ACTA UNIVERSITATIS AGRICULTURAE ET SILVICULTURAE MENDELIANAE BRUNENSIS

Volume 65 45 Number 2, 2017

https://doi.org/10.11118/actaun201765020419

SIMULATION OF FAILURE IN
GEARBOX USING MSC.ADAMS

Jan Furch1, Trung Tin Nguyen1

Department of Combat and Special Vehicles, Faculty of Military Technology, University of Defence, Kounicova 65,
1

662 10 Brno, Czech Republic

Abstract

FURCH JAN, TIN NGUYEN TRUNG. 2017. Simulation of Failure in Gearbox Using MSC.Adams.
Acta Universitatis Agriculturae et Silviculturae Mendelianae Brunensis, 65(2): 419–428.

Vehicle’s gearbox is regarded as one of the most crucial elements in a vehicle but it sustains a variety
of faults such as a broken tooth, misalignment, imbalance, looseness, and even a broken case. Using
accelerometers, which are mounted on the case to measure a vibration signal, we can detect faults
and monitor the condition but the signals are often complicated and difficult to interpret. Moreover,
it is costly and almost impossible to change the structure of a gearbox in order to survey, for instance,
the dependence of a vibration signal on the structure, or types of a fault. In this paper a dynamic
model of gearbox was developed with tooth breakage to study the capability of simulation using
MSC.Adams software. In this simulation we focus on the gearbox used in a common military vehicle.
The simulation was based on a nonlinear contact force to predict what happens to the gearbox in case
the tooth breaks.
The contact mechanics model of the meshing teeth is studied thoroughly by selecting contact
simulation parameters such as stiffness, force exponent, damping and friction coefficients. To
simulate the real working environment of the gearbox, simulated bearings were also built in the MSC.
Adams. The paper shows that it is possible to simulate vibration signals by the gearbox model created
in 3D CAD software and analyze the results in the multi‑body dynamics software MSC.Adams.

Keywords: gearbox, cracked tooth, simulation, model, vibration signal, contact force, MSC.Adams.

INTRODUCTION correctly the vibrational and acoustic performance

In practice the aim of modelling is to understand of the powertrain.
the physical phenomena, imitate the behaviour of The modelling of the dynamics is really useful
the examined machine in virtual environments, in case of conditional monitoring, since it allows
simulate it on a model and then affect its behaviour us to simulate a vibration signal for different
by a specific way. The virtual model is just an conditions of a gearbox which sometimes cannot be
approximation of reality since the real system can be checked in an experimental way. There are several
very complex and the model might not fit the system methods to simulate the behaviour of a gearbox
completely. such as mathematic modelling (e.g. Matlab
Modelling and simulation are two slightly Simulink), finite element methods (e.g. Ansys) and
different processes but usually carried out together. multi‑body dynamic methods (e.g. MSC.Adams).
They are used to make models based on a real object The last approach seems to be the most suitable
and allow us to experiment with these models (see for a vibration signal analysis, because it allows us
Fig. 1). to analyse contact force in the time domain even in
In the paper by Novotny, Prokop, Zubik and the spectrum which has the ability to express exactly
Rehak, 2016, there are introduced advanced the nonlinear effects of stiffness, gear backlash,
computational models suitable for developing rotation bearings, eccentricity, misalignment,
a modern powertrain and focusing on noise and unbalance and other nonlinear phenomena.
vibration. The aim is to decide to what extent MSC.Adams (Automatic Dynamic Analysis of
the model should be accurate in order to describe Mechanical System) is computing software used for

419
420 Jan Furch, Trung Tin Nguyen

Real object Computer model Model realization

Modelling Simulation

1: Diagram of modelling and simulation (Furch, Glos and Nguyen, 2016)

modelling, analysing and optimizing mechanical MATERIALS AND METHODS

systems called MBS (Multi‑Body System). Other
systems are Simpack and LMS Virtual.Lab Motion. Establishment of physical components
All of them allow us to execute the analysis of
This procedure needs to be accomplished by
three‑dimensional bodies. However, there are
3D‑model software such as ProEngineer, Catia,
many different algorithms proposed, e.g. teeth and
Solid Work and Inventor. The last software was
body of gear wheels are assumed to be rigid but
chosen to model components in the gearbox.
connected by elastic elements, or the calculation
The advantage of the software is the possibility of
based on estimating gear noise, or the use of
a parametric modelling process that refers to a set of
a nonlinear contact algorithm between teeth of
parameters. We can set up a structure, a shape, a size
gears and geometric defects, etc. MSC. Adams
and features of parts in the gearbox. In the software
consists of a number of different modules such
there are some standardized bodies such as gears,
as Adams/View, Adams/Car, Adams/Chassis, and
bearings and shafts. Using this method and selecting
Adams/Driveline. The MSC.Adams/View and
suitable parameters, we can create standardized
MSC.Adams/Postprocessing are the most common
bodies with different specifications and meet
modules. The former is used to carry out a virtual
the desired design. As long as basic parameters
experiment on a model and the latter is used to
(such as the number of teeth, modulus, pressure
analyse the results collected from the experiment
angle, helix angle, face‑width, diameter, and length)
(Mechanical Dynamics, 2008).
are inputted or modified, a 3D geometric model will
A virtual experiment in MSC.Adams is
be automatically generated or rebuilt. In the „spur
presented in the following study. The purpose of
gear component generator“ and „shaft component
the experiment is to simulate vibration signals
generator“, which are two among many interfaces of
when tooth breakage in the gearbox occurs. Tooth
Inventor, every dimension or other parameters such
breakage is one of the most common and dangerous
as the number of teeth, modulus, gear ratio, center
operating failures associated with increased noise
distance (see Tab. I) must be completed by entering
and vibration level, destruction of gears and other
precise values so that the software can create exactly
structures.
desired components and put them into the right
Basic procedures needed to accomplish
position. Because it is known that the parameters
the analysis of a vibration signal in MSC.Adams are
of a gear are very closely related to each other, this
introduced below:
work requires good knowledge of the gearbox.
• Create physical bodies including exact geometrical
Next, the model needs to be transferred from
shapes such as length, width, depth, dimension,
Inventor to MSC.Adams. There are two ways to
and distance between shafts.
perform the data conversion between Inventor and
• Apply physical characteristics such as rotational MSC.Adams:
inertia, mass and dynamic friction coefficient. a) Using intermediate types of format (Parasolid,
• Fulfil kinematics definition such as translation and Step, Iges, Sat etc.) that can help MSC.Adams to
rotation. identify exactly geometric shapes of components
• Simulation that relates to making motion, forces created by another software.
and observing model behaviour. b) Direct transfer using interface provided by MSC.
• Record and analyse results in the MSC.Adams/ Adams that is able to keep every component,
Postprocessing (Furch, Glos and Nguyen, 2016; which was created in another software, as
Mechanical Dynamics, 2008). independent part in MSC.Adams. This feature
is very useful for next steps and is adopted to
perform data conversion. In MSC. Adams click
File→Import→.iam, and the Inventor file can be
successfully imported into MSC.Adams.
 Simulation of Failure in Gearbox Using MSC.Adams 421

I:  Basic geometrical data of the gearbox used for modelling (Russia Ulyanovsk JSC, 2000)
Face width Number of
Description Gear Module (mm) Types of tooth Helix angle (0)
(mm) teeth
The constant-mesh Drive gear 3 20 Helical 28.85 Z0R = 15
gears Driven gear 3 20 Helical 28.85 Z0N = 32
Drive gear 3.5 20 Spur Z1R = 15
1st speed gear
Driven gear 3.5 20 Spur Z1N = 29
Drive gear 3 20 Helical 28.85 Z2R = 21
2nd speed gear
Driven gear 3 20 Helical 28.85 Z2N = 26
Drive gear 3 20 Helical 28.85 Z3R = 27
3rd speed gear
Driven gear 3 20 Helical 28.85 Z3N = 20
Drive gear 3.5 15 Spur ZRR = 15
Reverse gears
Driven gear 3.5 15 Spur ZRN = 19
Underdrive Drive gear 3 15 Spur ZER = 24
engaging gears Driven gear 3 15 Spur ZEN = 37
Drive gear 3 15 Spur ZAR = 27
Drive gears to real
Driven gear 3 15 Spur ZAN = 34
and front axle
Driven gear 3 15 Spur ZAN = 34

2:  Model of the gearbox in MSC.Adams (Two-wheel-drive mode)

Legend: A. Model with all constraints; B. Model with all cases;
1. Input shaft with constant-mesh drive gear; 2. Single row deep groove ball bearing SKF 6208; 3. Roller bearing SKF NU
2204 between the input shaft (1) and the output shaft (6); 4. 3rd speed driven gear; 5. 2nd speed driven gear; 6. Output shaft
of gearbox; 7. 1st speed driven gear; 8. Double row angular contact ball bearing SKF 3207BN; 9. Rear axle and underdrive
engaging gear; 10. Rear axle drive shaft with gear; 11. Single row deep groove ball bearing SKF 6307; 12. Layshaft of transfer
box with idle gear; 13. Front axle drive gear; 14. Front axle drive shaft with gear; 15. Transfer gearbox case; 16. Single row
deep groove ball bearing SKF 6306; 17. Layshaft (or countershaft) with 1st speed drive gear; 18. 2nd speed drive gear; 19. 3rd
speed driver gear; 20. Constant-mesh driven gear; 21. Single row deep groove ball bearing SKF 6305; 22. Gearbox case;
23. Reverse gears.
422 Jan Furch, Trung Tin Nguyen

When the gearbox model is imported with of a tooth by load during meshing, varies in time in
the above method, each component stands by itself real gears and is theoretically changed according to
in MSC.Adams and has no connection with another a parabolic function and determined by parameters
one. Because of this, they will not constitute a real of contact algorithm.
virtual model and will not work as required. Next, There are two types of contact that can be
two procedures need to be fulfilled:  modelled between the surfaces of the contacted
a) Adding material parameters, so that the physical bodies. The first one is a discontinuous contact such
data of gears and shafts such as centroid position, as a falling ball bouncing on the floor. The other
mass, stiffness and the rotation inertia defined one is a continuous contact where the contact is
as equation (1), can be obtained (Norton and defined as a nonlinear spring (see Fig. 3) (Meagher,
Karczub, 2003). WU and KONG, 2010). Both of them are respectively
available in MSC.Adams/View as the Restitution
 1 2 Method which calculates the collision force by
I xx = 2 mr a recovery coefficient, and the Impact Method

I= I= 1 m(3r 2 + L2 ) (1) which uses the stiffness and damping coefficients
 yy zz
12 to calculate the contact force. Considering
the essence of the problem, the latter is adopted.
Where In this algorithm gears and shafts are considered to
m ���������mass of the components, be rigid bodies but contact surfaces between teeth
r �����������radius of the components, are flexible. In order to simulate vibration signals
L ����������width of the components, during meshing from pairs of gears in the gearbox,
xx ��������rotating axis, contact force and Coulomb friction between teeth of
yy, zz ���perpendicular axes with rotating axis. two gear wheels were assumed.
In the Impact Method algorithm calculating
b) Adding torque and kinematic constraints. If the contact force can be defined as follows (Norton
the torque and the motion applied to the gears and Karczub, 2003):
need to be transferred to the shaft, a fixed joint
is added to every gear wheel and a corresponding K ( x0 + x )e + CSx x < x0
shaft. A revolute and cylindrical joint is added to F =
0 x ≥ x0
every shaft, so that it can rotate around its own (2)
axis. A constant rotational motion is applied to 0 x > x0
a revolute joint on the input shaft. This represents 
a virtual rotation from the engine and resistive S =[3 − 2( x0 − x )]( x0 − x )2 x0 − d < x < x0
torques which are applied on the output shafts as 1 x ≤ x0 − d

load from the axles. All the SKF virtual bearings,
which were chosen on the basis of the equivalent
In the equation (2), S is a step function, x0 − x is
parameters of the real Russian bearings, are also
the deformation in the process of contact‑collision.
added to the model (see Fig. 2).
This equation expresses that the contact will not
occur and F = 0 while x ≥ x0 (see Fig. 4). The contact
Determination of the contact force
will occur while x < x0 and the value of contact
parameters force is related to other parameters such as stiffness
Although numerous variants of gears dynamics K, deformation x0 − x, contact force exponent e,
modelling can be considered, we selected damping coefficient C and penetration depth d
the approach that takes into account the phenomena which is the maximum value of x0 − x.
occurring only inside a gearbox. This approach is The equation (2) also implies that the contact force
used for calculating dynamic contact forces between defined in MSC.Adams is composed from two parts,
teeth. By doing so, it can identify the signs of teeth an elastic component K(x0  − x)e acts like a nonlinear
failures. Meshing stiffness, which depends on spring and the other is the damping force CS(dx / dt)
the number of intermeshing teeth and the deflection which is a function of the relative deforming

3:  Contact force is defined as a nonlinear spring (Norton and Karczub, 2003).
 Simulation of Failure in Gearbox Using MSC.Adams 423

4:  The contact force is related to the deformation x, x0 (left) and the dependence of damping force F on penetration depth d (right) (Kong,
Meagher, Xu, Wu and Wu, 2008).

velocity or a cubic function of penetration depth Young’s modulus of two contacting bodies, i – gear
d. In order to avoid the function discontinuity ratio, d1 – diameter of standard pitch circle, α’t
caused by the dramatic variation of the damping and αt – transverse pressure angle at engaged and
force while contact‑collision occurs, the damping standard pitch circle, β and βb – helical angle at
force is set to zero when the penetration depth of the pitch and base circle, ν1 and ν2 – Poisson ratio of
the two contacted bodies is zero and approaches drive gear and driven gear, E1 and E2 – the Young’s
to a maximum value Fmax when the specified modulus of the material of two gears respectively. As
penetration depth d is reached. a result, the contact stiffness can be calculated and
All of the contact force parameters are considered listed in the Tab. II.
as follows: 2. The gears of the gearbox are assumed to be made
1. According to the Hertzian elastic contact theory, from alloy steel with Young modulus E  =  2.1  ×  105
the stiffness of the two contacted bodies could be N/mm2 and Poisson ratio ν = 0.29 (Fang, 2013).
described by a pair of ideal contacted cylindrical 3. Force exponent e = 1.5 is based on the calculated
bodies and could be defined as follows (Norton speed and it is the result of numerous trial
and Karczub, 2003): simulations.
 1 4. Penetration depth d = 0.1 mm is based on
1
 4 2 * 4  id1 cosα t tan α t'  2 the numerical convergence in MSC.Adams.
=K = R E  
3 3  2(1 + i)cos β b  5. Damping coefficient C = 3,000 Ns/mm because,

 generally speaking, the damping coefficient
2 2
 1 1 −ν 1 1 −ν 2 ranges from 0.1 to 1 % of the stiffness K (Wu,
= * +
E1 E2 (3) Sommer and Meagher, 2016).
E
 β = α tan(tan β cosα t ) 6. Between gears there is a friction force referred
 b to as Coulomb friction in MSC.Adams,
1 1 1 The variables such as dynamic or static friction
=R R
+
R
 1 2 coefficient and related velocities are chosen
In the equation (3), R1 and R2 – equivalent from mechanical handbooks; it is expected that
radius of two gear contact points, E* – equivalent the gearbox is lubricated well (see Tab. III).

II:  Contact stiffness of pairs of gear (K)

Pair of gears Values (N/mm3/2)
The constant-mesh gears 6.48 × 105
1 speed gear
st
6.84 × 105
2 speed gear
nd
5.59 × 105
3rd speed gear 5.12 × 105
Reverse gears 6.98 × 105
Underdrive engaging gears 7.22 × 105
Drive gears to real and front axle 7.56 × 105

III:  The variables for defining the friction force (Kong, Meagher, Xu, Wu and Wu, 2008).
Variables Values
Static friction coefficient (µs) 0.1
Static transonic speed (vs) 1 (mm/s)
Dynamic friction coefficient (µd) 0.08
Dynamic transonic speed (vd) 10 (mm/s)
424 Jan Furch, Trung Tin Nguyen

Algorithms for the dynamic simulation where x(t) is the function of contact force with period
MSC.Adams offers four solvers (the Gstiff, T, an and bn are constants called the coefficients
Wstiff, Dstiff and Constant‑BDF) to solve of the transform and given by the Euler formulas
the Differential‑Algebra Equation (DAE) for (Bakir, 2008 ):
the multi‑body dynamic simulation. All of them T /2
1
use multi‑step, variable order algorithms and apply a0 =
T
∫ x(t )dx (5)
one of these three integration formats including −T /2
the Index3 (I3), Stabilized Index 1 (SI1) and 2 T /2 2π nt   n = 1, 2, … (6)
Stabilized Index 2 (SI2). an = ∫
T −T /2
x(t )cos
T
dt
The Gstiff and the Wstiff use a variable step
and fixed coefficients. The former helps us to
2 T /2 2π nt
calculate faster with higher accuracy, but when bn = ∫ x(t )sin T dt   n = 1, 2, …
T −T /2
(7)
computing velocity, they can make an error which
might excite discontinuities in acceleration.
However, in practice, the function of contact force is
Because of this, the error must be controlled by
a set of data with discrete and finite values xn (n = 1, 2,
limiting the maximum step during the simulation.
…). To perform the analysis using these finite values
The latter is more useful and stable because it could
of discrete data, the discrete Fourier transform (DFT)
be modified according to variable steps without
should be applied (Bakir, 2008 ):
any accuracy loss, but it requires more calculated
time than those by the Gstiff. The Dstiff algorithm 1 N −1
is similar to the Wstiff, but it allows us to choose Xn = ∑ xke − j 2π nk / N n = 0,1,2 …N−1 (8)
N k =0
only the integration format Index3. Unlike them,
the Constant‑BDF algorithm uses fixed steps, so where N is the number of xn in a constant interval Δt
it is very useful when SI2 format is selected with and Xn is called DFT of the discrete values x0, x1 … xN−1.
short step. Although it does not calculate as fast as Equation (8) will transfer correspondingly finite
the Gstiff and Wstiff, it also reaches high accuracy values on the time axis to the discrete spectra on
and it is not as sensitive to the discontinuity of the frequency axis. DFT can also use real numbers
the acceleration and force as the Gstiff (Mechanical instead of complex ones (Bakir, 2008 ):
Dynamics, 2008).
 1 N −1 2π nk
Integration formats differ a lot, for instance the I3  An = ∑
N k =0
xk cos
N
monitors only the error of the displacement and 
other state variables of the differential equations, but  N −1
 n = 0,1,2….N−1 (9)
B = 1 2π nk
not the velocities and constrained reaction forces.  n N ∑ x k sin
N
Therefore, its accuracy when calculating velocities,  k =0
acceleration and constrained reaction forces is not as where Xn = An+jBn.
high as that of the others. The SI1 is able to monitor We apply a window function in the work in
all state variables such as displacement, velocity and order to provide the discrete values that appear to
Lagrange multiplier by introducing the velocity be continuous and periodic. Discontinuities are
constrained equations instead of acceleration “filled in” by forcing the function of contact force
constrained equations. Therefore, it calculates to be equal to zero at the beginning and the end of
quite accurately but it is very sensitive to the models the calculated period.
with friction and contact problems. Unlike the SI1, There are many available windowing functions
the SI2 is able to control the errors of the Lagrange such as Rectangular (it is equivalent to saying
multiplier and velocity by considering the velocity that no window was used), Gaussian, Hamming,
constrained equations, so more accurate result Blackman‑Harris and Hanning. If N is used to
could be obtained for the velocity and acceleration represent the width of a signal sample, these
computation. windows w[n] are defined in the range 0 ≤ n ≤ N − 1
Based on the above information about the solvers as (note that outside 0 ≤ n ≤ N − 1 then w[n] = 0 for
and integration formats, the Wstiff solver with SI2 all cases) (Chitode, 2008):
integration is adopted for the dynamic simulation of Rectangular window:
the gearbox.
w[n]=1, 0; (10)
Algorithms analysing the contact force
Hanning window:
Because the function of contact force is periodic,
2π n
a Fourier transform (FFT) will be used to decompose w[=
n] 0,5 − 0,5cos ; (11)
it into a sum of simple harmonic functions, namely N
sines and cosines. Theoretically, the FFT can be Hamming window:
defined as follows (Bakir, 2008 ):
∞ 2π n
2π nt 2π nt w=
[n] 0,54 − 0,46cos ; (12)
a0 + ∑ (an cos
x (t ) = + bn sin ) (4) N
n =1 T T
 Simulation of Failure in Gearbox Using MSC.Adams 425

Gaussian: RESULTS AND DISCUSSION

2
 2π n  Gear mesh frequencies (GMF)
w=
[n] exp[ −0,5   ] ; (13)
 N  The gear mesh frequency also called “tooth mesh
frequency” is the rate at which gear teeth mesh
Blackman‑Harris window:
together in a gearbox. It is equal to the number of
2π n 4π n teeth on the gear times the rotation speed of the gear
w[n] =
0,423 − 0,498cos + 0,0792cos .       (14)
N N (Norton and Karczub, 2003):

After many times of trial calculation, fm = f.Z, (15)

Blackman‑Harris windowing function is used for
the work. where fm is the gear mesh frequency (Hz), f is
rotational frequency of the gear (Hz) and Z is
The operating conditions of the gearbox the number of teeth.
during the simulation The number of teeth on the drive gear times
The gearbox is used to transfer the engine power the speed of the drive gear must equal the number of
and the moment to the axles of the vehicle. In teeth on the driven gear times the speed of the driven
the process of operation it could suffer from heavy gear. As the pinion rotates against the driven gear,
stimulation of the engine, the axles, the clutch and the individual cycles of the frequency generated are
other resources such as oil, backlash, clearance. a profile of the individual teeth meshing. Gear mesh
The stimulation makes the changes of the dynamic frequencies of gears in the gearbox were calculated
load very complex which is the main reason as follows (see Tab. IV):
of vibrations, impact, noise and other signs.
Simulation results of gear teeth contact force
The selected operating conditions are based on
the working characteristic of the gearbox in the field The top of Fig. 6 shows the graph of the gear teeth
and they are among the huge number of possible contact force for the pair of 3rd speed gears under
choices which can be applied to the virtual gearbox. normal conditions.
• Rotation speed n = 1,500 rpm is applied on It is easily observed that the above results seem
the input shaft (Fig. 2 position 1). to fit the theoretical values (in Tab. IV) well. Note
• Position of 3rd speed gear (i = 1.58) is chosen in that values of the GMF are always repeated with
the gearbox and two‑wheel‑drive mode is chosen an integer multiplier also known as harmonics.
in the transfer box. The second and the third harmonics are very
important. In the spectrum of 3rd speed gears (see
• Load torque can be calculated from the maximum
top of Fig. 6), the second harmonics of GMF (2fm3) has
torque of the engine used on the vehicle (166.7 Nm)
a high amplitude with respect to the fundamental
and the respective ratio (i = 1.58), then the load
(fm3) and the third harmonics (3fm3) which would
torques of 263.386 Nm is applied on the output
indicate a backlash‑type problem. In other words,
shaft (Fig. 2 position 6).
the gears may have too much backlash, or one of
• Other parameters are applied as the gears may be oscillating. It is known that there
follows: the simulation time (end time) 1 s and are 4 pairs of gears used to transfer power from
the simulation step (step size) 0.0001 s. the input shaft to the layshaft and the outshaft
(see Fig. 2), but not all of them can have the same
perfect mesh condition. In either cases, the gears
are meshing on both sides of the teeth. This double
mesh generates the second harmonics, the phase of
which is 1800 out of the phase with the fundamental.
Next, a part of a tooth is removed, as shown
in Fig. 5, on the drive gear of 3rd speed gear to

IV:  The GMF in 3rd speed gear, two-wheel-drive mode and 1,500 rpm on the input shaft
Description Gear Number of teeth Values (Hz)
Drive gear Z0R = 15
The constant-mesh gears fm0 = 375
Driven gear Z0N = 32
Drive gear Z1R = 15
1st speed gear fm1 = 175.78
Driven gear Z1N = 29
Drive gear Z2R = 21
2nd speed gear fm2 = 246.094
Driven gear Z2N = 26
Drive gear Z3R = 27
3rd speed gear fm3 = 316.406
Driven gear Z3N = 20
426 Jan Furch, Trung Tin Nguyen

5:  Virtual tooth breakage on the pinion of 3rd speed gear

simulate a tooth breakage in the gearbox. Thanks are fractional gear mesh frequencies at ½ fm3, 1½ fm3,
to the virtual model, the cracked tooth may be 2½ fm3, etc. which are the result of the pulse generated
modelled more accurately than the one expressed in by the broken tooth on the drive gear, more precisely
the Fig. 5. speaking, they are caused by amplitude modulation
Each type of fault produces a unique form in which occurs because of the broken tooth. This
the spectrum. As a general rule, the distributed graph also shows that the amplitude of the spectral
faults such as eccentricity and gear misalignment line does not indicate the presence of the broken
will produce sidebands and harmonics that have tooth, but the existence of harmonics and
high amplitude close to the GMF. Unlike this, sub‑harmonics does it. Theoretically, the presence
the localized fault such as a cracked tooth produces and the density of these harmonics reflect quite
sidebands that are spread more widely across accurately the existence of faults and the variability
the spectrum. Each gear has one tooth in mesh, in technical condition. By comparing the graphs in
one tooth going out of mesh, and one tooth going Fig. 6, it can be discovered that the gear tooth break
into mesh. When the broken tooth goes into mesh, fault has altered the harmonic distribution.
a pulse is generated. The next tooth is good and When the fault becomes more serious,
stops the system from vibrating. Therefore the pulse the amplitudes of harmonics are not unchanged
generated by a broken tooth is well damped. as illustrated at the bottom of Fig. 6. However, note
Fig. 6 describes the frequency spectra taken that beside the basic harmonics we can observe
from the 3rd speed gear. The spectral line at fm3 is the existence of the huge number of spectral lines
the fundamental GMF, the spectral line at 2  fm3 that have different frequencies of rotational speed
is the second harmonics, the spectral line at 3  fm3 of the layshaft. These spectral lines are quite clear in
is the third harmonics. It is also observed that there the graph and demonstrate the broken tooth.
 Simulation of Failure in Gearbox Using MSC.Adams 427

6:  The contact force of 3rd speed gear under a normal condition (top), a cracked tooth (middle), and a more seriously cracked tooth (bottom)

CONCLUSION
We can conclude that the use of multibody dynamic software is really effective and accurate to
simulate not only the operation and the fault of gears, but also any type of fault in a gearbox. However,
it is a very complicated process. Firstly, the 3D model of an object, which is accomplished in a CAD
program, must be transferred successfully into a multibody dynamic environment. Secondly, physical
characteristics and all of the constraints must be applied reasonably and accurately. Thirdly, a suitable
algorithm with a lot of parameters must be chosen exactly. Finally, the contact force can be analysed
and transferred by FFT using Post processing. The process allows us to choose many parameters of
meshing such as stiffness, damping, friction force. The results show that any fault, defect or damage in
the gearbox can be simulated and detected by the spectrum analysis.
Acknowledgement
The presented work has been prepared with the support of the Ministry of Defence of the Czech
Republic, Partial Project for Institutional Development and Specific Research, Department of Combat
and Special Vehicles, University of Defence, Brno.
428 Jan Furch, Trung Tin Nguyen

REFERENCES
BAKIR, P. G. 2008. Vibration based structural health monitoring. Berlin: Technische Universitat Berlin.
CHITODE, J. S. 2008. Digital signal processing. 1st Edition. Pune: Technical Publications Pune.
DABROWSKI, D., ADAMCZYK, J. and PLASCENCIA, H. 2012. A Multi‑model of gears for simulation of
vibration signals for gears misalignment. Diagnostyka- Applied structural health, usage and condition monitoring,
62(2):  15 – 22.
FANG, B. 2013. CAE Methods on Vibration‑based Health Monitoring of Power Transmission Systems. San Luis Obispo: 
California Polytechnic State University.
FURCH, J., GLOS, J. and NGUYEN, T. T. 2016. Vibration analysis of manual transmission using physical
simulation. In: Deterioration Dependalility Diagnostics. Brno, 11 – 12 October. Brno: University of Defence,
57 – 68.
FURCH, J., GLOS, J. and NGUYEN, T. T. 2016. Modelling and simulation of mechanical gearbox vibrations.
In: International scientific conference Transport means 2016. Juodkrante, 5 – 7 October. Kaunas: Kaunas University
of Technology, vol. 1, 133 – 139.
MEAGHER, J., WU, X. and KONG, D. 2010. A Comparison of Gear Mesh Stiffness Modeling Strategies.
In: Proceedings of the IMAC – XXVIII. Jacksonville: Society for Experimental Mechanics Inc.
MECHANICAL DYNAMICS. 2000. Building Models in ADAMS/View. Michigan: Mechanical Dynamics.
NORTON. M. P. and KARCZUB. D. G. 2003. Fundamentals of Noise and Vibration Analysis for Engineers. 2nd edition.
Cambridge: Cambridge University Press.
NOVOTNÝ, P., PROKOP, A., ZUBÍK, M. and ŘEHÁK, K. 2016. Investigating the influence of computational
model complexity on noise and vibration modeling of powertrain. Journal of Vibroengineering, 22(4): 378 – 393.
RUSSIA ULYANOVSK JSC. 2000. Automobiles UAZ‑31512, UAZ‑31514, UAZ‑31519, Parts Catalogue.
Ulyanovsk: JSC UAZ.
KONG, D., MEAGHER, J. M., XU, C., WU, X., and WU, Y. 2008. Nonlinear Contact Analysis of Gear Teeth
for Malfunction Diagnostics. In: IMAC XXVI Conference and Exposition on Structural Dynamics. Orlando, 4
February. Bethel: Society for Experimental Mechanics.
WU, X., SOMMER, A. and MEAGHER, J. 2016. Spectrum Diagnostics of a Damaged Differential Planetary
Gear during Various Operating Conditions. Physical Science International Journal, 9(3):  1 – 13.

Contact information
Jan Furch: [email protected]
Trung Tin Nguyen: [email protected]