Cv4201 - Civil Engineering Management Operational Research Techniques - Network Analysis

Download as pdf or txt
Download as pdf or txt
You are on page 1of 14

Dr Wong Wai Fan, Associate Professor, NTU

NANYANG TECHNOLOGICAL UNIVERSITY


School of Civil and Environmental Engineering
College of Engineering

CV4201 – CIVIL ENGINEERING MANAGEMENT

OPERATIONAL RESEARCH TECHNIQUES

- NETWORK ANALYSIS

Dr Wong Wai Fan


Associate professor

INTRODUCTION

Operational Research (OR) is often defined as a quantitative scientific approach to


the management of systems usually involving the allocation of scarce resources. OR
involves developing a quantitative model of the system with the objectives of
prediction, control or optimization. OR is used to assist existing decision-making
procedures and not to replace them. A scientific approach means that data
gathering, measurement, and testing are necessary. A mathematical model will link
the system variables, which are controllable. The model will also express
organizational objectives in terms of the controllable variables. The fact that scarce
resources need to be allocated between alternative and competing uses highlights
the economizing nature of OR problems. The word "economizing" means "to make
best use of resources" and not "minimal use" .The different phases of an OR study is
shown in Fig. 1.

Problem
definition

Data Model
gathering construction

Solution

Are the results No


reasonable?

Yes

Implementation

Figure 1: The phases of an OR study

1
Dr Wong Wai Fan, Associate Professor, NTU

DIFFERENT OR TECHNIQUES

A number of OR techniques are available, but all of them do not find ready
application in civil engineering problems. Some relevant techniques are:

1. Network analysis
2. General linear programming
3. Transportation & Assignment models.

NETWORK ANALYSIS

Network analysis covers a number of disciplines. Our interest is mainly in the use of
network analysis to project management in construction.

Network analysis in project management

Management of projects that involves a large number of activities poses complex


problems in planning, scheduling, and control especially because the activities must
be performed in a specified logical sequence. Network analysis affords the facility of
having the logical sequential requirement incorporated into any planning
computations.

In this context, two techniques are in use, namely:

1. PERT – Programme Evaluation and Review Technique.


2. CPM – Critical Path Method

PERT depends on a probabilistic estimation of activity durations leading to a


probabilistic network model. CPM, on the other hand, is based on deterministic
values for durations. CPM is used extensively and is discussed in the following
sections.

Network Planning

There are two different approaches to the preparation of the network model:

1. Arrow diagram method (Activity on the arrow)


2. Precedence diagram method (Activity on the node)

The arrow diagram method is discussed in this course.

Formulation of the arrow diagram network

A network is a model of the construction project. The arrow diagram model


comprises a series of stages called events. The movement from one event to the
next is achieved by performing one or more activities. Fig. 2 illustrates the principle.

2
Dr Wong Wai Fan, Associate Professor, NTU

Events

An event in network logic is a definite stage that occurs during the construction
process. Some examples are:

(a) Start of work


(b) Foundations complete, ready to start on superstructure.

An event is represented by a node in the arrow diagram network. It is the junction or


the intersection of two or more activities. The event is uniquely identified by a
numeral.

Figure 2 : Example of a network model

Activities

The work associated with a project is divided into a number of activities. Each activity
represents an independent item of work. A complete list must be drawn up.

For example, the following activities refer to work connected with constructing the
foundation of a building:

(a) Excavate trenches for foundation


(b) Place lean concrete
(c) Build foundation in brickwork.

Activities are designated uniquely by the immediately preceding event and the
immediately succeeding event. Thus activity linking event 1 and event 2 is
designated Act. 1-2.

An activity will need time and resources to be carried out. Time and resources are
two dependent variables as the duration of the activity will be determined by the
resources allocated to it.

3
Dr Wong Wai Fan, Associate Professor, NTU

Basic rules in drafting network

1. The logic in Fig. 3 implies that the activity B follows the activity A. Activity 2-3
cannot start until activity 1-2 is complete, and event 2 is reached.

A B
1 2 3

Figure 3

2. Fig. 4 means that both activities 2-3 and 2-4 cannot start until 1-2 is complete.

1 2

4
Figure 4

3. Fig. 5 indicates that activities 3-4 and 3-5 cannot start until both activities 1-3
and 2-3 are complete. Event 3 is reached only when both 1-3 and 2-3 are
complete.

1 4

2 5

Figure 5

4. Dummy activities: An arrow is sometimes required only for the purpose of


indicating the logical sequence. The arrow will not represent any item of work.
For this purpose, an arrow drawn in broken lines is used. It is known as a
dummy activity. It has zero duration and needs no resources. A dummy
activity is used in a number of ways:

(a) To express logical relationships that cannot be represented otherwise.


For instance, the logic intended in Fig. 6 cannot be indicated in any
other way.

4
Dr Wong Wai Fan, Associate Professor, NTU

Act. 3-6 can start after 1-3 is complete but 4-5 can start only after both
1-3 and 2-4 are complete. The dummy activity, 3-4 serves to denote
the logical dependency of 4-5 on 1-3.

2 4 5

1 3 6

Figure 6

(b) Dummies are also used to uniquely describe activities that can be
carried out in parallel between the same two events. The method
shown in Fig. 7(i) cannot be adopted as both activities A and B will
have the notation 1-3. The introduction of a dummy as shown in Fig.
7(ii) enables activity A to be designated as 1-2 and B as 1-3.

A
2
A
1 3

1 3
B B

(I) (II)

Figure 7

Establishing logical sequences

A network model can be constructed only after the correct logical sequences
between the different activities have been established. To do this, each activity must
be considered one by one, and answers must be evolved for the following three
questions:

Q1. Which activities must be complete before this activity can start?
Q2. Which activities cannot start until this activity is complete?
Q3. Which activities can be worked in parallel with this one?

5
Dr Wong Wai Fan, Associate Professor, NTU

Exercise

Listed below are the activities involved in constructing a foundation.

A. clear site
B. excavate for foundations
C. place lean concrete
D. fabricate formwork for base
E. cut and bend reinforcement for base
F. tie reinforcement in base
G. fix formwork
H. concrete base
I. cure concrete

Prepare an Arrow diagram network for the job.

Solution

Taking each activity one by one, it is sufficient to generate answers to Q1 first. From
these answers, it is possible to establish the logical sequence as in table below:

Answers
Q1 Q2 Q3
Activity
Immediately Immediately Activities in
preceding succeeding parallel
activities activities
A - B, D, E -
B A C D, E
C B F D, E
D A G B, C, E
E A F B, C, D
F C, E G D
G D, F H -
H G J -
J H - -

6
Dr Wong Wai Fan, Associate Professor, NTU

C
B

A E

F
D

G H J

Note: Although generating the table of logical sequences would appear tedious, a
planner with experience and intuitive judgment could perform the task with relative
ease. The network diagram can be drawn using the logical sequences as shown
above.

Analysis of Network

The network can be analyzed to obtain the following:

(a) a schedule of activities


(b) a schedule of activities and resources
(c) resource usage or resource histogram
(d) cost versus time curve

In order to carry out the analysis, data regarding each activity is essential. The
activity duration and the resources required are two basic information without which
network analysis cannot be proceeded with. They are mutually dependent and must
be estimated using past data and other productivity or performance values. However,
in the initial stages of planning approximate estimates are sufficient, and closer
attention is needed when final and detailed plans are prepared.

Analysis of network

Two types of analyses are carried out:

1. Time scheduling or time analysis


2. Resources scheduling

Time analysis is performed during preliminary stages of planning. For time analysis, it
is assumed that all resources are available whenever required.

Resources scheduling is done if there are limited resources and the available
resources are, at times, insufficient to work the activities to the schedule obtained by
time analysis.

7
Dr Wong Wai Fan, Associate Professor, NTU

Time analysis

The following computations are involved:

1. Calculating the Earliest possible time and Latest permissible time for each
event. These are called the Earliest Event Time (EET) and the Latest Event
Time (LET) respectively.

2. From the information obtained in (1), determining the

Earliest Start Time (ES) ,


Earliest Finish Time (EF) ,
Latest Start Time (LS),
and Latest Finish Time (LF)

of the activities.

3. Finding the Critical Path.

4. Computing the Total Float and Free Float of the activities.

Example

The network shown in Fig. 8 will be used to illustrate the Time analysis procedure.
The activity durations in days are given alongside the arrows.

10
4

1 2 5 6 7
1 6 2 2

15 6

Figure 8

Earliest Event Time (EET) - Definition

The earliest event time denotes the earliest possible time at which an event can
occur. All activities that terminate at a given event must be completed before that
event occurs.

8
Dr Wong Wai Fan, Associate Professor, NTU

Forward Pass

The process of computing the EET of all the events from the start of the project to the
finish is called the forward pass. The earliest event times are inserted in the left hand
boxes as in Fig. 9.

"End of Time" convention is used - day 0 means end of day 0.

Earliest time for event 1 = day 0


Earliest time for event 2 = 0+1 = 1
Earliest time for event 3 = 1 + 10 = 11
Earliest time for event 4 = 1 + 15 = 16

11 18

10
4

0 0 1 1 16 20 22 22 24 24

1 2 5 6 7
1 6 2 2

15 6

4
16 16

Figure 9: Time Analysis of Network

Event 5 has two activities 2-5 and 4-5 finishing on it.

From event 2 to event 5 = 1 + 16 = day 7 and


From event 4 to event 5 = 16 + 0 = day 16.

Therefore, earliest time for event 5 is day 16.

Similarly earliest time for event 6 is the latest of

11 + 4 = 15
16 + 2 = 18 = day 22.
16 + 6 = 22

Earliest time for event 7 = 22 + 2 = day 24.

Project duration = 24 days.

9
Dr Wong Wai Fan, Associate Professor, NTU

Latest Event Time (LET) - definition

The latest event time is the latest time by which an event must take place if the
project duration is not to be exceeded.

Backward Pass

Backward pass is the process of computing the latest event times. It is carried out in
the reverse direction starting from the last event and subtracting the activity
durations. LET is recorded in the right hand box. See Fig. 9.

If the project duration is not to be extended then


Latest event time for event 7 = day 24.

Latest event time for event 6 = 24 – 2 = 22.


Latest event time for event 5 = 22 – 2 = 20.

Latest event time for event 4 is the smallest of

20 - 0 = 20
day 16
22 - 16 = 16

Similarly other values can be found.

Activity Schedule Times

Earliest Start and Earliest Finish Times

Earliest start time (ES) of an activity is the earliest event time of the event from which
the activity starts. Earliest finish time (EF) of the activity is the time at which the
activity will be complete if it starts at its ES.

EF = ES + d (d - duration of the activity)

11 18 22 22
4
3 6

Figure 10

Thus, for activity 3-6 in Fig. 10 extracted from Fig. 9,

ES = day 11 and EF = 11 + 4 = day 15

Table 1 gives the ES and EF for all the activities.

10
Dr Wong Wai Fan, Associate Professor, NTU

Latest Start and Latest Finish Times

The latest finish time (LF) of an activity is the LET of the event on which the activity
ends. The latest start time (LS) of the same activity will be the time at which the
activity must start if it is to be complete at its latest finish time.

LS = LF – d

From Fig. 10, for activity 3-6,

LF = day 22 and LS = 22 - 4 = day 18.

Table 1 gives the LS and LF of all the activities.

Table 1: Schedule Times of Activities

Earliest Latest Float


Activity Duration
Start Finish Start Finish Total Free
1-2 1 0 1 0 1 0* 0
2-3 10 1 11 8 18 7 0
2-4 15 1 16 1 16 0* 0
2-5 6 1 7 14 20 13 9
3-6 4 11 15 18 22 7 7
4-5 dummy - - - - - -
4-6 6 16 22 16 22 0* 0
5-6 2 16 18 20 22 4 4
6-7 2 22 24 22 24 0* 0

* indicates critical activities

Critical Activities

Some activities have the same earliest and latest times, both start and finish. These
activities cannot be delayed and are regarded as critical to completing the project on
time. They are known as critical activities. The other activities are called the non-
critical activities. The critical activities form a continuous chain. The continuous
sequence of critical activities from start to finish is known as the Critical Path for the
project. In the example in Fig. 9, the critical path is 1 -2 -4 -6 -7. It is the longest path
or chain through the network (in terms of total duration). There can be more than one
critical path in the network. An examination of Fig. 9 will indicate that the events on
the critical path have their EET= LET. This is a necessary but not a sufficient
condition for identifying the critical path.

11
Dr Wong Wai Fan, Associate Professor, NTU

Float of Activities

Any non-critical activity will have its LS later than its ES. Even if this activity does not
start at its ES but start on or before the LS, the subsequent work will not be delayed.

The Spare time available (= LS – ES or LF - EF) is called the Total Float (TF) of the
activity and is defined as the amount of time by which the activity can be delayed
without extending the project duration. By implication, all critical activities will have
zero total float.

Figure 11: Schedule and float for a non-critical activity

Referring to Fig. 11,

TF = LF – EF

or TF = LS – ES

If a non-critical activity is delayed by a period equal to its total float, the project
duration will not be extended. But it will cause the rest of the activities in the same
chain to lose their spare times or float. If the activity is scheduled to start at its ES as
in Fig. 11 (i) then the full total float is retained, and the float of succeeding activities in
the same chain will not be .affected. But if it is scheduled for a latest start as in Fig.
11(ii) then the entire total float is consumed, and this will cause the float of
succeeding activities in the chain to be reduced correspondingly.

Free Float

Some non-critical activities may be delayed by a certain amount of time without


affecting the float of the subsequent activities. This introduces the concept of Free
Float (FF) for a non-critical activity. Free float of an activity is defined as the amount
of time that an activity can be delayed without changing the earliest start time of the
immediate successor.

12
Dr Wong Wai Fan, Associate Professor, NTU

EETi EETj 11 18 22 22
d 4
i j 3 6

(i) (ii)

Figure 12

Consider activity i-j for calculating free float.

EET of event j must not be changed.

FF = EET(j) – [ EET(i) + d ]

Alternatively, the ES of any of the immediately succeeding activities must not be


altered.

FF = (ES of the successor) – (EF of the activity considered)

Calculating the free float of act. (3-6) in Fig. 12(ii) (extracted from Fig. 9)

Using event times, FF = EET(6) – EET(3) – d = 22 – 11 – 4 = 7

Alternatively, the successor to act. (3-6) is act. (6-7).

ES of 6-7 = 22
EF of 3-6 = 15
FF = 22 – 15 = 7

Usually the last activity of a non-critical chain will have free float.

The values of the total float and the free float are also given in Table 1.

Information for Planning by CPM

The data from table 1 becomes the basic information required to formulate the final
details regarding activity scheduling, resources planning, and financial planning.
These aspects will be dealt with in the later lectures in construction management.

13
Dr Wong Wai Fan, Associate Professor, NTU

Some Common Characteristics of CPM Project Network

1. Each activity of an unbroken chain of non-critical path will have the same total
float. For example, as shown in the partial CPM project network below, Act 2-7
and 7-10 forms an unbroken chain, they share the same amount of total float (5
days).

2. Consumption of float upstream will generally affect the float available in the
downstream activities. (Refer to the partial network below, check what would
happen to the float available for the downstream activities 2-7-10 if Act 1-2 is
delayed by 2 days.)

60 60

1 5 10

0 0 15

25
2 7
15
25 28 40 45

3. If there is a delay in the Act 2-7, the total float available on this chain 2-7-10 will
be depleted accordingly. Eg. If 2-7 has a delay of 2 days, then the float available
to 7-10 will only have 3 days (5-2) left.

Updated: 2005

14

You might also like