It’s a question that has popped up in Zwift discussions many times over the years: do larger riders put out a bigger draft on Zwift? Conversely, is it harder to draft behind smaller riders?
It’s a logical question since anyone who has ridden road bikes outside knows there’s a big difference between drafting behind a larger rider or a small one.
It’s also a strategic question, because the answer could affect how you plan your next team time trial on Zwift.
Zwift calculates a CdA for each rider based on their height and weight – a key part of accurately simulating in-game speed. Years ago, I had a conversation with someone within Zwift (I forget who) who told me that the size/strength of the draft does indeed change based on the rider’s height and weight.
So I had always assumed that was the case. But I had also assumed that the difference was so small that I wouldn’t be able to measure it with the rudimentary methods I had at my disposal.
Then recently the question popped up again – twice in the same day, in fact – and I started thinking once about how I could test it. That’s when I realized Sauce for Zwift could help me make it happen.
Test Methodology
To test if the size of the front rider affects the draft received by the second rider, I devised the following simple test:
- Create a Meetup on Tempus Fugit, the flattest route in Zwift, and invite my two bots to the party. (A Meetup was used instead of a Club Event because this allowed me to dynamically change the height and weight of the front rider while remaining in the event.) No other riders were visible on course, and both bots were on the same basic bike: Zwift Carbon with 32mm Carbon wheels.
- Set the front rider to a very small size (152 cm, or 5′, and just 50kg in weight). Set that rider’s power so they are cruising as close to 40 kph on the flats as possible.
- Set my drafting bot to our standard dimensions (183 cm, or 6′, and 75kg in weight) then tweak this bot’s power until it is sitting on the first rider’s wheel at the lowest wattage possible.
I stayed in the Meetup (it was a 100-lap Meetup, after all) for quite a while, modifying the front rider’s dimensions and then tweaking both bots’ power until they were drafting nicely at 40 kph. Sometimes I changed that rider’s height, sometimes just the weight, sometimes both. I tested plenty of permutations until I was satisfied with the result.
With draft strength potentially differing by only a small amount, accuracy was crucial if my results were going to be useful at all. Zwift’s HUD shows no decimal points for speed, so I employed Sauce for Zwift in order to get at least one decimal place, which helped immensely!
I also used Sauce to show each rider’s current stats, and I kept the draft gauge up just out of curiosity.
Here’s a quick video summarizing what I saw:
Test Results
| Front Rider Height, Weight | Front Rider Power | Rear Rider Power | Speed +/- 0.1 kph | ||||||
| 152cm, 150kg | 378W | 225W | 40 kph | ||||||
| 152cm, 50kg | 212W | 225W | 40 kph | ||||||
| 198cm, 50kg | 254W | 225W | 40 kph | ||||||
| 198cm, 150kg | 437W | 225W | 40 kph | ||||||
| 198cm, 75kg | 306W | 225W | 40 kph | ||||||
| 152cm, 75kg | 267W | 225W | 40 kph |
Conclusions
The conclusion is clear: the strength of the draft does not change based on the size of the front rider.
I reached out to internal contacts at Zwift to confirm this conclusion, because I was surprised with the result after what I’d been told in the past. The contacts confirmed that this is in fact the case: height and weight are used to compute individual rider CdA, but that CdA does not affect the size/strength of the rider’s draft “shadow” in any way.
Not yet, at least. It sounds like Zwift may be looking into making some changes in this area. But of course, there’s nothing set in stone.
Zwift could simply change the overall draft strength based on rider dimensions – but that isn’t truly how it works outdoors, is it? Outdoors, what changes is the size of the draft. A taller rider casts a taller draft shadow. A wider rider casts a larger one. Could Zwift get so detailed that they’re actually changing the size of the draft shadow based on the rider’s size?
We’ll have to wait and see. I can only promise to report on any changes when they roll out!
Your Thoughts
Does this conclusion surprise you, or confirm what you already thought was true? Would you like Zwift to change the draft so it more accurately emulates the outdoor experience, or do you prefer how it works today? Share your thoughts below!
I’m not sure I understand your conclusions mainly related to how the outdoor draft works. While I might agree that the draft itself doesn’t change based on rider size/weight the effect that draft has on the rider behind certainly does change based upon the front riders size/weight. If you are riding behind a really large rider your power required to stay in the draft is less. Isn’t the effect on the trailing rider what’s important? This probably would have a big impact on Zwift Racing. In the Cats I race it seems almost always bigger guys almost always do well. Doesn’t happen that way outside. These bigger guys usually have very crappy positions on their bikes outdoors. Smaller people usually have better positions and they always benefit greatly from riding behind a larger rider. You seem to have demonstrated that Zwift favors larger riders over smaller riders if they both receive the same draft benefit. Which I think is what I’m seeing. I don’t think a 6′ 75kg rider outdoors drafting a very small person can go 40kph on 225 watts in the draft of the tiny person.
I think you misinterpreted the test. The conclusion is that even though drafting behind a larger rider IRL obviously provides a larger and stronger draft window to the following rider, it does not make any difference in Zwift at this time.
I think we are in agreement. I was simply trying to point out that the current way zwift is doing the draft greatly favors larger riders. Take it to an extreme. Very very small lead rider. So small they would IRL make almost no draft. In Zwift the 2nd guy gets the same draft regardless. 2nd very large guy should be putting out watts almost like they were on the front because the front rider is so small. It also seems that the 2nd rider regardless of how small they are had to put out 225 w to stay in the draft. That doesn’t favor small riders who should be putting out way less watts because they are both very small and in the draft. How many 90+kg recreational riders have you seen IRL have a “Decent” position on the bike? I’m trying to understand how Zwift does the draft if regardless of front riders size then 2nd “Standard” rider only has to do 225w. How do they arrive at 225w for the 2nd standard size rider? Is it a % of what he would do on the front at that same speed?
Drafting is so much better in real life than in Zwift.
How do you define “better” in this context?
Much more noticeable advantage in drafting in real life than in Zwift. Especially in the big bunches you can easily sit and spin in high speeds on the road, not so in Zwift where much more wattage is needed to stay in a bunch.
So that that also meant that the current Pack Dynamics still needs to be improved as staying in the bunch is still needs some effort just to keep up than benefit from the group draft?
Its suppose to be easier to be in a group than having to push big wattages to keep up, something I experienced last night in a group ride of Cs.
Correct, Dave.
In the past, Zwift has said they purposely don’t model the IRL ease of riding in a large pack, so people can still get a decent workout while riding together.
Or another way to look at it: if they made it as easy as IRL in big packs, Zwifters might need find ways to ride alone in order to get a good workout in in 30-60 minutes. Driving people away from the social thing is the opposite of what Zwift wants to do.
If that was important, why couldn’t they allow adjusting the draft effect individually? (i.e. I can change a setting that indicates that I only want 50% of the draft effect so I have to work harder in the group). Pretty simple to divide a formula by 1/2, no?
Thanks @Eric Schlange ! Do you think you could run this test the other way round, i.e. have a set height and weight for the first rider and permutate the drafting rider? As a tall and light rider I feel I’m often doing more W/kg than the people I’m drafting. But it might just be in my head…
We already know that height and weight changes your CdA in Zwift – those test results are already published here. I’m not sure what further testing would accomplish?
I was basically looking for CdA of riders of different shapes – in the draft. Thanks to your test we now know that irrespective of rider size in front, your standard rider requires 225W to do 40k in the draft. How about taller and shorter riders? Is everything there as it is in clean air, just offset by x? Or are there discontinuities? Has this been previously published?
CdA in the draft is the same as CdA without a draft. Just like outdoors.
The basic drag force equation is:
D=(1/2 Rho V^2) x (CdA).
The CdA term is the drag area, the (1/2 Rho V^2) term is the dynamic pressure (pressure due to velocity). In this simple drag model, the most straightforward way to implement a draft effect is to include a draft influence coefficient, call it i, in the equation. So the equation would become:
D= (i) x (1/2 Rho V^2) x (CdA).
The draft influence coefficient would take a value between zero and 1. The drag force is then directly reduced by the magnitude of the influence coefficient, i. For instance, if i=0.85, then drafting provides a 15% reduction in drag force vs. free stream riding. Because of the form of the equation, it doesn’t matter if you view this influence coefficient as reducing the CdA, or reducing the dynamic pressure, because math. But it makes a bit more physical sense to view it as a reduction in dynamic pressure on the following rider.
This makes sense to me – that the draft you feel is only based on things you know or can see on screen – i.e. your own stats and how many riders are in front of you.
Because of the way Zwift’s 5 avatars work, you can’t see or know the height or weight of other riders. So if the amount of draft you receive was impacted by the rider in front, you could sometimes get a greater draft from following a “small” avatar (if they were very tall, slightly thin) than a “large” avatar (if they were very short, slightly heavy).
What about the reverse? Does a riders size impact their ability to receive draft?
A rider’s weight and height determines their CdA, which is the main determiner of how many watts they need to push at a given speed in the draft (or without).
Maybe that’s a good idea for the next test, though. How many watts do different-sized riders save in the draft? Might be interesting.
Personally I have noticed that during double draft events it seems harder than would be expected to stay with the group. At 205cm I’m definitely a head above the rest… but in the real world I put the hurt on people on the flats… not the other way around (which I experience in swift)
Well the height disadvantage is also clearly visible in these tests, a 75kg rider at 152cm has to do 40W less (so 0.5W/kg) than a 198 cm rider with the same weight. Although there will definitely be a lower CdA for the smaller rider in real life, especially in TT events this wouldn’t be this big of s difference.
This also depends on the flexibilty of the rider. I’m a tall, slim and flexible rider that irl is more aero than my most of my shorter teammates.
But than again, Zwift is a game, not real live 🙂
Zwift’s height penalties are pretty astounding. All things being equal, a rider the same weight but different height will absolutely be more aero but the raw watt difference more than makes up for it… especially in a TT.
I have also wondered what the draft benefit of the bike is within the draft. I understand by yourself and nobody in front of you the bike will have a cda. Does that change based on the draft and if so by what percentage. If a race had a significant hill would it be beneficial to ride a lightweight bike and use the draft as an advantage as well as then having the bike weight beneficial on the climb. Or would I have to put out so much more power to keep the draft that the light bike would not be worth the cda penalty and hill advantage.
A simple test could be how much power does it take to hold the draft based on different bikes and/or wheels.
CdA doesn’t change based on drafting status.
I’m not sure I can test with enough accuracy to see a difference between two different road bikes in the draft, to be honest. Because it’s only a margin of a few watts, and these drafting tests probably have around 10W of wiggle room.
Eric,
Thanks so much for the response. Good to know that the bike would not make enough difference within the draft.
Thank you @Eric Schlange for this and all the other analyses you do. Helps satisfy my math nerdiness! 🤓
You and me both!
I believe Zwift uses a consistent method to calculate drafting effects, whether there’s just one rider or a group of ten ahead. Considering the size of each rider in front of you might not be practical, as the overall drafting benefit is an estimate anyway.
However, it could be possible to adjust the draft calculation method based on the number of riders in front. When there are only a few riders ahead, considering individual rider sizes might provide more accurate results. On the other hand, for larger groups, a simplified estimation that treats all riders equally could be more practical. Ultimately, the choice depends on the trade-off between accuracy and computational complexity.
As someone with a smaller stature who races in real life, this is a huge miss. It’s a massive tactical advantage for me to sit behind bigger riders or to go off in a break with bigger riders sitting behind me eating wind. Smaller riders typically have smaller power numbers and need advantages like this.
Small riders are being given the unrealistic advantage in zwift that they don’t have to jostle for position with much heavier riders, nor ever deal with crosswinds. Count your blessings and quitcher whinging.
I think it’s useful to know this, but as a smaller rider I still think it’s a disadvantage to other riders to be stuck in a group with me. Namely, a lighter rider has to work at a much higher wkg to be going the speeds you have posted there (4.24wkg for the 50-kg rider vs 2.5 for the 150-kg rider). It is rare if riding with a group of men that I am even able to get to the front, let alone pull at a noticeable wattage where they aren’t all just passing me again. So yeah, maybe that smaller rider will give you a draft, but only if they are really quite a bit stronger than everyone else in the pack!
Even if they don’t appreciate your lack of contribution on the long, fast, level miles, I bet those men really appreciate you slowing down and waiting for them on the climbs, though. You do slow down and wait on the climbs, don’t you? What?! NO?!? Darn you, Wee Folk.
This makes sense to me as a way to keep the network traffic & compute requirements lean. In ye olden dayes, Zwift had the client compute where a rider sat in a bunch, and the client was already receiving a lot of data around rider-counts, jerseys, avatars, bikes, wheels, etc, but while a lot of that was just visual data, taking rider sizes into account for computing draft models could have been too much for devices that already barely met the support requirements… or my deep-country internet provider’s 2Mbps throughput.
Now that all of this is computed server-side it’s probably more tenable, but would still put a significant tax on the cloud infrastructure (i.e., a lot more $$$). As it isn’t currently part of the Pack Dynamics, that would mean a whole new cycle of development, and we all know how much folks love changes to PD.
surprising and helpful info for sure!! thank you
To my mind It’s the size of the draft shadow that should be changing dramatically. At a given speed, a bigger rider in front of you/the pack creates a bigger ‘hole’ in which you can find shelter and be sucked along.
If you’re on a jetski drying to draft on the water behind another jetski it’s considerably different than if you’re trying to draft behind a yacht, but the difference is more about the size of the wake.
The type of shelter, whether you want to assume your pack is almost like a wing with laminar flow around you/the hole, or whether you want to model there’s a bunch of eddies being shedded favourably between riders, and where those eddies start and grow is not necessarily otherwise massively dependent on the size of the rider is. It’s probably just as dependent on the conditions and things like smoothed/textured zones (particularly given that these are now designed for this effect)
(P.S. One reason it might feel much easier outside behind a bigger rider is because it’s much easier to get into a nearly ideal part of the wake.)
Anyone drafting me would.ger a massive boost….. But I’m fat and old….
Great info Eric! Anyone who races or rides hard outside knows that Zwift is not a 100% replication of IRL riding. To create a platform that works for the majority there has to be compromises and this data set clearly illustrates that. As a bigger rider I can honestly say that I have never put out more power than a smaller rider on a flat road while in their draft unless there was a crosswind…that’s because 212w isn’t getting them to 40kph…try 260-275w. That is a compromise Zwift has to make so that small riders can contribute to the ride on flatter roads. In the bunch on a flat course all riders out of the wind will be within 1% of each other with regards to power output. So the issue on Zwift isn’t actually the draft but rather the power output of smaller riders on the front. Now this cuts both ways as the power smaller riders need to make going downhill is unrealistic…because bigger riders can’t do 300w downhill. With all that being said I love Zwift…it isn’t perfect…it isn’t real life…but it is close enough!
Zwift is becoming a game for nerds not for sporting.
You must be new around here…
I don’t understand why a taller rider at same weight has to produce almost 20% more power at 40km/h. I think that is far from reality. Tall, slim riders can be very aerodynamic when they make themselves small. At this point something in the calculation should definitely be changed.
Height, weight, bulk, mass, volume, density are physical properties. “Make themselves small” is nonsense akin to belief in witchcraft. That’s why.
Hey Eric, besides the draft topic this is also a (quite small but still) cool dataset for “How many W/kg do differently sized riders need for the same speed“. We already know that heavier riders are faster at constant height and W/kg. The same goes for smaller riders at constant weight and W/kg. The dataset shows exactly that, but the observed W/kg range for the same speed is crazy.
In the test the W/kg for riding at 40 kph on flat ground with one rider drafting range from 2.52 to 5.08 or Cat D to Cat A. Even without the 2 completely unrealistic combinations (50kg at 198cm and 150kg at 152cm) it is 2.91 to 4.24 W/kg. Based on actual categories it’s still mid Cat C to upper Cat B. Surely this is a very small dataset (with another purpose) and only one riding situation. It would be cool to see more such data. However, even so it supports that the actual categories are far away from good.
One could try to create a power based score for a better categorization with enough “same speed with different sizes“ data. (Enough data is likely much more than you could produce Eric.) Obviously, such a score has to include weight, height, and power (short term as well as long term). However, equal chances for riders of all sizes on all types of courses won’t work.
Ok better than the actual categories, which intentionally discriminate people in the no chances valley around 60kg, isn’t a huge challenge. That could even work without any testing, but better is not same as good …
Given in real life you get very little draft of smaller riders and a lot of larger riders it’s non sensical to standardise, but we had also noticed through zwift that it doesn’t change..
Cool test. But my takeaway is something unrelated. It’s how much weight affects speed so dramatically, particularly on completely flat road. 437w for 150kg? 215w for 50kg? Maybe if the gradient was 2-3% this may be an accurate difference, but that seems steep. Particularly silly when you consider that the 50kg rider in FRONT of the 75kg rider is doing LESS watts- not because the 75kg rider in the draft is receiving small draft, but simply because the 50kg rider weighs less? Damn. Seems like lighter riders have a significant advantage on Zwift, especially on flats.
By your own numbers, the orc is doing less than 3wkg and the hobbit is doing 4.3wkg.
That’s a funny kind of ‘significant advantage’ if, like me, your FTP is 3.5wkg.
So my popularity will be cratering when the news gets out. I’m 6’5″ 😉
Huge kudos to you Eric, for taking the trouble to investigate that scientifically
Thanks for your super useful site
I’d say they have more important fish to fry in this area before they start getting that detailed with draft. Height is drastically too important to CdA. Tall riders get way too harsh a penalty and short riders get far too much benefit.
Does the strength of the draft shadow increase with more riders out front? If 2,3,4 people were riding side by side does the person drafting behind them have to produce less and less watts to keep pace?
I’ve taken the liberty of doing a little bit of deeper analysis based on your data. Here are my results.
For approximately steady-state testing on mostly flat ground, such as you have done here, it is possible to calculate reasonable values for CdA using the steady-state cycling speed vs. power equation with drag and rolling resistance terms. I have done that calculation and the estimated values of CdA for your lead rider cases are shown below.
height weight power Calculated CdA (ft^2)
152 50 212 2.35
198 50 254 2.88
152 75 267 2.90
198 75 306 3.39
152 150 378 3.88
198 150 437 4.62
From the drafting rider speed and power, it is possible to calculate the “effective” CdA for the drafting rider. That calculation yields a value of effective CdA of the drafting rider in the draft of about 2.37 ft^2. (You can treat riding in the draft as an effective reduction in the CdA due to the form of the draft force equation).
Since you didn’t run the height/weight combination of the drafting rider in the lead position, the free stream CdA of the drafting rider cannot be determined directly, as it was for the front position riders. However, it can be estimated reasonably well analytically. That value is about 3.31 ft^2 based on a free stream power at 40kph of about 300 watts.
The difference between the free-stream CdA (3.31 ft^2) and the effective CdA in the draft (2.37 ft^2) for your 75kg, 183cm rider is about 28 percent. Your testing shows that the draft force on the 75kg, 183cm rider is reduced by 28 percent in the draft. The power savings (225 watts drafting vs. about 300 watts free stream) is about 25%.
The one thing it does confirm is that the effect of weight in Zwift, on the flat, is insanely inflated, it really shouldn’t require that much extra power, it’s not like a rider suddenly gets 3x wider.
Not really. If a rider becomes 3 times heavier, their volume increases by 3 times (assuming a constant human density). When you increase the volume of a specific 3 dimensional shape while keeping one of it’s dimensions (height) constant, the other dimensions increase by a factor of the dimension to the two thirds power. So you don’t become 3 times wider if your weight increases by a factor of three. You become 3 to the 2/3 power wider. 3^0.66 = 2. So a rider becomes 2 times wider (and 2 times thicker). Frontal area and surface area increase in the same proportion. The reason that viewing power required to maintain speed inflates the drag increase is because of the way that power varies with velocity due to the nature of drag. Power required to maintain a given speed increases with the cube of speed. As drag coefficient increases, the power required to maintain a certain steady state speed increases above and beyond the drag coefficient increase.
Eric, if you haven’t checked yet, please take a peek at the “Performance predictor” in Sauce fo Strava.
There is a capital “simulator” of drafting, its size, rotation, place in the group, etc.
example:
example 2:
example 3 (rotating group riders):
While I can still ride through the backs of folk, I would expect this nuance to be missing.
very interesting! Thanks!
I can swear, I could feel the difference when drafting behind a “small” cyclist… How powerful is our mind! 😀