Thoughts on intervals.icu data breakdown

Power is measured in kJ, and its as accurate as your power meter. Calories is an estimate derived from kJ work, based on some assumptions that may or may not be true for you. It can be a pretty good estimate, but it is an estimate.

Cycling is generally considered to be 21-24% efficient and 1 kcal (Calorie) = 0.239 kJ. So you can pretty much just use the kJ number for Calories. For every kJ of power you get to the cranks you are wasting at least 3 more in heat etc…

I don’t know how Garmin or TR go about calculating the Calories they report.

2 Likes

Simply put:

Energy = [ P x T ] / [ 4184 * ME ]

P = Power in Watts
T = Time in seconds
ME = Mechanical efficiency (assume ~23%)
4184 = constant J/kCal

2 Likes

I like it a lot. I had calculated FTP with ramp tests, but intervals.icu consistently estimated my FTP about 4% lower. Did a 20 minute test (first in years) and it was within 0.3% of intervals.icu’s estimate.

3 Likes

I am new to intervals.icu and had a couple of questions. Regarding the fitness chart. Last week I took an easy week and the fatigue line was consistently above the fitness line. This week as I did 3 hard sessions; 1 threshold 4.9, 1 sweet spot (4x12 @94%) and 1 sweet spot 2x40min @90%), other rides were easy, tempo type rides. This week the fatigue line is below the fitness, which is supposed to mean I am not in an optimal training zone. How can that be? Am looking at the fitness chart incorrectly?

Other question pertains to the eFTP. I did an 8min FTP test a few months ago and came out with 3.60w/kg, however, intervals.icu is estimating my FTP at 3.2x w/kg. Am I looking at the chart wrong or over performed on the 8min test?

I have never looked at my data in any meaningful way, but decided to give it a shot this year to see if helps my training. I am self-coached.

Hi David, tried using intervals again over the weekend and it isn’t detecting intervals the way I would like. Used the Actions > Use Laps at the bottom of the web page. That sort of worked. Two different problems described below:

Sunday Actual vs target in Garmin Mobile app (iPhone):

Strava simplified view:

and this is what I get with with Use Laps:

and

Intervals got the correct lap structure, but is not time aligned.

Here is Saturday:

Strava simplified:

Intervals are 1-min over, 9-min under, and 5-min recovery. Now here is intervals after applying Use Laps (repeatedly). Same thing on iPhone and MacBook Pro:

Lap structure is incorrect.

No problems with those same rides in Strava, TrainingPeaks, WKO, Garmin Connect.

And staying with that workout two more issues:

  • it never seems to detect short sprints, this is a season best 7-sec sprint:

Auto-detect instead of laps

  • Note it failed to detect the 1-min intervals on the 3rd and 4th sets. The power drop on 3rd set was at a 4-way intersection, which resulted in auto-detect splitting up that 9-min interval and detecting 5:42.

Unfortunately lap data from the Strava API gets mangled if you have a privacy zone setup. I am assuming that is the problem here because I don’t know who you are on on Intervals.icu to go look.

The auto detection works for nice square intervals without dips and so on. Interval detection was one of the first problems I tackled on Intervals.icu and I could certainly do a better job of it with what I know now. So for those you need to manually adjust by dragging the edges to extend and merge intervals.

For short sprints it expects at least 4 before it starts picking them up. This is to avoid random “non-intervals” in rides that you don’t really care about. You can hover over that position and press A to create an interval there.

Tx for giving it a try!

7 Likes

I think it’s great. I mostly use it for comparing my annual power PBs.

I’m a bit sceptical of its estimated FTP calculations - they seem way high.

What time frame are you using for E’FTP calculation? I think the default is 90s which can lead to scarily high numbers. Ive changed mine to 10mins and I haven’t really done any specific 10 min all out efforts so its a wee bit low for me at the moment.

1 Like

Intervals.icu estimates you FTP from a best effort over 3min by default. If 3min or 5min is a particular strength, it can be too high. Change the estimation duration in the settings.

Em, not sure - there’s a list of options for calculating e ftp or e critical power, I can’t see anywhere to choose the length of effort they use to estimate it?

All the options estimate high for me.

Somehwere near the bottom in the settings (from the menu on the left), it says 180s by default.

(Haven’t got it in front of me, and cant remember what its called atm)

Got it. Thanks! It was set at 5 minutes, which is my highest on all the comparison charts. Set it at 17 minutes (longest full gas effort) and its much more realistic now.

3 Likes

Hello David

Great work on intervals.icu.

Something on the power page to explain that you can adjust the time of interval used for the estimation of ftp would be a good addition. Maybe even make it possible to adjust the time from that page.

Cheers, D

1 Like

Ok, I deleted one Strava activity and manually uploaded the Garmin .fit file (transferred from 530 to laptop). Lap data was inaccurate. So then I revoked Strava access, which deleted everything on intervals.icu, requested export from Garmin, setup Intervals-Garmin sync, and Intervals.icu is now in the process of importing all the Garmin history.

Right now I can see most of 2022 on my intervals calendar, and again, quick spot checks of the earlier mentioned workouts (and a few others) shows the lap data is inaccurate. So the problem does not appear to be limited to imported Strava workouts and privacy zone (which I use) :man_shrugging:

I pulled up a recent example with six 10-second sprints followed by 50-sec rest, and ending with four all-out sprints. Nothing detected. And with the original workout synced from Strava, I could “Use Laps” and it displayed the laps correctly. But now the Strava activities are gone, and with the freshly imported Garmin fit file it will display those 10-sec / 50-sec laps. Weird. Ok not trying to sound negative but I give up, it can’t just be my data which is fine in every other platform I’m using (GC, Strava, TP, WKO).

Then a free ride with 4 full-gas ~30-sec sprints up a hill separated by about 20 minutes. Seems to have done better, except for this extra credit effort at the end:

The 10 seconds before the detected 9-sec interval has average power of 824W.

Going forward I’ll periodically check on progress, but things are so far off from any reasonable expectation I can’t be bothered to do more to help. I suspect you are aware of the limitations, or everyone is using detect Intervals and don’t care about using the actual recorded lap data. Hope that helps.

Is there a free way to get apple health data imported

Agree - this is the one thing that’s missing from Intervals.icu for me. Still have to use Garmin Connect to see the mile splits for my runs.

Its not free but HealthFit does a good job and is a once-off purchase.

@WindWarrior Please msg me a link to one of your activities with lap issues. I can’t really say anything much without looking at the data. People (well some) do use lap data successfully on Intervals.

2 Likes

Sent a lengthy note, despite saying I give up. Hope that helps.

Is there any way to easily compare the power and hr of a ride? Basically to see if HR is coming down of a specific HR