(99b) CFP Regenerator Model Development

Catalytic Fast Pyrolysis (CFP) is a promising technology for producing fungible transportation fuels from biomass. Just as in petroleum resid or vacuum gas oil (VGO) cracking in FCC units, carbonaceous coke deposits on the catalyst, and the heat from combusting the coke in a regenerator is used to drive the endothermic cracking reactions. However, biogenically derived coke is different from VGO and resid coke, and design of oxidative regenerators and regeneration processes must take these differences into account.

In coordination with the ChemCatBio, Advanced Catalyst Synthesis and Characterization (ACSC) and Catalyst Deactivation Mitigation (CDM) BETO consortia, the Consortium for Computational Science and Chemistry (CCPC) has been investigating the kinetics of coke generated on ZSM-5 based CFP catalysts and using these kinetics to build industrial scale reactor models. Two different ZSM-5 based catalysts have been investigated: a Geldart B (nominal 600-800 mm) coked to relatively high levels (12-15 wt% coke on catalyst), as would occur in a bubbling bed CFP reactor, and a Geldart A (FCC-type, nominal 80 mm) catalyst coked to much lower levels (< 2% coke on catalyst), as would occur in an FCC-type CFP reactor operating at typical catalyst circulation rates.

Kinetics model development was based on Temperature Programmed Oxidation (TPO) data generated at Pacific Northwest National Lab (PNNL), complimented by advanced microscopic characterization at Oak Ridge National Lab (ORNL). Fixed bed finite element computational models were used to assess mass and heat transfer limitations in the TPO experiments and incorporate learnings from characterization such as intrapellet coke deposition patterns. The fixed bed models were also used to test an ensemble of kinetic hypotheses including direct surface oxidation of C and H to form CO, CO₂ and H₂O (Eley-Rideal mechanisms) and homogeneous gas phase oxidation of CO to CO₂.

From the ensemble of models, the ones best fitting the experimental data were down selected, incorporated into the MFiX software suite and used to simulate a 10 ton per day (TPD) demonstration-scale regenerator. Parametric studies of different gas-solids drag closures, air flow rates, solids inventories etc. were conducted and the process was optimized. Furthermore, an industrial-size catalyst regenerator for a 2500 TPD scale CFP unit was simulated with MFiX-Exa, a true exascale capable code, producing the first (to our knowledge) exascale regenerator simulation. With the MFiX-Exa model, different configurations of air grids and various operational modes were studied. The study has high value for addressing the potential problem of “afterburn” (CO combustion in freeboard) common to many FCC units.