Combining COMIS and Potential Flow Theory to Predict Airflows in Buildings for Real-Time Analysis of Sensor Data

Speaker(s): 
Date: 
February 23, 2010 - 11:00am
Location: 
90-3075
Seminar Host/Point of Contact: 

Interpretation of data from air-monitoring sensors in real-time requires: a) simulations of many sets of airflow and building parameters to assess a representative range of events before the onset of an actual event; and b) rapid statistical (Bayesian) comparison of sensor data to simulated events. In this talk, I will show my implementation of a C++ program that calculates velocity fields in user-defined buildings using an innovative approach to potential flow theory (modified potential flow theory or MPFT) as part of my Ph.D. project. MPFT shapes flows near supply inlets as turbulent jets and approximates buoyant flow near people. Therefore, the fidelity of MPFT should fall between that of well-mixed codes / potential flow theory and computational fluid dynamics (CFD). However, MPFT is able to calculate airflow patterns in large structures in a fraction of the time required by CFD (e.g. minutes vs. several hours in FLUENT®). I hypothesize that individual-room velocity fields in large buildings calculated using MPFT and boundary conditions provided by COMIS (and subsequent concentration transport calculations) may provide improved fidelity over stand-alone COMIS simulations without being too computationally burdensome for this application (e.g. a CFD-COMIS pairing). The added resolution provided by MPFT may be used to refine the placement of sensors within rooms. Finally, I have experience in medical image analysis, including expectation-maximization algorithms, Bayesian updating, and machine learning / feature selection. I will present an overview of my work in these areas followed by a short discussion on possible translations of these techniques to indoor-air problems such as diagnosing sick building syndrome and generating rules-of-thumb for emergency response. For more information about this seminar, please contact: Michael Sohn(510) 486-7610

Add event to Google Calendar

Announcement List

For notification about future seminars, contact Erin Bertiglia.

Site Access

Schedule subject to change without notice. If you are coming from off-site, please call first to verify. UC staff and guests are welcome. LBNL shuttle buses stop every few minutes at marked sidewalk locations along Bancroft and Hearst Avenues and Rockridge BART.