PTQ Q3 2025 Issue

of the reactor, respectively. Afterwards, the graphical form of the Arrhenius equa- tion was used to retrieve both the apparent activation energy (Ea) and the loga- rithm of the pre-exponential factor (ln(A)). The measured k(N), represented by the symbols with the black bor- der in Figure 4c , were then plugged into the Arrhenius equation (k = A e -(Ea/RT) ) to obtain predicted tempera- ture values for each k(N) data point. The black/grey bars represent one, two, and three standard deviations from the mean. This exercise demon- strated that all values fell within ±1.5°C of the exper- imental temperature, which corresponds to three stand- ard deviations or less at every nitrogen concentration. The reason why this exercise was particularly important is that three standard deviations from the mean encompass 99.7% of the data points in a normal distribution, and such an interval corresponds to the total experimental error that can be expected for this type of data. In this test, the first 10 days were used to evaluate the pretreatment response to temperature changes. Subsequently, an N slip of

1: PT1 9: PT1 11: PT1 3: PT1

100

1: PT1 9: PT1 11: PT1 3: PT1

375

380

385

375

380

385

T_Rx (˚C)

8.0

7.5

7.0

6.5

6.0

5.5

5.0

374

376

378

380

382

384

386

388

Reaction temperature (˚C)

Figure 4 PT 1 evaluation: A: c(N) vs T_Rx, B: k(N) vs T_Rx, C: measured vs calculated k(N) vs T_Rx. The solid blue lines in panels A and B represent the average of the data points, and the shaded contours represent the confidence interval reported as standard deviations: one (dark blue), two (medium blue), and three (light blue) standard deviations. In panel C, the symbols with a solid black border represent the experimental data points, while the symbol without a border represents the data points calculated from the Arrhenius equation

Three different pretreatment catalysts were studied, but only PT1 has a dataset that is statistically meaningful. PT2 and PT3 were loaded only twice, so their average and standard deviations are not meaningful. Despite this, the activity of PT2 is clearly and significantly lower than the activity of PT3 and PT1. PT1 seems to be more active than PT3, but more repeats would be necessary to confirm this observation with greater certainty. To evaluate the error in terms of temperature, the con- centration of N for PT1 was transformed into the respective kinetic rate constant, as shown in Figure 4a and b , using the following equation:

10 ppm was targeted and achieved at a TOS of 10 days. Once the target N slip was reached, the hydrocracking sys - tems took five to seven days to line out. The SIMDIST was used to evaluate the yield of the prod - ucts of interest. In this case, the focus was the yield to middle distillate (MD), which included all hydrocarbon com - ponents boiling between 145°C and 375°C. Components boiling above 375°C were considered unconverted oil (more than 375 UCO). The performance of the catalysts in terms of yield to middle distillate and activity are shown in Figure 5 . PT1+HC1 is more active (panel B), but less selective, than PT1+HC2 (panel A). As both systems were loaded four times in different reactors using different heat- ers, a statistical evaluation was possible. In both plots, the standard deviation was reported either as bars or as a shadowed area around the mean. In terms of yield, the two systems are indistinguishable at conversions below 60%. They begin to show differences at

where k(N) is the apparent kinetic rate constant in h - 1, LHSV is the liquid hourly space velocity in h-1 and c(N) Feed and c(N) TLP are the concentrations of N at the inlet and outlet

104

PTQ Q3 2025