STELLA Surprises: Hitting Historical Targets

		Vol.1, Issue 4, Jul - Aug 2003

Return to the Front Page

STELLA Surprises: Hitting Historical Targets
Andrew Ford
Professor of Environmental Science
Washington State University
Have you ever made a simple change in a Stella model and found surprising improvement in the subsequent discussions of the model? Perhaps you rearranged the stocks and flows to show the flow of material in a clearer manner? Sometimes a simple change can pay big dividends in communicating our theories about a system.
I've had such pleasant surprises in my own work. And I've heard stories from colleagues who were surprised by the improvement in communication that followed upon a relatively simple change in a model. The "Stella Surprises" column will share such stories in The Connector. The surprises will involve simple changes that readers could easily implement in their own models.

	Andrew Ford is Professor of Environmental Science and Regional Planning at Washington State University. He teaches modeling with an emphasis on environmental problems in the west, and he is the author of the Island Press text on Modeling the Environment.
He uses the system dynamics approach to modeling, and he is the recipient of the Jay W. Forrester Award for the outstanding contribution to the field of system dynamics. Ford's recent research concentrates on the transition from regulation to competition in the electric industry. He uses System Dynamics and Stella to simulate the problems that have emerged in the restructured electricity system in California and in the western USA. Dr. Ford's students have developed models to aid our understanding of a variety of environmental problems. Recent examples include Stella models of the Yellowstone grizzly bear population and restoration of salmon habitat in the Pacific Northwest.
Andrew Ford Professor of Environmental Science Washington State University http://www.wsu.edu/~forda

Today's surprise comes from a project to simulate the boom and bust in power plant construction in the deregulated electricity industry. The model simulates market prices and power plant construction in the western USA and Canada, as explained in Figure 1 below.

The key finding from the western market model is that power plant construction is likely to appear in waves of boom and bust. The boom bust cycles are similar to cycles that appear in commodity markets and in real-estate. The unique requirements of electricity generation and transmission make this industry particularly vulnerable--- if new construction does not appear in a timely manner, prices can sky rocket and reliability can plummet. The results can be devastating, as western consumers learned during the years 2000-2001.

For today's "Stella Surprise," I turn to the challenge of checking the simulation results of the western market model against historical behavior. Information on power plant permits and construction is acquired by a number of permitting agencies, but converting the agencies' data into clearly understood "targets" for a simulation model requires considerable effort and explanation. Data on power plant construction is not necessarily available for each and every year, and some estimates are much less certain than others. The challenge is to display the historical targets in a clear manner. Then we wish to make it easy to check whether the simulation "hits the targets."

Stella's information buttons are particularly useful here. The info buttons may be positioned "on top of" a time graph, as shown in Figure 2 below. As the simulation proceeds through time, we can watch whether the model hits the targets. The buttons can be left closed for ease of viewing, or they can be opened when we wish to learn about the origin of the historical target.

The historical comparison in Figure 2 was a key step in building confidence in the theory of investor behavior in the western market model. Figure 3 shows the results when the simulation proceeds into the future. The simulation was presented at a November 2001 workshop to discuss alternative market structures for California. Figure 3 envisioned the system in 2001 at the crest of a building boom and on the verge of a bust. The system would then experience a lull in construction before the next building boom. Without fundamental changes in the wholesale markets, however, the next construction boom would come too late to prevent a return of the difficult conditions experienced in 2000-2001.

The results in Figures 2 and 3 have provided a base for important discussions of improvements in wholesale electricity markets in the west. At a key step in the development of these results, Stella's information button proved to be surprisingly useful in helping us understand and check the simulation results.

Adding information buttons to a time graph was a pleasant "Stella Surprise" - a simple change that made a big difference in communicating the results of a model.