【作者】 彭志威；

【导师】 刘子建；

【作者基本信息】 湖南大学 ， 机械工程， 2009， 博士

【摘要】 虹吸现象是一种常见的非稳态流动现象,虹吸式流道因其具有只要抬高液面高度差就能获得动力源的优点,在工程中具有广泛的应用,因此研究虹吸式流道的形状优化设计方法具有重要意义。国际国内已经在虹吸式流道的应用方面做了较多的研究,但是,针对非稳态虹吸过程的相关理论研究还较少见;在流体机械的形状优化设计方面,非稳态流动现象的优化设计是个长期存在的难点,这方面的研究工作才刚刚起步。因此,本论文首先在研究非稳态虹吸相关理论的基础上,提出产生虹吸的基本条件,再通过建立计算流体力学(CFD)模型深入分析非稳态虹吸的流体流动特性和启动特性,探究基于逆向工程的流道几何形状获取方法,结合提出的虹吸性能评价指标对复杂虹吸式流道进行形状优化设计与实验验证研究,并对复杂虹吸式流道的神经网络优化方法展开了研究。本论文主要研究内容与成果如下:1.研究了适用的计算流体力学模型及产生虹吸的所需条件根据虹吸原理分析了传统研究方法的缺陷,明确了非稳态虹吸基本概念,并以此确定了适用的CFD模型;同时采用定常形式的伯努利方程对稳态流动假设条件下的虹吸作定性研究。结果表明,定常形式的伯努利方程只能用于虹吸式流道的定性分析,对非稳态虹吸作定量研究宜采用CFD方法;压强是产生虹吸的必要条件,流态为产生虹吸的激发条件,在一定压强作用下,虹吸式流道出水段出现封闭水柱或局部满管流的流态都可能激发虹吸的产生;虹吸过程可视为一种活塞效应的结果。这些结论可为后续的非稳态虹吸研究奠定基础。2.研究了非稳态虹吸的流体流动特性及启动过程建立几何模型和CFD模型用于研究非稳态虹吸的流体流动特性,并以工程上常用到的虹吸滤池中的虹吸式流道为例,研究其非稳态虹吸的启动过程。研究明确了非稳态虹吸的流态、速度场和压强特性,指出通过调节虹吸式流道出水管道的长度可有效控制非稳态虹吸过程的最大平均流速、最大平均负压,证实了虹吸启动的关键是应尽快促使虹吸式流道中形成封闭水柱,并促成活塞效应将气体排出流道,促使全局满管流动的形成。3.研究了复杂虹吸式流道的几何形状获取方法及形状特征以坐便器中的虹吸式流道为例,研究如何获取复杂虹吸式流道的几何形状。结果表明,可采用逆向工程、计算机辅助设计(CAD)技术为复杂虹吸式流道的仿真分析和优化设计提供精确的CAD模型。虹吸式流道主要形状特征分析结果表明,流道的直径、α角、β角、水封高度和收缩口曲线是复杂虹吸式流道形状优化设计的关键,且在设计虹吸式流道时应考虑保持流道曲率总体上的光滑过渡,以减小流动过程中的阻力。4.研究了基于计算流体力学的坐便器虹吸式流道形状优化设计方法建立CFD模型并用粒子图像测速(PIV)实验来验证,用该模型预测复杂虹吸流道内部速度场和压强的变化,并根据预测值与多个评价虹吸性能的指标来研究流道的收缩口截面、流道上升段仰角α和下降段长度L等对虹吸性能的影响,以优化设计流道的形状。结果表明,建立的CFD模型具有较高精度,能很好地模拟复杂虹吸式流道中的流场变化;虹吸式流道的收缩口截面、流道上升段仰角α和下降段长度L对虹吸性能具有重要影响,当采用光滑曲线设计及α为39°、L为227 mm时,虹吸性能最佳,可使坐便器实际用水量从6升降至5升,具有节水效果;并证实累积负压是非稳态虹吸研究中最佳的虹吸性能评价指标。5.研究了神经网络的坐便器虹吸式流道形状优化设计技术通过CFD模型计算得到神经网络建模样本,以几何变量为输入、累积负压为输出,建立非稳态虹吸的神经网络模型,并在网络结构优化的基础上,对比分析复杂虹吸式流道使用Levengerg-Marquardt(L-M)神经网络和贝叶斯神经网络建模的优缺点。结果表明,L-M神经网络对训练样本有很好的拟合能力,贝叶斯神经网络则具有更好的泛化能力;L-M神经网络的最佳模型结构为2-4-1,贝叶斯神经网络有效节点数为3.67967,通过贝叶斯神经网络进行仿真,确定了流道形状的最佳设计参数(弯道半径为80 mm、出水段直径为45 mm);并构建了基于CFD和神经网络的虹吸式流道形状优化设计平台框架,以期为虹吸式流道形状的快速优化设计提供方法参考。6.研究了虹吸式流道优化设计过程中的实验验证方法针对坐便器中复杂虹吸式流道的形状优化设计建立相应的实验验证方法,评价优化设计效果。即,采用PIV实验测量虹吸式流道内部实际流场,结果证实了所建CFD模型的准确性;利用相似理论研制了用于制备不同形状虹吸式流道的实验装置,可根据模型仿真计算寻得的较优流道形状参数制备出相应的虹吸式流道实物,验证设计出的不同形状流道的虹吸效果,这可为类似的不规则流道实验研究提供方法参考;根据现代设计方法及理论构建了配套模拟实验装置,无需采用费时费力的修改企业模具这一传统方法,可更简便地进行不同形状虹吸式流道的虹吸性能测试和坐便器节水性能测试;结合节水实验,证实了本文构建的虹吸式流道形状优化设计方法确能有效实现流道形状的优化、改进流道虹吸性能。更多还原

【Abstract】 Siphon is a common flow phenomenon. Having the advantage of convenient acquisition method for power only by altitude difference, the siphon technology is widely used in engineering. It is very meaningful to study the shape optimization design methods of siphon channel. The current researches are mainly focused on the application of siphon without sufficient insight in the theory of unsteady siphon flow. In addition, the research about optimization design on unsteady flow has just started, which is difficult in shape optimization design. Considering the situation mentioned above, this dissertation makes the further studies on the basic theories, and the analysis of flow characteristic and siphon-generate in unsteady siphon based on computational fluid dynamics (CFD), and the method capturing the geometry of flow channel based on reverse engineering (RE), and the shape optimization design and experimental verification of complex siphon channel using performance indexes presented in this dissertation, and the optimization methods of complex siphon using Artificial Neural Network (ANN). The main contents and productions of this dissertation are as follows.1. Study on the CFD model for siphon research and the conditions of siphon-generate.Based on the analysis of the limitation on traditional research methods of siphon, the concept of unsteady siphon was defined, which guiding the selection of CFD models. Furthermore, the qualitative analysis of siphon was performed using Bernoulli equation on the condition of steady hypothesis. Results show that the Bernoulli equation can but used to the qualitative analysis of siphon and the quantitative analysis on unsteady siphon should be carried out using CFD method. It is presented that in the four conditions of siphon-generate (gravitation, pressure, geometry parameter and flow regime), the pressure is the necessary condition. The results also show that the flow regime is the excitation condition generating siphon, which indicates that a closed water column or a local full pipe flow should impel siphon- generate. Finally, the piston effect in siphon was presented. The conclusions above lay a foundation on unsteady siphon research.2. Study on the characteristic of unsteady siphon and siphon-generate.The geometry model and CFD model were constructed to investigate the flow characteristic of unsteady siphon. As an example, the process of siphon-generate on filter which usually used in engineering was studied. The researches have confirmed the three characteristics (flow reime, velocity field, pressure field) of unsteady siphon and indicated that both the max average velocity and pressure drop in unsteady siphon could be effectively controlled by adjusting the length of outlet pipe. It is also confirmed that the closed water column formed in flow is the key of siphon-generate, which could induce piston effect and could impel a globe full pipe flow by pumping out air.3. Study on the method capturing the shape of complex siphon channel and the form feature.The method capturing the shape of complex siphon channel on water closet (WC) was studied. The results show that the reverse engineering (RE) and computer aided design (CAD) can be used to build accurate CAD model for simulation and optimization design. The form feature analysis indicated that the keys of shape optimization design of siphon are diameter of flow channel, angleαandβ, water seal and the contract region. The curvature of siphon channel should be design to be smoothing to reduce the resistance in flow.4. Study on optimization design methods of complex siphon based on CFD.The CFD model verified by Particle Image Velocimetry (PIV) experiments was constructed and applied to predict the velocity field and pressure field in siphon. The predicted value combined with several performance indexes was used to studied the effects of contractive section, elevation angle (α) and the length (L) in outlet on siphonage. The CFD model was proved to be accurate in simulating the flow field in complex siphon. It was found that the siphon could be obviously influenced by contractive section,αand L. The design with smooth shape andαof 39°and L of 227 mm have improved the efficiency of siphon, which reduced the water consumption of WC from 6.0 to 5.0 liter. Results also indicated that the pressure drop could be the best index to assess the unsteady siphon flow.5. Study on optimization design methods of complex siphon based on Artificial Neural Network.Using the samples from CFD, the Artificial Neural Network (ANN) models were constructed with inputs of geometry and output of cumulative pressure drop. Comparing Levengerg-Marquardt ANN (L-M ANN) and Bayesian ANN built after network structure optimization, it is found that the first one is good at fitting training samples and the last one has more capability on generalization performace. The best structure of L-M ANN is proved to be 2-4-1 and in Bayesian ANN the effective number of parameters is 3.67967. Based on the simulation of Bayesian ANN, the best bend radius R and diameter D in outlet were found to be respectively 80 mm and 45 mm. An optimal design framework based on CFD and ANN was constructed to provide reference to optimization design of complex siphon.6. Study on experimental system of optimization design of complex siphon.The experimental system used to verify optimization design was constructed in allusion to optimization design of complex siphon. The PIV technique was used to validate CFD model. An experimental set up which can produce most shape of siphon was developed based on the theory of similarity and used to examine the performance of different siphon channel, provding refference to the experiment researches on abnormity flow channel. Based on modern design methodes, an experimental set up of water closet was designed and proved to be convenient to siphon test and water saving test of WC, avoiding the time consuming and laboursome methods by modifing mould. The water-saving test proved that the optimization design method in this dissertation is effective to shape optimization design and improvging siphon.更多还原