Mathematics and Cognition  Seminar

Spring 2002

Tuesdays 12:15 

(Supported in part by the Systems Science and Engineering Research Center)

Seminar Schedule:<http://math.la.asu.edu/~tom/cognition/math+cogsched.html>

The Mathematics and Cognition, Joint with the Stochastic Modeling Seminar Series, will
present our next lecture on Tuesday, April 16, at 12:15 PM in GWC 510. Light refreshments will be served at 12:00.

Our speaker will be Dr. Filiz Ozel, of the School of Architecture,
who will speak on the topic:

"Dynamic Simulation and Modeling in Architecture"

Abstract.

 
Buildings are complex systems comprised of components and spaces (rooms) as well as content such as goods and people, all of which have complex relationships with each other. BGRAF computer simulation is an emergency egress model that incorporates a stochastic model of the human decision making process as well a fire/smoke spread model into its structure.

The challenge in developing a simulation of dynamic processes in buildings lies in the modeling of at least three different aspects of the real world phenomenon:

1. Building model: Must include building components as well as their spatial and non-spatial relationships with each other.

2. Process model: The modeling of dynamic processes that act upon a building. Processes can not only affect the properties of a building's components over time in some manner, but can also be affected by the component properties, thus changing the nature of the evolving simulated process.

3. Decision models: Human intervention to the processes that are being simulated. Some dynamic processes develop without human intervention whereas others are very much affected by the decisions taken by humans, therefore human decision making processes must also be modeled in simulating dynamic processes in buildings. Decision models in many cases are stochastic, and require empirical data to run the simulation as well as to validate it.

BGRAF was developed at University of Michigan, and is based on an in-house CAD system for building definition. Not only algorithmic issues related to the navigation of people through spaces were resolved, but also a stochastic decision model was implemented based on data coming from reports on real world fire incidences. Two separate validation studies were undertaken by using detailed data provided by reports from National Fire Protection Association.