Department of Mechanical Engineering

MECH 411/501 Fall 21

Prof. F. H. Ghorbel

Dynamics & Control of

Mechanical Systems

Analytical Dynamics

Fall 2021
Copyright Fathi H. Ghorbel 2021
1
 Preview
• Newton s Law:
• Free body diagram (includes interacting internal forces)
• Vectorial Approach
• Forces and Accelerations difficult to obtain

• D Alembert s Principle:
Ø The principal of virtual work: statement of the Static equilibrium of a
mechanical system without considering the interacting forces
® ®
Ø Newton s 2nd Law: F- m a = 0 Static Problem

Ø Use the principal of virtual work to dynamic systems

Ø This permits the derivation of the equations of motion without
considering explicitly the interacting forces
Ø Still a Vectorial Approach
Fall 2021

Copyright Fathi H. Ghorbel 2021 2

 Preview (Continued)
• Lagrangian Dynamics or Analytical Dynamics
• Removes both objections to Newtonian Mechanics
• It permits the derivation of the equations of motion using 2
scalar quantities
üThe Kinetic Energy
üThe Potential Energy

• NO Free Body Diagrams

• NO Internal Forces

Fall 2021

Copyright Fathi H. Ghorbel 2021 3

 ANALYTICAL DYNAMICS
1. Generalized Coordinates & Degrees of Freedom
Consider system of N particles:
ü Each particle with mass: mi ; i=1, 2, …., N
→ → → →
ü Position vector: r i = x i i + y i j + zi k

If the rectangular coordinates are known functions of time,

xi = xi (t ), yi = yi (t ), zi = zi (t ) i = 1,2,..., N
€
then the position of the system is defined completely.
Coordinate Transformation:
Sometimes, it is more convenient to express position in a different set of coordinates, for
example, q1 , q2 ,...., qn where n = 3N
We write the relationship between the rectangular coordinates xi , yi , zi (i = 1,2,..., N )and the
new coordinates qk (k = 1, 2,..., n) with the following coordinate transformation :
x1 = x1 ( q1 , q2 ,..., qn )
y1 = y1 ( q1 , q2 ,..., qn )
z1 = z1 ( q1 , q2 ,..., qn )
x2 = x2 ( q1 , q2 ,..., qn )
y 2 = y2 ( q1 , q2 ,..., qn )
!
Fall 2021 z n = z n ( q1 , q2 ,..., qn )

Copyright Fathi H. Ghorbel 2021 4

Example 1: The planar motion of a particle

x = r cosq q = arctan( y, x)
y = r sin q r = (x2 + y 2 )
m

Let r = q1 , q = q2 Þ x = x(q1 , q2 ) = q1 cos q2 y

r
θ
y = y (q1 , q2 ) = q1 sin q2
x

Constraints:
Sometimes the particles are subject to constraints, restricting their freedom of motion.
Hence, the particles have constrained motion.

Example 2:
In the example above, the particle is constrained to move in the x-y plane only.
The constraint equation is: z=0

Fall 2021
Copyright Fathi H. Ghorbel 2021 5
Example 3: Two particles m1 & m2 are connected by a rigid, massless rod of length L.
z

L m1
(x1,y1,z1)

m2
x (x2,y2,z2)

The free motion of this system in space can be described in terms of the cartesian coordinates (x1, y1 , z1) of m1
and the corresponding coordinates (x2, y2 , z2) of m2 . The rod causes the separation between the particles to be a
constant distance L. Hence, the constraint equation is
( x1 - x2 ) 2 + ( y1 - y2 ) 2 + ( z1 - z2 ) 2 = L2 =
constant (*)

Conclusion: Not all six coordinates x1, y1 , z1, x2, y2 , z2 are independent. The motion is fully determined
by only five of these coordinates & equation (*) can be used to determine the sixth coordinate

The number 6-1=5Copyright

is Fathi
called the2021number of degrees of freedom.
Fall 2021
H. Ghorbel 6
Degrees of Freedom of a system & Generalized Coordinates
The number of d.o.f is the minimum number of coordinates which are used to specify the
configuration of the system minus the number of independent equations of constraints.

ü In general, if a system of N particles moving in a 3-D dimensional space is subject to c kinematical

constraints, then the motion of the system can be described completely by n-coordinates
q1 , q2 ,...., qn where
n = 3N - c

is the number of degrees of freedom of the system (d.o.f).

NOTE: The number of d.o.f is a characteristic of the system itself and does not depend upon the
particular set of coordinates used to describe its configuration. The n coordinates q1 , q2 ,...., qn are
known as the generalized coordinates.

Ø Generalized coordinates can refer to any of the commonly used coordinate systems. They
can also refer to any set of parameters which serve to specify the configuration of a system.

Ø Generalized coordinates may not always have physical meaning nor are they necessarily
unique.

Fall 2021

Copyright Fathi H. Ghorbel 2021 7

Example 4: (Continuation of Example 2)
z m1

θ (x1,y1,z1)
(x0,y0,z0)

y
ϕ
O
m2
(x2,y2,z2)
x
In this case, there are six coordinates (3N=6) and one equation of constraint (c=1). Therefore,

# of d.o.f = 6 – 1=5
The configuration of the system could be specified by

}
(x0, y0, z0) : position of the center of mass
θ, Φ : spherical coordinates 5 coordinates

Since each of these coordinates can vary freely without violating any constraints, there are no
equations of constraints.

NOTE: In many instances, it is preferable – as we have done so far - to choose independent generalized
coordinates so that
# of d.o.f = # of generalized coordinates . Hence, no equations of constraints.
Fall 2021
Copyright Fathi H. Ghorbel 2021 8
 Recapture
Ø System of N particles described by 3N cartesian coordinates
Ø c independent constraint equations
Hence, there are n= (3N-c) d.o.f
Ø The constrained system can be described by only n= (3N-c) of the coordinates and the remaining c
coordinates can be determined from the c-independent constraint equations.
Ø Sometimes, it is possible to find more convenient n= (3N-c) coordinates to describe the system.
Ø The n independent coordinates are called the generalized coordinates which are any set of the
n-parameters which serve to specify the configuration of the system

Ex: m1
y
(x1,y1) Free system: 2 particles of mass m1 & m2
L
θ (x1, y1) & (x2, y2) fully specify the free system (N=4)

(x0,y0)
Constraint eq: ( x2 - x1 ) 2 + ( y2 - y1 ) 2 = L2 (c=1) (*)

(x2,y2) m x → n=4-1=3 = d.o.f

2
O
Ø x1, y1, y2 describe the system without ambiguity, x2 is obtained from (*)
Ø They are independent while constraints are satisfied

Fall 2021

Copyright Fathi H. Ghorbel 2021 9

 Recapture (Continued)
Fix x1,y1 Fix x1,y2 Fix y1,y2

(x1,y1)

y y y

x x x

y2 moves freely y1 moves freely x1 moves freely

Ø Obviously (x0, y0,θ) are better generalized coordinates

Fall 2021

Copyright Fathi H. Ghorbel 2021 10

2. The Principle of Virtual Work
Consider a system of N particles with position vector
→ → → →
r i = x i i + y i j + z i k ; i = 1,...., N (3N coordinates)

subject to a constraint of the form

f ( x1 , y1 , z1 , x2 , y2 , z2 ,..., zn , t ) = c (*)

Definition: The virtual displacements δx1, δy1, δz1, ……, δ zn are defined as infinitesimal
changes in the coordinates x1, y1, z1, ……, zn .

They are consistent with the constraints of the system, take place without any change in time, so
that
f (x1 + δx1, y1 + δy1,z1 + δz1, x 2 + δx 2 ,...,zn + δzn ,t)= c (2∗)

but otherwise arbitrary.

€
t has not varied!

Fall 2021

Copyright Fathi H. Ghorbel 2021 11

2. The Principle of Virtual Work (Continued)

The virtual displacements obey the rules of differential calculus. Hence (2*) can be expanded
in a Taylor series about the positions x1, y1, z1, ……, zn as follows

N ! $
∂f ∂f ∂f
f (x1 , y1 , z1 , x2 , y2 , z2 ,..., zn ,t) + ∑# δ xi + δ yi + δ zi & + O(δ 2 ) = c (3∗)
i=1 " ∂xi ∂yi ∂zi % nonlinear terms

Ignoring the nonlinear terms in (3*) and compare with (*), we conclude that the virtual
displacements must satisfy

N
é ¶f ¶f ¶f ù
å ê ¶x d x + ¶y d y + ¶z d z ú = 0
i i i
i =1 ë i i i û

Fall 2021

Copyright Fathi H. Ghorbel 2021 12

Example 5: The two particles in Ex 3 &4 are subject to

( x1 - x2 ) 2 + ( y1 - y2 ) 2 + ( z1 - z2 ) 2 = L2
The virtual displacement satisfies

( x1 - x2 )(dx1 - dx2 ) + ( y1 - y2 )(dy1 - dy2 ) + ( z1 - z2 )(dz1 - dz2 ) = 0

In general, for a system of N particles subject to c constraints, hence possessing n=3N- c d.o.f,
only n virtual displacements are arbitrary

# of arbitrary virtual displacements = # of d.o.f

Virtual Work
Assume that each of the particles of the constrained system is acted on by a set of forces with resultant
® ®
R i (i = 1,....., N ) . When the system is in equilibrium, i.e. Ri = 0 , the virtual work performed by the resultant
®
forces on the ith particle over the virtual displacement vector d r i is

dWi = Ri • d ri = 0 i = 1,2,..., N
The virtual work of the entire system is
N 
  N
δWi = ∑δWi = ∑ Ri • δ ri = 0 system in equilibrium
so far trivial result! i=1 i=1

Fall 2021
Copyright Fathi H. Ghorbel 2021 13
Now let
→ !" !"
Ri = Fi + fi i =1, 2,..., N
applied forces constra int forces

The virtual work


 
 N N   
δW = ∑ Fi • δ r i + ∑ f i • δ r i (3*)

i=1    i=1  
virtual work virtual work
performed by performed by
applied forces constraint forces
Ø All of the systems of interest to us, when undergoing a virtual displacement consistent with the
constraints, have the virtual work performed by the constraint forces zero, that is,

∑ f • δr i i = 0 when δr i is consistent with constraints (4*)

i=1

N → workless constraints

⇒ ∑ F i • δri = 0
€
(3*) & (4*) Principle of Virtual
€
Work:
i=1
The work performed by the applied forces through virtual displacements
compatible with the system constraints is zero.
Example:
Fall 2021
Particle moving on a smooth surface
Copyright Fathi H. Ghorbel 2021 14
 Example:
 (i) Constraint forces (forces transmitted by the rod to the particles)
m2
δ r2 
  

f1 = − f 2 = − f ur
  
f2 (ii) Assume virtual displacements, δ r1 & δ r2
!
ur € m1 & m2 are constrained, hence displacement components along
the rigid rod must be equal.

    
€ → constraint eq.: ur • δ r1 = ur • δ r2
(iii) Virtual work
 
    
δ r1  δW = f1 • δ r1 + f 2 • δ r2
f1 
 

m1 = − f ur • δ r1 + f ur • δ r2
 
€ =0
€

Fall 2021
Copyright Fathi H. Ghorbel 2021 15
If the system is conservative (i.e. applied forces are conservative), then

N 
 
∑ Fi • δ ri = − δV = 0 (5∗)
variation in
 
i=1  the potential
virtual work energy

N
æ ¶V ¶V ¶V ö
V ( x1 , y1 , z1 , x2 , y2 , z2 ,..., xn , yn , zn ) Þ dV = å ç d xi + d yi + d zi ÷
i =1 è ¶xi ¶yi ¶zi ø

→(5*) becomes
N & ∂V N
∂V ∂V )
∑ (x
F δ
i xi

+

F

δ
yi yi
 
+

Fzi zi )

δ

= −∑('∂x i ∂y i ∂z δzi +* = 0
δ x + δ y + (6*)
i=1 i=1 i i i
cartesian components of Fi

Case 1: No constraints
i.e. δxi, δy
€ i, δzi (i=1,…, N) are all independent. For (6*) to hold for all possible values of the virtual
displacements

¶V ü
Fxi = - =0 ï
¶xi ï
¶V ï Equilibrium conditions for
Fyi = - =0 i = (1,2,..., N )ý
¶yi ï a system of N free particles
¶V ï
Fzi = - =0 ï
¶zi þ

Fall 2021
Copyright Fathi H. Ghorbel 2021 16
Case 2:
c constraints → n=3N-c d.o.f
Assume we have n independent generalized coordinates q1,q2,…,qn
N n
¶V
Virtual work : å F • dr = -variation
i id!V = -å dq
¶q
i =0
% "$"
i =1
# in
the potential
k =1 i
virtual work energy

n
∂V ∂V
∑ ∂q δ qi =0 ⇒ = 0 ; i = 1,2,...,n
k=1
 ∂qi
i independent

Example:
Two equal masses are connected by a
m1g
rigid massless bar. If all surfaces are
N1
m1 frictionless, solve for the force F2
required for static equilibrium of the
system
x1
a
45
m2g
x2
45 m2
F2
a N2
Fall 2021
Copyright Fathi H. Ghorbel 2021 17
 Solution
ü Assume that the masses do not leave the wall or floor, then the constraints N1 & N2 are
workless
ü Measure displacements x1 & x2 from the initial configuration

⇒ constraint equation: (a − x1 ) 2 + (a + x2 ) 2 = 2a 2
  
f ( x1, x2)

Ø virtual displacements consistent with the constraints satisfy

N
é ¶f ¶f ¶f ù
-2(a - x1 )d x1 + 2(a + x2 )d x2 = 0 å ê
i =1 ë ¶xi
d xi +
¶yi
d yi +
¶zi
d ziú =0
û
at equilibrium x1 = x2 = 0
Þ d x1 = d x2

Ø Apply principle of virtual work

0

   
F
∑ i • δ r = 0 ⇒ m1
gδ x1
+ m 2
g • δ x2 + F2δ x2 = 0

Þ F2 = -m1 g

Fall 2021
Copyright Fathi H. Ghorbel 2021
18
Example: (7.1 Textbook)

Mass on a smooth inclined plane attached to a

spring of stiffness k . Determine the equilibrium
position.

tio n
os i
ep

−mg

c
ren
e


ref
−kq

€ q

€ δq

q

!" θ €
f €
€
€
Ø Denote the displacement of the spring in the equilibrium position by the vector q as measured parallel
to the surface from the unstretched position.

Ø This system has 1 d.o.f along q

Fall 2021
Copyright Fathi H. Ghorbel 2021 19
 !"
Ø The constraint force f is workless. Why?

Ø Principle of virtual work:

kq • δ q + mg • δ q = 0 ⇒ (−kq + mgsin θ ) δq = 0

mg sin q
Þ -kq + mg sin q = 0 Þ q =
€ k

Ø Using Potential Energy

V = Vspring + Vgravitatio n

x =0
() ref
V r =ò !
r1
F •dr

òx=q
- kx • dx
= ∫
x=0

x=q
mgsin θ dx
x =0 conservative force
1 x=0
= - kx 2 = mgsin θ x x=q
2 x=q
= −mgq sin θ
1
= kq 2
2
1 2
So V = kq − mgqsin θ
2
¶V mg sin q
Virtual work dV = 0 or d q = 0 Þ kq = mg sin q or q =
¶q k

€ independent

Fall 2021
Copyright Fathi H. Ghorbel 2021 20
D Alembert s Principle
Newton s equation of motion for a system of particles mi (i=1,2,..,N)

  
Fi + f = mi 
ri ; i = 1,2,..., N (∗)
 i
applied constrained
forces forces
Acceleration of mass mi

(*) can be rewritten as


  
Fi + f i − mi 
r = 0; i = 1,2,..., N (2∗)
i
inertial force

(2*) can be regarded as representing the dynamic equilibrium of the system of particles.

   
Ø Virtual work for individual particles: ( Fi + f i − mi )
ri • δ r i = 0 ; i = 1,2,..., N
N

Ø Virtual work for entire system:

∑ (F + f
i i )
− mi ˙r˙i • δ r i = 0
i=1


Consider the class of systems for which the virtual work of the constraint forces over virtual displacements δ ri
€
compatible with the system constraints is zero, i.e.
N

åf
i =1
i •d ri = 0

Fall 2021

Copyright Fathi H. Ghorbel 2021 21

D Alembert s Principle (Continued)

N 
  
∑( Fi − mi  )
r i •δ ri = 0
i=1
The virtual work performed by the effective forces through virtual displacements compatible with the
system constraints is zero.

For systems of N rigid bodies and consider planar motion, d Alembert s principle takes the form
N 
     
"  %
  i=1
(F
∑$# i i − m 
r ci • δ )
r i + M ci(− I c
θ )
δθ i
'& = 0
whereδ r ci & δθ i are consistent with the system constraints.

Fall 2021

Copyright Fathi H. Ghorbel 2021 22

 Lagrange s Equations of Motion ®

Consider an n-d.o.f system of N particles and assume ri do not depend explicitly on

time. Write the coordinate transformation:

® ®
ü Positions : r = r (q , q ,..., q ) ; i = 1, 2, ...., N with qk (k = 1,2,..., N ) being the generalized coordinates
i i 1 2 n

   
ü Velocities: 
•
∂r ∂r ∂r n
∂r
ri = i q1 + i q2 + ... + i qn = ∑ i qk ; i = 1,2,..., N (2∗)
∂q1 ∂q2 ∂qn k=1 ∂qk

 
∂ri (2∗) ∂r i
= ; i = 1, 2,..., N; k = 1, 2,..., n (3∗)
∂qk ∂qk
ü By analogy with (2*)

n
¶r i
d ri = å dqk ; k = 1,2,..., n (4*)
k =1 ¶qk

Fall 2021

Copyright Fathi H. Ghorbel 2021 23

 D Alembert s Principle
N  

∑(Fi − mi ri )• δ ri = 0 (4.5∗)
i=1

Ø Consider

N  (4*) N  n ∂r i
∑ mi r i • δ ri = ∑ mi r i • ∑ ∂q δ qk
i=1 i=1 k=1 k

N n #  ∂r i &
= ∑∑% mi % r• (δ qk
i=1 k=1 $ ∂qk ('

   
 ∂r i d #  ∂r i &  d # ∂r i & ∂ri
mi 
ri • = % m ri • ( − m ri • % (=
∂qk dt %$ i ∂qk (' i dt %$ ∂qk (' ∂qk
* d # ∂ & ∂ -# 1   &
= , %% (( − /% mi ri • ri (

,+ dt $ ∂qk ' ∂qk /.$ 2
'
K .E.of particle i

Fall 2021

Copyright Fathi H. Ghorbel 2021 24

 # * # & - &
N   % n , d % ∂ ( ∂ /# N 1 • • &(
••
N (4∗) N n
∑ mi ri • δ ri = %∑, dt % • ( − ∂q /%%∑ 2 mi r i • r i ((( δ qk (5∗) δW = ∑ Fi • δ ri = ∑ F • ∑ ∂q
∂ri
δqk
i=1 % k=1 , %$ ∂q (' k /$ i=1 '( i
$ + k . ' i=1 i=1 k=1 k

* # & -
n &N ∂ri )
n
, d % ∂T ( ∂T / =∑ (∑ i F • + δqk (5.5∗)
∂q
= ∑, % • ( −
dt ∂q / δ qk k=1 'i=1 k*
k=1 , % ( k/ generalized forces
+ $ ∂qk ' . Q
k
• •
1 N
T = å mi ri • ri = T (q1 , q2 ,..., qn , q1 , q! 2 ,..., q! n )
n
where ( see 2*) = ∑ Qkδqk (6 ∗)
2 i =1 k=1

kinetic energy expressed as a function of

€
the generalized coordinates and velocities

Notes: -The generalized forces are not usually computed as above in actual situations. Instead,
they are derived by first identifying physically a set of generalized coordinates, and writing the
virtual work directly in the form (6*)

- generalized forces are not necessarily forces. Important thing Qk δqk has units of work

Fall 2021
Copyright Fathi H. Ghorbel 2021 25
 Ø Divide forces acting on the system into:

ü Conservative forces derived from potential energy V=V(q1, q2,…, qn)

ü Nonconservative forces

n
(5.5∗) ⇒ ∑ Qkδ qk •(5∗) & (7∗) in D ' Alembert ' s principle (4.5∗)
k=1
n
* $ ' -
, d & ∂T ) ∂T ∂V /

N  −∑, & • ) − + − Qknc /δ qk = 0 (8*)
k=1 dt & ∂qk ∂qk
∑ i • δ r i = δW
F ,+ % ∂q k )( /.
i=1

= δWc + δWnc
n •For (8∗)to hold
= −δV + ∑ Qkncδ qk $ '
k=1 d & ∂T ) ∂T ∂V
− + = Qk ;k = 1,2,...,n
% δV δV δV ( n dt && • )) ∂qk ∂qk 
= − '' δ q1 + δ q + ... + δ q * + ∑Q δ q % ∂q k ( nonconservative
& δ q1 δ q2 2 δ qn n *) k=1 knc k generalized force
(nc dropped )
n % (
δV ↑
= −∑'' − Qknc **δ qk
Lagrange equations of motion
k=1 & δ qk )

Fall 2021
Copyright Fathi H. Ghorbel 2021 26
 q j ( j = 1, 2,..., n)
In general, potential energy V does not depend on the generalized velocities

Ø Define the Lagrangian: L = T −V

→ Lagrange equations reduce to

" %
d $ ∂L ' " ∂L %
−$ ' = Qk ;k = 1,2,...,n
dt $$ • '' $# ∂qk '&
# ∂q k &

Ø Lagrange s equations for conservative systems (i.e. nonconservative forces involved)

" %
d $ ∂L ' " ∂L %
−$ ' = 0 ;k = 1,2,...,n
dt $$ • '' $# ∂qk '&
# ∂q k &

NOTES: 1.) Potential energy due to a conservative force

→
" → % r ref → →
V $ r ' = ∫ Fc • d r (3.40 in textbook)
# & →
r

2.) Always compute δWnc = Qkδ qk to determine Qk

Fall 2021
Copyright Fathi H. Ghorbel 2021 27
Example:
kΔ

L L
k k

x mg
m Δ
x=0
x
Lagrange s equation for a conservative system
Equations of motion: x is measured from equilibrium Q
" %
d $ ∂L ' " ∂L %
2 −$ ' = 0
1 æ•ö dt $ • ' # ∂x &
Kinetic Energy : T =K = mç x÷ #∂x &
2 è ø " %
y =0 " ∂L % ∂L •
d $ ∂L ' ••
y =0 1 2 y =0 $ ' = −kx ; • = m x ⇒ = mx
Potential Energy : V = òy = x ( mg - k ( y + D ) dy ) = mgy y=x
-
2
ky - k y D y = x # ∂x &
∂x
dt $ • '
#∂x &
••
1 m x+ kx = 0
= - mgx + kx 2 + kxD
2 *•
1 1 • , y1 = y2
= kx 2 + (k D - mg ) x = kx 2 y1 = x; y2 = x ⇒ + •
2 2
,y = − k y
- 2 m 1
2
1 æ•ö 1 " •% •• •
L = T - V = m ç x ÷ - kx 2 If there is a damper ⇒ δW = $ −c x 'δ x ⇒ m x+ kx = −c x
2 è ø 2 # &
Fall 2021
Copyright Fathi H. Ghorbel 2021 28
 Problem (3.4 Hulton):
m
A massless bar of length L is pinned at its center. It
carries a mass m at its top end and a mass 3m at the
θ
bottom. At the quarter point is attached a linear
spring with constant k. Find the equations of
motion for small oscillations.

k
d æ ¶L ö æ ¶L ö
ç ÷-ç ÷=0
dt çè ¶ q• ÷ø è ¶q ø
j
æ ¶L ö L2
ç ÷ = -k q
è ¶q ø 16
3m d æ ¶L ö d æ 1 2 • 3 2 • ö 1 2 •• 3 2 •• ••
ç • ÷ = ç mL q + mL q ÷ = mL q + mL q = mL2 q
dt çè ¶ q ÷ø dt è 4 4 ø 4 4
L2
•• •• k
Þ mL q + k q = 0 Þ q +
2
q =0
16 16m
2 2 2
1 æL ö 1 æL •ö 1 æL •ö
V = kç q ÷ ; T = mç q ÷ + (3m )ç q ÷
2 è4 ø 2 è2 ø 2 è2 ø x1 = q
2 2 2 •
1 æL •ö 3 æL •ö 1 æL ö x2 = q
L = T - V = mç q ÷ + mç q ÷ - k ç q ÷
2 è2 ø 2 è2 ø 2 è4 ø ì•
2 2 ï x1 = x2
æL •ö 1 æL ö í• Matlab implementation
= 2mç q ÷ - k ç q ÷ k
è2 ø 2 è4 ø ïx2 = - x1
î 16 m

Fall 2021
Copyright Fathi H. Ghorbel 2021 29