Srinivas Karra

PhD Candidate
(curriculum vitae)

Year Started: Fall 2005

 

B.S. in Chemical Engineering, National Institute of Technology, Warangal, India (2003).
M.S. in Chemical Engineering, Indian Institute of Technology Bombay, India (2005).

Email: srini.karra@ttu.edu
Phone: (806)282-7626
Fax: (806)742-3552

Honors and Awards:

PhD Research Topic:
Modeling, Identification and, Control of Complex Systems – A New Paradigm


Understanding the physical phenomena and ‘engineering’ the processes involve methods to correlate process parameters that determine the progression and state of the process.  These correlations form the basis for engineering analysis, design, optimization and control of the process of interest.  However, in practice there is no perfect mathematical model that can accurately represent a physical process.  This is because of various reasons namely; uncertainty in the process, nonlinearity associated with cause and effect relationships, and time varying external disturbances. 


Usage of linear dynamic models for describing such systems irrespective of their complex nature is pervasive in control engineering applications.  Such models are useful only in small ranges of operating conditions and may be no longer valid outside this window.  This can be mathematically perceived as a change in the input-output causality when the system operating conditions are different from that of the conditions when the linear dynamic model is built.    However, poor agreement between model predictions and output data does not imply that input-output causality has changed from the time of modeling, because the unmodeled component may not be necessarily from change in process dynamics.  Very often time-varying external disturbances that are entering the process may lead to plant-model mismatch.  For these reasons, model validation is an important tool which is used not only as a final ‘quality control station’ before controller deployment, but also as ‘supervisor station’ whiles the controller is active to detect any changes in process behavior.  While the model validation at the initial system identification stage is well addressed, on-line model validation using closed-loop data is still a challenging area. Even after isolating the proper root cause for plant model mismatch (PMM), it is a difficult task to perform system identification in presence non-stationary external disturbances.


In this work, an attempt has been made to resolve some of the issues posed by above mentioned complexities using simple linear dynamic models.  Following figure outlines the major contributions of my research.

Journal Publications:

Peer Reviewed Conference Proceedings:

Patents

Book Chapters