Effective Hybrid Algorithms for No-Wait Flowshop Scheduling Problem

International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056
Volume: 05 Issue: 01 | Jan-2018 www.irjet.net p-ISSN: 2395-0072
© 2017, IRJET | Impact Factor value: 6.171 | ISO 9001:2008 Certified Journal | Page 1473
Effective hybrid algorithms for no-wait flowshop scheduling problem
Habib Elyasi1
1Department of Management, Islamic Azad University, Marand Branch, Marand, Iran
---------------------------------------------------------------------***---------------------------------------------------------------------
Abstract - No-wait flowshop scheduling problems with
minimizing the total makespan is one of the significant
problems in production scheduling. In no-wait flowshop
problems, when the process of any job is initiated on any of
the machines, that job must pass through all the processes
and the machines without waiting until it leaves the last
machine. In other words, the process of a job on the first
machine would be put off until it is made sure it would not
face any delay throughout the process. A hybrid genetic
algorithm is proposed for solving such a problem that not
only will avoid local optimization, but it is also proven that
this algorithm is outstanding while facing NP-hard
questions. The results of the numerical example prove that
the hybrid genetic algorithm outperforms other algorithms
in the literature as well as the genetic algorithm.
Key Words: Flowshop, No-wait, Genetic Algorithm,
Hybrid Genetic Algorithm, Scheduling.
1. INTRODUCTION
Distributed multisite production has been placed a
premium on because of the competitive nature of
globalization. Because of their nature, optimization
problems have always been considered in the literature
(see for example: [1], [2], [3], [4]). Numerous variants of
the ﬂowshop scheduling problem (FSP) have been studied
[5] due to the nature of distinct industrial processes. One
of the varıants of FSPs is called no-wait FSPs (NWFSPs)
that are applied in numerous industries such as plastics,
metals and electronics. A no-wait flowshop problem
occurs when the operations of a job have to be processed
continuously from start to end without interruptions
either on or between machines. Hence, if needed, the start
of a job on a given machine is delayed in order that the
operation’s completion coincides with the start of the next
operation on the subsequent machine. Allahverdi [6]
reviewed more than 300 papers in scheduling problem
with no-wait constraints and classifies the problems based
on shop environment, performance measures, and other
factors. Two commonly used performance measures in the
scheduling literature are makespan and mean completion
time. Minimizing makespan is important in situations
where a simultaneously received batch of jobs is required
to be completed as soon as possible.
Callahan [7] was one of the first people who studied no-
wait flow shop problems. His study was focused on steel
industry and he studied no-wait problems with respect to
dependent parameters such as temperature. No-wait flow
shop problem with minimizing the total makespan is
proved to be an NP-hard problem by Röck [8]. Therefore,
instead of looking for exact methods, many heuristics have
been developed in order to generate solutions with high
calculation qualities (see for example: [9], [10], [11], [12],
[13]). Aldowaisan and Allahverdi [14] also suggested some
heuristics and proved that their proposed algorithms
outperform the previous ones. Grabowski and Pempera
[15] proposed several heuristics and compared the results
with the local search algorithm proposed by Schuster and
Framinan [16] and the algorithms proposed by Rajendran
[13]. A new heuristic which is based on analogy between
the machines was proposed by Framinan and Nagano [17].
Lin and Ying [18] proposed a mixed integer programming
mathematical model and an iterated cocktail greedy
algorithm as the solution method and they were proved to
be more effective than the known algorithms in the
literature. Nagano et al. [19] introduced a constructive
heuristic named QUARTS and they achieved
approximation solutions in a short CPU time.
Metaheuristic algorithms have also been proposed for
solving no-wait flowshop problem with the minimization
of the total makespan as the objective [see for example;
[20], [21]). Tseng and Lin [22] proposed a hybrid genetic
algorithm utilizing the base genetic algorithm and a
heuristic proposed by them. Zhu et al. [23] studied the
problem by implementation of a local search algorithm.
Aldowaisan and Allahverdi [14] studied six different meta-
heuristics consisting of three (simulated annealing (SA)
algorithms and three genetic algorithms (GA) to solve the
problem. The results revealed that the two of the
algorithms outperformed the algorithms proposed by
Gangadharan and Rajendran [12] and Rajendran [13] but
required more processing time. Schuster and Framinan
[16] proposed a hybrid algorithm that used both SA and
GA for solving the problem. Their studies proved that their
hybrid algorithm outperform the algorithm provided by
Rajendran [13]. Framinan and Schuster [24] improved the
algorithm proposed in Schuster and Framinan [16] using
an algorithm called complete local search with memory.
Grabowski and Pempera [15] developed some Tabu
search based algorithms that outperform the previous
findings including the one proposed by Rajendran [13].
Schuster [25] introduced a fast Tabu search algorithm for
solving the no-wait jobshop scheduling problem and found
their algorithm compared well to the algorithm proposed
by Schuster and Framinan [16]. Ying et al. [26] proposed
self-adaptive and ruin-and-recreate algorithm which
improved best known solutions in more than half of the
benchmarks. Lin and Ying [27] proposed a meta-heuristic

algorithm and that could handle very hard and large
problems. Zhao et al. [28] proposed a discrete water wave
optimization (DWWO) algorithm and a revised greedy
algorithm for the problem and proved that their algorithm
outperforms the algorithms in the literature.
The reminder of the paper is organized as follows.
Section 2 formally defines the problem. Section 3
describes the solution methods and proposed algorithms
in details. In section 4 computational results are proposed.
Finally, in section 5, final remarks as well as the
discussions and conclusions are provided.
2. PROBLEM FORMULATION
In this section no-wait flowshop problem (Fm ǀnwtǀCmax) is
defined. The following assumptions and notation are
considered for the problem.
2.1. Assumptions
i) No machine is allowed to have more than one job
at the moment and if any job is initiated it cannot
be interrupted until the procedure is finished
completely.
ii) Because of no-wait assumption in the problem, no
operation can be paused and no other operation
would be done while pausing the other operation.
iii) Machines can become idle during the process.
iv) None of the machines can face break down and
there is not maintenance time allowed.
v) There is only a machine of a kind and the jobs do
not have the option of choosing between the
machines.
vi) All jobs are considered to be known beforehand.
vii) The carrying time between the machines is
neglected and that time is considered as the
process time of the previous machine.
viii) The setup time is not considered in this problem.
ix) All the data in this problem are considered to be
deterministic.
2.2. Notations
Number of the jobs that are supposed to be
scheduled
Number of the machines
Process time of the ith job on the jth machine
Minimum delay on the first machine between the
start of the ith job and kth job
The job which is processed in ith place
The total time of the processing of the job in the
current place
The total time of the process of all jobs
Minimum delay on the first machine between the start of
the ith job and kth job and the total time of the processing
of the job in the current place can be calculated by utilizing
equations (1), (2) and (3) respectively:
∑ ∑ (1)
∑ (2)
∑ ∑ (3)
Assuming that all the jobs are available in the beginning of
scheduling horizon, total process time can be calculated as
follows:
∑ (∑ ∑ ) ∑
∑ ∑ ∑ ∑
∑ ∑ ∑
(4)
It is important to note that ∑ ∑ equals the total
sum of all of the jobs in all of the machines and it has a
constant value. Also, the value of depends on the
process time of the sequence of jobs (i,k). Therefore, the
sequence of two jobs that have the minimum value of
should be chosen as the first two jobs in the process.
3. PROPOSED ALGORITHMS
In this paper, a genetic algorithm and 2 hybrid algorithms
that utilize the local search algorithms proposed in
Aldowaisan and Allahverdi [14], and Grabowski and
Pempera [15] a hybrid algorithm utilizing the both local
search methods are proposed.
3.1. Population Size
The first population size (Ps) is calculated according to
equation (5):
(5)
3.2. Selection
The selection operation used here is a Tournament
selection. Tournament selection is a method of selecting
an individual from a population of individuals in a genetic
algorithm. Tournament selection involves running several
tournaments among a few chromosomes chosen at
random from the population. The winner of each
tournament (the one with the best fitness) is chosen.

3.3. Crossover
In this algorithm, a simplified partially matched crossover
(PMX) operator is used. The major merits of this operator
would be revealed in problems of a larger size. This is a
double-point crossover. Table 1 represents the PMX
crossover.
Table -1: PMX crossover
Parent 1 i h d e f g a c b j
Parent 2 h g a b c j i e d f
Offspring 1 i h d b c j a c b j
Offspring 2 h g a e f g i e d f
As it is presented in Table 1, j, c, b are repeated twice in
first offspring and the same problem has happened in
second offspring for g, f, e and this problem need to be
dealt with. After modification of the offsprings, the new
offsprings will be as presented Table 2.
Table -2: Offsprings generated from PMX crossover
Offspring 1 i h d b c j a f e g
Offspring 2 f g i b d c h j a e
3.4. Mutation
The mutation function used in this research, chooses two
random genes in one parent and replaces them and
generates a new offspring.
3.5. Termination Criteria
While dealing with the termination of an algorithm, two
opposing options are available:
1) Termination according to a number of iterations
2) Termination according to convergence
Therefore, logical termination criteria should be defined
for defined. Two termination criteria are defined here.
i. If the number of the chromosomes with minimum
TFT is more than 60% of the population, the
iteration is terminated.
ii. If the generation of the population is more than
100, the iteration is terminated.
3.6 Algorithms
3.6.1 Hybrid algorithm with Grabowski and Pempera
[15] (Hybrid 1)
1. Set the values of n and m.
2. Calculate population size.
3. m number of the population is calculated
according to local search presented by Grabowski
and Pempera [15].
4. The rest of the population is calculated randomly.
5. Calculated the value of TFT for each chromosome.
6. Calculated the fitness function as
7. Choose a pair of parents according to tournament
selection.
8. Utilize crossover and mutation operators over the
selected parents.
9. if termination criteria are satisfied, stop. Else,
move to step 3.
3.6.2 Hybrid algorithm with Aldowaisan and
Allahverdi [14] (Hybrid 2)
3. a single member of the population is calculated
according to local search presented by
Aldowaisan and Allahverdi [14].
selection.
selected parents.
move to step 3.
3.6.3 Hybrid algorithm with both algorithms (Hybrid
3)
3. m+1 number of the population is calculated
according to local search presented by
Aldowaisan and Allahverdi [14] and Grabowski
and Pempera [15].
selection.
selected parents.

move to step 3.
4. COMPUTATIONAL RESULTS
In order to evaluate the efficiency of the proposed
algorithms the benchmarks introduced by Taillard [32]
are used in this study. The process time of the jobs on the
machines is chosen from a set of data uniformly
distributed from 0 to 100. Utilizing the genetic algorithm,
and the hybrid algorithm of the genetic algorithm with the
algorithm proposed by Aldowaisan and Allahverdi [14]
and Grabowski and Pempera [15] as well as a hybrid
genetic algorithm with both of the algorithms proposed by
Aldowaisan and Allahverdi [14] and Grabowski and
Pempera [15], the results are presented and compared.
This comparison is carried out in between 12 different
problems with 20, 30 and 50 jobs on 5, 10, 15 and 20
machines. Each problem is solved for 10 times and in sum,
480 different instances are studied.
4.1. The results of minimum value of TFT
Table 3 represents the minimum value of TFT calculated
in all 4 algorithms. The lower the value, the better the
algorithm. As it is presented, the minimum values belong
to the hybrid genetic algorithm with Grabowski and
Pempera [15] and the hybrid algorithm with both of the
local search algorithms. Furthermore, as the size of the
problem increases the hybrid algorithms outperform the
basic genetic algorithm.
Table -3: Minimum value of TFT in 4 algorithms
Jobs Machines Genetic Hybrid 1 Hybrid 2 Hybrid 3
20 5 10676 10690 10909 10592
20 10 15593 15396 15466 15659
20 15 17671 17352 17356 17057
20 20 19281 19010 19282 19010
30 5 28283 26386 27681 26517
30 10 36566 33670 36333 33603
30 15 42019 38742 39565 38569
30 20 47342 44266 46315 44113
50 5 82897 75218 81962 75333
50 10 109154 100026 107356 104622
50 15 131974 119702 127467 123748
50 20 149203 135528 145781 135717
4.2. Average Relative Error
Average relative error represents the error between the
best-found answer and the calculated one. The percentage
of the average relative error can be calculated according to
equation (6):
(6)
Where represents the minimum value of TFT in each
sample and represents the mean of the TFT in each
sample for 10 instances.
Table -4: Average Relative Error in 4 algorithms
Jobs Machines Genetic Hybrid 1 Hybrid 2 Hybrid 3
20 5 6.7248 2.4074 5.3274 2.2290
20 10 4.5713 4.3706 4.8272 3.5035
20 15 6.1810 4.9223 5.4183 4.6145
20 20 7.4476 1.0231 5.4345 2.8963
30 5 10.1804 3.7644 7.6105 2.8015
30 10 10.9508 2.9092 10.4963 4.3710
30 15 12.1833 2.2671 7.6539 3.4226
30 20 11.5891 2.2093 7.6539 2.1765
50 5 13.4037 2.0737 11.6032 4.9577
50 10 13.3941 2.7663 10.1364 7.3764
50 15 13.0233 1.8111 9.7860 7.6779
50 20 13.7189 1.8846 9.4726 5.7290
Table 4 represents the average relative error in all 4
algorithms. It is made clear that the hybrid algorithm with
Grabowski and Pempera [15] and the one with Aldowaisan
and Allahverdi [14] have less average relative error in
comparison with two other algorithms. Moreover, if the
size of the problem is small, hybrid one with Aldowaisan
and Allahverdi [14] has less average relative error than the
one with Grabowski and Pempera [15] but when the size of
the problem increases hybrid 2 shows more efficiency.
Figure 1 represents the information in Table4 in
comparative way.
Fig -1: Representation of average relative error
5. DISCUSSION AND CALCULATION
In this paper a no-wait flowshop problem was defined and
formulalized. Following that, different algorithms were
proposed as the solution concept and their performance in
dealing with a benchmark problem was compared. As
initial population plays a pivotal role and since generating
the first population randomly has been prove to be not
that effective, different methods are being utilize in
generating first population. In this paper, local search

algorithms that are based on the best responding
algorithms in the literature are utilize for generating the
first population. The results prove that hybridizing the
genetic algorithm with different heuristics for generating
the first population results in better outcomes in most of
the cases. Furthermore, the following suggestions are
made for future studies:
1. Modeling the problem in cells of machines and
considering more than one machine in each cell.
2. Utilizing different genetic algorithm operators.
3. Utilizing different meta-heuristic algorithms.
4. Doing parameter design for the algorithm
according to statistical approaches.
5. Considering the breakdown of the machines in the
problem.
6. Considering the capacity of each machine cell.
ACKNOWLEDGEMENT
The author is grateful of the chief editor and anonymous
reviewers of the paper for their attention and
consideration.
REFERENCES
[1] N. Moradinasab, M. R. Amin-Naseri, T. J.
Behbahani, H. Jafarzadeh. "Competition and
cooperation between supply chains in multi-
objective petroleum green supply chain: A game
theoretic approach". Journal of Cleaner
Production, vol. 170, 2018, pp.818-841, doi:
10.1016/j.jclepro.2017.08.114.
[2] M. Elyasi, H. Jafarzadeh, F. Khoshalhan. An
economical order quantity model for items with
imperfect quality: a non-cooperative dynamic
game theoretical model. 3rd International
Logistics and Supply Chain Conference; 2012;
Tehran, Iran.
[3] M. Elyasi, H. Jafarzadeh, F. Khoshalhan. A non-
cooperative game theoretic model for EOQ model
for items with imperfect quality Proceedings of
the 5th International Conference of the Iranian
Society of Operations Research, Tabriz, Iran;
2012.
[4] M. Elyasi, F. Khoshalhan, M. Khanmirzaee.
"Modified economic order quantity (EOQ) model
for items with imperfect quality: Game-theoretical
approaches". International Journal of Industrial
Engineering Computations, vol. 5(2), 2014, pp.
211-222. doi: 10.5267/j.ijiec.2014.1.003.
[5] M. M. Yenisey, B. Yagmahan. “Multi-objective
permutation flow shop scheduling problem:
Literature review, classification and current
trends". Omega, vol. 45, 2014, pp. 119-135, doi:
https://doi.org/10.1016/j.omega.2013.07.004.
[6] A. Allahverdi. "A survey of scheduling problems
with no-wait in process". European Journal of
Operational Research, vol. 255(3), 2016, pp. 665-
686,doi:
https://doi.org/10.1016/j.ejor.2016.05.036.
[7] J. R. Callahan. “The Nothing Hot Delay Problem in
the Production of Steel [microform]: Thesis
(Ph.D.)--University of Toronto. (1971).
[8] H. Röck. “The Three-Machine No-Wait Flow Shop
is NP-Complete". Journal of the ACM, vol. 31(2),
1984, pp. 336-345, doi: 10.1145/62.65.
[9] H. Jafarzadeh, N. Moradinasab, A. Gerami. “Solving
no-wait two-stage flexible flow shop scheduling
problem with unrelated parallel machines and
rework time by the adjusted discrete Multi
Objective Invasive Weed Optimization and fuzzy
dominance approach". Journal of Industrial
Engineering and Management, vol. 10(5), 2017,
pp. 887, doi: 10.3926/jiem.2348.
[10] M. C. Bonney, S. W. Gundry. “Solutions to the
Constrained Flowshop Sequencing Problem".
Operational Research Quarterly (1970-1977), vol.
27(4), 1976, pp. 869-883, doi: 10.2307/3009170.
[11] J. R. King, A. S. Spachis. “Heuristics for flow-shop
scheduling". International Journal of Production
Research, vol. 18(3), 1980, pp. 345-357, doi:
10.1080/00207548008919673.
[12] R. Gangadharan, C. Rajendran. “Heuristic
algorithms for scheduling in the no-wait
flowshop". International Journal of Production
Economics, vol. 32(3), 1993, pp. 285-290, doi:
https://doi.org/10.1016/0925-5273(93)90042-J.
[13] C. Rajendran. “A No-Wait Flowshop Scheduling
Heuristic to Minimize Makespan". The Journal of
the Operational Research Society, vol. 45(4),
1994, pp. 472-478, doi: 10.2307/2584218.
[14] T. Aldowaisan, A. Allahverdi. “New heuristics for
no-wait flowshops to minimize makespan".
Computers & Operations Research, vol. 30(8),
2003, pp. 1219-1231, doi:
https://doi.org/10.1016/S0305-0548(02)00068-
0.
[15] J. Grabowski, J. Pempera. “Some local search
algorithms for no-wait flow-shop problem with
makespan criterion". Computers & Operations
Research, vol. 32(8), 2005, pp. 2197-2212, doi:
https://doi.org/10.1016/j.cor.2004.02.009.

[16] C. J. Schuster, J. M. Framinan. “Approximative
procedures for no-wait job shop scheduling".
Operations Research Letters, vol. 31(4), 2003, pp.
308-318, doi: https://doi.org/10.1016/S0167-
6377(03)00005-1.
[17] J. M. Framinan, M. S. Nagano. “Evaluating the
performance for makespan minimisation in no-
wait flowshop sequencing". Journal of Materials
Processing Technology, vol. 197(1), 2008, pp. 1-9,
doi:
https://doi.org/10.1016/j.jmatprotec.2007.07.03
9.
[18] S.-W. Lin, K.-C. Ying. “Minimizing makespan for
solving the distributed no-wait flowshop
scheduling problem". Computers & Industrial
Engineering, vol. 99, 2016, pp. 202-209, doi:
https://doi.org/10.1016/j.cie.2016.07.027.
[19] M. S. Nagano, H. H. Miyata, D. C. Araújo. “A
constructive heuristic for total flowtime
minimization in a no-wait flowshop with
sequence-dependent setup times". Journal of
Manufacturing Systems, vol. 36, 2015, pp. 224-
230,doi:
https://doi.org/10.1016/j.jmsy.2014.06.007.
[20] M. Elyasi, H. Jafarzadeh, F. Khoshalhan. Hybrid Ant
Colony Optimization (ACO) algorithm for no-wait
flow-shop makespan. Proceedings of the 5th
International Conference of the Iranian Society of
Operations Research, Tabriz, Iran; 2012.
[21] H. Jafarzadeh, N. Moradinasab, M. Elyasi. "An
Enhanced Genetic Algorithm for the Generalized
Traveling Salesman Problem". Engineering,
Technology & Applied Science Research, vol. 7(6),
2017, pp. 2260-2265.
[22] L.-Y. Tseng, Y.-T. Lin. “A hybrid genetic algorithm
for no-wait flowshop scheduling problem".
International Journal of Production Economics,
vol.128(1),2010,pp.144-152,doi:
https://doi.org/10.1016/j.ijpe.2010.06.006.
[23] J. Zhu, X. Li, Q. Wang. “Complete local search with
limited memory algorithm for no-wait job shops
to minimize makespan". European Journal of
Operational Research, vol. 198(2), 2009, pp. 378-
386,doi:
https://doi.org/10.1016/j.ejor.2008.09.015.
[24] J. M. Framinan, C. Schuster. “An enhanced
timetabling procedure for the no-wait job shop
problem: a complete local search approach".
Computers & Operations Research, vol. 33(5),
2006,pp.1200-1213,doi:
https://doi.org/10.1016/j.cor.2004.09.009.
[25] C. J. Schuster. “No-wait Job Shop Scheduling: Tabu
Search and Complexity of Subproblems".
Mathematical Methods of Operations Research,
vol.63(3),2006,pp.473-491, doi: 10.1007/s00186-
005-0056-y.
[26] K.-C. Ying, S.-W. Lin, W.-J. Wu. “Self-adaptive ruin-
and-recreate algorithm for minimizing total flow
time in no-wait flowshops". Computers &
Industrial Engineering, vol. 101, 2016, pp. 167-
176,doi:
https://doi.org/10.1016/j.cie.2016.08.014.
[27] S.-W. Lin, K.-C. Ying. “Optimization of makespan
for no-wait flowshop scheduling problems using
efficient matheuristics". Omega, vol. 64, 2016, pp.
115-125,doi:
https://doi.org/10.1016/j.omega.2015.12.002.
[28] F. Zhao, H. Liu, Y. Zhang, W. Ma, C. Zhang. “A
discrete Water Wave Optimization algorithm for
no-wait flow shop scheduling problem". Expert
Systems with Applications, vol. 91, 2018, pp. 347-
363,doi:
https://doi.org/10.1016/j.eswa.2017.09.028.
[29] E. Taillard. “Benchmarks for basic scheduling
problems”. European Journal of Operational
Research, vol. 64, 1993, pp. 278-285.

Effective Hybrid Algorithms for No-Wait Flowshop Scheduling Problem

More Related Content

What's hot (15)

Similar to Effective Hybrid Algorithms for No-Wait Flowshop Scheduling Problem (20)

More from IRJET Journal (20)

Recently uploaded (20)

Effective Hybrid Algorithms for No-Wait Flowshop Scheduling Problem