Dynamics of Mechanical Systems – prof. S. Bruni eng. S.

Zuin

Exercise session nr. 3

The third exercise session deals with the writing of the equations of motion of two mechanical systems with two
degrees of freedom. As in the second mechanical system there are non-linear kinematic relationships, its
equations of motion will be linearized around the given static position. In both cases the vector form of the
Lagrange’s equations will be used.

1. Exercise nr. 1
A disk is connected to the ground with a cart that allows both for rotation θ and for displacement x1 of its center.
A string is wound around the disk and its two extremities are connected to a body m2 that can move along a
horizontal plane; the upper extremity of the string is directly connected to the body m2, while the lower one is
connected through a spring and damper system. Write the equations of motion using the Lagrange’s equations.
Compute also the force transmitted to the ground by the spring and damper system (k1, c1).

m1 = 10 kg m2 = 2 kg
J1 = 0.2 kgm2 R = 0.5 m
k1 = 1500 N/m c1 = 10 Ns/m
k2 = 3000 N/m c2 = 0.2 Ns/m
k3 = 1500 N/m c3 = 10 Ns/m
𝐹𝐹(𝑡𝑡) = 𝐹𝐹0 𝑒𝑒 𝑖𝑖Ω𝑡𝑡

Solution of the exercise:

Step 1 - computation of the d.o.f.: there are two rigid bodies, three carts
and one string; the d.o.f. without constraints are six and the constraints
remove four d.o.f.; therefore the constrained system has two d.o.f.

Step 2 – sign convention: let’s assume the sign convention shown on the
right.

Step 3 – define the system’s independent coordinates: let’s assume the horizontal translation of the disk center
(x1) and that of the mass m2 (x2) as independent coordinates.

Step 4 – kinetic energy, dissipation function, potential energy and virtual work:

The kinetic energy is given by the sum of the kinetic energy of the disk and that of the mass m2. The first has two
terms because the disk translates and rotates, the latter has only the translational term.

The elastic potential energy and the dissipation function have two terms because there are three springs and
three dampers. The gravitational potential energy is constant because the height of the centers of mass is
constant:

1
Dynamics of Mechanical Systems – prof. S. Bruni eng. S. Zuin

1 1 1
⎧𝑇𝑇 = 𝑚𝑚𝑥𝑥̇ 12 + 𝐽𝐽1 𝜃𝜃̇ 2 + 𝑚𝑚𝑥𝑥̇ 22
⎪ 2 2 2
⎪ 1 1 1
𝐷𝐷 = 𝑐𝑐1 ∆𝐿𝐿̇ 21 + 𝑐𝑐2 ∆𝐿𝐿̇ 22 + 𝑐𝑐3 ∆𝐿𝐿̇ 23
2 2 2 (1.1)
⎨ 1 2
1 2
1 2
⎪𝑉𝑉𝑒𝑒𝑒𝑒 = 2 𝑘𝑘1 ∆𝐿𝐿1 + 2 𝑘𝑘2 ∆𝐿𝐿2 + 2 𝑘𝑘3 ∆𝐿𝐿3 ; 𝑉𝑉𝐺𝐺 = 𝑚𝑚𝑚𝑚ℎ𝐺𝐺1 + 𝑚𝑚𝑚𝑚ℎ𝐺𝐺2 = 𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐
⎪
�������⃗2 = 𝐹𝐹(𝑡𝑡)𝛿𝛿𝑥𝑥2
⎩𝛿𝛿𝛿𝛿 = 𝐹𝐹⃗ ∙ 𝛿𝛿𝑥𝑥

Step 5 – kinematic relationships:

The kinematic of this mechanical system is linear because it is given by the translation and rotation of the disk
and by the translation of the mass m2: being linear these elementary movements, all of the kinematic
relationships given by their combination will be linear as well. In order to write the kinematic relationships, we
will make use of the superimposition principle, which can be applied to linear systems or to non-linear system
which equations of motion are linearized.

In order to apply the superimposition principle to a two d.o.f. system, we will:

- Fix the second independent coordinate and move the system according to the first one, thus reducing
the system to 1 d.o.f; write the kinematic relationships that are needed;
- fix the first independent coordinate and move the system according to the second one; write the
kinematic relationships;
- obtain the overall kinematic relationships for a general system’s movement, namely when both
independent coordinates are different form zero, by summing term to term the partial kinematic
relationships.

In order to fix an independent coordinate, one additional degree of constraint must be added to the system.

In this exercise we have first to fix the translation of the m2 mass (Figure 1.1 and equation (1.2)) and, second, we
have to fix the translation of the disk center by substituting the cart with a hinge (Figure 1.2 and equation (1.3)).

In the first case, fixing the translation of the m2 mass, the Instantaneous Axis of Rotation (I.A.R.) of the disk is
point A because the velocity of this point is zero, being the upper part of the string fixed to the ground together
with the mass m2. When the position of the I.A.R. of a rigid body is known, it’s easy to compute the displacement
of any point of the body because it is proportional to the distance from the considered point and the I.A.R.; the
direction of the displacement is normal to the line that connects the point to the I.A.R.. In this case, Figure 1.1,
given the x1 displacement, the disk rotation is 𝑥𝑥1 /𝑅𝑅 (positive because a positive x1 displacement makes the disk
rotate counterclockwise – check the sign convention) and the spring k2 is compressed by 2x1.

In the second case, Figure 1.2, when the translation of the disk center is fixed, the I.A.R. if of course the hinge.
Given a positive x2 displacement, the displacement of point A is equal to x2 because the string is undeformable
along its axial direction; as a consequence, the disk rotation is 𝑥𝑥2 /𝑅𝑅 (negative because a positive x2 displacement
makes the disk rotate clockwise) and the left extremity of the spring moves towards left by x2. As the right
extremity of the spring moves towards right by x2, the total spring elongation is 2 x2.

2
Dynamics of Mechanical Systems – prof. S. Bruni eng. S. Zuin

𝒙𝒙𝟏𝟏 ≠ 𝟎𝟎; 𝒙𝒙𝟐𝟐 = 𝟎𝟎 𝒙𝒙𝟏𝟏 = 𝟎𝟎; 𝒙𝒙𝟐𝟐 ≠ 𝟎𝟎

Figure 1.1 Figure 1.2

𝑥𝑥1 𝑥𝑥2
⎧𝜃𝜃 = ⎧𝜃𝜃 = −
⎪ 𝑅𝑅 ⎪ 𝑅𝑅
∆𝐿𝐿1 = 𝑥𝑥1 (1.2) ∆𝐿𝐿1 = 0 (1.3)
⎨∆𝐿𝐿2 = −2𝑥𝑥1 ⎨∆𝐿𝐿2 = 𝑥𝑥2 − 𝑥𝑥𝐵𝐵 = 2𝑥𝑥2
⎪ ⎪
⎩∆𝐿𝐿3 = 0 ⎩∆𝐿𝐿3 = −𝑥𝑥2

Equations (1.4) are the kinematic relationships of the system, given by the sum of equations (1.2) and (1.3):
𝑥𝑥1 𝑥𝑥2
⎧𝜃𝜃 = −
⎪ 𝑅𝑅 𝑅𝑅
∆𝐿𝐿1 = 𝑥𝑥1 (1.4)
⎨∆𝐿𝐿2 = 𝑥𝑥𝐵𝐵 − 𝑥𝑥𝐴𝐴 = −2𝑥𝑥1 + 2𝑥𝑥2
⎪
⎩∆𝐿𝐿3 = −𝑥𝑥2

Being the independent coordinates more than one, the vector form of the Lagrange’s equations is more
convenient instead of the scalar one because it allows to define directly the structural matrices of the system:
𝑇𝑇 𝑇𝑇 𝑇𝑇 𝑇𝑇
𝑑𝑑 𝜕𝜕𝜕𝜕 𝜕𝜕𝜕𝜕 𝜕𝜕𝜕𝜕 𝜕𝜕𝜕𝜕
� � − � � + � � + � � = 𝑄𝑄 (1.5)
𝑑𝑑𝑑𝑑 𝜕𝜕𝑥𝑥̇ 𝜕𝜕𝑥𝑥 𝜕𝜕𝑥𝑥̇ 𝜕𝜕𝑥𝑥

where 𝑥𝑥 is the column vector collecting the independent coordinates:

𝑥𝑥1
𝑥𝑥 = �𝑥𝑥 � (1.6)
2

In order to make use of the Lagrange’s equations in the vector form, the kinetic energy, the potential energy, the
dissipation function and the virtual work have to be written in the matrix form. To this end we have to collect all
of the physical coordinates into vectors and all of the physical parameters into diagonal matrices (vector for the
virtual work): for example the vector 𝑦𝑦̇𝑚𝑚 collects the scalar velocities �𝑥𝑥̇ 1 , 𝜃𝜃̇ , 𝑥𝑥̇ 2 � and the diagonal matrix �𝑚𝑚𝑝𝑝ℎ �
collects the inertial parameters (𝑚𝑚1 , 𝐽𝐽1 , 𝑚𝑚2 ) used to define the kinetic energy. The vectors 𝑦𝑦̇𝑚𝑚 , 𝑦𝑦̇𝑐𝑐 , 𝑦𝑦𝑘𝑘 and
𝛿𝛿𝑦𝑦𝑄𝑄 (see equations (1.7), (1.8), (1.9) and (1.10)) are column vectors; the order used to collect the scalar physical

3
Dynamics of Mechanical Systems – prof. S. Bruni eng. S. Zuin

coordinates inside the vectors is arbitrary, but, once it has been defined, the same order has to be used to collect
the physical parameters into the matrices (and into the vector F):

1
𝑇𝑇 = 𝑦𝑦̇𝑚𝑚 𝑇𝑇 �𝑚𝑚𝑝𝑝ℎ �𝑦𝑦̇𝑚𝑚 𝑤𝑤ℎ𝑒𝑒𝑒𝑒𝑒𝑒: 𝑦𝑦̇𝑚𝑚 = {𝑥𝑥̇ 1 𝜃𝜃̇ 𝑥𝑥̇ 2 }𝑇𝑇 ; �𝑚𝑚𝑝𝑝ℎ � = 𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑(𝑚𝑚1 𝐽𝐽1 𝑚𝑚2 ) (1.7)
2

1
𝐷𝐷 = 𝑦𝑦̇𝑐𝑐 𝑇𝑇 �𝑐𝑐𝑝𝑝ℎ �𝑦𝑦̇𝑐𝑐 𝑤𝑤ℎ𝑒𝑒𝑒𝑒𝑒𝑒: 𝑦𝑦̇𝑐𝑐 = {∆𝐿𝐿̇ 1 ∆𝐿𝐿̇ 2 ∆𝐿𝐿̇ 3 }𝑇𝑇 ; �𝑐𝑐𝑝𝑝ℎ � = 𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑(𝑐𝑐1 𝑐𝑐2 𝑐𝑐3 ) (1.8)
2

1
𝑉𝑉𝑒𝑒𝑒𝑒 = 𝑦𝑦𝑘𝑘 𝑇𝑇 �𝑘𝑘𝑝𝑝ℎ �𝑦𝑦𝑘𝑘 𝑤𝑤ℎ𝑒𝑒𝑒𝑒𝑒𝑒: 𝑦𝑦𝑘𝑘 = {∆𝐿𝐿1 ∆𝐿𝐿2 ∆𝐿𝐿3 }𝑇𝑇 ; �𝑘𝑘𝑝𝑝ℎ � = 𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑(𝑘𝑘1 𝑘𝑘2 𝑘𝑘3 ) (1.9)
2

𝛿𝛿𝛿𝛿 = 𝛿𝛿𝑦𝑦𝑄𝑄𝑇𝑇 𝐹𝐹 𝑤𝑤ℎ𝑒𝑒𝑒𝑒𝑒𝑒: 𝛿𝛿𝑦𝑦𝑄𝑄 = {𝛿𝛿𝑥𝑥2 }; 𝐹𝐹 = {𝐹𝐹(𝑡𝑡)} (1.10)

The procedure to define the Jacobian matrices is based on the following steps:

- For what concern the Jacobian mass and damping matrices, the vectors 𝑦𝑦𝑚𝑚 and 𝑦𝑦𝑐𝑐 , collecting the physical
coordinates, have to be defined. Performing the first derivative of these vectors with respect time, we
get the Jacobian matrices [Λ 𝑚𝑚 ] and [Λ 𝑐𝑐 ] which collect the first partial derivatives of the physical
coordinates with respect to the independent one (see equations (1.11), (1.12), (1.15) and (1.16));
- Regarding the Jacobian stiffness matrix, the 𝑦𝑦𝑘𝑘 vector has to be written in series of power (limiting the
series to the first power): the first partial derivative of 𝑦𝑦𝑘𝑘 with respect 𝑥𝑥 is the Jacobian matrix [Λ 𝑘𝑘 ]
(equations (1.13) and (1.17));
- Regarding the Jacobian matrix of the external forces/torques, the 𝑦𝑦𝑄𝑄 vector, collecting the displacement
of the point at which the force F is applied, has to be defined. The computation of the differential
𝛿𝛿𝑦𝑦𝑄𝑄 allows to define the matrix �Λ 𝑄𝑄 � (equations (1.14) and (1.18)):

𝑑𝑑𝑦𝑦𝑚𝑚 𝜕𝜕𝑦𝑦𝑚𝑚 𝑑𝑑𝑥𝑥

𝑦𝑦𝑚𝑚 = {𝑥𝑥1 𝜃𝜃 𝑥𝑥2 }𝑇𝑇 = 𝑦𝑦𝑚𝑚 �𝑥𝑥� → 𝑦𝑦̇𝑚𝑚 = = = [Λ 𝑚𝑚 ]𝑥𝑥̇ (1.11)
𝑑𝑑𝑑𝑑 𝜕𝜕𝑥𝑥 𝑑𝑑𝑑𝑑

𝑑𝑑𝑦𝑦𝑐𝑐 𝜕𝜕𝑦𝑦𝑐𝑐 𝑑𝑑𝑥𝑥

𝑦𝑦𝑐𝑐 = {∆𝐿𝐿1 ∆𝐿𝐿2 ∆𝐿𝐿3 }𝑇𝑇 = 𝑦𝑦𝑐𝑐 �𝑥𝑥� → 𝑦𝑦̇𝑐𝑐 = = = [Λ 𝑐𝑐 ]𝑥𝑥̇ (1.12)
𝑑𝑑𝑑𝑑 𝜕𝜕𝑥𝑥 𝑑𝑑𝑑𝑑

𝜕𝜕𝑦𝑦𝑘𝑘
𝑦𝑦𝑘𝑘 = {∆𝐿𝐿1 ∆𝐿𝐿2 ∆𝐿𝐿3 }𝑇𝑇 = 𝑦𝑦𝑘𝑘 �𝑥𝑥� → 𝑦𝑦𝑘𝑘 = �𝑥𝑥 − 𝑥𝑥0 � = [Λ 𝑘𝑘 ]𝑥𝑥 (1.13)
𝜕𝜕𝑥𝑥

𝜕𝜕𝑦𝑦𝑄𝑄
𝑦𝑦𝑄𝑄 = {𝑥𝑥2 } = 𝑦𝑦𝑄𝑄 �𝑥𝑥� → 𝛿𝛿𝑦𝑦𝑄𝑄 = 𝛿𝛿𝑥𝑥 = �Λ 𝑄𝑄 �𝛿𝛿𝑥𝑥 (1.14)
𝜕𝜕𝑥𝑥

4
Dynamics of Mechanical Systems – prof. S. Bruni eng. S. Zuin

𝜕𝜕𝑦𝑦𝑚𝑚1 𝜕𝜕𝑦𝑦𝑚𝑚1 𝜕𝜕𝑥𝑥1 𝜕𝜕𝑥𝑥1

⎡ ⎤ ⎡ ⎤
⎢ 𝜕𝜕𝑥𝑥1 𝜕𝜕𝑥𝑥2 ⎥ ⎢𝜕𝜕𝑥𝑥1 𝜕𝜕𝑥𝑥2 ⎥ 1 0
⎢𝜕𝜕𝑦𝑦 ⎡ ⎤
𝜕𝜕𝑦𝑦𝑚𝑚2 ⎥ ⎢ 𝜕𝜕𝜕𝜕 𝜕𝜕𝜕𝜕 ⎥ 1 1⎥
[Λ 𝑚𝑚 ] = ⎢
𝑚𝑚2
⎥= ⎢ ⎥ = ⎢⎢ − ⎥ (1.15)
𝜕𝜕𝑥𝑥 𝜕𝜕𝑥𝑥2 ⎥ ⎢𝜕𝜕𝑥𝑥1 𝜕𝜕𝑥𝑥2 ⎥
⎢ 1 ⎢𝑅𝑅 𝑅𝑅 ⎥
⎢ ⎥ ⎢ ⎥ ⎣0
⎢𝜕𝜕𝑦𝑦𝑚𝑚3 𝜕𝜕𝑦𝑦𝑚𝑚3 ⎥ ⎢𝜕𝜕𝑥𝑥2 𝜕𝜕𝑥𝑥2 ⎥ 1 ⎦
⎣ 𝜕𝜕𝑥𝑥1 𝜕𝜕𝑥𝑥2 ⎦ ⎣𝜕𝜕𝑥𝑥1 𝜕𝜕𝑥𝑥2 ⎦

𝜕𝜕𝑦𝑦𝑐𝑐1 𝜕𝜕𝑦𝑦𝑐𝑐1 𝜕𝜕∆𝐿𝐿1 𝜕𝜕∆𝐿𝐿1

⎡ ⎤ ⎡ ⎤
⎢ 𝜕𝜕𝑥𝑥1 𝜕𝜕𝑥𝑥2 ⎥ ⎢ 𝜕𝜕𝑥𝑥1 𝜕𝜕𝑥𝑥2 ⎥ 1 0
⎢𝜕𝜕𝑦𝑦 ⎡ ⎤
𝜕𝜕𝑦𝑦𝑐𝑐2 ⎥ ⎢𝜕𝜕∆𝐿𝐿 𝜕𝜕∆𝐿𝐿2 ⎥
[Λ 𝑐𝑐 ] = ⎢ 𝑐𝑐2 ⎥= ⎢ 2
⎥ = ⎢−2 2 ⎥⎥
⎢
(1.16)
⎢ 𝜕𝜕𝑥𝑥1 𝜕𝜕𝑥𝑥2 ⎥ ⎢ 𝜕𝜕𝑥𝑥1 𝜕𝜕𝑥𝑥2 ⎥
⎢ ⎥
⎢ ⎥ ⎢ ⎥ ⎣ ⎦
⎢𝜕𝜕𝑦𝑦𝑐𝑐3 𝜕𝜕𝑦𝑦𝑐𝑐3 ⎥ ⎢𝜕𝜕∆𝐿𝐿3 𝜕𝜕∆𝐿𝐿3 ⎥ 0 −1
⎣ 𝜕𝜕𝑥𝑥1 𝜕𝜕𝑥𝑥2 ⎦ ⎣ 𝜕𝜕𝑥𝑥1 𝜕𝜕𝑥𝑥2 ⎦

𝜕𝜕𝑦𝑦𝑘𝑘1 𝜕𝜕𝑦𝑦𝑘𝑘1 𝜕𝜕∆𝐿𝐿1 𝜕𝜕∆𝐿𝐿1

⎡ ⎤ ⎡ ⎤
⎢ 𝜕𝜕𝑥𝑥1 𝜕𝜕𝑥𝑥2 ⎥ ⎢ 𝜕𝜕𝑥𝑥1 𝜕𝜕𝑥𝑥2 ⎥ 1 0
⎢𝜕𝜕𝑦𝑦 ⎡ ⎤
𝜕𝜕𝑦𝑦𝑘𝑘2 ⎥ ⎢𝜕𝜕∆𝐿𝐿 𝜕𝜕∆𝐿𝐿2 ⎥
[Λ 𝑘𝑘 ] = ⎢ 𝑘𝑘2 ⎥= ⎢ 2
⎥ = ⎢⎢−2 2 ⎥⎥ (1.17)
⎢ 𝜕𝜕𝑥𝑥1 𝜕𝜕𝑥𝑥2 ⎥ ⎢ 𝜕𝜕𝑥𝑥1 𝜕𝜕𝑥𝑥2 ⎥
⎢ ⎥
⎢ ⎥ ⎢ ⎥ ⎣
⎢𝜕𝜕𝑦𝑦𝑘𝑘3 𝜕𝜕𝑦𝑦𝑘𝑘3 ⎥ ⎢𝜕𝜕∆𝐿𝐿3 𝜕𝜕∆𝐿𝐿3 ⎥ 0 −1⎦
⎣ 𝜕𝜕𝑥𝑥1 𝜕𝜕𝑥𝑥2 ⎦ ⎣ 𝜕𝜕𝑥𝑥1 𝜕𝜕𝑥𝑥2 ⎦

𝜕𝜕𝑦𝑦𝑄𝑄 𝜕𝜕𝑦𝑦𝑄𝑄 𝜕𝜕𝑥𝑥2 𝜕𝜕𝑥𝑥2

�Λ 𝑄𝑄 � = � �= � � = [0 1] (1.18)
𝜕𝜕𝑥𝑥1 𝜕𝜕𝑥𝑥2 𝜕𝜕𝑥𝑥1 𝜕𝜕𝑥𝑥2

If N is the d.o.f. number, equal to the length of the vector 𝑥𝑥, and M is the length of the vectors 𝑦𝑦𝑚𝑚 , 𝑦𝑦𝑐𝑐 , 𝑦𝑦𝑘𝑘 and
𝑦𝑦𝑄𝑄 , the Jacobian matrices have M rows and N columns, therefore are rectangular matrices. They depend on the
vector 𝑥𝑥 when the kinematic relationships are non-linear, and, in this case, we will see with the next exercise
how to linearize them. Being linear the mechanical system under study, the Jacobian matrices are constant.

Step 6 – Structural matrices and Lagrange’s equations:

Substituting equations (1.11), (1.12), (1.13) and (1.14) in equations (1.7), (1.8), (1.9) and (1.10) respectively, the
structural matrices [𝑀𝑀], [𝐶𝐶], [𝐾𝐾𝐼𝐼 ] and the Q vector are easily defined:

𝑦𝑦̇𝑚𝑚 = [Λ 𝑚𝑚 ]𝑥𝑥̇
1 1
� 1 𝑇𝑇 → 𝑇𝑇 = 𝑥𝑥̇ 𝑇𝑇 [Λ 𝑚𝑚 ]𝑇𝑇 �𝑚𝑚𝑝𝑝ℎ �[Λ 𝑚𝑚 ]𝑥𝑥̇ = 𝑥𝑥̇ 𝑇𝑇 [𝑀𝑀]𝑥𝑥̇ ; [𝑀𝑀] = [Λ 𝑚𝑚 ]𝑇𝑇 �𝑚𝑚𝑝𝑝ℎ �[Λ 𝑚𝑚 ] (1.19)
𝑇𝑇 = 𝑦𝑦̇𝑚𝑚 �𝑚𝑚𝑝𝑝ℎ �𝑦𝑦̇𝑚𝑚 2 2
2

𝑦𝑦̇𝑐𝑐 = [Λ 𝑐𝑐 ]𝑥𝑥̇
1 1
� 1 𝑇𝑇 → 𝐷𝐷 = 𝑥𝑥̇ 𝑇𝑇 [Λ 𝑐𝑐 ]𝑇𝑇 �𝑐𝑐𝑝𝑝ℎ �[Λ 𝑐𝑐 ]𝑥𝑥̇ = 𝑥𝑥̇ 𝑇𝑇 [𝐶𝐶]𝑥𝑥̇ ; [𝐶𝐶] = [Λ 𝑐𝑐 ]𝑇𝑇 �𝑐𝑐𝑝𝑝ℎ �[Λ𝑐𝑐 ] (1.20)
𝐷𝐷 = 𝑦𝑦̇𝑐𝑐 �𝑐𝑐𝑝𝑝ℎ �𝑦𝑦̇𝑐𝑐 2 2
2

5
Dynamics of Mechanical Systems – prof. S. Bruni eng. S. Zuin

𝑦𝑦𝑘𝑘 = [Λ 𝑘𝑘 ]𝑥𝑥
1 1
� 1 𝑇𝑇 → 𝑉𝑉 = 𝑥𝑥 𝑇𝑇 [Λ 𝑘𝑘 ]𝑇𝑇 �𝑘𝑘𝑝𝑝ℎ �[Λ𝑘𝑘 ]𝑥𝑥 = 𝑥𝑥 𝑇𝑇 [𝐾𝐾𝐼𝐼 ]𝑥𝑥; �𝐾𝐾𝑒𝑒𝑒𝑒,𝐼𝐼 � = [Λ 𝑘𝑘 ]𝑇𝑇 �𝑘𝑘𝑝𝑝ℎ �[Λ𝑘𝑘 ] (1.21)
𝑉𝑉𝑒𝑒𝑒𝑒 = 𝑦𝑦𝑘𝑘 �𝑘𝑘𝑝𝑝ℎ �𝑦𝑦𝑘𝑘 2 2
2

𝛿𝛿𝑦𝑦𝑄𝑄 = �Λ 𝑄𝑄 �𝛿𝛿𝑥𝑥 𝑇𝑇 𝑇𝑇
� → 𝛿𝛿𝛿𝛿 = 𝛿𝛿𝑥𝑥 𝑇𝑇 �Λ 𝑄𝑄 � 𝐹𝐹 = 𝛿𝛿𝑥𝑥 𝑇𝑇 𝑄𝑄 → 𝑄𝑄 = �Λ 𝑄𝑄 � 𝐹𝐹 (1.22)
𝛿𝛿𝛿𝛿 = 𝛿𝛿𝑦𝑦𝑄𝑄𝑇𝑇 𝐹𝐹

As already mentioned, the gravitational potential energy is constant; therefore, the gravitational stiffness matrix
is zero. The second order elastic stiffness is a zero matrix as well because the elongations of both springs depend
linearly from the independent coordinates:

�𝐾𝐾 � = [0]
� 𝑒𝑒𝑒𝑒,𝐼𝐼𝐼𝐼 → [𝐾𝐾𝑡𝑡𝑡𝑡𝑡𝑡 ] = �𝐾𝐾𝑒𝑒𝑒𝑒,𝐼𝐼 � + �𝐾𝐾𝑒𝑒𝑒𝑒,𝐼𝐼𝐼𝐼 � + [𝐾𝐾𝐺𝐺 ] = �𝐾𝐾𝑒𝑒𝑒𝑒,𝐼𝐼 � (1.23)
[𝐾𝐾𝐺𝐺 ] = [0]

Hence, the equation of motion is:

[𝑀𝑀]𝑥𝑥̈ + [𝐶𝐶]𝑥𝑥̇ + [𝐾𝐾𝑡𝑡𝑡𝑡𝑡𝑡 ]𝑥𝑥 = 𝑄𝑄 (1.24)

Step 6 – force transmitted to the ground by the spring and damper system (k1, c1):

The spring and damper (k1, c1) transmit to the ground is the sum of the elastic and damping force: 𝐹𝐹𝐺𝐺𝐺𝐺 =
𝑘𝑘1 Δ𝐿𝐿1 + 𝑐𝑐1 Δ𝐿𝐿̇ 1.

2. Exercise nr. 2
A rigid slider m1 is connected to the ground by means of two carts (depicted as two small wheels). A rigid bar PO
with negligible mass and length 2L is connected to the body m1 through a hinge, and a point mass m2 is fixed to
the bar in the point P. The midpoint B of the bar is connected to the body m1 through a spring and damper system
(k2, c2). Two harmonic horizontal forces are applied, the first at the m1 slider, the second at point P. Write the
linearized equations of motion using the Lagrange’s equations.

m1 = 20 kg m2 = 4 kg
k1 = 3000 N/m k2 = 3000 N/m
c1 = 3 Ns/m c2 = 3 Ns/m
L = 1.0 m
F01 = 1 N F02 = 1 N

6
Dynamics of Mechanical Systems – prof. S. Bruni eng. S. Zuin

Solution of the exercise writing the non-linear kinematic relationships:

Step 1 - computation of the d.o.f.: there are two rigid bodies, two carts
and one hinge; the d.o.f. without constraints are six and the constraints
remove four d.o.f.; therefore the constrained system has two d.o.f.

Step 2 – sign convention: let’s assume the sign convention shown on

the right.

Step 3 – define the system’s independent coordinates: let’s assume the horizontal translation of the mass m1
(x) and the bar rotation (𝜃𝜃) as independent coordinates.

Step 4 – kinetic energy, dissipation function, potential energy and virtual work:

In this mechanical system the rotation of the PO bar makes some kinematic relationships non-linear. Even if we
could follow the faster approach, as only the linearized equations of motion have to be written, it’s worth writing
the first time the non-linear kinematic relationships in order to show the general non-linear form of the Jacobian
matrices.

As the OP bar has negligible mass, the kinetic energy is given by the sum of that of the m1 mass and that of point
mass m2. The elastic potential energy and the dissipation function have two terms because there are two springs
and two dampers. The gravitational potential energy is given by the sum of a constant term (that of the m1 mass)
plus the non-linear contribution due to the point mass. As the point mass moves together with point P, which is
the upper bar extremity, and taking into account that in the static configuration the hinge is below the point
mass, we can draw the conclusion that the point mass, together with the bar, is in the inverted pendulum
configuration and hence the second order gravitational stiffness must be negative:

1 1 1 1 1
⎧𝑇𝑇 = 𝑚𝑚1 𝑥𝑥̇ 2 + 𝑚𝑚2 𝑣𝑣𝑃𝑃2 = 𝑚𝑚1 𝑥𝑥̇ 2 + 𝑚𝑚2 𝑥𝑥̇ 𝑃𝑃2 + 𝑚𝑚2 𝑦𝑦̇ 𝑃𝑃2
⎪ 2 2 2 2 2
⎪ 1 1
𝐷𝐷 = 𝑐𝑐1 ∆𝐿𝐿̇ 21 + 𝑐𝑐2 ∆𝐿𝐿̇ 22
2 2 (2.1)
⎨ 1 2
1 2
⎪𝑉𝑉𝑒𝑒𝑒𝑒 = 2 𝑘𝑘1 ∆𝐿𝐿1 + 2 𝑘𝑘2 ∆𝐿𝐿2 ; 𝑉𝑉𝐺𝐺 = 𝑚𝑚𝑚𝑚ℎ𝐺𝐺1 + 𝑚𝑚𝑚𝑚ℎ𝐺𝐺2
⎪
⎩𝛿𝛿𝛿𝛿 = 𝐹𝐹⃗01 𝑒𝑒 𝑖𝑖Ω𝑡𝑡 ∙ ����⃗
𝛿𝛿𝛿𝛿 + 𝐹𝐹⃗02 𝑒𝑒 𝑖𝑖Ω𝑡𝑡 ∙ ������⃗
𝛿𝛿𝑠𝑠𝑃𝑃 = 𝐹𝐹01 𝑒𝑒 𝑖𝑖Ω𝑡𝑡 𝛿𝛿𝛿𝛿 + 𝐹𝐹02 𝑒𝑒 𝑖𝑖Ω𝑡𝑡 𝛿𝛿𝑥𝑥𝑃𝑃

Step 5a – kinematic relationships and Jacobian matrices:

As the non-linear kinematic relationships are sought, we can’t make use of the superimposition principle, but we
have to apply together a positive translation x and a positive rotation 𝜃𝜃. Being (xO, yO) the initial position of the
hinge with respect to a fixed reference system, the final position of point P is reported in eq. (2.2); the elongation
of the second spring depends only on the rotation 𝜃𝜃 and not on the displacement x because the latter moves the
left spring extremity (point A) together with its right extremity, being therefore zero the elongation:

7
Dynamics of Mechanical Systems – prof. S. Bruni eng. S. Zuin

𝑥𝑥𝑃𝑃 = 𝑥𝑥𝑂𝑂 + 𝑥𝑥 − 2𝐿𝐿 sin 𝜃𝜃

𝑦𝑦 = 𝑦𝑦𝑂𝑂 + 2𝐿𝐿 cos 𝜃𝜃
� 𝑃𝑃 (2.2)
∆𝐿𝐿1 = 𝑥𝑥
∆𝐿𝐿2 = −𝐿𝐿 sin 𝜃𝜃

Figure 2.1

The vector form of the Lagrange’s equations has the following expression:
𝑇𝑇 𝑇𝑇 𝑇𝑇 𝑇𝑇
𝑑𝑑 𝜕𝜕𝜕𝜕 𝜕𝜕𝜕𝜕 𝜕𝜕𝜕𝜕 𝜕𝜕𝜕𝜕
� � − � � + � � + � � = 𝑄𝑄 (2.3)
𝑑𝑑𝑑𝑑 𝜕𝜕𝑥𝑥̇ 𝜕𝜕𝑥𝑥 𝜕𝜕𝑥𝑥̇ 𝜕𝜕𝑥𝑥

Where 𝑥𝑥 is the column vector collecting the independent coordinates:

𝑥𝑥
𝑥𝑥 = � � (2.4)
𝜃𝜃

Now we have to define the 𝑦𝑦̇𝑚𝑚 , 𝑦𝑦̇𝑐𝑐 , 𝑦𝑦𝑘𝑘 and 𝛿𝛿𝑦𝑦𝑄𝑄 vectors and their correspondent matrices �𝑚𝑚𝑝𝑝ℎ �, �𝑐𝑐𝑝𝑝ℎ �, �𝑘𝑘𝑝𝑝ℎ � and
vector 𝐹𝐹, as already seen in the previous exercise:

1
𝑇𝑇 = 𝑦𝑦̇𝑚𝑚 𝑇𝑇 �𝑚𝑚𝑝𝑝ℎ �𝑦𝑦̇𝑚𝑚 𝑤𝑤ℎ𝑒𝑒𝑒𝑒𝑒𝑒: 𝑦𝑦̇𝑚𝑚 = {𝑥𝑥̇ 𝑥𝑥̇ 𝑃𝑃 𝑦𝑦̇ 𝑃𝑃 }𝑇𝑇 ; �𝑚𝑚𝑝𝑝ℎ � = 𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑(𝑚𝑚1 𝑚𝑚2 𝑚𝑚2 ) (2.5)
2

1
𝐷𝐷 = 𝑦𝑦̇𝑐𝑐 𝑇𝑇 �𝑐𝑐𝑝𝑝ℎ �𝑦𝑦̇𝑐𝑐 𝑤𝑤ℎ𝑒𝑒𝑒𝑒𝑒𝑒: 𝑦𝑦̇𝑐𝑐 = {∆𝐿𝐿̇ 1 ∆𝐿𝐿̇ 2 }𝑇𝑇 ; �𝑐𝑐𝑝𝑝ℎ � = 𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑(𝑐𝑐1 𝑐𝑐2 ) (2.6)
2

1
𝑉𝑉𝑒𝑒𝑒𝑒 = 𝑦𝑦𝑘𝑘 𝑇𝑇 �𝑘𝑘𝑝𝑝ℎ �𝑦𝑦𝑘𝑘 𝑤𝑤ℎ𝑒𝑒𝑒𝑒𝑒𝑒: 𝑦𝑦𝑘𝑘 = {∆𝐿𝐿1 ∆𝐿𝐿2 }𝑇𝑇 ; �𝑘𝑘𝑝𝑝ℎ � = 𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑(𝑘𝑘1 𝑘𝑘2 ) (2.7)
2

𝛿𝛿𝛿𝛿 = 𝛿𝛿𝑦𝑦𝑄𝑄𝑇𝑇 𝐹𝐹 𝑤𝑤ℎ𝑒𝑒𝑒𝑒𝑒𝑒: 𝛿𝛿𝑦𝑦𝑄𝑄 = {𝛿𝛿𝛿𝛿 𝛿𝛿𝑥𝑥𝑃𝑃 }𝑇𝑇 ; 𝐹𝐹 = {𝐹𝐹01 𝑒𝑒 𝑖𝑖Ω𝑡𝑡 𝐹𝐹02 𝑒𝑒 𝑖𝑖Ω𝑡𝑡 }𝑇𝑇 (2.8)

Following the same procedure described in the previous exercise, the Jacobian matrices are computed:

𝑑𝑑𝑦𝑦𝑚𝑚 𝜕𝜕𝑦𝑦𝑚𝑚 𝑑𝑑𝑥𝑥

𝑦𝑦𝑚𝑚 = {𝑥𝑥 𝑥𝑥𝑃𝑃 𝑦𝑦𝑃𝑃 }𝑇𝑇 = 𝑦𝑦𝑚𝑚 �𝑥𝑥� → 𝑦𝑦̇𝑚𝑚 = = = [Λ 𝑚𝑚 ]𝑥𝑥̇ (2.9)
𝑑𝑑𝑑𝑑 𝜕𝜕𝑥𝑥 𝑑𝑑𝑑𝑑

𝑑𝑑𝑦𝑦𝑐𝑐 𝜕𝜕𝑦𝑦𝑐𝑐 𝑑𝑑𝑥𝑥

𝑦𝑦𝑐𝑐 = {∆𝐿𝐿1 ∆𝐿𝐿2 }𝑇𝑇 = 𝑦𝑦𝑐𝑐 �𝑥𝑥� → 𝑦𝑦̇𝑐𝑐 = = = [Λ 𝑐𝑐 ]𝑥𝑥̇ (2.10)
𝑑𝑑𝑑𝑑 𝜕𝜕𝑥𝑥 𝑑𝑑𝑑𝑑

𝜕𝜕𝑦𝑦𝑘𝑘
𝑦𝑦𝑘𝑘 = {∆𝐿𝐿1 ∆𝐿𝐿2 }𝑇𝑇 = 𝑦𝑦𝑘𝑘 �𝑥𝑥� → 𝑦𝑦𝑘𝑘 = �𝑥𝑥 − 𝑥𝑥0 � = [Λ 𝑘𝑘 ]𝑥𝑥 (2.11)
𝜕𝜕𝑥𝑥

8
Dynamics of Mechanical Systems – prof. S. Bruni eng. S. Zuin

𝜕𝜕𝑦𝑦𝑄𝑄
𝑦𝑦𝑄𝑄 = {𝑥𝑥 𝑥𝑥𝑃𝑃 }𝑇𝑇 = 𝑦𝑦𝑄𝑄 �𝑥𝑥� → 𝛿𝛿𝑦𝑦𝑄𝑄 = 𝛿𝛿𝑥𝑥 = �Λ 𝑄𝑄 �𝛿𝛿𝑥𝑥 (2.12)
𝜕𝜕𝑥𝑥

𝜕𝜕𝑦𝑦𝑚𝑚1 𝜕𝜕𝑦𝑦𝑚𝑚1 𝜕𝜕𝜕𝜕 𝜕𝜕𝜕𝜕

⎡ ⎤ ⎡ ⎤
⎢ 𝜕𝜕𝑥𝑥1 𝜕𝜕𝑥𝑥2 ⎥ ⎢ 𝜕𝜕𝜕𝜕 𝜕𝜕𝜕𝜕 ⎥ 1 0
⎢𝜕𝜕𝑦𝑦 ⎡ ⎤
𝑚𝑚2 𝜕𝜕𝑦𝑦𝑚𝑚2 ⎥ ⎢𝜕𝜕𝑥𝑥 ⎥
𝜕𝜕𝑥𝑥𝑃𝑃 ⎥ ⎢ ⎥
�Λ 𝑚𝑚 (𝑥𝑥)� = ⎢ ⎥= ⎢ 𝑃𝑃 = ⎢1 −2𝐿𝐿 cos 𝜃𝜃 ⎥ (2.13)
⎢ 𝜕𝜕𝑥𝑥1 𝜕𝜕𝑥𝑥2 ⎥ ⎢ 𝜕𝜕𝜕𝜕 𝜕𝜕𝜕𝜕 ⎥ ⎢ ⎥
⎢ ⎥ ⎢ ⎥ ⎣0
⎢𝜕𝜕𝑦𝑦𝑚𝑚3 𝜕𝜕𝑦𝑦𝑚𝑚3 ⎥ ⎢𝜕𝜕𝑦𝑦𝑃𝑃 𝜕𝜕𝑦𝑦𝑃𝑃 ⎥ −2𝐿𝐿 sin 𝜃𝜃 ⎦
⎣ 𝜕𝜕𝑥𝑥1 𝜕𝜕𝑥𝑥2 ⎦ ⎣ 𝜕𝜕𝜕𝜕 𝜕𝜕𝜕𝜕 ⎦

𝜕𝜕𝑦𝑦𝑐𝑐1 𝜕𝜕𝑦𝑦𝑐𝑐1 𝜕𝜕∆𝐿𝐿1 𝜕𝜕∆𝐿𝐿1

⎡ ⎤ ⎡ ⎤ 1 0
⎢ 𝜕𝜕𝑥𝑥1 𝜕𝜕𝑥𝑥2 ⎥ ⎢ 𝜕𝜕𝜕𝜕 𝜕𝜕𝜕𝜕 ⎥
�Λ𝑐𝑐 (𝑥𝑥)� = ⎢ ⎥= ⎢ ⎥ = � � (2.14)
⎢𝜕𝜕𝑦𝑦𝑐𝑐2 𝜕𝜕𝑦𝑦𝑐𝑐2 ⎥ ⎢𝜕𝜕∆𝐿𝐿2 𝜕𝜕∆𝐿𝐿2 ⎥ 0 −𝐿𝐿 cos 𝜃𝜃
⎣ 𝜕𝜕𝑥𝑥1 𝜕𝜕𝑥𝑥2 ⎦ ⎣ 𝜕𝜕𝜕𝜕 𝜕𝜕𝜕𝜕 ⎦

𝜕𝜕𝑦𝑦𝑘𝑘1 𝜕𝜕𝑦𝑦𝑘𝑘1 𝜕𝜕∆𝐿𝐿1 𝜕𝜕∆𝐿𝐿1

⎡ ⎤ ⎡ ⎤
⎢ 𝜕𝜕𝑥𝑥1 𝜕𝜕𝑥𝑥2 ⎥ 1 0
⎢ 𝜕𝜕𝜕𝜕 𝜕𝜕𝜕𝜕 ⎥
�Λ𝑘𝑘 (𝑥𝑥)� = ⎢ ⎥= ⎢ ⎥ = � � (2.15)
⎢𝜕𝜕𝑦𝑦𝑘𝑘2 𝜕𝜕𝑦𝑦𝑘𝑘2 ⎥ ⎢𝜕𝜕∆𝐿𝐿2 𝜕𝜕∆𝐿𝐿2 ⎥ 0 −𝐿𝐿 cos 𝜃𝜃
⎣ 𝜕𝜕𝑥𝑥1 𝜕𝜕𝑥𝑥2 ⎦ ⎣ 𝜕𝜕𝜕𝜕 𝜕𝜕𝜕𝜕 ⎦

𝜕𝜕𝑦𝑦𝑄𝑄1 𝜕𝜕𝑦𝑦𝑄𝑄2 𝜕𝜕𝜕𝜕 𝜕𝜕𝜕𝜕

⎡ ⎤ ⎡ ⎤
⎢ 𝜕𝜕𝑥𝑥1 𝜕𝜕𝑥𝑥2 ⎥ 1 0
⎢ 𝜕𝜕𝜕𝜕 𝜕𝜕𝜕𝜕 ⎥
�Λ 𝑄𝑄 (𝑥𝑥)� = ⎢ ⎥= ⎢ ⎥ = � � (2.16)
⎢𝜕𝜕𝑦𝑦𝑄𝑄1 𝜕𝜕𝑦𝑦𝑄𝑄2 ⎥ ⎢𝜕𝜕𝑥𝑥𝑃𝑃 𝜕𝜕𝑥𝑥𝑃𝑃 ⎥ 1 −2𝐿𝐿 cos 𝜃𝜃
⎣ 𝜕𝜕𝑥𝑥2 𝜕𝜕𝑥𝑥2 ⎦ ⎣ 𝜕𝜕𝜕𝜕 𝜕𝜕𝜕𝜕 ⎦

The Jacobian matrices depend on the independent coordinates due to the non-linear kinematic relationships. In
order to linearize the equation of motion we have to compute the Jacobian matrices in the static configuration
in order to get constant matrices that represent the linearized kinematic relationships in the neighborhood of
the static configuration:

1 0
⎡ ⎤ 1 0 1 0 1 0
⎢ ⎥
[Λ 𝑚𝑚 ] = ⎢1 −2𝐿𝐿⎥ ; [Λ 𝑐𝑐 ] = � � ; [Λ 𝑘𝑘 ] = � � ; �Λ 𝑄𝑄 � = � � (2.17)
⎢ ⎥ 0 −𝐿𝐿 0 −𝐿𝐿 1 −2𝐿𝐿
⎣0 0 ⎦

Step 5b – Second order stiffness matrices:

The second order elastic stiffness is a zero matrix for two reasons: as the elongation of the first spring depends
linearly on x and the elongation of the second one depends on sin 𝜃𝜃, all of the second order derivatives evaluated
in the static configuration will be zero (this means that the Hessian is a zero matrix). Moreover, the second spring

9
Dynamics of Mechanical Systems – prof. S. Bruni eng. S. Zuin

cannot be preloaded in the static configuration (see also the solution of exercise nr. 2 of the Exercise session nr.
2, figure 2.2).

For what concern the second order gravitational stiffness, only the contribution of the point mass has to be taken
into account:

ℎ𝐺𝐺 = 2𝐿𝐿 cos 𝜃𝜃 (2.18)

Figure 2.2

The second order gravitation stiffness is given by the product of the point mass weight multiplied by the Hessian
matrix, which collects all of the second order derivatives of the ℎ𝐺𝐺 physical coordinate:

𝜕𝜕 2 ℎ𝐺𝐺 𝜕𝜕 2 ℎ𝐺𝐺 𝜕𝜕 2 ℎ𝐺𝐺 𝜕𝜕 2 ℎ𝐺𝐺

⎡ � � ⎤ ⎡ � � ⎤
⎢ 𝜕𝜕𝑥𝑥12 𝑥𝑥0 𝜕𝜕𝑥𝑥1 𝜕𝜕𝜕𝜕2 𝑥𝑥 ⎥
0
⎢ 𝜕𝜕𝑥𝑥 2 𝑥𝑥0 𝜕𝜕𝜕𝜕𝜕𝜕𝜕𝜕 𝑥𝑥 ⎥
0
0 0
[K 𝐺𝐺 ] = 𝑚𝑚2 g�H𝐺𝐺,2 � = 𝑚𝑚2 g ⎢ 2 2
⎥= ⎢ 2 ⎥ = 𝑚𝑚2 g � � (2.19)
⎢ 𝜕𝜕 ℎ𝐺𝐺 𝜕𝜕 ℎ𝐺𝐺 ⎥ ⎢ 𝜕𝜕 ℎ𝐺𝐺 𝜕𝜕 2 ℎ𝐺𝐺 ⎥
⎢𝜕𝜕𝑥𝑥 𝜕𝜕𝜕𝜕 � 0 −2𝐿𝐿
2 � ⎥ ⎢𝜕𝜕𝜕𝜕𝜕𝜕𝜕𝜕 � �
𝜕𝜕𝜃𝜃 2 𝑥𝑥 ⎥⎦
⎣ 2 1 𝑥𝑥 𝜕𝜕𝑥𝑥2 𝑥𝑥 ⎦ ⎣ 𝑥𝑥
0 0 0 0

According to the unstable configuration of the point mass (not taking into account the contribution of the
springs), it’s important to notice that the Hessian matrix has a negative element on its main diagonal, being
therefore not positive defined.

Step 6 – Structural matrices and Lagrange’s equations:

The computation of the structural matrices is reported in equations (2.20):

[𝑀𝑀] = [Λ 𝑚𝑚 ]𝑇𝑇 �𝑚𝑚𝑝𝑝ℎ �[Λ𝑚𝑚 ]

⎧
⎪[𝐶𝐶] = [Λ 𝑐𝑐 ]𝑇𝑇 �𝑐𝑐𝑝𝑝ℎ �[Λ𝑐𝑐 ]
𝑇𝑇 (2.20)
⎨[𝐾𝐾𝐼𝐼 ] = [Λ 𝑘𝑘 ] �𝑘𝑘𝑝𝑝ℎ �[Λ𝑘𝑘 ]; [𝐾𝐾𝑡𝑡𝑡𝑡𝑡𝑡 ] = �𝐾𝐾𝑒𝑒𝑒𝑒,𝐼𝐼 � + �𝐾𝐾𝑒𝑒𝑒𝑒,𝐼𝐼𝐼𝐼 � + [𝐾𝐾𝐺𝐺 ] = �𝐾𝐾𝑒𝑒𝑒𝑒,𝐼𝐼 � + [𝐾𝐾𝐺𝐺 ]
⎪ 𝑇𝑇
⎩𝑄𝑄 = �Λ 𝑄𝑄 � 𝐹𝐹

Hence, the equation of motion are:

[𝑀𝑀]𝑥𝑥̈ + [𝐶𝐶]𝑥𝑥̇ + [𝐾𝐾𝑡𝑡𝑡𝑡𝑡𝑡 ]𝑥𝑥 = 𝑄𝑄 (2.21)

10
Dynamics of Mechanical Systems – prof. S. Bruni eng. S. Zuin

Solution of the exercise using the faster approach (direct linearization):

Steps from 1 to 4 are exactly the same as those of the previous solution.

Step 5a – kinematic relationships and Jacobian matrices:

We aim directly at the evaluation of the linearized kinematic relationships; hence we can use the principle of
superimposition.

In this case we have to fix first the rotation of the bar by removing the hinge and connecting directly the bar to
the m1 body (Figure 2.3 and equation (2.22)); second, we have to fix the translation of the m1 mass (Figure 2.4
and equation (2.23)); in this case the linearized trajectory of point P will be considered, which is the horizontal
line passing through the static position of point P:

𝒙𝒙 ≠ 𝟎𝟎; 𝜽𝜽 = 𝟎𝟎 𝒙𝒙 = 𝟎𝟎; 𝜽𝜽 ≠ 𝟎𝟎

Figure 2.3 Figure 2.4

𝑥𝑥𝑃𝑃 = 𝑥𝑥 𝑥𝑥𝑃𝑃 = −2𝐿𝐿𝐿𝐿
𝑦𝑦𝑃𝑃 = 0 𝑦𝑦 = 0
�∆𝐿𝐿 = 𝑥𝑥 (2.22) � 𝑃𝑃 (2.23)
1 ∆𝐿𝐿1 = 0
∆𝐿𝐿2 = 0 ∆𝐿𝐿2 = −𝐿𝐿𝐿𝐿

Equations (2.24)Errore. L'origine riferimento non è stata trovata. are the kinematic relationships of the system,
given by the sum of equations (2.22) and (2.23). It can be noticed that, taking into account linearized trajectories,
the point mass displacement is horizontal (this is correct only for small displacements):

𝑥𝑥𝑃𝑃 = 𝑥𝑥 − 2𝐿𝐿𝐿𝐿
𝑦𝑦 = 0
� 𝑃𝑃 (2.24)
∆𝐿𝐿1 = 𝑥𝑥
∆𝐿𝐿2 = −𝐿𝐿𝐿𝐿

The �𝑐𝑐𝑝𝑝ℎ � and �𝑘𝑘𝑝𝑝ℎ � matrices, the 𝐹𝐹 vector and the 𝑦𝑦̇𝑐𝑐 , 𝑦𝑦𝑘𝑘 and 𝛿𝛿𝑦𝑦𝑄𝑄 vectors are not changed with respect to the
previous solution; the �𝑚𝑚𝑝𝑝ℎ � matrix and its correspondent 𝑦𝑦̇𝑚𝑚 vector have in this case only two rows, instead of
three, due to the fact that the point mass vertical displacement is zero:

1
𝑇𝑇 = 𝑦𝑦̇𝑚𝑚 𝑇𝑇 �𝑚𝑚𝑝𝑝ℎ �𝑦𝑦̇𝑚𝑚 𝑤𝑤ℎ𝑒𝑒𝑒𝑒𝑒𝑒: 𝑦𝑦̇𝑚𝑚 = {𝑥𝑥̇ 𝑥𝑥̇ 𝑃𝑃 }𝑇𝑇 ; �𝑚𝑚𝑝𝑝ℎ � = 𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑(𝑚𝑚1 𝑚𝑚2 ) (2.25)
2

11
Dynamics of Mechanical Systems – prof. S. Bruni eng. S. Zuin

The Jacobian matrices are now constant matrices, being linear the kinematic relationships used to their
computation:

𝜕𝜕𝑦𝑦𝑚𝑚1 𝜕𝜕𝑦𝑦𝑚𝑚1 𝜕𝜕𝜕𝜕 𝜕𝜕𝜕𝜕

⎡ ⎤ ⎡ ⎤
⎢ 𝜕𝜕𝑥𝑥1 𝜕𝜕𝑥𝑥2 ⎥ 1 0
⎢ 𝜕𝜕𝜕𝜕 𝜕𝜕𝜕𝜕 ⎥
[Λ 𝑚𝑚 ] = ⎢ ⎥= ⎢ ⎥ = � � (2.26)
⎢𝜕𝜕𝑦𝑦𝑚𝑚2 𝜕𝜕𝑦𝑦𝑚𝑚2 ⎥ ⎢𝜕𝜕𝑥𝑥𝑃𝑃 𝜕𝜕𝑥𝑥𝑃𝑃 ⎥ 1 −2𝐿𝐿
⎣ 𝜕𝜕𝑥𝑥1 𝜕𝜕𝑥𝑥2 ⎦ ⎣ 𝜕𝜕𝜕𝜕 𝜕𝜕𝜕𝜕 ⎦

𝜕𝜕𝑦𝑦𝑐𝑐1 𝜕𝜕𝑦𝑦𝑐𝑐1 𝜕𝜕∆𝐿𝐿1 𝜕𝜕∆𝐿𝐿1

⎡ ⎤ ⎡ ⎤
⎢ 𝜕𝜕𝑥𝑥1 𝜕𝜕𝑥𝑥2 ⎥ 1 0
⎢ 𝜕𝜕𝜕𝜕 𝜕𝜕𝜕𝜕 ⎥
[Λ 𝑐𝑐 ] = ⎢ ⎥= ⎢ ⎥ = � � (2.27)
⎢𝜕𝜕𝑦𝑦𝑐𝑐2 𝜕𝜕𝑦𝑦𝑐𝑐2 ⎥ ⎢𝜕𝜕∆𝐿𝐿2 𝜕𝜕∆𝐿𝐿2 ⎥ 0 −𝐿𝐿
⎣ 𝜕𝜕𝑥𝑥1 𝜕𝜕𝑥𝑥2 ⎦ ⎣ 𝜕𝜕𝜕𝜕 𝜕𝜕𝜕𝜕 ⎦

𝜕𝜕𝑦𝑦𝑘𝑘1 𝜕𝜕𝑦𝑦𝑘𝑘1 𝜕𝜕∆𝐿𝐿1 𝜕𝜕∆𝐿𝐿1

⎡ ⎤ ⎡ ⎤
⎢ 𝜕𝜕𝑥𝑥1 𝜕𝜕𝑥𝑥2 ⎥ 1 0
⎢ 𝜕𝜕𝜕𝜕 𝜕𝜕𝜕𝜕 ⎥
[Λ 𝑘𝑘 ] = ⎢ ⎥= ⎢ ⎥ =� � (2.28)
⎢𝜕𝜕𝑦𝑦𝑘𝑘2 𝜕𝜕𝑦𝑦𝑘𝑘2 ⎥ ⎢𝜕𝜕∆𝐿𝐿2 𝜕𝜕∆𝐿𝐿2 ⎥ 0 −𝐿𝐿
⎣ 𝜕𝜕𝑥𝑥1 𝜕𝜕𝑥𝑥2 ⎦ ⎣ 𝜕𝜕𝜕𝜕 𝜕𝜕𝜕𝜕 ⎦

𝜕𝜕𝑦𝑦𝑄𝑄1 𝜕𝜕𝑦𝑦𝑄𝑄2 𝜕𝜕𝜕𝜕 𝜕𝜕𝜕𝜕

⎡ ⎤ ⎡ ⎤
⎢ 𝜕𝜕𝑥𝑥1 𝜕𝜕𝑥𝑥2 ⎥ 1 0
⎢ 𝜕𝜕𝜕𝜕 𝜕𝜕𝜕𝜕 ⎥
�Λ 𝑄𝑄 � = ⎢ ⎥= ⎢ ⎥ = � � (2.29)
⎢𝜕𝜕𝑦𝑦𝑄𝑄1 𝜕𝜕𝑦𝑦𝑄𝑄2 ⎥ ⎢𝜕𝜕𝑥𝑥𝑃𝑃 𝜕𝜕𝑥𝑥𝑃𝑃 ⎥ 1 −2𝐿𝐿
⎣ 𝜕𝜕𝑥𝑥1 𝜕𝜕𝑥𝑥2 ⎦ ⎣ 𝜕𝜕𝜕𝜕 𝜕𝜕𝜕𝜕 ⎦

Apparently the Jacobian matrix given by equation (2.26) is different from that given by equation (2.17), but we
have to take into account that the third row of the first Jacobian matrix has only zero elements (according to the
fact that the linearized motion of the point mass is horizontal): when this matrix is multiplied by the
corresponding �𝑚𝑚𝑝𝑝ℎ � matrix, the third row won’t give any contribution to the mass matrix [𝑀𝑀].

As the last part of the solution (computation of the second order stiffness matrices and of the structural matrices)
is not affected by the second approach, we will make reference to the first solution, and, more in details, to
equations from (2.18) to (2.21) for this part.

12