New Insights into Resid Desulfurization ... - ACS Publications

Fixed-bed resid desulfurization (RDS) is an important industrial catalytic process to ... lighter fractions, sulfur content of the residual heavier fr...
0 downloads 0 Views 1MB Size
Article pubs.acs.org/EF

New Insights into Resid Desulfurization Processes: Molecular Size Dependence of Catalytic Performances Quantified by Size Exclusion Chromatography-ICP/MS Jérémie Barbier, Charles-Philippe Lienemann, Agnès Le Masle, Pascal Chatron-Michaud, Bertrand Guichard, and Mathieu Digne* IFP Energies nouvelles, Rond-point de l’échangeur de Solaize, BP 3, 69360 Solaize, France S Supporting Information *

ABSTRACT: Fixed-bed resid desulfurization (RDS) is an important industrial catalytic process to produce low sulfur fuel oil and to pretreat feed for resid fluid catalytic cracking (RFCC) unit. The catalytic system is based on a complex stacking of taskspecific catalysts. To improve the RDS performances, it is crucial to better understand and monitor the nature and amount of contaminants (especially sulfur and metals) along the process. In the present paper, an hyphenated technique based on Size Exclusion Chromatography separation (SEC) and Inductively-Coupled Plasma/Mass Spectrometry (ICP/MS) detection is applied to develop an innovative quantitative method: it allows to quantify the refractory compounds related to their molecular size. The analysis of the effluents arising from the different catalyst beds demonstrates the selectivity of each catalyst toward the molecular size of sulfur and vanadium containing molecules.



INTRODUCTION The upgrading of heavy oils and residues is a major industrial challenge for the next decades.1 Indeed, the market for resid fuel oil tends to decrease, whereas the demand for low contaminants transportation fuels is increasing. Among the resid fuel oils, only the demand for bunker fuel is forecasted to grow due the development of marine transportation. Bunker fuels are also planned to be submitted to more stringent sulfur specifications: the International Marine Organization (MARPOL Annex VI) asked a global sulfur content of 0.5 wt % in 2020 or 2025.2 An even more stringent specification of 0.1 wt % S is also mandatory for the so-called Sulfur Emission Control Areas (SECA) in 2015. Taking into these strong market and legislation tendencies, refiners have to adapt their resid conversion capacities. Several process technologies and processes combinations are possible to upgrade resid,3 either noncatalytic (such as solvent deasphalting or coking) or catalytic processes (resid fluid catalytic cracking RFCC or hydrotreatments HDT). Hydrotreatments can be divided into fixed bed, ebullated bed,4 or slurry systems. The optimal choice depends on several parameters: nature and contaminants content of the feedstock, required level of conversion into lighter fractions, sulfur content of the residual heavier fraction, investment costs. Among these available processes, the fixed-bed resid desulfurization (RDS) units are mainly used for two different tasks: the first one is the production of low sulfur fuel oil (typically lower than 0.5 wt % S). Fuel oil is for instance used as feedstock for electricity generation, but this use is declining, because of high environmental emissions, compared to natural gas power plants. Nevertheless, the industrial interest for the production of ever cleaner fuel oil could emerge again in the next years for the 0.5 wt % and 0.1 wt % S marine fuel production. The second task, which is currently the © 2013 American Chemical Society

predominant one, is the use of RDS unit as a pretreater for RFCC unit.5 Compared to the previous application, RDS has to maximize not only the hydrodesulfuration (HDS) activity but also the hydrodemetalation (HDM) activity, hydrodenitrogenation (HDN) activity, and Conradson carbon residue (CCR) removal to prepare suitable feed for RFCC. The RDS/RFCC can lead to more than 90% yields of feed conversion, mainly oriented into light fractions, LPG and gasoline. In the recent period, technological improvements allows RDS to treat heavier crude oil with higher metal content, by adding advanced guard reactors upstream the standard fixed bed reactors: the Onstream Catalyst Replacement (OCR) and the Up-flow Reactors (UFR) developed by Chevron Lummus Global,5,6 the PERNIS process (bunker flow-moving bed technology to replace the catalysts continuously online) by Shell,7 and the permutable reactor system (PRS) developed by Axens.8 Compared to other hydrotreatment processes, RDS exhibits several specificities: (i) severe operating conditions with temperature between 360 and 410 °C, total pressure between 100 and 160 bar and a liquid hourly space velocity LHSV between 0.15 and 0.4 h−1, (ii) multiple reactors with complex catalyst sequences encompassing a minimum of two different catalysts, but often at least three different catalysts, (iii) strong deactivation leading to short cycle length. For instance, standard cycle lengths for resid HDT is about 12 months, compared to 12−24 months for VGO HDT and 24−48 months for Diesel HDT.9 This is the reason why, on top of the intrinsic catalytic activities, the deactivation rate (usually expressed as the temperature increase per month needed to Received: August 2, 2013 Revised: October 10, 2013 Published: October 21, 2013 6567

dx.doi.org/10.1021/ef401540f | Energy Fuels 2013, 27, 6567−6574

Energy & Fuels

Article

maintain of global unit performances) is a crucial parameter to evaluate and select a catalysts system. Residues of distillations are the heaviest and the most complex petroleum fraction. They contain high levels of heteroatoms and metals which are poisons for refinery catalysts and contaminants for fuels. The most abundant and undesirable contaminants are sulfur, nitrogen, vanadium, and nickel. These contaminant are concentrated in asphaltenes which are the most complex and polar compounds of petroleum. Asphaltenes tend to self-associate in residues10,11 and are detrimental to residue HDT processes.12−14 Detailed chemical characterization of residues is awkward, and the choice of adapted analytical techniques is quite limited due to the very high complexity of the mixture.15 Nevertheless, the detailed chemical characterization and the quantification of residue compounds are needed in order to further understand their reactivity in the HDT process. Indeed, the knowledge of the compounds in feedstock and the effluent is crucial to improve the catalyst performances and to limit the catalyst deactivations. The goal of this work was to develop a new analytical approach to investigate the performance of a RDS catalysts system. Following HDS and HDM performance requires analytical techniques able to quantify trace concentrations of Ni, V, and S elements. Such techniques were deeply reviewed in 2009 by Caumette et al.16 Chromatographic separation with element specific detection, usually by inductively coupled plasma mass spectrometry (ICP MS), is an established analytical tool for element speciation in petroleum products. High-resolution (Fourier-Transform) MS is the ultimate instrumental technique for the identification of the metalcomplexes present. Nevertheless, the selectivity of the ionization step16 and the compositional and structural continuum of petroleum17 make it currently difficult to quantify contaminants. Based on these considerations, a hyphenated technique based on Size Exclusion Chromatography separation (SEC) and ICP/MS detection has been developed in a previous study.18 This analytical technique gives information on the molecular size distribution of the compounds containing contaminants (sulfur, nickel, and vanadium) from only one analysis. In this work, a new quantitative method has been developed in order to quantify contaminants in a range of molecular sizes. The quantitative method was applied to investigate a RDS catalysts system. The molecular size of contaminant-containing species converted or not during the RDS process was studied. Therefore, refractory compounds was quantified related to their molecular size. This new analytical approach leads up to the selectivity of catalysts in term of molecular size of contaminant-containing compounds. This information is very helpful and assists in improving catalyst performances and catalysts systems.



Table 1. Bulk Properties of the AH AR/AL VR Feed properties

unit

density @ 15 °C sulfur content Ni V total nitrogen content Conradson carbon residue (CCR) asphaltenes (C7)

g/mL wt % wt ppm wt ppm wt ppm wt % wt %

0.9861 3.97 21.6 67.9 2955 13.08 4.3

Table 2. Commercial Catalysts Properties catalyst

1

type porosity bulk density surface area mesopore diameter Mo Ni + Co metal capacity

g/mL m2/g nm wt % wt %

2

3

HDM mesoporousmacroporous 0.40−0.55 100−180 12−20

HDM/HDS mesoporous

HDS mesoporous

0.55−0.65 120−200 12−20

0.65−0.80 160−240 9−14

3−5 0.2−0.6 high

5−7 0.5−1.5 median

9−11 1.5−3.5 low

balances between HDM and HDS functions. The first catalyst is dedicated to HDM reactions and exhibits quite low HDS performances: catalyst 1. A second one has been incorporated after catalyst 1: catalyst 2. This catalyst displays balanced HDM and HDS functions and can be classified as a transitory catalyst. Finally, the third catalyst is dedicated to HDS reactions on demetallized effluent and achieves low HDM rates: catalyst 3. Hydrotreating Experiments and Sampling. Experimental Setup and Catalysts Loadings. A simplified scheme of the pilot plant unit is given in the Supporting Information (Figure S1). The feed is pumped and mixed with pure hydrogen and then flows through the feed heater of the reactors. The unit is equipped with two fixed-bed reactors, equipped with a sampling system between the two reactors. After the reactors, a high-pressure separator flashes the effluent into a hydrogen rich gas stream and a liquid phase stream. After the depressurization valve, a low-pressure separator allows a proper separation between the liquid stream and the sour gas. Both high and low pressure gas streams are analyzed by in situ-GC and quantified before being driven out of the unit’s limits. The liquid effluent can also be stripped with nitrogen in order to obtain a stabilized liquid effluent which is free of H2S, NH3, and light gases. Taking into account the two fixed-bed reactors of the pilot plan unit, two tests were carried out in order to investigate the reactivity and selectivity of the three catalysts: the first test (test 1) with the HDM catalyst (catalyst 1) in the first reactor and the intermediate catalyst (catalyst 2) in the second reactor. Each temperature of the reactor is increased to compensate the catalysts deactivation and to maintain the overall unit performances. This experiment is to give an overview on the deactivation of the HDM (catalyst 1) and intermediate (catalyst 2) catalysts. The second test (test 2) is composed of the HDM and the intermediate catalysts (catalysts 1 and 2) in the first reactor and the HDS catalyst 3 in the second reactor, to give an overview on the stability of the HDS catalyst. The time on stream of the collected sample is kept constant (2000 h) to avoid any interaction of the age of the catalysts with the interpretation of the catalysts performances. Operating Conditions. The pressure is decreased at the run pressure at 25 bar/h, and then the temperature is increased to reach the operating conditions under a SR Gas-oil flow with 2 wt % of DMDS (dimethyl disulfide) as additive. After sulfiding, the first feed AL AR (Arabian Light Atmospheric Residue) is processed for 300 h, and then the second feed (blending of Arabian Heavy Atmospheric

EXPERIMENTAL SECTION

Several samples were collected to measure the performances of the RDS process. In the Experimental Section, we describe first the materials (feedstocks and catalysts) used for the catalytic tests. Next, the pilot plant tests and the sampling are detailed. Finally the samples pretreatment and the SEC-ICP MS analysis are given. Materials. Feedstock. For the hydrotreatment tests, after sulfidation and nursing steps, a blend of Arabian Heavy Atmospheric Residue and Arabian Light Vacuum Resid 70/30 wt/wt was used as feed (Table 1 and Table S1 of the Supporting Information for more details). Catalysts. Three commercial catalysts were chosen for this study, and their properties are summarized in Table 2. They display different 6568

dx.doi.org/10.1021/ef401540f | Energy Fuels 2013, 27, 6567−6574

Energy & Fuels

Article

Residue and Arabian Light Vacuum Resid 70/30 wt/wt) is processed for about 3000 h to study the stability. Each reactor works in an isothermal mode: the start-of-run temperature is 370 °C for all the reactors. However the reactor temperatures have been increased to keep the same impurities levels in the hydrotreated effluents at the outlet of each catalyst bed. In test 1, the deactivations of catalysts 1 and 2 have been studied separately. To evaluate their stability, each catalyst was loaded in a reactor. The temperature of the HDM (1) and intermediate (2) catalysts have been increased to reach a metal content of around 35 wt ppm in the effluent from catalyst 1 and a sulfur content of around 1.2 wt % in the effluent from catalyst 2. At the same time, the outlet sulfur contents were checked and respectively kept at around 2.2 wt % sulfur (catalyst 1) and 20 wt ppm metals (catalyst 2). Then, in test 2, the deactivation of catalyst 3 (HDS) has been studied. To perform this evaluation, the catalyst 3 was loaded in the second reactor, whereas both catalysts 1 and 2 were loaded in the first reactor. As a consequence, the temperature of the catalyst 1 could not be increased to keep the same target as in test 1, but the temperatures of the catalysts 1 and 2 have been increased to keep the same outlet as in test 1. Nevertheless, the temperature changes for both catalysts were almost simultaneous according to test 1 (see the Results section). In this test 2, the temperatures of the HDM (1) and intermediate (2) catalysts on one part, and HDS catalysts (3) on another part, have been increased to reach a sulfur content in the effluents respectively around 1.2 and 0.25 wt %. Meanwhile, the metal content at the outlets were maintained closed to respectively 20 and 10 wt ppm. X-ray Fluorescence (XRF) spectroscopy was used to monitor the S, Ni, and V contents and was carried out on a PANALYTICAL AXIOS dispersive wavelengths spectrometer equipped with a Cr tube. The elements have been recorded at the respective angles: S at 110.68°, Ni at 48.62°, and V at 76.9° with a flow counter detector. Density measurements were performed on a ANTON PAAR DMA4100 density meter. The density is measured at a temperature of 70 °C at atmospheric pressure. Sampling. Samples were collected from the two hydrotreating tests 1 and 2 loaded with the catalysts 1, 2, and 3. Figure 1 specifies the location of the sample takes along the process flow:

Element XR sector-field-ICP/MS instrument (Thermo Fisher, Bremen, Germany). The SEC was provided with three columns connected in series with increasing porosity (100 Å, 1000 Å, and 10000 Å) and with polystyrene-divinylbenzene stationary phase. The columns were calibrated with 16 polystyrene standards from 380 Da up to 2630000 Da using refractometric detection. The mobile phase was THF at 0.7 mL·min−1. Samples were diluted by 160-fold in THF 24 h before injection of 20 μL. SEC separation was performed at 30 °C. The column eluted fraction was split to introduce 30 μL·min−1 into the ICP/MS. ICP parameters were already given elsewhere.20 The total exclusion time was measured at 27 min which corresponds to a molecular weight of 40 kDa according to polystyrene standard calibration (PS equivalent), and the complete permeation time was located at 42 min (100 Da PS equivalent).



DEVELOPMENT OF THE SEC-ICP/MS QUANTITATIVE METHOD Samples Characterization. The elementary content of samples is summarized in Table 3. These results allow estimating the distribution of elements during the deasphalting.

Table 3. Elemental Content of Sulfur (S), Nickel (Ni), and Vanadium(V) for Studied Samplesa origin feed

effluent of catalyst 1

effluent of catalyst 2

effluent of catalyst 3

a

sample

S (wt%)

Ni (ppm)

V (ppm)

F F-M F-A E1 E1-M E1-A E2 E2-M E2-A E3 E3-M E3-A

4.0 3.7 7.4 2.1 1.9 6.8 1.2 1.0 6.1 0.3 0.2 5.2

± ± ± ± ± ± ± ± ± ± ± ±

22 ± 1 13 ± 1 148 ± 8 13 ± 1 7±1 178 ± 9 7±1 3±1 173 ± 9 5±1 1±1 163 ± 8

68 ± 7 34 ± 4 552 ± 55 26 ± 3 9±1 463 ± 46 14 ± 1 4±1 426 ± 43 9±1 1±1 403 ± 40

0.1 0.2 0.2 0.1 0.1 0.2 0.1 0.1 0.2 0.1 0.1 0.2

M stands for maltenes and A for asphaltenes.

Figure 2 shows the normalized SEC-ICP/MS chromatograms of F and its maltenes (F-M) and asphaltenes (F-A) fractions. Continuous distributions of S, Ni, and V are detected from 40 kDa PS equivalent, and the signal reaches the baseline up to the complete permeation time. A previous study to validate SEC-ICP/MS procedure has shown that repeatability of the elution time of a porphyrin (used as a model molecule) was reproductible within 0.1%, and for petroleum product, the variations of the profile area on 6 injections were 0.5% for vanadium and 5% for nickel.19 Variations of the profiles for Ni and V over 5 days were 8 and 14% respectively. Due to this fairly high reproducibility caused by ICP signal variation, intensity corrections for Ni, V, and S signal were applied to reduce variation at about 5%. These intensity corrections were based on a control check sample, analyzed every day before samples analysis. In this study the check sample was analyzed 5 times. After correction, the total area of the check sample profile varied of 6%, 6%, and 4% for S, Ni, and V detection, respectively . These variations were used to estimate the uncertainties of the sample total area profiles. Therefore, the uncertainties were estimated respectively to about 12%, 12%, and 8% for S, Ni, and V detection. The previous work19 has shown that compounds which elute before the total permeation time are essentially separated according to a mechanic exclusion based on the hydrodynamic

Figure 1. Scheme of sample preparation. • Feed (sample F); • Effluents taken in between each catalytic beds (E): Samples E1, E2, and E3 correspond respectively to the samples taken after catalyst 1, 2, and 3. E1 and E2 were produced during test 1 and E3 was produced during test 2. The three effluents have the same time on stream (2000 h). For each sample, the asphaltene fraction was separated from the maltene fraction by precipitation of the sample into heptane according to the method NF T60-115. Maltenes and asphaltenes fractions of sample Ex were respectively noted EX-M and EX-A. The yields of the asphaltenes precipitation for each effluent are given in Figure 1. SEC-ICP/MS Analysis. The coupling of size exclusion chromatography with inductively coupled plasma mass spectrometry (SEC-ICP/ MS) was used to investigate the molecular distribution of sulfur, vanadium, and nickel. The SEC-ICP/MS method was developed and validated in a previous study.19 Chromatographic separations were achieved using an Ultimate 3000 binary HPLC pump (LC Packings, Amsterdam, The Netherlands) connected to a Thermo Scientific 6569

dx.doi.org/10.1021/ef401540f | Energy Fuels 2013, 27, 6567−6574

Energy & Fuels

Article

the column according to retention mechanism in addition to the mechanic exclusion. Therefore, no information about the size of compounds which elute after the complete permeation time can be obtained. Differences between the SEC-ICP/MS chromatogram of maltenes and asphaltenes fractions are observed for the three elements. Indeed, the asphaltenes and the maltenes fraction correspond essentially to the highest and the smallest fraction of the feedstock respectively. Therefore, the fractionation to asphaltenes and maltenes has produced samples enriched in compounds which have different size. The signal of the nickel chromatogram is highly noisily. This is consistent with the low Ni content in sample. Quantification by SEC-ICP/MS. In order to validate the quantification procedure, twelve different samples were used with different hydrotreating conditions (F, E1, E2, and E3) and various size distributions (maltenes and asphaltenes fractions). For each sample, the signal obtained with the SEC-ICP/MS for Ni, V, and S was integrated over the complete mass range. This integrated signal (total area) was compared to the elemental composition (S, Ni, and V) of the initial samples injected. Linear regressions were developed between the total areas obtained by integrating SEC-ICP/MS chromatograms and elemental quantities of Ni, V, and S of the initial samples according to eq 1 (1)

mX = a· A + b

where mX stands for S, V, and Ni mass content; A stands for the integration area chromatogram; and a and b are coefficient of the linear correlation. Figure 3 shows the correlations between the elementary content and the chromatogram area for sulfur (a), nickel (b), and vanadium (c). The confidence intervals were estimated from the reproducibility of the check sample analysis. Linear correlations well match the experimental data. In the case of sulfur, a global linear tend can be observed, but two different tendencies seem to appear when maltenes and asphaltenes fractions are compared. This may due to different response factors for the constituents of these fractions. Nevertheless, for the present study’s purpose, a linear correlation is satisfactory. Concerning nickel, its low content could explain the not well linear relationship between element content and chromatogram area. In the case of vanadium, the total areas are very well linearly correlated with the element content. However, according to the global linear trends, the entire set of samples seems to have the same response factor. Since we have shown that maltenes and asphaltenes have different molecular size distribution, we can assume that, in studied samples, the response factors are independent of the molecular size of analytes. With this hypothesis, the errors on the elementary quantifications by SEC-ICP/MS are estimated to about 14%, 12%, and 9% respectively for S, Ni, and V content. The performance of the hydrotreatment applied for X (X = S, V, Ni) is defined by eq 2

Figure 2. SEC-ICP/MS chromatograms of the feedstock (blue line), the maltenes fraction (red line), and the asphaltenes fraction (green line) for (a) sulfur, (b) nickel, and (c) vanadium.

volume of analytes, the heaviest analytes eluting before the smallest analytes. Further validation of this separation was obtained by the recovery and reinjection of SEC fractions. Each fraction were collected, evaporated to dryness, redissolved in THF, and reanalyzed by SEC-ICP/MS. The reinjection of the low molecular weight fraction (5 kDa PS equivalent) showed that the dominant part (71%) of the metal was present in the HMw elution area, indicating the existence of stable HMw metal complexes, as much as 29% appeared in the 100−5000 Da region. For the compounds eluted after the complete permeation time, the previous work19 has shown that they are retained on

HDX =

mX feed − mXeffluent × 100 mX feed

(2)

where mX stands for S, V, and Ni mass content Figure 4 compares the HDX calculated from element content with the estimation from the SEC-ICP/MS quantification. The confidence interval of the HDX (X = S, V, Ni) values was estimated at the most to 10% based on our experience of this pilot plan performance. Results show that SEC-ICP/MS 6570

dx.doi.org/10.1021/ef401540f | Energy Fuels 2013, 27, 6567−6574

Energy & Fuels

Article

response factor dependent on the chemical form. However, HDS estimations are on the whole satisfactory; therefore, we can assume that for studied sample, S and V can be quantified from SEC-ICP/MS chromatograms. Concerning HDNi, predictions are significantly different than experimental values, arising from the important noise of chromatograms due to the very low content of this element. Beside, since the response factor is considered independentlike of the analyte molecular size, we can assume that SECICP/MS allows for quantifying S and V in a molecular size interval from a chromatogram with an accuracy equal to respectively 14% and 9%.



APPLICATION OF THE QUANTIFICATION TO RDS PROCESS EFFLUENTS Reactivity of S and V According to the Molecular Size. In order to study reactivity of RDS according to the molecular size of the element treated, chromatograms obtained with SECICP/MS were segmented in a different zone. The limits of the intervals chosen are common for S and V and for all samples. HDXi is defined as the fraction of the total HDX contained between the limits of interval i. HDXi is then calculated according to eq 3 HDX i =

mX feed , i − mXeffluent , i × 100 mXeffluent

(3)

where mX,i stands for S and V mass content in the interval i. Taking into account the chromatographic profiles and as illustrated in Figure 5, three intervals have been defined: from

Figure 3. Calibration curves obtained for (a) sulfur, (b) nickel, and (c) vanadium content vs chromatography area.

Figure 4. Parity diagram comparing the global HDX (X = S, Ni, and V) obtained by typical elementary analysis method and those obtained by SEC-ICP MS analysis.

estimations are very consistent with the experimental HDV values. Concerning HDS estimations, only two samples are outside the confidence interval of 10%. They correspond both to maltenes fractions. This is consistent with the sulfur

Figure 5. Evolution of (a) sulfur and (b) vanadium distributions during the catalytic process. 6571

dx.doi.org/10.1021/ef401540f | Energy Fuels 2013, 27, 6567−6574

Energy & Fuels

Article

40 kDa PS equivalent to 7 kDa PS equivalent for the highest molecular weight compounds (HMw), from 7 kDa PS equivalent to 0.7 kDa PS equivalent for the intermediate molecular weight compounds (IMw), from 0.7 kDa PS equivalent to 0.1 kDa PS equivalent for the smallest molecular weight compounds (LMw). The last interval (Polar) is composed of the molecules that are eluted after the total permeation time and are composed of the most polar compounds retained on the divinylbenzene copolymer of the column. The SEC profiles for S are composed of two main families of compounds. A specific and intense peak around 5 kDa PS equivalent (IMw peak) is present with a second and less intense maximum at 20 kDa PS equivalent (HMw peak). The evolution of the S distribution can be monitored according to the interval i defined previously (HMw, IMw, LMw, Polar) in order to follow which family is degraded during HDSi. Typically, a drastic reduction of the IMw peak is observed between the feed F and the final effluent E3. The HMw peak is also reduced, but its final intensity in effluent E3 is similar to the IMw peak, showing less efficiency of the hydrodesulfuration process for this heavy molecules. Two different phenomena, with the hydrotreatment of the element within the oil and conversion of largest compounds to smaller size, are responsible for such changes in the distribution profiles of the element. The conversion of largest compounds to smaller size compounds will shift the distribution to smaller size, HMw will produce IMw and LMw compounds, and IMw will produce LMw compounds. However, the SEC-ICP/MS profiles show that the molecules in the IMw and LMw intervals are more efficiently treated during HDSi than HMw molecules. This clearly evidence differences in reactivity depending on the molecular size of the S molecules during the catalytic process. Based on the correlation developed between elemental content and the area of the SEC-ICP/MS profile for S, individual concentration for each interval (HMw, IMw, LMw, Polar) can be predicted. Figure 6a presents the quantification of S in the different size interval during the catalytic process. The study was performed not only on the total sample but also on the asphaltenes and maltenes fraction obtained from the feed and each effluent. The example shown here are based on the total sample in order to have a global view of the effluent treatment. Figure 6a illustrated clearly the efficiency of the hydrotreatment for each class of S compounds (HMw, IMw, LMw, Polar) after each catalytic section. Table 4 gives the hydrodesulfuration performance for each class of S compounds as defined by eq 3. Here again, the higher efficiencies are obtained for LMW and IMW compounds, with increasing hydrodesulfuration efficiency from catalyst E1 to E3 for all classes of compounds. Figure 6b shows the evolution of V distributions during the catalytic process and the performance of hydrodemetalation (HDXi) for each interval i (HMw, IMw, LMw, Polar) calculated with eq 3 is given in Table 5b. It is noted that the V chromatographic profiles is different than the S profiles, showing a greater distribution of V in the high molecular weight. In addition, a third peak with a maximum located LMw is present for V. The intensities of the different peaks (HMw, IMw, and LMw) have similar intensities, showing a homogeneous distribution of the V molecules all along the molecular size distribution. As already observed for S compounds, the IMw molecules containing V are more efficiently treated during the

Figure 6. Quantification of (a) sulfur and (b) vanadium in the different size interval during the catalytic process.

demetalation process. The demetalation of the LMw molecules is comparable to HMw for V, and a drastic decrease of the total amount of V is observed after the very first catalyst (E1). The developed model helps getting the elemental concentration for each size fraction, and this information is crucial to optimize the choice of the catalyst and the conditions used. Being able to identify which class of molecules is refractory to the process is very helpful to test and develop new classes of efficient solids. Our results show for the studied catalysts system that contaminant-containing compounds are eliminated with efficiency independently of the molecular size along the RDS process. Few content of contaminants were in the final effluent, and we have shown that they are properly distributed in all the molecular size range. Therefore, our results have shown that the studied catalyst system is very effective to the RDS process. Our new analytical approach will be used to optimize the selectivity of catalysts and their association according to the specificity of the feedstock.



CONCLUSION The hydotreatment of atmospheric and vacuum residues in a fixed bed RDS unit is an important process to upgrade the heavy oils fractions. A quantification method for S, Ni, and V based on molecular mass interval was developed using SEC-ICP/MS. A set of samples containing feed originating from the Middle East and effluent treated with fix-bed conditions, associated with asphaltenes and maltenes fractions, was used to develop a linear predictive model. This model helps to obtain the elemental concentration for each size fraction based on the integrated area of SEC-ICP/MS profiles. The response coefficients for each fraction are independent of the molecular 6572

dx.doi.org/10.1021/ef401540f | Energy Fuels 2013, 27, 6567−6574

Energy & Fuels

Article

Table 4. Hydrodesulfuration Performance (HDS) in SEC Chromatogram Interval during the Catalytic Process SEC chromatogram interval

HDS in E1 (wt%)

HDS in E2 (wt%)

HDS in E3 (wt%)

HMw IMw LMw Polar

6±1 32 ± 9 12 ± 3 2±1

7±1 46 ± 10 17 ± 4 3±1

8±1 60 ± 10 21 ± 4 5±1

Table 5. Hydrodemetalation Performance of Vanadium (HDV) in SEC Chromatogram Interval during the Catalytic Process SEC chromatogram interval

HDV in E1 (wt%)

HDV in E2 (wt%)

HDV in E3 (wt%)

HMw IMw LMw Polar

14 ± 2 34 ± 5 15 ± 2 4±1

17 ± 2 41 ± 5 17 ± 2 5±1

20 ± 2 44 ± 5 19 ± 2 9±1

mass for the set of samples studied. The developed method allows quantifying the performance of the process based on the molecular mass obtained from the SEC-ICP/MS profiles. This analytical method allowed the determination and the quantification of the refractory fractions of V and S and has the potential to bring crucial information to the molecular mass that are not reacting with the applied conditions. This information is not available with conventional approaches based on elemental analysis of the total amount of V and S. Results show that the studied catalyst system is effective in eliminating contaminants in residue independently of their molecular size. Therefore, this new analytical approach is an asset to improve catalyst performances and catalysts systems.





REFERENCES

(1) Huc, A. Y. Heavy crude oils, from geology to upgrading, an overview; Edition TECHNIP: Paris, 2010. (2) International maritime organization, International Convention for the Prevention of Pollution from Ships (MARPOL), Annex VI Prevention of Air Pollution from Ships, 2005. (3) Rana, M. S.; Samano, V.; Ancheyta, J.; Diaz, J. A. I. Fuel 2007, 86, 1216−1231. (4) Martinez, J.; Sanchez, J. L.; Ancheyta, J.; Ruiz, R. S. Catal. Rev.: Sci. Eng. 2010, 52, 60−105. (5) Threlkel, R.; Dillon, C.; Singh, U. G.; Ziebarth, M. J. Jpn. Pet. Inst. 2010, 53, 65−74. (6) Scheuerman, G. L.; Johnson, D. R.; Reynolds, B. E.; Bachtel, R. W.; Threlkel, R. S. Fuel Process. Technol. 1993, 35, 39−54. (7) Scheffer, B.; van Koten, M. A.; Röbschläger, K. W.; de Boks, F. C. Catal. Today 1998, 43, 217−224. (8) Kressmann, S.; Morel, F.; Harlé, V.; Kasztelan, S. Catal. Today 1998, 43, 203−215. (9) Vogelaar, B. M.; Eijsbouts, S.; Bergwerff, J. A.; Heiszwolf, J. J. Catal. Today 2010, 154, 256−263. (10) Eyssautier, J.; Espinat, D.; Gummel, J.; Levitz, P.; Beccera, M.; Shaw, J.; Barré, L. Energy Fuels 2012, 26, 2680−2687. (11) Mullins, O. C.; Sabbah, H.; Eyssautier, J.; Pomerantz, A. E.; Barré, L.; Andrews, A. B.; Ruiz-Morales, Y.; Mostowfi, F.; McFarlane, R.; Goual, L.; Lepkowicz, R.; Cooper, T.; Orbulescu, J.; Leblanc, R. M.; Edwards, J.; Zare, R. N. Energy Fuels 2012, 26, 3986−4003. (12) Le Lanic, K.; Guibard, I.; Merdrignac, I. Pet. Sci. Technol. 2007, 25, 169−186. (13) Ferreira, C.; Tayakout Fayolle, M.; Guibard, I.; Lemos, F.; Toulhoat, H.; Ramoa, R. Fuel 2012, 98, 218−228. (14) Gautier, T.; Danial-Fortain, P.; Merdrignac, I.; Guibard, I.; Quoineaud, A. A. Catal. Today 2008, 130, 429−438. (15) Merdrignac, I.; Espinat, D. Oil Gas Sci. Technol. 2007, 62, 7−37. (16) Caumette, G.; Lienemann, C.-P.; Merdrignac, I.; Bouyssiere, B.; Lobinski, R. J. Anal. At. Spectrom. 2009, 24, 263−76. (17) Panda, S. K.; Brockman, K. J.; Berter, T.; Schrader, W. Rapid Commun. Mass Spectrom. 2011, 25, 2317−2326. (18) Podogorsky, D. C.; Corilo, Y. E.; Nyadong, L.; Lobodin, V. V.; Robbins, W. K.; McKenna, A. M.; Marshall, A. G.; Rodgers, R. P. Energy Fuels 2013, 27, 1268−1276.

ASSOCIATED CONTENT

S Supporting Information *

Table S1 and Figure S1. This material is available free of charge via the Internet at http://pubs.acs.org.



LMw = low molecular weigh interval of the size exclusion chromatogram M = maltenes fraction Ni = nickel PS = polystyrene RDS = resid desulfuration S = sulfur SEC = size exclusion chromatography V = vanadium XRF = X-ray fluorescence spectroscopy

AUTHOR INFORMATION

Corresponding Author

*Fax: (+33) 437702745. E-mail: [email protected]. Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS The authors thank Nathalie Schildknecht for her longstanding support. They also thank Brice Bouyssière and Carine Arnaudguilhem (LCABIE-IPREM, Pau, France) for ICP-HR/ MS tuning and maintenance, which have helped to achieve this work, and Frédéric Poli for carrying out the catalytic tests.



NOMENCLATURE A = asphaltenes fraction E = effluent of the pilot plant test F = feedstock of the pilot plant test HDM = hydrodemetalation performance HDN = hydrodenitrogenation performance HDS = hydrodesulfuration performance HDT = hydrotreatment HMw = heavy molecular weigh interval of the size exclusion chromatogram i = interval of the size exclusion chromatogram ICP/MS = inductively coupled plasma/mass spectrometry IMw = intermediate molecular weigh interval of the size exclusion chromatogram LHSV = liquid hourly space velocity 6573

dx.doi.org/10.1021/ef401540f | Energy Fuels 2013, 27, 6567−6574

Energy & Fuels

Article

(19) Caumette, G.; Lienemann, C.-P.; Merdrignac, I.; Bouyssiere, B.; Lobinski, R. J. Anal. At. Spectrom. 2010, 25, 1123−9. (20) Chainet, F.; Le Meur, L.; Lienemann, C.-P.; Courtiade, M.; Ponthus, J.; Brunet-Errard, L.; Donard, O. F. X. Fuel Process. Technol. 2012, 104, 300−9.

6574

dx.doi.org/10.1021/ef401540f | Energy Fuels 2013, 27, 6567−6574