Search Results

Aerodynamic simulations using CFD is now a standard tool in the automotive industry, and is becoming more and more integrated in the aerodynamic design process of new vehicles. This process is distinguished by parallel development with wind tunnel experiments and CFD simulation results, which demands comparable results to be generated by the two development tools. As wind tunnel effects are not simulated in most industrial applications of CFD, the comparison with experimental results normally show differences partly due to wind tunnel effects and ground simulation effects. Therefore a deeper understanding of wind tunnel effects and methods to fully reproduce experimental values with CFD is necessary. In this paper, an extensive validation study with a detailed scale notchback model inside an open jet wind tunnel is presented. This study includes experimental data from the real wind tunnel as well as CFD simulation results with and without wind tunnel effects.

Wind tunnel interference effects are still considered to be negligible - or at least undesired - in automotive aerodynamics. Consequently, up to now there is no standard correction method which is used in everyday wind tunnel testing although a lot of research has been done in recent years. In most full-vehicle CFD simulations, wind tunnel interference effects are not simulated. The flow about the car is computed under idealized conditions. The wind tunnel is designed to simulate these conditions but fails to do so to some degree due to its limited size. Therefore a comparison of blockage-free CFD results and wind tunnel measurements is deficient. Hence CFD simulations including wind tunnel interference effects should be favored in the future for validation purposes. Furthermore, CFD offers new possibilities to investigate individual contributions to wind tunnel interference effects and therefore could help to increase the understanding of the flow in the wind tunnel.

This paper describes a method to simulate underhood temperature distributions in passenger cars. A simplified engine compartment simulation test rig is used to perform measurements with well known boundary conditions to validate the simulation strategy. The measurement setup corresponds to idle without working fan. The aim of this setup is to validate cases with strong natural convection, e.g. thermal soaking. A coupled steady-state CFD run and thermal analysis is undertaken to simulate the temperature distribution in the test rig. Convective heat transfer coefficients and air temperatures are calculated in StarCD™. The radiative and conductive heat transfer is considered in a RadTherm™ analysis. The strong coupling of flow field and wall temperature in buoyancy driven flows requires an iterative process. Calculated temperatures are compared to measured results in order to validate the simulation method as far as possible. Some of the results are reported in this paper.

This paper presents velocity and pressure measurements obtained around an isolated wheel in a rotating and stationary configuration. The flow field was investigated using LDA and a total pressure probe in the model scale wind tunnel at IVK/FKFS. Drag and lift were determined for both configurations as well as for the wheel support only. These results were used as a reference for comparing numerical results obtained from two different CFD codes used in the automotive industry, namely STAR-CD™ and PowerFLOW™. The comparison gives a good overall agreement between the experimental and the simulated data. Both CFD codes show good correlation of the integral forces. The influence of the wheel rotation on drag and lift coefficients is predicted well. All mean flow structures which can be found in the planes measured with LDA can be recognized in the numerical results of both codes. Only small local differences remain, which can be attributed to the different CFD codes.

In automotive wind tunnels with modern road simulation installations boundary layer pre-suction is a widely-used technique for boundary layer control. The consequence of boundary layer pre-suction is an additional pressure gradient in front of the model. In order to investigate the effects of the additional pressure gradient on drag, experiments were conducted with two different models (scale 1:5) in the IVK Model Wind Tunnel. In these experiments the suction velocity of the boundary layer pre-suction served as a parameter to change the static pressure gradient along the test section and was for this purpose adjusted higher and lower than the standard suction velocity. It is shown that the total drag increment due to boundary layer pre-suction consists of at least two parts: the ground simulation increment and the static pressure gradient increment. The ground simulation increment is due to a decrease in the boundary layer thickness and the resulting modified flow beneath the model.

This paper introduces a driving-torque measurement method for the determination of vehicle road load and its components. To increase the accuracy, the torque measurements are combined with rolling resistance measurements performed with a specially developed trailer. This method is a strictly experimental approach and does not use any mathematical models. The experimental techniques are described as well as the proceedings to compare test stand and road measurements. The results that are shown prove that this method is suitable for the investigation of single road load components. Furthermore, the comparison of different rolling resistance measurement devices shows the potential of the measurement trailer and the necessity to perform rolling resistance measurements on real road surfaces and not solely on test stands.

During the acquisition phase brake system supplier have to make predictions on a system's thermal behavior based on very few reliable parameters. Increasing system knowledge requires the usage of different calculation models along with the progress of the project. Adaptive modeling is used in order to integrate test results from first prototypes or benchmark vehicles. Since changes in the brake force distribution have a great impact on the simulation results fading conditions of the linings have to be integrated as well. The principle of co-simulation is used in order to use the actual brake force distribution of the system.

The development process of electronic control units (ECU) is increasingly supported by different tools. The target-specific code-generation for single micro-controllers becomes a standard technology. Thus a continuous tool support during the whole development cycle is possible. This extends from the specification of the functionality to the implementation of the software on the controller. The next generation of tool support is not only focused on single micro-controllers it also supports the design of systems consisting of different controllers connected via various communication entities. Thus the goal of the tool support is the automatic code-generation for such distributed embedded real-time systems including support of different communication buses (e.g., Controller Area Network CAN, Time Triggered Protocol TTP) and different processor targets.

The manufacturers of passenger cars increasingly assign development and production of complete subsystems to the supplying industry. A brake system supplier has to give predictions about system quality and performance long time before the first prototypical system is built or even before the supplier gets the order for system development. Nowadays, the usage of computer-aided system design and simulation is essential for that task. This article presents a tool designed to support the development process. A special focus will be on how to define quality. A formal definition of quality is provided, illustrated and motivated by two examples.

In automobile development, wind tunnel measurements are used to optimize fuel consumption and the vehicle's road behavior. The classic measuring technique is based on a stationary vehicle set up in the wind tunnel with stationary wheels. Relative movement between vehicle and road surface is therefore ignored. In more recent studies, measurements have been taken with improved ground simulation. For example, a belt is used instead of the stationary wind tunnel floor and the car wheels rotate. Ground simulation using a belt and rotating wheels generally leads to a reduction in flow angularity at the front wheels, in the same way as blocking the cooling air flow, whereby, as a matter of fact, the aerodynamic drag is reduced. Analogous air flow angle correlations can be established for the effect of underfloor panels.

This paper deals with the correction for interference effects on the flow about bluff bodies in open-jet wind tunnels. Altogether, five different interference effects in open-jet tunnels are identified and described by physical models. Besides the classical jet-expansion correction which has been treated repeatedly by various authors throughout the last 60 years, the new correction method also includes the effect of jet deflection due to the proximity of a vehicle to the nozzle. Furthermore, far-field interference effects originating from the nozzle and the collector are described. For this purpose nozzle and collector effects are treated as solid- wall blockage phenomena, and with the aid of Biot-Savart principles the induced velocity at the model position is calculated. Finally, the static- pressure gradient in an empty test section generates a horizontal buoyancy force when a model is placed in a wind tunnel stream.

The present paper reveals the design concept as well as results of experimental investigations, which were conducted in the early design stage of the planned AUDI Aero-Acoustic Wind Tunnel. This low-noise open-jet facility, featuring a nozzle exit area of 11 m2 and a top speed of approximately 60 m/s, enables aerodynamic as well as acoustic testing of both, full-scale and model-scale ground vehicles. Ground simulation is provided by means of a moving-belt rig. The surrounding plenum is designed as a semi-anechoic chamber to simulate acoustic free-field conditions around the vehicle. Fan noise will be attenuated below the noise level of the open jet. The work reported herein, comprises 1/8-scale pilot-tunnel experiments of aerodynamic and acoustic configurations which were carried out at the University of Darmstadt.

The range of applicability and the physical restrictions for the use of ground-simulation techniques in automotive wind tunnels are elucidated. The techniques considered are the moving-belt technique, as well as boundary layer control techniques like tangential blowing and distributed normal suction for use in wind tunnels with stationary ground boards. Attention has to be paid to the question of whether the flow to be simulated is of boundary layer or Couette type. In the case of boundary layer flow, interaction of the ground-floor boundary layer with the inviscid flow in the gap between a vehicle and the road can be fully simulated by introducing a negative transpiration velocity along the stationary ground plane. In practise however, angularity effects on the external flow result from mismatched control parameters. Very small relative ground clearances give rise to the formation of a Couette flow between the road and the vehicle.

With the help of theoretical considerations it is shown that the flow between a car and the ground is of boundary layer type, as long as there is no recirculation. Thus, boundary layer theory can be applied to evaluate the order of magnitude of typical effects like displacement and momentum-loss thicknesses of ground-plane boundary layer. If the boundary layer in a wind tunnel with stationary ground is to be controlled, either by distributed suction or by tangential blowing, to simulate on-road conditions, boundary layer theory can provide the orders of magnitude of modifications that have to be applied. Experiments with a ground-effect quarter-scale car with ground simulation by distributed suction and moving belt showed coincidence with theoretical predictions concerning the required suction rate, if integral coefficients (eg CD, CL) of both flow cases are matched.