TUE 11:00 AM Parallel: Modeling of the Electric Power Sector

Full report:

System Dynamics and Laboratory Experiments, by Santiago Arango, Yris Olaya, Jaime Andrés Castaneda Acevedo

The presentation was a general review of system dynamics and laboratory experiments.  The laboratory experiments centered on human decision making and the presentation considered three elements common to the experiments surveyed:  1) Goals or payoffs, 2) System environment and rules, and 3) Behavior as shown through subject's decisions.  In each experiment, the experimenter controlled the system and the goals.  Behaviors were recorded as subject's pursued their goals and decision rules are estimated from observed behavior.  Misperceptions of feedback contributed to poor performance. 

There was consideration of whether experiments can be effective in representing real-world conditions and, by extension, whether behaviors observed under experimental conditions serve as an accurate proxy for real-world behaviors.  The notion of parallelism argues that behavioral regularities will persist into new situations as long as the conditions are consistent.  Novel conditions require new experiments.  It was not clear how the distinctions between consistent and novel conditions can be rigorously determined.

It was concluded that system dynamics improves the external validity of the surveyed experiments and that more formal experimental protocol is needed for future experiments.

A system dynamics model for the German electricity market - model development and application, by Tobias Jaeger, Susanne Schmidt, Ute Karl

The development and validation of a model of the German electricity market was described.  Leveraging Zertsim, the model contained a short-term component that influenced demand, pricing and generation; a medium-term component that modeled capacity, investment and operating costs; and a long-term component that modeled technological progress and resource availability.  The model covered a time range from 1998 to 2025 (1998 began a new era of deregulation) and focused on Germany with exchange to neighboring countries.

Key inputs to the model included initial values and growth in prices for fuels, feed-in, tariffs and CO2 taxes, along with internal interest and plant construction.  Key outputs of the model included spot (daily) and average annual prices, capacity, and CO2 emissions.  The model provided a reasonable fit to the historic trends in spot price data but missed the magnitude of oscillations as well as some substantial discrete shocks to prices.  The fit to historic capacity was generally good.

From the model, the key drivers of price were determined to be environmental constraints, fuel prices, electricity demand and extended operating time for nuclear power plants in Germany.  Key drivers of CO2 emissions were determined to be environmental constraints, extended operating time for nuclear power plants in Germany, electricity demand, and fuel prices.

A System Dynamics Model of the Mauritian Power Sector, by Kailash Balnac, Chandradeo Bokhoree, Prakash Deenapanray, Andrea Bassi

While Mauritius boasts one of the highest GDPs in Africa, they face unique policy-making challenges due to the fact that most land is owned by foreigners and sugar producers wield considerable control over power production and water rights.  The continued rise in tourism adds another dimension to the policies of resource management.   Mauritius currently imports nearly 80% of their power generating resources, primarily oil and coal.  In planning a long-term sustainable energy policy, Mauritius hopes to avoid the risks of price volatility and resource scarcity that underly their current policies.

The model was based on existing Threshold-21 (T21) models and used a hybrid approach to combine system dynamics and optimization.  Electricity demand was calculated from available data.  GDP, income, population, and cultivated land were primary factors and were endogenous to the model. The model provided generally good fits to historical data using exogenous data to aid the fitting.  Work to incorporate more endogenous variables will be done in subsequent iterations.

Additional future work includes improved calculation of peak loads, including an endogenous pricing mechanism as a demand driver, incorporating the energy sector into T21 and analyzing the Mauritius vision in the model.

Michael Smith