Search Results

The purpose of this study is to investigate the turbulent mixing in a diesel spray by large eddy simulation (LES). As the first step for the numerical simulation of diesel spray by LES, the LES of transient circular gas jets and particle laden jets were conducted. The simulation of transient circular jets in cylindrical coordinates has numerical instability near the central axis. To reduce the instability of calculation, azimuthal velocity around the central axis is calculated by the linear interpolation and filter width around the axis is modified to the radial or axial grid scale level. A transient circular gas jet was calculated by the modified code and the computational results were compared with experimental results with a Reynolds number of about 13000. The computational results of mean velocity and turbulent intensity agreed with experimental results for z/D>10. Predicted tip penetration of the jet also agreed to experimental data.

Both the external and internal flows of cars are simulated simultaneously. A third-order upwind-difference scheme is used in these simulations. Computational grids are generated by a multi-block transformation and a trans-finite method. Engine compartments are modeled by grid systems but the heat exchanger is simulated as a pressure loss proportional to the dynamic pressure of the flow passing through it. First, the flow for a very simple test model with no wheels and nothing in its engine compartment is simulated and compared with experimental results in order to validate a simulation method for the engine compartment. Pressure distributions on the inner surfaces agree very well with measured values, while pressure distributions on the external surfaces show reasonable agreement except for the roof end and the leading edge of the floor. The predicted drag coefficient is 7% larger than the experimental value. This method is next applied to a prototype car.

Incompressible viscous unsteady airflow around three types of automobiles are simulated. A thirdorder upwind-difference scheme is used in these simulations. In all the analyses, computational grids are generated by a multi-block transformation and a transfinite method in each block. The first two types of automobile have almost the same shape in the front half of the body, and the accuracy of predicting the difference in drag coefficient is investigated. In this case, the bodies are simplified. They have flat under-floors and no wheels. These calculated results are compared with experiments using 1/5 scale models. The difference in drag coefficient between the two types agrees well with the experiments, and also the values themselves agree well. In the last case, a car with wheels and an under-floor resembling a production model is studied. Simulated results are compared with experiments using a real production car with closed front opening and without mirrors.

The acceleration performance of a car equipped with a turbocharged, intercooled engine is affected by the volume of cooling air that flows through the core of the intercooler. Additionally, the volume of cooling air entering the intercooler is influenced by the configuration of the air intake provided in the exterior design. Therefore, in planning a new model it is very important to be able to predict acceleration performance, at an early stage of the vehicle development process, in relation to vehicle styling and engine specifications. The procedures employed so far to predict the volume of air flowing through the intercooler have included two-dimensional finite-difference methods and a panel method. However, because of their simple nature, none of these approaches has provided sufficiently accurate results. This paper presents a new numerical analysis method that has been developed to overcome this problem.

The main function of automobile heat exchangers is to transfer residual heat from the engine to the open air so as to keep running the engine at its best condition. In order to efficiently transport heat from the heat exchanger to the air, extended surfaces (e.q., fins) are usually provided over the outer surface of the water tube in the heat exchanger. Moreover, many bent out plate louvers are manufactured on the fin safaces. In order to obtain higher rate of heat transmission in a more compact heat exchanger, many experimental and analytical studies have been carried out up to the present. However, there are many difficult problems to be resolved so far in finding the obtimum louver profile because of the complexity of air flow around small-scale louver assemblies. In the present paper, the characteristics of air flow as well as heat flow around louver assemblies are analysed by the numerical analysis code developed at the Central Engineering Laboratory of the Nissan Motor Co., Ltd.

This paper describes a preliminary study of the role of computational fluid dynamics in analyzing the aerodynamic characteristics of a heavy-duty truck body. Among truck related aerodynamic problems, we selected the soil problem on the vehicle side surfaces as the analysis subject. Because of computer capacity limitations, a half-truck-cab model with a tire and mud guard are used and created by using the multi-block transformation technique. The flow around the cab is simulated by directly integrating the Navier-Stokes equations, approximated by finite-difference equations. Calculated results on the flaw structures around the vehicle body surface where it becomes dirty under wet weather conditions provide some useful information in the search for understanding of soil problems.

The computation of the Navier-Stokes equations through the three torque converter components (i.e., the pump, the turbine and the stator) is shown. A third-order-upwind scheme is used in the computation. The flow in each component is first calculated individually. Then, the calculation results for each outlet condition are used as the inlet condition of the next component, and the flow in each component is calculated again. This iterative procedure is terminated when the loss of flow pressure in the three components reaches a steady state. The torque converter performance predicted with this method agrees well with experimental data.