During team time trials in road cycling changing schemes are used to spread the workload over the cyclists in the team. Models that provide predictions of race performance already exist for individual time trials. It is proposed that with a performance model for team time trials, the performance of different strategies can be compared and optimised.
In literature combinations of mechanical resistance models and physiological models are used to determine the performance of individual time trials. The aerodynamic interaction between cyclists is very important to the effectiveness of a strategy. Coefficients of drag reduction between cyclists in a team time trial are presented in several studies, however most studies use groups of only four cyclists, which is not useful for a team time trial with eight cyclists. Only two studies report data for groups up to eight cyclists. These two models show different behaviour and are both used to asses the performance of strategies. Also two physiological models were used.
In the model provided in this study the resistances are calculated from the kinematics resulting from the evaluated strategy. The mechanical resistance model, including the aerodynamic interaction model calculates the power required to perform the strategy. The physiological model calculates the physiology during the race, which determines if the cyclists are able to sustain the prescribed strategy.
Genetic algorithm optimisation is used to optimise the strategy parameters, such as initial position and times spend in first position. The velocity is optimised for each evaluated strategy configuration. A convergence test was performed to determine the parameters for the genetic algorithm, which are used in the optimisation of strategy.
Using the standard strategy, where cyclists only change from first to last position, different orders are compared. From this study it was determined that the mean velocity over a 30 km team time trial could be raised by a maximum of 0.228 m/s by improving the order, depending on the model configuration. It was found that the best performing orders were those where the mean performance difference of following cyclists was lowest.
Two different strategies have been assessed where cyclists still always change from first, but not necessarily last position have been assessed on their performance. With those more complex strategies the mean velocity could be increased with 0.358 m/s over a 30 km team time trial.
The model still lacks validity, but gives a relevant insight in the performance of different team time trial strategies. The validity can either improved by using track test to validate the drag reduction coefficients or by using power data from a team time trial to show that the model predicts realistic physiology. Of those two methods the last is preferred.