**Principal Investigator:**

Thorsten Lutz

**Affiliation:**

Institute of Aerodynamics and Gas Dynamics (IAG), University of Stuttgart (Germany)

**Local Project ID:**

WEAloads

**HPC Platform used:**

Hazel Hen of HLRS

**Date published:**

*As part of the WindForS project WINSENT two wind turbines and four met masts will be installed in the Swabian Alps in Southern Germany for research proposes. The results of highly resolved numerical simulations of this wind energy test site located in complex terrain are shown. By means of Delayed Detached Eddy Simulations (DDES) the turbulent flow above a forested steep slope is analyzed in order to evaluate the inflow conditions of the planned wind turbine in detail. The complex inflow conditions and production of turbulence due to the shape of the topography and the vegetation are evaluated. The intention of using supercomputers for these applications is to analyze the local atmospheric flow field in as much detail as possible.*

The current CFD study is performed by using the compressible block-structured flow solver FLOWer [1] from the DLR. This code is permanently extended for wind energy and other rotating body applications at the authors’ institute. For example the flow solver was extended by a WENO scheme with a spatial order of five [2]. This scheme improves the turbulent propagation by a better preservation of vortical structures to characterize the local flow field in much more detail. The CFD simulations were unsteady Delayed Detached Eddy Simulations (DDES) [3] with the Menter SST turbulence model and the Smagorinsky model as a subgrid scale model.

By using Reynolds Averaged Navier Stokes equation (RANS) simulations general effects characterizing complex terrain flows like a large inclination of the flow and a speedup due to the steep slope can be estimated. With the hybridization of RANS and LES, it is possible to resolve eddies detached from the wall [4] e.g. in the wake of the forest.

The simulation domain contains a complex terrain site with an escarpment of about 200 m height. The slopes are almost fully covered by a dense forest. Thus, the forest has to be considered in the CFD simulations by applying a canopy model [5]. The geometry of the CFD domain extents over a region of approximately 1200 m × 1800 m × 1100 m. In this case the atmospheric inflow at the valley is characterized by a turbulence intensity of 10% and a reference velocity of 8 m/s at 70 m above ground.

The physical effects of the local flow field become apparent by Figure 1. In this figure the instantaneous flow field of the fully turbulent flow at the terrain site over the forested steep slope is depicted. The surface of the terrain, colored by height above the sea level, the forest mesh geometry, some streamlines in the flow field, λ_{2 }vortex core structures, and a plane of the axial component of the local flow field on top of the plateau perpendicular to the inflow direction are highlighted.

The local flow field is divided into five characteristic zones to analyze the physical interactions between the wind flow and the forested topography separately. For clarity the respective effects are only visualized for some sections of the flow domain. Thus, it is possible to analyze all effects separately within a combined plot.

Zone 1 describes the turbulent flow just above the forest, which is influenced significantly by the topography. A large speedup of the flow due to the steep slope is visible. This is depicted by some streamlines approximately 50 m to 100 m above ground that become accelerated from 7 m/s to 10 m/s. This speed up becomes obvious by the red coloring of these streamlines.

The highly turbulent structures in the wake of the forest are shown by zone 2. Applying the WENO scheme, vortex structures within the flow domain can be resolved and propagated through the flow field with less dissipation. The interaction of the high speed flow of zone 1 and the strongly decreased wind speed of the forest wake lead to a shear layer that effects large fluctuations and a high turbulence intensity. This becomes apparent by comparing the few structures of the λ_{2 }vortex cores in the valley and the large field of turbulent fluctuations in the wake of the forested escarpment.

Zone 3 highlights the flow crossing the forested escarpment area near the ground. Due to the forest shape and the hilly 3D orography of the complex terrain, the streamlines are strongly crooked. Moreover, the blue coloring of these streamlines indicates the flow deceleration due to the forest drag force.

The turbulent upper flow field is described by zone 4 which is only marginally influenced by the topography. The streamlines fluctuate spatially and temporally due to atmospheric gusts but are not topographically inclined to a large extent. Consequently, in the mean there are no changes of the velocity in streamwise direction above the escarpment in contrast to zone 1.

A plane perpendicular to the axial flow field at the flat plateau on top of the hill shows the wake of the flow in zone 5, which has crossed the forested steep slope and the area above. The turbulent structures are visible in the whole domain of this plane. These structures show larger turbulence intensities due to the topography and the vegetation compared to the flow field in the valley. Colored in blue the highly turbulent forest wake becomes apparent, which highlights a large increase of turbulent fluctuations.

A typical job with 220 million grid points ran on 6,400 cores for 200,000 iterations. Hence, the total computional effort of a single job was around 1,000,000 CPU hours.

**References**

[1] Becker, K.,Kroll, N.,Rossow, C.C., Thiele, F: The MEGAFLOW project, Aerosp. Sci. Technol. 4 pp. 223-237 (2000)

[2] Kowarsch, U., Kessler, M., Krämer, E.: High Order CFD-Simulation of the Rotor-Fuselage Interaction, 39th European Rotorcraft Forum (2013)

[3] Weihing, P., Letzgus, J., Bangga, G., Lutz, T., Krämer, E.: Hybrid RANS/LES Capabilities of the Flow Solver FLOWer - Application to Flow Around Wind Turbines Notes on Numerical Fluid Mechanics and Multidisciplinary Design 137 pp. 369-380 (2018)

[4] Bechmann, A., Sørensen, N.N.: Hybrid RANS/LES applied to complex terrain, Wind Energ.14 pp. 225-237 (2011)

[5] Letzgus, P., Lutz, T., Krämer, E.: Detached Eddy Simulations of the local Atmospheric Flow Field within a Forested Wind Energy Test Site located in Complex Terrain, Journal of Physics: Conference Series 1037 (2018)

**Key Facts of a typical compute job:**

- 6400 cores
- 220 M grid points
- 4 TB data
- 200 000 iterations

**Scientific Contact:**

Patrick Letzgus

Institute of Aerodynamics and Gas Dynamics

University of Stuttgart

Pfaffenwaldring 21, D-70569 Stuttgart (Germany)

e-mail: patrick.letzgus[at]iag.uni-stuttgart.de

*HLRS Project ID: WEAloads*

*Date published: June 2019*