| We believe employing a systems perspective provides individuals, groups and organizations a greater likelihood of creating the future they desire with fewer unintended consequences. The endeavor is to promote further awareness and understanding of the power and value of employing a systems perspective and create content essential to enable one to develop and evolve the skill and ability to think and act systemically. |
System Dynamics Methodology
From SystemsWiki
Interest in System Dynamics is spreading as people appreciate its unique ability to represent the real world. It can accept the complexity, nonlinearity, and feedback loop structures that are inherent in social and physical systems.
On the other hand, several difficult steps in moving from problem to solution hamper system dynamics. First, and probably most elusive, little guidance exists for converting a real-life situation into a simulation model. At later stages, many system dynamics projects have fallen short of their potential because of failure to gain the understanding and support necessary for implementation. Systems thinking and soft operations research may help organize and guide group processes that must occur when system dynamics interfaces with people in the actual systems.
Fig. 1 illustrates the system dynamics process. An investigation starts at Step 1, motivated by undesirable system behavior that is to be understood and corrected. Understanding comes first, but the goal is improvement. System dynamics appeals to activists. It is undertaken for a purpose. At the first step, on the left in the diagram, the relevant system must be described and a hypothesis (theory) generated for how the system is creating the troubling behavior.
Step 2 begins formulation of a simulation model. The system description is translated into the level and rate equations of a system dynamics model. Creating the simulation model requires that the rather general and incomplete description of Step 1 be made explicit. As with every step, active recycling occurs back to prior steps. In Step 2, writing equations reveals gaps and inconsistencies that must be remedied in the prior description.
Step 3, simulation of the model, can start after the equations of Step 2 pass the logical criteria of an operable model, such as all variables being defined, none defined more than once, no simultaneous equations, and consistent units of measure. System dynamics software packages provide such logical checks. Simulation may at first exhibit unrealistic behavior. As a result, simulation leads back to the problem description and to refinement of the equations. Step 3 should conform to an important element of good system dynamics practice; the simulation should show how the difficulty under consideration is being generated in the real system. Unlike methodologies that focus only on an ideal future condition for a system, system dynamics should reveal the way we arrived at the present and then, in a later step, the path that leads to improvement. The first simulations at Step 3 will raise questions that cause repeated returns to Steps 1 and 2 until the model becomes adequate for the purpose under consideration. Note that “adequacy” does not mean proof of validity. There is no way to prove validity of a theory that purports to represent behavior in the real world.1 One can achieve only a degree of confidence in a model that is a compromise between adequacy and the time and cost of further improvement. The proper basis of comparison lies between the simulation model and the model that would otherwise be used. That competitive model is almost always the mental model in the heads of the people operating in the real system. A system dynamics model creates so much more clarity and unity, compared to prior mental models, that the “adequacy” decision usually generates little controversy among real-world operators who are under time and budget pressures to achieve improved performance. However, being noncontroversial does not mean acceptance in Steps 5 and 6.
Step 4 identifies policy alternatives for testing. Simulation tests determine which policies show the greatest promise. The alternatives may come from intuitive insights generated during the first three stages, from experience of the analyst, from proposals advanced by people in the operating system, or by an exhaustive automatic testing of parameter changes. I expect that system dynamics will continue to rest on experience, art, and skill for imagining the most creative and powerful policy alternatives. Automatic parameter searching will be of limited usefulness. In the more complex systems, there will be many competing criteria for defining success; also, there will be many peaks in the multi-dimensional behavior map so that the most favorable performance may depend on several simultaneous changes in the model. In addition, the best alternative behaviors will often come from changing the system structure.
Step 5 works toward a consensus for implementation. Step 5 presents the greatest challenge to leadership and coordinating skills. No matter how many people have participated in Steps 1 through 4, many others will become involved in ultimate implementation. The model will show how the system is causing the troubles that are being encountered. Almost always, the reasons will lie in policies that people know they are following and which they believe will lead to solutions to the troubles. Implementation often involves reversing deeply embedded policies and strongly held emotional beliefs. It is not that people disagree with the goals, but rather how to achieve them. Even with widespread intellectual agreement with a system dynamics model and with the recommended improved policies, there may still be great discomfort with the prospect of changing from traditional actions.2 To overcome both active and passive resistance requires sufficient duration and intensity of education and debate to reverse traditional practices. Questions will arise that require repeated recycling through Steps 1 through 5.
Step 6 implements the new policies. Difficulties at Step 6 will arise mostly from deficiencies in one of the prior steps. If the model is relevant and persuasive, and if education in Step 5 has been sufficient, then Step 6 can progress smoothly. Even so, implementation may take a very long time. Old policies must be rooted out. New policies will require creation of new information sources and training.
Evaluation of the policy changes comes after implementation. As with determining model adequacy, evaluation has no clear procedures, nor can one expect a conclusive outcome. While the new policies are being implemented and used, a process that can take several years, many other changes will have occurred in the system and its environment. Even when performance is unambiguously better, some people will claim that credit should go to changes, other than the new policies, that occurred during the system dynamics project. The evaluation may even rest on results other than those for which the project was undertaken. I recall a senior corporate officer who said, after a major system dynamics program had been operating in the company, “I can’t prove that it has made any difference on the profit and loss statement, but I know that we have a better understanding of what is happening and more confidence in what we are doing.” Evaluation will remain subjective. The weight of evidence will accumulate as system dynamics becomes the common thread through an accumulating sequence of successes.
[edit] Reference
- This content of this page is an excerpt, with permission, from System Dynamics, Systems Thinking, and Soft OR by Jay W. Forrester. 1994 and appeared in System Dynamics Review, Summer 1994, Vol. 10, No. 2. The rest of this paper is well worth reading and is highly recommended.
| Online Courses * Content Guide * Bookmarks Feedback to Gene Bellinger |
