MECHANICAL VIBRATIONS

MATLAB TEAMWORK PROJECT

A.A. 2023/2024

V6 90° ENGINE VIBRATION

ANALYSIS AND DISCUSSION

Group n. 3
 GROUP MEMBERS AND ROLES

• Chairman: Filippo Tullio

• Coordinator: Amedeo Moriconi
• Technical tasks
• Data Research
• Antonio Maria Macripò
• CAD Model
• Michele Palomba
• Amedeo Moriconi
• Theoretical developments
• Michele Chiomento
• Attilio Marasco
• Amedeo Moriconi
• Numerical Implementation
• Gianmarco Senzamici
• Michele Palomba
• Antonio Maria Macripò
• Graphics
• Filippo Tullio
• Attilio Marasco
• Michele Chiomento
• FEM and Multibody dynamics Modal Analysis
• Amedeo Moriconi

2
 INTRODUCTION
The primary aim of the project is to investigate the vibrations generated by the imbalance in a V6-90°
internal combustion engine, typical of nowadays Formula 1 competition cars.
The utilization of a Multi-degree of Freedom (N-DOF) model enables a more precise description of
physical problems, through the application of techniques that facilitate the conversion of an initial
problem, defined by partial differential equations
(PDEs) with infinite degrees of freedom. This
transformation is accomplished by employing
methodologies that yield a system characterized by a
specific number of coupled ordinary differential
equations (ODEs). These ODEs collectively serve as an
approximation of the original PDEs, thus enhancing the
accuracy of the system's representation and analysis.
As a simplification, the engine block is considered as a
rigid body supported by four mounts, through which
the vibration forces originating from the engine are
transmitted to the chassis. Each mount consists of three
springs and three linear dampers, as illustrated,
capable of bearing loads only in the main direction
(axial). A reference system positioned in the center of FIGURE 1: REPRESENTATION OF ENGINE AS A RIGID
mass with axes aligned as in Figure 1 is chosen. The BODY, 3 DAMPERS AND 3 SPRINGS ON EACH OF THE
system thus exhibits 6 degrees of freedom: 3 FOUR MOUNTS
translations along the principal axes x, y and z and 3 rotations around x (roll), y (pitch) and z (yaw)
respectively. In analyzing the behavior of the engine block, various stiffness and damping coefficients
for each mount, as provided by the manufacturer [1], are taken into consideration.
A V6 engine with a 90-degree angle between the cylinder banks provides several advantages, including
compactness, balance, and reduced vibrations during operation.
Although this configuration stands out for its application in the Formula 1 realm, owing to the limited
availability of accessible data, the group based its research and considerations drawing inspiration from
the “Nettuno” engine by Maserati, introduced in 2020 and featured in the cutting-edge MC20.
The aims delineated in the ensuing report encompass the formulation of a dynamic model to investigate
V6-90° vibrations, identification of inertial forces resulting
from reciprocating component motion, derivation of the
system's equation of motion, exploration of natural
frequencies and mode shapes, execution of ramp tests and
steady-state ramp analysis. Subsequently, the ultimate goal
of the project involves plotting and examining in detail the
displacements and accelerations of the center of mass, along
with the forces transmitted to the chassis, as functions of
both time and frequency across three distinct operational
regimes of interest. The regimes of 2000rpm, 3000rpm, and
7500rpm have been selected: the first corresponds to the
FIGURE 2: CAD MODEL
maximum highway cruising speed (130km/h) in the highest
gear, the second and third are chosen as they ensure the maximum torque (730Nm) and power (630hp)
that can be utilized by the engine, respectively.
The data found in literature is employed to build a reliable CAD model on SOLIDWORKS, including the
crankshaft. Thereafter, leveraging the "Evaluate mass properties" function within the software, we
successfully acquired the inertia matrix of the engine block and the precise position of the center of
mass. [2]

3
 THEORY
The theoretical development begins with the search for the motion equation using the Lagrange's
equations energetic approach, considering separately the terms of kinetic energy (T), potential energy
(U), and Rayleigh's dissipation function (D):
ⅆ 𝜕𝑇 𝜕𝑈 𝜕𝐷 (1)
( )+ + = 𝑄𝑖
ⅆ𝑡 𝜕𝑞̇ 𝑖 𝜕𝑞𝑖 𝜕𝑞̇ 𝑖
Where the counter ‘i’ ranges from 1 to n based on the number of degrees of freedom (in this case, six).
The system is initially considered damped, and the three coordinates of the COG, together with the three
rotations around the principal axes, are chosen as the basic Lagrangian coordinates: q1=xG , q2=yG , q3=zG,
q4=θx, q5=θy, q6=θz. 𝑄𝑖 represent the forcing vector ith component, which will be calculated later on.
Upon developing the motion equation, the following displacements ∆𝒊 of the mounts positioned in A, B,
C, and D are obtained as functions of the previous coordinates, considering the dimensions of the body.

FRONT LATERAL TOP

FIGURE 3: VIEWS OF THE SIMPLIFIED ENGINE BLOCK

𝛥𝑥𝐴 𝑥 − 𝜃𝑧 𝑏1 − 𝜃𝑦 ℎ 𝛥𝑥𝐵 𝑥 + 𝜃𝑧 𝑏2 − 𝜃𝑦 ℎ
∆𝑨 = [𝛥𝑦𝐴 ] = [ 𝑦 + 𝜃𝑧 𝑙1 + 𝜃𝑥 ℎ ] ∆𝑩 = [𝛥𝑦𝐵 ] = [ 𝑦 + 𝜃𝑧 𝑙1 + 𝜃𝑥 ℎ ]
𝛥𝑧𝐴 𝑧 − 𝜃𝑦 𝑙1 + 𝜃𝑥 𝑏1 𝛥𝑧𝐵 𝑧 − 𝜃𝑦 𝑙1 − 𝜃𝑥 𝑏2
(2)
𝛥𝑥𝐶 𝑥 − 𝜃𝑦 ℎ + 𝜃𝑧 𝑏2 𝛥𝑥𝐷 𝑥 − 𝜃𝑦 ℎ − 𝜃𝑧 𝑏1
∆𝑪 = [𝛥𝑦𝐶 ] = [ 𝑦 + 𝜃𝑥 ℎ − 𝜃𝑧 𝑙2 ] ∆𝑫 = [𝛥𝑦𝐷 ] = [ 𝑦 + 𝜃𝑥 ℎ − 𝜃𝑧 𝑙2 ]
𝛥𝑧𝐶 𝑧 − 𝜃𝑥 𝑏2 + 𝜃𝑦 𝑙2 𝛥𝑧𝐷 𝑧 + 𝜃𝑥 𝑏1 + 𝜃𝑦 𝑙2
At this point, initially neglecting the force analysis, presented later, the kinetic energy T is calculated
under the assumption of small translations and rotations as follows:
1 1
𝑇 = 𝑚(𝑥𝐺̇ 2 + 𝑦̇ 𝐺 2 + 𝑧̇𝐺 2 ) + [𝐽𝑥𝑥 𝜃̇𝑥2 + 𝐽𝑦𝑦 𝜃̇𝑦2 + 𝐽𝑧𝑧 𝜃̇𝑧2 + 2𝐽𝑥𝑦 𝜃̇𝑥 𝜃̇𝑦 + 2𝐽𝑥𝑧 𝜃̇𝑥 𝜃̇𝑧 + 2𝐽𝑦𝑧 𝜃̇𝑦 𝜃̇𝑧 ] (3)
2 2
The second term considered is the potential energy, calculated by summing up its respective
components in the three principal directions:
1 1
𝑈 = 𝑈𝑥 + 𝑈𝑦 + 𝑈𝑧 = 𝑘𝑥 (𝛥2𝑥𝐴 + 𝛥2𝑥𝐵 + 𝛥2𝑥𝐶 + 𝛥2𝑥𝐷 ) + 𝑘𝑦 (𝛥2𝑦𝐴 + 𝛥2𝑦𝐵 +
2 2 (4)
1
𝛥2𝑦𝐶 + 𝛥2𝑦𝐷 ) + 𝑘𝑧 (𝛥2𝑧𝐴 + 𝛥2𝑧𝐵 + 𝛥2𝑧𝐶 + 𝛥2𝑧𝐷 )
2
For simplicity, the stiffness of the four supports has been considered of equal value along the same axes.
Finally, the last term considered is the dissipation function (Rayleigh), subsequently employed to
determine the damping matrix:
𝑧 𝐷 𝐷 𝐷 𝐷
1 1 1 1
𝐷 = ∑ ∑ 𝐶𝑖 𝛥̇2𝑖𝑝 = 𝐷𝑥 + 𝐷𝑦 + 𝐷𝑧 = 𝐶𝑥 ∑ 𝛥̇2𝑥𝑝 + 𝐶𝑦 ∑ 𝛥̇2𝑦𝑝 + 𝐶𝑧 ∑ 𝛥̇2𝑧𝑝 (5)
2 2 2 2
𝑖=𝑥 𝑝=𝐴 𝑝=𝐴 𝑝=𝐴 𝑝=𝐴
The equation (1) can subsequently be reformulated in matrix form:

4
 𝑴𝒒̈ + 𝑪𝒒̇ + 𝑲𝒒 = 𝑸 (6)
Where the matrix M is the mass matrix, being directly associated with the expression of kinetic energy,
which remains consistently positive, is symmetric and positive-definite. C is known as the damping
matrix and is symmetric too. Conversely, the matrix K is known as the stiffness matrix that is both
symmetric according to Betti's reciprocity theorem and positive semi-definite as it is associated with the
potential energy of the rigid body, which can be positive or, at most, zero.
𝑚 0 0 0 0 0
0 𝑚 0 0 0 0
0 0 𝑚 0 0 0
𝑴= 0 0 0 𝐽𝑥𝑥 𝐽𝑥𝑦 𝐽𝑥𝑧 (7)
0 0 0 𝐽𝑦𝑥 𝐽𝑦𝑦 𝐽𝑦𝑧
[0 0 0 𝐽𝑧𝑥 𝐽𝑧𝑦 𝐽𝑧𝑧 ]
4𝑘𝑥 0 0 0 −4𝑘𝑥 ℎ 2𝑘𝑥 (𝑏2 − 𝑏1 )
0 4𝑘𝑦 0 4𝑘𝑦 ℎ 0 2𝑘𝑦 (𝑙1 − 𝑙2 )
0 0 4𝑘𝑧 2𝑘𝑧 (𝑏1 − 𝑏2 ) 2𝑘𝑧 (𝑙2 − 𝑙1 ) 0
𝑲= 0 4𝑘𝑦 ℎ 2𝑘𝑧 (𝑏1 − 𝑏2 ) 4𝑘𝑦 ℎ2 + 2𝑘𝑧 (𝑏12 + 𝑏22 ) 𝑘𝑧 (−𝑙1 𝑏1 + 𝑙1 𝑏2 + 𝑙2 𝑏1 − 𝑙2 𝑏2 ) 2𝑘𝑥 ℎ(𝑏1 − 𝑏2 ) (8)
−4𝑘𝑥 ℎ 0 2𝑘𝑧 (𝑙2 − 𝑙1 ) 𝑘𝑧 (−𝑙1 𝑏1 + 𝑙1 𝑏2 + 𝑙2 𝑏1 − 𝑙2 𝑏2 ) 4𝑘𝑥 ℎ2 + 2𝑘𝑧 (𝑙12 + 𝑙22 ) 2𝑘𝑥 ℎ(𝑏1 − 𝑏2 )
[2𝑘𝑥 (𝑏2 − 𝑏1 ) 2𝑘𝑦 (𝑙1 − 𝑙2 ) 0 2𝑘𝑦 ℎ(𝑙1 − 𝑙2 ) 2𝑘𝑥 ℎ(𝑏1 − 𝑏2 ) 2𝑘𝑥 (𝑏12 + 𝑏22 ) + 2𝑘𝑦 (𝑙12 + 𝑙22 )]

However, due to the lack of data regarding the coefficients of C, a more common approach in engineering
practice, known as proportional damping, is adopted instead of considering the dissipation function. In
this method, by setting a target value for the damping ratios ‘ζi’, the parameters α and β are adjusted to
distribute damping across both low and high frequencies and C matrix is then obtained:
1 𝛼 (9)
𝜁𝑖 = ( + 𝛽𝜔𝑟𝑖 ) , 𝑖 = 1, … , 𝑛
2 𝜔𝑟𝑖

𝑪 = 𝛼𝑴 + 𝛽𝑲 (10)
At this stage, natural frequencies and modal shapes can be computed by solving a non-standard
eigenvalue problem considering the undamped system, as shown:
ⅆ𝑒𝑡(𝑲 − 𝜔2 𝑴) = 0 (11)
Where the ωi2 (i=1, …, n) in the equation are the eigenvalues of the matrix and represent the square of
the system’s natural frequencies, while each vibrational mode shape is obtained through the eigenvector
ui associated to each eigenvalue, given by: [3]
(𝑲 − 𝜔𝑖 2 𝑴)𝒖𝒊 = 𝟎 (12)
Let’s now calculate the forces to complete the right-hand side
of the Lagrange's equation. First of all, as a simplification,
forces and torques deriving from the combustion phase are
not considered because they are essential for generating the
driving torque. Additionally, these forces are internal to the
system and have a net result of zero (they indeed only
produce a pure torque, which is the driving torque).
The force analysis begins with the displacement (s) of the
piston in the simplified system involving a single piston
connected to the connecting rod of length (l), which in turn
is linked to the crankpin with a radius (r), as illustrated. FIGURE 4: THRUST CRANK MECHANISM

𝑟 2
𝑠𝑖 = 𝑟 𝑐𝑜𝑠 Ɵ𝑖 + 𝑙 √1 − ( 𝑠𝑖𝑛Ɵ𝑖 ) 𝑓𝑜𝑟 𝑖 = 1: 6 (13)
𝑙
Considering the angular speed (ω) constant, the displacement is differentiated twice with respect to
time to obtain the piston’s acceleration that, due to the natural balancing of the inertial forces in the V6-
90° type engine, is evaluated by means of the Fourier expansion of the acceleration exact form:

5
 𝑟[𝑙 2 − 2𝑙 2 𝑐𝑜𝑠 2 (𝜔𝑡 + 𝜓𝑖 ) − 𝑟 2 𝑠𝑖𝑛4 (𝜔𝑡 + 𝜓𝑖 )]
𝑠̈𝑖 = −𝑟𝜔2 {𝑐𝑜𝑠(𝜔𝑡 + 𝜓𝑖 ) − 3 } = −𝑟𝜔2 𝑓𝑖 (𝜃𝑖 ) 𝑓𝑜𝑟 𝑖 = 1: 6 (14)
[𝑙 2 − 𝑟 2 𝑠𝑖𝑛2 (𝜔𝑡 + 𝜓𝑖 )]2
where:
𝑛=4 𝑛=4
𝑎0
𝑓𝑖 (𝜃𝑖 ) = + ∑ 𝑎𝑛 𝑐𝑜𝑠(𝑛𝜃𝑖 ) + ∑ 𝑏𝑛 𝑠𝑖𝑛(𝑛𝜃𝑖 ) 𝑓𝑜𝑟 𝑖 = 1: 6 (15)
2
𝑛=1 𝑛=1
with 𝜃𝑖 = 𝜔𝑡 + 𝜓𝑖 , where 𝜓𝑖 is defined as the angle between the rod-journals corresponding to the 1st
and the ith cylinder:
2 4 𝜋 7 11 𝑇 (16)
𝝍 = [0, 𝜋, 𝜋, , 𝜋, 𝜋]
3 3 2 6 6
The coefficients an and bn are computed for each cylinder through a dedicated MATLAB script.
Subsequently, this acceleration expression is employed to determine the inertial forces 𝐹𝑖:
𝐹𝑖 = 𝑚𝐵 𝑠̈ 𝑖 = −𝑚𝐵 𝑟𝜔2 𝑓𝑖 (𝜃𝑖 ) 𝑓𝑜𝑟 𝑖 = 1: 6 (17)
Where mb is the translational mass, calculated as the sum of one piston mass and one conrod translating
mass. At this point, we project the forces along the y and z axes:
√2
𝐹 = (∑6 𝐹 − ∑3 𝐹 ) 𝐹 = (∑6 𝐹 )
√2 (18)
𝑦 2 𝑖=4 𝑖 𝑖=1 𝑖 𝑧 2 𝑖=1 𝑖

FIGURE 6: FORCES LEVER ARMS FIGURE 5: REFERENCE SYSTEM

The centrifugal forces are now analyzed: the centrifugal force of a single V2 is considered because two
connecting rods are placed in the same rod-journal:
𝐹𝐶𝑖 = 𝑚𝑎 𝜔2 𝑟(𝑠𝑖𝑛(ωt + 𝜑𝑖 ) + 𝑐𝑜𝑠(ωt + 𝜑𝑖 )) 𝑓𝑜𝑟 𝑖 = 1: 3 (19)
Where ma is the rotational mass, defined by the sum of two con-rods and the journal rotational masses,
φi stands for the phase shift between the three V2s. It's possible to see that the total centrifugal force is
given by the sum of three rotating vectors, phase-shifted by 120° and having the same magnitude: this
means that the total centrifugal force is always zero.

Now it is possible to calculate the inertia moment acting along the x-axis:
1 (−1 + √3)𝑟 9𝑟
𝑀𝑖𝑛,𝑖 = 𝑚𝑏 𝑟 2 𝜔 2 [ 𝑠𝑖𝑛(𝜔𝑡 + 𝜓𝑖 ) + 𝑠𝑖𝑛(2𝜔𝑡 + 2𝜓𝑖 ) − √3 𝑐𝑜𝑠(2𝜔𝑡 + 2𝜓𝑖 ) − 𝑠𝑖𝑛(3𝜔𝑡 + 3𝜓𝑖 ) + (20)
2 4𝑙 2𝑙
9𝑟
+ 𝑐𝑜𝑠 (3𝜔𝑡 + 3𝜓𝑖 )] 𝑓𝑜𝑟 𝑖 = 1: 6
2𝑙

It’s now possible to define the Q vector as follows:

6 6 6 6 6 6 𝑇

𝑸 = [𝐹𝑥 𝐹𝑦 𝐹𝑧 𝑀𝑥 𝑀𝑦 𝑇
𝑀𝑧 ] = [0 ∑ 𝐹𝑦,𝑖 ∑ 𝐹𝑧,𝑖 ∑ 𝑀𝑖𝑛,𝑖 + ∑ 𝐹𝑦,𝑖 ∙ ⅆ ∑ 𝐹𝑧,𝑖 ∙ 𝑙𝑖 ∑ 𝐹𝑦,𝑖 ∙ 𝑙𝑖 ] (21)
𝑖=1 𝑖=1 𝑖=1 𝑖=1 𝑖=1 𝑖=1

6
 NUMERICAL IMPLEMENTATION
This section addresses the numerical implementation of the theory discussed in the previous paragraph.
The numerical implementation is divided into 3 different scripts: the first computes the forcing
calculation, the second one regards the steady state condition and the last one analyzes the ramp test.
FORCING CALCULATION: the first script under analysis is “MV_project_forcing.m”, a script which
calculates the expression of pistons accelerations by means of the Fourier series expansion. From line
16, the integration data is defined: the integration interval [0, 2π], integral limits and number of
harmonics are defined. From line 26 to line 53 the accelerations of the pistons in the first three
cylinders, which belong to the right bank of the engine, are computed. Firstly, the angular displacements
between the respective crankpins and the expressions of the harmonic components of the piston
accelerations, denoted in theory as 𝒇𝒊 (𝜽𝒊 ), are defined. Subsequently, the a0 coefficient and an initial
value to fx, that represents the Fourier series expansion of 𝒇𝒊 (𝜽𝒊 ), are assigned. Finally, through a for-
end loop, the coefficients an and bn, along with fx, are calculated for each cylinder. In the successive lines,
the same calculations are performed for the other cylinders.

STEADY STATE RAMP: according to the “MV_project_ss.m” script, engine data and the motion equation
matrices M and K are defined, with a verification of their symmetry using the Cholesky decomposition.
Between lines 44 and 48, the time interval for the ramp is defined, and subsequently, it is subdivided
into 100 time intervals. A vector comprising the 100 corresponding time values (in respect to the ramp
test) and their respective angular velocities of interest is subsequently defined.
From line 115 to 149 the proportional damping matrix is defined. Firstly, the damping ratio values are
set equal to 0.1 for each DOF. Then, by solving of the Ordinary Least Square problem, the alpha and beta
values are obtained. As a solving strategy, the QR numerical solver is implemented.
To solve the motion equation, a modal analysis is performed for decoupling the initial ODEs in time,
which are then solved by means of the convolution integral.
MODAL ANALYSIS
From line 102 to line 114, the eigenvalues and eigenvectors are calculated by means of the eig function,
which gives as output two 6x6 matrix: the first matrix (U) columns are the eigenvectors, while the
second one is a diagonal matrix containing the eigenvalues, from which the natural frequency are
obtained.
From line 150 to line 171, the modal analysis is carried on. After the normalization of the eigenvectors
matrix, it’s possible to take advantage of the orthonormality conditions to verify that Λ is a diagonal
matrix containing the eigenvalues and I is an identity matrix:
𝑼𝑇 𝑴𝑼 = 𝑰 𝑼𝑇 𝑲𝑼 = 𝚲 (22)
In the next section, the problem discretization is performed. Firstly, the time interval is discretized into
small time intervals dt=0.0001s and the time vector is subsequently defined, being sure that the
Shannon-Nyquist theorem is respected, to avoid the aliasing phenomena: the sampling frequency
(reciprocating of dt) must be at least twice the maximum signal frequency.
After this section, some frequency specifications are defined, which will be used for the fft computation.
In the “calculation of the displacement” section starting from line 187, θ is computed for every regime,
the forcing matrix is converted to the modal space, then the responses to the unitary impulse are
computed for every DOF and for every instant of time by means of a for-end loop cycle. After these
calculations, the displacements are calculated by means of the conv MATLAB function.
100 REGIMES ANALYSIS
From line 357 to 470 the maximum and root mean square value of the 100 regimes displacements are
computed, by means of the max and rms MATLAB function.

7
In the next section, the three significant regimes spectrum is evaluated. First of all, the 3-dimensional
q_ss vector is reduced to a 2D one by means of the squeeze MATLAB function, then a fft is performed
for every DOFs.
Before plotting the spectrum (for simplicity only for the first DOF), a fftshift is performed so the zero-
frequency component is shifted to array center, so only the positive frequencies are considered.
From line 588 to 683, the velocities and accelerations of the COG are computed for all the regimes (and
plotted for the three significant regimes), by ‘squeezing’ q_ss and using the gradient MATLAB function.
In the next section, the forces transmitted to the chassis in all the regimes are computed. For the three
regimes of interest, they are also plotted. These forces are defined as:
𝑇𝑖,𝑝 = 𝑐𝑖 ∆̇𝑖,𝑝 + 𝑘𝑖 ∆𝑖,𝑝 𝑖 = 𝑥, 𝑦, 𝑧; 𝑝 = 𝐴, 𝐵, 𝐶, 𝐷 (23)

RAMP TEST: analyzing the script “MV_project.m”, from line 14 to 34 a selector is implemented, to
choose whether to analyze the ramp up (from 0 to 8000 rpm) or ramp down (from 8000 to 0rpm).
The engine data, the motion equation, and the time vector are defined in accordance with the steady
state ramp script.
To solve the motion equation, two different approaches are adopted:
• Modal analysis, decoupling of ODEs in time domain and solving using the convolution integral.
• Direct numerical integration of the ODEs in time domain.
The modal analysis, the convolutional integral, velocities and accelerations of the center of mass are
computed in accordance with the steady state ramp script, with the exception that the squeeze function
is not required, because all the matrices are two-dimensional.
DIRECT NUMERICAL INTEGRATION: the numerical solver ODE45 admits as input only ordinary
differential equation of the first order, so a vector u needs to be defined:
𝒖 = [𝒒 𝒒̇ ]𝑇 (24)
By rearranging the equation of motion, a system of two ODEs can be written:
ⅆ𝒒
= 𝒒̇
{ ⅆ𝑡 (25)
ⅆ𝒒̇ −1 −1 −1
= 𝑴 𝑪𝒒̇ − 𝑴 𝑲𝒒 + 𝑴 𝑸
ⅆ𝑡
By expressing the system in matrix form and applying the u vector definition:
𝒖̇ = 𝑨𝒖 + 𝑩 (26)
Within the MATLAB script, the Q components are initially converted from symbolic functions into time-
dependent functions using the matlabFunction command. Subsequently, matrices A and B are
computed. Note that 𝒖 is a 12-component vector, with the first six components representing the DOFs
of the system and the remaining six components representing their derivatives.
At line 507, the ode45 solver is employed to compute a solution with homogeneous initial conditions.
For the rising ramp test, the system starts from an equilibrium condition with zero initial displacements
and velocities, while for the ramp down case, the homogeneous IC are a simplification due to the initial
non-zero velocities. This translates in an initial incorrect behavior of the latter, which becomes correct
(verified because the trend is mirrored in respect to the rising ramp test result) after the transient finish.
The solver output consists of the time vector and the displacement vector, which are plotted in the
following lines.
In the next section, the forces transmitted to the chassis are defined as before and plotted.
Subsequently, from line 644 the normalized modes (eigenvectors) are plotted.
MODES PLOT: the modes plot, last section of the script, serve the purpose of giving a graphic
visualization of the interaction between every mode and its DOFs excitation contribute.

8
 RESULTS
FORCING CALCULATION
The script “forcing_calculation.m” gives as output the
matrices of the coefficient an and bn, and the harmonic
component of the piston acceleration plot, see figure xx.
As expected, due to the odd firing order of the Nettuno
engine, the piston accelerations for a fixed value of the
angular speed are shifted by 30° and 90°.

STEADY STATE RAMP: The first result obtained by the

steady state ramp script are the coefficient defining the FIGURE 7: ACCELERATIONS TREND
damping matrix:
𝛼 = 9.1183 𝛽 = 0.000894 (27)
Subsequently, the script validates the theory development by confirming that the M and K matrices are
symmetric and positive definite.
MODAL ANALYSIS: the most relevant results obtained from the modal analysis section are the
eigenvectors, stored as columns of the modes matrix and the natural circular frequencies 𝜔𝑛,𝑖 :
𝑟𝑎ⅆ (28)
𝝎𝒏 = [57.78 61.86 85.28 138.18 159.63 180.98]𝑇 [ ]
𝑠
MAX AND RMS VALUES: after solving for the displacement vector q by means of the convolutional
integral, the MAX and RMS values for each of the 100 regimes are plotted. For simplicity, only the first
two DOFs are shown here:

FIGURE 8: MAX (UP) AND RMS (DOWN) VALUE

Near the resonance frequencies, the MAX and RMS value increase rapidly, depending on which
frequency and which DOF is considered, because for every mode of the system one or more DOFs are
more excited than the others.
SIGNIFICANT REGIMES: The next results are regarding the three regimes of interest: the COG
displacement and acceleration are computed and plotted to visualize their trend along time. In figure 10
the displacement and accelerations of the first three DOFs are plotted. The max values are consistent
with the ones found before at those regimes. The COG displacements are a linear combination of
different periodic functions (whose can be seen in the displacement spectrum) and their amplitude
depends on the proximity of the regime considered to a resonance frequency and on the corresponding
force and moments magnitude. The most relevant value is recorded along the x direction.

9
 FIGURE 9: COG DISPLACEMENT AND ACCELERATIONS
SPECTRAL ANALYSIS: to visualize the harmonic components of the COG displacement, the spectrum of
q is plotted (for simplicity only for the first DOF, but it’s easy to plot also the other DOFs spectrum) for
the 3 significant regimes. The main contributes are found near to the considered regime frequency.

FIGURE 10: SPECTRUM AND DISPLACEMENT FOR 7500 RPM

TRANSMISSIBILITY: The highest value of displacement is along the z direction at 2000 rpm, which is the
closest regime to a resonance frequency. From the transmissibility plot, a consistent result is obtained:
the highest force value is in the z direction, at 2000 rpm, of about 775N.

RAMP TEST: The results are discussed of the ramp up subcase. The same considerations apply for the
ramp down subcase. The first interesting result of the ramp test is the forces trend in time, where the
magnitude grows quadratically and it’s the same for every cylinder.
After the force plot, the solutions obtained by the convolution integral in the modal space and the direct
numerical solution are computed and plotted, see figure 12

10
As expected, every DOFs displacements is
characterized by its highest values in
correspondence of the natural frequency that
has the highest modal participation for that
DOF. A similar behavior is found in some
accelerations plot.
Due to the increasing rotational speed, the COG
displacements frequency increases with time.
So, their second derivatives, which are the COG
accelerations along the DOFs, grow with time.
To visualize the difference between the two
methods, the displacements along the x
direction are plotted in the same chart. The FIGURE 11: INERTIAL FORCES TREND
results are satisfying, with a slightly error of
value about 0.015mm only in correspondence on the peaks of the signal.

FIGURE 12: ODE45 NUMERICAL SOLUTION

TRANSMISSIBILITY: The transmissibility trend is consistent with the displacement one, with the
maximum value of about 2kN in correspondence of the resonance frequency crossed during the ramp.
After crossing these critical frequencies, the force magnitude decreases very quickly and then rises
slowly. [5]

FIGURE 13: FORCES TRANSMITTED TO THE CHASSIS IN X DIRECTION

11
 FEA AND MULTIBODY DYNAMICS VALIDATION
To validate the modal analysis results from the MATLAB
script, a normal modes analysis has been performed by
means of a multibody dynamics software MSC Adams [6]. The
engine has been modeled as a rigid block, where the COG
position, the mass and the inertia matrix has been defined
from the data collected in the CAD model. Subsequently,
every engine mount has been simulated by three linear
springs, directed along the three-reference system axis. The
normal modes analysis gave as a result the natural
frequencies and the modal participation factors, which are
scalars that measure the interaction between the modes and
the directional excitation in a given reference frame. Larger
values indicate a stronger contribution to the dynamic
response.
FIGURE 14: ADAMS MODEL

FIGURE 15: MODAL PARTECIPATION FACTOR

To visualize the modal shapes and to validate the modal participation
factors, a finite element modal analysis of the engine has also been
performed. The mesh size ranges in between 1mm and 5mm,
depending on the most critical dimension of the component
considered. The material properties have been defined with the
correct values of densities but also with infinite stiffness values, to
obtain the correct inertia matrix and to satisfy the rigid body
assumption. The engine components have been connected by contact
surfaces, while the engine mounts have been modeled by 1D CBUSH
elements. A property defining the stiffness value along the x, y and z
direction has been assigned to every CBUSH element. [7]
FIGURE 16: FEM MODEL
The natural frequency obtained from the MATLAB SCRIPT, the
multibody dynamics software and the FEA are comparable, as figure 17 states. [8]

FIGURE 17: NATURAL FREQUENCIES COMPARISON

From the FEA results, it has been possible to observe the mode shape of the engine, by means of a
graphic post-processor. The results obtained are consistent with the modal participation factors
values obtained from the multibody dynamic analysis and with the MATLAB modes plot.

12
 CONCLUSION
This analysis is fundamental to analyze the vehicle NVH behavior, which is essential for achieving a
pleasurable driving experience and enhancing the overall driver sensations.
The steady state analysis allows to better understand the vibration magnitude and trend when cruising
at constant speed, while the ramp test is useful to analyze the vibration behavior when accelerating or
decelerating.
As expected, the inertial forces magnitude is quite small due to the natural forces balancing of the 90-
degree V6, while the moments intensity is significative.
Consequently, the displacements of the engine COG are negligible if compared to the rotations.
The natural frequency and the resonance frequency are close to each other due to the low damping ratio
value chosen. The analysis of those frequencies is crucial, because if the engine is operated in such a way
that the inertia forces and moments have the same frequencies, severe vibration amplitude take place,
which can cause noise and failures.
The magnitude of the forces transmitted to the chassis exhibits the maximum values in the resonance
frequency range, between 550 and 1800 rpm, and for frequencies near to it. After the resonance
frequencies are crossed, the force amplitude is much lower. To improve the vehicle low-speed
drivability, it’s possible to reduce those forces in the low rpm range by adding countershafts and tuning
the stiffness and damping values of the engine mounts.
The main problem found is that, when cruising on the highway at the speed legal limit in the top gear,
the engine rotates at 2000 rpm, which is a critical regime: the forces transmitted to the chassis and the
engine COG displacement exhibit great magnitude, mainly because the rotation frequency is in the
resonance frequencies range. This needs to be solved like discussed previously or by tuning the top gear
ratio to change the rotation speed frequency.
In a deeper analysis, it’s possible to not adopt the rigid body hypothesis, to include the engine stiffness
(and so its deformations) into the study.

REFERENCES
[1] Maserati S.P.A.
[2] Dassault Systems, 2023 SOLIDWORKS Student Edition.
[3] F. Pellicano, Mechanical Vibration, 2020 edition.
[4] M. Callegari, P. Fanghella, F. Pellicano, Meccanica applicata alle macchine, CittàStudiEdizioni,
2017.
[5] MathWorks, MATLAB R2023b
[6] Hexagon AB, MSC Adams Student Edition.
[7] Altair S.p.a., Hyperworks 2023 suite Student Edition
[8] Microsoft Corporation, Office 365 Excel

13