Cost 6

Planning project costs A Tulip jetting realistic plans for the expendi~re on a project can cause much guessing to be done in the initial stages. These guesses are more vague if only a little is known about the projects being studied. The resource smoothing method as it can be applied to cost aspects of project planning is described. Keywords: project management, resource-smoothing algorithm

cost

planning,

When a manager is confronted with a project for the first time, there are many questions to be answered. In the current economic situation, the rate at which money is required often becomes a regulating factor. It is, therefore, important to set the budget requests at realistic levels while there is the minimum of detailed information available. In setting these levels, a variety of possibly conflicting objectives should be achieved. They should be as low as possible without unduly delaying the project milestones. Wild fluctuations in the budget should be restricted as much as possible. Wages may need to be treated separately from costs incurred in materials and machines. It may be necessary to phase the progress to suit other projects.

REQUIREMENTS The algorithm of resource smoothing originally described by Tulip’ has proved a reliable and adaptable method of scheduling. For each ‘item’ to be scheduled, this system requires four pieces of data: l l l l

the the the the

earliest time the work can start, latest time it is needed, estimated duration, estimated rate of spend.

Honeywell Information Systems, Road, Uxbridge, Middlesex, UK

194

Boundary

0263-78631831040194-03

House,

$03.00 0

Crickettield

If the different types of money involved in the project need to be identified because of conflicts, then this last item would need to be repeated for each category. At the initial stages of project formulation there is not much more information available. Perhaps there could be some idea of the sequence of work even though the work items have not been specified in detail. In the original specification of the method, a full project network was required but experience has shown that this, while useful, is not essential. Consider, as an example, a draft project that is expected to last for three years. During the initial specification, a number of high-level tasks (almost certainly no more than 30) will be defined. These tasks will have a description such as ‘develop marketing strategy’ or ‘construct plant on site’. For the actual planning and control, they will need to be divided into manageable units, but for this exercise a job lasting six to nine months is probably the best that current information will allow.

METHOD OF WORKING Four steps in the smoothing process are cycled until all the tasks have been scheduled.

Schedule critical jobs If a task available, time and scheduled

has time constraints so that no float is then it is scheduled beginning at its earliest lasting for the estimate. All such jobs are before proceeding.

Find the most important job The second step is to scan all those tasks that remain to be scheduled in order to find the best one to study. Arithmetically, this can be stated as the task with the highest criticality factor where criticality factor = total cost/float available. If two or more activities end up with the same criticality factor then the manager needs to select one. It may be the one with the earliest or latest start date, or the shortest or longest duration.

1983 Butterworth

& Co (Publishers) Ltd

Project Management

If the manager is looking at classes of money e.g. wages and materials, then there are two options for adjusting this factor. Either all of one type of money can be studied to the exclusion of all the others, or some sort of weighting factor can be used in determining the cost of the job. For example, wage costs may be considered doubled when compared to materials, using the excuse that wages must be paid but outside suppliers can wait. The easier of these options is the former. While the most important type of money is being scheduled to the exclusion of all others, it does have implications for the others as will be seen. It is also a system of ranking that a manager can be expected to do more easily than estimate the qualitative weighting between types.

Schedule this task The scan of all the tasks will select the best activity to schedule. The obvious next step is to schedule it in the most favourable time slot. Three sets of data should be taken into account in this determination. l l l

the profile of the cost requirement of the tasks scheduled so far, the range between the earliest and latest times of this task, the cost requirement of this task.

When trying this method for the first time, most students can intuitively locate the best position within the given time range. This can be stated as the slot that gives the lowest overall profile at this time. A mathematical statement of this is, however, a little longer. The earliest and latest times that constrain the scheduling of the activity give a finite set of possible start times. The activity is scheduled at each of these start times in turn. For each of these possibilities, the requirement for each time period is squared and then summed. The position that gives the lowest sum of squares is then chosen as the best schedule time for this activity. This method of squaring forces the scheduling of new jobs to avoid peaks of the previously committed cost. In other words, it is always looking for the valleys in order to fill them.

Adjust times Although there may not be a network in the true sense, there will often be a set of imprecise linkages. The activity that has just been scheduled may have implications on the activities yet to be done. For example, if it was scheduled at its earliest time, then anything that had to precede it may need its late time adjusting. This is not a major task to do quickly with a pencil and rubber for the set of activities in question. Should there be any further activities to schedule, these four steps are repeated. If some types of cost were ignored and all jobs involving the important cost type have been scheduled, then the process can be continued with the next category. In this case, all the tasks that have already been scheduled will be considered as critical in the initial stage of the process.

Vol 1 No 4 November

1983

OBSERVATIONS

ON THE METHOD

A result of using this approach in the early days of the study is that a cost profile is produced as an output rather than as an input item. Naturally, the cost requirements will vary from period to period. The overall shape of the final cost curve will help the manager in setting his own level of cost flow. He may, for example, ignore a peak in one period knowing that he can use the valley nearby. Normally, computer allocation algorithms need the maximum level to be supplied as a data item and also the rules to be observed when it tries to schedule over this limit. Although the smoothing algorithm has been computerized in at least two computer programs, it is not necessary for it to be done in this way. Used manually, it accelerates the manager’s understanding of his project by highlighting the sensitive portions. It also allows him to try alternatives quickly. In one exercise, the time required for determining a provisional budget was cut from two weeks to two days. The benefits of this are numerous, but the prime one must be that more real discussion can take place before a commitment is made. If the manager considers the categories of cost in turn using this method, it is only necessary to produce an ordered list. The sequence of cost types on this list will almost certainly not be the same as the sequence of these costs when looked at from the company viewpoint. The most important cost should be the one that is necessary to keep the project progressing. It can be tempting to pick a cost that occurs on one or two tasks only. In scheduling these tasks, the manager will then throw away much of the flexibility that he needs later in the process. Some managers are perturbed by the idea of needing a late time for the tasks. The scheduling process uses that as a boundary but will not necessarily delay tasks up to that date. However, the more flexibility is given to the project by extending the late times, the lower will be the overall cost profile.

MULTIPROJECT

ENVIRONMENT

It is sometimes necessary to consider the proposed project in the light of the currently committed load. For example, during the sales campaign for a new contract, it is no good quoting unrealistic dates and costs. An early date would commit the company to a cost rate greater than it could handle and a late date would upset the client. In practice, starting with the costs committed to other projects, it was found that a very competitive and realistic estimate was supplied to clients by using the smoothing process to the new project. The sales team can normally supply all the information needed. Even the late times are usually obtained early in the discussions. To use the procedure, committed cost is taken as the starting profile and then the four steps outlined previously can be done for the proposed project. The exercise can be repeated using a number of variations on the proposal so that management can evaluate the best strategy to take in the negotiations. The sales team will also have an indication of the implications to the

195

company and to other projects of the various requests coming from the client. This gives them an advantage over competing organizations.

REFERENCES 1 Tulip, A ‘Resource Proc.

smoothing

Internet Conf Stockholm

theory and practice’ (1972)

Internet 80s -

dP

the supremely flexible management system

*

easily adaptable to users’ specific requirements

*

machine independent avoids conversion problems

*

integrated planning/financial modeling capability

*

interactive input - prompted - menu driven

*

advanced network plotting interactive graphics

CRD

Computation Research and Development 12-15 Dartmouth Street, London SWlH Telephone: (01) 222 9822

196

Arthur Tulip graduated from University College, London, in 1959. He worked for a number of years on simulation methods in navigation and atomic energy before joining Honeywell, where his work consists of advbing customers in the use of management techniques and developing, maintaining and supporting the computer programs associated with them. He was involved in the formation of the UK Association of Project Managers and in the past has held the offices of both secretary and treasurer for the Association.

Ltd., 9BL

Data Processing is the international journal for computer managers

DATA PROCESSING is specifically tailored to meet the business and professional interests of managers of computer installations and company executives who procure computing equipment. It contains news and features on industry developments, DP management issues, advances in technology, applications and new products and services.

Further details and sample copy can be obtained Christine Mullins Butterworth - Journals PO Box 63, Bury Street, GU2 SBH,

from:

Scientific Limited Division, Westbury House, Guildford, Surrey UK.

Telephone 0483 3 126 1 Telex: 859556 SCITEC G

Project

Management