(412c) Some Uses and Misuses of FCC Catalyst Testing Experimental Data

The objective of this work is to develop detailed models of state-of-the-art catalyst evaluation procedures for the Fluid Catalytic Cracking (FCC) process. The key properties desired of the process models are that they can be practically implemented in everyday catalyst evaluation procedures, that they use the same reaction kinetics network and catalyst deactivation functions for different catalyst testing reactors (SCT-MAT, ACE unit, and DCR unit) and that they can provide theoretical insights to decoupling catalyst activity and selectivity from the design and hydrodynamic regime of the various laboratory units used for FCC testing. Experimental data from different laboratory catalyst testing units (SCT-MAT, ACE and DCR) are used for model development and validation.

Typically, new and promising catalysts are tested in batch laboratory-scale reactors, such as the Short-Contact-Time Micro-Activity-Testing (SCT-MAT) unit and the ACE unit. Catalyst testing is performed using a few representative feedstocks and similar operating conditions to get an initial screening of catalyst activity and selectivity. This initial screening is then used for estimating the potential benefits of using the catalyst in a commercial riser reactor by:

- assuming that catalyst selectivity will remain the same in the commercial riser,

- using kinetic constants calculated from modeling the batch laboratory reactor using the kinetic model of a continuous reactor to predict the selectivity and activity in a commercial riser,

- performing experiments in pilot-scale riser reactors using the catalyst that appears most promising from the initial batch reactor experiments,

- using all the available experimental data in unit-specific models capable of scaling-up catalyst selectivity.

Of course, the last of the aforementioned approaches is the appropriate way of analyzing catalyst testing data. However, this type of detailed analysis is not always easy. Catalyst evaluation experiments can be analyzed with consistency, only if one uses models that capture the peculiarities of the catalyst testing reactor used while using the same kinetic and catalyst decay constants for all different reactors. Fixed bed reactors, such as the SCT-MAT unit, resemble the plug flow conditions of commercial risers, but in a time averaging way: layers of catalyst come in contact with the product of gradually cracked feedstock and coke can still be formed on the inactive diluent used to establish a standard bed volume. Similar modeling challenges exist in the modeling of the ACE unit: modeling of the today's state-of-the-art unit in catalyst evaluation is not an easy task. The ACE unit operates as a fluid bed in regimes ranging from bubbling to turbulent or even slugging. It can be modeled as a spouted bed reactor [1], but this involves a large uncertainty in the determination of the contact efficiency between catalyst and feedstock inside the unit. Moreover, several different protocols are applied to its operation. In the original protocol [2] the catalyst inventory is kept constant and the experiment time is the only operational parameter changing. This approach has the advantage that fluid dynamic properties of the unit are kept unchanged between experiments at different C/O ratios, but has the disadvantage that the space time is always the same. Such a protocol is significantly different than the actual operation of riser reactors. A different protocol suggests changing the catalyst inventory, which should resemble riser conditions but has an effect on the hydrodynamics of the unit, hence the contact efficiency. Using a diluent creates a better hydrodynamic profile in the reactor, but again involves uncertainties coming from the difference in density between catalyst and quartz particles and the possibility of thermal coke production on the surface of the diluent. Nonetheless, catalyst activity and selectivity are catalyst properties and should the same regardless of the protocol applied; their profiles should of course be different depending on the hydrodynamic regime of the reactor. The challenge in analyzing and interpreting catalyst testing experiments performed in the ACE unit involves accurate modeling of the hydrodynamics of the unit. Similar considerations exist in modeling pilot riser reactors. The small diameter of these reactors impacts the slip between gas and catalyst creating an annular flow regime inside the reactor, which again has a significant effect in the contact efficiency between catalyst and reactants [3].

The objective of this paper is to model different reactors and the different protocols applied in their operation using the same kinetic constants for a given catalyst and feedstock. This can be achieved by decoupling the catalytic results from the hydrodynamics of each process. The first and most important difference between a riser and a batch reactor is that in a batch reactor one needs to calculate the time averaging integral:

where y_iis the yield to each product and t_c is the catalyst residence time in the reactor. As a first approach to demonstrate this time averaging effect we assume that the 3-lump model of Weekman and Nace [4] can simulate accurately the results of a batch FCC reactor.

Figure 1: The 3-lump model of Weekman and Nace.

The problem to be studied here is the possibility of a fixed bed reactor to give different rankings for the gasoline selectivity of two catalysts with the same activity (K₁=22.9 as in Ref. [4]) but different gasoline selectivity constants (K₁₂=5-22, K₂=0.1-5). In Figure 2(a) the effect of kinetic constants describing gasoline production and consumption at 70wt% conversion is presented. The vertical axis shows the difference between the model prediction for a fixed bed reactor and a riser. The difference between the two shows a maximum of ~2.5wt% of gasoline yield within the range of K₁₂ and K₂ studied here. Figure 2(b) presents the typical plot of gasoline vs. conversion for two hypothetical catalysts. Again catalyst activity is assumed the same (K₁=22.9 and identical catalyst deactivation functions for both catalysts) and gasoline selectivity constants are different (solid lines: K₁₂=18, K₂=2.75; dashed lines: K₁₂=21, K₂=5). Blue lines illustrate model predictions for the riser reactor and red lines show predictions for a fixed bed reactor.    

Figure 2: Effect of kinetic constants related to gasoline production (K₁₂) or consumption (K₂) on gasoline selectivity in a fixed bed unit and a riser reactor.

Figure 2(b) shows that in the riser reactor the maximum of gasoline selectivity is estimated to occur at different conversions for the two catalysts but the maximum gasoline yield is practically the same. Depending on the operation profile of the commercial unit in which the catalysts will be used, Catalyst A may be a promising catalyst (if for instance diesel production at maximum gasoline selectivity is of interest). On the other hand, in a batch reactor the same two catalysts behave differently. The maximum gasoline yield appears much higher for Catalyst B, making this catalyst a clear winner in catalyst testing. Keep in mind that these results assume that the simple 3-lump model can describe these two catalysts perfectly, experimental error has not been considered and activity is the same for both catalysts. Relaxing all these assumptions (in a real catalyst evaluation analysis) would make catalyst evaluation an extremely difficult task that might lead to wrong conclusions about the catalyst ranking.

In this presentation the results of applying a more detailed kinetic model will be presented and the effect of the aforementioned assumptions regarding the impact of unit operating profile on catalyst selectivity will be identified. The effect of catalyst deactivation profiles caused by the different operating profiles and contact efficiency inside different types of catalyst testing units will be explored. Experimental data from all three different types of reactors using the same catalyst and feedstock will be used to illustrate the accuracy of the proposed models and identify the pitfalls that can be caused by the usual over-simplifications applied to every-day catalyst testing procedures. This work leads to a series of practical tools for analyzing catalyst evaluation experiments, identifying reactor-independent kinetic constants (that include the effect of catalyst on selectivity) and scaling-up catalyst performance from laboratory units to riser reactors.  The pitfalls of translating catalyst performance measured in batch laboratory reactors to catalyst activity and selectivity of commercial FCC risers will be examined and rigorous models for interpreting laboratory batch data will be proposed.

References

1. Vieira RC, Pinto JC, Biscaia Jr EC, Baptista CMLA, Cerqueira HS. Simulation of catalytic cracking in a fixed-fluidized-bed unit. Industrial and Engineering Chemistry Research. 2004;43:6027-6034.

2. Kayser Technology, Inc. http://kaysertech.com/downloads.htm.

3. Bollas GM, Vasalos IA, Lappas AA, Iatridis D. Modeling small-diameter FCC riser reactors. A hydrodynamic and kinetic approach. Industrial and Engineering Chemistry Research. 2002;41:5410-5419.

4. Weekman Jr VM, Nace DM. Kinetics of Catalytic Cracking Selectivity in Fixed, Moving, and Fluid Bed Reactors. AIChE Journal. 1970;16:397-404.