Catalysis 2026 Issue

affect hydrogen transfer and should not be included in the analysis. Figure 1 can be used to estimate a likely net heat of reac- tion based on the conversion and the RE content of the cat- alyst. An upward adjustment should be applied when using more than 10% ZSM-5 additive. This estimated value can be compared against the model-calculated net heat of reaction. Differences greater than 10% from the estimate indicate an inconsistency that should be investigated. After the mass balance and hydrogen in coke are confirmed, air rate and feed temperature values should be reviewed next, as they exert the greatest influence on the heat balance. Model usage best practices Shifting focus from process data to model usage, another common source of error is incorrect data entry into the model. These errors are often difficult to trace. They may occur when the user cannot locate a desired variable in a process stream or meter object and subsequently adds a similar var - iable. This can create a conflict, resulting in the model using an unintended value during calibrations or predictions. Another common scenario arises during what-if evalu - ations, when the user overwrites the data historian linked with a constant. If the model file is saved afterwards, that overwritten constant persists the next time the model is opened and executed. Prior to running any predict cases, a recommended best practice is to begin with the model’s calibration case and run a prediction using the same independent variable values and calibration factors as in the calibrated case. The results from this prediction should exactly match the calibration results. If this does not happen, it is a clear signal that an overriding value is present within the model, and the user must rectify this before proceeding with model usage. When reviewing data fed to a kinetic model, consider whether the input value is directly measured or calculated. Some parameters, such as catalyst addition rate, may be based on a totaliser calculation that may not be aligned with the data historian’s averaging period. A mismatch between totaliser reset time, calculation time, and calibration time may result in model inputs that deviate significantly from their true values. Many refiners lack control mechanisms for managing kinetic models. Unit engineers often inherit models from predecessors or find them at random network locations. These uncontrolled models, lacking formal review, cannot be assured to function properly. They may be obsolete with respect to either process configuration or data historian mapping. In some cases, the creator of the model and their overall experience level are unknown, or the model may have been developed for a narrow focus and may never have been validated for use outside that limited scope. Implementing a control process covering model review, validation, approval, versioning, and distribution is easily jus - tified when weighed against the costs of making incorrect decisions based on the recommendation of obsolete, broken, or otherwise untrustworthy models. When considering model evolution and version control, it is important to consider that not all users will be fully trained

240

220

200

180

160

Hydrogen transfer

140

120

100

Conversion

calculation. Feed rate is easily validated by the mass bal- ance; remaining riser-side sources of error often stem from incorrect feed temperature values used in the model. Heat to steam is the next largest riser heat consumer. Although smaller than the heat absorbed by vaporising and heating the feed, errors here can still affect the heat balance. The most common error found is incorrect units of measure applied to the steam flow rates. The remaining riser heat consumers are generally small enough that measurement issues have little effect on the calculated heat of reaction, except when incorrect units of measure are used. The net heat of reaction cannot be directly measured in the FCC; rather it is calculated by subtracting from the total heat of combustion all the heat consumers present in the regenerator(s) and riser(s). For this reason, the calculated net heat of reaction is perhaps the most important single-quan- tity indicator of data consistency on an FCC unit, where the calculated net heat of reaction only makes sense when all the quantities in the heat balance are reasonable. There are many simultaneous reactions occurring in the riser, with the most significant being cracking and hydrogen transfer. The net heat of reaction represents the combined effects of all reactions. Cracking is endothermic, as apparent by the observation that temperature decreases rapidly as cracking proceeds up the riser. Figure 1 shows that increases in conversion, accomplished by driving riser severity, increase the absolute value of the heat of reaction. This is because smaller mol- ecules need more energy to crack. ZSM-5 zeolites, which crack naphtha-range olefins into LPG olefins, require the greatest heat. Units utilising large concentrations of ZSM-5 may exhibit apparent net heats of reaction exceeding 200 Btu/lb (110 kcal/kg). Hydrogen transfer is exothermic. The rate of hydrogen transfer cannot be directly measured, but it is evident in the product distribution. Catalyst formulation influences the rel - ative amounts of cracking and hydrogen transfer. Rare earth (RE) elements (primarily lanthanum) are used to stabilise Y-zeolites and increase their activity. However, they also increase hydrogen transfer. Low RE catalysts (~0.5 wt% RE) exhibit the least hydrogen transfer capability. Catalysts with more than 3% RE are considered high in hydrogen trans - fer. Rare earths from SOx additives and metals traps do not Figure 1 Absolute net heat of reaction as functions of con- version, hydrogen transfer (catalyst RE), and ZSM-5

Catalysis 2026