Home
About Us
News
People
Research
Industry
Learning
Grad Studies
Contact
Downloads
Login
Jump To Topic...
Theme Topic
Model Predictive Control of Batch Processes and Bio-Systems

Several chemical processes, including specialty chemicals, polymers and pharmaceutical are operated in batch fashion. The control and operation of batch processes, is however intrinsically different from continuous operation, because the control objective is not that of stabilization, but achieving desired values of the process variables at the end of the batch. The desired end-point conditions are not necessarily an equilibrium point of the process, and therefore, the large body of advanced control tools developed for continuous operation are not directly applicable to batch processes. Existing results on batch process operation either rely on open-loop operation policies or set point trajectory tracking for closed--loop operation, or utilize repeated implementation of model predictive control calculations. Existing results, however, do not allow for an explicit computation of the set of initial conditions from where a desired end-point can be reached. The absence of such a characterization prevents determining, short of solving an optimization problem, whether uncertainties in the batch feeds can be tolerated. The lack of such characterization also prevents the design of fault-handling frameworks for batch systems.

Motivated by these considerations, we recently addressed the problem of controlling batch processes to achieve a desired final product quality subject to input constraints and faults in the control actuators. Specifically, faults are considered that cannot be handled via robust control approaches, and preclude the ability to reach the desired end-point, necessitating fault-rectification. A safe-steering framework is developed to address the problem of determining how to utilize the functioning inputs during fault rectification in order to ensure that after fault-rectification, the desired product properties can be reached upon batch termination. To this end, first a novel reverse-time reachability region (we define the reverse-time reachability region as the set of states from where the desired end point can be reached by batch termination) based MPC is formulated that reduces online computations, as well as provides a useful tool for handling faults. Next, a safe-steering framework is developed that utilizes the reverse-time reachability region based MPC in steering the state trajectory during fault rectification to enable (upon fault recovery) the achieving of the desired end point properties by batch termination. The proposed controller and safe-steering framework were illustrated using a fed-batch process example (Aumi and Mhaskar, 2009).

Dr. Prashant Mhaskar
Professor and Canada Research Chair (Tier II)
Handling multi-rate and missing data in variable duration economic model predictive control of batch processes
AIChE J (2017)  -  [ Publisher Version ]
Subspace Model Identification and Model Predictive Control Based Cost Analysis of a Semicontinuous Distillation Process
Computers & Chemical Engineering, 103 39-57 (2017)  -  [ Publisher Version | Open Access Version (free) ]
Multi-rate modeling and economic model predictive control of the electric arc furnace
J Process Control, 40 50-61 (2016)  -  [ Publisher Version ]
Data-driven model predictive quality control of batch processes
AIChE J, 59 (8) 28522861 (2013)  -  [ Publisher Version ]
Data-based Modeling and Control of Nylon-6,6 Batch Polymerization
Aumi, S.Corbett, B.Mhaskar, P., Clarke-Pringle, T.
IEEE Trans. Contr. Syst. Tech., 21 94 - 106 (2013)  -  [ Publisher Version ]
Integrating Data-Based Modeling and Nonlinear Control Tools for Batch Process Control
AIChE J., 58 (7) 2105-2119 (2012)  -  [ Publisher Version ]
Energy Efficient Model Predictive Temperature Control
Wallace, M., McBride, R., Aumi, S.Mhaskar, P., Salsbury, T., House, J.
Chem. Eng. Sci., 69 45-58 (2012)