MINE SYSTEMS ENGINEERING

(MN 6035)

DR. AMIT KUMAR GORAI

ASSOCIATE PROFESSOR
DEPARTMENT OF MINING ENGINEERING,
NIT, ROURKELA
 Lecture 14
Network Analysis in Project Planning (Contd.)
Programme Evaluation and Review Technique (PERT)
 PERT was devised in 1958 by the US Navy to optimize the time schedule of a Polaris
missile program.
 The method has the ability to accommodate the uncertainty involved in the completion
time of the activities.
 PERT is used for project scheduling in a similar way to that of CPM. However, the
duration of the completion of an activity in PERT is probabilistic in nature, but the
completion time of any activity in CPM is deterministic in nature.
 It is assumed that this probability distribution is continuous, uni-modal with two non-
negative abscissa intercepts.
 These intercepts denoted by ‘a’ and ‘b’ (a < b) are the minimum and maximum time
necessary to perform the job. In the context of PERT calculation ‘a’ is referred to as
the optimistic time and ‘b’ is known as the pessimistic time. In addition, the modal
value m is commonly referred to as the most likely time also used in PERT calculation.
 The statistical distribution which ideally suits these assumptions is known as a β-
distribution. The expected duration (te) and the standard deviation (σ) of the duration
of an activity can be approximated by
a  4m  b ba
te   
6 6
In the above equation,

a: optimistic time of completion of an activity, when all the conditions are favourable for
early completion than the expected. This is the shortest possible time in which an activity can
be completed in ideal conditions.
b: pessimistic time of completion when all the conditions are against the normal working
condition. This is the maximum possible time in which everything goes wrong and
abnormal situations prevailed.
m: most likely time of completion of an activity, when all the conditions are normal as
expected.
Once the mean and variance of the duration of the project is completed, the probability of
completion of a project in the pre-specified time, T, can be determined as

𝑡𝑒 − 𝐸 𝑡𝑒 𝑇 − 𝐸 𝑡𝑒 𝑇 − 𝐸 𝑡𝑒 𝑇 − 𝐸 𝑡𝑒
𝑃 𝑡𝑒 ≤ 𝑇 = 𝑃 ≤ =𝑃 𝑧≤ =𝑃 𝑧≤
𝑣𝑎𝑟 𝑡𝑒 𝑣𝑎𝑟 𝑡𝑒 𝑣𝑎𝑟 𝑡𝑒 𝜎

In the above equation, z represents standard normal random variable whose mean and
standard deviations are 0 and 1 respectively. The reason for consideration of normal
distribution is that the project completion duration, te, is the sum of random independent
variables.
Example 1: A mining project has 5 tasks
(A, B, C, D, and E). The sequence and the
time required to finish the individual task
in days are shown in the network diagram
as given below. Determine the expected
time of completion of the project. Given
the uncertainties in accurately estimating
activity durations, what is the probability of
completing the project by the deadline of
35 days?
Computation of expected duration and standard deviation of individual task
Activity Optimistic Most likely Pessimistic te = (a + 4m + b)/6 σ = (b - a)/6
time (a) time (m) time (b)
A 7 10 16 10.5 1.5
B 6 10 15 10.2 1.5
C 4 6 10 6.3 1.0
D 6 8 12 8.3 1.0
E 8 10 14 10.3 1.5
The network showing the expected
duration of each activity is shown here.
The computation of expected duration of
completion of the project will be done
using the similar approach as that of
CPM.
Computation of Project Completion Time
Activity te ES EF LS LF Slack
A* 10.5 0 10.5 0 10.5 0
B 10.2 0 10.2 0.3 10.5 0.3
C 6.3 10.5 16.8 12.5 18.8 2
D* 8.3 10.5 18.8 10.5 18.8 0
E* 10.3 18.8 29.1 18.8 29.1 0
Note: *Critical activity/task

Thus, the critical tasks involves in the project are A, D and E and the expected duration of
the project is 29.1 days.
The critical path is A-D-E.
Therefore, the variance of the project completion time (VT) = Sum of the variance of the
critical activities = σA2 + σD2 + σE2 = 1.52+ 12 + 1.52 = 5.5 (days)2

Standard deviation of duration of the project σ = VT = 5.5 = 2.345 (days)

Determination of the probability that the project will complete within 35 days
Expected Mean duration of project completion (te) = 29.1 days
Project time standard deviation (σ) = 2.345 days
Probability that the project will finish within 35 days (t = 35 days) is given by

𝑡 − 𝐸 𝑡𝑒
𝑃 𝑡𝑒 ≤ 𝑡 = 𝑃 𝑧 ≤
𝜎
𝑡 − 𝑡𝑒 35 – 𝑡𝑒
𝑃 𝑡 ≤ 35 = 𝑃 𝑡 − 𝑡𝑒 ≤ 35 – 𝑡𝑒 = 𝑃 ≤
𝜎 𝜎
35 – 29.1
=𝑃 𝑧≤ = 𝑃 𝑧 ≤ 2.515 = 0.994
2.345
[Note: Refer the normal distribution table. The value is corresponding to 2.5 in column and
0.01 in row]
Probability distribution function: The probability distribution for a discrete random
variable is represented by spikes of probability values correspond to the random
variable.
The cdf, F(x), represents the probability that x is less than or equal to k, for a discrete
random variable is given by
𝑃 𝑥 ≤ 𝑘 = 𝐹 𝑥 = σ𝑘𝑥=0 𝑃(𝑥)

On the other hand, the probability distributions of

continuous random variables are represented by
smooth curves. The cdf of a continuous random
variable, F(x), is a non-decreasing function with a
maximum value of 1. The cdf, F(x) represents the
probability that x is less than or equal to k for a
continuous random variable is given by
𝑘
𝑃 𝑥 ≤ 𝑘 = 𝐹 𝑥 = ‫׬‬0 𝑓 𝑥 𝑑𝑥
Where f(x) is a probability density function of random
variable, x. This can be measured by the ordinate of
the probability curve.

(a) Probability density function of a normal distribution

function (b) Cumulative probability density function of
a normal distribution function showing the probability
of x ≤ k
Similarly, the probability that a value of ‘x’ lies within the interval a, b is given by
𝑏
𝑃 (𝑎 ≤ 𝑥 ≤ 𝑏 ) = 𝐹 (𝑥 ) = 𝑓(𝑥)𝑑𝑥
𝑎
This is represented by the area under the probability curve y = f(x) between the values
x = a and x = b under the conditions f(x) ≥ 0 for all values of x
b

 f ( x)dx  1 , if (a, b) is in the range of x

a

(a) Probability density function of a normal distribution function (b) Cumulative

probability density function of a normal distribution function showing the probability
of a≤ x ≤ b
Probability of the random variable taking a value exactly equal to a given value is zero,
because
m
P(x = m) = P (m ≤ x ≤ m) =  f ( x)dx  0
m

The probability density is given by

d
f ( x)  F ( x)
dx
Common probability distribution functions
Uniform distribution: A uniform distribution, also known as a rectangular
distribution, is a distribution that has a constant probability.
The probability density function for a continuous uniform distribution on the interval
[a,b] are
0, 𝑥<𝑎
1
𝑓 (𝑥 ) = , 𝑎≤𝑥≤𝑏
𝑏−𝑎
0, 𝑥>𝑏
The cumulative distribution function for a continuous uniform distribution on the
interval [a,b] can be determined as
𝑥
𝑃 (𝑎 ≤ 𝑥 ) = 𝐹 (𝑋 ) = 𝑓 (𝑥 )𝑑𝑥
𝑎
0, 𝑥<𝑎
𝑥−𝑎
𝐹 (𝑥 ) = , 𝑎≤𝑥≤𝑏
𝑏−𝑎
1, 𝑥>𝑏
Example 1: The daily demand of explosives in an opencast mine is uniformly
distributed between 2500 and 3250 kg. The explosive tank store with a capacity of
3000 kg is refilled daily. What is probability that the tank will be empty just before
fill?
For uniform distribution of demand in the range of a (=2550) to b (=3250), the function
can be represented as
1 1 1
𝑓 (𝑥 ) = = =
𝑏 − 𝑎 3250 − 2550 700

𝑏 3250 3250
1 1 1 3250
𝑃 (𝑘 ≤ 𝑥 ≤ 𝑏 ) = 𝑓(𝑥)𝑑𝑥 = 𝑑𝑥 = 𝑑𝑥 = 𝑥 3000
𝑘 3000 700 700 3000 700
= 0.355
Normal Distribution: Although many types of distribution functions are developed
and applied for different purposes, but the normal distribution or the Gauss
distribution is widely used. The probability density function of the normal distribution
is given by
1 (𝑥−𝜇 )2
−
𝑓 (𝑥 ) = 𝑒 2𝜎 2
2𝜋𝜎
where,
µ = mean of the distribution
σ2 = variance of the distribution
𝒙𝝐(−∞, ∞)
An arbitrary normal distribution can be converted to a standardized normal density by
changing variables to𝑧 = (𝑥 − 𝜇) 𝜎 , so dz=dx/σ, yielding
1 −𝑍 2
𝑓 (𝑧 ) = 𝑒 2
2𝜋
In the normalized PDF, μ=0 and σ=1. The CDF of a standardized normal density
function is given by:

𝑍 𝑍 𝑍2
1 − 1 𝑧
𝐹 (𝑧 ) = 𝑓(𝑧)𝑑𝑧 = 𝑒 2𝑑𝑧 = 𝑒𝑟𝑓
0 2𝜋 0 2 2
where erf is an error function. Neither F(z) nor erf can be quantified with finite
additions, subtractions, multiplications, and root extractions, and therefore the
functions need to be computed numerically or otherwise approximated.
𝑃 𝑧<𝑎 =𝜙 𝑎 𝑓𝑜𝑟 𝑎 > 0

𝑃 𝑧>𝑎 +𝑃 𝑧<𝑎 =1
𝑃 𝑧 > 𝑎 = 1 − 𝑃 𝑧 < 𝑎 = 1 − ɸ(𝑎)
Therefore, ɸ −𝑎 = 1 − ɸ(𝑎)
𝑃 𝑧 < −𝑎 = 1 − ɸ(𝑎)

𝑃 𝑎 < 𝑧< 𝑏 = 𝑃 𝑧 < 𝑏 −𝑃 𝑧 < 𝑎

= ɸ 𝑏 −ɸ 𝑎