Variable pacing on terrain in road cycling time trials
What modelling studies say
Key Finding
Zignoli 2020 (abstract): a rider-dynamics model optimised over simulated 40-km time trials with varied slope, wind, corners and tyre-road friction, with maximal power constrained by the critical-power model.When technical sections made up 25% of the course, road conditions meaningfully changed finishing time and the peak power required, but not the pacing strategy; time lost in slow technical sections could not be regained on fast straight sections, even when those sections were used to restore anaerobic capacity. The author calls for experimental tests of these predictions.
Yamamoto 2018: optimal pacing under a nonlinear power constraint (no abstract available; content not checked here).
Practical Application
In a critical-power-constrained simulation of 40-km time trials (Zignoli 2020), corners and slippery road conditions changed finishing time and the peak power needed but not the optimal pacing strategy, and time lost in slow technical sections could not be won back on the straights — on a technical course, carry speed through the corners rather than planning to recover the loss later.
The common advice to push harder on climbs and into headwinds and ease on descents and tailwinds comes from this modelling literature; the size of the surge (e.g. "+10-20% on climbs") is not supported by the checked sources and is not given here.
These are model predictions, not measured race outcomes.
Limitations
Simulation only; no experimental or race data in the checked abstract.
The Yamamoto 2018 source has no abstract available and was not checked.
How this entry was checked and changed
Why this level
regraded 2026-09-27 from B+: modelling studies, predictions not tested experimentally
Content correction 2026-09-27
"Power increase on climbs (+10-20% above target)" removed -> not in the checked source; Zignoli 2020 reports that road conditions change time and peak power but not pacing strategy, and that time lost in technical sections is not regained.
The "supporting meta" had no identifier and was removed from the sources line.
Level B+ -> C (unvalidated modelling, per the rubric).
2026-09-29: DOI 10.1177/1754337120974872, already named in the entry's Representative sources line, attached as the entry's DOI after verification on Crossref (title and author match Zignoli 2020). All findings and the practical lead come from that abstract (checked 2026-09-27).