ELSEVIER Contents lists avalable at SciVerse ScionceDiroct, Fuel journal homepage: www.elsevier.com/locate/fuel [EVER Predicting SARA composition of crude oil by means of NMR Felipe Sanchez-Minero**, Jorge Ancheyta®, Guadalupe Silva-Oliver*, Sergio Flores-Valle ‘bepatament de gener Quimica Petters, SIE ttt PltcnicsNecna. Za Micon del eres, Ke Cental Ltr Clrdenas 152 Mesa iy 07720, Mesa “aborts de canis y teres, ESIQE, nse Potenice Nacional ace, Meio Cy Q7738, Mek > We characterize five crude ils by means of NMR and SARA analysis. > We obtain aromatcty factors fom NMR data > We propose an equations system to precct the SARA concentration ARTICLE INFO ABSTRACT ‘Accepted 11 October 2012 ‘valle enine 25 October 2012 obtained, with ors were used ‘romtiety Five crude ois with a wide range of API gravties were characterized by means of nuclear magnetic re= fnance (°H NMR and "°C NMR) and SARA (saturates, aromatics, resins, and asphaltenes) composition From NMR analyses, the concentrations of ct the eoresponding aromaticty factors Fy and Fe, Were caeolated, Aromatity Tac- develop corelations to predict the SARA concentration of crude oils. The developed cor lations showed good aecuraey when compating experimental and calculated values omati-hyérogen (He) and aroraticecarbon (Cy) Were 2012 Elsevier Li Allright reserved 1. Introduction Characterization of petroleum is typically cartied out by stan- dardized ASIM methods, such as true boiling point distillation, APL gravity, and elemental analysis (C, H.N, and $). The results of all these analytic techniques are reported as crude il assay, which ‘ean go just from a few analyses up to very detailed characterization ofthe oil and ofits cistillated fractions [1,2], SARA (saturates, aro- ‘matics, resins, and asphaltenes) analysis is frequently done as part ‘of assay of petroleum, which among other applications is used for predicting the compatibilty/stability of blends of crude oils to anticipate problems of sedimentation during storage and transpor- tation [3]. SARA composition has been also recently reported to be used for kinetic modeling studies in the molecular reconstruction ‘when dealing with heavy oil hydrocracking [45] Other specialized techniques, such as nuclear magnetic reso- nance (NMR), are not common for characterizing crude oils: how- ‘ever some studies have revealed that NMR can indeed be used for such a purpose no matter the API gravity ofthe petroleum sample mal ares sanchenespasne Sanchez tne) 0016-2561/5- see rat mater ©2012 Eee Ll iht teserved [6.7]. This isa great advantage since due to the heaviness of some Tow API gravity oils; various conventional methods are not suitable for analyzing this type of samples [8-13] The experimental results of structural characterization as deter ‘mined by NMR have been used to develop correlations to caleulate diferent properties of petroleum and ts fractions. IH atomic ratio and contents of paraffins, naphthenes and aromatics are some of, the properties that are reported to be calculated from *H and "C [NMR data 14-16]. As for SARA composition, Molina eta. [17] re- ported correlations to predict it from 'H NMR, but their study was limited to vacuum residva Developing correlations for predicting SARA composition of crude oils with a wide range of API gravity (10-33°) is the main objective of this work. To do that, NMR results were used to obtain the concentration of aromatic-hydrogen and aromatic-carbon, from which aromaticity factors were computed and correlated to SARA analysis. 2. Experimental Five crude oils with different API gravity were used. The crudes were labeled as A (32.8° API) B (27.1° APD, € (21.2° APD, D(13.1° APD) and E (103° APD, The crude oils were characterized by using the following methods: ‘Elemental composition (hydrogen and carbon) was obtained with a Leco TCH-600 model analyzer according to ASTM 'D5291 method, Sulfur concentration was determined following the ASTM D424, using a Tanaka RX-306SH model analyz “= SARA concentration was determined in two steps. Firstly asphaltenes were precipitated with n-heptane by means of ASTM 3279 method. The remaining maltenes fraction was separated in accordance with ASTM 2700 method, obtaining inthis way the concentration of resins, aromatics and saturates. The *H and °C NMR spectra of crude oils were determined in a Varian Oxford NMR spectrometer operating at 'H and PC reso- nance frequency of 300 and 75 MHz, respectively, accumulating 694 scans for hydrogen and 1000 scans for carbon spectra, NMR spectra were obtained in deuterated chloroform (CDCI) solu- tions. in accordance with ASTM D5292 method. Tetramethylsil- ane Was used as intemal reference in these solvents for assigning the 0.0 parts per milion (ppm) chemical shit postion in both " and ¥C NMR. Integrals of the aromatic hydrogen band were corrected by subtracting the signal of residual chlo- roform, from 7.25 ppm chemical shift, in the predominantly chloroform-d solvent For 'H spectra, samples were prepared by dissolving about 30 mg of oil in 0.8 mL of chloroform-d, to seta concentration close {0 5% viv, Solid particles were removed by filtration to obtain a homogeneous solution by common procedures. For "C spectra, samples were prepared by dissolving 1 ml of oil and 1.0 mof deu- {erated chloroform. Samples were filtered through a cotton-packed, Pasteur pipette. A fip angle of 30° was set up and tube diameter of, ‘Smm were used for samples. Spectra were recorded from ~2 14 ppm for hydrogen and from ~10 ta 250 ppm for carbon, Rotor spin was set up at 20 Hz. Delay time was set up at 15 for and 90's for Bc 53. Results and discussion 31, Elemental analysis The five crude ols used in this study show a wide range of APL sravity values (10.3-32.8°), with very different composition as can be observed in Table 1. The H/C molar ratio calculated from the ele- mental analysis indicates that crude A (the lightest one, 32.8° API) exhibits higher concentration of saturated hydrocarbons (high va~ lue of H/C ratio), while crude F (the heaviest one, 10.3° API) is con- centrated in unsaturated hydrocarbons, such as asphaltenes, In the case of sulfur, the alteady-Known typical tendency was found, that is the heavier the crude oils the higher its sulfur content. This behavior has also been reported by others [18.19] Flemeatal an SARA analyses of anaus cud as 3.2 SARA analysis ‘The results of SARA analysis are also reported in Table 1. The concentration of saturates and aromatics (mono-,di- and tr-rings) ecteases as the API gravity of the crude oil diminishes, whereas the concentration of resins and asphaltenes exhibits the opposite tendency. Thus, the higher concentration of saturates and aromat- des in the lightest crude oll is related toa great content of naphtha, whereas the higher concentration of resins and asphaltenes in the heaviest crude oil is related (o a great content of residue. 3.3. Nuclear magnetic resonance analysis NMR spectra of crude oils were analyzed to obtain the concen- tration of aromatic- and aliphatic-hydrogen, as well as aromatic and aliphatic-carbon. For 'H NMR spectrum, the total aromatic- hydrogen was obtained from the integral in the region from 6 to ‘ppm, whereas for the total aliphatic-hydrogen it was obtained in the region of 05-5 ppm. For "°C NMR spectrum, the total aro- ‘matic-cazbon and the (otal aliphatic-carbon were obtained in the region of 110-160 ppm and 10-60 ppm, respectively. Both spectra are shown in Fig. 1, where the different regions are illustrated. From NMR analysis *H NMR and °C NMR), the aromaticity factors of yerogen (Fa) andl carbon (Fes) for each crude oil were obtained with the following equations [20.21 : i ll , | | ental ana Hic molseracio tery ass ara Sah analysis Satrser (ots) S162. 21S 3230538200 Asphatenes wt abo 80 Wo” wo G0 1008060 4d DO pm Fig 1 itand CNM specs af deo0 Sanches Mineo ea Ft 10 (2012) 218-221 Table ‘romatcy face fcr as Gneeol * c o p Aromavy factor O) @ where He is the total aromatic hydrogen, Hy the total aliphatic hydrogen. Cy the total aromatic carbon, and Cy the total aliphatic carbon ‘The values of Fux and Fea calculated with Eqs. (1) and (2) are presented in Table 2, Both factors decrease with the increased [APL gravity of the crude oil, This trend is similar to that observed for the concentration of resins and asphaltenes of the crude oll ‘which indicates that these fractions strongly affect the aromaticity| actos. 43.4, Development of correlations SARA concentration of the crude oils was correlated with the aromaticity factors, Fig. 2 presents the concentration of aromatics, resins and asphaltenes (ARA) for each crude ol as a function of Fun and Fe, factors. The fraction of saturates was not plotted since it does not contain aromatic structures, Aromaticity factors have a direct dependency with resins and asphaltenes contents while ‘with aromatics content it is inverse, This means that the types of ‘molecules that contribute to increase the aromaticity of a crude oil are not those with a few aromatic rings but those with highly concentrated aromatic structures such as asphaltenes ‘Then, the results of Fig. 2 were used to develop the correlations to predict SARA composition as function of aromatieity factors. The tendency curve observed for each ARA component was selected to propose the correlations, based on their higher values of? (>0.98}, The best set of parameters of the proposed correlations was ob- tained by using the Powell minimization method with the Scientist of MicroMath software. The following correlations were derived |ARA concentration as function of Fy 40903 Fly +310 Fa +329 8) % Aromatics Fh Resins — 4146 Fig 49 @ % Asphaltenes = 561.6 Fax ~ 105 6) AARA concentration as function of Fes % Aromatics = ~748.5 Fi, + 198.3 Fux + 168 6 Fh Resins = 113.9 Fe +238 a % Asphaltenes = 301.2 Fi, +32.0 Fy —28 @ “The concentration of saturates can be determined by difference: 5 Resins ~ % Asphaltenes @ 5 Saturates = 100 — % Aromatics Fig. 3 presents the comparison between calculated and expesi- ‘mental data of SARA concentration for each crude oll. The results cleanly illustrate that the developed correlations do not tend to either over estimate or under estimate the SARA composition. In & fe : i i. i 7 : i FHa é z g E 68 3 3 Fg. 2 AKA concentration a¢ fonction af Fue (aromatics (2), phates (0) she Fa aromas (reso) and arphalene general, haf ofthe points are on the right-hand side of the 45" line and the other half on the left-hand side. ‘On the other hand, Table 3 shows the K? values for each corre- lation. The correlation to predict the concentration of resins pre- sents the highest value of R2, whereas the correlation to predict the concentration of asphaltenes exhibits the lowest value. in addi- tion, the correlations as function of the aromaticity factors show that Fag factor reaches higher precision than Fog factor, To verity this, the mean absolute percentage error (MAPE) was calculated using the following equation: are 1005 Ya ao) Where mis the total number of experimental data, Yo isthe exper- imental value and Yeu isthe calculated value. The correlations that use Fyq factor present a MAPE value of 5.2%, whereas the correlations that use Fe factor reaches a MAPE value of 11.5% This is mainly due to the better prediction of asphaltenes concentration from Fyn compared with Fes, Finally the ranges obtained from a residual analysis were -1.48;+1.78 and ~3.56/+3.42 for Fg and Fes, respectively. This also confirms the better prediction with Fi It is worth to mention that the developed correlations were generated using five crude oils of a particular tegion with APL