A. Spentzos 1, G. Barakos 1, K. Badcock 1 P. Wernert 2, S. Schreck 3 & M. Raffel 4 1 CFD Laboratory,...

Post on 28-Dec-2015

214 views 1 download

Tags:

transcript

A. Spentzos1, G. Barakos1, K. Badcock1

P. Wernert2, S. Schreck3 & M. Raffel4

1 CFD Laboratory, University of Glasgow, UK 2 Institute de Recherche de Saint Louis, France3 National renewable energy laboratory USA4 DLR - Institute for Aerodynamics and Flow Technology, Germany

Numerical Simulation of 3D Dynamic Stall

Outline

Background and Objectives Past efforts in 3D dynamic stall CFD requirements for validation Summary of selected tools 2D dynamic stall Validation cases and results Conclusions

Motivation and Objectives

DS is encountered in rotorcraft and highly maneuverable aircraft

Complex problem – prediction of loads and flow structure 3D studies are rare Study 3D DS, use a variety of turbulence models and

simulation (LES) Improve existing turbulence models Understand flow physics Validate CFD so that industry can exploit Take things a bit further…

Background

What is Dynamic Stall? Experimental and CFD work on DS The majority of the work performed on DS (experimental

and CFD) has been done on 2-D Most CFD has been done for code validation rather than

investigation of the flow physics. 2D CFD suggested that turbulence modelling is a key

issue if fidelity is required Missing: 3D, centrifugal effects, dM/dt, interaction with

wake

CFD requirements for validation

Surface pressure Integral loads Boundary layers Information for turbulence levels in the tunnel

and transition Higher Mach numbers Near-tip and flow-field measurements Measurements on rotating blades Measurements on more complex geometries

Summary of experiments Most experiments on DS are 2D 3D work has been done by the following:

Selected Validation Cases

CFD solver PMB solver of the Univ. of Glasgow Control volume method Parallel (distributed memory) Multi-block (complex geometry) structured grids Moving grids Unsteady RANS - Variety of turbulence models – LES Implicit time marching Osher's and Roe's schemes for convective fluxes MUSCL scheme for effectively 3rd order accuracy Central differences for viscous fluxes Conjugate gradient linear solver with pre-conditioning Validation database

www.aero.gla.ac.uk/Research/CFD/validation

2D Results for Ramping and Oscillating Aerofoils

CFD results for dynamic stall of helicopter sections

Flow Field Comparison

Sinusoidal pitch, k=0.15, Re=373,000, M=0.1

a) 22 Deg (upstroke) b) 23 Deg (upstroke) c) 24 Deg (upstroke)

Geometry – Grid Generation

Geometry – Grid Generation

One-block extruded tip

Geometry – Grid Generation

C-O topology

4-block extruded tip

Grid and Time Convergence

Three levels of refinement: 120k, 800k, 1,800k

Grid and Time Convergence

Two levels of time refinement resolving frequencies up to 20 Hz and 40Hz

Experimental evidence of the -shaped vortex

3D CFD

Schreck & Hellin2D CFD

Coton et al.

Surface Pressure

Ramping motion,Re=69,000, M=0.1, K=0.1Incidence 40.9 degrees

Experiment CFD

Close the loop – AnalysisONERA model

Cz Cz

Cz

C C Cz z z 1 2

C C C k C k C k C kz z z s z c1 z s z c2 20 2 1 2 2 2 2 22 2 sin cos sin cos C C C k C kz z z s z c 0 sin cos

Close the loop – AnalysisONERA model

Conclusions

Experimentalists like CFD pictures! Are keen to collaborate and look in their

databases for measurements They developed the ability to understand

much about the flow from a small number of measurements

They are getting used to the idea of CFD…or at least looking at CFD results

Conclusions

CFD developers are always looking for good data and have many requirements

Have sometimes to make a first step Have to be open about any limitations of

their methods Perform simulations, validation,

comparisons and maybe …some analysis!