The aerodynamic effects on a cornering Ahmed body

Highlights

• Cornering affected the aerodynamic characteristics of a simple vehicle shape.

• Large Eddy Simulations were used to model three different radius corners.

• C-pillar vortices and separated flow regions become asymmetric.

• Drag coefficient increased as corner radius decreased.

Abstract

As a vehicle travels through a corner, the flowfield observed from the vehicle׳s frame of reference becomes curved. This condition results in the relative flow angle and freestream velocity changing both across the width and along the length of the body. Wall-resolved Large Eddy Simulations were used to simulate a simple vehicle shape through three different radii corners. The variable flow angle and acceleration affected the pressure distribution along either side of the body and caused an increase in the size of the outboard C-pillar vortex, and an inboard decrease. Furthermore, an outboard extension of the separation bubble at the bluff trailing face resulted in a gentler downwash angle off the backlight surface, with the opposite occurring inboard. At a Reynolds number of 1.7×10⁶, a 19.2% increase in aerodynamic drag occurred for a five car-length radius corner when compared to the straight-line condition. In addition, a yawing moment acted against the rotation of the body through the corner, and a side force acted towards the centre of the corner. An exponential trend related the curvature of a vehicle׳s path to the increase in aerodynamic drag, with a linearity exhibited for the increase in yawing moment and side force.

Introduction

Technical development in the automotive industry continues to increase the performance capabilities of vehicles. As dynamic capabilities improve, tighter manoeuvres can be achieved at higher speeds, and will result in the aerodynamic effects having a greater contribution towards overall performance. Some vehicles use these higher speeds beneficially to create large amounts of aerodynamic downforce and further enhance cornering speeds (Toet, 2013, Katz, 2006). Despite this, designs will typically only be analysed in the straight-line condition and yaw (Toet, 2013). Recent studies in the field have considered a wider variety of conditions to assess real-world aerodynamic performance and this has resulted in investigations into conditions such as transient cross-winds and travelling through a corner (Tsubokura et al., 2012, Okada et al.,, Nara et al., 2014, D׳Hooge et al., 2014, Keogh et al., 2015a). Corners, in particular, have proved problematic due to the difficulty in recreating the conditions experimentally (Toet, 2013). There have been previous investigations into the cornering condition, but none have detailed the range of specific flow mechanisms responsible for the change in aerodynamic forces observed.

Fig. 1 shows the freestream flow characteristics for a vehicle in the steady-state cornering condition. The car itself is considered to have a constant angular velocity about a fixed external point – which is representative of a constant radius, steady-state corner. The velocity of the flow relative to the car, and the relative dynamic pressure of the flow, will increase with distance from this external point. In addition the relative angle (effectively yaw) of the flow will vary along the length of the car. In the specific condition shown it can be observed that the front and rear have opposite angles of yaw. How a vehicle travels through a corner will vary for different vehicles and driving styles. The condition investigated in the present work positions the body tangent to the curvature of its path at its centre, as can be observed in Fig. 1.

There have been previous experimental attempts to replicate the cornering condition in a wind tunnel with the use of bent models (Gordes, 2005), and curved test sections (Bird et al., 1951), but these methods are not capable of representing all aspects of true cornering flow. More recently there have been developments toward a unique new type of wind tunnel which is designed to be capable of accurately recreating the cornering condition (Keogh et al., 2016) however this project remains in its infancy, and leaves aerodynamicists to rely on numerical simulation (Keogh et al., 2015b).

Previous published investigations into the cornering condition have demonstrated the importance of evaluating the high speed cornering condition during the aerodynamic design phase for a passenger vehicle (Tsubokura et al., 2012, Okada et al.,). These studies considered two medium-sized sedan geometries and focussed on the yawing moment which damped the rotation of the vehicle through its curved path, as well as the side force. Large Eddy Simulation was used for the numerical component of this investigation, with a Smagorinsky Subgrid-Scale model. The most influential geometric variation between the two vehicle geometries was the amount of open space around the wheel in the wheel well. This simple geometric difference caused a 49% variation in Yawing moment – with the increased space around the wheels increasing pressure losses, which enhanced the moment. The investigation related all findings to an instantaneous yaw angle at the vehicle centre, and neglected to address the freestream curvature effects caused by this type of motion.

Another prior numerical study considered an isolated inverted wing travelling in the path of a corner (Keogh et al., 2015a). The work identified significant changes to the structure of the vortical wake forming downstream and attributed these to the yaw angle of the flow across the endplates. A side force, yawing moment, and rolling moment also occurred due to a local Reynolds number increase across the span and the angle of the flow. The potential cumulative effects this could have for an entire car were highlighted through an analysis of the downstream vortex paths in the wake.

In the public domain research in automotive aerodynamics has typically favoured simple bodies over detailed car geometries. This ensures that conclusions are not likely to be geometry specific and that discoveries will likely be common aerodynamic characteristics that are widely applicable (Le Good and Garry, 2004). The Ahmed body is one of the most commonly studied simple bodies, first analysed experimentally in 1984 (Ahmed et al., 1984). The geometry was designed to maintain attached flow over the front of the vehicle and permit detailed investigation into flow features occurring over the sloping rear face, which is a common feature on hatchbacks/fastbacks. As this is the geometry selected for the present study it is appropriate to introduce the main flow features.

The wake is highly unsteady due to the blunt trailing face, but the time-averaged structure has been investigated in detail and is closely related to the pressure component of the drag (Ahmed et al., 1984, Strachan et al., 2007, Lienhart and Becker, 2003, Krajnović and Davidson, 2004, Krajnović and Davidson, 2005a, Krajnović and Davidson, 2005b, Minguez et al., 2008, Wang et al., 2013, Guilmineau et al., 2011, Serre et al., 2013, Bayraktar et al., 2001, Conan et al., 2011, Thacker et al., 2012, Joseph et al., 2012). This structure consists of both longitudinal and spanwise structures indicated in Fig. 2.

The flow down the centre of the rear sloping angle, referred to as the backlight, detaches at the start of the surface and reattaches prior to the trailing edge, forming a separation bubble, as shown in Fig. 2. Due to the lower pressure over the backlight, the shear layer from the side of the body forms a longitudinal vortex at the C-pillar location. From a backlight angle of 12.5–30°, the size of this vortex increases and aids in promoting the reattachment of the separation bubble. At 35° the separation bubble fails to reattach and the vortex bursts (Ahmed et al., 1984, Strachan et al., 2007, Lienhart and Becker, 2003). The most commonly investigated backlight angle is 25° and this is due, in part, to the difficulty associated with accurately capturing the flowfield – particularly the separation and reattachment of the flow over the rear angle. The popularity of this angular configuration made it favourable for the present study and enabled any changes that were recognised due to cornering could be compared to the findings of a wide range of previous studies (Ahmed et al., 1984, Strachan et al., 2007, Lienhart and Becker, 2003, Krajnović and Davidson, 2004, Krajnović and Davidson, 2005a, Krajnović and Davidson, 2005b, Krajnović and Davidson, 2005, Minguez et al., 2008, Wang et al., 2013, Guilmineau et al., 2011, Serre et al., 2013, Bayraktar et al., 2001, Conan et al., 2011, Thacker et al., 2012, Joseph et al., 2012).

Subsequent experimental studies have adopted similar methodologies to that of the initial study by Ahmed et al. (1984). (Lienhart and Becker, 2003) conducted an experimental study of the original configuration with the benefit of a more modern facility and found close correlation with the initial study.

Through Wall-Resolved Large Eddy Simulation (LES) (Krajnović and Davidson, 2004, Krajnović and Davidson, 2005a, Krajnović and Davidson, 2005b) were the first to identify the existence of lower vortices, and present an accurate three-dimensional understanding of the flow structure. These lower vortices occur along either side of the lower surface due to an increase in pressure from boundary layer growth along the underside of the body. Experimental studies predominantly place support struts underneath the model which inhibit the development of these structures, and hence their identification. These were initially dismissed as being of little aerodynamic significance, but have received more attention in recent publications (Strachan et al., 2007, Wang et al., 2013).

Various LES and DES models have been used to simulate the unsteady flow structure around the Ahmed body. Particularly LES has proved to be superior in capturing both longitudinal and spanwise structures (Krajnović and Davidson, 2004, Krajnović and Davidson, 2005a, Krajnović and Davidson, 2005b, Minguez et al., 2008). The use of DES over the more computationally demanding LES has aided in achieving a solution in a much shorter time-frame, but it has failed to offer the same degree of accuracy – particularly in capturing the separation and reattachment of the flow down the backlight surface (Guilmineau et al., 2011, Serre et al., 2013).

The present work focusses specifically on the aerodynamic changes that occur during cornering and how these differ from the more-familiar straight-line condition. Through this analysis an investigation into the underlying physical causes of time-averaged cornering-specific flow phenomena was conducted.