diff --git a/exawind/NREL_Phase_VI_Turbine/README.md b/exawind/NREL_Phase_VI_Turbine/README.md index a5f808e..dff846a 100644 --- a/exawind/NREL_Phase_VI_Turbine/README.md +++ b/exawind/NREL_Phase_VI_Turbine/README.md @@ -1,8 +1,7 @@ - -## NREL Phase VI Turbine -ExaWind simulations are performed for NREL’s Unsteady Aerodynamics Experiment Phase VI rotor[^1]. The experimental conditions are summarized below: +## NREL Phase VI Rotor +ExaWind simulations are performed for NREL’s Unsteady Aerodynamics Experiment Phase VI rotor[^1]. The experimental conditions of the rotor are summarized below: - Tested in the 80 X 120 ft wind tunnel at NASA Ames Research Center - Two bladed rotor with a diameter of 10.529 m @@ -15,35 +14,33 @@ ExaWind simulations are performed for NREL’s Unsteady Aerodynamics Experiment - ExaWind driver version: [a38a4d5f96e4d3b42b52f41280e2d8d28c57ef25]( https://github.com/Exawind/exawind-driver/commit/a38a4d5f96e4d3b42b52f41280e2d8d28c57ef25) - Nalu-Wind version: [f3cecafbdc05e61d0550ff41a30307425ef8197b](https://github.com/Exawind/nalu-wind/commit/f3cecafbdc05e61d0550ff41a30307425ef8197b) - Turbulence / Transition model: SST-2003 with the 1-eq Gamma transition model + - Four Picard iterations - AMR-Wind version: [8bad127f62cf2fd2f0d0ae16f2df47fdd0d069f8]( https://github.com/Exawind/amr-wind/commit/8bad127f62cf2fd2f0d0ae16f2df47fdd0d069f8) ## Freestream conditions - Simulations are performed for both fully turbulent and laminar-turbulent transition conditions at two wind speeds: 7 m/s and 15 m/s, which represent speeds below and above the rated wind speed, respectively. -The tunnel’s turbulence intensity was reported to be below 0.5%, but the precise measurement was unavailable during the experiments. In this work, a turbulence intensity of 0.1% was assumed, which is typical for wind tunnels. The freestream conditions for the k and ω account for the decay of turbulent variables from the input to the blade, with the inlet set 100m upstream of the rotor. The conditions are as follows: +The tunnel’s turbulence intensity was reported to be below 0.5%, but the precise measurement was unavailable during the experiments. In this work, a turbulence intensity of 0.1% was assumed, which is typical for wind tunnels. The freestream conditions for the k and ω account for the decay of turbulent variables from the inlet to the rotor, with the inlet set 100m upstream of the rotor. The conditions are as follows: - Inflow conditions at 7 m/s - - U∞=7.0m/s, ρ=1.246kg/m3, µt/µ=4.5 + - U∞=7.0 m/s, ρ=1.246 kg/m3, µt/µ=4.5 - k∞= 0.007350, ω∞= 115.044281 - Inflow conditions at 15 m/s - - U∞=15.0m/s, ρ=1.246kg/m3, µt/µ=9.7 + - U∞=15.0 m/s, ρ=1.246 kg/m3, µt/µ=9.7 - k∞=0.033750, ω∞=245.071186 ## CFD mesh generation - -The near-body, Nalu-Wind mesh was created using Pointwise from the CAD model. The two blades are connected with a cylinder at the center, while other components such as the spinner or tower were not included in the CFD model. +The near-body (Nalu-Wind) mesh was created using Pointwise from the CAD model. The two blades are connected with a cylinder at the center, while other components such as the spinner or tower were not included in the CFD mesh. - Mesh topology: O-O typed structured mesh - 500 points in the chordwise direction - Initial wall normal spacing: 5e-6m - Size of the overset boundaries: 1.5m from the blade surface -- Wall normal growth rate: 1.15 +- Wall-normal growth rate: 1.15 - Total cell counts: 23,192,978 -Off body -AMR-Wind mesh is generated using the built-in capability of AMR-Wind. Off-body mesh information is summarized below -- Mesh topology: Carstensen with AMR +Off-body (AMR-Wind) mesh was generated using the built-in capability of AMR-Wind. Off-body mesh information is summarized below +- Mesh topology: Cartesian with AMR - Domain in x= -100 to 150m, y=-100m to 100m, z=-100m to 100m - Initial grid size: 0.78125m - Finest cell size: 0.1953 m with 4 AMR levels @@ -51,23 +48,21 @@ AMR-Wind mesh is generated using the built-in capability of AMR-Wind. Off-body m - Total cell counts: 45,527,040 ## Results - The rotor simulations are performed in four sequential stages with reduced time step sizes as follows: - - Rev. 1 and 2: 0.25° - Rev. 3 and 4: 0.125° - Rev. 5 and 6: 0.0625° - Rev. 7 and 8: 0.03125° -This approach was particularly important for the wind speed of 15 m/s, which exhibits highly separated flow, with the sectional angle of attack distribution ranging between 15° and 35°. +This approach was particularly important for the wind speed of 15 m/s to acheive the time-step size conevergence, which exhibits highly separated flow, with the sectional angle of attack distribution ranging between 15° and 35°. The figure below compares the rotor performance (thrust and torque) against experimental data and available other numerical results[^2]. For the Nalu-Wind results, the rotor performance was averaged over the last one revolution. Cf -For the thrust, the transition simulations predict slightly higher values than the fully turbulent simulations, although the differences are minor. For the torque, both fully turbulent and transition simulation results fall within the standard deviation of the experimental measurements. At 7 m/s, the transition simulation predicts slightly higher torque than the fully turbulent simulations due to lower drag from more laminar flow. On the other hand, at 15 m/s, the fully turbulent simulations predict higher torque. This could be attributed to higher turbulent eddy viscosity prediction in the fully turbulent simulation, resulting in more attached flow on the blade section. Overall, both fully turbulent and transition simulations provide reasonable predictions. +For the thrust, the transition simulations predict slightly higher values than the fully turbulent simulations, although the differences are minor. For the torque, both fully turbulent and transition simulation results fall within the standard deviation of the experimental measurements. At 7 m/s, the transition simulation predicts slightly higher torque than the fully turbulent simulations due to lower drag from more laminar flow. On the other hand, at 15 m/s, the fully turbulent simulations predict higher torque. This could be attributed to higher turbulent eddy viscosity in the fully turbulent simulation, resulting in more attached flow on the blade section. Overall, both fully turbulent and transition simulations provide reasonable predictions. -The simulations were performed using 1,440 cores on NREL's Kestrel HPC cluster, 1152 cores for Nalu-Wind, and 288 cores for AMR-Wind. The simulations took approximately 55 hours for the transition simulations. Nalu-Wind used four Picard iterations. +The simulations were performed using 1,440 cores on NREL's Kestrel HPC cluster: 1152 cores for Nalu-Wind, and 288 cores for AMR-Wind. The transition simulations took approximately 55 hours for the total 8 revolutions. ## Reference [^1]: M. H. Hand et al, "Unsteady Aerodynamics Experiment Phase VI: Wind Tunnel Test Configurations and Available Data Campaigns," NREL/TP-500-29955, 2001.