Drafting plays a huge role in road cycling and thus was built into Zwift’s physics engine from the early days. Of course, Zwifters (especially racers) have debated the “correctness” of Zwift’s draft since the early days as well!
Perfecting a virtual draft is no mean feat, especially when you consider that there is no way to steer or brake on a trainer (not yet, anyway). Because of these restrictions, Zwift’s draft has a bit of “stickiness” to it, so if you approach a rider from behind, and your speed is close to theirs, you will get “stuck” in their draft. This allows riders to get into the draft and stay there more easily, but can be a bit annoying when you’re trying to come around another rider.
There are, perhaps, some improvements which could be made to the draft. In the past I’ve advocated for Zwift to let us “feel” the draft by changing resistance based on our draft status. And Zwift themselves implemented “double draft mode” back in 2018 – an optional event mode which increases the draft effect substantially so it’s more in line with outdoor levels. (It’s still used in some events to help groups stay together, but racers have found that it makes races less exciting since it raises overall pack speeds, making it more difficult for attacks to stick.)
We had already done some testing of the draft effect, but only with two riders. Our initial draft tests showed a wattage savings of approximately 25% when drafting behind a single rider. We also looked at draft savings up climbs and down descents. But what everyone really wants to know is how the draft behaves in larger groups! How much power am I saving if I’m riding 5th wheel, instead of 2nd? What’s the most efficient power plan for a team time trial?
Here’s a screenshot I grabbed from a recent GCN race broadcast. It shows a special secret “draft viewing mode” – fascinating! But just how powerful is that draft?
It’s much harder to test the group draft, because you have to have multiple Zwift accounts set up, each on separate machines, and be able to control each rider individually on a flat course without interference from outdoor riders. Zwift certainly has ways to do this internally, but for the rest of us mortals it means cobbling together a lot of Zwift accounts, computers, and ANT sticks… then figuring out a way for just those accounts to ride by themselves!
That last bit was the sticky part for us – there was no way to hold a “private” group ride, until Zwift released “Meetup-Only View” this week. Now we can arrange a Meetup on any Zwift course, and only have those we invite visible. Perfect! Let’s test the group draft!
This initial group test set out to answer just two questions:
- Is it more efficient to ride in a standard TTT format (single-file line), or for riders to simply hold high power and “churn” continually on the front like we see in most Zwift races?
- If it’s more efficient to ride single-file, what sort of power savings occurs in each successive rider in a typical TTT formation?
All of the test riders were set to 183cm height, 75kg weight, and rode Zwift Carbon bikes with 32mm Zwift wheels.
Interesting side-note: Zwift’s draft effect actually takes rider height and weight into account – similar to outdoors! So a taller rider will create a stronger draft than a shorter rider, and a heavier rider a stronger draft than a light one. Zwift computes an estimate of a rider’s frontal area and uses this to compute the wind resistance they encounter, as well as the draft “wake” they produce.
Tests were done in “Meetup-Only View” on Watopia’s Tempus Fugit route because it’s the flattest on Zwift, and it has a timed section (Fuego Flats Reverse, 4.4 miles long) which could be used to measure the speeds of each test formation.
All of the tests were done with four riders. Because I ran out of ANT sticks and computers!
Tests and Results
- All riders @ 300W
Segment time 10:14.8
Speed: 41.34 kph (25.64 mph)
- All four riders continually “churned” on the front, alternating between poking their nose into the wind, then getting slowed so another rider could come around to the front. This is what you typically see at the front of a non-TT Zwift race.
The second test had the lead rider holding 300W, with the other three riders in single file behind, holding the minimum wattage possible to stay in formation. This is what you would see in an outdoor team time trial:
- Rider 1 @ 300W, Rider 2 @231W, Rider 3 @ 205W, Rider 4 @ 199W
Segment time: 10:47.7
Speed: 39.24 kph (24.34 mph)
- The “minimum wattages” stated for riders 2-4 on this test and other tests below should be considered approximations, as it is impossible to figure out the precise wattage required to hold formation due to Zwift’s dynamic physics engine and very small undulations in terrain, even on Fuego Flats.
- Riders received power savings of 23%, 32%, and 34%. So the more riders you’re behind, the bigger the draft effect.
- In a TTT situation with all riders taking equal pulls on the front at these wattages, each rider would average 234W.
- Test 2’s segment time was 23.1 seconds slower than Test 1’s, despite riders holding no higher than 300W in both tests. This may seem odd at first, but it’s a well-known result of the way Zwift’s physics engine works at the front of packs. In the “churn”, riders are speeding up while in the draft, then shooting ahead into the wind, only to be slowed and have another rider shoot past them. This little speed boost accounts for a significant time difference, as we see here!
Next, I tried to guess what wattage the lead rider would need to hold in order to beat the time set by the churning group of 300W monsters from test 1. I settled on 320 watts – here are the results:
- Rider 1 @ 320W, Rider 2 @244W, Rider 3 @ 225W, Rider 4 @ 209W
Segment time: 10:22.4
Speed: 39.83 kph (25.33 mph)
- Riders received power savings of 24%, 30%, and 35% (2nd, 3rd, and 4th rider respectively). This lines up with the power savings seen in other tests.
- In a TTT situation with all riders taking equal pulls on the front at these wattages, each rider would average 250W.
- This group didn’t quite beat the 300W churn, but it was close.
This time around I wanted to make sure we beat the 300W churn group. So I bumped the lead rider up to 350W:
- Rider 1 @ 350W, Rider 2 @270W, Rider 3 @ 244W, Rider 4 @ 235W
Segment time: 10:01.2
Speed: 42.28 kph (26.23 mph)
- Riders received power savings of 23%, 30%, and 33% (2nd, 3rd, and 4th rider respectively). This lines up with the power savings seen in other tests.
- In a TTT situation with all riders taking equal pulls on the front at these wattages, each rider would average 275W. This is crucial to understand: that even with Zwift’s “speed churning” from test 1, the four riders in this test significantly beat test 1’s time by riding efficiently in single file formation.
For the fifth test, I bumped power up to around the max we might see in a TTT. I also wanted to see how the power savings (in terms of %) changed with higher power on the front:
- Rider 1 @ 400W, Rider 2 @306W, Rider 3 @ 277W, Rider 4 @ 261W
Segment time: 9:31.9
Speed: 44.44 kph (27.57 mph)
- Riders received power savings of 23.5%, 31%, and 35% (2nd, 3rd, and 4th rider respectively). This lines up with the power savings seen in other tests.
- In a TTT situation with all riders taking equal pulls on the front at these wattages, each rider would average 311W. So just 11 watts higher than all riders held in test 1: but much, much faster due to smart use of drafting efficiency!
For the final test, I reduced power down to the 3 w/kg line, where many D and C riders may ride a TTT effort:
- Rider 1 @ 225W, Rider 2 @173W, Rider 3 @ 156W, Rider 4 @ 144W
Segment time: 11:57.9
Speed: 35.4 kph (21.96 mph)
- Riders received power savings of 23%, 31%, and 36% (2nd, 3rd, and 4th rider respectively). This lines up with the power savings seen in other tests.
- In a TTT situation with all riders taking equal pulls on the front at these wattages, each rider would average 175W.
Power savings in a single file line of riders is quite consistent in terms of a percentage, even when the front rider’s wattage varying from 225-400W in the tests. Savings for the 2nd rider ranged from 23-24%, 3rd rider 30-32%, and 4th rider 33-36%.
Clearly, the single-file TTT formation is the most efficient way to travel in a group on Zwift. Racers competing in TTT competitions would be smart to practice riding in formation with a focus on holding that minimum wattage needed to stay on the wheel of the rider ahead. This is more challenging on Zwift compared to outdoors due to delayed response times and a lack of steering/braking, but riders can become very proficient with practice. And that will pay off in a big way in terms of TTT race results!
Got comments or questions? Share below!
Perfect day to release the info from this test! The WTRL Team Time Trial teams are all over this with teams of 4-8. http://www.wtrl.racing
This is fascinating and goes against general consensus for TTTs. This could change everything! Thanks for the tests
Really?! What’s the general consensus?
See reply below – specific to zwift TTTs
I’m curious to know in what way you think this goes against consensus, and are you speaking of Zwift TTT’s or IRL?
Specific to zwift – that a tight blob moves faster than a line (where IRL a line is faster). Check out the top teams on wtrl. Mind you, there are very few 4 man TTTs. It will be interesting to know the best formation for 8 man.
So when does rider weight come into effect…
It’s ALWAYS in effect.
I thought that theoretically, weight isn’t as much of a factor on the flats, and instead it’s actual wattage.
With that said, a heavier rider has more surface area, and therefore more drag, which means that they’d have to put out more power to overcome the wind resistance.
Or am I way off?
The heavier rider has less surface area relative to their weight. So it’s the lighter rider who is more disadvantaged assuming same w/kg.
I’m really curious about the details of how you do all of your testing. Do you generate fake ANT+ data to get consistent power outputs?
Pretty much supports what we have learning in the WTRL TTT series, though the churn effect like from test 1 was an interesting tidbit. We have also found that keeping the first 3-4 in line helps but after that in groups of 7-8 you can have the riders bunch up a bit to good effect, and it’s less likely a rider gets gaped off the back
I have also seen some strange draft effects with regards to front vs back. When I’m riding just behind the front rider(s) in a bigger field, it feels easier than if I am riding at the back of the field. It should have been the opposite. Has anyone else experienced this behavior or have some explanation? One explanation could off course be that I feel like I have a chance when I’m just behind the front for a few seconds and gets a boost from that 🙂
Great test, thanks ! You mention “delayed response times”, do you see that with your ANT+ simulators too ? I was under the impression it was worse on the cheaper HTs like my Vortex vs the DDs.
There are multiple links in the chain. Better trainers are more responsive. But it’s not the same as turning a pedal and feeling your tires speed up like outdoors.
There is a slight modification test for some time when you want to come back to this. Have a lead rider or two and then have the rest of the group stay bunched up and in the draft behind them. In Zwift, it is hard to hold that pace line has been mentioned. For the groups I have rode with in the WTRL TTT, we have gone to this formation and it at least helps us both recover and get the next lead rider out front. There is no doubt it is not optimal. It might not give the same… Read more »
Interesting test. One of the things I’ve notices is that there is a propensity to get “stuck” and grab the tire of a slower rider the group is passing. The only way to completely avoid this is to significantly bump your power output when the group is passing. This seems relatively new in Zwift and is a negative in my opinion. Any thoughts on that? Also, for me it seems like it is easier near the front, but not in the front, which seems counter to the figure in your presentation. Maybe this is because the groups are often much… Read more »
It’s always been that way, but more noticeable now that there are more zwift users.
Michael: “Often, in big group rides it feels faster to ride just ahead of the beacon than just behind it…. Not sure I understand what is going on there.”
I have also experienced this phenomenon. Maybe this is something TTT teams can take advantage of? One or two in the front and the rest just behind taking turns at the front. I guess this is pretty hard to simulate then.
Thanks for another great set of data!
Have you analyzed the effect of distance you are behind a rider on draft effect. Zwift has that visual prompt 5m, 4m etc, is there no draft effect beyond 5m? Is the draft effect at 1m greater than at 5m (I am assuming so).
Also, when in a blob I sometimes get bumped to the side of the blob and out of the draft by other riders – any tips on how to avoid this?
This is excellent! Well done!
1) I get it that your resources are limited but can you estimate drag for a lollipop formation in a larger team? 1-2-Blob is much easier to manage.
2) How far back can you fall before breaking the line draft effect?
What makes zwift exciting is how realistic it is compared to riding in the street.Drafting is important specially in headwind and crosswind situations..
Super interesting. Thanks.
Great test! Thank you! However, your data doesn’t quite align with my experience. In a race, in the middle of a fast moving blob of 30 or more riders, if I drop power below 250 w my avatar seems to move steadily toward the back of the blob and eventually off the back. According to your data, even if the riders at the front are holding 350 w, I should be able to stick with them at 235 w (and I’m much further back than 4th wheel so I’d expect that even lower power should be sufficient to stay with… Read more »
Could it be that you’re exceptionally tall and/or heavy? That would explain it.
My other explanation is simply that it’s pretty much impossible to see what’s actually happening at the front of a busy race. You can see ever-changing w/kg, but you’re not even sure who is on the front, or what they weight (and therefore what their actual wattage is).
I’m close to your test riders in size and weight. It is difficult to see what’s happening at the front of the blob, and even if I can see those riders, I can only see w/kg, not watts. Maybe the riders at the front are taking turns pulling at 450 w. But if the blob is moving at ~42 km/h, then that’s in line with your tests. 350 w at the front, rider 4 has to hold 235 w to keep up. There are 20 riders between my avatar and the front, and yet I’m slowly drifting back while holding… Read more »
I experience this too on a regular basis, I take a pull at front for example, back off maybe 10% wattage to try and sit in and ideally slowly move toward back or better yet middle, at least in my mind, and it seems like I literally get shot to the back of the back and sometimes have to go 4 w/kg to get back on, get nowhere near 30% draft or whatever.
All your ‘experience’ data is talking about dropping backwards through the pack. If the pack is going 42km/hr and you are dropping back at 40km/hr, you have to re-accelerate back up to 42km/hr or more to not get popped off the back. The Testriders in test2-test6 in these experiments are all holding the same speed as the lead rider. They don’t have to re-accelerate. In my experience, it’s much easier to hold that 235w or slightly above it while you are moving up through a pack doing 300w because you’ve already accelerated. Never rest while falling back in Zwift. Rest… Read more »
Great tip Eric – Resting/sitting up in the back of a blob and suddenly losing contact is an experience to be avoided!
Interesting….. “Rest while moving up.”
I’ve tried that, but can’t make it work consistently. How do you deal with the Stickyness? In order to move up through the pack, I seem to have to put in a massive effort to break the stick to the rider Zwift has stuck me to. And then I seem to just get stuck to another rider two places ahead.
Moving backwards through the pack, the stickyness doesn’t seem to work.
Kind of exactly the opposite of what you’d want to happen.
your comment about practicing single file formation made me question whether all four formations consistently appeared the same on all four monitors. How did they compare to each other?
They were pretty similar. Front to back location is pretty consistent across monitors in Zwift, side to side is what changes. But if you’re keeping your wattage dialed in, you should be spaced out enough that Zwift keeps you in a single-file formation.
Tests 1 and 2 are Interesting for racing in general and kind of what I’ve thought was a big problem with racing. Breakaways are really hard to establish and keep because the pack don’t have to even try and chase and the churning does most of the work to bring you back even when you are putting out a lot more power. This is especially pronounced on descents. In a real pack if the front rider decides they’ve had enough of pulling and no one comes around the pack speed stalls.
Thanks Eric. Did you test riders taking turns in front but the rest being in a blob rather than a line?
That’s really interesting Eric thanks a lot.
Any chance of a height effect in game at weight n and a calculation of CDa? It would be interesting to see how much the effect is.
Interesting reading, thanks!
Would it be possible to test draft effect for different bikes? I think that’d be fascinating.
Not sure if its placebo effect – I feel that the Tron provides the best draft in race conditions. Even though the Venge/Disc combo is faster on flat courses etc as per your testing, I think I feel that I work less and find it easier to hold the blob in races when I use the Tron rather than a traditional road bike/wheel combo. Might be in my head – not sure.
Rolling around watopia on my TT bike doing workouts, I get the same feeling when looking at who draftee and what wkg they need to hold. Maybe it’s just because the trim stands out, but it does seem they can get some pretty low wkg and stay in the draft.
Can’t edit, but you get my point – tron.
IRL, when you are in front and someone is drafting behind you, you increase your aerodynamics resulting in higher speed for the same power output (seems weird, but the bike behind you act as a tail reducing your drag coefficient or Cx). Do you see such phenomena on Zwift? (speed of a solo rider at 300w versus speed of the same rider at 300w with a pack following him?)
This is great work, Eric, and provides really useful data. I don’t know why Zwift don’t publish their rider speed algorithm. There’s nothing particularly difficult about modelling the various power input, air resistance, gravity, rolling resistance, drive chain and draft parameters, but the way they have chosen to implement them is of interest to the community. It’d also save you having to reverse engineer the black box, which you do remarkably well!
Thanks Eric, this is fun! Now I know why I feel so guilty as a 110Kg rider on the flats toodling along at 1.8w/kg while the front is working their asses off! And double drafts are kind of like being sucked along in a vacuum. However, a bit of virtual team coaching input. I think your Test 5 riders are slacking off. If Test 1 represents the maximum energy capacity of the team, then I see about 777kJ available. In Test 5 they only expended 711kJ so coasting along comfortably at about 90% of capacity. I think you can lash… Read more »
Great work on the site and all the testing & results – this is awesome. I do have one concern here: On tests #2,3,4,5 I feel like: the tests are performed with no position swapping, where each rider holds the stated power throughout. conclusion is extrapolated to “riders taking equal pulls” (and therefore rotating) to an ‘average power’ number that is lower. Personally, I am not convinced that actually rotating the riders would result to the split times that were obtained in those tests with static positions. Also, brings the question about what would be the best way to perform… Read more »
This is spectacular work. I wonder where you got the ANT+ testers. One thing is that average power doesn’t necessarily capture the effort of time-variable efforts. Coggan’s normalized power metric is the 4th root of the average of the sum of watts to the 4th power, after time-averaging. So neglecting the time-averaging (which yields an upper bound on NP), the powers you reported for the “near match”: rider 1 @ 320W, Rider 2 @244W, Rider 3 @ 225W, Rider 4 @ 209W AP = 249.5 W NP = 260.8 W Of course with each rider doing 300W constant, NP =… Read more »
No rider will be holding exactly constant watts when at the front, chances are even the pull powers are going to vary in time due to terrain changes and so on. So what you’re proposing is taking the normalized power of average pull powers which probably is not going to be correct anyway, unless you took normalized powers of the normalized pull powers. I would not want to use Coggan’s metric here and just keep things simple.
Additional question for future research – what about a TT bike leading the group of 4 (or 8 for WTRL TTT)? What’s the potential speed increase/power savings for the other riders from a TT bike leading for the first 10-15 minutes of a race at a strong clip.
The team finally got around to trying this – after a new member with real world experience of TTT made us (!). It took some practice to prevent surging, and switching front riders smoothly is a massive challenge but your experience with bots bears out with a team of 7 real riders too. It’s hard to be precise because of team members changes but I believe we got a 10-15% improvement this week from running a line instead of a loosely managed blob. I attached the link to the Strava record below.
Is there more sticky draft impact on a light rider than heavy, when the lead rider finishes a pull and drifts back through the line but still holding significant watts? (the best TTT teams seem to have the ex-front rider drop watts right down to or near zero, and go to the back fast, so minimise imposing their sticky draft impact on the rest of their team).
So tough to test… I don’t really know. Taller/heavier riders do give a stronger draft.
Bit late to the party, but in a 6man TTT, what savings do you expect the 5th and 6th rider in a line formation to have? Question in reference to Zwift model.
Based on the results above, I would guess 5th and 6th would save around 35%.
Hi Eric, I’d love to explore your side note to help me further complicate my current TTT spreadsheet… Do you know if height differences between riders change the power savings when in the draft? Taking rider 1 from test 2 (75kg, 183cm, 300W) I presume his solo effort would be the same as the team result in test 2. We also know a second identical rider would benefit from a 23% power saving from being in the draft of rider 1. What if the 2nd rider was 153cm? From your other article on height we could estimate this rider only… Read more »
Great work. Keeping the single file in Zwift is in my experience almost impossible. So it would be interesting to see the results of a group actually changing leader every 30 sec. It would probably be from more of a bunch than a single file then. Also because it takes effort to brake the stickiness when overtaking.
Any chance you could rerun this test with the new pack dynamics?
it would be great if we could have this analysis with the new pack dynamics… keep the formation on a TTT seems harder now
@Eric Schlange Bump
Some have mentioned there a definite push BACKWARDS st some point that’s very unlike real world. In a blob it strongly feels like closer to the front is way better. But don’t pop through the front and if you do keep pedaling (or best, stop for 1 second then resume) or you get rocketed out the back way than than IRL. It seems that the sticky draft sort of cumulatively has a backwards negative force that negates part of the draft effect at least in a blob. So different than IRL. As you pointed out in a sweet spot you… Read more »
The lighter riders seem to have been penalised with the new pack dynamics, I’m 66kg pushing higher w/kg than other came cat riders who are heavier but still not able to stay with the group!? Was that intentional or just a bi product of the new dynamics? My stats have gone off a cliff edge since the change and I’m not the only light rider to suffer this?
65kg here and ditto. I no longer enter races since I’ll be easily dropped by a heavier rider putting less wkg factor as my ratio has to be a good deal higher. Hate to say it but reverse weight doping might be the best way around it.
Just wondering what distance you tried to use to draft or if you changed from overlapping the back wheel and dropping 5m back ??
I tried to stay within 1m of the rider ahead. No overlapping, and no dropping too far off. It wasn’t perfect, but it was pretty close.