【作者】 庄亚明；

【导师】 陈晓平； 刘道银；

【作者基本信息】 东南大学 ， 热能工程， 2018， 博士

【摘要】 随着计算机科学的发展,基于CFD（Computational Fluid Dynamics）的数值模拟方法已经成为研究气-固流化床的重要手段之一。经过几十年的研究,诸如CFD-DEM（Discrete Element Method）等模型日臻成熟,已经能够较为准确的模拟出流化床中详细的气固流动,但计算负荷高和计算速度慢的缺陷一直是此类模型进一步应用于大型工业装置的瓶颈。而应用于颗粒系统的MCM（Markov chain method）统计学模型则具有计算简单、高效及适用性强等特点,发展流化床颗粒运动MCM模型,并进一步耦合气相运动、化学反应、传热及传质,不失为一条突破流化床传统数值模拟方法发展瓶颈的潜在路线。本文结合流化床统计学模型的发展难点以及流化床富氧燃烧碳减排技术这一热点问题,对流化床颗粒运动和焦炭燃烧的统计学模型进行了深入的研究和探索,取得的主要创新性研究成果总结归纳如下:（1）为克服以往流化床颗粒运动MCM模型无法给出颗粒具体位置信息的缺陷,以CFD-DEM模型的计算结果作为采样样本,从中提取颗粒运动统计学规律,基于流化床物理空间划分网格,定义颗粒的Markov状态空间,建立了模拟鼓泡流化床和循环流化床提升段颗粒运动的MCM模型。就流化床颗粒运动是否具有Markov特性,以及MCM模型参数的独立性进行了探索性的讨论。相比CFD-DEM模型,MCM模型的计算速度几乎提高了两个数量级,且对计算负荷的需求很低。MCM模型能够比较准确的模拟出流化床中颗粒运动的宏观规律,并能够跟踪到每一颗颗粒。（2）为改进颗粒运动MCM模型无法模拟出鼓泡流化床气泡运动以及瞬时性颗粒运动规律的缺点,提取鼓泡流化床CFD-DEM模型计算结果中的气泡信息,基于统计学理念创新性的建立了气泡随机发展模型（Stochastic bubble developing model,SBDM）,并将其与MCM模型耦合。SBDM-MCM模型具有和MCM模型近似的高计算速度和低运算负荷,且成功将气泡对颗粒运动的影响引入了MCM模型中,重现了样本数据中颗粒运动的瞬时性规律。（3）为改进颗粒运动MCM模型无法模拟出循环流化床提升段颗粒絮团运动以及瞬时性颗粒运动规律的缺点,提取循环流化床提升段CFD-DEM模型计算结果中的颗粒絮团信息,基于统计学理念创新性的建立了絮团随机发展模型（Stochastic cluster developling model,SCDM）,并将其与MCM模型耦合。SCDM-MCM模型同样高效且节约计算资源,并且在模拟结果中有效体现了絮团对颗粒运动特性的瞬时影响,对比表明SCDM-MCM模型对样本数据中颗粒运动规律的拟合程度要明显高于MCM模型。（4）为解决SBDM-MCM模型无法模拟出气相信息,导致后续耦合化学反应存在困难的问题,创新性的将经典的流态化两相模型对气相流动和相间传质的描述引入SBDM-MCM模型,并耦合单颗粒焦炭燃烧模型,创建了鼓泡流化床SBDM-MCM反应模型。SBDM-MCM反应模型的计算速度可达到CFD-DEM反应模型的115倍左右,且计算负荷较低。不同O₂浓度的O₂/N₂和O₂/CO₂气氛下,SBDM-MCM反应模型和CFD-DEM反应模型对流化床焦炭燃烧的数值模拟结果对比表明,SBDM-MCM反应模型不仅对样本数据中石英砂和焦炭颗粒的运动规律具有很高的拟合度,其预测的流化床内焦炭的碳消耗速率,以及流化床整体的气相分布规律都能够与CFD-DEM反应模型的计算结果吻合的很好。更多还原

【Abstract】 With the development of the computer science,the CFD based numerical simulation has become one of the most important methods for the study of gas-solid fluidized beds.After the decades of research,many popular numerical approaches based on the CFD have the ability to predict the detailed gas-solid flow,such as the CFD-DEM.However,the long computating time and the high computational load are always the bottlenecks of these kinds of methods for practical application.While the Markov chain method（MCM）based statistic model applied in granule processes has the advantages of simple theory,easy program and fast calculation.Thus,developing the MCM model of particle movement in fluidized beds is a potential method to break through the bottlenecks of the traditional numerical methods.In this work,the development of the MCM for fluidized beds and the oxy-fuel fluidized bed combustion which is a hot issue of CO₂ capture,are both taken into consideration.The deep study and exploration is made for the statistic model of the particle movement and char combustion in fluidized beds.The significant results are summarized as follows:（1）In order to overcome the deficiency of the MCM established by other researchers which can not provide the detailed particle movement,the statistic laws of particle movement are extracted from the CFD-DEM calculated results,and the fluidized bed is discretized into physical cells,which represent the Markov states of particles.The MCM models for the particle movement in both the 2D bubbling fluidized bed（FB）and the circulating fluidized bed（CFB）riser are established.The Markov property of the FB and the independence of the MCM parameters are discussed.Results show that the calculating speed of the MCM is faster than that of the CFD-DEM by about 2 orders of magnitude,and the MCM costs a low computation load.Besides,the MCM can accurately predict the macroscopic property of particle movement and track every single particle.（2）In order to improve the shortcomings of the MCM which can not simulate the bubble movement and the instantaneous property of particle movement in the FB,the statistic laws of bubble movement are extracted from the CFD-DEM calculated results through the graphic recognition method.A stochastic bubble developing model（SBDM）,which is coupled with the MCM,is established based on the statistic theory.The SBDM-MCM also has a high calculation speed and a low computation load,which successfully introduces the influence of bubbles on the particle movement into the MCM and has the ability to reproduce the instantaneous fluctuation of particle movement.（3）In order to improve the shortcomings of the MCM which can not simulate the particle cluster movement and the instantaneous property of particle movement in the CFB riser,the statistic laws of particle cluster movement are also extracted from the CFD-DEM calculated results through the graphic recognition method,and a stochastic cluster developing model（SCDM）coupled with the MCM is established based on the statistic theory.The SCDM-MCM calculates quickly too.Results show that the SCDM-MCM successfully predicts the instantaneous influence of the clusters on the particle movement.The SCDM-MCM results fit the sample date provided by the CFD-DEM much better than the MCM results of the CFB riser.（4）In order to solve the thorny problem that the SBDM-MCM can not afford the information of the gas phase,which makes it difficult to be coupled with the chemical reaction,the fluid sub-model and the mass transfer sub-model of the traditional fluidization two phase model along with the char combustion model are introduced into the SBDM-MCM.The calculation speed of the SBDM-MCM reaction model is almost 155 times faster than that of the CFD-DEM reaction model.The results show that the SBDM-MCM reaction model not only simulates the movement of bed material and char particles well,but also predicts the similar char reaction rate and gas component distribution to compare with the CFD-DEM reaction results.更多还原