Tài liệu Project Planning and Control Part 4 pdf

17

Computer analysis

Most manufacturers of computer hardware, and

many suppliers of computer software, have

written programs for analysing critical path

networks using computers. While the various

commercially available programs differ in detail,

they all follow a basic pattern, and give, by and

large, a similar range of outputs. In certain

circumstances a contractor may be obliged by his

contractual commitments to provide a computer￾ized output report for his client. Indeed, when a

client organization has standardized on a partic￾ular project management system for controlling

the overall project, the contractor may well be

required to use the same proprietary system so

that the contractor’s reports can be integrated into

the overall project control system on a regular

basis.

History

The development of network analysis techniques

more or less coincided with that of the digital

computer. The early network analysis programs

were, therefore, limited by the storage and

processing capacity of the computer as well as

the input and output facilities.

Project Planning and Control

The techniques employed mainly involved producing punched cards (one

card for each activity) and feeding them into the machine via a card reader.

These procedures were time consuming and tedious, and, because the

punching of the cards was carried out by an operator who usually understood

little of the program or its purpose, mistakes occurred which only became

apparent after the printout was produced.

Even then, the error was not immediately apparent – only the effect. It then

often took hours to scan through the reams of printout sheets before the actual

mistake could be located and rectified. To add to the frustration of the planner,

the new printout may still have given ridiculous answers because a second

error was made on another card. In this way it often required several runs

before a satisfactory output could be issued.

In an endeavour to eliminate punching errors attempts were made to use

two separate operators, who punched their own set of input cards. The cards

were then automatically compared and, if not identical, were thrown out,

indicating an error. Needless to say, such a practice cost twice as much in

manpower.

Because these early computers were large and very expensive, usually

requiring their own air-conditioning equipment and a team of operators and

maintenance staff, few commercial companies could afford them. Computer

bureaux were therefore set up by the computer manufacturers or special

processing companies, to whom the input sheets were delivered for punching,

processing and printing.

The cost of processing was usually a lump sum fee plus x pence per activity.

Since the computer could not differentiate between a real activity and a

dummy one, planners tended to go to considerable pains to reduce the number

of dummies to save cost. The result was often a logic sequence, which may

have been cheap in computing cost but was very expensive in application,

since frequently important restraints were overlooked or eliminated. In other

words, the tail wagged the dog – a painful phenomenon in every sense. It was

not surprising, therefore, that many organizations abandoned computerized

network analysis or, even worse, discarded the use of network analysis

altogether as being unworkable or unreliable.

There is no doubt that manual network analysis is a perfectly feasible

alternative to using computers. Indeed, one of the largest petrochemical

complexes in Europe was planned entirely using a series of networks, all of

which were analysed manually.

128

Computer analysis

The PC

The advent of the personal computer (PC) significantly changed the whole

field of computer processing. In place of the punched card or tape we now

have the computer keyboard and video screen, which enable the planner to

input the data direct into the computer without filling in input sheets and

relying on a punch operator. The information is taken straight from the

network and displayed on the video screen as it is ‘typed’ in. In this way, the

data can be checked or modified almost instantaneously.

Provided sufficient information has been entered, trial runs and checks can

be carried out at any stage to test the effects and changes envisaged. Modern

planning programs (or Project Management systems, as they are often called)

enable the data to be inputted in a random manner to suit the operator,

provided, of course, that the relationship between the node numbers (or

activity numbers) and duration remains the same.

There are some programs which enable the network to be produced

graphically on the screen as the information – especially the logic sequence

– is entered. This, it is claimed, eliminates the need to draw the network

manually. Whether this practice is as beneficial as suggested is very

doubtful.

For a start, the number of activities which can be viewed simultaneously

on a standard video screen is very limited, and the scroll facility which

enables larger networks to be accommodated does not enable an overall

view to be obtained at a glance. The greatest drawback of this practice,

however, is the removal from the network planning process of the team

spirit, which is engendered when a number of specialists sit down with the

planner round a conference table to ‘hammer out’ the basic shape of the

network (see Chapter 20). Most problems have more than one solution, and

the discussions and suggestions, both in terms of network logic and

durations, are invaluable when drafting the first programs. These meetings

are, in effect, a brainstorming session at which the ideas of the various

participants are discussed, tested and committed to paper. Once this draft

network has been produced, the planner can very quickly input it into the

computer and call up a few test runs to see whether the overall completion

date can, in fact, be achieved. If the result is unsatisfactory, logic and/or

duration changes can be discussed with the project team before the new data

are processed again by the machine. The speed of the new hardware makes

it possible for the computer to be part of the planning conference, so that

(provided the planner/operator is quick enough) the ‘what if’ scenarios can

be tested while the meeting is in progress. A number of interim test runs

129