∑ ian Ei y + TL y + Tθ
F F F F Fy
 t 
y z n n n
r n =1
•
θi  • 
ω
M θ y = Rθ θ i y = Rθ K 2..6  r , (i = 1..6)
i
(35) • •
 t  FI z = ω x EωFz + ω y EωFz + ω z EωFz + t x EtFz + t y EtFz +
x y z x y

( )
• 6 (45)
t z EtFz − Tmp
Fz
+ ∑ ian EiFz + TLFz + TθFz
z n n n
n =1

Copyright © 2008 Praise Worthy Prize S.r.l. - All rights reserved International Review of Mechanical Engineering, Vol.
xx, n. x
 Đ.Yıldız, V.E.Ömürlü, A.Sağırlı

•
V Li = ω x Eωvi + ω y Eωvi + ω z Eωvi + t x Etvi +  b11 b12 ..... b16 
x y z x b b22 ..... b26 
• •
(46)
[B ]6 x 6 =  21
 M
t y Etvi + t z Etvi + iai Eivi + Texvi , (i = 1..6) M ..... M 
y z i
 
Equations (40), (41), (42), (43), (44), (45), and (46) b61 b62 ..... b66 
should be plug into equation (24).  c11 c12 ..... c16 
These equation series can be expressed by matrix c c22 ..... c26 
norm as, [C ]6 x 6 =  21
 M
ω x  ω x  M ..... M 
     
ω y  ω y  c61 c62 ..... c66 
ω z  ω z 
 •   • 
 tx   tx 
 •   •  
 ty   ty  I = diag [IP]3x3 [mP]3x3 m1u [Ib1]3x3 m2u [Ib2]3x3 [Ib3]3x3
 
d  t•  [A]6 x6
z = +
[ ]
[B]6 x6   t•   I~ −1 6 x6 0 
[U ]12 x 6

dt   [C ]6 x 6 [D]6 x6   i z   0 [La ]6 x6  (47)
m4u [Ib4]3x3 m5u [Ib5]3x3 m6 [Ib6]3x3 ]30x30
 ia1   a1 
i  i 
 a2   a2  ...
ia3   ia3 
   
ia 4  ia 4 
ia   ia 
 5  5
ia6  ia6 
Equation (47) represents the state-space equation for
a stewart platform mechanism.

V. Conclusion
Dynamic characteristics of mechanical systems take
important place for efficient control. Parallel
mechanisms, especially stewart platform mechanisms,
have complex system dynamics and are completely
nonlinear. Therefore, in the present study, kinematics
and dynamics of a general stewart platform mechanism
are obtained by Bond Graph method using a novel
visualization technique and nonlinear state-space
representation of the system dynamics is developed. An
important problematic point of obtaining dynamic
equations is the possible derivative causality problem. If
there is a disregarded derivative causality problem
during calculations, numerical problems and errors
occur when simulating the system. Thus, transforming
inertial elements of dependent velocities to the ports of
independent velocities, derivative causality problem is
solved. At the end of the study, a state-space model of
the mechanism is provided.

Appendix
 a11 a12 ..... a16 
a a ..... a 
[A]6 x6 =
21 22
 M M ..... M 
26 

 
a61 a62 ..... a66 

Copyright © 2008 Praise Worthy Prize S.r.l. - All rights reserved International Review of Mechanical Engineering, Vol.
xx, n. x
 Đ.Yıldız, V.E.Ömürlü, A.Sağırlı

~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~
a11 = I11−1Eωmxx + I12−1Eω xy + I13−1Eωmxz + I14−1EωFxx + I15−1Eω xy + I16−1EωFxz b11 = I11−1 Eimx + I12−1 Ei y + I13−1Eimz + I14−1 EiFx + I15−1Ei y + I16−1 EiFz
m F m F
1 1 1 1 1 1

~ ~ m ~ ~ ~ F ~ ~ ~ ~ ~ ~ ~
a12 = I11−1Eωmyx + I12−1Eω yy + I13−1Eωmyz + I14−1EωFyx + I15−1Eω yy + I16−1EωFyz b12 = I11−1 Eim x + I12−1 Ei y + I13−1 Eimz + I14−1 EiFx + I15−1Ei y + I16−1EiFz
m F
2 2 2 2 2 2

M M
~ ~ m ~ ~ ~ F ~ ~ ~ ~ ~ ~ ~
b16 = I11−1 Eim x + I12−1 Ei y + I13−1 Eimz + I14−1 EiFx + I15−1Ei y + I16−1EiFz
m F
a16 = I11−1Etmz x + I12−1Et z y + I13−1Etmz z + I14−1EtFz x + I15−1Et z y + I16−1EtFz z 6 6 6 6 6 6

~ ~ m ~ ~ ~ F ~ ~ ~ ~ ~ ~ ~
a21 = I 21−1Eωmxx + I 22−1Eω xy + I 23−1Eωmxz + I 24−1EωFxx + I 25−1Eω xy + I 26−1EωFxz b21 = I 21−1 Eim x + I 22−1 Ei y + I 23−1 Eim z + I 24−1 EiFx + I 25−1 Ei y + I 26−1EiFz
m F
1 1 1 1 1 1

~ ~ m ~ ~ ~ F ~ ~ −1 m x ~ −1 m y ~ −1 m z ~ −1 Fx ~ −1 Fy ~ −1 Fz
a22 = I 21−1Eωmyx + I 22−1Eω yy + I 23−1Eωmyz + I 24−1EωFxy + I 25−1Eω yy + I 26−1EωFyz b22 = I 21 Ei + I 22 Ei + I 23 Ei + I 24 Ei + I 25 Ei + I 26 Ei
2 2 2 2 2 2

M M
~ ~ ~ ~ ~ ~
~ ~ m ~ ~ ~ F ~ b66 = I 61−1 Eim x + I 62−1 Ei y + I 63−1 Eim z + I 64−1 EiFx + I 65−1 Ei y + I 66−1EiFz
m F
a66 = I 61−1Etmz x + I 62−1Et z y + I 63−1Etmz z + I 64−1EtFz x + I 65−1Et z y + I 66−1EtFz z 6 6 6 6 6 6

 1 0 0 0 0 0 
 0 1 0 0 0 0  c11 = La1Eωv1x , c12 = La1 Eωv1y , c13 = La1 Eωv1z ,

 0 0 1 0 0 0  c14 = La1 Etv1 , c15 = La1 Etv1 , c16 = La1 Etv1
  x y z

 0 0 0 1 0 0 
 0 c21 = La2 Eω x , c22 = La2 Eω y , c23 = La2 Eω z ,
v2 v2 v2
0 0 0 1 0 
 
 0 0 0 0 0 1  c24 = La2 Etv2 , c25 = La2 Etv2 , c26 = La2 Etv2
 K L1 L1 
x y z
L1 L1 L1 L1
K12 K13 K14 K15 K16
 11 θ1 θ1 θ1 θ1 θ1 θ1
 M
 K11 K12 K13 K14 K15 K16 
 θ1 θ1 θ1 θ1 θ1 θ1  c61 = La6 Eωv6 , c62 = La6 Eωv6 , c63 = La6 Eωv6 ,
 K 21 K 22 K 23 K 24 K 25 K 26  x y z

θ1
 K 31 θ1 θ1 θ1 θ1 θ1  c64 = La6 Etv6 , c65 = La6 Etv6 , c66 = La6 Etv6
K 32 K 33 K 34 K 35 K 36
 L2 L2 L2 L2 L2 L2 
x y z

 K11 K12 K13 K14 K15 K16   d11 d12 ..... d16 
 K θ2 θ2 θ2 θ2 θ2 θ2  d
K12 K13 K14 K15 K16 d 22 ..... d 26 
 11 θ
 K 212
θ2
K 22 θ2
K 23 θ2
K 24 θ2
K 25 θ2
K 26


[D ]6 x 6 =  21
 M M ..... M 
 K θ2 θ2 θ2
K 32 K 33 K 34 K 35θ2 θ2 θ2 
K 36  
 31   d 61 d 62 ..... d 66 
 K L3 L3
K12 L3
K13 L3
K14 L3
K15 L3
K16 
TI =  11 θ3 θ3 θ3 θ3 θ3 θ3 
 K11 K12 K13 K14 K15 K16 
d11 = La1 ERv1 , d12 = 0, d13 = 0, K, d16 = 0
θ3
 K 21 θ3 θ3 θ3 θ3 θ3 
K 22 K 23 K 24 K 25 K 26 1
 θ3 θ3 θ3 θ3 θ3 θ3 
 K 31 K 32 K 33 K 34 K 35 K 36  d 21 = La2 ERv2 , d 22 = 0, d 23 = 0, K, d 26 = 0
2
 K L4 L4
K12 L4
K13 L4
K14 L4
K15 L4 
K16
 11  M
θ θ4 θ4 θ4 θ4 θ4
 K114 K12 K13 K14 K15 K16  d 61 = La 6 ERv6 , d 62 = 0, d 63 = 0,K, d 66 = 0
 K θ4 θ4
K 22 θ4
K 23 θ4
K 24 θ4
K 25 θ4 
K 26 
6
 21
θ4 θ4 θ4 θ4 θ4 θ4
 K 31 K 32 K 33 K 34 K 35 K 36 
 L5 L5 L5 L5 L5 L5 
~ I −111
~
I −112 .....
~ −1 
I 16
 K11 K12 K13 K14 K15 K16   ~ −1 ~ ~ −1 
 K θ5
 11
θ5
K12 θ5
K13 θ5
K14 θ5
K15 θ5 
K16 [I~]−1
6 x6 =
 M
I 21 I −122 ..... I 26 
M 
θ θ5 θ5 θ5 θ5 θ5  M .....
 K 215 K 22 K 23 K 24 K 25 K 26   ~ −1 ~ −1 ~ −1 
 K θ5 θ5
K 32 θ5
K 33 θ5
K 34 θ5
K 35 θ5 
K 36  I 61 I 62 ..... I 66 
 31 L6 L6 L6 L6 L6 L6
  La1 0 ..... 0 
 K11 K12 K13 K14 K15 K16   0
 θ6 θ6 θ6 θ6 θ6 θ6  La2 ..... 0 
 K11 K12 K13 K14 K15 K16  [La ]6 x6 =
 M M ..... M 
θ6
 K 21 θ6 θ6 θ6 θ6 θ6 
K 22 K 23 K 24 K 25 K 26  
 θ6 θ6 
 K 31
θ6
K 32 θ6
K 33 θ6
K 34 θ6
K 35 K 36   0 0 ..... La6 
[U ] = [u1 u2 u3 u 4 u5 u6
Tvv1 Tvv2 Tvv3 Tvv4 Tvv5 Tvv6
ex ex ex ex ex ex
]
T

Copyright © 2008 Praise Worthy Prize S.r.l. - All rights reserved International Review of Mechanical Engineering, Vol.
xx, n. x
 Đ.Yıldız, V.E.Ömürlü, A.Sağırlı

6 6 6   ~
− R K n K n −
L L

∑ TLm ∑ TL Eωmzz = ∑  Ln θ 13 θ 13 θ θ
my my
u1 = x
+ Tθmx , u2 = + Tθ θ n θ n  − G33
 Rθ ( K 31 K13 + K 32 K 23 + K 33 K 33 ) 
n =1 
n n n n n n n n
n =1 n =1
6 6
u3 = ∑ TLm z
+ Tθmz , u4 = ∑ TLF x
+ TθFx 6 
− R K n K n −
L
Etmx z = ∑  Ln θ 13 θ 14 θ θ
L  ~
θ n θ n  − G34
n n n n
n =1 n =1
6 6  R
n =1  θ ( K n
K
31 14
n
+ K n
K
32 24
n
+ K 33 34 
K ) 
∑ ∑ TLF
F F
u5 = TL y + Tθ y , u6 = TmFz + z
+ TθFz
n n p n n
n =1 n =1 6   ~
− R K n K n −
L L
Etmy z = ∑  Ln θ 13 θ 15 θ θ θ n θ n  − G35
6 − R ( K Ln )2 −  ~  Rθ ( K 31 K15 + K 32 K 25 + K 33 K 35 ) 
n =1 
n n n n

Eωmx =
x ∑R ((LKθ 11)2 + Kθ Kθ
n
n n n θn
+ K11 K31 θn
)  − G11

n =1  θ 11 11 21 6 
− R K n K n −
L L  ~
  Etmz z = ∑  Ln θ 13 θ 16 θ θ θ n θ n  − G36
6 − RLn K11 K12 −
Ln Ln
 Rθ ( K 31 K16 + K 32 K 26 + K 33 K 36 )
n =1 
n n n n
  ~
Eωmx
y
= ∑ θn 2
  ( K12
R
θn θn
) + K12 K 22 +  − G12 − H z
n =1  θ  θ n θ n  6   ~
− R K n K n −
L L
K
  12 32 K  EωFxx = ∑  Ln θ 14 θ 11 θ θ θ n θ n  − G41
 R
n =1  θ ( K n
K n
+ K n
K n
+ K 16 31 
K ) 

6 − RLn K11 K13 −
Ln Ln  14 11 15 21
  ~
Eωmx ∑ 
θ 2 θ θ
=   ( K13n ) + K12n K 23n +  − G13 + H y

6 
− R K n K n −
L L  ~
R EωFyx = ∑  Ln θ 14 θ 12 θ θ
z
n =1   θ n θ n  − G42
θ  n n
θ θ
  K12 K 33   Rθ ( K14 K12 + K15 K 22 + K16 K 32 ) 
n =1 
n n n n

6   ~
− R K K −
Ln Ln
= ∑  Ln θ 11 θ 14 θ θ
Etmx x
 R
n =1  θ K n
(
K
11 14
n
+ K n
K
11 24
n
+ K θ n θ n  − G14
11 34 
K  ) 6 
− R K n K n −
L
EωFzx = ∑  Ln θ 14 θ 13 θ θ
L  ~
θ n θ n  − G43
 Rθ ( K14 K13 + K15 K 23 + K16 K 33 ) 
n =1 
n n n n
6 
− R K K −  ~
Ln Ln
= ∑  Ln θ 11 θ 15 θ θ
Etmy x
 R
n =1  θ K n
(
K n
+ K n
K n
+ K θ n θ n  − G15
11 35 
K  ) 6 
− R K n K n −
L L  ~
EtFx x = ∑  Ln θ 14 θ 14 θ θ
11 15 11 25
θ n θ n  − G44
6− RL K11 K16 −
Ln Ln  ~  R
n =1  θ ( K n
K n
+ K n
K n
+ K 16 34 
K ) 
∑ (
14 14 15 24
Etmx = 
) − G16
n
θn θn θn θ n θn θ n
n=1 Rθ K11 K16 + K11 K26 + K11 K36 
z
6   ~
− R K n K n −
L L
EtFx = ∑  Ln θ 15 θ 15 θ θ θ n θ n  − G45
− RL K12 K11 −
Ln L n  y
 Rθ ( K14 K15 + K15 K 25 + K16 K 35 ) 
n =1 
n n n n
6
 n
θn θn θn θn  ~
= ∑ Rθ ( K 21 K11 + K 22
my
Eω K 21 +  − G21 + H z
θ θ 6   ~
− R K n K n −
x L L
n =1  K K ) 
EtFx = ∑  Ln θ 14 θ 16 θ θ
n n
 23 31  θ n θ n  − G46
 Rθ ( K14 K16 + K15 K 26 + K16 K 36 ) 
n =1 
z n n n n

6   ~
− R K n K n −
L L
Eω yy = ∑  Ln θ 12 θ 12 θ θ
m
θ n θ n  − G22
6   ~
− R K n K n −
L L
Eω y = ∑  Ln θ 15 θ 11 θ θ
F
 R ( K K + K K + K 23 32 
K ) θ n θ n  − G51
n n n n
n =1  θ 21 12 22 22 
 Rθ ( K 24 K11 + K 25 K 21 + K 26 K 31 ) 
n =1 
x n n n n

− RL K12 K13 −
Ln Ln 
6
 n
θn θn θn θn  ~
= ∑ Rθ ( K 21 K13 + K 22 6   ~
my
Eω K 23 +  − G23 − H x − R K n K n −
L L
Eω y = ∑  Ln θ 15 θ 12 θ θ
F
θ θ θ n θ n  − G52
z
n =1  K n K n )   + +
R ( K K K K K 26 32 
K )
y n n n n
n =1  θ 
 23 33  24 12 25 22

6   ~
− R K n K n −
L L
6 − R K Ln K Ln −  ~
Eω y = ∑  Ln θ 15 θ 13 θ θ
F
= ∑  Ln θ 12 θ 14 θ θ θ n θ n  − G53
m
Et x y θ n θ n  − G24
+ +  Rθ ( K 24 K13 + K 25 K 23 + K 26 K 33 ) 
n =1 
z n n n n

n =1  θR ( K n
K
21 14
n
K n
K
22 24
n
K 23 34 
K ) 
6   ~
− R K n K n −
6  L L
− R K n K n −
L L  ~
= ∑  Ln θ 15 θ 14 θ θ
Fy
Et y y = ∑  Ln θ 12 θ 15 θ θ θ n θ n  − G54
m
θ n θ n  − G25
Et
 Rθ ( K 21 K15 + K 22 K 25 + K 23 K 35 )   Rθ ( K 24 K14 + K 25 K 24 + K 26 K 34 ) 
n =1 
x n n n n
n =1 
n n n n

6   ~
− R K n K n −
6  L L
− R K n K n −
L L  ~
= ∑  Ln θ 15 θ 15 θ θ
Fy
Et z y = ∑  Ln θ 12 θ 16 θ θ θ n θ n  − G55
m
θ n θ n  − G26
Et
 + + 26 35 
 Rθ ( K 21 K16 + K 22 K 26 + K 23 K 36 ) 
y
R ( K n
K n
K n
K n
K K )
n =1  n =1  θ 
n n n n
24 15 25 25

6   ~
− R K n K n −
L L
− RL K13 K11 −
Ln L n  Fy
= ∑  Ln θ 15 θ 16 θ θ
6
 n
 ~ Et θ n θ n  − G56
θn θn
= ∑ Rθ ( K 31 K11 + K 32θn θn
+  − G31 − H y  Rθ ( K 24 K16 + K 25 K 26 + K 26 K 36 ) 
n =1 
z n n n n
Eωm z K 21
x
θ θ
n =1  K n K n ) 
 33 31  6 
− R K n K n −
L L  ~
EωFz = ∑  Ln θ 16 θ 11 θ θ θ n θ n  − G61
 Rθ ( K 34 K 11 + K 35 K 21 + K 36 K 31 ) 
n =1 
x n n n n
− RL K13 K12 −
Ln Ln 
6
 n
θn θn θn θn  ~
Eωmz
= ∑ Rθ ( K 31 K12 + K 32 K 22 +  − G32 + H x 6   ~
− R K n K n −
L L
n =1  K θ n K θ n ) EωFz = ∑  Ln θ 16 θ 12 θ θ
y
θ n θ n  − G62

 33 32
 y
 R
n =1  θ ( K n
K
34 12
n
+ K n
K
35 22
n
+ K 36 32 
K ) 

Copyright © 2008 Praise Worthy Prize S.r.l. - All rights reserved International Review of Mechanical Engineering, Vol.
xx, n. x
 Đ.Yıldız, V.E.Ömürlü, A.Sağırlı

6   ~
− R K n K n − TLFx = − K14
L5
mu5 g sin(θ53 ) , TLFx = − K14
L6
mu6 g sin(θ 63 )
L L
EωFz = ∑  Ln θ 16 θ 13 θ θ θ n θ n  − G63 5 6
 Rθ ( K 34 K13 + K 35 K 23 + K 36 K 33 ) 
n =1 
z n n n n

mu1 g sin(θ13 ) , TL y = − K15 mu2 g sin(θ 23 ) ,

F F
6 − R K Ln K Ln −  ~ TL y = − K15
L1 L2

EtFz = ∑  Ln θ 16 θ 14 θ θ θ n θ n  − G64
1 2

 R ( K K + K K + K 36 34 
K )
mu3 g sin(θ 33 ) mu4 g sin(θ 43 ) ,
x n n n n F F
n =1  θ 34 14 35 24  TL y = − K15
L3
TL y = − K15
L4
3 , 4
6 − R K Ln K Ln −  ~
mu5 g sin(θ 53 ) , TL y mu6 g sin(θ 63 )
F F
= ∑  Ln θ 16 θ 15 θ θ TL y = − K15
L5
= − K15
L6
E tFz θ n θ n  − G65
 Rθ ( K 34 K15 + K 35 K 25 + K 36 K 35 ) 
5 6
n =1 
y n n n n

6 
− R K n K n −
L L  ~ TLFz = − K16
L1
mu1 g sin(θ13 ) , TLFz = − K16
L2
mu2 g sin(θ 23 ) ,
EtFz = ∑  Ln θ 16 θ 16 θ θ θ n θ n  − G66
1 2

 Rθ ( K 34 K16 + K 35 K 26 + K 36 K 36 )
n =1 
z
mu3 g sin(θ 33 ) TLFz mu4 g sin(θ 43 ) ,
n n n n
TLFz = − K16
L3
= − K16
L4
3 , 4
Eimx = L1
K11 K t , Eimx = K11
L2
K t , Eimx = L3
K11 Kt ,
1 2 3
TLFz = − K16
L5
mu5 g sin(θ 53 ) , TLFz = − K16
L6
mu6 g sin(θ 63 )
Eimx = L4
K11 K t , Eimx = K11
L5
K t , Eimx = L6
K11 Kt 5 6
4 5 6

θ1
my my my Tθmx = K11 mu1 gn1L1 cos(θ13 ) ,
Ei = K12
L1
K t , Ei = K12
L2
K t , Ei = K12
L3
Kt , 1
1 2 3
θ2
= K11 mu2 gn2 L2 cos(θ 23 ) ,
m m m mx
Ei y = L4
K12 K t , Ei y = L5
K12 K t , Ei y = L6
K12 Kt Tθ
2
4 5 6

θ3
Tθmx = K11 mu3 gn3 L3 cos(θ 33 )
Eimz = K13
L1
K t , Eimz = K13
L2
K t , Eimz = K13
L3
Kt , 3 ,
1 2 3
θ4
Eimz = K13
L4
K t , Eimz = K13
L5
K t , Eimz = K13
L6
Kt Tθmx
= K11 mu4 gn4 L4 cos(θ 43 ) ,
4 5 6 4

θ1
mu1 gn1L1 cos(θ13 ) ,
my
Tθ = K 23
EiFx = K14
L1
K t , EiFx = K14
L2
K t , EiFx = K14
L3
Kt , 1

θ6
1 2 3

EiFx = K14
L4
K t , EiFx = K14
L5
K t , EiFx = K14
L6
Kt Tθmx = K11 mu6 gn6 L6 cos(θ 63 )
4 5 6 6

Fy Fy Fy θ1
mu1 gn1L1 cos(θ13 ) ,
my
Ei = K15
L1
K t , Ei = K15
L2
K t , Ei = K15
L3
Kt , Tθ = K 23
1 2 3 1
Fy Fy Fy
Ei = K15
L4
K t , Ei = K15
L5
K t , Ei = K15
L6
Kt θ2
mu2 gn2 L2 cos(θ 23 ) ,
my
4 5 6 Tθ = K 23
2

EiFz = K16 K t , EiFz = K16 K t , EiFz = K16 θ3

mu3 gn3 L3 cos(θ 33 )
L1 L2 L3 my
Kt , Tθ = K 23
1 2 3 3 ,
EiFz = K16
L4
K t , EiFz = K16
L5
K t , EiFz = K16
L6
Kt θ4
= K 23 mu4 gn4 L4 cos(θ 43 ) ,
my
4 5 6
Tθ
4

TLmx = − K11
L1
mu1 g sin(θ13 ) , TLmx = − K11
L2
mu2 g sin(θ 23 ) , θ5
mu5 gn5 L5 cos(θ 53 ) ,
my
1 2
Tθ = K 23
5

TLmx = − K11
L3
mu3 g sin(θ 33 ) TLmx = − K11
L4
mu4 g sin(θ 43 ) , Tθ
my θ6
= K 23 mu6 gn6 L6 cos(θ 63 )
3 , 4 6

TLmx = − K11
L5
mu5 g sin(θ 53 ) , TLmx = − K11
L6
mu6 g sin(θ 63 )
, θ1
Tθmz = K33 mu1 gn1L1 cos(θ13 ) ,
5 6

1
m
= − K12
L1
mu1 g sin(θ13 ) , TL y
m
= − K12
L2
mu2 g sin(θ 23 ) θ2
TL y
1 2
, Tθmz
= K 33 mu2 gn2 L2 cos(θ 23 ) ,
2

θ3
mu3 g sin(θ33 ) TL y = − K12 mu4 g sin(θ 43 ) ,
m m
TL y = − K12 Tθmz = K33 mu3 gn3 L3 cos(θ 33 )
L3 L4
3 , 4 3 ,
θ4
mu5 g sin(θ 53 ) , TL y mu6 g sin(θ 63 ) = K 33 mu4 gn4 L4 cos(θ 43 ) ,
m m mz
TL y = − K12
L5
= − K12
L6
Tθ
4
5 6

θ5
Tθmz = K33 mu5 gn5 L5 cos(θ 53 ) ,
TLmz = − K13
L1
mu1 g sin(θ13 ) m
, TL z = − K13
L2
mu2 g sin(θ 23 ) , 5

θ6
= K 33 mu6 gn6 L6 cos(θ 63 )
1 2 mz
Tθ
TLmz = − K13
L3
mu3 g sin(θ33 ) TLmz = − K13
L4
mu4 g sin(θ 43 ) , 6
3 , 4
θ1
TLmz = − K13
L5
mu5 g sin(θ53 ) , TLmz
6 = − K13 mu 6 g sin(θ 63 )
L6
TθFx = K16 mu1 gn1L1 cos(θ13 ) ,
5 1

θ2
Tθ = K16
Fx
mu2 gn2 L2 cos(θ 23 ) ,
TLFx = − K14
L1
mu1 g sin(θ13 ) , TLFx = − K14
L2
mu2 g sin(θ 23 ) , 2

θ3
Tθ = K16 mu3 gn3 L3 cos(θ 33 )
1 2 Fx

TLFx = − K14
L3
mu3 g sin(θ 33 ) TLFx = − K14
L4
mu4 g sin(θ 43 ) , 3 ,
, 4 θ4
Tθ = K16 mu4 gn4 L4 cos(θ 43 ) ,
3 Fx
4

Copyright © 2008 Praise Worthy Prize S.r.l. - All rights reserved International Review of Mechanical Engineering, Vol.
xx, n. x
 Đ.Yıldız, V.E.Ömürlü, A.Sağırlı

θ5
TθFx = K16 mu5 gn5 L5 cos(θ 53 ) , n1 =
2 2 2 2 2
, n 2 = , n3 = , n 4 = , n5 = , n 6 =
2
5
3 3 3 3 3 3
θ6
TθFx = K16 mu6 gn6 L6 cos(θ 63 )
6

Acknowledgement
θ1
mu1 gn1L1 cos(θ13 ) ,
F
Tθ y = K 26
1 This work is supported by The Scientific and
θ2 Technological Research Council of Turkey (TUBITAK)
Tθ = K 26 mu2 gn2 L2 cos(θ 23 ) ,
Fy
2 grant 3501-105M192.
θ3
mu3 gn3 L3 cos(θ33 )
Fy
Tθ = K 26
3 ,
TθFy θ4
= K 26 mu 4 gn4 L4 cos(θ 43 ) , References
4
θ5 [1] Stewart, D.,A Platform with Six Degrees-of-freedom,
mu5 gn5 L5 cos(θ 53 ) ,
Fy
Tθ = K 26 Proceedings of Mechanical Engineering Part I, Vol.180, pp
5
371-386, 1965-1966.
θ6
mu6 gn6 L6 cos(θ 63 )
F
Tθ y = K 26 [2] Karnopp, D.C., Rosenberg, R.C., Introduction to Physical
6 , System Dynamics McGraw Hill Inc., New York, 1983.
[3] Thoma, J.U., Simulation by Bond Graphs: Introduction to a
θ1
TθFz = K 36 mu1 gn1L1 cos(θ13 ) , Graphical Method Springer-Verlag, Berlin, 1990.
1 [4] Margolis, D. And Shim T., A Bond Graph Model Incorporating
θ2 Sensors, Actuators, And Vehicle Dynamics For Developing
Tθ = K 36
Fz
mu2 gn2 L2 cos(θ 23 ) , Controllers For Vehicle Safety, Journal of the Franklin
2

θ3 Institute, Vol. 338, n. 1, 21-34, 2001.

Tθ = K 36
Fz
mu3 gn3 L3 cos(θ 33 ) [5] Allen, R.R., Dubowsky, S., Mechanisms as Components of
3 ,
Dynamic Systems: A Bond Graph Approach, ASME Journal of
θ4
Tθ = K 36 mu4 gn4 L4 cos(θ 43 ) ,
Fz Engineering for Industry, Vol. 99, No. 1, pp: 104-111, 1977.
4 [6] Bos, A.M., Tiernego, M.J.L., Formula Manipulating in the Bond
θ5
Tθ = K 36
Fz
mu5 gn5 L5 cos(θ 53 ) , Graph Modeling and Simulation of Large Mechanical Systems,
5 Journal of Franklin Institute, Vol. 319, n. 1/2, pp. 51-55, 1985.
θ6 [7] Fahrenthold, E.P., Worgo, J.D., Vector Bond Graph Analysis of
Tθ = K 36
Fz
mu6 gn6 L6 cos(θ 63 ) Mechanical Systems, Transactions of ASME, Vol. 113, pp. 344-
6
353, 1991.
[8] Karnopp, D., Approach to Derivative Causality in Bond Graph
TmFz = −m p g Models of Mechanical Systems, Journal of the Franklin
p
Institute, Vol. 329, n. 1, 1992.
[9] Do, W. Q. D. and Yang, D. C. H., Inverse Dynamic Analysis
Eωv1 = − K11
L1
K t , Eωv2 = − K11
L2
K t , Eωv3 = − K11
L3
Kt , and Simulation of a Platform Type of Robot, Journal of Robotic
x x x
Systems , Vol. 5, n. 3, pp. 209-227, 1988.
Eωv4 = − K11
L4
K t , Eωv5 = − K11
L5
K t , Eωv6 = − K11
L6
Kt [10] Dasgupta B., Mruthyunjaya T.S., Closed-form dynamic
x x x
equations of the general Stewart Platform through the Newton-
Euler approach. Mechanism and Machine Theory, Vol. 33, n. 7,
Eωv1 = − K12
L1
K t , Eωv2 = − K12
L2
K t , Eωv3 = − K12
L3
Kt , pp. 993-1012, 1998.
y y y
[11] Geng, Z., Haynes, L. S., Lee, J. D. and Carroll, R. L., On the
Eωv4 = − K12
L4
K t , Eωv5 = − K12
L5
K t , Eωv6 = − K12
L6
Kt
y y y Dynamic Model and Kinematic Analysis of a Class of
Stewart Platforms, Robotics and Autonomous Systems, Vol. 9,
pp. 237-254, 1992.
Eωv1 = − K13
L1
K t , Eωv2 = − K13
L2
K t , Eωv3 = − K13
L3
Kt , [12] Liu, K., Fitzgerald, M., Dawson, D. W. and Lewis, F. L.,
z z z

Eω = − K13
v4 L4
K t , Eω = − K13v5
L5
K t , Eω = − K13
L6
Kt v6
Modeling and control of a Stewart platform manipulator, ASME
z z z
DSC Vol. 33, Control of Systems with Inexact Dynamic Models,
Etv1 = − K14
L1
K t , Etv2 = − K14
L2
K t , Etv3 = − K14
L3
Kt , 1991, pp. 83-89.
x x x
[13] Sagirli A., Bogoclu M.E., Omurlu V.E., Modeling a Rotary
Etv4 = − K14
L4
K t , Etv5 = − K14
L5
K t , Etv6 = − K14
L6
Kt Crane by Bond-Graph Method, Analysis of Dynamical Behavior
x x x
and Load Spectrum Analysis (Part I), Nonlinear Dynamics, Vol.
33, pp. 337-351, 2003.
Etv1 = − K15
L1
K t , Etv2 = − K15
L2
K t , Etv3 = − K15
L3
Kt , [14] I. Davliakos, P. Chatzakos, and E. Papadopoulos, Development
y y y
of a Model-based Impedance Controller for Electrohydraulic
Etv4 = − K15
L4
K t , Etv5 = − K15
L5
K t , Etv6 = − K15
L6
Kt Servos, In L.A. Gerhardt (Ed.), Robotics and Applications,
y y y
(ACTA Press, 2005, 498-061).
Etv1 = − K16
L1
K t , Etv2 = − K16
L2
K t , Etv3 = − K16
L3
Kt , [15] I. Davliakos, E. Papadopoulos, A Model-Based Impedance
z z z

Etv4 = − K16
L4
K t , Etv5 = − K16
L5
K t , Etv6 = − K16
L6
Kt Control of a 6-dof Electrohydraulic Stewart Platform,
z z z Proceedings of the European Control Conference, Greece, 2007,
pp. 3507-3514.
E Rv1 = R1 , E Rv2 = R2 , E Rv3 = R3 ,
E Rv4 = R4 , E Rv5 = R5 , E Rv6 = R6
Authors’ information
1
Yıldız Technical University, Mechanical Engineering Department,
Tvvi = Vexi , i = 1..6 34349, Besiktas, Istanbul/Turkiye, Fax: 90-212-2616659, e-mail:
ex
iyildiz@yildiz.edu.tr

Copyright © 2008 Praise Worthy Prize S.r.l. - All rights reserved International Review of Mechanical Engineering, Vol.
xx, n. x
 Đ.Yıldız, V.E.Ömürlü, A.Sağırlı

2
Yıldız Technical University, Mechatronic Engineering Department,
34349, Besiktas, Istanbul/Turkiye, Fax: 90-212-2616659, e-mail:
omurlu@yildiz.edu.tr
3
Yıldız Technical University, Mechanical Engineering Department,
34349, Besiktas, Istanbul/Turkiye, Fax: 90-212-2616659, e-mail:
sagirli@yildiz.edu.tr

Đbrahim Yıldız was born in Konya/Turkiye on

26th of August. He received his M.Sc. degree in
Yildiz Technical University of Istanbul/Turkiye
about the topic of dynamics of stewart platform
mechanisms on year 2007.
He is currently working on parallel
mechanisms, biomechanics and aerial vehicles.
Mr. Yildiz is currently the research/teaching
assistant in Yildiz Technical University and is working towards his
Ph.D. degree.

Vasfi E. Ömürlü was born in Istanbul/Turkiye

on 4th of January, 1969. He received his M.Sc.
degree in Yıldız Technical University of
Istanbul/Turkiye about vehicle dynamics on year
1993 and completed his Ph.D. studies about
surgical robotics in Ohio State University of
Columbus, Ohio/USA on year 2002.
He published several papers on the topics of
vehicle dynamics and control, dynamics of robotic and crane systems,
and surgical robotics. He is currently working on parallel robotic
manipulators, autonomous aerial vehicles and automation systems.
Dr. Ömürlü is currently the vice chair of recently established
Mechatronic Engineering Department of Yıldız Technical University.

Ahmet Sağırlı was born in Erzincan/Turkiye

on 28th of November, 1964. He received his
M.Sc. degree in Yıldız Technical University of
Istanbul/Turkiye about machine design on year
1988 and completed his Ph.D. studies about the
dynamics of telescopic rotary cranes in Yıldız
Technical University of Istanbul/Turkiye on
year1996.
He published several papers on the topics of rotary crane dynamics
and is currently working on parallel robotic manipulators, wind
affects on steel constructions and dynamics of elevator systems.
Dr. Sağırlı is currently the faculty of Mechanical Engineering
Department of Yıldız Technical University.

Copyright © 2008 Praise Worthy Prize S.r.l. - All rights reserved International Review of Mechanical Engineering, Vol.
xx, n. x