Case Study

The Cakes’ shop Problem

An owner of a bakery shop would like to determine how many 10-inch birthday
cakes he should produce each day in order to maximize his profit. His present
method of determining the quantity to bake is based on his best guess. For
example if he estimates the daily demand will be five cakes, then five cakes will
be produced for the day’s operation. Since it is more economical to process all the
cakes in one batch, all five cakes would be produced early in the morning.

The production costs are $2.00 per cake. And the profit for each cake is
$2.5. However, If over estimates the daily demand, some cakes will be left over at
the end of the day. The policy is to sell all leftover cakes to a local store that
specializes in day-old items. He is currently receiving $1.50 per cake for the
surplus cakes, thus incurring a loss of $0.50 per cake.

Unfortunately, daily decision of how many cakes to produce is difficult because

the daily demand is uncertain. The owner has experienced some days when there
was no demand, and yet one day he experienced a demand of eight cakes of the
birthday cakes.

Ideally, he wanted a recommendation on the day production quantity for the

birthday cakes that would maximize his profit over the long run.

What he needs is a way to carry out his trial-and-error procedure without actually
performing the experiments.

Actually, computer simulation can be used to experiment with production sizes in

just this manner. In the simulation technique we will need to develop a
mathematical model and a computer program that describe or recreate the
bakery’s shop operation for the sale of white birthday cakes. In the simulation
model we will choose a trial production quantity, simulate a number of daily
demand, and then compute the resulting total profit. Then by choosing other
production quantities, we can continue this trial-and –error procedure until we
find what appears to be best profit maximizing production quantity. While this
trial-and –error procedure will not guarantee an optimal solution, the simulated
“best” solution should be close to the optimal solution and thus a very good
production-quantity decision.

Model development:

Let

x = number of 10-inch birthday cakes produced

d =daily demand for 10-inch birthday cakes

z =daily profit associated with producing x 10-ich birthday cakes

When writing a mathematical model to describe how profits are related to the
production quantity x and d , we need only consider two separate case:

1) The production quantity is less than or equal to the daily demand

2) The production quantity is greater than the daily demand.

Case 1: The production quantity is less than or equal to demand

If x<= d, z = 2.5x

Case 1: The production quantity is greater than the demand

If x > d z = 2.5d + (x-d) (-0.5)

Z = 3.00 d- 0.5 x

To develop a general model that are appropriate for similar problems, we must
introduce some additional notation.

Let

p = selling price for each cake

c = cost of each unit

s = day-old price
Case 1: The production quantity is less than or equal to demand

If x<= d, z = (p-c) x

Case 1: The production quantity is greater than the demand

If x > d z = (p-c) d + (x-d) (s-c)

Z = (p-s) d – (s-c) x

Generating daily demand

Since we want the model to be a good representation of the real situation, it is

important that the generated daily demand be a good representation of the acual
daily demands that exist for the birthday cakes.

Assume that we have available data showing the daily demand during the past
month (10 days of operation) This history of daily demand is shown in the table
below
frequency _ of _ observation
Where relative _ frequency =
total _ number _ of _ observations

Daily demand d Frequency of Days observed Relative frequency

0 1 0.05
1 2 0.1
2 1 0.05
3 2 0.1
4 3 0.15
5 6 0.3
6 3 0.15
7 1 0.05
8 1 0.05
Total 1

If we believe that this relative frequency distribution is representative of the

future pattern of daily demand, we can use it as the basis for generating
hypothetical daily demand in our simulation model. This is a critical part of any
simulation study and needs serious consideration by both the analyst and the
decision maker. If the demand distribution is not an accurate representation of
the actual demand the simulation results will be of little value. In general decision
models based on inaccurate or non representative inputs will not provide useful
output information for the decision maker. We must now develop a method of
simulating daily demand d=0, d=1,d=2,…etc. Our method should be in the long
run generate a demand of o units approximately 5% of the time, a demand of 1
unit approximately 10% of the time, and so on. First we explain that will a simple
method, where you need nothing more than a paper and a pencil.

Manual Simulation:

Take a sheet of paper and cut it into twenty equal pieces. Following the historical
daily demand frequency in the table, write the number zero on one piece. On two
of the remaining pieces write the number one, which stands for the demand of
one unit. Then on one piece write the number two, on two pieces write the
number three… and so on.

Check the numbers you have written carefully, because this “deck” of twenty
pieces will be used in our simulation of the process. Note that your deck has 5%
0’s. 10%1’s, 5% 2’S, and so on. And represents the historical relative frequency
distribution of the above table Now shuffle your deck of twenty pieces of paper
so as to thoroughly mix up the numbers. Since the slips are all the same size, if we
select one slip of paper at random, we will have simulated drawing a specific daily
demand from the distribution shown in the table.

Let us now see how we can use this “deck” in the simulation of the shop
operation. The first step is the selection of the production quantity, Assume (x=3).
The second step in the simulation process is to select a “hypothetical” daily
production quantity. Now use the deck of twenty slips of paper to generate a
demand by selecting one slip of paper at random. Suppose the first slip drawn has
a 5 written on it. We shall then use a demand of 5 cakes for the first simulated
day of bakery shop operation. This level of demand corresponds to an
underproduction of 2 cakes. Since x<d, we can computer our first day’s profit
using the expression 2.5x=25(3)=$7.5. i.e. Total profit of $7.5.
We generate a hypothetical demand for the second day of our simulation by
returning the previously selected slip of paper to the deck, reshuffling all twenty
pieces of paper, and then drawing another slip at random. And compute the
profit z of the second day then add it to the profit of the first day to upgrade the
total profit.

And so on. Here is an example of 10-day simulation results for a production

quantity of x=3
Day Generated demand Daily profit Total profit
1 5 7.5 7.5
2 1 1.5 9
3 6 7.5 16.5
4 3 7.5 24
5 4 7.5 31.5
6 4 7.5 39
7 3 7.5 46.5
8 0 -1.5 45
9 5 7.5 52.5
10 6 7.5 60

Now Consider changing the production rate. Using the same set of hypothetical
daily demand for x=1,2,3,4,5,6,7,8.

Here is an example of a Ten-Day Simulation Results for the x’s

Production Size Ten Day Simulated profit $
1 25
2 44
3 60
4 79
5 90
6 93
7 91
8 89

From the table it is clear that the best production quantity that maximizes the
profit is at x=6. The results are based on only 10-day simulation.

A much longer simulation period is required to develop confidence in our best

production quantity conclusion