1st Meeting Steps of Modeling and Problem Definition 6th System Dynamics Summer School, July 2019 Andrew Ford, Instructor, Intermediate Track 1st Meeting Steps of Modeling and Problem Definition Introductions: Students, Assistants, Instructor Schedule and Book Website
Steps of Modeling Step 1 A is for get Acquainted Get acquainted with the client, the system and the problem Step 2 B is for Be specific Be specific about the dynamic problem Step 3 C is for Construct Construct the sock and flow diagram Step 4 D is for Draw Draw the causal loop diagram Step 5 E is for Estimate Estimate the parameter values Step 6 R is for Run Run the model to get the reference mode Step 7 S is for Sensitivity Conduct sensitivity analysis Step 8 T is for Test Test the impact of policies
Another View
Parameter Estimation To decide how much effort to put into estimating a given parameter value, one ought to know how sensitive the behavior of the model is to the value of that parameter. Yet to know that, one must run the model, and that requires parameter values. To resolve the dilemma, the modeler picks some values rather quickly and simulates the model. Initial estimates are made carefully, to be sure, with as much concern for accuracy as can be easily mustered, but keeping in mind that one can always go back and estimate more carefully if it makes a difference.
Sources of Information the measurement of real values may be difficult or even impossible. Under these circumstances the value used in the model, legitimately, may be a straight-out guess. Of course, such a value cannot be regarded and treated in the same way as other parameters in the model. It must be treated tentatively and its role can be evaluated using techniques of sensitivity analysis. (Kitching, Systems Ecology) Include your guess if it is better than zero
Good modeling is an iterative process of trial and error Why don’t we just get it right the first time? Iteration reveals mistakes Iteration brings out the good ideas in the group Iteration increases group understanding Good Advice from Ed Roberts: Complete first iteration within 25% of time
A good pattern of Iterative Modeling This is Fig. 13.6 from MTE and from the PhD Thesis by Allyson Beall-King
Alternative style of modeling (NOT recommended)
Concluding Suggestions for the modeling process: Iterate early and often Two, Four, Six, Eight, Don’t forget to Iterate
First Two Steps in Modeling A is for get Acquainted with the client, the system and the problem B is for Be specific about the Problem
Problem Definition Identify the client or clients group Ask the clients: What is the Problem? (the dynamic problem!!) Work with client group to describe the “Reference Mode” or “Target Pattern” Does a “Target Pattern” seem troubling to you?
S-Shaped Growth vs. The Overshoot: Often the goal of sustainability policies
Example of S-Shaped Growth Annual Visitors to Glacier National Park
More Examples of S-Shaped Growth Yellowstone Nat. Park Yosemite Nat. Park
Tips to arrive at a useful Reference Mode Learn from separate drawings by all members of the Client Group Which pattern should be selected to start model building ? Keep other patterns in mind for later in the project (they may guide the development of a support* model with an entirely different time scale and target pattern.) *For examples, see “Simulating fast and slow dynamics” from the 60th Anniversary Issue of the System Dynamics Review, July 2018