Evolution Analysis Of Internal Flow Coherent Structure Of Francis Turbine

In recent years,with the rapid growth of installed capacity of renewable new energy sources such as wind and light,more and more hydropower stations have begun to operate as regulators.This puts forward higher requirements for the stable operation of the hydraulic turbine,which requires it to work in a relatively large range of working conditions.However,as the core component of the hydropower station to convert water flow energy into electrical energy-the Francis turbine,its optimal operating range is relatively short.Small,it will often complicate the internal flow field when it deviates from the optimal operating condition,and cause the stability of the turbine.There are many inducing factors for the stable operation of hydraulic turbines,mainly including electromagnetic,mechanical and hydraulic.Among them,hydraulic factors are the most serious.The influence of hydraulic factors is mainly reflected in the cyclical development of the internal flow field（such as blade vortex,wake vortex,Karman vortex,etc.）.The fluctuating pressure frequency induced by the structure is consistent with the natural frequency of the structure,causing resonance in the unit and even the plant.The existing vortex visualization criteria can identify the vortex structure,but the coherent structure of different frequencies cannot be extracted by these methods.For this reason,this paper adopts the method of dynamic modal decomposition to decompose the flow field of the runner single flow channel and the straight cone section of the draft tube.The coherent structure and dynamic information corresponding to each single frequency are studied,with a view to the hydropower station Stable operation and optimal design of hydraulic turbines provide guidance and reference.The thesis first derives the dynamic modal decomposition（DMD）algorithm,writes and optimizes the DMD algorithm code,and takes the flow around a cylinder as an example,carries out the dynamic modal decomposition algorithm modal decomposition analysis and reconstruction and predicts the flow field ability The error analysis verifies the accuracy and reliability of the DMD algorithm application.Secondly,pre-processing the Fracis-99 turbine model,including geometric modeling and meshing.Then,three different working conditions（partial load PL,optimal working condition BEP,high load HL）Francis turbine full flow numerical simulation are carried out.The accuracy of the numerical simulation results is verified through the comparison of the external characteristics,the mean value of the fluctuating pressure at the monitoring point and other parameters with the experimental data.On this basis,the analysis of the variation law of the draft tube vortex zone and curl is carried out.The results show that the numerical calculation structure and the experimental data are basically consistent under the three working conditions.The PL vortex has the largest scale,the highest curl,and downwards along the axis,the scale of the vortex gradually becomes larger,but the energy of the spiral downward development is getting smaller and smaller.The size of the BEP vortex is medium,with the smallest curl,and the size of the vortex gradually becomes larger along the axial direction.The HL vortex has the smallest scale and medium curl,and the scale of the vortex gradually becomes smaller along the axial direction.Finally,using the numerical simulation results and data as samples,the three-dimensional coherent structure and dynamic information of the runner single flow channel and the draft tube cone section of the three working conditions are analyzed using the DMD method.The results show that the BEP coherent structure of the three different working conditions is the most stable,followed by HL,and the coherent structure of the PL flow field has the worst stability;the development process,position and shape of the coherent structure corresponding to each mode are captured;The DMD method can accurately reconstruct the flow field with an error of about 10^-12.The error of the predicted flow field will increase stepwise relative to the error of the reconstructed flow field,but the error is finally stable at about 10^-2.