Soft Sensor Development for Monitoring ASTM-D86 Index: Effect of Feed Flow Rate Change

Changes in a crude oil flow rate to an atmospheric distillation unit can influence the quality of the products. This paper presents a modification method for a soft sensing model, including an update term, which makes it compatible with industrial variations. A modified soft sensing structure is adopted using the lookup table (LUT) method where steady-state soft sensing models are performed. The steady-state soft sensing models are proposed based on the local instrumental variable (LIV) technique for an industrial atmospheric distillation unit (ADU) at Shiraz refinery, Iran. The LIV-based soft sensors use tray temperature measurements to monitor the ASTM-D86 index of side products for a nominal flow rate (60,000bbl/day). The lookup tables have been developed based on the difference between the predicted values of the ASTM-D86 index and the corresponding simulation values to make update terms in different feed flow rates. The results present improvement in the predictions of LIV-based soft sensors, as well as acceptable control performance in feed flow rate variations. The comparison of soft sensing results with/without the lookup tables demonstrates that the proposed update term helps to predict product quality more precisely and is suitable for advanced monitoring scheme due to no complexity and low computational time.