【作者】 梁健；

【导师】 舒立春；

【作者基本信息】 重庆大学 ， 电气工程， 2017， 博士

【摘要】 我国受季风性气候影响明显,冬季风力资源特别丰富,但低温造成的叶片覆冰严重地制约着风电产业的发展。叶片覆冰对其空气动力特性造成极大的影响,导致严重的风力机功率损耗;叶片上附着的冰块在旋转过程中因离心力作用而被甩落,对机组及附近工作人员的安全造成极大的隐患。国内外对于风力机叶片覆冰模型的研究多基于经典的叶素动量理论(BEM),将三维旋转问题转化成具有特定攻角的二维翼型进行研究,且缺乏相应的试验验证。由于风力机旋转过程中包含沿叶展方向的速度,三维模型能更加准确地描述风力机的覆冰过程。因此,从覆冰试验与三维数值计算模型相结合的角度研究风力机叶片的覆冰增长过程,提出基于气象参数的叶片覆冰预测模型,对于风电场的冬季维护预警、优化防冰除冰方法具有重要的工程指导意义和实用价值。论文首先试验研究了风力机叶片的覆冰增长过程,然后从建立三维数值计算模型的角度出发,将其分为空气流场计算、水滴碰撞系数计算、传质传热计算以及冰形计算四部分,在二维Messinger传质模型的基础上对三维覆冰算法进行了改进,在建立模型的过程中综合考虑了湍流模型、表面水膜流动以及冰面粗糙度对覆冰过程的影响,得到覆冰过程中碰撞系数、对流换热系数、冻结系数以及冰形沿叶片的分布规律及其影响因素,最后通过试验验证了覆冰预测模型。论文的具体工作及成果如下:(1)在人工气候室及雪峰山自然覆冰试验站搭建试验平台得到了风力机叶片覆冰的分布规律,并根据显微成像法及旋转多圆柱监测法研究了环境参数对覆冰荷载的影响。结果表明,覆冰主要集中在叶片前缘,沿展向覆冰逐渐增多;风速、液态水含量(LWC)的增大以及温度降低都会导致更加严重的覆冰。(2)采用多参考坐标系(MRF)法模拟了叶片旋转,基于S-A、标准k-ε和k-ωSST三种湍流模型的数值模拟,并引入LEWICE的势流理论求解叶片边界层速度,研究了湍流模型对叶片表面压力系数、边界层速度的影响。计算结果表明,k-ωSST湍流模型具有较高的计算精度,与试验结果吻合较好;S-A模型在前缘吸力面存在压力系数偏小的情况,标准k-ε模型对吸力面的压力系数变化不够灵敏。(3)以机翼、圆柱和球体为算例,分别采用拉格朗日法及欧拉法模拟了过冷却水滴的撞击过程,并用于计算叶片的碰撞系数分布特性及其影响因素。结果表明,水滴碰撞区域主要集中在叶片前缘附近,碰撞系数沿叶展方向逐渐增大;碰撞系数和碰撞区域随着水滴中值直径(MVD)的增大而增大;风力机转速下降或风速上升都会造成水滴的撞击区域整体向压力面移动。(4)试验得到了叶片粗糙度对覆冰过程以及冰面粗糙度的影响规律,并在显微镜下观测了叶片表面冰面粗糙度随时间以及沿弦向的分布变化,依据观测结果建立了叶片冰面粗糙度模型。试验结果表明,叶片粗糙度对长时间的覆冰过程影响可以忽略;冰面粗糙度在覆冰增长初期随时间有显著的函数关系;通过拍摄到的沿弦长方向冰面粗糙度变化,可将粗糙度的发展过程分为水膜、水珠和细流三个阶段。(5)将二维Messinger传质模型拓展到三维叶片,结合水膜的流动开发了三维数值计算的算法,基于传质传热平衡方程建立了叶片覆冰冻结热力学模型,得到了对流换热系数、冻结系数沿叶片的分布规律,并研究了环境参数对两个参数的影响。结果表明,沿叶展方向对流换热系数逐渐增大、冻结系数逐渐减小;MVD的增大会导致对流换热系数的明显增大;冻结系数随着温度、MVD和LWC的降低而升高,风速变化仅对驻点附近的对流换热系数和冻结系数有一定影响。(6)采用网格节点的线性插值、B样条曲线光顺冰层边界以及动弹网格法重构网格等方法对冰形计算进行了改进优化,提出了完整的叶片覆冰预测模型及三维数值计算流程,并通过时间多步法模拟了叶片覆冰的动态变化过程,提出了确定时间步长的判定依据。覆冰预测模型与FENSAP-ICE的数值模拟以及覆冰试验结果进行了对比,其计算结果比较可靠。基于覆冰预测模型研究了网格节点数、迭代冰层数对数值计算的影响,并研究了环境参数对冰形的影响。结果表明,增大网格节点数可以有效提高计算精度,可以通过覆冰程度和覆冰增长速率来确定计算的时间步长;低温造成叶片雨淞覆冰前缘加厚、覆冰区域减小;水滴颗粒越小越趋于雾凇覆冰,前缘覆冰越薄;LWC增大导致更加严重的覆冰,而一定范围内的风速变化对覆冰的影响较小。更多还原

【Abstract】 In winter,wind resources are rich in China due to the monsoon climate,but the icing problem seriously restricts the development of wind power industry.Ice on blades changes the aerodynamic performance,causing power loss to the wind turbine.Ice adhered to the blade is easily thrown out by centrifugal force in the process of rotation,which is harmful to the crew and staff near turbine.Ice accretion model for wind turbine was mainly on 2-D airfoil,where the 3-D rotating case was transformed into a specific angle of attack by the classical Blade Element Momentum theory(BEM).Besides,most research lacked experimental verification.Considering the vector of spanwise direction of the rotating blade,3-D model is more accurate on describing the process of wind turbine icing.Thereby,3-D numerical calculation model is combined with icing test,and the blade ice prediction model is based on the meteorological parameter.This has important engineering significance and practical value on wind farm warning and optimization of anti-icing and de-icing method in winter.In this paper,firstly,the ice accretion process is experimentally studied.Then,the 3-D ice accretion model is established and contains four parts: air flow computation,water collision efficiency computation,heat and mass transfer and ice shape simulation.The 3-D ice accretion algorithm is improved on the basis of 2-D Messinger model.The influence of turbulence model,the motion of water film on the surface and the ice roughness on the ice accretion are all taken into consideration.The distribution of collision efficiency,heat transfer coefficient,freezing fraction and ice shape and their environmental influences are also studied in this paper.Finally,the ice accretion model is verified with icing tests.The work and achievements are shown as follows:(1)The test platform is built in both the artificial climate chamber and Xuefeng Mountain natural icing test station.The ice distribution on the blade is experimentally studied and the environmental effects on ice load are studied by both microscopic imaging method and rotating multi-cylinder monitoring method.Results show that ice mainly accumulates at the leading edge and increases along the spanwise direction.The increase of wind velocity,liquid water contents(LWC),and temperature decline will all lead to the heavier icing.(2)The blade rotation is performed by multiple reference frame(MRF)method.Based on S-A,standard k-ε and k-ω SST,the influence of turbulence model on the pressure coefficient of the blade is studied.The local edge velocity is computed by potential flow theory of LEWICE.Numerical results show that k-ω SST has higher accuracy,which agrees well with the experiment.S-A shows smaller pressure coefficient at the leading edge of suction side and Standard k-ε is not sensitive to the change of pressure coefficient on the suction side.(3)In the case of wings,cylinder and sphere,the supercooled droplet impingement is simulated by both lagrangian and eulerian method.This is applied to calculate the distribution characteristics of collision efficiency of blades and its influence factor.Numerical results show that the collision region mainly accumulates at the leading edge,and collision efficiency increases along the spanwise side.Collision efficiency increases and collision area expands with media volume diameter(MVD).Lower rotor speed and higher wind velocity would both lead to the collision region moving to pressure side of the blade.(4)The influence of blade roughness of clean blades on ice accretion and ice roughness is experimentally studied,and the ice roughness changes with time and its distribution are observed under microscope.The ice roughness model of blades is proposed on the basis of observation.Results show that blade roughness can be ignored during long time icing process.Ice roughness shows strongly functional relationship with time.Based on the images of the ice roughness along the chordwise,the development of ice roughness is divided into three stages: water film,bead and rivulet.(5)By expanding 2-D Messinger model to the 3-D blade and combining the water film motion,the 3-D numerical calculation algorithm is developed.The blade freezing thermodynamic model is established based on mass and heat transfer balance equations.The distribution of heat transfer coefficient and freezing fraction on the blade and environmental effects are numerically studied.Results show that from the root to tip,the heat transfer coefficient increases and freezing fraction decreases.The heat transfer coefficient significantly increases with MVD.The freezing fraction decreases with temperature,MVD and LWC.Wind velocity effect on heat transfer coefficient and freezing fraction is obvious only at the stagnation point.(6)The ice shape calculation is developed by the linear interpolation on the gird node,B-spline smoothing ice boundary and spring grid remesh method.The blade ice prediction model and whole numerical calculation process are proposed.The dynamic change process of blade icing is simulated by multi-step method and the judging method of time step is proposed as well.The ice prediction model is reliable by the verification of FENSAP-ICE simulation and icing test.The influences of grid node number and ice layer number of iterations on numerical calculation model are discussed based on the ice prediction model.The environmental influences on ice shape are numerically studied as well.Results show that more nodes help to improve the calculation accuracy.The time step can be determined by the ice degree and ice growth rate.As temperature reduced,more glaze ice is found at the leading edge,and ice area significantly decreases.Smaller droplets lead to rime and thinner ice at the leading edge.Larger LWC leads to more serious ice,but ice accretion is less affected by wind velocity when wind turbine is on the rated operation.更多还原