W' estimate from power curve vs Wbal depletion difference


I have a rather technical question re. W’ that I’d like to use to set VO2 training interval power.

I recently did a 40m TT effort that ended with full depletion, I wasn’t able to move - and intervals estimated my Wbal depletion at 24300 J with the eFTP/eCP based on that ride. However, when checking my power curve, it estimates by W’ at 19320 J based on the eCP model, and 12420 J based on the eFTP model.

Can someone help me understand the difference? Why does the power curve estimate my W’ at a lower value than what I achieved? Not sure I fully understand the concept here, maybe one of you know this topic better.


A TT effort is just around FTP so if your W’ decreases by that amount, your FTP is not correctly set.
W’ can only be of value for efforts well above FTP. It can be used to “estimate” the target power for consecutive ~4min intervals or shorter. By definition, if you ride exactly at FTP, your W’ will remain 100%… You will be completely exhausted after an hour or less while your W’ hasn’t been lowered at all.
Might sound unrealistic but that’s the shortcoming of that model. This stuff isn’t easy to understand, it requires good knowledge of when it can or can’t be useful.

Here’s a reply on a user question from an article on CP/W’ by a good reference site:

Thanks for your kind words! The critical power model has a number of assumptions, and so will never be perfectly accurate, but in theory you can’t deplete your W’ below 0.
I’m guessing that you’re using a Garmin or similar to calculate your W’ Balance or W’bal within a particular training session? This is the percentage of your W’ that you have remaining (or in other words, the amount of ‘charge’ remaining in your W’ battery). If so, then this model also makes some pretty sizeable assumptions about the rate at which you can ‘reconstitute’ (i.e. recharge) your W’ when riding below your critical power. The W’ reconstitution model doesn’t work as well as the more basic W’ estimation model because there are more variables that have to be assumed. So you might be seeing negative W’ values either because your W’ and CP are calculated incorrectly, or because the W’ reconstitution model is not operating correctly (it’s probably a combination of the two in reality!).
Hopefully that helps clear things up for you. At the moment, I would only ever use the W’bal metric as a fairly rough estimate of the amount of ‘charge’ you have in your W’ battery. I wouldn’t recommend relying heavily on W’bal, because the modelling just isn’t quite good enough yet.

From Critical Power and W’ Explained For Cyclists (inc. Critical Power Calculator) — High North Performance


Treat W’ in the same way you look at HRV, lots of data over time, eg. If you do VO2 intervals, you should literally empty the tank by the end of each interval, rest and do it again.

If you can’t finish the full set at the same average power, then you started too hard. If you finish feeling you could have gone harder, you obviously didn’t go hard enough. If executed well, you will see the W’ drop very close to zero each time. The more you do it, the better you get.

Get to know how hard VO2 efforts are, and you will see the value you have. W’ is simply the kJ you can produce relative to time. It’s not perfectly linear, but very close.


Thanks for clarifying - and the quote is really helpful in understanding the other side of the equation (the reconstitution). Never thought that that might be the factor with errors. Appreciate taking your time!

Seems somewhat counterintuitive to me, but I don’t have experience doing it like that.
If you empty W’ on each interval, you need very long recovery in between intervals. Probably 2-3 times the duration of the VO2max intervals? How many work intervals do you recommend? This way of doing VO2max work, targets the VO2max Power (pushing up the ceiling) while more intervals with shorter rest (2min) will rather target the VO2max capacity (pulling up the floor).
From what I’ve seen and read, the recommendation is to use the second way (short rest - slightly lower power) until you notice a plateau, and then do a couple of VO2 workouts in the first way to recreate room above that value.
Do you agree or do you have, as a coach, other experiences?

1 Like

On this note: If I assume that my CP is estimated well from the power curve, can I use the Ramp test to figure out my W’, or it needs to be a constant power effort to get to the right W’? I’d assume that going till exhaustion with the ramp depletes the W’ fully, and therefore identify my ceiling - and then adjust workouts based on MedTechCD’s approach potentially

Maybe I should say, that of the data I have seen, it’s true.

The VO2 intervals are done as maximal efforts for the whole set, with a 1:1 work:recovery ratio, but is not set at exactly 5 minutes. It’s all feel. If you feel like a 6-min recovery is needed, then take 6. Max is 8 minutes.

In one block, after the full base period is complete and we get ready for race efforts, it would be as follows:

Wk1: 3x4m
Wk2: 4x4m
Wk3: 5x4m
Wk4: recovery
Wk5: 3x5m
Wk6: 4x5m
Wk7: 5x5m
Wk8: Recovery

All at close to 120% of FTP as possible. There shouldn’t be anything left in the tank at the end of the last interval. The average power for the full set is used to set a goal average for the next week.

This is not for a beginner, but rather someone that has been doing 2-3 years of racing, plus has a solid aerobic base. They need to know what 4-5m at 120 feels like. It might stretch to 6min, but not at 120%.

The W’ drops very close to zero each time, and is consistent across the intervals.

The funny thing, is that 2 sets of AC intervals of 3x2m, 3x2m and 4x2m very close to 150% of FTP, yields a similar full emptying of the W’

It’s better to do a a range of tests, on different days from 3min to 20min; all of them an all out effort, but evenly paced.

There are formulae available to calculate the CP from a 3-min and 12-min, but many modeling software programs can automate it. W’ is also estimated from using the PDC generated from the tests.

Obviously the more one tests, the better one gets at getting close to the value. Remember that it is never the same each day, as we aren’t robots capable of producing exact numbers day in and day out. That is why trends and averages across a range is better.

You could check out Xert - I (and it seems many) have found their modeling of efforts above threshold and predictions of failure points to be very accurate (requires good data, and there is a bit of a learning curve, but not so bad). A big advantage is that they don’t require you to do constant power efforts for fixed durations to develop a curve - they analyze all of the points at which you’ve ‘failed’ before, including real life variable efforts, to identify parameters to represent aerobic, anaerobic and sprint capability (they use different terms, but that’s how I understand it). They also have workouts based on that modeling, which I think would suit what you’re looking for (or you can build your own based on that modeling). Am not affiliated with them btw.

Re other comments, most will find going ‘to failure’ (W’ to zero) at the same power output repeatedly near enough impossible if threshold and W’ are ‘correct’. Maybe possible for the very highly trained (as noted, to be fair), but for most amateurs even depleting half your W’ each time x5 is a tough workout already. And that may actually feel like it’s all-out anyway, even if not technically bringing W’ to zero.

Also, whether 120% of threshold for 5 minutes fully depletes your W’ is highly individual, depending on the balance of aerobic and anaerobic capacity. As a ballpark starting point can work, but I wouldn’t get hung up on that exact number.

1 Like