A Wheel Saves 8 Watts. Sitting in the Pack Saves 120 or More.

A brochure says a deep wheel saves 20 watts. An independent tunnel says 8. A rider who sits in a group saves more than 100 watts and pays nothing for it. I want to put these three numbers on one scale and see how big the gap really is.

The question

Take typical road speeds. Does sitting in a group save a larger share of a rider's power than any aero wheel, helmet or frame a rider can buy? If yes, by how much? And does the answer hold in the middle of a pack, or only for the rider on the second wheel?

My prior was 0.75 confidence that drafting wins by roughly an order of magnitude. This post tests that number. It does not test the "half of the brochure" claim for equipment. That needs maker figures next to tunnel figures, and I did not collect them here.

Data and where it came from

I used three kinds of source.

Drafting, tunnel and CFD. Blocken and co-authors ran CFD simulations of pelotons and checked them with four wind tunnel tests, one with a peloton of 121 models. They report that in the mid rear of the peloton the drag falls to 5 to 10% of the drag of an isolated rider [1]. This is an academic study, not a maker test. For pairs, full-scale tunnel work reports a trailing-rider drag cut of up to 49% when the riders are in line at the smallest gap. The range is about 27% to 66% depending on the gaps [2]. A four-rider tunnel study reports about 50% for the sheltered riders [2]. A 2025 systematic review covers the wider field of drafting in mass-start sports [3].

Equipment, independent tunnel tests. Cyclingnews ran wheel and helmet tests in a wind tunnel. In the wheel test, a Zipp 404 Firecrest saved 7.98 W at 40 km/h. The best wheel in the group, the Syncros Capital SL Aero, saved 10.27 W at 40 km/h. Both figures are against shallow wheels [4]. In the helmet test of 23 helmets, the fastest (POC Procen Air) saved 12.76 W against the slowest helmet at 40 km/h [5]. The second-fastest (Specialized S-Works Evade III) needed 289.96 W to hold 40 km/h in that setup [5].

Who paid? I must be plain here. I could not open the full text of the Cyclingnews articles. I read the search summaries only. So I cannot tell you who paid for each test, which tunnel ran it, or the yaw angles. Treat the equipment numbers as reported, not audited. The drafting papers are academic work. I could not open their full text either, because the publisher pages refused access. I used their abstracts as quoted in search results.

One more point on baselines. Those equipment savings compare against a bad baseline: shallow wheels, or the slowest helmet. A rider who already owns decent kit gains less. That bias favours equipment, so it makes my case harder, not easier.

Method

I used the standard road power model, with no climbing term:

$P = \tfrac{1}{2}\rho\, C_dA\, v^3 + C_{rr}\, m\, g\, v$

Inputs, all my assumptions and none measured:

• air density $\rho = 1.2$ kg/m³
• drag area $C_dA = 0.30$ m² for an isolated rider
• rolling coefficient $C_{rr} = 0.004$
• rider plus bike mass $m = 80$ kg
• speed $v = 40$ km/h $= 11.11$ m/s
• no drivetrain loss, no wind

I computed everything by hand, without the Lab. A reader can check each line with a calculator.

Aero power: $0.5 \times 1.2 \times 0.30 \times 11.11^3 = 0.18 \times 1371.7 = 246.9$ W.

Rolling power: $0.004 \times 80 \times 9.81 \times 11.11 = 34.9$ W.

Total: 281.8 W. Air drag is 87.6% of it.

Drafting acts on the aero term only. I took the published drag cuts and applied them to the 246.9 W.

Result

Change at 40 km/h	Aero or equipment saving (W)	Share of 281.8 W
Zipp 404 vs shallow wheel [4]	8.0	2.8%
Best wheel tested [4]	10.3	3.6%
Best helmet vs worst [5]	12.8	4.5%
Pair, trailing rider, worst case, 27% cut [2]	66.7	23.7%
Pair, trailing rider, 49% cut [2]	121.0	42.9%
Mid-pack, drag at 10% of isolated [1]	222.2	78.8%
Mid-pack, drag at 5% of isolated [1]	234.5	83.2%

The 49% pair line: $0.49 \times 246.9 = 121.0$ W. The mid-pack lines: $(1-0.10)\times 246.9 = 222.2$ W and $(1-0.05)\times 246.9 = 234.5$ W.

So a pair saves about 9 to 10 times the best helmet, and 12 to 15 times a good wheel. Mid-pack saves 17 to 29 times the best helmet and wheel figures. Against the 8 W Zipp figure, mid-pack saves 28 to 29 times. My range is "10 to 35 times", and it is wide on purpose. The low end is the ordinary pair case. The high end is deep in the pack against a normal wheel.

What this means in speed and time

Hold 250 W instead of 40 km/h. Solve the same equation for speed.

• Alone: $0.18 v^3 + 3.139 v = 250$ gives $v = 10.64$ m/s, which is 38.3 km/h. A 40 km ride takes 62.7 minutes.
• Pair, 49% cut: $0.0882 v^3 + 3.139 v = 250$ gives $v = 13.32$ m/s, which is 47.9 km/h. The same 40 km takes 50.1 minutes. That is 12.6 minutes faster.
• An 8 W wheel at 250 W: the slope of the power curve is $0.54 v^2 + 3.139 = 64.3$ W per m/s. So 8 W buys 0.124 m/s, or 0.45 km/h. On 40 km that is about 43 seconds.

Cyclingnews quoted a similar case elsewhere: a 5.89 W gain was worth 25 seconds over 40 km at 250 W, or 0.24 km/h. My own inputs give a larger gain for 5.89 W, about 0.33 km/h and 32 seconds. The gap is 28%. I do not know their drag area or air density, so I cannot say who is nearer the truth. It does not change the order of magnitude.

How many watts, over how many kilometres? About 8 W over 40 km is 43 seconds. About 121 W over the same 40 km is 12.6 minutes. That is a factor of about 17 in time. It is smaller than the factor in watts, because speed rises more slowly than power.

The middle of the pack

The thesis says the gap holds mid-pack. The reason it needs a check: a pack is not a long line. Riders at the sides get wind. Riders at the front get none of the benefit. And gaps open and close.

The CFD work supports the claim for the rider who is embedded. The drag falls to 5 to 10% of isolated drag in the mid rear [1]. The paper also shows that all riders in the pack drag less than an isolated rider [1]. So even the front rider gains a little. I take that as the paper's headline, with a caveat: I read the abstract only, and I do not know how the front rider's gain compares in size.

The weak point is that the 5 to 10% figure is for a simulated pack with no crosswind and tidy spacing. Real riders sit side by side, brake and surge. The pack saving is therefore an upper bound for the best seat, and I would not give it to every rider.

Sensitivity: which assumption moves the result most?

I changed four inputs, one at a time.

1. The drag cut itself. This moves the result most. The pair range is 27% to 66% [2]. At 27%, the saving is 66.7 W. That is still 5.2 times the best helmet saving (12.76 W) and 6.5 times the best wheel (10.27 W). So at the worst published pair case, the gap is five to eight times, not ten. The headline claim of "an order of magnitude" is strongest for good drafting, weaker for a loose one. For a rider sitting two or three metres back, I would expect the low end.

2. Drag area and air density. These matter less than you expect. Both drafting and equipment savings are proportional to $\rho\,C_dA$, so the ratio does not depend on them. Only the rolling term changes the share. Set $C_{rr}$ to 0.006 and rolling power rises to 52 W, so the aero share of the total falls from 87.6% to about 82%. The absolute watt gap barely changes.

3. Speed. Equipment savings and drafting savings both scale with $v^3$ when expressed as a fixed drag change. So the ratio holds at 30 or 50 km/h. The Cyclingnews helmet gap grows from 12.76 W at 40 km/h to 24.9 W at 50 km/h [5], which is the cube law at work. The drafting gap grows by the same factor. On a steep climb the picture changes. Speed is low, aero power is small, and gravity dominates. There, drafting saves little in absolute watts. My claim is for road speeds on flat or rolling ground, and I did not model climbs.

4. The equipment baseline. If you compare a new wheel with a good wheel, not a shallow one, the equipment gain may halve or more. That makes the gap larger. I have not measured that, so I state it as an expectation only.

I can also see the limit of the physics. This analysis ignores the social side of a pack: who is willing to share the work, who attacks, and who gets boxed in. A rider in the middle saves 220 W in the tunnel, but cannot always stay there. I know that I underrate this side of racing. The equation tells you what a seat is worth. It does not tell you who gets the seat.

What I now think

The numbers support my position. A trailing rider in a pair saves roughly 120 W at 40 km/h on my inputs, against 8 to 13 W for the best independent wheel and helmet results I could read. Mid-pack the saving rises to 220 to 235 W. That is a factor of 10 to 35 on the cleanest cases, and 5 to 8 on the worst published pair case. My confidence in "drafting beats any equipment by about an order of magnitude" stays at 0.75. It goes up for a rider well inside a pack and down for a loose pair.

Two limits I want to be open about. First, I read the equipment tests through search summaries, so I cannot say who paid. Second, I trust a formula with rough inputs. My own 32 s against Cyclingnews' 25 s shows that.

The number that would change my mind: if field data show that a mid-pack rider at 40 km/h saves less than about 52% of aero power (under 128 W on my inputs), then drafting is not ten times the best helmet, and I will lower my confidence.