Coker naphtha hydrotreating

Highly exothermic olefins saturation and silica contamination can occur when hydrotreating coker naphtha
Rasmus Breivik and Rasmus Egebjerg Haldor Topse A/S

n recent years, the application of residue upgrading technologies has grown in a market with increasing prices for light sweet crudes and a decreasing demand for heavy residual fuel oil. Carbon-rejection technologies such as delayed cokers have been the preferred choice, since crude quality declines with increasing metal content and Conradson carbon residue (CCR). Installation of a new coker unit, resulting in an increase in cracked distillate, poses some additional challenges to downstream hydrotreaters. This is particularly the case for naphtha hydrotreaters, since the properties of coker naphtha are very different from those of straight-run (SR) naphtha.

Coker naphtha contaminants

Typically, coker naphtha contains 1020 times more sulphur and higher amounts of olefins, nitrogen and silica than SR naphtha. Target levels of nitrogen in the reformer feed are around 0.10.5 wtppm to avoid ammonium chloride deposition. This implies that the coker naphtha hydrotreater needs to be operated at high severity to meet the required product specifications on nitrogen. However, it is practically impossible to increase operating severity with a higher temperature, since sulphur recombination takes place at high temperatures. The origin of the silica can be traced back to the silicone oil added to the heavy residue feed to the coker. As a result of gas formation, silicone oil is added to the coker drums to suppress foaming. Excess quantities of silicone oil will crack or decompose to form modified silica gels and fragments. These are mostly distilled in the naphtha range and carried to the downstream hydrotreaters together with coker naphtha.

Unfortunately, silica poisoning of the catalyst in the downstream hydrotreaters reduces HDN activity and results in large catalyst volumes being required to ensure simultaneous removal of nitrogen and silica. The required space velocity to match the turnaround schedule of the reformer/coker unit will often be below 0.2hr-1 when using standard hydrotreating technology and catalyst. Another important consideration for coker naphtha hydrotreating is control of the temperature increase from olefins saturation, since this reaction takes place readily and is highly exothermic. When olefin saturation is not properly controlled, this may lead to a premature shutdown, as excessive coke formation will take place due to gum formation/polymerisation in the top layer of the catalyst bed. In the future, it is expected that there will be a higher ratio of cracked components in the refinery slate. For new grassroots units, a three-reactor layout is typical. In the first reactor, diolefins are mainly saturated at low temperature. In the second, silica is adsorbed on high-surface area catalysts known as silica guard. Simultaneously, most of the olefins are saturated, while a relatively large degree of HDS and a small degree of HDN take place. Finally, a reactor with high HDS and HDN activity catalyst ensures the sulphur and nitrogen specifications are met. Even when operating with only two reactors, these three steps must be carried out.

Controlled removal of conjugated olefins

Olefins formation is a result of high-temperature conversion reactions in the delayed coker. Upon contact with air, olefins and diolefins may form gum that complicates the transportation and processing of coker naphtha. Diolefins and, in

www.digitalrefining.com/article/1000153

PTQ Q1 2008 1

Selectivity

Selectivity

Catalyst B Catalyst A
Conversion

Conversion

Figure 1 Selectivity of diolefin saturation reaction

Feed

Feed

660

Product Product

Figure 2 Bypass silica reactor (left) and lead/lag reactor system (right)

particular, conjugated diolefins present in coker naphtha must be saturated in order to stabilise the feed. The stabilisation takes place in a separate reactor, since the required operating conditions are quite different from those of the HDS and HDN reactions. The saturation of diolefins is a fast reaction and can therefore be carried out at high LHSVs and at low temperatures. Conjugated diolefins polymerise at normal hydrotreating conditions, and the polymers cause fouling of the reactor, resulting in pressure drop build-up. The polymer formation potential of coker naphtha is about 300 times the potential for SR naphtha. A generally accepted way of controlled saturation of conjugated olefins is to use a hydrotreating catalyst operated at a low temperature (160220C) in the presence of hydrogen. With the growing market for coker technology, focus is put on the reliability of the coker unit,

resulting in more coker units being designed with ISBL facilities to remove conjugated olefins. The benefit of making the removal of conjugated olefins independent of the coker naphtha hydrotreater is the possibility of sending the coker naphtha directly to storage if the coker naphtha hydrotreater offers an unforeseen shutdown. Many units are currently designed independently of the coker naphtha hydrotreater with oncethrough hydrogen, which facilitates the change of catalyst on-line to increase coker unit reliability. Since the exothermic diolefin reactions produce large amounts of heat, it is necessary to operate at low temperatures using a selective catalyst to control the temperature increase. High selectivity means the catalyst should saturate diolefins only and not mono-olefins, which might lead to too high a temperature increase in the reactor. Topse has carried out studies of catalyst selectivity by processing a model feed containing 1, 3-hexadiene. The diolefin-to-olefin reaction selectivity was studied as a function of diolefin conversion, as shown in Figure 1a. As expected, the selectivity decreased with conversion, since the two reactions (saturation of 1,3-hexadiene to hexenes and saturation of hexenes to hexane) are consecutive. The effect of pressure was also investigated. In Figure 1b, the selectivity is shown as a function of the reactor pressure for two catalysts. At all conditions, the hexadiene is completely converted, but the selectivity towards hexane increases with decreasing pressure. Furthermore, the two catalysts tested have different selectivity and slightly different responses to changes in pressure. This means that even though the saturation reaction is fast, there are a number of knobs that can be turned to control the reactions to ensure complete saturation of diolefins and high selectivity towards olefins are achieved. This

2 PTQ Q1 2008

www.digitalrefining.com/article/1000153

knowledge is used to optimise the selection of catalyst and process conditions to maximise the run length.

Silica removal and olefin saturation

The simultaneous removal of silica by adsorption and saturation of olefins requires a proper design of both the catalyst system and process schemes to control temperature increases and pressure drop build-up. The amount of silicon in coker naphthas is dependent on the economics of the coker unit. Typically, a coker naphtha fraction contains about 5 wt ppm silicon, but silicon contents of up to 10 wt ppm can be seen on some occasions. Unfortunately, silica poisoning of the catalyst in the downstream hydrotreaters reduces the HDN activity of the catalyst to a large degree. Typically, when processing feedstocks in the past with low silica contents, the approach was to allow the silica to be absorbed on the bulk catalyst, since only a small amount of catalyst was required to remove silica. However, with the increasing amount of coker material to be processed, the content of silica and nitrogen has increased, and it is nowadays desirable to confine the silica deposition to the guard material to allow the use of a high-activity catalyst as a bulk catalyst. Different configurations can be selected for the silica reactors, dependent on feedstock and refinery-specific considerations (Figure 2): Bypass silica reactor Lead/lag reactor system. Typically, the bypass reactor configuPolydimethylsiloxane ration is selected for hydrotreaters operating with a low content of coker CH3 CH3 CH3 CH3 naphtha and a high amount of silica. . . . . . . . Si O Si O Si (CH3)3 The main problem with a bypass reac- (CH3)3 Si O Si O Si tor is that the heat integration should CH3 CH3 CH3 CH3 be prepared for switching the temperature increase from the silica guard CH3 CH3 reactor to the main hydrotreater reacSi H3C O H C CH 3 3 tor during catalyst change-out. This Si O Si limits the fraction of coker naphtha to CH3 CH 3 be processed with this reactor O O O Si CH3 H3C configuration. CH3 O Si Si Si In a lead/lag reactor configuration, H3C CH3 CH3 CH3 O two reactors are connected in series Hexamethylclotrisiloxane Octamethylcyclotetrasiloxane (Figure 2). When the lead reactor is saturated with silica, the catalyst may be replaced while still running the unit Figure 3 Structure of silicone oil (PDMS) and typical decomposition at full capacity. The flow is then redi- products (cyclic siloxanes) present in coker naphthas

rected to have the fresh catalyst in the lag position and the partially saturated silica guard in the lead position. The lead/lag reactor system is preferred for hydrotreater units processing high amounts of coker naphtha. This solution has the advantage that a very high temperature increase can be allowed across the silica guard system. With the lead/lag reactor system, it is possible to process a larger fraction of coker naphtha than with the bypass reactor configuration. Typically, the fraction of coker naphtha is controlled by recycling part of the product to dilute the coker naphtha fraction. Allowing a larger fraction of coker naphtha to be processed would therefore decrease the actual throughput of the high-pressure reactor section, reducing both investment and operating cost. Another benefit would be the full utilisation of the silica guard, since a large silica slip can be tolerated to the lag reactor, ensuring the maximum silica pick-up in the lead reactor is reached before the catalyst change-out. Extensive studies have been conducted to determine the mechanism of catalyst deactivation when subjected to coker naphtha feedstocks containing silicon. These studies investigate the nature of the silicon deposition chemistry and the effect of silica contamination on HDS and HDN activity. Silica in coker naphthas originates from silicone oil added to the heavy residue feed in upstream processes. Typically, the silicone oil

www.digitalrefining.com/article/1000153

PTQ Q1 2008 3

surface groups SiOH and Si(OH)2 and partly of modified silica gels with methylated surface species. The deposition of Si is similar to the deposition of coke. No chemical bonds to the alumina or metal sulphides are formed. The deactivation is caused by adsorption on surface reaction sites reducing the amount of active sites that the sulphur and nitrogen species Temperature, C Temperature, C have access to. However, the loss of activity due to Si is irreversible Figure 4 Influence of catalyst-specific surface area (left) and operating and cannot be restored by catatemperature (right) on the Si uptake capacity lyst regeneration. The deposition has been shown is a polydimethylsiloxane (PDMS). At the to be an activated and diffusionally controlled elevated temperature of the delayed coker, PDMS reaction catalysed by the surface alumina sites. decomposes, and Topse has identified a homo- This means the silicon uptake capacity is higher logue series of cyclic siloxanes in coker naphthas, for catalysts with higher specific surface areas examples of which are shown in and that, at higher bed average temperatures, Figure 3. the silica capacity of the catalyst will be higher The amounts of these siloxanes causing the (Figure 4). In order to fully utilise the silica catalyst deactivation are determined by a quanti- capacity of the guard, it is thus necessary to tative Si-sensitive GC analysis. Furthermore, the design the heat profile of the entire unit properly interaction of these siloxane species with the from SOR to EOR. catalyst surface as well as a characterisation of The influence of Si deposition on catalyst activspent catalysts have shed light on the deactiva- ity is more pronounced for HDN than for HDS tion mechanism. The cyclo-siloxanes are quickly (Figure 5), and the HDN activity can thus be adsorbed on the catalyst surface. A detailed char- followed to track silicon contamination. These acter-isation of the Si deposits on aged coker findings have implications for catalyst design naphtha HDT catalysts was carried out by NMR with respect to maximising Si capacity and HDS/ spectroscopy. Si is present in the form of modi- HDN activity. Since the surface alumina sites are fied silica gels consisting partly of bulk Si02 with responsible for the decomposition of the siloxane species, a low metal loading and a high surface area are optimum for Si uptake. However, low metal loadings result in a low HDS/HDN activity. The knowlHDS edge acquired from fundamental studies of silicon deposition chemistry combined with canister tests in industrial coker HDN naphtha hydrotreaters has led to the development of Topse TK431, TK-437, and TK-439 catalysts, allowing an optimum design of the silicon guard catalyst system. This catalyst series SiO2 on catalyst, wt% consists of catalysts with varying silica capacities and HDS/HDN Figure 5 Influence of SiO2 deposition on HDS and HDN catalyst activity activities. Depending on
Si capacity of TK-431 Si capacity of TK-437
Activity relative to fresh, %
Si on spent catalyst, % Selectivity

4 PTQ Q1 2008

www.digitalrefining.com/article/1000153

feedstock reactivity and operating conditions, the correct catalyst system can be selected, maximising cycle length.

Interbed cooling
The combination of high olefin content and the low-end boiling point of coker naphtha results in large temperature increases, which are typically seen across the first catalyst bed. For high concentrations of olefins, temperature increases of more than 100C have been observed in industrial units. The combination of a high temperature increase and the possibility of sulphur recombination, which requires a low temperature, results in a substantial need for interbed cooling. In the past, interbed cooling was carried out by quenching with cold gas or liquid. However, with increasing energy prices, alternative ways of cooling need to be considered. Topse recently developed a new technology utilising the high temperature increase to improve heat integration by applying interbed heat exchangers. This new concept not only improves heat integration, but also increases catalyst activity, since the treat gas rate and fugacity can be controlled independently of the required cooling requirements. The concept has already been applied in the design of several coker naphtha hydrotreaters.

energy is required, resulting in a high temperature increase in the topmost part of the catalyst bed even when utilising low-activity catalysts. The key to solving this challenge, in combination with the activity grading, is to design the reactors and the rate of treat gas to allow a sufficient heat transfer in the axial direction, thereby avoiding hot spots and resulting crust formation. This problem has been overlooked in the past, but is becoming more important as the amount of olefins in feeds increases. However, even with the best activity grading, some crust and coke formation will still occur in combination with contaminants introduced with the feedstock. This calls for additional protection to make sure there is a sufficient void for these types of contaminants. Typically, this is done by controlling the bed void size. Controlling the bed void size is done by loading the reactor with different sizes and shapes of grading material: high void topping material followed by larger rings on top of smaller rings, and underneath these the bulk catalyst. The reason for using rings of different sizes is to create a loading that has a differentiated filtering effect, so the largest contaminants introduced with the feed are trapped in the upper layers and the smaller particles are deposited in the lower layers.

Grading
The high olefin content of coker naphthas increases the risk of coking on the catalyst, and good grading becomes more important to avoid excessive pressure drop build-up. Typically, a hard crust layer leading to rapid pressure drop build-up would be experienced in the top part of the catalyst bed if the grading and reactor dimensions were not designed to handle the higher amount of olefins. The solution to the problem comes from a fundamental understanding of crust formation with respect to graded catalytic activity; heat transfer; bed void size; and rate of olefin saturation versus temperature. In the past, part of the solution was to grade the top of the catalyst bed by catalytic activity only (ie, gradually increasing catalytic activity from the inert top layer down to the main bed by several increments instead of having one step from inert to maximum activity). However this is not sufficient for high olefinic feedstocks, since only a very small activation

Achieving sub-ppm sulphur and nitrogen levels

After the removal of silica and the saturation of olefins, the removal of sulphur and nitrogen to very low product levels remains. The light-boiling sulphur components present in coker naphthas are generally quite readily removed. However, mercaptan formation through the recombination of olefins and H2S may limit HDS conversion. For nitrogen removal, a high-activity HDN catalyst should be used to obtain reformergrade naphtha. The catalyst selection depends on feedstock and product specifications. Major challenges are to select the correct pressure level and to avoid recombination to limit the cycle length.

Hydrogen partial pressure

One of the key parameters to define is the minimum required hydrogen partial pressure. Below a certain hydrogen partial pressure it becomes impossible to reduce the product nitrogen to the level required for the reformer feedstock, even when increasing the operating temperature. The

www.digitalrefining.com/article/1000153

PTQ Q1 2008 5

increase in reactor temperature cannot make up for the loss in catalyst activity when approaching EOR. To understand the fundamental chemistry behind the recombination Equilibrium reactions, pilot plant experiments have been carried out. Using l-hexene as representative of olefins in the liquid feed and a Low WHSV H2S/H2 treat gas mixture, three High WHSV mercaptans were formed: 1-, 2- and 3-hexanethiol. This observation strongly implies that both a radical Temperature, C addition pathway and an elecFigure 6 Recombination of H2S and hexene as a function of temperatures. trophilic addition pathway are Q1 At low temperatures, the reaction is kinetically controlled; at high 2008 possible reaction routes. To temperatures, it is equilibrium controlled prove that not only terminal olefins may undergo recombination, an actual reactor outlet hydrogen partial pressure is experiment with cyclohexene was carried out, a function of several parameters: total pressure; and only a single thiol (cyclohexanethiol) was degree of feedstock vapourisation; amount of observed. It is thus necessary to take all olefin circulated treat gas; and type of quench, liquid isomers present in the reactor into account when product or treat gas. All these need to be consid- modelling the recombination reactions. As was ered along with the ammonia inhibition and demonstrated, l-hexene readily isomerises into fugacity of the reaction components to reach the the more stable 2- and 3-hexenes at typical reacoptimum catalyst activity. tion conditions. Hydrogen partial pressure used to be The recombination reactions may reach equiconsidered the only factor affecting catalyst librium at high temperatures, as demonstrated activity. However, the fugacity of the sulphur in Figure 6, where the measured equilibrium components is dependent on the amount of treat constant (Pthiol/ Phexene PH2S) is shown and gas being circulated. Since the treat gas-to-oil compared with the theoretical equilibrium line. ratio has pronounced effects on the fugacities of At low temperatures, the reaction is kinetically all reacting molecules, a more thorough analysis limited, while at high temperatures the reaction in the design phase is needed to define the opti- is controlled by equilibrium. Such fundamental mum treat gas amount that results in maximum studies help determine to what extent recombination reactions take place in the reactor catalyst activity. effluent train and to take measures to avoid Sulphur recombination these. The reaction of H2S with olefins to form mercaptans is known as recombination and is of Conclusion particular importance in coker naphtha hydrot- Hydroprocessing of coker naphtha material reating due to the high amounts of olefins and requires specialised catalysts and technologies to organic sulphur in these feedstocks. The forma- overcome the challenges of the high olefin and tion of olefins from dehydrogenation of paraffins silica content, and to produce a product meeting is an endothermic reaction. At high reactor outlet the very stringent sulphur and nitrogen specifitemperatures, mercaptans may be encountered cations. A careful selection of process conditions in the product, even though the recombination as well as a proper integration of the different reaction itself is exothennic. Thus, contrary to sections within the unit make longer runs possicommon practice, it may be necessary to lower ble and ensure a better profitability. During the the reactor temperature to reach the product last two years alone, Topse has designed more sulphur specification. The recombination reac- than ten units processing coker naphtha tion may then limit the cycle length, since an feedstock.
Recombination equilibrium constant

6 PTQ Q1 2008

www.digitalrefining.com/article/1000153

This article was presented at the ERTC 12th Annual Meeting 2007 in Barcelona, Spain. Rasmus Breivik is proposal and technology manager for refining technology at Haldor Topse A/S in Lyngby, Denmark. Breivik earned a BSc in chemical engineering from the Technical University of Denmark. Rasmus Egebjerg is with Haldor Topse A/S in Lyngby, Denmark.

Links
More articles from the following categories: Catalysts & Additives Delayed Coking

www.digitalrefining.com/article/1000153

PTQ Q1 2008 7