OPERATIONS RESEARCH

Chapter 7
Network Analysis

Prof. Bibhas C. Giri

Department of Mathematics
Jadavpur University
Kolkata, India
Email: [email protected]
 MODULE - 3: Program Evaluation and
Review Technique (PERT)

3.1 Program Evaluation and Review Technique (PERT)

PERT was developed to handle projects where the time duration for each activity
is not known with certainty but is a random variable which is characterized by β-
distribution. Three estimates for each activity are required to calculate the expected
completion time of the project:

1. Optimistic time (to ) - The shortest possible time in which an activity can be
performed assuming that everything goes well.

2. Pessimistic time (tp ) - The longest possible time required to perform an activ-
ity under extremely bad conditions. However, such conditions do not include
natural calamities like earthquake, flood, etc.

Fig. 3.1:

3. Most likely time (tm ) - The time that would occur most often to complete an

2
 activity, if the activity is repeated many times under the same conditions. Obvi-
ously, it is the completion time that would occur most frequently.

4. Expected time (te ) - It is the average time an activity takes if it is repeated a large
number of times. The expected time is given by the formula
to + 4tm + tp
te =
6
and the variance for the activity is given by the formula
( )2
2
tp − to
σ = .
6

The main difference in PERT calculation is that instead of activity duration, expected
time te for the activity is considered. With each node, variance is associated. Thus, the
duration of the project is the mean expected time with variance.

3.1.1 Rules for finding variance of event

(i) Set the variance for initial event is zero, i.e., V1 = 0

(ii) Vj , the variance for succeeding event j is obtained by adding activity’s variance
to the variance of predecessor event except at merge points, i.e., Vj = Vi + σij2

(iii) At merge points, the variance is computed along longest (critical) path. In the
case of two paths having the same length, the larger of the two variances is chosen
as the variance for that event.

Here, it is assumed that two activities are independent of each other and hence their
variances can be added.

For example, Suppose that a project’s expected duration is 19 days and variance of
this path is 50/36. If the exact probability distribution of the path is known, it would
have been easy to find out the probability of completing the project in a given time.
Using Central Limit Theorem, the probability distribution of time for each event can
be considered as normal. This assumption greatly simplifies calculations. Assuming
normality, the probability of the project being completed by a certain date can be eval-
uated easily.
Prob[P roject duration 20 days]

[ ]
D − µ 20 − µ
= P rob[D ≤ 20] = P rob ≤
σ σ
where µ is the mean of the distribution and σ is the standard deviation.
Now, the value (D − µ)/σ is normalized value and usually written as z. Then
[ ]
20 − 19
P rob z ≤ √ = P rob [z ≤ 85] = 0.80 (from standard Normal distribution table.)
50/36
The probability of finishing the job in less than or equal to 20 days is 0.80. The
physical meaning of this statement is as follows. If this job is done hundred times
under same conditions, then there will be 80 occasions when this job will have taken
20 days or less to complete it.
One of the main advantages of PERT approach to the management of a large scale
project is binding for contractual dates to finish the project. A designer would like to
know the duration of the project that will have 95% chance of being completed. Let
Ts be the scheduled duration such that

P rob[t ≤ Ts ] = 0.95
[ ]
t − µ Ts − µ
or, P rob ≤ = 0.95
σ σ
[ ]
Ts − µ
or, P rob Z ≤ = 0.95
σ
From standard normal table, we have z0.95 = 1.64. Therefore,
Ts − µ 7.07
= 1.64 gives Ts = 19 + × 1.64 = 20.90 = 21 days
σ 6

3.1.2 PERT Algorithm

Step 1 For a PERT network, denote the most likely time by tm , the optimistic time to
and pessimistic time by tp .

Step 2 Use β-distribution for the activity duration. Compute the expected time te for
each activity by using the formula te = (tp + 4tm + to )/6.

Step 3 Calculate the earliest start time and the latest finish time, and the total float as
the difference of these two.

Step 4 Identify the critical activities and the critical path, and the expected date of
completion of the project.

Step 5 Use the values of tp and to and compute the variance (σ 2 ) of each activity’s time
estimate by the formula σ 2 = ((tp − to )/6)2
Step 6 Compute

Due date - Expected date of completion

z0 = √
Project variance

Step 7 Use standard normal table to find the probability P (z ≤ z0 ) of completing the
project within the scheduled time, where z = N (0, 1).

Example 3.1: A project is represented by the network shown below and has the fol-
lowing data :
Task : A B C D E F G H I
Least time : 5 18 26 16 15 6 7 7 3
Grestest time : 10 22 40 20 25 12 12 9 5
Most likely time : 8 20 33 18 20 9 10 8 4

Fig. 3.2:

Determine the following :

(i) expected task time and their variance,

(ii) the earliest and latest expected times to reach each node,

(iii) the critical path, and

(iv) the probability of node occurring at the proposed completion date if the original
contract time of completing the project is 41.5 weeks

Solution: (i) Using the formula te = (to + 4tm + tp )/6, σ 2 = (tp − to )2 /36, we compute
results given in the following table :

(ii) We find the earliest times in usual notations: E1 = 0, E2 = 0 + 7.8 = 7.8, E3 =

0 + 20 = 20, E4 = 0 + 33 = 33, E5 = 7.8 + 18 = 25.8, E6 = max[7.8 + 20, 20 + 9] = 29,
E7 = max[33 + 9.8, 25.8 + 8, 29 + 4] = 42.8.
 Activity to tp tm te σ2
(1-2) 5 10 8 7.8 0.69
(1-3) 18 22 20 20.0 0.44
(1-4) 26 40 33 33.0 5.43
(2-5) 16 20 18 18.0 0.44
(2-6) 15 25 20 20.0 2.78
(3-6) 6 12 9 9.0 1.00
(4-7) 7 12 10 9.8 0.69
(5-7) 7 9 8 8.0 0.11
(6-7) 3 5 4 4.0 0.11

Table 3.1

Moving backward, we calculate the latest times as follows: L7 = 42.8, L6 = 42.8 −

4 = 38.8, L5 = 42.8 − 8 = 34.8, L4 = 42.8 − 9.8 = 33, L3 = 38.8 − 9 = 29.8, L2 =
min[34.8 − 28, 38.8 − 20] = 16.8, L1 = min[16.8 − 7.8, 29.8 − 20, 33 − 33] = 0.

(iii) To find the critical path, we calculate slack time by taking difference between
the earliest expected time and the latest allowable time. Results are given in the
following table and the critical path is shown by double line in figure 3.3.

Node (i) te Ei Li Slack Var. σi

2 7.8 7.8 16.8 9.0 0.69
3 20.0 20.0 29.8 9.8 0.44
4 33.0 33.0 33.0 0.0 5.42
5 18.0 25.8 34.8 9.0 1.13
6 9.0 29.0 38.8 9.8 1.44
7 9.8 42.8 42.8 0.0 6.12

Table 3.2

(iv) The scheduled time of completing the project is 41.5 weeks. Therefore, the dis-
tance in standard deviations that the schedule time from the earliest expected
time Ei is given by

ST − Ei 41.5 − 42.8
Di = √ i = √ = −0.52
[V ar.(i)] (6.12)

where STi denotes the schedule time.

 Fig. 3.3:

Therefore, P (z ≥ −0.52) = 1 − P [z ≤ 0.52] = 1 − 0.70 = 0.30 (from Normal table)

which is the area under the standard normal curve bounded by ordinates at x = 0,
and x = 0.52.

From this, we conclude that if the project is performed 100 times under the same
conditions, there will be 30 chances when this job would take 41.5 weeks or less
to complete it.

Example 3.2: A small project consists of the following activities and time estimates :

Activity Most optimistic Most likely Most pessimistic

time time time
1-2 4 8 12
1-3 4 10 12
1-4 8 14 24
2-5 5 8 10
3-4 2 5 8
3-5 2 4 8
4-5 6 10 14
5-6 1 3 6

Determine the following :

(i) Construct the operational network.

(ii) Locate the critical path.

(iii) Calculate the mean and standard deviation for the critical path.
(iv) What is the probability of completing the project in more than 26 weeks?

(v) Along the non-critical path 1 − 4 − 6, what is the probability of exceeding 26

weeks?

(vi) What is the probability of exceeding 26 weeks along non-critical path 1−2−5−6?

Solution: The operational network is as shown below.

Fig. 3.4:

We now note the time estimates of the activities as shown in the following table:

to +4tm +tp tp −to

Activity to tm tp te = 6 σ= 6
A1−2 4 8 12 8 1.3
A1−3 8 10 12 10 0.6
A1−4 8 14 24 14.6 2.6
A2−5 5 8 10 7.8 0.8
A3−4 2 5 8 5 1
A3−5 2 4 8 4.3 1
A4−5 6 10 14 10 1.3
A5−6 1 3 6 3.1 0.8

Route Time (weeks) Total time (weeks)

1-2-5-6 8 +7.8 + 3.2 19
1-3-5-6 10 + 4.3 + 3.2 17.5
1-3-4-5-6 10 + 5 +10 + 3.2 28.2∗
1-4-5-6 14.7 + 10 +3.2 27.9

It is evident from the above table that the critical path is 1 − 3 − 4 − 5 − 6 and te = 28.2
weeks.
 Activity te σ σ2
A1−3 10 0.7 0.47
A3−4 5 1 1.00
A4−5 10 1.3 1.69
A5−6 3.2 0.8 0.64
√
Thus we have te = 28.2, σt2e = 3.82 and σte = 3.82 = 1.954.
Hence the mean and the standard deviation for the critical path are respectively 28.2
weeks and 1.954.
Taking TD = 26, we find
26 − 28.2 2.2
z= =− = −1.12.
1.954 1.954
Therefore, P (z ≤ −1.12) = 13.6%
Hence the probability of completing the project in more than 26 weeks is (100−13.6)%
= 86.4%.

Along the noncritical path 1-4-5-6, we have the following results:

Activity te σ σ2
A1−4 14.7 2.7 7.29
A4−5 10 1.3 1.69
A5−6 3.2 0.8 0.64
√
Thus we have te = 27.9, σt2e = 9.62 and σte = 9.62 = 3.101.
Taking TD = 26, we find
26 − 27.9 1.9
z= =− = −0.6.
3.101 3.101
Therefore, P (z ≤ −0.6) = 27.4%
Therefore, the probability of exceeding 26 weeks along the non-critical path 1-4-5-6
is (100 − 27.4)% or, 72.6%.

For the noncritical path 1-2-5-6 we have the following results:

Activity te σ σ2
A1−2 8 1.3 1.69
A2−5 7.8 0.8 0.64
A5−6 3.2 0.8 0.64
 √
Thus we have te = 19, σt2e = 2.97 and σte = 2.97 = 1.724.
Putting TD = 26, we find
26 − 19 7
z= =+ = 4.06
1.724 1.724
From standard normal table, we find that the probability of completing the project
along the path 1-2-5-6 in less than or equal to 26 weeks is cent percent. Hence the
probability of exceeding 26 weeks along the noncritical path 1-2-5-6 is nil.