A NUMERICAL AND AN ANALYTICAL APPROACH TO APPROXIMATE DYNAMIC MODELING OF DISTILLATION TOWERS (RECYCLE, TIME-CONSTANTS, SHORT-CUT, LINEAR-ANALYSIS).

摘要

This study discusses the need for developing short-cut modeling methods as an alternative to dynamic simulation of chemical process systems. The resulting approximate models developed through short-cut techniques are useful for operability analysis and operator training. For example, some companies use approximate models to decide operability questions such as whether or not surge tankage is needed on the feed to a distillation tower and what types of control schemes to use.;The above conclusion of using a perturbed steady state has been applied to towers by two methods developed in this thesis. The first method is a numerical approach for predicting the transient responses of the product compositions. This approach is relatively simple to use and less time consuming compared to the dynamic simulation technique. The second method is a purely analytical technique for predicting the transient behavior of distillation towers. Simple analytical expressions are derived that relate the state of the towers to their dynamic parameters. Further, the analytical technique helps in identifying the key sections of a tower that are the major contributors to its non-linear transients.;The methodology of modeling a chemical/petroleum plant by decomposing it into various processing configurations is employed in this study. In particular, recycle processing configurations and their effect on the dynamic behavior of distillation columns are analysed. Distillation towers are inherently recycle processes. Linearization techniques have been used to estimate approximate tower time constants. For high purity towers the resulting estimates are known to be substantially in error when compared to either actual or simulated tower responses. In this thesis it is shown that the recycle structure of distillation towers is the reason that the linearized estimates are in error. It is further demonstrated that substantially better estimates of tower dynamics can be obtained if a perturbed steady state rather than the design steady state is used in the linearization analysis for high purity towers. A number of other low and moderate purity towers with a variety of characteristics have been studied. The key conclusion about using a perturbed steady state appears valid for all towers.