An experimental software package is being used as an extension to the DYNAMO IV compiler to linearize the model at any point during a simulation, compute the eigenvalues and eigenvectors of the linearized system, identify the levels important in producing each behavior mode, and compute the elasticity of a given eigenvalue (corresponding to elasticiy of period and damping) with respect to all model parameters. The package is intended to help modelers understand the causes of behavior in very complex models, both for debugging implausible behavior, and for presenting the causes of plausible behavior more convincingly. The package is able to work for the System Dynamics National Model, a model of around 300 levels. Practical experience has uncovered some difficulties in making the analysis useful,, but these are being surmounted. The experience suggests that mathematical methods should be used extensively “in the field” before being offered as candidates for expanding the paradigm of System Dynamics modeling.
A number of challenges face firms that need to decide when and whether to convert from technologies to new computer-based technologies. Such is the case with lithographic setup shops, which prepare photos for color printing; they must choose between continuing with traditional craft methods or acquiring digital image-processing equipment. Pioneering firms can be saddled with experimental, undependable, and expensive prototype systems. Rapid technological changes still occurring in digital systems can allow competitors who invest later to obtain cheaper, more effective equipment. But firms investing later may find themselves paying for the large investment just when most competitors are established in the new technology and competition has forced prices and profits to low levels.In order to create an organizing framework for analyzing and developing conversion strategies for these firms, we worked in collaboration with Inter/Consult, the project's sponsor, to build a system dynamics model of the color process industry, its market, and a typical firm. The primary purpose of the moel is to provide a clear understanding of the impact these major capital investments will have on the profit structure of lithographic setup shops and to help these shops develop effective conversion strategies. A secondary purpose of the model is to aid digital image-processing equipment suppliers in understanding their market and to provide them with a toll for generating alternative scenarios given different assumptions about economic trends, technological developments, prices, market size and composition. The model serves as a strategy support system that allows clients to derive scenarios explicitly from causal assumptions and to evaluate alternative investment strategies.
Several interactive computer graphics technologies are now available that can provide powerful tools which enhance our ability to conceptualize, implement, and communicate complicated system dynamics model structure and behavior thereby giving us opportunities to improve our effectiveness as researchers, consultants and educators. This paper gives an overview of several projects utilizing interactive computer graphics and evaluates their significance for system dynamics. Included in this discussion are: 1) computer aided design systems for “automagic” design and updating of overview, policy structure, flow, and causal loop diagrams, 2) computer teaching games and self-paced interactive computer aided instruction packages designed for personal computers; 3) review of the new Micro-DYNAMO and Hewlett-Packard plotting software from Pugh-Roberts, 4) computer networks, computer conference based academic programs for the general public, and network indexed video cassette extension libraries of system dynamics presentations and seminars; 5) interactive computer driven video disk processors with touch sensitive screens allowing a modeller multimodal access to overview, subsystem, policy-structure, causal loop and flow diagrams, table functions, documentors, and DYNAMO equations on the same system; and 6) two- and three- dimensional representations and animations of model behavior on multicolor dynamic displays driven by computer and video disks. These developments are assessed with respect to their possible contribution to the growth of system dynamics as a field, dissemination of system dynamics methodologies and to the implementation of policy recommendations. Because of falling prices for software and hardware, the explosion in interest in personal computers, the exponential growth in their functionality, and the current state of the field, we believe the next two decades will be the phase of most rapid growth for system dynamics.
In 1981 a preliminary system dynamics model was developed for the Norwegian State Railways to study passenger and freight traffic for the complete network. In addition, a particular model was developed to study commuter traffic in Oslo. This paper describes modeling work done for the Norwegian State Railways (NSB).
Certain medical interventions may result in reducesd costs to society. Others, however, by keeping people alive longer, may cause higher costs to be incurred for continuing health care and disability and retirement payments. A generic disease process model for projecting the implications of various medical interventions is presented. The model is applied to myocardial infarction in the U.S. male population and results of simulating several interventions specific to that disease process are discussed. Conclusions are drawn and it is argued that this model is useful for identifying interventions that result in higher costs to society in order that adequate resources be set aside to cover these costs. The work reported in this paper was funded by a grant from the Kaiser Family Foundation.
The model described in this report is meant to show how some of the practical problems of combining hydrological and biological processes can be addressed, how models can be used to examine specific questions, and along what lines the present model ought to be developed to eventually arrive at a useful policy tool.
At the last System Dynamics research conference held in the United States, we presented a paper which described a computer simulation model of an elementary school. The purpose of the model was to examine the structural differences between schools which are effective and ineffective for what we have come to call “initially low-achieving children.” In that paper (Clauset & Gaynor, 1981), in a subsequent paper (Clauset and Gaynor, 1982), and in a book manuscript (Clauset and Gaynor, in preparation), we have described in varying degrees of details tests which examined a number of school improvement policies. Policies testes included the following: Changing policies affection time allocations, Improving teacher skills, Encouraging teachers to place more emphasis on low achievers, Raising teacher expectations for low-achievers, Improving classroom of school-wide behavior, Changing class size, Changing the demographics of the student body (e.g., size low achievers).
This paper will address the relative utility of employing the linguistic structure used by system dynamics compared to translating the modeler’s perception of reality into other symbolic language systems. The first section will review the relation of language to the method of scientific inquiry. This will include a discussion of the debate over the problem of evaluating policy alternatives of social systems. The final section of the paper will specifically identify some of the differences between the imposed linguistic structure of system dynamic models and the symbolic language systems often employed in orthodox economic analysis.
Eigenvalue analysis of dominant feedback loops promises to be a powerful new tool for identifying the structural origins of behavior in system dynamics models. Traditional simulation methods for dominant loop analysis are time-consuming and error-prone. A new technique permits calculating the marginal contribution of each feedback loop to each mode of behavior in a model. The technique computes the numbers showing the percentage change in natural frequency and damping of each eigenvalue resulting from a one percent change in loop gain. The magnitude of an elasticity measures the overall importance of a loop to a mode of behavior. The magnitudes can be used to rank loops by relative dominance over each mode, or to rank; modes by relative importance to each loop. The techniques can be used to analyze both linear and some nonlinear behavior modes.
Sensitivity testing, according to the glossary of terms in a Congressional manual on simulation modeling, is defined as the “running of a simulation model by successively changing the states of the system…and comparing the model outputs to determine the effects of these changes” (Congress 1975, p. 129). Sensitivity testing is generally viewed as an important part of the modeling process because it helps researchers narrow down those areas where more data gathering would be useful. In our introductory remarks, we argue that detailed sensitivity testing is particularly important in system dynamics modeling efforts, and we list several obstacles that make detailed sensitivity testing difficult. We introduce a set of testing procedures developed at the Los Alamos National Laboratory and verified by the Control Data Corporation that can help system dynamicists perform detailed sensitivity testing on a routine basis. In the body of the paper, we present an illustrative application of the testing procedures, and we list six specific uses of the procedures. We describe the availability of the testing package, and we conclude with a set of practical guidelines for investigators wishing to make use of this unique set of procedures.