ssg

Agent-based modeling seems to be gaining a lot of attention. Bo Parker, PricewaterhouseCoopers, presented how agent-based modeling is helping transform business enterprise by modeling and acurately predicting the outcome of complex adaptive systems. The presentation was given at a packed audience at the Nouvelle Theatre during Oracle Open World 2009. The material applies to nearly all aspects of an enterprise.

Complex adaptive systems relate to social networks, customer behaviour, game theory and process improvements. Agent-based modeling is used to predict the outcome of the systems as-they-are or when changes are introduced into the system. It has been used primarily for academic purposes but with new visualization based tools such as AnyLogic, XJ Technologies, they are becoming more common-place in industry as well.

Mr. Parker presented several case-studies involving capacity planning for health care facilities and predicting tragic events as occurred at a British-Patroleum refinery in Texas City, Texas.

March 2005, an explosion rocked the small town of Texas City, killing 15 people and injured 150. The agent-based model was used to model simple human behavoiur mechanism such as how long it takes to return phone calls and distribute emails. The simulation was run over several scenarios to determine how and why a company like BP with good process controls and good safety records, fell prey to such a catastrophe. The models showed how insignificant behaviour like responding to email messages and phone calls in relation to process improvement initiatives, caused the failure. Without such models something as seemingly insignificant would have been over-looked

Mr. Parker sees a growing trend to use these models to transform the enterprise allowing to be more adaptive, anticipate change; without adverse affects. A company like 3M would have benefited by this when it began 6-sigma improvements.

In the 1990′s after major successes in innovation, 3M began suffering from poor profits. They brought in a new CEO with proven 6-sigma credentials who cut costs, improved quality and regained profitability; only to see a sudden drop in innovation. Agent-based modeling could have predicted and avoided this harmful side-affect.

As agent-based models become prolific, Mr Parker expects a convergence of models. Models for Operations, Enterprise Architectures and Business Process Management will converge as they take advantage of the predictive abilities of agent-based models. This will be evident, Mr Parker believes, as the semantics of these tools begin to blur the current boundaries of these tools.