OFFICIAL (CLOSED) \ NON-SENSITIVE

Lesson 12
Manufacturing Cells and
Travel Network
E210 – Operations Planning

SCHOOL OF ENGINEERING E210 – OPERATIONS PLANNING I

 OFFICIAL (CLOSED) \ NON-SENSITIVE

E210 Operations Planning Topic Tree

2
 OFFICIAL (CLOSED) \ NON-SENSITIVE

Mini-Lecture (1)
What is Grouping or Clustering?
 OFFICIAL (CLOSED) \ NON-SENSITIVE

Purpose of Clustering
 An approach in which people or objects are grouped together
based on similar attributes or traits.

 For people, attributes like age and interests can be used as

grouping criteria. For objects, it can be functionalities, size,
weight, etc. depending on the purpose

 The objective of grouping is to exploit the similarity between

entities to be grouped. Applications include:

 Facility layout – placing departments or machines with similar

functionalities close by.
 Image processing and segmentation (extraction of objects of
interests from an image)
 Transmission tracing (e.g. tracing of disease, gene mutation)
 Social network analysis and formation of communities
 Marketing and sales – consumer segregation and targeted
marketing 4
 OFFICIAL (CLOSED) \ NON-SENSITIVE

How to Group?

 How to group people/ objects (entities) into specific

groups so that differences in attributes are minimized?

1. Use a measure (such as dissimilarity index) to determine the

degree of ‘likeness’ among the entities.

2. Construct a minimum spanning tree to connect the entities

using their dissimilarity index.

3. Group the entities of the minimum spanning tree. Grouping

criteria can be minimization of total dissimilarity, number of
groups, balanced number of people (objects) within groups
or some other attributes.

5
 OFFICIAL (CLOSED) \ NON-SENSITIVE

Mini-Lecture (2)
Dissimilarity Index
 OFFICIAL (CLOSED) \ NON-SENSITIVE

Dissimilarity Index
 A measure of dissimilarity, dij between the entity i and j can be
expressed as:

nij
dij = 1 −
nij + mij

nij is the number of common attributes between entities i and j

mij is the number of attributes of either entity i or entity j only
 In network representation

 Dissimilarity measure dij - Arcs

 Entities (objects or people) - Nodes

7
 OFFICIAL (CLOSED) \ NON-SENSITIVE

Dissimilarity Index (Example)

Friend Countries visited
Alan Malaysia, Thailand, Australia
Bob China, Australia, Japan, South Korea
Cliff China, Japan, New Zealand

Calculate the Dissimilarity between friends based on the countries visited

• Dissimilarity between Alan and Bob,

• DAB = 1 – 1/(1+5) = 0.833
Which pair of
friends are most
• Dissimilarity between Bob and Cliff,
similar?
• DBC = 1 – 2/(2+3) = 0.6

• Dissimilarity between Alan and Cliff,

• DAC = 1 – 0/(0+6) = 1
 OFFICIAL (CLOSED) \ NON-SENSITIVE

Mini-Lecture (3)
Intro to Network Analysis

MRT Network
 OFFICIAL (CLOSED) \ NON-SENSITIVE

Network Definitions
7 E
B
8
5 6

4
A D
3 4
C

Node Arc

A network consists of:

• A set of points - nodes (or vertices) and
• A set of lines connecting certain pairs of the nodes –
arcs (or links, branches, edges).
10
 OFFICIAL (CLOSED) \ NON-SENSITIVE

Network Definitions - Arcs

• Arcs are labeled by naming the
nodes at either end, e.g. AB A
Arc AB
B

• The arcs of a network may have a

flow of some type through them
• If flow through an arc is allowed in
only one direction (e.g. one-way
street), the arc is said to be a A B

directed arc.
• If flow through an arc is allowed in
either direction, the arc is said to be A B
undirected.

11
 OFFICIAL (CLOSED) \ NON-SENSITIVE

Network Definition - Path

• A path between two nodes is a
sequence of distinct arcs B
E

connecting the nodes E.g. path 5 6

4
between nodes A and E, AB →
A D
BD→DE

• A directed path from node i to B

E

node j is a sequence of 5 6
4
connecting arcs whose direction A D
(if any) is towards node j, so that
only the flow from node i to node j
along this path is feasible.

12
 OFFICIAL (CLOSED) \ NON-SENSITIVE

Network Definition - Path

• An undirected path from

E
node i to node j is a B

sequence of connecting 5 6
4
arcs whose direction (if any) A D

can be either toward or

away from node j.

7 E
• A path that begins and ends B

at the same node is called a 6

cycle. D
4

13
 OFFICIAL (CLOSED) \ NON-SENSITIVE

Network Definition
• Two nodes that are connected by an arc are
called adjacent nodes.
• Two nodes are said to be connected if the
network contains at least one undirected path
between them.
• A connected network is a network where every
pair of nodes is connected.

14
 OFFICIAL (CLOSED) \ NON-SENSITIVE

Components of Typical Networks

Nodes Arcs Flow
Intersections Roads Vehicles
Airports Air lanes Aircraft
Switching points Wires, Channels Messages
Pumping stations Pipes Fluids
Work Centres Materials-handling route Jobs

Network optimization problems include:

• Transportation
• Minimum cost flow OP1
• Minimum spanning tree
• Shortest path
• Maximum flow OP2
• Routing and scheduling
• Critical path PM 15
 OFFICIAL (CLOSED) \ NON-SENSITIVE

Mini-Lecture (4)
Minimum Spanning Tree
Prim’s algorithm
Kruskal’s algorithm
 OFFICIAL (CLOSED) \ NON-SENSITIVE

Spanning Tree
B
7 E A spanning tree is a tree
5 6
8
composed of all the nodes
A D
4
and some of the arcs in the
3 C 4
network. Two nodes are
B
7 E connected by exactly one
5 6
8 path, no cycles are present
A D
4 in a spanning tree.
3 C 4

Examples of spanning tree (in blue)

7 E
B
8
5 6
4
A D
3 C 4

Not a spanning tree, has cycle BDE

17
 OFFICIAL (CLOSED) \ NON-SENSITIVE

Minimum Spanning Tree

• The Minimum Spanning Tree (MST) problem
is to connect all nodes in a network so that the
total branch lengths are minimized.
• There are two algorithms commonly used,
Prim’s algorithm and Kruskal’s algorithm, both
of which are greedy algorithms.

18
 OFFICIAL (CLOSED) \ NON-SENSITIVE

Recall: Minimum Spanning Tree Model

 Spanning tree – a connected network with no loops allowed

 Minimum spanning tree – nodes are connected such that the
total length of the arcs are minimized
• Two greedy algorithms commonly used – Prim’s algorithm and
Kruskal’s algorithm. Both will give optimal solutions.

19
 OFFICIAL (CLOSED) \ NON-SENSITIVE

Prim’s Algorithm
Step 1: Select any starting node
Step 2: Select the node closest to the starting
node to join the spanning tree
Step 3: Select the closest node not presently
in the spanning tree
Step 4: Repeat Step 3 until all nodes have
joined the spanning tree

20
 OFFICIAL (CLOSED) \ NON-SENSITIVE

Prim’s Algorithm - Example

Network Description Solution

Set

Node A is selected as a starting node A

7 E
for convenience
B
8
5 6
4
A D
3 C 4

Node C is added to the solution set as AC (3)

7 E
it is the closest to the starting node,
B node A.
8
5 6
4 Fringe nodes are nodes not
A D connected to nodes in the solution
3 C 4
set.

21
 OFFICIAL (CLOSED) \ NON-SENSITIVE

Prim’s Algorithm - Example

Network Description Solution

Set

Node D is added to the solution set AC (3)

7 E as it is closest to nodes in the solution
B CD (4)
8 set (A and C).
5 6
4
A D
3 C 4

Node E is added to the solution set as AC (3)

7 it is closest to nodes in the solution CD (4)
E
B set (A, C, D) DE (4)
8
5 6
4 Arc CE becomes redundant as node
A D C is connected to node E via path CD
3 C 4
-> DE

22
 OFFICIAL (CLOSED) \ NON-SENSITIVE

Prim’s Algorithm - Example

Network Description Solution

Set

Node B is added to the solution set as AC (3)

7
B
E it is closest to nodes in the solution CD (4)
8 set (A, C, D, E). DE (4)
5 6
4 AB (5)
A D Since all the nodes have joined the
3 C 4 solution set, stop the algorithm.

The minimum spanning tree is

highlighted in red.

Minimum length:
3(AC)+4(CD)+4(DE)+5(AB) = 16

23
 OFFICIAL (CLOSED) \ NON-SENSITIVE

Kruskal’s Algorithm
• Kruskal’s algorithm is to find the set of arcs
that results in the minimum cost.
• At each stage, add the shortest arc that
connects two nodes that are not originally
connected into the solution set.

24
 OFFICIAL (CLOSED) \ NON-SENSITIVE

Kruskal’s Algorithm - Example

Network Description Solution Set

Shortest arc AC is added into the solution AC (3)

7 set.
E
B
8
5 6
4
A D
3 C 4

CD and DE are the shortest arcs, with AC (3)

7 lengths 4, and arc DE has been arbitrarily DE (4)
E
B chosen to be added to the solution set.
8
5 6
4
A D
3 C 4

25
 OFFICIAL (CLOSED) \ NON-SENSITIVE

Kruskal’s Algorithm - Example

Network Description Solution Set

Shortest arc CD is added to the solution set. AC (3)

7 E DE (4)
B
8
Arc CE is marked as a redundant arc as it CD (4)
5 6 would form a cycle CD -> DE -> EC
4
A D
3 C 4

Shortest arc AB is added to solution set. AC (3)

7 DE (4)
E
B Minimum spanning tree is found. Compare CD (4)
8 this tree with the tree found using Prim’s
5 6 AB (5)
4 algorithm.
A D
3 C 4

Minimum length:
3(AC)+4(DE)+4(CD)+5(AB) = 16 26
 OFFICIAL (CLOSED) \ NON-SENSITIVE

Learning Outcomes
• Define a network using nodes and arcs.
• Identify a minimum spanning tree problem and its
applications.
• Apply Prim’s and Kruskal’s algorithms to solve typical
minimum spanning tree problems.
• Define dissimilarity index in grouping of people or
objects.
• Apply dissimilarity index and minimum spanning tree
in grouping of people or objects.

27
 OFFICIAL (CLOSED) \ NON-SENSITIVE

Overview of E210 Operation Planning Module

Decision
Operations LP – Integer Making using
Planning Programming Decision Tree
Overview
LP -
Transportation Network
Model Diagram and
Decision
Making with Min. Spanning
Dependent Tree
Decisions

LP formulation LP – Binary
Process Strategy and EXCEL Integer
& Capacity Solver with Programming Decision
Requirements Sensitivity Making using
analysis Utility Dissimilarity
Function Index and MST

Linear
Programming Decision
(LP) Graphical Making under
Method uncertainty
and risk

End of
OP! 28