Volume 7, Issue 12, December – 2022 International Journal of Innovative Science and Research Technology

ISSN No:-2456-2165

Implementation of the Sine - Cosine Algorithm to the

Pressure Vessel Design Problem
Seth Anim Frimpong and Kwaku Forkuoh Darkwah
Department of Mathematics
Kwame Nkrumah University of Science and Technology, Kumasi, Ghana

Abstract:- Engineering design problems usually include survival of the fittest and the mechanism of natural selection
large-scale, nonlinear, or constrained optimization are the foundations of the GA. The three most significant
problems. Under a given circumstance, optimization stages in GA are mutation, crossover, and selection.
methods aim to find the optimum solutions that give the
extremum of a function. There are numerous methods Eberhart and Kennedy (1995) proposed the Particle
for solving optimization problems. Some of these Swarm Optimization (PSO). The PSO algorithm imitates the
problems are solved by heuristic or evolutionary group and individual foraging behaviors of a herd of
approaches. Genetic Algorithms (GA) and Particle animals and a flock of birds. The optimization process starts
Swarm Optimization (PSO) are two of the state-ofart with a set of solutions generated at random, just like the GA
heuristic optimization techniques. Additionally, one of algorithm. There is a second set called the velocity set in
the recently developed heuristic population-based addition to the set of solutions generated that is used to
optimization technique is the Sine-Cosine Algorithm. define and store the speed of particle motion.

The pressure vessel design problem has been solved Kannan and Kramer (1994) described the augmented
using various methods in the literature. Fruit Fly Lagrange multiplier method in conjunction with Powell’s
Algorithm gives the best minimum design cost, which is zeroth order method and, alternatively, the Fletcher and
5896.9489. In this work, the Sine-Cosine Algorithm was Reeves Conjugate Gradient method as a general approach
used to optimize the design cost of the pressure vessel for resolving mixed discrete, integer, zero-one, continuous
and obtained 5888.5213, which is better than the optimization problems. Augmented Lagrangian method is
literature’s best reported design cost. one of the techniques for solving constrained optimization
problems. They resemble penalty methods in the sense that
Keywords:- Genetic Algorithm, Particle Swarm they add a penalty term to the objective and swap out a
Optimization, Sine - Cosine Algorithm, Pressure Vessel and constrained optimization problem for a sequence of
American Society of Mechanical Engineers. unconstrained ones. The augmented Lagrangian approach,
however, adds still another term that is meant to resemble a
I. INTRODUCTION Lagrange multiplier. Although they are not identical, the
augmented Lagrangian and the Lagrange multiplier
Mechanical design encompasses the designing of a approach are linked.
variety of machine components, such as gears, bearings,
clutches, and fasteners (Kannan & Kramer, 1994). The key Sandgren (1990) proposed an algorithm that combined
design requirements for mechanical systems comprise wear, the Branch and Bound method with a quadratic
maintenance, liability, weight, size, functionality, safety, programming method, and an exterior penalty function
dependability, and manufacturability. Some basic design method. A branch and bound approach involve enumerating
concerns are relevant to all mechanical systems, despite the potential candidate solutions step by step while scouring the
fact that the number and importance of standards and whole search domain. A rooted decision tree with all of the
criteria differ from product design to product design. These potential solutions is first created. The entire search space is
factors include the capacity for loading, deformation, represented by the root node. Each child node is a
stability, and durability. To optimize products, the modeling component of the solution set and a partial solution. Based
and analysis of the dependency of these assessment criteria on the ideal solution, we build an upper and lower bound for
on design variables are required. The mechanical design a specific problem prior to building the rooted decision tree.
should satisfy the specifications provided by the American
Society of Mechanical Engineers (ASME) (Sandgren, Montes et al. (2007) developed the differential
1990). evolution (DE) approach. The algorithm explores the design
space by keeping track of a population of potential solutions
Numerous researches have been conducted on the (individuals), and by combining potential solutions in
pressure vessel design problem utilizing a variety of accordance with a predetermined method, it generates new
techniques, including the genetic algorithm, particle swarm solutions. The candidates with the best objective values are
optimization algorithm, augmented lagrange multiplier retained in the algorithm’s subsequent iteration in order to
method, branch and bound, differential evolution approach improve each candidate’s new objective value and include it
and fruit fly algorithm (Ke, et al., 2016). into the population.
Holland (1975) initially introduced the core ideas of Pan (2011) developed a new algorithm called Fruit Fly
the genetic algorithm (GA). GA employs techniques drawn Optimization Algorithm. The algorithm is built on two
from natural evolution to tackle optimization issues. The primary foraging techniques: using the osphresis organ to

IJISRT22DEC721 www.ijisrt.com 1069

Volume 7, Issue 12, December – 2022 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165
smell the food supply and sensitive vision to move toward et al.,2016), which is the literature’s best method for
the appropriate area to find food. In terms of sensory obtaining the minimum design cost.
perception, the fruit fly performs better than other species
particularly in vision and osphresis. Fruit flies have This paper is looking for a method which will do better
osphresis organs that can identify a variety of smells in the than the current best method. Therefore, the sine - cosine
atmosphere. When it is near a source of food, it uses its keen algorithm is employed to examine its performance against
eyesight to locate both the food source and where its the known methods in the literature.
companions are congregating before flying in that direction.
II. DESIGN ANALYSIS AND FORMULATION
The design of pressure vessels is a crucial component
of structural engineering optimization, and it frequently A. Design Analysis
looks for ways to reduce costs across board, including those The cylindrical pressure vessel with hemispherical heads
associated with forming, welding, and material costs on each of its ends is considered (Sandgren, 1990). The shell
(Cagnina, 2008). There has been a lot of solutions to the is constructed from two rolled plates that will be joined by
Pressure Vessel Design Problem. Branch and Bound two longitudinal welds to form a cylinder. Each head is
algorithm yielded $8129.8000 (Sandgren, 1990), Genetic forged, after which it is welded to the shell. Single-welded
Algorithm yielded $6288.7445 (Coello, 2000), Particle butt joints with a backing strip are used for all of the welds.
Swarm Optimization algorithm yielded $6059.1313 (Hu et The pressure vessel is set up so that the cylindrical shell’s
al., 2003) and Fruit Fly algorithm obtained $5896.9489 (Ke axis will be vertical.

Fig. 1: Schematic of Pressure Vessel

B. Data Approximate cost of welding (Cw) = $8.00/lb.

The data used in this work was taken from (Sandgren,
1990). Approximate material cost for the shell plate (Cs) =
$0.35/lb.
The material used in the vessel is carbon steel ASME
SA 203 grade B. Approximate material cost for the hemispherical head
plate (Ch) = $1.00/lb.
Density of the carbon steel (D) = 0.2830 lb/in3.

C. Data Processing
The total cost comprises of the welding cost, material cost and forming cost.
a) Welding Cost:
1. Longitudinal welding cost:
Longitudinal welding cost = Vl× D × Cw (1)

where,

(2)
Eqn(2) into (1):

(3)
2. Circumferential welding cost:
(4)
Circumferential welding cost = Vs × D × Cw

where,

IJISRT22DEC721 www.ijisrt.com 1070

Volume 7, Issue 12, December – 2022 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165

(5)
Eqn(5) into (4):

(6)
The total welding cost is the sum of the longitudinal and circumferential welding cost.

(7)

b) Material and Forming Costs:

Material and forming costs = 2πDCsRTsL+ 2πDChR2Th (Sandgren,1990) (8) = 0.6224RTsL + 1.7781R2Th (9)

D. Formulation of the Problem control the exploration and exploitation during the search
The objective is to minimize the total cost of material, process.
forming, and welding. Ts(thickness of the shell, y1),
Th(thickness of the head, y2), R (inner radius, y3) and L
(length of cylindrical section of the vessel, y4) are the four
design variables. Tsand Thare integer multiples of 0.0625
inch, the available thickness of rolled steel plates, and R and where t is the current iteration and Tmaxis the maximum
L are continuous. The total cost which we intend to number of iterations.
minimize is then given as follows:
The parameter r2 lies in the interval (0,2π) and
specifies how far the movement should be toward or away
(10) from the destination point. The parameter r3 has the interval
[0,2] and gives random weight to the Pit which focuses on
The minimum wall thicknesses must be constrained by the exploration (r3 >1) and exploitation (r3 <1). The
the constraint set. The minimum value of the tank and the parameter r4 lies in the interval (0,1). and toggles between
length of the cylindrical shell are both constrained by the sine and cosine components. Finally, Pit is the best
Tsand Thfrom the ASME codes (Sandgren, 1990). These destination point obtained so far and || indicates the absolute
constraints are listed as: value.
Constraint of circumferential stress: g1(y) = B. The algorithm for sine - cosine technique:
−y1 + 0.0193y3 ≤ 0 (11) 1. Initialization of the set of search agents
{X1,X2,···,Xn} using
Constraint of longitudinal stress: g2(y) =
xij= lb+ rand(0,1) × (ub− lb)
−y2 + 0.00954y3 ≤ 0 (12)
where xijis the components of each search agent,
lband ubare the lower and upper boundary
Constraint of volume: respectively. (13)

Constraint of length: 2. Fitness of each search agent is evaluated, f(Xi )

g4(y) = y4 − 240 ≤ 0 (14)
3. Memorize the best destination point,
1 × 0.0625 ≤ y1,y2 ≤ 99 × 0.0625, 10 ≤ y3,y4 ≤ 200
4. The Algorithm parameter r1 is initialized and updated
III. METHODOLOGY at every iteration using

A. Sine - Cosine Algorithm (SCA)

The Sine - Cosine Algorithm was developed in 2016
(Mirjalili, 2016). The sine cosine algorithm begins by 5. Update each search agent using the SCA search
generating random solutions known as search agents. The equation.
sine - cosine algorithm is tuned using four variables Xi =Xit + rit× sin(r2) × |r3Pit− Xit | r4 <0.5
t+1

(r1,r2,r3,r4) and is given by:

Xit+1 =Xit + rit× cos(r2) × |r3Pit− Xit | r4 <0.5
Xit+1=Xit+ rit× sin(r2) × |r3Pi t− Xit| r4 <0.5
6. Evaluate the fitness of each updated search agent
Xi t+1
=Xi + ri × cos(r2) × |r3Pi − Xi | r4 <0.5
t t t t f(Xit).

where Xitis the position of the current search agent in 7. Repeat steps 3, 4, 5 and 6 until the termination
i−thdimension at t−thiteration. The r1,r2,r3 and r4 are random criteria is fulfilled.
values. The parameter r1 uses the expression below to

IJISRT22DEC721 www.ijisrt.com 1071

Volume 7, Issue 12, December – 2022 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165
The sine-cosine algorithm begins the optimization IV. COMPUTATIONAL PROCEDURE
process with a set of random solutions. The algorithm then
stores the best solutions thus far, labels it as the destination Sine - Cosine Algorithm was used in this work. All the
point, and modifies all other solutions in respect to it. The codes were run 10 times with 5000 iterations in each run in
ranges of the sine and cosine functions are updated as the Octave environment on a computer with the following
iteration counter increases to emphasize exploitation of the specifications:
search space. Operating System: Windows 10 Pro Education 64 - bit.
Processor: Intel(R) Celeron(R)
By default, when the iteration counter exceeds the N4020 CPU @ 1.10GHz (2CPUs).
preset number of iterations, the SCA algorithm terminates Memory: 4096MB RAM.
the optimization process. Any additional termination criteria
may be taken into account, including the maximum number Input for the Sine - Cosine Algorithm:
of function evaluations or the precision of the discovered Search Agents (X) = 150
global optimum. Number of iterations (Tmax) = 5000
Lower boundary (y1,l,y2,l,y3,l,y4,l) = (0.0625,0.0625,10,10),
Upper boundary (y1,u,y2,u,y3,u,y4,u) = (99 × 0.0625,99 ×
0.0625,200,200), (Sandgren, 1990) Dimension = 4

V. RESULTS

The following are the values of the best design variables obtained after running the program for the Sine - Cosine Algorithm:
y1 = 0.8259
y2 = 0.3814
y3 = 42.7444
y4 = 168.7212
f(X) = 5888.5213

The Figure 1 below describes how the fitness value converges towards the 5000 iterations.

Fig. 1: The Convergence Vs Iteration Graph

From the graph or simulation, it was observed that the optimal cost of 5888.5213 was attained at the 4500th iteration after
which it was maintained for the rest of the iterations.

IJISRT22DEC721 www.ijisrt.com 1072

Volume 7, Issue 12, December – 2022 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165
VI. DISCUSSION

From the results section, we have:

Thickness of the shell (y1) = 0.8259 inch.
Thickness of the head (y2) = 0.3814 inch.
Inner radius (y3) = 42.7444 inch.
Length of the shell (y4) = 168.7212 inch.
Minimum cost f(X) = $5888.521

The table 1 below shows the optimal cost calculated by the various methods found in theliterature as compared to the value o
btained in this paper

Table 1: Comparison of the best solution for the pressure vessel design problem

From table 1, the difference between the results Computational simulations indicate that the proposed
obtained from the Artificial Bee Colony technique and the SCA approach achieves the best result in terms of objective
Gausian Quantum Behaved Particle Swarm technique is function (total design cost) minimization, being 5888.5213
0.5063. Also, the difference between the results obtained which is 8.4276 (0.142%) better than fruit fly algorithm
from the Mixed Integer technique and the New Particle method, with 5896.9489, which is the best literature’s best
Swarm technique is 0.0313. These differences in the results reported objective function value.
were considered significant.
It can be concluded that by using the sine-cosine
The current work optimizes the parameters such as the algorithm, the pressure vessel’s optimal design parameters
thickness of the shell, length, and radius of the pressure are found, and the objective of cost minimization by
vessel using the Sine - Cosine Algorithm. The results are reducing pressure vessel weight is accomplished. The
compared to various works which used other optimization application of the sine-cosine algorithm to a pressure vessel
methods and are shown in Table (6.1). It has been found that problem with four design constraints and four variables has
the optimal design cost, $5888.5213, obtained by the Sine - been demonstrated in this paper.
Cosine Algorithm is better as compared to the literature’s
best cost of $5896.9489. REFERENCES

The choice of the 5000 iterations, 150 search agents [1.] Akay, B. and Karaboga, D. (2012). Artificial bee
and the currency in dollars is that the comparing methods colony algorithm for large-scale problems and
used the same information and therefore, the study did not engineering design optimization, Journal of
want to introduce any variation to conflict the results. Also, Intelligent Manufacturing, vol 23, pp.1001-1014.
the results hold for Carbon Steel ASME SA 203 grade B. [2.] Cagnina, L., C., et al. (2008). Solving engineering
optimization problems with the simple constrained
VII. CONCLUSION particles swarm optimizer. Informatica, vol.32,
pp.319-326.
In this paper, the optimal design cost of a pressure [3.] Coelho, L., S. (2010). Gaussian quantum-behaved
vessel was carried out through the sinecosine algorithm particle swarm optimization approaches for
(SCA) by optimizing parameters which include thickness of constrained engineering design problems. Expert
the shell and head, length of the shell and inner radius of the System with Applications, vol37, pp.1676-1683.
pressure vessel. The results obtained are compared with the [4.] Coello, C., A., C. (2000). Use of a self -adaptive
results of other optimization methods in the literature penalty approach for engineering optimization
applied to the pressure vessel design problem. problems. Computers in Industry, vol.41, pp.13-127.

IJISRT22DEC721 www.ijisrt.com 1073

 1074
[5.] Coello, C., A., C. and Montes, E., M. (2002). [21.] Montes, E., M. and Coello, C., A., C. (2008). An
Constraint- handling in genetic algorithms through empirical study about the usefulness of evolution
the use of dominance-based tournament selection. strategies to solve constrained optimization problems.
Advanced Engineering Informatics, vol.16, pp.193- International Journal of General Systems, vol.37, pp.
203. 443-473.
[6.] Deb, K. and Gene, A., S. (1997). A robust optimal [22.] Pan, W., T. (2011). A new Fruit Fly Optimization
design technique for mechanical component design; Algorithm: Taking the financial distress model as an
Evolutionary Algorithms in Engineering example. KnowledgeBased System, vol.26, pp.69-74.
Applications, Springer, Berlin , pp.497-514. [23.] Sandgren, E. (1990). Nonlinear integer and discrete
[7.] Eberhart, R., C. and Kennedy, J. (1995). A new programming in mechanical design. Proceedings of
optimizer using particle swarm theory. Proceedings the ASME Design Technology Conference, F.L.
of the sixth international symposium on micro Kissimine, vol.95-105.
machine and human science, pp. 39-43. [24.] Zhang, C. and H. P. Wang, H., P. (1993). Mixed-
[8.] Gandomi, A., H. et al. (2003). Cuckoo search discrete nonlinearoptimization with simulated
algorithm: a meta heuristic approach to solve annealing. Engineering Optimization, vol.21, pp.277-
structural optimization problems. Engineering with 291.
Computers, vol.29, pp.17-35.
[9.] He, S. et al. (2004). An improved particle swarm
optimizer for mechanical design optimization
problems. Engineering Optimization, vol.36, pp.585-
605, 2004.
[10.] He, Q. and Wang, L. (2007). An effective co -
evolutionary particle swarm optimization for
constrained engineering design problems;
Engineering Applications of Artificial Intelligence,
vol.20, pp.89-99.
[11.] Holland, J., H. (1975). Cognitive systems based on
adaptive algorithms. ACM SIGART
[12.] Bulletin.
[13.] Hu, X., H., et al. (2003). Engineering optimization
with particle swarm; Proceedings of the 2003 IEEE
Swarm Intelligence Symposium, pp.53-57.
[14.] Kannan, B., K. and Kramer, S., N. (1994). An
augmented lagrange multiplier based method for
mixed integer discrete continuous optimization and
its applications to mechanical design; Transactions of
the ASME, Journal of Mechanical Design, vol.116,
pp.405-411.
[15.] Kaveh, A. and Talatahari, S. (2009). Engineering
optimization with hybrid particle swarm and ant
colony optimization. Asian journal of civil
engineering (building and housing), vol.10, pp.611-
628.
[16.] Kaveh, A. and Talatahari, S. (2010). An improved ant
colony optimization for constrained engineering
design problems. Engineering Computations, vol.27,
pp.155-182.
[17.] Ke, X., et al. (2016). Solving design of pressure
vessel engineering problem using a fruit fly
optimization algorithm. International Journal of
Simulation: Systems, Science and Technology,
17(43), 5.1-5.7
[18.] Lee, K., S. and Geem, Z., W. (2005). A new meta-
heuristic algorithm for continuous engineering
optimization: harmony search theory and practice;
Computer Methods in Applied Mechanics and
Engineering, vol.194, pp.3902-3933.
[19.] Mirjalili, S. (2016). SCA: A Sine Cosine Algorithm
for Solving Optimization Problems. Knowledge-
Based Syst. 96, pp. 120-133
[20.] Montes, E., M., et al. (2007). Multiple trial vectors in
differential evolution for engineering design.
Engineering Optimization, 39, 567-589.

IJISRT22DEC721 www.ijisrt.com 1074