• Energy Research

The operating conditions of modern high-speed trains require efficient aerodynamic shape optimization methods. Many of the interesting aerodynamic features can be determined by means of RANS simulations. The objective of the study, presented in this paper, is to determine how far a shape optimization procedure using adjoint-based methods with continuous approach in conjunction with filtered gradients and CAD free mesh morphing based on radial basis function interpolation is capable of meeting the requirements of modern train development. First applications of the developed process chain consider shape optimization with regard to the generated pressure wave.

Fast passing trains induce aerodynamic forces which can be dangerous for trackside objects and for passengers on station platforms. To develop theoretical and numerical approaches for optimizing the shape of train heads with the objective of minimizing the amplitude of the induced pressure wave and the resulting forces on trackside objects is the aim of the present work. In the first part, we consider shape optimization of a potential train head modeled by superposing monopole sources to a uniform flow. To demonstrate the reliability of the potential flow model, we determine the aerodynamic loads on a sphere induced by the head pressure pulse of a passing train and compare them with results measured in a moving model rig, the so-called “Tunnel Simulation Facility” (TSG). Since good agreement is obtained , we conclude that the potential flow model is suitable to predict the pressure induced forces on track side objects. Then, the validated simplified potential model is used to select the shape of a train head which minimizes the sucking force on a cylindrical object passing the train at a specified lateral distance. In the second part, we apply the continous adjoint optimization approach, first on the potential train body and then on a more detailed shape of a train performing Reynolds-averaged Navier-Stokes (RANS) simulations. The focus of this ongoing research is to develop a process chain using the continous adjoint-based shape optimization approach in conjunction with a potential flow on the one hand and with RANS simulations, filtered gradients and CAD-free mesh morphing based on radial basis function interpolation on the other hand. In the RANS study, shape optimization is performed with the objective of minimizing the amplitude of the generated pressure wave. This part of the paper includes new results of the work started in Jakubek et al..

Direct Numerical Simulations (DNS) of turbulent channel flow are performed in a computaional domain and for the friction Reynolds number Reτ = 180 realized by Kim with the aim to reduce the skin friction by controlling the near- wall transport processes based on surface modifications. The latter are determined by solving the adjoint Navier-Stokes equations which depend on time-averaged DNS flow fields. It is well known that the wall-normal momentum Transport is organized in sweep and ejection events in the boundary layer which controls skin friction. Bewley et al. used adjoint optimization for suction and blowing in turbulent channel flow. They mitigated the effect of the wall-normal transport by opposing near-wall vertical motions of the fluid with an equal and opposite control velocity at the wall. The aim of this work is to achieve this with wall surface modifications. The idea is further to damp the spanwise fluctuations of streaky structures and thus, to reduce the associated skin friction.

Very-large-scale motions (VLSM) in turbulent pipe flow become increasingly energetic with higher Reynolds numbers and contribute significantly to the turbulent kinetic energy budget. Although several investigations regarding energy mechanisms in wall-bounded turbulence were carried out in the past, the origin of the energy at the largest wavelengths has not been studied rigorously, yet. Consequently, we analysed the budget equation of the filtered streamwise velocity field with very large filter lengths to identify energy sources and sinks related to VLSM. While there was no inverse energy cascade found towards VLSM on average, conditional sampling revealed an inverse energy cascade related to low-speed structures, whereas high-speed structures were subject to a forward cascade. This behaviour is at odds with the mechanism relevant for the near-wall small-scale motions, where the inverse cascade correlates with high-speed fluid.