Group-Type Analysis of Heavy Crude Oils Using Vibrational

Jan 19, 2005 - Twenty heavy and/or particle-rich crude oils have been quantitatively fractionated into saturates, aromatics, resins, and asphaltenes (...
2 downloads 0 Views 155KB Size
Ind. Eng. Chem. Res. 2005, 44, 1349-1357

1349

Group-Type Analysis of Heavy Crude Oils Using Vibrational Spectroscopy in Combination with Multivariate Analysis Andreas Hannisdal,* Pål V. Hemmingsen, and Johan Sjo1 blom Ugelstad Laboratory, Department of Chemical Engineering, Norwegian University of Science and Technology (NTNU), N-7491 Trondheim, Norway

Twenty heavy and/or particle-rich crude oils have been quantitatively fractionated into saturates, aromatics, resins, and asphaltenes (SARA) by asphaltene precipitation in n-hexane and highperformance liquid chromatography (HPLC). The newly developed and fully automated HPLC method has a sample capacity corresponding to 0.6 g of heavy crude oil. The crude oils have been characterized by vibrational spectroscopy in the near-infrared (NIR) and infrared (IR) regions. Principal component analyses (PCA) of the data sets from IR and NIR were performed so that exploratory data analyses could be conducted. Partial least-squares (PLS) regression models were built for each SARA component from IR and NIR data to predict the amounts of SARA components. These models successfully fitted the experimental data from NIR analyses and showed good predictive ability for the crude oil composition. The regression models from IR data were not modeled properly for aromatics and asphaltenes but were modeled excellently for saturate and resin components. For SARA determination, NIR spectroscopy appears to be a favorable alternative to the more time-consuming fractionation method. 1. Introduction The overall character of the feedstocks entering refineries has changed to such an extent that the difference can be measured by a decrease of several points on the API (American Petroleum Institute) gravity scale.1 This change has and will continue to make new demands with respect to both improved process technology and understanding of the crude oil system. With the necessity of processing heavy oil, there has been the recognition that knowledge of the constituents of these higher-boiling feedstocks is of great importance. Indeed, the problems encountered in processing the heavier feedstocks can be equated to the chemical character and the amount of complex, higherboiling constituents.1 Heavy oil is a complex mixture of hydrocarbons containing a small amount of heteroatoms (N, O, and S). The classification of heavy oil is in some respect indistinct but has usually been restricted to the more viscous part of conventional petroleum, having an API gravity of less than 20°. Because of its complex nature, it is not possible to determine its individual molecular constituents, and compositional studies are usually done by fractionation into predefined chemical families. The SARA group type fractionation separates the crude oil into the following classes: saturates, aromatics, resins, and asphaltenes. This fractionation method has found great utility in combination with highperformance liquid chromatography (HPLC).2-6 However, because this technique is time-consuming and requires expensive laboratory equipment, some attempts have been made recently to find alternative analytical options. Various applications of vibrational spectroscopy have been introduced with success.7-10 In this context, it is particularly noteworthy that the work done by Aske and co-workers11 has shown the ability of * To whom correspondence should be addressed. Fax: +47 73 59 40 80. E-mail: [email protected].

near-infrared (NIR) and infrared (IR) spectroscopy in combination with partial least-squares regression to predict SARA components in lighter crude oils and condensates. In this study, we have extended the range of application from light crude oils and condensates to heavy and particle-rich crude oils. This has introduced additional aspects with respect to both sample fractionation and spectroscopic analysis. We have developed a fully automated preparative HPLC procedure with a sample capacity corresponding to 0.6 g of heavy crude oil. The large sample capacity makes it possible to perform further analyses on the individual SARA fractions. 2. Experimental Section 2.1. Materials. Crude oil samples at ambient temperature and pressure were received from exploration sites on the Norwegian Continental Shelf and sites located in Brazil, France, the South China Sea, the Atlantic Ocean, and the Gulf of Mexico. From these, 20 crude oils were chosen to form a sample set of oils differing in physical properties from conventional petroleum to heavy or particle-rich oil. Some samples contained a lot of water (up to 50 wt %), which had to be removed prior to analyses. Other samples with lower water content were kept in their original state because of very stable water-in-oil systems. To reduce the effect of oxidation, 1 L of each of the crude oils was stored in UV-protective containers under nitrogen atmosphere.12 The separation of crude oils into SARA fractions is well accepted and will not be presented in detail here. Speight1 provides a thorough introduction to the chemistry and structure of crude oils. In short, saturates are defined as the saturated hydrocarbons ranging from straight-chain paraffins to cycloparaffins (naphthenes), whereas the aromatic fraction includes those hydrocarbons containing one or more aromatic nuclei that might be substituted with naphthenes or paraffins. Asphaltenes are defined as the solubility class of crude oils that

10.1021/ie0401354 CCC: $30.25 © 2005 American Chemical Society Published on Web 01/19/2005

1350

Ind. Eng. Chem. Res., Vol. 44, No. 5, 2005

Figure 2. Scheme of the flow channel selection valves. Valve A controls the flow direction through columns (normal/backflush), and valve B includes or excludes the amino precolumn.

Figure 1. Schematic representation of the HPLC system with main modules.

precipitate in the presence of aliphatic solvents (here, n-hexane), whereas the resin fraction is defined as the fraction soluble in light alkanes but insoluble in liquid propane. Asphaltenes and resins are known as large, polar, polynuclear molecules consisting of condensed aromatic rings and heteroatoms such as sulfur, nitrogen, and oxygen. Asphaltenes and resins have gained increased interest with the knowledge of the large effect of these constituents in particular on the overall performance of heavy crude oils. 2.2. Group-Type Fractionation. (a) Precipitation of Asphaltenes. The precipitation yield of asphaltenes from a crude oil in liquid hydrocarbons has been shown to be dependent on both contact time and the degree of dilution.1 Precipitation is done in three parallel trials, each of 4 g of crude oil diluted in 160 mL of n-hexane, left for mixing overnight and filtered through a 0.45µm membrane filter (Millipore HVLP). Asphaltene content is determined gravimetrically after solvent removal. (b) HPLC Method: Fractionation of Maltenes. The high-performance liquid chromatography (HPLC) system used in this study, shown schematically in Figure 1, was built from the following modules from Shimadzu Inc.: helium degassing unit (DGU-10B), subcontroller (FCV-130AL), preparative pump (LC-8A), automatic sample injector (SIL-10AP), two high-pressure flow channel selection valves (FCV-12AH), dualmode UV-vis detector (SPD-10), refractive index detector (RID-10A), fraction collector (FRC-10A), system controller (SCL-10A). The columns were purchased from Phenomenex: unbonded silica 15-µm 21.2 × 250 mm, amino 10-µm 21.2 × 50 mm. Dichloromethane (99.8%) and n-hexane (95%) were used as mobile phases. Preparative separations of organic samples favor the use of unmodified silica as the column packing material.13 However, experiments on a preparative silica column showed that dichloromethane did not have the necessary solvent strength to elute the most polar components in crude oils. Instead, an amino column, weaker in column strength, was used in the multidimensional column setup shown in Figure 2. The amino precolumn and the silica preparative column could be reloaded individually. The new setup reduced the time of analysis from 70 to 26 min. A sample of 5 mL of asphaltene-free crude oil in n-hexane (corresponding to 0.6 g of crude oil) is injected into the 20 mL/min flow (isocratic) of filtered and degassed n-hexane. Crude oil components travel through the columns and are retained mainly depending on

Figure 3. Typical chromatogram from preparative HPLC analysis. The solid line represents the RI signal, and the broken line represents the UV chromatogram at 254 nm. Peaks are cut for better visualization.

polarity. Saturates, having no retention on the columns, are collected from refractive index (RI) signals (3-5 min). After 15 min, all aromatics have left the amino precolumn, but some yellow-colored aromatics have not yet eluted from the silica column. These are collected from the preparative column by a dichloromethane backflush (