[CCoE Notice] PhD Defense: Building Models of Process Systems: Applications in Control & Design

[Grayson, Audrey A]

PhD DEFENSE STUDENT: Shobhit Misra
DATE: Friday, October 27, 2017
TIME: 2:00 PM
PLACE:  MREB Building, Conference Room #320
DISSERTATION CHAIR: Dr. Michael Nikolaou
________________________________
TITLE:
Building Models of Process Systems: Applications in Control & Design

Traditional or new methods for building models of process systems can combine physical understanding with statistical analysis so that models can be developed for use in many fields of engineering.  Numerous engineering applications are taking advantage of models derived from data that is either recorded during the normal course of process operations or obtained in a deliberately designed experiment.  This work focuses on developing and refining methods for identification of models that find applications in a) designing recipes for cementing natural gas wells that avoid leaking to water aquifers or the atmosphere, and b) design of automatic control systems for a variety of processes.
Availability of large amounts of data that is recorded during hydrocarbon well construction makes it possible to “learn” from such data, i.e. build a model that describes how various (over two dozen) factors involved in the design of wellbore preparation and cementing jobs affect the tendency of a cemented gas well to leak or not.  In addition, such models can offer significant insight by ranking the most important factors that affect leakages.  Capturing the effect of all such factors using first-principles models would be overly complicated.  Addressing gas leakage problems has significant environmental implications, as leaking wells pose serious pollution threats to both groundwater and air.
Model quality can also be improved by using data from experiments deliberately designed for building a model with particular properties.  Explicit use of models in multivariable control systems is one such application.  The control-relevant model building method entails two steps, namely estimation of (a) model structure, i.e. model order, and (b) values of model parameters.  Of the two steps, the first one is crucial, as it affects the second one and consequently the overall quality of the model.  A new approach to rigorous design of experiments (DOE) is proposed that accurately estimates the model order even for challenging systems.  Simulations on realistic industrial models exemplify the proposed approach.

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://Bug.EGR.UH.EDU/pipermail/engi-dist/attachments/20171025/761b5dd2/attachment.html