Simulation Of Scale-up Effects Of Methanol-to-olefins Fluidized Bed Reactors By Using Combination Of Multiscale CFD And PBM

The newly proposed MTO(methanol-to-olefins)process is an economical route to produce light olefins such as ethylene and propylene.Since methanol can be easily obtained from coal and natural gas,the MTO process thus links the coal industry to petrochemical industry,and is expected to become the primary route for the production of light olefins in the near future.Nowadays,the MTO process has been commercialized since 2013,but the related fundamental research on chemical reaction engineering is seldom reported,especially the current reactor scale-ups which still rely upon the experience and step-by-step experiments.Traditional scaling theories focus on hydrodynamics similarity and neglect the coupling between reactions and hydrodynamics,therefore they are insufficient to comprehensively and exclusively scale up fluidized bed reactors.With the development of Computational Fluid Dynamics(CFD)and multiphase models,a "CFD-aided" scale-up approach is expected to speed up the "experiment-based" scale-up process with lower cost.As the reactor scale up is associated with flow regime transition and concomitant change of reaction behaviors,it poses a big challenge to CFD modeling:1)since the flow regime transition always happens in the scale-up process,how to choose a drag model which considers the suitable meso-scale structures to characterize different flow regimes is recognized as a key factor to reliable CFD simulations;2)for the reactive simulations of large reactors,the lumped-species kinetic model is widely used to describe the reaction kinetics.Since the lumped model is generally based on experiments over micro-scale reactors which filter the influence of dynamic flow structures,whether such a lumped model is suitable for larger-scale reactors running over different flow regime remains an open problem;3)to simulate a large reactor economically,coarse-grid resolution with adequate accuracy is always necessary.However,even using coarse meshes,it is still formidable to afford when the investigated system needs a long period of time to evolve such as coke deposition in MTO reactions.A reactor model was proposed by Lu et al.[1]to provide the initial distribution of coke for the reactive simulation of a pilot-scale MTO reactor and found to help the simulation quickly reach the steady state.Whether this reactor model is still effective for harnessing the speedup of simulations of larger-scale MTO reactors needs to be further investigated.To investigate the above issues,this study focuses on the reactor scale-up effects through a series of simulations of different-sized MTO fluidized-bed reactors.The first Chapter introduces the development of the MTO process,the process features,simulation approaches as well as the previous work on simulations of MTO reactors.Then,the content of this study will be demonstrated.In the second Chapter,a series of simulations of four different-sized DMTO(MTO process developed by Dalian Institute of Chemical Physics)reactors are carried out.The two-fluid model(TFM)is used to describe the flow hydrodynamics.For the drag force which plays the predominant role over the other interaction forces,the EMMS/bubbling model and its newly developed two-step version are chosen to simulate the bubbling fluidized bed and turbulent fluidized bed reactors,respectively.The seven-lumped parallel kinetic model is adopted to describe the reaction kinetics.Then,the flow and reaction behaviors with scaling up of reactor are investigated.It is found that the simulation can successfully predict the typical flow structures and methanol conversion in the different-sized reactors,however,the prediction on the selectivity of light olefins(ethylene and propylene)gradually deviates from the experiment as the dimension of reactor increases.Changing the seven-lumped parallel model to the cross reaction model with considering product conversion,seems not helpful to improve the prediction of light olefins.The possible reason is the insufficient prediction on the distribution of coke content by using current TFM-based simulation,since the distribution of coke content which is closely related to the product selectivity is filtered or averaged by TFM.In order to consider the influence of coke content distribution in the simulation,the third Chapter proposes to use PBM to describe the distribution of catalyst particles with different coke contents.Firstly,a CSTR model is proposed to provide the distribution of residence time which is combined with coke formation rate to give the distribution of coke content.Secondly,the coke content distribution is taken as the initial value and described by PBM in the CFD simulation.The simulation is then performed on a two-dimensional demo-scale DMTO reactor to investigate the influence of coke distribution on the reaction behaviors.It is found that with assistance of initial distribution from the reactor model,the introduction of PBM into TFM simulation improves the prediction of product selectivity at limited computational cost.In Chapter 4,a summary and the future work on improving the reactor scale-up simulation are delivered.