International Journal of Pure and Applied Mathematics
Volume 119 No. 16 2018, 4885-4894
ISSN: 1314-3395 (on-line version)
url: http://www.acadpubl.eu/hub/
Special Issue
http://www.acadpubl.eu/hub/
Optimization of Permutation Flowshop by using
Fisherman Algorithm
1
R. Pugazhenthi, 2R. Franklin Issac, 3M. Ishwariya and 4A. Parthiban
1
Department of Mechanical Engineering,
Vels Institute of Science,
Technology & Advanced Studies,
Chennai, India.
[email protected] 2
CEG,
Anna University,
Chennai, India.
[email protected] 3
Department of Computer Science and Engineering,
Madha Engineering College,
Chennai, India.
[email protected] 4
Department of Mechanical Engineering,
Vels Institute of Science,
Technology & Advanced Studies,
Chennai, India.
[email protected] Abstract
The emerging trend in manufacturing time management is very
important in all aspects, especially in scheduling plays a vital role to
minimize the makespan. Scheduling is a nondeterministic polynomial
hardest problem, which is thorny to reach the optimal sequence by
mathematically. This research article put forward a heuristic approach to
solve the Flow Shop Problem (FSP) with the computational support of the
F-Distribution technique. The Snedecor's F distribution is a continuous
higher class of probability distribution that arises a mathematical
4885
International Journal of Pure and Applied Mathematics Special Issue
distribution model which helps to distribute the frequency in the processing
time of each job through M machines. This distribution makes comfortable
to search the optimal sequence by slope index technique. The proposed
heuristic validated by the Taillord benchmark problems, the results of the
proposed heuristic compared to classical heuristic approaches and it yields
better results.
Key Words:Fisherman distribution, heuristic, flow shop, makespan,
optimization.
4886
International Journal of Pure and Applied Mathematics Special Issue
1. Introduction
The current modern manufacturing scenario every minute and second is very
important the loss of time makes the loss of productivity and growth rate of any
firm. In this view minimizing makespan, minimizing idle time of the machine,
minimizing total tardiness, etc., are the major objective of any manufacturing
firm. So time management is very important to improve the productivity and
profit of the production environment, in this virtue scheduling plays a very
important role in time management. The scheduling problems generally broadly
categorized by the static problem and dynamic problem; it means batch size is
fixed and stochastic process type problems. In the industrial scheduling problem
mostly the batch sized Permutation Flow Shop (PFS) is processed, in this a set of
machines were processed a fixed batch size of jobs. The jobs were processed
through the series of the arranged machine, the order sequence of the job is
known as job sequence which decides the minimized makespan, as well as the
elimination of job or machine, is strictly avoided. All the jobs must be processed
through all the machines no jumping of sequence allowed.
2. Literature Survey
Industrial scheduling is best time management and resource management
technique, the best scheduling makes the industry become high productivity one.
The productivity makes a more profitable, so the industrial scheduling is very
important in all the aspects. Scheduling is a tedious task of any production
operation managers in any production industry.
Solving of scheduling problems is non-polynomial, generally it is ‘n!’ possible
problem, so still, many researchers were focused in this area even after over the
six decades [1].
In scheduling the mathematical models and methods is more complex one, the
simulation of the models more costly as well time-consuming process, so for
industrial need a quick and easy solution it can be achieved only by heuristics.
Because, the heuristics are very simple, easy and more effectively find the
optimal sequences with minimum makespan.
In 1954, Johnson proposed a simple heuristic for solving of ‘n’ jobs processing
only two machines industrial scheduling problem it achieves the best solution [2].
The non-polynomial nature of the industrial scheduling problems is discussed
widely by Quan and Ling [3].
Palmer proposed a slope index heuristic in 1965, it is he first slope index heuristic
in industrial scheduling it achieves the best result [4].
A similar work was carried out by Campell, Dudek, and Smith it is popularly
known as CDS method. The CDS method yield best results, in this method, it
splitting the all the ‘m’ machines were split into two categories and they were
4887
International Journal of Pure and Applied Mathematics Special Issue
considered two machine industrial scheduling problem and applied the Johnson’s
heuristic or Johnson’s rule [5].
Dannenbring, developed a new algorithm which is the good inheritance of both
Palmer’s slope index and CDS procedures; It is a with raid access based heuristic
[6]. The traveling salesman problems with the radically difficult approach are
used by Stinson and Simith [7], it yields the best result. The outcome of the
Stinson and smith performs very well than Palmer and CDS but it needs the
computational effort in addition [8]. Still the industrial scheduling is
non-polynomial and more popular it had the nature of jobs ‘n’ factorial
combinations required to reach the optimal makespan. The heuristics are more
active in industries scheduling, a semi-finite distribution technique is developed
to reach the near-optimal solutions for the PFS problem [9].
Various search techniques are used to find an optimal makespan in the industrial
scheduling problems, they initially generate a solution and then based the
multiple lustrations or searches they find the optimal or nearer optimal solutions
[10].
The neighborhood search techniques are initiated from the current locations of
the previous better solution this can generate a better moving mechanism [11]. In
this research article proposed a new heuristic NFH to solve the permutation
flowshop, it is the combination of both mathematical and computational effect
which is using semi finite Fisherman distribution technique.
3. Methodology -Proposed Heuristic
3.1. Assumptions
1. Any job operation started in any one of the machines it can’t stop and
after completion of the previous job then only next job to be engaged in
the machine.
2. All the jobs are readily available for processing.
3. The processing time of all the jobs in the machines are well known.
4. The sequence of job processing is independent of each other.
5. All the machines are readily available to processing the jobs.
6. The first machine is assumed to be ready to process any job is to be
processed at first.
7. Generally, machines should be engaged, it may be idle for sometimes.
8. All the jobs must to process all the machine, no exception to give to the
job.
9. All the jobs process only one time in all the machines.
10. Splitting or separation of the job is not allowed.
3.2. Notations
i = Machine from 1to M.
j = Job from 1 to N.
Tij = processing time of jth job in ith machine.
Fj = index value for jth job.
4888
International Journal of Pure and Applied Mathematics Special Issue
3.3. Pseudocode
Step 1: Assign the number of machines ‘M’ and number of jobs ‘N’.
Step 2: Allocate the processing time of ‘N’ jobs in ‘M’ machines and form the
permutation flowshop problem by N x M matrix form.
Step 3: The F-index value to be calculated by using the equation for N jobs.
Step 4: The F index values are sorted by ascending order.
Step 5: Based on the sorted value, the jobs to be sequenced.
3.4. MATLAB Code
[M,N]=size(DATA);
SLOPE=zeros(N);
for i=1:N
for j=0:(M-1)
SLOPE(i)=SLOPE(i)+ FINV((DATA(j+1,i)/100),M,j+1);
end
end
SLOPE1=sort(SLOPE1,'ascend');
SEQUENCE=zeros(N);
for i=1:N
for j=1:N
if(SLOPE1(i)==SLOPE(j))
SEQUENCE(i)=j;
end
end
end
4. Results and Discussions
In any research work, the validation only decides the characterization the
proposed methodology in this research article a new fisherman algorithm based
heuristic is proposed.
The ‘F’ distribution is applied on the job processing time and with help of a slope
index method is used to find the optimal sequence. The proposed heuristic is
validated with the help of Taillard benchmark problems [12].
The ‘F’ distribution based new NFH heuristic validated with the 20, 50 & 100
jobs through 5, & 10 machines of Taillard benchmark problems which are
compared to the CDS heuristic, Palmer heuristic; The MAT LAB programming
environment is used for validating the heuristics.
The makespan of the palmer, CDS, NFH are compared with the lower bound vale
of the appropriate Taillard seeds, they are shown in table 1 to table 6.
4889
International Journal of Pure and Applied Mathematics Special Issue
Table 1: Makespan Value of 10 Jobs Process through 10 Machines
Taillard Seed Lower Bound Palmer CDS NFH
873654221 1232 1384 1409 1358
379008056 1290 1439 1424 1469
1866992158 1073 1162 1255 1275
216771124 1268 1490 1485 1534
495070989 1198 1360 1367 1422
402959317 1180 1344 1387 1429
1369363414 1226 1400 1403 1566
2021925980 1170 1313 1395 1387
573109518 1206 1426 1360 1470
88325120 1082 1229 1196 1347
Table 2: Makespan Value of 10 Jobs Process through 10 Machines
Taillard Seed Lower Bound Palmer CDS NFH
587595453 1448 1790 1829 1935
1401007982 1479 1948 2021 2108
873136276 1407 1729 1773 1856
268827376 1308 1585 1678 1720
1634173168 1325 1648 1781 1804
691823909 1290 1527 1813 1816
73807235 1388 1735 1826 1677
1273398721 1363 1763 2031 1864
2065119309 1472 1836 1831 2030
1672900551 1356 1898 2010 1841
Table 3: Makespan Value of 50 Jobs Process through 50 Machines
Taillard Seed Lower Bound Palmer CDS NFH
1328042058 2712 2774 2920 3142
200382020 2808 3041 3032 3114
496319842 2596 2777 3034 2957
1203030903 2740 2860 3156 3101
1730708564 2837 2957 3253 3023
450926852 2793 3090 3154 3053
1303135678 2689 2845 2969 3003
1273398721 2667 2826 3236 2950
587288402 2527 2733 3255 2877
248421594 2776 2915 3167 3043
Table 4: Makespan Value of 50 Jobs Process through 10 Machines
Taillard Seed Lower Bound Palmer CDS NFH
1958948863 2907 3478 3660 3803
575633267 2821 3313 3645 3508
655816003 2801 3321 3659 3514
1977864101 2968 3511 3707 3620
93805469 2908 3427 3664 3503
1803345551 2941 3323 3584 3634
49612559 3062 3457 3806 3708
1899802599 2959 3356 3758 3564
2013025619 2795 3414 3548 3512
578962478 3046 3404 3964 3665
4890
International Journal of Pure and Applied Mathematics Special Issue
Table 5: Makespan Value of 100 Jobs Process through 5 Machines
Taillard Seed Lower Bound Palmer CDS NFH
896678084 5437 5749 5592 6008
1179439976 5208 5316 5657 5693
1122278347 5130 5325 5619 5604
416756875 4963 5049 5286 5581
267829958 5195 5317 5623 5795
1835213917 5063 5274 5259 5474
1328833962 5198 5376 5557 5645
1418570761 5038 5263 5509 5569
161033112 5385 5606 5821 5824
304212574 5272 5427 5740 5643
Table 6: Makespan Value of 100 Jobs Process through 10 Machines
Taillard Seed Lower Bound Palmer CDS NFH
1539989115 5759 6161 6858 6779
655816003 5345 5889 6284 6107
960914243 5623 6127 6609 6430
1915696806 5732 6313 6783 6777
2013025619 5431 6070 6436 6608
1168140026 5246 5870 6138 6072
1923497586 5523 6442 6456 6115
167698528 5556 6168 6602 6837
1528387973 5779 6081 6356 6540
993794175 5830 6259 6852 6656
Table 7: Comparison of the Heuristics Nearer to the Percentage of Lower
Boundary
MXN Palmer CDS NFH
5 X 20 86.5 79.3 80.4
10 X 20 68.88 65.46 65.22
5 X 50 88.83 85.01 88.45
10 X 50 77.50 73.31 76.59
5 X 100 90.52 89.71 90.45
10 X 100 80.98 82.87 83.69
Average 82.20 79.27 80.80
The Table 7 shows the cumulative % of the success of NFH compared with
Palmer and CDS traditional heuristic in reaching the Lower bound value. The
results shows that the new proposed NFH performs well when compared to the
CDS method in the total average performance as well as the NFH performance
very closer to the palmer heuristic.
The fig. 1 shows the performance of the NFH, CDS and Palmer heuristics it is the
comparison of heuristics based on the % nearer to LB.
4891
International Journal of Pure and Applied Mathematics Special Issue
Fig. 1: Comparison of Heuristics based on the % Nearer to LB
Fig. 2: Overall Comparison of Heuristics based on the % Nearer to LB
The fig.2 shows the Overall Comparison of heuristics based on the % nearer to
LB, its seems to be that Palmer heuristic yield an average performance 82.2%
nearer to lower bound value. The CDS method performance is an average
performance of 79.27% nearer to lower bound value and the proposed NFH
heuristic results that the average performance of 80.8% nearer to lower bound
value. It results that the NFH gives a better result in average performance of the
percentage nearer to the lower bound value.
5. Conclusion
The optimization of flowshop is one of the crucial issues of any production
environment; this aim is achieved by the proposed heuristics based on Fisherman
Algorithm. The proposed heuristic performance enriched by a higher class of ‘F’
distribution is applied on the processing time of each job through M machines
and slope index was found. The ultimate aim of the flowshop is minimizing the
makesapn, the proposed slope index heuristic method found an optimal sequence
and it yields best results comparatively the traditional methods. The proposed
heuristic is evaluated by Taillard benchmark problems in MATLAB program
environment and which is compared to Palmer and CDS methods. The proposed
NFH performs very well when compared to CDS and almost equal to Palmer
heuristics.
4892
International Journal of Pure and Applied Mathematics Special Issue
References
[1] Pugazhenthi, R., and M. Anthony Xavior. "A heuristic to minimise
the idle time of the rental/critical machine in a
flowshop." International Journal of Services and Operations
Management 22.4 (2015): 395-412.
[2] Johnson, S.M. “Optimal Two and Three stage Production
schedule with Setup Times Included,” Naval Research Logistics
Quarterly, Vol.1, No.1 (March, 1954).
[3] Quan-Ke Pan and Ling Wang: Effective heuristics for the blocking
flowshop scheduling problem with makespan Minimization.
OMEGA, Vol. 40(2), 218-229 (2012).
[4] Palmer, D.S.: Sequencing Jobs through a Multi-Stage Process in
the Minimum Total Time – A Quick Method of Obtaining a near
Optimum. Operations Research, Vol.16, 101-107 (1965).
[5] Campbell, H.G., Dudek, R.A., and Smith, M.L. ,”A heuristic
Algorithm for the n- job m- machine Sequencing Problem”,
management Science, Vol. 16, No.10 (June, 1970).
[6] Dannenbring, D.G.: An Evolution of Flow-Shop Sequencing
Heuristics. Management Science, Vol. 23, 1174-1182 (1977).
[7] Stinson, Simith, D.T. and Hogg, G.L.: A state of art survey of
dispatching rules for manufacturing job shop operations.
International Journal of Production Research, Vol. 20, 27 – 45
(1982).
[8] Palmer, K.: Sequencing rules and due date assignments in a job
shop. Management Science, Vol. 30 (9), 1093 – 1104 (1984).
[9] Taillard, E., “Some efficient heuristic methods for the flow shop
sequencing problem” EJOR, 47, pp. 65-74, (1990).
[10] R. Pugazhenthi and M Anthony Xavio, "Optimization of
Permutation Flow Shop with Multi-Objective Criteria" International
Journal of Applied Engineering Research 8.15 (2013): 1807-1813.
[11] Franklin Issac, R., Pearlin Helina, R., Illakiya, T., Chitra, S.,
Yogalakshmi, M., and Ishwariya, M., “Computational Inverse
Branch and Bound Heuristic for Flowshop Problem”, International
Journal of Innovative Research in Technology, Science &
Engineering (IJIRTSE), Vol. 2 (6), 65-72, (2016).
[12] Taillard, Bench Marks for Basic Scheduling Problems, European
Journal of Operational Research, vol. 64, no. 2, pp.278-285,
(1993).
4893
4894
