(111c) Molecular Composition Modelling of Vacuum Gas Oils Via a Molecular Reconstruction Approach

A Vacuum Gas Oil (VGO) is a complex hydrocarbon mixture of several thousand different chemical species. Describing the chemical composition of the feedstocks is essential to the design and operation of these refining processes. Although the most advanced analytical techniques, such FT-ICR/MS, enable to identify a large number of compounds and classes of chemical families, a molecular composition of the VGO cuts cannot be obtained directly, even by means of cutting-edge analytical techniques. To overcome this drawback, a molecular representation of the feedstock can be built using a molecular reconstruction algorithm. The basic idea behind this type of algorithms is to generate a discrete set of molecules from overall petroleum analyses, model hypotheses and chemical knowledge.

During previous work [1-3], a two-step molecular reconstruction algorithm was developed to model the composition of VGO cuts by using two methods in series: Stochastic Reconstruction (SR) [4] and Reconstruction by Entropy Maximization (REM) [5]. The SR step uses a Monte-Carlo approach to generate the initial set of molecules. In the REM, the mole fractions of the molecules are adjusted by maximizing an information entropy criterion.

Even that this algorithm can successfully reconstruct the VGO cuts, the SR step still remains a computationally demanding algorithm.

This work focuses on the development of an alternative approach of this base algorithm which allows reducing the computational burden when several VGO cuts need to be reconstructed. The idea of this alternative approach, called indirect two-step molecular reconstruction (SR-REM) algorithm, consists of using the SR method to build a reference database of molecules for all VGO cuts. The abundance of each molecule of the database is then adjusted via the REM method in order to obtain a synthetic mixture whose properties are closer to the analytical data of the VGO to be represented.

The indirect SR-REM algorithm was applied to eight different VGO cuts that are either straight run VGOs obtained from crude oils from different geographical origins or conversion VGOs from various refining processes. The VGOs have been characterized by means of elementary analysis (C,H, S, N), density, detailed Fisher mass spectrometry, liquid chromatography S.A.R. distribution as well as sulfur and nitrogen (total, basic and non-basic nitrogen) simulated distillation.

The reference database was built by collecting the mixtures of 5000 molecules for three VGO cuts. In this work, VGO molecules are constructed from 16 structural attributes which enable to generate acyclic molecules (paraffins, iso-paraffins) and cyclic molecules, the latter being composed of napthenic rings, aromatic rings and heterocyclic rings (thiophenes, pyrolles and pyridines). The structural attributes are described by three type of PDFs: Histograms, Exponential distributions and Gamma distributions.

The application of the indirect molecular reconstruction algorithm yields a synthetic mixture of 15 000 molecules that mimics very well the properties of all VGOs. The results of the indirect reconstruction are similar with those obtained with the direct molecular reconstruction algorithm. This molecular reconstruction algorithm advantageously combines the power of the SR method to generate appropriate molecules with the flexibility of the REM method, while completely eliminating the computational burden when several VGOs need to be reconstructed.

References

[1] J. J. Verstraete, N. Revellin.,H. Dulot, D. Hudebine, Preprints Paper - ACS, Division Fuel Chemistry. 49 (2004) 20-21

[2] N. Charon-Revellin, H. Dulot,C. López-García, J. Jose, Oil & Gas Science and Technology. 66 (2011) 479-490

[3] A. Alavarez-Majmutov, R. Gieleciak, J. Chen, Enery & Fuels. 29 (2015) 7931-7940

[4] M. Neurock , A. Nigam, D.M. Trauth, M.T. Klein, Chem. Eng. Sci., 49 (24) 4153-4177

[5] D. Hudebine, J.J. Verstraete Chem. Eng. Sci., 59 (2004) 4755