Production Scheduling P.C. Chang, IEM, YZU. 1 Flow Shop Scheduling
Feb 01, 2016
Production Scheduling P.C. Chang, IEM, YZU.1
Flow Shop Scheduling
Production Scheduling P.C. Chang, IEM, YZU.2
Definitions
• Contains m different machines.
• Each job consists m operators in different machine.
• The flow of work is unidirectional.
• Machines in a flow shop = 1,2,…….,m
• The operations of job i , (i,1) (i,2) (i ,3)…..(i, m)
• Not processed by machine k , P( i , k) = 0
Production Scheduling P.C. Chang, IEM, YZU.3
Flow Shop SchedulingBaker p.136
The processing sequence on each machine are all the same.
1
2.....M
2 3 1 5 4
2 3 1 5 4
Flow shop
Job shop
n! - flow shop permutation schedulen!.n! …….n! - Job shop
m)!n(k
)!n( m
k : constraint( routing problem)∵
1 3 2 4 5
or
Production Scheduling P.C. Chang, IEM, YZU.4
Workflow in a flow shop
Machine1
Machine2
Machine3
MachineM-1
MachineM
….
Input
output
Machine1
Machine2
Machine3
MachineM-1
MachineM
….
Input
outputoutputoutputoutputoutput
Input Input Input Input
Type 1.
Type 2.
Baker p.137
Production Scheduling P.C. Chang, IEM, YZU.5
Johnson’s Rule
Note:Johnson’s rule can find an optimum with two machinesFlow shop problem for makespan problem.
Baker p.142
.filledaresequenceinpositionsalluntil1steptoreturnandionconsideratfromjobassignedtheRemove:3Step
.3steptogo.sequenceinpositionavailablefirsttheinjobtheplace,2machinerequiresmintheIf:2Step
.3steptogo.sequenceinpositionavailablefirsttheinjobtheplace,1machinerequiresmintheIf:2Step
Find:1Step
tb
ta}t,{tmin i2i1i
Production Scheduling P.C. Chang, IEM, YZU.6
Ex.
j 1 2 3 4 5
tj1 3 5 1 6 7
tj2 6 2 2 6 5
Stage U Min tjk Assignment Partial Schedule
1 1,2,3,4,5 t31 3=[1] 3 x x x x
2 1,2,4,5 t22 2=[5] 3 x x x 2
3 1,4,5 t11 1=[2] 3 1 x x 2
4 4,5 t52 5=[4] 3 1 x 5 2
5 4 t11 4=[3] 3 1 4 5 2
Production Scheduling P.C. Chang, IEM, YZU.7
Ex.
3 1 4 5 2
3 1 4 524
M1
M2
The makespan is 24
2
Production Scheduling P.C. Chang, IEM, YZU.8
The B&B for Makespan Problem
The Ignall-Schrage Algorithm (Baker p.149)- A lower bound on the makespan associated with any comp
letion of the corresponding partial sequence σ is obtained by considering the work remaining on each machine. To illustrate the procedure for m=3.
For a given partial sequence σ, let
q1= the latest completion time on machine 1 among jobs in σ.
q2= the latest completion time on machine 2 among jobs in σ.
q3= the latest completion time on machine 3 among jobs in σ.
The amount of processing yet required of machine 1 is 'j
1jt
Production Scheduling P.C. Chang, IEM, YZU.9
The Ignall-Schrage AlgorithmIn the most favorable situation, the last job
1) Encounters no delay between the completion of one operation and the start of its direct successor, and
2) Has the minimal sum (tj2+tj3) among jobs j belongs to σ’
Hence one lower bound on the makespan is
A second lower bound on machine 2 is
A lower bound on machine 3 is
The lower bound proposed by Ignall and Schrage is
}tt{mintqb 3j2j'j'j
1j11
}t{mintqb 3j'j'j
2j22
'j
3j33 tqb
}b,b,bmax{B 321
Production Scheduling P.C. Chang, IEM, YZU.10
The Ignall-Schrage Algorithm
M1
M2
M3
tk1
tk2
tk3
……..
……..
……..
q1
q2
q3 b1
M2
M3
tk2
tk3
……..
……..
q2
q3 b2
Job in σ’
Production Scheduling P.C. Chang, IEM, YZU.11
Ex. B&B
j 1 2 3 4
tj1 3 11 7 10
tj2 4 1 9 12
tj3 10 5 13 2
m=3For the first node: σ =1
37)37,31,37max(B372017b
312227b376283b
boundlowerThe17tttq
7ttq3tq
3
2
1
1312113
12112
111
212
139
51
min6
Production Scheduling P.C. Chang, IEM, YZU.12
Ex. Partial Sequence
( q1 , q2 , q3 ) (b1,b2,b3) B
1xxx ( 3 , 7 , 17 ) ( 37 , 31 , 37 )
37
2xxx ( 11 , 12 , 17 ) ( 45 , 39 , 42 )
45
3xxx ( 7 , 16 , 29 ) ( 37 , 35 , 46 )
46
4xxx ( 10 , 22 , 24 ) ( 37 , 41 , 52 )
52
12xx ( 14 , 15 , 22 ) ( 45 , 38 , 37 )
45
13xx ( 10 , 19 , 32 ) ( 37 , 34 , 39 )
39
14xx ( 13 , 25 , 27 ) ( 37 , 40 , 45 )
45
132x ( 21 , 22 , 37 ) ( 45 , 36 , 39 )
45
134x ( 20 , 32 , 34 ) ( 37 , 38 , 39 )
39
1 2
1 2
1 2
0 3 14
7 15
17 22 45
212
139min)107(14
}tt{mintqb 3j2j'j'j
1j11
Production Scheduling P.C. Chang, IEM, YZU.13
Ex. B&B
P0
1xxx 2xxx 3xxx 4xxx
B=37
B=45
B=46
B=52
12xx 13xx 14xx
B=45
B=45
B=39
132x 134x
B=45
B=39
Production Scheduling P.C. Chang, IEM, YZU.14
Refined Bounds
'j
1j122 ]tmin[q,qmax'q
Modification1:
'j
2j1j1'j
2j233 ]ttmin[q,]tmin[q,qmax'q
The use of q2 in the calculation of b2 ignores the possibility that the starting time of job j on the machine 2 may be constrained by commitments on machine1. Hence:
consider idle time
Production Scheduling P.C. Chang, IEM, YZU.15
Refined Bounds
Previous : Machine-based boundModification2 : Job-based bound
}b,b,Bmax{'B
t,tminttmax'qb
t,tmintttmaxqb
54
kj'j
3j2j3k2k'k
25
kj'j
3j1j3k2k1k'k
14
Modification2: (McMahon and Burton)
Production Scheduling P.C. Chang, IEM, YZU.16
Refined Bounds
Obviously, B’>=B, This means that the combination of machine-based and job-based bounds represented by B’ will lead to a more efficient search of the branching tree in the sense that fewer nodes will be created.
Production Scheduling P.C. Chang, IEM, YZU.17
Hw.
a. Find the min makespan using the basic Ignall-Schrage algorithm. Count the nodes generated by the branching process.
b. Find the min makespan using the modified algorithm.
j 1 2 3 4
tj1 13 7 26 2
tj2 3 12 9 6
tj3 12 16 7 1
Consider the following four-job three-machine problem
Production Scheduling P.C. Chang, IEM, YZU.18
Heuristic Approaches
Traditional B&B:
• The computational requirements will be severe for large problems
• Even for relatively small problems, there is no guarantee that the solution can be obtained quickly,
Heuristic Approaches
• can obtain solutions to large problems with limited computational effort.
• Computational requirements are predictable for problem of a given size.
Production Scheduling P.C. Chang, IEM, YZU.19
Palmer
Palmer proposed the calculation of a slope index, sj, for each job.
1,j2,j2m,j1m,jm,jj t)1m(t)3m(t)5m(t)3m(t)1m(s
Then a permutation schedule is constructed using the job ordering
]n[]2[]1[ sss
Production Scheduling P.C. Chang, IEM, YZU.20
Gupta
Gupta thought a transitive job ordering in the form of follows that would produce good schedules. Where
}tt,ttmin{
es
3j2j2j1j
jj
Where
3j1j
3j1jj ttif1
ttif1e
Production Scheduling P.C. Chang, IEM, YZU.21
Gupta
Generalizing from this structure, Gupta proposed that for m>3, the job index to be calculated is
}tt{min
es
1k,jjk1mk1
jj
Where
jm1j
jm1jj ttif1
ttif1e
Production Scheduling P.C. Chang, IEM, YZU.22
CDS
Its strength lies in two properties:
1.It use Johnson’s rule in a heuristic fashion
2.It generally creates several schedules from which a “best” schedule can be chosen.
The CDS algorithm corresponds to a multistage use if Johnson’s rule applied to a new problem, derived from the original, with processing times and . At stage 1, 1j't 2j't
jm2j1j1j t'tandt't
Production Scheduling P.C. Chang, IEM, YZU.23
CDS
In other words, Johnson’s rule is applied to the first and mth operations and intermediate operations are ignored. At stage 2,
1m,jjm2j2j1j1j tt'tandtt't
That is, Johnson’s rule is applied to the sums of the first two and last two operation processing times. In general at stage i,
i
1k1km,j2j
i
1kjk1j t'tandt't
Production Scheduling P.C. Chang, IEM, YZU.24
Ex.
Palmer:
j 1 2 3 4 5
tj1 6 4 3 9 5
tj2 8 1 9 5 6
tj3 2 1 5 8 6
3712453
22468
2211
54321
1313
M
sssss
tttmtms jjjjj
Gupta:
36M2143511
1s
13
1s
12
1s
2
1s
10
1s 54321
CDS: 3-5-4-1-2 M=35
Production Scheduling P.C. Chang, IEM, YZU.25
HW.
Let
1. Use Ignall-Schrage & McMahon-Burton to solve
2. Use Palmer, Gupta, CDS to solve this problem.
j 1 2 3 4 5
tj1 8 11 7 6 9
tj2 3 2 5 7 11
tj3 6 5 7 13 10
}3,1{
2 21 2 3 4 5 13 31, , , , ,xxx xxxb b b b b of P P