Bhaskar Bhatnagar and Gerald W. Recktenwald
Presentated at the 2004 ASME Heat Transfer/Fluids Engineering Summer Conference Charlotte, North Carolina, July 11-15, 2004
Presentation slides (PDF, 9.7 MByte)
We have used a commercial CFD program and boundary layer analysis to design the inlet of an open circuit wind tunnel for aerodynamic testing of full scale, class eight trucks. The wind tunnel has many novel features, not the least of which is that it is being constructed at a fraction of the cost of wind tunnels of comparable size.
The goals of the study were to obtain the combination of wall shape, contraction length, and settling screens that provided a uniform velocity profile upstream of the vehicle under test. Most of the analysis is performed with a commercial CFD code. A primary concern is to avoid separation of the boundary layer in the inlet. To provide an independent check on the CFD model, the boundary layer analysis procedure of Mehta and Bell is used to verify that the boundary layer does not separate.
This paper presents the results of analyzing a quarter section model of the S-shaped inlet section. The CFD model has a highly graded, body-fitted mesh of 1.6 million cells. The effect of settling screens is included. Simulation results were obtained with a low Reynolds number turbulence model and a high order convection modelling scheme (MARS).
The boundary layer analysis uses free stream data from running the CFD model on a coarse mesh with slip boundary conditions on the wall. The velocity and pressure of cells outside the boundary layer are used to specify the variation of free stream velocity for a Thwaites analysis of the wall boundary layer. The primary output of the boundary layer model is the variation of friction coefficient along the wall. A positive friction coefficient at all points along the wall indicate that the boundary layer does not separate.
We present velocity profiles upstream of the vehicle under test for various design configurations of the inlet. The recommended design has a velocity variation of less than one percent outside of the boundary layer. The contraction introduces a weak secondary flow in the corners of the wind tunnel, but the cross stream components do not lead to significant distortion of the axial profile. The Thwaites analysis confirms that the boundary layer does not separate in the recommended design.