What is System Dynamics?

System dynamics originated in the 1960s with the work of Jay Forrester and his colleagues at the Sloan School of Management at the Massachusetts Institute of Technology. They developed the initial ideas by applying the concepts from feedback control theory to the study of industrial systems (Forrester 1961).

One of the best-known applications of the 1960s was Forrester's (1969) Urban Dynamics. It explained the patterns of rapid population growth and subsequent decline that have been observed in American cities like Manhattan, Detroit, St. Louis, Chicago, Boston and Newark (Schroder and Strongman 1974). Forrester's simulation model portrayed the city as a system of interacting industries, housing and people (Forrester 1969).

One of the most widely known applications of system dynamics appeared a few years later in a best-selling book entitled The Limits to Growth (Meadows et al. 1972). This study looked at the prospects for human population growth and industrial production in the global system over the next century. A computer model was used to simulate resource production and food supply to keep up with the growing system. The authors concluded that the world could not support the present rates of economic and population growth much beyond the year 2100. The study was not about a pre-ordained future - it was about making choices to influence the future.

This very brief and incomplete overview of the beginning of system dynamics shows that system dynamics is a method to study the world around us. The central concept is to understand how all the objects in a system interact with one another. System dynamics looks at a system as a whole. A system can be a bank account, a game herd, a population, and a company.

System dynamics attempts to understand the basic structure of a system, and therefore to understand the behavior it can produce. Many of these systems and problems can be built as a computer model. The advantage is that the model on the computer is flexible and can carry out many simulations. Hence, many future development paths can be evaluated.

What Is a Causal Loop / Feedback Loop?

With the availability of the stock-and-flow software programs, we can concentrate on the realism of the model rather than on its analytical tractability. The inclusion of non-linear relationships is one of the most important improvements compared to the numerical models. Furthermore, the emphasis of system dynamics on the role of information feedback. The causal loop diagram shows the technique to portray the information feedback in a system. Causal refers to a cause-and-effect relationship. The word loop refers to a closed chain of cause-and-effect.

Feedback is a process through which an indicator goes through a chain of causal relations to re-affect itself. There are positive and negative feedback loops. On the one hand, a feedback is positive if an increase in a variable, after a certain delay, leads to a further increase in the same variable. Positive feedback is found in systems that produce exponential behavior. On the other hand, a feedback is negative if an increase in a variable leads to a decrease of the same variable. Negative feedback drives balancing or stabilizing systems that produce asymptotic or oscillatory behavior.

Figure: Flow diagram and corresponding causal loop diagram

Example:
The diagram shows a population that is fed by the flow of births and drained by the flow of deaths. The causal loop diagram shows the two feedback loops. The loop on the left is a positive feedback loop (+). It shows a closed chain of cause-and-effect in which a larger population leads to more births, and more births leads to a still larger population. The loop on the right side is a negative feedback loop (-). An increase in the population will tend to increase the number of deaths which will reduce the size of the population

The Model Language

• Stocks are accumulations and hold the current state of the system.
• Flows change. They increase or decrease the stocks. The cloud represents a stock that is outside the system boundary, so we are not concerned to keep track of it.
  Direction of flows are:
  - one stock into another (e.g., maturation)
  - a stock into a cloud (e.g., deaths)
  - a cloud into a stock (e.g., births)
• The converters modify stocks and flows. When you see converters without incoming arrows, you know the converter is specified by the model builder. These are sometimes called model inputs.
• A connector shows the flows inside the model. Connectors enable the creation of FEEDBACK loops.

Software for Modeling and Simulating Dynamical Systems
A Selection

The simulation language Dynamo was developed in the late 1950s by a group working with Jay Forrester at MIT (http://web.mit.edu/sdg/www/). Modern, structural-diagram-oriented descendants of DYNAMO are:

Vensim® - a visual software to develop, analyze, and test dynamic feedback models. Models are constructed graphically or in a text editor. Vensim® was created by Ventana® Systems, Inc., Harvard, MA (http://www.vensim.com/). Vensim® Personal Learning Edition (PLE) (http://www.vensim.com/venple.html) was designed for educational purposes to lower the barriers for the beginning system dynamics modeler and is free of charge.

Stella and ithink (http://www.hps-inc.com/edu/stella/stella.htm) - the first system dynamics software with graphical model input on the level of structural diagrams. It was created by High Performance Systems, Inc. (HPS).

Powersim (http://www.powersim.com/default_home.asp) - (newer) models have the same data-format as Stella.

Simple system dynamics models can also be calculated using a simple spreadsheet program. Spreadsheets are the preferred approach to perform a complex calculation for a single point in time. The limits are clear. If you look for insights into the system's behavior, you should use a system dynamics software.

Web Resources

People with more detailed interests in system dynamics should visit the System Dynamics Homepage at MIT - Massachusetts Institute of Technology (Cambridge, MA, USA). The MIT website, Tom Fiddaman's library and Günther Ossimitz mega link list provide helpful information, such as system dynamics models which can be download free of charge, links to system dynamics courses and programs, system dynamics programs, system dynamics institutions, contact addresses, teaching manuals, on-line literature and recommended readings, events and simulation games.

Readings - A Selection

Ford, Andrew. 1999. Modeling the Environment. An Introduction to System Dynamics Modeling of Environmental Systems. Washington, D.C.: Island Press.

Forrester, Jay. 1961. Industrial Dynamics. Walthan, MA: Pegasus Communications.

Forrester, Jay. 1969. Urban Dynamics. Portland, OR: Productivity Press.

Forrester, Jay. 1971. World Dynamics. Walthan, MA: Pegasus Communications.

HPS. 1993. Stella II: Technical Documentation. Hanover, NH: High Performance Systems, Inc.

Kirkwood, Craig. 1995. An Overview of Methods for Applied Decision Analysis. Interfaces 22, No. 6 (November): 28-39.

Meadows, Donella, Dennis Meadows, Jorgen Randers, and William Behrens. 1971. The Limits to Growth. New York, NY: Universe Books.

Meadows, Dennis, William Behrens, Donella Meadows, Naill Rogers, Jorgen Randers, and Erich Zahn. 1974. Dynamics of Growth in a Finite World. Walthan, MA: Pegasus Communications.

Powersim Corp. 1996. Reference 2.5. Reference Manual. Herndon, VA: Powersim Corporation Inc.

Richardson, George and Alexander Pugh. 1981. Introduction to System Dynamics Modeling with Dynamo. Walthan, MA: Pegasus Communications.

Schroeder, Walter and John Strongman. 1974. Adapting Urban Dynamics to Lowell. Chapter 16 in Readings in Urban Dynamics, Vol.1, edited by Nathaniel Mass. Walthan, MA: Pegasus Communications.

Sterman, John D. 2000. Business Dynamics. Systems Thinking and Modeling for a Complex World - plus one CD-ROM. Boston: Irwin.

Ventana Systems. 1995. Vensim User's Guide. Ventana Systems, Inc., 149 Waverley Street, Belmont, MA 02178.

Ventana Systems. 1996. Vensim Personal Learning Edition, User's Guide. Ventana Systems, Inc., 149 Waverley Street, Belmont, MA 02178.