Increasing power duration curve

It seems that the sampling rates/algorithm used to create the power duration curve at the end of the workout is buggy - looking at the picture below, it shows my power increasing at some points, which doesn’t actually make sense for a power duration curve.

I just completed a zwift ride where I was following a workout my coach gave me, but was on the “just ride” mode on zwift. The workout had two main sets of intervals - 10 x 30 seconds hard, 15 seconds easy (I can share the specifics if that would help). I suspect that the power sampling rates for different durations lined up oddly with with the time intervals - if you look closely, there are 2 sets of 10 “peaks” grouped together as bumps here on the power duration curve, corresponding perfectly to the intervals I did.

Not sure if this is worth spending the time for zwift engineers to fix, as it’s probably non-trivial, but at the very least it’s very interesting and definitely not the intended behavior.

Mine never seems right either so I just ignore and check it later on Strava or ZwiftPower if I’m that interested for a particular ride.

It’s completely normal. Power can go up on a power-duration curve, and it will most likely do if you’re doing intervals. This is because sometimes, by moving right on the curve, you can include a portion of a new high-intensity interval that couldn’t be included before.

For example, let say you do 20 minutes 300w, then 5 at 100, then again 20 at 300. Max power for 20 minutes will be 300w. Max power for 25 minutes will be 260 ((20x300+5x100)/25 since you will always have those 5 minutes at 100w in any 25’ interval). Max power for 45 minutes will be 278w ((40x300+5x100)/45).