January 2025

Catalysts

Sensible diesel hydrotreating catalyst selection: An independent catalyst testing approach

This article presents a case study that illustrates how independent catalyst testing was utilized to identify the most effective diesel hydrotreating catalyst. 

IP: 18.191.171.26

The best performing catalysts usually offer better product yields with lower utility consumptions and often enable refiners to process cheaper (lower quality) feedstocks while maintaining the targeted cycle length. The costs associated with independent catalyst testing are easily outweighed by the performance gains from selecting the best catalysts.¹ 

While assessing hydrocracking catalysts is commonly considered to be highly challenging, selecting the best diesel hydrotreating catalyst is more complex due to: 

• Diesel hydrotreating net conversion is typically < 5.0 wt% without dewaxing (also depending on the feed). A well-cut, straight-run feed (e.g., from a crude distillation column) would easily achieve > 95 wt% diesel yield. As such, it is much more difficult to distinguish the best diesel hydrotreating catalyst based on the yield gaps alone. 

• Hydrogen (H2) consumption is considerably lower than that of hydrocracking, which makes the assessment more challenging because the differences between vendors can be minimal. 

• Predicting product properties with different estimators makes direct comparisons less useful. These properties are crucial for economic analysis, particularly in terms of volume swell and product blending. 

• Catalyst activity is estimated using kinetic models, but each vendor has its own models and assumptions, leading to comparison biases. 

Diesel hydrotreating catalyst comparison is particularly important for current diesel hydrotreating units with limited reactor volumes. In these cases, the best approach to maximize profitability is to select the optimal diesel hydrotreating catalyst that balances both operating and economic criteria. Examples of economic benefits include: 

• Increased refining profit by processing more refractory and less-expensive feed components. In a similar fashion to upgrading units, light cycle oil (LCO) can be converted into high-quality diesel rather than fuel oil via diesel hydrotreatment. 

• Maximized unit cycle length. Unplanned unit shutdowns cause severe refinery-wide margin losses. 

• Improved product quality and yield: lower product sulfur/nitrogen/density/total aromatics, and higher cetane number. Some refineries have cheap/surplus H2, making aromatics saturation activity an important economic driver by increasing volume swell. 

• Reduced energy consumption by minimizing the energy required to initiate the chemical reactions and maximize heat recovery. A higher-activity catalyst requires a lower reaction temperature and generates more heat of reactions, which can be recovered by feed-effluent heat exchangers.  

Given these potential economic benefits, selecting a diesel hydrotreating catalyst should be considered as crucial as it is for hydrocracking or naphtha reforming. In fact, it is more difficult to differentiate the best catalyst—choosing a diesel hydrotreating catalyst based solely on vendor predictions is almost like guessing. Without robust independent testing, the end user cannot accurately determine the real performance of the chosen catalyst system, resulting in an uncertain cycle after the catalyst changeout. 

This article presents a case study that illustrates how independent catalyst testing was utilized to identify the most effective diesel hydrotreating catalyst. 

Realistically scaling down diesel hydrotreaters. The first step in designing an effective catalyst testing program involves a comprehensive review of the commercial operation or process.  

FIG. 1 shows a block diagram of the diesel hydrotreating unit with a visual illustration how the commercial operation can be translated into the lab-scale experiment. 

FIG. 1. A block diagram of the diesel hydrotreating unit. *The presence of an amine scrubber ensures low (negligible) hydrogen sulfide (H2S) content in the recycle gas feed.**Dimethyl disulfide (DMDS) is required to reflect H2S content in the recycle feed if no amine scrubber is present in the commercial unit. 

This unit processes a blend of straight-run gasoil (SRGO) and LCO. In this case, an amine scrubber is used to remove the produced H2S in the recycle gas to enhance the desired chemical reactions [e.g., hydrodesulfurization (HDS)]. It is well-known that H2S inhibits HDS reactions and must be removed before recycling the remaining H2 back to the reactor inlet. In this case, the H2S concentration after the amine scrubber is generally < 30 ppmv. As such, there is no real need to add easily decomposable sulfur compound (e.g., DMDS) in the liquid feed to simulate the H2S concentration. 

Note: There are also diesel hydrotreaters without the amine scrubber in the recycle gas circuit. In those units, other process parameters are designed/adjusted to compensate the high H2S level (e.g., using a larger reactor or higher H2 partial pressure). To realistically scale down such specific units, the test designer should add the easily decomposable sulfur compound to reflect the actual H2S level in the recycle gas returning to the reactor inlet, as demonstrated in FIG. 1. When processing the data, some process variables should also be corrected to exclude the byproducts from these additives [e.g., excluding methane produced from DMDS from the total methane detected by gas chromatography (GC)]. 

The next considerations are H2 partial pressure and the H2-to-oil ratio. These two parameters should be compensated for the fact that H2 purity is not 100%. Additionally, the flowmeter used in the commercial process monitoring might not be 100% correct due to deviating gas composition/process conditions or incomplete meter calibration. In this case, the H2 partial pressure and H2-to-oil ratio were corrected to reflect actual operations. 

As the lab/pilot-scale experiment is generally a once-through operation, the H2 partial pressure and H2-to-oil ratio should reflect the reactor inlet conditions of the commercial-scale process (recycle operation). Normally, the recycle gas is first mixed with fresh H2 to maintain the reactor circuit pressure—as part of the H2 will be consumed in the reactor—making the inlet H2 purity different from the cold separator (where the recycle gas is separated from the reactor products).  

Another pitfall often encountered when designing a pilot-scale experiment is the inclusion/exclusion of the demetallization catalyst. Generally, the volumetric percentage of the grading/demetallization catalysts is 10%–15% and may be ignored from the lab/pilot reactor. However, some diesel hydrotreaters are processing high metal-content feeds (e.g., gasoil distilled from condensates) and the volume of grading/demetallization catalysts could take up to 50% of the total reactor volume. As such, the demetallization catalysts should be included when the volume fraction is significant. 

This approach better represents the overall activity of the reactor and, to a certain extent, helps decouple metal deactivation from coke deactivation during the experiment. Lastly, the scaling down should be based on weight-hourly space velocity (WHSV), as small inert solid diluent material is used in the lab/pilot scale to improve catalyst wetting, axial/radial dispersion and isothermality. The presence of this small inert solid material fundamentally makes liquid-hourly space velocity (LHSV) an invalid catalyst loading approach due to differences in catalyst concentration per unit volume compared with the commercial scale reactor. Additionally, catalyst vendors offer varieties of catalysts with varying densities and loading techniques (sock vs. dense loading), making the catalyst weights differ for a fixed reactor volume.  

When designing a test, LHSV must be first converted into WHSV using an accurate feed and catalyst density. However, since the catalyst density varies from loading scheme to loading scheme, the WHSV will be different for a constant LHSV. 

The test design summary of this diesel hydrotreating catalyst benchmarking is shown in TABLE 1. 

The authors’ company uses a single-pellet single-string reactor (SPSR) as a default to minimize the axial dispersion. In theory, an SPSR offers the lowest theoretical catalyst pellet concentration per unit volume, making it the most efficient loading technique to minimize axial dispersion (from both reactant/product concentration gradients), as illustrated in FIG. 2. 

FIG. 2. A graphical illustration of how an SPSR can minimize axial dispersion. 

In addition, an SPSR provides excellent temperature control and reproducible reactor loading because the diameter of the extrudates is slightly smaller than the reactor diameter. In addition, extrudates automatically align as a string of extrudates which, in combination with the narrow reactor, avoids the maldistribution of gas and liquid over the catalyst bed, thereby eliminating catalyst bed channeling and incomplete catalyst wetting. When an inert diluent is used, it can be introduced after catalyst pellets are loaded over the full length of the tube to fill the gap between extrudates and the reactor wall. 

The authors’ company’s micro-pilot plant allows for the highly efficient testing of catalysts for fixed bed processes, producing the highest data quality (repeatability, reproducibility and scalability) with low amounts of feed (less waste generated). The company’s proven high throughput parallel testing technologya was used.² A detailed description of the micro-pilot plant is offered in literature.3 The technology offers the possibility to test up to 16 different catalysts systems and/or process options side-by-side. Refineries can now cost-effectively explore more options or catalyst systems. 

With high-throughput technology, the authors’ company can run replicate reactors and significantly increase the precision on the anticipated activity, as well as the selectivity of catalyst systems under a variety of predefined conditions, providing relevant results to evaluate the economic impact of the catalysts’ performance (e.g., using the refinery economical/optimization models). 

Catalyst loading design strategy for LCO processing.

In general, LCO contains more refractive sulfur (e.g., di-alkyl benzo thiophenes) and nitrogen species compared with SRGO. Organic nitrogen compounds [especially basic nitrogen compounds with six-membered rings (e.g., pyridine derivatives)] are known to inhibit HDS and HDA reactions. From TABLE 2, LCO contains significantly higher sulfur, nitrogen and aromatics compared with SRGOs.

As such, it often requires the use of nickel-molybdenum (NiMo) catalyst (higher HDN/HDA activity to create a low-nitrogen zone in the reactor where HDS and HDA are accelerated). Despite the advantages, several drawbacks are related to the use of NiMo-based catalysts: 

• They tend to consume more H2 at a fixed product sulfur target, especially < 500-ppm product sulfur, where the reaction pathway selectivity starts to shift towards non-direct sulfur removal (pre-aromatic saturation of the first aromatics ring following by sulfur abstraction). 

• Since their functionality revolves around aromatics saturation, they do not function well under low-H2 partial pressure (requiring a minimum of 40 bara as a rule of thumb) and high-temperature operation, as the aromatic saturation equilibrium temperature decreases with lower H2 partial pressure. 

Depending on H2 availability and cost, NiMo catalysts can be also used in conjunction with cobalt-molybdenum (CoMo) catalysts to balance the H2 consumption and maintain product specifications throughout the operating cycle. The use of CoMo catalysts also offers better stability toward the end-of-run (EOR): it better promotes direct HDS, so it is less susceptible to high temperature and low-H2 partial pressure (e.g., EOR or near the reactor outlet). In a stacked-bed design, CoMo is the preferred choice if the H2 availability is limited near the reactor outlet (as part of the H2 is consumed along the reactor length).  

Note: There are also scenarios where the use of 100% NiMo is also desirable to maximize the product volume swell (e.g., cheap H2 in natural gas-rich regions). Apart from the catalyst activity and H2 consumption, system robustness must also be considered, as described below. 

By nature, LCO contains more olefinic compounds resulting from fluid catalytic cracking (FCC) processing. Olefin saturation is rapid and can cause local H2 starvation, which leads to subsequent polymerization/coking if the catalyst activity near the reactor inlet is too high. As such, low-activity catalyst should be used first to strategically saturate olefinic compounds. This grading catalyst layer serves the main purpose for particulate and metal trapping, but also to prevent gumming/coking as described. 

Also, both LCO and SRGO are the product cuts next to the heavy residue in the distillation column. In most cases, the wash zone (to clean up the draw-off product by using scrubbing liquid) operates at a relatively low internal reflux rate. As such, both LCO and SRGO can be contaminated with poly nuclear aromatics (PNAs) from the heavy residue portion by either boiling point overlap or entrainment. To handle PNAs, it is preferrable to use catalysts with lower metal content, high metal dispersion and a wider pore mouth near the reactor inlet to handle coke precursors.  

In theory, smaller metal slab (high dispersion) has a lower affinity towards PNA adsorption, and so a lower chance of coking near the pore mouth. In conjunction with a wider pore mouth, it reduces the risk of pore mouth blocking in the subsequent higher-activity catalyst layers (narrower pore diameter, higher surface area) by allowing PNAs to penetrate deep into the center of the wider pore mouth catalyst while being converted into lower aromatics. Even if PNAs undergo aromatics condensation (coking) under certain conditions, coke will deposit deep inside the pellet and still allow more PNAs to enter the pore channel. Note: PNAs have a lower aromatic saturation equilibrium temperature compared with lower carbon number aromatics, so they are more susceptible to aromatics condensation (coking) even at lower temperature (e.g., near reactor inlet). FIG. 3 graphically summarizes a typical catalyst loading strategy for LCO processing.  

FIG. 3. A graphical summary of catalyst loading strategy. 

It must be noted that applying dense loading might be considered to improve the overall activity; however, this is also subject to hydraulic availability, as dense loading can significantly increase the overall pressure drop. 

Proving the challenging operating target. In this case study, catalyst vendors offered various combinations of base metals and stacking strategies. Some vendors proposed 100% NiMo for the maximum HDS/HDA activity, while others used combinations of CoMo/NiMo to balance the H2 consumption and provide more HDS stability over the operating cycle. In total, nine catalyst loading schemes (including the incumbent loading scheme) were compared.  

A conventional catalyst test system will take much longer to finalize the results, as only a few catalyst systems can be tested at the time. With 16 parallel reactors in the authors’ company’s proprietary high throughput catalyst testing systema, the whole study was conducted in less than 2 mos with comparable results to the commercial scale unit. 

The project was executed in accordance with catalyst vendor protocols, with good mass balance and data consistency. For quality assurance and to ensure good repeatability, the authors’ company used duplicate reactors for the catalyst systems. 

The key highlight here is to find the catalyst system that can process the feed blend of 25% LCO (see TABLE 2 for a summary) over the 60-mos catalyst cycle length while remaining within the H2 consumption limit (dictated by the size of the make-up H2 compressor). This amount of LCO is relatively high compared to standard diesel hydrotreating units, since the refinery had never processed this high a percentage of LCO before. This study can also be considered proof of concept—whether the economic drive from the linear programming (LP) planning tool can be implemented in the real world. 

Detailing the results. As discussed earlier, the predicted weight-averaged bed temperatures (WABTs) were based on different kinetic models and assumptions. While some catalyst vendors offered attractive catalyst activity on paper with others being more conservative (e.g., higher WABT), the reality turned out to be different, as demonstrated in FIG. 4. 

FIG. 4. WABT requirements for 8-ppm product sulfur, prediction vs. actual. 

During the proposal stage, Catalyst Scheme F was conservatively estimated with 9°C poorer activity than Catalyst Scheme B. In reality, Catalyst Scheme F was by far the best performer with 25°C better activity compared with Catalyst Scheme B (which is also among the poorest performers). The same observation applies to Catalyst Schemes E and G to a certain extent, where their actual activity is top tier while being underestimated on paper.  

Considerations for catalyst selection. While catalyst activity is an important selection factor, other catalyst aspects should be considered, as they all can play roles in the economic analysis. 

As an example, catalyst schemes with higher HDS activity tend to consume more H2, as supported by chemistry explanations below.  

With LCO processing, the inhibiting nitrogen species become more complex—this is mostly associated with aromatic rings. It must be noted that most of the basic nitrogen species (e.g., quinoline) are adsorbed on FCC catalysts (acidic) and subsequently destroyed in the regeneration process. As such, most of the remaining nitrogen compounds in LCO are non-basic (e.g., carbazoles). 

To accelerate HDS reactions, these nitrogen compounds must first be removed to a low enough level (usually < 10 ppm), which requires a catalyst system with better HDN activity. Since the removal of a nitrogen atom requires pre-saturation of the first aromatic ring (FIG. 5), this implies that the catalyst system with high aromatic saturation activity will also better convert nitrogen into ammonia.  

FIG. 5. Nitrogen removal mechanisms. 

Therefore, stronger overall HDS activity often means higher H2 consumption. This is in line with one of the observations during this case study, where Catalyst Scheme F consumed by far the highest H2 at different fixed temperatures, followed by Catalyst Schemes E and G, as illustrated in FIG. 6. In most cases, H2 is expensive and consumption must be optimized. Additionally, too high H2 consumption can also cause operational troubles if the H2 compression capacity/availability is limited, especially for old units where the processing capacity is often pushed to or already beyond the original design. Therefore, refineries should carefully select the right catalyst system for each unique scenario, considering economic and operational aspects. 

FIG. 6. H2 consumption vs. WABT. 

While processing more LCO and extending the catalyst cycle length are always beneficial, this inevitably requires higher activity catalysts. As discussed earlier, higher activity catalyst schemes often possess higher aromatic saturation activity (to remove nitrogen), which in turn converts more feedstock into the naphtha range by H2 addition.  

Moreover, catalyst manufacturers also include additives like phosphorous or boron to further improve HDS/HDN activity.4 Apart from metal morphology/dispersion, these additives also modify the surface acidity of diesel hydrotreating catalysts and, therefore, catalytic cracking activity. It is important to note that gamma alumina is mildly acidic even without these additives.  

Apart from accelerating HDS/HDA, the removal of organic nitrogen compounds to low levels can enhance catalytic cracking. Theoretically, lower nitrogen levels increase catalytic activity, and some catalyst vendors have noticed ring-opening of naphthenic compounds in certain hydrotreating applications.  

Lastly, thermal cracking can also play a role (to a certain extent), as each catalyst scheme requires a different temperature to reach the product sulfur target. 

Although the diesel density decreases with higher aromatic saturation activity, the density reduction does not always compensate for the loss in diesel mass yield—as seen in this case. FIG. 7 shows that Catalyst Schemes E, F and G, which exhibit some of the highest HDS activity, produced relatively low volumetric diesel yields at fixed temperatures compared to other catalysts. This further indicates that cracking, whether catalytic or thermal, can lead to an overall reduction in volumetric diesel yield. Additionally, the diesel yield difference between catalyst vendors could exceed 1 vol%. 

FIG. 7. Volumetric diesel yield vs. WABT. 

This is often ignored when evaluating diesel hydrotreating catalyst: not only does catalyst activity matter, but also how well the catalyst can preserve the diesel yield.  

It is best to take a comprehensive approach to evaluate the catalyst, considering all key aspects. Ultimately, the choice of catalyst depends on economic incentives, cost and operability. 

Takeaways. Confident selection of diesel hydrotreating catalyst includes the following factors: 

• Without independent catalyst testing, it is extremely difficult to select the right diesel hydrotreating catalyst for any application. Real-world catalyst testing reveals the true catalyst performance and provide refiners reliable data for their economic evaluation. 

• A parallel catalyst testing system also benefits both refiners and catalyst vendors by allowing one catalyst vendor to offer/test more than one catalyst loading scheme. This increases the chance of getting better catalyst loading schemes that better suit the application. 

• The data obtained from this pilot plant testing was also used for kinetic modeling, translating the isothermal to adiabatic operation and gaining further insights into actual catalyst performance.5 

• Independent validation of multiple catalyst systems enabled the refinery to objectively choose the best catalyst. The client performed detailed economic analyses and confidently selected the catalyst system that best fit the strategy for the next cycle. The authors’ company remains in contact with the refinery client to continuously validate and improve pilot vs. commercial unit results. 

 

NOTES 

^a Avantium’s Flowrence system 

ACKNOWLEDGEMENT 

The authors would like to acknowledge the project execution team for their support:  

Theo Groen and Mario Strihavka, Flowrence unit operation; Tom Oosterhoff and Nicholas Roffey, catalyst loading and sample analysis; and Rene Vreuls, Manfred van Zuijlen and Imko Juffermans, analytic setup.  

LITERATURE CITED 

1 Pongboot, N. and T. Vilela, “Common pitfalls in refinery catalyst selection,” The Catalyst Review, October 2022. 

2  https://rds.avantium.com/products/flowrence 

The Authors

Related Articles

From the Archive

Comments