Still waiting on an ultra light rider update

Racing
45

I agree that this is definitely an issue which needs addressing and it has been raised time and again on this forum @DavidP

46

This is the problem in every iTT or TTT…riders that are +90kg and under 175cm are not AERO but in Zwift they are super FAST and win every TT

The speed are accurate if you take average cyclist into account
example 1: 50kg and 165cm
example 2: 60kg and 170cm
example 3: 70kg and 180cm
example 4: 80kg and +185cm

The formula FAILs when you have:
Example 1: 90kg and -185cm
Example 2: 80kg and -180cm
Example 3: 70kg and -170cm

The height advantage in speed calculation is to big…because the smaller and heavier riders are more AERO in Zwift but not IRL.

3 Likes
47

I am actually quite positive about how much Zwift in general is listening to their customer base, but I also think there’s some room for improvement regarding tweaking the height and weight component in the model. Perhaps this picture will help to visualize.

image

Now first add some height and reconsider the loss of aero, and then add some weight and reconsider the loss of aero.

And apart from all that:
37kg at 6,0 w/kg for an hour… it’s hard to believe that’s possible. I think the Race organizer should DQ until proven this was legit. At TFC Mad Monday we had a rider A+ with Pogacar values recently and we simply DQ-ed this person. I think race organizers can have a role in this well to improve the Zwift racing!

1 Like
48

Thanks Dejan. Is this linear? What happens for a 60 kg and 160 cm rider; does this fail too?

Also, what about the ultra low weight examples (< 45 kg). Are there any heights at which sub-45 kg speeds are realistic?

49

Any rider exceeding 6w/kg for 20mins indoors is exceptional. Doing this at 37kgs is not something I’ve ever seen.

True, but regardless of the CdA model estimation using height/weight, the real issue is, was the user lying about his weight? :thinking:

I’d like to see an automated system that flagged performances such as this and immediately contacted the rider to address the validity

I’d like to see that too. I hope we can get to that, but until then we still have a lot of other problems to take care of.

50

I’ll take a look at this when I can. But can’t make promises at the moment.

6 Likes
51

Just to be clear. Do you mean only the issue with heavier riders on a TT-bike or are you thinking about looking at this from more sides, like the light rider issue also?

52

You have to bare in mind that shorter riders have smaller bikes, and the A component of CdA often would be smaller if riders of all shapes and sizes adopted the same position, which Zwift (probably sensibly) assumes.

In reality chunkier riders typically can’t hold more aero positions (see Kristian Blummenfelt), but should Zwift really be modelling that?

I agree that it seems like at the very extremities (children) the model tends to break down, but then also, as David alludes to, the chances are their power isn’t accurate anyway, which is probably 9/10ths of the problem.

You could actually probably argue that the complete flip of this argument is true. I have done a 2-up TT with a friend a few times who is 60kg to my 73kg and about 170cm to my 182cm. My raw watt FTP is higher by around 20w and our positions are similar, although undoubtedly my area is larger with a larger frame. He absolutely buries me. I am doing close to threshold when I’m on his wheel, but he is doing easy watts behind my larger frame.

I’d leave this alone personally, I don’t think the ‘problem’ has been clearly documented, it’s all anecdotal, and you could end up solving the wrong thing.

53

The aerodynamically messy parts (wheels, legs moving) are pretty much the same for all riders. Longer legs are longer for sure, but probably do not contribute to the total wind resistance of the rider+bike system in a linear proportion to the added surface area, and the effect of a longer upper body in the aero position is (almost?) negligible.

1 Like
54

Someone like George Goodwin would disagree when he came 3rd at Challenge Daytona on a pan-flat race track putting out well under 300w compared to the 360w+ of his competitors. Smaller bike, smaller frame.

55

Since Zwift isn’t real life, surely it would be “easy” for someone at Zwift to study the weight vs height vs speed relationships with bots to see where the model breaks down?

I’m surprised Eric hasn’t looked into this?

1 Like
56

Found an interesting video about this: https://youtu.be/0NHEuSUiDIM
Why frontal area does not scale linearly with mass etc.

57
1 Like
58

He is only testing the weights in the normal range (75kg and 82kg).

Conclusion: reducing your body weight by 1kg will save you ~9 seconds over an hour-long flat race effort on Zwift.

I’m not sure if this can be seen as proof that Zwift works like IRL.
It would be interesting to see tests with weight from 30-100kg.

59

“I’d leave this alone personally, I don’t think the ‘problem’ has been clearly documented, it’s all anecdotal, and you could end up solving the wrong thing.”

The problem is clearly documented, and it is not just a problem for very small riders. Look at the comments from the link you posted.

Also look at the first post I did on this thread: CdA dependency on height issue

In summary: Comparing the 170cm guy with the 185cm guy there is a difference in speed of 2.3%, running the numbers in a bike speed calculator gives a difference in speed of 0.77% with hands in drops and 0.86% difference with hands in hoods.

1 Like
60

I’m still not seeing it. What is the issue (as in the thing that is ‘wrong’, with evidence) and what should the fix be? This forum is full of anecdotes that things are wrong, usually such anecdotes are directly contradicted by other anecdotes in other threads. There may well be an issue with rider speed when applied to outliers (e.g. children) but I haven’t even seen that categorically demonstrated.

Edit for your edit: Does the bike speed calculator take in to account the fact that the 170cm guy would be on a smaller and lighter frame? Do you happen to be quite tall, by any chance?

1 Like
61

I have just seen others reporting abnormalities for years, but it is hard to get proof.

If Zwift (or preferably an independent third party) could show us that their algorithms work for the whole range of weight and height for their users, they would not get these questions and it would be an advantage over other competing platforms.

1 Like
62

You could do it yourself? Run Zwift offline, ride in ERG, change your height and weight and record the impact. I’d rather ZHQ were pointing their limited resource towards activities with a tangible benefit, rather than:

  • Forum user X thinks Y is a problem.

  • Does he have data to demonstrate it so we know exactly what to fix?

  • No.

  • OK could you spend hours running tests to see if that anecdote holds true?

  • Ummm, no. I’m going to do something useful.

2 Likes
64

Yes, of course, I could do it myself :slight_smile:

Zwift wants to be seen as a serious racing platform. They should spend some time proving that it really is.

65

Why, when none of the competition are threatening that space? What a waste of time. RGT don’t model physical differences at all!