Prediction the individual component distillation curves of the blended feed using a hybrid GDM-PcLE method

A comprehensive knowledge of the properties and characterisations of the individual
component in the blended feed is primary importance because different feedstock
blending yields different products palate. Crude oil / condensate distillation unit
optimization is an uphill task because unavailability of cheaper and reliable on line
feed and product analyzers. Furthermore, laboratory analysis for feedstock
characterization is very costly and time consuming. Alternatively, feed synthesis
technique is used to reconcile the entire range of feed distillation curves by back
blending the product streams from the actual column operation. The TBP and SG
correlation are widely been used to estimate other bulk properties because they give
the most accurate results. Due to highly nonlinear behaviour, methods like linear
regression, non linear regression and rigorous models are adopted to predict TBP and
SG distillation curves. The latter could give better accuracy results, but it is more
complex, lengthy and costly to be implemented. In addition, the rigorous model
commercially available such as PetrosimTM and Hysis 3.1TM are only being used to
predict blended feed distillation curves, not for the individual component. Thus, a
hybrid approach is proposed to overcome the deficiency of current methods and
practices. The proposed method integrates the most versatile General Distribution
Model (GDM) with a Pseudo-component Linear Equation (PcLE) method to predict
the entire range individual component TBP and SG distillation curves of the blended
feed from the readily available plant data, which are routinely taken by refiners. The
predicted results given by hybrid GDM-PcLE model are almost agreeable with the lab
results. A case study using the proposed short cut feed synthesis procedure and hybrid
GDM-PcLE model showed additional 5% Naphtha yield can be achieved by changing
the current feed blending ratio and product cut points. The accuracy of the predicting
results can be improved if the distillates samples are to be carried out simultaneously
and the flow meters are calibrated and corrected the measurements to density and
temperature of the measuring devices. Since PcLE method is simple and open
application, it can be easily integrated with iCONTM to enhance its application
predicting the pure component TBP and other distillation curves from blended feed.