Aero drag is 80-90% of the overall resistance affecting a rider. Roughly 80% of this aero drag is the rider, the remainder is the bike. Aerodynamics are important!
Dr. Andrew Coggan Ph.D., “The individual pursuit: demands and preparation”
For one example, in a 4km pursuit, 2% of power is spent overcoming drivetrain friction, 5% rolling resistance, 7% kinetic energy (change in velocity, in this case, acceleration), and 86% aerodynamic drag. That's on a flat velodrome, with high acceleration at the standing start, yet aero is clearly still the largest source of resistance.
Aerodynamics are still important in other riding conditions. Drafting in the bunch, you need about 30% less power. (Jeukendrup, High-Performance Cycling). This doesn't mean there's no benefit from riding an aero bike in the peloton, it just means you'll keep about 70% of your aero bike's aero benefit when you're in the draft. (By the way, if you don't believe in aerodynamics then try riding your next road race without drafting.)
This is why we are so concerned with reducing aerodynamic drag in our designs.
To understand aerodynamics, we have to understand how air flows around a body moving through it. Initially the air is undisturbed. As the body passes through the air, it divides the air to the left and right (e.g., a bicycle) , or above and below (e.g., an airplane wing), forcing its way through. The air presses on the body as the body forces its way through, and the component of this pressure that faces aft is called pressure drag.
Air, like all fluids, has viscosity. Thus the air molecules that come in contact with the body stick to the surface; they are stationary with respect to the body. As the body continues through the air, other air molecules pass by those stuck molecules as they flow around the body in layers, following parallel paths. This is the laminar boundary layer. The viscous nature of air means this passing creates a shear force, or friction drag.
At some point on nearly all bodies, laminar flow cannot be maintained and the air molecules tumble and mix instead of flowing smoothly. The transition point is where this turbulent boundary layer begins.
Due to its mass, moving air has inertia and therefore cannot easily follow quick shape changes, like high curvature surfaces. Under these conditions air separates from the body, creating low pressure regions that "pull" on the surface, contributing to more drag.
If the trailing edge of the body fails to guide the air neatly back together again, there is a stalled flow, a region of recirculating air that has low pressure. This low pressure again pulls on the surface, adding still more drag.
Understanding air flow around a body helps understand how drag is created, but how does the measured drag change in response to variables such as shape, frontal area, air density and speed?
- FD = Resistive Force
- ρ = Air Density
- V2 = Velocity Squared
- Cd = Coefficient of Drag
- A = Frontal area
In the drag formula, drag force is what we want to minimize. The "half" is a constant required to make the formula work in real life. Rho is the air density. On race day, we assume every athlete has the same air density, but at any rate we cannot change it. V is velocity; we want to maximise this. Cd is the coefficient of drag, the inherent drag related to any particular shape. A is the frontal area, the area included in a front view silhouette of the body.
As bicycle engineers, we can't change most of these variables; in fact, the only parameters affecting drag that we control are shape and frontal area.
Let's look first at a few different shapes. To make a comparison of just the shape's effects on drag, let's give them all the same frontal area.
Aero Shapes and Relative Drag
Any shape has its own Cd and for the range of speeds common for bikes, it does not change. As you can see from the chart above, different shapes have different effects on drag. For example, a round shape has roughly 24 times the drag of a TrueAero shape of the same frontal area. This sounds bad, and it is, but remember most round-tube bikes don't present the tubes at right angles to the airflow - the seat and down tubes for example, have an angle that means the air really doesn't see a circular section, but closer to an ellipse. Looking now at the ellipse, you can see from the chart that for the same frontal area, it has only about 4 times the drag of a TrueAero shape.
Clearly shape matters a lot.
The other parameter we control is frontal area. This is quite intuitive: drag goes up for bigger things. This is reflected in the drag equation above, with "A" as a factor on the right side of the equation. Double the area, double the drag.
So that's how drag is created, and how we can control how our designs generate more (or hopefully less) drag. Now let's look at some ways we can measure the effectiveness of our designs.
We measure drag in two main ways: wind tunnel testing and Computational Fluid Dynamics (CFD).
Wind tunnel testing allows us to create a situation close to “real world” conditions in a controlled environment. Precision balances allow accurate measurement of drag forces (and side & lifting forces, plus rotation about all three axes). Repeatability between tunnels & methods is limited, so we use baseline standards: bikes that we never change, so we know we're testing a constant object and thus how (and whether!) the tunnels can be compared. A good wind tunnel gives the drag value, but not why a shape is good or bad (though you can get some insight with smoke, tufts, etc). A good wind tunnel is also expensive.
Of course to know how a bike performs in the real world, it's good to include in the wind tunnel as many factors from the real world as possible, like a rider. But for a human rider a wind tunnel can be uncomfortable. Even in a "comfortable" tunnel, its simply difficult to hold precisely the same position all day, day after day (in our case; most companies don't spend weeks at the wind tunnel like we do). Nevertheless, we still use live riders — when we're positioning our pros. When we're engineering bikes, we can't tolerate the imprecision of live riders.
We solved the problem of rider discomfort — and also the problem of consistency — by creating a mannequin of U.S. National time trial champion and pro rider Dave Zabriskie. We call this mannequin "Foam Dave," and use him when comparing various frame designs. Foam Dave is 100% consistent and repeatable, whether on the same frame or on different frames. Our consistent mannequin means that measured drag differences are due only to changes in bike design, not the rider.
We frequently have hundreds of ideas to reduce drag. To measure the effect of each idea properly, we need to make only one change per test. It's difficult & expensive to make so many frames that are the same except for one detail. It's also time-consuming to change out so many frames during tesing and swap over the live rider or test dummy.
Solution: Cervelo's test mule frame.
We designed and built a frame specifically to test our ideas in the wind tunnel, a test mule. It has a steel core so it can be used with either the test mannequin or a live rider. The exterior skin is SLA plastic, rapid prototyped parts that bolt on to the steel core. Different parts of the frame can be changed to test their aerodynamics. Test shapes are produced directly from the computer model using stereo-lithography (SLA).
Together, Foam Dave and our test mule let us test more ideas faster. We arrive more quickly at the most advanced designs.
Remember the wind tunnel has its limitations; a good wind tunnel gives an accurate drag value, but it's not easy to understand why different bikes have different drag. Enter CFD.
CFD = Computational Fluid Dynamics
CFD is a “virtual wind tunnel”. It's a software program that can show details of airflow that are not easily visible or measurable with normal wind tunnel techniques. But CFD analysis is only as accurate as the model, so Cervélo was the first to make an accurate rider model, again from rider Dave Zabriskie. In fact, the same scan data was used to create "Foam Dave" and "Virtual Dave," which means our CFD model is exactly the same as our wind tunnel model. This allows Cervélo to understand the effects of airflow on the rider and bike together. The results of a CFD analysis can show airflow, pressure distribution, drag values, where drag is created and what causes it.
Above is an image from one of our CFD analyses. The colors on the surface indicate pressure, with high pressure shown in red and low pressure in blue. Bonus points if you noticed there are no handlebars shown. Obviously this isn't easy to do in a wind tunnel, since normally the rider leans on the bars! However, in CFD's virtual world we can do nearly anything, including test without bars. Nevertheless, in this analysis, the CFD simulation was run including the aerobar, so the results mean something to a real rider, but we've just hidden the bar from view so we can more easily see the air's effects on the rider.
Recall from the drag equation that drag force increases with increasing speed; clearly aerodynamics are important for fast riders like Dave Zabriskie. But what about the rest of us? Take a look at our Slow vs. Fast Riders article.