Capacity Planning

(Break-Even Analysis)
Business Decision
Making under
Uncertainty
 Capacity Planning

Capacity is the upper limit on the load that an operating unit

can handle (measured as maximum production or throughput rate)

Measuring Capacity

Break-Even Analysis
 Measuring Capacity
Design Capacity: Maximum obtainable output rate
under ideal conditions.

Effective Capacity: Maximum output rate given work

breaks, product mix, scheduling difficulties, equipment
maintenance, delays, employee breaks, and other
realities (i.e. standard output rate).
 Measuring Capacity
Efficiency: The ratio of actual output rate to effective
capacity

Utilization (Of a unit of a resource during a period) : Uptime

(time machine is in actual operation) divided by
available time.
Capacity Consideration: Bottleneck in a
ti o n
n o p era h o se
eck : A on s w
o tt le n
o p e ra ti t o f th e
B e of n th a
e n c t h a
sequ y is lower
it
capac perations.
o
other
 Break-Even Analysis
Break-even analysis focuses on the relationship
between costs, revenue, and volume.
TC  FC  VC
VC  Q  v
TR  R  Q
P  TR  TC  R  Q  FC  v  Q 
P  Q ( R  v )  FC
P  FC
Q 
Rv
FC
QBEP 
Rv
Break-Even Analysis Schematic
Break-Even Analysis: Example 1
Break-Even Analysis: Example 1
Break-Even Analysis: Example 1
Break-Even Analysis: Example 1a
 Break-Even Analysis in Dollars
Decision makers might be interested in knowing the
break-even point in dollars rather than units
FC
QBEP 
( R  v)

QBEP $  RxQBEP  R ( FC ) /( R  v)  FC /(1  v / R )

FC  Total Fixed Cost

when  1 type of output use : R  Total Revenue
FC v  variable cost per unit
QBEP $ 
sum[(1  vi / Ri )Wi ] QBEP $  break even volume in dollars
Wi  proportion of revenue due to product i (index)
 Break-Even Analysis in Dollars: Example 1d
A manager at a bakery wants to use new equipment to make Quiche. Quiche will have a
variable cost of $3 and will sell for $10 each. Making pies, alternatively, cost $2 per pie
and sells at $7 each. The cost of leasing a special equipment for Quiche/pie is $6000 per
month. Calculate the break-point in dollars?
Nothing is said about proportion of
Quiche/pie, so we assume 50/50

= $8,484.85
 Business Decision Making under
Uncertainty: Intuitive Quick Approach Doesn’t Work
What is suggested first to the decision maker influences the decision.

Most people choose to do nothing (i.e., they prefer the status quo).

There is a tendency to waste good money after bad decisions in order to

avoid admitting to having made a bad decision.
 Business Decision Making under
Uncertainty: Intuitive Quick Approach Doesn’t Work
People are risk averse when a random variable (gamble) is posed in
terms of gains,
but risk seeking when it is posed in terms of avoiding a loss, even though
the two gambles are equivalent.

People are overconfident in estimating the range of a random variable

(i.e., they choose a small estimation range).
Also, they base their estimate on a dramatic unusual past event that they
recall.
 Business Decision Making under
Uncertainty
A set of possible future conditions (of a
States of random variable) which the decision maker
Nature has no control, but will have a bearing on
the results of the decision

Decision
Elements Alternatives A list of alternatives for the manager to
choose from

Payoffs
A known payoff for each alternative under
each possible future condition
Business Decision Making under Uncertainty

Influence Diagrams Sensitivity Analysis

Decision Tables/Trees
Modeling
Preferences/Utilities
Value of information
 Influence Diagrams
Graphical represent of a decision problem. Answers the question:
Should we introduce a new product, and at what price?
New Product Decision
Example
Expected Monetary Value: (EMV)
EMV (sometimes referred to as EV) = sum of (payoff for an alternative *
probability of occurrence of each possible state of nature)

Risk Neutral decision makers

Average amount that would be achieved if

a large number of identical decisions were
to be made (Long-run average payoff)
 EMV Payoff: Example 2
Possible Future Demand
Alternative Low Moderate High
(p=0.3) (p=0.5) (p=0.2)
Small 10* 10 10
Facility
Medium 7 12 12
Facility
Large (4) 2 16
Facility

Which would you choose?

 EMV Payoff: Example 2
Possible Future Demand
Alternative Low Moderate High Multiply
(p=0.3) (p=0.5) (p=0.2) probability by
Small Facility 10 10 10 payoff (i.e. possible
Medium Facility 7 12 12 future demand)
Large Facility (4) 2 16

EVSMALL  0.3  10   0.5  10   0.2  10   10

EVMEDIUM  0.3  7   0.5  12   0.2  12   10.5
EVLARGE  0.3   4  0.5  2   0.2  16   3
 Decision Trees
Tree-like graphical representation of the decisions, random
variables and their probabilities, and the payoffs

States of
Alternatives Decision
nature
ure a re
s o f nat
State ted with
ia
assoc ilities
b
proba
 Decision Trees

ti ves
e rna
a l t
are
c hes
B ran

F o r e ac h
d ur e: a t w ill
proc e ve t h
a c k rn a ti
f o l ding b e the alte ) p a y off
The n , c ho os ex pe cted
o (
decisi e greatest
e l d t h al c ulating
yi dure: C
c e
b a c k pro ree
e fo l ding e c is i on t
Th n the d
’s i
EMV
 Decision Trees: Folding Back Procedure
Very For each decision, choose the alternative that will
important to yield the greatest (expected) payoff
know the
difference
between For each chance node, calculate the EV of the payoffs
decision (can for all its states of nature
control) and
chance event
(cannot Continue folding back until the best alternative for all
control) decisions is determined.
 Decision Trees: Example 3
A manager must decide the size of a video arcade to construct. The manager has narrowed
the choices to two: small or large. Information has been collected on payoffs and the
following decision tree has been constructed. Determine which alternative (build small or
build large) should be chosen in order to maximize the expected value

For a small facility with high demand,

there are two choices: do nothing, or
expand.

Start from the Because expand has higher payoff ($55 > $40), you
very right of the would choose it.
tree
For a large facility with low demand,
there are two choices: do nothing, or
reduce prices.

Because reduce prices has higher payoff ($50 >

$10), you would choose it.
 Decision Trees: Example 3
A manager must decide the size of a video arcade to construct. The manager has narrowed
the choices to two: small or large. Information has been collected on payoffs and the
following decision tree has been constructed. Determine which alternative (build small or
build large) should be chosen in order to maximize the expected value

Eliminate the far branches: Replace

the squares with actual payoffs (we’ve
gotten in previous slide)

Calculate EVM for both initial

branches: Build Small and Build Big
 Decision Trees: Example 3
A manager must decide the size of a video arcade to construct. The manager has narrowed
the choices to two: small or large. Information has been collected on payoffs and the
following decision tree has been constructed. Determine which alternative (build small or
build large) should be chosen in order to maximize the expected value

Build Small EV: 0.4*40 + 0.6*55 = $49

Build Big EV: 0.4*50 + 0.6*70 = $62
 Decision Trees Steps
Draw the decision tree

Assign probabilities to the states of nature

Estimate payoffs for each possible combination of decision alternatives

& states of nature.

Working from right to left,

calculate EMV for each state of nature (circular node).

Choose alternative with highest EMV (or lowest expected cost) for
each decision point (square node)
 Value of Information
In certain situations, it is possible to buy information about
the states of nature of a random variable (e.g. performing
market survey)

Expected value of perfect information (EVPI):

difference between the expected payoff with perfect information
minus expected payoff without the information.

_
 Value of Information
Expected value of perfect information (EVPI):

Expected Value Under Certainty (perfect Information) =

best payoff in each state * probability
 Value of Information: Example 4
Given the data and EV result from Example 1: Calculate the EVPI ?

Possible Future Demand

Alternative Low Moderate High
(p=0.3) (p=0.5) (p=0.2)
Small Facility 10 10 10
Medium Facility 7 12 12
Large Facility (4) 2 16
Value of Information: Example 4
t h e best EV = 10.5 (from example 1; expected payoff under risk)
Pick for each Possible Future Demand
Alternative payoffLow ility Moderate High
o b a b
pr
(p=0.3) (p=0.5) (p=0.2)
Small Facility 10 10 10
Medium Facility 7 12 12
Large Facility (4) 2 16
Expected Value Under Certainty = best payoff in each state x probability
= .30($10) + .50($12) + .20($16) = $12.2

Expected value of Expected payoff Expected payoff

 
perfect informatio n under certainty under risk

EVPI  12.2  10.5  1.7

 Value of Sample Information
Expected value of sample information (EVSI):
The difference between expected payoff with sample
(imperfect) information and expected payoff without sample
information.
Bayes' Rule is used to update the (prior) probability
distribution of the states of nature of the random
variable
 Value of Sample Information
Bayes’ Rule for determining conditional probability

The probability, wherein you condition on the data, is posterior

probability (conditional probabilities)
In decision trees, it’s all the conditional probability at the end
of the branches.
Value of Sample Information: Example 5a
A manager is considering random drug testing of her employees. However,
the test is not 100 percent accurate. The conditional probabilities are as
follows: If a person uses drugs, then the test will be positive 93 percent of
the time, whereas if he does not, the test will be negative 97 percent of the
time. Suppose that 5 percent of the employees are drug users. What is the
probability that a drug user is correctly identified after drug testing?

Let B1 = The event that an employee is a drug user, B2 = The event that
he is not a drug user.

A+ = The event that the test result is positive.

 Value of Sample Information: Example 5a
A manager is considering random drug testing of her employees. However, the test is not 100
percent accurate. The conditional probabilities are as follows: If a person uses drugs, then the
test will be positive 93 percent of the time, whereas if he does not, the test will be negative 97
percent of the time. Suppose that 5 percent of the employees are drug users. What is the
probability that a drug user is correctly identified after drug testing?

So they are given us P(A+ | B1) = 0.93 But we want P(B1 | A+)

B1 = The event that an employee is a drug user, B2 = The event that he is not a drug user.

They also given us P(A-| B2) = 0.97 A+ = The event that the test result is positive.

We have P(B1) = 0.05 and hence P(B2) = 1 − 0.05 = 0.95.

We have P(A+ | B1) = 0.93

 Value of Sample Information: Example 5a
A manager is considering random drug testing of her employees. However, the test is not 100
We want P(B1 | A+)
percent accurate. The conditional probabilities are as follows: If a person uses drugs, then the
test will be positive 93 percent of the time, whereas if he does not, the test will be negative 97
percent of the time. Suppose that 5 percent of the employees are drug users. What is the
probability that a drug user is correctly identified after drug testing?
B1 = The event that an employee is a drug user, B2 = The event that he is not a drug user.

A+ = The event that the test result is positive

A+) = )*P(B1)/[()*P(B1))+ ()*P(B2))]

P(A-| B2) = 0.97
This is the probability of A+ not (A-) given the
P(A+ | B2) = 1 − 0.97 = 0.03 employee is not a drug user : P(A- | B2) ; we want
P(A+ | B2)
 Value of Sample Information: Example 5a
Let B1 = The event that an employee is a drug user, B2 = The event that he is not a drug user.
P(A+ | B1) = 0.93 P(A+ | B2) = 0.03 Let A+ = The event that the test result is positive
A manager is considering random drug testing of her employees. However, the test is not 100 percent accurate.
The conditional probabilities are as follows: If a person uses drugs, then the test will be positive 93 percent of
the time, whereas if he does not, the test will be negative 97 percent of the time. Suppose that 5 percent of the
employees are drug users. What is the probability that a drug user is correctly identified after drug testing?

P(B1) = 0.05 P(B2) = 0.95

So, the probability that a drug user is identified is probability given a positive result a
drug user is identified: P ( B1|A+ )

A+) = )*P(B1)/[()*P(B1))+ ()*P(B2))]

P ( B1|A+ ) = ( 0.93 ( 0.05 ) ) / ( 0.93 ( 0.05 ) + 0.03 ( 0.95 ) ) = 0.62;

 Value of Sample Information: Example 5b
Let B1 = The event that an employee is a drug user, B2 = The event that he is not a drug user.
P ( B1|A+ ) = 0.62 Let A+ = The event that the test result is positive

If the cost of not identifying (and dismissing) a drug user (i.e., a false
negative test result) is $1,000, cost of falsely accusing a non-user (i.e.,
a false positive test result) is $200, and other costs are zero, what is
the maximum the manager should be willing to pay for a drug test
(expected value of sample information)?
 Value of Sample Information: Example 5b
If the cost of not identifying (and
dismissing) a drug user (i.e., a false
negative test result) is $1,000, cost of
e r 0 .6 2 0
No charge if test
falsely accusing a non-user (i.e., a false Us correctly identifies
user
positive test result) is $200, and other
costs are zero, what is the maximum Non-U
the manager should be willing to pay se r 0 .3 8
for a drug test (expected value of os itive Test incorrectly
P -200 identifies a non-user
sample information)?
as user: $200 cost

Neg Test does not

t ative -1000
Tes User identify a user:
$1000 cost

Non-U
se r
No charge if a non-
No 5 0 user is correctly
Tes
t 0 .0 -1000 No Test, so user is
not identified: identified as non-
r
Use $1000 cost user

Non
-Use
r
0 .9
5
 Using Table User
0 .6 2 0
Value of Sample 0 .0 7
5 Non-Use
r
iti ve 0 .3 8
s
Information: Po -200
Example 5b
N eg
Tes
t ative
User -1000

Non-U
se r
0 . 05 -1000 0
No
Tes
t User
P(B2) = 0.95 P(A+ | B1) = 0.93
Non
-Use
r 0 .9 5
P(B1) = 0.05 P(A+ | B2) = 1 − 0.97 = 0.03

Condition Prior Conditional Joint Posterior

User 0.05 0.93 = 0.05*0.93 = 0.0465 0.0465/0.075= 0.62

Positive Non-User 0.95 0.03 = 0.95*0.03 = 0.0285 0.0285/0.075= 0.38

Sum of Joint = 0.0465 + 0.0285 = 0.075

 Value of Sample 0 .6 2
Us e r 0
Information: Example 5b 0 .0 7
5
Non-U 0 .3 8
s iti ve se r
Po -200
Neg
ative
t 0 .9 2 0 .0 0 3 8 -1000
Tes 5 Us e r
Non-Us
e r 0 .9 9 6
2
0
No 0.05 -1000
Tes
t User
Non
-Use 0 .9 5
r

Condition Prior Conditional Joint Posterior

User 0.05 0.07 = 0.05*0.07 = 0.0035 0.0035/0.925 = 0.0038

Negative Non-User 0.95 0.97 = 0.95*0.97 = 0.9215 0.9215/0.925 = 0.9962

Sum of Joint = 0.0035 + 0.9215 = 0.925

 Value of .6 2
= - 76 Us e r 0 0
Sample ( - 200 )
5
.38 0 .0 7 Non-U
Information +
) 0 tive se r 0 .3 8
2 ( 0 Posi
0.6 -200
: Example 5b 0.996 Neg
ative
t 2(0) 0 .9 2 0 .0 0 3 8 -1000
Tes = - 3. + 0.0 5 Us e r
8 038
(-100 Non-Us
e r 0 .9 9 6
0) 2
-1000 0
No
Tes
t User

Non Perform the Folding

-Use
r Back method
Condition Prior Conditional Joint Posterior
User 0.05 0.93 = 0.05*0.93 = 0.0465 0.0465/0.075= 0.62

Positive Non-User 0.95 0.03 = 0.95*0.03 = 0.0285 0.0285/0.075= 0.38

Sum of Joint = 0.0465 + 0.0285 = 0.075

 Value of Sample
Information: Example 5b
5
0 .0 7 -76
s iti ve
Po

N eg 0.075(-76) + 0.925 (-3.8) = -9.2150

ative
t 0 .9 2
Tes 5
-3.8

No 0 .0 5 -1000
Tes User
t
Non
-Use
r 0 .9 5 0.95(0) + 0.05 (-1000) = -50
 Value of Sample
Information: Example 5b

0.075(-76) + 0.925 (-3.8) = -9.2150

t
Tes

No
Tes
t
0.95(0) + 0.05 (-1000) = -50

EVSI = Difference between Expected payoff with sample (imperfect) information

& expected payoff without sample information.

The maximum the manager should pay for the drug test is 50 –
9.215 = $40.79
 Sensitivity Analysis
Since payoffs and probabilities are estimated values.
Decision maker might want to know how sensitive the choice or payoff
of the best alternative is to changes in one or more of these values.
 Decision Trees: Example 6
Determine the course of action that has the highest expected value for the following
decision tree. Values in parentheses are negative.
 Decision Trees: Example 6
Determine the course of action that has the highest expected value for the following decision tree.
 Decision Trees: Example 6
Determine the course of action that has the highest expected value for the following decision tree.
Decision Trees with Posterior Prob. : Example 7

Find the posterior probabilities: Probabilities at tree ends

Hiring consultant is not
a choice (consultant
already hired) row th
G

Dec
lin e
iti ve
Pos

Find the
N eg
ativ
e G row th posterior
probabilities:
Probabilities at
Dec
Only states of natures in lin e tree ends
this tree
Decision Trees with Posterior Prob. : Example 7
Decision Trees with Posterior Prob. : Example 7

To get joint prob., multiply the

prior and conditional prob.

Sum joint probabilities for

each state of nature.
Decision Trees with Posterior Prob. : Example 7

Calculate Posterior
Probabilities by dividing the
joint probability by the sum of
the joint.
Decision Trees with Posterior Prob. : Example 7

0.59
row th
G

Decline
e
Positiv
0.41

Nega 0.25
tive row th
G

Declin
e
0.75
More Exercises
 Example A
Princely Prizes is considering 3 different models for its new suit. This will involve changes
in capacity and thus different volumes will have different unit costs. They estimate that
the average demand per week will range between 80 and 120. They choose 3 possible
levels of weekly demand and calculate profits for each. Given the data below, draw a
decision tree, calculate the EMV, and calculate the EVPI?

Possible Average Weekly Demand

80 (p=.4) 100 (p=.2) 120 (p=.4)

Model William $10,000 $15,000 $14,000

Model Harry $ 8,000 $18,000 $12,000
Model Charles $ 6,000 $16,000 $21,000
Example A : Draw decision tree
c h e s from
Bran n odes
l a r
circu nt
se
repre e demand
l
possib
of
0 0 , o r 120
80, 1 eek
w
suits/
 Example A: Calculate EV

Choose
alternative
with highest
expected
payoff

cv
 Example A: Calculate EVPI
EV = 14,000

EVPI = .4(10,000) + .2(18,000) + .4(21,000) - 14,000 = $2,000

 Example B
The director of social services of a province has learned that a new act has mandated
additional information requirements. This will place an additional burden on the agency.
The director has identified three alternatives to handle the increased workload: (1)
reassign present staff, (2) hire and train two new workers, and (3) redesign current
practice so that workers can readily collect the information with little additional effort. An
unknown factor is the caseload for the coming year when the new data will be collected
on a trial basis. The estimated costs for various options and caseloads are shown in the
following table:
The director has decided that reasonable caseload
probabilities are 0.10 for moderate, 0.30 for high, and
0.60 for very high.
a. Which alternative will yield minimum expected cost?
b. Construct a decision tree for this problem and solve it.
c. Determine the expected value of perfect information.
 Example B; part a
The director has decided that reasonable caseload probabilities are 0.10 for
moderate, 0.30 for high, and 0.60 for very high.
a. Which alternative will yield minimum expected cost?

Reassign: .10(50) + .30(60) + .60(85) = $74

New Staff:.10(60) + .30(60) + .60(60) = 60*
[minimum]

Redesign: .10(40) + .30(50) + .60(90) = 73

 Example B; part b
The director has decided that reasonable caseload probabilities are 0.10 for
moderate, 0.30 for high, and 0.60 for very high.
b. Draw a tree diagram (place all the numbers calculated in part A)
 Example B; part c
The director has decided that reasonable caseload probabilities are 0.10 for
moderate, 0.30 for high, and 0.60 for very high.
c. Determine the expected value of perfect information.
EV = 60

EVPI = Expected Cost with perfect info.– Expected Cost without perfect info.
= 60 – [.10(40) + .30(50) + .60(60)] = 60 – 55 = 5 or $5,000
 Example C
A diagnostic test of a certain disease has 95 percent probability that the test will be positive
for a patient with the disease and 96 percent probability that the test will be negative for a
healthy person. Only 1 percent of the population has the disease in question. If the diagnostic
test reports that a person chosen at random tests positive, what is the conditional probability
that the person does, in fact, have the disease?
P(+test | disease) = 0.95 P(+test | disease) = 0.96 P(disease) = 0.01

We need to use Bayes’ rule

P(disease | +test) = (P(+test | disease)P(disease))/(P(+test |disease)P(disease) +
P(+test | no disease)P(no disease))

= 0.1935
 Example E
A law firm is representing a company that is being sued by a customer. The decision is
whether to go through litigation or settle the case out of court. The cost of losing the case is
estimated to be $1 million, whereas the cost of settling the case is $200,000. Winning the case
in court would result in no loss. What is the minimum chance of winning for which the
company should contest the case?
1-p p

Let p = prob. of winning

Expected cost of litigation (in millions) = 1(1-p) + 0p = 1 – p

Expected cost of settling (in millions) = 0.2(1-p) + 0.2p = 0.2

 Example E
A law firm is representing a company that is being sued by a customer. The decision is
whether to go through litigation or settle the case out of court. The cost of losing the case is
estimated to be $1 million, whereas the cost of settling the case is $200,000. Winning the case
in court would result in no loss. What is the minimum chance of winning for which the
company should contest the case?
Expected cost of litigation (in millions) = 1 – p
Expected cost of settling (in millions) = 0.2

We are looking for when the cost of litigation is less than settlement
ne q uality
be r the i de by
m i
Reme en you div
1-p < 0.2 p > 0.8 h
flips w ve numbe
ati
r
a neg

Company should litigate only if probability of winning in court is

greater than 0.8
 Example E
A law firm is representing a company that is being sued by a customer. The decision is
whether to go through litigation or settle the case out of court. The cost of losing the case is
estimated to be $1 million, whereas the cost of settling the case is $200,000. Winning the case
in court would result in no loss. What is the minimum chance of winning for which the
company should contest the case?
Expected cost of litigation (in millions) = 1 – p
Expected cost of settling (in millions) = 0.2

We are looking for when the cost of litigation is less than settlement
ne q uality
be r the i de by
m i
Reme en you div
1-p < 0.2 p > 0.8 h
flips w ve numbe
ati
r
a neg

Company should litigate only if probability of winning in court is

greater than 0.8
 Example F
A team of five students found that the proliferation of a new technology, called DPC12, would
improve the chances of their product success. The worst case, which highly unlikely, scenario
is that the product does not sell well (i.e. failure). The team is thinking about delaying the
launch of the product to reduce this risk. If the team waits six months, the probability of
DPC12 proliferating is 0.44. The failure rate is 0.01%. If DPC12 doesn’t proliferate, the
probability of competitors beating them to the market is 0.06 in that case and the failure rate
would be 0.64%. If the competitor doesn’t beat them to the market by 6 months, the failure
rate is 0.09%. If they launch right away, the probability of DPC12 proliferation is 0.44 while the
failure rate is 0.04%. The failure rate when DPC12 proliferation is not present is 0.09 percent.
Draw the decision tree and determine the best course of action in this case.

Draw the tree diagram

 Example F
Draw the decision tree and determine the best course of action in this case.

Expected value of node 3 = .06(.64) + .94(.09) = .123%

 Example F
Draw the decision tree and determine the best course of action in this case.

Expected value of node 3 = .06(.64) + .94(.09) = 0.123%

Expected value of node 2 =.44(0) + .56(.123) = .0689%
Example F

Expected value of node 3 = .06(.64) + .94(.09) = 0.123%

Expected value of node 2 =.44(0) + .56(.123) = 0.0689%
Expected value of node 4 = .44(.04) + .56(.09) = 0.068%
Example F

Expected value of node 3 = .06(.64) + .94(.09) = 0.123%

Expected value of node 2 =.44(0.01) + .56(.123) = 0.0733%
Expected value of node 4 = .44(.04) + .56(.09) = 0.068%
Because .068% < .0733%, choose to launch and not wait
 Example G-a
An offshore underwater structure has tested positive for oil by two exploratory wells drilled
near the crest of the structure. Management is now trying to determine whether it is wise to
make a definite commitment to begin building a platform, and, if so, what size, or to defer the
decision to allow for the drilling of an additional exploratory well. The principal uncertainty is
the size of the field. If the structure is nearly filled with oil, the area of the field will be about
15 square miles and it would require installation of a large platform costing $50 million. If the
structure is only partially filled with oil, the area will be much smaller and the field could be
depleted with fewer wells and a smaller platform costing $30 million. Based on the subjective
judgment of the company’s geologists, the likelihood of the structure being completely filled is
0.4 (i.e., field is large). If a small platform is built and the field turns out to be large, a second
small platform will have to be built at the same cost. If another exploratory well is drilled
along the flanks of the structure, there is a 10 percent chance that the result will be
erroneous. However, management will base its platform size decision on the result of the
exploratory well. A) The exploratory well will cost $2 million. Calculate the posterior
probabilities given that the explanatory well indicates a small field?
Example G-a
 Example G-b
An offshore underwater structure has tested positive for oil by two exploratory wells drilled
near the crest of the structure. Management is now trying to determine whether it is wise to
make a definite commitment to begin building a platform, and, if so, what size, or to defer the
decision to allow for the drilling of an additional exploratory well. The principal uncertainty is
the size of the field. If the structure is nearly filled with oil, the area of the field will be about
15 square miles and it would require installation of a large platform costing $50 million. If the
structure is only partially filled with oil, the area will be much smaller and the field could be
depleted with fewer wells and a smaller platform costing $30 million. Based on the subjective
judgment of the company’s geologists, the likelihood of the structure being completely filled is
0.4 (i.e., field is large). If a small platform is built and the field turns out to be large, a second
small platform will have to be built at the same cost. If another exploratory well is drilled
along the flanks of the structure, there is a 10 percent chance that the result will be
erroneous. However, management will base its platform size decision on the result of the
exploratory well. A) The exploratory well will cost $2 million. Calculate the posterior
probabilities given that the explanatory well indicates a small field? B) Which decision has the
lowest expected cost? Draw the decision tree.
Example G-b