Operations Research: Project Management Using Network Analysis
Operations Research: Project Management Using Network Analysis
com
OPERATIONS RESEARCH
PROJECT MANAGEMENT USING NETWORK ANALYSIS
‘Project’ is not a new word for any of us. We read about new technology development (3G
mobiles, vaccine for H1N1), Implementation of new transport system (Metro, Petronet),
Quality improvement program (TEQIP), New civil construction (stadium, factory), etc. The
common features among these entitle them to be called as Projects.
Several analytical techniques such as PERT, CPM, PEP, RAMPS have been developed to aid
real time project management.
The phases that comprise PM can be identified as (i) Project planning, (ii) Project scheduling
and (iii) Project executing and Controlling.
1
www.getmyuni.com
WBS: Break down the project into constituent activities such that each activity is clearly
identifiable and manageable
Activity: This is a physically identifiable part of the project that consumes time and
resources. It is represented by an arrow (in AOA diagrams)
Events (node): These are the beginning or end points of an activity. Event is a point in time
and does not consume any time or resource and is represented by a circle.
Path: This is a continuous chain of activities from the beginning to the end of the project
Network: A graphical representation of logically and sequentially arranged arrows and nodes
of a project. It indicates the interrelationships between the activities of a project
AOA (Activity-On-Arrow) Diagram: A network with
activities represented on arrows and event on nodes.
Often dummy arrow is needed to establish precedence
relationship which makes the network a little
cumbersome and requires greater computation. But it is
easily understandable.
2
www.getmyuni.com
Network Construction:
3
www.getmyuni.com
Open bonnet D -
Check battery G D
Collect payment L K
CPM was developed by E I du Pont de Nemours & Co. (1957) for construction of new
chemical plant and maintenance shut-down. CPM does not incorporate uncertainties in job-
times, thus usable for projects with activities having single time estimates, which can be
arrived at with prior experience on similar projects. It assumes activity time is proportional to
the resources allocated to it (within a certain limit). CPM is mostly suitable for construction
type projects. The objective of using CPM is to determine Critical path, Minimum project
duration and Floats available with each activity.
This technique was developed by the U S Navy (1958) for the POLARIS missile program.
The project involved coordination of thousands of contractors and agencies. With the help of
PERT, the project got completed 2 years ahead of schedule. PERT is suitable for Non-
repetitive projects (ex. R & D work), where job-times are not estimable with certainty a
priori. PERT uses multiple estimates of activity-time (probabilistic nature). The technique
emphasizes on the completion of various stages of a project. Jobs that may cause delays are
known in advance in terms of their variability.
Problem 3. Suppose a robot building firm plans the following project. Draw the n/w and find
the Critical path
Project Network:
Problem 4. (solved in ppt). A small maintenance project consists of the following 10 jobs
whose precedence relationships are identified by their node numbers. Draw an AOA diagram
and identify the CP by enumeration method.
5
www.getmyuni.com
1 2 2
2 3 3
2 4 5
3 5 4
3 6 1
4 6 6
4 7 2
5 8 8
6 8 7
7 8 4
Limitations of enumeration method: (i) Very difficult to use when the complexity of network
increases, (ii) No information on flexibility available with respect to activities, (iii) Difficult
to schedule activities in complex networks
Structured approach:
Forward pass: To determine the Earliest Occurrence (EO) times of events. The
computations start at Node 1 and advance recursively to the last Node n
Backward Pass: To determine the Latest Occurrence (LO) times of events. The computations
start at the last Node n and end at Node 1
6
www.getmyuni.com
1. Consider separately, each node (say, j) to which an activity is directly reaching from
node i
2. Subtract activity time tij from the corresponding LO(j)
3. Select the minimum of them as LO(i)
The Backward pass is complete when LO(1) is computed
Problem 5. Determine the Critical Path for the activity data given below. All durations are in
days.
1 2 5
1 3 6
2 3 3
2 4 8
3 5 2
3 6 11
4 5 0
4 6 1
5 6 12
7
www.getmyuni.com
Calculate the floats associated with each activity using the above information
8
www.getmyuni.com
C D 8
C E 2
C F 7
D G 5
E G 6
E H 4
F H 4
G H 6
G I 2
H I 6
9
www.getmyuni.com
In many situations it becomes ne necessary to cut down the project duration. How ccan it be
done? Activities that are criticall nneed to be crashed in order to reduce the project
ct duration as
it is these activities that determin
ine the project duration. But this has got its own cost
c
implications. Reduction in projec ject duration calls for more resources to be pumpeped in and
hence, the direct costs increase.. W Whereas indirect costs such as equipment rent,, supervision
s
charges, etc. reduce. Thus, it beco
ecomes necessary to identify a project duration up to which the
project can be crashed so that ove ver all project costs are minimum.
2-3 4 2 40
2-4 7 3 10
3-4 5 2 20
10
www.getmyuni.com
Activity Days Proj. Cost of crashing Total Cost Over- Total Cost
crashed saved duration of crashing heads
The lowest cost schedule is the plan corresponding to project duration of 7 days.
Problem.8. Table below gives the time and cost data with respect to normal and crash
periods of a project. (a) Draw the n/w of the project, (b) What is the normal duration and cost
of the project? (c) Determine the project cost if all activities are crashed indiscriminately (d)
Determine the optimum project duration, if the indirect cost is Rs. 150/day
Critical path : 1 – 2 – 3 – 4 – 5 – 6
11
www.getmyuni.com
d) Now draw a squared network as shown below. Choose the activities on the critical path to
crash such that the present critical path continues to remain as (at least one of) the critical
path. Also the cost of crashing/day shall be the least among available options at any stage.
Thus the final crashed network appears as below. The associated cost table is also shown.
The table shows that the project duration of 19 days is most economical and optimum.
12
www.getmyuni.com
Used in such projects where the activities are, to some extent, governed by States-of-nature
and the organization does not have full control over the activity duration and hence the time
estimates of activities are probabilistic. For example, R&D projects, projects exposed to seasonal
variations.
In PERT, we use three time estimates for an activity which reflect the degrees of uncertainty
to = Optimistic time: When environment is very favorable
tm = Most likely time: When environment is just normal
tp = Pessimistic time: When environment is very unfavorable
There can be a number of time estimates between to and tp
The frequency distribution curve of activity time ‘t’ is assumed to be a distribution which
is unimodal at tm and has extremes at to and tp. But tm need not be the mid point between to and
tp.
The expected (average) time te of an activity is calculated as a weighted average of these three
estimates.
How reliable te is? How confident are we that the activity gets over in time te? This depends
upon the spread of the distribution of ‘t’ or variability of ‘t’. Larger the variation, lesser
would the confidence. Example: Distributions with same mean but different spreads
Variance ( 2)’ and ‘Standard deviation ( )’ are two important measures of variability. In
2
PERT, variance of an activity is calculated as: = [(tp- to)/6]2
Using te s, the earliest and latest occurrences of each event are calculated from which the CP
is determined
There is a possibility of alternate Critical paths. In such cases we need to consider the critical
path with greater variance. As the analysis is based on the probabilistic time estimates even
near Critical paths should be carefully examined.
When activity times (‘tj’ for j= 1, 2,…n) are independent and identically distributed random
variables, then (by Central limit theorem) project duration (T) will follow normal distribution
13
www.getmyuni.com
with expected project duration Te and project variance p2 . With Te and p we can compute
the probability of completing the project within a due date using normal PD tables.
Problem 9. Given the list of activities in a project and their time estimates (in days):
a) Draw the project network,
b) Determine the critical path(s) and the expected project duration
c) What is the probability that project will be completed in 35 days?
d) What due date has 90% chance of being met?
Activity t t t
o m p
1-2 6 12 30
1-3 3 6 15
1-4 3 9 27
2-6 4 19 28
3-5 3 9 27
3-6 2 5 8
4-5 1 4 7
5-6 6 12 30
a)
14
www.getmyuni.com
b)
Activity t
e
Critical path = 1 – 3 – 5 – 6
1-2 12
1-3 7 Expected project duration T = 32 days
e
1-4 11
2-6 18
3-5 11
3-6 5
4-5 4
5-6 14
c)
2
Activity t t t CA
o m p
1-2 6 12 18 4
1-3 3 6 15 1-3 4
1-4 3 9 27 16
2-6 4 19 28 16
3-5 3 9 27 3-5 16
3-6 2 5 8 1
4-5 1 4 7 1
5-6 6 12 30 5-6 16
15
www.getmyuni.com
2
Variance = 36 days
p
Standard deviation = 36 = 6 days
p
Expected duration of the project Te = 32 days
Std. deviation of project duration p = 6 days
Scheduled duration Ts= 35 days
We know, Z = (Ts – Te)/ p
= (35-32)/6 = +0.5
The area under the normal curve (from standard normal PD table) up to Z = +0.5
= 0.6915 (i.e. 69.15 % chance)
The probability that project will be completed in 35 days is 0.6915
d)
We know, Z = (Ts – Te)/ p
For 90% chance (probability = 0.9) area under the std. normal curve, we have Z = +1.28
+1.28 = (Ts - 32)/6
Hence, Ts = 39.68 days
Project duration of 39.68 days has 90% chance of being met
Advantages:
• Simple to understand and use
• Show whether the project is on schedule; or behind/ ahead of the schedule
• Identify the activities that need closer attention (critical)
• Determine the flexibility available with activities
• Show potential risk with activities (PERT)
• Provide good documentation of the project activities
• Help to set priorities among activities and resource allocation as per priority
Limitations:
• Demand clearly defined and stable activities
• Precedence relationship should be known in advance
• Overemphasis on Critical path
• Activity time estimates are subjective
• Activity times in PERT may not follow Beta PD in reality
• Cost of crashing an activity may not be linear
16