December 2022

Special Focus: Catalysts

Refinery catalyst selection: Facts and fictions every refiner should know

Most refiners carefully evaluate their refining catalysts to ensure maximum benefits over the catalyst cycle length.

Pongboot, N., Avantium; Upienpong, T., PTT Global Chemical

Most refiners carefully evaluate their refining catalysts to ensure maximum benefits over the catalyst cycle length. Without a doubt, pilot plant testing is the best method to evaluate true catalyst performance. Often, the incremental catalyst cost is trivial compared with the economic benefits from improved catalyst performance.

To the authors’ surprise, many refiners only consider testing some refining catalysts, such hydrocracking, fluid catalytic cracking (FCC) or naphtha reforming units. This seems a reasonable line of thought, as these refining units are the critical profit makers of a typical refinery. So, why risk not testing?

Conversely, these refiners somehow believe testing other refining catalysts (e.g., diesel hydrotreating catalysts) is unnecessary. They argue that other refining catalysts are less critical and a paper-based evaluation is usually sufficient. This notion seems sensible on the surface, and many refiners follow this practice. However, is this really the case? Should refiners consider an 86,000-bpd diesel hydrotreater (like the one shown in FIG. 1) as not critical? Where is the line defining the criticality? Is a paper-based evaluation even a valid method? This article will answer these questions and emphasize why refiners should remain skeptical about skipping refinery catalyst testing.

FIG. 1. The diesel hydrotreater is one of the most important-yet-ignored refining units when it comes to refinery catalyst testing.

Reliance on a paper-based catalyst evaluation

Many refiners think that comparing catalyst proposals from different catalyst vendors is a completely valid evaluation technique. However, this is flawed approach.

A typical catalyst proposal is the result of a proprietary kinetic model developed by an individual catalyst vendor. Although most (if not all) kinetic models share the same core principles (e.g., Langmuir-Hinshelwood kinetics), each of them is unique if you look further into the assumptions and the details. Also, these kinetic models are not 100% accurate, so they are likely to yield different results despite the same feed properties.

Secondly, the feed properties provided by the refiner in the catalyst tender documents are mostly incomplete, inconsistent with each other or even over-specified. In the authors’ experience, a set of feed properties is often a mix of data from various sources (e.g., original design data plus past operating data), as shown in TABLE 1.

It is not uncommon to see catalyst vendors making certain feed assumptions (typically based on institutional knowledge) during the feed characterization step (e.g., nitrogen, sulfur or aromatics distribution). A catalyst vendor might assume 35% of total nitrogen is basic nitrogen for Middle Eastern crudes while another uses 1/3 as an assumption. Which number is more accurate? How can we know if all catalyst vendors are adopting the same assumptions? Additionally, a typical crude diet is often a mix of various crudes from all around the world, and not only from a specific region. What if South American or West African crudes are dominant in the crude slate?

To add to the challenges, some catalyst vendors are more aggressive than others with their yield estimates. It must be remembered that a catalyst vendor’s priority is to secure the sale. The authors have experienced cases where it became apparent that a catalyst vendor “tuned” its kinetic model to produce yields for its proposal that were more attractive. For example, a diesel hydrotreating catalyst vendor once proposed an almost nil offgas yield to a refiner without justifying or explaining this anomaly. When challenged, the catalyst vendor just stated that it was the result of the kinetic model. To add insult to injury, the same catalyst vendor also proposed a suspiciously aggressive diesel product density (e.g., a much lower than usual diesel density). Although the same catalyst vendor later revised the new diesel density because the refinery expressed strong disbelief in the performance predictions, the vendor did not make any changes in aromatic contents in the diesel product and overall hydrogen consumption. The illogical revision was clear proof that this catalyst vendor deliberately tuned the catalyst performances presented in its proposal.

Without a reasonable explanation from the catalyst vendor regarding the “too good to be true” catalyst performance, the refiner decided to negatively weigh the results from the abovementioned catalyst vendor due to the uncertainty over the catalyst performance claims, resulting in the other bidders coming out ahead in the catalyst evaluation. However, how did the refiner know the vendor did not offer decent catalysts without pilot plant testing? While the estimates might be suspicious, the catalyst R&D and actual performance might actually prove accurate.

Another example of how vendors may “manipulate” kinetic models is an accidental internal email in a catalyst company that discusses how to alter the original pretreating temperature to make it believable during a hydrocracking catalyst bidding. This catalyst vendor later realized that the refinery contact was mistakenly included in the internal communication, although that contact never mentioned it.

These real stories are perfect examples of how refiners should always challenge the data presented in vendors’ proposals and how paper-based evaluations can be highly subjective.

Inherent flaws in paper-based catalyst evaluations exist and should be avoided, even when selecting naphtha hydrotreating catalysts. The authors have seen a naphtha hydrotreating unit that chronically suffered from off-specification product due to sulfur recombination reactions as a result of lower-than-expected catalyst activity (e.g., the reactor temperature exceeded the olefin saturation equilibrium). In another example, a kerosene hydrotreater with more than 150% of gas plus overhead liquid products—compared to the design (and catalyst specification) and at much lower reactor temperatures than the design—overloaded the top section of the stripper column.

The myth of hydrocracking catalyst selection

It is vital to test hydrocracking catalysts to reduce the risks of selecting a suboptimal hydrocracking catalyst package. As this major conversion unit is a significant profit center, a slight performance gap can easily cause multi-million-dollar losses. Nonetheless, the risk level associated with hydrocracking catalyst selection is not necessarily higher than other refining catalysts, as discussed in the following sections.

Clear distinctions exist between good and average/poor hydrocracking catalysts. As demonstrated in FIG. 2, catalyst Package C is clearly the best by both paper-based and pilot plant-based evaluation. While Package E seems a relatively weak second-best on paper, it performed well and almost won in the pilot plant testing—a $10 MM/cycle benefit gap, although seemingly significant, is considered relatively close in hydrocracking. According to the authors’ survey, many refiners agree that the catalyst vendors that designed catalyst Packages C and E are technically fine based on past operating experiences.

FIG. 2. Relative cycle benefit of various hydrocracking catalyst packages for a 54,000-bpd hydrocracker. Catalyst Package D was a baseline for a paper-based evaluation, while catalyst Package B took the same role in pilot plant testing.

Conversely, catalyst Packages A, B and D are inferior on both paper and pilot plant testing. The only difference is that while catalyst Package D offered the lowest cycle benefit on paper, it exhibited equivalent or better potential in pilot plant testing when compared to catalyst Packages A and B.

These results are supported by technical factors. One is the choice of metals. Principally, hydrocracking catalysts are bi-functional, comprising cracking and hydrogenation functions. For the cracking function, various blending ratios of amorphous silica-alumina (lower acidity/activity) and zeolites (higher acidity/activity) are possible. Concurrently, several metal choices with different hydrogenation strengths are also available, as summarized in TABLE 2. Catalyst manufacturers can adjust the ratio between the catalyst’s cracking and hydrogenation strength to optimize overall catalyst activity and selectivity. A higher hydrogenation-to-acidity strength ratio results in better middle distillate yield, while a lower ratio produces more naphtha.1

In this example, catalyst Package B utilized nickel-molybdenum (Ni-Mo) as the hydrogenation function for the proposed hydrocracking catalysts, while others relied on nickel-tungsten (Ni-W). As Ni-W is stronger than Ni-Mo when it comes to hydrogenation, Ni-W-based hydrocracking catalysts can better preserve primary cracking products (e.g., kerosene and diesel) by suppressing secondary cracking, assuming the same type of zeolite and zeolitic content, as illustrated in FIG. 3. This observation is in line with the fact that catalyst Package B achieved the lowest middle distillate yield during the pilot plant testing, among other factors. Additionally, hydrocracking catalysts with a stronger hydrogenation function help improve product qualities [e.g., smoke point, cetane number and unconverted oil (UCO) viscosity index].

FIG. 3. Typical hydrocracking reaction pathways: both metal and acid sites are essential to complete the hydrocracking process.

Conversely, the top performer, the supplier of catalyst Package C, claimed to have a high degree of mesoporosity in the zeolite framework (FIG. 4). This structural advantage aids molecular diffusion and helps reduce the degree of secondary cracking (i.e., the primary cracking products are released more rapidly from the zeolite network).2 Additionally, this catalyst manufacturer employs a specially engineered catalyst shape where the diffusion path length is shorter than conventional ones, therefore minimizing over-cracking. Lastly, this competitor utilizes stronger Ni-W as the hydrogenation function for its hydrocracking catalysts. These described catalytic features partially explain why the company outperformed other hydrocracking catalyst manufacturers in terms of the total liquid and middle distillate yields.

FIG. 4. Mesoporization of the zeolite structure can improve the acid site accessibility and reduce the degree of over-cracking in heavy oil hydrocracking by shortening the diffusion path length within the zeolite structure. Source: Zeopore Technologies.

Although this is just one example, the authors have seen several hydrocracking catalyst evaluations with similar results (i.e., the winner on paper is also the winner in pilot plant testing). In other words, selecting catalyst Package C over Package E is simply choosing a very good catalyst package over an almost-as-good catalyst package. In truth, selecting catalyst Package E over Package C for alternative reasons is still not a bad choice in terms of economics.

In most cases, catalyst vendors must guarantee key performance parameters (e.g., product yield or hydrogen consumption). As such, it would be risky for them to over-exaggerate their proposal just to win the contract. When the catalyst vendors know that the catalyst evaluation is pilot plant-based, it provides more incentive to be more realistic with the yield estimates in their proposals.

For instance, if the middle distillate yield based on the original estimation is 70 wt% for a middle distillate selective application, it can be almost certain that no catalyst vendor would aggressively put 75 wt% in the catalyst proposal—it would obviously fail during the pilot plant testing or performance test run, and random excuses will be insufficient. In the authors’ experience, the middle distillate yield gap between “the best” and “the worst” hydrocracking catalyst in the market could be greater than 5 wt%. Therefore, the hydrocracking catalyst price tends to play a minor role in the hydrocracking catalyst evaluation.

For these reasons, the risks associated with hydrocracking catalyst selection are not necessarily as high as perceived. At a minimum, this prevents the selection of poor hydrocracking catalysts. Nonetheless, pilot plant testing is still the best method to evaluate hydrocracking catalysts, as the gap between the best and second-best hydrocracking catalyst can be significant. More importantly, it is also possible that the catalyst ranking might change if the catalyst performances are close (i.e., catalyst Packages C and E in FIG. 2).

Higher-activity pretreating catalysts improve the middle distillate selectivity of hydrocrackers

In principle, an overall conversion is a result of both pretreating (hydrotreating) and hydrocracking reactions. The primary focus of the pretreating section is to lower the organic nitrogen concentration to an acceptable level (typically < 50 ppm for modern designs) to prevent nitrogen inhibition effects in the hydrocracking section. In parallel, the pretreating reactor also removes other contaminants like metals, sulfur, oxygen and halides, as well as saturates olefins and aromatics (FIG. 5).

FIG. 5. Reaction classes in a typical hydrocracking process.

These hydrotreating reactions reduce the boiling point of the original hydrocarbon molecules, also known as pretreating conversion (FIG. 6). The gross pretreating conversion can be up to 30% for modern pretreating catalysts with a relatively low nitrogen slip target (e.g., < 10 ppm). In other words, the hydrocracking conversion could be as low as 69% for a fixed overall conversion of 99%.

FIG. 6. A demonstration of how heteroatom removals and hydrocarbon saturations reduce hydrocarbon boiling points. In this case, a di-benzothiophene (DBT) molecule is hydro-treated and ends up as bi-cyclohexane (BCH).

Targeting a higher pretreating conversion improves product selectivity, especially middle distillates, as pretreating catalysts are typically much more middle distillate selective than hydrocracking catalysts. A higher-activity pretreating catalyst saturates more aromatic compounds (increased hydrogen consumption), thus higher pretreating conversion and better overall middle distillate selectivity, assuming the same nitrogen slip target. With a lower pretreating temperature, the extent of thermal cracking (which results in poor liquid yields) will also be reduced.

Additionally, higher-activity pretreating catalysts will permit refiners to operate at the lower nitrogen slip level, allowing the addition of more middle distillate selective hydrocracking catalysts (lower zeolitic content).

Unfortunately, this best practice is not always the case. The authors once tried adding a small layer of ultra-high activity trimetallic unsupported catalyst (Ni-Mo-W) into a pretreating reactor to improve the overall middle distillate yield, as there was excess hydrogen available then. According to the catalyst supplier, this modification should enhance the overall middle distillate yield by another 0.5 vol%. As this unsupported catalyst had just been recently launched to the market and had few references, the authors requested the catalyst supplier to conduct pilot plant testing to confirm the benefit claims.

In contrast with the already explained conversion balance principle, the pilot plant testing campaign revealed that the trimetallic unsupported catalyst increased light ends production, thus reducing the overall middle distillate yield by 0.6 vol%. Although it also lowered kerosene and diesel density, the density improvement was nonetheless insufficient to compensate for the middle distillate weight yield loss. The catalyst vendor offered no clear explanation for this phenomenon—perhaps the vendor did not know, or it did not want to reveal flaws in its newly launched catalyst.

One hypothesis for this phenomenon is that the trimetallic unsupported catalyst developed by the concerned catalyst supplier promoted excessive C-C bond cleavage by metal components, also known as hydrogenolysis. This metal-catalyzed hydrogenolysis mechanism results in high C1 and C2 hydrocarbons yields, along with light n-paraffins and a near absence of iso-paraffins.

Fast forward to the subsequent bidding: this catalyst supplier decided not to include this trimetallic catalyst in the offered loading scheme, but returned to a conventional Ni-Mo hydrotreating catalyst. This implies a performance trade-off in the supposedly high-performing trimetallic hydrotreating catalyst. Otherwise, the supplier would have continued to propose it in the following cycle to gain performance advantages in the bidding.

“There is no need to test diesel hydrotreating catalysts.”

The authors actually heard a diesel hydrotreating catalyst vendor say this after a refiner suggested that pilot plant testing might be used to better evaluate diesel hydrotreating catalysts in the next bid process.

The vendor’s reasoning was that such a “less critical refining unit” did not require pilot plant testing. The authors disagree with this catalyst vendor and question whether they will lose a bid again in a fair process.

In contrast with hydrocracking units, diesel hydrotreaters produce much lower light-ends yields (gas and naphtha), mainly from hydrotreating reactions of sulfur, nitrogen, olefin and aromatic compounds (see the previous section). In diesel hydrotreating applications, hydrocracking reactions are negligible unless dewaxing catalysts (zeolitic-based) are used for cloud point reduction.

Hydrotreating conversion is less than 3 wt% without dewaxing, although a decently cut straight-run feed (e.g., from a crude distillation column) would easily achieve > 98 wt% diesel yield. This is one reason the selectivity of diesel hydrotreating catalysts is rarely discussed. A more popular discussion topic is the optimal ratio between Co-Mo and Ni-Mo catalysts and how to stack them in a reactor.

It is difficult for refiners to accurately quantify the light-ends yields in an operating environment due to inherent flow measurement errors. Naphtha flow is only a tiny fraction of total product flow, while offgas flow is notoriously inaccurate due to varying gas composition. It is not uncommon to see a raw mass feed flowrate with a flow reading less than that of the diesel product.

As discussed previously, the diesel yield usually lies somewhere between 98 wt% and 100 wt% for most straight-run diesel hydrotreaters, which is actually less than the possible orifice flowmeter error of +/– 3 wt%–5 wt%. Unsurprisingly, an error in diesel flow measurement could easily mask a lower-than-expected diesel yield. In the authors’ experience, some catalyst vendors have exploited this loophole to benefit them in catalyst biddings by proposing an unrealistic diesel yield. The authors once saw a catalyst vendor that boldly proposed 99.9 wt% diesel yield for ultra-low sulfur diesel (ULSD) operation with reasonable sulfur and aromatic contents in the feed.

Another popular excuse is the actual feed quality, which is never exactly the same as what is in the bid tender—catalyst vendors can always argue that the performance gaps are from deviations in feed qualities. Any deviations in product properties (e.g., diesel density) will be disputed.

In many cases, it is difficult for refiners to claim any penalties (usually not worth it, compared with potential damages) due to a failed unit test run. The authors have seen a catalyst vendor that failed every performance test run over 6 yr that was selected repeatedly just because its customer relied on a paper-based evaluation to choose diesel hydrotreating catalysts.

For the described reasons, diesel hydrotreating catalyst selection is no less critical than other refining catalysts; it is difficult to distinguish the best catalyst package from the average/poor ones without pilot plant testing due to different kinetic models used in performance predictions and the potential for the catalyst vendor sales team to over-exaggerate the performance of their catalysts. Rather than differentiating the performances of various catalysts, refiners may potentially be wasting their time trying to identify frequently unmeasurable/debatable gaps in the operating environment.

According to Hoekstra Trading LLC, a one-tier improvement in performance or a 5°C reduction in the start-of-run temperature is worth $1 MM/yr–$5 MM/yr in a typical ULSD unit (TABLE 3). A lower start-of-run temperature means a higher federate, the ability to use more difficult (cheaper) feeds in the ULSD unit, and lower light gas yields from less severe thermal cracking. The benefits of a better diesel hydrotreating catalyst package, especially for a larger unit, easily outweigh the service fee of an independent pilot plant testing program. Generally, the service fee for diesel hydrotreating applications is only half of that for hydrocracking catalyst testing.

Naphtha reforming is a monopoly business. The authors have seen one company have repeated success in pilot plant testing, so it could be assumed that pilot plant testing is unnecessary since every benchmarking campaign the authors have seen by this company had the same result. This might indicate that this company is ahead of its competitors regarding naphtha reforming catalyst development, but an expert in refinery catalyst testing detailed cases where other naphtha reforming catalyst suppliers also had successful pilot plant testing campaigns. Naphtha reforming catalysts perform differently with varying feeds, operating conditions and product specification requirements.

Importantly, all catalyst suppliers continue to develop their products, and there is no guarantee that today’s prominent market leaders will retain their position. Pilot plant testing is still highly recommended for the evaluation and selection of the best naphtha reforming catalyst to suit refiners’ applications.


While both conventional and high-throughput approaches can be used to distinguish the best-performing refinery catalyst from average ones, the high-throughput technique offers some great advantages over the conventional method in several circumstances.3

The author’s company’s high-throughput platforma with 16 parallel reactors produces consistently high data quality in terms of repeatability, reproducibility, scalability and mass balance quality, thanks to the company’s patented gas distribution technology, active liquid distribution (ALD) system and single-pellet string reactor (SPSR) catalyst loading approach. Additionally, the required catalyst volume is less than 1 mL. As such, the required catalyst and feed sample volumes are minimal, making the platforma one of the most cost-effective commercial catalyst testing platforms available. HP


a Avantium’s Flowrence® platform


The authors would like to thank Tiago Vilela from Avantium and George Hoekstra from Hoekstra Trading LLC for useful insight and information; Shaun Dyke from PetroQuantum for editing and helpful comments; Kurt du Mong from Zeopore Technologies for an image; and Kongphop Boonwong from GlobalR&D for final editing.


  1. Scherzer, J. and A. J. Gruia, Hydrocracking science and technology, 1st Ed., CRC Press, Boca Raton, Florida, 1996.
  2. Garcia-Martinez, J., K. Li and M. Davis, Mesoporous zeolites: Preparation, characterization and applications, Wiley, 2015.
  3. Pongboot, N., T. Upienpong and T. Karunkeyoon, “Hydroprocessing catalyst selection—Part 1: Planning and selecting the catalyst evaluation method,” Hydrocarbon Processing, May 2022.

The Authors

Related Articles

From the Archive



{{ error }}
{{ comment.comment.Name }} • {{ comment.timeAgo }}
{{ comment.comment.Text }}