Ansys Fluent Day 2, Turbulence & Meshing

Turbulence Validation: NACA0012 Airfoil

• Let's confirm the k-omega SST Turbulence model in Fluent
• From Workbench, drag/drop a Fluent Component (not Fluent Meshing)
• Open in 2D mode, and import the NACA 0012 Airfoil here: NACA0012_0.msh
• (Reference: https://turbmodels.larc.nasa.gov/naca0012_val.html)

Setup:

• Audit the model
• Note the Mesh density!
• Inlet = 106 m/s
• Initialize and solve 150 iterations

Post-Process:

• Review the velocity, note the distance between the airfoil and the model limits
• Check for Mesh Independence
• Create a Pressure Coefficient (Cp) Contour

Post Process:

• Discuss the definition of Cp
• Plot the Cp along the top edge of the airfoil
• Correlate the Cp ratio to the NASA Data
  • Report: Vertex Max, Pressure, Cp
  • Report: Vertex Min, Pressure, Cp
  • Report: Expression = Max Cp/Min Cp
• The answer should be 0.41!

Investigate:

• Let's use Fluent to determine the stall Angle of Attack of the NACA0012 airfoil.
• During this investigation:
  • Post-process the turbulence values
  • Observe the convergence for signs of oscillation
•  Download the pre-modeled mesh files here: NACA0012_aoa_sweeps.zip

Turbulence:

• “Unpredictable” changes in the flow characteristics: velocity and pressure
• Characterized by vortices (eddies) within the flow
  • Vortices can be large or small
  • Vortices can be fast or slow
  • Vortices can live for a long time for a short time
  • This all happens at the same time
• Turbulence is constantly changing in Length AND Time
  • Video Example 1 and Example 2

Turbulence:

• Very computationally expensive, so its "modeled, not meshed"
• CFD codes are put in categories based on how they handle Turbulence:
  • RANS – Reynold’s Averaged Navier Stokes (Time-Averaged Turbulence)
  • DES – Detached Eddy Simulation (Combo of RANS & LES)
  • LES – Large Eddy Simulation (Ignores small vortices)
  • DNS – Direct Numerical Simulation (Resolving flow on time and length scales!) (EXAMPLE)

Reynolds Averaged Navier-Stokes (RANS):

• In the boundary layer, the mesh must be accurately sized (not too big, and not too small!)
• The Law of the Wall:
  • “The average velocity of turbulent flow is proportional to the log of the distance from the wall”
• We use the “y+” value to determine the proper wall layer height:
  • y+ = y * Friction Velocity/Kinematic Viscosity
  • y+ = distance from the wall * flow characteristics/fluid properties

Reynolds Averaged Navier Stokes (RANS):

• k-epsilon: Standard turb model used in many applications (30<y+<300)
• SST k-omega: Recommended for external Aero, detached flows (1<y+<10)
• SAS: Vortex shedding, variable wake
• DES: Separation/High Reynold’s Number
• RNG: Reattachment (Flow over a backward facing step)
• Low Re k-e: 1,500<Re<5,000, and “jets”
• Mixing Length: Designed for Internal Natural Convection (gases)
• Eddy Viscosity: For low speed turbulent flow if the k-epsilon won’t work…
• GEKO – Generalized K-Omega
• More Information in Ansys Help
  • User's Guide Chapter 13 Modeling Turbulence
    • 13.2 Choosing a Turbulence Model
  • Theory Guide Chapter 4: Turbulence

Steady State Subsonic External Flow - The Vehicle in Ground Effect

• Performance characteristics are defined by the forces acting on the vehicle
  • Drag, Thrust, Lift, Gravity
• Lift forces come from: Fluid being redirected and the Bernoulli Effect
• Drag Forces come from:
  • Form: Creation of turbulence and low pressure zone in wake
  • Skin Friction Drag: Viscous Shear forces in Boundary Layer
  • Lift-Induce Drag: Creation of turbulent vortices in the wake

Steady State Subsonic External Flow - The Vehicle in Ground Effect

• Fdrag = Cd*(1/2*rho*V^2)*A
• Flift = Cl*(1/2*rho*V^2)*A
• Typically, Cd & Cl are obtained experimentally
  • Cd is relatively easy to research for passenger vehicles ~0.30 (wiki)
  • Cl not so much… (so let’s use CFD!)
• Simple Convertible Calcs:
  • A = 17.7 ft2
  • Cd = 0.38 (Top down)
  • V = 100mph

Prepare:

• Use SpaceClaim to prepare the geometry for meshing
• Fix "dirty" CAD
  • Extra Edges; Inexact Edges
• Create Named Selections
• Download model here: miata_air.x_t

Mesh:

• Import into the Watertight Workflow
• Create a volume mesh using the default mesh sizes
• Improve the mesh quality as needed
• Help: Fluent User Guide 24.2 Quality Measurements

Setup:

• Inlet - 100 mph
• Start with K-Epsilon Turbulence model
  • 30<y+<300
• Start with 1e-3 residuals
• Start with 200 iterations
• Solve & Review

Refine:

• The turbulent wake needs to be resolved finer
• Utilize a Body of Interest (BOI) in SpaceClaim & Fluent Meshing
• What surfaces need a surface refinement?

Assess for Accuracy in Turbulence Modeling:

• Review the y+, does it correspond to our Turbulence Model?
• Calculate the drag force, does it correlate to the published Cd value?
• Would a K-Omega turbulence model be more accurate?

Optimize:

• Can we use the results of this analysis to make the design better?
• How might we reduce the drag?
• Pick an idea and try it!

Brief Overview on CFD Validation:

• In a Wind Tunnel
  • Quantitatively: Measuring pressure with a Pitot Tube
  • Qualitatively: Using smoke trails
• In the real world
  • Quantitatively: Pitot tubes, Coast Down test
  • Qualitatively: Flow viz paint

© Fastway Engineering. All Rights Reserved. May not be duplicated without express written consent.