Reynolds Stress Modeling of Separated Turbulent Flows over Helicopters

Open Access
Alpman, Emre
Graduate Program:
Aerospace Engineering
Doctor of Philosophy
Document Type:
Date of Defense:
January 25, 2006
Committee Members:
  • Lyle Norman Long, Committee Chair
  • Barnes Warnock Mccormick Jr., Committee Member
  • Joseph Francis Horn, Committee Member
  • Savas Yavuzkurt, Committee Member
  • Yousry Y Azmy, Committee Member
  • Reynolds Stress Model
  • Computational Fluid Dynamics
  • Parallel Processing
  • Helicopter Fuselage Aerodynamics
  • Separated Flow
A numerical investigation of inviscid and viscous flows around three-dimensional complex bodies is made using unstructured meshes. Inviscid flow solutions around an RAH-66 Comanche helicopter fuselage are performed to analyze the aerodynamics of ducted tail rotors in low-power, near-edgewise flow conditions. A numerical solution of the Euler Equations is obtained for the flow over the Comanche fuselage with a uniform actuator disk and blade element models for the FANTAILTM; the main rotor is excluded in this study. The solutions are obtained by running the PUMA2 computational fluid dynamics code with an unstructured grid with 2.8 million tetrahedral cells. PUMA2 is an in-house computer code written in ANSI C++. Excellent correlation between the calculations and a variety of static test data are presented and discussed. The dynamic relationship between the antitorque thrust moment and applied collective pitch angle is studied by changing the pitch angle input by five degrees at a rate of 144 degrees per second. Dynamic fan thrust and moment response to applied collective pitch in hover and forward flight are presented and discussed. In order to remove the deficiency of the Euler equations in predicting separated flows, which is mostly the case in helicopter fuselage aerodynamics, a concurrent study is performed to simulate turbulent flows around three-dimensional bodies. Most of the turbulence models in the literature contain simplified assumptions which make them computationally cheap but of limited accuracy. Dramatic improvements in the computer processing speed and parallel processing made it possible to use more complete models, such as Reynolds Stress Models, for turbulent flow simulations around complex geometries, which is the focus of this work. The Reynolds Stress Model consists of coupling Reynolds transport equations with the Favre-Reynolds averaged Navier-Stokes equations, which results in a system of 12 coupled nonlinear partial differential equations. The solutions are obtained by running the PUMA_RSM computational fluid dynamics code on unstructured meshes. Results for high Reynolds number flow around a 6:1 prolate spheroid, a sphere and a Bell 214ST fuselage are presented. For the prolate spheroid basic flow features such as cross-flow separation are simulated. Predictions of mean pressure and circumferential locations of cross flow separation points are in good agreement with experiment. Most of the separation location predictions are in less than five percent discrepancy with measurements. The effects of the freestream turbulence intensity and turbulent Reynolds number are analyzed and discussed. A grid refinement study is performed to improve the computations. The fine mesh solution predicted locations of primary and secondary separation points with errors of roughly two and zero degrees, respectively. Mean pressure and skin friction predictions for the sphere solutions are also in good agreement with the measurements. The computed separation location is very close to the measured one, the error is less than one degree. The distribution of turbulent stresses shows that the turbulent flow around a sphere is highly anisotropic and supports the notion that using anisotropic turbulence models are necessary for three-dimensional separated flows. Flow simulations around a Bell 214ST fuselage are performed for an isolated fuselage at three different flight conditions and helicopter with rotors modeled using momentum theory in forward flight. Predicted pressure and drag force correlate well with the wind tunnel data. For a high angle of attack case drag was able to be predicted with a less than ten percent error. Drag predictions show not only the abilities of Reynolds Stress Model against eddy viscosity models but also its relative speed compared to Large Eddy Simulation. Three-dimensional flow solutions require a considerable amount of computational time and memory requirements. In order to compensate for this, parallel processing is applied with the MPI communication standard. The codes are run on Beowulf clusters. The parallel performance of the code PUMA_RSM is analyzed and presented.