EFIM20005: Management Science

Video Lecture Part A: LP IV Network Models -

The Transportation Problem

Dr Bangdong Zhi
Lecturer in Operations Management and Analytics
Objectives of Today’s Lecture

By the end of this lecture, you should be able to:

• Formulate models for the Transportation problem

• Use Excel and Solver to find the solution to such models (second video)
Network Models

Transportation Models
• Minimize total cost of directly shipping supplies from source nodes to destination nodes

Capacitated Transshipment Model

• Minimize total cost of shipping supplies from source nodes to destination nodes via some

intermediary nodes

Assignment Model
• Assign workers to jobs at minimum cost
Flow

• When two nodes are connected by an arc, a flow can occur directly between them.

• Sometimes the flow is restricted by a maximum capacity.

Transportation Networks

Transportation problems involve the distribution of goods and

services from several limited supply locations to satisfy several
demand locations in the most cost-effective manner.
Model Requirements

• There are 𝒎 sources

➢ The supply of a resource at source 𝑖 is 𝑆𝑖 .

• There are 𝒏 destinations

➢ The demand for the resource at destination 𝑗 is 𝐷𝑗

• The unit shipping cost between nodes 𝒊 and 𝒋 is 𝑪𝒊,𝒋

Goal: Minimize the total shipping cost of supplying the destinations with the
required demand from the available supplies at the sources.
Assumptions of Model

• The per unit shipping cost remains constant regardless of the number of units

shipped.

• All the shipping from the sources to the destinations occurs simultaneously (or

within some fixed time frame).

• The product can be shipped only between sources and destinations.

• The total requirement at the destinations is equal to the total supply from the

sources.
Example Transportation Problem

A Company has three manufacturing plants producing its product based at London, Paris and Berlin. Its
four regional distribution centers are based in Brussels, Athens, Rome and Madrid. Production
capacities at each manufacturing plant, demand requirements of each distribution center and
transportation costs are given in the tables:

Find the transportation schedule that minimises total transportation costs.

Network Representation

Demand
Supply
LP Formulation
Solution
Answer Report:
Sensitivity Report:
Reduced Costs

Reduced Costs:

➢As this is a minimization problem, the reduced costs tell us how much
transportation costs would have to be reduced before that route would be
considered for inclusion in the solution.

➢For example, the reduced cost of 4 for Berlin to Athens means that the cost of
transportation would have to drop below 5 − 4 = 1 before it would become a
used route.
Shadow Prices

• For supply constraints shadow prices give the saving that would be

achieved if an extra unit could be supplied from this source.

• For demand constraints shadow prices give the saving that would be

achieved if one less unit was demanded by this destination.

Allowable Increase / Decrease

• For the Coefficients:

➢Give the range within which the cost values may change and the solution will
not alter.

• For the Constraints:

➢Give the range of change that could be made in supply totals and demand
requirements for which the shadow price information is valid.
Problem Variation: Total Supply Not Equal to Total
Demand

• In reality, total supply does not equal total demand in most cases.

• If total supply exceeds total demand, then no modification is necessary as

excess supply will appear as slack in the LP problem.

• Slack for any particular source can be interpreted as unused supply not

transported from that origin.

Problem Variation: Total Supply Not Equal to Total
Demand

• If total supply < total demand, then the LP will not have a solution.

• We modify the formulation to include a dummy source with a supply equal to the difference

between the total demand and total supply.

• We assign a cost of zero to unit transportation costs via this route as in reality no transportation

will take place.

• When the optimal solution is found, destinations receiving supply from the dummy source will

in practice be experiencing a shortfall or unsatisfied demand.

Example
Balancing Supply and Demand
Problem Variation: Problem Variation: Maximization
Objective Function

• It is possible that the Transportation Problem will not be expressed in terms of

costs per unit transported, but as profit per unit transported.

• In this case we simply maximize the objective function.

Problem Variation: Route Capacities

• Suppose the amount that can be transported via a route is limited maybe due

to limited space available on the mode of transportation.

➢for example, source 3 to destination 1 has a capacity of 1000

• This would be dealt with by adding the constraint:

𝑋3,1 ≤ 1000
Problem Variation: Route Minima

• Suppose there is a minimum requirement that must be transported via a

specific route, maybe due to contract or minimum amounts that can be

transported by particular modes of transport.

➢for example, demand 2 has a contract with source 3 to supply at least 1500

• This would be dealt with by adding the constraint:

𝑋3,2 ≥ 1500
Problem Variation: Unacceptable Routes

• Establishing a route from every source to every destination may not be possible.

• To model this, we:

➢drop the corresponding arc from the network diagram

➢remove the corresponding variable from the LP: 𝑋𝑖,𝑗 = 0.

 Sheet1 (3) Sheet1 (2) Answer Report 1 Sensitivity Report 1 Sheet1
European Transportation Problem

Unit Cost Data

To
Brussels Athens Rome Madrid
London 3 2 7 6
From Paris 7 5 2 3
Berlin 2 5 4 5

Shipments
To
Brussels Athens Rome Madrid Total Shipped <= Capacity
London 3500 1500 0 0 5000 <= 5000
From Paris 0 2500 2000 1500 6000 <= 6000
Berlin 2500 0 0 0 2500 <= 2500
Total Received 6000 4000 2000 1500 13500
= = = = = =
Demand 6000 4000 2000 1500

Minimum cost 39500

EFIM20005: Management Science

Live Lecture: Transshipment and Assignment Models

Dr Bangdong Zhi
Lecturer in Operations Management and Analytics
Objectives of Today’s Lecture

By the end of this lecture, you should be able to:

• Formulate models for transshipment and assignment models

• Use Excel and Solver to find the solution to such models

Capacitated Transshipment Networks

• Sometimes shipments take place by first transporting goods through one or more

transshipment nodes before reaching their final destination.

• Additionally, an upper limit may be placed on the amount of flow on the arcs in the

network.

• Such a problem is called the Capacitated Transshipment or General Network

Model.
Capacitated Transshipment Problem

• The constraints of the problem are expressed in terms of the Net Flow Out (NFO)

of a node.

• NFO = Flow Out - Flow In

• The constraints require:

➢ For each supply node, the NFO must not exceed supply.

➢ For each transshipment node, the NFO must be 0.

➢ For each demand node, the NFO must equal the negative of the demand (𝑁𝑒𝑡 𝐹𝑙𝑜𝑤 𝐼𝑛𝑡𝑜 𝑎 𝑛𝑜𝑑𝑒 =
− 𝐷𝑒𝑚𝑎𝑛𝑑).
Example
• Depot Max has 6 stores located in the UK. It is Saturday night and stores in London and Bath

have found themselves running low on the Model 65A Arcadia workstation that will be
advertised in the Sunday papers. They request 12 and 13 workstations, respectively, be shipped
to them to cover the anticipated increase in demand from the ad.

• Management has identified stores in Manchester and Newcastle to be able to supply 17 and 10

workstations respectively. However, the stores can only utilise space available on delivery
trucks currently scheduled to transport other supplies between stores.

• The unit shipping costs and the maximum number of workstations that can be shipped between

stores is shown in the Network. Depot Max wishes to transport the available workstations from
Manchester and Newcastle to London and Bath at minimum total cost.
Network
LP Formulation
LP Formulation (cont.)
Assignment Networks

• The Assignment Problem addresses the task of allocating individuals to tasks so as

to make the most effective or cost-efficient assignment.

• It is a special case of the Transportation Problem where each source supplies 1 unit

(worker) and each demand point requires 1 unit (job).

Features of Assignment Problem

• M workers are to be assigned to M jobs

• A unit cost (or profit) 𝐶𝑖𝑗 is associated with worker 𝑖 performing job 𝑗.

• Goal: Minimize the total cost (or maximize the total profit) of assigning workers to

jobs so that each worker is assigned one job and each job is performed.
Example

A consultancy firm has three consultants available and three clients to whom they
are to be assigned. Each consultant possesses different skills and would therefore be
more efficient at different jobs. The table below gives cost per hour of each
consultant if they were assigned to each job. Find the allocation of consultants to
jobs that minimizes total cost
Example (cont.)

• Represent the problem as a network

• Formulate the problem as a LP

• Find the optimal allocation

Network
LP Formulation
Excel Solution
Problem Variation: Number of Workers Not Equal to
Number of Jobs

• If the number of workers exceeds the number of jobs, we introduce dummy jobs

(with zero costs) to make the number of workers and jobs equal.
➢ Workers allocated to dummy jobs will not, in reality, be employed

• If the number of jobs exceeds the number of workers, we introduce dummy

workers (with zero costs) to make the number of workers and jobs equal.
➢ Jobs allocated to dummy workers will not, in reality, be performed
Problem Variation: Maximization Objective Function

• Assignment problems may be stated as maximization problems.

• For example, suppose a department store has four locations unassigned and is

considering five departments that might occupy these locations. If estimates are
available of expected profitability if each department were to be situated in each
location, then we have an assignment problem to maximize profit.
Example

Problem: Formulate as
an Assignment Problem
Solution
Department Constraint
Location Constraint
Problem Variation: Unacceptable Assignments

• It is possible that certain workers cannot be assigned to certain jobs.

• If this is the case, we remove this variable from the LP with the extra constraint

𝑋𝑖𝑗 = 0.
 Management Science Week 10 – Questions on Network Models

Question 1

Klein Chemicals Inc produces a special oil-base material that is currently in short
supply. Four of Klein’s customers have already placed orders that together exceed the
combined capacity of Klein’s two plants. Klein’s management faces the problem of
deciding how many units it should supply to each customer. Since the four customers
are in different industries, different prices can be charged because of the various
industry pricing structures. However, slightly different production costs at the two
plants and varying transportation costs between the plants and customers make a “sell
to the highest bidder” strategy unacceptable. After considering prices, production
costs and transportation costs, Klein has established the following profit per unit for
each plant-customer alternative.

Customer
Plant D1 D2 D3 D4
Clifton Springs £32 £34 £32 £40
Danville £34 £30 £28 £38

The plant capacities and customer orders are as follows:

Plant Capacity (units) Distributor Orders (units)

Clifton Springs: 5000 D1: 2000
Danville: 3000 D2: 5000
D3: 3000
D4: 2000

Formulate a network and linear programming model to represent the problem of

finding the schedule that maximises profits. Set up and solve your model in a
spreadsheet.

Question 2

Scott and Associates Inc is an accounting firm that has three new clients. Project
leaders will be assigned to the three clients. Based on the different backgrounds and
experiences of the leaders, the various leader-client assignments differ in terms of
project completion times. The possible assignments and the estimated completion
times in days are

Client
Project Leader 1 2 3
Jackson 10 16 32
Ellis 14 22 40
Smith 22 24 34

a Develop a network representation of this problem.

b Formulate the problem as a linear program, and solve. What is the total time
required?

Question 3

The distribution system for the Herman Company consists of three plants, two
warehouses and four customers. Plant capacities and shipping costs (in £) from each
plant to each warehouse are

Warehouse
Plant 1 2 Capacity
1 4 7 450
2 8 5 600
3 5 6 380

Customer demand and shipping costs per unit (in £) from each warehouse to each
customer are

Customers
Warehouse 1 2 3 4
1 6 4 8 4
2 3 6 7 7
Demand 300 300 300 400

a Develop a network formulation of this problem

b Formulate a linear programming model of the problem
c Find the optimal shipping plan
 Management Science – Week 10 Solutions

Question 1

Let Xij = amount transported from plant i to customer j (i=1,2,3,; j=1,2,3,4)

(Note supplier 3 is a dummy supplying the excess demand of 4000)

Min C = 32X11 + 34X12 + 32X13 + 40X14 + 34X21 + 30X22 + 28X23 + 38X24

s.t.
X11 + X12 + X13 + X14 ≤ 5000
X21 + X22 + X23 + X24 ≤ 3000
X31 + X32 + X33 + X34 ≤ 4000
X11 + X21 + X31 = 2000
X12 + X22 + X32 = 5000
X13 + X23 + X33 = 3000
X14 + X24 + X34 = 2000
(Clifton supply)
(Danville supply)
(Dummy supply)
(Customer 1 demand)
(Customer 2 demand)
(Customer 3 demand)
(Customer 4 demand)

and Xij ≥ 0 (i=1,2,3; j=1,2,3,4)

2000

5000 D1
32
Clifton 34 5000
32
3000 40 D3
Demand
34
Supply

30
Danville
28 3000
38 D3
4000 0 0
0 2000
Dummy
0
D4
Question 2

Let Xij = 1 if project leader i is assigned to client j (i=1,2,3; j=1,2,3); 0 otherwise

Min Time = 10X11 + 16X12 + 32X13 + 14X21 + 22X22 + 40X23 + 22X31 + 24X32 +
34X33

s.t. X11 + X12 + X13 = 1 project leaders assigned

X21 + X22 + X23 = 1 to one job
X31 + X32 + X33 = 1

X11 + X21 + X31 = 1 clients assigned to

X12 + X22 + X32 = 1 one project leader
X13 + X23 + X33 = 1

and Xij binary i=1,2,3; j=1,2,3

1 1
10
Jackson Client 1
16
1 32
1

Demand
14
Supply

Ellis 22 Client 2
40
1 22 1
24
Smith 34 Client 3
Question 3

Let Xij = flow from node i to node j (i=1,2,…,9; j=1,2,…,9)

Min C = 4X14 + 7X15 + 8X24 + 5X25 + 5X34 + 6X35 + 6X46 + 4X47 + 8X48 + 4X49 +
3X56 + 6X57 + 7X58 + 7X59

s.t.
X14 + X15 ≤ 450
X24 + X25 ≤ 600 (supply nodes)
X34 + X35 ≤ 380

X46 + X47 + X48 + X49 – X14 –X24 – X34 = 0 (transshipment

X56 + X57 + X58 + X59 – X15 –X25 – X35 = 0 node)

-X46 – X56 = -300

-X47 – X57 = -300
-X48 – X58 = -300
-X49 – X59 = -400

and Xij ≥ 0 i=1,2,…9; j=1,2,…,9

D6 = 300

6
S1 = 450
D7 = 300
1. 4 T4 = 0 6 7
4
S2 = 600 7 4 8
8 D8 = 300
2.
4
5 6 8

T5 = 0 3
S3 = 380 5 7
D9 = 400
3
6 5
7 9
 Question 1 Question 2 Question 3
Klein Chemicals Inc Problem

Unit Profit Data

To
D1 D2 D3 D4
Clifton 32 34 32 40
From Danville 34 30 28 38
Dummy 0 0 0 0

Shipments
To
D1 D2 D3 D4 Total Shipped Capacity
Clifton 0 4000 0 1000 5000 <= 5000
From Danville 2000 0 0 1000 3000 <= 3000
Dummy 0 1000 3000 0 4000 <= 4000
Total Received 2000 5000 3000 2000 12000
= = = =
Demand 2000 5000 3000 2000

Maximum Profit £282,000.00