Power Pacing Case Study:  Fitchburg TT

2003

words and images by Kraig Willett

Awhile back on the older, better, more open dialogued wattage list, a question was posed about time-trial pacing strategies at the 2003 Fitchburg race.  The basic question was essentially: "which is better - a constant power strategy or a variable power strategy".

The debate intrigued me then, and recently I set out to see if I could reproduce what the literature had already demonstrated - that a variable power strategy is best when the course conditions are variable.

Method

For simplicity, only the course profile was digitized (course heading and ambient wind conditions were neglected).  Furthermore, a simplified power model was used - inertial effects and drivetrain efficiencies were ignored.  The power model only looked at overall aerodynamic drag, gravitational effects, and rolling resistance.

Figure 1: The Fitchburg TT course profile that was provided by a list contributor.

Figure 2: Digitized course used for analysis.

The way I approached the numerical solution of the problem was as follows:

·         Digitize course (300+ segments were generated)
·         generate a power profile based on road angle constraints (theta +/- 0.01, 0.03, 0.05 radians), variable power assumptions (+/- 5% and 15% of average power), and maintaining the same average power for each trial (275 W and 400 W)
·         iteratively solve power equation for the velocity/time during each individual course segment
·         sum the segment times to determine the total time

The process was a bit tedious, but in no way excruciatingly difficult.

Results

The results I generated were pretty consistent with the literature, which reported a 2 second benefit on a 16 km course when using a variable power pacing strategy (+/-5% deviation from mean power) compared to a constant power strategy.  For the ~7.5 mile Fitchburg course, a +/-5% deviation in power when the road angle was greater than theta = 0.05 radians, resulted a time savings of 4 seconds for the 400W case (0.5% of total time) and 6 seconds for the 275W case (0.5% of total time).

In the figure below, the +/- 15% of average power case is shown.  While not necessarily realistic in terms of

physiological limits (bike racers are robots, though, aren't they?) this figure shows that when the road goes up, one rides harder, and when the road goes down, one eases off a bit.  In the end the average power is still 275W.

Figure 3:  Variable power profile (red) and course profile (blue).

In figure 4 below, it can be seen that even though the constraint used when determining the power profile was based on road angle, lower rider speeds trended with the higher powers.  Perhaps a different set of power constraints would improve performance even further.

Figure 4: Power and speed relationship for the +/- 15% of mean power case.
All the results from the variables investigated have been tabulated below:

Table 1: Time savings (in seconds) of variable power pacing strategy for all cases tested.

Conclusion

On the Fitchburg TT course, better performance can be had by using a variable power pacing strategy as opposed to a constant power pacing strategy.  If one is really serious, training should be done on the course over a period of time using a power meter and trying to see what is actually sustainable/realistic for the individual.