The parameters of the power-duration relationship, critical power (CP) and W′ can be used to accurately predict severe-intensity (>CP) exercise performance. Both endurance-based sports such as cycling and team sports involve severe-intensity exercise. Thus, the optimization of severe-intensity exercise performance is highly relevant. It is advantageous if athletes can monitor their energy usage and energy availability either during an event or in retrospect, in order to inform tactics and pacing strategy. Mathematical models of energy expenditure have been developed, based on the CP concept and designed to measure the dynamic balance of W′ (W′BAL) during exercise. The accuracy of the current W′BAL models is equivocal and may be influenced by day-to-day variability in estimates of CP and W′. Additionally, it is known that dynamic changes in the power-duration parameters can occur during exercise – such changes are currently unaccounted for in the W′BAL models. The purpose of this thesis was two-fold. Firstly, to investigate the accuracy with which the current W′BAL models were able to characterise W′BAL and the influence of inherent variability in CP and W′ on the accuracy of the W′BAL models. Secondly, to examine whether dynamic changes in the power-duration parameters occurred during team sport type exercise. Study 1: W′BAL was modelled for a 16.1-km road TT, using both the point (Best individual fit; BIF) estimates of CP and W′ and the upper and lower 95% CIs of both parameters. Upon completion of the 16.1-km TT, predicted end W′BAL was -14.7 ± 26.4, - 0.82 ± 5.52, and -14.2 ± 20.0 kJ for Morton’s W′BAL model (W′BAL-MORTON), the integral W′BAL model (W′BAL-INT) and the differential W′BAL model (W′BAL-ODE) respectively, with no significant differences (P > 0.05) between models. When accounting for the 95% CIs of CP and W′, the ‘bandwidth’ in predicted end W′BAL was equivalent to 278, 134 and 292% of starting W′ (W′0) for the W′BAL-MORTON, W′BAL-INT and W′BAL-ODE models respectively. These results indicate that: (i) the current W′BAL models have poor predictive accuracy and; (ii) That inherent variability in the power-duration parameters significantly influences the accuracy of the W′BAL models. Study 2: CP, W′ and the speed at which W′ recovered following full depletion (W′REC) were measured in a fresh condition (no exercise prior to fatiguing bout) and following; (i) one 40 min block and; (ii) two 40 min blocks separated by a 15 minute ‘half-time’ interval, of simulated match play. CP and W′REC did not significantly decline (P > 0.05) relative to the fresh condition following either 40 or 80 min of match play. W′ was significantly lower (P < 0.05) in comparison to baseline in

3
both the 40- and 80-min conditions but did not significantly differ (P > 0.05) between the 40- and 80-min conditions, indicating that the 15-minute recovery interval permitted full recovery of W′. The above results indicate that significant dynamic changes in the power-duration parameters do not occur during team sport type exercise. Taken collectively, the findings of both studies inform us of potential sources of error in W′BAL modelling.