Anda di halaman 1dari 11

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/271020225

DESIGN OPTIMISATION OF TWO STAGE GEARBOX WITH HELICAL GEARS

Conference Paper · May 2006


DOI: 10.13140/RG.2.1.5193.7363

CITATIONS READS
0 3,485

2 authors, including:

Metin Zeyveli
Karabuk University
13 PUBLICATIONS   42 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Joining of different material by Clinching rivet View project

All content following this page was uploaded by Metin Zeyveli on 19 January 2015.

The user has requested enhancement of the downloaded file.


Proceedings of 5th International Symposium on
Intelligent Manufacturing Systems, May 29-31, 2006: 724-733
Sakarya University, Department of Industrial Engineering

DESIGN OPTIMISATION OF TWO STAGE GEARBOX WITH


HELICAL GEARS

Metin ZEYVELİ a, Cevdet GÖLOĞLU b,*


a
Division of Mechanics and Machine Design, Faculty of Technical Education,

University of Zonguldak Karaelmas, 78050 Karabuk, Turkey.

Tel: +90 370 433 82 10, Fax:+90 370 433 82 04

E-mail: m_zeyveli@hotmail.com

b
Division of Design and Machine Building, Faculty of Technical Education,

University of Zonguldak Karaelmas, 78050 Karabuk, Turkey.

Tel: +90 370 433 82 10, Fax:+90 370 433 82 04

E-mail: cgologlu@hotmail.com

Abstract

Engineering design is an iterative process that requires to be dealt with all feasible design
solutions in order to arrive at desired objective. Proper design of gearbox has a significant
place in power transmission applications. Traditional methods used in its design do not
have ability in automating the process. Thus, an attempt to automate preliminary design of
gearbox has been accomplished in the paper. A software to automate preliminary design
of gearbox with spur, helical and bevel gears was developed. In the software Genetic
Algorithm (GA) was applied to the problem with the objective function of minimising of
volume of gear trains. The objective function was constrained by bending strength,
contact stress, face width and number of pinion and gear teeth. The preliminary design
parameters module, number of teeth, and width of teeth for pinion and gear pairs of the
stages were optimised and gear ratios were determined in respect to the objective
function and design constraints. Design optimisation of a two stage gearbox by using GA
was accomplished by readily supplying the design parameters requested.

Keywords: Design optimisation, Gearbox, Helical gears, Genetic Algorithm


Submission area; Intelligent Manufacturing Systems
1. Introduction
Engineering design is an iterative process that is started with a poorly defined problem,
refined and then developed a model, finally arrived at a solution. Due to nature of
engineering design there could be more than one solution, therefore a search should
be conducted in order to find the best solution. As a mechanical design problem,
design of gearbox is very complex because of multiple and conflicting objectives.
Moreover, the design variables encountered could be in forms of continuous, zero-one,
integer and discrete variable.

A gearbox utilises a group of gears to achieve a gear ratio between the driver and
driven shafts. The material volume of gear trains is the main determination factor in
sizing of this power transmitting units. Trail-and-error method is mainly used in
traditional design of gearbox. Researchers have developed several applications using
different design and calculation methods. A gearbox producing the required output
speed was designed by GA [1]. The objective function of the study was to determine
the number of shafts and number of teeth for each gear by GA. The objective function
was constrained by number of teeth of gear, maximum transmission ratio and
maximum number of shafts. An algorithm automating preliminary design of multi stage
gear drives was proposed [2]. The algorithm having four steps was run iteratively until a
desirable solution is found. The steps in the algorithm were mainly conducted
manually, by random search and generate and test methods. In last step, simulated
annealing algorithm has been used for minimising geometrical volume of a gearbox by
means of integrating dimensional and configuration design process. An optimal weight
design problem of a gear pair system was studied using GA [3]. The system was able
to find the number of design variables considering specified constraints. A generalised
optimal design formulation to gear trains was presented [4]. In the research, tradeoff
between surface fatigue life and minimum volume using a basic multiobjective
optimisation procedure was studied. The results were presented as a Pareto optimal
set representing a collection of optimal designs. A computer aided design of gears
approach was proposed [5] to optimise one stage gear pair. GA was employed for
minimising gear volume by reducing the distance between the centres of gear pairs,
and other parameters such as transmitting power, reduction ratio. In another study an
expert system composing of a GA module has been developed [6]. The GA module
was used for optimising volumes of pitch cylinders of gears for a single reduction
gearing system.

In this research, a genetic algorithm was used for preliminary design of gearbox. The
objective function was chosen as minimum material volume of gearbox. The
constraints of bending strength, contact stress, number of pinion and face width
parameters of helical gear pairs were employed as restrictions on design criteria.

2. GA formulation of the problem


In this study a part of the developed software, gearbox optimisation with helical gears
will be addressed. In optimising volume of gear trains module, number of teeth and
face width are selected as effective design parameters. The module variable mn is
normal module in helical gears and has been defined at interval of 1 to 5.5 with total of
16 variables. They are mapped to integer values due to being discrete values; binary
representation of them is given in Table 1 as four bits. In use of calculations they have
again been converted into real module values. In design of gear pairs, materials of
pinion and gear have accepted same and design process is carried out based on
pinion. Thus number of teeth of gear is defined subject to number of teeth of pinion and
gear ratio. The interval for number of teeth of pinion is accepted based on the premise

725
of 17 ≤ z1,3 ≤ 24 where z1 and z3 are the numbers of teeth of pinions of first and
second stages respectively. Binary representation of the number of teeth of pinion is
also given in Table 1 as three bits. As it starts from the teeth number 17 so the method
at the representation of modules was used here. The interval for number of teeth of
gear is accepted as 35 ≤ z 2, 4 ≤145 where z2 and z4 are the numbers of teeth of gears
of first and second stages respectively. For binary representation of the gear teeth
number of z2 the method of Lin and Hajela [7] was used and it was presented with
seven bits. The face width b is determined at the beginning of design process based on
face width factor 19 ≤ ϕ m ≤ 40 and it is defined as b = ϕ m mn . The face width values
are presented as five bits in binary code at represented.

Table 1 Binary representation of module and number of pinion teeth values


Decimal Num. of pinion Decimal
Module Binary code Binary code
equiv. code teeth equiv. code
1 0 0000 17 0 000
1.125 1 0001 18 1 001
1.25 2 0010 19 2 010
1.375 3 0011 20 3 011
1.5 4 0100 21 4 100
1.75 5 0101 22 5 101
2 6 0110 23 6 110
2.25 7 0111 24 7 111
2.5 8 1000
2.75 9 1001
3 10 1010
3.5 11 1011
4 12 1100
4.5 13 1101
5 14 1110
5.5 15 1111

Representation of design variables in GA is very important and wrong representation of


variables affects to performance and working period in negative manner. Combining of
design variables at the length of 31 bits for a two stage helical gear trains forms a
solution vector. The composition elements are depicted in Table 2.

Table 2 Design variables and forming of a solution vector

Design variables m1 z1 b1 z2 m2 z3 b2
4 bits 3 5 bits 7 bits 4 bits 3 5 bits
String length
bits bits
Sample design variables 1010 101 10010 1011100 0111 011 11001
Module m1 (1st stage)
Number of teeth of pinion z1 (1st stage)
Tooth (face) width b1 (1st stage)
Number of teeth of gear z2 (1st stage)
Module m2 (2nd stage)
Number of teeth of pinion z3 (2nd stage)
Tooth (face) width b2 (2nd stage)
A sample of design solution
1010101100101011100011101111001
(chromosome)

726
3. The objective function
In the study, minimisation of material volume is the objective function Fobj and it is
defined as:

Minimum material volume, mn1 , z1 , b1


Fobj =  (1)
Minimum material volume, mn 2 , z 3 , b2

where mn1 is the module of pinion and gear at first stage and mn2 is the module of
pinion of gear at second stage. z1 and z3 are the number of teeth of pinion at first stage
and number of teeth of pinion at second stage respectively. The width of gear pairs at
stages are presented by b1 and b2. The material volume of a helical pinion F1 is defined
as:

π 2 
F1 =  (mn1 z1 ) b1  (2)
4 

In a two stage gearbox configuration, total volume of two pairs of helical gears FT can
be written as a function of module, number of teeth and width of gear (Eq. (3)).

π π 
4
( 2

4
) ( 2
)
FT = F1 + F2 + F3 + F4 =  (mn1 z1 ) + (mn1 z2 ) b1 + (mn 2 z3 ) + (mn 2 z4 ) b2  (3)
2 2

In minimum volume objective function a number of constraints are introduced. The


constraints are used to provide suitable design choices or as sub functions to restrict
the objective function such a way that suitable contents are incorporated. These are as
follows:

Bending strength constraint,


Surface fatique constraint,

g ( j ) = Face width constraint, (4)
Module constraint,

 Number of pinion teeth constraint,

Then, the global objective function with constraints can be presented by:

Nconst
Fglabalobj = FT + ∑ rj (max [0, g j ( x)] ) (5)
j =1

By using Eq. (5), constraint optimisation problem has been become unconstrained
optimisation problem by introducing static penalty functions to the objective function.
The penalty coefficients, rj, for the j-th constraint have to be wisely selected. These
leads to the solutions are selected by trusting on proper values of penalty coefficients.
The weighting constraints by penalties are difficult to adjust and assigning right values
are essential to solutions compromised. By subtracting the function from a large
enough positive number F [8], the function becomes as

727
 Nconst 
Fobj = F −  Fi + ∑ r j (max[0, g j x)])  (6)
 
 j =1 

where Fi presents 1 to 4 pinion and gears.

3.1 Constraining the objective function


The formula of bending strength, contact stress, face width, and teeth number have
been used for forming constrains for gearbox with helical gears. The constraints are
thus classified into the bending stress constraints, the contact stress constraints, the
face width, and teeth number constraints. The constraints g1 and g2 present the
bending stress of the first pair gears and the bending stress of the second gear pairs
respectively.

g1 ( x) = K d K fe K c Ft1 − ε b1 mn1 (0.55 σ K ) ≤ 0 (7)


g 2 ( x) = K d K fe K c Ft 2 − ε b2 mn 2 (0.55 σ K ) ≤ 0 (8)

The constraints g3 and g4 present similarly first and second pairs’ contact stress.

 z + z1 
g 3 ( x) = K d (K m K α K ε K β )2 Ft1  2  − b1 mn1 z 2 (Palw )2 ≤ 0 (9)
 z1 
 z + z3 
g 4 ( x) = K d (K m K α K ε K β )2 Ft 2  4  − b 2 mn 2 z 4 (Palw )2 ≤ 0 (10)
 z3 

The constraint g5 represents the pinion width constraint and g6 represents the gear
width constraints for the first stage.

g5 ( x) = ((20 mn1 ) − b1 ) ≤ 0 (11)


g 6 ( x) = (b1 − (40 mn1 )) ≤ 0 (12)

Similarly, g7 and g8 represent the pinion width and gear width constraints for the
second stage respectively. Here, a premise of the width factor for helical gears is
chosen between 20 to 40 has been adopted in Eqs. (13)-(14).

g 7 ( x) = ((20 mn 2 ) − b2 ) ≤ 0 (13)
g8 ( x) = (b2 − (40 mn 2 )) ≤ 0 (14)

The constraints of g9 and g10 represent the number of teeth of pinions for first and
second stages respectively.

g 9 ( x) = (17 − z1,3 ) ≤ 0 (15)


g10 ( x) = ( z1,3 − 24) ≤ 0 (16)

The number of teeth of gears for first and second stages is constrained by g11 and g12
respectively.

728
g11 ( x) = (35 − z 2, 4 ) ≤ 0 (17)
g12 ( x) = ( z 2, 4 − 145) ≤ 0 (18)

4. A sample design optimisation


The coefficients used in strength formulas and input values used for a sample design
optimimisation are given in Table 3.

Table 3 Coefficients and input values for sample design practise

Transferred power (kW) 5.5 Working factor, K o 1.25


Input speed (rpm) 1440 Overlap ration, ε 1.6
Output speed (rpm) 144 Stress concentration factor, K c 1.5
Total gear ratios (i) 10 Material factor, K m (N/mm2) 271.11
Material (Cementite steel) 16MnCr5 Flank transverse coefficient, K α 1.76
Manufacturing process Fine Tooth overlap factor, K ε 0.79
work
Usage Electricity Helical angle, β 18º
motor
Brinell hardness (N/mm2) 1460 Tooth slope factor K β 1
Ultimate tensile strength, 1100
σ K (N/mm2)
1, 1.125, 1.25, 1.375, 1.5, 1.75, 2, 2.25, 2.5, 2.75, 3,
Available modules (mm)
3.5, 4, 4.5, 5, 5.5

The GA parameters in the study were chosen as initial population = 50; number of
generations = 100, 200, and 500; crossover probabilities = 0.5, 0.7, and 0.9, mutation
probabilities = 0.1, 0.01, and 0.001. As crossover operator two-point crossover was
employed (Fig. 1) and as a selection mechanism, tournament selection was chosen
with 6 of tournament size. The GA was run with the mentioned number of generations
and it was seen that the objective function curve do not change at the generations after
100. Thus, the generation number 200 was admitted as a compromised one.

Two chromosomes before crossover


1 0 1 1 0 1 1 1 1 0 1 1 1 0 0 1 0 0 0 1 0 1 1 0 1 1 0 0 0 1 1
↑ ↑
1st cross. point 2nd cross. point
↓ ↓
0 0 0 0 0 0 0 1 1 0 0 0 1 0 0 1 0 1 1 1 0 1 1 1 1 0 1 1 0 0 0

Two chromosomes after crossover


1 0 1 1 0 1 1 1 1 0 0 0 1 0 0 1 0 1 1 1 0 1 1 1 1 1 0 0 0 1 1
↑ ↑
1st cross. point 2nd cross. point
↓ ↓
0 0 0 0 0 0 0 1 1 0 1 1 1 0 0 1 0 0 0 1 0 1 1 0 1 0 1 1 0 0 0

Fig. 1. Two point crossover

729
The crossover probabilities of 0.9, 0.7, and 0.5 were tried and the last crossover
probability of 0.5 was seen as the most appropriate (see Fig. 2).

19000
Cross. prob. of 0.9
17000 Cross. prob. of 0.7
Cross. prob. of 0.5
15000

13000

11000
Fitness

9000

7000

5000

3000
0 50 100 150 200
Generations

Fig. 2. Fitness values at mutation probability of 0.001

Based on the crossover probability of 0.5, the mutation probabilities of 0.1, 0.01, and
0.001 were tested. Fig. 3 shows the average fitness values obtained by the different
mutation probabilities. The most successful was seen as the mutation probability of
0.001. The minimum fitness values at crossover probability of 0.5 were given in Fig. 4.
Minimum and average fitness values at minimum crossover probability of 0.5 and
minimum mutation probability of 0.001 were depicted in Fig. 5.

73000 Mutation prob. of 0.1


Mutation prob. of 0.01
63000
Mutation prob. of 0.001
53000

43000
Fitness

33000

23000

13000

3000
0 50 100 150 200
Generations

Fig. 3. Average fitness values at crossover probability of 0.5

730
8000

7500 Mutation prob. of 0.1


Mutation prob. of 0.01
7000 Mutation prob. of 0.001
6500
Fitness
6000

5500

5000

4500

4000
0 50 100 150 200
Generations

Fig. 4. Minimum fitness values at crossover probability of 0.5

10000

Average fitness
9000
Minimum fitness
8000

Crossover prob. of 0.5


7000
Fitness

Mutation prob. of 0.001

6000

5000

4000
0 20 40 60 80 100 120 140 160 180 200
Generations

Fig. 5. Minimum and average fitness values

The result of GA gives the preliminary design optimisation variable values of two stage
gearbox (see Table 4).

Table 4. Results of optimised preliminary design parameters


Preliminary design parameter values
Module m1 (1st stage), mm 2.25
st
Number of teeth of pinion z1 (1 stage) 22
Tooth (face) width b1 (1st stage), mm 46.77
Number of teeth of gear z2 (1st stage) 79
Module m2 (2nd stage), mm 3.5
Number of teeth of pinion z3 (2nd stage) 23
Tooth (face) width b2 (2nd stage), mm 67.10
Minimum material volume, cm3 4513.29

731
Fig. 6 and Fig 7 show initial design parameter determination menu and the GA results
menu of the developed software respectively.

Fig. 6. Initial design parameter determination menu

Fig. 7. GA results menu

5. Conclusion

Most engineering design problems are optimised with traditional methods as


gearboxes. A GA approach with the objective of minimum material volume has been
employed to design optimisation of gearbox with the introduced constraints. In the
study a part of the developed software, gearbox optimisation with helical gears has
been addressed. The preliminary design parameters module, number of teeth, and
width of teeth for pinion and gear pairs of the stages are readily issued in optimised
values to the designers.

732
Nomenclature

σb , Bending stress (N/mm2) HB , Brinell hardness


stress, σ alw = σ D K c
σ alw , Allowable Pmax , Maximum surface pressure
2
(N/mm ) (N/mm2)
Dynamic stress, σ D ≅ 0.55σ K
σD, Km , Material factor
(N/mm2)
σK , Ultimate tensile strength (N/mm2) Kε , Tooth overlap factor
Kc , Stress concentration factor Kα , Flank transverse coefficient
Ft , Tangential force (N) Kβ , Tooth slope factor
ε, Overlap ratio d, Pitch diameter
Kd , Dynamic velocity factor mt , Transverse module
K fe , Form factor z, Number of teeth
Allowable surface pressure,
Palw, β, Helix angle
palw = 0.25 H B (N/mm2)

References
[1] D.T. Pham, Y. Yang, Optimization of multi-model discrete functions using genetic
algorithms. Proc. Instn. Mech. Engrs. 207 (1993) 53-59
[2] T.H. Chong, I. Bae, G.J. Park, A new and generalized methodology to design multi
stage gear drives by integrating the dimensional and configuration design process.
Mechanism and Machine Theory 37 (2002) 295-310
[3] T. Yokota, T. Taguchi, M. Gen, A solution method for optimal weight design
problem of the gear using genetic algorithms. Computers ind. Engng. 35 (3-4)
(1998) 523-526
[4] D.F. Thompson, S. Gupta, A. Shukla, Tradeoff analysis in minimum volume design
of multi-stage spur gear reduction units. Mechanism and Machine Theory 35
(2000) 609-627
[5] J.L. Marcelin, Genetic optimisation of gears. Int J Adv Manuf Technol. 17 (2001)
910-915
[6] B. Abersek, V. Popov, Intelligent tutoring system for training in design and
manufacturing. Advances in Engineering Software 35 (2004) 461–471
[7] C.Y. Lin, P. Hajela, Genetic algorithms in optimization problems with discrete and
integer design variables. Engineering Optimization 19 (1992) 309–327
[8] D.E. Goldberg, Genetic Algorithms in Search, Optimization and Machine Learning,
Addison Wesley, New York, 1989

733

View publication stats

Anda mungkin juga menyukai