Optimizing Power Distribution in Zwift: Insights from Simulation and Testing

General Discussion
1

Hello Zwifters,

I wanted to share some insights from an experiment I conducted to better understand how power distribution affects ride times in Zwift. During some rides, I noticed that certain riders could complete courses faster despite having lower average power and lower watts per kilogram. This intriguing observation led me to investigate what might be happening and how power distribution could influence performance.

To dive deeper, I downloaded a Zwift ride’s .fit file containing power, speed, and elevation data. Using this information, I estimated the drag and rolling resistance coefficients (weight with bike = 87.61159 kg (my weight without bike is 81.6 kg (180 lb), so tron bike is about 6kg, Roll = 0.0025, Drag = 0.3086613, free fall = 9.81, air density = 1.225 (all in metric units). These parameters allowed me to solve the differential equations of bike motion and replicate Zwift’s speed outputs under various power strategies.

With the model calibrated, I focused on a simplified elevation profile—a single trapezoidal hill with a 5% grade (total length 6 km). The goal was to find the shortest travel time achievable while maintaining the same total energy expenditure as a constant 200W effort. Here’s what I learned:

  1. Start Strong to Reach Cruising Speed: At the beginning of the ride, applying as much power as possible helps quickly overcome inertia and achieve a stable cruising speed. This initial burst is energy-efficient and pays dividends over the course of the ride.
  2. Rest on Downhills: When descending, gravity does much of the work. It’s often more efficient to reduce power output significantly or even coast, letting momentum carry you forward. This strategy conserves energy for more demanding sections of the course.
  3. Attack the Uphills: On inclines, applying maximal sustainable power is key. Uphill sections disproportionately affect overall travel time, so maximizing effort here yields significant time savings. Importantly, don’t ease off immediately when the grade flattens—continue pushing until you regain cruising speed.

This might be obvious to more seasoned Zwift riders, but I was surprised to learn that easing off immediately when the grade flattens is a bad strategy, as it is better to keep pushing until you reach cruising speed. Pushing harder on climbs and maintaining the effort through the transition to flatter terrain until cruising speed is reached seems to be the key for optimal energy distribution.

In this example, it takes 710 sec to finish the course with a constant 200W power, but with optimal power distribution the same exact course can be finished in 588 sec (2 min difference) using the same amount of energy. This explains how riders with lower average power can sometimes outperform those with higher average power but suboptimal distribution.

bike_power
bike_power934×519 32.8 KB

Power decomposition showing contribution of drag, gravity and roll.

power_decomp
power_decomp1015×429 42.7 KB

The effect of drafting is not included in this analysis, which is the key in races.Also this is a numerical simulation that models power profiles as piecewise linear with 31 knots (i.e., 200 m between knots), which might lead to some discretization errors.

I plan to refine this analysis by testing more complex elevation profiles and more flexible power distribution models.

Ride On!

Alex

2

Thanks, this is quite useful. I always used the hairpins as moments of rest and to refuel. I guess I shouldn’t and do that after regaining momentum

Edit: but what also might be useful is calculate normalised power, because surges although maybe efficient from an energy perspective can come at a greater metabolic cost

3

Drafting just makes it even more important to hit the hills hard, because you have the chance to save power everywhere else.

The only slight problem (potentially) with the model is this:

Let’s say your 200w average power represents 90% of FTP. Good old sweetspot. I can do that for an hour.

So my FTP would be 222w. 600w is 270% of FTP. How long am I meant to be holding this for!?!

4

Thanks for your feedback! You’re absolutely right that power surges, while optimal can be unsustainable, and calculating normalized power is a good idea to better understand this tradeoff.

As for using hairpins to rest, I used to think the same way, but I’ve noticed that easing off too early can lead to getting dropped from the peloton. Maintaining power through those flatter sections and transitions until you hit cruising speed makes a significant difference. Computer simulation seems to confirm that speed matters more than the grade itself in these situations.

It would also be great if Zwift implemented a feature to display the ‘par’ time for completing a course with constant average power. This would allow riders to gauge how efficiently they distributed their efforts during the ride.

1 Like
5

You should be able to do FTP for an hour; that’s the purpose of that number.

SS you should be able to handle for… several hours with “ease.”

Granted, holding FTP for an hour is purely a mental game not everyone is capable of, but; just to clear the air, the purpose of FTP is still at the end of the day, the power you should be able to handle theoretically for an hour.

As far as iTT’s goes, this is literally nothing new, this is just simply “how” one should do an iTT, and what makes someone good or bad on power alone, and specifically why some folks can, as OP stated, go faster with less power&w/kg.

Though it’s going to be all about knowing your limits, and knowledge on the route (knowing what’s coming up).

Same goes for racing a car; faster the exit speed of a turn, the faster your lap will be. Slow in fast out is exactly that; a well handled turn but a higher exit speed means a higher averaged speed down the next straight, and that’s where all the time is made… it’s not made diving deeper into a turn.

6

I didn’t say I could do sweetspot for an hour and no more, Andrew. If I can do it for two hours or 90 minutes, I can also do it for an hour, yes? :wink:

I could nitpick your definition of FTP as 1 hour power, but I won’t, because that’s been done to death too many times already.

7

:crazy_face:
the war is not over!

(Though, still, FTP as a number should be possible to produce for 1 hour [not saying it IS 1 hour power, just stating it’s a theoretical number that should be obtainable]; how it’s calculated depends on the source, but, assuming it’s correct, then anyone should be able to do it. Whether they want to or not however is the reality of it.)

That’s not to say FTP “is” 1 hour power. Just saying it’s what one should be able to do.
If you cannot hold FTP for an hour, then either the test method doesn’t suit you, or you’re greatly out of form to test it.