Computational modeling of the aerodynamics of the car-truck overtaking process by means of eddy-resolving turbulence models
From the fluid mechanics point of view, overtaking a truck by a car is an extremely complex process, characterized by a multiplicity of flow phenomena and cannot be successfully solved without a prior knowledge. Before a flow simulation of the overtaking maneuver is carried out, the problem of the flow around the vehicle must first be understood and, subsequently, a numerically relevant simulation model must be selected. One of the primary goals of the present work is therefore the checking of the predictive capabilities, credibility and reliability of the adopted eddy-resolving modeling strategy in connection to the corresponding numerical procedure in such a complex flow situation.
Preliminary, a grid study has been carried out by means of computing a somewhat simplified BMW 5 series car model (down-scaled to 1:2.5) regarding the effects of spatial and temporal resolution on the quality of results. As part of this "must study", the sensitivity of the simulation results to different grid arrangements and grid sizes can be examined relatively fast. The results obtained are mutually compared. A suitable grid is then selected that contains as few cells as possible, but ensuring at the same time an adequate results quality. The resulting grid setting was then adopted and a simulation accounting for a detailed BMW 5-series car geometry (among other things regarding the underbody) has been performed.
The reference experiment, carried out in the BMW AG wind tunnel, with which the simulation results have been compared, considered at first an isolated single BMW 5-series vehicle model, but then also a "quasi-stationary" overtaking process, in which eight fixed positions of the car relative to the truck were mapped. A flow was simulated with and without accounting for the boundary layer suction. Finally, a road measurement of the overtaking process with moving real vehicles was carried out, which was also numerically reproduced on a model scale.
All these cases were numerically simulated according to the experimental operating conditions. The boundary layer suction is represented by a suitable boundary condition. A dynamic overtaking maneuver on the road is reproduced by utilizing the code feature dealing with a deforming and moving grid, in accordance with the method based on the so-called "space conservation law". The car-truck interactions in the static and dynamic considerations possess different effects having a corresponding influence on the drag coefficients and moments, which was also discussed in more detail. Numerically, two considerations were pursued with regard to the flow domain to be taken into account: on the one hand, the calculation domain designed in the form of a regular hexahedron with appropriately defined inlet, outlet and symmetry boundary conditions was adopted and, on the other hand, a flow domain was designed coinciding exactly with the shape and dimensions of the wind tunnel.
As indicated above, an adequate turbulence model in the context of a suitable simulation strategy represents an important part of this work. The work on the development of the so-called hybrid RANS/LES (Reynolds-Averaged Navier-Stokes / Large-Eddy simulation) models experienced great intensification in recent years. The highly complex, interacting fluid mechanics phenomena that characterize the transient vehicle aerodynamics under fully turbulent conditions put high demands on turbulence modeling. For this reason, an advanced RANS four-equation model, based on the concept of eddy viscosity, was adopted, which describes the dynamics of the unresolved residual turbulence in the context of the so-called eddy-resolving PANS method (Partially-Averaged Navier Stokes). The model represents one of the modeling highlights of the FIRE code by the AVL List GmbH (Graz, Austria), which is based on the finite-volume numerical method, and with which all calculations were carried out. For comparison purposes, both stationary and unsteady RANS calculations were carried out.
With respect to the results evaluation, the main focus was on the distribution of the surface pressure and the resulting aerodynamic forces, which are interpreted in the form of their coefficients and acting moments. These are the properties that most directly determine the aerodynamic behavior of the vehicle configuration including vehicle stability. The capability of the PANS method in capturing the fluctuating turbulence to a corresponding level, is reflected in the quality of the results obtained, when compared to the outcome of conventional RANS calculations. The PANS results show a high degree of agreement with the experimentally determined data. The interpretation of the results is further enriched by the presentation and associated discussion of the instantaneous and time-averaged flow fields (iso-surfaces; vortex visualization through the Q criterion) in their entirety. The presentation of the results was accompanied by corresponding quality assessments with respect to the numerical method accuracy (such as the order of the spatial and temporal discretization) as well as to the correct capturing of the physics of the turbulence (in regard, among other things, to the relationship between the characteristic grid spacing and the Kolmogorov's length scale). | English |