Module 7 – Sequencing and scheduling

1. Define sequencing and scheduling?

Sequence refers to the order of carrying out activities. Scheduling is the timing (or
timetable) to carry out the activities

2. Classify production systems as shops based on the arrangement of machines?

Manufacturing shops are classified as single machine shops, flow shops and job
shops based on arrangement of machines. As the name suggests, single machine
shops have a single machine or parallel machines which means multiple copies of the
same machine. In a flow shop all jobs have the same route of visit of machines, while
a job shop, each job has a specific route of machines.

3. What are the general assumptions in sequencing and scheduling?

Some of the assumptions are: All jobs are available at the start of the scheduling
period. All machines are available right through the period. Processing times are
known and deterministic. No job splitting. A machine processes at most one job at a
time. A job once taken is completed fully before another job is taken up.

4. Mention six measures of performance in sequencing and scheduling?

Some important measures of performance are minimizing makespan, minimizing
total flow time, minimizing total tardiness, minimizing maximum tardiness,
minimizing number of tardy jobs and minimizing sum of earliness and tardiness.

5. Classify the measures of performance based on internal and external measures?

Makespan and flow time are internal measures while tardiness, earliness, tardy jobs
etc are external measures.

6. What is due date?

Due date is the time at which a task is to be completed. Any delay results in
tardiness.

7. Define lateness, earliness and tardiness? Write equations to represent them?

The term lateness is a general term used to represent the deviation from the due
date. Lateness Lj = Cj –Dj where Dj is the due date for job j and Cj is the completion
time of job j. If Lj is tardy it is early and negative it is early. Earliness Ej = Max {0, Dj –
Cj} and tardiness Tj = Max {0, Cj – Dj}

8. Mention two popular rules used in single machine sequencing? Which

performancemeasures do they attempt to minimize?
Two popular rules for single machine sequencing are the Shortest Processing time
(SPT) and Earliest Due Date (EDD) rules. While the SPT rule minimizes total flow time,
the EDD rule gives a good sequence to minimize tardiness related measures.
9. What is a flow shop?
A flow shop is an arrangement of machines such that all the jobs visit machines in
the order in which they are arranged. This is applicable to a scheduling system where
all jobs are required to visit all the machines in the same order.

10. What is a permutation flow shop?

A permutation flow shop is a flow shop where all jobs visit the machines in the same
order. If we take any machine, the order of visit of jobs is the same.

11. Name the algorithm to minimize makespan in a two machine flowshop?

The Johnson’s algorithm is used to minimize makespan on a two machine flowshop.

12. Define the steps in Johnson’s algorithm? What is the simple principle based on
whichthe algorithm performs?
The steps in Johnson’s algorithm are: From the job set, find the job with the smallest
processing time. If this is on the first machine, place the job in the first available
position from the left side and if it on the second machine, place the job in the first
available position from the right. The principle is that jobs with smaller processing
times on the first machine come early in the sequence and jobs with small
processing times on the second machine come later in the sequence.

13. Under what conditions does the Johnson’s algorithm give the optimum makespan in
thethree machine flowshop scheduling problem?
When the largest processing time on the second machine is not more than the
smallest processing time in the first or third machine, we can modify the three
machine flow shop problem into equivalent two machine problem and solve using
Johnson’s algorithm to get the optimum sequence.

14. What is a branch and bound algorithm?

The branch and bound algorithm is a form of implicit enumeration algorithm used to
solve difficult problems involving large computations. Here some of the feasible
solutions are explicitly evaluated and some are implicitly evaluated. A branch and
bound algorithm is quick when more solutions are implicitly evaluated.

15. Explain the machine based bounds used in the branch and bound algorithm?
We consider the branch and bound algorithm for the n job m machine flow shop
scheduling problem to minimize makespan. Considering a 3 machine case, there are
3 bounds (LB1 to LB3). In our usual notation, ∑ ( ),
( ) ∑ and ( ) ∑ . The
lower bound LB = max {LB1, LB2, LB3}.
16. When do we fathom a node in the branch and bound algorithm?
We fathom a node if we get a feasible solution or a complete sequence. This is called
fathoming by feasibility. When a node does not have a full sequence, we can fathom
it if its LB is more than the makespan of the best available full sequence. This is
called fathoming by bound.

17. What is a heuristic and how is it different from an optimal algorithm?

A heuristic algorithm does not guarantee optimum solutions but gives good solutions
within reasonable (polynomial) computing effort.

18. Why do we require heuristics to solve sequencing and scheduling problems?

We require heuristics to solve sequencing and scheduling problems when the
problem becomes “hard”. A hard problem does not have an optimum algorithm that
runs in polynomial time.

19. Mention two heuristics to solve the flow shop scheduling problem to
minimizemakespan?
Two heuristics that can be used to solve the flow shop scheduling problem (n jobs, m
machines) are Palmers’ heuristic and the one by Campbell Dudek and Smith (CDS
heuristic). The Palmer’s heuristic computes an index for each job and creates a
sequence by arranging the jobs in decreasing order of the index. The CDS heuristic
evaluates m-1 solutions using ideas from Johnson’s algorithm and chooses the best
sequence.

20. What is a job shop and how is it different from a flow shop?
A job shop scheduling problem is one where each job has its own set of machines
that it visits in a given order. This is different from a flow shop where all the jobs visit
all the machines in the same order.

21. What is a dispatching rule and what is a tie breaking rule in the context of job
shopscheduling?
A dispatching rule is use to choose the next job to be scheduled on a machine from
the jobs in front of it. A tie breaking rule is used when the dispatching rule when
applied does not give a unique job but gives a set of jobs from which one is chosen
using the tie breaking rule.

22. What is a Gantt chart and how is it useful in scheduling?

A Gantt chart is a chart that can represent a schedule. It is a convenient tool used in
any sequencing and scheduling process.

23. What is a non delay schedule?

 A non-delay schedule is one where a machine is not kept idle when there is a job
waiting in front of it.

24. Mention five dispatching rules based on processing times?

Five dispatching rules based on processing time are: Shortest Processing time,
Shortest total processing time, Shortest remaining processing time, Longest
processing time, Longest remaining processing time.

25. Mention five dispatching rules that do not involve processing times?
Five dispatching rules not involving processing times are: First-in-first out, random,
Minimum remaining number of operations, Last-in-first out, earliest due date.

26. Explain the shifting bottleneck heuristic for job shop scheduling?
The shifting bottleneck heuristic starts with a network representation for the job
shop scheduling problem. It identifies the bottleneck machine and solves a problem
to find the sequence of jobs on this machine that minimizes maximum tardiness. It
then solves the tardiness minimization problem on all the machines and finally gives
a feasible schedule for the job shop scheduling problem.

27. Does the shifting bottleneck algorithm give the optimum solution or better
solutionthan any dispatching rule based solution always?
The shifting bottleneck heuristic, as the name suggests is a heuristic procedure and
does not guarantee the optimum makespan always. It is expected to give better
solutions than dispatching rules but this also cannot be guaranteed always.

28. The single machine sequencing problem solved in the shifting bottleneck heuristic is
ahard problem? How do we still use the shifting bottleneck heuristic?
The single machine sequencing problem solved for every machine in the shifting
bottleneck heuristic is indeed a hard problem. Very efficient branch and bound
algorithms exist for this problem. These methods give quick solutions for reasonable
sized problems though they are implicit enumeration algorithms in the worst case.
Considering that they are fast for large problem instances they are used.

29. Can the shifting bottleneck heuristic be adapted to solve the flowshop
sequencingproblem to minimize makespan? Explain?
Yes. It can be adapted. It also gives good results.

30. How does the longest path on the network relate to makespan in the shifting
bottleneckheuristic?
The longest path gives the maximum time taken to complete all the tasks and reach
the finish. The rest of the activities are done in parallel and hence represents the
makespan.
Problems

1. Consider six jobs to be processed on a single machine with processing times 15, 10,
8, 17, 12, and 9. Find the sequence that minimizes total flow time? Compute the
total flow time?
SPT sequence minimizes total flow time. The SPT sequence is J3-J6-J2-J5-J1-J3. The
completion times (flow times) are 8, 17, 27, 39, 54, 71. The total flow time is 216.

2. Consider six jobs to be processed on a single machine with processing times 15, 10,
8, 17, 12, and 9. The due dates are 10, 20, 15, 20, 30 and 35. Find the EDD sequence
and thecorresponding flow time and total tardiness?
The EDD sequence provides a good sequence to minimize total tardiness. The EDD
sequence is J1-J3-J2-J4-J5-J6. The completion times are 15, 23, 33, 50, 62 and 71. The
corresponding due dates are 10, 15, 20, 20, 30, 35. All jobs are tardy. The tardiness
values are 5, 8, 13, 30, 32 and 36. Total tardiness = 124.

3. Consider six jobs to be processed on a single machine with processing times 12, 10,
18, 17, 10, and 9. The due dates are 10, 20, 15, 20, 30 and 35. Find the SPT sequence
and the corresponding flow time, total tardiness, number of tardy jobs, total
earliness and number of early jobs?
The SPT sequence is J6-J2-J5-J1-J4-J3. The completion times are 9, 19, 29, 41, 58, 76.
The corresponding due dates are 35, 20, 30, 10, 20, 15. Three jobs J6, J2, J5 are early
and the total earliness is 26 + 1 + 1 = 28. Three jobs J1-J4-J3 are tardy and the total
tardiness is 31 + 38 + 61 = 130.

4. Create a two job single machine sequencing problem with processing times and due
dates such that EDD rule is not optimum?
Consider p1 = 10, p2 = 8, d1 = 7, d2 = 8. EDD sequence is J1-J2 with CT = 10, 18. Both
jobs are tardy. Total tardiness = 3 + 10 = 13. Sequence J2-J1 has CT = 8, 18. DD = 8, 7
total tardiness = 11. EDD sequence is not optimum

5. Consider a five job two machine flowshop scheduling problem. The processing times
for the five jobs are J1 (27, 20), J2 (12, 9), J3 (11, 10), J4 (8, 13) and J5 (4, 16). Find
the makespan for the sequence J1‐J3‐J5‐J4‐J2? Also find the sequence that has
minimum makespan?
For the sequence J1-J3-J5-J4-J2 has job completion times 47, 57, 73, 86 and 95. The
makespan is 95. Johnson sequence is J5-J4-J1-J3-J2 with optimum makespan = 78.

6. Consider a five job two machine flowshop scheduling problem. The processing times
for thefive jobs are J1 (27, 20), J2 (12, 9), J3 (11, 10), J4 (8, 13) and J5 (4, 16). The due
dates for thejobs are 25, 30, 40, 35 and 50. Find the number of tardy jobs and the
total tardiness for thesequence J1‐J3‐J5‐J4‐J2? Also find the number of tardy jobs
and the total tardiness for the sequence that minimizes makespan?
 The sequence J1-J3-J5-J4-J2 has job completion times 47, 57, 73, 86 and 95. The
corresponding due dates are 25, 40, 50, 35, 30. All five jobs are tardy. The total
tardiness is 22 + 27 + 23 + 51 + 65 = 188. The sequence that minimizes makespan is
J5-J4-J1-J3-J2 with job completion times 20, 33, 59, 69 and 78. The corresponding
due dates are 50, 35, 25, 40, 30. Three jobs are tardy and the total tardiness = 34 +
29 + 48 = 111.

7. Consider a five job three machine flowshop scheduling problem. The processing
times forthe five jobs are J1 (27, 10, 8), J2 (12, 9, 10), J3 (11, 10, 8), J4 (8, 13, 12) and
J5 (14, 12, 9)
(a) Find the lower bound to makespan?
(b) Find the lower bound for a partial sequence J3‐J5‐J1?
(c) Modify the one of the processing time on M1 such that Johnson’s algorithm can
be applied to get the optimum solution? What should be the minimum change?
LB1 = 72 + 18 = 90; LB2 = 8 + 54 + 8 = 70; LB3 = 21 + 47 = 68. Lower bound = 90. For
the partial sequence J3-J5-J1, the completion times for J1 in three machines are 52,
62, 70. The three bounds are 72+18, 62+9+13+10, 70+10+12. The LB = 94.
Maximum on M2 is 13. At least two values on M1 has to change (processing time of
J1 and J3 on M1).

8. Consider a five job three machine flowshop scheduling problem. The processing
times forthe five jobs are J1 (27, 10, 8), J2 (12, 9, 10), J3 (11, 10, 8), J4 (8, 13, 12) and
J5 (14, 12, 9).Find a heuristic solution using the Campbell Dudek Smith (CDS)
heuristic? Evaluate thegoodness of the heuristic?
We computed LB = 90. We consider M1 and M3 and create the Johnson solution J5-
J4-J1-J3-J2 with makespan = 90. This is optimum.

9. Consider a job shop scheduling problem with three jobs and three machines. The
routes andprocessing times (in the usual notation) for the jobs are J1 – M1 (8), M2
(9), M3 (12). J2 –M3 (9), M2 (9), M1 (10), J3 – M3 (8), M1 (10), M2 (10) and J4 – M1
(12), M3 (11), M2 (8).Solve the job shop scheduling problem using SPT rule as the
dispatching rule and FIFO as tiebreaking rule. Find the makespan, mean flow time
and utilization of machines.
We draw the Gantt chart. M1 has the jobs visiting in the order J1-J3-J4-J2 with
completion times 8, 18, 30, 40. M2 has the jobs visiting in the order J1-J2-J3-J4 with
completion times 17, 26, 36, 49. M3 has the jobs visiting in the order J3-J2-J1-J4 with
completion times 8, 17, 29, 41. The completion times for the four jobs are 29, 40, 36
and 49. Makespan = 49; Mean flow time = 154/4 = 38.5; Utilizations are M1 = 40/49
= 81.6%, M2 = 36/49 = 73.5%, M3 = 40/49 = 81.6%.

10. Consider a job shop scheduling problem with three jobs and three machines. The
routes andprocessing times (in the usual notation) for the jobs are J1 – M1 (8), M3
(12), M3 (8). J2 –M3 (9), M1 (9), M2 (10), and J3 – M3 (14), M1 (12), M2 (15). Solve
the job shop schedulingproblem using SPT rule as the dispatching rule and FIFO as tie
breaking rule. Find the makespan, mean flow time and utilization of machines. If the
 due dates are 40, 45 and 50 for the three jobs, find the total tardiness and number of
tardy jobs?
We draw the Gantt chart. M1 has the jobs visiting in the order J1-J2-J3 with
completion times 8, 18, 35. M2 has the jobs visiting in the order J1-J2-J3 with
completion times 20, 30, 50. M3 has the jobs visiting in the order J2-J3-J1 with
completion times 9, 23, 31. The completion times for the three jobs are 31, 30 and
50. Makespan = 50; Mean flow time = 111/3 = 37; Utilizations are M1 = 29/50 = 58%,
M2 = 37/50 = 74%, M3 = 31/50 = 62%.

11. Consider a job shop scheduling problem with three jobs and three machines. The
routes andprocessing times (in the usual notation) for the jobs are J1 – M1 (8), M3
(12), M3 (8). J2 –M3 (9), M1 (9), M2 (10), and J3 – M3 (14), M1 (12), M2 (15). The
due dates are 40, 50 and 45 for the three jobs. Solve the job shop scheduling
problem using EDD rule as thedispatching rule and FIFO as tie breaking rule. Find the
makespan, mean flow time andutilization of machines, the total tardiness and
number of tardy jobs?
We draw the Gantt chart. M1 has the jobs visiting in the order J1-J3-J2 with
completion times 8, 26, 35. M2 has the jobs visiting in the order J1-J3-J2 with
completion times 20, 41, 51. M3 has the jobs visiting in the order J3-J2-J1 with
completion times 14, 23, 31. The completion times for the three jobs are 31, 51 and
41. Makespan = 51; Mean flow time = 123/3 = 41; Utilizations are M1 = 29/51 =
56.86%, M2 = 37/51 = 72.55%, M3 = 31/51 = 60.78%.

12. Consider a job shop scheduling problem with three jobs and three machines. The
routes and processing times (in the usual notation) for the jobs are J1 – M1 (8), M3
(12), M3 (8). J2 –M3 (9), M1 (9), M2 (10), and J3 – M3 (14), M1 (12), M2 (15). You are
given that thesequence on M1 and M3 are J1‐J3‐J2 and J3‐J2‐J1.
(a) Draw the network for the shifting bottleneck heuristic for the given information?
(b) Solve the 1/rj/Lmax problem using the branch and bound algorithm?
(c) Find the makespan using the information in (a) and (b). Draw the corresponding
Gantt Chart?
 a) The network is shown above. The longest path is Start – J3M3 – J3M1 – J2M1 –
J2M2 – Finish with length = 45.
b) We solve the 1/rj/Lmax problem on M2. The pj values are 12, 10, 15; dj values are
37, 45, 45 and rj values are 8, 35, 26. The sequence J1-J3-J2 has CT = 20, 41, 51
with maximum tardiness = 6. Makespan = 45 + 6 = 51
c) The Gantt chart is drawn with the job sequences for the machines: M1: J1-J3-J2
with completion times 8, 26 and 35; M2: J1-J3-J2 with completion times 20, 41
and 51 and M3: J3-J2-J1 with completion times 14, 23 and 31. Makespan is 51.

13. Consider a job shop scheduling problem with three jobs and three machines. The
routes and processing times (in the usual notation) for the jobs are J1 – M1 (8), M3
(12), M3 (8). J2 –M3 (9), M1 (9), M2 (10), and J3 – M3 (14), M1 (12), M2 (15). The
due dates are 40, 45 and 50 for the three jobs. Solve using the shifting bottleneck
heuristic to get the makespan? Find the total tardiness for the schedule?
The network is drawn. The machine loads are 29, 37 and 31 which gives us the order
M2, M3 and M1. We start with makespan = 37 (maximum load). Solving a single
machine problem on M2 gives us L = 8 and the sequence J1-J2-J3 on M2. The
makespan becomes 45 and solving for M2, we get the sequence J2-J3-J1 with L = 5.
The makespan becomes 50 and we solve for M1. The sequence is J1-J2-J3 with L = 0.
The final makespan is 50. The job completion times are J1 = 31, J2 = 30 and J3 = 50