Robustness E-Laboratories

9700 S. Cass Avenue

Argonne, IL 60439

Robust systems often exhibit adaptive coherence in the face of chaotic change (Fontana and Wagner, 2001). Building robust systems requires both an understanding of robustness and empirical tools to ensure that this understanding has been correctly applied. For some applications, the necessary empirical tools include agent-based models and simulations (ABMS). Where appropriate, ABMS can be used as electronic laboratories (“e-laboratories”) to test systems for robustness before problems are experienced or exploited in the real world. Example application areas include the development of effective electric power pricing policies and the protection of critical infrastructures.

Electricity pricing policies define how electric power producers are paid for their inputs to the power grid. A variety of pricing policies are currently being used in electricity markets (NYISO, 1999; Ofgem, 1999). Determining an effective pricing policy is a complex endeavor that has important ramifications for market robustness. ABMS has been used to study market behavior and robustness for markets as varied as power systems (Bower and Bunn, 2000; North, 2001), equities (LeBaron, 1999), and foreign exchange (Yang, 2000). In particular, Argonne National Laboratory has created several electricity market ABMS including the Electricity Market Complex Adaptive Systems Model (EMCAS).

In EMCAS, the diverse participants in the electricity market are represented as “agents,” each of which has its own set of objectives, decision-making rules, and behavioral patterns. Unlike conventional electric system models, EMCAS does not postulate a single decision-maker with a single objective for the entire system. Rather, each agent is allowed to establish individual objectives and apply individual decision rules. As the simulation progress, the agents adapt their strategies based on the success or failure of their previous efforts. EMCAS acts as an e-laboratory for pricing policy robustness testing by allowing different policies to specified and then tried, before they are tried in the real world. EMCAS is a Java-based ABMS that uses the Recursive Porous Agent Simulation Toolkit (RePast).

Protecting critical infrastructures, such as electric power and natural gas systems, requires “thinking the unthinkable” and “doing the undoable.” Potential natural and man-made failures of robustness must be discovered and corrected before they occur in the real world. Argonne National Laboratory has created several critical infrastructure robustness e-laboratories including the Swarm-based Spot Market Agent Research Tools.

The Spot Market Agent Research Tools include agents that simulate electric power marketing and transmission; agents that simulate natural gas marketing and distribution; and agents that simulate links between these systems. The tools also include an interactive virtual reality interface that features large-scale fully immersive three-dimensional imaging. An image from this virtual reality system is shown in Figure 1. These tools have been used as robustness e-laboratories to discover the effects that natural or man-made failures in one infrastructure might have on other infrastructures.

Figure 1: The Spot Market Agent Research Tools Immersive Virtual Reality Interface

As demonstrated by EMCAS and the Spot Market Agent Research Tools, robustness e-laboratories can be constructed for complex engineered systems. These robustness e-laboratories allow complex systems to be tested for robustness before problems are experienced or exploited in the real world.

References

Bower, J., and Bunn, D. W. A Model-based Comparison of Pool and Bilateral Market Mechanisms for Electricity Trading, Energy Journal, Volume 21, Number 3, July 2000.

Fontana, W. and A. Wagner, Mutational Robustness, Modularity and Evolvability, Available as http://www.santafe.edu/sfi/research/focus/robustness/projects/mutationalRobustness.html, Santa Fe Institute: 2001.

LeBaron, B., “Building Financial Markets With Articial Agents: Desired goals, and Present Techniques,” Forthcoming in Computational Markets, Karakoulas, G. ed., MIT Press, Boston, Massachusetts: 1999.

North, M.J., Agent-Based Infrastructure Modeling, Social Science Computer Review, Sage Publications, Thousand Oaks, California: Fall 2001.

New York Independent System Operator (NYISO), NYISO Day Ahead Scheduling Manual, NYISO, New York, New York: 1999.

The Office of Gas and Electricity Markets (Ofgem), The New Electricity Trading Arrangements: A Draft Specification for the Balancing Mechanism and Imbalance Settlement, Ofgem, United Kingdom: 1999.

Yang, J., “The Efficiency of an Artificial Double Auction Stock Market with Neural Learning Agents,” Proceedings of the Workshop on Agent Simulation: Applications, Models, and Tools, Argonne National Laboratory, Argonne, IL: 2000.