Chapter 4

Forms of the Dynamic Equations

 The problems in dynamics can be classified into:
o Direct Dynamic Problem: Forces are the input. Solve the system as set of 2nd order
differential equations
o Inverse Dynamic Problem: Motion is the input. Solve the system as set of algebraic
equations
 In this chapter, we will present a systematic method for deriving the equations of motion of inter-
connected mechanical systems.
 Solving these equations will be presented later.

Direct Dynamic Problem

Equations of
Force/ Motion Motion
Torque
Inverse Dynamic Problem
Equations of
Motion
Motion Force/
Torque

1
D’Alembert’s Principle
Fiy m i ai y

Mi Ji d2i/dt2
Fix
mi aix

Jean le Rond d'Alembert

(1717-1783) (Source:
Wikipedia)

For an unconstrained mass moving in a plane, the following equations govern the motion:
𝑚𝑖 𝑎𝑖 𝑥 = 𝐹 𝑖 𝑥
} Newton’s Equations
𝑚𝑖 𝑎𝑖 𝑦 = 𝐹 𝑖 𝑦
𝐽𝑖 𝜃̈ 𝑖 = 𝑀𝑖 Euler’s Equation

Isaac Newton (1642-1727) Leonhard Euler (1707-1783)

(Source: Wikipedia)
The left hand sides of these equations are the inertia (effective) force.

2
Classical Kinematics
The classical approach depends on combining free body diagrams and kinematic equation to derive
equations of the motion of the mechanical system, which will tend to be second order differential
equations whose number is equal to the degree of freedom of the system.

While we should be comfortable with the process from earlier dynamic courses, this process becomes
increasingly more complex if we deal with multiple interconnected bodies. In the following, we present
an overview of the process.

3
Example
Develop equation for the dynamics of a carriage carrying a stick (inverted pendulum). Disregard the
effects of the carriage and wheels dynamics (they provide motion (kinematic) input only).
We will solve the problem using a sequence of steps.
1. Free Body Diagrams

3
M3
J3 d23/dt2
Y1
Y2
3
m3a3x
3
R2 O mg
1
X2 m3a3y
X 2
3 3
1
O F23x O

F23y

Based on the free body diagram above, we can obtain the Newton’s Equations for body 3:
𝑚3 𝑎3 𝑥 = 𝑚3 𝑅̈ 3 𝑥 = 𝐹𝑥 23
𝑚3 𝑎3 𝑦 = 𝑚3 𝑅̈ 3 𝑦 = 𝐹𝑦 23 − 𝑚3 𝑔
The Euler’s Equation is obtained by taking the moment around the center of mass of body 3:
𝑖 𝑗 𝑘
3 3
𝑙 𝑙
𝐽3 𝜃̈ 3 𝑘 = 𝑀3 𝑘 + 𝑟 3 × 𝐹 23 = 𝑀3 𝑘 + ||− cos 𝜃 3 − sin 𝜃 3 0||
2 2
23
𝐹𝑥 𝐹𝑦 23 0
3 3
3 ̈3 3 23 𝑙 3 23 𝑙
𝐽 𝜃 = 𝑀 + 𝐹𝑥 sin 𝜃 − 𝐹𝑦 cos 𝜃 3
2 2
Note: We used the local vector connecting O and the center of mass of body 3 for the Euler’s Equation.

4
2. Kinematics
The above equations are not sufficient (why?). They have to be complemented by kinematics.
From kinematics (assuming that input motion 𝑅 2 𝑥 is fully known), the center of mass of link 2 can be
expressed using these equations:
𝑙3
𝑅3𝑥 = 𝑅2𝑥 + cos 𝜃 3
2
𝑙3
𝑅3𝑦 = sin 𝜃 3
2
Differentiating,
𝑙3
𝑅̇ 3 𝑥 = 𝑅̇ 2 𝑥 − 𝜃̇ 3 sin 𝜃 3
2
𝑙3
𝑅̇ 3 𝑦 = 𝜃̇ 3 cos 𝜃 3
2
𝑙3 3 2
𝑙3
𝑅̈ 3 𝑥 = 𝑅̈ 2 𝑥 −𝜃̈ 3 3
sin 𝜃 − (𝜃 ) ̇ cos 𝜃 3
2 2
𝑙3 2𝑙
3
𝑅̈ 3 𝑦 = 𝜃̈ 3 cos 𝜃 3 − (𝜃̇ 3 ) sin 𝜃 3
2 2

5
3. Combining the Equations
Substituting the acceleration equations into Newton’s Equations,
𝑚3 𝑅̈ 3 𝑥 = 𝐹𝑥 23
𝑚3 𝑅̈ 3 𝑦 = 𝐹𝑦 23 − 𝑚3 𝑔
The result is,
𝑙3 2𝑙
3
𝑚3 (𝑅̈ 2 𝑥 −𝜃̈ 3sin 𝜃 3 − (𝜃̇ 3 ) cos 𝜃 3 ) = 𝐹𝑥 23
2 2
𝑙3 2𝑙
3
𝑚3 (𝜃̈ 3 cos 𝜃 3 − (𝜃̇ 3 ) sin 𝜃 3 ) = 𝐹𝑦 23 − 𝑚3 𝑔
2 2
The reactions can be then substituted into the Euler Equation
𝑙3 𝑙3
𝐽3 𝜃̈ 3 = 𝑀3 + 𝐹𝑥 23 sin 𝜃 3 − 𝐹𝑦 23 cos 𝜃 3
2 2
The result is,
𝐽3 𝜃̈ 3 = 𝑀3
𝑙3 2𝑙
3
𝑙3
+ (𝑚3 (𝑅̈ 2 𝑥 −𝜃̈ 3 sin 𝜃 3 − (𝜃̇ 3 ) cos 𝜃 3 )) sin 𝜃 3
2 2 2

𝑙3 3 2
𝑙3 𝑙3
− (𝑚 (𝜃̈ 3
3 3 ̇
cos 𝜃 − (𝜃 ) sin 𝜃 ) + 𝑚 𝑔) cos 𝜃 3
3 3
2 2 2
Simplifying and rearranging to have the acceleration terms on the left hand side,
2
𝑙3 𝑙3
(𝐽 + 𝑚 ( ) ) 𝜃̈ 3 = 𝑀3 + 𝑚3 (𝑅̈ 2 𝑥 sin 𝜃 3 − 𝑔 cos 𝜃 3 )
3 3
2 2

 The above equation is a second-order differential equation that can be solved to obtain 𝜃 3 and 𝜃̇ 3 .
 Once the angular acceleration is determined, the linear accelerations of the center of mass of body 3
and the reaction forces can be calculated.
 The accelerations can be integrated in terms of time to derive the velocity and displacement
equations.

6
Homework
Derive the equations of motion for this system. M3
Both links are subject to external torques as
shown below. Gravity acts in the vertical down
direction. 3
M2
2

7
D’Alembert’s Principle and Newton-Euler Equation

ri dVi (d2r/dt2)
Yi
Y i
ui X
i

ri Oi
Ri

X
Newton Equations
By equating the inertia forces to the applied forces

∫ 𝑟̈ 𝑖 𝑑𝑚𝑖 = ∫ 𝜌𝑖 𝑟̈ 𝑖 𝑑𝑉 𝑖 = 𝐹 𝑖
𝑉𝑖 𝑉𝑖

Fi is the vector of the applied forces acting on body i

From classical kinematics,
r̈ i = R̈i + ωi × (ωi × ui ) + αi × ui
ui = Ai u̅i
If the reference point is selected to be the center of mass of the body, one can conclude,

∫ 𝜌𝑖 ui 𝑑𝑉 𝑖 = ∫ 𝜌𝑖 Ai u̅i 𝑑𝑉 𝑖 = Ai ∫ 𝜌𝑖 u̅i 𝑑𝑉 𝑖 = 0
𝑉𝑖 𝑉𝑖 𝑉𝑖

Substituting in the acceleration equation

∫ 𝜌𝑖 R̈i 𝑑𝑉 𝑖 = 𝐹 𝑖
𝑉𝑖

R̈i 𝑚𝑖 = 𝐹 𝑖

8
Euler Equation
The moment of the inertia force of a particle about the reference point

∫ ui ×̈ 𝑟 𝑖 𝑑𝑚𝑖 = ∫ 𝜌𝑖 ui × r̈ i 𝑑𝑉 𝑖 = 𝑀𝑖 𝑅
𝑉𝑖 𝑉𝑖

This is a one equation in the direction that is normal to the plane.

∫ 𝜌𝑖 (ui × (R̈i + (ωi × (ωi × ui )) + (αi × ui ))) 𝑑𝑉 𝑖 = 𝑀𝑖 𝑅

𝑉𝑖

If the reference point is selected to be the center of mass of the body, one can conclude,

∫ 𝜌𝑖 ui × (αi × ui ) 𝑑𝑉 𝑖 = 𝑀𝑖 𝑅
𝑉𝑖

WHY???
 ωi × (ωi × ui ) is parallel to ui

 The term, ∫𝑉 𝑖 𝜌𝑖 ui × R̈i 𝑑𝑉 𝑖 , will disappear because the integration is about the center of mass
In the case of planar motion,

2 2
𝜃̈ 𝑖 ∫ 𝜌𝑖 (𝑥̅ 𝑖 + 𝑦̅ 𝑖 ) 𝑑𝑉 𝑖 = 𝜃̈ 𝑖 𝐽𝑖 = 𝑀𝑖
𝑉𝑖

9
Constrained Dynamics
 Joints or specified motion trajectories add restrictions to the motion of the system.
 For modeling the dynamics of a multi-body system, selection of coordinate systems is an important
part of setting the equations.
 Each joint is associated with:
o Loss of degrees of freedom and
o A constraint force/moment (reaction forces/moments).
 The number of constraint forces is equal to the number of constraint equations.
 For example, a hinge is associated with two force constraints and the loss of the two translations.

10
Example:

3
M3

X
2

2
Y

X
3
M2

Y1 2

2
X1

A F23x A

F23y M3 J3
M2
J2 F23y m3y m3a3x
C C F23x
m2a2x
2
mg
m2a2y m3g
m3 a 3 y
2 2
O F12x O

F12y

11
Using Principles of Classical Kinematics
Dynamic Equilibrium of Link 2
Newton’s Equations
𝑚2 𝑅̈ 2 𝑥 = 𝐹𝑥 12 − 𝐹𝑥 23
𝑚2 𝑅̈ 2 𝑦 = 𝐹𝑦 12 − 𝐹𝑦 23 − 𝑚2 𝑔
It may be useful to express the Newton’s equations in matrix form
Euler’s Equations (Moment is taken around center of mass)
𝑖 𝑗 𝑘 𝑖 𝑗 𝑘
2 2 2 2
𝑙 𝑙 𝑙 𝑙
𝐽2 𝜃̈ 2 𝑘 = 𝑀2 𝑘 + ||− cos 𝜃 2 − sin 𝜃 2 0|| + || cos 𝜃 2 sin 𝜃 2 0||
2 2 2 2
𝐹𝑥 12 𝐹𝑦 12 0 −𝐹𝑥 23 −𝐹𝑦 23 0
𝑙 𝑙 𝑙 𝑙
𝐽2 𝜃̈ 2 = 𝑀2 + 𝐹𝑥 12 sin 𝜃 2 − 𝐹𝑦 12 cos 𝜃 2 + 𝐹𝑥 23 sin 𝜃 2 − 𝐹𝑦 23 cos 𝜃 2
2 2 2 2
Similarly, Dynamic Equilibrium of Link 3
𝑚3 𝑅̈ 3 𝑥 = 𝐹𝑥 23
𝑚3 𝑅̈ 3 𝑦 = 𝐹𝑦 23 − 𝑚3 𝑔
𝐽3 𝜃̈ 3 = 𝑀3
We have:
 Six equations
 Ten unknowns: six accelerations and 4 reaction forces
These equations are complemented by kinematic equations. Based on classical kinematics, the
acceleration equations are:
𝑅̈ 2 𝑥
𝑅̈ 2 = { ̈ 2 } = ω2 × (ω2 × u2 CO ) + α2 × u2 CO
𝑅 𝑦
𝑅̈ 2 𝑥 2𝑙
2 2 𝑙2 2
𝑅̈ 2 = { ̈ 2 } = −(θ̇2 ) {𝑐𝑜𝑠𝜃2 } + θ̈2 {−𝑠𝑖𝑛𝜃2 }
𝑅 𝑦 2 𝑠𝑖𝑛𝜃 2 𝑐𝑜𝑠𝜃
𝑅̈ 3 𝑥
𝑅̈ 3 = { ̈ 3 } = ω2 × (ω2 × u2 AO ) + α2 × u2 AO
𝑅 𝑦
𝑅̈ 3 𝑥 2 2
𝑅̈ 3 = { ̈ 3 } = −(θ̇2 ) 𝑙 2 {𝑐𝑜𝑠𝜃2 } + θ̈2 𝑙 2 {−𝑠𝑖𝑛𝜃2 }
2
𝑅 𝑦 𝑠𝑖𝑛𝜃 𝑐𝑜𝑠𝜃
These accelerations are equal to the four constraint equations.
The system can be then reduced to two differential equations in terms of 𝜃 2 and 𝜃 3 and their derivatives.
Homework: Reduce the above equations to the two equations of motion of the system.
12
Computational Kinematics
Equations of Motion
Objective: Form the equations into a standardized form that can be automated

The equations of motion can be expressed as a set of differential equations.

𝑀𝑞̈ = 𝑄𝑒 + 𝑄𝑐
where, M is the mass matrix
Qe is the vector of applied forces
Qc is the vector of constraint (reaction) forces

13
As we did in Chapter 3, the kinematics of a mechanical system can be described using a number of
algebraic equations that represent the constraints imposed on the system.

These equations are related to the independent constraint forces of the system.
𝐶(𝑞, 𝑡) = 0
Differentiating the above equation,
𝐶𝑞 𝑞̇ + 𝐶𝑡 = 0
Or,
𝐶𝑞 𝑞̇ = −𝐶𝑡
Differentiating one more time,
𝐶𝑞 𝑞̈ + (𝐶𝑞 𝑞̇ )𝑞 𝑞̇ + 2𝐶𝑞𝑡 𝑞̇ = −𝐶𝑡𝑡
Rearranging,
𝐶𝑞 𝑞̈ = − ((𝐶𝑞 𝑞̇ )𝑞 𝑞̇ + 2𝐶𝑞𝑡 𝑞̇ + 𝐶𝑡𝑡 ) = 𝑄𝑑
Already discussed in Chapter 3

14
Return to the example

3
M3

X
2

2
Y

X
3
M2

Y1 2

2
X1

Similar to earlier work, we need to start by the free body diagrams and the associated equation. The six
equilibrium equations are:
𝑚2 𝑅̈ 2 𝑥 = 𝐹𝑥 12 − 𝐹𝑥 23
𝑚2 𝑅̈ 2 𝑦 = 𝐹𝑦 12 − 𝐹𝑦 23 − 𝑚2 𝑔
𝑙 𝑙 𝑙 𝑙
𝐽2 𝜃̈ 2 = 𝑀2 + 𝐹𝑥 12 sin 𝜃 2 − 𝐹𝑦 12 cos 𝜃 2 + 𝐹𝑥 23 sin 𝜃 2 − 𝐹𝑦 23 cos 𝜃 2
2 2 2 2
3 ̈3 23
𝑚 𝑅 𝑥 = 𝐹𝑥
𝑚3 𝑅̈ 3 𝑦 = 𝐹𝑦 23 − 𝑚3 𝑔
𝐽3 𝜃̈ 3 = 𝑀3

15
Rearranging these equations in the canonized (standard) form,
𝑀𝑞̈ = 𝑄𝑒 + 𝑄𝑐
𝑇
where, 𝑞 = [𝑅𝑥2 𝑅𝑦2 𝜃2 𝑅𝑥3 𝑅𝑦3 𝜃3]
We will get:
𝑚2 0 00 0 0
0 𝑚2 00 0 0
0 0 𝐽2 0 0 0
𝑀=
0 0 0 𝑚3 0 0
0 0 0 0 𝑚3 0
[ 0 0 0 0 0 𝐽3 ]
0
−𝑚2 𝑔
𝑄𝑒 = 𝑀2
0
−𝑚3 𝑔
{ 𝑀3 }
𝐹𝑥 12 − 𝐹𝑥 23
𝐹𝑦 12 − 𝐹𝑦 23
12 𝑙 2 12 𝑙 2 23 𝑙 3 23 𝑙
𝑄𝑐 = 𝐹𝑥 sin 𝜃 − 𝐹𝑦 cos 𝜃 + 𝐹 𝑥 sin 𝜃 − 𝐹𝑦 cos 𝜃 3
2 2 2 2
𝐹𝑥 23
𝐹𝑦 23
{ 0 }

16
The constraint equations are:
R2 + A2 u̅2O
𝐶={ 2 } = 01 (4x1)
−R − A2 u̅2A + R3
where,
2⁄ ) (𝑙 2 ⁄2) 2
u̅2O = {−(𝑙 2 } u̅2A = { } 𝐴2 = [cos 𝜃2 − sin 𝜃 2 ]
0 0 sin 𝜃 cos 𝜃 2

Velocity:
𝑇
Differentiating with respect to q vector 𝑞 = [𝑅𝑥2 𝑅𝑦2 𝜃2 𝑅𝑥3 𝑅𝑦3 𝜃3] yields the Jacobian matrix.
1 0 +(𝑙 2 ⁄2) sin 𝜃 2 0 0 0
𝐼 A2 𝜃 u̅2O 0 0 0 1 −(𝑙2 ⁄2) cos 𝜃 2 0 0 0
𝐶𝑞 = [ ]= (4x6)
−𝐼 −A2 𝜃 u̅2A 𝐼 0 −1 0 (𝑙 2 ⁄2) sin 𝜃 2 1 0 0
[0 2
−1 −(𝑙 ⁄2) cos 𝜃 2
0 1 0]
𝐶𝑡 = 0
Therefore,
𝐶𝑞 𝑞̇ = −𝐶𝑡
Or,
𝑅̇𝑥2
1 0 +(𝑙 2 ⁄2) sin 𝜃 2 0 0 0 𝑅̇𝑦2
0 1 −(𝑙 2 ⁄2) cos 𝜃 2 0 0 0 𝜃̇ 2
=0
−1 0 (𝑙 2 ⁄2) sin 𝜃 2 1 0 0 𝑅̇𝑥3
[ 0 −1 −(𝑙 2 ⁄2) cos 𝜃 2 0 1 0] 𝑅̇𝑦3
{ 𝜃̇ 3 }

𝑅𝑥2 − (𝑙 2 ⁄2) cos 𝜃 2

𝑅𝑦2 − (𝑙 2 ⁄2) sin 𝜃 2
1
Therefore, C= =0
−𝑅𝑥2 − (𝑙 2 ⁄2) cos 𝜃 2 − 𝑅𝑥3
2 2 2 3
{ −𝑅𝑦 − (𝑙 ⁄2) sin 𝜃 − 𝑅𝑦 }

17
Acceleration:
Differentiating one more time,
𝐶𝑞 𝑞̈ + (𝐶𝑞 𝑞̇ )𝑞 𝑞̇ + 2𝐶𝑞𝑡 𝑞̇ = −𝐶𝑡𝑡
Rearranging,
𝐶𝑞 𝑞̈ = − ((𝐶𝑞 𝑞̇ )𝑞 𝑞̇ + 2𝐶𝑞𝑡 𝑞̇ + 𝐶𝑡𝑡 ) = 𝑄𝑑

In this case,
𝐶𝑡𝑡 = 𝐶𝑞𝑡 = 0
𝑅𝑥2̇
1 0 +(𝑙 2 ⁄2) sin 𝜃 2 0 0 0 𝑅𝑦̇
2 𝑅𝑥2̇ + 𝜃 2̇ (𝑙 2 ⁄2) sin 𝜃 2
0 1 −(𝑙 2 ⁄2) cos 𝜃 2 0 0 0 𝜃 2̇ 𝑅𝑦2̇ − 𝜃 2̇ (𝑙 2 ⁄2) cos 𝜃 2
𝐶𝑞 𝑞̇ = =
−1 0 (𝑙 2 ⁄2) sin 𝜃 2 1 0 0 𝑅𝑥3̇ −𝑅̇ 𝑥2 + 𝜃 2̇ (𝑙 2 ⁄2) sin 𝜃 2 + 𝑅𝑥3̇
[ 0 −1 −(𝑙 2 ⁄2) cos 𝜃 2 0 1 0] 𝑅 3̇
{−𝑅̇ 𝑦2 − 𝜃 2̇ (𝑙 2 ⁄2) cos 𝜃 2 + 𝑅𝑦̇ }
3
𝑦
{𝜃 3̇ }
𝑅𝑥2̇ 2
(𝜃 2̇ ) (𝑙 2 ⁄2) cos 𝜃 2
0 0 +𝜃 2̇ (𝑙 2 ⁄2) cos 𝜃 2 0 0 0 𝑅𝑦 2̇
2
0 0 +𝜃 2̇ (𝑙 2 ⁄2) sin 𝜃 2 0 0 0 𝜃 2̇ (𝜃 2̇ ) (𝑙 2 ⁄2) sin 𝜃 2
(𝐶𝑞 𝑞̇ )𝑞 𝑞̇ = =
+𝜃 2̇ (𝑙 2 ⁄2) cos 𝜃 2 0 0 𝑅𝑥3̇
2
0 0 0 (𝜃 2̇ ) (𝑙 2 ⁄2) cos 𝜃 2
[0 0 +𝜃 2̇ (𝑙 2 ⁄2) sin 𝜃 2 0 0 0 ] 𝑅𝑦3̇ 2
{ (𝜃 2̇ ) (𝑙 2 ⁄2) sin 𝜃 2 }
{𝜃 3̇ }
Since, 𝐶𝑞 𝑞̈ = 𝑄𝑑 , we get:

𝑅𝑥2̈ 2
𝑅𝑦2̈ (𝜃 2̇ ) (𝑙 2 ⁄2) cos 𝜃 2
1 0 +(𝑙 2 ⁄
2) sin 𝜃 0 0 2
0 2
0 2
1 −(𝑙 ⁄2) cos 𝜃 0 02
0 𝜃 2̈ (𝜃 2̇ ) (𝑙 2 ⁄2) sin 𝜃 2
=−
−1 0 (𝑙 2 ⁄2) sin 𝜃 2 1 0 0 𝑥𝑅 3̈ 2
(𝜃 2̇ ) (𝑙 2 ⁄2) cos 𝜃 2
2 2
[ 0 −1 −(𝑙 ⁄2) cos 𝜃 0 1 0] 𝑅 3̈ 2
𝑦
{ (𝜃 2̇ ) (𝑙 2 ⁄2) sin 𝜃 2 }
{ 𝜃 3̈ }
Or,
2
𝑅̈𝑥2 + (𝑙 2 ⁄2) 𝜃 2̈ sin 𝜃 2 + (𝑙 2 ⁄2)(θ̇2 ) cos 𝜃 2
2
𝑅̈𝑦2 − (𝑙 2 ⁄2) 𝜃 2̈ cos 𝜃 2 + (𝑙 2 ⁄2)(θ̇2 ) sin 𝜃 2
2 =0
−𝑅̈𝑥2 + (𝑙 2 ⁄2) 𝜃̈ 2 sin 𝜃 2 + (𝑙 2 ⁄2)(θ̇2 ) cos 𝜃 2 + 𝑅̈𝑥3
2
{ −𝑅̈𝑦2 − (𝑙 2 ⁄2) 𝜃 2̈ cos 𝜃 2 + (𝑙 2 ⁄2)(θ̇2 ) sin 𝜃 2 + 𝑅̈𝑦3 }
This acceleration balance equations for the two joints. We can reach the same result by differentiating
the constraint equation twice. The above approach however automates the process.

18
Homework:
Derive the equations for the dynamics a Revolute / Prismatic arm that we studied in Chapter 3 using the
computational kinematic approach discussed in this section.

19
Lagrange Multipliers

Joesph-Louis Lagrange (1736-1813)

(Source Wikipedia)

 Lagrange Multiplier Method presents an approach for including generalized constraint forces with
the Newton-Euler equations (Augmenting the Equations).
 The process starts by creating the Jacobian Matrix of the Constraint Matrix.
 The method is based on augmenting the equations of the system dynamics by the constraint
equations. Additional unknowns are also added.
 In this case, the constraint forces that restrict the system. These forces are lumped together as the
Lagrange Multipliers such that:
𝑄𝑐 = 𝑈𝜆
 The equations of motion based on the free body diagrams can be rearranged as,
𝑀𝑞̈ − 𝑄𝑐 = 𝑄𝑒
 Substituting the first equation in the second one and expressing the results in matrix form,
𝑞̈
[𝑀 𝑈] { } = 𝑄𝑒
𝜆
 We claim that U is related Cq since constraint force and velocities have duality.
 If the above bullet is true, the overall system is expanded to become:
𝑀 𝐶𝑞 𝑇 𝑞̈ 𝑄
[ ] { } = { 𝑒}
𝐶𝑞 0 𝜆 𝑄𝑑

20
Continue the previous example.

3
M3

X
2

2
Y

X
3
M2

Y1 2

2
X1

From the previous step, we ended with:

𝑀𝑞̈ = 𝑄𝑒 + 𝑄𝑐
Or,

𝑅𝑥2̈ 𝐹𝑥 12 − 𝐹𝑥 23
𝑚2 0 0 0 0 0 0 𝐹𝑦 12 − 𝐹𝑦 23
0 𝑚2 0 0 0 0 𝑅𝑦2̈ −𝑚2 𝑔 2 2 2 2
0 0 𝐽2 0 0 0 𝜃 2̈ 𝑀 2 12 𝑙 2 12 𝑙 2 23 𝑙 3 23 𝑙
= + 𝐹𝑥 sin 𝜃 − 𝐹𝑦 cos 𝜃 + 𝐹 𝑥 sin 𝜃 − 𝐹𝑦 cos 𝜃 3
0 0 0 𝑚3 0 0 𝑅𝑥3̈ 0 2 2 2 2
0 0 0 0 𝑚3 0 −𝑚3 𝑔 𝐹𝑥 23
𝑅𝑦3̈
[ 0 0 0 0 0 𝐽3 ] { 𝑀3 } 𝐹𝑦 23
{𝜃 3̈ } { }
0
and,
𝐶𝑞 𝑞̈ = 𝑄𝑑
These equations can be rewritten as,
𝑅𝑥2̈ 2
𝑅𝑦2̈ (𝜃 2̇ ) (𝑙 2 ⁄2) cos 𝜃 2
1 0 +(𝑙 2 ⁄
2) sin 𝜃 0 0 0 2
2
0 2 2
1 −(𝑙 ⁄2) cos 𝜃 0 0 0 (𝜃 2̇ ) (𝑙 2 ⁄2) sin 𝜃 2
𝜃 2̈
=−
−1 0 (𝑙 2 ⁄2) sin 𝜃 2 1 0 0 𝑥
2
𝑅 3̈
(𝜃 2̇ ) (𝑙 2 ⁄2) cos 𝜃 2
[ 0 −1 −(𝑙 2 ⁄2) cos 𝜃 2 0 1 0] 𝑅 3̈ 2
𝑦
{ (𝜃 2̇ ) (𝑙 2 ⁄2) sin 𝜃 2 }
{ 𝜃 3̈ }

21
Rearranging the first equation,
𝑀𝑞̈ − 𝑄𝑐 = 𝑄𝑒
𝐹𝑥 12 − 𝐹𝑥 23
2 0 0 0 0
𝑚 0 𝐹𝑦 12 − 𝐹𝑦 23 0
2
0 𝑚2 0 0 0 0 2 2 2 2 −𝑚 𝑔
0 0 𝐽2 0 0 0 12 𝑙 12 𝑙 23 𝑙 23 𝑙 2
2 2 3 3
𝑞̈ − 𝐹𝑥 2 sin 𝜃 − 𝐹𝑦 2 cos 𝜃 + 𝐹𝑥 2 sin 𝜃 − 𝐹𝑦 2 cos 𝜃 = 𝑀
0 0 0 𝑚3 0 0 0
0 0 0 0 𝑚3 0 𝐹𝑥 23 −𝑚3 𝑔
[ 0 0 0 0 0 𝐽3 ] 𝐹𝑦 23 { 𝑀3 }
{ 0 }
Qc matrix can be rearranged as,
𝑚2 0 0 0 0 0 1 0 −1 0 0
0 𝑚2 0 0 0 0 𝐹𝑥 12 −𝑚 2
𝑔
0 1 0 −1 12
0 0 𝐽2 0 0 0
𝑞̈ − (𝑙 2⁄ )
2 sin 𝜃 2 −(𝑙 2 ⁄2) cos 𝜃 2 (𝑙 2 ⁄2) sin 𝜃 2 −(𝑙 2 ⁄2) cos 𝜃 2 𝐹𝑦 = 𝑀 2
0 0 0 𝑚3 0 0 0 0 1 0 𝐹𝑥 23 0
3
0 0 0 0 𝑚3 0 0 0 0 1 23 −𝑚 𝑔
[ ] {𝐹𝑦 }
[ 0 0 0 0 0 𝐽3 ] 0 0 0 0 { 𝑀 }
3

Or,
𝑚2 0 0 0 0 0 1 0 −1 0 0
0 𝑚2 0 0 0 0 𝐹𝑥 12 −𝑚 2
𝑔
0 1 0 −1 12
0 0 𝐽2 0 0 0 (𝑙 2⁄ )
2 sin 𝜃 2 −(𝑙 2 ⁄2) cos 𝜃 2 (𝑙 2 ⁄2) sin 𝜃 2 2⁄ )
−(𝑙 2 cos 𝜃 2 𝐹𝑦 𝑀 2
𝑞̈ + =
0 0 0 𝑚3 0 0 0 0 1 0 𝐹𝑥 23 0
3
0 0 0 0 𝑚3 0 0 0 0 1 23
𝐹𝑦 } −𝑚 𝑔
[ ] {
[ 0 0 0 0 0 𝐽3 ] 0 0 0 0 { 𝑀3 }

If we transpose the coefficient matrix of the constraint forces (Qc), it will be

1 0 (𝑙 2 ⁄2) sin 𝜃 2 0 0 0
0 1 −(𝑙 2 ⁄2) cos 𝜃 2 0 0 0
−1 0 (𝑙 2 ⁄2) sin 𝜃 2 1 0 0
[ 0 −1 −(𝑙 2 ⁄2) cos 𝜃 2 0 1 0]
This is very similar to Cq,
1 0 +(𝑙 2 ⁄2) sin 𝜃 2 0 0 0
0 1 −(𝑙 2 ⁄2) cos 𝜃 2 0 0 0
−1 0 (𝑙 2 ⁄2) sin 𝜃 2 1 0 0
[ 0 −1 −(𝑙 2 ⁄2) cos 𝜃 2 0 1 0]
Can you identify any difference?

22
This means that Cq can represent the force constraints, which is what we started from (loss of motion is
associated with reaction forces)
𝑚2 0 0 0 0 0 12 0
0 𝑚 2 0 0 0 0 −𝐹𝑥 −𝑚2 𝑔
12
0 0 𝐽 2
0 0 0 −𝐹𝑦 𝑀2
𝑞̈ + 𝐶𝑞 𝑇 =
0 0 0 𝑚 3 0 0 𝐹𝑥 23 0
3
0 0 0 0 𝑚 3
0 𝐹 23 −𝑚 𝑔
{ 𝑦 }
[ 0 0 0 0 0 𝐽 ] 3 { 𝑀 }
3

The equations can then be assembled as,

𝑞̈
[𝑀 𝐶𝑞 𝑇 ] { } = 𝑄𝑒
𝜆
where  are the Lagrange multipliers,
𝜆 = [𝐹𝑥 12 𝐹𝑦 12 𝐹𝑥 23 𝐹𝑦 23 ]
The overall system is expanded to become:
𝑀 𝐶𝑞 𝑇 𝑞̈ 𝑄
[ ] { } = { 𝑒}
𝐶𝑞 0 𝜆 𝑄𝑑

23
Homework:
Determine the equations of motion and acceleration equations (based on kinematic constraints) for an
internal combustion engine using the Lagrange Multiplier approach.
Notes:
 Neglect gravity
 Neglect friction
 Center of mass of body 4 (crank) is at O

B
M4
4 3

O
P2
A
2

24
Generalization of the Lagrange Multiplier
From the system equations,
𝑀 𝐶𝑞 𝑇 𝑞̈ 𝑄
[ ] { } = { 𝑒}
𝐶𝑞 0 𝜆 𝑄𝑑
The first equation,
𝑀𝑞̈ + 𝐶𝑞 𝑇 𝜆 = 𝑄𝑒
This equation has two sets of unknowns:
𝑞̈ System accelerations
𝜆 Lagrangian Multipliers, which represent the constraint forces

𝑞̈ can be represented in terms of 𝑞̈ 𝑖 (independent accelerations). These are the accelerations of the
independent motions of the system. The relation between 𝑞̈ and 𝑞̈ 𝑖 is:
𝑞̈ = 𝐵𝑖 𝑞̈ 𝑖 + 𝛾𝑖
Bi velocity transformation matrix
𝛾𝑖 matrix representing quadratic terms of acceleration (normal accelerations). 𝛾𝑖 is a function of
velocities.

25
Before proceeding, we need to look at the above equation closely. Remember that the acceleration
equation is,
𝐶𝑞 𝑞̈ = 𝑄𝑑
Since we are studying dynamics Cq is not square. Actually, the number rows is always less than the
number of columns (why?). We can rearrange the 𝑞̈ matrix as follows,
𝑞̈
[𝐶𝑞𝑑 𝐶𝑞𝑖 ] { 𝑑 } = 𝑄𝑑
𝑞̈ 𝑖
Or,
𝐶𝑞𝑑 𝑞̈ 𝑑 + 𝐶𝑞𝑖 𝑞̈ 𝑖 = 𝑄𝑑
Cqd is a square matrix. Therefore, we can solve the above equation as,
𝑞̈ 𝑑 = 𝐶𝑞𝑑 −1 (𝑄𝑑 − 𝐶𝑞𝑖 𝑞̈ 𝑖 )
−1
𝑞̈ 𝑑 = −𝐶𝑞𝑑 −1 𝐶𝑞𝑖 𝑞̈ 𝑖 + 𝐶𝑞𝑑 𝑄𝑑
If we add 𝑞̈ 𝑖 to the left hand side to form 𝑞̈ ,
𝑞̈ −1
𝐶 −1 𝑄𝑑
𝑞̈ = { 𝑑 } = {−𝐶𝑞𝑑 𝐶𝑞𝑖 } 𝑞̈ 𝑖 + { 𝑞𝑑 }
𝑞̈ 𝑖 𝐼 0
Or,
𝑞̈ = 𝐵𝑖 𝑞̈ 𝑖 + 𝛾𝑖
Therefore,
−1
𝐵𝑖 = {−𝐶𝑞𝑑 𝐶𝑞𝑖 }
𝐼
𝐶𝑞𝑑 −1 𝑄𝑑
𝛾𝑖 = { }
0

26
Substituting the acceleration relation into the equation of motion. However, the above equations should
be rearranged before using it in the dynamic formulations (why?).
𝑀(𝐵𝑖 𝑞̈ 𝑖 + 𝛾𝑖 ) + 𝐶𝑞 𝑇 𝜆 = 𝑄𝑒
Or (in matrix form) after rearranging,
𝑞̈
[𝑀𝐵𝑖 𝐶𝑞 𝑇 ] { 𝑖 } = 𝑄𝑒 − 𝑀𝛾𝑖
𝜆
𝑞̈
Matrix [𝑀𝐵𝑖 𝐶𝑞 𝑇 ] is square (why?), which means that { 𝑖 } can be solved as,
𝜆
𝑞̈ 𝑖 𝑇 −1
{ } = [𝑀𝐵𝑖 𝐶𝑞 ] {𝑄𝑒 − 𝑀𝛾𝑖 }
𝜆

27
In the previous example,

3
M3

X
2

2
Y

X
3
M2

Y1 2

2
X1

Kinematic analysis:
Since
𝐶𝑞 𝑞̈ = − ((𝐶𝑞 𝑞̇ )𝑞 𝑞̇ + 2𝐶𝑞𝑡 𝑞̇ + 𝐶𝑡𝑡 ) = 𝑄𝑑
Based on the kinematic results presented earlier,
1 0 +(𝑙 2 ⁄2) sin 𝜃 2 0 0 0
0 1 −(𝑙 2 ⁄2) cos 𝜃 2 0 0 0
𝐶𝑞 =
−1 0 (𝑙 2 ⁄2) sin 𝜃 2 1 0 0
2
[ 0 −1 −(𝑙 ⁄2) cos 𝜃 0 1 0] 2
2
(𝜃 2̇ ) (𝑙 2 ⁄2) cos 𝜃 2
2
(𝜃 2̇ ) (𝑙 2 ⁄2) sin 𝜃 2
𝑄𝑑 = − 2
(𝜃 2̇ ) (𝑙 2 ⁄2) cos 𝜃 2
2
{ (𝜃 2̇ ) (𝑙 2 ⁄2) sin 𝜃 2 }
2
In this case, 𝑞𝑖 = {𝜃 3 }, which are the third and sixth rows of q
𝜃
𝑅𝑥 2
𝑅𝑦 2
Therefore, 𝑞𝑑 = ,
𝑅𝑥 3
3
{𝑅𝑦 }
We will divide Cq accordingly,
(𝑙 2 ⁄2) sin 𝜃 2 0
−(𝑙 2 ⁄2) cos 𝜃 2 0
𝐶𝑞𝑖 =
(𝑙 2 ⁄2) sin 𝜃 2 0
[−(𝑙 2 ⁄2) cos 𝜃 2 0]
1 0 0 0
𝐶𝑞𝑑 = [ 0 1 0 0]
−1 0 1 0
0 −1 0 1
28
We now have all the matrices needed to create this equation we derived earlier
𝑞̈
𝑞̈ = { 𝑑 } = 𝐵𝑖 𝑞̈ 𝑖 + 𝛾𝑖
𝑞̈ 𝑖
and,
−1
𝐵𝑖 = {−𝐶𝑞𝑑 𝐶𝑞𝑖 }
𝐼

𝐶𝑞𝑑 −1 𝑄𝑑
𝛾𝑖 = { }
0
Homework:
 Derive the matrices 𝐵𝑖 , 𝛾𝑖 , and. Rearrange them to fit the standard q matrix.
 Derive [𝑀𝐵𝑖 𝐶𝑞 𝑇 ]. Show that it is invertible
State your observations

29
Embedding Techniques
 While the above work organizes the equations, it does not resolve the main issues we have seen so
far in terms of deriving a manageable system.
 Embedding Techniques can be used to eliminate constraint equations
 The resulting set of equations will have the same number of equations as the degrees of freedom of
the system

Start with,
𝑞̈
[𝑀𝐵𝑖 𝐶𝑞 𝑇 ] { 𝑖 } = 𝑄𝑒 − 𝑀𝛾𝑖
𝜆
Pre-multiply both sides by BiT
𝑇 𝑞̈
[𝐵𝑖 𝑇 𝑀𝐵𝑖 𝑇 𝑇
𝐵𝑖 𝑇 𝐶𝑞 ] { 𝑖 } = 𝐵𝑖 𝑄𝑒 − 𝐵𝑖 𝑀𝛾𝑖
𝜆
𝑇
It is interesting to notice that, 𝐵𝑖 𝑇 𝐶𝑞 = 0 Why??
Therefore, the first equation becomes:
[𝐵𝑖 𝑇 𝑀𝐵𝑖 ]𝑞̈ 𝑖 = 𝐵𝑖 𝑇 𝑄𝑒 − 𝐵𝑖 𝑇 𝑀𝛾𝑖
Or
𝑀𝑖 𝑞̈ 𝑖 = 𝑄𝑖

where,
𝑀𝑖 = [𝐵𝑖 𝑇 𝑀𝐵𝑖 ] Generalized Inertia Matrix
𝑄𝑖 = 𝐵𝑖 𝑇 𝑄𝑒 − 𝐵𝑖 𝑇 𝑀𝛾𝑖 Generalized Force Matrix
The final objective is to reach this equation:
𝑞̈ 𝑖 = 𝑀𝑖 −1 𝑄𝑖

Homework:
𝑇
Show that 𝐵𝑖 𝑇 𝐶𝑞 = 0 for this system
3

M3
Y

X
2

2
Y

3
M2

Y1 2

2
X1

30
(skip Amalgamated Formulation)
Modeling Examples
Modeling Open-Chain Systems (skip Equilibrium of the Separate Bodies)
Equilibrium of the Separate Bodies

F3y
F3x
P X3

Y3 3
R3
Y2 M3
2 X2
M
Y1
R2

O 2 X1

F3y
A F23x F3x
P
F23y
M2
C C
M3
m2g
m3g 3
2
F12x A F23x
O

F12y
F23y

31
The process starts by considering the dynamic equilibrium of link 2.
The force equations for this link are,
𝐹12 𝑥 −𝐹 23 𝑥 = 𝑚2 𝑎𝑥2
𝐹12 𝑦 − 𝐹 23 𝑦 − 𝑚3 𝑔 = 𝑚2 𝑎𝑦2

Applying D’Alembert’s Principle

𝑖 𝑗 𝑘 𝑖 𝑗 𝑘
(𝑀2 )𝑘 + |−𝑙𝑂2 cos 𝜃 2 −𝑙𝑂2 sin 𝜃 2 2
0| + |𝑙𝐴 cos 𝜃 2 𝑙𝐴2 sin 𝜃 2 0| = (𝐽2 𝜃̈ 2 )𝑘
𝐹12 𝑥 𝐹12 𝑦 0 −𝐹 23 𝑥 −𝐹 23 𝑦 0
Expanding,
𝑀2 + (𝐹12 𝑥 𝑙𝑂2 sin 𝜃 2 − 𝐹12 𝑦 𝑙𝑂2 cos 𝜃 2 ) + (𝐹 23 𝑥 𝑙𝐴2 sin 𝜃 2 − 𝐹 23 𝑦 𝑙𝐴2 cos 𝜃 2 ) = 𝐽2 𝜃̈ 2

The same approach can be repeated for link 3.

The force equations for this link are,
𝐹 23 𝑥 = 𝑚3 𝑎𝑥3
𝐹 23 𝑦 − 𝑚3 𝑔 = 𝑚3 𝑎𝑦3
Taking the moment for link 3 about its center of mass and applying D’Alembert’s Principle
𝑖 𝑗 𝑘 𝑖 𝑗 𝑘
(𝑀3 )𝑘 + |𝑙𝑃3 cos 𝜃 3 𝑙𝑃3 sin 𝜃 3 3
0| + |−𝑙𝐴 cos 𝜃 3 −𝑙𝐴3 sin 𝜃 3 0| = 𝐽3 𝜃̈ 3 𝑘
𝐹3𝑥 𝐹3𝑦 0 𝐹 23 𝑥 𝐹 23 𝑦 0
Expanding,
𝑀3 + (−𝐹𝑥 3 sin 𝜃 3 + 𝐹𝑦 3 cos 𝜃 3 )𝑙𝑃3 + (𝐹 23 𝑥 𝑙𝐴3 sin 𝜃 3 − 𝐹 23 𝑦 𝑙𝐴3 cos 𝜃 3 ) = 𝐽3 𝜃̈ 3

32
 2 3
Based on the previous example and assuming that, 𝑙𝑂2 = 𝑙𝐴2 = 𝑙 ⁄2 𝑎𝑛𝑑 𝑙𝐴3 = 𝑙𝑃3 = 𝑙 ⁄2, the system
equations are:
𝑀𝑞̈ = 𝑄𝑒 + 𝑄𝑐
Or,
𝑚2 0 0 0 0 0 𝑅 2̈ 𝑥
0 𝑚 2
0 0 0 0 𝑅 2̈ 𝑥
0 0 𝐽2 0 0 0 𝜃 2̈
𝑅 3̈ 𝑥
3
0 0 0 𝑚 0 0
3
0 0 0 0 𝑚 0 𝑅 3̈ 𝑥
[ 0 0 0 0 0 𝐽3 ]
{ 𝜃 2̈ }
0
−𝑚2 𝑔
𝑀2
= 𝐹𝑥 3
𝐹𝑦 3 − 𝑚3 𝑔
3 3
{𝑀3 + (−𝐹𝑥 sin 𝜃 3 + 𝐹𝑦 cos 𝜃 3 )(𝑙 3 ⁄2)}
𝐹𝑥 12 − 𝐹𝑥 23
𝐹𝑦 12 − 𝐹𝑦 23
(𝐹𝑥 12 sin 𝜃 2 − 𝐹𝑦 12 cos 𝜃 2 )(𝑙 2 ⁄2) + (𝐹𝑥 23 sin 𝜃 2 − 𝐹𝑦 23 cos 𝜃 2 )(𝑙 2 ⁄2)
+
𝐹𝑥 23
𝐹𝑦 23
{ (𝐹𝑥 23 sin 𝜃 3 − 𝐹𝑦 23 cos 𝜃 3 )(𝑙 3 ⁄2) }

33
Kinematic Analysis:
To assemble the equations of the system, we need to start from the constraint equations
𝑅 2 + 𝐴2 𝑢̅2 𝑂
𝐶={ 2 }=0
−𝑅 − 𝐴2 𝑢̅2𝐴 + 𝑅 3 + 𝐴3 𝑢̅3𝐴
Therefore,
𝐼 𝐴2𝜃 𝑢̅2 𝑂 0 0
𝐶𝑞 = [ 3 3 ] 4X6
−𝐼 −𝐴2𝜃 𝑢̅2𝐴 𝐼 𝐴𝜃 𝑢̅ 𝐴
2
In this case, 𝑞𝑖 = {𝜃 3 }, which are the third and sixth rows of q
𝜃
𝑅𝑥 2
𝑅𝑦 2
Therefore, 𝑞𝑑 = ,
𝑅𝑥 3
3
{𝑅𝑦 }
We will divide Cq accordingly,
𝐴2𝜃 𝑢̅2 𝑂 0
𝐶𝑞𝑖 = [ 2 2 3 3 ]
−𝐴𝜃 𝑢̅ 𝐴 𝐴𝜃 𝑢̅ 𝐴
𝐼 0
𝐶𝑞𝑑 = [ ]
𝐼 𝐼
We now have all the matrices needed to create this equation we derived earlier
𝑞̈
𝑞̈ = { 𝑑 } = 𝐵𝑖 𝑞̈ 𝑖 + 𝛾𝑖
𝑞̈ 𝑖
and,
−1
𝐵𝑖 = {−𝐶𝑞𝑑 𝐶𝑞𝑖 }
𝐼

𝐶𝑞𝑑 −1 𝑄𝑑
𝛾𝑖 = { }
0
These equations result in the following system of equation,
𝑀𝑖 𝑞̈ 𝑖 = 𝑄𝑖
where,
2
𝑙2 𝑙3
𝑚2 ( 2 ) + 𝑚3 (𝑙2 )2 + 𝐽2 𝑚3 (𝑙 2 ) ( 2 ) cos(𝜃 2 − 𝜃 3 )
𝑀𝑖 = [𝐵𝑖 𝑇 𝑀𝐵𝑖 ] = [ 2 ]Generalized Inertia Matrix
𝑙3 𝑙3
𝑚3 (𝑙 2 ) ( 2 ) cos(𝜃 2 − 𝜃 3 ) 𝑚3 ( 2 ) + 𝐽3

𝑄𝑖 = 𝐵𝑖 𝑇 𝑄𝑒 − 𝐵𝑖 𝑇 𝑀𝛾𝑖 Generalized Force Matrix

Remember that our objective is not to derive this equation. We are satisfied with having the matrices.
We are doing this to convince ourselves that the approach is sound.

34
Homework:
For the 2R manipulator:
𝑇
 Prove that 𝐵𝑖 𝑇 𝐶𝑞 = 0
 Derive 𝑀𝑖 and 𝑄𝑖

35
Modeling Closed-Chain Systems (skip Equilibrium of the Separate Bodies)
Equilibrium of the Separate Bodies

F3 B
3 f
A

M2
4
4
M
2

Q2
O C

The approach is similar to what is used in the case of the open-chain systems. In this case, we are
developing the equations in terms of 𝑞𝑖 = 𝜃 2 since it is the input motion.

36
 A
F32x

C2 F23y
M2
m2g
Q2 12
O F x
F12y
The force equations for this link are,
𝐹12 𝑥 −𝐹 23 𝑥 = 𝑚2 𝑎𝑥2
𝐹12 𝑦 − 𝐹 23 𝑦 − 𝑚3 𝑔 = 𝑚3 𝑎𝑦2

The moment equation about the center of mass is,

𝑖 𝑗 𝑘 𝑖 𝑗 𝑘
(𝑀2 )𝑘 + |−𝑙𝑂2cos 𝜃 2 −𝑙𝑂2 sin 𝜃 2 2
0| + |𝑙𝐴 cos 𝜃 2 𝑙𝐴2 sin 𝜃 2 0| = (𝐽2 𝜃̈ 2 )𝑘
𝐹12 𝑥 𝐹12 𝑦 0 −𝐹 23 𝑥 −𝐹 23 𝑦 0
Expanding,
𝑀2 + 𝐹12 𝑥 𝑙𝑂2 sin 𝜃 2 − 𝐹12 𝑦 𝑙𝑂2 cos 𝜃 2 + 𝐹 23 𝑥 𝑙𝐴2 sin 𝜃 2 − 𝐹 23 𝑦 𝑙𝐴2 cos 𝜃 2 = 𝐽2 𝜃̈ 2

37
Taking the moment for link 3 about its center of mass and applying D’Alembert’s Principle.

F3 B
23 F34x
F y C3 f

F23x A F34y
Q3 m3g
The force equations for this link are,
−𝐹 23 𝑥 + 𝐹 34 𝑥 + 𝐹 3 𝑥 = 𝑚3 𝑎𝑥3
𝐹 23 𝑦 + −𝐹 34 𝑦 + 𝐹 3 𝑦 − 𝑚3 𝑔 = 𝑚3 𝑎𝑦3
The moment equation about the center of mass is,
𝑖 𝑗 𝑘 𝑖 𝑗 𝑘
(𝑀3 )𝑘 + |−𝑙𝐴3 cos 𝜃 3 −𝑙𝐴3 sin 𝜃 3 3
0| + |𝑙𝐵 cos 𝜃 3 𝑙𝐵3
sin 𝜃 3 0| = 𝐽3 𝜃̈ 3 𝑘
𝐹 23 𝑥 𝐹 23 𝑦 0 −𝐹 34 𝑥 −𝐹 34 𝑦 0
Expanding,
𝑀3 + 𝐹 23 𝑥 𝑙𝐴3 sin 𝜃 3 − 𝐹 23 𝑦 𝑙𝐴3 cos 𝜃 3 + 𝐹 34 𝑥 𝑙𝐵3 sin 𝜃 3 − 𝐹 34 𝑦 𝑙𝐵3 cos 𝜃 3 = 𝐽3 𝜃̈ 3

38
The third set of equations can be obtained by studying links 4 together. Taking the moment about its
center of mass:

B
F34x

F34y
4

M4 m4g

Q4
C
F41x
F41y

The force equations for this link are,

−𝐹 41 𝑥 + 𝐹 34 𝑥 = 𝑚4 𝑎𝑥4
−𝐹 41 𝑦 + 𝐹 34 𝑦 − 𝑚4 𝑔 = 𝑚4 𝑎𝑦4
The moment equation about the center of mass is,
𝑖 𝑗 𝑘 𝑖 𝑗 𝑘
4
(𝑀4 )𝑘 + |𝑙𝐵4 cos 𝜃 4 4
𝑙𝐴𝐵sin 𝜃 4 0| + |−𝑙𝐶 cos 𝜃 4 −𝑙𝐶4
sin 𝜃 4 0| = 𝐽4 𝜃̈ 4 𝑘
𝐹 34 𝑥 𝐹 34 𝑦 0 −𝐹 41 𝑥 −𝐹 41 𝑦 0
Expanding,
𝑀4 + 𝐹 34 𝑥 𝑙𝐵4 sin 𝜃 4 − 𝐹 34 𝑦 𝑙𝐵4 cos 𝜃 4 − 𝐹 41 𝑥 𝑙𝐶4 sin 𝜃 4 + 𝐹 41 𝑦 𝑙𝐶4 cos 𝜃 4 = 𝐽4 𝜃̈ 4

Homeworkd
 Combine the above equations in this form:
𝑀𝑞̈ = 𝑄𝑒 + 𝑄𝑐
 Use Lagrange’s Multiplier and embedding techniques to reach:
𝑞̈ 𝑖 = 𝑀𝑖 −1 𝑄𝑖

39
Kinematic Analysis
To assemble the equations of the system, we need to start from the constraint equations.
We need to separate the depended variables from the independent variable, 𝜃 2 or,
𝑞𝑖 = { 𝜃 2 }

𝑅 2 + 𝐴2 𝑢̅2 𝑂
−𝑅 2 − 𝐴2 𝑢̅2𝐴 + 𝑅 3 + 𝐴3 𝑢̅3𝐴
𝐶= =0
−𝑅 3 − 𝐴3 𝑢̅3 𝐵 + 𝑅 4 + 𝐴4 𝑢̅4 𝐵
{ −𝑅 4 − 𝐴4 𝑢̅4 𝐶 + 𝑢1 𝐶 }
Therefore,
𝐼 𝐴2𝜃 𝑢̅2 𝑂 0 0 0 0
−𝐼 −𝐴2𝜃 𝑢̅2𝐴 𝐼 𝐴3𝜃𝑢̅3𝐴 0 0
𝐶𝑞 = 8X9
0 0 −𝐼 −𝐴3𝜃 𝑢̅3 𝐵 𝐼 𝐴4𝜃 𝑢̅4 𝐵
[0 0 0 0 −𝐼 −𝐴4𝜃 𝑢̅4 𝐶 ]
𝐶𝑡 = 0
𝐼 0 0 0 0
3 3
−𝐼 𝐼 𝐴𝜃 𝑢̅ 𝐴 0 0
𝐶𝑞𝑑 =
0 −𝐼 −𝐴𝜃 𝑢̅ 𝐵 𝐼 𝐴4𝜃 𝑢̅4 𝐵
3 3

[0 0 0 −𝐼 𝐴4𝜃 𝑢̅4 𝐵 ]
𝐴2𝜃 𝑢̅2 𝑂
2 2
𝐶𝑞𝑖 = −𝐴𝜃 𝑢̅ 𝐴
0
[ 0 ]
𝐶𝑞 𝑞̈ = −(𝐶𝑞 𝑞̇ )𝑞 𝑞̇ − 2𝐶𝑞𝑡 𝑞̇ − 𝐶𝑡𝑡 = 𝑄𝑑

𝐶𝑞𝑡 = 𝐶𝑡𝑡 = 0
Therefore,
𝐶𝑞 𝑞̈ = −(𝐶𝑞 𝑞̇ )𝑞 𝑞̇ = 𝑄𝑑

where,
𝑅̇ 2 + 𝜃̇ 2 𝐴2𝜃 𝑢̅2 𝑂
−𝑅̇ 2 − 𝜃̇ 2 𝐴2𝜃 𝑢̅2𝐴 + 𝑅̇ 3 + 𝜃̇ 3 𝐴3𝜃 𝑢̅3𝐴
𝐶𝑞 𝑞̇ =
−𝑅̇ 3 − 𝜃̇ 3 𝐴3𝜃 𝑢̅3 𝐵 + 𝑅̇ 4 + 𝜃̇ 4 𝐴4𝜃 𝑢̅4 𝐵
{ −𝑅̇ 4 − 𝜃̇ 4 𝐴4𝜃 𝑢̅4 𝐶 }
0 −𝜃̇ 2 𝐴2 𝑢̅2 𝑂 0 0 0 0
0 𝜃̇ 2 𝐴2 𝑢̅2𝐴 0 ̇3 3
−𝜃 𝐴 𝑢̅3𝐴 0 0
𝑄𝑑 = −(𝐶𝑞 𝑞̇ )𝑞 𝑞̇ = − 𝑞̇
0 0 0 𝜃̇ 3 𝐴3 𝑢̅3 𝐵 0 −𝜃̇ 𝐴 𝑢̅4 𝐵
4 4

[0 0 0 0 0 𝜃̇ 4 𝐴4 𝑢̅4 𝐶 ]

40
We now have all the matrices needed to create this equation we derived earlier
𝑞̈
𝑞̈ = { 𝑑 } = 𝐵𝑖 𝑞̈ 𝑖 + 𝛾𝑖
𝑞̈ 𝑖
and,
−1
𝐵𝑖 = {−𝐶𝑞𝑑 𝐶𝑞𝑖 }
𝐼

𝐶𝑞𝑑 −1 𝑄𝑑
𝛾𝑖 = { }
0
This equation should be rearranged before using it in the dynamic formulations:
𝐵𝑖 𝑇 (𝑀(𝐵𝑖 𝑞̈ 𝑖 + 𝛾𝑖 ) − 𝑄𝑒 ) = 0
𝐵𝑖 𝑇 𝑄𝐶 = 0
Expanding
𝐵𝑖 𝑇 𝑀(𝐵𝑖 𝑞̈ 𝑖 + 𝛾𝑖 ) − 𝐵𝑖 𝑇 𝑄𝑒 = 0
Or,
−1
𝑞̈ 𝑖 = (𝐵𝑖 𝑇 𝑀𝐵𝑖 ) 𝐵𝑖 𝑇 (𝑄𝑒 − 𝑀𝛾𝑖 )

41
Homework:
Determine the equations of motion and acceleration equations (based on kinematic constraints) for an
RP arm using the Embedding technique.
Notes:
 Include gravity
 Neglect friction
 Center of mass of body 3 (crank) is at its geometric center

Homework:
Determine the equations of motion and acceleration equations (based on kinematic constraints) for an
internal combustion engine using the Embedding technique.
Notes:
 Neglect gravity
 Neglect friction
 Center of mass of body 4 (crank) is at O

B
3
4

O
P
A
2

42
Summary
We went through this approach to show that the equations can be derived using the computational
approach.

Of course, when the system becomes more complex, intuition will be limited as the need for a
systematic approach becomes evident.

43