Search

The problem of synthesizing the control structure of a complete chemical plant is addressed. Due to the fact that this topic has found virtually no coverage in the past, a qualitative outline of the different necessary steps is presented initially. The multilayer-multiechelon concept is used to decompose the overall problem into manageable subproblems using developed quantitative criteria to classify disturbances (in order to split the control task into an optimizing and a regulatory one) and to estimate the loss caused by insufficient coordination. Feedback optimizing control is discussed and a criterion for its attractiveness is derived. Alternative regulatory control structures are generated by an algorithmic procedure based on extensions of the concepts of structural controllability and observability. The method requires simple structural models only and includes the development of the usually encountered feedback structures where the number of control inputs is equal to the number of measured outputs. For the case that some variables cannot be measured easily on line, criteria for the selection of secondary measurements are developed. Those indices give a measure of the static and the dynamic error when the estimation is performed through a simple linear combination of the measurements. As a dynamic compensation for this kind of static estimator, the Luenberger observer and the Kalman filter are modified to account for the presence of unknown persistent disturbances. The well known augmentation of the state vector is shown to lead to observability problems when the number of unmeasured inputs exceeds the number of observations and a method is presented to overcome those difficulties in a steady state optimal fashion.