Drag and Lift Validation of Wing Profiles STAR European Conference 2010 London By: Dr Martin van Staden Aerotherm Computational Dynamics 14th IAHR Conference December 2009
Outline of Presentation Background Detailed Fan Modelling 2-D Wing Section Modelling Comparison with Experimental data Mesh & Turbulence model sensitivity Summary & Conclusions
Background Two large Coal fired power stations are currently being built in South Africa which make use of Air cooled Condensers for cooling. These ACC s Consist of 384 fans each with a diameter of 10.4m (34ft). Detailed fan modelling has become an important part of predicting the thermal performance of such a power station.
Background. CFD provides us with the ideal tool to model an ACC in order to understand the complex flows around as well as within the ACC s A-frames. One of the most important parts of the ACC modelling process is to understand how the fans react to poor inflow conditions. Fan performance can therefore be analysed as they will be installed in situ and tested under real operating conditions. This data can then be used in global ACC models to model the entire ACC in order to evaluate it s response to changing wind conditions. An important outcome of the CFD analysis is the predicted fan power for a given blade angle setting.
Detailed Fan Modelling Requirements for fan modelling Global ACC model: Must predict the volume flow rate accurately Must represent system pressure losses accurately Must take into account the affect flow rate as a function of varying pressure losses Must take into account the affect of skewed inflow conditions Must be able to accurately predict fan power consumption Accurate prediction of the Fan power is important as the fan power for a given flow rate is a contracted value and has a heavy financial penalty coupled to exceeding of the contract value.
Detailed Fan Modelling What is a detailed fan model? The fan blades are explicitly modelled Fan is rotated (explicitly or implicitly) Cell sizes as small as 1mm All support structures such as the A-frame, I-beams, fan screen supports, steam ducts, fan bridge, motor and gearbox, fan inlet bell etc are explicitly modelled.
Detailed Fan Modelling The lift and drag is explicitly calculated based on the 3D flow field and pressure filed which develops around the fan blades as a result of the rotation. Rotation is achieved through steady state MRF or explicitly rotation of the mesh (transient) i.e. using a sliding mesh. A test facility was modelled in order to compare the outcome of the CFD results with experimental results
Outcome of test facility simulation Good agreement was found with Pressure vs. Volume flow rate Power was over predicted by the CFD models by more than 11%
2D Wing profile comparison Tests were conducted on 2D wing sections in order to evaluate the expected accuracy for relatively coarse meshes used in detailed fan models. The aim of the study was to identify which modelling parameters were the main cause for large discrepancy in fan power predicted by the CFD models.
Comparison of CFD vs. Experimental data The laser scanned fan wing section profile was matched with a Wortmann FX60-126 wing section. Prof. Ewald Krämer from Stuttgart University was kind enough to provide us with the aerodynamic data for comparison with 2D CFD simulations of the lift and drag.
Initial mesh Low Rey poly mesh (15 boundary layer cells) No wake refinement
Simulation assumptions Free stream velocity of 50m/s was used (Rey=1.94E+6) Inlet turbulence intensity of 0.01 Turbulent viscosity ratio of 10 Used all y+ approach in all turbulence models where this option is relevant
Initial mesh Y+ values
Lift Coefficien Cl Comparison of lift curve FX60-126 - Aerofoil data Cl from CFD 2D profile simulation 2.00 1.50 1.00 Stall point 0.50 Cl CFD Coarse k-e Cl Stuttgart : FX 60-126 - Rey=2e6 0.00-10 -5 0 5 10 15 20 25-0.50 Angle of attack (α)
Drag Coefficient Cd Comparison of drag curve FX60-126 - Aerofoil data Cl and Cd from CFD 2D profile simulation 0.06 Cd CFD Coarse k-e 0.05 Cl Stuttgart : FX 60-126 - Rey=2e6 0.04 0.03 0.02 0.01 0-10 -5 0 5 10 15 20 25 Angle of attack (α)
Finer mesh
Finer mesh y+ values
Flow field at 10º & 15º angle of attack 10º Onset of stall 15º
Lift Coefficien Cl Turbulence models - lift FX60-126 - Aerofoil data Cl and Cd from CFD 2D profile simulation 2.50 2.00 10º 15º 1.50 1.00 Cl CFD Coarse k-e Cl Stuttgart : FX 60-126 - Rey=2e6 0.50 k-e low Rey mesh k-w low Rey mesh Spalat-Almaris 0.00 RS-2l -10-5 0 5 10 k-e_v2-f 15 20 25 k-w_trans -0.50 Angle of attack (α)
Drag Coefficient (Cd) Turbulence models - drag 0.06 FX60-126 - Aerofoil data Cl and Cd from CFD 2D profile simulation Cd CFD Coarse k-e Cl Stuttgart : FX 0.05 60-126 - Rey=2e6 k-e low Rey mesh k-w low Rey mesh-refine1 Spalat-Almaris 0.04 RS-2l k-e_v2-f k-w_trans 0.03 10º 15º 0.02 0.01 0-10 -5 0 Angle 5 of attack (α) 10 15 20 25
Skin Friction Coef. Pressure Coefficient 5000 Pressure Coefficient 0 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1-5000 -10000 k-e low rey -15000 k-w -20000 Spalat-Almaris -25000-30000 RST_2l k-e v2f -35000 Position (m)
Skin Friction Coef. Skin Friction Coefficient 300 Skin Friction Coefficient 250 k-e low rey 200 k-w 150 Spalat-Almaris RST_2l 100 k-e v2f 50 0 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 Position (m)
Summary Pressure vs. Shear Pressure vs. Shear drag @ 10deg & 15deg angle of attack 120% 100% Pressure Shear 10 deg Average 68% 32% 15 deg Average 86% 14% 15deg Pressure 10deg Pressure 15deg Shear 10deg Shear 80% 60% 40% 20% 0% k-e low Rey mesh RS_2l k-e_v2-f k-w low Rey mesh k-w_sst Spalat-Almaris
Pressure vs. Shear On average at 15º angle of attack the drag due to pressure accounts for 86% of the drag as opposed to 68% for an angle of attack of 10º. This leads one to suspect that the pressure drag may be the component which is mainly to blame Further mesh refinement studies could confirm and/or quantify this assumption. Better definition of the profile geometry could reduce the shear drag. Sensitivity to inlet turbulence levels still has to be investigated
Summary and Conclusions Lift and stall point are predicted with a high degree of accuracy even with relatively coarse meshes using the Realizable κ-ε turbulence model with an all Y+ wall function formulation. Drag however is highly over predicted with all turbulence models that were tested. This coincides with the higher power predicted in detailed fan simulations. Further work has to be performed on mesh sensitivity studies as only 2 mesh sizes have presently been investigated. Experimental work is underway at Universities to evaluate drag and lift forces explicitly in order to obtain a further set of independent aerodynamic data
Thank you for your time! Questions? 14 th IAHR Conference December 2009
14 th IAHR Conference December 2009
Mesh Sensitivity on Detailed Fan COARSE MESH (1.7 million cells) FINE MESH (5.3 million cells)