【作者】 宋鹏；

【导师】 李绍武； 解鸣晓；

【作者基本信息】 天津大学 ， 水利工程（专业学位）， 2018， 硕士

【摘要】 由于世界上的天然河流都是紊流,且床面通常由泥沙、砾石等粗糙元素组成,因此研究明渠粗糙紊流具有重大的工程意义。近年来,SPH这种无网格粒子法在模拟流动方面具有很大的潜力,但是在模拟明渠粗糙紊流上仍面临挑战。为此,本文开展了不同粗糙床面下明渠紊流的数值模型研究工作,主要研究成果总结如下:(1)详细研究了弱可压缩光滑粒子法(WCSPH)和明渠粗糙紊流最前沿的理论成果。明确了WCSPH数值模型的研究难点主要在于:首先SPH模型缺乏适合的紊动模型使其闭合,其次很难在模型中模拟靠近粗糙床面处的流动结构。(2)分别选用基于Smagorinsky假定和混合长度理论的SPS紊动模型进行数值模型实验,发现后者更加适合进行明渠紊流的模拟。(3)开发了用于处理粗糙床面的壁面函数模型,并应用其对水力光滑流、水力粗糙流和过渡流进行模拟,均得到不错的结果。为保证模拟精度,只将其应用于小尺度粗糙床面的模拟,而对于大尺度粗糙床面则采用改进后的拖曳力模型进行模拟。(4)为了方便壁面函数模型的应用,利用基于遗传算法的BP神经网络进行公式拟合,得到了能够应用于不同紊流流态下的壁面函数公式。更多还原

【Abstract】 Because all natural river flows around the world are turbulent and the channel beds are often composed of rough elements such as sand and gravel particles,the study of turbulent open channel flows over rough beds is of significant engineering interest.In recent years,mesh-free particle modelling approaches,such as SPH,have been investigated for their potential in simulating open channel flows,but their potential has not been fully explored for the turbulent channel flows over rough wall boundaries.Hence,the numerical modeling of turbulent open channel flows over rough beds is studied.The main achievements of the present paper are summarized as follows:(1)Theoretical achievements of the WCSPH method and turbulent open channel flows over rough beds are studied in detail.The difficulties of the WCSPH numerical model are identified.There are two difficulties for this,firstly the lack of adequate turbulence models which can be used to close with the Lagrangian SPH equations,and secondly the difficulty in modelling the flow structure near wall boundaries,especially when the wall is composed of rough elements.(2)The SPS turbulence model based on the Smagorinsky assumption and the mixing length theory is selected to carry out numerical model experiments,and it is found that the latter is more suitable for the simulation of open channel turbulence.(3)A wall function model for rough bed surface treatment was developed and applied to simulate hydraulically smooth flow,hydraulically rough flow and hydraulically transitional flow.Good results were obtained.In order to ensure the simulation accuracy,it is only applied to the simulation of small-scale rough bed,while the improved drag force model is used to simulate the large-scale rough bed.(4)In order to facilitate the application of the wall function model,the BP neural network based on genetic algorithm is used to fit the formula,and the wall function formula which can be applied to different turbulent flow states is obtained.更多还原