A Framework for Multi-layer Data Reconciliation in Chemical Process Systems

Jixin Qian

2009-01-01

Abstract:Reliable and consistent process data are very important in chemical process system.As a(result) of random and possibly gross errors,data reconciliation and gross error detection are needed to(minimize) the measurement errors.As the measured variables are usually different types in chemical process systems,the problem of data reconciliation for different types of measured variables is waited to be solved.In this paper,we present a framework for multi-layer data reconciliation in process systems,which can(adjust) the model to be used for data reconciliation.Data reconciliation for different types of measurements is classified into three layers in this framework.The first layer is based on the total mass balances.The second layer is based on not only the total but also the component mass balances.The last layer is based on the rigorous model.In each layer,a weighted least square objective function will be used.The strategy for gross error detection in the framework will be proposed,which is robust and efficient.With the framework,whatever types of measurements in the chemical process system present,it can adjust the model to be fit for data reconciliation.The effectiveness of this framework is demonstrated on the simulation of distillation column system and air separation system.Result shows that the framework is flexible and useful.