Catalysis 2026 Issue

Catalyst post-audit predict case setup

Model-corrected catalyst post-audit evaluation results

Average delta Denali Action over Action during all periods

Group

Independent Incumbent

Incumbent

Denali Action

Denali Action Denali Action

variables

FCC User std. conversion by vol, vol% C₃+ volume expansion, vol%

Calibration

Incumbent

Denali Action

Incumbent

1.7 0.2

factors

average

Delta coke, wt%

-0.060

Table 4

Activity, wt%

-0.6

Component net product yields H₂+C1+C₂+C₂0, wt%

-0.2

Prior to performing any model work, the observed yield shifts to Denali Action are summarised in Table 1 . The higher conversion and lower C 4 olefin yield are initially counter to the anticipated benefits. Reflecting upon the changes in independent variable values (see Table 2 ) shows substan- tial changes in feed rate, ZSM-5 additive addition rate, pur- chased E-cat activity, addition rate, and E-cat metals levels across the two catalyst periods. These differences account for the higher observed conversion and lower LPG olefins yields and olefinicity. Since fresh and purchased E-cat are added simultaneously, fluidised standard test (FST) yields, a bench-scaled test measuring the full yield profile of catalyst over a standard feed at constant reaction severity, were examined for both the combined E-cat and the purchased E-cat separately, as shown in Table 3 . On the surface, the combined E-cat (consisting of Denali Action and the purchased E-cat) shows higher conversion and lower LPG olefins. When only the purchased E-cat is tested, significant differences, including higher conversion and lower ZSM-5 activity, are evident. While FST testing helps account for some operating impacts, it alone is not sufficient to completely evaluate catalyst performance, especially in situations where a blend of fresh and purchased E-cat is added to the unit daily. The kinetic model can isolate catalyst effects from the pro- cess variable and purchased E-cat addition rate changes. The first step is to collect as many relevant cases as possible for each catalyst period. Calibration factors are reviewed, and any cases showing unexplainable deviations from the others are excluded. After verifying no mechanical changes across the periods, the differences in calibration factors can be considered to reflect only catalyst formulation changes, and each period’s factors are averaged. Four groups of predictions are run. Groups 1 and 4, shown in Table 4, are re-predicts of calibration cases using the aver- aged calibration factors. Their results should closely match the calibration results and serve as a consistency check. Groups 2 and 3 are cross-predicts, where the calibration factors from the opposite period are applied to the independent variable inputs. Group 2 predicts what would have occurred if Denali Action catalyst had been added during the incumbent period. Group 3 predictions are what would have occurred had the decision been made to remain on the incumbent catalyst. This methodology delineates the effects of catalyst for- mulation changes from feed and process variable changes. The benefit of switching to the new catalyst is calculated by averaging the difference in results seen between Groups 2 and 1 and Groups 4 and 3.

Propene, vol% Iso-butane, vol%

1.5 0.3 0.7

Total C₄ olefins, vol%

C₅ standard naphtha, vol%

-0.9 -0.4 -1.3

Standard cycle oil

Standard FCC bottoms, vol% C₅ standard naphtha octane RONC

0.6 0.3

MONC

Table 5

The model-corrected results show the anticipated bene- fits: increased LPG olefin yield and improved delta coke (see Table 5 ). The higher conversion and lower bottoms cracking leave opportunity to lower the catalyst addition rate or riser temperature to maintain conversion at the previous levels. Conclusion Kinetic FCC models are powerful tools that enable engineers to make more informed and insightful decisions and optimise FCC unit performance. Increased confidence in model results is obtained through data consistency checks, such as eval- uating the hydrogen in coke and the calculated net heat of reaction. Prior to running predictions, it is essential to vali- date that the model re-predicts its own calibration. Only validated models with known pedigrees should be used. Unnecessary complexity should be avoided. Data qual- ity should be reviewed, and results compared back to inputs to ensure intended values are maintained. Best practices for catalyst addition rate and activity inputs, coupled with veri- fication of purchased E-cat predictive capability, help ensure the model accurately captures catalyst behaviour. Frequent calibration and systematic review of calibration factors are critical in maintaining model performance. Adherence to these best practices allows the model to be confidently used for sophisticated analysis, such as catalyst post-audits. The example presented in this article demon- strates how cooperation between the catalyst supplier and the refiner delivered the expected delta coke and LPG yield benefits despite the obfuscation caused by many simulta - neously changing feed and process variables. By following these best practices and methods, refiners can leverage kinetic FCC models to make more confident, data-driven operational, and strategic decisions.

DENALI ACTION and ACTION are marks of Ketjen. Alan Kramer is FCC Modelling Senior Advisor at Ketjen.

Patrick McSorley is Senior FCC Technical Service Consultant at Ketjen. Bridget Cadigan is a chemical engineer at Marathon Garyville Refinery.

Catalysis 2026