旋风分离器数值模拟分析与优化设计研究

发布时间：2018-03-08 02:14

本文选题：旋风分离器　切入点：数值模拟　出处：《西南石油大学》2017年硕士论文　论文类型：学位论文

【摘要】：采用数值模拟的方法分析了旋风分离器各个几何结构尺寸(入口高度、入口宽度、排气管直径、排气管插入深度、排气管外延伸长度、总高、圆柱段高、排尘口直径)对旋风分离器性能的影响;结果表明,随着入口高度与宽度的增大,旋风分离器压降降低,旋风分离器分级效率曲线逐渐变平缓,分离能力逐渐变差,切割粒径d50逐渐增大,入口宽度对压降的影响比入口高度大;排气管直径对旋风分离器压降和切割粒径d50影响均很大,而排气管插入深度、排气管外延伸长度、排尘口直径对旋风分离器性能的影响很小;随着旋风分离器高度(圆柱段或圆锥段)的增大,压降减小,而圆柱段高度的变化对压降的影响程度明显大于圆锥段,切割粒径随着旋风分离器高度的增大而减小,圆锥段高度的变化对旋风分离器分离能力的影响明显强于圆柱段。通过搜集整理文献中的98组实验数据,应用BP神经网络训练出一组能够准确预测旋风分离器几何结构尺寸与无量纲压降(欧拉数Eu)之间复杂的非线性关系的神经网络,结合神经网络的预测结果对一些常见的经验模型、理论模型检测;结果表明,Shepherd-Lapple模型预测的排气管直径Dx、入口宽度b和入口高度aa变化对旋风分离器欧拉数影响趋势一致,但数值大小以及影响程度预测偏低;而Casal-Benet模型在数值大小与变化趋势上均不能正确预测,Iozia模型和Barth模型在预测入口宽度对欧拉数影响上,存在与实际情况不符合的情况,而且在数值上也与BP神经网络预测输出值存在一定偏差;至于MM模型在数值变化趋势预测上最接近于BP网络的输出值。结合神经网络的预测值与MM理论模型,通过响应曲面法拟合出两组关于旋风分离器欧拉数Eu、切割粒径d50的二次多项式,并对二次多项式系数回归分析,得到旋风分离器各个几何结构尺寸对其性能影响的显著性关系;结果表明,对旋风分离器性能影响程度依次为排气管直径、入口宽度、入口高度、总高;入口高度与排气管直径、入口宽度与排气管直径的交互项影响显著,而其余交互项影响不显著。通过统计学软件Design expert8.0中的优化模块对旋风分离器几何尺寸优化设计,得出一组新的旋风分离器结构尺寸分布,并对新尺寸的旋风分离器与Stairmand高效旋风分离器数值模拟对比分析;结果表明,优化后旋风分离器切割粒径d50与Stairmand高效旋风分离器基本相同,但压降值却小于Stairmand旋风分离器,其降低达到38.3%。
[Abstract]:The geometry dimensions of cyclone separator (inlet height, inlet width, exhaust pipe diameter, exhaust pipe insertion depth, exhaust pipe extension length, total height, cylindrical section height) were analyzed by numerical simulation. The results show that with the increase of inlet height and width, the pressure drop of cyclone separator decreases, the classification efficiency curve of cyclone separator becomes more and more gentle, and the separation ability becomes worse. The influence of inlet width on pressure drop is greater than that of inlet height, the diameter of exhaust pipe has great influence on pressure drop and cutting diameter d50 of cyclone separator, while the depth of exhaust pipe insertion and the extension length of exhaust pipe, The diameter of dust outlet has little effect on the performance of cyclone separator, and the pressure drop decreases with the increase of cyclone height (cylinder or cone), and the influence of the height of cylinder on pressure drop is obviously greater than that of cone. The cutting particle size decreases with the increase of cyclone height, and the influence of the height of cone section on the separation ability of cyclone separator is obviously stronger than that of cylindrical section. By collecting and sorting 98 groups of experimental data in literature, BP neural network is used to train a set of neural networks which can accurately predict the complex nonlinear relationship between the geometry size of cyclone separator and dimensionless pressure drop (Euler number Eu). Combined with the prediction results of neural networks, some common empirical models and theoretical models are tested, and the results show that the variation of exhaust pipe diameter Dx, inlet width b and inlet height AA predicted by Shepherd-Lapple model has the same effect on the Euler number of cyclone separators. However, the numerical value and the degree of influence are low, and the Casal-Benet model can not correctly predict the influence of the inlet width on the Euler number in the Casal-Benet model and the Barth model, which is inconsistent with the actual situation. Moreover, there is a certain deviation between the predicted output value of BP neural network and the predicted value of BP neural network, and the MM model is closest to the output value of BP neural network in the prediction of numerical variation trend, combining the predicted value of neural network with the MM theoretical model, Based on the response surface method, two groups of quadratic polynomials about the Euler number Euand cutting diameter D50 of the cyclone separator are fitted, and the regression analysis of the coefficient of the quadratic polynomial coefficient shows that the significant relationship between the size of the geometry structure of the cyclone separator and the performance of the cyclone separator is obtained. The results show that the influence degree on the performance of cyclone separator is as follows: the diameter of exhaust pipe, the width of inlet, the height of inlet, the total height, the interaction between inlet height and diameter of exhaust pipe, and the interaction between inlet width and diameter of exhaust pipe. However, the other interaction items have no significant effect. Through the optimization module in Design expert8.0, a new group of cyclone separator structure size distribution is obtained by optimizing the design of the cyclone separator geometry dimension. The numerical simulation of the new size cyclone separator and the Stairmand high efficiency cyclone separator shows that the cutting diameter D50 of the cyclone separator after optimization is basically the same as that of the Stairmand high efficiency cyclone separator, but the pressure drop value is smaller than that of the Stairmand cyclone separator. The decrease reached 38.3%.
【学位授予单位】：西南石油大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TQ051.8

【参考文献】