Study On Reduced-order Modeling Of Cryogenic Air Separation

Accurate and rapidly converged process model is necessary toreal-time optimization and control of industry plant. The thesis focuses onreduced-order model of distillation column in a large-scale cryogenic airseparation unit.The study mainly includes the following three aspects：1. Consider of the high dimension and slow computation speed ofrigorous model, aggregate model is use to simplify the simulation of adistillation column. The performance equation of Group Method is used todescribe the component distribution and some approximate equations areestablished after analyzing the degree of freedom. The Reduced-ordermodel of the distillation column is built for equation-oriented simulation.According to the simulation examples, with the increasing of the traynumber, dimension and convergence time of the model are both obviouslylower than those of the rigorous model, but the accuracy of the compositionconcentration is not reached.2. In order to improve the accuracy of the reduced order model, a ruleof how to comparting the column according to the distribution characteristics of absorption factors in high purity column is put forward.The recovery factors of high purity column-argon sidearm column studiedto analyze the effect of the number and the location of the column sections.The results show that three sections model can fit the rigorous model wellwith this kind of absorption factors profile. A method which piecewiselinearize the recovery factor profile is presented. Three sections linearmodel is much suitable to argon sidearm column as a representative of highpurity column.3. Use the improved reduced-order model to simulate thethermo-coupled lower column C1and upper column C2of air separationunits. Reduced-order model can provide obviously simple simulation of thecomplex process. The reduced dimension model can describe thecharacteristics of distillation column well.