Hydrodeoxygenation of fatty acid methyl ester in gas oil blend–NiMoS/alumina catalyst

2.1 Feed material

The refined FAME feedstock was supplied by Preem AB (Göteborg, Sweden), which contains fatty esters and acids with traces of neutrals and resin acids. The FAME feedstock contains: octadecadienoic acid C₁₈:2 (1.6 wt%), octadecenoic acid C₁₈:1 (1.0 wt%), octadecanoic acid C₁₈:0 (0.4 wt%), methyl palmitate (3.6 wt%), methyl linolenate (8.6 wt%), methyl linoleate (45.6 wt%), methyl oleate (20.7 wt%) and other esters (17.9 wt%).

LGO was the diesel fraction supplied by Preem AB. The LGO had 17.6 (v/v%) aromatics, 295 (ppmw) sulfur and C₈ to C₁₆ hydrocarbon fractions [22]. Table 1 provides the physicochemical properties of the LGO.

2.2 Hydroprocessing experiments

The Trilobe HDN-60 (NiMo/γ-Al₂O₃) catalyst (Criterion Catalysts, Farehem, UK), consisting of Ni-Mo/γ-Al₂O₃ with 2.5–3% Ni, 12.5–13.5% Mo and 83–85% γ-Al₂O₃, was used for the experiments [22]. The catalyst consisted of 1/32-in extrudes (calcined at 400°C overnight) and 31 ml (27.43 g) was loaded in the fixed-bed reactor at the middle section for in-site sulfidation before the hydroprocessing experiment. The catalyst surface properties are presented in Table 2. Sulfidation of the catalyst was carried out at 400°C and achieved by introducing H₂S (10 v/v% sulfur in H₂) at 200 mln/min for 3 h to the catalyst and kept overnight under N₂ presence [22]. After sulfidation of the catalyst, pure LGO was fed for 10 days at 360°C and 5 MPa in order to achieve a stabilized coke deposition on the catalyst surface, enabling stable activity during the experiments [30].

The experimental set-up consisted of a feed tank, pump, trickle-bed reactor, separator tank, gas flow meter and product collector [22], [30]. Both sides of the reactor were fastened with heavy bolts to a thickness of 19.3 mm, which were connected with inlet and outlet pipes. The down flow fixed-bed reactor, insulated with three independent heating zones, had an internal diameter of 1.8 cm; the void space above and below the catalyst bed was filled with borosilicate glass pellets with a diameter of 2 mm. The reciprocal dossier pump was used to pump liquid feed with H₂ gas in co-current flow. Products were withdrawn periodically from the separator tank. The reaction conditions were controlled through a proportional-integral-derivative controller and products were taken subject to the manual time control at ambient conditions [30].

2.3 Distillation and analysis

The product was collected at 20 h intervals, and subsequently placed in the distillation reboiler in N₂ atmosphere [30]. A 2 l reboiler in a simple distillation column operates at a desirable cut-off temperature of 162°C; water and lighter hydrocarbons were separated from products and precaution was taken not to loss middle distillates [22], [30]. The distillate containing lighter hydrocarbons and water was condensed through a water cooler (to keep the losses low). Between 90 and 99 vol% gas oil was recovered (as shown in Figure 1) from the residue and analyzed with respect to properties of the upgraded products.

Figure 1:

Volume % of gas oil recovered from the product by simple distillation on a cut point temperature 162°C. The distillate contains water and naphtha fractions (majority are C₅ to C₁₂ and traces of C₁₃ to C₁₆ hydrocarbons).

The upgraded product before and after distillation was analyzed using a gas chromatograph (GC) technique. Simulated distillation (ASTM D2887) was performed using a GC technique (Varian 3400, Agilent Technologies, CA, USA) equipped with a packed column (10% silicon OV-101, 80–100 mesh, 1 m×3.175 mm×2.00 mm) and a flame ionization detector (FID). The injector and detector were maintained at 300°C and the initial column temperature was maintained at 40°C and heated to 325°C (18°C/min). Finally the temperature was maintained at 325°C for 10 min. A Varian 4270 integrator was used for data computation.

The fuel gas collected from the top outlet of the separating tank was analyzed using the Clarus 500 online GC (Perkin Elmer, MA, USA). This GC had a thermal conductivity detector (maintained at 200°C with oven heat-up of 40–60°C at 2°C/min in helium gas) and FID detectors (maintained at 60°C in nitrogen gas), and four valves actuated by nitrogen gas at 0.4 MPa and connected with 600 link switch controllers that interpret signal to integrator [22], [30]. The specified detectors were a thermal conductivity detector to analyze CO, CO₂ and CH₄ and lighter hydrocarbon fractions were analyzed in an FID.

The physicochemical properties of the residue were analyzed at Preem refinery in Göteborg, Sweden. The spent catalyst was purified by extracting the feed solvent (LGO) with m-xylene for more than 22 h and washing in a Sohxlet apparatus. The washed catalyst was dried at 120°C in nitrogen atmosphere and the degree of coke deposition was analyzed at Karlshamn Kraft AB (Karlshamn, Sweden) using an ASTM D 5291 standard test method.

3.1 HDS reaction

In order to remove sulfur from the LGO feed, an NiMoS/γ-Al₂O₃ catalyst was used for the HDS reaction forming H₂S. The reaction was run at 360°C [7], [10], [11], and the LGO space velocity variation resulted in varying physicochemical properties of the products. The desired properties for LGO are aromatic composition with polyaromatic hydrocarbon (PAH), distinguished distillate recovery on boiling point and other properties (e.g. sulfur and nitrogen content, density, viscosity and cloud point) in accordance with ASTM standard (Table 3). The boiling point of the product was in the range of the LGO standard grade, i.e. between 180°C and 289°C. The aromatic content of the LGO feed was 17.8 v/v%, however, when the space velocity was increased from 0.4 1/h to 1.5 1/h, the aromatic content varied from about 16.2 v/v% to 17.9 v/v%, which is much higher than the LGO grade standard of ≤5 v/v%. The cloud point, density and viscosity were in the same range as the LGO standard grade. The sulfur content was reduced from 295 ppmw to less than 27 ppmw. However, the sulfur content in the upgraded product was strongly influenced by the LHSV. For instance at an LHSV of 0.5, the sulfur level was 1 ppmw, while increase in space velocity caused an increase in sulfur content in the product. Space velocities around 0.35–0.5 1/h result in high cracking, which leads to high yields of petroleum gas and naphtha with a distillate range below 170°C. Thus, the operating range of LHSV was considered between 0.75 and 1.5 for further results in orientation and discussion.

3.2 Co-processing FAME and LGO

LGO with 10% FAME (10%F) was co-processed using the experimental parameters as mentioned previously. When mixing FAME (feed) and LGO, a slight precipitation was observed that was dissolved by warming, without loss of feedstock. The analytical results of the upgraded products at elevated temperature and LHSV are shown in Table 3. We examined the conversion, selectivity and yield, such that:

Conversion= convertedfeed×100suppliedfeed,Selectivity=desiredproductundesiredproduct,Yield= desiredproduct × 100suppliedfeed

Upon deoxygenation of FAME, the desired product is C₁₈, in order to obtain a high cetane number of upgraded LGO. The formation of C₁₈ during upgrading of 10%F decreased for increasing LHSV and temperature, from 5 wt% to 4 wt% (concentration in the product) at elevated LHSV (0.75–1.5) and 5–2 wt% for elevated temperature (300–390°C) (Table 3). This formation was 5 wt% to 3 wt% and 4 wt% to 6 wt% for elevated LHSV and temperature, respectively. For different H₂/10%F ratios, there are no variations in viscosity, density or aromatic properties, except for a slight decrease in the naphtha concentration from 2.1 wt% to 1.6 wt% for increasing H₂/10%F ratios (Table 4).

CO and CO₂ are mainly formed when a carbonyl group is detached from the carbon chains of FAME, as the C=O bond is stronger than the C-C and C-H bonds, which is due to configuration stability [2], [18]. The degree of carbonyl detachment in the upgrading reaction can hence be followed in the product formation (as CO and CO₂). Consequently, H-β scission by thermal decomposition leads to formation of carbocations in paraffin chains [29], [30]. Hydrogen redistribution in paraffin chains has a tendency to further inhibit H-β scission in forming smaller carbonium ion groups, such as lighter hydrocarbons of low boiling chemical species.

Deoxygenation and decarbonylation reactions, that involve partial cracking, mainly occur at the surface and in the pores of the catalyst [22], [23], [24], [25], [26]. This is confirmed by the decrease in naphtha concentration as the H₂ content was increased, for constant LHSV. The aromatic content did not change upon varying the H₂ ratio owing to the equilibrium between dehydrogenation and hydrogenation, which was mainly dependent on temperature. The C=C reactivity in large molecules in FAME, such as methyl linoleate and methyl linolenate, depend on the type and position of the fatty acid component in the molecule. Thus, mainly linoleic and oleate compounds undergo cracking at the C=C bond position via carbonium ion scission (at mass transfer limitation) [2]. However, the C-(CO) (α-bond) bond will likely break more easily than the C=C bond in unsaturated hydrocarbon chains, whereas at higher temperatures the C=C bonds (e.g. C=C bonds 9 or 12 in methyl oleate) will likely break in monounsaturated compounds [2], [30]. Similarly, decomposition and H-β scission have both been observed for C-C and C-H bonds, which have the same endothermicity. Hence, Osmont et al. [2] suggest that the two reaction pathways in cracking (i.e. decarbonylation and decarboxylation) are equally probable for initiating the process [23], [31], [32], [33]. The thermal decomposition of FAME, occurring around 350°C, could for example be α bond breaking resulting in C₁₇. The probability for cracking to occur at the bond of the seventh position and ninth position is generally low because of the weak pi-bonding over acidic sites or on alumina [30], [34], [35].

The aromatic content in the product increases steadily for the studied temperature from 16 v/v% to 18 v/v%. In particular, for temperatures above T=350°C, the aromatic content in the product increases (from 17 v/v% to 22.8 v/v%) as can be seen in Table 3, which is above the thermodynamic equilibrium. The feed (10%F) contains 0.58 v/v% PAH, which is higher than the standard limit which should be less than 0.02 v/v% in conventional Swedish diesel. Yet, as shown in Table 3, the PAH content in the product increases from 0.32 v/v% to 0.55 v/v%, for temperatures between 300°C and 370°C. The shorter residence time (higher LHSV) reduces the diaromatic or PAH (from 0.64 v/v% to 0.36 v/v%) but increases the monoaromatic content. By varying the H₂ partial pressure in the system, the PAH formation was unchanged in products (Table 4). Change in H₂ resulted in increased PAH (0.98 v/v%). These results indicate that an increasing temperature and a decreasing space velocity cause an increase in PAH and a reduction in monoaromatic and diaromatic compounds (Table 3). Therefore, it is important to identify the optimum conditions for this process, e.g. altering the dehydrogenation mechanism by decreasing the H₂ concentration in the feed, as the H₂ content directly influences the aromatization. Moreover, the presence of monoaromatic compounds in the feed requires a high residence time over the catalyst sites in order to undergo a de-aromatization reaction within the hydrogenation conditions used in the present study.

3.3 Catalyst deactivation

Generally, the sulfur content in the upgraded product is less than 10 ppmw (Figure 2). Initially, a few product samples contained about 25 ppmw sulfur and after some time on stream, the samples showed less than 10 ppmw of sulfur. The explanation may be that the active sites have a weak Fermi level over transition metals [35], which prevailed to form H₂S by excess hydrogen. The vacant sites that are replaced by sulfur atoms have a high reaction rate for H₂S formation with adjacent H-donor, however this is dependent on the interfacial mass transfer on the catalytic surface [36]. Thus, continuous addition of sulfur on vacant sites significantly lowers the deactivation of the catalyst [36]. Preferably, branched cyclic compounds in the LGO fraction had steric hindrances; desulfurization of the reaction rate reduced by hindrance of presences of adjacent S atoms near to the alkyl group, for instance alkyl-benzo thiophene, in middle distillate. This scission determines the reaction rate over the active sites at the surface [37].

Figure 2:

C₁₈/C₁₇ ratio versus hydrogenation period of 10% fatty acid methyl ester (FAME) with light gas oil (LGO) at 360°C with liquid hourly space velocity (LHSV)=1 1/h and P=5 MPa.

Table 2 shows the porosity and surface area of a fresh catalyst (fresh NiMo) and a hydrotreated spent catalyst (spent NiMo). The pore volume of the fresh and spent NiMo catalysts is 0.31 cm³/g and 0.12 cm³/g, respectively. Similarly, the surface area is 156 m²/g and 54 m²/g for fresh and spent NiMo catalysts, respectively. The pore plugging and diminishing surface area are indications of coke deposition. The selectivity of the product measured as the C₁₈/C₁₇ ratio is 0.5 over the fresh catalyst, while after an extended period of reactor run, this ratio increased to 1.3 (Figure 3). This low ratio may be due to domination of the decarboxylation/decarbonylation reaction route as compared to a deoxygenation reaction pathway, in both pores and on the surface of the catalyst. Here, a major part of the decomposition product indicates that the active sites were blocked by coke formation in the pores. The selectivity towards the upgraded products is mainly determined by the catalyst structure [38] and the reaction conditions. Thus, -Mo-S- on the surface contributes highly to the deoxygenation reaction compared to other types of promoters [39]. The presence of sulfur and H₂ induces a higher degree of deoxygenation, which tends to increase the C₁₈ presence in the product. The FAME deoxygenation on support material, γ-Al₂O₃, has acid sites, and its ability to induce cracking of hydrocarbon on the surface and pores is comparatively higher than in Mo active sites [39]. The sulfur content in the product varies between 1 ppmw and 8 ppmw for all samples measured (Table 2), suggesting that there is low deactivation over the catalyst. By contrast, high PAH (0.64 v/v% in feed LGO) from the feed has a tendency to increase the coke formation in the pores, which is in agreement with previous results [40].

Figure 3:

Amount of sulfur content versus hydroprocessed 10% fatty acid methyl ester (FAME) in light gas oil (LGO) for all conditions.

3.4 Lighter hydrocarbon

For C₁₈ formation, the ester group is the most active for hydrogenation of the carboxylic acid group, indicating that the hydrogenation reaction occurs through the carboxyl group following the route via the aldehyde group forming alcohol and, finally dehydration to form C₁₈. Methane is the main gas product that detached from the ester group. Furthermore, there is evidence for CO₂ and CO formation (Figure 4). Here, as the temperature increases, formation of CO by decarbonylation is between 0.2 wt% and 2.5 wt%, with a simultaneous increase of CO₂ in the gas phase, although marginally from 0.1 wt% to 0.3 wt%. Decarbonylation is predominantly higher than decarboxylation during thermal decomposition. This thermal effect on the bond scission of H₃C-O is energetically favored over the O-(CO) bond [2], [30], [41]. Thus, both the H₃C-O* and *O-(CO) radicals form alcohol; further alcohol dehydrated into water and hydrocarbon. In the case of the *O-(CO) radical, decomposition occurs by α-bond cleavage, which leads to formation of CO and C₁₇ compounds. Figure 4 depicts high methane formation at low space velocity and relatively high temperatures. With high hydrogen partial pressure in the reactor, methanation possibly occurs, as CO disproportionation reaction produces C radicals and forms methane. Dissociation of CO is a rate-determining step of methanation [42], hence the activation energy of the hydrogenation rate increases at low space velocity. We found that it is not viable to form CO from CO₂ in the methanation process, whereas, only CO hydrogenation favors high methane formation. The results indicate that the decarbonylation depends only on temperature and there is no effect by space velocity of feed, whereas decarboxylation depends on both temperature and space velocity.

Figure 4:

CO, CO₂ and CH₄ concentration in the gas outlet with elevated temperature and space velocity at 5 MPa pressure. Here, liquid hourly space velocity (LHSV) was ■ 0.7, ● 1.0, ▲ 1.2 and ▼ 1.5.