Design Feasibility Analysis and Optimization under Uncertainty - A Bayesian Optimal Decision Framework

October 7, 2003 - 12:00pm
Bldg. 90
Seminar Host/Point of Contact: 

A new approach to the problem of identifying design feasibility and optimality under uncertainty is introduced. Based on the Bayesian concepts of predictive probability and expected utility, the method can quantify the feasibility of a process design and identify the optimal operation conditions when there are uncertainties in the process parameters. The use of Bayesian statistics enables the treatment of a very wide class of parameter uncertainties, including simple bounds, analytic probability density functions, correlation structures and empirical distributions. Numerical solution of the feasibility and optimization problems is accomplished using Markov Chain Monte Carlo (MCMC) numerical techniques. Examples used to illustrate the new approach are drawn from: chemical kinetics, process operability and the design of unit operations where the information on the model parameters is obtained from statistical analysis of experimental data.

Add event to Google Calendar

Announcement List

To be added to our seminar annoucement list please, email ETA Seminars. Seminars may change without notice, stay informed by joining our seminar email list. 

Site Access

For site access please email your request to ETA Seminars. To arrive at Building 90 via the LBNL shuttle, take "Blue Uphill". Click here for the shuttle schedule. To ensure your site access is set-up send your request 36 hours in advance of the seminar.

Remote Access

Not able to attend in person? Remote participants can use ReadyTalk Audio & Net conference to phone in to listen, while viewing the presentation via the ReadyTalk site.

Web Access Code: 4864835

USA & Canada call-in number:
1 (866) 740-1260
Access Code: 4864835#

International callers:
Please visit this website to see if a toll-free call-in number is available for your international location.

If not, you will need to pay for the call yourself and use this phone number:

(303) 248-0285.