PURPOSE: This study (i) investigates the effect of recovery power (Prec) and duration (trec) on the recovery of the curvature constant (W') of the power-duration relationship, (ii) compares the experimentally measured W' balance to that predicted (W'bal) by two models (SK2 and BAR), and (iii) presents a case of real-time performance optimization using the critical power (CP) concept. METHODS: Seven competitive amateur cyclists performed a ramp test to determine their V˙O2peak and gas exchange threshold, two to four 3-min all-out tests to determine CP and W', and nine intermittent cycling tests to investigate W' recovery. The intermittent cycling tests involved a 2-min constant work-rate interval above CP, followed by a constant work-rate recovery interval below CP (Prec and trec were varied), followed by a 3-min all-out interval. RESULTS: There was a significant two-way interaction between Prec and trec on W' recovery, P = 0.004 (η = 0.52). Simple main effects were present only with respect to Prec at each trec. The actual W' balance at the end of the recovery interval was less than the W'bal predicted by both SK2 (P = 0.035) and BAR (P = 0.015) models. The optimal strategy derived from the subject-specific recovery model reduced the race time by 55 s as compared with the self-strategy. CONCLUSIONS: This study has shown that in a recovery interval, Prec has a greater influence than trec on W' recovery. The overprediction of W'bal from SK2 and BAR suggests the need for individualized recovery parameters or models for sub-CP exercise. Finally, the optimal strategy results provide encouraging signs for real-time, model-based performance optimization.
PURPOSE: This study (i) investigates the effect of recovery power (Prec) and duration (trec) on the recovery of the curvature constant (W') of the power-duration relationship, (ii) compares the experimentally measured W' balance to that predicted (W'bal) by two models (SK2 and BAR), and (iii) presents a case of real-time performance optimization using the critical power (CP) concept. METHODS: Seven competitive amateur cyclists performed a ramp test to determine their V˙O2peak and gas exchange threshold, two to four 3-min all-out tests to determine CP and W', and nine intermittent cycling tests to investigate W' recovery. The intermittent cycling tests involved a 2-min constant work-rate interval above CP, followed by a constant work-rate recovery interval below CP (Prec and trec were varied), followed by a 3-min all-out interval. RESULTS: There was a significant two-way interaction between Prec and trec on W' recovery, P = 0.004 (η = 0.52). Simple main effects were present only with respect to Prec at each trec. The actual W' balance at the end of the recovery interval was less than the W'bal predicted by both SK2 (P = 0.035) and BAR (P = 0.015) models. The optimal strategy derived from the subject-specific recovery model reduced the race time by 55 s as compared with the self-strategy. CONCLUSIONS: This study has shown that in a recovery interval, Prec has a greater influence than trec on W' recovery. The overprediction of W'bal from SK2 and BAR suggests the need for individualized recovery parameters or models for sub-CP exercise. Finally, the optimal strategy results provide encouraging signs for real-time, model-based performance optimization.