- Email: [email protected]
Hydroprocessing of fatty acid methyl ester containing resin acids blended with gas oil
Fuel Processing Technology 126 (2014) 435–440
Contents lists available at ScienceDirect
Fuel Processing Technology journal homepage: www.elsevier.com/locate/fuproc
Hydroprocessing of fatty acid methyl ester containing resin acids blended with gas oil Shanmugam Palanisamy, Börje S. Gevert ⁎ Chalmers University of Technology, Göteborg, Sweden
a r t i c l e
i n f o
Article history: Received 14 March 2013 Received in revised form 13 May 2014 Accepted 19 May 2014 Available online xxxx Keywords: Hydrodeoxygenation Decarboxylation FAME Fatty acids Abietic acid Resin acids
a b s t r a c t Fatty acid methyl ester (FAME) with resin acids was blended with light gas oil (LGO) from 10 to 30 wt.% and then investigated using elevated temperature (300 to 370 °C) at 5 MPa on NiMoS/alumina in a trickle-bed reactor (TBR). Hydroprocessing of blended LGO showed lower S content (less than 8 ppm) at 100% conversion and reached the target cloud point on middle distillate diesel fuel. In this study a significant change in the physical and chemical properties of LGO was observed. For the distillate product, viscosity, distillation curve, nitrogen content and density were all in the expected standard grades of diesel. Here, a presence of sulfur in LGO significantly reduced catalyst deactivation. Higher concentrations of aromatic content and polycyclic aromatic hydrocarbons degraded to reach the EU diesel standard; in particular, the resin content in FAME could enable up to 50% aromatization. Elevated experimental temperatures increased decarboxylation as compared with deoxygenation for all compositions of FAME and resin acids in this upgrading process. Carbonyl compounds like CO and CO2 formation in lighter products result from resin acids decarbonylation and decarboxylation reaction pathway. © 2014 Published by Elsevier B.V.
1. Introduction The on-road legislation by EU member states is in force from 2012 for gas oil's and diesel's sulfur not to exceed 10 ppmw. A large number of studies have investigated desulfurization of middle distillates to adopt in EU member states refineries. It achieved an impressive milestone on commercialization of the hydrodesulfurization (HDS) process, particularly for middle distillate in refineries. These middle distillate fuels were vital for the transportation sectors [1–4]. One of the middle distillates – light gas oil (LGO) – had a high value because of properties and energy content that were dependent on natural and fuel adaptability. The cetane number was the main property that needed to be considered in the qualitative analysis to substitute renewable fuel with LGO [5–8]. The current standard cetane index of EU is at the range of 46 to 51 according to EN 596 standards. Sometimes, the premium diesel exceeds as high as 60. Addition of carbon rich substitutes like FAME increases cetane number. An increase in the cetane number of LGO resulted in a lower of shorter ignition delay, which tended to lessen combustion noise, leading to increased energy
⁎ Corresponding author at: Ecotraffic ERD3, Kempross AB, Larsereds lyckor 14, 42539 Hisings kärra, Sweden. Tel.: +46 702195834. E-mail address: gevert@ecotraffic.se (B.S. Gevert).
http://dx.doi.org/10.1016/j.fuproc.2014.05.015 0378-3820/© 2014 Published by Elsevier B.V.
efficiency and power output for power engines [9]. However, an increase in cetane number proved to be non-friendly for easier starting in severe winter weather and reduced both smoke and odor, which is in need of upgrading further [9–11]. Although kerosene and turbine fuels have similar properties (such as LGO), they had different specifications corresponding to their intended use. LGO mostly had paraffin and olefin type hydrocarbons with a boiling range between 125 and 300 °C, sometimes mixed with various cracked distillate fractions to increase volume requirement [12–14]. An increase in octadecane (C18) and heptadecane (C17) in production by hydrodeoxygenation (HDO) of fatty acid methyl ester (FAME) in LGO gives a better cetane number for diesel fuel compared to if no n-C17 and n-C18 was added [15–17]. Presently, combining HDO and HDS in ultra-low sulfur diesel production is recognized in some refineries through NiMoS catalyst [4,18]. Resin acids (e.g., abietic acids) present in tall oil were the main renewable sources. These acids have a complication of tri-ring structure making hydrogenation difficult in downstream products [7,19]. A primary aim of research has been to understand the fundamental aspects of the NiMoS catalytic reaction [18,20–23], develop/design a suitable condition and demonstrate the physico-chemical properties of deoxygenation products rich in carbon content [7,24]. The main objective of this study was to examine the effect of hydroprocessing FAME with resin acids (FAME-resin) blended with LGO (referred to as middle distillates). In this study, LGO was blended with FAME-resin and hydrogenated on a NiMoS catalyst for temperatures ranging between 300 and 370 °C at 5 MPa and H2/feed (mln/mln) = 80.
436
S. Palanisamy, B.S. Gevert / Fuel Processing Technology 126 (2014) 435–440
2. Experimental
temperature for removal of water and lighter hydrocarbons from the downstream product at a cut point temperature of 162 °C. The residue was analyzed to check for essential features of physicochemical properties.
2.1. Feed material The feed contained 78% FAME and 20% resin acids of the carboxylic group on tri-ring cyclic hydrocarbon with the rest as neutral fractions. The classified fatty group with resin acids is shown in Table 1. The feed was FAME-resin blended with LGO with three ratios: 10, 20 and 30 wt.%. The properties of the LGO were aromatic content 17.8 (v/v %), sulfur (S) content 295 (ppmw) and paraffin fraction from C8 to C16. Table 1 provides additional information about the LGO. 2.2. Reactor system A fixed trickle-bed and semi-batch reactor with a sulfidized NiMo/ alumina catalyst was investigated. The major reaction parameters included liquid hourly space velocity (LHSV = mln of liquid/ml of cat. ∗ h), feed mixture, hydrogen flow and reaction temperature (T) (°C). For the investigation, the commercial catalyst Trilobe HDN-60 (NiMo/ γ-Al2O3) from Criterion Catalysts was used and consisted of NiMo/γAl2O3 with 2.5–3% Ni and 12.5–13.5% Mo with 83–85% γ-Al2O3. The catalyst had the shape of 1/32-in extrudes with a surface area of 156 m2/g and a pore size of 7–8 nm calcined at about 400 °C for overnight: 31 ml (27.43 g) was loaded in the center of trickle-bed reactor (TBR) for insitu sulfidation. The catalyst was heated to 400 °C at the rate of 10 °C per minute and the temperature maintained for 2 h under nitrogen flow of 100 mln per minute before sulfidation. Later, sulfidation was achieved by allowing H2S as 10% by volume of S in H2 at 200 mln per minute for 3 h and kept overnight under a small nitrogen flow. At 360 °C and 5 MPa, LGO was flooded for 10 days, until stabilized coke developed over catalyst and steady state was achieved for further experimentation. The reactor set-up consisted of a feed tank, Dossier pump, TBR, separator tank, gas meter and collector. The down flow TBR, insulated with three independent heating zones, had an internal diameter 1.8 cm, with borosilicate glass pellets of 2 mm in diameter filled in void space. The feed that was fed by the reciprocal dossier pump was mixed with H2 or N2 gas before being allowed on a co-current flow into the reactor: all products were withdrawn periodically from the separator tank [25]. The experiment used a semi-batch reactor made up of stainless steel with 300 ml capacity that was connected to a gas inlet and outlet. The thermo couple, heating system and magnetic stirrer was integrated in an eight microprocessor-based process controller to maintain the operating condition steadily. The experiments were carried out with the same stirring speed. In this reactor, 5 wt.% abietic acid in hexadecane was used at temperatures of 300 to 350 °C and 5 MPa experimented on a NiMoS/alumina catalyst. The reactor was loaded with 150 ml of solution that includes 1 g of a defined amount of NiMoS catalyst and 3 g of tetradecane as internal standard. The product collected from the reactor was placed in the distillation re-boiler on N2 flow at 100 mln per minute. The simple distillation column consists of 2 L capacity re-boiler and heated to a desired
2.3. Analysis Simulated distillation [SIM-DIST] (ASTM D2887) was performed on the upgraded product by gas chromatograph (GC) technique. Data computation was performed with a Varian 4270 integrator. The flue gas from the reactor was collected in a gas bottle at ambient condition and subsequently analyzed using the Clarus 500 online GC. This GC has an inlet and outlet sampling values with a thermal conductivity detector (TCD) and flame ionization detector (FID), as well as four valves actuated by nitrogen gas at 0.4 MPa. The detector out signal was connected with a 600 link switch controller that interprets the signal to an integrator. The detectors are TCDs for analyzing CO, CO2 and CH4, and FID for lighter hydrocarbon fractions. Helium was used as carrier gas for the TCD, which was maintained at 200 °C with oven heat-up of 40 to 60 °C at 2 °C per minute and nitrogen as carrier gas for the FID at 60 °C as oven temperature. The cloud point, aromatic content, density, nitrogen content, sulfur content, SIM-DIST and viscosity properties were analyzed from the residue, which was performed by Preem refinery, Sweden. 3. Results and discussion 3.1. Remarks In all experiments, the general hydrogenation of FAME produced normal hydrocarbons (e.g., n-C17 and n-C18) as reported in previous studies [26,27]. The ratio of liquid to gas inlet was held constant throughout the experiments with comparability of mass balance and thermodynamic calculation. Molar balances based on the gas product analyses were calculated (error elimination was less than 5% on average). The assumption regarding hydrogenation includes deoxygenation and cracking, which were summed in the result description. The recovered residue of products mainly contains hydrocarbons between octane and octadecane with traces of high hydrocarbons. Moisture content in the products was excluded. 3.2. Aromatic composition The feed of three compositions were tested at elevated temperatures and space velocity in the TBR. The aromatic content in resin acid was done in order to study the intra-couple of deoxygenation and de-aromatization process in ensuring the product property. LGO contains 17.8 vol.% aromatics by the addition of FAME-resin aromatic compositions and retained at 17.7 vol.% for all three feed types. The upgraded products have aromatic content between 17 and 18 vol.% (see Table 2) for different blended feed compositions. This observation led us to understand that there was neither notable aromatization nor cyclic scission. However, reaction mechanism
Table 1 Classified composition of resin acids with fatty acid methyl esters supplied by Preem AB. Components
wt%
Components
wt%
Components
wt%
Octadecadienoic acid C18:2 Octadecenoic acid C18:1 Octadecanoic acid C18:0 Methyl palmitate Methyl linolenate Methyl linoleate Methyl oleate Other esters Total free fatty acids and esters
1.25 0.80 0.28 2.82 6.77 35.75 16.26 14.05 78.41
Isopimaric acid Levopimaric acid Palustric acid Neoabietic acid Dehydroabietic acid Sandaracopimaric acid Pimaric acid Abietic acid Total resin acids
1.37 0.21 1.32 0.34 4.45 0.42 2.40 8.69 19.29
β-Sitosterol Squalene Other
0.16 0.34 1.79
Total neutrals
2.29
S. Palanisamy, B.S. Gevert / Fuel Processing Technology 126 (2014) 435–440
437
Table 2 Physico-chemical properties of LGO and different blend products obtained by HDS on Ni–Mo/alumina catalyst at 5 MPa pressure. Parameters
FAME-resin + LGO at To = 350 °C LGO
Compositions Aromatic (vol%) PAH (vol%) ≤C8 (wt %) C9–C16 ΣC17 ΣC18 Heavy fraction
17.8 0.64 1 99
Distillation temperature V/V rec. (°C) (SIM-DIST) IBP 204 10%. 208 50% 215 65% 235 90% 265 95% 283 FBP Others properties Cloud point S in feed (mg/kg) S in the product (mg/kg) Density at 15 °C (kg/m3) Viscosity at 40 °C (mm2/s)
−40 295 819 1907
To (°C) at 30 wt.% FAME-resin + LGO 30%
20%
17 ** 1.0 90.1 4.0 4.5 0.4
16.9 ** 2.0 80.8 7.6 8.3 1.3
213 220 237 244 265 284
208 217 233 243 269 288
205 218 246 258 297 308
211 220 246 258 299 312 332
209 220 246 258 298 311 327
202 220 246 258 296 309 325
206 217 244 256 294 307 324
−18 276 6 821 1928
−10 236 4 820 2031
−6 207 2 821 2179
−5 207 b1 821 2220
−5 207 5 821 2212
−6 207 1 821 2165
−9 207 1 822 2105
17.7 0.63 2.3 75.2 10.1 10.2 2.2
330 17.6 0.45 0 78.4 8.4 10.7 2.5
340 17.8 0.53 1.6 75.2 9.3 11.6 2.3
360
Reference test method
10%
18.3 0.73 2.5 75.9 10.1 9.6 1.9
370 19.5 0.91 2.9 75.4 10.3 9.1 2.3
SS 155116:1993
ASTM D86 ASTM D86 ASTM D86 ASTM D86 ASTM D86 ASTM D86 ASTM D86
ASTM D2500-05 ASTM D5453-00 ASTM D5453-00 ASTM D4052-09 ASTM D7042-04
Note: ** not available.
between mono- and poly-aromatic hydrocarbons (PAH) of resin acids in LGO is not briefly estimated, because of deducible complication in combined hydroprocessing, cracking, hydrogenation and (de) aromatization. Resin de-aromatization on the resin group was tested in the semibatch reactor with a model compound of 5 wt.% abietic acid in solvent and 1 g of NiMo/Al2O3 catalyst, as shown in Table 3. The final product contains a total aromatic content of 2.1 and 2.4 wt.% for 3-hour residence time at 340 and 350 °C, respectively. Table 3 shows that 2 wt.% of aromatic content was mono group with 0.3 wt.% di-aromatic and 0.1 wt.% PAH, with 100% conversion of abietic acids. Above values estimated a 50% of the resin acid that favored dehydrogenation or aromatization in the above-mentioned condition. The C\C scission of α-bond scission in abietic acid (for reference in Fig. 5) results to form aromatics by dehydrogenation. The lack of H2 donor on catalytic surface tends to Table 3 Analysis of abietic acid for different temperatures at 5 MPa pressure for residence time = 3 h in batch reactor. Parameters
T (°C) 320
340
Gas composition (vol%) H2 CO2 CO C1 C2/C= 2 C3/C= 3 C6/C5/C4
96.30 0.71 2.33 0.48 0.07 0.08 0.02
95.88 0.52 2.25 0.95 0.13 0.24 0.02
Liquid paraffins (wt%) b320 320–360 360b
45 43 10
88 12 0
Properties Aromatics (wt%) Mono aromatics (wt%) Di-aromatic content (wt%) Poly-aromatic content (wt%) Density at 15 °C (kg/m3)
** ** ** ** **
Note: ** not available.
2.1 1.9 0.1 0.09 780.9
350 93.26 0.66 2.4 1.14 1.5 0.07 0.99
100 0 0
2.4 2 0.3 0.17 781.6
increase re-distribution of H through cyclic group. The Π-bond interaction between the benzene group and catalytic surface induces the H2 re-distribution on vacant sites at equilibrium condition. Thus, the high aromatic content formation indicates de-hydrogenation and alkylbenzene formation was appropriate in this defined temperature region. At 30 wt.% FAME-resin, total PAH increased by 0.46 vol.% subject to temperature (Table 2). Total aromatic content increased from 17.6 to 19.5 vol.% with a temperature between 330 and 370 °C. Observation of lower aromatic composition at the lower temperature (i.e. 330 °C) corresponds with the theory that the equilibrium condition of aromatization has not been attained. Both the catalytic and thermal effects had a tendency to form cyclization or aromatization, implying a more appropriate condition [28]. Metal site reactivity of aromatic formation influenced the reduction of S inhibition. The thermodynamic dependency of aromatic compounds was identified as an increase in aromatic concentration with increased temperature by attaining an equilibrium condition. However, as seen in Table 2, different compositions of FAME-resin had little effect on aromatic content compared with temperature. As a result, resin aromatic content did not change under all experimental conditions. It clarified that the resin group did not induce any dehydrogenation or polymerization, which neglected the hydrogen re-distribution or simultaneous hydrogen source availability for the deoxygenation process. Dehydrogenation induces PAH formation in LGO and FAME fractions which are free radicals formed by β-H scission over alkane chain of alkyl-benzene with respect to H2 deficient due to mass transfer resistance [29]. 3.3. Upgraded LGO properties In SIM-DIST analyses, it was observed that the final boiling point (FBP) decreased gradually from 332 to 324 °C with an increase in reaction temperature (Table 2). However, decomposition was unavoidable at higher temperatures. A cracking process, from higher hydrocarbon to lighter hydrocarbon, was clearly observed in all tested condition with different compositions. So, cloud point analyses focused without lighter hydrocarbon by removing boiling point range hydrocarbons not exceeding 165 °C. Cloud point changed from − 5 to − 9 °C with the tested elevated temperature. Further, the presence of olefins in FAME and LGO was found to influence cloud point to some extent.
Variation in paraffin and olefin composition depends on partial pressure and H-donor. Here, in our tested conditions, hydrogenation favors to form paraffin [20]. So this condition emphasizes that olefin is limited in cloud point variation. Further, it clears that isomeric products formed from resin acids had direct influence in cloud point. Sulfur content in the product was found to be lower with increased FAME-resin in the feed. The product had below 8, 5 and 2 ppmw S for 10, 20 and 30% blend FAME-resin with respective S in feed contains 276, 236 and 207 ppmw for all tested conditions. However, estimation to less than 1 ppmw of S concentration can be achieved by upgrading through a maximal of 200 ppmw S in feed, as shown in Table 2 in the comparison of feed and the product S. Instead of bypassing S in the product, higher residence time of feed or lower concentration of S in feed has a tendency to remove as H2S. Lower Fermi bond energy of catalytic active sites induces higher activity at the surface [30]. The inhibition effect of H2S discussed in some investigations [30] suggests that there was a vacancy by naked MoS2 at the corner of the active sites induced to increase hydrogenation through Π-bonding. The density of the product related to all the experiments was between 819 and 821 kg/m3. The viscosity was lower for elevated temperatures and increased with an increase in the blend ratio. Thus, relative value of viscosity raises 100 mm2/s for every 10% increase in FAME-resin composition in LGO. 3.4. Deoxygenation/decarboxylation Decomposition can be directly observed with a decrease in the C18/ C17 ratio, where the existence of a decarboxylation/decarbonylation [27] effect was evidently demonstrated (Fig. 1). Senol et al. [30,31] suggest that both deoxygenation and decomposition have a conversion ratio of 1:1. Our experiments showed deoxygenation was higher and thus more active on the carboxylic acid group to form C18 compounds by excess inhibition of S and H2. The conversion for 30% FAME-resin has between 8.4 and 10.7 wt.% in the product of C18 and C17 compositions, respectively, and heavy fraction (above C18) has 2.5 wt.% relative to temperature (see Table 2). Further, middle distillates (C6–C16) had a composition of about 75 wt.% with the rest as lighter hydrocarbons. As depicted in Fig. 1, selectivity determination indicated that the ratio of deoxygenation/decarboxylation reduced from 1.3 to 0.9 with increased temperature (from 330 to 370 °C). In the case of 10 and 20 wt.% FAMEresin, C18/C17 remains at 1.3 to 1.1. At lower temperatures, high selectivity of C18 in the product denoted the better selectivity from the reaction path. This C18 was mainly achieved through direct hydrogenation and minimized the decomposition effect. Direct hydrogenation depends on two criteria: activity with the Lewis sites at the catalyst [32] and a
2,5
100
2,0
80
1,5
60
1,0
40
0,5
20
0,0 320
340
350
360
370
0 380
To (oC) Fig. 2. CO, CO2, H2 and light hydrocarbons concentration in gas outlet with various temperature for 30% blend in LGO, Po = 5 MPa and LHSV = 1. Here, ●: CO2, ■: CO, ▲: CH4, ♦: H2 and ☼: C1–C6 (flue gas) in gas phase.
temperature dependent C\C bond cleavage. As temperatures rise, a gradual decrease in C18 and an increase in C17 in the product drive a kinetically mobilized rate of the decomposition pathway. Thus, an increase in temperature obviously influences the thermal decomposition through dissociation of the carboxylic group as CO and CO2. This lower decomposition finalized that the rate of hydrogenation was high on the hydroxyl group of acid (\OH bond), followed by ester formation, resulting in the aldehyde group (C_OH), i.e. (CO)\OH N (CO)\O\C N HC_O. However, the source of the CO2 and CO molecules in lighter gas varied with bond cleavage in H3C\O\(CO)\C\, which rose up to 0.5 vol.% as the total in flue gas. The carbonyl gas observation for both elevated temperature and FAME-resin compositions is displayed in Figs. 2 and 3. The main compositions of flue gas consisted of CO, CO2, CH4 and a small fraction of lighter hydrocarbon with the rest containing H2. Decarboxylation was found constant with the same fraction of CO2, which was notable decomposition on the carboxylic group at resin acid (Figs. 2 and 3). This finding demonstrated that CO2 formation from abietic acid observed the same fraction for all tested temperatures (Table 3). Even longer residence of abietic acid resulted in the formation of higher distillates by a preliminary path of deoxygenation and decarboxylation (Fig. 4). The CO fraction increases with an increase in 100
1,8 1,6
10% FAME-resin 20% FAME-resin 30% FAME-resin
Gas conc. (vol%)
1,3
80
1,4
1,2
1,1
1,0
1,2
60 1,0 0,8
40 0,6 0,4
Gas conc. (vol%)
1,4
C18/C17
330
Gas conc. (vol%)
S. Palanisamy, B.S. Gevert / Fuel Processing Technology 126 (2014) 435–440
Gas conc. (vol%)
438
20
0,2
0,9 0
0,0 10% blend
0,8 330
340
350
360
370
Temp. (oC) Fig. 1. C18/C17 vs. temperature (°C) of different blended LGO at LHSV = 1 and P = 5 MPa.
20% blend
30% blend
tall oil blend with gas oil Fig. 3. CO, CO2, H2 and light hydrocarbons concentration in gas outlet for different blends in LGO, Po = 5 MPa, To = 350 °C and LHSV = 1. Here, ●: CO2, ■: CO, ▲: CH4, ♦: H2 and ☼: C1–C6 (flue gas) in gas phase.
S. Palanisamy, B.S. Gevert / Fuel Processing Technology 126 (2014) 435–440
80
decarbonylation has noted interest about the properties of the product. Result in Fig. 4 shows that the cyclic group's presence at a SIMDIST ranges of 320–360 °C increases steadily. Abietic acid (N 360 °C) decreases drastically at shorter residence time and decreases steadily afterwards. Mono-/di-cyclic or iso-alkanes are the end product of the resin group with substantial removed oxygen in \COOH to form B and C products (see in Fig. 5). The resin groups deoxygenate to form a cyclic group and hydrogenation of unsaturated ring followed (Fig. 5). The scission of iso-alkanes from cyclic group results in lighter gas products. So, no changes in concentration of products after 15 min residence time, have noticed at SIM-DIST range below 320 °C. Thus indication of SIM-DIST range of 320–360 °C of middle distillates exponentially increases for residence times in Fig. 4, explains cracking of cyclic group and reformulation of aromatic compounds like products D, E, F and G in Fig. 5.
70
Distillate boiling point
Concentration (wt%)
60
below 320 oC 320 to360 oC above 360 oC
50 40 30 20 10 0 0
30
439
60
90
120
4. Conclusion
Time (min) Fig. 4. Abietic acid hydrogenation in batch reactor at T = 320 °C and P = 5 MPa on NiMoS/alumina.
Aromatic content showed a steep increase on upgraded products at elevated temperatures. Furthermore, it is clear from the data that the resin group had no influence on this increase. Paraffins to aromatic in LGO were found to significantly increase and the PAH concentration had a particularly strong influence on the quality of LGO. Upgrading resin acids and FAME increased the cetane number of LGOs, but domination of decarboxylation and decarbonylation is needed to be controlled in elevated conditions. Upgraded LGO with FAME and resin acids can be processed to obtain b1 ppmw of S. From this experiment, it could be concluded that deoxygenation and decarbonylation had domination on resin acids. However, the PAH resin structure led to mono-aromatic by cyclic scission to some extent. Dehydrogenation, isomerization and polymerization are limited or not existing with the used conditions in the process.
temperature through the influence of the decarbonylation path. This event occurs because of the tendency of a reaction path on dehydration by the formation of aldehyde from ester, which led to decomposition with a loss of 1 mol of carbon and 2 mol of oxygen. Dependency of CO and CO2 fraction with different FAME-resin compositions had a notably linear increase. The estimation of the CO and CO2 in this experiment showed that deoxygenation and decarbonylation influenced the reaction mechanisms. The CH4 formed in flue gas of 1 to 2.5 vol.% consisted of different paths: the large fraction of CH4 gas mainly evolved from methyl group in FAME (Fig. 3) and methanation process by converting CO to CH4 due to the presence of excess hydrogen favored to increase methane in lighter gas. Other sources by cracking which influenced by temperature tend to raise CH4 and other lighter hydrocarbons (C2–C6) (Fig. 2).
References [1] D. Mohan, C. Pittman, P. Steele, Pyrolysis of wood/biomass for bio-oil: a critical review, Energy & Fuels 20 (2006) 848–889. [2] S. Czernik, A.V. Bridgwater, Overview of applications of biomass fast pyrolysis oil, Energy & Fuels 18 (2) (2004) 590–598. [3] M. Lapuerta, O. Armas, J. Rodríguez-Fernández, Effect of biodiesel fuels on diesel engine emissions, Progress in Energy and Combustion Science 34 (2) (2008) 198–223. [4] Teh C. H, Deep HDS of diesel fuel: chemistry and catalysis, Catalysis Today 98 (1–2) (2004) 3–18.
3.5. Reaction mechanism of resin acid The main reaction path for deoxygenation, decarbonylation and decarboxylation in FAME hydroprocessing has been discussed previously in Section 3.5. The model resin acids (abietic acid) deoxygenation and
-CO/CO2
A
F a
a
B 2H2
O
HO
Mono -cyclic Di-cyclic group n-Alkane n-Alkenes
2H2
-H2O
G
2H2 2H2
a
C D
E
Fig. 5. Abietic acid hydrogenation mechanisms. (A = abietic acid, B = decarboxylic/decarbonylic path and C = deoxygenate path and F, G, E = (de)aromatization), (note: a = α-bond).
440
S. Palanisamy, B.S. Gevert / Fuel Processing Technology 126 (2014) 435–440
[5] A. Demirbas, Progress and recent trends in biodiesel fuels, Energy and Combustion Science 33 (2007) 1–18. [6] J.T.G. Monnier, D.W. Soveran, A. Wong, E.N. Hogan, M. Stumborg, USA 5705722, Natural Resources Canada, 1998. [7] Robert Coll, Siddharth Udas, William A. Jacoby, Conversion of the rosin acid fraction of crude tall oil into fuels and chemicals, Energy and Fuels 15 (2001) 1166–1172. [8] T. Laza, A. Bereczky, Basic fuel properties of rapeseed oil-higher alcohols blends, Fuel 90 (2) (2011) 803–810. [9] P. Ghosh, Predicting the effect of cetane improvers on diesel fuels, Energy & Fuels 22 (2) (2008) 1073–1079. [10] K. Hashimoto, M. Ikeda, M. Arai, M. Tamura, Cetane number improvement of diesel fuel by autoxidation, Energy & Fuels 10 (6) (1996) 1147–1149. [11] A. Demirbas, Relationships derived from physical properties of vegetable oil and biodiesel fuels, Fuel 87 (2008) 1743–1748. [12] T. Marker, J.P., T. Kalnes, M. McCall, D. Mackowiak, B. Jerosky, B. Regan, L. Nemeth, M. Krawczyk, S. Czernik, D. Elliot, D. Shonnard, Opportunities for biorenewables in oil refineries, http://www.osti.gov/energycitations/product.biblio.jsp?osti_id=861458 2005 UOP, Des Plaines, Illinois. [13] V.A. Yakovlev, S.A. Khromova, O.V. Sherstyuk, V.O. Dundich, D.Yu. Ermakov, V.M. Novopashina, M.Yu. Lebedev, O. Bulavchenko, V.N. Parmon, Development of new catalytic systems for upgraded bio-fuels production from bio-crude-oil and biodiesel, Catalysis Today 144 (2009) 362–366. [14] J.S.J. Alonso, J.A.L. Sastre, C. Romero-Avila, E. Lopez, A note on the combustion of blends of diesel and soya, sunflower and rapeseed vegetable oils in a light boiler, Biomass and Bioenergy 32 (9) (2008) 880–886. [15] Paivi Maki-Arvela, Iva Kubickova, Mathias Snare, Kari Eranen, Dmitry Yu Murzin, Catalytic deoxygenation of fatty acids and their derivatives, Energy & Fuels 21 (2007) 30–41. [16] K. Breda, Biodiesel usage at low temperature, Fuel 87 (7) (2008) 1306–1317. [17] X. Chen, L. Yongbin, Qun Shao, Study on cold flow properties of biodiesel from typical materials, Shiyou Yu Tianranqi Huagong 39 (6) (2010) 491–493 497. [18] H. Topsoe, B. Hinnemann, J.K. Norskov, J.V. Lauritsen, F. Besenbacher, P.L. Hansen, G. Hytoft, R.G. Egeberg, K.G. Knudsen, The role of reaction pathways and support interactions in the development of high activity hydrotreating catalysts, Catalysis Today 107–108 (2005) 12–22. [19] F. Bauer, J.C., P. Johansson, LindenJ. , M. Nyström, L. Ruden, Tallolja som biobaserad dieselersättning — Katalytisk hydrogenering av hartssyror. KBTX01-10-6. BSc, Chalmers University Of Technology, Göteborg, Sweden, 2010.
[20] M. Snare, I. Kubickova, P. Maki-Arvela, D. Chichova, K. Eranen, D.Yu. Murzin, Catalytic deoxygenation of unsaturated renewable feedstocks for production of diesel fuel hydrocarbons, Fuel 87 (2008) 933–945. [21] D. Kubicka, J. Horacek, Deactivation of HDS catalysts in deoxygenation of vegetable oils, Applied Catalysis A: General 394 (1–2) (2011) 9–17. [22] M. Snare, I. Kubickova, P. Maki-Arvela, K. Eranen, J. Warna, D.Yu. Murzin, Production of diesel fuel from renewable feeds: kinetics of ethyl stearate decarboxylation, Chemical Engineering Journal 134 (2007) 29–34. [23] R. Badilla-Ohlbaum, David Chadwick, D.G. Gavin, Activity of NiMo catalysts for hydroprocessing coal-derived liquids, Fuel 60 (5) (1981) 452–453. [24] R.G. Graham, et al., Commercial-scale rapid thermal processing of biomass, Biomass and Bioenergy 7 (1–6) (1994) 251–258. [25] F.X. Long, B.S. Gevert, Modeling initial decay of hydrodemetallization catalyst with simultaneous adsorption and reaction mechanism, Journal of Catalysis 222 (1) (2004) 1–5. [26] B. Donnis, R.G. Egeberg, P. Blom, K.G. Knudsen, Hydroprocessing of bio-oils and oxygenates to hydrocarbons. Understanding the reaction routes, Topics in Catalysis 52 (2009) 229–240. [27] Iva Kubickova, M.S., Kari Eranen, Paivi Maki-Arvela, Dmitry Yu Murzin, Hydrocarbons for diesel fuel via decarboxylation of vegetable oils, Catalysis Today 106 (2005) 197–200. [28] Raphael O. Idem, Sai P.R. Katikaneni, Narendra N. Bakhshi, Catalytic conversion of canola oil to fuels and chemicals: roles of catalyst acidity, basicity and shape selectivity on product distribution, Fuel Processing Technology 51 (1997) 101–125. [29] A. Osmont, L. Catoire, P. Dagaut, Thermodynamic data for the modeling of the thermal decomposition of biodiesel. 1. Saturated and monounsaturated FAMEs, The Journal of Physical Chemistry. A 114 (11) (2009) 3788–3795. [30] O.I. Senol, E.M.R., T.R. Viljava, A.O.I. Krause, Effect of hydrogen sulphide on the hydrodeoxygenation of aromatic and aliphatic oxygenates on sulphided catalysts, Journal of Molecular Catalysis A: Chemical 277 (2007) 107–112. [31] O.I. Senol, T.R. Viljava, A.O.I. Krause, Hydrodeoxygenation of methyl esters on sulphided NiMo/γ-Al2O3 and CoMo/γ-Al2O3 catalysts, Catalysis Today 100 (2005) 331–335. [32] P.L. Hanks, W.E. Lewis, K.Y. Cole, Process for Preparing Diesel Fuels Using Vegetable Oils or Fatty Acid Derivatives, in: A.C.Society (Ed.), 2010, p. 30, (USA).
Contents lists available at ScienceDirect
Fuel Processing Technology journal homepage: www.elsevier.com/locate/fuproc
Hydroprocessing of fatty acid methyl ester containing resin acids blended with gas oil Shanmugam Palanisamy, Börje S. Gevert ⁎ Chalmers University of Technology, Göteborg, Sweden
a r t i c l e
i n f o
Article history: Received 14 March 2013 Received in revised form 13 May 2014 Accepted 19 May 2014 Available online xxxx Keywords: Hydrodeoxygenation Decarboxylation FAME Fatty acids Abietic acid Resin acids
a b s t r a c t Fatty acid methyl ester (FAME) with resin acids was blended with light gas oil (LGO) from 10 to 30 wt.% and then investigated using elevated temperature (300 to 370 °C) at 5 MPa on NiMoS/alumina in a trickle-bed reactor (TBR). Hydroprocessing of blended LGO showed lower S content (less than 8 ppm) at 100% conversion and reached the target cloud point on middle distillate diesel fuel. In this study a significant change in the physical and chemical properties of LGO was observed. For the distillate product, viscosity, distillation curve, nitrogen content and density were all in the expected standard grades of diesel. Here, a presence of sulfur in LGO significantly reduced catalyst deactivation. Higher concentrations of aromatic content and polycyclic aromatic hydrocarbons degraded to reach the EU diesel standard; in particular, the resin content in FAME could enable up to 50% aromatization. Elevated experimental temperatures increased decarboxylation as compared with deoxygenation for all compositions of FAME and resin acids in this upgrading process. Carbonyl compounds like CO and CO2 formation in lighter products result from resin acids decarbonylation and decarboxylation reaction pathway. © 2014 Published by Elsevier B.V.
1. Introduction The on-road legislation by EU member states is in force from 2012 for gas oil's and diesel's sulfur not to exceed 10 ppmw. A large number of studies have investigated desulfurization of middle distillates to adopt in EU member states refineries. It achieved an impressive milestone on commercialization of the hydrodesulfurization (HDS) process, particularly for middle distillate in refineries. These middle distillate fuels were vital for the transportation sectors [1–4]. One of the middle distillates – light gas oil (LGO) – had a high value because of properties and energy content that were dependent on natural and fuel adaptability. The cetane number was the main property that needed to be considered in the qualitative analysis to substitute renewable fuel with LGO [5–8]. The current standard cetane index of EU is at the range of 46 to 51 according to EN 596 standards. Sometimes, the premium diesel exceeds as high as 60. Addition of carbon rich substitutes like FAME increases cetane number. An increase in the cetane number of LGO resulted in a lower of shorter ignition delay, which tended to lessen combustion noise, leading to increased energy
⁎ Corresponding author at: Ecotraffic ERD3, Kempross AB, Larsereds lyckor 14, 42539 Hisings kärra, Sweden. Tel.: +46 702195834. E-mail address: gevert@ecotraffic.se (B.S. Gevert).
http://dx.doi.org/10.1016/j.fuproc.2014.05.015 0378-3820/© 2014 Published by Elsevier B.V.
efficiency and power output for power engines [9]. However, an increase in cetane number proved to be non-friendly for easier starting in severe winter weather and reduced both smoke and odor, which is in need of upgrading further [9–11]. Although kerosene and turbine fuels have similar properties (such as LGO), they had different specifications corresponding to their intended use. LGO mostly had paraffin and olefin type hydrocarbons with a boiling range between 125 and 300 °C, sometimes mixed with various cracked distillate fractions to increase volume requirement [12–14]. An increase in octadecane (C18) and heptadecane (C17) in production by hydrodeoxygenation (HDO) of fatty acid methyl ester (FAME) in LGO gives a better cetane number for diesel fuel compared to if no n-C17 and n-C18 was added [15–17]. Presently, combining HDO and HDS in ultra-low sulfur diesel production is recognized in some refineries through NiMoS catalyst [4,18]. Resin acids (e.g., abietic acids) present in tall oil were the main renewable sources. These acids have a complication of tri-ring structure making hydrogenation difficult in downstream products [7,19]. A primary aim of research has been to understand the fundamental aspects of the NiMoS catalytic reaction [18,20–23], develop/design a suitable condition and demonstrate the physico-chemical properties of deoxygenation products rich in carbon content [7,24]. The main objective of this study was to examine the effect of hydroprocessing FAME with resin acids (FAME-resin) blended with LGO (referred to as middle distillates). In this study, LGO was blended with FAME-resin and hydrogenated on a NiMoS catalyst for temperatures ranging between 300 and 370 °C at 5 MPa and H2/feed (mln/mln) = 80.
436
S. Palanisamy, B.S. Gevert / Fuel Processing Technology 126 (2014) 435–440
2. Experimental
temperature for removal of water and lighter hydrocarbons from the downstream product at a cut point temperature of 162 °C. The residue was analyzed to check for essential features of physicochemical properties.
2.1. Feed material The feed contained 78% FAME and 20% resin acids of the carboxylic group on tri-ring cyclic hydrocarbon with the rest as neutral fractions. The classified fatty group with resin acids is shown in Table 1. The feed was FAME-resin blended with LGO with three ratios: 10, 20 and 30 wt.%. The properties of the LGO were aromatic content 17.8 (v/v %), sulfur (S) content 295 (ppmw) and paraffin fraction from C8 to C16. Table 1 provides additional information about the LGO. 2.2. Reactor system A fixed trickle-bed and semi-batch reactor with a sulfidized NiMo/ alumina catalyst was investigated. The major reaction parameters included liquid hourly space velocity (LHSV = mln of liquid/ml of cat. ∗ h), feed mixture, hydrogen flow and reaction temperature (T) (°C). For the investigation, the commercial catalyst Trilobe HDN-60 (NiMo/ γ-Al2O3) from Criterion Catalysts was used and consisted of NiMo/γAl2O3 with 2.5–3% Ni and 12.5–13.5% Mo with 83–85% γ-Al2O3. The catalyst had the shape of 1/32-in extrudes with a surface area of 156 m2/g and a pore size of 7–8 nm calcined at about 400 °C for overnight: 31 ml (27.43 g) was loaded in the center of trickle-bed reactor (TBR) for insitu sulfidation. The catalyst was heated to 400 °C at the rate of 10 °C per minute and the temperature maintained for 2 h under nitrogen flow of 100 mln per minute before sulfidation. Later, sulfidation was achieved by allowing H2S as 10% by volume of S in H2 at 200 mln per minute for 3 h and kept overnight under a small nitrogen flow. At 360 °C and 5 MPa, LGO was flooded for 10 days, until stabilized coke developed over catalyst and steady state was achieved for further experimentation. The reactor set-up consisted of a feed tank, Dossier pump, TBR, separator tank, gas meter and collector. The down flow TBR, insulated with three independent heating zones, had an internal diameter 1.8 cm, with borosilicate glass pellets of 2 mm in diameter filled in void space. The feed that was fed by the reciprocal dossier pump was mixed with H2 or N2 gas before being allowed on a co-current flow into the reactor: all products were withdrawn periodically from the separator tank [25]. The experiment used a semi-batch reactor made up of stainless steel with 300 ml capacity that was connected to a gas inlet and outlet. The thermo couple, heating system and magnetic stirrer was integrated in an eight microprocessor-based process controller to maintain the operating condition steadily. The experiments were carried out with the same stirring speed. In this reactor, 5 wt.% abietic acid in hexadecane was used at temperatures of 300 to 350 °C and 5 MPa experimented on a NiMoS/alumina catalyst. The reactor was loaded with 150 ml of solution that includes 1 g of a defined amount of NiMoS catalyst and 3 g of tetradecane as internal standard. The product collected from the reactor was placed in the distillation re-boiler on N2 flow at 100 mln per minute. The simple distillation column consists of 2 L capacity re-boiler and heated to a desired
2.3. Analysis Simulated distillation [SIM-DIST] (ASTM D2887) was performed on the upgraded product by gas chromatograph (GC) technique. Data computation was performed with a Varian 4270 integrator. The flue gas from the reactor was collected in a gas bottle at ambient condition and subsequently analyzed using the Clarus 500 online GC. This GC has an inlet and outlet sampling values with a thermal conductivity detector (TCD) and flame ionization detector (FID), as well as four valves actuated by nitrogen gas at 0.4 MPa. The detector out signal was connected with a 600 link switch controller that interprets the signal to an integrator. The detectors are TCDs for analyzing CO, CO2 and CH4, and FID for lighter hydrocarbon fractions. Helium was used as carrier gas for the TCD, which was maintained at 200 °C with oven heat-up of 40 to 60 °C at 2 °C per minute and nitrogen as carrier gas for the FID at 60 °C as oven temperature. The cloud point, aromatic content, density, nitrogen content, sulfur content, SIM-DIST and viscosity properties were analyzed from the residue, which was performed by Preem refinery, Sweden. 3. Results and discussion 3.1. Remarks In all experiments, the general hydrogenation of FAME produced normal hydrocarbons (e.g., n-C17 and n-C18) as reported in previous studies [26,27]. The ratio of liquid to gas inlet was held constant throughout the experiments with comparability of mass balance and thermodynamic calculation. Molar balances based on the gas product analyses were calculated (error elimination was less than 5% on average). The assumption regarding hydrogenation includes deoxygenation and cracking, which were summed in the result description. The recovered residue of products mainly contains hydrocarbons between octane and octadecane with traces of high hydrocarbons. Moisture content in the products was excluded. 3.2. Aromatic composition The feed of three compositions were tested at elevated temperatures and space velocity in the TBR. The aromatic content in resin acid was done in order to study the intra-couple of deoxygenation and de-aromatization process in ensuring the product property. LGO contains 17.8 vol.% aromatics by the addition of FAME-resin aromatic compositions and retained at 17.7 vol.% for all three feed types. The upgraded products have aromatic content between 17 and 18 vol.% (see Table 2) for different blended feed compositions. This observation led us to understand that there was neither notable aromatization nor cyclic scission. However, reaction mechanism
Table 1 Classified composition of resin acids with fatty acid methyl esters supplied by Preem AB. Components
wt%
Components
wt%
Components
wt%
Octadecadienoic acid C18:2 Octadecenoic acid C18:1 Octadecanoic acid C18:0 Methyl palmitate Methyl linolenate Methyl linoleate Methyl oleate Other esters Total free fatty acids and esters
1.25 0.80 0.28 2.82 6.77 35.75 16.26 14.05 78.41
Isopimaric acid Levopimaric acid Palustric acid Neoabietic acid Dehydroabietic acid Sandaracopimaric acid Pimaric acid Abietic acid Total resin acids
1.37 0.21 1.32 0.34 4.45 0.42 2.40 8.69 19.29
β-Sitosterol Squalene Other
0.16 0.34 1.79
Total neutrals
2.29
S. Palanisamy, B.S. Gevert / Fuel Processing Technology 126 (2014) 435–440
437
Table 2 Physico-chemical properties of LGO and different blend products obtained by HDS on Ni–Mo/alumina catalyst at 5 MPa pressure. Parameters
FAME-resin + LGO at To = 350 °C LGO
Compositions Aromatic (vol%) PAH (vol%) ≤C8 (wt %) C9–C16 ΣC17 ΣC18 Heavy fraction
17.8 0.64 1 99
Distillation temperature V/V rec. (°C) (SIM-DIST) IBP 204 10%. 208 50% 215 65% 235 90% 265 95% 283 FBP Others properties Cloud point S in feed (mg/kg) S in the product (mg/kg) Density at 15 °C (kg/m3) Viscosity at 40 °C (mm2/s)
−40 295 819 1907
To (°C) at 30 wt.% FAME-resin + LGO 30%
20%
17 ** 1.0 90.1 4.0 4.5 0.4
16.9 ** 2.0 80.8 7.6 8.3 1.3
213 220 237 244 265 284
208 217 233 243 269 288
205 218 246 258 297 308
211 220 246 258 299 312 332
209 220 246 258 298 311 327
202 220 246 258 296 309 325
206 217 244 256 294 307 324
−18 276 6 821 1928
−10 236 4 820 2031
−6 207 2 821 2179
−5 207 b1 821 2220
−5 207 5 821 2212
−6 207 1 821 2165
−9 207 1 822 2105
17.7 0.63 2.3 75.2 10.1 10.2 2.2
330 17.6 0.45 0 78.4 8.4 10.7 2.5
340 17.8 0.53 1.6 75.2 9.3 11.6 2.3
360
Reference test method
10%
18.3 0.73 2.5 75.9 10.1 9.6 1.9
370 19.5 0.91 2.9 75.4 10.3 9.1 2.3
SS 155116:1993
ASTM D86 ASTM D86 ASTM D86 ASTM D86 ASTM D86 ASTM D86 ASTM D86
ASTM D2500-05 ASTM D5453-00 ASTM D5453-00 ASTM D4052-09 ASTM D7042-04
Note: ** not available.
between mono- and poly-aromatic hydrocarbons (PAH) of resin acids in LGO is not briefly estimated, because of deducible complication in combined hydroprocessing, cracking, hydrogenation and (de) aromatization. Resin de-aromatization on the resin group was tested in the semibatch reactor with a model compound of 5 wt.% abietic acid in solvent and 1 g of NiMo/Al2O3 catalyst, as shown in Table 3. The final product contains a total aromatic content of 2.1 and 2.4 wt.% for 3-hour residence time at 340 and 350 °C, respectively. Table 3 shows that 2 wt.% of aromatic content was mono group with 0.3 wt.% di-aromatic and 0.1 wt.% PAH, with 100% conversion of abietic acids. Above values estimated a 50% of the resin acid that favored dehydrogenation or aromatization in the above-mentioned condition. The C\C scission of α-bond scission in abietic acid (for reference in Fig. 5) results to form aromatics by dehydrogenation. The lack of H2 donor on catalytic surface tends to Table 3 Analysis of abietic acid for different temperatures at 5 MPa pressure for residence time = 3 h in batch reactor. Parameters
T (°C) 320
340
Gas composition (vol%) H2 CO2 CO C1 C2/C= 2 C3/C= 3 C6/C5/C4
96.30 0.71 2.33 0.48 0.07 0.08 0.02
95.88 0.52 2.25 0.95 0.13 0.24 0.02
Liquid paraffins (wt%) b320 320–360 360b
45 43 10
88 12 0
Properties Aromatics (wt%) Mono aromatics (wt%) Di-aromatic content (wt%) Poly-aromatic content (wt%) Density at 15 °C (kg/m3)
** ** ** ** **
Note: ** not available.
2.1 1.9 0.1 0.09 780.9
350 93.26 0.66 2.4 1.14 1.5 0.07 0.99
100 0 0
2.4 2 0.3 0.17 781.6
increase re-distribution of H through cyclic group. The Π-bond interaction between the benzene group and catalytic surface induces the H2 re-distribution on vacant sites at equilibrium condition. Thus, the high aromatic content formation indicates de-hydrogenation and alkylbenzene formation was appropriate in this defined temperature region. At 30 wt.% FAME-resin, total PAH increased by 0.46 vol.% subject to temperature (Table 2). Total aromatic content increased from 17.6 to 19.5 vol.% with a temperature between 330 and 370 °C. Observation of lower aromatic composition at the lower temperature (i.e. 330 °C) corresponds with the theory that the equilibrium condition of aromatization has not been attained. Both the catalytic and thermal effects had a tendency to form cyclization or aromatization, implying a more appropriate condition [28]. Metal site reactivity of aromatic formation influenced the reduction of S inhibition. The thermodynamic dependency of aromatic compounds was identified as an increase in aromatic concentration with increased temperature by attaining an equilibrium condition. However, as seen in Table 2, different compositions of FAME-resin had little effect on aromatic content compared with temperature. As a result, resin aromatic content did not change under all experimental conditions. It clarified that the resin group did not induce any dehydrogenation or polymerization, which neglected the hydrogen re-distribution or simultaneous hydrogen source availability for the deoxygenation process. Dehydrogenation induces PAH formation in LGO and FAME fractions which are free radicals formed by β-H scission over alkane chain of alkyl-benzene with respect to H2 deficient due to mass transfer resistance [29]. 3.3. Upgraded LGO properties In SIM-DIST analyses, it was observed that the final boiling point (FBP) decreased gradually from 332 to 324 °C with an increase in reaction temperature (Table 2). However, decomposition was unavoidable at higher temperatures. A cracking process, from higher hydrocarbon to lighter hydrocarbon, was clearly observed in all tested condition with different compositions. So, cloud point analyses focused without lighter hydrocarbon by removing boiling point range hydrocarbons not exceeding 165 °C. Cloud point changed from − 5 to − 9 °C with the tested elevated temperature. Further, the presence of olefins in FAME and LGO was found to influence cloud point to some extent.
Variation in paraffin and olefin composition depends on partial pressure and H-donor. Here, in our tested conditions, hydrogenation favors to form paraffin [20]. So this condition emphasizes that olefin is limited in cloud point variation. Further, it clears that isomeric products formed from resin acids had direct influence in cloud point. Sulfur content in the product was found to be lower with increased FAME-resin in the feed. The product had below 8, 5 and 2 ppmw S for 10, 20 and 30% blend FAME-resin with respective S in feed contains 276, 236 and 207 ppmw for all tested conditions. However, estimation to less than 1 ppmw of S concentration can be achieved by upgrading through a maximal of 200 ppmw S in feed, as shown in Table 2 in the comparison of feed and the product S. Instead of bypassing S in the product, higher residence time of feed or lower concentration of S in feed has a tendency to remove as H2S. Lower Fermi bond energy of catalytic active sites induces higher activity at the surface [30]. The inhibition effect of H2S discussed in some investigations [30] suggests that there was a vacancy by naked MoS2 at the corner of the active sites induced to increase hydrogenation through Π-bonding. The density of the product related to all the experiments was between 819 and 821 kg/m3. The viscosity was lower for elevated temperatures and increased with an increase in the blend ratio. Thus, relative value of viscosity raises 100 mm2/s for every 10% increase in FAME-resin composition in LGO. 3.4. Deoxygenation/decarboxylation Decomposition can be directly observed with a decrease in the C18/ C17 ratio, where the existence of a decarboxylation/decarbonylation [27] effect was evidently demonstrated (Fig. 1). Senol et al. [30,31] suggest that both deoxygenation and decomposition have a conversion ratio of 1:1. Our experiments showed deoxygenation was higher and thus more active on the carboxylic acid group to form C18 compounds by excess inhibition of S and H2. The conversion for 30% FAME-resin has between 8.4 and 10.7 wt.% in the product of C18 and C17 compositions, respectively, and heavy fraction (above C18) has 2.5 wt.% relative to temperature (see Table 2). Further, middle distillates (C6–C16) had a composition of about 75 wt.% with the rest as lighter hydrocarbons. As depicted in Fig. 1, selectivity determination indicated that the ratio of deoxygenation/decarboxylation reduced from 1.3 to 0.9 with increased temperature (from 330 to 370 °C). In the case of 10 and 20 wt.% FAMEresin, C18/C17 remains at 1.3 to 1.1. At lower temperatures, high selectivity of C18 in the product denoted the better selectivity from the reaction path. This C18 was mainly achieved through direct hydrogenation and minimized the decomposition effect. Direct hydrogenation depends on two criteria: activity with the Lewis sites at the catalyst [32] and a
2,5
100
2,0
80
1,5
60
1,0
40
0,5
20
0,0 320
340
350
360
370
0 380
To (oC) Fig. 2. CO, CO2, H2 and light hydrocarbons concentration in gas outlet with various temperature for 30% blend in LGO, Po = 5 MPa and LHSV = 1. Here, ●: CO2, ■: CO, ▲: CH4, ♦: H2 and ☼: C1–C6 (flue gas) in gas phase.
temperature dependent C\C bond cleavage. As temperatures rise, a gradual decrease in C18 and an increase in C17 in the product drive a kinetically mobilized rate of the decomposition pathway. Thus, an increase in temperature obviously influences the thermal decomposition through dissociation of the carboxylic group as CO and CO2. This lower decomposition finalized that the rate of hydrogenation was high on the hydroxyl group of acid (\OH bond), followed by ester formation, resulting in the aldehyde group (C_OH), i.e. (CO)\OH N (CO)\O\C N HC_O. However, the source of the CO2 and CO molecules in lighter gas varied with bond cleavage in H3C\O\(CO)\C\, which rose up to 0.5 vol.% as the total in flue gas. The carbonyl gas observation for both elevated temperature and FAME-resin compositions is displayed in Figs. 2 and 3. The main compositions of flue gas consisted of CO, CO2, CH4 and a small fraction of lighter hydrocarbon with the rest containing H2. Decarboxylation was found constant with the same fraction of CO2, which was notable decomposition on the carboxylic group at resin acid (Figs. 2 and 3). This finding demonstrated that CO2 formation from abietic acid observed the same fraction for all tested temperatures (Table 3). Even longer residence of abietic acid resulted in the formation of higher distillates by a preliminary path of deoxygenation and decarboxylation (Fig. 4). The CO fraction increases with an increase in 100
1,8 1,6
10% FAME-resin 20% FAME-resin 30% FAME-resin
Gas conc. (vol%)
1,3
80
1,4
1,2
1,1
1,0
1,2
60 1,0 0,8
40 0,6 0,4
Gas conc. (vol%)
1,4
C18/C17
330
Gas conc. (vol%)
S. Palanisamy, B.S. Gevert / Fuel Processing Technology 126 (2014) 435–440
Gas conc. (vol%)
438
20
0,2
0,9 0
0,0 10% blend
0,8 330
340
350
360
370
Temp. (oC) Fig. 1. C18/C17 vs. temperature (°C) of different blended LGO at LHSV = 1 and P = 5 MPa.
20% blend
30% blend
tall oil blend with gas oil Fig. 3. CO, CO2, H2 and light hydrocarbons concentration in gas outlet for different blends in LGO, Po = 5 MPa, To = 350 °C and LHSV = 1. Here, ●: CO2, ■: CO, ▲: CH4, ♦: H2 and ☼: C1–C6 (flue gas) in gas phase.
S. Palanisamy, B.S. Gevert / Fuel Processing Technology 126 (2014) 435–440
80
decarbonylation has noted interest about the properties of the product. Result in Fig. 4 shows that the cyclic group's presence at a SIMDIST ranges of 320–360 °C increases steadily. Abietic acid (N 360 °C) decreases drastically at shorter residence time and decreases steadily afterwards. Mono-/di-cyclic or iso-alkanes are the end product of the resin group with substantial removed oxygen in \COOH to form B and C products (see in Fig. 5). The resin groups deoxygenate to form a cyclic group and hydrogenation of unsaturated ring followed (Fig. 5). The scission of iso-alkanes from cyclic group results in lighter gas products. So, no changes in concentration of products after 15 min residence time, have noticed at SIM-DIST range below 320 °C. Thus indication of SIM-DIST range of 320–360 °C of middle distillates exponentially increases for residence times in Fig. 4, explains cracking of cyclic group and reformulation of aromatic compounds like products D, E, F and G in Fig. 5.
70
Distillate boiling point
Concentration (wt%)
60
below 320 oC 320 to360 oC above 360 oC
50 40 30 20 10 0 0
30
439
60
90
120
4. Conclusion
Time (min) Fig. 4. Abietic acid hydrogenation in batch reactor at T = 320 °C and P = 5 MPa on NiMoS/alumina.
Aromatic content showed a steep increase on upgraded products at elevated temperatures. Furthermore, it is clear from the data that the resin group had no influence on this increase. Paraffins to aromatic in LGO were found to significantly increase and the PAH concentration had a particularly strong influence on the quality of LGO. Upgrading resin acids and FAME increased the cetane number of LGOs, but domination of decarboxylation and decarbonylation is needed to be controlled in elevated conditions. Upgraded LGO with FAME and resin acids can be processed to obtain b1 ppmw of S. From this experiment, it could be concluded that deoxygenation and decarbonylation had domination on resin acids. However, the PAH resin structure led to mono-aromatic by cyclic scission to some extent. Dehydrogenation, isomerization and polymerization are limited or not existing with the used conditions in the process.
temperature through the influence of the decarbonylation path. This event occurs because of the tendency of a reaction path on dehydration by the formation of aldehyde from ester, which led to decomposition with a loss of 1 mol of carbon and 2 mol of oxygen. Dependency of CO and CO2 fraction with different FAME-resin compositions had a notably linear increase. The estimation of the CO and CO2 in this experiment showed that deoxygenation and decarbonylation influenced the reaction mechanisms. The CH4 formed in flue gas of 1 to 2.5 vol.% consisted of different paths: the large fraction of CH4 gas mainly evolved from methyl group in FAME (Fig. 3) and methanation process by converting CO to CH4 due to the presence of excess hydrogen favored to increase methane in lighter gas. Other sources by cracking which influenced by temperature tend to raise CH4 and other lighter hydrocarbons (C2–C6) (Fig. 2).
References [1] D. Mohan, C. Pittman, P. Steele, Pyrolysis of wood/biomass for bio-oil: a critical review, Energy & Fuels 20 (2006) 848–889. [2] S. Czernik, A.V. Bridgwater, Overview of applications of biomass fast pyrolysis oil, Energy & Fuels 18 (2) (2004) 590–598. [3] M. Lapuerta, O. Armas, J. Rodríguez-Fernández, Effect of biodiesel fuels on diesel engine emissions, Progress in Energy and Combustion Science 34 (2) (2008) 198–223. [4] Teh C. H, Deep HDS of diesel fuel: chemistry and catalysis, Catalysis Today 98 (1–2) (2004) 3–18.
3.5. Reaction mechanism of resin acid The main reaction path for deoxygenation, decarbonylation and decarboxylation in FAME hydroprocessing has been discussed previously in Section 3.5. The model resin acids (abietic acid) deoxygenation and
-CO/CO2
A
F a
a
B 2H2
O
HO
Mono -cyclic Di-cyclic group n-Alkane n-Alkenes
2H2
-H2O
G
2H2 2H2
a
C D
E
Fig. 5. Abietic acid hydrogenation mechanisms. (A = abietic acid, B = decarboxylic/decarbonylic path and C = deoxygenate path and F, G, E = (de)aromatization), (note: a = α-bond).
440
S. Palanisamy, B.S. Gevert / Fuel Processing Technology 126 (2014) 435–440
[5] A. Demirbas, Progress and recent trends in biodiesel fuels, Energy and Combustion Science 33 (2007) 1–18. [6] J.T.G. Monnier, D.W. Soveran, A. Wong, E.N. Hogan, M. Stumborg, USA 5705722, Natural Resources Canada, 1998. [7] Robert Coll, Siddharth Udas, William A. Jacoby, Conversion of the rosin acid fraction of crude tall oil into fuels and chemicals, Energy and Fuels 15 (2001) 1166–1172. [8] T. Laza, A. Bereczky, Basic fuel properties of rapeseed oil-higher alcohols blends, Fuel 90 (2) (2011) 803–810. [9] P. Ghosh, Predicting the effect of cetane improvers on diesel fuels, Energy & Fuels 22 (2) (2008) 1073–1079. [10] K. Hashimoto, M. Ikeda, M. Arai, M. Tamura, Cetane number improvement of diesel fuel by autoxidation, Energy & Fuels 10 (6) (1996) 1147–1149. [11] A. Demirbas, Relationships derived from physical properties of vegetable oil and biodiesel fuels, Fuel 87 (2008) 1743–1748. [12] T. Marker, J.P., T. Kalnes, M. McCall, D. Mackowiak, B. Jerosky, B. Regan, L. Nemeth, M. Krawczyk, S. Czernik, D. Elliot, D. Shonnard, Opportunities for biorenewables in oil refineries, http://www.osti.gov/energycitations/product.biblio.jsp?osti_id=861458 2005 UOP, Des Plaines, Illinois. [13] V.A. Yakovlev, S.A. Khromova, O.V. Sherstyuk, V.O. Dundich, D.Yu. Ermakov, V.M. Novopashina, M.Yu. Lebedev, O. Bulavchenko, V.N. Parmon, Development of new catalytic systems for upgraded bio-fuels production from bio-crude-oil and biodiesel, Catalysis Today 144 (2009) 362–366. [14] J.S.J. Alonso, J.A.L. Sastre, C. Romero-Avila, E. Lopez, A note on the combustion of blends of diesel and soya, sunflower and rapeseed vegetable oils in a light boiler, Biomass and Bioenergy 32 (9) (2008) 880–886. [15] Paivi Maki-Arvela, Iva Kubickova, Mathias Snare, Kari Eranen, Dmitry Yu Murzin, Catalytic deoxygenation of fatty acids and their derivatives, Energy & Fuels 21 (2007) 30–41. [16] K. Breda, Biodiesel usage at low temperature, Fuel 87 (7) (2008) 1306–1317. [17] X. Chen, L. Yongbin, Qun Shao, Study on cold flow properties of biodiesel from typical materials, Shiyou Yu Tianranqi Huagong 39 (6) (2010) 491–493 497. [18] H. Topsoe, B. Hinnemann, J.K. Norskov, J.V. Lauritsen, F. Besenbacher, P.L. Hansen, G. Hytoft, R.G. Egeberg, K.G. Knudsen, The role of reaction pathways and support interactions in the development of high activity hydrotreating catalysts, Catalysis Today 107–108 (2005) 12–22. [19] F. Bauer, J.C., P. Johansson, LindenJ. , M. Nyström, L. Ruden, Tallolja som biobaserad dieselersättning — Katalytisk hydrogenering av hartssyror. KBTX01-10-6. BSc, Chalmers University Of Technology, Göteborg, Sweden, 2010.
[20] M. Snare, I. Kubickova, P. Maki-Arvela, D. Chichova, K. Eranen, D.Yu. Murzin, Catalytic deoxygenation of unsaturated renewable feedstocks for production of diesel fuel hydrocarbons, Fuel 87 (2008) 933–945. [21] D. Kubicka, J. Horacek, Deactivation of HDS catalysts in deoxygenation of vegetable oils, Applied Catalysis A: General 394 (1–2) (2011) 9–17. [22] M. Snare, I. Kubickova, P. Maki-Arvela, K. Eranen, J. Warna, D.Yu. Murzin, Production of diesel fuel from renewable feeds: kinetics of ethyl stearate decarboxylation, Chemical Engineering Journal 134 (2007) 29–34. [23] R. Badilla-Ohlbaum, David Chadwick, D.G. Gavin, Activity of NiMo catalysts for hydroprocessing coal-derived liquids, Fuel 60 (5) (1981) 452–453. [24] R.G. Graham, et al., Commercial-scale rapid thermal processing of biomass, Biomass and Bioenergy 7 (1–6) (1994) 251–258. [25] F.X. Long, B.S. Gevert, Modeling initial decay of hydrodemetallization catalyst with simultaneous adsorption and reaction mechanism, Journal of Catalysis 222 (1) (2004) 1–5. [26] B. Donnis, R.G. Egeberg, P. Blom, K.G. Knudsen, Hydroprocessing of bio-oils and oxygenates to hydrocarbons. Understanding the reaction routes, Topics in Catalysis 52 (2009) 229–240. [27] Iva Kubickova, M.S., Kari Eranen, Paivi Maki-Arvela, Dmitry Yu Murzin, Hydrocarbons for diesel fuel via decarboxylation of vegetable oils, Catalysis Today 106 (2005) 197–200. [28] Raphael O. Idem, Sai P.R. Katikaneni, Narendra N. Bakhshi, Catalytic conversion of canola oil to fuels and chemicals: roles of catalyst acidity, basicity and shape selectivity on product distribution, Fuel Processing Technology 51 (1997) 101–125. [29] A. Osmont, L. Catoire, P. Dagaut, Thermodynamic data for the modeling of the thermal decomposition of biodiesel. 1. Saturated and monounsaturated FAMEs, The Journal of Physical Chemistry. A 114 (11) (2009) 3788–3795. [30] O.I. Senol, E.M.R., T.R. Viljava, A.O.I. Krause, Effect of hydrogen sulphide on the hydrodeoxygenation of aromatic and aliphatic oxygenates on sulphided catalysts, Journal of Molecular Catalysis A: Chemical 277 (2007) 107–112. [31] O.I. Senol, T.R. Viljava, A.O.I. Krause, Hydrodeoxygenation of methyl esters on sulphided NiMo/γ-Al2O3 and CoMo/γ-Al2O3 catalysts, Catalysis Today 100 (2005) 331–335. [32] P.L. Hanks, W.E. Lewis, K.Y. Cole, Process for Preparing Diesel Fuels Using Vegetable Oils or Fatty Acid Derivatives, in: A.C.Society (Ed.), 2010, p. 30, (USA).