You are on page 1of 32

Flowshop Scheduling Heuristics

to Minimize Makespan

PRSCHED
Eric A. Siy
DLSU I.E. Dept.
Johnson’s Rule for 2 machines and
extension for 3 machines
Job 1 Job 2 Job 3 Job 4
M1 6 4 10 2
M2 7 5 8 5

J4 – J2 – J1 – J3

Siy, De La Salle University Dept of Industrial Engineering 2


Sample Flowshop Problems:
minimize makespan
Job 1 Job 2 Job 3 Job 4 Job 1 Job 2 Job 3 Job 4
M1 6 4 10 2
M1 6 4 10 2
M2 7 5 8 5
M3 5 14 12 6
M2 7 5 8 5
M3 5 14 12 6
M4 3 2 9 11

Siy, De La Salle University Dept of Industrial Engineering 3


Heuristic methods in Flowshop scheduling
minimizing makespan
• Palmer’s Slope index method
• Dannenbring’s Heuristic
• Campbell, Dudek and Smith method
• Nawaz, Enscore and Ham (NEH) Heuristic

Siy, De La Salle University Dept of Industrial Engineering 4


Palmer’s Slope Index
• A job priority function based on job processing times which could be
used to compute a priority number for determining the job processing
sequence in flow shop scheduling problems.
• This approach was taken up by Palmer (1965), in the specification of a
job priority function which he called the "slope index" for the job. The
form of priority function was deliberately chosen to give priority to
jobs that have a tendency to progress from short to long processing
times in their passage through the machines.

https://ttu-ir.tdl.org/bitstream/handle/2346/60699/31295007107591.pdf?sequence=1&isAllowed=y

Siy, De La Salle University Dept of Industrial Engineering 5


Palmer’s Slope Index

The sequence is then determined by ranking the jobs on the


basis of the slope order index, by ordering jobs in non-
increasing order of Sj. (Descending order)

Siy, De La Salle University Dept of Industrial Engineering 6


Palmer’s Slope Index
The sequence is then determined by ranking the jobs
on the basis of the slope order index, by ordering jobs
in non-increasing order of Sj. (Descending order)
Three machine example:
factor Job 1 Job 2 Job 3 Job 4
M1 -2 6 4 10 2
M2 0 7 5 8 5
M3 +2 5 14 12 6
Sj =sumproduct(factor, Pij) -2 +20 +4 +8
-2(6)+0(7)
+2(5)

Recommended sequence:
J2-J4-J3-J1
Siy, De La Salle University Dept of Industrial Engineering 7
Palmer’s Slope Index M1
factor
-2
Job 1 Job 2 Job 3 Job 4
6 4 10 2
M2 0 7 5 8 5
Recommended sequence: M3 +2 5 14 12 6
=sumproduct(f
J2-J4-J3-J1 Three machine example: Sj actor, Pij) -2 +20 +4 +8

Siy, De La Salle University Dept of Industrial Engineering 8


Palmer’s Slope Index
The sequence is then determined by ranking the jobs
on the basis of the slope order index, by ordering jobs
in non-increasing order of Sj. (Descending order)
Four machine example:
factor Job 1 Job 2 Job 3 Job 4
M1 -3 6 4 10 2
M2 -1 7 5 8 5
M3 +1 5 14 12 6
M4 +3 3 2 9 11
Sj =sumproduct(factor, Pij) -11 +3 +1 +28
-3(6)-1(7)+
1(5)+3(3) Recommended sequence:
J4-J2-J3-J1
Siy, De La Salle University Dept of Industrial Engineering 9
https://www.youtube.com/watch?v=pTIqEu9_sYg
Job 1 Job 2 Job 3 Job 4
Palmer’s Slope Index
factor
M1 -3 6 4 10 2
M2 -1 7 5 8 5
Recommended sequence: M3 +1 5 14 12 6
J4-J2-J3-J1 M4 +3 3 2 9 11
Makespan Cmax= 51 =sumproduct(fac
Sj tor, Pij) -11 +3 +1 +28

Siy, De La Salle University Dept of Industrial Engineering 10


Dannenbring Heuristic
• Using a multiplier for each job’s machining times to find two sum of
products Aj and Bj

• Then apply Johnson’s Rule for two machines as if A and B are the
machines’ processing times.

Siy, De La Salle University Dept of Industrial Engineering 11


• Using a multiplier for each job’s machining times to find two sum of products Aj and Bj

Dannenbring Heuristic
• Then apply Johnson’s Rule for two machines as if A and B are the machines’ processing
times.

Factor for A Factor for B Job 1 Job 2 Job 3 Job 4


M1 4 1 6 4 10 2
M2 3 2 7 5 8 5
M3 2 3 5 14 12 6
M4 1 4 3 2 9 11
A =sumproduct(factor, Pij) 58 61 97 46
B =sumproduct(factor, Pij) 47 64 98 74

Applying Johnson’s Rule gives the sequence: J4-J2-J3-J1


Siy, De La Salle University Dept of Industrial Engineering 12
Dannenbring recommends: J4-J2-J3-J1
Same as Palmer’s method.

Siy, De La Salle University Dept of Industrial Engineering 13


Campbell Dudek and Smith (CDS) Heuristic
• Similar approach as Johnson’s Rule for three machines, but extended for
more machines
• Partial sums A and B of Pij’s per job from
the endpoint machines up to (M-1) deep in
each direction.

• Schedule jobs based on Johnson’s Rule with two machines (A and B sums)

Siy, De La Salle University Dept of Industrial Engineering 14


Campbell Dudek and Smith (CDS) Heuristic

Trial 1: A: M1 B: M4
Trial 2: A: M1+M2 B: M3+M4
Trial 3: A: M1+M2+M3 B: M2+M3+M4

Siy, De La Salle University Dept of Industrial Engineering 15


Campbell Dudek and Smith (CDS) Heuristic

Trial 1: Sequence 4312 Cmax= 53


Trial 2:
Trial 3:

Siy, De La Salle University Dept of Industrial Engineering 16


Campbell Dudek and Smith (CDS) Heuristic

Trial 1: Sequence 4312 Cmax=53


Trial 2: Sequence 4231 Cmax=51
Trial 3:

Siy, De La Salle University Dept of Industrial Engineering 17


Campbell Dudek and Smith (CDS) Heuristic

Trial 1: Sequence 4312 Cmax=53


Trial 2: Sequence 4231 Cmax=51
Trial 3: Sequence 4321 Cmax=54

Siy, De La Salle University Dept of Industrial Engineering 18


Campbell Dudek and Smith (CDS) Heuristic
Recommends

Trial 1: Sequence 4312 Cmax=53


Trial 2: Sequence 4231 Cmax=51
Trial 3: Sequence 4321 Cmax= 54

Siy, De La Salle University Dept of Industrial Engineering 19


Nawaz, Enscore and Ham (NEH) Heuristic
• NEH algorithm is also categorized as a constructive heuristic. To minimize the makespan of n-jobs and m-
machine problems, NEH sorts the n-jobs (j = 1, 2, 3… n) in decreasing order of the sum of their total
processing times (P ji ) on all machines, and for the kth job (k = 1, …, n) from the kth possible it locates the
position that minimizes the partial sequence makespan, and then inserts it into the position. The iterative
procedure stops after inserting the remaining job into the possible sequence.

Step 1: Sort the n-jobs in decreasing order of the sum of their processing time in machines.

Step 2: Pick the first two jobs and form two partial sequences by interchanging the position of the two jobs.
Compute makespan of the partial sequences, and call the partial sequence with a minimum value for
the next iteration.

Step 3: For k = 3 to n do step 4.

Step 4: Insert the kth job at the position, where it minimizes the partial sequence makespan among the k
possible ones.
https://www.cogentoa.com/article/10.1080/23311916.2018.1430007
Siy, De La Salle University Dept of Industrial Engineering 20
NEH Heuristic M1
Job 3
10
Job 2
4
Job 4
2
Job 1
6
Step 1: Sort the n-jobs in decreasing order of the M2 8 5 5 7
sum of their processing time in machines.
M3 12 14 6 5
Step 2: Pick the first two jobs and form two M4 9 2 11 3
partial sequences by interchanging the position Sum 39 25 24 21
of the two jobs. Compute makespan of the partial
sequences, and call the partial sequence with a
minimum value for the next iteration.
(J2-J3) vs (J3-J2) J2-J3 Cmax=44
J3-J2 Cmax=46

Siy, De La Salle University Dept of Industrial Engineering 21


NEH Heuristic M1
Job 3
10
Job 2
4
Job 4
2
Job 1
6
Step 1: Sort the n-jobs in decreasing order of the M2 8 5 5 7
sum of their processing time in machines.
M3 12 14 6 5
Step 2: Pick the first two jobs and form two M4 9 2 11 3
partial sequences by interchanging the position Sum 39 25 24 21
of the two jobs. Compute makespan of the partial
sequences, and call the partial sequence with a
minimum value for the next iteration.
(J2-J3) vs (J3-J2) J2-J3 Cmax=44
J3-J2 Cmax=46

Siy, De La Salle University Dept of Industrial Engineering 22


Nawaz, Enscore and Ham (NEH) Heuristic
• NEH algorithm is also categorized as a constructive heuristic. To minimize the makespan of n-jobs
and m-machine problems, NEH sorts the n-jobs (j = 1, 2, 3… n) in decreasing order of the sum of
their total processing times (P ji ) on all machines, and for the kth job (k = 1, …, n) from the kth
possible it locates the position that minimizes the partial sequence makespan, and then inserts it
into the position. The iterative procedure stops after inserting the remaining job into the possible
sequence.

Step 3: For k = 3 to n do step 4.

Step 4: Insert the kth job at the position, where it minimizes the partial sequence makespan among the
k possible ones.

Always J2 before
J3:
Where to insert J4?
(J4) J2 J3
J2 (J4) J3
J2 J3 (J4)
Siy, De La Salle University Dept of Industrial Engineering 23
NEH Heuristic
Always J2 before
J3:
Where to insert J4?
(J4) J2 J3 Cmax(423)=48
J2 (J4) J3
J2 J3 (J4)

Siy, De La Salle University Dept of Industrial Engineering 24


NEH Heuristic
Always J2 before
J3:
Where to insert J4?
(J4) J2 J3 Cmax(423)=48
J2 (J4) J3 Cmax(243)=50
J2 J3 (J4)

Siy, De La Salle University Dept of Industrial Engineering 25


NEH Heuristic
Always J2 before
J3:
Where to insert J4?
(J4) J2 J3 Cmax(423)=48
J2 (J4) J3 Cmax(243)=50
J2 J3 (J4) Cmax(234)=55

Siy, De La Salle University Dept of Industrial Engineering 26


NEH Heuristic
Always J4 before J2 before
J3:
Where to insert last J1?
(J1) J4 J2 J3 Cmax(1423)=59
J4 (J1) J2 J3
J4 J2 (J1) J3
J4 J2 J3 (J1)

Siy, De La Salle University Dept of Industrial Engineering 27


NEH Heuristic
Always J4 before J2 before
J3:
Where to insert last J1?
(J1) J4 J2 J3 Cmax(1423)=59
J4 (J1) J2 J3 Cmax(4123)=55
J4 J2 (J1) J3
J4 J2 J3 (J1)

Siy, De La Salle University Dept of Industrial Engineering 28


NEH Heuristic
Always J4 before J2 before
J3:
Where to insert last J1?
(J1) J4 J2 J3 Cmax(1423)=59
J4 (J1) J2 J3 Cmax(4123)=55
J4 J2 (J1) J3 Cmax(4213)=53
J4 J2 J3 (J1)

Siy, De La Salle University Dept of Industrial Engineering 29


NEH Heuristic
Always J4 before J2 before
J3:
Where to insert last J1?
(J1) J4 J2 J3 Cmax(1423)=59
J4 (J1) J2 J3 Cmax(4123)=55
J4 J2 (J1) J3 Cmax(4213)=53
J4 J2 J3 (J1) Cmax(4231)=51

Siy, De La Salle University Dept of Industrial Engineering 30


Job 4 Job 2 Job 3 Job 1
NEH Heuristic M1 2 4 10 6
M2 5 5 8 7
NEH Recommends: M3 6 14 12 5
J4-J2-J3-J1 Cmax=51 M4 11 2 9 3

Siy, De La Salle University Dept of Industrial Engineering 31


Heuristic methods in Flowshop scheduling
minimizing makespan
• Palmer’s Slope index method
• Dannenbring’s Heuristic
• Campbell, Dudek and Smith method
• Nawaz, Enscore and Ham (NEH) Heuristic

Siy, De La Salle University Dept of Industrial Engineering 32

You might also like