IJISET - International Journal of Innovative Science, Engineering & Technology, Vol. 2 Issue 1, January 2015.

www.ijiset.com
ISSN 2348 – 7968

Finding an Optimal Sequence in the Flowshop Scheduling

Using Johnson’s Algorithm
Prabhu M.S1, Sankar D2 and Paramaguru V3
1 ,2,3
Department of Aeronautical Engineering, Hindusthan Institute of Technology,
Coimbatore, Tamilnadu-641032, India

Abstract The goal is to minimize the total completion time of all the
This work is concerned about the minimization of the make span activities and the approach which are used is a linear
and also to establish the idle time for the machines in a flow shop programming which dominates the heuristic models which
scheduling environment by using Johnson’s algorithm for two mostly used for the NP-Hard problems. to do this, first of
machine problems and Extension of Johnson’s algorithm for
all we convert the production system into the network form,
three machine problems. The article also focuses on time-in and
time-out strategy for two machines as well as for three machine
then we find the critical activities which affects the total
problem. The work proposes to find the optimal sequence in completion time (makespan), then we assign some budget
which the job will be scheduled to flow in the machines using to the activities to crash them, by assigning some budget to
linear programming model which dominates the heuristic models some of the operations, (Hojjati & Sahraeyan, 2009) the
used for NP hard problems. operation time of these activities reduces and affects the
Keywords: makespan, idle time, flow shop, scheduling, total completion time of all the operations and because of
Johnson’s algorithm, linear programming, heuristic models, NP the shortage of the budget, the problem is solved and
hard. determines which activities are better to absorb the limited
budget to minimize the makespan.
1. Introduction
2. Job Shop Production System
The production scheduling in general are three forms, Job
shop production scheduling , Flow shop production A job shop is a type of manufacturing process structure
scheduling and project production scheduling. where small batches of a variety of custom products are
made. In the job shop process flow, most of the products
In the traditional flow shop scheduling problem, it is produced require a unique set-up and sequencing of
assumed that there is only one machine at each stage to processing steps. Similar equipment or functions are
execute passing jobs. With the development of hardware, grouped together, such as all drill presses in one area and
software, and theory in parallel computing, the traditional grinding machines in another in a process layout. The
model of flow shop scheduling is becoming somewhat layout is designed to minimize material handling, cost, and
unrealistic. Defined to capture the essence of parallel work in process inventories.
computing is the so-called hybrid flow shop model, in
which each job has to go through multiple stages with Job shops use general purpose equipment rather than
parallel machines instead of a single machine (Havill & specialty, dedicated product-specific equipment. Digital
Mao, 2002). numerically controlled equipment is often used to give job
shops the flexibility to change set-ups on the various
Scheduling is one of the most important decisions in machines very quickly. Job shops compete on quality,
production control systems. Every production system speed of product delivery, customization, and new product
should have a kind of production scheduling, no matter introduction, but are unlikely to compete on price as few
whether it is managed and organized traditionally or have a scale economies exist. When an order arrives in the job
systematic and scientific approach to the planning in the shop, the part being worked on travels throughout the
production system. If a scientific approach to production various areas according to a sequence of operations. Not
planning is organized, we can be sure that a better usage of all jobs will use every machine in the plant. Jobs often
the resources especially the machinery and the manpower travel in a jumbled routing and may return to the same
are considered and a better situation for competition are machine for processing several times. This type of layout is
formed in the market. In this we try to use a mathematical also seen in services like department stores or hospitals,
optimization model for doing this job. where areas are dedicated to one particular product or one
type of service.

350
 IJISET - International Journal of Innovative Science, Engineering & Technology, Vol. 2 Issue 1, January 2015.
www.ijiset.com
ISSN 2348 – 7968
A job is characterized by its route, its processing 3. Methodology Selection
requirements, and its priority. In a job shop the mix of
products is a key issue in deciding how and when to Makespan is one of the most important criteria in every
schedule jobs. Jobs may not be completed based on their production systems; it is equal to the total completion time
arrival pattern in order to minimize costly machine set-ups of all the activities. Minimizing this criterion caused better
and change-overs. Work may also be scheduled based on usage of the resources specially machinery and manpower.
the shortest processing time. In both simple and hybrid flow shop, the methodology is to
convert the flow shop into a network form, then a linear
programming model with the objective of minimizing the
Capacity is difficult to measure in the job shop and total completion time of all the activities are constructed.
depends on lot sizes, the complexity of jobs, the mix of Minimizing total completion time of all the activities is
jobs already scheduled, the ability to schedule work well, equivalent to minimizing makespan in the production
the number of machines and their condition, the quantity system. The result is that the sequencing and scheduling of
and quality of labor input, and any process improvements. all the activities are determined. The data is analyzed and
the understood there is enormous opportunities for
2.1 Job shop characteristics improvement. The methodologies are considered for the
implementation Flow shop Scheduling using Johnson’s
 Low volume, high variety customized products Algorithm and Extension of Johnson’s rule
 Flexible resources
 Skilled human resources
3.1 Johnson’s algorithm
 Jumbled work flows
 High material handling Johnson's algorithm is a way to find the shortest paths
 Large of inventories between all pairs of vertices in a sparse directed graph.In
 Long flow time operations research Johnson's rule is a method of
 Highly structured information system scheduling jobs in two work centers. Its primary objective
 High cost per unit of production but low is to find an optimal sequence of jobs to reduce makespan
investment (the total amount of time it takes to complete all jobs). It
also reduces the number of idle time between the two work
2.2 Scheduling centers. Results are not always optimal, especially for a
small group of jobs.

Scheduling is the allocation of start and finish time to each

particular order. Therefore scheduling can bring 3.2 Techniques
productivity in shop floor by providing a calendar for
processing a set of jobs. It is nothing but scheduling  The time for each job must be constant.
various jobs on a set of resources (machines) such that  Job times must be mutually exclusive of the job
certain performance measures are optimized. sequence.
 All jobs must go through first work center before
going through the second work center.
2.3 Shop scheduling  There must be no job priorities.

3.3 Johnson’s rules

In flow shop scheduling problem, there are ‘n’ jobs; each
requires processing on ‘m’ different machines. The order  List the jobs and their times at each work center.
in which the machines are required to process a job is  Select the job with the shortest activity time. If
called process sequence of that job. The process sequences that activity time is for the first work center, then
of all the jobs are the same. But the processing times for schedule the job first. If that activity time is for
various jobs on a machine may differ. If an operation is the second work center then schedule the job last.
absent in a job, and then the processing time of the Break ties arbitrarily.
operation of the job is assumed to be zero.  Eliminate the shortest job from further
consideration.

351
 IJISET - International Journal of Innovative Science, Engineering & Technology, Vol. 2 Issue 1, January 2015.
www.ijiset.com
ISSN 2348 – 7968
 Repeat steps 2 and 3, working towards the center the objective of minimizing the sum of absolute deviation
of the job schedule until all jobs have been using the processing times on the second machine where
scheduled. the ready times are actually the completion times of the
jobs on the first machine. The completion times on the first
3.4 Procedure machine (or the release times for the second machine) are
computed such that the jobs are scheduled with no idle
STEP 1: Find the minimum among various t i1 and t i2 . time on the first machine since the insertion of idle time on
STEP 2a: If the minimum processing time requires the first machine does not improve the objective function
machine 1, place the associated job in the first available value.
position in sequence. Go to STEP 3
STEP 2b: If the minimum processing time requires 3.7 Sequencing
machine 2, place the associated job in the last available
position in the sequence. Go to STEP 3 The most important thing in finding sequence is to find out
STEP 3: Remove the assigned job from consideration and the flow of the jobs in which sequence it has to load to the
return to Step 1 until all positions in sequence are filled. machines. The sequence can be found out by their
(Ties may be broken randomly) processing times of the jobs in machine 1 and machine 2.it
also allows finding out the idle time of the machines. The
3.5 Extension Of Johnson’s Rule jobs will be in a flow for the first machine when it comes
to the second machine it will moving according to the jobs
Extension of Johnson’s algorithm is applicable only if the completed on the machine 1.if the job is completed before
machine exceeds more than 2 and it also should satisfy the the first machine then the next job should continue with the
condition (min t i1 ≥ maxt i2 , mint i3 ≥ maxt i2 ). The sequence end time of the first machine, if the job exceeds the time on
can be found out by the same procedure instead we have to the first machine then the next job should continue the
make the three machine problem to a two machine by proceedings. Sequence can be found out by minimum
combining (Machine1 and Machine 2) (Machine 2 and processing time, it should be calculated according to the
Machine 3).the combination of Machine 1 and Machine 2 ascending order. If the minimum processing time is on
will be considered as machine 1 and then the combination machine 2 means it should start from right to left and if it
of Machine 2 and Machine 3 as Machine 2.then it is is on machine 1 means it will be from right to left. In the
proceeded in the usual as Johnson’s algorithm. In this table 1.1 the minimum processing time is machine 2 and
algorithm machine 3 will also be included and idle time job is 1, so it is started from the right. And the next
will be calculated for Machine 3 also. The jobs can be minimum processing time is 4 and job is 4, so that is also
many in number but if the machine exceeds more than started from right. If the processing times on both the
three then we should use another heuristics to solve the machines are same means we can select randomly the
conditions. machines. An example for the sequencing is shown from
the table 1.
3.6 Heuristic Algorithms Sequence = 3 – 2 – 5 – 4 – 1
Machine1 Machine 2
Choi and Kim developed two simple heuristic algorithms,
two hybrid algorithms, and three constructive algorithms
for m-machine re-entrant flow shop, where the key idea of
3 2 5 4 1
a hybrid algorithm is a combination of simple heuristic
algorithms and the essence of a constructive algorithm is
an improvement of an initial sequence which can be 3.8 Optimal Sequences:
obtained randomly or by simple heuristic algorithms.
Optimal sequencing is the term where there will be a
sequence of flow in the jobs in which the make span and
3.6 Heuristic solution
the idle time will be less so that in that sequence of flow of
Given a sequence of jobs on both machines, the model jobs can be allocated to the machines.in the flow shop
transforms to a linear programming problem. When the problem the sequences will be in factorial depending upon
sequence is known, the binary variables are fixed. The the machine in which the jobs will be allotted. For the
model can determine a schedule given a sequence. problem in table in there are five jobs so there will be 5
Furthermore, given a sequence, the problem can be factorial so many no of sequences are possible. The
reformulated as a single machine problem with release calculation of all the sequences should be necessary, so
times. The formulation for the single machine would have that we can find out the optimal sequences.an example for
the sequence 3-2-5-4-1 is shown in figure 1,where the

352
 IJISET - International Journal of Innovative Science, Engineering & Technology, Vol. 2 Issue 1, January 2015.
www.ijiset.com
ISSN 2348 – 7968
make span or completion time of the jobs is 30 and the idle Table 3 Flow shop scheduling for combination of two
time is 3.like this the make span and the idle time should machines
be calculated for all the sequences,so that we can find out Lathe + Drilling Drilling + Boring
Jobs
the optimal sequence and then that sequence is alloted to (mins) (mins)
the machine.the main disadvantage in this is it will take
J1 13 9
sometime to find out the optimal sequence.
J2 16 15
4. Results and Discussion
J3 8 10

4.1 Flowshop Scheduling For Two Machines J4 10 9

Condition: This problem has two machines and five jobs J5 15 9

with processing times. Machine 1 drilling operation and
machine 2 with boring operation. The processing times
will vary with jobs. Their make span and idle time are
calculated below using the Gantt chart and time-in and
time-out strategy

Table 1Flow shop scheduling for two machines

Drilling Boring
Jobs
(mins) (mins)
J1 4 1
J2 5 8
J3 3 9
J4 6 4
J5 7 5

Figure 1 Gantt Chart, Tables And Graphs

4.2 Extension Of Johnson’s Algorithm For Three
Machines
Condition: this problem has three machines and five jobs
with processing times. Machine 1 turning operation, Table 4 Time-In And Time-Out Strategy For Two
machine 2 with drilling operation, machine 3 with boring Machines
operation. The processing times will vary with jobs. Their
make span and idle time are calculated below using the Process to Process to Idle
Gantt chart and time-in and time-out strategy Drilling Machine Boring Machine time in
Jobs Time Time Time Time Boring
Table 2 Flow shop scheduling for three machines In Out In Out Machine
Lathe Drilling Boring (mins) (mins) (mins) (mins) (mins)
Jobs
(mins) (mins) (mins)
3 0 3 3 12 3
J1 8 5 4
J2 10 6 9 2 3 8 12 20 0
J3 6 2 8
5 8 15 20 25 0
J4 7 3 6
J5 11 4 5 4 15 21 25 29 0

1 21 25 29 30 0

353
 IJISET - International Journal of Innovative Science, Engineering & Technology, Vol. 2 Issue 1, January 2015.
www.ijiset.com
ISSN 2348 – 7968

Table 5 Sequences

Make Idle Make Idle

Sequence Span Time Sequence Span Time
(mins) (mins) (mins) (mins)
1-2-3-4-5 35 8 2-1-5-4-3 35 8

1-2-3-5-4 35 8 2-1-5-3-4 34 7

1-2-5-4-3 35 8 3-1-2-4-5 30 3

1-2-4-3-5 35 8 3-1-2-5-4 30 3

1-2-5-3-4 35 8 3-1-4-5-2 33 6

1-2-4-5-3 36 9 3-1-4-2-5 31 4

2-1-3-4-5 32 5 3-1-5-2-4 31 4 Figure 3 Sequences & Idle Time

2-1-3-5-4 32 5 3-1-5-4-2 33 6

2-1-4-3-5 33 6 4-1-3-5-2 35 8

2-1-4-5-3 36 9 4-1-3-2-5 35 8

Figure 4 Gantt Chart For Three Machines

Figure 2 Sequences & Make Span

354
 IJISET - International Journal of Innovative Science, Engineering & Technology, Vol. 2 Issue 1, January 2015.
www.ijiset.com
ISSN 2348 – 7968
Table 6 Time-In And Time-Out Strategy For Three Machines
Process to Drilling
Process to Lathe (mins) Process to Boring (mins) Idle time in Idle time in
(mins)
Jobs drilling machine boring machine
(mins) (mins)
Time In Time Out Time In Time Out Time In Time Out

3 0 6 6 8 8 16 6 8

2 6 16 16 22 22 31 8 6

5 16 27 27 31 31 36 5 0

1 27 35 35 40 40 44 4 4

4 35 42 42 45 45 51 2 1

[8] G. Finke, H. Jiang. A variant of the permutation flow shop

model with variable processing times, Discrete Applied
5. Conclusion & Future Guidelines Mathematics 76 (1997) 123-140.
[9] Tamer Eren a,*, ErtanGu¨ner b,1. A bicriteriaflowshop
From this study it is clear that for two machine and more scheduling with a learning effect, Applied Mathematical
than two machine problems, Johnsons and extended Modelling 32 (2008) 1719–1733.
[10] Harm Bart , Leo Kroon , Rob Zuidwijk . Quasicomplete
Johnsons algorithm are used to find the makespan and idle
factorization and the two machine flow shop problem, Linear
time for a particular factorial series. The prime critic Algebra and its Applications 278 ( 1998) 195-219.
observed in this algorithm is, it will take some time to find [11] Ching-Jong Liao, Li-Man Liao. Improved MILP models for
the optimal sequences. On practicing this strategy to three two-machine flowshop with batch processing machines,
components and results are really good and we suggested Mathematical and Computer Modelling 48 (2008) 1254–1264.
to follow the system to all the parts with continuous [12]
monitoring and improvement. This work can also be Caixia Jing, Guochun Tang, XingsanQian, Heuristic algorithms
extended to design a algorithm to minimize the idle time. for two machine re-entrant flow shop, Theoretical Computer
Science 400 (2008) 137–143.
[13] T. SAWIK, An Exact Approach for Batch Scheduling in
References Flexible Flow Lines with Limited Intermediate Buffers,
[1] Vitaly A. Strusevich. Two machine flow shop scheduling Mathematical and Computer Modelling 36 (2002) 461-471.
problem with no wait in process: Controllable machine speeds, [14] P. DAMODARAN AND K. SRIHARI, Mixed Integer
Discrete Applied Mathematics 59 (1995) 75-86. Formulation to Minimize Makespan in a Flow Shop with Batch
[2] Eugeniusz Nowicki and CzeslawSmutnicki. New results in Processing Machines, Mathematical and Computer Modelling 40
the worst-case analysis for flow-shop scheduling, Discrete (2004) 1465-1472.
Applied Mathematics 46 (1993) 21-41. [15] MilindDawande, SrinageshGavirneni, Ram Rachamadugu.
[3] Yu.N. Sotskov, N.V. Shakhlevich. NP-hardness of shop- Scheduling a two-stage flowshop under makespan constraint,
scheduling problems with three jobs, Discrete Applied Mathematical and Computer Modelling 44 (2006) 73-84.
Mathematics 59 (1995) 237-266. [16] D. Hajinejad , N. Salmasi, R. Mokhtari. A fast hybrid
[4] A. Agnetis, D. Pacciarelli, F. Rossi. Batch scheduling in a particle swarm optimization algorithm for flow shop sequence
two-machine flow shop with limited buffer, Discrete Applied dependent group scheduling problem, ScientiaIranica E (2011)
Mathematics 72 (1997) 243-260. 18 (3), 759–764.
[5] Sternerg. The Connection Between The Bump Number [17] X. Li, M. Yin . A discrete artificial bee colony algorithm
Problem And The Flow-Shop Scheduling With Precedence with composite mutation strategies for permutation flow shop
Constraints, Computers Math. Applic.Vol. 20, No. 11, pp. 13-18, scheduling problem, ScientiaIranica E (2012) 19 (6), 1921–1935
1990. [18] Lin-HuiSun , Lin-YanSun, Ming-ZhengWang , Ji-BoWang.
[6] Johann Hurinka, Sigrid Knust. Makespan minimization for Flow shop makespan minimization scheduling with deteriorating
Flow-shop problems with transportation times and a single robot, jobs under dominating machines, Int. J. Production Economics
Discrete Applied Mathematics 112 (2001) 199–216. 138 (2012) 195–200.
[7] K.N. Kashyrskikha;1, C.N. Pottsb, S.V. Sevastianov. A 3/2 [19] Arun Guptaaand Sant Ram chauhan, A heuristic algorithm
approximation algorithm for two-machine! ow-shop sequencing for scheduling in a flow shop environment to minimize
subject to release dates, Discrete Applied Mathematics 114 makespan. International Journal of Industrial Engineering
(2001) 255–271. Computations 6 (2015).

355