by Matt Weaver

(Peter Ross was reasonably concerned with the influence of wind on HPV performance, and inquired my opinion on several points. The following is a slightly revised version of the e-mail response I sent Peter. Peter presented my response to Human Power for publication. Matt Weaver.)


"Stubby Wings"

Fully streamlined bicycle bodies, especially those with fairly long or tall trailing edges and low ground clearance, act much like half of a stubby wing turned on edge (the other "imaginary" half extends into the ground). As the ground clearance increases, the body's sensitivity to the wind decreases. But even with good ground clearance, if there is still an appreciable trailing edge the body will produce definite lift in a side wind. Typically it is not too efficient, but by virtue of high vehicle speeds and low body drag, the forward thrust is seldom negligible.

Estimated sailing characteristics

These lift characteristics can be well approximated for certain body shapes, and others can ultimately be plugged into a computer if coordinates are available (assuming the overall body flow is "well attached" - which can be checked by the computer, and which is valid for the top streamlined bikes, but not for many automobiles). I am familiar with both methods and have tools and know the math, though I've looked at side-wind effects primarily to compute the lift and "moment" forces on the body for stability and steering-geometry considerations (regarding the "moment" - most bodies like to torque away from the side wind, as if the lifting force is acting on an imaginary point beyond the front of the nose).

Some lift and drag relations

Some bodies are designed with low-drag "natural laminar flow" airfoil sections. These airfoils have low drag at zero flow incidence (i.e. no side wind). As the side wind increases beyond a certain point, the pressure distribution over the section changes such that the boundary layer on the downwind side of the airfoil becomes largely turbulent instead of extensively laminar. At that stage the drag of say a 50-60% natural-laminar-flow airfoil section will roughly double. The lower-drag part of the characteristic is sometimes referred to as the "low-drag bucket" on lift/drag plots of various airfoils. You can shape an airfoil to have a "wider" low-drag bucket (that is, it would cover a wider range of side winds) but the bucket becomes less deep (i.e., the low-drag portion is increased).

"Turbulent" as used here refers to the state of the thin blanket of "boundary-layer" (BL) air near the body surface. It is not the same thing as "separation" - which is what happens at the rear of blunt bodies such as most automobiles. A "laminar" BL has much lower shear or "skin friction" rubbing against the surface of a body than a comparable turbulent BL. A BL starts in a laminar state, and may irreversibly "trip" (a bit like a wave breaking on the beach) into a "turbulent" BL. The skin friction increases locally as much as ten-fold as the BL transitions from laminar to turbulent, and drag of identical-looking sections can vary as much as three-fold, depending on the overall state of the BL.

Lift of symmetric sections is independent of drag

Interestingly, regardless of whether the boundary layer is laminar or not, or what the airfoil section profile is (large nose-pointy tail, or pointy nose-shorter tail...), the lift coefficient of a symmetric airfoil section is largely a function of the incidence of the flow alone.

Lift, drag, and forward thrust

As the side wind increases with low-drag "natural laminar flow" sections, the lift keeps rising, but the forward thrust will reduce temporarily when the body drag jumps up, but typically any wind produces a net forward thrust for the body at all times regardless of whether or not the body drag has gone up (i.e. the body drag has increased, but the forward or "thrust" component of the lift has increased more than the drag and thus gives a net propulsion). Not considering changes in drag coefficient, the potential forward "thrust" from the wind goes roughly with the square of the side-wind speed.

Estimates of sailing characteristics of the "White Hawk"

I'll spare the summary of math details here because they go on for a page or so, and just give you numbers for a single case. The "White Hawk" as well as the "Tomahawk" look like fairly good vehicles for side winds with their low ground clearance and fully developed height and trailing edges. Other fast streamlined bicycles exhibit similar estimates: slightly greater sensitivity for the Gold Rush and Cheetah, and slightly less for the Varna Mephisto and Cutting Edge. Estimates for a 5-MPH (2.24 m/s) side wind at 50-MPH (22.35 m/s) vehicle speed are as follows based on a model from images of the "White Hawk",

  • 2.56 pounds (11.4 N) forward component of lift (thrust)
  • 1.84 pounds (8.18 N) induced drag (in the form of a vortex cone coming off the upper trailing edge)
  • net forward thrust of 0.72 pounds (3.2 N)
  • corresponding wind power slightly under 0.10 horsepower (75 W)
  • vehicle lean angle of about 7 degrees (fairly modest)
  • At 50 mph (22.35 m/s) and 0.50-HP rider power output, velocity increment would be about 3 MPH (1.3 m/s).

Course geometry and side winds

A linear "out and back" closed course is sensitive to wind direction; a circular course is not. Closed courses of varying eccentricity naturally fit somewhere in between. Giving respective weights, with 1.0 = perfect perpendicular side wind, the following holds:

  • linear with perfect perpendicular side wide = 1.0
  • linear with head/tailwind = <0.0 (you don't get back downwind all that you put in upwind)
  • circular = 0.45-0.50 (the math is interesting - I did it for the whole loop; this is the range where it ends up. There is a slight penalty for the "headwind/tailwind" region of the loop, else it would be basically 0.50) The principle is a little like that of a "Darrieus" or "egg-beater" wind turbine.
So, for a circular course, you get a little less than half of what you'd get perfectly aligned with the side wind at all times, which is pretty good considering the ideal linear course is not a likely scenario.

Considering the previous case listed above with an average 5-MPH (2.24-m/s) wind (regardless of direction) and a circular course, wind power would amount to about 0.045 horsepower (34 W), or about 1.5-MPH (0.67-m/s) increase in vehicle speed near 50 mph (22.35 m/s).

Wind and hour records

I hadn't considered the matter closely until you asked, nor did I even know there was an issue with Lar's record!

"5 to 10 MPH winds reasonably significant"

I would consider the above estimations reasonable - of about 1.5-MPH speed increase in a closed-loop hour run near 50 MPH with a 5-MPH side wind. For winds much higher than 10 mph, the vehicle might start to have a hard time, but I would consider increases in average speed as much as 3 MPH or a little more very likely before handling begins to deteriorate.

"High winds"

Higher winds - such as the 15-to-25-MPH (6.7-11.2-m/s) range as experienced in Yreka - are adverse for most streamlined bicycles. Furthermore, the lifting characteristics may be nearly maxed out or stall separating - or in other words - lift is no longer increasing much, but drag is. The bicycle gets buffeted around and likely ends up running slower.

"Displacement of HPV records."

Considering the history of the hour record for the last ten years, the average increment between records has been less than 1.5 MPH (0.67 m/s), and thus a modest wind in the 5-to-10-MPH range could result in a displacement of an otherwise true "human powered vehicle" record. I think such displacements generally stifle efforts to set records, and ultimately slow the improvement of both the vehicles and the records. Such displacements are bound to happen, and have, and the effects are evident too.

Considering that, I'd say the wind is relevant if you are seeking to know what is the most efficient "human-powered vehicle" rather than say the most effective "local-natural-power vehicle" (e.g. local muscle, wind, gravity, solar). The question then is it worth the trouble to measure the wind, or how might you measure it?

Measuring the wind

An "instant-wind-speed" anemometer is next to useless (unless it feeds into a data-logger). What you want to know is how much total wind (flow distance) has occurred within the hour period and ideally sub-periods within it. The average wind (total distance/ time) is a good approximation. A more exact calculation would be to have additionally the total distance every 2 to 5 minutes or so or even arbitrary weighted sub-periods, and do a "velocity-squared" average. Interestingly, the power assist of the wind for a streamlined bike goes with the velocity squared - the reason in part being that the vehicle velocity is largely independent of the wind speed, unlike say a windmill. For a windmill, the "velocity-cubed" power average is the most accurate (as is commonly done in assessing windmill sites using wind speed distribution (Weibull curve, etc...) characteristics of the site.

Large tracks.

Interestingly, for a large site - such as two-mile (three-km) or larger track, the instantaneous wind speed at one end of the track tends to correlate poorly with the opposite end, but the average wind speeds are remarkably consistent (assuming one end is not grossly "shadowed" by some structure). This makes average-wind-speed measurements for a long event like the hour far more meaningful than wind-speed measurements for say, a top-speed sprint.


The top-speed sprint covers a large distance (often two miles or more of acceleration) and the wind speed for the brief instant the vehicle passes through the time trap has little to do with the wind speed over the entirety of the course. In other words, the wind speed at the end of the course may or may not correlate well with the wind the vehicle experiences over the acceleration distance, but at least it gives a probable idea, especially if multiple runs and readings are performed. It would make more sense in the case of a sprint to measure the wind speed for the entire duration of the run-up and sprint and optionally take a velocity-squared average. This can be done by simply reading the time and feet of wind when the bicycle starts rolling and when it finishes and then divide total distance/total time. Good runs are then actually less likely to get ruled out because of a little gust and vice-versa (i.e. if it is generally calm, everyone will have good runs, rather than some getting ruled-out).

Probability of getting "Legal" winds

A quick analysis of available weather-station data for a number of locations suggests that the probability is high to get "legal" (sub 1.66 m/s) winds as previous hour records required - in the early morning hours. Typically a window of three hours or so after sunrise exists. After that, winds are likely to persist until midway through the following night. Later in the day, such low winds are far less certain and each site must be considered individually.

So, if you want to set an hour record with low winds, plan on running in the early morning.

Key "power assist" areas

For streamlined bikes, there are three major areas of "power assist" that are most likely to occur.

  • Gravity
  • Wind
  • Pressure gradient
Gravity is significant - it is one of the primary keys to fast runs recorded in the past - typically more so than altitude. A hardly discernible and "legal" downhill assists dramatically at higher speeds. (For instance, gravity adds a steady additional 0.25 HP (187 watts) at 65 MPH on a (legal) 2/3% slope for a rider-plus-vehicle weight of 215 lb., 98 kg.)

Wind as discussed - looks as if it gives modest yet significant assist by default to leading streamlined bicycles before it grows too large and hinders handling.

Pressure gradient, the unrecognized aerodynamic advantage, becomes probable if a "pursuit" automobile is following behind a streamlined bicycle. The effect is far more dramatic than most HPV enthusiasts would suspect. An automobile can approach a streamlined bicycle from behind and gently accelerate, and long after the cyclist stops pedaling the streamlined bicycle will continue to "ride" the pressure-gradient "wave" extending significantly several car lengths in front of the automobile. The streamlined bicycle essentially accelerates on its own. (For instance, a truck approaching a streamlined bike from behind may never be able quite to catch up no matter how fast it can go!) This effect holds true up to Mach numbers well over 0.5, and eventually diminishes to zero approaching the speed of sound.

There are some neat research references dating back to the 1930s well documenting this effect. Dolphins know it well, often racing for hours directly in front of ships. It also applies to a crouched racing cyclist on a standard bicycle, but it is far less dramatic than with a fully streamlined bicycle.. I have carefully tested this after witnessing some unusual performances in several time trials in the Tour de France. Even with a compact car the effect is very repeatable and correlates well with some "unexplainable" margins.

At least one of the exceptional performances in a trans-continental race may have utilized the pressure-gradient advantage. Without going into specific incidents, the general application has been to follow the cyclist closely with a large support vehicle, and to go so far as to substantially increase the frontal area (and extent of the pressure field) of the vehicle by mounting a large "billboard" banner extending several feet above the top front edge of the vehicle.

I should add one interesting quotation that may relate to this question, and that appeared in the San Francisco Chronicle (Monday, February 8th, 1993): "I felt like I was dying," "At [x] miles per hour I'd gone anaerobic, but I forced myself to keep turning the cranks and our speed kept rising. It felt like the [n] was accelerating on its own, a real testimony to aerodynamics."

The "felt like I was dying" anaerobic state is strongly associated with a decay in power output, which is associated with a fundamental reduction in speed with few exceptions, not "accelerating on its own." A rather interesting quote!

The "HPV".

Anyway, I like to call an "HPV" an HPV if it really is what its title claims - a "human-powered vehicle." Then you can look at the very fundamental contest that combines solely the power of a human's muscles and a human's mind to pure movement, speed and efficiency going from A to B - the very stuff so central to our existence and so pervasive in our daily lives. Such a contest reveals vehicle efficiencies far exceeding all other vehicles I know of - for a given speed times occupancy, nothing does as much with as little.

To clarify, in my opinion the Du Pont prize for top speed was brilliantly won fair and square by Freddy Markham in Gardner Martin's Gold Rush at 65.48 MPH (29.27 m/s), but the fastest official "human-powered-vehicle" record now stands at 62.51 MPH (27.94 m/s) set in 1999 by Sam Whittingham in Georgi Georgiev's Varna Mephisto in dead-flat near-zero-wind conditions.

To increase this "HPV" record to, say, 70 MPH (31 m/s) will require a nearly 40% increase in rider output or vehicle efficiency.

There is lots of energy all around us in the form of wind, and surely we should better utilize it. But I consider racing a "hybrid" sail-HPV, from which phenomenal performance is possible, a different contest.

Milwaukee hour

Just a note on the Milwaukee hour you mentioned - I crashed for the first time at about 48 mph (21 m/s) earlier that day, and I was a bit uneasy to race at all. I decided I'd run if I rode on the outside of the track, and we measured and estimated I traveled over 3 miles (5 km) extra over the course of the run.

Not to mention I started about 1/2 lap late as my windshield came loose just before the start of the race! I honestly didn't notice the wind. My hydration system failed, my head was overheating with the ventilation messed up due to the last-minute windshield fix, and so I was more or less persisting to the finish!

To my surprise, I actually passed the Gold Rush on the last lap. I believe he was ahead of me at the 60-minute mark, but the rules required us to complete our last lap. Gardner ran down the track and signaled to Freddy to ride through the last lap. If you have video of the race, you might see this all happen. The officials said I was a lap behind Fred, and it took Freddy to tell them that I was in fact on the same lap as he before they'd admit their error. I was happy with Freddy being the winner since he led most of the race, so no matter what the rules said there. Interestingly they somehow managed to get our marks recorded to the nearest 0.001 mile for the record! (I wonder what measurement they used?!)

The next day was faster - I wish they recorded the lap times. Other than the caution lap, I averaged just shy of 48 MPH (21 m/s), and would have been happy to keep going at that rate except there were more trees off to the side of the course than I cared for so I limited my speed except for the last lap which I sped up and averaged just under 52 MPH (23 m/s). I wanted to go fast enough to "wear down" Freddy so that he wouldn't take me at the finish with some mean sprint, but not so fast that I might crash and get hurt. Freddy and Paul Swift and Bobby Livingston were all real excited after that event - they all said they wanted to ride the Cutting Edge and see who could go the fastest! I'd love to see what they could do! The only problem was that the bike was too long for all but Paul Swift's legs!


1. Modest winds appear significant - reasonable and confident estimations for leading streamlined bicycles indicates speed increments of about 1.5 MPH (0.67 m/w) with a 5-MPH (2.2 m/s) randomly oriented side wind at steady hour-record speeds (50 MPH, 22 m/s) on a closed circular course. Increments as much as 3 MPH (1.3 m/s)are probable before significant notice of wind effects by the rider on most vehicles. Handling problems may eventually arise with winds greater than about 10 to 15 MPH with detriment to performance.
2. A 1.5-MPH increment in vehicle speed is larger than the average increment in the world hour record over the last four or so records set in the last ten years.
3. Circular courses - yield about 45% of the power assist a perfect side wind would give.
4. The probability is high to obtain "legal" (sub 1.66 m/s or 3.71 MPH) winds in the early morning hours for most courses, but the probability of having "legal" winds is not very high at other times of daylight.
5. Instantaneous wind measurements may correlate poorly at different points on the course, but averages are remarkably consistent over several miles or more in flat areas assuming no wind shadowing at the point of measurement.
6. Wind measurement is relatively simple and meaningful if done as "total distance" of wind per hour. More precise readings would involve a "velocity squared" average for land vehicles (not velocity cubed as for windmills).
7. Weather data of Frankfurt/Main suggest that there may have been wind in the 5-10 MPH (2.2 - 4.5 m/s) range at the 8 PM European time Aug 7, 1999 of the latest hour record attempt.
8. Check if the track has a weather station as most test tracks do, or other nearby station evidence of the conditions. If the track weather station confidently shows legal wind at the time of the race, then that measurement should be respected and considered official.
9. Previous records and other teams have been held to certain requirements that appear to have significant effect on performance. Until the rules are changed, it is a simple matter to uphold this well-known and simple-to-measure requirement if an official record is desired. Otherwise, to exempt generously the latest claim simultaneously rejects the previous official record holder's title and may also displace potential future HPV record achievements.
10. The position of any "chase" or support vehicle behind the streamlined bicycle should be clearly documented demonstrating following distances in excess of 100 feet at all times except maybe momentarily at the finish for marking purposes only.
11. Arbitrary wind should be allowed for long-duration records two (two hours or more) from a practical standpoint, but should still be recorded and maybe velocity-squared power averaged for purposes of meaningful comparison of achievements (not unlike the documentation of altitude or course slope).

Matt Weaver,

The above Technical Note was originally published in Human Power, the IHPVA technical Journal, issue Number 49, Winter 1999-2000.

It is posted here with the permission of Matt Weaver, Copyright © 1999, 2000

Return to Main Page
Copyright © 2000, C.C. Broome 
Contact: Chris Broome