Scribd
Upload
11 views3 pages

Networksuncertain

The document discusses the use of PERT/CPM methods for project management when activity durations are uncertain, emphasizing the need for probabilistic data and estimates for optimistic, most likely, and pessimistic times. It outlines the calculations for expected project duration and variance, as well as the assumptions necessary for determining probabilities of completion times. Additionally, it presents a decision-making framework for evaluating alternative plans based on expected net benefits and costs, using a case study of a company aiming to accelerate project completion to gain competitive advantage.

Uploaded by

Ngân Võ Trần Tuyết
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOC, PDF, TXT or read online on Scribd
11 views3 pages

Networksuncertain

The document discusses the use of PERT/CPM methods for project management when activity durations are uncertain, emphasizing the need for probabilistic data and estimates for optimistic, most likely, and pessimistic times. It outlines the calculations for expected project duration and variance, as well as the assumptions necessary for determining probabilities of completion times. Additionally, it presents a decision-making framework for evaluating alternative plans based on expected net benefits and costs, using a case study of a company aiming to accelerate project completion to gain competitive advantage.

Uploaded by

Ngân Võ Trần Tuyết
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOC, PDF, TXT or read online on Scribd
You are on page 1/ 3

Project Networks and PERT/CPM

Activities with uncertain durations


If times required to complete activities cannot be predicted with certainty (and this is usually the case for research and development
projects), then predictions of project length (completion time, etc.) will require some use of probability, and the data needed will be
probabilistic data.

The sort of information needed will be the same as in any project network, except that we need information on the distribution of
possible times (not just a single time) required for each activity. The method is used for predicting the critical path and for dealing
with questions about completion time – what is the probability the project can be completed in three months, what is the probability
the required time will be more than two years, what is the most optimistic estimate of the required time, etc.

To use the method, we need three estimates for the duration of each activity [generally obtained from experts in the field]:
most likely time (m) – exactly what it says
optimistic estimate (o) – there is less than a 1% chance that the activity will take less time than this
Pessimistic time (p) – there is less than a 1% chance the activity will take longer.

We assume (based on some theory, some experience) that the distribution of times will follow a beta distribution - somewhat
skewed to the right (there is a decreasing probability of very long times, a high probability of times near the most likely time)
For a beta distribution, we have formulas, based on the optimistic, most likely, and pessimistic times, for the mean (average)
duration and the standard deviation (average deviation) of the durations:
Expected (mean) duration  =
Standard deviation of durations  =
variance of durations 2 =

We can calculate the total duration for the project based on mean times - this gives our "on the average" (expected) duration and
mean critical path. We can also find all optimistic and all pessimistic durations and critical paths (critical paths may well be
different).
It is more realistic to expect that some activities will go quickly, some more slowly, some near the mean - we would like to find
probabilities for different completion times.

Calculating probabilities for completion times.


The calculations are based on three simplifying assumptions:

Assumption 1. The mean critical path will, in fact, be the critical path even with variation in times.[Usually a good approximation -
if other paths become longer, it's usually because they are close to the mean critical path in length]
Assumption 2. The durations are statistically independent - the variations in any two of them are completely unrelated (if one
activity takes longer, that gives no information about whether another activity might take longer).

These two assumptions allow us to calculate the mean (expected) duration for the project p and the variance for completion time
p2:
p = sum of means of the durations on the mean critical path
p2 = sum of the variances of the durations on the mean critical path.

Assumption 3. The possible total times fit a normal or gaussian distribution (this works well if there are a large number of
activities on the critical path and if Assumption 1 holds)

Then for any deadline we care about, say d , we can find the probability that completion time T will be less than or equal to d as

P(T  d) = P(Z  ) [Z is the standard normal variable - tables are available for these probabilities - a short one that will do for
our work is on p.265 of the text]
Computer implementation:
The estimates can be made in Excel –see the "PERT_CPMtemplates.xls" file. [Note: the “NORMDIST(d,, )” command
calculates P(Z  ) (under our assumptions, this is P(completion time ≤ d) and the “NORMINV(a, , )” command calculates the
value T for which P(duration<T) = a if duration is normally distributed with mean  and standard deviation ]

C E
B

G D H
A F

J
I

Klonepalm example for PERT


Time On
Estimates Mean
Activit Slac Critica
y Pred o m p m s2 ES EF LS LF k l Path
12 9 53.7777777
A 76 86 0 0 8 0 90 0 90 0 Y
1 10 11
B A 12 15 18 5 1 90 5 95 0 5
0.11111111 11 11 11
C B 4 5 6 5 1 105 0 0 5 5
2 14 12 14
D G 15 18 33 0 9 129 9 9 9 0 Y
2 17 17 19
E D 18 21 24 1 1 149 0 3 4 24
2 5.44444444 11 11
F A 16 26 30 5 4 90 5 90 5 0 Y
1 12 11 12
G C,F 10 13 22 4 4 115 9 5 9 0 Y
2 1.77777777 17 16 19
H D 24 28 32 8 8 149 7 6 4 17
3 21.7777777 12 11 14
I A 22 27 50 0 8 90 0 9 9 29
4 13.4444444 19 14 19
J D,I 38 43 60 5 4 149 4 9 4 0 Y

E,H, 19 19 19
Finish J 0 0 0 0 0 194 4 4 4 0 Y
Project
Duration 19
Data = 4
Results

Mean Critical
Path
m= 194
85.6666666
s2 = 7
9.25562891
s= 8
P(T<=d) = 0.06519153
where
d= 180
20

If we want
P(T<=C)
to be 0.99
215.531812
then C= 7

Balancing alternative plans


Decision-,making is based on consideration of the net benefit/net cost of the alternatives – basically calculating costs/benefits (using
our computer model) for each of the alternatives, and comparing.

Most often, we look at the Expected value criterion – comparing the net expected value (expected profit minus expected cost) of
the alternatives. Other criteria are used, but we will concentrate on this one.

For example (see p. 286ff in text):


Klone has competitors in the market for palm computers. If they can bring out their new model in less than 180 days, they will beat
both competitors and gain an additional $1 million in profit. If they can bring out the model in 180 to 200 days, they will beat one
competitor, but not the other, and gain an additional $400,000 in profit. The present expected time for completion is 194 days, but
there is a possibility of bringing the project in under 180 days, and a chance of finishing in 180 to 200 days.
The chances could be improved by speeding up activities – but this would cost money, and there would still be uncertainty:
By spending $200,000 (salaries, travel, overtime, etc.) Klone could reduce the optimistic, most likely and pessimistic times for sales
training (Activity H) to 19, 21 and 23 days, respectively.
By spending $250,000 (same sorts of expenses), Klone could reduce the times for technical training (Activity F) to 12, 14, 16 days,
respectively.
Should they spend either or both of these amounts?

They have four alternatives:


A..) Stay with the current plan
B.) spend only the $200,000 on sales training
C.) Spend only the $250,000 on technical training
D.) Spend both amounts and accelerate both training processes.

To make a decision, we consider the expected value of net additional profit under each of the four scenarios.
Expected additional profit = (amount 1) x (probability of amount 1) + (amount 2) x (probability of amount 2) + ...
and Net expected additional profit = expected additional profit – expected additional costs

You might also like

pFad - Phonifier reborn

Pfad - The Proxy pFad of © 2024 Garber Painting. All rights reserved.

Note: This service is not intended for secure transactions such as banking, social media, email, or purchasing. Use at your own risk. We assume no liability whatsoever for broken pages.


Alternative Proxies:

Alternative Proxy

pFad Proxy

pFad v3 Proxy

pFad v4 Proxy