REF-847 · Wells & Marwood, 2016 · European Journal of Sport Science, Vol. · Paper · PubMed

Effects of Power Variation on Cycle Performance during Simulated Hilly Time-Trials

Cweak evidence: opinion or lab dataMeasured on a mixed groupWritten before October 2026, being re-checked

Because riders cannot hold constant power on hilly courses, the authors modelled what power variation costs.

A hypothetical 70 kg cyclist on a 10 kg bicycle was simulated over three 40-km courses made of repeated 2.5-km uphill and downhill sections at 1%, 3% and 6%, with power varied by 5-15% over distances of 0.31-20.00 km at mean powers of 200-600 W, and compared with constant power.

Variation that did not follow the gradient was typically detrimental, more so with larger swings and steeper gradients; varying power every 1.25 km was worst, costing 43.90 s at 200 W (±15%, 6% gradient). Varying power in parallel with the gradient instead saved 188.89 s at 200 W (±15%, 6% gradient).

Participants

No human participants. One hypothetical cyclist (body mass 70 kg, bicycle mass 10 kg).

Protocol

Mathematical model of cycling performance including the effects of acceleration.
Three hypothetical 40-km courses of repeated 2.5-km uphill and downhill sections, gradients 1%, 3% and 6%.
Power variation: amplitude 5-15%, distance of variation 0.31-20.00 km, mean power 200-600 W; each compared with constant power.
A further condition varied power in parallel with changes in gradient.

Key Results

Power variation not matched to the gradient was typically detrimental; the cost grew with the amplitude of variation and the severity of the gradient.
Varying power every 1.25 km was the most detrimental: at a mean of 200 W, performance was impaired by 43.90 s (±15% variation, 6% gradient).
On the steepest gradients the cost of variation was relatively independent of the distance over which power varied.
Varying power in parallel with gradient changes improved performance by 188.89 s (±15% variation, 6% gradient, 200 W).
These adverse effects were substantially larger than those previously observed during flat, windless time trials.
Results at higher mean powers are not given numerically in the abstract.

Practical Application

In a model of a 70 kg rider on hilly 40-km courses, holding the same mean power but pushing ±15% with the gradient (harder up, easier down) saved 188.89 s at 200 W on 6% slopes, while swings of the same size out of step with the gradient cost up to 43.90 s. On a hilly time trial, surge on the climbs and ease on the descents; random surges and fades cost more on hills than they do on the flat.
The steeper the course and the bigger the swings, the more it matters that power follows the terrain rather than the rider's mood.
Inference, not tested here: a rider who fades on the climbs and surges on the flats is doing the costly version; review a hilly time-trial file for where the high-power sections fell, not only for the average.
The model does not know the rider's physiology: whether a ±15% surge on every climb can be held for 40 km without fading is not tested here (REF-252 shows variable power costs more oxygen and lactate at the same mean).

Limitations

Simulation only: one hypothetical rider, stylised courses of repeated 2.5-km climbs and descents; no measured rides.
Constrains mean power, not physiological cost; the tolerability of the power pattern was not modelled.
Only the 200 W results are given numerically in the abstract.
Written from the abstract and record metadata only (full text not read).
How this entry was checked and changed
Why this level

modelling: predictions not tested on riders