(19b) Near-Infrared Modeling of Conjugated Diolefins in Selective Hydrogenation Units

The increasing worldwide demand of fuels requires conversion of heavy petroleum cuts into valuable products such as gasolines or diesels. With over 350 units in operation, Fluid Catalytic Cracking (FCC) is currently the dominant refinery conversion process that allows the upgrading of heavy cuts to produce gasolines. FCC gasolines contain unsaturated aliphatic compounds such as olefins and diolefins. Olefins are valuable compounds since they have high octane numbers. However, conjugated diolefins are undesired compounds since they are reactive to polymerization, provoking plugging problems and deteriorating gasoline stability. To remove these diolefinic compounds, FCC gasolines are treated in Selective Hydrogenation Units (SHU). The main objective of this process is to selectively hydrogenate the conjugated diolefins while minimizing the olefins hydrogenation. To determine the performances of the SHU units, it is essential to verify the conjugated diolefins content of the gasoline leaving the process. The conjugated diolefins are usually measured via indirect methods such as the Maleic Anhydrid Value (MAV) that is based on the Diels-Alder reaction between the conjugated diolefins and maleic anhydride. However, this analysis is impractical for continuous monitoring of the process since it is very time consuming (> 5 hours per sample) and not suitable for automation. For these reasons, Near InfraRed spectroscopy (NIR) was used to model the MAV content in SHU effluents. This technique has the advantages of being fast and repeatable but it is generally not adapted for trace analysis. Since the diolefins contents in the SHU effluents are very low (<1 wt%), a specific spectra preprocessing methodology was adapted to model the MAV content in these samples. The results of the Partial Least Squares (PLS) regression demonstrate that the NIR model can accurately predict the MAV: 98% of the calibration samples are predicted in the confidence interval of the chemical MAV method. A very good agreement between experimental and predicted MAV was also obtained for the external validation data (Figure 1). The NIR model was tested for SHU pilot plant monitoring. The model can correctly predict the variations of MAV observed when the process operating conditions are modified (temperature, hydrogen flowrate). Therefore, the NIR prediction of the conjugated diolefins content in selectively hydrogenated gasolines via the MAV is a very interesting tool to determine in real time the performances of the selective hydrogenation units.