Wind Tunnel Simulation Algorithms and 3D Course Modeling: A Scientific Field Guide to Building Dynamic Power Pacing Matrices with Best Bike Split

1. Introduction and Cutting-Edge Research Background

In the wave of scientific training for cycling, the proliferation of power meters has moved athletic performance from “going by feel” to a precise era of “letting the data speak.” However, most cyclists still apply power data only within the confines of laboratory-based Functional Threshold Power (FTP) percentage zones. Once out on real-world undulating courses with continuous climbs, descents with sharp corners, headwinds, crosswinds, and temperature fluctuations, a pacing strategy based on fixed power zones often proves inadequate. In recent years, stage physics simulation engines led by Best Bike Split (hereafter BBS) have sought to bridge this gap—transforming static power training into a dynamic “power route map” that adjusts in real time to terrain and environment.

The technical prototype of BBS can be traced back to a cycling aerodynamic simulation tool developed by engineers at the National Aeronautics and Space Administration (NASA) in the early 2010s. Through multiple iterations, it gradually integrated weather forecast APIs, OpenStreetMap elevation databases, and global race databases. By the mid-2020s, platforms of this kind could simultaneously handle real-time stage simulation requests from tens of thousands of riders, outputting FIT files (Flexible and Interoperable Data Transfer Format) within seconds—containing target power for every 100 meters, recommended gear ratios, and estimated finish times—ready to be loaded onto bike computers. The breakthrough of this technology lies in the fact that it no longer treats the rider as a thermodynamic engine with constant output, but rather views the rider-bicycle system as a six-degree-of-freedom rigid body moving within a variable flow field, solving its equations of motion using numerical methods.

From a sports science research perspective, over the past five years, several empirical studies published in the Journal of Sports Sciences and the International Journal of Sports Physiology and Performance have validated the significant advantages of dynamic pacing strategies over constant power strategies in hilly stages and multi-day races. For example, a 2022 study simulating the cobblestone sectors of “Paris-Roubaix” found that riders employing Terrain-Aware Pacing finished on average 2.3% faster, while their subjective ratings of perceived exertion (RPE) dropped by approximately 12%. Such research confirms the feasibility of stage modeling moving from theory to practice, and provides a solid academic foundation for the technical details this article aims to explore.

2. Core Mechanisms of Exercise Physiology and Biomechanics

2.1 Equations of Motion and Energy Conservation Models

The core computational backbone of BBS is grounded in the application of Newton’s Second Law of Motion to bicycle dynamics. The system treats the rider and bicycle as a combined mass ( m ) (in kilograms), with the net force acting along the tangent of the course determining the acceleration ( a ). The complete equation of motion can be expressed as:

[
m \cdot \frac{dv}{dt} = F_{drive} - F_{rolling} - F_{aero} - F_{gravity} - F_{bearing}
]

Where:

• ( F_{drive} ) is the driving force applied by the rider to the rear wheel, determined by power ( P ) and instantaneous velocity ( v ): ( F_{drive} = \frac{P \cdot \eta}{v} ), where ( \eta ) is the drivetrain efficiency (typically taken as 0.97).
• ( F_{rolling} ) is the rolling resistance, proportional to the normal force: ( F_{rolling} = C_{rr} \cdot m \cdot g \cdot \cos(\theta) ), where ( C_{rr} ) is the coefficient of rolling resistance (approximately 0.003 to 0.005, depending on tire and road surface), and ( \theta ) is the grade angle.
• ( F_{aero} ) is the aerodynamic drag, proportional to the square of the relative wind speed: ( F_{aero} = \frac{1}{2} \cdot \rho \cdot C_dA \cdot (v + v_w \cdot \cos(\phi))^2 ), where ( \rho ) is the air density (varying with barometric pressure, temperature, and humidity), ( C_dA ) is the effective frontal area coefficient, ( v_w ) is the wind speed, and ( \phi ) is the angle between the wind direction and the direction of travel.
• ( F_{gravity} ) is the gravitational component: ( F_{gravity} = m \cdot g \cdot \sin(\theta) ), acting as resistance during climbs and becoming part of the driving force during descents.
• ( F_{bearing} ) accounts for bearing and chain friction losses, typically incorporated into the drivetrain efficiency.

2.2 Dynamic Coupling of Air Density and Meteorological Parameters

The key to BBS’s modeling accuracy lies in the real-time calculation of air density ( \rho ). The density formula, corrected based on the ideal gas law, is:

[
\rho = \frac{P_{atm}}{R_{specific} \cdot T_{air}} \cdot \left(1 - 0.378 \cdot \frac{RH \cdot P_{sat}(T_{air})}{P_{atm}}\right)
]

Where ( P_{atm} ) is the atmospheric pressure (in Pascals), ( T_{air} ) is the temperature converted from Celsius to Kelvin, ( RH ) is the relative humidity, and ( P_{sat} ) is the saturation vapor pressure (calculable via the Magnus formula). For every 1,000 meters of elevation gain, barometric pressure drops by approximately 12% and air density by about 10%. This means that at the summit of Wuling at 3,275 meters, aerodynamic drag is only about 65% of that at sea level. If a rider fails to adjust their power output strategy, the reduced air resistance will cause speed to increase, potentially pushing them beyond their aerobic metabolic threshold.

2.3 CdA Variation Curves and Dynamic Modeling of Riding Position

CdA (coefficient of effective frontal area) is not a constant value; it dynamically changes with riding position, grade, wind direction, and fatigue level. BBS allows users to input multiple CdA values corresponding to different scenarios such as “aero tuck on flats,” “standing climb position,” “descending aero position,” and “low headwind crouch.” Data from the platform’s built-in wind tunnel database shows that a male rider 175 cm tall and weighing 70 kg has a CdA of approximately 0.22 m² in the “flat road aero bar position,” rising to 0.35 m² in the “standing climb position,” and dropping to 0.18 m² in the “time trial aero position.” At a speed of 40 km/h, these differences result in aerodynamic drag surging from 72.6 Newtons (N) to as high as 115.5 N—a difference of nearly 60%—making the impact on power requirements substantial.

3. Key Parameter Field Testing and Comparative Analysis

To concretely illustrate the sensitivity of input parameters in BBS stage modeling, the following uses Taiwan’s classic “Westbound Wuling” climb (from the Geographic Center Monument at the start to the Wuling parking lot, approximately 55 km in total length with about 2,800 meters of total elevation gain) as an example for parameter comparison analysis. Simulation conditions are set as: rider weight 65 kg, bicycle weight 7.5 kg, FTP 250 watts, temperature 20°C, barometric pressure 1013 hPa, no wind.

3.1 Impact of Different CdA Settings on Finish Time and Average Power

As the table above shows, merely by dynamically adjusting CdA through posture changes—without increasing average power—a rider can save approximately 20 minutes in finish time, a remarkably significant benefit. This also explains why top amateur riders in recent years still place great emphasis on maintaining an aero position during descents and flat sections, even in climbing races.

3.2 Impact of Wind Direction and Speed on Power Requirements

Next, we simulate the differences in power requirements under various wind conditions for the “One-Day Taipei to Kaohsiung” (North-South) 360 km flat course. The rider is set to output a constant 200 watts, with CdA fixed at 0.25 m² and temperature at 25°C.

This data reveals a critical fact: in the North-South race, if a rider encounters a sustained headwind of 5 m/s (approximately 18 km/h) for the entire course and insists on a constant power output of 200 watts, the finish time will extend from 10.2 hours to 12 hours—an increase of nearly 2 hours. In this scenario, BBS’s dynamic pacing matrix would recommend the rider slightly increase power to 210-215 watts to maintain a balance between economy and finish time.

4. Periodized Training Plans and Equipment Setup and Adjustment Guide

4.1 Data Collection and Parameter Calibration Before Stage Modeling

To ensure BBS simulation results accurately reflect individual capability, the following parameter calibration process must be completed first:

1. FTP Field Test: Perform a 20-minute time trial and multiply the average power by 0.95 to obtain an FTP estimate.
2. CdA Field Test: On a clear, windless day, perform 4 out-and-back aero lab test runs on a flat, fixed course in a time trial position, back-calculating CdA from speed and power.
3. Rolling Resistance Coefficient Confirmation: Confirm tire model and tire pressure settings, referencing the manufacturer’s provided ( C_{rr} ) data, or conduct a decay test on laboratory rollers.
4. Weight Registration for Rider and Equipment: Rider weight, bicycle weight, water bottle weight, and nutrition supply weight must all be entered precisely.

4.2 8-Week Pre-Race Periodized Training Plan (Targeting the Wuling Stage)

4.3 Equipment Adjustment and FIT File Loading

After completing the simulation, BBS outputs a FIT file with target power marked for every 100 meters. It is recommended that cyclists load this file onto their Garmin, Wahoo, or Bryton bike computer and set power zone alerts (e.g., target power ±10% as the allowable range). Additionally, based on the maximum power requirement for climbing sections in the simulation results, adjust the cassette gear ratio configuration. For example, if the simulation shows that the steepest section (the final 5 km of Wuling with an average grade of 8%) requires outputting 280 watts at a cadence above 55 rpm, ensure the gear combination allows the rider to maintain a comfortable pedaling cadence on that grade.

5. Race Nutrition, Environmental Adaptation, and Race-Day Strategy

5.1 Quantified Carbohydrate and Energy Intake Strategy

Dynamic power pacing means energy expenditure is not uniformly distributed. Using BBS simulation output data for the Westbound Wuling as an example, total energy expenditure for the entire course is approximately 2,800 kJ, of which the first 20 km (gentler grades) consumes about 700 kJ, the middle 20 km (average grade 5%) consumes about 1,100 kJ, and the final 15 km (average grade 8%) consumes about 1,000 kJ. According to sports nutrition recommendations, 60-90 grams of carbohydrates should be consumed per hour to maintain blood glucose stability and muscle glycogen synthesis efficiency. The specific strategy is:

• 3 hours before the race: Consume 1-1.2 grams of carbohydrates per kilogram of body weight (approximately 70-80 grams for a 65 kg rider).
• Every hour during the race: Consume 60-80 grams of carbohydrates, alternating between energy gels (25 grams of carbs per packet) and sports drinks (30 grams of carbs per 500 ml bottle).
• Within 30 minutes after the race: Consume 1.2 grams of carbohydrates and 0.4 grams of protein per kilogram of body weight to promote muscle glycogen replenishment.

5.2 Hydration and Electrolyte Balance

For every 1,000 meters of elevation gain, respiratory water loss increases by approximately 15%. Therefore, during the Wuling stage, even with cooler temperatures, fluid loss remains substantial. It is recommended to consume 150-200 ml of electrolyte drink every 15 minutes, adjusting based on individual sweat rate (measurable through pre-race weight differences). If race-day temperatures exceed 25°C, add an extra 200 ml of fluid intake per hour, and pay attention to sodium supplementation (approximately 1.5 grams of sodium per liter of sweat).

5.3 Environmental Adaptation and Weather Response

For courses with strong winds like Yangmingshan Fengzhongjian (approximately 75 km total with about 1,100 meters of cumulative elevation gain), BBS simulations place particular emphasis on the impact of wind direction changes on stability during descents. Race-day strategies include:

• Log in to the day’s weather forecast 1 hour before the race to confirm hourly changes in wind direction and speed.
• Adopt a low, aero position during headwind sections, adjusting CdA to its minimum value.
• Slow down in advance during crosswind sections (especially on the Yangjin Highway descent) to avoid control instability caused by gusts.
• If temperatures are below 15°C, wear a windproof vest and arm warmers to prevent a drop in core temperature from affecting muscle power output.

6. Common Operational Pitfalls and Debunking Scientific Myths

6.1 Myth 1: “As long as FTP is high enough, stage simulation is unnecessary”

Many riders believe FTP is the sole determinant of race performance. However, BBS simulations reveal another dimension: among riders with the same FTP, different CdA and weight combinations can result in finish time differences of 15-20%. FTP is the engine displacement, but CdA and weight are the transmission and body weight that convert engine power into forward speed. Ignoring stage modeling is like driving a high-horsepower SUV with a terrible drag coefficient on a race track—you get half the results with twice the effort.

6.2 Myth 2: “You should completely rest on descents and let gravity accelerate you”

BBS simulation data shows that on descents with grades exceeding -6%, if a rider stops pedaling, speed will reach equilibrium due to aerodynamic drag (approximately 55-60 km/h). However, if the rider continues pedaling with a light load of 50-100 watts, the equilibrium speed can be raised to 65-70 km/h while maintaining muscle temperature and neural activation. Furthermore, on the subsequent climb following a long descent, riders who kept pedaling can save approximately 30-60 seconds of physiological adaptation time.

6.3 Myth 3: “CdA is exclusively for professional riders and irrelevant to the average cyclist”

This is the most erroneous myth. According to wind tunnel and field test data, an amateur rider can reduce CdA by 5-8% simply by wearing a well-fitted cycling jersey, moving water bottles behind the saddle, and adjusting arm position. On a flat section at 30 km/h, this equates to saving 10-15 watts of power output. Accumulated over a 4-hour race, this is equivalent to saving approximately 150-200 kJ of energy—enough to support a final 5 km attack.

6.4 Myth 4: “Wind speed data from weather forecasts can be applied directly”

BBS’s weather module does not directly use ground-level wind speeds measured at weather stations. Instead, it uses a boundary layer model to convert forecast wind speeds into the “rider-height wind speed” actually experienced (typically about 70-80% of the weather station measured value). Ignoring this correction will lead to overestimating power requirements in headwind sections, thereby affecting pacing decisions.

7. Expert FAQ

Q1: How is the accuracy of BBS simulations verified? How can I confirm the simulation results are trustworthy?

A1: Verification is divided into two levels. The first level is “physical model verification”: conduct a field test on a flat course with no wind and constant temperature, then compare the actual finish time with the BBS simulated time—the error should be within 2%. The second level is “personal parameter verification”: confirm whether the input FTP, CdA, and weight are accurate. If the error exceeds 5%, first examine whether the CdA setting is overly optimistic or conservative. It is recommended to perform an Aero Lab Test once per season to recalibrate.

Q2: If race-day weather differs from the conditions set during simulation, how should I adapt?

A2: BBS’s advanced features allow riders to re-import the latest weather forecast within 24 hours before the race and regenerate the FIT file. If sudden weather changes occur during the race (such as increased gusts), it is recommended to manually adjust power targets: when headwinds strengthen, lower the target power by 5-10% to avoid premature exhaustion; when tailwinds strengthen, keep the target power unchanged and enjoy the extra speed dividend.

Q3: My FTP is only 200 watts. Is BBS simulation still valuable for me?

A3: Absolutely. The core value of BBS lies in the “optimal allocation of relative resources,” not the prediction of absolute speed. Even with a lower FTP, stage modeling can still identify the best strategy for “which sections to go all out and which sections to conserve energy.” In practice, a rider with an FTP of 200 watts can finish 5-8% faster with dynamic pacing compared to a constant power strategy—a significance that is even greater in lower-intensity events.

Q4: Does BBS simulation apply to trail running or triathlon events?

A4: BBS itself focuses on cycling stages, but its underlying physics engine (equations of motion, CdA models) can be extended to the cycling leg of a triathlon. For trail running, due to the complexity of terrain friction and gait dynamics being far higher than cycling, no commercial platform currently offers simulation of equivalent precision. However, BBS’s “power route map” concept has been borrowed by some ultramarathon coaches for planning speed strategies across different elevation zones.

Q5: Will training with FIT files generated by BBS make training mechanical and take away the fun?

A5: This is a concern shared by many cyclists. It is recommended to adopt a “flexible pacing” philosophy: treat BBS target power as a “recommended range” rather than an “absolute command.” On training days, allow yourself to deviate from the target power by ±15% to listen to your body’s feedback; on race days, strictly adhere to the target power within a ±5% range. This way, you can enjoy the exploratory fun of the training process while maximizing the benefits of scientific pacing in key races.