Modeling and Simulation of the Dynamics of the Quick Return

Mechanism: A Bond Graph Approach

Anand Vaz1* and Thommen G.K.2
1,2
Department of Mechanical Engineering, Dr. B.R. Ambedkar National Institute of Technology, Jal-
andhar
*
Anand Vaz (email: [email protected])

formulation of system equations which govern the be-

havior of the system, help in pointing out any physical
Abstract impossibility or system property we may have failed to
take into account in the modeling stage.
This paper applies the multibond graph approach for Simulation has become an indispensible analytical
rigid multibody systems to model the dynamics of gen- tool using which one can experiment with a system at
eral spatial mechanisms. The commonly used quick re- little or no expense. It is also used in design and devel-
turn mechanism which comprises of revolute as well as opment of appropriate control systems. In this work, a
prismatic joints has been chosen as a representative ex- multibond or vector bond graph approach is used to
ample to demonstrate the application of this technique model the dynamics of the quick return mechanism [3].
and its resulting advantages. In this work, the links of Using this Bond Graph model, we simulate the dynamic
the quick return mechanism are modeled as rigid bodies. behavior of the system with appropriate codes written in
The rigid links are then coupled at the joints based on MATLAB [4]. This establishes a very effective method
the nature of constraint. This alternative method of for- of predicting the behavior of systems, which, in a num-
mulation of system dynamics, using Bond Graphs, offers ber of applications can prove to be economical as well as
a rich set of features that include pictorial representation time saving.
of the dynamics of translation and rotation for each link In general, mechanical systems can be treated as a
of the mechanism in the inertial frame, representation finite number of rigid bodies, interconnected suitably,
and handling of constraints at the joints, depiction of with appropriate constraints imposed at the joints. Kin-
causality, obtaining dynamic reaction forces and mo- ematics of such mechanisms is usually available in most
ments at various locations in the mechanism and so on. texts and reference books on machines and mechanisms
Yet another advantage of this approach is that the coding [1, 5]. However, the dynamics is rarely presented alt-
for simulation can be carried out directly from the Bond hough it is extremely important from the design perspec-
Graph in an algorithmic manner, without deriving sys- tive. Using the Bond Graph approach, simulation can be
tem equations. In this work, the program code for simu- conveniently carried out in MATLAB, and the dynamic
lation is written in MATLAB. The vector and tensor quantities like reaction forces and torques at various
operations are conveniently represented in MATLAB, locations on the system can be determined, plotted and
resulting in a compact and optimized code. The simula- analyzed. Also, the actuator forces and torques required
tion results are plotted and discussed in detail. to produce desired behavior can be determined.
The primary aim of this work is to demonstrate the mod-
Keywords: Bond Graph, Modeling, Simulation, Quick
eling of the dynamics of the quick-return mechanism
return mechanism and then discuss the results obtained using a realistic
computer model of this system to simulate its response
by means of appropriate programs written in MATLAB.
The system equations derived from the bond graphs are
1 Introduction in a form which can directly be used to write the pro-
gram code for the simulation of the system in MATLAB.
Bond graphs [1] are graphical tools which can be used to The results of the simulation are then discussed in detail.
model and analyze the dynamic behavior of various mul-
ti-energy systems. The application of this technique re-
sults in a number of advantages, as described in [1, 2]. 2 Modeling
In addition, causality establishes a cause and effect rela-
tionship between the flow and effort elements of the In this work, we develop a multibond graph model of
bond. The notion of causality, apart from aiding with the the quick return mechanism, representing the translation
and rotation for each rigid link of the system. The quick 2.1 The Quick Return Mechanism
return mechanism is treated as one comprising of five
rigid moving links, having relative motion with respect The quick return mechanism is used to convert the rota-
to each other, and also with respect to a stationary 0 tional motion of the crank to translational motion of the
frame. This inertial frame does not possess any transla- slider. The return stroke of the slider takes a lesser span
tional or rotational motion. The center of mass of each of time to complete, as compared to the forward stroke.
link is assumed to be located at its corresponding geo- The individual components of the mechanism are con-
metric center. The translational effect is concentrated at sidered as rigid bodies [13-14], connected at the joints.
the center of mass of each link, while the rotational ef- The mechanism consists of a crank (link 1), connected to
fect is considered in the inertial frame itself by consider- a slider (link 2) which slides along a rocker arm (link 3).
ing the inertia tensor for each link about its respective The rocker arm is connected to the final sliding link
center of mass and expressed in the inertial frame. Ref- (link 5) through a connecting rod (link 4). Reference
erence frames are fixed on each link, using the Denavit- frames are fixed on each link, that is, frame 1 is fixed on
Hartenberg convention [6]. Fig. (1) shows the different link 1, frame 2 on link 2 and so on. A fixed inertial
links of the system, along with their corresponding cen- frame 0, whose origin coincides with that of frame 1, is
ters of mass and associated frames. The links are inter- chosen. However, it will neither translate nor rotate. C1,
connected with each other by imposing suitable con- C2, C3, C4 and C5 are the centers of mass of the respec-
straints at the joints. A problem of differential causality tive links. As mentioned earlier, the Denavit-Hartenberg
[1, 2] arises while modeling the joints, which is rectified convention is used to fix the reference frames on each of
by introducing suitable stiffness and damping elements the links. The dynamics of the system of fig. (1) is mod-
while imposing constraints. These elements make the eled in the multibond graph shown in fig. (2). The model
model more realistic by bringing in the effects of com- depicts rotational as well as translational dynamics for
pliance and dissipation at the joints, within definable each link in the system. The left side of the bond graph
tolerance limits. The multibond graph is then causaled is associated with the rotational dynamics of the system,
and the codes for simulation in MATLAB are directly while the right side represents the translational dynamics.
derived from it. Bond graph modeling of rigid multibody We restrict any relative translational motion between the
systems has been presented using scalar bond represen- origin of the inertial frame 0 and point O1 on link 1 by
tations [7, 8], and using multibond graphs [9-15]. How- applying source of flow Sf as zero. Similarly, the relative
ever, a clear approach derived from first principles has motion at point A, distinguished by A1 on link 1, A2 on
been presented in [14]. The simulation results obtained link 2 and A3 on link 3, is restricted by applying a source
are then plotted, analyzed and discussed in detail. of flow Sf equal to zero. A similar approach is used to
model the joints between link 3 and 4, at point A33, and
between link 3 and the inertial frame, at point O3. The
slider, which is link 2, is constrained to translate along
the X3 direction only. In order to constrain the motion of
link 2 in the moving frame 3, all the flows and efforts
which were initially expressed in the inertial frame are
expressed in frame 3. This is achieved by using a modu-
0
lated transformer, having a modulus equal to − ⎡ 3 R ⎤ .
⎣ ⎦
Differential causality which arises at the joint of the
sliding link (link 2) is eliminated by setting the K3C(1,1)
element in the stiffness matrix [K3C] between frames 0
and 3 as zero. The multibond is split into its scalar com-
ponents and the translation along Y3 and Z3 directions is
constrained by applying source of flow Sf equal to zero
for these components. A similar treatment is carried out
while modeling the joint at the second slider (link 5). It
is also observed that link 2 and link 3 have the same
angular velocity, as seen from the inertial frame. Thus,
in order to constrain any relative angular motion be-
tween link 2 and link 3, we apply a source of flow Sf
  equal to zero, between the rotational sides of these two
links. The viscoelastic elements used at the joints are
Fig. 1: The Quick Return Mechanism represented by C and R elements. For the simulation,
the crank is made to rotate about point O1 with an angu-
The mechanism has five moving links with reference lar velocity equal to 5 rad/s. This is achieved by apply-
frames fixed to each link. The centers of mass of each of ing a source of flow Sf = 5 rad/s to the fixed end of link
the links are also shown. X0-Y0 and X0’-Y0’ are inertial 1 about the z direction.
frames of reference.
 Table-1: Link parameters used for simulation
Con-
Rock nect-
Cran Slid- Slid-
er ing
k er er 2
Arm Rod

Mass 0.5 kg 0.1 kg 0.7 kg 0.3 kg 0.1kg

Lengt
h 0.01 0.01
0.2 m 0.7 m 0.4 m
along m m
x-axis
Lengt
h 0.01 0.01 0.01 0.01 0.01
along m m m m m
y-axis
Lengt
h 0.01 0.01 0.01 0.01 0.01
along m m m m m
z-axis

Table-2: Values of stiffness and damping of different

couplings used for simulation
Stiffness , K Damping , R

Translational
coupling be- K01 = 100000 R01 = 20
tween crank N/m N·s/m
and fixed frame

Translational
coupling be- K12 = 100000 R12 =20
tween slider N/m N·s/m
and crank

Translational
coupling be-
K03 = 100000 R03 =20
tween rocker
N/m N·s/m
arm and fixed
frame

Translational
coupling be- K3C = 100000 R3C =20
tween slider N/m N·s/m
Fig. 2: Multibond graph model of the quick return and rocker arm
mechanism
Rotational
In this model, the inertia elements are denoted by I, and coupling be- K23r = 100 R23r= 0.5
the stiffness and damping elements by K and R tween slider N·m/rad N·m·s/rad
respectively. and rocker arm

3 Simulation Results Rotational

coupling be-
tween source of KC = 100 RC = 100
The Bond Graph model has been simulated using
constant angu- N·m/rad N·m·s/rad
MATLAB. The simulation results are shown in this
lar velocity and
section. Table 1 shows the values of the various link
crank
parameters used in the simulation, whereas table 2
shows the values of the stiffness and damping of the
various couplings used.
 Translational
coupling be-
K34 = 100000 R34 = 20
tween rocker
N/m N·s/m
arm and con-
necting rod

Translational
coupling be-
K45 = 100000 R45 = 20
tween connect-
N/m N·s/m
ing rod and
slider 2

Translational
coupling be- K5C = 100000 R5C = 20
tween slider 2 N/m N·s/m
and fixed frame Fig. 4: Variation of angular momentum of link 1 with
time.

3.1 Simulation Plots As indicated in the Fig. (4), the z component of angular
momentum is almost constant with time. This is due to
the constant value of source of flow Sf imposed on the
The simulation plots for the different links have been
crank. The small variation can be explained as effects of
discussed in the following sections.
the compliance element KC. The initial transients which
arise due to the sudden imposition of source of flow Sf
3.1.1 Dynamics of the Crank die down after a brief initial period due to damping.

Fig. 3: Variation of orientation matrix of link 1 with

time

The crank rotates in the x-y plane. The orientation ma- Fig. 5: Variation of force at fixed end of crank with time
trix ⎡⎣ 10 R ⎤⎦ represents the projection of unit vectors of
frame 1 (the crank) on frame 0 (the stationary frame).
The unit vectors in x and y directions move in a circular
path, while the unit vector in the z direction is stationary.
Fig. (3) shows the variation of the unit vectors with time
in a 3D plot.
Fig. 6: Variation of force at fixed end of crank with rad/sec at the crank. This source of flow generates a
crank angle torque which rotates the crank and hence transmits
forces to the rest of the links to actuate the entire mech-
The crank is pivoted at the point O1. As the crank rotates, anism. Fig. (7) and fig. (8) indicate the variation of this
the motion of the crank causes forces to be transmitted generated torque with time and crank angle respectively.
on the fixed end. Fig. (5) shows the variation of these The crank starts from an initial angle of 90o or 1.57
forces with time and fig. (6) shows the variation with radians. As seen in the force plots, we observe initial
crank angle. In both the figures, we observe initial tran- transients in the torque plots. This is observed at 1.57
sients for a short period after the system is set into mo- radians in fig. (8), due to the sudden imposition of a
tion. These transients occur due to sudden application of source of flow. Torque is maximum at 1.57 radians.
source of flow Sf = [0 0 5]' rad/s on the crank from t = 0. This is required to start the motion of the system from
In the simulation, an initial crank angle of 90° or 1.57 its initial position of rest. These transients gradually die
radians is given. So, in fig. (6), the transients start at an down and we observe the forced part of the response.
angle of 1.57 radians, that is, the angle from which the Thus, the actuator specifications can be determined.
crank starts rotating. The forces reach a maximum value
during this initial period and gradually, the transients die 3.1.2 Dynamics of the Slider
down and thereafter, force variation continues in a peri-
odic manner. The x, y and z components of the force
vector are plotted separately. The plot lines at the ends
show a discontinuity. This is perhaps not part of the re-
sponse, but due to the angle exceeding one revolution.
These force plots can be used as a guideline for design
purposes, as these represent the actual dynamic reaction
forces at the crank shaft axis during its working.

Fig. 9: Variation of force at pin connecting crank and

slider at point A2 on slider with time

Fig. 7: Variation of torque Tc on the crank with time

Fig. 10: Variation of force at pin connecting crank and

slider at point A2 on slider with crank angle

The point at which the crank is connected to the slider is

Fig. 8: Variation of torque Tc on the crank with crank point A2. Due to interaction with other links such as the
angle slider, forces get transmitted to the point A2. Fig. (9)
indicates the variation of the forces with time and fig.
We have considered a constant source of flow of 5 (10) indicates the variation of forces with crank angle.
Transients exist for a short period after the system is set The variation of the translational momentum of the
into motion, after which it gradually dies down due to rocker arm is indicated in fig. (12). The x-component
damping. We observe the forced response of the system and y-component of the translational momentum vector
thereafter. The plot lines at the ends show a discontinui- vary with time and the z-component is 0, as the rocker
ty due to the angle exceeding one revolution. The z- arm rotates about the z-axis. Transients are observed at
component is zero and the forces exist in the x-y plane the beginning of the plot. These transients occur due to
only. sudden imposition of the source of flow, or motion, on
the crank. As observed in the previous plots, the transi-
3.1.3 Dynamics of the Rocker Arm ents die down after some time and we observe a periodic
variation of the translational momentum, as shown in the
figure, which represents the forced response of the sys-
tem. An interesting observation is that the peak values of
the translational momentum for both x and y coordinates,
have greater magnitude during the return stroke (part of
the graph with positive values of momentum) as com-
pared with the forward stroke (part of the graph with
negative values of momentum). This shows the quick
return phenomenon.

Fig. 11: Variation of coordinates of center of mass of

rocker arm with time

The center of mass of the rocker arm oscillates about the

z-axis. Fig. (11) indicates the variation of the compo-
0
nents of the position vector r of center of mass of
0 C3
the rocker arm. The z-component is constant as the cen-
ter of mass of the rocker arm oscillates in the x-y plane,
about the z-axis. From fig. (11), it can be observed that Fig. 13: Variation of force at fixed end of rocker arm
the time variation of the position of the x and y co- with time
ordinates of the center of mass is steeper during the re-
turn stroke and less steep during the forward stroke. This
implies that the return stroke takes a shorter duration to
complete as compared with the forward stroke. This is
due to the quick-return action of the mechanism.

Fig. 14: Variation of force at fixed end of rocker arm

with crank angle

The rocker arm is pivoted at the point O3. As link 3 is

set into motion by the crank, forces develop at the point
O3. The force at O3 is plotted in fig. (13) and fig. (14).
Fig. 12: Variation of translational momentum of rocker Fig. (13) and fig. (14) shows the variation of the force
arm with time with time and with crank angle respectively. As ob-
served in the other force plots, the initial transients gen-
erated due to a sudden imposition of a source of flow, the center of mass is steeper during the return stroke and
dies down after a period of time. The maximum forces less steep during the forward stroke. This is due to the
that occur at the point O3 can be determined and the quick-return action of the mechanism.
system can be designed and materials be decided accord-
ingly.

3.1.4 Dynamics of the Connecting Rod

Fig. 17: Variation of the components of translational

momentum with time

As seen from the previous graphs, it is seen that the slid-

er 2 reciprocates in the x direction. Its motion in the y
and z directions is constrained. This is reflected in fig.
Fig. 15: Variation of orientation matrix of link 4 with (17) which shows the variation of the x component of
time momentum with time. It is seen that the slider 2 attains
higher peak values of the x component of translational
The orientation matrix ⎡⎣ 40 R ⎤⎦ represents the projection
momentum during the return stroke as compared to the
of unit vectors of frame 4 (the connecting rod) on frame peak values in the forward stroke. This is due to the fact
0 (the stationary frame). The unit vectors in x and y di- that it has a greater velocity during the return stroke as
rections move in a circular arc, while the unit vector in compared to the forward stroke, due to the quick return
the z direction is stationary. Fig. (15) shows the varia- action of the mechanism.
tion of the unit vectors with time in a 3D plot.

3.1.5 Dynamics of Slider 2 (Link 5)

Fig. 18: Variation of force at the center of mass of slider

2 with time
Fig. 16: Variation of coordinates of center of mass of
slider 2 with time

The center of mass of the slider 2 reciprocates along the

x direction. The y-component and z-component is con-
stant as the center of mass of the slider 2 reciprocates
along the x-axis. From fig. (16), it can be observed that
the time variation of the position of the x co-ordinates of
 [5] S.S. Rattan, Theory of Machines, Second edition
Tata McGraw Hill Publishing Company Limited, New
Delhi, 2005
[6] J.J. Craig, Introduction to Robotics: Mechanics and
Control, 3rd ed., Pearson Education Inc, 2005.
[7] A.A. Zeid, J.L. Overholt, “Singularly Perturbed
Formulation: Explicit Modeling of Multibody Systems,”
J. Franklin Institute, vol. 332B, No. 1, pp. 21-45, 1995.
[8] D. Karnopp, “Understanding Mulitbody Dynamics
Using Bond Graph Representations,” J. Franklin Insti-
tute, vol. 334B, No. 4, pp. 641-642, 1997
[9] M. J. L. Tiernego and A. M. Bos, “Modelling the
Dynamics and Kinematics of Mechanical Systems with
Multibond Graphs,” Journal of The Franklin Institute,
vol. 319, No. 1/2, pp. 37-50, 1985
Fig. 19: Variation of force at center of mass of slider 2
with crank angle [10] A. M. Bos and M. J. L. Tiernego, “Formula Manip-
ulation in the Bond Graph Modelling and Simulation of
This point O5 is a point on the slider 2 that moves along Large Mechanical Systems,” Journal of The Franklin
parallel to the X5 axis. As the crank rotates, due to the Institute, vol. 319, No. 1/2, pp. 51-65, 1985.
interaction of link 5 with the other links, forces are
transmitted to point O5. Fig. (18) shows the variation of [11] E. P. Fahrenthold and J. D. Wargo, “Vector and
these forces with time and fig. (19) shows the variation Tensor Based Bond Graphs for Physical Systems Mod-
of the forces with crank angle. As seen in the force plots eling,” J. Franklin Institute, vol.328, No. 5/6, pp. 833-
of the other links, initial transients can be observed. The 853, 1991
initial transients die down and thereafter, the force varia- [12] J.Jang and C. Han, “Proposition of a Modeling
tion is found to be repetitive. The maximum force occur- Method for Constrained Mechanical Systems Based on
ing at this point can be determined and the joint can be the Vector Bond Graph,” J. Franklin Institute, vol.335B,
accordingly designed. No. 3, pp. 451-469, 1998
[13] Anand Vaz and Shinichi Hirai, “Modeling a Hand
4 Conclusions Prosthesis with Word Bond Graph Objects”, Proceed-
ings of The International Conference on Integrated
The quick return mechanism has been modeled and Modeling & Analysis in Applied Control & Automation
its dynamics simulated in MATLAB using the Bond (IMAACA 2004), Oct. 28-31, 2004, Genoa, Italy, vol. 2,
Graph approach. The codes for simulation have pp. 58-67
been directly obtained from the Bond Graph model.
[14] Anand Vaz, “Lecture notes: Bond Graph Modeling
This approach has not only enabled us to conven-
for Rigid Body Dynamics,” 2003
iently extract information regarding the kinematic
aspects of each link of the mechanism, but also the [15] Germán Filippini, Diego Delarmelina, Jorge Pagano,
dynamic aspects, which is very important from the Juan Pablo Alianak, Sergio Junco and Norberto Nigro
design point of view. “Dynamics of Multibody Systems With Bond Graphs”
Mecánica Computacional Vol XXVI pp. 2943-2958,
2007

References
[1] A.Mukherjee, R. Karmakar, Modeling and Simula-
tion of Engineering Systems Through Bondgraphs,
Narosa Publishing House, New Delhi, 2000
[2] D.C. Karnopp, D.L. Margolis, and R.C. Rosenberg,
System Dynamics: Modeling and Simulation of Mecha-
tronic Systems third edition, Wiley-interscience, 2000
[3] K. J. Waldron, G. L. Kinzel, “Kinematics, Dynamics,
and Design of Machinery,” 2nd ed., John Wiley & Sons,
Inc., USA, 2004
[4] www.mathworks.com