Applied CFD around/inside complex geometries

Despite extensive research and significant progress in CFD over the last 30 years, there are still several technical challenges with respect to accuracy and computational efficiency when computing aerodynamic flows around and inside aeronautical configurations. CFD simulations of engineering interest include high Reynolds numbers and complex geometries, and are mostly under-resolved with respect to the grid. There is significant progress in the framework of LES, however, RANS will continue to be used in engineering design studies, at least in the foreseeable future. We have developed a CFD software suite, called Azure [1,2], which aims to address some of the present CFD challenges in a threefold way: (i) enable the assessment of accuracy and efficiency of different numerical schemes and physics models by incorporating them in the same computational framework, based on block-structured [3,4] or unstructured/hybrid grids [5,6]; (ii) implement the most suitable computational approach for each problem; (iii) offer a high-fidelity CFD framework that offers implicit large eddy simulations (ILES); Reynolds-Averaged Navier-Stokes (RANS) models; and Detached Eddy Simulation (DES).

The common basis of the RANS, ILES and DES methods in the framework of Azure is the use of high-resolution and high-order methods, which has been implemented in a variety of flows and grid topologies. We have investigated the accuracy and efficiency of Azure for problems ranging from fundamental turbulent flows and aicraft wings to complex airctaft configurations.

Unstructured surface mesh of the DLR-F6 with CFM56-type nacelles and results from RANS simulations; Mach number of 0.75, zero angle of attack, and Reynolds number of 3x106; [1,2] using methods of [5,6].

Pressure contour plots around the DLR F11 high-lift configuration; comparisons between experiment and Azure simulations for ten spanwise cross-sections along the wing [1,2].

Using high-fidelity CFD based on Implicit Large Eddy Simulations, we have studied [6] the time-varying airwake which is shed behind any ship which is moving relative to the wind. The behaviour of this airwake is of key importance to helicopter pilots, who must land safely on the flight deck in condi-tions which are potentially lethal. During the approach to the ship the pilots typically have to deal with adverse weather conditions, rolling and pitching of a very small landing deck and the presence of turbulence generated by the ships superstructure. To enable pilots to decide if it is possible to land on a ship in a given weather condition, it is common practise to define ship/helicopter operating limits (SHOLs) for a given ship-helicopter combination. The operating limits are given as a function of the wind over deck (WOD) velocity. Determining the SHOLs is extremely costly, requiring availability of the ship and test pilot time. The ship/air interface framework (SAIF) aims to enable experienced test pilots to predict the SHOLS for a given naval vessel years before the real test flights are undertaken and to give improve-ments to the quality of the operating limits, using advanced flight simulators An important component in the SAIF project is the provision of a time-accurate airwake into the simulation framework. The aim was to verify if benefits can be gained by replacing with ILES the existing time-averaged data produced using RANS computations. Although existing simulations included a fluctuating velocity added to the time averaged data, this fluctuating velocity could not be expected to have the correct spectral content as compared to a real airwake. In reality, the airwake consists of complex, interacting coherent structures. These coherent structures are, of course, dependent on the actual physical structure of a given ship. The vast majority of the previous works use compressible upwind numerical methods to simulate the flow field, usually rescaling the flow velocity up to Mach 0.3. Most simulations run at Mach 0.3 because a) it is more computationally efficient (i.e. a smaller number of time steps per simulation) and b) upwind methods are excessively dissipative at low Mach numbers. With regard to the second reason, it has been shown [7] that the dissipation of turbulent kinetic energy in an upwind scheme increases as the inverse of the Mach number. Hence if the simulation was run at the true Mach number then the upwind method would dissipate the majority of the turbulent structures. However, even when rescaling the flow Mach number to 0.3, the instabilities which seed turbulent fluctuations occur at a much lower Mach number, meaning that they will still be excessively damped. This prevents the natural turbulent breakdown of large structures and the growth of very small structures, producing very dissipative results.

The current work differs notably from previous studies in that it employs a very high order accurate structured multi-block method instead of an unstructured approach in the framework of Implicit Large Eddy Simulation, which also incorporates low Mach correction to reduce the numerical dissipation at low speed flows.

Visualisations of isosurfaces illustrating the flow features at wind-over-deck angles 00 to 450

To enhance our understanding of complex dynamics governing a helicopter flow, several test cases, corresponding to different flight conditions parameters, we have performed simulations over helicopter configurations using RANS and DES. The predictive capabilities of the employed CFD solvers were assessed through comparisons with experimental data [8]. Furthermore, we have worked on the applications of high-order methods and turbulence models for modelling dynamic stall around helicopter blades.

Third-order high-resolution Detached Eddy Simulations around a helicopter fuselage (left); evolution of dynamic stall vortex (at two different time instants) around an oscillating helicopter blade (right).


  • A.F. Antoniadis , P. Tsoutsanis, Z. Rana, I. Kokkinakis, D. Drikakis , Azure: An Advanced CFD Software Suite Based on High-Resolution and High-Order Methods, AIAA Aerospace Sciences Conference, 2015, in preparation.
  • A.F. Antoniadis , P. Tsoutsanis, D. Drikakis, High-Order RANS Solutions for Full Aircraft and High-Lift Devices, Proceedings of the Royal Aeronautical Society Applied Aerodynamics Conference, Bristol, UK, 2014, in preparation.
  • D. Drikakis, Advances in turbulent flow computations using high-resolution methods, Progress in Aerospace Science, 39, 405-424, 2003.
  • Drikakis, D., Hahn, M., Mosedale, A., Thornber, B., Large Eddy Simulation Using High Resolution and High Order Methods, Philosophical Transactions Royal Society A, 367, 2985-2997, 2009.
  • Tsoutsanis, P., Antoniadis, A., and Drikakis, D. WENO schemes on arbitrary unstructured meshes for laminar, transitional and turbulent flows, Journal of Computational Physics, 256, 2014, 254-276.
  • P. Tsoutsanis, V.A. Titarev, D. Drikakis. WENO schemes on arbitrary mixed element unstructured meshes in three space dimensions, Journal of Computational Physics, Vol. 230, 4(20), 1585-1601, 2011.
  • B. Thornber, M. Starr, D. Drikakis, Implicit large eddy simulation of ship airwakes, The Aeronautical Journal, 114, 1162, 715-736, 2011.
  • B. Thornber, D. Drikakis, R. Williams, D. Youngs, On Entropy Generation and Dissipation of Kinetic Energy in High-Resolution Shock-Capturing Schemes, Journal of Computational Physics, 227, 4853-4872, 2008
  • A.F. Antoniadis, D. Drikakis, B. Zhong, G. Barakos, R. Steijl, M. Biava, L. Vigevano, A. Brocklehurst, O.Boelens, M. Dietz, M. Embacher, W. Khier, T. Renaud, Assessment of CFD methods against experimental measurements for helicopter flows, Aerospace Science and Technology, 19, 1, 86-100, 2012