Thursday, December 28, 2006

TSS vs. kJ

Recently I've noticed a few discussions in the on-line forums about using the CPU read-out on power meters showing work performed (usually expressed as kilojoules) as a proxy for cumulative TSS. Many have noticed a roughly linear relationship appears to exist between the two measures.

This comes from a desire by some to know their cumulative TSS while riding.

But is the kJ read-out a valid proxy for TSS?

For those that use the newer ergomo power meter, they already have this feature built into their CPUs, so they have no need for using kJ as a proxy for TSS. However, for the rest of us, unless you do the same type of ride every time you head out the door or get on the trainer, it is unlikely that this linear relationship will hold true.


The Practical Example

While the mathematics shows this relationship to be a spurious one, there's nothing like real data to emphasise a point.

The above chart is a plot of kJ vs. TSS for 161 rides so far this season.

While the relationship is roughly linear, there are major differences in work performed for rides of a similar TSS, even rides with TSS around 100-130, the energy spread is greater than 1,000kJ or up to three times the difference for the same TSS!

So I won't be using the kJ display on my power meter as a proxy for TSS.

Besides, I'm better at estimating what TSS to expect per hour for certain types of rides. While you can do this by estimating the Intensity Factor (IF) of a given ride (or sections of ride), I simply know from experience what to expect since I have probably already done similar rides as a part of my regular training and racing.

The Maths

Here's a quote from Andy Coggan on the matter to explain the maths:

"The relationship between TSS and total work not only varies with the individual, but also with the nature of the workout. That is evident by looking at the respective formulae:

Total work = duration x average power

TSS = duration x (normalized power/functional threshold power)^2

Thus, during long workouts the ratio of TSS to total work will tend to be lower, since normalized power and average power will tend to be closer together. Conversely, during shorter, higher intensity workouts (especially, e.g., with trackie style "go hard, puke, go home" level 6 intervals) the ratio of TSS to total work will tend to be higher, since normalized power will be much higher than average power. Of course, if practically all of your workouts are "middle of the road", then these differences won't be apparent and you may find a fairly high correlation between total work and TSS. As should be evident from the formulae given above, however, the two are not equivalent, i.e., such a correlation is simply spurious."

So until TSS, IF etc are incorporated into the CPUs of more brands of power meter, I guess most of us will just need to get to know our own data better....

2 comments:

dl6kbg said...

Hi Alex ! Thanks for this post. I did the same Calculation for my 48 Rides i did on my indoor trainer (tacx i-magic). i noticed, that there is a relationship on short (1 hour) intensiv rides. But maybe have a look here: http://powercycling.blogspot.com/2006/12/tss-training-stress-score-vs-in.html

Regards, Oliver

wilf said...

TSS and kJ are not the same thing, but if you construct a PMC-type chart using kJ (same smoothing constants) it will give you useful info. For example, during base period or during a taper, the kJ chart will (should) tend to "diverge" from the TSS chart. Or if you feel tired but TSS TSB is +10 but based on kJ it is -20, it probably gives you a hint about why you're feeling tired.

TSS is part of the story, but so is kJ. Different stories but both can be useful.