Seminar 2x45mins
Abstract
Dr. Songbo Cong
A coordinated control strategy is presented to fit the comprehensive requirement of petrochemical
process, especially of separation processes, where there frequently are many manipulated variables
available. In a FAT system with more inputs than outputs, the controller will first select more
reasonable manipulated variables by using optimization logic, then a predictive controller or ordinary PID
controller is used to make controls available. In this system, gas end point and diesel pour point are two
controlled variables which are calculated by online observers. Control results are
illustrated by the data collected from the controlled units.
2) Structural observability and petroleum distillation column
An analysis method of structural observability and controllability for processes with
time delay is given. The result was used to analyze the observability of petroleum
distillation column, guiding the online calculation of non-measurable variable in FCCU
main fractionator.
In structural observabilty and controllability analysis, no numerical dynamic model is
needed, only is the model structure necessary, which makes it convenient for a control
system to be designed for a unit before it being put into use.
This method can not only be used in linear systems but also used in non-linear systems.
3) Model based coordinated control
A model based multivariable coordinated control algorithm is presented, which
brings both the control performance for controlled variables and the Ideal Resting
Values (IRV) for manipulated variables into consideration. In a system with more input
than output, the controller will first makes full use of effort to drive the controlled
variables to their setpoints, then, if it is possible, makes some of the manipulated
variables approaching to their IRV in economic consideration. In a system with less
inputs than outputs, the controller enforce different control efforts to the controlled
variables according to their importance.
Both linear system and non-linear systems are included. For a linear system, the
condition for unique solution, steady state property and stability of the controller are
given. For a non-linear system, a SQP algorithm is used to solve the non-linear
programming program problems.
A FCCU fractionator is used here as a case studied. Simulated results
demonstrated the performance of this control algorithm.
--=====================_895656023==_-- ------------------------------------------------------------------------ To cancel your subscribtion/subscribe to this list: Send mail to majordomo@chembio.ntnu.no with "UN-/SUBSCRIBE prost" in the messagebody.