Numerical Simulation Of Axial Compressor And Construction Of Entropy Generation Loss Model Based On The Hybrid RANS/LES Method

Efficiency is one of the most important optimization target in the design of the compressor.Since 1940s,many semi-experience loss models have been established to optimize the efficiency of compressor and to analyze the flow loss mechanisms through the 2-dimensional cascade experiment.However,with the development of the compressor,the traditional loss models are incompetent with the design and flow mechanism analysis of the modern compressor.According to the second law of thermodynanics,reducing the flow irreversible loss is the basic way to improve the efficiency and the entropy generation rate is the only parameter to represent the flow irreversible loss.Based on these physical concepts,this thesis presents a local loss model,an integral loss model and an integral loss coefficient of rotor using the hybrid RANS/LES method.Meaningwhile,a simulation of an axial compressor were carried out through two new in-house compressible CFD codes,which is used to test the reliability of the entropy generation loss model.There are four main research contents in this thesis:1.Based on the compressible pressure algorithm,use OpenFOAM to develop compressible CFD solvers in the rotational reference frame;2.Test the solvers by some basic cases;3.In terms of the compressor cascade,make simulation of Hybrid RANS/LES method.Establish and validate the flow loss models proposed based on the concept of entropy generation rate;4.In terms of the compressor rotor,make simulation of Hybrid RANS/LES method.Validate and optimize the entropy generation loss models.In the first part,the official incompressible solver of OpenFOAM,pimpleFoam,was used to introduce OpenFOAM how to process the computational algorithm of Navier-Stokes equations,linearization of the nonlinear equations,calculation of equations with huge matrix.After that,the computational algorithms of the new in-house compressible solver,TurboSimpleRFoam and TurboPISOFoam were introduced.TurboSimpleRFoam was developed based on the compressible SIMPLER algorithm.TurboPISOFoam was developed based on the compressible PISO algorithm.The TurboSimpleRFoam is used to calculate the initial condition of the TurboPISOFoam.They all have the ability to make all-speed simulation.Using the parallel module of the OpenFOAM,they can carry out the massively parallel computation.Meanwhile,they have access to the mainstream forms of structure or unstructure meshes,which make the solvers have ability to simulate complex geometries.At last,the turbulent models and boundary conditions used in this thesis were introduced.In the second part,the classic test cases:a low-speed 3-dimension backstep flow,the transonic RAE 2822 profile flow and NASA Rotor 37,were used to test TurboSimpleRFoam and turboPISOFoam.It is shown that the solvers can simulate the low-speed flow with second order difference accuracy robustly.The solvers can give reasonable RANS,URANS and hybrid RANS/LES results.The hybrid RANS/LES results are more accuracy than the ones of the other two methods.Especially,in the simulation of the separation flows behind the backstep,the hybrid RANS/LES method can give a more accurate axial shear stress,static pressure coefficient on the wall and more detailed flow structures.In terms of the simulation of the transonic flows,the solvers can give correct total characteristics.However,the difference accuracy and the ability of shock capture should be improved.In the third part,a local loss model and an integral loss model based on the hybrid RANS/LES model were evaluated theoretically.Using a well-tested mesh,the accuracy of computational results and the reliability of the local and the integral loss models were validated.The local and integral loss distribution varying with flow incidence angles were studied in detail.The local loss-intensity distribution described by the local loss model can be explained well by the 3D flow structures.The trend of the total flow loss calculated by the integral loss model is the same as that exhibited by the total pressure loss coefficient with the incidence angle.The boundary layer shear loss is almost uninfluenced by the incidence angle.The secondary flow loss in the wake flow and blade-passage regions changes dramatically with the flow condition due to the occurrence of corner stall.Insight can be derived using the iso?surface contours of Lc to identify the 3D loss distribution.With increasing incidence angle,three types of secondary flow loss change dramatically.They are the passage shear loss,which is located at the interface region between the main flow and the outside of the corner separation flow;the corner wake-flow shear loss,which is located in the mixing region between the boundary layer flow of the pressure side and the corner separation flow;and the tornado vortex trailing loss,which represents the loss generated at the top of corner separation flow where the main flow structure is the tail of the corner tornado vortex.In the forth part,using the TurboPISOFoam solver,a low-speed axial compressor was numerically studied by the hybrid RANS/LES method.It is shown that,using the SSTDDES model,TurboPISOFoam can present reasonable flow field of the low-speed compressor.The relation between the pressure ratio and mass flow coefficient,the pressure distribution at the wake region and the axial shear force on shroud are reasonable.In terms of the single flow condition of rotor,the integral loss model and the local loss model can give reasonable local loss distribution.In the high performance condition and the low mass flow condition,the distribution of the integral loss along the radial direction can be related with the tip leakage flow and the corner separation flow.In the region between 90%span and the shroud,the integral loss along the axial direction and the distribution of the local loss coefficient are consistent with the load distribiton near the blade tip.The integral loss coefficient of rotor,which takes the power output capability into account,can be used to analyze the contribution of flow loss to the variation of adiabtic efficiency in a user-difined region when the flow condition is changed.