Field of the invention

The invention relates to process for converting glycerol to propane. Further, the invention relates to a process for improving the yield of liquefied petroleum gas, and in particular propane, in a petroleum refinery.

Background

Propane, normally traded as liquefied petroleum gas or LPG, is a good energy carrier, especially in rural areas. The propane used today is typically a derivative of crude oil. Similar to natural gas, LPG thus contributes to a net emission of greenhouse gases when used. Biogas has emerged as a green alternative to natural gas. Also, for liquid hydrocarbon fuels, e.g. gasoline and diesel, there is a major interest in renewable resources decreasing the carbon footprint. Apart from biodiesel, e.g. methyl esters of fatty acids derived from tri-glycerides, also conversion of e.g. tall oil to hydrocarbons to be used on their own, or to be blended with diesel derived from crude oil, has emerged as way of reducing the carbon footprint. As an example, in US 2010/0228068 a process for production of normal alkanes by hydrotreating mixtures of triglycerides and vacuum gasoil over a catalyst is disclosed.

There is a similar desire to move towards using renewable resources also for the production of LPG or propane. Examples in the art addressing this need include processes for converting glycerol, obtained as side product in biodiesel production, to ethane, ethane, or propane (cf. WO 2010/052208 and US 2009/0054701). There is, however, much capital and infrastructure needed for setting up dedicated plants for producing propane from sources not contributing to net emissions of greenhouse gases when used, hereon forth called renewable resources.

It would thus be of interest to lower the cost for producing propane from renewable resources. Summary

An object of the invention is to alleviate, at least one of the above stated problem. Thus, there is according to a first aspect provided a process for converting glycerol to propane in a petroleum refinery.

In petroleum refineries, crude oil is fractioned and refined to eventually provide various hydrocarbon fractions, e.g. LPG, gasoline, kerosene, diesel etc. Apart from the initial fractioning of crude oil by distillation to provide petroleum refinery intermediate streams (e.g. naphtha, kerosene (middle distillate), light gas oil, heavy gas oil, or vacuum gas oil), petroleum refinery typically includes other unit operations, e.g. hydrotreatment and cracking (cf. Fig. 1 showing an exemplary flow scheme for petroleum refining).

As recognized in the art, hydrotreatment is a process to reduce sulfur, aromatics, nitrogen, and oxygen in petroleum refinery intermediate streams. In petroleum refining, hydrotreatment is used to enhance the combustion quality, density and smoke point of distillates, as it serves to reduce sulfur, aromatics, nitrogen, and oxygen. Compounds comprising sulfur, nitrogen, and oxygen and/or being aromatic are by hydrotreatment converted into more volatile compounds, e.g. hydrogen sulfide, ammonia, and water, which may be removed from the petroleum refinery intermediate stream. In hydrotreatment in a petroleum refinery, a petroleum refinery intermediate stream (e.g. gas oil (sometimes denoted atmospheric gas oil), or vacuum oil) is reacted with hydrogen over a catalyst at elevated temperature (330 to 420°C) and pressure (60 to 120 bar).

Cracking is a process in which complex organic, often aromatic, molecules, such as kerogens, or long-chain hydrocarbons are broken down into smaller and simpler molecules, such as lighter hydrocarbons. Cracking typically lowers the amount of aromatics. Further, the boiling point of long-chain hydrocarbons, by cracking them into hydrocarbons with shorter carbon chains, is lowered. In short, cracking is the breaking of carbon-carbon bonds in the starting material.

In addressing the need to lower the cost for producing propane from renewable resources, e.g. glycerol from a methyl transesterification process, it was realized that it may be possible to utilize the existing infrastructure at a petroleum refinery and whereby increasing the yield of LGP. Glycerol may be converted to propane in existing unit operations in a petroleum refinery, thereby dispensing with the need and costs for establishing a new production site. In order to efficiently convert propane to glycerol, it may be advantageous to apply some modifications to the standard unit operations in the petroleum refinery. However, there is no need for any major modifications. As an example, replacing existing catalysts does not require any major investment. Further, the catalysts are anyhow regularly replaced.

As already outlined, compounds comprising various heteroatoms (e.g. oxygen nitrogen, and sulfur), may be converted to hydrocarbons by hydrotreatment. Feeding glycerol to a hydrotreater in a petroleum refinery will also be able to reduce the glycerol to propane. It was concluded that glycerol may be hydrotreated as a mixture with a petroleum refinery intermediate stream, though some adjustments to the standard process may be required to provide an effective process. Treating, glycerol as a mixture with a petroleum refinery intermediate stream may further be advantageous, as dilution of the glycerol will attenuate the oligomerization and/or polymerization of glycerol monomers upon heating the glycerol stream, thus improving the yield of propane and lowering de-activation of the hydrotreatment catalyst

According to an embodiment, the process for converting glycerol to propane in a petroleum refinery comprises the steps of:

- mixing glycerol with a petroleum refinery intermediate stream, such as vacuum gas oil and/or gas oil (atmospheric gas oil), to provide a mixed stream comprising glycerol;

- refining the mixed stream comprising glycerol by hydrotreatment in a reactor in the petroleum refinery with hydrogen over a porous catalyst, the average pore diameter of the catalyst exceeding 60 Angstrom, wherein the hydrotreatment reducing glycerol to propane, to provide a refined, mixed stream having an increased relative content of propane; and

- separating a fraction comprising propane from the refined, mixed stream.

In order to minimize oligomerization and/or polymerization of glycerol, it is preferred if the glycerol is mixed with the petroleum refinery intermediate stream before being introduced into the reactor. However, according to a less preferred embodiment, the glycerol and the petroleum refinery intermediate stream, respectively, are fed separately to the reactor and mixed therein. In order to provide efficient hydrotreatment, it is preferred if the amount of the petroleum refinery intermediate exceeds the amount of glycerol. Thus, the glycerol may be mixed with the petroleum refinery intermediate stream in a ratio of less than 50:50 on a weight basis. The glycerol may be mixed with the petroleum refinery intermediate stream in a ratio of from 10:90 to 40:60 on a weight basis.

The catalyst may be a supported, porous heterogeneous catalyst comprising nickel, or cobalt, and molybdenum. The support may be alumina, such as delta-(6)- alumina. The catalyst may comprise 1 to 6 wt.% nickel or cobalt. Further, the catalyst may comprise 5 to 15 wt.% molybdenum. According to an embodiment, the catalyst comprises nickel and molybdenum. According to such an embodiment, the catalyst may comprise 2 to 5 wt.% nickel, and 5 to 10 wt.% molybdenum. The support may, according to such an embodiment be delta(6)-alumina. The catalyst is typically used in its sulfide form.

According to an embodiment, the support of the supported, heterogeneous catalyst is aluminum oxide, AI2O3 (alumina). The crystal structure of the alumina is preferably delta phase. Further, preferably the alumina comprises;

- at least 96 wt.% AI2O3;

- not more than 0.030 wt.% SiO;

- not more than 0.050 wt.% Fe2Cb:

- not more than 0.015 wt.% Na20;

The alumina support may be in the form of an extrudate having a diameter of

3.15 ± 0.15 mm and length of 3 to 9 mm.

According to an embodiment, the catalyst is a heterogeneous catalyst obtainable by a two-step procedure, starting from a pre-shaped, e.g. extrudated, support of porous delta-alumina. The first step is an incipient wetness impregnation, where the desired amount of ammonium molybdate is deposited on the catalyst using a water solution that is allowed to fill the catalyst pore system entirely. The second step is an incipient wetness impregnation with a water solution containing the nitrate or acetate salt of nickel or cobalt, with citric acid added up to a stoichiometric amount to the nickel or cobalt. The catalyst may then be transferred to its sulfide state by reaction with a sulfur-containing hydrocarbon or hydrogen sulfide before use.

According to another embodiment, the catalyst is a heterogeneous catalyst obtainable by procedure outlined below. In a first step, porous delta-alumina, acting as support, is impregnated with aqueous solution of ammonium molybdate to fill the catalyst pore system. Subsequently, the catalyst is dried and calcined. In a second step, the alumina support with molybdate is impregnated with an aqueous solution comprising a nitrate or acetate salt of cobalt or nickel and citric acid (added in ratio of 0.5 to 1 mol citric acid per mol cobalt or nickel) to fill the catalyst pore system.

Subsequently, the catalyst is once more dried and calcined. The provided catalyst is used in its sulfide form. The catalyst may be transferred to its sulfide state by reaction with a sulfur-containing hydrocarbon or hydrogen sulfide before use. By adding citric acid, the pH is lowered, whereby changing the adsorption properties of cobalt and nickel to enhance the activity of the final catalyst.

In order to provide a catalyst with high activity, the support is typically porous to increase the ratio of active surface area per weight unit. The alumina support, defining the structure of the catalyst, is thus porous. Further, smaller pore size does typically provide higher efficiency. However, for the present process, a catalyst with somewhat larger pores is preferred. It was found (cf. examples 4 and 5) that a conventional hydrotreatment catalysts with smaller pores (diameter of about 60

Angstrom) is by far less efficient compared to a catalyst with larger pores (e.g. diameter of about 120 Angstrom). Seemingly, the catalyst with smaller pores is more prone to deactivation resulting from carbonization, which, without being bound to any theory, may explain its lower activity. Accordingly, the average pore diameter of the catalyst thus exceeds 60 Angstrom. Preferably, the average pore diameter of the catalyst is at least 80 Angstrom, or even at least 100 Angstrom, such as about 120 Angstrom. Thus, the average pore diameter of the catalyst may be 80 to 200 Angstrom such as 100 to 140 Angstrom. The average pore diameter of the catalyst may be determined using nitrogen physisorption at liquid nitrogen temperatures and calculated using e.g. the BJH (Barret, Joyner, and Halenda) method assuming cylindrical pores (cf. Barrett EP, Joyner LG, Halenda PP (1951) The Determination of Pore Volume and Area Distributions in Porous Substances. I. Computations from Nitrogen Isotherms. J Am Chem Soc 73 (l):373-380. doi: 10.1021/ja01145al26).

According to an embodiment, the average pore diameter is determined from the pore size distribution determined by gas adsorption in accordance with ISO 15901- 2:2006. In short, a representative sample is extracted using the procedure of ISO 8213 : 1986. The sample is first degassed under vaccuum and at temperatures above

100°C. The analysis is then performed using one of four methods: static volumetric method, flow volumetric method, carrier gas method or gravimetric method. According to an embodiment, the analysis is performed using the static volumetric method.

Irrespective of the method, the same analysis is performed. The adsorbed gas (typically nitrogen) at different pressures/partial pressures (the adsorption/desorption isotherm) at liquid nitrogen temperature (77 Kelvin), is used for determining the amount of adsorbate taken up by the sample. First the t-plot method is used for determining if the sample contain mesopores. Thereafter one of two methods can be employed, algebraic methods, such as the Barrettt, Joyner and Halenda method [J Am Chem Soc 73 (1):373- 380. doi: 10.1021/jaOl 145al26], or non-localised denisity function theory, as described in ISO 15901-3:2007. According to an embodiment, the non-localised density function theory as described in ISO 15901-3:2007 is used. According to another embodiement, the algebraic method of Barrettt, Joyner and Halenda method [J Am Chem Soc 73 (l):373-380; doi: 10.1021/jaOl 145al26] is used. The calculated values for incremental pore volume are then expressed as a function of pore diameter. The average poor diameter is calculated from the thus obtained pore size distribution. As recognized by the skilled person, the average poor diameter is the pore diameter with 50% of the total pore volume above as well as below the pore diameter value.

According to an embodiment, the average pore diameter is determined from the pore size distribution determined by gas adsorption in accordance with ISO 15901-

2:2006 using the static volumetric method and the non-localised density function theory described in ISO 15901-3:2007. According to another embodiment, the average pore diameter is determined from the pore size distribution determined by gas adsorption in accordance with ISO 15901-2:2006 using the static volumetric method and the algebraic method of Barrettt, Joyner and Halenda method [J Am Chem Soc 73 (l):373-380;

doi: 10.1021/ja01145al26]

According to an embodiment, the alumina support thus has average pore diameter exceeding 60 Angstrom. Preferably, the average pore diameter of the alumina support is at least 80 Angstrom, or even at least 100 Angstrom, such as about 120 Angstrom. Thus, the average pore diameter of the alumina support may be 80 to 200 Angstrom such as 100 to 140 Angstrom. The average pore diameter of the alumina support may be determined using nitrogen physisorption at liquid nitrogen temperatures and calculated using e.g. the BJH (Barret, Joyner, and Halenda) method assuming cylindrical pores (cf. Barrett EP, Joyner LG, Halenda PP (1951) The Determination of Pore Volume and Area Distributions in Porous Substances. I. Computations from Nitrogen Isotherms. J Am Chem Soc 73 (l):373-380. doi: 10.1021/ja01145al26).

According to an embodiment, the average pore diameter is determined from the pore size distribution determined by gas adsorption in accordance with ISO 15901- 2:2006. In short, a representative sample is extracted using the procedure of ISO 8213 : 1986. The sample is first degassed under vaccuum and at temperatures above

100°C. The analysis is then performed using one of four methods: static volumetric method, flow volumetric method, carrier gas method or gravimetric method. According to an embodiment, the analysis is performed using the static volumetric method.

Irrespective of the method, the same analysis is performed. The adsorbed gas (typically nitrogen) at different pressures/partial pressures (the adsorption/desorption isotherm) at liquid nitrogen temperature (77 Kelvin), is used for determining the amount of adsorbate taken up by the sample. First the t-plot method is used for determining if the sample contain mesopores. Thereafter one of two methods can be employed, algebraic methods, such as the Barrettt, Joyner and Halenda method [J Am Chem Soc 73 (1):373- 380. doi: 10.1021/jaOl 145al26], or non-localised denisity function theory, as described in ISO 15901-3:2007. According to an embodiment, the non-localised density function theory as described in ISO 15901-3:2007 is used. According to another embodiement, the algebraic method of Barrettt, Joyner and Halenda method [J Am Chem Soc 73 (l):373-380; doi: 10.1021/jaOl 145al26] is used. The calculated values for incremental pore volume are then expressed as a function of pore diameter. The average poor diameter is calculated from the thus obtained pore size distribution. As recognized by the skilled person, the average poor diameter is the pore diameter with 50% of the total pore volume above as well as below the pore diameter value.

According to an embodiment, the average pore diameter is determined from the pore size distribution determined by gas adsorption in accordance with ISO 15901-

2:2006 using the static volumetric method and the non-localised density function theory described in ISO 15901-3:2007. According to another embodiment, the average pore diameter is determined from the pore size distribution determined by gas adsorption in accordance with ISO 15901-2:2006 using the static volumetric method and the algebraic method of Barrettt, Joyner and Halenda method [J Am Chem Soc 73 (l):373-380;

doi: 10.1021/ja01145al26]

According to an embodiment, the average pore diameter is determined from the pore size distribution determined by gas adsorption in accordance with ISO 15901- 2:2006 using the static volumetric method and the non-localised density function theory described in ISO 15901-3:2007. According to another embodiment, the average pore diameter is determined from the pore size distribution determined by gas adsorption in accordance with ISO 15901-2:2006 using the static volumetric method and the algebraic method of Barrettt, Joyner and Halenda method [J Am Chem Soc 73 (l):373-380;

doi: 10.1021/ja01145al26]

Further, the pore volume of the support may be at least 0.75 ml/g. The pore volume of the alumina support may be determined using nitrogen physisorption at liquid nitrogen temperatures and calculated using the method suggested by Barret et al (cf. above). According to an embodiment, the alumina support has a surface area of 120 ± 10 m ² /g. The density of the alumina support may be 0.42 to 0.56 g/ml.

The hydrotreatment takes place in a rector to which glycerol, a petroleum refinery intermediate stream, and hydrogen is fed. As already outlined, the glycerol and the petroleum refinery intermediate stream may be fed as a mixture (preferred) or separately to the reactor. In the reactor, the catalyst is present. The reactor may be operated as a fixed bed reactor. Typically, the reactor is operated at a total pressure of 60 to 120 bar and at a temperature of 330 to 420 °C. The partial pressure of hydrogen may be 20 to 70 bar.

As already discussed, the initial distillation of crude oil in petroleum refinery results in various petroleum refinery intermediate streams. These streams are outlined in Table 1 below with their typical boiling point range at atmospheric pressure.

Table 1 - Petroleum refinery intermediate streams

Petroleum refinery intermediate stream Boiling point (°C)

Light naphtha up to 150

Heavy naphtha 150-205

Kerosene (middle distillate) 205-260

(atmospheric) Light gas oil 260-315

(atmospheric) Heavy gas oil 315-425

Vacuum gas oil_ 425-600

Various petroleum refinery intermediate streams may be used in the present process. According to an embodiment, the petroleum refinery intermediate stream is a petroleum refinery intermediate stream as defined in Table 1, such as heavy naphtha, kerosene (middle distillate), light gas oil (atmospheric light gas oil), heavy gas oil (atmospheric heavy gas oil), or vacuum gas oil. The petroleum refinery intermediate stream is preferably a stream already being subject to hydrotreatment step in the petroleum refinery. According to a preferred embodiment, the petroleum refinery intermediate stream is light gas oil (atmospheric light gas oil), heavy gas oil

(atmospheric heavy gas oil), vacuum gas oil, as defined in Table 1, or any mixture thereof. Especially, the petroleum refinery intermediate stream may be light gas oil (atmospheric light gas oil) or vacuum gas oil, as defined in Table 1.

As shown in Fig. 1 (cf. arrows), glycerol may be added to any hydrotreatment step in a petroleum refinery. Further, glycerol may also be added to other unit operation(s) in a petroleum refinery able to convert glycerol to propane, such as to a hydro cracker and/or to a catalytic cracker.

Preferred reaction conditions for the hydrotreatment step already have been defined herein above. Though the reactor is operated at elevated temperature, the mixed stream of glycerol and the petroleum refinery intermediate stream may still be a liquid stream if being high boiling. The mixed stream may thus be hydrotreated in liquid state. Alternatively, the mixed stream may be hydrotreated in gaseous state. Further, the process may comprise the step of pre-heating the glycerol before being mixed with the petroleum refinery intermediate stream. As the petroleum refinery intermediate-stream may be fed from the initial distillation of crude oil, this stream may already be heated.

The petroleum refinery intermediate-stream is very lipohilic, whereas glycerol is highly hydrophilic. In liquid state, they are thus not miscible, whereas the gaseous forms are.

Though the mixed stream may be hydrotreated as a two-phase system, it is preferred to affect the solubility of glycerol in the petroleum refinery intermediate stream. According to an embodiment, a surfactant is thus present in the step of mixing the glycerol with the petroleum refinery intermediate stream. According to such an embodiment, it is preferred if the surfactant is mixed with the glycerol before mixing the glycerol with the petroleum refinery intermediate stream. Typically, a quite low amount of the surfactant is added. Thus, the surfactant may be added in a weight ratio, with respect to glycerol, of 1 : 10 or less. However, preferably the ratio is lower than 1 :10. The surfactant may thus be added in a weight ratio, with respect to glycerol, of 1 :25 or less such as of 1 : 100 or less. The surfactant may thus be added in a weight ratio, with respect to glycerol of 1 :25 to 1 :500, such as 1 :50 to 1 :250, or 1 :75 to 1 :150.

As the he surfactant is to be present in the hydrotreatment, it should be essentially free from metal. There is a low tolerance to metals entering the

hydrotreatment, and a maximum of 10 ppm is considered for the combined metals stemming from the surfactant and the rest of the feed.

According to an embodiment, the surfactant is a polymeric surfactant.

Polymeric surfactants are macromolecules, which contain both hydrophilic and hydrophobic parts in their structure. One type of polymeric surfactants is derived by the polymerization of a surface-active monomer. Such polymeric surfactants also known as polysoaps. Another type of polymeric surfactants is derived by copolymerization of a hydrophobic and a hydrophilic monomer. This kind of copolymers can thus have a random, a gradient, or a block structure. Amphiphilic diblock copolymers are basically the macromolecular transposition of low-molecular weight surfactants and,

consequently, they are commonly referred to as macrosurfactants. Examples of polymeric surfactants comprises of EO/PO block copolymers, acrylic/styrene copolymers, methacrylic copolymers, poly hydroxystearate derivatives and alkyd PEG resin derivatives. A preferred polymeric surfactant is Hypermer B246-SO sold by Croda. The source of the glycerol is not essential in terms of successfully converting the glycerol. However, in terms of providing a process for producing propane from a renewable resource, at low cost, the source of the glycerol is of importance. According to an embodiment, the glycerol thus stems from glycerol obtained as side product in biodiesel production. Especially, the glycerol to be mixed with the petroleum refinery intermediate stream may be provided from a methyl transesterification process, wherein a triglyceride and methanol are the starting materials. Thus, the glycerol to be mixed with the petroleum refinery intermediate stream may be aqueous glycerol comprising at least 75 wt.% glycerol and up to 25 wt.% water. The aqueous glycerol may comprise at least 80 wt.% glycerol and up to 20 wt.% water, at least 85 wt.% glycerol and up to 15 wt.% water, or at least 90 wt.% glycerol and up to 10 wt.% water.

Further, given the sensitivity of the employed catalyst, salts, e.g. inorganic salts, present in the crude glycerol obtained in the methyl transesterification process have to be removed. They may be removed by use of a combination of anion and cation exchange resins. Anion and cation exchange resins are preferred over, e.g. traditional distillation process as they require much lower energy input for the purification.

Purification by ion exchange resins is enabled since there is no need for removing organic contaminants before feeding the glycerol to the refinery, as organic

contaminants may be feed with the glycerol to the hydrotreater without negatively affecting the hydrotreatment. Preferably, the aqueous glycerol comprises less than 100 ppm inorganic salts, such as less than 50 ppm inorganic salts, less than 25 ppm inorganic salts, or less 15 ppm inorganic salts.

Once the mixed stream has been hydrotreated, thereby increasing the relative amount of propane, a fraction comprising propane is separated. Given the low boiling point of propane, a fraction comprising propane is typically easily separated from other hydrocarbons originating from the petroleum refinery intermediate stream, having significantly higher boiling point, e.g. by distillation. According to an embodiment, the separated fraction comprising propane is a liquid petroleum gas fraction.

In a petroleum refinery, typical unit operations, apart from fractional distillation and hydrotreatment, typically include cracking. Cracking serves to crack higher hydrocarbons to lower hydrocarbons. Further, cracking may serve to crack aromatics to hydrocarbons. Examples of cracking include catalytic cracking and/or hydro cracking. According to an embodiment, the refined, mixed stream is subject to cracking before separating a fraction comprising propane from the refined, mixed stream. The cracking may be catalytic cracking and/or hydro cracking. Such cracking may further increase the content of propane.

As already outlined, it was found to be advantageous to use a catalyst with somewhat larger pores. Hence, according to a second aspect there is provided use of an alumina supported, heterogeneous and porous catalyst comprising nickel, or cobalt, and molybdenum, and having an average pore diameter exceeding 60 Angstrom, to convert glycerol to propane in a petroleum refinery by hydrotreatment of a mixture of glycerol and a petroleum refinery intermediate stream with hydrogen over the catalyst. As already described, the mixture of glycerol and the petroleum refinery intermediate stream may further comprise a surfactant to provide one-phase system.

The average pore diameter of the catalyst may be at least 80 Angstrom, such as at least 100 Angstrom. Further, the average pore diameter of the catalyst may be 80 to 200 Angstrom, such as 100 to 140 Angstrom. The average pore diameter of the catalyst may be determined as has been describe herein above. Accordingly, the average pore diameter is, according to an embodiment, determined from the pore size distribution determined by gas adsorption in accordance with ISO 15901-2:2006.

Further, the alumina support may have average pore diameter exceeding 60 Angstrom. Preferably, the average pore diameter of the alumina support is at least 80 Angstrom, or even at least 100 Angstrom, such as about 120 Angstrom. The average pore diameter of the alumina support may be 80 to 200 Angstrom, such as 100 to 140 Angstrom. The average pore diameter of the alumina support may be determined as has been describe herein above. Accordingly, the average pore diameter is, according to an embodiment, determined from the pore size distribution determined by gas adsorption in accordance with ISO 15901-2:2006.

Furthermore, the pore volume of the support may be at least 0.75 ml/g. The pore volume of the alumina support may be determined using nitrogen physisorption at liquid nitrogen temperatures and calculated using the method suggested by Barret et al. (cf. above). According to an embodiment, the alumina support has a surface area of 120 ± 10 m ² /g. The density of the alumina support may be 0.42 to 0.56 g/ml.

The catalyst is a supported, heterogeneous and porous catalyst comprising nickel, or cobalt, and molybdenum. The support may be alumina, such as delta-(6)- alumina. The catalyst may comprise 1 to 6 wt.% nickel or cobalt. Further, the catalyst may comprise 5 to 15 wt.% molybdenum. According to an embodiment, the catalyst comprises nickel and molybdenum. According to such an embodiment, the catalyst may comprise 2 to 5 wt.% nickel, and 5 to 10 wt.% molybdenum. The support may, according to such an embodiment be delta(6)-alumina. The catalyst is typically used in its sulfide form.

As already described, the catalyst may be a porous and heterogeneous catalyst obtainable by:

- impregnating porous delta-alumina, to act as support, with an aqueous solution of ammonium molybdate to fill the catalyst pore system and subsequently drying and calcining the alumina support impregnated with molybdate; and

- impregnating the molybdate impregnated alumina support with an aqueous solution comprising a nitrate or acetate salt of cobalt or nickel and citric acid, the citric acid being present in ratio of 0.5 to 1 mol citric acid per mol cobalt or nickel, to fill the catalyst pore system and subsequently, drying and calcining the alumina support impregnated with molybdate and cobalt or nickel. Further, the mixture of glycerol and the petroleum refinery intermediate stream may comprise a surfactant.

Without further elaboration, it is believed that one skilled in the art can, using the preceding description, utilize the present invention to its fullest extent. The preferred specific embodiments described herein are, therefore, to be construed as merely illustrative and not limitative of the remainder of the description in any way whatsoever. Further, although the present invention has been described above with reference to specific embodiments, it is not intended to be limited to the specific form set forth herein. Rather, the invention is limited only by the accompanying claims and, other embodiments than the specific above are equally possible within the scope of these appended claims, e.g. different from those described above.

In the claims, the term "comprises/comprising" does not exclude the presence of other elements or steps. Additionally, although individual features may be included in different claims, these may possibly advantageously be combined, and the inclusion in different claims does not imply that a combination of features is not feasible and/or advantageous.

In addition, singular references do not exclude a plurality. The terms "a", "an", “first”,“second” etc. do not preclude a plurality.

Drawings

Fig. 1 depicts an exemplary flow scheme for petroleum refining. Examples

The following examples are mere examples. They should by no means be interpreted to limit the scope of the invention. Rather, the invention is limited only by the accompanying claims.

Example 1 - mixed stream comprising glycerol and light gas oil

In a first example, a 10% by weight glycerol solution in a light gas oil was prepared by mixing 1 g of surfactant (Hypermer B246-SO sold by Croda) with 100 g of aqueous glycerol of 95% purity and mixing it slowly with 900 g of light gas oil (boiling point 260 to 315°C). The mixing was performed using moderate stirring at 25°C and by slowly adding the glycerol and surfactant to the mixture.

Example 2 - mixed stream comprising slvcerol and vacuum gas oil

In a second example a 10% by weight glycerol solution in a vacuum gas oil was prepared by mixing 1 g of surfactant (Hypermer B246-SO sold by Croda) with 100 g of aqueous glycerol of 95% purity and mixing it slowly with 900 g of vacuum gas oil (boiling point 425 to 600°C). The mixing was performed using moderate stirring at 50°C and by slowly adding the glycerol and surfactant to the mixture. Example 3 preparation of large pore hydrotreatment catalyst

The catalyst was prepared in two steps, starting from a pre-shaped support of porous delta-alumina provided by Sasol GmbH. The alumina support had an average pore diameter of 120 Angstrom, a pore volume of 0.77 ml/g and a specific surface area of 118 m ² /g. The support was in a granular form sieved to between 10 and 20 mesh, resulting in 0.5 to 1 mm particles of irregular shape.

The first step was an incipient wetness impregnation where the desired amount of ammonium molybdate was deposited on the catalyst using a water solution that was allowed to entirely fill the catalyst pore system, followed by drying and calcination. The concentration of the ammonium molybdate tetra hydrate as purchased from Sigma Aldrich (99% purity) was 190 g/liter. The catalyst was dried at ambient temperature for 4 h and then at 120°C for 12 h. The temperature was thereafter increased by 4°C/min from 120°C to 500°C.

The second step was an incipient wetness impregnation with a water solution containing the nitrate salt of nickel, with citric acid added up to a stoichiometric amount to the nickel, followed by drying and calcination The purpose of the citric acid is to lower the pH to change the adsorption properties of the nickel or cobalt to enhance the activity of the final catalyst.

The second step of catalyst preparation was performed starting from the catalyst intermediate from step one after it has been allowed to cool to ambient temperature. The second incipient wetness impregnation was performed using a mixture of deionized water, nickel nitrate hexahydrate (Sigma Aldrich, 99% purity) and citric acid mono hydrate (Sigma Aldrich, 99% purity). The concentrations of nickel nitrate hexahydrate was 209 g/liter, and citric acid monohydrate was 106 g/liter respectively. The catalyst was dried at ambient temperature for 4 h and then at 120°C for 12 h. The temperature was thereafter increased by 4°C/min from 120°C to 500°C. The catalyst is then transferred to its sulfide state by reaction with sulfur-containing hydrocarbons or hydrogen sulfide before use. In this case, hydrogen sulphide at 1,000 ppm concentration in hydrogen was used at a gas hourly space velocity of 500 starting at ambient temperature and increasing the temperature with 2°C/min to 400°C and holding at this temperature for 4 h.

The resulting catalyst had an average pore diameter of 117 Angstrom, a pore volume of 0.75 ml/g, a surface area of 113 m ² /g, a Ni concentration of 3.1 wt% and a Mo concentration of 8.9 wt%. Example 4

In the third example, two catalyst consisting of 3 wt% Ni and 8 wt% Mo supported on alumina, one commercial (with 60 Angstrom in average pore diameter) and the another one with 117 Angstrom in average pore diameter (cf. example 3), were used. The two catalysts were loaded in one reactor each with a 15 mm inner diameter and operated as a fixed bed with a liquid hourly space velocity of 1.5 h ¹ using a simulated vacuum gas oil feedstock mixed with 10% by weight glycerol as prepared in example 1. The reactor was operated for 200 h using each catalysts at 100 bar and 360°C.

The operation of the reactor with the small pore catalyst showed an increase in propane production and the yield was calculated to 88%. For the other catalyst with larger pores, the propane production corresponded to a yield of 93%, i.e. slightly higher yield. Interestingly, post mortem analysis of the catalysts showed that the small-pore catalyst was containing 6.3% by weight heavy carbon residue, while the corresponding number for the open-pore catalyst was only 1.2% by weight. This is indicative for the open-pore catalyst maintaining its catalytic activity of far longer time. Example 5

In a fourth example, two new batches of the catalyst described in example 4 were put through the same test, using a feedstock prepared as in example 2.

The operation of the reactor with the small-pore catalyst showed an increase in propane production and the yield was calculated to 76%. For the other catalyst, the propane production corresponded to a yield of 91%. Post mortem analysis of the catalysts showed that the small pore catalyst was containing 10.8% by weight heavy carbon residue, while the corresponding number for the open-pore catalyst was 1.7% by weight. As can be seen, difference between the two catalysts was even more pronounced for hydrocarbon feedstock with higher boiling point (light gas oil vs vacuum gas oil).

Example 6

In this example, another catalyst was prepared following the same method as in example 3, but omitting the citric acid from the second impregnation solution. This resulted in a much lower dispersion of the active Ni and Mo phase. The catalysts were analyzed using hydrogen chemisorption. Firstly, the catalysts were reduced at 400°C using a gas stream consisting of 4% hydrogen in He. Thereafter, an isotherm adsorption at 40°C was measured twice on the same sample. The difference between the two were calculated and represent the chemisorbed hydrogen. The volume adsorbed by the catalyst prepared without citric acid was 0.109 ml of hydrogen adsorbed per gram of catalyst, while the catalyst prepared with citric acid was 0.189 ml of hydrogen adsorbed per gram of catalyst. There is thus a significant effect on the dispersion by adding the citric acid.