1. Introduction
High-quality biofuels, in particular, synthetic bio-jet fuel and synthetic biodiesel, can be produced by hydrotreating feedstocks containing triglycerides of higher fatty acids [
1,
2]. These raw materials include various vegetable oils, waste cooking oils and animal fats. Hydrotreating products are often referred to as hydrotreated vegetable oil (HVO) or green diesel in the case of diesel fuel or as hydroprocessed esters and fatty acids (HEFA) in the case of bio-jet fuel. These products mainly contain paraffinic hydrocarbons and almost no aromatics and sulphur and have excellent thermal oxidation stability and a high cetane number [
1,
3].
The most common renewable feedstock materials for the HVO production contain triglycerides formed by fatty acids with 14–22 carbon atoms, and maximally, three double bonds. The first step of the triglyceride hydrotreating is the hydrogenation of the double bonds present in the aliphatic chains of the acyl groups. The next step involves the hydrogenolysis of the ester bonds with the formation of diglycerides, monoglycerides, and finally, carboxylic acids and propane [
4,
5,
6]. According to some authors, the intermediates then decompose into hydrocarbons via hydrodeoxygenation (HDO) and hydrodecarboxylation (HDCx) [
7,
8]. The CO
2 generated by HDCx can be partially reduced by the hydrogen to CO and water (Equation (1)). In the opinion of other authors, in addition to HDO and HDCx, hydrodecarbonylation (HDCn) is also involved in the decomposition of the carboxylic acid into hydrocarbons [
4,
5,
6].
Hydrodeoxygenation removes oxygen from the carboxylic acids in the form of water, and n-alkanes with the same number of carbon atoms as the starting fatty acids. Hydrodecarboxylation removes oxygen in the form of CO
2, while hydrodecarbonylation removes oxygen in the form of CO and water. In both cases, the produced n-alkanes have one carbon atom less than the original carboxylic acids [
4,
5,
6].
The molar ratio of the C
17/C
18 n-alkanes in the liquid products is usually used to determine the dominant pathway of the triglyceride conversion, i.e., whether HDO or HDCx/HDCn predominates. The reaction pathways of the triglyceride conversion are influenced by the reaction conditions. Under high hydrogen pressure, HDO occurs at a higher extent than HDCx/HDCn [
9]. The CO
2/CO ratio in the gaseous products is sometimes used to determine the selectivity of the HDCn and HDCx reactions, but it should be taken into account that methanation reaction, where the CO
2 or CO reacts with hydrogen (Equations (2) and (3), respectively) to form methane, and the reaction between the carbon dioxide and hydrogen which forms carbon monoxide and water (Equation (1)) may affect the composition of gaseous products and hydrogen consumption [
4,
5,
6].
Some oxygenates, such as carboxylic acids, alcohols, and esters, are usually formed under moderate reaction conditions when the conversion of the feedstock to hydrocarbons is not complete [
8,
10]. In some cases, hydrocarbons with a lower carbon number than that of the corresponding fatty acid are formed, due to the hydrocracking of the carbon chains of the intermediates originating from the triglyceride conversion, or due to the hydrocracking of the formed n-alkanes. The extent of hydrocracking depends on the acidity of the catalyst and the reaction conditions. Hydroisomerisation, cyclisation, and dehydrocyclisation with the formation of isoalkanes, alkyl cycloalkanes and alkyl aromatics, are also possible reactions. Their extent depends on the type of the catalyst and the reaction conditions [
1,
2]; however, n-alkanes are usually the predominant products of hydrotreating vegetable oil/animal fats. The hydroisomerisation of the n-alkanes takes place at the acidic centres of the catalysts and leads to an improvement in the low-temperature properties of the liquid product. However, the catalyst should not be too acidic because undesirable cracking reactions also take place at the acidic centres [
1,
2]. As hydroisomerisation requires lower reaction temperatures than the conversion of triglycerides to n-alkanes, it is, therefore, usually carried out in the second stage in the commercial production of HVO/HEFA [
11,
12].
A wide variety of catalysts for triacylglyceride hydrotreating have been investigated. Conventional catalysts consist of Mo and/or W with Co and Ni as promoters on various supports [
2,
13,
14,
15]. Various types of γ-alumina with different textural properties (surface area, pore volume and size) are the most common types of support. Recently, other supports, such as SiO
2, TiO
2, mixed oxides, zeolites, mesoporous supports based on SiO
2, and carbon, have also been investigated [
2,
13,
14,
15]. Conventional catalysts typically provide high yields of hydrocarbon liquid fractions boiling in the diesel fuel range, but their low-temperature properties (cloud point and CFPP—cold filter plugging point) are often poor. These properties can be improved either by the additional processing on a hydroisomerisation catalyst or by using multifunctional catalysts directly for the hydrotreating [
2,
13,
14,
15].
Unsupported transition metal sulphide catalysts, mainly unsupported Ni/Co-Mo/W sulphide catalysts, were studied for the hydrotreating of petroleum fractions [
16]. The performance studies identified several unsupported catalysts with higher activity and/or selectivity than traditionally supported Ni-Mo/γ-Al
2O
3 and Co-Mo/γ-Al
2O
3 catalysts. Even if the unsupported catalyst contains Group VIII and Group VI metals only, they are more expensive than the supported ones, because they contain a high concentration of the metals [
16,
17]. The commercial use of the unsupported catalysts is mainly limited to NEBULA catalysts [
17].
The hydrotreating of a feedstock based on the triglycerides of fatty acids can be performed either in a separate unit or in a unit integrated into a conventional oil refinery (coprocessing or cohydrotreating) [
18,
19,
20,
21]. The commercial success of each process also depends on the development of an efficient catalyst, which must exhibit a stable activity in the presence of a relatively large amount of water that is generated as a by-product of the triglyceride conversion.
The coprocessing of vegetable oils/animal fats together with a conventional refinery feedstock provides flexibility to the refineries. However, it is not applied too often, mainly due to the inhibition of the hydrodesulphurisation (HDS) of the petroleum feedstock caused by the presence of large amounts of oxygen compounds or the inhibition of the hydrodeoxygenation (HDO) of triglycerides caused by the presence of sulphur compounds in the petroleum feedstock [
21].
The hydrogen consumption during the hydrotreating of middle distillates depends on the composition of the feedstock, in particular, on the contents of sulphur, nitrogen and polyaromatics. The presence of triglycerides in the feedstock for cohydrotreating has a strong influence on hydrogen consumption. It mainly depends on the concentration of triglycerides in the feedstock and the preferred routes of the conversion of triglycerides to alkanes. The origin and structure of the triglycerides (mainly the amount of double bonds) also plays an important role. In the case of rapeseed oil with four double bonds in the molecule, 7 moles of hydrogen are consumed if 1 mole of rapeseed oil is converted by the HDCx route, in comparison with 10 and 16 moles of hydrogen consumed during the conversion of the same 1 mole of rapeseed oil via the HDCn and HDO routes, respectively.
The hydrogen consumption for the hydrotreating of the triglyceride-based feedstock into HVO is higher than the fossil middle distillates. The hydrogen consumption determined by Al-Darous and Ali [
22] during the deep desulphurisation of gas oil (S: 1.0 wt %) over an Ni-Mo/Al
2O
3 catalyst was around 37 m
3∙m
−3. The hydrotreating of light gas oil (S: 1.27 wt %, N: 170 mg·kg
−1, polyaromatics: 12 wt %) and its mixtures with rapeseed oil (15 and 25 wt %) over an Ni-Mo/Al
2O
3 catalyst was compared by Donnis et al. [
23]. The hydrogen consumption was 47 m
3·m
−3 in the case of the pure gas oil hydrotreating and increased to 83 m
3·m
−3 and 103 m
3·m
−3 if the feedstocks contained 15 and 25 wt % of rapeseed oil, respectively. The hydrogen consumption for palm oil hydrotreating was estimated [
24] to be 210 dm
3·kg
−1. Similarly, the hydrogen consumption for various triglycerides hydrotreating of 178–234 dm
3·kg
−1 was reported by Carmona et al. [
25].
Although the investigation of the coprocessing of vegetable oils/animal fats with a conventional petroleum feedstock has already brought a number of new findings, there are still technical and operational problems, and further research is needed to solve these problems. It is necessary to investigate the compatibility of various vegetable oils and animal fats with different petroleum feedstocks, the long-term stability of the catalysts (over 1 year), the optimisation of the reaction conditions and the hydrogen consumption while maintaining the required product quality, the mutual influence of the heteroatoms contained in the petroleum feedstock (S, N), and the oxygenates derived from the biomass.
This work is focused on the investigation of the effect of the reaction conditions on the activity of a sulphided Ni-Mo/Al2O3 catalyst during the cohydrotreating of rapeseed oil and petroleum middle distillates in a 20:80 mass ratio to produce diesel fuel containing more than 10 % of biocomponent. Moreover, petroleum middle distillate that was processed contained 10 % of low-quality light cycle oil (LCO) and corresponded, thus, to a real refinery feedstock for the hydrotreating.
Not only the influence of the reaction temperature and the WHSV of the feedstock were studied, but also the influence of hydrogen to feedstock flow rate ratio, including coprocessing with lower hydrogen excess, was tested. The paper also brought a detailed overview of hydrogen consumption that is so rarely reported, especially in the field of vegetable oil coprocessing. Based on the mass flows of the individual components of the raw materials and the hydrotreating products formed from them, the approximate hydrogen consumptions for the individual reactions taking place during the hydrotreating were calculated.
3. Materials and Methods
3.1. Materials
Straight run gas-oil (SRGO) from crude oil distillation, light cycle oil (LCO) from fluid catalytic cracking and commercial rapeseed oil (RO) were used as components of the two types of feedstock in this work. The first type of feedstock (F0) was a mixture of SRGO and LCO in a mass ratio of 90:10. The second type of feedstock (F20) was composed of SRGO, LCO and RO in a mass ratio of 72:8:20. For the given SRGO and LCO quality, the usual amount of LCO in the feedstock for the production of summer diesel fuel is 10–12 wt %. The properties of both feedstocks are summarised in
Table 11.
The commercial RO contained 4 mg∙kg
−1 of sulphur and 8 mg∙kg
−1 of nitrogen. The C
18 acyl groups represented more than 91 wt % of the total acyl groups in the RO (
Table 12).
3.2. Hydrotreating
The fossil feedstock (F0) was hydrotreated at a pressure of 4.0 MPa, with a commonly used hydrogen/feedstock flow rate ratio of 240 m
3∙m
−3 (hydrogen flow rate of 28 dm
3∙h
−1), a weight hourly space velocity (WHSV) of about 1.0 h
−1, and temperatures around 320, 330, 340 and 350 °C (
Table 13).
The hydrotreating of the F20 feedstock was carried out at the same reaction pressure and similar temperatures as the F0 feedstock. The hydrogen flow rate of 28 dm
3∙h
−1 was kept constant for all the experiments. The WHSV of 0.5, 1.0, 1.5 and 2.0 h
−1 were tested, and thus, the hydrogen/feedstock flow rate ratio varied from 480 to 120 m
3∙m
−3. Sixteen liquid products from the hydrotreating of the F20 feedstock were obtained in total and are listed in
Table 14.
The hydrotreating was carried out in a tubular fixed-bed reactor with the cocurrent flow of the feedstock and hydrogen. A simplified schematic diagram of the preheater and electrically heated reactor system was described in a previous article [
31].
The total length of the reactor tube was 658 mm, and its internal diameter was 30 mm. The reactor was divided into three zones: An upper preheat zone, a catalytic bed zone, and a bottom zone below the catalytic bed. The catalytic bed zone was filled with a commercial hydrotreating Ni-Mo/Al
2O
3 catalyst with a particle size reduced to the range of 0.25–0.42 mm. The total catalyst volume in the bed was 95 cm
3, corresponding to a catalyst mass of 99 g. The catalyst was mixed with silicon carbide (particle size of 0.25–0.30 mm) in the volume ratio of 1:1. The activation, composition and properties of the catalyst were described in detail by Vozka et al. [
31].
The whole experiment lasted about 250 h. The catalyst activation lasted for about 24 h; it was followed by the stabilisation of the catalyst activity, in which the SRGO hydrotreating was performed, and the sulphur content in the liquid products was measured every 4 h. When the sulphur content of the products did not change for at least 20 h, the individual experiments were performed, each lasting about 10 h. The setting and control of the reaction conditions took about 1 h, rinsing the reactor and separator with the reaction products took 6–8 h, the collection of the liquid product took 1.5–4.0 h (ca 200 ml of the liquid product was collected). The gaseous products were also simultaneously collected into Tedlar sampling bags with a volume of 5 dm3 during the collection of the liquid product. After the last hydrotreating conditions, the conditions from the first experiment were adjusted to verify the same catalyst activity at the beginning and the end of the experiment. Both samples showed the same level of desulphurisation.
3.3. Processing of the Liquid Products
The liquid products were processed considering their use for diesel fuel production. At first, they were purged with hydrogen with a flow rate of 0.5 dm3∙min−1 for 2 h to remove the hydrogen sulphide and ammonia (the products of refining the hetero compounds). If the hydrogen sulphide remained in the product, it would be oxidised into elemental sulphur during further processing, which is undesirable because of the evaluation of the desulphurisation efficiency.
The fraction boiling up to 150 °C was afterwards removed from the liquid products by the distillation in a Fischer HMS 500 distillation apparatus. The liquid product was first distilled at atmospheric pressure and then under a reduced pressure of 5 kPa until the temperature at the head of the column reached 64 °C, which is the boiling point corresponding to 150 °C under atmospheric pressure. The obtained stabilised liquid products were consequently analysed.
3.4. Analysis of the Liquid and Gaseous Products
The conversion of the rapeseed oil to hydrocarbons was verified by the simulated distillation of the liquid products using a TRACE GC ULTRA gas chromatograph (Thermo Fisher, Milano, Italy) according to the extended ASTM D2887 standard. The parameters of the simulated distillation are listed in
Table 15. The cetane index was calculated according to EN ISO 4264 using the density and the simulated distillation data converted to the ASTM D86 equivalent results.
The nitrogen and sulphur contents in the liquid products were determined on a Xplorer-NS (Trace Elemental Instruments, Delft, The Netherlands) according to the ASTM D4629 and the ASTM D5453 procedures, respectively. The density was measured according to EN ISO 12185 via an DMA 4000 (Anton Paar, Graz, Austria), the kinematic viscosity was measured according to ASTM D7042 via an SVM 3000 (Anton Paar, Graz, Austria) and the cold filter plugging point (CFPP) was measured according to EN 116 on a Callisto 100 (Anton Paar, Blankenfelde-Mahlow, Austria) coupled to a FL 601 cryostat (Julabo, Seelbach, Germany).
The content of the n-alkanes in the liquid products was determined using an Agilent 6890N gas chromatograph (Agilent Technologies, Santa Clara, CA, USA) equipped with a flame ionisation detector (GC-FID). The n-alkanes were identified by comparing their retention times with those of a standard mixture of the C
6-C
30 n-alkanes analysed under the same conditions as the samples. The content of each n-alkane was calculated as the relative ratio of the individual n-alkane area to the total product area. The GC-FID parameters were provided in a previous article [
32].
The group-type composition of the liquid products was determined according to the EN 12916 procedure via high-performance liquid chromatography (HPLC) with a refractometric detection and a normal phase arrangement (Shimadzu Corporation, Kyoto, Japan).
The analysis of the gaseous products was performed using an Agilent 6890N gas chromatograph (Agilent Technologies, Santa Clara, CA, USA) equipped with two detectors: A flame ionisation detector (FID) for the detection of the hydrocarbons (C
1–C
5) and a thermal conductivity detector (TCD) for the detection of the permanent gases. The effluent flowing out of the column was split into the detectors using a Y-piece Siltek MXT Connector (Restek, Bellefonte, PA, USA). The experimental parameters of the analysis are listed in
Table 16.
The results of the determination of the CO, CO2, and hydrocarbons were corrected to zero content of nitrogen and oxygen, which entered the sampling bags in small quantities during the collection of the gaseous hydrotreating products. As hydrogen provides a small response on the TCD detector if helium is used as a carrier gas, its direct determination can have a relatively large error. Therefore, the hydrogen content was calculated as the rest to 100 vol%.
3.5. Calculation of Hydrogen Consumption
The hydrogen and other compounds contained in the gaseous products can be dissolved in varying amounts in the liquid hydrotreating products. It can affect the correct determination of these compounds. Therefore, the consumption of hydrogen could not be precisely calculated from the amount and composition of the gaseous products. The consumption of the hydrogen was, thus, calculated from the mass balance and from the composition of the feedstocks and the liquid and gaseous products (excluding the hydrogen content). Moreover, the calculation made it possible to determine the approximate hydrogen consumption for the individual reactions taking place during the hydrotreating and is described in the
Supplementary Material.
4. Conclusions
This study compared the hydrotreating of the mixture of petroleum middle distillates and the same mixture containing 20 wt % of rapeseed oil (RO). Experiments were carried out in a fixed-bed reactor over a commercial Ni-Mo/γAl2O3 catalyst at a temperature range of 320–350 °C, a pressure of 4 MPa, a WHSV of 1.0 h−1 and at a ratio of hydrogen to feedstock of 240 m3·m−3.
The total conversion of the RO to hydrocarbons, CO, CO2, CH4, propane and water was achieved at all the tested reaction temperatures. The content of the saturated hydrocarbons was, therefore, higher in the hydrotreating products of the RO containing feedstock compared to the hydrotreating products of the neat petroleum feedstock. A higher reaction temperature favoured the hydrodecarbonylation and/or hydrodecarboxylation of the RO prior to its hydrodeoxygenation. As a result, the content of n-heptadecane in the liquid products increased with an increasing reaction temperature at the expense of the n-octadecane. The slight inhibition of desulphurisation, caused by the oxygen-containing compounds arising from the hydrotreating of the RO was observed at reaction temperatures of 340 and 350 °C. In the hydrotreating of the pure petroleum feedstock, hydrogen was mainly consumed for the dearomatisation of the feedstock. In the hydrotreating of the feedstock containing 20 wt % of RO, a large amount of hydrogen was consumed for the dearomatisation as well, but also for the saturation of the double bonds in the RO and its hydrodeoxygenation.
The hydrotreating of the feedstock containing 20 wt % of the RO under a variable WHSV (0.5, 1.0, 1.5 and 2.0 h−1) and a constant hydrogen flow rate of 28 dm3∙h−1 (a variable hydrogen/feedstock flow rate ratio in the range of 120–480 m3∙m−3) was further compared. At a WHSV of 0.5 h−1, the hydrogenation of the diaromatics into monoaromatics and the monoaromatics into saturated hydrocarbons reached its maximum rate, due to the considerable time for the reactions and the largest hydrogen/feedstock ratio. It was proved that for the deep desulphurisation of the feedstock to the sulphur level below 10 mg∙kg−1, a reaction temperature of at least 350 °C and a WHSV not higher than 1 h−1 are necessary. On the other hand, the highest tested WHSV of 2.0 h−1 combined with the lowest tested hydrogen/feedstock ratio of 120 m3∙m−3 were still sufficient for the complete RO conversion and to get diesel fuel with the acceptable other parameters (density, kinematic viscosity, cetane index and CFPP) required by the EN 590 specification.
The hydrogen consumption related to 1 m3 of the feedstock decreased slightly with an increasing WHSV and decreasing hydrogen to feedstock ratio, mainly due to the decreased extent of the hydrodearomatisation and methanation reactions. Depending on the increasing hydrogen consumption, the hydrogen excess decreased with an increasing WHSV.