Skip to content
By continuing to browse NWTF site you are agreeing to our use of cookies. Review cookie page for details.

Wake of an Isolated Rotating Wheel

Date/revision:

31 March 2026

Authors:

Dipesh Patel 1 LinkedIn Logo Andrew Garmory 1 ✉ Martin A. Passmore 1 LinkedIn Logo

Affiliation: 1Loughborough University, Epinal Way, Loughborough, LE11 3TU, United Kingdom
Email-address: a.garmory@lboro.ac.uk

Introduction

The carbon dioxide (CO2) emissions from a car are inherently linked to the aerodynamic drag acting on the vehicle. This force which is acting to reduce the speed of the vehicle is constantly being overcome by the engine, resulting in increased fuel consumption. At high speed, the drag is more significant as the force is proportional to the square of the speed of the vehicle (U2). Whilst designers attempt to reduce the drag by shaping the body to be more aerodynamically efficient, one area that contributes significantly that cannot be easily influenced is around the wheels.

In automotive bodies, there are two types of wheel configuration: closed-wheel which is common in everyday vehicles and open-wheel which is more prominent in motorsports. Both configurations have a significant impact on the drag characteristics of a vehicle, but open-wheel cars are more adversely affected than closed-wheel. In the context of motorsport, tailpipe emissions are less important than overall performance. Drag is targeted in the motorsports context to increase vehicle speed and reduce weight through reduction in fuel consumption (or energy consumption in the case of electric racing series) which would reduce the size of the fuel tank/battery.

Research into isolated stationary and rotating wheels spans the last five decades and is well-documented. Agreement regarding the expected surface pressure distribution was reached early in the field, as highlighted by the works of Fackrell and Harvey[1] and Cogotti[2]. However, research into the flow structures contained in the wake generated by an isolated rotating wheel reveals weaker agreement. This is of particular importance to CFD engineers who may be attempting to validate transient automotive models that include open wheels, as the flow structures play a vital role in the overall vehicle performance. This concern is not limited to open wheel vehicles as the formation of wheel wake is equally applicable to closed wheel applications, where drag may be of particular interest.

Far-field wake measurements taken of an isolated stationary and isolated rotating wheel highlight that the stationary wheel produces a shorter and wider wake[3]. The enhanced downwash of the stationary wheel relative to the rotating one was partially attributed to the delayed separation point. Unsteady CFD using a RANS model revealed separation near the top of the rotating wheel, with instantaneous results revealing periodic vortex shedding in the generated wake[4]. Nearer to the ground, a large region of perturbed flow was observed and attributed to a "jetting"-type behaviour[1], theorised as the result of the wheel and ground boundary layers forming a convergent nozzle which resulted in acceleration of the flow. The jet was subsequently deflected by the mainstream and proceeds past the wheels. Consideration of the instantaneous flow field illustrates the wake as a composition of seemingly random eddies; absent of a coherent structure on the macro-scale[5].

The study by Patel et al.[6] from which this dataset originated sought to identify the primary structures present in the time-averaged wake and the alteration of these structures in the instantaneous flow field through experimental and numerical means. The details of the numerical simulation are given in the original work[6], with the experimental PIV results provided below. The motivation of the study was to provide the following:

  1. To provide a data set for the flow around a wheel with realistic shoulder radii for which well-defined and publishable geometry is available
  2. To provide unsteady and mean flow field data from the geometry

Model Geometry

A 50% scale variant, generic non-deformable wheel geometry representative of a treadless (slick) tyre from open wheel motorsport was used in the experiments as shown in Figure 1. The wheel is 330 mm in diameter and 180 mm in tread width, with a double-radius wheel shoulder. The wheel was connected to a rotating drum such that it could rotate about its central axis to match the freestream conditions of the tunnel. The origin in the model is located at the centre of the wheel, X is in the freestream direction, Y is in the horizontal direction following a right-handed coordinate system, and Z is in the vertical direction (upwards positive).

The wheel geometry used in the experiments of Patel et al.

Figure 1: The wheel geometry used in the experiments of Patel et al. [6] in (a) front view and (b) side view.

A pair of supports either side of the wheel are used to mount the model to the ground of the tunnel. To enable rotation whilst maintaining contact, a separate fixed ground plane with a roller set into the floor of the tunnel was employed and used to drive the wheel. 3D-printed inserts were installed around the contact patch to limit the gap to the cavity below the contact patch. Patel et al.[6] showed that removing these supports in CFD simulations resulted in minor changes to the flow field when compared to the full geometry without the experimental mounting supports.

Measurement Locations and Techniques

Particle Image Velocimetry (PIV) was used as the primary experimental method for attaining instantaneous and time-averaged velocity measurements in the wake. The technique is non-invasive and uses a combination of light and flow tracing through the use of a fluid with the appropriate Stokes Number to track the flow without altering it. A schematic of the setup used is shown in Figure 2.

An example schematic of the PIV setup used to measure planar velocity.

Figure 2: An example schematic of the PIV setup used to measure planar velocity.

Two measurement planes were taken downstream of the wheel at X = 330 mm and 495 mm (corresponding to x/dw = 1 and 1.5, respectively.) where X = 0 corresponds to the centre of the axis of rotation. In these planes, the velocities refer to crosswise flows (VY), vertical flows (VZ), and velocity through the plane (VX). Additional planes were taken across the wheel tread at Y = 0 (centreline) and Y = 50 mm where streamwise (VX) and vertical (VZ) velocities were measured. A schematic of the planar slices relative to the wheel are shown in Figure 3.

Measurement planes shown relative to the wheel geometry.

Figure 3: Measurement planes shown relative to the wheel geometry.

In each plane, 2000 image pairs were taken at a sampling rate of 15 Hz, resulting in a total sampling time of 134 seconds[6]. Post-processing was based on a multi-pass scheme with reducing window size. Initially, a window of 128 × 128 pixels was used with a 50% overlap and square weighting that reduced to a final window of 24 × 24 pixels with circular weighting and an overlap of 75%. Consequently, the mean flow-field contains an uncertainty of ±2% based on a confidence interval of 99%[7].

Experimental Facility

The experimental work was conducted at the Large Wind Tunnel facility at Loughborough University; an open-loop tunnel with a working section of 1920 × 1320 × 3600 mm (W × H × L). The resulting blockage ratio from the model in the working section was 3.3%. The typical turbulence intensity in the working section was measured as 0.2%. The flow was seeded with particles upstream of the model to enable PIV. In this instance, Di-Ethyl-Hexyl-Sebacat was used as the flow tracer, producing atomised 1 μm particles that are both spherical and neutrally buoyant.

Flow Conditions

Free-stream velocity, V∞ = 30 m/s
Free-stream turbulence ≈ 0.2%
Wheel-diameter based Reynolds number, Red = 0.67 × 106
PIV sampling frequency = 15 Hz
Sampling time = 134 s

Available Geometry

To aid in the accuracy of simulations, the geometry used in the wind tunnel facility is provided in STEP, IGS and STL formats. This geometry represents the 50% scale model used in the physical experiments conducted by Patel et al.[6] and includes the mounting geometry. The false floor with roller inset has not been included in these models.

Datasets

The datasets below contain either four or five columns of data. The first two relate to the position on the plane, with nomenclature consistent with the global co-ordinate system in Figure 1, measured in metres (m). The following two/three columns present the velocity measured at the location on the plane in the directions denoted by the subscript, again relating to the co-ordinate system in Figure 1 measured in metres per second (m/s). The complete dataset can be downloaded here.

Plane Average Data Root Mean Square (RMS) Data
X = 330 mm X330mm_Mean.csv
X = 495 mm X495mm_Mean.csv
Y = 0 mm Y0mm_Mean.csv Y0mm_RMS.csv
Y = 50 mm Y50mm_Mean.csv Y50mm_RMS.csv

Sample Plots

An example of the dataset as analysed to produce an interpolated plane of data is shown in Figures 4 and 5. Here, a linear interpolation of the data between sample points was conducted to give a complete picture of the velocity on the plane, rather than a series of points. A python sample code for both planes is provided based on the AeroX class and data standard, which is available for download here. Similarly, a MATLAB script is available which performs a similar function

An example interpolated plot of streamwise velocity on the X = 330 mm plane.

Figure 4: An example interpolated plot of streamwise velocity on the X = 330 mm plane. The colour bar relates to the streamwise velocity.

An example interpolated plot of streamwise velocity on the Y = 0 mm plane.

Figure 5: An example interpolated plot of streamwise velocity on the Y = 0 mm plane. The colour bar relates to the streamwise velocity.

Open Access

This metadata is provided under the Creative Commons Attribution Non-Commercial 4.0 International License https://creativecommons.org/licenses/by-nc/4.0/). This license allows for unrestricted use, distribution, and reproduction in any medium, provided that proper credit is given to the original author(s) and the source. Also provide a link to the license and indicate if any changes were made. Furthermore, this license does not allow the use of this material for commercial purposes.

Acknowledgements

The NWTF acknowledge the support for the metadata work from the UKRI Research Grant, EP/Y036476/1. The original dataset was funded by Loughborough University and the EPSRC on grant EP/P 020232/1.

Citation

If the user wants to cite the data presented here, then please cite both the NWTF metadata and the corresponding original work (reference 6).

References

  1. Fackrell, J.E., Harvey, J.K., 1975. The aerodynamics of an isolated road wheel. In:
    Proceedings of the Second AIAA Symposium of Aerodynamics of Sports and
    Competition Automobiles. Web Link
  2. Cogotti, A., 1983. Aerodynamic characteristics of car wheels. Int. J. Veh. Des. 3,
    173–196. Special Publication SP3. Web Link
  3. Bearman, P.W., De Beer, D., Hamidy, E., Harvey, J.K., 1988. The Effect of a Moving Floor
    on Wind Tunnel Simulation of Road Vehicles. Society of Automotive Engineers.
    Paper 880245. Web Link
  4. Bruckner, J.H., 2011. The Aerodynamics of an Inverted Wing and a Rotating Wheel in
    Ground Effect. Ph.D Thesis. University of Southampton. Web Link
  5. Axerio, J., Laccarino, G., 2009. Asymmetries in the wake structure of a Formula 1 tire. In:
    Sixth International Symposium on Turbulence and Shear Flow Phenomena. DOI
  6. Patel, D., Garmory, A. and Passmore, M., 2021. On the wake of an isolated rotating wheel: An experimental and numerical investigation, Journal of Wind Engineering & Industrial Aerodynamics 227 (2022). Paper 105049. DOI
  7. Hollis, D., 2004. Particle Image Velocimetry in Gas Turbine Combustor Flow Fields,. PhD
    Thesis. Loughborough University. Web Link