The Hydrotreating Process Part 3

Other Process Aspects

The operation of hydrotreating reactors is considered to be very close to adia-batic because the heat losses from the reactor are usually negligible compared with the heat generated by the reactions. The exothermality of hydrotreating reactions, predominantly HDS and the hydrogenation of aromatics, can cause (LHSV: liquid hourly space velocity) or a weight basis (WHSV: weight hourly space velocity). LHSV and WHSV are calculated as follows:

LHSV and WHSV are related to each other by the equation an increase in reactor temperature beyond the design limits, depending, of course, on the conversion level, reaction conditions, and feed properties— which is why an appropriate temperature control system is required. Temperature control is essential to achieve an economically acceptable catalyst cycle length and to obtain the product quality required. Hence, in the sense of temperature control of multiple catalytic beds, injection of quench fluids and/or heat integration of the effluent from each bed come together.

Figure 3.9. Effect of LHSV on product sulfur content during hydrotreatment of various middle distillates (P = 54 kg/cm2, H2/oil = 2000scf/bbl, T = 360°C).

The design of the quench system is based on establishing an appropriate reactor temperature profile by determining the number of catalyst beds and their respective lengths in order to retain the required product quality and an economically acceptable catalyst cycle life. This is accomplished by solving the reactor mass and energy balances simultaneously until the optimal configuration of the system is obtained. An important aspect when designing a quench system is to consider a maximum allowable temperature which limits the reactor bed length and is commonly about 30°C or less above the inlet reactor temperature. Figure 3.10 illustrates the quench concept for a catalytic fixed-bed reactor with several quenches (Alvarez et al., 2007a).

Commonly, control of the reaction temperature in hydrotreating reactors is achieved by introducing part of the hydrogen recycle stream between the catalytic beds, called quenching or cold shot cooling. The use of quench liquids has also been reported. Quenching fluids are introduced in the quench zone or quench box, which is typically a mixing chamber where the bed effluent is mixed with the cooling medium. The flow of fluid injected to each quench location is adjusted to achieve the desired temperature profile and is specified to limit the temperature rise below the maximum allowable temperature.

 

Figure 3.10. General representation of quench in a hydrotreating reactor.

Another vital aspect of fixed-bed hydrotreating reactor performance is the reactor internal hardware design. Reactor internal hardware allows for efficient catalyst utilization by means of effective reactant distribution, quenching performance, and fouling protection. Most of the fixed-bed hydrotreating reactors currently in operation in worldwide petroleum refineries have been built and designed over the past 30 years. These units have been experiencing under-performance with the increasing supply of heavier oils to refineries, the tightening environmental legislation, and poor reactor internal design. Some of these problems were partially solved with increases in reaction severity, which reduced considerably the catalyst cycle life due to enhanced catalyst deactiva-tion. Mechanical constraints in reactor design and product demand were other problems that refiners had to face when trying to increase reactor temperature and reduce feed flow rate (i.e., decreased space velocity). In addition, excessive pressure drops were present due to fouling caused by solids contained in the feed (iron scale, salts, coke fines, etc.) and reaction products (coke and metals).

Over the years, many strategies have been proposed to meet current product specifications and at the same time to keep the catalyst cycle life at acceptable levels. Those strategies are based on the development of new highly active catalysts, tailoring reaction conditions (e.g., temperature, LHSV, hydrogen partial pressure) and designing new reactor configurations (e.g., multibed reactors with interstage quenching, reactors in series, and counterflow reactors); for fouling abatement, improved procedures for catalyst loading, low activity mesoporous materials, and graded-bed designs were developed. However, the experience has shown that improving catalyst performance and maximizing its volume within an existing unit are the most cost-effective options for improving unit performance. Two key parameters have been identified as possible solutions of these problems (Patel et al., 1998): (1) increasing catalyst activity and (2) efficient distribution of the reactants through the catalytic bed by means of proper reactor internal design.

Quench Systems

Conventional Quenching Quenching with hydrogen is used widely in most hydrotreatment units. Hydrogen, being the main reactant in hydrotreating, has the advantage of replenishing some of the chemically consumed hydrogen in the catalytic beds, decreasing the hydrogen sulfide and ammonia partial pressures in the reactor, which reduces the inhibition effect on HDT reactions and keeps the catalyst clean by inhibiting coke formation. The availability of quench hydrogen depends on the H–oil ratio along the reactors, which is a design condition that influences product quality. The value of the H2/oil ratio depends primarily on the compressor capacity within the plant. High H-/oil ratios improve the product quality and increase the quench availability; for example, a staged hydrocracking unit that operates at a H2/oil ratio of ~10,000scf/bbl may have up to five gas injection points. However, high H2/oil ratios also imply higher hydrogen recycling rates, and therefore larger compressors and equipment in the separation section are required, which increases the investment costs (Munoz et al., 2005).

Quenching with Liquids Processes that use liquid quench are not as common as gas-based quenching processes. That is why most of the information reported in the literature is related to hydrotreating reactors with hydrogen quench systems. However, quench hydrogen is not always the best option, due to its availability in refineries and compression requirements. In such cases quench liquids may be more attractive, due to their higher heat capacity and lower compression costs; nevertheless, it may require more reactor volume or a lower liquid-hourly space- velocity (i.e., more reactor volume to achieve the same conversion). The way in which a liquid quench is introduced into a reactor is different from that of a gas quench, and special reactor components and liquid quench injection devices are needed to provide an efficient contact between the gas and liquid phases. The processes that use liquid quench streams may be classified in two general categories:

1. Multiplefeedprocesses. Processes with multiple feeds are characterized by introducing several liquid hydrocarbon streams of different composition and properties at the top and between the beds of a reactor. Generally, the hydrocarbon feed is first fractionated, then the heaviest fraction is fed at the top of the reactor and lighter fractions are introduced as a side feed (Figure 3.11a). By this approach the side feeds act as quench streams and at the same time are processed together with each bed effluent in the following catalytic bed.

Figure 3.11. Examples of liquid quench-based processes.

2. Product recycleprocesses. In processes with product recycle, a portion of the reactor effluent is separated and cooled via heat exchange in order to be introduced between beds as quench stream. Generally, the recycle stream used as a quench comes from the bottom of the high- or low-pressure separators, as presented in Figure 3.11b. In this way, the portion of the heaviest treated fraction has a second-pass opportunity through the reaction system.

Comparison of Quench Approaches The use of quench fluids may have either a positive or a negative impact on the process configuration and the product quality, depending on several factors, such as quench fluid properties, flow rate and temperature, and injection points. Table 3.2 summarizes the advantages and disadvantages of using each quench fluid or method. The use of recycle hydrogen as quench always has a positive effect on product quality at the expense of high compression costs. Quenching with liquids may reduce such a disadvantage, but to achieve the same conversion degree will require more reactor volume or a decrease in LHSV. The choice between each alternative requires detailed process studies in order to select the most cost- effective option. For example, Bingham and Christensen (2000) evaluated the use of liquid quench versus recycle gas quench as well as other process aspects in order to revamp a two-stage HDS/HDA unit. For this particular unit, they determined that the most cost-effective alternative was the recycling of hydrotreated product from the high-pressure separator as liquid quench combined with state-of-the-art reactor components. Nevertheless, in other types of processes, such as hydrocracking, hydrogen will be the cooling medium of choice due to its effect on the product composition and quality and coke formation.

TABLE 3.2. Advantages and Disadvantages of Quench Fluids

Type of
Quench Fluid	Advantages	Disadvantages
Hydrogen	Replenishes consumed hydrogen	Low heat capacity
	Reduces H2S and NH3 partial	Increases requirements of
	pressures	equipment for recycling
		hydrogen
	Reduces coke formation	High pressure drops
	Improves distribution tray	Increases height of the reactor
	performance by increasing gas
	velocity
Liquid	High heat capacity	Increases LHSV and
		decreases reaction severity
	Reduces equipment requirements	Increases height and diameter
		of the reactor
	Reduces viscosity of the mixture	In case of vaporization,
		hydrogen partial pressure
		decreases
	Provides treatment to liquid	Increased costs due to
	quench	fractionation of the feed
	Allows for adjusting the hydrogen	May increase heat of reaction
	distribution in the fractions of
	the product
	Provides second-pass opportunity
	to unreacted species

In some cases it is possible to employ both liquid and gas quenches in the reaction system as described by Bradway and Tsao (2001) – who proposed a method for quenching HDT reactions by injecting recycled hydrogen along with liquid hydrocarbons. The proposal combines the multiple feed or product recycle process schemes along with hydrogen recycling in order to reduce the high-pressure drops in the reactor generated by the large gas volumes originally required to cool the reaction zone. Thus, the process scheme proposed makes it possible to enjoy the advantages of each quench method and lessen the disadvantages of each.

Reactor Internals The majority of the hydrotreating units operating at the end of the twentieth century relied on rudimentary reactor internal designs, such as sieve trays, chimney trays, conventional bubble cap trays, and impingement quench boxes; even worse, in some cases reactors may have any components at all. The design of those distributor trays was strongly influenced by hardware employed in fractionation columns, which is not necessarily adequate for trickle -bed reactors. Inappropriate reactor internal design caused poor catalyst utilization, due to flow maldistribution of the reactants at the inlet of the catalyst bed. Flow maldistribution was also enhanced by the increase in reaction severity, which eventually led to its detection by the high radial temperature differences measured at catalyst bed outlets. The main problems generated by flow maldistribution are the overuse of a part of the catalyst inventory and the formation of hot spots; meanwhile, the rest of the catalyst becomes underused, leading to poor product quality and shorter cycle lengths. This fact has increased the awareness of the importance of reactor internal design and its role in efficient catalyst utilization (Alvarez et al., 2007b ).

Based on the idea that the efficient catalyst utilization is governed by reactor components, a proper design must perform the following functions (Ouwerkerk et al., 1999):

• Uniform volumetric and thermal distribution of liquid and gas reactants over the cross-sectional area of the catalyst bed

• Inter- and intraphase mixing

• Quench performance

• Fouling resistance

• Space efficiency

• Ease in maintenance and installation

Reactor internal hardware may be located at the reactor inlet, interbed zones, and at the reactor outlet. The hardware at the reactor inlet provides an initial distribution of the reactants and protection against fouling; this is achieved by means of distributor trays together with fouling abatement trays and/or top-bed grading materials. For high-hydrogen-demanding feeds, where a large AT value is generated due to reaction exothermality, multibed reactors with interstage quenching are employed to limit the temperature rise; quench zones located between catalyst beds comprise a reactant collection system, a quench fluid-injection device, a chamber for mixing the cooling medium with the hot reactants, and a reactant redistribution tray. Finally, the reactor outlet contains hardware for fluid collection along with catalyst retention.

Figure 3.12 shows the hydrotreating reactor internal fundamentals according to the previous description for a unit with two catalytic beds and one quench.Axial and radial AT profiles within the reactor are also illustrated. The axial AT represents the typical temperature rise caused by the exothermality of the hydrotreating reactions in the catalytic bed. It allows for establishing the catalytic bed length when the reactor temperature reaches a maximum allowable temperature and for determining the number of beds for a required impurity removal. On the other hand, the radial AT value reflects the performance of reactor internals. The figure shows radial AT values for good and poor reactor internal performance; the former is characterized by low radial temperature differences after distribution trays and quench zones, and the latter exhibits gradual widening in radial AT, which provides evidence of flow maldistribution. It is worth mentioning that maldistribution has a cumulative character if the distributor trays and quench boxes are not working adequately; thus, in multibed reactors the poorest catalyst utilization will be in the last bed, which is reflected in the widest radial temperature differences.

Distributor Trays Certainly, the most relevant reactor internal feature is the distribution system, whose purpose is to establish radial liquid distribution across the catalyst bed, and thus it determines the performance of a trickle-bed reactor. To date, most hydrotreating units have used the original distributor designs, such as sieve trays, chimney trays, and bubble cap trays, the last two being the most successful.

Sieve trays are the most rudimentary systems, being simple and cheap in construction; they comprise a great number of liquid downcomers (perforations) across the tray and sometimes widely spaced chimneys for separating gas flow. This type of tray has been used more as a predistribution system, followed by chimney or bubble cap trays, than as a principal distributor. Chimney trays are basically descendants of sieve trays; their main feature is that of evenly spaced chimneys with lateral apertures for the liquid and a top aperture for the gas, thus allowing independent flow of both fluids. A great number of chimney designs are available, differing mainly in the number or type of lateral apertures, such as traditional chimney distributors, chimneys with triangular notches, and multiaperture chimneys. An alternative design of chimney trays comprises gas chimneys and triangularly notched liquid chimneys. Bubble cap trays are essentially similar to those employed in distillation columns, but with a different function. They are characterized by the contact and mixture of gas and liquid reactants, creating a mixed phase that flows through the slots of the bubble cap. The wide range of bubble cap trays goes from the early designs to the latest high-performance designs. Many types of the distribution systems described above have been patented over the years; however, they are simply variations of the original systems, delivering little improvement, and in many cases promote liquid maldistribution (Patel et al., 1998). Even though technical information on these designs is available in expired patents, the reasons for their underperformance were not well understood until recently.

Figure 3.12. Fundamentals of internals of hydrotreating reactors.

The study of distribution systems has been of great interest to mayor oil companies, resulting in the current state-of-the-art distributors, such as Shell GSI’s HD (high distribution) tray (Den Hartog and van Vliet, 1997; Altrichter et al., 2004), Tops0e -s vapor–ift tray (Yeary et al., 1997- Seidel et al., 2002),Exxon’s spider vortex technologies (Davis, 2002; McDougald et al. 2006), Akzo Nobel – s duplex tray (Akzo Nobel, 2003), and Fluor -s swirl cap tray (Jacobs et al., 2000). The development of these high -performance distributor trays, using sophisticated high-pressure cold flow units in combination with computational fluid dynamics (CFD), led to a better understanding of the flaws of the original designs. The meticulous evaluation of those designs highlighted the importance of specific parameters, such as a liquid source layout (i.e., tray spacing and wall coverage), discharge pattern, tray levelness, sensitivity to plugging, and flexibility in operation.

1. LIQUID-SOURCE LAYOUT A liquid-source layout is characterized by tray or center-to-center spacing and wall coverage capability. Tray spacing is referred to the distance between the centers of two drip points. This parameter is directly proportional to the catalyst particle diameter and must be optimized so that radial mixing, provided by the grading material, compensates for maldistribution.

Uniform liquid distribution can be achieved closer to the top of the catalyst bed with narrower tray spacing: in other words, a larger number of liquid point sources. On the other hand, wide tray spacing reduces catalyst utilization and requires more bed depth to correct liquid distribution by means of radial dispersion. In this matter, original tray designs do not have optimal tray spacing, as discussed earlier by Patel et al. (1998); for example, bubble cap trays are known for having the worst tray spacing, due to their relatively large size (50 to 100% larger than a chimney).

Wall coverage capability is the other layout parameter that influences reactor performance. Conventional distributors present dead zones without liquid sources near the reactor wall, as in the case of bubble caps. Poor wall coverage together with a disk discharge pattern contributes to flow bypassing, leaving a great percentage of unused catalyst near the reactor wall vulnerable to hot-spot formation.

2. DISCHARGE PATTERN The most important design parameter of distributor trays is perhaps the liquid discharge pattern. Along with tray spacing, it determines the percentage of wetted catalyst across the top of the catalyst bed and, consequently, overall catalyst utilization. For the last decade, distributor tray development has been focused on providing an efficient discharge pattern, which in this context refers to uniform distribution closer to the top of the catalyst bed. Conventional distributors, such as chimney trays and bubble cap trays, produce a disk type of discharge pattern, which wets only the catalytic surface right beneath the discharge point. This type of discharge pattern is very inefficient because it leaves a great percentage of unused catalyst at the beginning of the bed. However, commercial state- of-the- art trays provide a very wide spray discharge pattern, covering almost 100% of the catalyst bed. The liquid discharge pattern is governed by the hydrodynamics present in the discharge points. Liquid flow in sieve and chimney trays is governed by the overflow principle, where the liquid accumulated over the tray drips down through the sieves or the apertures on the chimneys, generating a disk-type discharge pattern, while gas enters separately through the top of the chimneys. In addition to the inefficient discharge pattern, these designs provide poor vapor-liquid contact, and therefore large temperature gradients may be observed. On the other hand, the gas-assist principle takes advantage of the high gas velocity to drag the liquid held on the tray, forming a dispersed liquid phase which is discharged through a central downcomer, as in the case of bubble caps and state-of-the-art distributors, although for the former this does not produce an efficient discharge pattern. This operation principle provides excellent vapor-l iquid contact, reducing interphase temperature differences by up to 90% (Ballard and Hines, 1965).

3. TRAY LEVELNESS Tray tilt or levelness is another important factor that must be considered during installation. When a tray is not leveled properly, the liquid will gravitate toward the lowest area of the tray, resulting in preferential liquid flow, causing poor distribution.

4. L IQUID – L OADING S ENSITIVITY A proper distributor design must be able to provide satisfactory performance over a broad range of liquid loads. Variations in liquid loading, such as those presented at start- and end-of-run conditions, may affect the functioning of distributor trays.

Quench Zones Hydrotreating fixed-bed reactors requires a quench system to control the temperature rise caused by the exothermality of the reactions. The main consequences of temperature runaway are hot-spot formation, leading to enhanced catalyst aging by coke formation and sintering, poor product yields due to excessive hydrocracking, and sometimes damage to the reactor vessel. As discussed earlier, controlling reaction temperature in hydrotreating reactors is achieved by introducing quench fluids into the quench zone located between catalytic beds. This interbed zone makes it possible to inject cooling medium, mixed with the hot reactants from the previous bed, and to redistribute the liquid and gas reactants across the following catalyst bed (Ouwerkerk, 1999).

Early interbed hardware designs included an impingement quench box together with a redistribution tray such as those discussed earlier (Ballard and Hines, 1970- Peyrot, 1987). The impingement quench box system comprises a quench tube that allows for injecting cold hydrogen, a liquid collector tray, a mixing box where the fluid impingement occurs, a perforated tray for collecting fluids coming from the mixing box, and a bubble cap redistributor tray. The operating principle of the quench box is based on:

• Division of the downflowing fluids (hot reactants and quench gas) into two streams which enter through the openings of the collector tray to separate chambers of the mixing box

• Directing the flow of the streams by means of baffles located in each chamber, causing the fluids to impinge in a turbulent mixing zone

• Discharge of the mixture toward the redistribution system

However, impingement mixing is known to provide ineffective gas-liquid mixing, due to poor interphase contacting, resulting in large gas-hquid temperature differences. The latter defect, along with an inappropriate redistribution tray, results in poor quench zone performance. This is characterized by wide radial AT values after each interbed zone, which persists or in the worst case grows as the fluids move down the reactor.

Nevertheless, the failures of conventional systems were corrected in new interbed designs which are constituted by vortex-type mixers and high-performance distributors [e.g., Shell GSI's UFQ (ultraflat quench), ExxonMobil's spider vortex quench zone, Chevron-Lummus's nautilus reactor technologies, Isomix - s internals, Fluor - s swirl zone vortex mixer]. The main feature of this type of system is the swirling motion of the fluids generated within the mixing box, which enhances gas-hquid contact. The performance of vortex mixers is explained well by Litchfield et al. (1996) and Pedersen et al. (1995) when describing operation of a proprietary quench zone design. The authors stress the difficulty in achieving effective mixing due to the large density differences between the quench gas and process gases, and the importance of the quench zone configuration in order to maximize intra- and inter-phase contact. The primary parts of a quench zone are the quench fluid-injection device, which imparts radial and perpendicular mixing of the process fluids and quench gas, and the vortex mixer arrangement, which provides turbulent swirling motion to the fluids. Injection devices include traditional quench pipes for direct injection into the mixing chamber, concentric manifolds with nozzles that surround the mixing chamber (e.g., UFQ quench ring) for radial inward injection, or the spider, which is a small manifold located at the center of the quench zone (spider vortex) for radial outward injection. Vortex mixers vary in the arrangement of vanes and baffles within the chamber, which create passageways and constrictions, which impart a swirling motion and turbulence to the fluids. A different approach to vortex mixers is the Albermarle Q- Plex quench mixer (Albermarle, 2006), where the quench gas and process fluids are passed through a single orificei thus, the constriction provides intimate intra-and interphase mixing. The operation is carried out in three mixers in series.

One important aspect of these systems is their reduced height in comparison to conventional systems. In commercial hydrotreating units it is extremely important to minimize the vertical dimensions of interbed internals to reduce the height of the required reactor vessel, especially in hydrocracking units, which may have more than two quench zones. Large reactors with wall thicknesses of 20 to 40 cm (high-pressure operation) represent considerably heavy reactor vessels, which in return increases the total cost due to the larger supporting structure required and difficult transportation and installation. Of all the commercial technologies, perhaps Shell’s technical papers emphasize this aspect most strongly. For example, after installing the UFQ and HD internals in a conventional hydrocracker, where 67% of the reactor vessel volume was occupied by catalyst, catalyst utilization grew up to about 86%, due to the reduced catalyst-to-catalyst distance of the UFQ (1 to 1.4 m) and elimination of the grading material due to the HD tray effectiveness (Swain and Zonnevylle, 2000). On the other hand, Albermarle’s Q-Plex has been reported to have a total height of about 0.5 m, which is much smaller than that of most vortex mixers (about 1 m).

However, reduced-height components compromise flexibility for operating at variable liquid-gas loads, especially at high loads, where flooding may occur or the residence time in the mixer is not enough for effective fluid mixing (Litchfield et al., 1996). It has also been reported recently that for optimal fluid mixing over a wide range of liquid-gas loads (33 to 200%) the vortex mixer should be 0.35 to 0.65 times the inner reactor diameter, the total quench zone length being less than 1.5 times the inner diameter (Van Vliet et al., 2006).