## Modeling and Simulation of Reactive Three-phase Flows in Fluidized Bed Reactors: Application to the SE-SMR Process

##### Abstract

In this thesis a reactive binary-particle multi-fluid model for sorption enhanced steam methane reforming (SE-SMR) fluidized bed reactors has been derived. The kinetic theory of granular flows (KTGF) is a applied to close the governing equations for the particle phases. The species mass and heat transport phenomena have been incorporated in deriving the KTGF constitutive equations. The conventional single phase k − ε turbulence model has been applied for the gas phase. Special effort has been given to the particleparticle drag coupling. The collisional and frictional particle - particle drags have been considered separately. A collisional particle-particle drag model has been derived based on the KTGF concepts, and a semi-empirical approach has been employed to derive a constitutive relation for the frictional particle-particle drag. Both the collisional and frictional drag models have been validated for binary fluidized bed simulations. The proposed particle-particle drag relations give reasonable particle-particle velocity/momentum coupling approximations. Finally, a reactive model is applied to study the flow and chemical performance of the SE-SMR bubble bed reactors.
A description of the numerical solution strategy employed in order to solve the governing equations has be presented. The solution strategy is based on the extended SIMPLE algorithm for multi-phase flow. The finite volume method (FVM) is used to discretize the model equations. The Bi-Conjugate gradient (BCG) method is used to solve the algebraic equations obtained. For the coupled momentum equations and molecular temperature equations, a coupled solver is applied. All the steps in the solution algorithm have been outlined.
The cold flow binary model has been applied for both dilute riser flows and dense bubble bed flows. The main results and conclusions are:
(1) Model validation using data from Mathiesen et al. (2000) measured for a binary riser flow shows that the novel particle-particle drag model gives reasonable estimates of the particle-particle velocity coupling. Reasonable predictions have been obtained for the particle volume fraction, the particle velocities and the velocity fluctuations. A model equations development analysis reveals that the basic particle velocity fluctuations constitute 2 terms: the velocity fluctuations of the discrete particles, and the velocity fluctuations of the continuous fluid flow. Furthermore, the simulation results show that the velocity fluctuations of the continuous fluid flow are dominant in a binary riser flow.
(2) The segregation behavior of two types of particles with approximately the same particle diameters, but different particle densities, was studied and validated using the experimental data from Formisani et al. (2008). Detailed information regarding the gas phase behavior, the particle velocity profiles, the distributions of the particle volume fractions and the flotsam-to-total particle volume fraction ratios is presented. The simulation results show that the simulated axial average flotsam-to-total particle volume fraction ratio distribution agrees reasonably with the experimental data of Formisani et al. (2008). The binary particle velocities are closely coupled though the segregation phenomena exist.
The segregation behavior and the particle velocity profiles are superficial gas velocity dependent. The number and distribution of particle velocity vortices change dramatically with superficial gas velocity: at a comparatively low superficial gas velocity, the particles mainly segregate axially, and at a comparatively high superficial gas velocity, the particles segregate both axially and radially.
(3) Seven binary particle drag closures found in the literatures, derived based on the kinetic theory of granular flows (KTGF) for the application to fluidized beds, have been investigated using a binary KTGF model. Experimental data for dynamic size segregation from the literature is used to validate the simulation results. The validations show that all the seven binary particle-particle drag relations under-predict the binary particle coupling in the dense fluidized bed. On the other hand, satisfactory results could be obtained if an extra semi-empirical frictional binary particle - particle drag term was included. The semi-empirical frictional binary particle - particle drag considering the long term particleparticle contact effects, represents a correction of the short term collisional frictional binary particle-particle drag model proposed by Syamlal (1987). Furthermore, numerous comprehensive calculations show that the three - fluid model using the suggested semiempirical frictional binary particle-particle drag model can fairly well predict the particle segregation rates and the bed heights of the individual particles, with a change of gas fluidization velocity, the initial bed height, and the small particle ratio using the same fixed correction coefficient value. Therefore, it is proposed that a frictional binary particle - particle drag term that represents the binary particle momentum exchange due to the long term particle-particle contacts (sliding/rolling) must be included in order to model dense binary fluidized bed flows.
A reactive flow model has been applied in order to model the flow behaviors and chemical performance for the SE-SMR process, as operated in bubble bed reactors. The main results and conclusions are:
(1) The SE-SMR process, as operated in a laboratory scale fluidized bed reactor has been investigated using the three-fluid model derived in the thesis. The binary sorbent and catalyst particles may segregate due to the density difference between them. Initially, the light sorbent particles tend to rise and the heavy catalyst particles tend to sink. As the process proceeds, the sorbent particles adsorb more CO2 and become heavier, and the density difference between the binary particles will become smaller, thus the binary particle mixture tends to be well-mixed. As the sorbent particles are either at the upper sections of the bed or well-mixed with the catalysts, the adsorption of CO2 may always induce a sorption enhancement effect, thus the hydrogen purity at the outlet is between 98- 99% before the breakthrough, which is much higher than the characteristic 73-74% for a steam methane reforming (SMR) process. Due to the exothermic CO2 adsorption reaction and the mixing of the gas particle flows, a comparatively uniform gas/particle temperature distribution is found in the whole bed. In general, the hydrogen purity obtained in the simulations agrees fairly well with the experimental data of Johnsen et al. (2006).
(2)A 1m-high laboratory scale and a 4m-high industrial scale sorption enhanced steam methane reforming (SE-SMR) fluidized bed reactors have been simulated using a three fluid model. The performance of the SE-SMR process has been compared with the steam methane reforming (SMR) process operated in fluidized bed reactors. The influences of the superficial gas velocities and the solid loading (packed bed heights) on the reactor performance (hydrogen purity) have been studied. The simulation results show that the sorbent particles can adsorb CO2 in these reactors, thus a higher purity of the hydrogen product can be obtained in a SE-SMR reactor than in a conventional SMR reactor. The superficial gas velocity is an important parameter. A pseudo- steady SE-SMR bed performance can be obtained provided that the bed is operated at certain optimized medium level superficial gas velocities. In the present study, it has been found that the binary sorbent-catalyst particles are well mixed when the bed is operated at 0.2m/s. The sorbent can adsorb CO2 steadily, thus the dry mole fraction of the hydrogen product can get above 0.95 in the 1m laboratory scale bed, and above 0.97 in the 4m industrial scale bed. However, when the laboratory scale bed is operated at a lower superficial gas velocity of 0.15m/s, the binary sorbent-catalyst particles are segregated. In this case, the heavy sorbent particles accumulate at the bottom of the bed, and the light catalyst particles tend to rise and become distributed at higher sections in the bed. For this reason, the sorbent can not adsorb CO2 in an optimized manner, and the hydrogen product purity is decreased compared to the well-mixed bed case. When the bed is operated at a higher superficial gas velocity of 0.3m/s, the process work load is increased, and the gas residence time in the reactor is decreased. Therefore, the hydrogen product purity is further decreased. The simulation results obtained for the 4m industrial scale bed show that a dry mole purity of hydrogen above 0.97 can be obtained due to the increased bed height. These simulation results also show that there is an optimal bed height limit, at which further increase of the packed bed height can not increase the hydrogen purity.