 Feature articles
Microelectronic nose detects and
recognizes gases to save lives
Sidebar: Smart Sensor Developer Kit sees things in the air Sweet solution to a salty situation is also environmentally friendly Modeling world turned upside down with new simulation software Argonne Update
|
Modeling world turned upside down
with new simulation
software
by Evelyn Brown
Putting a spin on traditional modeling techniques, Argonne researchers are building simulations of complex systems from the bottom up to create a closer match with the real world.
 IMPRESSING THE BOSS – Department of Energy Office of Science Director Ray Orbach (foreground) describes Argonne's CASS program to President George W. Bush during their July 22 visit to the laboratory. House Speaker Dennis Hastert (left) looks on. |
As one of the country's leading research groups in this burgeoning field of Complex Adaptive Systems Simulation (CASS), Argonne is developing software, teaching the simulation technique to business and government, and pioneering complex-systems modeling as diverse as electric power deregulation, an endangered species habitat and an ancient Mesopotamian civilization. In 2000, one of Argonne's CASS models predicted the possibility of California's electricity-price spikes months before they occurred.
Modeling complexity
Complex adaptive systems evolve and adjust to a changing environment. The CASS programming approach allows researchers to examine the behavior and interactions of individual agents – a single decision-making entity such as a consumer, company or product – within a larger system. Traffic is an example of a complex system – it involves multiple variables and individual decision makers or agents, each with his or her own goals and strategies.
Other complex systems include stock markets, ecosystems, pipelines, industrial supply chains and electrical distribution systems central to national and global concerns. Small changes in these dynamic networks may bring about severe effects. Business and government need better tools to predict the behavior of these systems, especially ones that could become unstable or fail.
CASS takes advantage of today's greater computer power and new software tools, but it also represents a change in perspective.
"We simulate the process of agents interacting," said Charles Macal, group leader for modeling, simulation and visualization in Argonne's Decision and Information Sciences (DIS) Division, "instead of solving equations based on long-run average behavior."
Most computer modeling looks at the aggregate – the big picture – and examines averages. "Agent-based simulation," said DIS software engineer Mike North, "builds models up from the individual level and looks at the patterns of behavior that emerge or the presence of extreme events. Using individual building blocks is like understanding chemistry by looking at atoms and molecules instead of a solid lump of coal."
Macal said "building the simulations from the bottom up allows interactions between the agents to determine the system's primary behavior. The agent interactions create patterns and structures we would not anticipate from looking at the simple components. This technique is particularly useful in simulating the behavior of agents that adapt to changing conditions."
Macal and North are just two of a cadre of researchers in DIS exploring the power of this simulation tool.
Power modeling
For the Illinois Commerce Commission (ICC), researchers are studying electric power deregulation using the Argonne-created Electricity Market Complex Adaptive System (EMCAS) model. This effort builds on more than three decades of Argonne experience in modeling the electric power industry.
 MARKET COMPLEX – Diagram of the agents in Argonne's Electricity Market Complex Adaptive System model. |
To avoid many of the mistakes California experienced in its electric power deregulation, Argonne is simulating problems that could arise as Illinois deregulates its electric power industry. The results will help the ICC understand potential effects of the deregulation process, which may be implemented as soon as January 2007. The findings could range from identifying the need to build new power lines to determining possible market conditions that affect electricity prices.
DIS is modeling the system from the ground up using such agents as:
- generators,
- generation companies,
- major power lines,
- residential, industrial and commercial consumers,
- demand companies that sell to consumers,
- brokers,
- regulatory agencies, and
- random events caused by weather or outages.
The Argonne simulation will probe the system for power outages, market manipulation or other behavior outside of the norm. Each agent has its own set of objectives, decision-making rules and behavioral patterns. As the simulation progresses, each agent adapts its strategies based on the success or failure of the previous efforts. EMCAS allows decision makers to test the effects of alternative policies and rules.
Argonne researchers are relying on a detailed model of the state's physical power grid provided by the University of Illinois at Urbana-Champaign.
SMART predictions
Moving beyond the state, another DIS group – Argonne's Infrastructure Assurance Center – studies the complex relationships of the nation's interconnected systems. Their work was highlighted in logos Vol. 17, No.2. Center scientists have created several CASS models called SMART, for Spot Market Agent Research Tool, to investigate the electric power and natural gas infrastructures and their interdependencies.
When President George W. Bush visited Argonne in July 2002 to learn about homeland-security technologies, he was briefed on Argonne's SMART tools. The research is supported by the Department of Defense.
SMART II simulates electric power generation and consumer agents interconnected through the power-transmission grid. The program considers economic issues including production cost, investment capital, demand growth, new generation capacity for profitable producers and bankruptcy for noncompetitive organizations.
While more work is needed to validate and calibrate the model, software engineer North reported in early 2000 that SMART II indicated certain power transmission line configurations – like those in California's electrical grid – could encourage price spikes.
"SMART II provided us with qualitative insights," North said. "Greater electrical market price stability may be gained by avoiding specific configurations that encourage instabilities."
Within months of the prediction, such price spikes occurred in California.
Another program, SMART II+ is more complex. It adds natural gas-fired electric generators to SMART II to more accurately reflect current trends in the energy market. Natural gas-fired turbine generators are the nation's fastest growing electrical power source.
The transmission or distribution capacity of each line or pipe is also added to the interconnected infrastructures. SMART II+ even allows components to be disabled to simulate failures.
"SMART II+ showed us back in 2000," North said, "that the rising natural gas-fired electrical generator market would increase interdependence between the electric power system and the gas supply system. The electric power and natural gas markets would be pitted against one another during simultaneous failures since both are competing for the same natural gas. We recently saw this situation in the market after the recent El Paso natural gas pipeline explosion."
Recreating Mesopotamia
Working with the University of Chicago's Oriental Institute, John Christiansen is using a complex adaptive systems approach to create an ancient Mesopotamian village circa 2,500 B.C.E.
 MODELING ANCIENT MESOPOTAMIA – The modern town of Qaraqosh in northern Iraq is being used as a basis for the Virtual Ancient Mesopotamia simulation. This town has roughly 3,700 agricultural fields cultivated by 1,000 households. |
Mesopotamia, the land that was between the Tigris and the Euphrates rivers, relied on irrigation to grow crops. But, the irrigation eventually turned the soil toxic making it unable to support crops. Eventually the civilization disappeared leaving information – including that of the poisoned fields – on cuneiform tablets.
The model – Virtual Ancient Mesopotamia – addresses the complex issues of sustainability, growth and decline in the face of increasing environmental stress. Farmers in the San Joaquin Valley and other places are experiencing similar irrigation-related problems today. It simulates natural processes such as weather, crop growth and hydrology as well as societal ones such as farming practices and kinship-driven behaviors interweaving on a daily basis across 200-year-long scenarios.
In the model, different households respond to repeated droughts in different ways, including:
- emigrating,
- manuring fields,
- clearing more distant fields,
- violating rules about letting fields lie fallow, or
- shifting temporarily toward herding over farming.
Virtual Mesopotamia uses an archeological site now known as Tell Al-Hawa, near Mosul in northern Iraq as its basis. The 1,000 households with about 4,000 farm fields interact and conduct their activities in an environment that responds to their actions and updates itself daily.
The National Science Foundation is funding this research.
Environmental modeling
CASS is also being used to aid military land managers working to re-establish endangered species. Army land managers at large facilities have to balance military training with support of endangered species growth and other natural resource management as required by law.
To support the Army's work, DIS researchers built a model of the endangered Red-Cockaded Woodpecker population. The woodpecker lives in the southeastern United States in old, longleaf pines. In addition to being the only known woodpecker to nest and roost in living pine trees, it is also unique because some of the males remain in their birth area and act as nonbreeding helpers until the breeding male dies.
Argonne researchers working with colleagues in the U.S. Army Construction Engineering Research Laboratory created a CASS model to study the birds' dispersal and mortality. The model includes such attributes as gender, age, location, territory, role, population and lineage.
Researchers simulated seasonal changes over four years of woodpecker life, projecting population growth through space and time. The simulation is being validated in the pinelands by the land manager at Fort Benning, Ga.
"We want to expand the model to include other natural processes, land use and land management influences that act within the ecosystem," said computer modeler Pam Sydelko.
Foreign Affairs
CASCADE is another CASS program that concurrently addresses socioeconomic, psychological and environmental factors to support countermeasure analysis. The first implementation is a counter-drug simulation. The Joint Chiefs of Staff is funding the project to assist drug-interdiction analysts planning and justifying operations and force structure to combat the South American cocaine trade.
Argonne researchers are also exploring terrorist-network modeling for homeland security. The scenarios the CASS software develops go beyond what researchers can imagine and provide insight for homeland security decision making and training.
Refining RePast
In addition to creating application-specific CASS, Argonne researchers are working with the University of Chicago to extend and improve the university's RePast software. DIS is adding distributed computing and other capabilities to RePast. The Java-based modeling program runs on anything from a PC to a supercomputer.
RePast replays situations with altered assumptions and supports the modeling of agents, organizations, institutions and belief systems as recursive social constructions.
University of Chicago political scientist John Padgett is using RePast to understand the history-altering economic, technical and political developments of Renaissance Florence between 1300 and 1500. His simulation will use data sets on 50,000 persons and 12,000 political-office elections as well as records of loans, marriages, tax records and business partnerships.
Agent education
As a leader in the CASS community, Argonne has sponsored simulation courses and workshops annually since 1999. In conjunction with the University of Chicago and Sante Fe Institute, Argonne created a week-long, agent-based simulation course for industry and government software designers in defense, transportation and communications. Argonne and the university also hold annual workshops such as the October 2002 meeting "Social Agents: Ecology, Exchange and Evolution."
For more information, please contact Evelyn Brown (630/252-5510 or eabrown@anl.gov) at Argonne.
Go to next article: ANL-W, INEEL to be core of advanced nuclear power research