DYNAMICA DIGITAL ARCHIVE
Abstract: Since the publication of Industrial Dynamics there have been very few published accounts of the industrial application of the methodology. Indeed, the main use of the technique has been in large models of socio-economic problems. This paper discusses the practical application problem, and describes several cases of real industrial studies. It puts forward general criteria for successful practical studies, which apply to socio-economic as well as industrial problems.
Abstract: Problems concerned with capacity acquisition are discussed. A simple production model is considered to explain how System Dynamics Methodology could be used in planning capacity acquisition. The consequences of a discrete type of capacity ordering policy which reacts only to capacity requirements above a specified minimum value is presented; capacity requirements are computed by linear extrapolative forecast. Effects of proportional, integral and derivative control on capacity are compared.
Abstract: System Dynamics Methodology (S.D.M) has proved its applicability in several types of policy design (Coyle, 1974). This paper attempts to explore its usefulness in the formulation of financial policies with an application to cash management. After a brief introduction dealing with the difference between SDM and simulation techniques, the structure of the model is developed and described, followed by a discussion of the potential of the proposed model for long-rang cash planning and control. Finally, a short summary is given of other possible applications of this methodology in the financial field.
J. A. Sharp
Abstract: The reasons for being interested in the effects of perturbations on System Dynamics models are discussed. Various important types of perturbations are indicated. A first order model of the effects of system perturbations is then discussed and its implications considered.The possibility of classifying models by their response to perturbations is then considered in the light of the guidance they may give as to the utility of the model.
Abstract: In the past years many developing countries tried to establish or expand a national merchant fleet. However, there seems to exist a great discrepancy between the statement and the implementation of their shipping polices. This is mainly caused by the scarcity of their financial resources and the complexity of the maritime industry (e.g. shipping, shipbuilding, ports). Some of the secrets of the maritime industry have been subject of detailed research. However, the dynamics of the whole industry has never been studied. This paper presents a simple system dynamics model, that attempts to illustrate the dynamics of the maritime industry – in particular the national fleet sector – in a hypothetical developing country.
Abstract: The economic difficulties facing a small dairy farm have been modeled in a System Dynamics format which is validated by reference to a particular farm. The policy area of calving schedules has been investigated in a search for alternative dynamic behaviour in the level of feedstuff-debt which seems more likely to aid survival. Simulations have been produced for the years 1974-79 for a variety of calf schedules, two of which are compared with the actual policy of the modelled farm.
Abstract: This paper presents and discusses two tentative models of the U.K. natural resource usage system.
Abstract: Some examples showing the way that such forecasting characteristics as bias, errors, sampling and delays affect the dynamic behaviour and performance of corporate systems…
Abstract: A basis of co-operation in System Dynamics research is a common language for system managers and investigators. The system Dynamics model may be this language.
A form of structural model is proposed, taking as its base Forrester’s flow diagram notation. Some features of structural models are:
They present the whole of the structure in a compact clear way.
Business interpretation of the model is easy, owing to its verbal components.
Use of the formally defined elements enables direct transformation of the structural form of the model into a mathematical form.
Abstract: The techniques of Demography are well developed but their application in simulation modelling has been somewhat restricted. The classical techniques were developed in an age of manual computation and the direct application of these techniques to dynamic simulation model is somewhat cumbersome. This paper sets out the basic dynamics of population systems, its simulation and the display of the vast quantity of simulation output in such a manner that the behaviour of such systems can be easily comprehended.
Abstract: This paper attempts to explain a result that has been noted in several system dynamics studies, namely that the performance of a properly designed system that depends on a forecast is relatively insensitive to forecast error. This is shown for continuous linear systems to be a consequence of the design criteria used in system dynamics modeling.
The role of Perfect Forecasts in such systems is discussed and the means by which normal system dynamics design procedures produce systems that are sensitive to forecast error elucidated.
Abstract: The balance (or imbalance) between supply and demand for cargo-carrying capacity has a very important influence on the behaviour of ship-owners and operators. Conversely, the behaviour of the decision-makers is, itself an influence on the supply/demand ratio. This paper shows that the shipping industry, in particular the dry cargo market, behaves as a dynamic system and a simple model of this system is described.
Abstract: Given a simple system with orders flowing into a backlog at a varying rate of OR units per unit time and being executed at a varying rate PR units per unit time to determine the length of time τ which has been spent in the backlog by the order currently being executed.
James C. Hershauer, William A. Ruch
Abstract: “Productivity” is an oft-discussed and vitally important concept; however, there exists a dearth of models concerning productivity and the factors which influence it. A fundamental model of “organizational productivity” depicting the dynamics of the system is presented to provide a pragmatic and pedagogical structure for analyzing productivity.
A.K. Ratnatunga, J.A. Sharp
Abstract: The utility of linearization and order reduction of System Dynamics models is discussed.
Two methods are presented for linearizing a System Dynamics model and the results of implementing them on an actual model are given.
Abstract: The demand for new ships is known to fluctuate widely, due to both external political factors and to market forces. In these circumstances the pricing strategy adopted by a particular shipbuilding company has a profound effect not only on the profitability of the company, but also on its growth prospects and stability of operation.
This paper suggests that the problem of designing a strategy for a company may be regarded as a problem in the design of a suitable management control system, set up to meet given management objectives. This type of system design problem may be most effectively studied by means of the techniques of system dynamics.
Abstract: Bias, intended or accidental, can, and certainly does exist in forecasts, and the effects of such bias on corporate performance is studied in this paper. Using system dynamics simulations, experiments were performed to examine the sensitivity of a complex crude-oil supply system to forecast bias, and to find bias factors which minimized a simple cost performance indicator. Forecast biases of the order of ±10% produced variation in cumulative expenditure on the tanker fleet of only about ½ % - a very robust system. Minimum cost was obtained with very unlikely values for biases in some forecasts. Though not definitive, these results suggest that managers should think again when blaming poor system performance on the forecasting function.
Abstract: In System Dynamics work concerned with modelling whole industries it is frequently necessary to model the aggregate ordering policy of a group of companies. The purpose of this note is to derive suitable rules for the case of an industry in which all the companies operate a Cyclic Review or (s, t) system.
Abstract: System dynamics models can be heuristically optimized but the process of optimization usually implies that an objective function has already been selected.
This paper demonstrates the use of a new tool, called SDRDYN, which allows a model builder to experiment with various objective functions from a time-sharing terminal. A solved application example with artificial decision parameters indicated that both a control approach and an economic approach were needed for the best objective function of those attempted.
Abstract: The reasons that led to the identification of a need for a package that could carry out the dimensional analysis of a System Dynamics model are discussed.
A dimensional analysis package has been developed for use with DYSMAP compiler and example of the way it is used are given. A preliminary assessment of the usefulness of such a package to System Dynamics model is also made.
C. J. Stewart
Abstract: As a result of work carried out on the Dynamic of Inter Library lending, a new model to aid strategic planning at the national level has been developed.
Further development and investigation are necessary to bring this project to fruition resulting in a tool capable of aiding policy formulation and assessment in the economics of the national system.
Abstract: The paper shows the connections between system dynamics and the systems approach. The basis is a comparison between features of the systems approach and features of classical system dynamics as applied in the Sprague Electronics Company study.
R. G. Coyle
Abstract: In some industrial models the firm maintains a number of products at the market and the average age of the product line is one of the factors affecting the firm’s sales. Typically …
Abstract: This is a paper of a more technical nature, and it examines the effects of various planning horizons and of erroneous cost on the model that was presented in the first part of the paper.
The study demonstrates perhaps counter-intuitively that mixed policies were more economical than more logical ‘pure’ policies of the traditional system dynamics type.
Abstract: The aim of this paper is to make a brief description of a model of the policies for achieving an adequate level of capacity in the Electricity Supply Industry. The purpose of the model at this stage is to reproduce the effect of those policies and of exogenous events on the performance of the industry.
This problem is an important one because the industry has found itself with a considerable excess of installed capacity for meeting a slackening demand.
Dante H. Montaldo
Abstract: This paper describes the main features of a System Dynamic Model of a mining firm which manages a single underground. The model includes mining, development and milling sectors, together with cash flow and pays special attention to the feedback relations between all components. In particular it shows how the management policies of individual company sectors interact and affect the overall system performance.
The aim of the work is to demonstrate that analysis of a company’s structure and behavior when subjected to the perturbing effects of strong and frequent fluctuations from the metal market sector, can lead to recommendations for improved mine management policies. The model is applicable to existing mines and could be a major aid when analyzing the feasibility of new projects.
Abstract: The reconstruction of the DYSMAP compiler undertaken in the last year has covered three main areas:
Edward B. Roberts
Abstract: This paper illustrates the effectiveness of System Dynamics modeling in treating apparel manufacturer. The study was one part of a several years effort by my consulting firm, Pugh-Roberts Associates Inc., sponsored by the Textile and Apparel Technology Center of the U.S. Department of Commerce. The contracts developed systems analyses of a board variety of textile and apparel industry problems and produced trial applications of the techniques to one major textile company and one medium-sized apparel firm.
In conducting the case study of the cooperating apparel company we began with a general familiarization phase, including interviews with company personnel and the study of operating data over a several year period. During this first phase ideas were formulated as to how individual management practices might be related structurally to produce the company trends observed. As certain aspects of the firm began to appear influential on the firm’s overall performance were translated into a System Dynamics computer model. The resulting computer model was then tested using computer simulation to see if the hypothesized model structure produced the same type of behaviour that was actually observed in the firm’s history. In particular, the case study model was tested against the company’s performance over the previous five years. Once the model produced the basic underlying trends and changes that were apparent in the real situation, it became a useful tool with which to understand the problems of the apparel firm.
R. G. Coyle
Abstract: The typical International Mining Company (IMC) is a cross between a merchant bank and a consulting firm and usually has only a minority equity stake in its operating companies, which are often large corporations themselves.
It is often argued that what is good for the subsidiaries is good for the IMC but this has been called into increasing question because of the different corporate objectives which may be pursued by the IMC and the operating companies, especially in the face of an extremely unstable economic environment.
The article describes a model of a mining group which shows how conflicts can arise between Head Office and subsidiaries and how what is good for the IMC is not necessarily good for the subsidiaries and vice versa.
Traditionally, the mining industry has not engaged in corporate planning but has tended to react to investment opportunities as they arise. The model looks at some obvious planning and shows that they may be harmful to the corporate entity.
Abstract: This report describes the principle features of a computer model, BOOM1, designed to simulate the “boom town” impacts that may result from locating large energy facilities near small, isolated, communities. Model outputs include population, permanent and mobile homes, public facilities and municipal debt, local property tax rate and state transfer payments, construction work force, construction productivity, and retail and service facilities. The model can be used to simulate the behaviour of these variables over a time span covering the pre-boom, construction, operation, and retirement phases of an energy project. The model runs may be interrupted in any year to simulate the effectiveness of assistance measures that have been proposed by national, state, local, and energy company officials. In these ways, the model can be used to help deal with boom-town problems and so minimize social and economic disruption from future energy construction projects.
A. J. Taylor
Abstract: In the construction of any type of model there is the problem of identifying and justifying relationships between variables. For certain types of models the structures of which are well-defined this problem reduces to one of determining the quantifiable aspects of relationship (for linear programming models this is often the case).
However, in the construction of a tailored system model, as with System Dynamics, the problem is two dimensional comprising the aspects of identification as well as quantification. When the system under study originates in an area of research the resolution of the problem is often very difficult if the model-builder is inexperienced. The purpose of the present paper is to demonstrate the resolution of the “problem” by describing an example from the authors own research.
Abstract: In order to enable optimization of system dynamics models three hierarchies are proposed: I. Model, II. Optimizing procedures, III. (artificial) Intelligence. Using these levels two iterations are proposed. The optimization procedures of level II leads to “small” iteration. When this has ended a transition is made via level III to other optimization procedures – “grand iteration”. This hierarchical approach leads to a division of labour in system dynamics, as model structuring can now be made an automatic process.
R. G. Coyle
Abstract: In modelling corporate problems one often needs to incorporate a relationship between a variable such as Advertising Spending and the resulting Market Share, it being assumed that there is a connection. There are various approaches to this, for example Coyle (1977) relates Market Share to Market Share Spending (which includes advertising, technical service, etc.) by a table function. The table function is calibrated against the firm’s existing Market Share Spending and the assumed ‘shelf life’ of Market Share, i.e. the rate at which it would decline if no further spending took place, and hence the rate of spending needed just to prevent that decline. There is, of course, a time delay to reflect the perception time of the customers.
Abstract: A series of pilot studies have indicated that a computerized world game can have a significant effect in increasing the players’ sense of control over their future. These studies were carried out with students from the fifth grade through graduate school. These optimistic results have led the author to suggest that a complex game, which uses the computer as a computational tool, and where the students compete against “nature” rather than each other, are important elements in game design. Moreover, a game in which reasonable decisions produce satisfying results will tend to increase the sense of future efficacy among its players. Sophisticated System Dynamics models are often excellent game material as well as motivating teaching strategies.
Abstract: It is sometimes useful in comparing one simulation run with another to calculate a performance Index for a System Dynamics Model. The construction of such indices using a weighted combination of Final Values and Instability Penalties is discussed. The method is illustrated using an example from a study of Mining Companies carried out by the author.
Abstract: The different approaches to measuring parameter sensitivity in system dynamics models are discussed and a practical evaluation attempted. For certain types of model behaviour, a procedure is proposed for generating performance measures which converge to values independent of simulated time. Using these types of measure in conjunction with local sensitivity analysis, most likely values of the performance measures may be estimated and then used to calculate the most likely values of performance improvement following system redesign.
Abstract: A problem frequently encountered in system modelling is to ensure a negligible response to disturbances of a specific period. This need often arises, for instances in production systems where orders show strong seasonal components of 12 months, 6 months, 3 months, etc. In other systems it may be similarly required to reduce response to Business Cycle fluctuations with a four to five year period. The standard way of achieving this in a System Dynamics is to make use of the SMOOTH function to reduce the effect of disturbances which the system should not respond to. Unfortunately the SMOOTH function is a rather crude instrument for this purpose and at least for systems subject to seasonal variation tends to lead to worse performance than is necessary.
Roger I. Hall
Abstract: It has been observed that decision makers with complex policy problems grossly simplify the mental maps that they use to structure their decisions. The individual and group processes that are brought into play in combatting complexity and uncertainty in the policy domain can lead to deficiencies in a decision making group’s collective map of causality characterized by “balance”, lack of feedback loops, beliefs conditioned by “superstitious learning”, and resistance to change. Whereas these human mechanisms of uncertainty resolution help decision makers keep track of a large number of causal assertions, they can also lead to systematic pathologies as organizations become trapped in policy cul-de-sacs or are victims of policy side effects due to the system dynamics of unseen feedback loops. Three simple techniques based on influence diagramming are suggested for constructing causal maps or models to aid this kind of policy analysis. They are all amenable to simulation and analysis by computers, but the major difficulties are gaining the confidence of senior managers and access to information. The models so constructed do initially make decision making more difficult because of the feedback loops and etc. uncovered. However, it has been demonstrated that they can lead to a search for more creative decisions.
Charles H. Braden
Abstract: A highly aggregated model with eight principal levels is constructed in order to investigate the concept of a geeignet (appropriate) population for a society. Although expressed in terms of an agrarian society in which technology supports agriculture, the model is intended to have broader implications. The model incorporates a price structure and a decision procedure to adjust production mechanisms so as to minimize the marginal production cost. This report is mainly concerned with equilibrium conditions for fixed population levels, but the model has been developed to demonstrate stable dynamical behaviour. Selected results are discussed which exemplify alterations in societal conditions with alterations in the parameter set that characterizes a society.
Ali N. Mosleshirazi
Abstract: One of the factors influencing decisions on any new Capital Expenditure to be incurred in the future is the availability of financial resources up to the Planning Horizon to finance such investment. This is determined after setting aside the resources needed for projects already commissioned but not fully paid for.
This paper presents a method of calculating expenditures on already commissioned projects up to the Planning Horizon through calculating coefficients representing the amount of progress that will be made on those projects. The method assumes progress stage payments which is customary both in industry and practical System Dynamics modelling.
J. A. Sharp
Abstract: This paper deals with the ways in which the ideas of Optimal Control Theory are relevant to System Dynamics. The salient features of Policy Design and Parameter estimation problems in System Dynamics are defined. The major features of Optimal Control Theory are given and it is shown how they can be applied to these problems. The computational aspects of Optimal Control methods and their link with Dynamic Programming are discussed.
Abstract: We have had System Dynamics available to use for 20 years. It seems to fulfil a need which is not, and cannot be met by standard planning and programming approaches, namely that of providing for the concept of controllability when the unforeseen happens. Despite this it is an unseen technique.The explanation seems to lie in the areas of previous applications of the method, both good and bad, and the extent to which the need it satisfies was actually felt. A review of applications helps to characterise the System Dynamics method in various circumstances and leads to ideas for its improved application and further development.
Charles H. Braden
Abstract: A decision procedure is described which has been employed to control the evolution of a system dynamics societal model by minimization of the marginal production cost for food. Although explicitly discussed within this framework, the procedure should have more general applicability when optimization of the marginal value of some controlling figure of merit is desired. The basic decision procedure is supplemented by several refinements in order to modify the dynamical behavior of the model.
Tero Ansio, Erkki Mattila
Abstract: Three alternative strategies for pollution control are studied. From the results of this paper one of these is recommended as the best strategy. The method of this study is system dynamics which has been used in three ways. First, an analytic approach has been used in formulating some decision rules for the system. Secondly, the heuristically optimizing SDRDYN has been used so that the optimal path of one SD variable has been found. Thirdly, SDRDYN is used in sensitivity analysis when strategies are compared.
We proceed as follows: First, three different strategies for pollution control are introduced, then the model and analytic solution of production and pollution is presented. In the third section the model for finding the optimal path in system dynamics is introduced and after that the results of simulation are given. The sensitivity of the results for parameter changes is studied and conclusions and suggestions for further research are given.
David A. Lopez, George W. Watson
Abstract: A number of different methodologies have been applied to the study of manpower planning systems. This paper presents and develops a computer simulation approach through the use of System Dynamics in study of the manpower requirements of the U.S. Marine corps. A number of different recruitment policies were studied to observe their impact on the total manpower system. In addition, the ramifications of the economy, the unemployment rate, the socio-biological trend of a reducing population segment of recruitable age, and the increasing political uncertainty on the combat effectiveness of the first line combat infantry force are studied.
J. A. Sharp, R. M. Henry
Abstract: This paper illustrates the use of two methods devised by Ziegler and Nichols to design feedback controls for System dynamics models. The methods were originally devised through experiments with hardware systems. For this reason, they give good basic designs with very little effort.
R. G. Coyle
Abstract: This paper deals with the modelling of accounting depreciation on both a historic cost and an inflation adjusted basis.
Abstract: The purpose of this paper is to describe work undertaken, within the System Dynamics Research Group at Bradford Management Center, into the development and application of a System Dynamics model of an international mineral industry; aimed at providing a basis for analysis of the inter-relationship of supply, consumption and price over time, under different production and stocking policies. A review is presented of some of the major characteristics and problems associated with such industries. This is followed by an outline description of the model and the way it has been applied to examine various supply responses, in the light of changing consumption patterns. Demonstration results deal with model validation and with analysis of international buffer stocks and conflicting supply policies on the historic situation in the copper industry.
George P. Richardson, David F. Andersen
Abstract: System Dynamicists must find innovative ways to combine teaching and research in System Dynamics, using teaching to draw a pool of interested students into the subject and using research to teach advanced modeling techniques to intermediate and advanced students.
Four formal designs are presented for combining research and teaching interests. The pitfalls and advantages of each of these strategies are discussed and examples given of the results that may be expected from each approach. Three critical classes of considerations are found to influence the success of joint research-teaching programs: faculty roles and responsibilities, student readiness, and project selection and management. Each class of considerations is discussed in detail with reference to concrete examples.
Graham W. Winch
Abstract: A dynamic simulation model, using the methods of system dynamics, has been developed to study how an individual Building Society can control its cash flows in order to produce a smooth flow of mortgage funds. Of particular interest is the role that forecasting plays in the decision process.
It has been concluded that the use of a forecast of net deposits by investors with a five month time horizon can improve performance significantly over existing practice of using current values only. The improvement persists even when large errors are present in the forecast.
Abstract: This paper is concerned with the use of “Loop Analysis” in the evaluation and validation of System Dynamics model. The part that loop analysis plays in the overall validation process is described, and it is shown, by reference to the analysis of a complex model, how the model’s behaviour may be investigated. The analysis proceeds by identifying some of the elements behaviour such as damping, phase-shifts and gain/delay factors, and, in the process, highlights those feedback loops that are pseudo-positive and/or unconformable. In addition to the validation aspect of the use of loop analysis, there is a valuable contribution towards the knowledge and insight of the model-builder, enabling him better to identify areas of the system which may benefit from re-design.
Abstract: System Dynamics explores the use of scarce resources to achieve some goals in a feed-back and possible feed-forward framework. Information (including orders), material, human resources, fixed capital and money have been the resources to choose from. Information network, however, should always be included in an SD model.
E.F. Wolstenholme, R.G. Coyle
Abstract: It is sometimes argued that discrete and random events are very hard to model in the DYNAMO and DYSMAP syntax, and that the logical facilities in these languages are inferior to those of FORTRAN. This paper demonstrates, by examining a rather complicated modelling problem, that neither of those propositions is necessarily correct. It is shown that, once one understands the underlying problems, DYSMAP is usually far more flexible, efficient, and friendly to the user than is FORTRAN.
A subsidiary aim of the paper is to demonstrate how equations can be built up, piece by piece, to produce required dynamic behaviour, rather than simply emerging somehow from the mind of the modeler.
Pratap K.J. Mohapatra
Abstract: This paper discusses an algebraic and a diagrammatic method to highlight the structural equivalence between Control System Theory and System Dynamics. An analogue scheme of visual representation to SD models is suggested which makes it possible to express the SD model in the form of a vector-matrix state differential equation. An obvious duality of this representation with the signal flow graph helps computing the elements of the system matrix.
Pratap K.J. Mohapatra
Abstract: This paper classifies nonlinearities in System Dynamics Models into three types. Procedures are indicated to resolve these nonlinearities and transform the models to linear ones.
Jack William Jones, E. Leigh Secrest
Abstract: This paper reports on the development of a system dynamics model to enhance the economic decision-making processes involved in the identification, evaluation, and selection of chemical-based enhanced oil recovery projects. Model structure is described and an example illustrates usage of the model in project evaluation activities.
G. D. Craig
Abstract: This paper describes a methodology for constructing financial planning models using a computer package developed by the author. The package is designed to provide general management with a flexible and easy-to-use modelling facility capable of supporting the development of dynamic simulation models of both financial and non-financial systems.
A brief summary of the nature of the methodology, and its underlying concepts is presented initially, followed by an application. First a simple form of financial model is developed, followed by an extended version of the same model. The appendices contain details of the two computer programs used in setting up the models, together with output from a run of the simple version.
Abstract: With the recent advent of new technology, involving the extensive use of micro computers, for centralized monitoring of the state of underground conveyor belts and bunkers, the scope for total automatic, real time control of mineral clearing systems has been enhanced. This paper outlines the result of research aimed at providing compatible advances in methods of designing total system control rules, by which information collected can be fed back to correct the observed system states. The technique used is that of continuous feedback simulation (System Dynamics) and the way this has been used to model an underground conveyor belt system incorporating realistic production generation patterns is described. The model is used to test out and improve the design of alternative policies for bunker discharge rates under a wide range of system parameters. The final policy evolved in this way has general applicability and is shown to maximise conveyor belt utilisation, be independent of limitations in the maximum bunker discharge rates and to require only the monitoring of bunker levels. Finally, the model is used to quantify the benefits of such improved control in terms of savings in physical capacity required to obtain maximum system efficiency.
Abstract: The purpose of this paper is to investigate the different parts of a paper company which affects its dynamic performance.
The demand, which behaves in a cyclic manner, and the foreign competition have resulted in the decline of the market share of U.K. paper companies over the last ten years.
The problems of plant utilisation level and plant efficiency are of great importance, it is argued that such problems may be regarded as a result of policies followed by management.
A system dynamics model for a hypothetical paper company which uses only waste paper as a raw material, has been constructed and the dynamic behaviour of the company is examined under typical historical managerial policies. The behaviour of the model is shown to give rise to many of the phenomena observed in the real world such as production rates, capacity levels and plant efficiency.
By designing new policies, the model is leading to greatly improved dynamic behaviour.Finally the practical problem of implementing such a study is examined, and the benefits of it are discussed.
Abstract: Broadly speaking, the essence of a system is its complex connective structure and its dynamical behavior. Many conventional analysis, however consider these fundamental features in a restricted way, and, in so doing, they undermine our ability to fully comprehend the system of interest, and, perhaps more significantly, they reduce our ability to control it in a meaningful way. Given particular interest in the complex interconnections between a system’s sub-systems, for example, it is somewhat surprising that techniques, such as conventional multivariate statistical approaches founded on constrained (often linear) functions (which are a special type of relation), are frequently employed to analyse this characteristic. Unfortunately, by their very nature, these taxonomic procedures destroy much of the structure of interest. The basic argument is that theoretical deductions may be as much a result of the specific approach and representation applied as of the system’s features. Specifically, attention is drawn to the fact the employment of linear functions is inhibiting.
Jean D. Lebel
Abstract: In a companion paper published in the Simulation 77 Proceedings (Lebel 1997a), we hope to have provided the reader with an elementary background in System Dynamics.
In this paper we would like to try to put a new light on
the “raison d’être” of System Dynamics
its underlying basic assumptions
the question of uncertainty
the problem of validity
R. Joel Rahn
Abstract: The development of a system dynamics model for a long-range planning group in a large electric utility is described. A distinction is drawn between decision-making and decision-preparing models. Some conclusions regarding the implementing process for a decision-preparing model are proposed.
Mario T. Tabucanon, Anwarul Islam
Abstract: The purpose of this paper is to evaluate the characteristics of a Product Life Cycle (PLC) against various factors especially the effect of advertisement on it. The study took into consideration of the technology based products which are durable in nature and also changing their life styles and forms due to rapid technological change. The model assumes that consumer behavioral impact on life is negligible compared to other factors. This consumer behavioral factor acts quite randomly under constantly changing technological, social and political environment. As the marketing system is very complex in nature having feedback causal relationships among the marketing factors considered, System Dynamics modeling methodology was utilized. Sensitivity analysis was done to see the behavior of PLC under varying factor conditions. The results obtained from the model was compared with actual data and other calculated values.
Donella H. Meadows
Abstract: This paper resulted from thinking-at-the-typewriter in an attempt to accomplish two goals. The first goal was to be useful in the design of a new research project at the international Institute of Applied Systems Analysis (IIASA). The project was to deal with “resources, technology, and the environment in agricultural production”. The second goal was to design my own next research project in a way that was efficient, relevant, and in alignment with the project at IIASA. The comments that follow were addressed more to myself than to anyone else. If they sometimes sound critical or impatient, it is because parts of myself need that kind of talk from other parts of myself, not because I was trying to tell someone else what to do. Since then my students and other analysts have found the paper useful and have encouraged me to publish it. I have generalized it somewhat and left in just enough references to “resources, technology, and the environment” and to IIASA to use that research project as an illustration of my general points about problem definition.
Fiona Chen, David Andersen, Tanette Nguyen
Abstract: A system dynamics simulation of state aid to education policies is presented. The model contains a state taxation sector, a state aid distribution sector, and a representation of the taxation and expenditure policies of four aggregate local districts. Preliminary results from system simulations suggest that policies directed at equalizing educational expenditures may be effective in the short run (1 to 2 years) but will have little, if any, impact in the long run (7 plus years). Furthermore, local expenditure patterns may be very sensitive to factors (such as changes in the state’s non-educational expenditures) that have been excluded from formal analyses of state aid to education.
Abstract: A new system dynamic transportation model deals with important aspects of passenger transport and communication which interact with other aspects of society. The model’s analyses indicate substantial reduction of the vulnerability of transport systems to increasing oil prices on the world market as a result of taking certain measures. However, it is important to introduce the new policies as early as possible. It is not rational to wait and see how the oil market develops. The long-term effects of measures taken are, however, only slightly dependent on oil price developments in the first few years.
Abstract: This paper introduces an aggregate view of factors and policies that can influence the development of military forces in two international alliances which see each other as potential adversaries. The growth of forces observed in the NATO and Warsaw Pact alliances is taken as a reference mode. A conceptual System Dynamics Model is described which can accommodate a number of different perspectives on this issue.
Donald R. Drew
Abstract: This is the first of a three part article concerning Systems Management. In part I the conceptual systematic framework underlying the subject is discussed together with various current approaches adopted to translate the ideas into reality. The role of system dynamics in this framework is emphasised. In part II and III, to appear later, the author explores the breadth of application of system dynamics through a series of examples and presents a case study of the application of system dynamics to national development planning in the Lebanon.
Abstract: In the fall of 1956 Jay Forrester founded the Industrial Dynamics Group at the Sloan School, M.I.T. In the ensuing twenty five years many people have learned System Dynamics method and used them to study a wide variety of systems. On the occasion of this silver anniversary it is appropriate to celebrate our many accomplishments as well as to reflect on our present condition and future aspiration as a professional field of knowledge and practice. Perhaps, as a result of that reflection some of our weakness may be recognized and our research, teaching and professional practice may be revitalized, coordinated and redirected in ways that will produce an even better future.
R. Y. Cavana
Abstract: This note is a contribution to the discussion of academic training requirements for System Dynamics modellers. In particular, it suggests that training in Strategic Management can provide the System Dynamics modeller with some essential complementary tools and a top management perspective (or systems viewpoint), which is needed to define problems of real managerial interest.
R. G. Coyle
Abstract: The paper discusses two scenarios for a third World War in Europe, and argues that it is often convenient to supplement a scenario by a formal model. The problem of creating such a model are examined, and a model is formulated for land, air and sea combat in Europe and Atlantic, using the System Dynamics approach, which has not hitherto been widely employed for military analysis.
The model confirms the results of the two scenarios, and its use for the analysis of alternative force configurations is illustrated. Some implications in the interpretation of Soviet military literature are discussed, and the paper ends with speculations on the role and value of such models.
Donald R. Drew
Abstract: This is the second of a three part article concerning Systems Management. In part I the concepts underlying the subject were developed. Part II is concerned with the scope of application of system Dynamics and is a forerunner to a case study concerning National Development in the Lebanon which will appear in part III.
John D. W. Morecroft
Abstract: During the development of the field of System Dynamics, the causal loop diagram has risen to prominence as the diagraming tool for conceptualizing feedback system models. This paper challenges the prominent role of causal loops in conceptualization and offers instead two new tools, the subsystem diagram and the policy structure diagram.
Causal loop diagram are argued to be weak tools of conceptualization. They do not correspond closely to common mental models of social and industrial systems, and are therefore inefficient as a channel of communication between a formal model and its database of descriptive information. Causal loop diagrams do not adhere to the basic structuring principles of feedback systems, nor do they represent explicit decision-making processes and therefore lack the organizing power that should be expected of a conceptualizing tool. Their main strength is in providing an overview of loop structure, which is most useful in behavior analysis, not conceptualization.
Two new diagramming tools are proposed for overcoming the weakness of causal loop diagrams. The subsystem diagram shows major organizational divisions in a social or industrial system and is useful in boundary definition. The policy structure diagram shows the stock-and-flow structure diagram of a subsystem and major polices with their supporting information flows. The policy structure diagram is designed to improve the efficiency of communication between formal and mental models, and adheres strictly to the structuring principles of feedback systems. The use of the new conceptualizing tools is illustrated using material derived from a corporate marketing strategy projects.
Jack B. Homer
Abstract: Historian have presented a variety of hypotheses to explain the spectacular socio-economic changes in late 18 th and 19 th century England known collectively as the Industrial Revolution. This paper seeks to clarify the feedback structures underlying these diverse theories and show how they may be viewed as separate pieces of one overall puzzle. Causal-loop diagrams are used to help explain the basic mechanisms necessary for rapid economic development and the roles of investment, agriculture, income demand, raw materials, and technological innovation in that process. It is suggested that feedback thinking can be a powerful tool for unifying the typically broad spectrum of ideas concerning set of events as complex as the Industrial Revolution.
R. G. Coyle, J. C. Rego
Abstract: The purpose of System Dynamics modelling is the study of the mechanism, especially including managerial policies, which govern the evolution of a socio-economic system through time, and in the face of a chasing external environment. In particular, the analyst seeks to identify the processes which enable the system to benefit from opportunities and defend itself against threats. The concept of designing policy structures for controllable systems is the essence of the approach, allied to powerful methods of system description, and an efficient simulation technique.
Abstract: This paper is essentially a reply to Willard Fey’s challenge of how System Dynamics might be used to assist its own development. It builds on the basic ideas on suggested by Fey for the representation of the field of System Dynamics as a human feedback system. Additional perspectives on the field are derived using a reformulation of the System Dynamics method and it is concluded that a more thorough understanding of the environment of the field is required as a prerequisite to further growth.
G. P. Richardson
Donald R. Drew
Abstract: This is the third paper of a three article concerning System Management and presents an application of System Dynamics to development planning in the Lebanon.
D. E. Probert
Abstract: British Telecom’s Strategic Modelling Group has been actively involved in system dynamics modeling since 1976 when it was decided to develop a dynamic corporate model of British Telecom with a planning horizon of around 30 years. This research project was successfully carried out in collaboration with the Department of Control and Management Systems, Cambridge University. We have since developed a strategic control module which allows us to track a set of corporate objectives and also to analyse the robustness of the business under various imposed crises.
Richard H. Day
Abstract: This paper shows how irregular, more-or-less random fluctuations arise in deterministic economic models. The heart of the matter lies in the intrinsic properties of nonlinear feedback, a phenomenon emphasized by Forrester as causing fundamental difficulties in conventional dynamic analysis.
Fahriye H. Sancar, Bulent Ozgul
Abstract: This paper deals with the application of System Dynamics to user participation in departmental policy making in a university. A systematic method for the estimation of multipliers using worth assessment and delphi techniques is discussed. Expected behavioral responses of the users to changes in the major state variables are incorporated into a simulation model.
R. Joel Rahn
Abstract: Many feedback systems, when tested by applying exogenous inputs are able to maintain their characteristic behavior in the face of moderate changes in their internal relations. The robustness of behavior is usually due to the presence of compensating feedback. One class of systems seems to show a much lower resistance to external disturbances. The behavior of systems of this type - unity – gain positive feedback systems – depends in detail on the exogenous influences that impinge on the levels. In particular, the equilibrium state that results from a given disturbance depends on the size and duration of the disturbance. In this paper we show that these systems have a constant of the motion which accounts for their unique behavior when subject to exogenous disturbances. As well, the constant of motion permits a clear definition of the polarity of a large class of feedback structures.
Abstract: This paper relates the various controversies of economic development to the limitations of the models underlying the specialist judgments. The paper further relates the limitations of these models to the limitations of the methodological tools used and advocates a radical change in methodological thinking for improving the practical utility of the analyses of economic development.
David F. Andersen
Abstract: Recent experiences with the statistical evaluation of social programs suggest that when the programs being evaluated have impacts that are dynamic and possibly involve feedback effects, the statistical models being used to evaluate the programs may produce conclusions that are not fully justified. This paper proposes a research program that will use feedback simulation models to test the robustness of proposed evaluation design before data collection begins.
David F. Andersen
John Rohrbaugh, David F. Andersen
Abstract: The creation of workable and reliable objective functions has been an elusive goal of economics, policy analysts, and dynamic modelers since the beginning of quantitative analyses of social policy. The development of such functions would produce several dramatic benefits for the formation of social policy. Not only would objective functions provide a precise index of system performance, but they would also clarify and explicate the criteria being used in the process of policy formation. Perhaps more importantly, objective functions would allow analysts to rank order preferred sets of policy alternatives.
R. Joel Rahn
Abstract: Analysis of some typical structures found in system dynamics models of organizational and economic systems has shown that it is possible to define a contrast of the motion associated with positive feedback loops having unity gain, and that such a constant is a sensitive indicator of the under-lying dynamic nature of the multiple loop structure. In this paper, a general treatment of a class of positive feedback structures is developed in which the unity-gain examples appear as singular cases. The treatment includes development of a canonical form of the structure, transformation of the structure to reduce the order of the system, discussion of the eigenvalues of the linearized structure, some properties of open-loop and closed-loop step and pulse gains, general formulas for the constants of the motion, an additivity property of the structure and a brief discussion of the effects of stochastic influences on the generic structure.
Abstract: The agrarian economies of the developing countries have displayed throughout history many wage and income distribution patterns, which are the bases for the various models of the rural social class structure. Notable among these models are the neo-classical economic theory, the peasant agricultural system, the mixed peasantry and tenancy system, the feudal system, and the dual system. This paper attempts to show that these models only represent various phenomena which may arise when the social organisation of an agrarian economy functions under different socio-technical conditions. Thus, organizational variable appear to be promising instruments for policy intervention for economic development, which should aim at affecting the role behaviour of the actors in the social organization instead of merely attempting to alleviate symptoms of under-development.
Erik Mosekilde, Steen Rasmussen
Abstract: Using examples from physics and operational research it is shown how DYNAMO can be applied to systems for which the development in time depends upon a series of random events. It is further discussed how the notion of a Markov process, fundamental to stochastic systems analysis, completely concurs with the idea of a state determined system underlying System Dynamics. Non-Markovian systems can usually be brought into Markovian from through a redefinition of state space, and we therefore conclude that System Dynamics has a general applicability to stochastic systems.
Nancy Roberts, David Andersen, Ralph Deal, Michael Garet, William Shaffer, Tanette Nguyen and Marian Steinberg
Abstract: A curriculum for teaching system dynamics at an introductory level is described. The materials are designed to make instruction in system dynamics available to a wider audience than has hitherto been possible. The curriculum consists of seven self-teaching units: Basic Concepts of System Simulation; Structure of Feedback Systems; Graphing and Analyzing the Behavior of Feedback Systems; Analyzing Less Structured Problems; Introduction to Simulation; Formulating and Analyzing Simulation Models; and Formulating More Complex Models. The packages have been field-tested at the graduate and undergraduate levels as well as at six secondary school sites in the Boston metropolitan area.
Henry Birdseys Weil
Abstract: Virtually all System Dynamics practitioners are dependent on the continued support of clients or sponsors. Achieving greater client satisfaction is of the utmost importance. It is not only a matter of good business, but also of professional satisfaction, the credibility of System Dynamics, and the field’s ability to attract top people…
Stephen G. Boyce
Abstract: This paper describes the use of system dynamics methods to examine forestry polices. The important results are a change in the question posed by forest managers and the use of white boxes to make assumptions explicit to all interested parties. Behaviour of people is improved because complexity is reduced and communications for all interested parties are improved.
Raimo Keloharju, Ari Luostarinen
Abstract: This paper describes how redundant relations from a system dynamics model can be removed by using a special purpose software. The new methodology developed was applied to a recent publication of a forest study.
Harold A. Linstone
Abstract: This paper centers on problems which have arisen in the linking of technology with management. It is postulated that technological based issues require different perspectives and paradigms from human based issues and in Part I of the paper presented here some common paradigms are presented and criticised. Part II of the paper, which will appear in the next, edition of DYNAMICA, introduces the need for, and an assessment of, multiple perspectives in dealing with the management of technology.
Abstract: Using the principle of System Dynamics a software package of an integrated Steel Plant has been designed for computing its possible futures states, carrying out “what if” experiments, policy and testing. The package includes a set of models to estimate general Indian economic conditions, all India demand of flat and tubular steel products, availability of coking coal and power, and the performance of Steel Plant both in tones of output and financial indicators. The software package has been used to identify would-be bottlenecks in the next ten years and for evaluation of rationalization, debottlenecking and modernisation projects already contemplated by the management.
E. F. Wolstenholme
Abstract: The purpose of this paper is to highlight and enhance recent development in system dynamics by presenting the subject as currently perceived by the System Dynamics Research Group at Bradford Management Center. Emphasis is placed on the role of System Dynamics in system description, problem identification and qualitative analysis. It is concluded that the subject satisfies many of the requirement of a general within the system field, as well as being a sophisticated dynamic modeling technique.
W. E. Cundiff
Abstract: In the School of Social and Industrial Administration at Griffith University, students concentrating in “Decision Analysis and Support Systems” undertake a defining course in Simulation and Modelling. This course provides third year students with a theoretical and orientation for subsequent, more advanced management science course work or further development of simulation in particular. The first course aims to convey the scope and breadth of the field while at the same time leaving the student with some in-depth skill and appreciation of simulation. The course is divided into two parts focusing each one the familiar continuous and discrete approaches.
Chanoch Jacobsen, Habert Law-Yone
Abstract: The most basic problem of sociology as an empirical science is the difficulty of replicating studies within reasonable time limits and in genuinely comparable situations. It is the problem of controlled experimentation. Sociologists want to make correct predictions based on verified causal relationships, but cannot, because the nature of macro-social phenomena precludes experiments with adequate controls.
System dynamics promises a way out of this dilemma. The proposed strategy involves four phases. (1) Formulating the theory as a causal loop diagram. (2) Stating the variables involved in the functioning of the system, building the model and calibrating it until it is consistent with the theory. (3) Refining and adjusting the constants until the model can reproduce known time-series of relevant data on a number of data sets. (4) Systematically varying each constant while controlling the others. The last phase is, in fact, the quasi-experimental procedure for testing the conditions under which theory will stand or fall.
An illustrative example of the proposed strategy is given, with encouraging results relative to two data sets.
James W Kirchner
Abstract: The scientific technique known as the method of multiple hypotheses can be adapted to suit the purposes of system dynamics policy modeling. This method would allow determination of a model’s value through comparison with other competing models. It would also diminish modelers’ emotional attachment to any single theory. But in adopting this method, system dynamicists would need to develop a new philosophy of model evaluation, emphasizing disproof over verification and comparison among theories over improvement or elaboration on a single model.
Khalid Saeed, Arit Arayman Irdamidris
Abstract: This paper suggests continuous mathematical functions that can be used to replace DYNAMO supplied TABLE functions for most modelling applications where non-linear functions are needed. The proposed functions are selected for the ease of control of their slopes and limit parameters. Their use may help to increase the size of the model that can be handled on small systems while also avoiding the discontinuities of the TABLE functions they replace.
Harold A Linstone
Abstract: Part I of this paper appeared in the last edition of DYNAMICA and dealt with a presentation and critique of some commonly used paradigms and perspectives in the problems relating to the management of technology.
In this section of the paper the need for simultaneous, multiple perspectives is introduced and this philosophy is developed through a discussion of currently available methods, including the work of the author.
Abstract: DYNANET is a network modeling language which lays the foundation for interactive graphical-input specification of System Dynamic models. DYNANET is written in FORTRAN to make it portable and to be compatible with other modeling techniques. The special symbols which a DYNANET user would employ to represent System Dynamics concepts are defined. An example is given which illustrates each of the small set of special symbols.
Mary E. Floyd, George P. Richardson
Done M. M. Booker
Abstract: A large body of recent work in information system design models has encouraged the use of structured data flow models (Yourdon and Constantine, 1978; Gane and Sarson, 1977; Jackson, 1975). For the past several years, the computing and information systems curriculum at Pace University has emphasized incorporating the data flow approach to systems analysis and design in the two term introductory management information system design course, the on-line system design course, and to a lesser extent the decision support system design course.
John D Sterman
Abstract: System Dynamics modelers are often faulted for their reluctance to employ formal measures of goodness-of-fit when assessing the historical behaviour of models. As a result, the validity of system dynamics models is often questioned even when their correspondence to historical behaviour is quite good. This paper argues that the failure to present formal analysis of historical behaviour creates an impression of sloppiness and unprofessionalism. After reviewing the theory of validity in system dynamics, the paper proposes a simple set of summary statistics appropriate for system dynamics models. The statistics allow the error due to individual behaviour modes to be analysed, do not require the use of formal parameter estimation procedures, and can be conveniently computed. A large model of the U.S. economy is used to illustrate the use of the statistics.
Peter L. Galbraith
Abstract: This paper describes aspects of the construction and operation of a model designed to illustrate possible alternative futures for the higher education system in Australia.
After the context of the problem has been set some representative mechanisms are discussed for operations within the school, university, and labour force sectors of the model.
A discussion of simulation output samples from the range of policy evaluations and system conditions that were encompassed in the study. These include the effects of postulated changes in social and labour force conditions such as reduced working lifetimes, changes in the pattern of student progression from school to tertiary education, and the effects of various policies such as those which relate to the balance between school leavers and mature age enrolments. Implications of the simulation data for the system are considered.
Pieter W. Uys
Abstract: Table functions are frequently used in System Dynamics. An analytical function capable of representing a large class of such functions is defined and discussed.
Peter M Senge
Abstract: Diverse evidence suggests that western society may be in the midst of evolving fundamental assumptions, beliefs, and perceptions more consistent with a systemic world view. Pulled by this undercurrent, tools like system dynamics can focus the forces of change and bring them to bear more directly on pressing societal problems.
Abstract: It is well known that System Dynamics (SD), from its very beginning, has been reluctant to employ statistical and econometric tools for estimation and evaluation of model parameters and equations. What has changed up to now is mainly a shift in the justification for this refusal. For more than the first decade of SD, Forrester’s philosophy of selecting reasonable parameter values and of judging a model’s validity has expressed the dominant attitude. Because of the insensitivity of model behavior to most parameter values and the direct observability of parameters from the real system statistical methods were held to be superfluous. GRAHAM describes an arsenal of SD methods for parameter formulation and estimation.
F. J. Torrealdea, M. Grana
Abstract: In this paper some of the ideas of Ortega y Gasset about the dynamics of history have been gathered and organized according to the system dynamics paradigm. A cyclic process, characteristic of every normal course of history, is described as well as the feedback hypothesis responsible for it. Human life, as far as it affects history, is shown as being composed of five age groups each of them covering fifteen years of life. Two of these groups, two generations acting simultaneously in the field of history, are presented as taking the main responsibility for the dynamics of history.
[ Home ]