Computational Fluid Dynamics Optimisation Based On Formula 1 Methods
Computational Fluid Dynamics (CFD) Aerodynamic simulation in a computer.
One of the major developments of the Hadron Aero wheel concept is a detailed aerodynamics optimisation study. Swiss Side’s goal is clear. To design an absolute top level performing aerodynamic wheel set in its category but as always, to bring it to the market at a price 40% less than the “Big Brands”.
The first steps in this process are to simulate, understand and optimise the wheel profile shape using CFD. Wind tunnel tests will follow this. Jean-Paul Ballard, co-founder of Swiss Side, explains the reasons, pros, cons and tricks behind the CFD optimisation process with his insider knowledge from over 13 years of working in Formula 1 Aerodynamics.
CFD OPTIMISATION APPROACH
Targets and Performance Function:
In our approach, we wanted to make our aerodynamic development as realistic as possible. Therefore, considering real world wind conditions which riders actually see on their bikes is very important for determining the targets for our aerodynamic development. Most importantly, what air speeds do the wheels see in reality and in particular, what are the extremes of cross wind (aka yaw) angles which occur? In addition, what percentage of riding time is spent in which range of cross wind angles? Identifying these key parameters was fundamental for us in defining our performance function which we would use to evaluate each design.
Our performance function considers aerodynamic drag over a range of yaw angles from 0 – 20 degrees. Each angle range is weighted differently in its contribution to overall performance. For example, the lower yaw angles occur much more frequently than the higher yaw angles and are consequently weighted more heavily. Also included in our performance function is the effect of the movement of the centre of pressure on the front wheel with increasing yaw angle. Since the rider’s body is, by far, the largest contributor to aerodynamic drag, shifts in centre of pressure of the front wheel cause the rider to correct the steering and consequently their body position. This is quite detrimental to the overall drag so on a gusty day this is another factor to consider for the overall aerodynamic efficiency. Overall, our “Performance Function” looks like this:
Our goal is simple: To produce a wheel which gives the best all round real world aerodynamic efficiency and consequently the best overall real world performance. It might not necessarily result in the lowest drag number at a single high yaw point e.g. 20 degrees, but will offer the lowest overall combined drag during a riding session.
This approach to performance weighting is exactly the technique we use in Formula 1 car aerodynamic development. The various key aerodynamic attitudes and speeds of the car are considered such as acceleration, braking, end of straight, low speed cornering, high speed cornering etc. Each condition is weighted in a performance function in order to optimise the car performance for the best lap time around a track.
Parameters For Evaluation:
In our CFD development, we investigate a broad range of geometric shape parameters of the rim and the boundary conditions for simulations. The geometric parameters include the rim height, width, curvature, position of wide point and more. As for the boundary conditions, these include the air speed and yaw angle.
Another important point we are considering which draws on our Formula 1 knowhow is tyre shape. On an F1 car, the tyre shapes and the associated flow fields are one of the most dominant features for aerodynamics. This is no different for a bike rim design. Considering the tyre and rim as a wing element, the tyre forms both the leading edge at the front of the wheel, as well as the trailing edge, half a rotation away at the rear of the wheel. The interaction of the tyre with the shape of the rim profile is very important, especially at higher yaw angles. Because we understand that our customers have their own preference in tyres, we consider a range of tyre shapes in our optimsation process with the intention to minimise the sensitivity.
We perform our optimisation primarily on a front wheel, without spokes, on a partial bike model. Using a complete bike model with rider would simply drive up the model size, complexity, time and cost drastically without any significant benefit. Similarly, the effect of thin elliptical spokes like the Sapim CX-Ray spokes we will use on the Hadron is minimal and does not influence the rim shape development and would simply increase the complexity and mesh size of the model. Most important for the simulation is the presence of the fork and frame down tube blockages. In our final performance evaluation step we consider a full bike with rider.
For our CFD work, we are using OpenFOAM, an excellent open source CFD software which is now used by numerous F1 teams, commercial and academic organisations. However as we do not have the computing power in house to achieve the level of speed and meshing accuracy we require for our simulations, we have outsourced the crunching to a leading F1 consultant. For our CFD tech savvy followers out there, we perform the majority of our calculations using a steady state solver with a RANS based turbulence model. Some final transient unsteady state simulations will be made in our final performance evaluation step.
THE A-B-C OF C-F-D
Some further background info:
What is it? CFD is now well and truly one of the fundamental development tools in aerodynamics. It allows us to simulate aerodynamic flows in a computer and analyze and understand the flow field in a far more detailed way than is possible in a wind tunnel. Not only forces like lift and drag can be calculated but the flow and its effects can be visualized in many ways, such as in the form of pressure plots, streamlines and surface flows. It is even possible to model complex fluid flows involving heat transfer, chemical reactions, solid dynamics, and electromagnetics using CFD.
How does it work? Keeping it simple, the fundamental physical laws of fluid dynamics are programmed into a software code. A CAD model is produced of whatever shape wants to be analysed. The volume around this shape is transformed into a mesh. Various boundary conditions are entered into the code such as what the fluid is (water, air etc), temperature, pressure, onset flow velocities, etc. Using this, the solver of the CFD software calculates how the flow moves between each node in the mesh. The more fine the mesh, the better the resolution of what is happening. For example, on the surface of the shape typically a much finer mesh is applied.
One of the great advantages of CFD is that structurally impossible solutions can be evaluated. For example, a wheel without spokes, a bike with one wheel, a car with no wheels. There are many cases when such simplifications can make sense for reducing the size and complexity of a model, or for understanding the contribution of any individual component on the flow field.
As computer processor speeds have improved, the viability of CFD as a development tool has boomed. A single simulation which 20 years ago took weeks on a supercomputer, today takes just a number of hours on a powerful PC. Due to the ever increasing computing power, the speed of calculation is constantly dropping and meshes have been able to become much more refined enabling ever more accurate simulation results.
However, like any instrument, CFD is just a tool and requires specialised knowledge and experience in aerodynamics for setting up and interpreting the solutions. Poorly performed simulations can be inaccurate and even downright wrong.