A consistent mesh refinement study, relating to the prediction of aerodynamic forces about an experimentally 13 validated reference train geometry, is presented in this paper. The flow about a high-speed train has a multi-scale 14 character which poses challenges for the design of computationally effective meshes. The purpose of this study is to 15 assist in the development of guidelines for effective drag prediction of high-speed trains using numerical simulation. 16 These guidelines should assist CFD practitioners by identifying the regions of the mesh that are critical for the correct 17 estimation of drag as well as providing information on appropriate mesh characteristics, such as volume and surface 18 element length scales. Numerical assessments are validated against an experimental drag measurement program and 19 the extent to which RANS is sufficiently predictive for industrial design is discussed. The results obtained in the work 20 suggest that the mesh about the train nose is essential for the proper assessment of the aerodynamic drag acting on the 21 vehicle 22 1 Introduction 24 Stringent safety requirements over a wide range of operational conditions are applied to modern high speed trains. 25 An understanding of the aerodynamic forces acting on a vehicle is mandatory, especially under crosswind con-26 ditions, in order to construct useful operational safety constraints. The measurement of force coefficients for 27 full-scale vehicles is optimal but expensive, and normal practices are geared towards the use of small-scale mod-28 els that can be tested inexpensively in wind tunnel experiments or by using full-scale-in-service vehicles (Baker, 29 2010). However experimental methods are limited in scope with respect to the study of such questions as the 30 optimization of vehicle shape over a range of design parameters. In comparison, computational methods have the 31 potential to provide detailed flow information at a cost that is comparatively inexpensive over a much wider range 32 of operational conditions. For example these methods can be used to determine optimal shape forms in terms of 33 stability and drag constraints. 35The use of computational methods to assess the aerodynamic loading on trains has been recognized by the transport 36 industry. For example the German standard EN 14067-6 (DB Netz AG, 2010) permits evaluation of aerodynamic 37 forces by means of computational fluid dynamics (CFD) simulations for full-scale or reduced model geometries. 38The guidelines for CFD in EN 14067-6 using RANS (Reynolds-averaged Navier-Stokes) equations are stringent 39 and give a specific error criterion that CFD calculations must satisfy. In particular the standard requires that com-40 puted integral forces cannot be accepted for certification work if variations against accepted reference values (i.e. 41 experiment) differ by more than three percent. A major challenge in satisfying EN 14067-6 requirements is due 42 to the multi-scale nature of the flow problem which is characterized by a large ra...

No abstract

Extreme statistics, Gaussian statistics, and superdiffusion in global magnitude fluctuations in turbulence Phys. Fluids 24, 105103 (2012) Diffusion in grid turbulence of isotropic macro-particles using a Lagrangian stochastic method: Theory and validation Phys. Fluids 24, 103303 (2012) Symmetry analysis and self-similar forms of fluid flow and heat-mass transfer in turbulent boundary layer flow of a nanofluid Phys. Fluids 24, 092003 (2012) Turbulent concentration diffusion in multiphase flow Phys. Fluids 24, 093301 (2012) Additional information on Phys. FluidsThe coefficient C 0 , which determines the effective turbulent diffusion in velocity space, is fundamental in Lagrangian modeling. Others, for example, Yeung and Pope ͓J. Fluid Mech. 207, 531 ͑1989͔͒ have investigated this coefficient numerically with direct numerical simulation ͑DNS͒ by analyzing the dispersion of discrete particles. In our paper we estimate the coefficient C 0 by using DNS to study the initial evolution of continuous passive scalar fields. Using an equivalent probability density function ͑pdf͒ and conditional moment analyses we examine the initial transient behavior for passive scalar mixing, with both linear and Gaussian-type initial distributions in the velocity space. Our estimates of C 0 are found to be consistent with the data found in the literature.

Safety assessments of cross-wind influence on high-speed train operation require a detailed investigation of the aerodynamic forces acting on a vehicle. European norm 14067-6 permits the derivation of required integral force and moment coefficients by experiments as well as by numerical simulation. Utilizing the DLR's Next Generation Train 2 model geometry, we have performed a case study comparing simulations with varying turbulence modeling assumptions. Because of its relevance for actual design, a focus lies on steady RANS computations, but more expensive unsteady RANS (URANS) and delayed detached eddy simulations (DDES) have also been carried out for comparison. Validation data for the exact same model configuration and moderate Reynolds numbers 250,000 and 450,000 is provided by side wind tunnel experiments. Particular emphasis is laid on simulating a yaw angle of 30• , for which a major vortex system on the leeward side of the train leads to sizeable uncertainties in predicted integral coefficients. At small to intermediate wind angles the flow remains attached and absolute errors in integral quantities decline with decreasing yaw angles. However, a consistent relative difference to the experimental results greater than 10% raises doubts about the general reliability of CFD methods, that are not capable of capturing laminar-turbulent transition, which is observed for scaled models in industry type wind tunnel experiments.1

No abstract