IOP Conference Series: Materials Science and Engineering

PAPER • OPEN ACCESS You may also like

- Design and topology optimization of a rack
Investigation on Multiple Algorithms for Multi- and pinion steering system using structural
and vibrational analysis
Objective Optimization of Gear Box T. Narendiranath Babu, E Rajkumar,
Tanmay Joshi et al.

- Hybrid and electric vehicle tribology: a

To cite this article: R Ananthapadmanabhan et al 2016 IOP Conf. Ser.: Mater. Sci. Eng. 149 012049 review
Hemanth G, Suresha B and
Ananthapadmanabha

- Experimental and numerical study of

contact fatigue for 18CrNiMo7-6 and
View the article online for updates and enhancements. 20MnCr5 carburized gear tooth
Oriana Palma Calabokis, Yamid Enrique
Núñez de la Rosa, Samara Paulin de
Moraes et al.

This content was downloaded from IP address 103.237.59.146 on 01/08/2023 at 07:01

IConAMMA-2016 IOP Publishing
IOP Conf. Series: Materials Science and Engineering 149 (2016) 012049 doi:10.1088/1757-899X/149/1/012049

Investigation on Multiple Algorithms for Multi-Objective

Optimization of Gear Box
1
R Ananthapadmanabhan, 2S Arun Babu, 3KR Hareendranath,
4
C Krishnamohan, 5 S. Krishnapillai and 6 Krishnan A
1,2,3,4
Department of Mechanical Engineering, Amrita Vishwa Vidyapeetham, Kollam,
Kerala-690525, India.
5
Professor, Department of Mechanical Engineering, IIT Madras, Chennai, Tamil
Nadu,India
6
Corresponding author, Assistant Professor, Department of Mechanical Engineering,
Amrita Vishwa Vidhyapeetham, Kollam, Kollam, Kerala- 690525, India E-mail:
[email protected]

email:[email protected], [email protected], [email protected],

3
[email protected],[email protected]

Abstract. The field of gear design is an extremely important area in engineering. In this work a
spur gear reduction unit is considered. A review of relevant literatures in the area of gear
design indicates that compact design of gearbox involves a complicated engineering analysis.
This work deals with the simultaneous optimization of the power and dimensions of a gearbox,
which are of conflicting nature. The focus is on developing a design space which is based on
module, pinion teeth and face-width by using MATLAB. The feasible points are obtained
through different multi-objective algorithms using various constraints obtained from different
novel literatures. Attention has been devoted in various novel constraints like critical scoring
criterion number, flash temperature, minimum film thickness, involute interference and contact
ratio. The output from various algorithms like genetic algorithm, fmincon (constrained
nonlinear minimization), NSGA-II etc. are compared to generate the best result. Hence, this is
a much more precise approach for obtaining practical values of the module, pinion teeth and
face-width for a minimum centre distance and a maximum power transmission for any given
material.

1. Introduction
Mechanical design includes an optimization process in which designers consider certain objectives
and constraints. Design of gear box is an art as well as a science of engineering. Designers, based on
the principles and their knowledge about the gears, lay out a gear for a particular application. The
community of engineers know that the principles of engineering alone cannot suggest a good design. It
is the designer’s expertise suggests good design. The problem with the conventional design procedures
is that it gives out only a single solution. Optimization is the act of obtaining the best result for the
given circumstances. However, design optimization of a complete mechanical assembly leads to a
complicated objective function with a large number of design variables. So, it is a better practice to
apply optimization techniques on individual components or intermediate assemblies rather than a
complete assembly.

Traditional gear design involves computations based on tooth bending strength, tooth surface
durability, tooth surface fatigue, interference, efficiency, and so on. Gear design involves a number of
empirical formulas, various graphs and tables which lead to a complicated design procedure. Manual
design is very difficult considering the above facts and thus, there is a need for the computer aided
design for gears. With the help of computers, iterative designing can be performed. Also, the design

Content from this work may be used under the terms of the Creative Commons Attribution 3.0 licence. Any further distribution
of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI.
Published under licence by IOP Publishing Ltd 1
IConAMMA-2016 IOP Publishing
IOP Conf. Series: Materials Science and Engineering 149 (2016) 012049 doi:10.1088/1757-899X/149/1/012049

variables which satisfy the given conditions can be determined. As the optimization problem involves
objective functions and constraints that are not stated as explicit functions of the design variables, it is
hard to solve it through classical optimization methods. Moreover, the increasing demand for compact,
efficient, and reliable gears forces the designers to use optimal design methodology.

The design of gears is a highly complicated task. The need to develop light weight, quiet and more
reliable gear designs has resulted in a variety of changes in the design process. Here, mathematical
programming techniques, generally known as numerical optimization methods, are investigated to
provide a reliable design methodology for gears. These techniques offer a logical approach to design
automation. It also handles a wide variety of design variables and constraints which are difficult to
visualize using graphical methods. Design optimization of spur gears is to reduce the size and weight,
and increase the power transmission capacity of the gear. As a part of this study, gear optimization
programs have been developed for spur gears. The programs evaluate a wide variety of functions
required by the optimizer. Typical design variables include the number of pinion teeth, module and
face width. Design objectives are maximum power and minimum centre distance. Constraints may be
imposed on flash temperature, film thickness, involute interference, scoring and contact ratio.

The problem is solved by following two ways:

(i) Using optimization toolbox of MATLAB,
(ii) Using code developed for multi-objective optimization (NSGA-II) technique.

2. Algorithms for Optimization

2.1. Genetic Algorithm
Genetic engineering is a field of tremendous growth. It is applicable to a variety of areas. People
are more interested in this field because of its possibility to build up the next generation of easier and
faster algorithms. They are also being used in multi-disciplinary design methods for optimal design.
Genetic algorithm (GA)is a set of evolutionary algorithms, and non-deterministic and random search
methods. It is an adaptive search optimization algorithm. It is based on the principles of natural
selection and natural genetics. It utilizes the theories of evolution and natural selection in order to
solve a problem within a complex solution space. In the process of evolution, reproduction
concentrates on high fitness individuals by exploiting the available fitness information. Hence, each
individual in the population receives a measure of its fitness in the environment.

Genetic Algorithm maintains a population of structures that are evolved upon the rules of selection
and other operations generally known as ‘search operators’ like recombination and mutation.
Recombination and mutation disturb those individuals providing general trial and error method for
exploration. GA begins with a population of randomly generated strings that represents the problem
and there possible solutions. Thereafter, each of these strings is evaluated to find its fitness. If a
satisfactory solution, based on the acceptability or search stoppage criterion exists, the search is
terminated. If not, the initial population is subjected to genetic evolution to procreate the next
generation of candidate solutions. The genetic method of procreation uses the population as the input.
The members of the population are processed by the four main GA operators – reproduction,
crossover, mutation and inversion –to create the progenies for the next generation of candidate
solutions. The resultant progenies are then evaluated and tested for termination.

2.2. NSGA-II
In this case, the problem is considered as a multi-objective problem. So, both objectives are treated
together. In general, in case of multi-objective optimization, the objectives are conflicting. So, a single
solution cannot be accepted as the best solution. Instead, a set of solutions is obtained which are better
than the other solutions in terms of both objectives which are called Pareto optimal solutions. Since
evolutionary algorithms are population based, they are the natural choice for solving this kind of

2
IConAMMA-2016 IOP Publishing
IOP Conf. Series: Materials Science and Engineering 149 (2016) 012049 doi:10.1088/1757-899X/149/1/012049

problem. In NSGA-II, the iterative procedure starts from an arbitrary population of solutions and
gradually the algorithm converges to a population of solutions lying on the Pareto optimal front with
higher diversity. The operators applied are the same as those of GA, namely, selection, crossover, and
mutation. The tournament selection operator is applied which also takes care of constraints. However,
in case of multi-objective optimization, additional task is to obtain solutions, which are as diverse as
possible. For that, the sharing function approach is used. Crossover and mutation operators are applied
as usual. The major problem with the inbuilt ‘fmincon’ function of MATLAB is that it considers all
the variables real. As a result, one has to round the optimum value of integer variable to the nearest
integer. However, GA can deal with both types of variables, integer and real, very easily by choosing
appropriate string length.

2.3. Fmincon
Fmincon finds the minimum of a constrained non-linear multivariable function. It is based on a
sequential quadratic programming. It finds a constrained minimum of a scalar function of several
variables starting from an initial estimate. This is commonly known as constrained non-linear
optimization. Fmincon may only give local solutions. When the problem is infeasible, fmincon
attempts to minimize the maximum constraint value. The objective functions and the constraint
functions must both be continuous and real-valued, that is why they cannot return complex values.

3. Design optimization
In this optimization problem for spur gear design, objective functions can be stated as:
Maximize: Power transmission
Minimize: Centre distance

Subjects to limit on
1) Flash temperature
2) Specific film thickness
3) Scoring criterion
4) Contact ratio
5) Involute interference

4. Design variables
The selection of the independent design variables, that are necessary to characterize the design of
the system ,is the key element in formulating an optimization problem. Generally, it is good to choose
those variables that have a significant impact on the objective function. Based on this criterion design
variables are chosen to be the number of pinion teeth (z1), module (m) and face-width (b).

5.Objective function
5.1.Centre distance
In general, the most desirable gear set is the smallest one that will perform the required job.
Because, smaller gears are less expensive, easier to manufacture, requires less material and needs only
less space to operate.
C = m*z1*(1+G)/2
m = module
z1 = no. of teeth on pinion
G = gear ratio

5.2.Power
It is desirable to increase the power output.

3
IConAMMA-2016 IOP Publishing
IOP Conf. Series: Materials Science and Engineering 149 (2016) 012049 doi:10.1088/1757-899X/149/1/012049

Ft ∗m∗z1∗2∗π∗N1
P=
2∗60
Ft = Tangential load(N)
N1 = RPM of pinion

6. Constrains
6.1. Flash temperature
Flash temperature should be within the limit, since the oil properties change when it exceeds the
temperature limit.
𝐶𝑡 ∗𝑓∗𝐹𝑡 ∗(𝑉1−𝑉2 )
Tf = Tb +
B
b∗(√V1 +√V2 ∗√ cos θ
2

Where,
Tb = Temperature of blank surface in contact zone (0F)
Ct = Material constant for conductivity, density and specific heat
(0.0528 for straight petroleum oil)
f = Coefficient of friction
Ft = Tangential driving load (lb)
b = Face width (inches)
V1 = Rolling velocity of pinion at point of contact tips (fps)
V2 = Rolling velocity of gear at point of contact tips (fps)
Vs = (V1-V2) = Sliding velocity (fps)
Fe = Face width in contact (in)
B = Width of band of contact (in)
Ѳ = Pressure angle
C = Centre distance in inches

𝑊 ∗ 𝜌 ∗ 𝜌2
B = 0.00054 √𝐹 ∗( 𝜌𝑡 + 𝜌1
𝑒 𝑝 𝑔 ) cos 𝜃

np ∗𝜋∗𝜌1
V1 = 360
np ∗𝜋∗𝜌2
V2 = 360

𝜌1 = 𝜌𝑝 and 𝜌2 = 𝜌𝑔 − 𝑧 ;

𝜌𝑝 = √𝑟02 − (𝑟 cos 𝜃)2 r0 = r +addendum

𝜌𝑔 = √𝑅02 − (𝑅 cos 𝜃)2 R0 = R + addendum

Length of the line of action,
z = 𝜌𝑝 + 𝜌𝑔 – CsinѲ

6.2. Specific film thickness;

To avoid scoring failure, specific film thickness > 1
Specific film thickness,

λ = (hmin/σ)
Where,
σ = √𝜎𝑝2 + 𝜎𝑔2

4
IConAMMA-2016 IOP Publishing
IOP Conf. Series: Materials Science and Engineering 149 (2016) 012049 doi:10.1088/1757-899X/149/1/012049

σp = Average roughness of pinion (rpm)

σg = Average roughness of gear (rpm)
hmin = Minimum oil film thickness (in)

Minimum elastohydrodynamic (EHD) film thickness,

2.65∗α0.54 ∗(V0 ∗u)0.7 E−0.03 R0.43
hmin= W0.13

Where,
α = Pressure viscosity coefficient
V0 = Absolute velocity
W = Wt/b = Specific loading,
1 1 1−µ21 1−µ22
𝐸
= [
2 𝐸1
+ 𝐸2
]
1
u = 2 [𝑢1 + 𝑢2 ]

u1and u2 are the rolling velocity of pinion and gear at the point of contact.
𝜋𝑛𝑝
u1 = 30
𝜌1
𝜋𝑛𝑔
u1 = 30
𝜌2
𝜌1 ∗𝜌2
R= 𝜌1 +𝜌2

6.3. Scoring criterion number

To avoid scoring failure,
Scoring criterion number < Critical scoring number.

Scoring criterion number = [(Ft/b)(3/2)]*[(np)(1/2)/(Pd)(1/4)]

Where,
Ft = Tangential driving load (lb)
b = Face width (in)
np = Speed of pinion (rpm)
Pd = Diametric pitch
6.4. Contact ratio
For standard spur gear, contact ratio should be greater than 1.4 and normally less than 2.
i.e., 1.4 < x < 2

2 −𝑟 2 +√𝑟 2 −𝑟 2 −𝐶 sin 𝜃) 𝑃
(√𝑟𝑎1 𝑏1 𝑎2 𝑏2
mp = 𝜋 cos 𝜃

Where,
ra1 = r + (1/Pd) r=radius of pinion
ra2 = R + (1/Pd) R=radius of gear
rb1 = r * cosѲ p=circular pitch
rb2 = R* cosѲ

5
IConAMMA-2016 IOP Publishing
IOP Conf. Series: Materials Science and Engineering 149 (2016) 012049 doi:10.1088/1757-899X/149/1/012049

6.5. Involute interference

Involute interference occurs when the pinion tooth makes contact with the gear tooth before the
involute portion of the pinion comes within the range i.e., when the contact occurs below the base
circle of the pinion on the non-involute portion of the flank.

For no involute interference,

2 2
CsinѲ>√𝑟𝑎2 − 𝑟𝑏2

7. Formulation of problem
Inputs:
Permissible bending stress = 191.6Mpa
Permissible contact stress = 455Mpa
Pinion RPM (N1) = 960
Gear RPM (N2) = 320

Design vector: x= (m,z1,b)

3 ≤ m ≤ 10
18≤ Z1 ≤ 24
30 ≤ b ≤ 70 (mm)

Objective function to be minimized: F(1) = m*z1*(1+G)/2

Ft ∗m∗z1∗2∗π∗N1
Objective function to be maximized: F(2) =
2∗60

Constrains to be satisfied:
C (1): Flash temperature ≤ 176
C (2): 1 ≤ specific film thickness ≤ 10
C (3): Scoring criterion number ≤ 8000
C (4): 1.4 ≤ contact ratio ≤ 2
C (5):For no involute interference,

2 2
CsinѲ>√𝑟𝑎2 − 𝑟𝑏2

8. Results and Discussion

Table 1. Comparison of outputs from Algorithms.

Variables GA1 GA2 Fmincon NSGA-II

Module 6.061 6.853 6.441 7.642
Number of Pinion Teeth 18.91 23.35 18.00 19.87
Face width(mm) 70.00 66.89 66.84 69.99
Power(kW) 24.32 46.49 23.69 43.65
Centre distance (mm) 229.5 320.2 231.7 304.0

GA1 – single objective genetic algorithm.

GA2 – multi objective genetic algorithm.

6
IConAMMA-2016 IOP Publishing
IOP Conf. Series: Materials Science and Engineering 149 (2016) 012049 doi:10.1088/1757-899X/149/1/012049

Figure 1. Feasible results for NSGA II Figure 2. Feasible results for Genetic Algorithm

X-axis: 1/power
Y-axis: Centre distance

Fig 1 and fig 2 shows the pareto-front obtained for NSGA II and multi-objective genetic algorithm. X-
axis represents reciprocal of power and Y-axis represents centre distance. Since both the algorithms
are designed for minimization problems and our objective is to maximize power, we took reciprocal of
power instead of power. By taking reciprocal of power in X-axis, we got perfect minimization curve as
desired. From the pareto-front obtained from both algorithms, we found the utopia point, which gives
the best-optimized result for our problem. Utopia point is the point, which is nearer to origin in the
minimization curve obtained for the given problem. From the utopia point in the NSGA II graph, we
got the best value of power as 43.469KW and centre distance as 304 mm. From the utopia point in GA
multi-objective, we got value of power as 46.49 KW and centre distance of 320 mm.

9. Conclusion

The single-speed spur gear design optimization problem considered in this paper provides an adequate
challenge to almost every aspect of optimization. The design variables for the multi –objective
optimization are module, no of teeth and face width of pinion. Objective functions include reducing
the center distance and maximizing the power transmitting capacity of the gear box. Both the objective
functions are conflicting in nature, which creates challenge to designer to choose the values of design
variables. Various novel constraints like flash temperature, minimum film thickness, scoring criterion
number, involute interference and contact ration are used to find the optimum solution. In the present
study we use algorithms like genetic (single and multi-objective), NSGA II and fmincon for obtaining
optimized results.
By comparing four algorithms for the optimization process, we found that multi-objective Genetic
algorithm provides the optimized results. This provides the maximum power output with an optimized
centre distance for the gears. GA multi-objective provides 49% increase in power when compared to
fmincon and 6% increase in power when compared to NSGA II. fmincon algorithm depends on
starting point, and thus the results depend on different starting points that are given by the user. Even
though the centre distance provided by single objective genetic algorithm is minimal, it shows a
remarkable loss in power output.

7
IConAMMA-2016 IOP Publishing
IOP Conf. Series: Materials Science and Engineering 149 (2016) 012049 doi:10.1088/1757-899X/149/1/012049

10. References
[1] G. N. Vanderplats, Xiarig Chen, and Ning-tian Zhang, 1988 NASA Contractor Report 4201,Gear
Optimization,4.0,5-20.
[2] N.Anifantis, University of Patras, Greece, A.D.Dimarogonas, Washington University,
St.Louis,U.S.A, 2.0,160-164.
[3] V. B. Bhandari, Design of Machine Elements, Tata McGraw-Hill, 2010.
[4] Kalyanmoy Deb, Sachin Jain, Multi-speed gearbox design using multi-objective evolutionary
algorithms, IIT Kanpur.
[5] Hindawi publishing corporation, multi-objective optimization of two stage helical gears using
NSGA-II
[6] Carroll, R. K., and Johnson, G. E., 1984, ‘‘Optimal Design of Compact Spur
[7] Gear Sets,’’ ASME J. Mech., Transm., Autom. Des., 106, pp. 95–101
[8] Dr. Shankar Krishnapillai, Introduction to Matlab Optimization Toolbox.
[9] Joseph Edward shigley & John Joseph Uicker, JR “Theory of Machines and Mechanisms”
[10] D E. Dudley, Gear handbook. McGraw-tlill, New York (1962).
[11] R. (iauvm, P. Girard and I[. Ycllc, ASME ,)td Int. Power Transmission and Gearing Conf, San-
Fancisco CA. [8 21 Aug. (19~10).
[12] R. C. Juvinall and K. M. Marshek, Fundamentals of Machine Component Design, John Wiley &
Sons, 2011.
[13] G. Maitra, Handbook of Gear Design, Tata McGraw-Hill, 2nd edition, 2003.
[14] Design Catalog of Hi-Tech Drive Pvt. Ltd. Plot No. 443/A, GIDC, V. U. Nagar, Gujarat, India.
[15] J. H. Holland, Adaptation in Natural and Artificial Systems, University of Michigan Press, Ann
Arbor, Mich, USA, 1975
[16] Y. K. Mogal and V. D. Wakchaure, “A multi-objective optimization approach for design of
worm and worm wheel based on genetic algorithm,” Bonfring International Journal of Man
Machine Interface, vol. 3, pp. 8–12, 2013.
[17] T. Yokota, T. Taguchi, and M. Gen, “A solution method for optimal weight design problem of
the gear using genetic algorithms,” Computers & Industrial Engineering, vol. 35, no. 3- 4, pp. 523–
526, 1998.
[18] A. D. Dimarogonas and G. N. Sandor, Wear 14, 153--170 (1969).
[19] http://www.mathworks.in/matlabcentral/fileexchange/31166- ngpm-a-nsga-ii-program-in-
matlab-v1-4.
[20] O. Buiga and C.-O. Popa, “Optimal mass design of a single-stage helical gear unit with genetic
algorithms,” Proceedings of the Romanian Academy Series A—Mathematics Physics Technical
Sciences Information Science, vol. 13, no. 3, pp. 243–250, 2012
[21] Goldberg, D. E., 1989, Genetic Algorithms for Search, Optimization, and Machine
Learning, Addison-Wesley, Reading, MA.
[22] Rao S. S., and Eslampour H. R., 1986, ‘‘Multistage Multiobjective Optimization
of Gearboxes,’’ ASME J. Mech., Transm., Autom. Des., 108, pp. 461-468