Most Common Reactors For Hydrotreating Biology Essay

Published: Last Edited:

This essay has been submitted by a student. This is not an example of the work written by our professional essay writers.

These reactors have been used widely for decades and are the most common reactors for hydrotreating. The technology is well-developed and can handle high throughputs (G. Biardi, 1999) using continuous operation. The system is simple to operate for hydrotreatment of light feeds and further developments now facilitate processing of heavy feeds (Furimsky, 1998) (J. Ancheyta, 2007). Gas phase fixed bed reactors are used for two-phase processes, in which the feed is fully vaporised and passes over the solid catalyst particles. Trickle bed reactors are used for gas-liquid-solid systems, consisting of H2 gas, a partially vaporised or liquid hydrocarbon and a solid catalyst (J. Ancheyta, 2007). The catalyst bed is often split into several sections, shown in Figure ‎3 .1.

Figure ‎3.1: Diagram of a fixed bed reactor (Furimsky, 1998).

The reactions that occur are exothermic, thus cool H2 is used to quench the reactor between beds, enabling temperature control (Furimsky, 1998). Another method is to dissipate the excess heat by partial vaporization of the liquid reactant (Stitt, 2002). Commonly reactor designs allow for a maximum temperature rise of 33oC, above this quenching would be necessary. As the catalyst activity declines, the temperature in the reactor should be increased, typically by 1oC per month (D. S. J. Jones, 2006). This occurs until the maximum design rating is reached, at which point the catalyst will need to be replaced or regenerated.

Some multiple bed reactors have been developed which facilitate the scrubbing of ammonia between beds (Furimsky, 1998). Other advantages of using multiple beds include increased vapour/liquid mixing and equal flow distribution. Therefore the catalyst is used more efficiently and hot-spot formation is avoided (D. S. J. Jones, 2006) (J. Ancheyta, 2007). To further help distribution; baffles, diffusers and dispersion trays can be used (F. M. Dautzenberg, 1985). Stagnant zones can also be prevented by using reactors with high length to diameter ratios (F. M. Dautzenberg, 1985). To prevent fouling and remove contaminants passing through the fixed bed reactor, filters are often used on top of the initial bed (Furimsky, 1998).

To withstand the hydrotreating conditions of moderately high pressure and temperature, reactors are now built from a Cr-Mo metal base, lined with austenitic stainless steel. These materials give both the high strength and corrosion protection required. However, to avoid corrosion cracking, the stainless steel must not be exposed to air, thus a nitrogen purge should be used (D. S. J. Jones, 2006).

Benefits of fixed bed reactors compared with less conventional reactor types are; continuous operation, no moving parts, lower investment and operating costs, ability for larger reactors, large amount of catalyst per unit volume and minimal back-mixing in down-flow mode. However, disadvantages are; impracticality for heavy feeds, high pressure drop and larger particle size (J. Ancheyta, 2007) (M. H. Al-Dahhan, 1997). The large particle size leads to; insufficient gas-liquid mixing, thus low efficiency of mass transfer and possible hot-spot formation. Consequently distributors are needed in the reactor and a higher reaction pressure is required (L. B. Datsevich, 2004). Figure ‎3 .2 shows the three types of fixed bed reactors, which will be discussed in more detail.

Figure ‎3.2: Types of fixed bed reactor (M. P. Dudukovic, 1999).

Trickle bed reactors

These are the most widely used fixed bed reactors commercially. Liquid hydrocarbon enters in the top of the reactor and trickles down, contacting the catalyst particles, to be withdrawn as effluent at the bottom (J. Ancheyta, 2007). H2 gas may be passed through co-currently in down-flow mode, or counter-currently. Trickle bed reactors can be operated in plug-flow mode, which increases particle wetting and helps prevent hot-spot formation. This also leads to higher conversions than fluidised reactors for the same reactor volume (R. H. Perry, 2007). Industrial trickle bed reactors typically have bed depths of 3-6m and a bed diameter of 3m (J. M. Winterbottom, 1999). Operating costs are considered to be low (Stitt, 2002).


Traditionally trickle bed reactors operate in down-flow mode. However, this results in high concentrations of H2S in the outlet at the bottom of the reactor. At this concentration, H2S combines with olefins, which inhibits the residual small concentrations of sulphur from being removed (I. V. Babich, 2003). The main problem with down-flow mode is that the catalyst may not be completely utilised due to the possibility of incomplete catalyst wetting. However, down-flow mode enables faster transport of gas to the catalyst, thus is preferred for gas-limited reactions (M. P. Dudukovic, 1999).


This system can be used where very stringent sulphur limits are required, which cannot be reached with a down-flow trickle bed reactor. A technology called SynSat improves on this mode of operation by removing H2S between catalyst beds (I. V. Babich, 2003). Using counter-current operation it is possible to remove sulphur down to 1ppm (A. Chander, 2001). The same amount of hydrodesulphurisation is possible with less catalyst than in the down-flow mode.

It is possible to combine both co-current and counter-current operation, as shown in figure 5.

Figure ‎3.3: Diagram of a counter-current fixed bed reactor, showing H2 quench (I. V. Babich, 2003).

A problem with counter-current operation is the occurrence of flooding at low flow-rates, making it unsuitable for industrial operation. For example, flooding occurs when using commercially practical values; a gas velocity of 0.15-3 m/s, a liquid velocity of 0.001-0.02 m/s, a particle size of 0.8-3 mm in a bed with porosity of 0.4 (E. G. Derouane, 2000). Flooding can be made to occur at a higher velocity by arranging packing, rather than using a randomly packed reactor, or through shaping catalyst particles (F. S. Mederos, 2007).

Packed bed up-flow reactors

In an up-flow fixed bed reactor the gas and liquid travel upwards co-currently. However, these are not frequently used commercially (M. H. Al-Dahhan, 1997). Up-flow reactors are better for liquid-limited reactions because there is fast transport of the liquid reactant to the catalyst and the catalyst particles are fully wetted (M. P. Dudukovic, 1999). If total wetting of catalyst particles can be achieved in a down-flow scheme the performance of both is equal (M. H. Al-Dahhan, 1997).

A problem with up-flow reactors is the possible formation of stagnant zones of liquid in the catalyst bed. Another problem is the possibility that high liquid hold-up reduces the availability of H2 gas to the catalyst particles (A. Chander, 2001). The high liquid hold-up results in a higher pressure drop compared with the down-flow reactor, due to the hydrostatic head of liquid (G. Biardi, 1999). However, the benefit of higher liquid hold-up is that up-flow reactors are more suitable for highly exothermic reactions because the excess heat will be absorbed into the liquid (G. Biardi, 1999).

Non-traditional fixed bed reactors

For feedstocks with high metal contents (25-150 ppmw), multiple catalyst systems are used. A reactor and suitable catalyst for hydrodemetallisation is followed by a reactor and catalyst for hydrodesulphurisation and other hydrotreating processes (J. Ancheyta, 2007).

Non-traditional reactors include periodically operated trickle-bed reactors, which are not currently used commercially. The operation alternates between minimum and maximum liquid flowrate because maximum flowrate facilitates fast liquid transport to the catalyst, whereas draining the liquid periodically allows rapid gas transport to the catalyst (G. Biardi, 1999). Up-flow reactors can incorporate structured packing. Catalyst particles are packed between wire mesh sheets in a structure that results in cross-flow, giving faster mass transfer (G. Biardi, 1999).

Monolith reactors produce higher mass transfer coefficients, compared with trickle-bed reactors and packed bed up-flow reactors because the gas and liquid flows in alternate slugs. This results in uneven distribution of gas and liquid but this may be overcome by sectioning the monolith (A. Cybulski, 2006). The pressure drop is lower than for the packed bed reactor and scaling up is simpler (A. Cybulski, 2006). However, these reactors are comparatively expensive (G. Biardi, 1999).

Moving bed reactors

These reactor systems use the most important features of both the plug-flow scheme in the fixed bed reactor and the easy catalyst replacement in the ebullated bed reactor as discussed in section 3.2.4. This results in a system that enables periodic catalyst replacement without stopping the process (Furimsky, 1998). To replace the catalyst, fresh catalyst is added at the top and slowly moves down the reactor until it is withdrawn at the bottom, as shown in Figure ‎3 .4.


Fresh/regenerated catalyst

Figure ‎3.4: Diagram of moving bed 'bunker' reactor (Furimsky, 1998).

Consequently, this system provides higher tolerance to impurities, such as metals, than the fixed bed reactor because these will be withdrawn with the catalyst. The need for moving bed reactors may be limited in hydrotreating, due to lower contaminant levels.

Compared with fixed bed reactors, moving bed reactors use higher temperatures (~400oC-430oC) and pressures (200atm), which will raise the operating cost (Furimsky, 1998).

Ebullating bed reactors

Ebullated or expanded bed reactors, similar to moving bed reactors, can process more difficult feedstocks than fixed bed reactors (Furimsky, 1998). These include vacuum residues and heavy oils, often containing a significant amount of contaminants, such as metals, sediments and asphaltenes (Furimsky, 1998).

A type of commercial expanded bed reactor, called the LC-Fining reactor, passes feed and H2 upwards through a distributor plate at high velocity, which results in the catalyst being expanded and operating in the turbulent flow regime (Furimsky, 1998). The catalyst is withdrawn from the bottom and fresh catalyst is added, shown in Figure ‎3 .5 .

Figure ‎3.5: Diagram of expanded LC-Fining reactor (Furimsky, 1998).

The main advantage of the ebullated bed reactor is that catalyst can be withdrawn, along with contaminants, whilst the reactor continues to operate (Furimsky, 1998). For this reason it is more suited to hydrotreating of heavier feeds, such as heavy gas oil, because coke is formed more quickly and therefore catalyst regeneration or replacement is needed sooner (A. Chander, 2001).

There is less catalyst hold-up per unit volume in an ebullated bed reactor, compared with a fixed bed reactor, due to the expansion of the catalyst during operation (F. M. Dautzenberg, 1985). The larger catalyst interparticle distance in the ebullated bed reactor (of 1.6 mm (F. M. Dautzenberg, 1985)) enables contaminants to settle to the bottom without accumulating. This avoids problems such as plugging or a higher pressure drop (Furimsky, 1998). The reaction rate is higher than in a fixed bed reactor because smaller catalyst particles of diameter less than 1 mm are used (Furimsky, 1998), (M. S. Rana, 2006). The small particle size enables the suspension of the catalyst by the liquid phase (M. S. Rana, 2006). The average temperature in the reactor is higher and these factors lead to a catalyst activity in the ebullated bed reactor of between 1.5 and 2.0 times higher than for the fixed bed trickle reactor (F. M. Dautzenberg, 1985).

Slurry phase reactors

In slurry phase reactors, either mechanical stirring or bubbling gas is used to suspend the solid catalyst particles in the liquid phase, shown in Figure ‎3 .6. In the mechanically stirred slurry phase reactors, the heat and mass transfer efficiencies are high, however, stirring leads to catalyst attrition and backmixing. The bubble column reactors exhibit a reduction in heat and mass transfer efficiencies. However, they can operate at higher catalyst loadings (30 vol% compared to 5% in the stirred tank), use less power and reduce the amount of backmixing (R. V. Chaudhari, 1979) (Stitt, 2002).

Figure ‎3.6:Mechanically stirred and bubble column slurry reactors (S. Whitaker, 1986).

Slurry phase reactors are operated in a batch mode, thus are more suited to processes with lower throughputs than the fixed bed reactor (E. G. Derouane, 2000). Uses in continuous mode are limited because of problems with separating the product and catalyst.

This main advantage of this operation is that the catalyst particles are smaller than in conventional reactor systems, resulting in a smaller interparticle diffusion length (F. M. Dautzenberg, 1985). There are more catalyst particles per unit volume of liquid (2.4x109 particles per cm3 compared with approximately 250 for ebullated bed and 120 for fixed bed trickle reactor (F. M. Dautzenberg, 1985)). This results in intensive mass transfer.

Effective temperature control is also possible due to the higher heat capacities and heat transfer coefficients in slurries. Maintaining isothermal operation is easier because of the large liquid volumes (R. V. Chaudhari, 1979). Good temperature control, combined with good mass transfer, facilitates a lower pressure to be used in comparison with fixed bed reactors. Other advantages are the reliability of the reactors and ease of scale-up (E. G. Derouane, 2000) as well as saving the cost of pelletizing the catalyst, which is required for fixed beds (R. V. Chaudhari, 1979).

The disadvantages of slurry bed reactors are; catalyst attrition resulting in difficult separation of product from the catalyst, erosion of process equipment due to continuously moving solid particles and the complexity of a continuous slurry phase process (E. G. Derouane, 2000).


Several types of reactors are available for hydrotreating. The main factor influencing the choice of reactor is the nature of the feedstock. This project will be focusing on hydrotreatment of distillate feed and light cycle oil. These relatively light feeds are most commonly treated in fixed bed reactors. This is mainly owing to their well-developed technology, ease of operation and simple design, compared with other reactors. The feeds do not require moving bed, ebullated bed or slurry phase reactors, mainly because they do not contain metal impurities and are not difficult to process. Counter-current operation is an option where highly stringent sulphur limits are required, however, flooding at industrial flow-rates may render this too problematic for effective use.

Alternative catalysts

Conventional commercial catalysts

Catalysts conventionally used for hydrodesulphurisation consist of a carrier support, such as alumina, magnesia or silica, impregnated with a metal oxide belonging to group 6 of the periodic table which is usually molybdenum oxide or tungsten oxide. In addition, the catalyst contain an active metal belonging to group 9 or 10 of the periodic table, usually cobalt or nickel. The catalyst is usually sulphided just before operation. Sulphiding converts the nickel or cobalt to the sulphide and molybdenum or tungsten oxides to disulphides by deposition of sulphur onto the catalyst's homogeneous surface, enhancing its performance. Incomplete sulphiding could result in an inactive catalyst. Alumina is generally preferred as a support because it is has a larger surface area, higher affinity to sulphide for high dispersion, better pore size control, and is significantly cheaper (Isao Mochida, 2004).

Catalyst consumption generally varies from 0.003 to 0.02 kg/m3 of feed depending on the severity of the operation and the metal content in the feed. The catalyst performance depends greatly on its pore diameter and particle size but in the case of hydrodesulphurisation of distillate feed it depends on the chemical composition of the catalyst. Some of the key parameters in choosing a catalyst are the metal dispersion, acidity and the catalyst deactivation rate.

Studies have shown that it is the MoS2 edges that are relevant to catalysis and here the molybdenum centre can stabilise a coordinatively unsaturated site (Gates & Knozinger, 2006). Substrates, such as thiophene, bind to this site and undergo a series of reactions that result in both C-S scission and C=C hydrogenation. The reaction mechanism is shown in

Figure ‎3.7: Reaction mechanism for thiophene on MoS2 (Gates & Knozinger, 2006)

Catalyst deactivation is usually caused by the formation of coke which may be stopped by adding surplus high-pressure hydrogen. H2S present in the recycle gas also inhibits the desulphurisation activity of the catalyst. Hence it is advisable for the recycle gas to be scrubbed of H2S (Parkash, 2003)

The most commonly used commercial catalysts for hydrodesulphurisation reactions are cobalt molybdenum and nickel molybdenum supported on alumina (Segawa, Takashi, & Satoh, 2000).

Cobalt molybdenum sulphide supported on alumina

CoMo sulphide is generally preferred for the hydrodesulphurisation processes, as it has a higher selectivity for sulphur than most other catalysts. They have the lowest hydrogenation activity hence consume the least amount of hydrogen for a given sulphur removal (D. S. J. Jones, 2006) and consequently have the lowest denitrogenation performance. CoMo catalysts require low operating pressures approximately below 40bar. They are thus best suited for desulphurisation at low pressure or when the hydrogen supply is limited.

Nickel Molybdenum

Nickel molybdenum is used when nitrogen removal is required or when more refractory sulphur compounds must be desulphurised. This is as a result of nickel molybdenum catalyst having a higher hydrogenation activity than cobalt molybdenum hence at similar operating conditions, the use of NiMo would result in a greater saturation of aromatic rings. Nitrogen is generally more difficult to remove than sulphur from hydrocarbon streams; hence any treatment that significantly reduces the concentration of nitrogen in a distillate stream would significantly reduce the excess sulphur. The cetane index of aromatic feed can be greatly improved using this catalyst at a pressure of 50bar and temperature range of 360oC - 380oC. The main disadvantage is deactivation by removal of the promoter which is not easily recoverable by regenerative processes.

Tungsten based

These catalysts are generally used in treating feeds where higher hydrogenation activity is required than with CoMo or NiMo. Their desulphurisation activity is generally poor at the pressures used for hydrotreating but their hydrocracking ability is enhanced which favours higher pressures (D. S. J. Jones, 2006). Tungsten based catalysts are chosen only when high activity for aromatic saturation is required coupled with activity for sulphur and nitrogen removal. They however have a low pore volume and an intermediate diameter. In most cases NiW would be chosen due to its higher activity than CoW for hydrogenation.

Unconventional catalysts

This project requires the reduction of sulphur content to 0.05% however, conventional catalysts are only capable of reducing the sulphur content to 0.25% (Segawa, Takashi, & Satoh, 2000). This implies that efforts need to be made to desulphurise the less reactive refractory sulphur species. Such compounds include the alkyl substituted dibenzothiophenes (DBT) such as 4-methyl-DBT (4-MDBT) and 4,6-dimethyl-DBT (4,6-DMDBT). Hence, there is a need for catalysts that can desulphurise these species and which are more resistant to contaminants such as hydrogen sulphide, ammonia, nitrogen and aromatics.

Two main reaction pathways have been proposed for the hydrodesulphurisation of alkyl substituted DBT which have been illustrated in Figure ‎3 .8 below. The direct desulphurisation route (hydrogenolysis) forms a biphenyl compound. The hydrogenation route forms hexahydrodi-DBT and tetrahydrodi-DBT as intermediates which are desulphurised to the corresponding cyclohexylbenzene derivatives. It is also known that hydrogenation of neighbouring phenyl groups reduces the steric hindrance caused by the methyl groups (Segawa, Takashi, & Satoh, 2000).

Figure ‎3.8: Reaction path for hydrodesulphurisation of DBT

(Li, Wang, Wang, & Chen, 2002)

The influence of positions of methyl groups on the conversion rate were studied (Segawa, Takashi, & Satoh, 2000). The work showed that methyl groups situated on the 4th or 4th and 6th carbon atoms provide a steric hindrance retarding the C-S bond, hence reducing conversion. Methyl groups arranged in alternative configurations such as the 2nd and 8th carbon atoms avoids the steric hindrance. Hence it is important to use a catalyst which; has a higher isomerisation activity and/or hydrogenation activity for deep hydrodesulphurisation (Segawa, Takashi, & Satoh, 2000), enhances hydrogenation of the aromatic rings and removes inhibiting substances (Song, 2003).

The sulphur content could possibly be reduced by using modified operating conditions for hydrotreaters with respect to the reaction temperature and liquid hourly space velocity (LHSV). However, higher reaction temperature results in coke formation on the catalyst and rapid catalytic deactivation, and lower LHSV results in reduced hydrotreating efficiency, thus, requiring an additional or larger reactor. In this case the best way of achieving deep HDS without changing the operating conditions and in a cost-effective manner is to develop a catalyst which has a very high HDS activity and high selectivity for the alkyl dibenzothiophenes (Fujikawa, Kimura, & Kiriyama, 2006).

Different methods of improving the catalyst include:

Improving the activity and/or selectivity by using different supports (MCM-41, carbon, HY, TiO2, TiO2-Al2O3) for preparing supported CoMo, NiMo and NiW catalysts.

Increasing loading level of active metal (Mo, W) by modifying the preparation procedure. This includes the use of different precursors, additives or different steps of metal loading.

Using additives or additional promoters (P, B, F) by adding one more base metal (e.g. Ni-CoMo, Co-NiMo, Nb).

Incorporating a noble metal (e.g. Pt, Pd, Ru).

Catalyst supports

The strength of the interaction with the support controls the dispersion, reducibility, acidity and catalytic activity. The support mesoporosity is important for improving the dispersion of sulphide layer. The nature of the support affects sulphidation of the active species, leading to better promoted active sites and dispersion of the catalysts. Figure ‎3 .9 shows the available catalyst support options.

Figure ‎3.9: Catalyst support options for refractory sulphur compound removal (Bej, Maity, & Turaga, 2004)

Supports mixed with alumina

The effects of additives to conventional alumina supports in is shown in Figure ‎3 .10. The resulting combined supports are used to perform the reactions shown by utilising the improvements in either acidity or dispersion.

Figure ‎3.10: Effects of additives on the properties of alumina supported catalyst (Bej, Maity, & Turaga, 2004)

Oxide supports

Titanium oxide

These supports have a higher catalytic activity for hydrodesulphurisation compared to alumina supported molybdenum catalysts and a higher reducibility to a lower valence state (Gates & Knozinger, 2006), (Kunisada, Choi, Korai, Mochida, & Nakano, 2003). However, TiO2 supports have no pore system, their specific surface areas are very small and they are difficult to pelletize. As a result, TiO2 supports cannot be used on their own industrially. Alumina coated in TiO2 leads to the formation of TiO2-Al2O3 composite supports which produces a useable catalyst with:

Higher activity & longevity: Increase in surface area leads to a higher number of active sites and the catalyst ageing through sintering is hindered, which leads to longer life of the catalyst.

Improved handling & mechanical stability: Titanium promotion leads to higher crush strength of the catalyst (less dusting and attrition of the catalyst) which results in improved handling and reduced pressure drop over the lifetime of the catalyst.

Enhanced breakthrough protection: Titanium promoted catalysts are easier to sulphide and remain sulphided for longer which results in additional operational security and enhanced breakthrough protection at no additional cost.


Zeolitic supports have been shown to reduce the sulphur content of diesel to 10ppm (Kunisada, Choi, Korai, Mochida, & Nakano, 2003). Zeolite is acidic which enhances the activity by increasing hydrogenation, methyl group migration and lowering the hydrogen sulphide inhibition. However, this high acidity causes easy deactivation and the occurrence of side reactions such as cracking and isomerisation. Therefore an alumina-zeolite composite support is preferred which has a much longer lifetime than any other zeolite supported catalyst.

The best way to achieve deep hydrodesulphurisation is to use two layers of catalysts. The first layer eliminates all of the reactive sulphur species and most of the refractory ones then the second layer can reduce the refractory sulphur compounds from 100-500ppm to less than 10ppm in the presence of the inhibitors. The choice between CoMo or NiMo for the first layer is of insignificant consequence.


Attempts have been made to add boron to alumina supports. The acidity of B/Al2O3 increases the hydrodenitrogenation activity and hydrocracking ability of the Ni-Mo catalyst. However, B/Al2O3 shows only a small change in the hydrodesulphurisation activity of the refractory compounds when compared to using the alumina support on its own at the same conditions. This suggests that the high acidity of B/Al2O3 has no significant contribution to the hydrodesulphurisation activity of Co-MoS2 and Ni-MoS2 catalysts as shown in Table ‎3 .1 (Li, Sato, Imamura, Shimada, & Nishijima, 1998). It can thus be concluded that using boron-alumina supports would not form a suitable catalyst for hydrodesulphurisation.

Table ‎3.1:Effect of adding boron to alumina based catalysts


These supports have a very high efficiency for the removal of nitrogen species and some refractory sulphur compounds, but are not very effective in removing sulphur species REFERENCE.

Siliceous MCM-41 (Mobile Crystalline Material)

Supporting NiW and CoW with siliceous MCM-41 leads to much higher desulphurisation than using conventional NiW and CoW supported with alumina. The catalyst activity for hydrodesulphurisation of DBT can be increased by up to 1.7 times, in the case of Co-Mo/Al/MCM-41 (Wang, Li, Chen, & Han, 2001).

Research was previously focused on aluminium containing MCM-41 supports considering that the strong acidity of alumina may help to crack the polyaromatic sulphur containing compounds, which would improve the hydrodesulphurisation activity. However, incorporating aluminium resulted in no significant change in the hydrodesulphurisation activity. It was observed that using MCM-41 without aluminium could increase the level of hydrodesulphurisation.

MCM-41 has an extremely high specific surface area, hence can easily accommodate the NiW or CoW. Ni-W/MCM-41 showed higher HDS activity than Co-W/MCM-41, which could imply that the difference in activity may be attributed to the promoters. Ni-W sulphides are more active in hydrogenation than Co-W sulphides.

Alternative acid gas (H2S) removal options

In general, the processes can be split into main categories; direct conversion, batch, solvents and other novel processes which are detailed in the subsequent sections. The physiochemical solvent techniques currently dominate the industry (Jensen & Webb, 1995), but many developing technologies are being considered in an attempt to improve the economics.

The main factors to consider when making the choice of correct processing technique are; acid gas concentration and volume, contaminants, operating conditions, composition, treated gas specification and economics (Jensen & Webb, 1995). The key parameter in choosing the process is the sulphur content in the feed gas (Manning & Richard, 1991) and the need for high H2S selectivity (Astarita, Savage, & Bisio, 1983). Figure ‎3 .11 shows the choice of amine treatment as a function of inlet and outlet H2S partial pressure.

Figure ‎3.11: Range of treatment processes (Astarita, Savage, & Bisio, 1983).

Direct conversion processes

Direct conversion processes are termed as such as they extract acid gas by direct conversion into elemental sulphur. This process makes the need for the Claus plant redundant and is therefore unsuitable for this particular service, but has been included for completion. The presence of the Claus plant means extra capital expenditure on the current plant to produce elemental sulphur is unnecessary and a poor economical choice. Direct conversion processes include dry bed processes and propriety direct conversion processes such as Stretford and Lo-Cat (Jensen & Webb, 1995).

Dry bed processes includes the iron sponge process using iron oxide and the molecular sieve technique using a reagent of crystalline alkali-metal aluminosilicates (Jensen & Webb, 1995). The Stretford process produces elemental sulphur from acid gas using sodium carbonate, sodium vanadate, anthraquinone and disulphonic acid (Jensen & Webb, 1995). The Lo-Cat process utilises iron ions to perform direct conversion of H2S into elemental sulphur (Newman, 1985). The operation is flexible and can operate over wide H2S content and process condition ranges and has demonstrated resistance to the presence of contaminants (Newman, 1985).

Batch processes

These include iron sponge and caustic soda treatment (traditional flue gas desulphurisation), however, because the reactant is discarded, it is only suitable for removing small sulphur concentrations (Manning & Richard, 1991).

Iron sponge is the most widely used form of non-regenerable scavenger (Houghton & Bucklin, 1994) and are of use for small quantities of sulphur (180kg/day or less) (Kidnay & Parrish, 2006). Batch processes have lower capital investment compared to continuous, however, pseudo-continuous operation requires two systems in parallel (one operating and one regenerating) (Manning & Richard, 1991). The small-scale nature of these operations make them unsuitable for the current design where large gas quantities are expected and continuous operation is demanded as part of the design philosophy.

Solvent processes

These processes have relatively high chemical and catalyst purchase and disposal costs (Jensen & Webb, 1995). Chemical solvents are corrosive (Burr & Lyddon), requiring specialist construction materials. Regeneration of chemical solvents is usually via the application of heat (Burr & Lyddon) and requires more energy than physical solvents (Kidnay & Parrish, 2006) since the heat of desorption is much lower for physical solvents than chemical.

These processes are implemented in a continuous operation over wide varieties of services and are the most popular method of acid gas removal. Technological advancements are still developing and a competitive marketplace will be advantageous to the designer of such a plant in order to achieve a good price for the technology.

Physical solvents

These solvents rely on physical interactions to extract H2S via physical absorption. Many physical absorption processes have been developed under various trade names and are shown in Table ‎3 .2. Since many of these processes are of a proprietary nature; royalty payments may be required (Burr & Lyddon) which needs to be considered when performing economical analysis. Physical solvents tend to be favoured when the acid gas concentrations or partial pressures are high (Kidnay & Parrish, 2006), around 200psia (Newman, 1985). Physical solvents have heavy hydrocarbon (C5+) co-absorption characteristics and are not usually used when there is a high hydrocarbon concentration (Burr & Lyddon). This contamination should be avoided since any hydrocarbons present will be absorbed and carried over into the solvent regeneration system and downstream processing such as the Claus plant. The particular solvents shown in table 2 are non-corrosive and relatively non-toxic (Newman, 1985).

Table ‎3.2: Physical solvent processes (Burr & Lyddon) (Jensen & Webb, 1995)

Process Name



Aqueous sulpholane and di-isolpropanolamine


dimethyl ether of polyethylene glycol





The Sulphinol process is good for feeds when a chemical solvent would be limited by high circulation rates and duty requirements (Hobson & Pohl, 1973). The operating conditions and unit operations are very similar to that of amine absorption (Hobson & Pohl, 1973).

The Selexol process is the selective removal of H2S using a combination of dimethyl ethers of polyethylene glycol. The solvent is manufactured by several companies including Dow (Burr & Lyddon). The higher viscosity of the solvent compared to others means reduced mass transfer (Burr & Lyddon) leading to increases in the required contacting surface area in processing equipment requiring more trays or packing. The process operates at relatively high temperatures up to 175°C (Burr & Lyddon).

The Purisol process, licensed by Lurgi AG, has the highest selectivity (Burr & Lyddon) and can produce a high purity treated gas stream. It is recommended that the treated gas is washed with water to prevent solvent loss due to a high vapour pressure (Burr & Lyddon). The operating range may require the use of refrigeration which can be expensive and is the economical barrier for its use.

The Rectisol process was the earliest commercial organic solvent process and is widely implemented in synthetic gas applications (Burr & Lyddon). The process is much more complicated compared to other solvent processes and the final flow scheme can differ vastly depending on the final product specifications (Burr & Lyddon). The solubility of H2S in methanol is higher than for the Selexol process and is further enhanced by the low temperature (Burr & Lyddon). Since methanol has a high vapour pressure, refrigeration is required to limit solvent losses in the treated gas (Newman, 1985) by operating at sub-ambient temperatures. Since such a low temperature is required the plant requires the use of stainless steel and expensive refrigeration systems, dramatically increasing the capital and operating cost. The main advantage of Rectisol is the better recovery of CO2 (Burr & Lyddon) which is negated in this case as only H2S is removed; therefore the process offers no advantage over Selexol.

The solvent regeneration process is usually performed by pressure reduction (flashing) or inert gas stripping (Burr & Lyddon). However, many also require thermal regeneration when the H2S content is high to achieve the highest purities (Burr & Lyddon).

Chemical solvents

Treatment with alkanolamine solvents, known as the Girbotol process (Hobson & Pohl, 1973), is the market leader and most popular process available (Jensen & Webb, 1995) and has been in use for many decades (Newman, 1985). Figure ‎3 .12 shows a typical amine treatment system.

Figure ‎3.12: A common amine treatment system (Newman, 1985)

Table ‎3 .3 shows selected solvents and associated properties. These solvents are preferred when the acid gas feed partial pressure is low (Teng & Mather, 1991). The use of alkanolamines is usually conducted at ambient temperature and follows the reaction below once physical absorption has occurred:

2ΡΝÎ-2 + Î-2S = (ΡΝÎ-3)2S (Hobson & Pohl, 1973)

The choice of amine is a function of; gas composition, H2S content and the desired purity (Hobson & Pohl, 1973). The selection of amine is of primary importance and the choice should consider; circulation rates, concentration, regeneration duty, selectivity, corrosion and losses (Newman, 1985).

Table ‎3.3: Selected amine solvents (Newman, 1985)






Solution strength (wt%)





Acid gas loading (mole/mole)





Regeneration Duty (BTU/lb)





Preferentially absorb H2S?


Under certain conditions


Under most conditions

Table ‎3 .3 highlights the common, most useful properties of selected amines required when deciding on an appropriate choice. MDEA (methyl diethanolamine) is usually the most concentrated solution used, as a result of it being a tertiary amine which is the least basic and therefore the least reactive. This means that the solution will be more expensive to purchase for constant amine volumes. This disadvantage is offset by the fact that the gas loading is unlimited meaning a lower circulation rate can be used. DEA (diethanolamine), followed by MDEA have the cheapest regeneration costs requiring the smallest duties.

MEA (monoethanolamine) is a primary amine and is therefore the strongest base and most reactive (Jensen & Webb, 1995) and has a high solution capacity (Kidnay & Parrish, 2006). The disadvantages of MEA are; a high vapour pressure means that a high proportion is lost via vaporisation (Kidnay & Parrish, 2006), high heats of reaction with H2S means high regeneration energy is required (Kidnay & Parrish, 2006) and relatively high corrosion rates (Kohl & Nielsen, 1997).

DEA and DGA (diglycolamine) are a secondary amines which mean a reduced reactivity compared with MEA. The advantage is a lower vapour pressure facilitating higher permissible concentrations (Kidnay & Parrish, 2006), due to less potential losses, decreasing the circulation rate. DEA has much lower heats of reaction and therefore the lowest regeneration duty but DGA has the highest heats of reaction (Newman, 1985) leading to the most expensive duty requirements. A major advantage is a reduction in corrosion leading to cheaper construction equipment and less operational maintenance.

Tertiary amines such as MDEA are the least reactive but have the lowest vapour pressure. This means the system needs more amine but less is lost through vaporisation leading to cheaper inventory costs over time. It has low rates of degradation (Newman, 1985) and therefore the least corrosion. MDEA is also H2S selective meaning good absorption is experienced (Hobson & Pohl, 1973). Figure ‎3 .13 shows the H2S loading capacity for selected amine solutions; it shows MDEA has the highest loading capacity of the pure amine solutions. Higher loadings are only possible using hybrid solvents of amine and methanol as discussed below. The solubility of amines is shown in Figure ‎3 .14; it shows an increase from primary through to tertiary and a general increase with increasing partial pressure.

Figure ‎3.13: H2S loading capacity (Newman, 1985)

The main problem with amine solutions is the formation of heat stable salts by anions from impurities in system make-up water or amine degradation by-products (Kadnar & Rieder, 1995). These salts prevent solvent regeneration (Bord, Cretier, Rocca, Bailly, & Souchez, 2005) and once the concentration reached 500ppm, the system experiences poor H2S removal, corrosion and foaming (Kadnar & Rieder, 1995). The formation of any heat stable salts should be minimised and the maximum concentration should not exceed 10% of the total amine concentration (DuPart, Bacon, & Edwards, 1993). Foaming can be prevented by addition of inhibitors such as oleyl alcohol or pre-washing the feed with water to remove foaming causing impurities such as organic acids (Hobson & Pohl, 1973). Corrosion forms iron sulphide particles in a fine dust. Large amounts of this can result in a sludge forming in the column. The use of a filter is recommended to prevent this (Hobson & Pohl, 1973).

An alternative to amine solutions is the Shell Phosphate Process which involves the use of tripotassium phosphate and proceeds by the following reaction:

K3PO4 + H2S = K2HPO4 + KHS (Hobson & Pohl, 1973)

Tripotassium phosphate is highly H2S selective and is a stable salt meaning no losses or contamination leading to very little make-up being required and a reduction in corrosion (Hobson & Pohl, 1973). Regeneration is via reboiling in a steam stripper resulting in a very high duty requirement (Hobson & Pohl, 1973).

Alkaline salt reactions with compounds such as potassium or sodium carbonate, known as the Benfield Process (Haring, 2008) are also used (Jensen & Webb, 1995). However, these only possess moderate H2S solubility (Hobson & Pohl, 1973) and extraction capacity and would not be suited to the current application.





Figure ‎3.14: Acid gas solubility (Teng & Mather, 1991)

Hybrid solvents

Hybrid mixtures are a combination of chemical and physical solvents. These include Ucarsol and Flexsorb. These solvents are used for high acid gas loadings and are also capable of removing organic sulphur present (Manning & Richard, 1991). An example of this is the AMISOL process, using a mixture of diethanolamines and methanol (Newman, 1985).

Other processes

This section includes processes which are implemented in small-scale and novel services and those in current stages of initial development. The processes detailed below are not suitable for the desired design in their current state but may become applicable in the future and have been included for completeness.


These processes have undergone development as a result of the search for more economical acid gas removal processes (Jensen & Webb, 1995). These processes reduce the energy consumption by operating at ambient temperature and pressure (Jensen & Webb, 1995). Since the development is still ongoing, only very few of the possible processes have been scaled-up and optimised for industrial application (Jensen & Webb, 1995). The majority of the microbiological processes produce either elemental sulphur or the SO42- ion as a product (Jensen & Webb, 1995) which are unsuitable for Claus processes and therefore cannot be used in the particular plant.


The use of membranes is applicable for high pressure gas with high acid gas concentrations (Burr & Lyddon). Membrane technology advantages are; low capital costs compared with solvent systems (Kohl & Nielsen, 1997) ; simple operation not involving any chemical reactions or solvents which greatly improves reliability and the ability to mount the systems on skids (Kohl & Nielsen, 1997) simplifying the construction and fabrication processes. However, the modular nature means the process does not benefit from economy of scale and the ease at which membranes can be fouled means pre-treatment would be required. High pressure drops are commonly experienced and may require additional compression or pumping.


From the work presented here it can be concluded that the most applicable technologies are solvent processes. These processes are able to handle the large quantities of gas associated with hydrodesulphurisation and will be able to reliably reach the outlet specifications.

Amine solutions are the most applicable solvents as determined from estimates using Figure ‎3 .11 and by evaluating the relative merits of each process. Amine technology is widely used and understood and can provide the plant with the required degree of acid gas recovery. It also provides the correct feed to the Claus plant.

In order to decide the correct amine solution, testing using HYSYS must be performed on the design specific case to determine the most effective amine for removing the required gas quantity. This work must also take into account factors of cost, corrosion and degradation. The information presented here indicates MDEA would be the most appropriate solution for this design, due to low corrosion, degradation and vaporisation losses.

Alternative separation techniques

There are many possible alternatives for separating the reactor effluent before further processing. This section gives a brief overview of some of these alternatives.

Separator drum configurations

The most common method for removing hydrogen is by using a cooler and cold high pressure separator, illustrated in Figure ‎3 .15. To improve the heat recovery in the system, a hot gas separator can be added before the cooler as shown in (Sepehr Sadighi, S.Reza Seif Mohaddecy, Omid Ghabouli, Majid Bahmani, 2009). It can be concluded that installing a hot gas separator before cooling the feed will reduce the heat exchanger pressure drop and required duty. The installation of a cold low pressure separator allows improved gas recovery, as shown in Figure ‎3 .17 (Les Harwell, Sam Thakkar, Stan Polcar, R.E. Palmer, 2003). This configuration also provides an added layer of protection between the high pressure and low pressure parts of the plant.

Figure ‎3.15: Use of cooler and cold high pressure separator

Figure ‎3.16: Introduction of hot high pressure separator

Figure ‎3.17: Introduction of a cold low pressure separator (Les Harwell, Sam Thakkar, Stan Polcar, R.E. Palmer, 2003).

Refinery fuel gas separation

A hydrogen recovery unit can be used to reclaim hydrogen from the hydrotreater purge gas that would otherwise be burned. It is important to recover the hydrogen because hydrogen demand in the petroleum industry is increasing by approximately 9% annually (U.S Department of Energy, 2009). Figure ‎3 .18 shows the methods available to perform the required hydrogen separation.


Hydrogen flowrate (nm3/hr)

Figure ‎3.18: Methods for hydrogen separation based on flowrate and recovery (Grasys)


Membrane technology, as shown in Figure ‎3 .19 can be used for this purpose. Feed gas to the membrane unit, containing aromatics and olefins must be processed to prevent condensation within the membrane unit as hydrogen is recovered. Hydrocarbon condensation on the membrane fibres would form a liquid film, resisting hydrogen transport and impeding recovery. Pre-treatment options include; feed gas refrigeration, scrubbing the feed using lean oil, or preheating the feed. Feed preheat also improves permeation rate. Since hydrogen permeates through the membrane, the hydrogen rich stream is produced at a low pressure while the hydrogen deficient purge stream is produced at a pressure only nominally lower than the feed pressure (R.L. Schendel, C.H. Mariz and J.Y. Mak, 1983).

Figure ‎3.19 Pilot and full-scale membrane and module (Liu, 2006).

The advantages using membrane separation are:

High purity hydrogen obtained

Compact and low weight

Minimum operating costs

Simplicity of installations and maintenance

Increase in catalyst service life

Pressure swing adsorption (PSA)

Pressure swing adsorption (PSA) is a technology used to separate some gas species from a mixture of gases under pressure according to the species' molecular characteristics and affinity for an adsorbent material. It operates at near-ambient temperatures using special adsorptive materials as a molecular sieve, such as zeolites. These preferentially adsorb the target gas species at high pressure. The process then swings to low pressure to desorb the adsorbent material The process then swings to low pressure to desorb the adsorbent material.

The PSA process consists of vessels packed with selective molecular sieves that adsorb hydrocarbons and other impurities to produce a 99%+ purity hydrogen product. When the sieve beds are depressurized, the hydrocarbon and residual hydrogen are released as waste gas. Lower waste gas pressure improves recovery, since a lower desorption pressure reduces residual hydrogen in the bed.

For a typical hydrotreatment plant, membrane separation is more favourable than PSA gas separation. Study results show that the total cost for hydrogen recovery is lower in the membrane separator case than the PSA case. Increases in power cost reduces the benefit shown by the membrane case, as would a longer payout period REFERNCE.