Collaborative Research: Design and Model-based Control of Nonlinear Chemical Processes

点击翻译按钮获取中文摘要

ABSTRACTPI: Masoud Soroush Institution: Drexel University Proposal Number: 0101133To achieve greater profitability, process designers have been creating design in regions involving complex nonlinearity where process controllers continue to face stiff challenges. Steady state multiplicity, limit cycles, chaos, and parametric sensitivity are manifestations of the nonlinearity. In addition to the nonlinearity, there are many process designs that have unstable and/or non-minimum-phase steady states. In many cases, operation is more profitable at an unstable steady state, or at a stable steady state in the close proximity of an unstable steady state, often involving non-minimum-phase behavior (inverse response). Examples of process designs that can show such behavior are chemical reactors, fermentation reactors, fluidized catalytic crackers, reactor-separator-recycle plants, azeotropic distillation towers, and reboilers. Processes with such designs are known to be more challenging to control than processes with stable and minimum-phase steady states.This University of Pennsylvania-Drexel University joint research project is aimed at addressing more efficient and easier operation of processes with unstable and/or non-minimum-phase steady states. It is a study of the interactions between design and control in the processes, and strategies to develop process designs that make control of the processes easier. The designs of existing chemical and biochemical processes with unstable and/or non-minimum phase steady states will be studied to identify the design features that give rise to such process behavior. Alternative new process designs, if possible, will then be developed for the processes such that the same level of profitability is maintained but the instability and non-minimum-phase behavior are eliminated. For processes for which such desired alternative designs cannot be found, new differential-geometric, model-based control laws will be developed that are applicable to general nonlinear processes, whether minimum-phase or non-minimum-phase. Initially, control laws will be developed for general, nonlinear, continuous-time processes without deadtimes or input-saturation constraints. Subsequently, the latter two complications will be addressed, and the methods will be extended to multi-input-multi-output (MINO) systems and tested for chemical, petrochemical, and biochemical processes. A prototype symbolic-manipulation software that, given a process model, automatically generates differential geometric control laws will be developed to simplify their industrial implementation and testing.