Researchers from Exa Corp. and Navistar compared a model in wind tunnel (WT) testing to computational fluid dynamic (CFD) WT over the entire yaw curve. Their study also examined impacts of the tunnel environment compared with real-world wind conditions.

WT tests were performed on a 1/8^th scale long sleeper and standard 53-ft dry-box trailer with trailer skirts in the rolling wind tunnel at the Auto Research Center (ARC). Figure 1 shows a representative WT test tractor model. The same tractor geometry was used for both the wind tunnel and CFD simulation.

CFD simulation was performed using a commercial tool that employs the Boltzmann equation to solve for velocity and pressure field. The Lattice Boltzmann formulation, which is equivalent to solving the time-dependent compressible Navier-Stokes equation, can be used to impose fluid boundary conditions on solid surfaces. The software employs the VLES (very large eddy simulation) turbulence model to solve the resolvable-scale flow scale, and it models the subgrid scale using RNG (re-normalization group) k-epsilon.

The far field and grids were set up per SAE J2966. Three CFD simulation environments were used: CFD WT – 1/8^th scale model; open-road (OR) – full scale, and the simulation domain was an idealized domain with negligible blockage representing the open road; and real-world (RW) – full scale and 7% upstream turbulence to replicate the standard wind gust, and added road/tire surface roughness.

The WT testing was performed with a yaw sweep from -9° to +9° yaw. CFD WT simulations were conducted with the identical CAD model from the scale model used in the WT test and with the same test setup.

Figure 2 illustrates the accumulated delta drag coefficient between 6° and 0° yaw condition for the different environments. Alternatively, the figure reveals the influence of yaw.

Three different dynamic correction methods were investigated: nozzle method, plenum method, and modified nozzle method. Two of them, the nozzle and modified nozzle method, agreed well with results of the WT test. The nozzle method predicted within -0.8% C_DWA (wind averaged total drag coefficient) of the WT test. The study also illustrated the differences in three aerodynamic environments—1/8^th scale wind tunnel, open road, and open road with realistic wind condition.

Asymmetric drag polar existed between the positive and negative yaw angles due to the different clearances to the side of the test section itself and/or to some extent of the test vehicle upper body and underhood components. A significant shift in the normalized C_D (total linear drag coefficient) between the tunnel simulation and the OR simulation was seen. The tunnel boundary conditions affected the shape of the yaw curve and reduced the model’s sensitivity to yaw effects. Drag effects of yaw were less pronounced in the WT.

An overall higher C_DWA occurred in the WT environment due to a slightly wider shear layer, which increased the pressure drag around the vehicle, and also higher skin frictional drag due to lower Reynolds number. Total normalized drag coefficient predicted at WT, CFD WT (per nozzle method), and OR were lower than the RW drag coefficient because real-world flow presence of upstream turbulence and surface friction on the tires as well as on the floor increased the drag coefficient. By considering several effects present during WT testing, this study helped to close the gap between simulation drag prediction and wind tunnel testing results, and led to a greater understanding of simulation methodologies that can be used to reproduce the aerodynamic losses in the real-world environment.

This article is based on SAE technical paper 2016-01-8140 authored by Devaraj Dasarathan and Matthew Ellis of Exa Corp. and Ashraf Farag of Navistar. The paper will be presented at the SAE 2016 Commercial Vehicle Engineering Congress.