Two examples of plant design are described and an abstract model of the process of design is suggested. The model is used to discuss the search for possible solutions, the strategies for their examination, and the rules for choosing between them. It seems likely that the model applies only to problems requiring novel solutions and not to those for which the form of solution is known, but the choice of parameters to meet conflicting objectives is difficult.
Contemporary models of research and development are incomplete in that they ignore the many reappraisals and budgeting decisions that occur in the time between a project's proposal and its commercialization. The sequential decision aspects of project budgeting are particularly important since 1) the research expenditure is usually an order of magnitude less than the irrevocable investment for commercialization and 2) an allocation to a project today does not presuppose continuation of the project into future periods. The research and development budgeting problem is structured to take into account the sequential decision characteristic. Utilizing the technique of dynamic programming, methods are developed to determine optimal project budgets when the aggregate research and development budget is either constrained or unconstrained, These models also suggest a rational explanation of the patterns of project expenditures over time that one observes in practice. Finally, some of the shortcomings of the developed methods which inhibit their practical application are discussed.
The structure of an R & D project is described in terms of an Effort-Distribution Array. A logistic model is presented for the cumulative times series of effort devoted to a project, and problems of using such curves for prediction are discussed. A set of rules is given, which, with the aid of computer simulation, can be used to generate a project schedule. The consequences of several combinations of rules and restrictions are examined.
Several writers on the history of science have argued that changes in the institutional location of science will be accompanied by several major changes in scientific attitudes and values. A sample of approximately 30 per cent of the working scientists in several disciplines was interviewed in the Minneapolis-St. Paul area. Those interviewed worked in government, industrial, and university laboratories. A number of indicators of scientific attitudes and values were examined.
Real-time digital systems are largely a technical innovation of the past decade, but they appear destined to become more wide spread in the future. They monitor or control a real physical environment, such as an air-traffic situation, as distinguished from simulating that environment on an arbitrary time scale. The complexity and rapid variation of such an environment necessitates use of a fast and versatile central-control device, a role well suited to digital computers. The usual system will include some combination of sensors, communication, control, display, and effectors. Although many parts of such a system pose no novel management problems, their distinguishing feature, the central digital device, frequently presents unusually strict requirements for speed, capacity, reliability and compatibility, together with the need for a carefully deisgned stored program. These features, particularly the last, have implications that are not always foreseen by management. An attempt is made to point out specific hazards common to most real-time digital systems and to show a few ways of minimizing the risks associated with them.
The use of the probability of a successful proposal in manpower forecasts is discussed. Formulas for the expected requirements and the standard deviation are presented. An illustrative example of the use of these formulas is given.
The term ``engineering management'' is intended to be roughly descriptive of the management of technical matters, merely as contrasted with such other areas of management as financial, sales, personnel, etc. Decision-theory, which seems not to be a theory after all, is presented in broad survey as a collection of analytical tools and methods, and an effort is made to see if these have any particularly happy applications to engineering management problems.
The evolution of general types of multidisciplinary aerospace systems is considered during three phases: Research, Engineering, and Sub-system Integration. Management methods and personnel utilization techniques most applicable during these phases are discussed. This paper is not concerned with management during the operational use of aerospace systems. In this context it does not discuss technical management beyond the early part of the Acquisition Phase as defined in Air Force Regulation No. 375-1. It is indicated that technical management must be considerably limited in degree during the Research Phase where ideas and concepts, based to a large extent on individual talent and inspiration, are of primary importance. Management techniques must be applied with discretion during the Engineering Phase. Here the breadth of technical coverage required for subsystem engineering is such that teams or groups of technical experts provide a useful tool. The talent and initiative of the individual team members is quite important during this phase. Scheduling is also important and cannot be overlooked. A maximum degree of technical management is required during the Subsystem Integration Phase of the over-all system evolution. Here programming or scheduling is of primary importance. Procedures such as PERT are useful management aids and should be applied as much as practicable during this phase.
The moral values associated with scientific research were studied by means of interviews with 57 academic researchers of faculty status. The areas examined were freedom in research, impartiality, suspension of judgment until sufficient evidence is at hand, absence of bias, diffusion of information, and group loyalty. Two dimensions of creativity were measured in the respondents: rate of publication and strength of motivation toward research as judged by peers. Neither of these measures was significantly related to acceptance of the classical position in any of the six areas of value. It is concluded that the classical morality of science is not associated to any important degree with productive research.
The question of whether PERT is of significant value to project managers will not be resolved until the technique is understood to be a tool for a newly recognized function rather than a new tool for an old function. The newly recognized function is decision making as opposed to the old function of planning and control. The need for the performance of the new function arises from the dynamic and complex character of modern R and D projects. PERT as a decision-making tool can best be understood within the framework of the decision-making process. The decision-making framework indicates that assumptions must continuously be made in four areas. The information needed to develop these assumptions constitutes the information necessary for decision making. The information obtained by PERT provides a portion of this information.
Many indices of profitability for research and development (R and D) ventures have appeared in the literature. Most of these are calculations of the estimated economic value of a project, if successful. Their greatest utility has been in the development-type project where the probability of success has been relatively large (pr ≧ 0.25). This paper considers the optimum utilization of a scarce resource, professional man power, among many alternative research projects. Other parameters and restrictions of the model are: the economic value of a successful project, the probability of success, the man-hours required per test or screen per project, the total available man-hours, the cost per man-hour, and the available raw materials (compounds, components, etc.). These factors are used to construct a linear programming model. The solution indicates the optimum allocation of professional man power over the most attractive projects to maximize the return to the corporation. Further aspects of the model are discussed.
A summary is presented of a study of decision-making in the area of research budgeting and project selection. Data were obtained from three major chemical companies. The objective of the study was to determine if scientific analysis could be employed to derive objective and quantitative procedures in these areas, currently dominated by a mixture of intuition, judgment, and experience. A two-fold classification of R and D is used- ``product research'' and ``process research.'' A solution is presented for the general R and D budgeting problem, involving a budgeting model. A computational procedure for this general problem is illustrated. The special problem of allocation of the budget to specific projects is treated briefly in this paper.
A survey was made of the role of the accountant in the control of Research and Development, primarily through a mailed questionnaire and a limited number of interviews. Data from 51 responding companies in a variety of industries include locus of responsibility for R & D budget preparation, basis for the annual budget, frequency of revision of budget, and other factors.
The anticipated total cost of a complex project is placed on a simple statistical quality control basis, through a natural extension of the PERT system of management control. The method is designed to provide management with the earliest possible warning of a potential budget overrun, consistent with the uncertainties inherent in the basic data. The procedure, when applicable, can readily be made to augment standard PERT programs, thus providing an integrated package for the simultaneous (probabilistic) control of project time and project cost.
The evolution of network planning techniques is described. The main lines of development and the variations around them are indicated. Those discussed include: Critical Path, PERT, PEP, CPM, LESS, PACT, SPECTROL, and SCANS. A table is presented of areas within industrial or military organizations where such techniques are in operation or appear suitable for application.
Two costly mistakes are frequently made in the management of the support program for a major weapon or space system: 1) Considerations of support requirements are neglected until it is too late to accomplish a reasonably scheduled minimum cost program. 2) In an ardent effort to avoid the first mistake, hardware and software programs are promulgated before support concepts or prime system design have been adequately frozen. The purpose of this paper is to illuminate these pitfalls and indicate how they may be avoided.
The name SCANS is an acronym derived from Scheduling and Control by Automated Network System. The system is based upon the network techniques developed for PERT and might well be considered to be one of the many versions of PERT which have evolved since its introduction. Although SCANS is described as an automated system and as a man-machine system, it is the men in the system who perform the most important functions of planning, decision making, and control. The machines only perform the functions which would have to be delegated to clerical and computational personnel if the machines were not available. SCANS is not intended to be a substitute for the human mental process. It can only hope to facilitate this process and improve its effectiveness by providing the best information in an efficient format at the time it is required.
Realization of the full potential of critical path techniques involves six essential aspects. These are 1) network planning, 2) time and resource planning, 3) optimum scheduling and allocation, 4) progress reporting, 5) analysis and control of progress, and 6) statistical analysis to improve time and resource planning. All of these aspects are not currently being employed and thus full benefits are not accruing to the user. The principal reason for this is the lack of data on which to base time-cost trade-off relationships. This in turn is due to the present structure of accounting and control systems which do not relate costs with the work accomplished. Several efforts employing enumerative cost models are currently in existence or under development which largely overcome the shortcomings of present accounting and control systems. The enumerative cost model collects cost by network activity and compares actual expenditure and schedule progress with that planned in order to determine program status in an accurate and unambiguous fashion. In addition, this type of model will provide the raw materials for construction of the time-cost relationships required for optimization provided that the non-comparability of activities problem can be overcome.
This paper discusses the environment in which analytical methods of program evaluation and review (PERT, PEP) are most useful. The conditions under which these techniques are most applicable are discussed. A number of limitations are pointed out. Aspects of the cost of using these techniques are discussed.