You are on page 1of 4

New England Root Beer Distributors

I. Introduction
Operations research deals with decision problems. It is the representation of real-world systems
by mathematical models together with the use of quantitative methods (algorithms), with a
view towards optimizing. There are three phases of an operations research project. They are as
follows:

A. Pre-modelling phase
1. Problem Identification
Diagnosis of the problem from its symptoms, and delineation of the problem.

2. Quantification of the Problem


For an identified problem, establish objectives, understand constraints and requirements, and
quantify them.

B. Modelling Phase
1. Formulation of a Mathematical Model
Any mathematical model has:
a. An objective function such as maximization o profit or minimization of cost.
b. Decision variables for optimization.
c. Constraints such as availability of man-hours, availability of machine hours, availability of
funds, demand etc.
d. Parameters such as time taken to paint a chair, number of units produced per person, machine
hours required per one unit of a product.

2. Data Collection
Data is necessary to generate a solution from the model. Data can be collected from various
sources such as historical data, research, time study or subjective estimate.

3. Algorithm Validation and Generating Model Solution


Algorithm validation involves running the algorithm for the model on the computer in order to
ensure that:
a. The input data is free from errors
b. The computer program is bug-free
c. The computer program correctly represents the model we are attempting to validate.

4. Sensitivity Analysis and Model Validation


a. In practice, a “solution” often involves many solutions under varying assumptions to
establish sensitivity. For ex. If we vary the input data, how would it affect the values of the
decision variables?
b. The speed of processing by computer systems.
c. After performing sensitivity analysis, choose the most likely scenario and validate the model
by conducting hypothesis test to ensure that the expected gains are significantly more than
current gains.

C. Post Modelling Phase


This phase may involve data interpretation and decision making as well as implementation,
monitoring and control.

1
New England Root Beer Distributors

1. Data Interpretation and Decision Making


Based on sensitivity analysis and model validation results, the interpretation is done and the
decision is taken. For ex., in case of a product mix decision, the decision may be taken to
produce say 50 numbers of Style I shirts and 100 numbers of Style II shirts.

2. Implement, Monitor and Control


a. Implementation of the model involves initiating execution as per decision. With minor
variation in available resource, if there is significant variation in profitability, then we can say
that the model is very sensitive.
b. For a sensitive model, we should monitor very closely from time to time as well the best
available manpower should be deployed for such work. Also planned versus actual
performance should be reported on a timely basis, in order to initiate corrective action, if
deviations are reported.
c. With major variation in available resource, if there is insignificant variation in profitability,
we can say that the model is very robust. For a robust model, very close monitoring and
reporting is not required. You can deploy fresh trainees as well as lesser experience persons
for such work.

II. Business Problem


For the purposes of our project, we have chosen the warehousing case of New England Root
Beer Distributors (NERBD), which is discussed as follows:

NERD serves the highly profitable quality root beer soda market and currently sells their
product in twelve cities across New England namely Boston, Brattleboro, Concord, Hartford,
Manchester, Nashua, New Haven, New London, Portsmouth, Providence, Springfield,
Worcester. Out of those 12 cities they have decided to have 5 candidate distribution centers
(DCs) namely Boston, Nashua, Providence, Springfield & Worcester. Capacity of each
distribution centre is 2000.

ID City Name Dj
BO Boston 425
BR Brattleboro 12
CO Concord 43
HA Hartford 125
MN Manchester 110
NA Nashua 86
New
NH 129
Haven
New
NL 28
London
PO Portsmouth 66
PR Providence 320
SP Springfield 220
WO Worcester 182
Table 1. Demand in each city

2
New England Root Beer Distributors

Distance matrix
Boston Nashua Providence Springfield Worcester
(miles) Dij

ID City BO NA PR SP WO

BO Boston 0 37 42 82 34

BR Brattleboro 93 65 106 59 68

CO Concord 69 33 105 101 72

HA Hartford 98 103 73 27 66

MN Manchester 55 20 92 93 60

NA Nashua 37 0 72 79 41

NH New Haven 128 137 94 63 98

NL New London 95 113 57 57 71

PO Portsmouth 62 48 104 127 85

PR Providence 42 72 0 68 38

SP Springfield 82 79 68 0 47

WO Worcester 34 41 38 47 0
Table 2. Distribution matrix for each warehouse

Assume, transportation cost is 1 $/mile and fixed cost for opening DC is $10,000.

Questions
Q1. Where should NERD set up its distribution centres to serve all these cities in order to
minimize expected transportation costs?
Q2. Find the right balance that improves profitability and customer satisfaction.
Q3. Find the optimal number of DCs to get the optimum trade off.
Q4. Do sensitivity analysis with no of DCs.

III. Solution
A. Pre-modelling Phase
The objective here is to minimise transportation problems by finding the right number of
warehouses. There is a trade-off between profit and transportation cost. The problem can be
quantified as:
Let us formulate the above problem as a mathematical optimization model. Consider n
customers i = 1, 2, 3 ….n and m distribution centers j = 1, 2, 3 ….m. Define continuous
variables as the amount serviced from facility j to demand point i, and binary variables yj = 1
if a facility is established at location j, yj = 0 otherwise. An integer-optimization model for the
capacitated facility location problem can now be specified as follows:

3
New England Root Beer Distributors

Minimize: ∑𝑚 n m
𝑗=1 𝑓𝑗 𝑦𝑗 + ∑i=1 ∑j=1 cij xij

Subject to: ∑𝑚𝑗=1 𝑓𝑖𝑗 for i = 1….n


𝑛
∑𝑖=1 𝑥𝑖𝑗 ≤ 𝑀𝑗 𝑦𝑗 for j = 1….m
𝑥𝑖𝑗 ≤ 𝑑𝑖 𝑦𝑗 for i = 1….n; for j = 1….m
𝑥𝑖𝑗 ≥ 0 for i = 1….n; for j = 1….m
𝑦𝑗 ∈ {0,1} for j = 1….m

The objective of the problem is to minimize the sum of facility activation costs and
transportation costs. The first constraints require that each customer’s demand must be
satisfied. The capacity of each facility j is limited by the second constraints: if facility j is
activated, its capacity restriction is observed; if it is not activated, the demand satisfied by j is
zero. Third constraints provide variable upper bounds; even though they are redundant, they
yield a much tighter linear programming relaxation than the equivalent, weaker formulation
without them.

B. Modelling Phase
The problem has a mathematical model in the form of: Minimize: ∑𝑚 n m
𝑗=1 𝑓𝑗 𝑦𝑗 + ∑i=1 ∑j=1 cij xij .
We used Excel Solver in order to find out the number of DCs that should be opened. From the
solution it is clear that DCs should be established at Boston, Providence and Springfield. The
total cost would be kept down to $66,285. We have attached the Excel Solver Optimization
File with the business case.

C. Post-Modelling Phase
The post modelling phase would involve decision making and implementation, monitoring and
control by NERBD.

IV. References
1. Phatak, Girish G. Operations Research (2016), Pune, Techmax Publications.

You might also like