You are on page 1of 37

PERT

What does PERT stand for?

Program Evaluation
E al ation and Review
Re ie Technique
Techniq e
How and why did it come about?
• For some projects time is critical but cost
is not
– Specifically
p y for R&D p
projects
j in defence,,
space related research etc
• Many of these projects are very unique in
nature
– There is no prior basis for estimating activity
duration
• Many of these projects are also extremely
complex
– Conventional
Con entional PM tools do not fit the bill
Events leading up to the
development of PERT

• Russia launches the


Sputnik Space Missile
in 1957, before the
Americans

• The Americans are


embarrassed
Events leading up to the
development of PERT
• The Americans
retaliate by planning
to launch the Polaris
Ballistic Missile

• They cannot be
embarrassed further
by delaying the
launch
Events leading up to the
development of PERT
• Expected durations of activities were
g y unknown
largely
• Project Planning and Control was initially
lacking on the Polaris project
• Actual costs and project durations in the
initial stages exceed estimates by as much
as 50%
Events leading up to the
development of PERT
• U
United
it d St
States
t Department
D t t off Defense
D f hires
hi B
Booz
Allen Hamilton, to develop a Project
Management Technique for this project
• Booz Allen Hamilton invents the PERT
technique in 1958
1958.
What is PERT?
• Technique to assist in project planning
when historical cost and time data are not
available or reliable

• Usually associated with R&D and new


product development projects where the
uncertainty is great
What is PERT?
• PERT statistically treats uncertainty
– Estimates the likelyy duration time
– Estimates the likelihood of meeting certain
milestones
• Rarely used in construction
– Can be used on projects where quantity
and productivity uncertainties exist
Comparing PERT and CPM
PERT CPM
• Assumes variability in • Assumes fixed (or
productivity
d ti it rates
t and d very mildly
ildl varying)
i )
therefore variability in activity durations
activity times
• Milestone or event • Duration or activity
focused f
focusedd
• Probabilistic • Deterministic
calculations C l l ti
Calculations
How PERT works
• Si
Since activity
ti it titimes are uncertain,
t i PERT uses
three time estimates for an activity
– An
A optimistic
ti i ti d
duration
ti ((assumes hihigh
h productivity)
d ti it ) - To
– A pessimistic duration (low productivity) - Tp
• Not based on disasters etc
• Based more on late material delivery, equipment
breakdowns, bad weather etc
– A “most
“ t likely”
lik l ” d
duration
ti - Tm
• Not the average of the Optimistic and Pessimistic durations
• It is the amount of time the activity will require most of the
time, if repeated many times
Probability Basics
• Concrete Placement Data for Columns in a
Building

0 1 1 4 6 9 12 16 10 4 1 0 0 0
Frequency
36 38 40 42 44 46 48 50 52 54 56 58 60 62
Minutes to
Place
Histogram
Distribution of Concrete Placement Time

20

15
ncy
Frequen

10

0
36 38 40 42 44 46 48 50 52 54 56 58 60 62
Minutes to Place
Continuous Distribution
Distribution of Concrete Placement Time

20

15
ency
Freque

10

0
36 38 40 42 44 46 48 50 52 54 56 58 60 62
Minutes to Place
Normal Distributions
Normal Distribution
• Al
Also called
ll d th
the “b
“bellll curve””
• Approximates many distributions we see in real
life and on construction sites
• It can be defined by a mean (μ) and a standard
deviation (σ)
• The normal probability density function is
characterized by a unit area, with a mean of 0
and a variance of 1

• F(x; μ, σ2) =

–∞<x<∞; σ2 is called the “variance”


Unpacking the normal distribution
• Mean μ – average of all the values in the
distribution
– For a probability density function μ is the
mean median and mode
mean,
• Standard Deviation σ – a measure of the
di
dispersion
i off the
th ddata
t

σ=
Standard Deviation

• Approximately 68% of values lie within 1 σ from


the mean
pp o ate y 95% o
• Approximately of values
a ues lie
e within
t 2 σ from
o
the mean
pp
• Approximately y 99.7%
% of values lie within 3 σ
from the mean
Means and Standard Deviations
• μ1 = μ2 = μ3

• μ 4 < μ1
1

4 • σ1 < σ2 < σ3

2 • σ 4 > σ1; < σ2,


3 although μ1 = μ2
The Central Limit Theorem
• Completion times for each activity could follow
a variety of distributions
• The sum of these distributions (i.e the
overall project
j time)) will tend to follow a
normal distribution
• So we assume that
1. The mean of the sum is the sum of the means
2 The variance of the sum is the sum of the variances
2.
3. The distribution of the sum is a normal distribution
regardless
g of the shape
p of individual distributions
Cumulative Distribution Function
• Gives the probability that a random variable X is
less than a given value x
Using the cumulative distribution
function
1 C
1. Convertt the
th distribution
di t ib ti iinto
t a standard
t d d normall
distribution (μ=0 and σ =1)
2 For
2. F a random d variable
i bl X with
ith a mean off μ and
d
a standard deviation of σ, the standardized
random variable is

3. Use the normal distribution table to find out the


probability of an event
Calculating the probability of an
event
• Approject
j is normally
y distributed and has a
mean duration of 4 days and a standard
deviation of 2.

1. What is the p
probability
y that the p
project
j will be
completed in 7 days?

2. In 3 days?
Completing within 7 days

• μ 4, σ =2
μ=4 2, X(variable)=7
X(variable) 7
• Z = (7-4)/2 = 1.5
• F(1 5) = 0.933
F(1.5) 0 933
• P(completing within 7 days) = 93.3%
Completing within 3 days
• μ=4,
4 σ =22
• X(variable)=3
( )
• Z = (3-4)/2 = - 0.5
• F( z) = 1-F(z)
F(-z) 1 F(z)
• F(-0.5) = 1-F(0.5)
• F(0.5) = 1-0.691 = 0.309
• P(completing within 3 days) = 30.9%
30 9%
Activity Properties in PERT
• Activities follow a Beta distribution
– Characterized byy modal and extreme values
Activity properties contd..
contd
• Empirically it is accepted that
• Expected duration of an activity is
– Te = (To + 4Tm + Tp)/6
• Standard
S deviation ffor the activity is
– σ = ((Tp - To))/6
• Variance is
– υ = σ2
The PERT algorithm
1 F
1. For eachh activity,
ti it calculate
l l t ththe expected
t d
duration, standard deviation and variance
2 Use
2. U th the expectedt ddduration
ti as ththe b
basis
i ffor
performing a forward and backward pass to
obtain critical path
3. Sum the variances to find the variance of the
entire project
4. Use the cumulative normal distribution table to
determine probability of event or project
completion
Use in construction
• Can be used instead of CPM when
productivity
p y is unknown or variable

• Ti
Time d
data
t is
i often
ft collected
ll t d bbased
d on
experience

• Can be used
sed to predict the likelihood of
various events and milestones being met
An example

D F
A

C
Start Finish

B E G
Data table
Activity To Tm Tp
A 2 4 6
B 3 5 9
C 4 5 7
D 4 6 10
E 4 5 7
F 3 4 8
G 3 5 8

Find expected project duration.


What is the likelihood that the project will finish within 20 days?
Step 1 - Calculate expected
duration and variances
Activity To Tm Tp Te σ υ
A 2 4 6 4 0 66
0.66 0 44
0.44
B 3 5 9 5.33 1 1
C 4 5 7 5.17 0.5 0.25
D 4 6 10 6 33
6.33 1 1
E 4 5 7 5.17 0.5 0.25
F 3 4 8 4.5 0.83 0.69
G 3 5 8 5 17
5.17 0 83
0.83 0 69
0.69
Step 2 - Calculate Critical Path
0 0 0 0 0 0

A4 D 6.33 F 4.5
0 0 0 0 0 0

0 0 0 0 0 0

Start C 5.17 Finish


0 0 0 0 0 0

0 0 0 0 0 0

B 5.33 E 5.17 G 5.17


0 0 0 0 0 0
Critical Path Calculations
0 4 4 10.3 10.3 14.8

A4 D 6.33 F 4.5
0 0 0 8.66 4.66 15 15 4.66 19.5

0 0 4 9.17 19.5 19.5

Start C 5.17 Finish


0 0 4 0 9.17 19.5 0 19.5

0 5.33 9.17 14.3 14.3 19.5

B 5.33 E 5.17 G 5.17


3.83 3.83 9.17 9.17 0 14.3 14.3 0 19.5
Step 3 - Cumulative Progress
Activity To Tm Tp Te (ES) σ υ
A 2 4 6 4 0 66
0.66 0 44
0.44
B 3 5 9 5.33 1 1
C 4 5 7 9.17 0.5 0.69
D 4 6 10 10 33
10.33 1 1 44
1.44
E 4 5 7 14.34 0.5 0.94
F 3 4 8 14.83 0.83 2.13
G 3 5 8 19 51
19.51 0 83
0.83 1 63
1.63
Probability of finishing in 20
days
• μ=19.51,
19 51 σ2 = 1.63
1 63 ((critical
iti l path
th variance)
i )

• X(variable)=20

• Z = (20-19.51)/1.27 = 0.38

• F(0.38) = 0.648

• P(completing within 20 days) = 64.8%


Thank You

You might also like