Sport Engineering - Road to World Championships

In order to improve my racing performance, within the sports of Triathlon, Duathlon and Time Trial cycling, I looked towards my bike. This presented an optimisation problem, with coefficients including drag, comfort (with a relationship to power output), friction and how to fuel my body for sustained efforts in races.

Throughout this process, I used skills and techniques such as CFD (computational fluid dynamics) and CAD and relied heavily on knowledge of tribology and mechanics. These modifications were carried out from November 2018 up until the European Duathlon Championships in March 2020 - the last race of the international calendar before COVID-19 lead to widespread cancellations and postponed races.

The work helped me go from amateur racing, performing reasonably well at local races, through to qualifying for and racing in two European championships (2019 and 2020) as well as qualifying for the 2020 World Multi-sport Championships (postponed to September 2021 due to COVID-19). At the 2020 European Duathlon Championships, I recorded the fastest cycle time at the event, covering the 20.1km distance in 29 minutes and 3 seconds according to GPS data, in between two running stages.

Biomechanical Optimisation

The first step in this process was to use common cycling principles, to maximise my power output, in the aerodynamic position afforded by the geometry of my bike.

The bike in question is a Felt B16 2019 Model, in size 58 (the length of the top tube and effective seat tube).
My height is 190cm, with an inside leg measurement of 88cm.

With the fixed geometry of the bike drawn in a Solidworks sketch, this left my body with 9 degrees of freedom, on 4 points of contact with the bike. These are:

Crank Length
Saddle Height
Saddle Tilt
Saddle Position
Stem Angle
Stem Height
Stem Length
Aerobar Angle
Aerobar Height

There are suggested values available from professional bike fitters, such as Aerocoach, for measurements like the angle you foot should be from your leg when pedalling, the angle at which your arms should be bent in order to maximise comfort on the arm pads, etc.

My main aim here was to increase what is known as the hip angle (the angle between your thigh and your torso) at the highest point. This reduces physical strain on glutes and quads when pedalling, allowing for a greater power output, as well as reducing the peak acceleration of your femur during each stroke, in theory reducing overall strain on the legs.

I modelled this scenario as a 4-bar linkage. My foot must remain at a 110 degree angle to my shin, and at full extension my knee should still retain a 20 degree bend to provide the most efficient stroke with minimal risk of injury. I can modify the length of the cranks, my saddle height, saddle tilt, saddle position (how far forward, altering the effective seat tube angle).

The interactive Solidworks sketch allowed me to roate the cranks through full strokes, with these fixed parameters, ad eventually I found an 800mm saddle height, -9 degree saddle angle, and a 76 degree effective seat tube angle to fit my anthropometrics.

With the lower half optimised, the upper came next. Aerocoach suggest that around 60% of your back should ideally be parallel to the ground, allowing a close fitting helmet tail to lay upon it, reducing flow separation. An angle between 110-115 degrees at the shoulder is commonly found on riders with the fastest Ironman times, for example, Jan Frodeno. I have found that an 80 degree angle at the elbow suits me well, reducing weight on my forearms and placing it nearer to the joint, reducing leverage. Again using the interactive sketch along with standard stem lengths and angles, and common bar angles and stack heights, I concluded that a 35 degree bar angle, 5cm stack height, 100mm -17 degree stem would fit my proportions.

Custom Parts

Hydration System
Another consideration to improve my racing performance is the need for hydration and fuel. This is only a concern for Olympic distance triathlons and up, with a race time of around 2 hours or more. Sprint distance races tend to be under an hour, whereby my body is capable of performing without need for water or glucose.

The standard solution to this is to use bottle cages mounted to the down tube. This however is not very aerodynamic. Data from XLab USA - a specialist in aerodynamic hydration systems for triathlon - suggests that a standard mounted bottle requires an additional 2 watts to sustain 40 kmh.

They found that just by moving this horizontally, and placing between the forearms, this was reduced by 64.7% via extensive CFD analysis.

I proceeded to go further, designing a bottle that integrates in line with the stem, head tube, and acts as a fairing over the front brake caliper. Initial CFD analysis showed that this actually reduced drag on the frame of the bike, by closely hugging the top of the front wheel, and providing a smoother path for air flow around the front brake, and onto the teardrop shaped head tube. Additionally, the use of a hydration tube coming up from the 1.035L internal reservoir allows me to remain in my race position while taking on fluids. Standard bottle cages would require me to break position to reach the bottle, and stay out of position while taking on fluids, significantly increasing the drag due to my more upright position.

This was designed in Fusion 360, and printed in black ABS on an Ultimaker 3. I printed this in 5 components, in order to fit easily on the print bed, minimise overhangs, reduce finishing work required and to allow the fill valve to be removed for cleaning.

This was assembled, the tongue and groove fittings printed were bonded with food-safe epoxy, and then the inside lined with a lower viscosity food grade epoxy to ensure it was water tight, and easy to clean. This was then secured to my stem face plate, meaning the bottle can be easily unbolted and taken off for races when additional weight is not needed.
The surface was then sanded down to 400 grit, painted black, and lacquered to provide a smooth, glossy surface finish.

Stem Transition
The next modification to the cockpit was the addition of a fairing behind the stem (the part which connects the handlebars to the steerer tube). This part, even at it's lowest most "slammed" position, sits about 40mm above the top tube, leaving a vertical face that air flows directly over.

XLab found that without any form of fairing this could detract up to 1.8 Watts at 40kmh. They then showed that by adding in a gradually sloped fairing this could be reduce to as little as 0.3 Watts.
The Felt B16 however does not come with any such part available to purchase. Some brands, such as the Canyon Speedmax and Giant Trinity come with these fairing as standard, and actually use them as a storage box for nutrition / energy gels.

In order to replicate this design on my bike, I bought a Profile Design ATTK IC. I milled out the bottom of this box and added a slot to allow the cockpit cables to go through this box and into the frame of the bike.
This was then bonded to the frame using epoxy putty, filling the voids between the walls and the top tube to create smooth transitions for the airflow.

This creates a fully integrated cockpit, with hydration and nutrition storage, fairings around the stem and front brake, and with internal cable routing, theoretically saving 5.5 Watts.

XLab USA Demonstration of a standard bottlecage setup between the downtube and seat tube.

XLab USA Demonstration of airflow over a standard stem, and a stem followed by a storage box, like the profile design ATTK IC

Chain Optimisation

Chain Alignment

While manufacturers always place emphasis on the aerodynamic performance of their frames, and weight savings between generations of products, something they don't focus much on is chain performance. The company Ceramicspeed are pioneers in this field however. Their research has shown that the standard chain, chainring and rear cassette system is around 90-97% efficient depending on use.

Their research with FrictionFacts has shown that "Lateral chain misalignment creates frictional losses, and the losses increase as the angle of misalignment increases". Therefore, the first aim of chain optimisation is to ensure that for my projected race pace and preferred cadence, my chain is perfectly aligned.

From over 18 months of training data, I have inferred that my optimal cycling cadence is between 95 and 100 rpm, depending on terrain and traffic. In addition, My FTP (functional threshold power) tests on a Wattbike Atom training bike, suggest that with optimal positioning and a perfect bike setup, my race fitness will allow me to hold around 41.5 kmh for a 40 km ride in an Olympic distance triathlon (holding ~280 Watts).

Using a laser guide, I concluded that the Vision Trimax cranks (of the length specified in the bio-mechanics analysis) lead to a straight chain to the 6th cog on my rear cassette, right in the middle of the 11 speed setup common on high end road bikes.

Shimano provide an 11 speed cassette with the felt B16, this ranges from 11 teeth on the smallest cog, to 28 on the largest. With this range in mind, and my optimal cadence, speed and chain-line, I used an online calculator to work out what front chainring size I need.
In order to hold 100 rpm, with my optimal chain-line using the 17 tooth cog on the rear, a 56 tooth front chainring provides the closest match to my 41.5 kmh target.

This bike came with a 50 tooth front chainring as standard, which would lead to a shift of 2 cogs in order to hold my ideal cadence and target speed, which according to CeramicSpeed and FrictionFacts could increase the power loss in my drive-train by over 4%.

Gear Size

This report also stated that a measurable efficiency increase was noticed when larger gears were used (with equivalent gear ratios and perfect chain alignment), allowing for a smaller rate of change in link angle, and therefore lower friction. Running a 56-17 combo is already a step in the right direction with this, rather than running the 50-15 which would be the nearest standard equivalent. The pulley wheels are also mentioned, something Ceramicspeed have researched heavily.
They concluded that the standard pulley wheels offered on the Shimano 105 groupset fitted to my bike costs 1.175 Watts at a 95rpm rider cadence. Meanwhile, changing to an OSPW (Over-sized pulley wheel) with ceramic bearings brings this down to 0.033 Watts. This does mean a slightly longer, and therefore heavier chain is needed to fit the larger pulley wheels, along with larger front chainring, however given most time-trails and triathlons are flat, this causes negligible penalty.

By Making the modifcations to my chainline, chainring size, and pulley wheel, the research suggests this will lead to around a 4 Watt saving, at 280 Watts at 95-100 RPM.

Wheel Optimisation

Rolling Resistance
The first consideration with wheels is rolling resistance. With all bike wheels having fairly high quality bearings above a certain price, the main variable is tyre choice.

Of course every brand claims to be the fastest, most comfortable, best gripping tyre on the market, but the website BicycleRollingResistance.com has an extensive range of independent tests for market leading tyres.
The only caveat to these tests, is that they do not contain the full range of tyre widths, and some tests are performed on Tubeless tyres (no inner tubes), while others are Latex or Butyl inner tubes. These inherently have differing rolling resistance values, whereby Latex and Tubeless are roughly equivalent, while Butyl tubes tend to add around 2-3 watts of rolling resistance at 40kmh. These tests are performed at a speed of 28.8kmh, at a temperature of 22.0 +/- 0.5 degrees Celsius. So while these may not exactly compare to race conditions, they provide a comparison.
Additionally, the wheels that I have (ZUUS Pro Carbon 88mm) are Clinchers, meaning they can take Tubeless and Clincher tyres, but not Tubular.

From this chart, The Vittoria Corsa Speed G+ 2.0 is a clear winner, in terms of speed. As a tubeless tyre, this can be run on my wheels with Latex inner tubes at fairly equal rolling resistance. These are a common choice for professional cyclists and triathletes who are willing to sacrifice puncture resistance and durability for speed.
For my weight, speed and chosen pressure, this equates to 9.63 Watts of rolling resistance per tyre (19.3 Watts total). Comparing this with my training tyre, and Continental's previous generation flagship tyre at 14.34 watts per tyre (28.7 Watts total), shows a saving of 9.4 Watts, or 33% reduction.

Bicycle Rolling Resistance .com ranking of tyre rolling resistance for tyres at 120 psi (my preferred race pressure)

Bicycle Rolling Resistance .com measurement of Vittoria Corsa Speed G+ 1.0 tyre (23mm Spec)

Tyre Width
Tyre width is a contentious issue within cycling. The pro peloton riders in recent years have been moving to wider tyres, for increased control and recent studies suggest that wider tyres may in fact be faster on non-perfect road surfaces. On Ideal road surfaces however, it is still widely accepted that a 23mm tyre will perform almost exactly the same, if not slightly faster than a 25mm. A more important trend identified by pro time trial riders is the interface between the tyre and the rim, and how that affects the way air flows over the surface of the wheel.

Given the Corsa Speed G+ 2.0s are available in both 23 and 25mm stated diameters, I decided to use the measurements available online to run simulations of how air would flow over these tyres on my wheel rims. BicycleRollingResistance has tested a 25mm version of the Speed G+ 2.0 and a 23mm version of the Speed G+ 1.0 - I assumed the measurements will be comparable across generations.

Conveniently, the internal rim width on my Zuus pro Carbon wheels and the wheels used in the had very similar internal rim widths (the area in which the bead of the tyre interfaces with the rim) at 17.4 and 17.8mm respectively. Therefore, I assumed there would be negligible difference in the BicycleRollingResistance measurements and how they would fit on my rims.

I drew the wheels up on Solidworks, and then added the tyres with the dimensions specified. The wheels most importantly had a 25.5mm external rim width. It stood to reason in my mind that matching the tyre width to the rim width would provide the smoothest transition.

After running this through Solidworks FlowSimulation, it suggested that the 25mm tyre would provide a 13.2% increase in drag at 40kmh, over the 23mm. It seems that the 23mm tyre (25mm at it's widest point) provides a smoother transition to and from the 25.5mm rim, with smaller deviations in the flow lines. This equates to around a 2 Watt saving. This is only relevant to the front wheel however, due to the design of the B16 frame, and the fact that the air will almost certainly have transitioned to turbulent flow by the rear of the bike, the savings will be negligible.

Bicycle Rolling Resistance .com measurement of Vittoria Corsa Speed G+ 2.0 tyre (25mm Spec)

Flow lines and pressure indicators for 40kmh windspeed across the Vittoria Corsa G+ 2.0 tyre with Zuus Pro Carbon 88mm wheels. Reduced resistance was also observed at 5, 10 and 15 degree yaw angles, while more sophisticated studies suggest a closer matching profile also aids with vortex shedding at greater yaw angles.

Race Ready

Below is an image of the setup as of March 2020, a few days before the 2020 European Duathlon Championships. During this race, I achieved the fastest cycle leg in the competition, out of over 120 international athletes, cycling the 20.1km route in 29 minutes and 3 seconds, an average speed of 41.5kmh on a course with 4 about turns requiring complete deceleration.