NP & math question

[SolarEnergy]

Hi

I'm currently writing an xl app to help me planning sessions/blocks/weeks and such.

I might have a question or two once in a while.

Right now I have 3 questions :

1) How should I account for passive recovery periods? I was thinking of maybe allocating "number of minutes off" * 10w or an other low wattage value". The idea is to have a clear picture of the NP, IF (including rest) and TSS of a workout as a whole.

2) What would be the impact of ignoring these off bike recovery periods?

3) Given an FTP of 220. Given the following workout : 3x20 at 100% FTP followed by 10min off (after each 20).
The fact of allocating (a fake) 10w ap for the rest duration lowers the overall NP (which isn't bad), lowers the IF (of course) but has a exagerated impact on TSS

I think I got the maths all wrong. Because by allocating only 10w during the rest period I endup with a TSS of 122, by allocating 100w during the rest period I end up with a TSS of 123. Everything under 100w doesn't have a significant impact on TSS. But I find that 30min "off" shouldn't raise the TSS from 100 (when not considering the rest) to 120.

God (and maybe Frenchyge) knows how bad I am with maths so please "dum" your explanations down for me.

 

[Old Junker]

Hi

I'm currently writing an xl app to help me planning sessions/blocks/weeks and such.

I might have a question or two once in a while.

Right now I have 3 questions :

1) How should I account for passive recovery periods? I was thinking of maybe allocating "number of minutes off" * 10w or an other low wattage value". The idea is to have a clear picture of the NP, IF (including rest) and TSS of a workout as a whole.

2) What would be the impact of ignoring these off bike recovery periods?

3) Given an FTP of 220. Given the following workout : 3x20 at 100% FTP followed by 10min off (after each 20).
The fact of allocating (a fake) 10w ap for the rest duration lowers the overall NP (which isn't bad), lowers the IF (of course) but has a exagerated impact on TSS

I think I got the maths all wrong. Because by allocating only 10w during the rest period I endup with a TSS of 122, by allocating 100w during the rest period I end up with a TSS of 123. Everything under 100w doesn't have a significant impact on TSS. But I find that 30min "off" shouldn't raise the TSS from 100 (when not considering the rest) to 120.

God (and maybe Frenchyge) knows how bad I am with maths so please "dum" your explanations down for me.

i think it works like this anything at ft=1x the minutes divided by 60 so 1x20/60 (1 third of an hour)= .333, so for every 20 minutes you get a tss of .333, since you have 3 of them you get .999.. or 1.00 then just for easy math you take the recovery at 110 watts (or half of the FT) for half an hour so you get .50 for half the FT and multiply by .5 for half an hour which equals .25 so the total is 1 plus .25= 1.25, so your 100 would be .455x.5=.228 so you are right the 10 watts would be 10/220=.045 x.5=.023 total 1.023, you must have left out a decimal point.. must be all the mandatory french classes we had to take

 

[SolarEnergy]

i think it works like this anything at ft=1x the minutes divided by 60 so 1x20/60 (1 third of an hour)= .333, so for every 20 minutes you get a tss of .333, since you have 3 of them you get .999.. or 1.00 then just for easy math you take the recovery at 110 watts (or half of the FT) for half an hour so you get .50 for half the FT and multiply by .5 for half an hour which equals .25 so the total is 1 plus .25= 1.25, so your 100 would be .455x.5=.228 so you are right the 10 watts would be 10/220=.045 x.5=.023 total 1.023, you must have left out a decimal point.. must be all the mandatory french classes we had to take

I think I understand what you mean (after having read it few times )

When I calculate a workout as
1x30min at 10w, I don't even get 1 tss.

When I calculate a workout as
1x60min at FTP, I get 100 tss (of course)

When I calculate a workout as
1x60 at FTP followed by 30min at 10w, I really get 123 TSS.

It must be my something somewhere in my custom power related functions.

Thanks anyway !

 

[rmur17]

I think I understand what you mean (after having read it few times )

When I calculate a workout as
1x30min at 10w, I don't even get 1 tss.

When I calculate a workout as
1x60min at FTP, I get 100 tss (of course)

When I calculate a workout as
1x60 at FTP followed by 30min at 10w, I really get 123 TSS.

It must be my something somewhere in my custom power related functions.

Thanks anyway !

Hi there,
Well I haven't checked your math as I think you've stumbled across what Frank Holt of the Wattage list coined "RATSS" or Recovery Augmented TSS.

Bottom line is that "yes" long rest/downhill intervals at low power relative to FT seem to artificially boost TSS relative to the "core" workout. In fact, I personally don't take TSS across the entire workout but simply add the segments to avoid this dodgy area. Whether or not it's a big deal depends on how much really low power work is interspersed with "real" work.

So for 1x60 @FTP followed by 30min at low power, I add TSS for both segments ... which results in something just a shade over 100 pts ... as I believe it should.

rmur

 

[SolarEnergy]

Hi there,
Well I haven't checked your math as I think you've stumbled across what Frank Holt of the Wattage list coined "RATSS" or Recovery Augmented TSS.

No I haven't stumbled across Frank's statement about RATSS. Well actually I might have (I don't know what stumble means )

But why is it like that. Is it because of the number of seconds having too much weight? (s * W * ...). Not that I complain anyway.

I noticed that even the normalized power doesn't get much affected (negatively) by active rest periods. But I actually find it pretty cool. Because frankly a short while riding easy won't change the fact that you got tired from riding hard just before.

So in other words you're recommending that I don't include passive (and not even active) recovery periods in the overall cacluation of my workouts. Since I had to work a bit harder to be able to include them in, that shouldn't be a problem to not consider them.

Thanks !

 

[SolarEnergy]

On the other hand, even recovery time is done at an intensity that is higher than when sitting on a couch.

I mean, I expect glycogen utilisation to be higher when resting between 2 or 3 20min intervals. But how much higher?

Would it be closer from zero or from what TSS returns?

 

[RapDaddyo]

How should I account for passive recovery periods? I was thinking of maybe allocating "number of minutes off" * 10w or an other low wattage value". The idea is to have a clear picture of the NP, IF (including rest) and TSS of a workout as a whole.

Why not use zero watts for the recovery duration off the bike?

 

[tk_bike]

Are you getting active recovery (keeping riding at a low intensity) mixed up with passive recovery (off the bike sat down)???

 

[Old Junker]

Hi

I'm currently writing an xl app to help me planning sessions/blocks/weeks and such./QUOTE]

now that i think about it if its for planning sessions why are you using rest/recovery, if i look at my entire year my tss was only over 2000 in one month, if i take the recovery parts out my tss would be way higher, you now have given me something to think about for planning my next year

 

[rmur17]

No I haven't stumbled across Frank's statement about RATSS. Well actually I might have (I don't know what stumble means )

But why is it like that. Is it because of the number of seconds having too much weight? (s * W * ...). Not that I complain anyway.

I noticed that even the normalized power doesn't get much affected (negatively) by active rest periods. But I actually find it pretty cool. Because frankly a short while riding easy won't change the fact that you got tired from riding hard just before.

So in other words you're recommending that I don't include passive (and not even active) recovery periods in the overall cacluation of my workouts. Since I had to work a bit harder to be able to include them in, that shouldn't be a problem to not consider them.

Thanks !

well I'm not a coach or ex. phys. and can only claim to have looked at the math in detail but given those caveats I'd recommend the 'segment' approach that I mentioned. There's little doubt that the TSS math (which I won't try to demonstrate here) causes long low-power segments which are recorded by the PM (i.e. while the PM is active) to artificially inflate TSS scores to a certain degree. This is exactly what you've seen when you manually calculate TSS for your sample workouts.

Especially when using the more advanced applications (PMC), I think it's important to not exaggerate the TSS load.

I suspect for 90% of typical training time (L1-L4) that there's very little difference between the base and my approach. But if doing very hard efforts/intervals with considerable rest intervals (for example) then I believe the rest period augmentation effect should be discounted.

I try not to nitpick or criticize but this is ONE issue on which I feel I'm on firm ground.

rmur

 

[SolarEnergy]

Are you getting active recovery (keeping riding at a low intensity) mixed up with passive recovery (off the bike sat down)???

I'm not doing all this for me (worthless ).

But while you're asking... I take passive recovery only during winter turbo trainer session.

Why not use zero watts for the recovery duration off the bike?

That didn't have any impact. using the aforementioned example : 3x20 followed by 10min at 0w = 122 TSS. 2x20 followed by 10min at 10w = 122 TSS.

if i look at my entire year my tss was only over 2000 in one month, if i take the recovery parts out my tss would be way higher

Are you sure it wouldn't be the opposite? That is leaving the recovery parts artificially boosts TSS rolling average values?

I try not to nitpick or criticize but this is ONE issue on which I feel I'm on firm ground.

It's understood that we're not nitpicking or criticizing here.
I take a good note of your recommendation. You're probably right on all counts. But I can't help but wondering about how "off" TSS really is though.

I'll try to setup my app to report using both methods.

Take our example with a 220 FTP rider. To get 20 TSS in 30min (more or less the value allocated for passive recovery periods), this rider would need to ride at 140w for 30min. I wonder how the ressource usage during such a workout (30min at 140w) compares with the ressource usage while resting between the intervals.

 

[frenchyge]

(I don't know what stumble means )

Stumble litterally means to trip (while walking) over something unexpected in the path like a rock or root. Figuratively, it means to discover something unexpectedly (ie, stumble upon something).

But why is it like that.

The 4th-power averaging used to determine NP means that the high power efforts are weighted much more heavily in the NP calculation. When comparing work vs. rest power levels, the drastic difference in power levels means that a 300w interval is 810,000x more heavily weighted than a 10w recovery period in determining the overall NP of the workout.

The NP is calculated as above, which is then divided by FTP, squared and multiplied by the duration to determine TSS. The duration is equally affected by work and rest period lengths, which means that in the extreme cases that you're referring to, TSS can be thought of as essentially taking the power level from the work period times the duration of the work + rest periods. Not that there's anything wrong in the TSS calculation, it's just designed to model the stresses of a more typical cycling workout (as opposed to drastic difference in power with short work periods and long rest periods).

Rmur17 has already offered up some good suggestions, and I'd add that it isn't really an issue unless one prefers to take unusually long rests in between intervals.

 

[rmur17]

Rmur17 has already offered up some good suggestions, and I'd add that it isn't really an issue unless one prefers to take unusually long rests in between intervals.

yeah I was 'reared' on the Lemond (Koechli) school in the late 80's ... lots of 2-4min intervals but with quite long recovery periods (2 to 4 times the work period).

When I first started L5 work with the power-meter, NP, TSS etc., my pattern was say 5min WI (work interval) and 10min or more RI (rest interval). That was a prescription for TSS inflation ... and probably a prescription for wasting workout time as well

For you engineering types if you consider an interval 'work' to 'work + rest' duratiion to be the duty-cycle, then inflation is a function of (1/DC). As DC approaches unity for steady-state work or very short rest periods, inflation tends towards zero. As DC approaches 0.5 (for a pure square-wave workout) or 0.25, inflation subtantially increases.

rmur

edit: Had to check in my old file but the function is square-root, i.e. inflation=SQRT(1/DC). 1.41 for 50% duty-cycle and 2.00 for 25% duty cycle. Okay I'll shut up now

 

[SolarEnergy]

Hi

I (too) am working on few analytical tools for my own use.

Powertap generated datafiles, processed through cyclingpeaks and then exported in csv, show a sampling rate of 1.26sec. That means that every record shows what was the power every 1.26 second.

In converting such a file in 1sec steps, should I be using a linear interpolation? Would you have suggestions better than that?

Thanks a lot.

 

[rmur17]

Hi

I (too) am working on few analytical tools for my own use.

Powertap generated datafiles, processed through cyclingpeaks and then exported in csv, show a sampling rate of 1.26sec. That means that every record shows what was the power every 1.26 second.

In converting such a file in 1sec steps, should I be using a linear interpolation? Would you have suggestions better than that?

Thanks a lot.

can I ask why you need to convert to a 1-sec time base? Why not let the sample time dictate it?

 

[SolarEnergy]

can I ask why you need to convert to a 1-sec time base?

Of course you can. I'm very bad at math (I think I mentionned it earlier). But still, I want to do my best calculating NP related stuff. So I want to stick to the 30s rolling average.

Right now my approach is to process a datafile using a recursive function, creating nodes as I find them, to later return those nodes probably in a tree view. The logic I'm writing at this moment (while being aware that I might turn my back later on and start with something different), involves navigating from this second to the next, then to the next, and then back and forth...

Why not let the sample time dictate it?

I don't understand your question.

 

[RapDaddyo]

Of course you can. I'm very bad at math (I think I mentionned it earlier). But still, I want to do my best calculating NP related stuff. So I want to stick to the 30s rolling average.

Right now my approach is to process a datafile using a recursive function, creating nodes as I find them, to later return those nodes probably in a tree view. The logic I'm writing at this moment (while being aware that I might turn my back later on and start with something different), involves navigating from this second to the next, then to the next, and then back and forth...

I don't understand your question.

I would think it would be easier to deal with the actual PT time line (1.26s) than to write an interpolation routine just to get 1s data. The only thing you really need to do is to convert your desired time gap to 1.26s observations. The other advantage of leaving the data in its native form is that you have a perfect audit trail back to the original ride file. It's maddening to try and find the same segment in two files, one in 1s durations and the other in 1.26s durations. I think it should be the sobriety test administered by highway patrolmen. Of course, we'd get a high ratio of false positives, but what the hell?

FWIW, one option I would not recommend is to export a PT ride file as an SRM .txt file. While conceptually appealing (it is converted to 1sec durations), I have recently been working with CP and WKO+ export files of various formats and I have found that if you export a ride file from power meter x in power meter y format that it (the export file) can have erroneous data.

 

[SolarEnergy]

I would think it would be easier to deal with the actual PT time line (1.26s) than to write an interpolation routine just to get 1s data. The only thing you really need to do is to convert your desired time gap to 1.26s observations.

I think I understand what you mean. Let me still try my original idea, as most of my logic is already done using this metric 1s time line. But I'll keep your warning in mind.

In fact, I might need to roll this back though, probably because of this here..

I have recently been working with CP and WKO+ export files of various formats and I have found that if you export a ride file from power meter x in power meter y format that it (the export file) can have erroneous data.

I didn't think about this idea. It does seems appealing indeed. But thinking about this, I believe that I might create the same issue in trying to convert the time line myself. If not, then it may mean that CyclingPeaks developers would possibly benefit from using a linear interpolation themselve in converting from PT to SRM.

That's the first thing I will evaluate Rap. I'll convert then navigate through the whole file and see if I can compute the average power within an acceptable error margin range.

 

[RapDaddyo]

I think I understand what you mean. Let me still try my original idea, as most of my logic is already done using this metric 1s time line. But I'll keep your warning in mind.

In fact, I might need to roll this back though, probably because of this here..
I didn't think about this idea. It does seems appealing indeed. But thinking about this, I believe that I might create the same issue in trying to convert the time line myself.

That's the first thing I will evaluate Rap. I'll convert then navigate through the whole file and see if I can compute the average power within an acceptable error margin range.

And, if you're working with PT data, don't overlook blank watts (as distinct from zero watts). I haven't see blank watts with SRM files, but they are regularly a part of PT files. Isn't data fun?

P.S., You can find the blanks yourself in WKO+. Just go down the ride file an observation at a time. Blank watts show as "---" whereas zero watts are displayed as "0".

 

[SolarEnergy]

And, if you're working with PT data, don't overlook blank watts (as distinct from zero watts). I haven't see blank watts with SRM files, but they are regularly a part of PT files. Isn't data fun?

I've seen some too. What did you do with them?

My understanding was that they too would require some sort of interpolation. Or let's put it this way. Blank HR are very commun right? I sometimes get a lot of them but yet, they don't seem to affect the calculated average negatively. This suggests that they're considered as Null (or absence of) value as opposed to a Zero value.