Transportation Problems
Transportation Problems
The distinction between them is of no consequence at this level because if a learner understands
and can model one type then he/she should be able to model all the others. We will define the
three but confine our illustration and discussion to Transportation problems only.
Transportation problem: this often involves minimizing the cost of transporting goods from a
set of origins to a set of destinations.
For instance, New KCC has eight major milk processing factories, and consumers are
everywhere in the country and beyond. How much milk should be transported from each of these
factories to each of the major towns in the country? Think about how you would make the
decision if you were a manager in New KCC.
What variables would you consider? Please note your answer here, which you should confirm or
dismiss after the learning.
Assignment problem: the assignment of agents or persons to tasks, assigning jobs to machines,
assigning sales personnel to sales territories, assigning contracts to bidders, among others. One
needs to come up with a set of assignments that will optimize a stated objective. For instance,
under certain constraints, sales costs may be minimized by sending Otieno to Kapenguria and
Kitale and Kyalo to Bomet and Migori, as sales assistants.
1
3.1.2 Objectives
You have now commenced the third of the three chapters prescribed in the course outline.
Chapter One was introductory, and covered the course objectives, various definitions, origins
and nature of OR, Problem solving and decision making, the quantitative analysis process, and
finally OR in practice and the role of OR in decision making
Chapter Two presented an introduction to Linear Programming (LP). It should be noted that LP
is not synonymous with OR, but is just one of the tools used in quantitative analysis in a decision
making process. LP is an OR tool, but OR is not always LP. However, a learner who follows and
understands LP will manage many other tools of OR. In the chapter we covered graphical
solutions, computer solutions and algebraic (Simplex) solutions. We discussed Sensitivity
Analysis (SA) in all the solution approaches. SA is always important in decision making,
especially in agriculture and agribusiness where systems are relatively more risky and are
generally dynamic. Once you make a decision based on certain parameters, what if at least one
of the parameters changes? If you decide to grow and sell beans based on a projected profit of sh
100 per kg., what if the neighbouring countries flood our market and dampen the actual profit?
Take note that “What if” is the essence of SA, and can be applied in any decision making
process, even outside LP.
The Lecture 3.1 will require that you recall how to formulate and solve linear programming
problems
3.1.4 The formulation and solution of a transportation problem
The transportation problem arises frequently in planning for the distribution of goods and
services from several supply locations (origins) to several demand locations (destinations). The
2
objective of a transportation problem is to minimize the cost of transporting goods from the
origins to the destinations. (Caution: watch out for a different objective, for instance
maximization of profit that will be dependent on the set of routes taken).
Let us consider the transportation problem faced by Mboga Transporters Limited, involved in the
transportation of assorted Asian vegetables from three growing areas (Molo, Kibwezi and
Kinangop).
Production capacities for the three areas over the next one month planning period for the
vegetable assortment are as follows:
Mboga Transporters Ltd distributes the vegetables to urban markets in four towns, Kisumu,
Nakuru, Mombasa and Nairobi. The one-month forecast of the demand for the distribution
centres is as follows:
Take note of an unusual phenomenon here, where demand = supply; in reality often this is not
the case. Many times Demand>supply and a few times Demand <supply.
The vegetable producers would like to determine how much should be transported from
each farmland to each market
The objective is to determine the routes to be used and the quantities to be transported via each
route that will provide the minimum total transportation cost associated with meeting all
distribution centre demands.
As an illustration: Question - will all the 6000kg demand in Kisumu come from just one
farmland or from a combination of two or more? Think about it! What should guide this? Which
variables should guide the decision process?
Clue: transport of goods is normally charged on distance and quantity (weight or volume)
For Mboga Ltd., the cost in Kshs. of each kg transported on each route is given below:
3
Destination
Origin
Kisumu Nakuru Mombasa Nairobi
Molo 3 2 7 6
Kibwezi 7 5 2 3
Kinangop 2 5 4 5
How do we solve?
Prepare a graphical representation called a network of the problem
Formulate an LP model of the problem
Solve using any optimization computer software (we will use Lindo)
Please try doing the network where we stopped in class before you read further, then compare
with what is presented below.
The network diagram has errors in two places, please see if you can identify the errors
4
5
The circles are referred to as nodes, representing each origin and each destination
The lines connecting the nodes are arcs, representing each possible transportation route
The vegetables transported from the origins to the destinations represent the flow in the network;
direction of flow is indicated by arrows
The network is complete; we now move to the next step of formulating an LP model
In general, the decision variables for a transportation problem having m origins and n
destinations are written as xij
xij is the number of units or the quantity transported from origin i to destination j, where
i= 1, 2, …, m and j = 1, 2, …, n
Take note: there is one decision variable for each arc. If we find that x24=500, it means
transporting 500 units from origin 2 (Kibwezi) to destination 4 (Nairobi)
The answers to the three bullets are 3, x11, and 3x11 respectively. If you managed to solve before
reaching here then you are doing well.
The objective function here is minimizing total transportation cost; now let us define total
transportation costs from each origin
6
Transportation costs for units from Molo: 3x11 + 2x12 + 7x13 + 6x14
Are you able to do the ones from Kibwezi and from Kinangop? Try then compare with this:
Transportation costs for units from Kibwezi: 7x21 + 5x22 + 2x23 + 3x24
Transportation costs for units from Kinangop: 2x31 + 5x32 + 4x33 + 5x34
Min 3x11 + 2x12 + 7x13 + 6x14 + 7x21 + 5x22 + 2x23 + 3x24 + 2x31 + 5x32 + 4x33 + 5x34
Now we have all the parts of the LP model we were building. We should now put the parts
together to make the model:
7
Min 3x11 + 2x12 + 7x13 + 6x14 + 7x21 + 5x22 + 2x23 + 3x24 + 2x31 + 5x32 + 4x33 + 5x34
St x11 + x12 + x13 + x14 ≤ 5000
x21 + x22 + x23 + x24 ≤ 6000
x31 + x32 + x33 + x34 ≤ 2500
x11 + x21 + x31 = 6000
x12 + x22 + x32 = 4000
x13 + x23 + x33 = 2000
x14 + x24 + x34 = 1500
xij ≥ 0, for i = 1, 2, 3 and j = 1, 2, 3, 4
The LP model is complete. The presentation here where I have made a specific column for each
decision variable is desirable if the solution will be done through Simplex procedure. The
arrangement, however, is not necessary when using computer solution.
Observe:
All the information needed in the LP is in the network
There is one constraint for each node
There is one variable for each arc
The sum of variables from an origin is less than or equal to the supply; you supply only
what you have or less
The sum of variables to a destination is equal to the demand; demand is specific
Computer solution
Now apply what we learnt in class to get the computer solution
I will share here the solution and its interpretation. I will, however, not discuss the Sensitivity
Analysis. Please try it first before checking what is presented here
8
We will skip slack/surplus because there is none (but there are dual prices! What do they
represent?) We skip sensitivity analysis also for now, but you may just explore it as you revise
on your own. Let us now interpret the solution
Question: what is the meaning of value of a variable? How about reduced costs? Refer to your
notes on LP
Problem variations
The case we have discussed is fairly general, with total supply being equal to total demand.
There are, however, possible variations which include the following:
(i) Total supply ≠ total demand (total supply not equal to total demand)
(ii) Maximization objective function rather than minimization
(iii) Route capacities or route minimums
(iv) Unacceptable routes
Solving problems with the respective variations may be easily accommodated with slight
modifications
9
Suppose in the Mboga Ltd problem Kibwezi could supply only 5000kgs instead of 6000, there
would be a shortfall of 1000. In the respective network, a fourth node would be added and named
dummy, with arrows to each of the four demand locations. A zero per unit cost is assigned to
each arc leaving the dummy origin.
The value of the optimal solution will represent the transportation cost for the units actually
transported (there will be NO transportations from the dummy)
The destinations showing transportations being received from the dummy will be experiencing
shortfalls or unsatisfied demand
Subject to:
; i=1, 2, …, m: supply
; j = 1, 2, …, n: demand
11