Catálogo de publicaciones - libros

The flow over a series of two-dimensional hills has been computed by partners ICSTM and UMIST with 8 models. The flow separates smoothly after the first hill and reattaches naturally before the second, making it ideal to test the ability of the different models to predict these difficult features. Two equation models generally fail early separation or are affected by slow post-attachment recovery. The wall resolved SSG and the AJL NLEVM provide better predictions.

The A-airfoil has served as a validation test case in previous EU projects, EUROVAL, ECARP and LESFOIL, using RANS approaches for the first two projects and LES methods in LESFOIL. Nevertheless, and demonstrated below, there is still considerable interest in further investigation of that airfoil case for both turbulence model validation and testing of numerical schemes and methods. In that respect, new RANS computations for the full three-dimensional geometry (wing in wind tunnel) resulted in obvious 3D effects t hat put the two-dimensional approaches “somewhat” in question, or vice versa.

The objective of the present study is to analyse the main compressibility terms involved in the modelling of the turbulence kinetic energy equation for the upper transonic flow regime that is of priority interest for the aeronautical industry. It is well known that mostly the modelling of this equation comes from incompressible flow assumptions and for a long time it was admitted that compressibility effects become significant only beyond Mach numbers higher than 4 (Morkovin hypothesis). However, this stands only for steady flows. In case of unsteady boundary-layer-shock wave interaction and of appearance of recirculation regions near the wall, the van Driest law and Morkovin hypothesis for the turbulent fluctuations are no longer valid. For these reasons, it is also well known that the attempts to predict transonic flows in the upper transonic regime by simply ‘transposing’ the turbulence modelling hypotheses from the incompressible flow give very poor prediction of the shock position and of the drag coefficient. There are few developments for modelling of compressibility effects (Sarkar, (1992), (1995), Zeman, (1990)) among other. However, these developments had given poor improvements in the prediction of transonic flows around bodies. The aim of the present study is to address the terms to be modelled in the energy equation as well as to evaluate these terms in their exact form and according to the different suggested modelled assumptions in order to investigate the efficiency of the modelling assumptions. This can be done essentially by means of a DNS database, because the experimental measurement of compressibility effects is very difficult, although there remains the restriction for DNS concerning the low Reynolds number range of their realisation. IMFT has provided in the FLOMANIA programme a detailed DNS data base for the unsteady transonic flow around a NACA0012 wing at zero angle of incidence. Furthermore, the study of onset of the turbulent motion in its early stages has been carried out for the incompressible flow around a NACA0012 wing at 20°. In previous studies of the same research group (Bouhadji and Braza (2003a,b)) it had been shown that the increase of the Mach number in the upper transonic regime causes a natural amplification of a von Kàrmàn instability and of a strong shock-boundary layer and shock-vortex interaction near the trailing edge. Therefore, this test case is an ideal one for the investigation of compressibility terms under the effect of unsteadiness, for aerodynamic flows around bodies. Thanks to a fully parallelised version of the EMT2/IMFT code ICARE and to the use of a consequent grid size and small time step, it has been possible to perform DNS at Reynolds numbers 5,000 to 10,000 for the flow around the NACA0012 wing. The outcomes of this study are described in the following.

This chapter deals with the validation of a number of turbulence models for industrial relevant configurations, i.e. the DLR F6 wing-body model with and without pylon and through-flow nacelles. In the comparison, which employs structured and unstructured grids of different resolution, some light is shed on the performance of the models, the influence of transition and of the grids used.

The final chapter aims to summarise the experience gained with respect to the performance of turbulence models over the variety of applications considered in the FLOMANIA project. To achieve a level of summary whilst adhering to the case-specific nature of the problem, the discussion is based on groups of flows. This may also assist the potential users seeking to apply the knowledge gained in FLOMANIA to their own applications. As a significant amount of effort has been dedicated to hybrid methods, in particular DES, background models used by these methods are addressed in addition to pure RANS investigations. Furthermore, any derivable best practice guidelines are presented, and the inherent limitations of the applied evaluation procedure are outlined.

External aerodynamics has not been widely established in the rail-vehicle industry until recent years. Nonetheless, the subject is of fundamental importance in some respects, e.g. aerodynamic loads due to the head-wave or the slip-stream of a train, running resistance and cross-wind stability. The latter is the dominating safety issue when attention is drawn to high cruising speeds. The objective of this study is to scrutinise the predictive prospects of unsteady, scale resolving (DES) for cross-wind train aerodynamics. Attention is restricted to a mirrored pair of generic end cars of the German ICE2 high-speed train. The example included refers to a yaw angle of and a Reynolds number - based on the length of the first car - of ≈ 10. Computational results are reported for DES, supported by companion steady and unsteady RANS simulations and windtunnel measurements. More comprehensive consequences on the stability of the vehicle are briefly addressed by means of a quasi-static mechanical analysis. The aerodynamic loads obtained from the DES approach are in fair agreement with experimental data and outperform RANS predictions slightly. Results indicate that — in terms of the maximum allowable cross-wind speed — the predictive failure returned by DES corresponds roughly to only 1% of the actual value. Moreover, DES provides some insight into potential risks for an excitation of natural frequencies due to cross winds, which might detract the riding comfort.