Investigation of Delta Wing Time Dependent Flow Characteristics with Lattice-Boltzmann Method Gauss Centre for Supercomputing e.V.


Investigation of Delta Wing Time Dependent Flow Characteristics with Lattice-Boltzmann Method

Principal Investigator:
Erol Oezger

Technische Hochschule Ingolstadt (Germany)

Local Project ID:

HPC Platform used:
Hermit and Hornet of HLRS

Date published:

The project DWLBM – Investigation of Delta Wing Time Dependent Flow Characteristics with Lattice-Boltzmann Method is carried out by the Technische Hochschule (TH) Ingolstadt, Faculty of Mechanical Engineering, at the High Performance Computing Center Stuttgart (HLRS). It focuses on the numerical investigation of the delta wing flow under various aspects such as sweep angle variation, sharp or round leading edges, as well as high and lower Reynolds numbers with a Lattice-Boltzmann PowerFLOW solver provided by Exa.

The Lattice-Boltzmann approach shows advantages in non-stationary flow phenomena and is normally used in the car industry. There are numerous delta wing flow calculations performed with classical Navier-Stokes approaches. Therefore, this investigation can be considered as the first one to survey the flow over a delta wing configuration in a systematic manner with this Lattice-Boltzmann approach. One purpose is thus to evaluate the computing performance of the Lattice-Boltzmann approach and its prediction capability comparing the simulation results with experimental wind tunnel data.

Delta wing flow is vortex dominated where primary, secondary, and tertiary vortices are produced at the leading edge, interacting with each other and the wing boundary layer. At increasing angles of attack, the vortex layers detach at the leading edge to form free shear layers. The figure below shows an extract of the simulation results with the flow solver.

Computational Resources used on Cray XE6 Hermit and the Cray XC40 Hornet platform

For the vortex flow on a delta wing the correct prediction of all vortex structures, even the small ones, determine the force and moment characteristics on the wing, especially at higher angles of attack, round leading edges or lower Reynolds numbers.

Due to the challenge of resolving very fine vortex structures such as the secondary or even the tertiary vortex the number of grid points becomes so large that adequate computational power is needed to determine a solution in acceptable time scales. The table below gives an overview on some configurations their required computation time (kCPUh), number of processors and the simulation time.


The full analysis of all results has not yet been finished at this point because for a detailed analysis the amount of data gathered requires more time than planned. In a first step, it has to be shown that the new solver can be verified with experimental data.

The figures shown below show the predicted drag, lift and pitching moment coefficients (C_D, C_L, C_my) of various grid configurations compared with experimental data from the ONERA wind tunnel gathered at angle of attack 13° and a leading edge sweep angle of 65°. It can be shown that with the optimal grid configuration the experimental values is captured.

Further computations of a configuration with 65° leading edge sweep angle at angle of attack 23° are performed and compared with surface pressure measurements done at the Technische Universität München. The following figures show the comparison of the surface pressure profiles at various planes along the x-axis of the delta wing (x/c=0.2, 0.4, 0.6, 0.8, 0.95). Apart from the first plane the flow solver is able to predict the experimental pressure profiles. This makes clear that there is still modelling effort needed at the apex region of the delta wing where the vortex arises whereas the vortex physics downstream are better captured.

The next months will be used to further analyze the effect of sweep angle variation, leading edge radius, or Reynolds numbers. A further experiment at the TH Ingolstadt at low Reynolds numbers will provide the necessary experimental results to further compare the simulation results with the experimental ones. One key to correctly predict the delta wing flow is a turbulence model which captures the flow physics when and where exactly the vortices are generated, convected downstream and burst eventually. The location of the primary and secondary vortices on the wing is of paramount importance for the moment coefficients.

Scientific Contact:

Erol Oezger
Faculty of Mechanical Engineering
Technische Hochschule Ingolstadt
Esplanade 10, D-85049 Ingolstadt (Germany)
e-mail: Erol.Oezger [at]

TH Ingolstadt, Faculty of Mechanical Engineering

Tags: HLRS CSE Technische Hochschule Ingolstadt