In the petroleum industry, the oil fractions are usually complex mixtures containing several hundreds up to several millions of different chemical species. For this reason, even the most powerful analytical tools do not allow to separate and to identify all the species that are present. Hence, petroleum fractions are currently characterized either by using average macroscopic descriptors (density, elemental analyses, Nuclear Magnetic Resonance, etc.) or by using separative techniques (distillation, gas or liquid chromatography, mass spectrometry, etc.), which quantify only a limited number of families of molecules however. Reconstruction methods for the petroleum cuts are numerical tools, which allow to evolve towards a molecular detail and which are all based on the following principle: defining simplified but consistent mixtures of chemical compounds from partial analytical data and from expert knowledge of the process under study. Thus, the reconstruction method by entropy maximization, which is proposed in this article, is a recent and powerful technique which allows to determine the molar fractions of a predefined set of chemical compounds by maximizing an entropic criterion and by satisfying the analytical constraints given by the modeler. This approach allows to reduce the number of degrees of freedom from several thousands (corresponding to the molar fractions of the compounds) to several tens (corresponding to the Lagrange parameters associated with the analytical constraints) and to greatly decrease the CPU time required to perform the calculations. This approach has been successfully applied to reconstruct FCC gasolines by precisely predicting the molecular composition of this type of feedstocks from a distillation and an overall PIONA analysis (Paraffins, Isoparaffins, Olefins, Naphthenes and Aromatics). The extension to other naphthas (Straight Run naphthas, Coker naphthas, hydrotreated naphthas, etc.) is straightforward.
1H DOSY NMR experiments were used to investigate the macrostructure of the asphaltenes of Maya, Athabasca, and Buzurgan feedstocks in toluene-d8 at 20 °C. The influence of the concentration of asphaltenes on their diffusion coefficients is presented for the three asphaltenes. A separation between two classes of aggregates—one diffusing quickly and one diffusing more slowly—was observed at an onset concentration that is dependent upon the origin of the sample. This illustrates that the chemical interactions (and, hence, the chemical structures) are different for the three asphaltenes. Because asphaltenes are a continuum of both archipelago and continental asphaltenes, the interactions in the solutions differ, depending on the repartition between archipelago- and continental-type asphaltenes in the feed. Maya and Buzurgan asphaltenes show similar diffusion properties in the dilute regime, while Athabasca asphaltenes diffuse more slowly. Results obtained from DOSY experiments data seem to indicate that Buzurgan asphaltenes show a more continental character than the two other asphaltenes, while Athabasca asphaltenes seem to contain more archipelago asphaltenes. 1H and 13C NMR experiments were also performed to determine the average structural parameters of asphaltenes. Average sizes and molecular weights were determined from the 1H-DOSY NMR diffusion coefficients and compared to size exclusion chromatography (SEC) data.
Heavy crude oils are more and more of interest for the oil industry to meet the growing worldwide energy demand. Asphaltenes, which are the heaviest and least reactive molecules in crude oils, have received great attention over the last few decades, because they are responsible for many problems occurring during hydrotreatment and hydroconversion. 1H diffusion-ordered spectroscopy (DOSY) nuclear magnetic resonance (NMR) based on pulsed field gradient (PFG) sequences is a powerful tool to analyze polydisperse samples. The key advantage of such a technique compared to traditional NMR diffusion sequences, such as pulsed field gradient spin−echo (PFGSE), is to obtain both physical and chemical information in a single experiment. 1H DOSY NMR has been carried out on different types of samples to obtain a deeper insight into the physicochemical properties of petroleum samples, because diffusion coefficients are both sensitive to molecular weights and sizes. The application of DOSY NMR to analyze hydrocarbon mixtures and asphaltenes is assessed. It is shown that both solute and solvent diffusion coefficients decrease with an increasing concentration of solute. Different types of intermolecular interactions were observed on petroleum samples depending upon the sample concentration. Dilute and semi-dilute regimes have also been detected. 1H DOSY spectra applied to diesel and asphaltene samples in toluene are presented and interpreted to demonstrate the potential of DOSY techniques to analyze heterogeneous petroleum samples. The data obtained for the diesel sample enable us to establish that monoaromatics were connected to long alkyl chains, whereas di- or triaromatic molecules were linked to shorter hydrocarbon chains. However, a clear conclusion could not be reached for the asphaltene sample because there were only a few aromatic protons that were not detected. This observation is most consistent with a continental model of these asphaltenes.
This work concerns the development of a methodology for kinetic modelling of refining processes, and more specifically for vacuum residue conversion. The proposed approach allows to overcome the lack of molecular detail of the petroleum fractions and to simulate the transformation of the feedstock molecules into effluent molecules by means of a two-step procedure. In the first step, a synthetic mixture of molecules representing the feedstock for the process is generated via a molecular reconstruction method, termed SR-REM molecular reconstruction. In the second step, a kinetic Monte-Carlo method (kMC) is used to simulate the conversion reactions on this mixture of molecules. The molecular reconstruction was applied to several petroleum residues and is illustrated for an Athabasca (Canada) vacuum residue. The kinetic Monte-Carlo method is then described in detail. In order to validate this stochastic approach, a lumped deterministic model for vacuum residue conversion was simulated using Gillespie's Stochastic Simulation Algorithm. Despite the fact that both approaches are based on very different hypotheses, the stochastic simulation algorithm simulates the conversion reactions with the same accuracy as the deterministic approach. The full-scale stochastic simulation approach using molecular-level reaction pathways provides high amounts of detail on the effluent composition and is briefly illustrated for Athabasca VR hydrocracking.
Modeling of refining processes using metal-acid bifunctional catalysts involves an exponentially increasing number of species and reactions, which may rapidly exceed several thousands for complex industrial feedstocks. When building a model for such a process, a priori lumped kinetic models by chemical family do no longer meet the current requirements in terms of simulation details, predictive power and extrapolability. Due to the large number of elementary steps occurring in bifunctional catalysis, it would be quite unrealistic to manually build a detailed kinetic network of this scale. Hence, computer generation of the reaction network according to simple rules offer an elegant solution in such a case. Nevertheless, it remains difficult to determine and solve the kinetic equations, mainly due to the lack of analytical detail required by a detailed model. For several refining processes, however, reasonable assumptions on the equilibria between species allow to perform an a posteriori relumping of species, thus reducing the network size substantially while retaining a kinetic network between lumps that is strictly equivalent to the detailed network. This paper describes a network generation tool and the a posteriori relumping method associated with the single-event kinetic modeling methodology. This a posteriori relumping approach is illustrated for and successfully applied to the kinetic modeling of catalytic reforming reactions.
Many applications in the oil and gas industry require modeling physicochemical properties of complex mixtures. In this work, we propose a methodology to predict the interfacial tension of water/crude oils by modeling the composition of crude oil samples using a combined approach of experimental characterization, molecular representation (surrogate), and mesoscopic simulations such as dissipative particle dynamics (DPD). The methodology for molecular representation is based on the experimental analysis by separation of crude oil according to the number of carbon atoms in molecules into two fractions: C20– and C20+. A lumping approach was applied to the C20– fraction and a stochastic reconstruction approach was employed on the C20+ fraction. The influence of the different variables (chemical diversity and number of molecular types in the C20+ fraction) of the models was analyzed to propose surrogates based on building units with different functional groups. Based on a previous work (Steinmetz et al. J. Chem. Theory Comput. 2018, 14, 4438–4454), a thermodynamically consistent methodology was applied to obtain the DPD interaction parameters of the different chemical building units. DPD simulation on the model crude oil provides predictive values of the interfacial tension that are in good quantitative agreement with the experimental data.
This work concerns the study of sediment formation in ebullated-bed hydroconversion of vacuum residues and investigates ways to avoid their formation at high conversion operation. Tests were carried out in a bench-scale unit at high temperature using a vacuum residue feedstock. Feeds and effluents were characterized to follow sediment formation, as well as conversion, yield structure, and hydrotreating performances. It was found that co-processing a vacuum residue feedstock with low amounts (<15 wt %) of heavy cycle oil from a fluidized catalytic cracking (FCC) unit can significantly improve the stability of the unconverted effluents. A detailed characterization of feedstock and products was carried out to understand the stabilization mechanisms. The effect of heavy cycle oil (HCO) on sediment reduction was attributed to the polycondensed tri-, tetra-, and penta-aromatics. It was also found that these species do not affect the hydroconversion performances and are refractory to conversion.
