Distillation end point estimation in diesel fuel production

Soft sensors for the on-line estimation of kerosene 95 % distillation end point (D95) in crude distillation unit (CDU) are developed. Experimental data are acquired from the refinery distributed control system (DCS) and include on-line available continuously measured variables and laboratory data which are consistently sampled four times a day. Additional laboratory data of kerosene D95 for the model identification are generated by Multivariate Adaptive Regression Splines (MARSplines). Soft sensors are developed using different linear and nonlinear identification methods. Among the variety of dynamic models, the best results are achieved with Box Jenkins (BJ), Output Error (OE) and Hammerstein-Wiener (HW) model. Developed models were evaluated based on the Final Prediction Error (FPE), Root Mean Square Error (RMSE), mean Absolute Error (AE) and FIT coefficients. The best results for diagnostic purposes show BJ model. For continuous estimation of D95, OE and HW models can be used. Key words: Crude distillation unit, distillation end point, soft sensor, identification