409898 Validation and Sensitivity Analysis of a Continuous Lumping Model for Hydrocracking on a Zeolite Catalyst

Thursday, November 12, 2015: 1:54 PM
Salon E (Salt Lake Marriott Downtown at City Creek)
Per Julian Becker, IFPEN, LYON, France, Benoit Celse, Process Division, IFP, Vernaison, France, Victor Costa, Process Design and Modeling Division, IFP, Vernaison, France and Denis Guillaume, IFP-LYON, Solaize, France

Validation and Sensitivity Analysis of a Continuous Lumping model for Hydrocracking on a Zeolite Catalyst

Per Julian Becker1, Benoit Celse1*, Victor Costa1, and Denis Guillaume1

1. IFP Energies Nouvelles, Rond-Point de l’Echangeur de Solaize, 69360 Solaize, France

*E-mail: benoit.celse@ifpen.fr


• Validation of a Continuous Lumping Model.

• Parameter fitting on 57 experimental runs.

• Good prediction of conversion (370+ cut) and yield structure.

• Improved performance compared to correlative simulator.

• Sensitivity study of operating conditions and feed characteristics.


Catalytic hydrocracking is used to convert heavy Vacuum Gas Oil (VGO) residue to more valuable middle distillate (150-370°C) and/or naphta (>150°C) cuts. A robust kinetic model allows the optimal process design and operating conditions to be chosen to maximize the desired cuts and product characteristics. Hydrocracking of VGO residue performed in a two-step process: 1) a hydrotreatment step in reactor R1, which serves mainly to remove nitrogen and sulfur from the feed; 2) a hydrocracking step in the second reactor (R2), which performs the main hydrocracking on a zeolite catalyst.

A continuous lumping model for R2 is proposed and validated in this work. This approach groups the individual components of the hydrocarbon mixture according to a continuous distribution with normalized boiling point C(θ,t). The cracking kinetics are governed by a continuous reactivity function f(θ), and a yield distribution g(θ,θ*). The inhibition of the zeolite catalyst due to residual organic nitrogen and ppNH3 are taken into account by IHDN.


A total of 12 model parameters are adjusted using 57 experimental runs in a pilot plant, with different Iranian/Arabian feeds. About half the feeds were used in a two-step process (R1+R2) where gasses (H2S, NH3, and hydrocarbon C1-C4) are carried over from the first reactor. The remainder of the tests used pre-treated feeds, where the ppNH3 was simulated by adding an equivalent amount of aniline additive. Feed characteristics (SIMDIS, sulfur/nitrogen content, aromatic carbon content etc.) for the feed of the second reactor were known. Temperature was varied between 360 – 400°C, and liquid hourly space velocities (lhsv), which corresponds to the inverse of contact time, was varied between 1.0 – 3.5 h-1. Experiments were performed at 100 and 140 bar, with a ppNH3 between 0.1 – 1.5 bar and a H2/HC of 600 – 1400 l/l. An additional set of 15 experimental runs was used to validate the model. These tests used similar feeds and operating conditions as the ones used in the calibration database. Parameter identification was performed on a supercomputer, using a set of 96 randomized initial parameters, in order to ensure convergence of the optimizer to a global, rather than a local, minimum.


The comparison with a correlative model (with more than 30 parameters) for the experimental validation database (15 points) on conversion and yield was carried out. It shows that the continuous lumping model provides better simulation results. The parity graphs show that the validation points are well predicted, even for relatively high values of lhsv (>3 h-1).

A sensitivity study allows to further validate the model. The influence of variations of the most important operating conditions on the model predictions for the conversion X370+ is quantified. The operating conditions studied are : 1) lhsv (liquid hour space velocity), 2) H2/HC, 3) the nitrogen content of the feed of R1 (NR1), 4) the residual nitrogen content of the feed of R2 (NR2). The shifts in ΔT (to obtain the right conversion), shown below, are within the ranges expected from experimental observations. More global sensitivity analyses using Sobol index may be carried out.


 [1] PJ Becker, B Celse, D Guillaume, H Dulot, V Costa, Hydrotreatment modeling for a variety of VGO feedstocks: a Continuous Lumping Approach, Fuel 139 (2015) 133-143

Extended Abstract: File Not Uploaded
See more of this Session: Process Modeling and Identification II
See more of this Group/Topical: Computing and Systems Technology Division