Вы находитесь на странице: 1из 7

Tire Pitch Sequence Optimization using HyperStudy

Ujjwal Kumar Dr. Jyoti Prakash G. Unnikrishnan Tom Thomas


Assistant Manager, Rath Vice President Executive Director
Advance Engineer- Head, R&D Technology and
ing(R&D) Advance Engineer- CEAT Limited, Halol, Projects
CEAT Limited, Halol, ing(R&D) Pin: 389350,INDIA CEAT Limited, Halol
Pin: 389350,INDIA CEAT Limited, Halol Pin: 389350,INDIA
Pin: 389350,INDIA

Abbreviations: GA-Genetic Algorithm, MOGA-Multi-Objective Genetic Algorithm

Keywords: Tire, Noise, Pitch sequence optimization, Air-pumping, Optimization, Genetic Algorithm,
HyperMath, HyperStudy
Abstract:

The tread of a tire is a combination of same pattern units (pitch) of different sizes. This arrangement contributes to various tire
performance parameters viz. noise, handling, wear etc. The challenge is to have an optimized pitch sequence that corresponds to
desired extrema of each performance parameter. An optimized pitch sequence distributes the acoustic energy to a broad range of
frequencies, reduces wear and provides good and stable handling. This necessitates a multi objective optimization engine. The
solver neutral interface of HyperStudy provides many optimization engines like Adaptive Response Surface Method, Method of
Feasible Directions, Multi-Objective Genetic Algorithm (MOGA), Gradient-Based Multi-Objective Method for Optimization (GMMO)
etc. Genetic algorithm works on finding global extremum of a single function. MOGA can search global extrema even of more than
one function. A comparative study was carried out for these optimization engines. The solution provided by genetic algorithm
based optimization was found to be promising. In order to have an optimized pitch sequence each performance parameter (noise,
handling, wear) was assigned as an objective function and desired global extremum of every objective function was found.

Introduction
Occupant safety and comfort are the major expectations of a customer from a passenger car. Automo-
tive industry always pays attention towards occupant comfort, safety and environmental issues. Noise
emission from a car has a major role in determining its comfort aspect. Vehicle manufacturers have
been particularly successful for significant noise reduction in engine, exhaustion and transmission sys-
tems. Tire noise is considered as a big culprit in deteriorating the comfort of occupant. A tire causes 80
percent of total vehicle noise in steady rolling and 23 percent in acceleration [1]. Also, many environ-
mental regulations (mostly in Europe) are directed towards tire noise emission. Tire component, tread,
comes in contact with the road. Tread consists of basic elements around the periphery of the tire, which
are called pitch. Tires tread is a major contributor in total tire noise emission.

Noise spectra of rotating bodies always exhibit very high amplitudes at discrete frequencies. These
peaks are attributed to the even spacing of the working elements. It is a well established fact that a rotat-
ing object having unevenly spaced elements exhibits reduced peak heights. Similarly in tires, uneven
sequencing of pitch (basic elements) reduces the height of peaks in noise spectrum [2]. These peaks
are termed as tonal noise. As far as tire noise quality is concerned, a broadband spectrum is always a
desired output. In a broadband spectrum, total acoustic energy is distributed over a wide range of fre-

Simulate to Innovate 1
quencies. On the other hand, a spectrum consisting of lofty peaks, acoustic energy is concentrated at
certain specific frequencies. This feature makes the emitted sound unpleasant. In the present work GA
based optimization techniques were utilized to arrive at better noise spectrum.
Along with noise, the tread pattern also influences many tire performance characteristics like wear, trac-
tion force and moment generation for handling, hydroplaning and rolling resistance [3]. Many of these
performance parameters are of conflicting nature. This conflicting nature necessitates the induction of
optimization process in pitch sequencing. GA based optimization techniques are effective and getting
popular for pitch sequence optimization. A passenger car tire (PCR) of size 155/70R13 is taken for this
study. The tires periphery (tread) is divided in to 65 equal pitches (Figure1). The resulting pattern noise
was simulated with the help of an in house developed code. This noise spectrum (Figure 1) was show-
ing peaks at harmonic number(s) 65 and its multiples. A new tread was designed having randomly dis-
tributed pitches of three different sizes (Figure1). The tread noise simulation result (Figure 2) shows a
broader spectrum.

Figure 1: Pitch distribution for the size 155/70R13

Simulate to Innovate 2
Figure 1: Pattern noise spectrum for tire size 155/70R13

Genetic Algorithm

Computer programs that "evolve" in ways that resemble natural selection can solve complex problems
even their creators do not fully understand-John H. Holland, pioneer of genetic algorithm. A genetic
algorithm (GA) is an optimization engine used for solving complex optimization task. It mimics the me-
thod of natural evolution. A classical algorithm solves a problem. In the case of genetic algorithm, solu-
tion is evolved. Natural evolution works on a very strict principle: Select the Best, discard the Rest. A
clear understanding of natural evolution process helps in understanding the GA optimization process.
Cells are the basic units of all living organisms. All cells are made of same set of chromosomes.
Chromosomes are an array of DNA. These blocks of DNA (genes) are responsible for the characteristics
e.g. hair colour. Nature evaluates the fitness of a population on the basis of some fitness function(s)
(ability to sustain climate change, ability to safeguard children from predator etc.).The new population
contains chromosomes of parents having better fitness. The new chromosomes are not a mere copy of
the parental chromosomes.

Simulate to Innovate 3
Figure 3: Flowchart for genetic algorithm

Crossover and mutation are the two processes which the offspring chromosomes will differ from the pa-
rental chromosomes. Crossover is a process in which most promising individuals of a population mate.
This could be understood from the Figure 3. Genes line up, a point along the strings is selected at ran-
dom and the portions to the left of that point are exchanged to produce two offspring: one containing the
symbols of the first string up to the crossover point and those of the second beyond it, and the other,
containing the complementary cross [4].

Figure 4: Crossover

When a chromosome undergoes the mutation process, one or more of its genes will be changed. In
mutation, the solution may change entirely from the previous solution. Mutation operation is explained in
Figure 5.

Simulate to Innovate 4
Figure 5: Mutation

Pitch sequence optimization

Tire noise generation is a combined result of various phenomena viz. whole body translation, cavity re-
sonance, stick-slip, air-pumping, pipe resonance, horn effect. Air pumping comes from the flow of air
trapped in between the road surface and tread pattern grooves. In a rolling tire, tread pattern undergoes
a loading-unloading cycle. This results in air being pumped out of the leading edge of the tire footprint
and sucked into the trailing edge. This pumping of air causes sound generation. This air pumping could
be characterized as an acoustic monopole [5]. The wave form can be expressed by a sinusoidal function
with exponential decay.[1]

f(t)=0 t<ti
(1)
-at
f(t)=e *sinbt t>ti

Where a and b are constants. The values of these could be measured from experiment. ti is the time
taken by signal in travelling up to measurement station.

After doing a Fouriers transformation the noise spectrum could be obtained in frequency domain. With
the help of HyperMath a function was formulated which could give the value of amplitude of this signal
for a particular harmonic number. The maximum of amplitude was termed as Noise Index. Value of
Noise Index depends on the pitch sequence. Higher value of Noise index leads to a high sound pres-
sure level (SPL). Hence, in our exercise of pitch sequence optimization, this noise index was kept as
objective function. Genetic algorithm (GA) tried to minimize this objective function. HyperStudy, a solver
neutral platform, provides optimization engine based on GA. The HyperMath code was registered as
solver. HyperStudy provides options for setting the operating environment for different GA operators Viz.
crossover rate, mutation probability, constraint violation threshold etc. proper study should be carried out
for defining these values. For example, mutation probability should not be too high otherwise it will
cause a slow convergence. A large crossover probability causes the optimum sequences disappear-
ance in the next generation [1].Either after sufficient number of iterations or finding convergence GA will
stop the optimization process. The result should be a new pitch sequence having smaller value of the
Noise Index.

Results & Discussion

Existing pitch sequence for tire size 155/70R13 was taken as the initial solution for optimization engine.
Noise Index corresponding to optimized pitch sequence was calculated using the above mentioned
HyperMath code. The amplitude harmonic number curve were plotted (Figure 6) for both the pitch se-
quences using HyperMath plot option. The optimized pitch sequence corresponds to a lower peak hence
low Noise Index Value.

Simulate to Innovate 5
Figure 6: Noise index comparison between existing and optimized pitch sequence

Benefits Summary

1. Very rich and diverse syntax library of HyperMath helped to formulate the objective function for
noise optimization. Hence, no separate software was required to do a pattern noise simulation.
2. The option of registering HyperMath as a solver in HyperStudy provides opportunity for pitch
sequence optimization.
3. This exercise demonstrates the manifold enhancement in HyperWorks capabilities if its different
applications (viz. HyperMesh and HyperStudy or HyperMath and HyperStudy) are used togeth-
er.

Challenges

Converting an exponentially decaying function into an objective function was a difficult task. Hyper-
Maths capabilities were a great help.

Future Plans

Expansion of this method in the field of wear and handling is planned.

Conclusions

HyperMath provides a platform where users could overcome the difficulty faced due to limitations or un-
availability of commercial codes. For pitch sequence optimization there are very limited codes. Hyper-
Math and HyperStudy, combined together provide new opportunity for further work.

ACKNOWLEDGMENTS

The authors would like to thank the CEAT management for allowing us to publish this work. We would
also like to thank Altair Technical Support and ATC2013 for providing us an opportunity to present this
work

Simulate to Innovate 6
REFERENCES

[1] Nakajima, Yukio; and Abe, Akihiko, Application of Genetic Algorithms for Optimization of Tire Pitch Sequences, Japan J.
Indust. Appl. Math., 17, (2000) 404-426.
[2]Hoffmeister, K. M. and Bernard, J. E., "Tread Pitch Arrangement Optimization Through the Use of a Genetic Algorithm," Tire
Science and Technology, TSTCA, Vol. 26, No. 1, January-March 1998, pp. 2-22.
[3] Prakash, J., Kumar, U., Unnikrishnan, G., & Thomas, T. Facilitation of symmetric result transfer and convergence of a non-
linear steady state simulation of tire through scripting.
[4] http://www2.econ.iastate.edu/tesfatsi/holland.gaintro.htm
[5] Hayden, R. E., "Roadside Noise from the Interaction of a Rolling Tire with the Road Surface, "Proceedings of the Purdue Noise
Control Conference, Pro'due University, W. Lafayette,Ind., 1971, p. 59.

Simulate to Innovate 7

Вам также может понравиться