Intervals miss calculating load

Bug Reports
1

hello @david

could you please see this workout?

https://intervals.icu/activities/6697786391

Intervals is under estimating the load. Garmin & strava & xert tell me around 170 but intervals calculates 90? is it due to the short duration of the intervals maybe?
everytime I do this kind of workout I get undervalue load

thank you

2

The quick guess (no one besides David can see your linked activity) is that your FTP is not set up correctly?

Note that the FTP has 2 parts, indoor and outdoor version in intervals.

1 Like
3

I think Intervals.icu has it right. 90 TSS for a 1h05m workout is a lot. If that was 170 it would mean your FTP was set way too low. TSS is normalised so that a 1h effort at FTP is 100 TSS.

The normalized power is 253w which is the same as reported by Strava Sauce and that is what is used to calculate the load.

1 Like
4
  1. Do you have FTP set correctly on all platforms?
  2. Are you looking specifically at TSS or are you comparing to some proprietary number? Garmin for example has TSS but also Exercise Load.
    image
    image508Ă—539 25.1 KB
5

My watch doesn’t have that. Had to. look that up as well.

Exercise Load and EPOC

After you complete a workout, your body continues to use oxygen at a higher rate than it would at rest (the “afterburn” effect). This phenomenon is known as excess post-exercise oxygen consumption, or EPOC. Since the oxygen your body consumes relates directly to the amount of energy it uses, EPOC is an excellent measure of Exercise Load. It quantifies how much work your body has to do to get back to normal.

1 Like
6

It’s part of the Firstbeat analysis parameters, so you only have it on units with those metrics.

2 Likes
7

I have a Fenix 5x+ so that is w/ FirstBeat for sure, however, Garmin (even tho they now own FirstBeat) they don’t backport the new stuffs into their old stuffs.

so yeah…

8

thank you for your answers. Ill check my data

9

I have seen the same issue where Intervals is under estimating Load:

Duration 2:03 hrs
NP: 188W
Avg Power: 129W
FTP (across all platforms): 277
Weight: 155 lbs
Climb: 3,035ft
Strava, TrainerRoad, Training peaks: 105-106 TSS
Intervals Load: 86

I don’t see this significant gap on other rides… I was pushing hard on DH segments where I wasn’t pedaling as much and wonder if the other platforms consider some HR or speed factors.

Thanks!

LeTabiq

10

Yeah… sometimes the load in intervals.icu can be off a little (1-5TSS) and then there are times when the load is diff by 10-20 TSS and this usually comes down to how intervals.icu calc moving time.

I’ve gone down this rabbit hole a few times and that’s my perception / findings.

11

Intervals.icu only counts moving time (using the moving stream for Strava rides) when calculating training load from power. Otherwise adding 20m coffee stop in the middle of a ride actually makes the TSS higher which really doesn’t make sense.

You can test this out in the workout builder.

12

I’d like to understand this better.
Let’s assume Load Calculation priority = Power, HR, Pace. And no wheel sensor on the bike (distance and speed from GPS). That’s probably a very common situation.
In the case of Power Based Load, Power = 0 can happen under the following conditions:

  • Coasting: moving → velocity <> 0 and power = 0
  • Stopped: not moving → velocity and power = 0
  • Longer stop (coffee stop): Power meter probably falls asleep and is no longer connected, or it is out of reach. So power = 0 or N/A, velocity = 0 (or below threshold)

The first 2 cases should only rarely cause any problem, although I had my power meter disconnect once on a longer relatively straight descent, coasting for prolonged time.
What happens in the third case if you keep the activity running?
If the Power Meter disconnects, does Intervals switch to HR load? Or does it continue with what was available at the start of the activity (and assumes power = 0)? If it switches to HR load and you have bad GPS signal, this might accumulate quite some load. When you get back on the bike and wake up the Power meter, I suppose it switches back to Power load?

Off course, the best thing to do is to pause your activity on longer stops to avoid the hassle of having to cut a part of your activity.

13

For power load:

If there is a moving stream (data from Strava) then load is counted for all the time that you are moving according to the moving stream.

Otherwise (no moving stream) all time counts except for recording gaps (e.g. auto or manual pauses) bigger than 2 x the median time between samples for the stream.

14

Thanks David, I wonder what other assumptions is Trainer Road doing.

I tried the workout builder exercise and found a smaller variation, but still some (94 vs 87)

The command was: - 123m 188W , power is less important than getting the torque

188W was my. normalized Power and the moving time was 2:03… driving the 94 Load. Because I had a lower avg power, do you think the load is being affected?

There is one more scenario that might be overlooked:

  • Not moving from the GPS perspective (maybe due to tree canopy messing up with the signal, I know this happens on my area where the “auto pause” feature is triggered even on 5 KPH climbs) but the Power entry does exist… would that be accounted for?

Thanks again for this amazing tool and support.

LeTabiq