engleski

Soft sensor for cold filter plugging point estimation

Soft sensors for quality estimation of diesel fuel in the crude distillation unit (CDU) are developed. Due to the growing fuel quality norms and needs for various quality gradations of diesel fuel, continuous laboratory testing and product quality control are necessary. One of the key diesel fuel properties is cold filter plugging point which is determined only by laboratory assays. On the basis of available continuous measurements of temperatures and flows of appropriate process streams, several soft sensor models for estimating cold filter plugging point have been developed and tested. Experimental data are obtained from the refinery distributed control system (DCS) and include on-line available measured variables and laboratory assays. Data preprocessing included detecting and outlier removal, generating additional data by Multivariate Adaptive Regression Splines (MARSplines) algorithm, offset removal i.e. detrending data and filtering data. Soft sensors are developed using different linear and nonlinear identification methods. From variety of developed models (autoRegressive Moving Average with eXogenous inputs (ARMAX), Output Error (OE), Nonlinear AutoRegressive model with eXogenous inputs (NARX), Hammerstein–Wiener (HW) and neural network models) the best achieved results are shown. Statistical data analysis has been carried out and the results were critically judged. The results of this research indicate that soft sensor models for refinery product quality estimation of diesel fuel can be successfully applied as an alternative for laboratory testing.

soft sensor; crude distillation unit; dynamic modeling; cold filter plugging point

nije evidentirano