The Case for Manual Analysis
Abstract
This chapter discusses the advantages of analyzing networks manually before entering the data into a computer for a fuller report on early and late starts and finishes.
Keywords
Manual analysis; critical path analysis
Chapter Outline
Although network analysis is applicable to almost every type of organization, as shown by the examples in Chapter 29, most of the planning functions described in this book have been confined to those related to engineering construction projects. The activities described cover the full spectrum of operations from the initial design stage, through detailing of drawings and manufacture, up to and including construction. In other words, from conception to handover.
In this age of specialization there is a trend to create specialist groups to do the work previously carried out by the members of more conventional disciplines. One example is teaching where teaching methods, previously devised and perfected by practising teachers, are now developed by a new group of people called educationalists.
Another example of specialization is planning. In the days of bar charts, planning was carried out by engineers or production staff using well-known techniques to record their ideas on paper and transmit them to other members of the team. Nowadays, however, the specialist planner or scheduler has come to the fore, leaving the engineer time to get on with his engineering.
The Planner
Planning in its own right does not exist. It is always associated with another activity or operation, i.e., design planning, construction planning, production planning, etc. It is logical, therefore, that a design planner should be or should have been a designer, a construction planner should be familiar with construction methods and techniques, and a production planner should be knowledgeable in the process and manufacturing operations of production – whether it be steelwork, motors cars, or magazines.
In construction, as long as the specialist planner has graduated from one of the accepted engineering disciplines and is familiar with the problems of a particular project, a realistic network will probably be produced. By calling in specialists to advise him in the fields with which he is not completely conversant, he can ensure that the network will be received with confidence by all the interested parties.
The real problem arises when the planner does not have the right background, i.e., when he has not spent a period in design or has not experienced the holdups and frustrations of a construction site. Strangely enough, the less familiar a planner is with the job he is planning, the less he is inclined to seek help. This may well be due to his inability to ask the right questions, or he may be reluctant to discuss technical matters for fear of revealing his own lack of knowledge. One thing is certain: A network that is not based on sound technical knowledge is not realistic, and an unrealistic network is dangerous and costly, since decisions may well be made for the wrong reasons.
All that has been said so far is a truism that can be applied not only to planning but to any human activity where experts are necessary in order to achieve acceptable results. However, in most disciplines it does not take long for the effects of an inexperienced assistant to be discovered, mainly because the results of his work can be monitored and assessed within a relatively short time period. In planning, however, the effects of a programme decision may not be felt for months, so it may be very difficult to ascertain the cause of the subsequent problem or failure.
The Role of the Computer
Unfortunately, the use of computers has enabled inexperienced planners to produce impressive outputs that are frequently utterly useless. There is a great danger in shifting the emphasis from the creation of the network to the analysis and report production of the machine, so that many people believe that to carry out an analysis of a network one must have a computer. In fact, of course, the computer is only a sophisticated number cruncher. It does not see the whole picture, including access problems, political or cultural restraints, labour issues, and staff idiosyncrasies. The kernel of network analysis is the drafting, checking, refining, and redrafting of the network itself, an operation that must be carried out by a team of experienced participants of the job being planned. To understand this statement, it is necessary to go through the stages of network preparation and subsequent updating.
Preparation of the Network
The first function of the planner in conjunction with the project manager is to divide the project into manageable blocks. The name is appropriate since, like building blocks, they can be handled by themselves and shaped to suit the job but are still only a part of the whole structure to be built.
The number and size of each block is extremely important since, if correctly chosen, a block can be regarded as an entity that suits both the design and the construction phases of a project. Ideally, the complexity of each block should be about the same, but this is rarely possible in practice since other criteria such as systems and geographical location have to be considered. If a block is very complex, it can be broken down further, but a more convenient solution may be to produce more than one network for such a block. The aim should be to keep the number of activities down to 200–300 so that they can be analysed manually if necessary.
As the planner sketches his logic roughly, and in pencil on the back of an old drawing, the construction specialists are asked to comment on the type and sequence of the activities. In practice, these sessions – if properly run – generate an enthusiasm that is a delight to experience. Often consecutive activities can be combined to simplify the network, thus easing the subsequent analysis. Gradually the job is ‘built’, difficulties are encountered and overcome, and even specialists who have never been involved in network planning before are carried away by this visual unfolding of the programme.
The next stage is to ask each specialist to suggest the duration of the activities in his discipline. These are entered onto the network without question. Now comes the moment of truth. Can the job be built on time? With all the participants present, the planner adds up the durations and produces his forward pass. Almost invariably the total time is longer than the deadlines permit. This is when the real value of network analysis emerges. Logics are re-examined, durations are reduced, and new construction methods are evolved to reduce the overall time. When the final network – rough though it may be – is complete, a sense of achievement can be felt pervading the atmosphere.
This procedure, which is vital to the production of a realistic programme, can, of course, only be carried out if the ‘blocks’ are not too large. If the network has more than 300 activities it may well pay the planner or project manager to re-examine that section of the programme with a view to dividing it into two smaller networks. If necessary, it is always possible to draw a master network, usually quite small, to link the blocks together.
One of the differences between the original PERT program and the normal CPM programs was the facility to enter three time estimates for every activity. The purpose of the three estimates is to enable the computer to calculate and subsequently use the most probable time, on the assumption that the planner is unwilling or unable to commit himself to one time estimate. The actual duration used is calculated from the expression known as the β distribution:
where te is the expected time, a the optimistic time, b the pessimistic time, and m the most likely time.
However, this degree of sophistication is not really necessary, since the planner himself can insert what he considers to be the most probable time. For example, a foreman, upon being pressed, estimated the times for a particular operation to be:
The planner will probably insert 7 days or 8 days. The computer, using the above distribution, would calculate:
With the much larger variables found in real-life projects, such finesse is a waste of time.
A single time estimate by an experienced planner is all that is required.
Typical Site Problems
Once construction starts, problems begin to arise. Drawings or other data arrive late on-site, materials are delayed, equipment is held up, labour becomes scarce or goes on strike, underground obstructions are found, the weather deteriorates, etc.
Each new problem must be examined in the light of the overall project programme. It will be necessary to repeat the initial planning meeting to revise the network, to reflect on these problems, and to possibly help reduce their effects. It is at these meetings that ingenious innovations and solutions are suggested and tested.
For example, Figure 22.1 shows the sequence of a section of a pipe rack. Supposing the delivery of pipe will be delayed by four weeks, completion now looks like week 14. However, someone suggests that the pump bases can be cast early with starter bars bent down to bond the plinths at a later date. The new sequence appears in Figure 22.2. Completion time is now only week 11, a saving of three weeks.
Figure 22.1
Figure 22.2
This type of approach is the very heart of successful networking and keeps the whole programme alive. It is also very rapid. The very act of discussing problems in the company of interested and knowledgable colleagues generates an enthusiasm that carries the project forward. With good management, this momentum is passed right down the line to the people who are actually doing the work at the sharp end.
The NEDO (National Economic Development Office) Report
Perhaps the best evidence that networks are most effective when kept simple is given by the NEDO report, which is still applicable today even though it was produced way back in the early 1980s.
The relevant paragraphs are reproduced below by permission of HM Stationery Office.
1. Even if it is true that UK clients build more complex plants, it should still be possible to plan for and accommodate the extra time and resources this would entail. By and large the UK projects were more generously planned but, none the less, the important finding of the case studies is that, besides taking longer, the UK projects tended also to encounter more overrun against planned time. There was no correlation across the case studies between the sophistication with which programming was done and the end result in terms of successful completion on time. On the German power station the construction load represented by the size and height of the power station was considerable, but the estimated construction time was short and was achieved. This contrasts with the UK power stations, where a great deal of effort and sophistication went into programming, but schedules were overrun. On most of the case studies, the plans made at the beginning of the project were thought realistic at that stage, but they varied in their degree of sophistication and in the importance attached to them.
2. One of the British refineries provided the one UK example where the plan was recognized from the start by both client and contractor to be unrealistic. None the less, the contractor claimed that he believed planning to be very important, particularly in the circumstances of the UK, and the project was accompanied by a wealth of data collection. This contrasts with the Dutch refinery project where planning was clearly effective but where there was no evidence of very sophisticated techniques. There is some evidence in the case studies to suggest that UK clients and contractors put more effort into planning, but there is no doubt that the discipline of the plan was more easily maintained on the foreign projects. Complicated networks are useful in developing an initial programme, but subsequently, though they may show how badly one has done, they do not indicate how to recover the situation. Networks need, therefore, to be developed to permit simple rapid updates, pointing where action must be taken. Meanwhile the evidence from the foreign case studies suggests that simple techniques, such as bar charts, can be successful.
3. The attitudes to planning on UK1 and the Dutch plant were very different, and this may have contributed to the delay of UK1 although it is impossible to quantify the effect. The Dutch contractor considered planning to be very important, and had two site planning engineers attached to the home office during the design stage. The programme for UK1 on the other hand was considered quite unrealistic by both the client and the contractor, not only after the event but while the project was under way, but neither of them considered this important in itself.
On UK 1 it was not until the original completion date arrived that construction was rescheduled to take 5 months longer. At this point construction was only 80% complete and in the event there was another eight month’s work to do. Engineering had been 3 months behind schedule for some time. A wealth of progress information was being collected but no new schedule appears to have been made earlier.
Progress control and planning was clearly a great deal more effective on the Dutch project; the contractor did not believe in particularly sophisticated control techniques, however.
Clearly, modern computer programs are more sophisticated and user-friendly and have far greater functionality, but it is precisely because these programs are so attractive that there is the risk of underestimating, or even ignoring, the fundamental and relatively simple planning process described earlier in this chapter.