PTQ Q4 2025 Issue

Apparent conversion (%)

Yield (diesel) (%) - HC G Yield (diesel)_BoF (%) - HC G Yield (diesel) (%) - HC H Yield (diesel)_BoF (%) - HC H

Yield (diesel) (%) - HC E Yield (diesel)_BoF (%) - HC E Yield (diesel) (%) - HC F Yield (diesel)_BoF (%) - HC F

Yield (diesel) (%) - HC E Yield (diesel) (%) - HC F Yield (diesel) (%) - HC G Yield (diesel) (%) - HC H

Yield (diesel) (%) - HC A Yield (diesel) (%) - HC B Yield (diesel) (%) - HC C Yield (diesel) (%) - HC D

Figure 5 Selectivity plot for all of the systems (left) and for the most interesting systems (right). On the right, the Yield of Middle distillate is indicated with a solid black border if the yield was calculated based on the mass flow of oil at the inlet of the reactor

Ultimately, the activity plots are created by plotting the N slip against the temperature or the apparent conversion, as shown in Figure 4. The pretreatment reactors were ranked as follows in terms of activity: HT F ≥ HT G >> HT E > HT C ≈ HT B > HT H >> HT D >>> HT A. The hydrocracking reactor activity ranking was: HC G ≈ HC E ≥ HC H > HC F > HC A ≈ HC C > HC B >> HC D. hte units are able to clearly distinguish activity differences above 5°F (3°C), even with- out the use of duplicates. Selectivity The selectivity plot is obtained by plotting the yield of the most valuable cut for the refinery, which is the middle distil - late in this case, against the apparent conversion. The selec- tivity plot for all the catalyst sets tested is reported in the left-hand panel of Figure 5. The selectivity ranking was as follows: HC A ≥ HC D > HC C ≈ HC B ≈ HC H ≈ HC F > HC E > HC G. The shadow below each curve passing through the experimental data represents the confidence interval. These curves are typically used to interpolate selectiv- ity at a given conversion, so it is important to understand the variability of these interpolated values. The confidence interval was calculated using bootstrapping, where the dataset was resampled multiple times to assess statistical variability. This approach allows for more confident com - parisons of selectivity data and helps determine whether differences in selectivity are significant. CITGO engineers were interested in maximising the activity of the catalyst without compromising too much on selectiv- ity due to the previously mentioned temperature limitations. Catalyst sets A and D were the most selective, but the least active by far, and were the least promising of the candidate pool. Catalyst E was the incumbent and ranked among the most active for both PT and HC. Catalysts HC B and HC C are also ranked as less promising, as they are considerably less active than the incumbent catalyst. It is possible to see that only catalyst sets HC F and HC H were very close in cracking activity as compared to the incumbent catalyst.

The ranking among the three remaining systems was made considering the selectivity of the different catalysts. To correctly compare those catalysts, the results were plot- ted both as yield (Yield (Middle distillate)) and yield based on feed (Yield (Middle distillate)_BoF). The yield of middle distillate is calculated as follows: the total mass of middle distillate produced over the total mass of hydrocarbons at the outlet of the reactors, while the yield of middle distil- late based on feed is the mass of middle distillate produced divided by the mass of hydrocarbons at the inlet of the reactor. Yield (Middle distillate)_BoF is affected by the variability of the mass balance, while Yield (Middle distillate) normal- ises the mass balance to 100% for all of the systems. It is possible to see from the right panel of Figure 5 that the ranking in order of selectivity was the same regardless of the type of yield considered. It is clear that HC F and HC H were the most selective catalyst systems, and they were not very different from one another. The close second was HC E, and the least selective among the most active four systems was the catalyst HC G. According to the screening test, HC F was the most promising catalyst thanks to the very high activity of the pretreat and selectivity of cracking. As part of hte’s com - prehensive reporting, detailed analyses of unconverted oil, middle distillate, and naphtha fractions, generated at the most relevant conversion level for each catalyst, are col- lected in the myhte relational database and delivered to CITGO engineers in a fully visualised format. The final deci - sion is made by CITGO using its own modelling tools with the experimental data obtained through this testing. Based on this catalyst benchmarking, CITGO chose the most promising catalyst load and, after the changeout, started the hydrocracker with the selected load. Commercial unit data showed that the start-of-run (SOR) temperature of the commercial unit deviated by less than 5°F from the SOR temperature determined by hte in the pilot plant test.

PTQ Q4 2025