Human Power Editorials lar group, first. Other countries are also The technical journal of the having outbreaks of right-wing advocacy International Human-Powered Vehicle Internationalism and violence. There is at present a high- Association The IHPVA is often criticized, per- ly embarrassing war of words between a David Gordon Wilson, editor haps justly, for not being sufficiently in- small minority of public people in Japan 21 Winthrop Street Winchester, MA 01890-2851, USA ternational. It is something we all have and the USA following a disastrous state Phones: 617-729-2203 (home) to work on constantly. All the "higher" visit by the president and a group of 617-253-5121 (MIT) animals are territorial. Mankind easily over-paid auto-company executives. 617-258-6149 (FAX) reverts to this pattern. As I write this, (Why couldn't he have taken some of one of the US presidential challengers is our HPV pioneers - people we could Associate editors trumpeting the call of "America First!" I have been proud of?) Toshio Kataoka, Japan thought that that had disappeared in the These reflections are partly the re- 1-7-2-818 Hiranomiya-Machi Hirano-ku, Osaka-shi, Japan 547 thirties, a time when I remember my pri- sult of another act of kindness by a Japa- Theodor Schmidt, Europe mary schoolteacher hugging me and say- nese friend, in this case Akira Naito. Hoheweg 23 ing "Aren't we lucky to belong to the We were very grateful to him for giving CH-3626 Hunibach best country on earth?" Pride in one's us his paper on HP helicopters in the last Switzerland country and in one's group or team is issue. He was appreciative of the help Philip Thiel, watercraft good so long as it doesn't lead to a dis- we had given in editing the paper. He 4720 7th Avenue, NE Seattle, WA 98105, USA paragement of other countries and has just sent a stunningly beautiful pre- IHPVA groups. In Britain it was only slightly sentation of micro-origami - Parade of P.O. Box 51255 humourous to refer in those days to peo- Cranes. (The folded-paper cranes have Indianapolis, IN 46251, USA ple unlucky enough not to have been wingspans from about 25 mm to 0.7 Phone: 317-876-9478 born in Britain as "the great unwashed". mm, too small when mounted on the Officers A famous London Times headline is re- points of sewing needles to be seen with- Dave Kennedy, president to have stated "Fog in the Chan- out a strong magnifying glass). I hope Adam Englund, secretary puted Bruce Rosenstiel, treasurer nel: Continent cut off' On my first visit to present it to a museum so that others Marti Daily, exec. director to the USA I was entranced at the great- may appreciate the craftsmanship and Paul MacCready, int'l president er interest in and acceptance of people artistry. Doug Milliken, VP water of other countries than I saw in my own I also feel, as I did when Ellen and I Glen Cole, VP land country. Foreigners were treated as peo- visited Japan, that in the face of many Chris Roper, VP air ple having habits that were no so much beautiful and subtle courtesies shown to Matteo Martignoni, VP ATV Theodor Schmidt, VP hybrid power strange as interesting. At the same time us we may have been perceived as we were guilty of appalling racism. We clumsy, even boorish. (No one gave us Board members in the USA have apparently progressed any impression of perceiving us that Allan Abbott in this area so far in the last thirty years way). The point that I want to make is Marti Daily that we feel smug when confronted by that, if we are to make the IHPVA truly Peter Ernst occasional renewed appearances of international we have to keep on our Chet Kyle the guard not to disparage any practices of Gardner Martin throwback movements proclaiming Gaylord Hill need to put the country, or some particu- other groups, but to do our utmost to Dennis Taves communicate. compete and cooperate in David Gordon Wilson In this issue fun and friendship. Human Power is published quarter- Editorials Dave Wilson 2 De-cavitator HPH Mark Drela, Almost DTP ly by the International Human- Human Power has relied in the past Powered Vehicle Assoc., Inc., a non- Marc Schafer & Matt Wall 3 on a network of volun- profit organization devoted to the for its production Letters to the editor 4 teers and of one or two people who did study and application of human mus- Book reviews - Bob Stuart 5 cular potential to propel craft through professional work at well below market the air, in and on the water and on "Paddlewheels" - a correction 9 rates. Drafts and diskettes were shuttled land. Membership information is In search of the massless flywheel from place to place around the country, available by sending a self-addressed John S. Allen 10 sometimes suffering some delays and stamped business-sized envelope to the What is an HPV? Rob Price 13 always surprising the editor somewhat. I IHPVA address above. Aditional copies of Human Power Cycling Science 3/3-4 - a review 18 would never know, until the final prod- may be purchased by members for More on hull shapes for pedal power uct turned up, exactly what the composi- $3.50 each, and by nonmembers for Augustus Gast 19 tor had managed to squeeze in. For this $5.00 each. Advances in flow visualization issue I'm trying the experiment of pro- Material in Human Power is copy- most of it myself. I've bought a a review (Dave Wilson)20 ducing righted by the IHPVA. Unless copy- new, fast PC, Microsoft Windows, DOS righted also by the author(s), complete Wind + HP quadracycle Wally Flint 21 5.0, and Lotus Ami Pro. (If this sounds ar, ticles or respresentative excerpts Tricanter HPV energy modelling may be published elsewhere if full like an advertisement, let me reassure credit to the author(s) and the IHPVA J.KRaine & M.R.Amor26 you that it has a purpose. On the title is prominently given. Richards' Ultimate Bicycle Book page of my last book I printed "Pro- We are indebted to the authors, to a review (Dave Wilson)33 duced on Lotus Manuscript". When it Marti Daily and to Carolyn Stitson, Second int'l HP-submarine race didn't work as advertised, Lotus was fan- whose dedicated help made this issue Cory Brandt 34 tastic. It brought out a new version of possible. Dave Wilson l Uawp. 'in fluurlllrfuuAS-/flLsfU Urt P. V/ Cover design by Bruce adndmore p. 2 Human Power, fall & winter, 1991-2, vol. 913 & 9/4

DECAVITATOR PxjrraPoweredHy'ofoil

27 October91

0 1 2 3 4 5f

Figure 2 Three-view drawings of Decavitator, withoutfairings Drawn by Mark Drela requires only a modest anaerobic effort en's racing kayak, with the deck lowered the overall vehicle drag if large-chord for a few seconds. Once flying on the by about 2 inches / 50 mm. A similar wings are used at low takeoff speeds. hydrofoils, the vehicle can be sustained design is employed for the monohull An earlier version of the Decavitator had by a fit cyclist at 9-10 knots / 4.5-5 m/s Hydroped vehicle. Molded composite a rather large takeoff wing of 5", 127 with aerobic power levels. A maximum construction with a hard gelcoat finish mm, average chord, and required exces- effort produces about 15 knots (7.5 m/s). gives very nearly the lowest drag attain- sive takeoff power due to the 2-D wave- Very High Speed. After unlocking a able. Although such exotic pontoons drag mechanism - as clearly evidenced safety latch, the rider has the option to might seem frivolous on a hydrofoil by the dramatic wave train set up behind pivot the large wing up and out of the boat, their low drag is in fact crucial to the wing. Reducing the wing area by water, much like on one of the more re- top-speed capability of the vehicle. nearly half gave a larger Froude number, cent Hydroped variants. The wing piv- the oting is accomplished by accelerating Reducing pontoon drag permits higher and produced a large power reduction the vehicle to at least 14 knots / 7 m/s (a takeoff speeds, which in turn permit despite the higher takeoff speed. fairly hard effort), and then suddenly in- smaller wings and higher maximum A further advantage of higher takeoff creasing the angle of attack of the entire speeds. speeds is that it permits optimizing the wing system via the left lever, which Higher takeoff speeds also have the propeller for higher maximum speeds. drives the vehicle upwards. When the important effect of reducing wave drag One useful feature of a racing-kayak hull upper large wing breaks the water sur- associated with the two-dimensional shape is that it retains its low-drag char- face, rubber cords pivot it together with wave train set up behind a lifting airfoil. acteristics when partially raised out of its mounting struts forward and up into a This is quite independent of the "inverse the water. This permits a very gradual streamlined receptacle. The sequence is ground effect" mechanism of the free and low-power transition to the foil- shown in figure 3. If the high power is surface which increases the induced drag borne mode, where the wings gradually sustained, the vehicle then rapidly accel- of a 3-D lifting wing. As described in lift the pontoons as the speed is in- erates on the remaining small wing to its Hoerner [4], the 2-D wave drag scales creased. The use of a rider-adjustable propeller be- maximum speed. The air inversely with the square of the chord- angle of attack of the wings is also im- comes very efficient in this operating and exponentially portant, as it permits the pontoons to re- mode. based Froude number low-drag Pontoons with the square of the depth-based main at a nearly-level, 2 2 orientation at all speeds. Each 17-foot / 5.2-m pontoon hull is Froude number: C.avJCL = 0.5gc/V 2 drag can dominate shaped like a modern open-water wom- exp(-2gh/V ). This

- __ ------__ - ---- ______p. 6 Human Power, fall & winter, 1991-2, vol. 9/3 & 9/4 Drive System lift-to-drag ratios are fairly good. Ordi- bows. The larger 60x2.35-in / The rider is seated in a semi-recumbent narily, a substantial blade CL at high 1520x60-mm (span x mean chord) wing position on an adjustable seat with Ke- speeds result in a very large blade CL at is placed about 6 in / 150 mm below the vlar cloth webbing. The pedals are lower speeds, stalling the blades and pontoon bottoms, and the smaller linked to the two-bladed 10-ft/ 3-m di- making transition to the hydrofoils diffi- 30x1.4-in / 760x35-mm wing is placed ameter air prop via a 1/4-in / 6-mm pitch cult. However, because of the high disk another 6 in / 150 mm lower. Each wing stainless-steel chain-drive with a 2:1 loading, the prop has a very substantial is supported by two slender struts placed gear ratio. The propeller is of a self-induction, or "slip", at low speeds 26 in / 660 mm apart. The advantage of minimum-induced-loss type, and has (i.e. it draws air into itself). Together using two struts is that they do not need been designed with algorithms similar to with the modest takeoff-power require- to carry significant bending moment, those of Larrabee [5]. The propeller is ments of the low-drag pontoons, this and hence can be made much smaller designed to rotate at 250 rpm (125 rpm self-induction is sufficient to prevent the and have a lower overall drag than an at the pedals) with maximum power at blades from stalling above speeds of 5-8 equivalent single strut. Using two struts 20 knots / 10 m/s. Its pitch can be dock- knots / 2.5-4 m/s, depending on the geo- also greatly relieves bending moments adjusted to optimize its performance at metric pitch setting. on the wings, and permits much smaller lower speeds and power levels, and to Another very large advantage offered wings to be used for a given material- compensate for wind direction. If the by the air propeller is that the wing stress limit. The wings employ a cus- air/water density ratio is accounted for, struts do not need to enclose any drive tom 14%-thick airfoil which has been the 10-ft / 3-m air propeller is equivalent system, and can be sized as small as tailored for the operating Reynolds- to a 4-in / 100-mm-diameter water prop material-stress and buckling limitations number range of 150 000 - 400 000, us- in terms of the non-dimensional thrust permit. Where it attaches to the small ing the design principles and numerical coefficient T, = 2T/rhoV2piR 2 , which wing, each strut has only a 1-in / 25-mm simulation methods employed for the determines the induced or "slip" losses. chord and a 0.15-in / 4-mm thickness. A Daedalus-wing airfoils [6, 7]. The struc- At low takeoff speeds, the 10-ft air prop strut enclosing a chain or shaft transmit- tural merit of the relatively thick airfoil gives high disk loadings (large Tc) and ting 1 hp / 750 W would need to be far allows smaller wing areas and less over- poor efficiency relative to what could be larger. In addition, the exposed hardware all drag than the 10-12%-thick airfoils obtained with an effectively larger 8-in / associated with an air propeller has neg- more commonly employed at these low 200-mm water prop, say. At speeds ligible air drag, while a housing for an Reynolds numbers. The thick airfoil also close to 20 knots, however, T, becomes underwater propeller mount typically gives the rather wide usable lift- sufficiently small to give efficiencies has a substantial drag penalty. coefficient range 0:2 < CL < 1: 1, which close to 90% even at maximum power. Hydrofoil/Strut System translates to low wing drag over a wide This high efficiency is also due to the The hydrofoil system consists of two range of speeds. The ability of the large prop blade lift coefficients being reason- fully-submerged high-aspect-ratio wings wing to perform well from 7 to 15 knots ably high at CL=0.6 (the Daedalus prop under the rider, and two skimmer- is particularly important for the Decavi- airfoil is used), so that the blade-profile actuated trim surfaces on the pontoon tator as it is brought to its maximum-

/q-

(a) (b) (c) (d)

Figure 3 Transition sequencefrom double- to single-wing mode: the large wing pivots out of the water into a streamlined re- ceptacle. Human Power, fall & winter, 1991-2, vol. 9/3 & 9/4, p.7 speed mode. Each of the two 9x0.85-in tial since the small wing dimensions Smaller takeoff wing. Since the effort / 230x22-mm front trim surfaces is push material stresses to the limit. The required to lift the pontoons off the wa- mounted at the bottom of a slender rud- small wing, for example, experiences ter is quite modest, the area of the large der in an inverted-T configuration. Each 100 000 psi / 690 MPa material stress wing could be decreased somewhat. The rudder pivots on two axes in a gimbal with a 140-lb / 64-kg rider at 2 g, and areas of the front trim surfaces could be mounted on the pontoon bow. The pitch hence could not be safely built even out decreased proportionately as well. The axis is controlled by a surface skimmer of aircraft-grade solid aluminum. The reduction in wetted area would reduce cantilevered forward from the gimbal, struts connecting the pontoons are oven- the considerable effort needed to achieve while the steering axis is controlled by baked tubes made of pre-preg carbon sufficient speed for the transition to the the rider via cables linked to a right side- fiber formed around aluminum man- single-wing mode. The rider would then stick. The geometry of the drels. These are also highly stressed, and have more energy available at maximum skimmer/trim-surface mechanism is set the use of carbon fiber gives greater stif- speed. up to lift the pontoon bow a few inches fness as well as weight reductions of Aero fairing. Although all major ex- off the water surface at speeds over 6 many pounds over equivalent aluminum posed tubes and struts have already been knots / 3 m/s. This height is firmly main- tubes. carefully faired, the aerodynamic drag tained at all higher speeds, so that the Each pontoon shell is a pre-preg near 20 knots / 10.3 m/s still consumes vehicle is stabilized in depth and roll, glass/carbon/Nomex/glass sandwich, and between 25% to 35% of the propulsive and can pivot only in pitch about the was baked inside a mold for an open- power, most of this being drag on the pontoon bows. This pitching alters the water women's kayak owned by Com- rider. Enclosing the rider in a high- wing's angle of attack relative to the wa- posite Engineering of West Concord, quality aerodynamic shell would ter surface, so that for any given speed MA. The top of each pontoon is perma- theoretically push the maximum speed the boat rapidly seeks the one unique nently sealed off with a glass/No- past 20 knots. Naturally, for record- pitch attitude where the wing lift equals mex/glass deck. Internal plywood setting runs it is desirable to operate the the vehicle weight. By altering wing bulkheads hold the strut-attachment vehicle with the fastest legal tailwind angle of attack relative to the boat via bolts. The fuselage frame supporting (3.22 knots / 1.67 m/s) to reduce the air the left lever, the rider can therefore pre- the rider and drive system is constructed drag to an absolute minimum. cisely control the pitch attitude and of thin-walled large-diameter aluminum Larger rider. The benefits of increas- hence the wing submergence depth. At tubes joined with Kevlar/epoxy lashings ing rider size on a hydrofoil vehicle are low speeds, a large submergence depth in lieu of welds. Carbon-fiber tubes were significantly smaller than on a bicycle. is best to keep the large profile- and rejected for the frame from durability The actual benefits depend on the rela- induced-drag contributions of the free considerations. In retrospect, a carbon- tive fractions between profile and in- surface in check. At high speeds, the tube frame clearly would not have sur- duced drags. With the maximum legal viscous profile drag of the support struts vived the numerous modifications and tailwind, the Decavitator's induced drag becomes more dominant, and a very general abuse seen by the frame over the is about 27% of the total at 18 knots / 9 small submergence depth is optimal. vehicle's three-year lifetime. The seat is m/s, and 20% at 20 knots / 10 m/s, so a The minimum workable depth is set by likewise constructed of lashed thinwall larger rider would have some advantage. the need to avoid ventilating the wing by aluminum tubing with a Kevlar cloth However, the vehicle's hydrofoil system an errant wave trough. Loss of lift due to webbing, and employs adjustable is already very highly stressed with the ventilation immediately drops the ve- mounts for different-sized riders. The 140-lb / 64-kg design rider weight, and a hicle onto the pontoons. The pivoting of drive system employs standard bicycle significantly heavier rider would require the large wing out of the water is an es- cranks and pedals, lightened somewhat larger underwater surfaces to provide sential feature of the Decavitator's hy- by drilling. The chainwheels and sprock- greater structural strength. Also, the drofoil system. Removal of the large ets for the 1/4-in- / 6-mm-pitch chain heavier rider would need to expend dis- wing reduces the total underwater were custom-made from high-strength proportionately more power to lift the wetted area by a factor of three, giving a 2024-T4 aluminum plate by pontoons and when preparing for the roughly proportional reduction in profile numerically-controlled machining. single-wing operating mode, unless the drag. This is partially offset, however, Each propeller blade is a hollow shell wing areas are increased. In either case, by a substantial increase in the induced with a hard Rohacell-foam shear web, much of the larger rider's advantage dis- drag due to the loss in total loaded span. bonded to an aluminum-tube root stub. appears. Larger propeller. As men- Overall, a speed increase of about three The shell surface is a Kevlar/Roha- tioned earlier, the air propeller is knots is realized for the same power lev- cell/Kevlar sandwich, laid-up wet and relatively inefficient at lower speeds due el. vacuum-bagged in a female mold. to excessive disk loading. Increasing the Construction Carbon-fiber rovings are incorporated diameter would therefore give more The Decavitator makes extensive use into the shell sandwich for bending thrust at low speeds for the same power of structural and manufacturing strength. The propeller shaft is a thin- input, giving faster transition and accel- technology developed at MIT in the walled large-diameter aluminum tube. eration. This may significantly conserve course of numerous human-powered- Further developments the rider's energy and hence permit a aircraft projects. All underwater sur- Possibilities for further increasing the higher power level to be sustained over faces are made via wet lay-up of solid Decavitator's performance include the the 100-meter course, although this is carbon/epoxy vacuum-bagged in female following. difficult to quantify. Offsetting this po- molds. The use of carbon fiber is essen- tential benefit is the increase in size and p. 8 Human Power, fall & winter, 1991-2, vol. 9/3 & 914

energy storage or release in steady-state IN SEARCH OF THE MASSLESS FLYWHEEL operation, but does increase rolling and aerodynamic friction, and can lead to by John S. Allen cause backlash in a low-rpm human- powered system in which the flywheel must be geared up to run at high rpm. A Human power, as commonly trans- HP-alternator problem flywheel also makes both starting and mitted through pedals and rotating I became aware of the importance of stopping more difficult. cranks, peaks once every 180 degrees of the flywheel effect after I built a human- It does not matter for a conventional crank rotation, with deep "dead spots" in powered device that lacked it. This was flywheel when or in what amounts ener- between. Placing the two cranks 180 an automotive alternator mounted on a gy is input and extracted. Short or long, degrees out of phase results in in two bicycle, with a two-stage chain drive to light or heavy pulses of energy input or power peaks per turn of the crank. In bring the rpm up to the required range. output make little difference to it, as addition, the muscles of the leg can Unfortunately, an automotive voltage long as it has sufficient capacity to ac- drive the foot in any direction, if weakly regulator works to keep the alternator's commodate them. If the pulses are too in most -- so the dead spots aren't as bad output constant regardless of how fast it long and/or heavy, you move up to a as with single-cylinder steam engines, is turning. Consequently, the input larger, more massive or faster-spinning treadle sewing machines and the like, torque requirement increases as the gen- flywheel. But in a human-powered sys- that can actually hang up at dead center erator turns more slowly. This works tem, especially a human-powered ve- and must be hand-started. The small fine with an automobile engine, which hicle, we want a lighter flywheel. If we amount of torque that the rider can apply has a multicylinder engine and a fly- design a device that requires a particular at the top and bottom of the pedal stroke wheel, but it worked poorly for me: the input and output -- tailored to the human makes it possible for a bicycle's drive- pedals were difficult to push over the engine and the task at hand -- we can train to include a freewheel, without the dead centers or to accelerate to running achieve this goal. need for a flywheel in the crankset to speed. bring the cranks over the dead spots. The dead-center problem -- though A no-flywheel alternative In a bicycle, the unavoidable mass of not the startup problem -- can be solved The lightest flywheel is no flywheel the bicycle and rider takes the place of a by a conventional flywheel, using rota- at all, and is realizable in the case of our flywheel, maintaining a relatively steady tional inertia. I've seen at least one de- electric generator charging a battery. forward speed despite the uneven power scription of this venerable idea in All that is necessary is to add a tachome- application. This works very well at Human Power, in an electrical generat- ter sensor to the voltage regulator, so it normal riding speeds, but poorly when ing system used on a sailboat to power cuts out the generator below a certain climbing a very steep grade at a speed an automatic steering system. rotational rate. This idea is nothing new below three or four miles per hour (two But my generator was mounted on a to people who build wind generator sys- m/s) -- so the rider compensates with a bicycle intended also to be ridden: one tems. One approach is to use a "pigeon walk," throwing upper-body goal in assembling it was to promote hu- centrifugally-actuated mechanoelectrical mass backward and forward to even out man power in a parade from the seat of switch borrowed from electric typewrit- the bicycle's forward speed and to put my moving bicycle, through a powerful ers. (See "Marshall Price's 'Basement- added downward force on the rear wheel public-address system. I also wanted to Built' Windplant," in Mother Earth during the power phase of the pedal be able to pedal the generator when the News, Issue 100, July-August 1986, p. stroke. At very low speeds, and espe- bicycle was stationary, so I could not 103). cially on soft surfaces, the ability to shift drive it directly off the rear wheel and Taking advantage of the chain-and- body mass gives an upright bicycle some use my own mass as the flywheel. A sprocket drive, we can also use the prin- advantage over a recumbent. Unfortu- conventional flywheel would be heavy, ciple of the Hammond organ: a coil of nately, no type of bicycle gives any making the bicycle harder to pedal up wire wound around a bar magnet with climbing advantage to a rider with ex- hills, and would act as a gyroscope, af- one pole placed near passing steel cess mass! fecting steering and balancing. The sprocket teeth will generate an alternat- The flywheel effect requires low roll- search for a solution to these problems ing current increasing in frequency and ing resistance and a rigid connection led me to consider some alternatives to intensity as the sprocket spins faster. through the drivetrain to the road sur- flywheels. Rectifying and smoothing this output face, as you rapidly discover when ped- will generate a dc control signal that can aling through deep gravel or climbing a Conventional flywheels cut in the generator at the desired rota- slippery hill. (Nonetheless, only a dirt- A conventional flywheel has a high- tional rate. Just as with a wind-power track motorcycle or an all-wheel-drive mass high-speed rim, so that variations system, an ammeter and state-of-charge vehicle can outclimb a bicycle on a slip- in energy input and output cause only meter also must be supplied and heeded; pery surface, since the bicycle benefits small percentage changes in its rotation- since the system is designed to use input from having most of its weight over the al rate. The energy which the flywheel power as available, it cannot raise elec- drive wheel when climbing, and from stores and releases corresponds only to trical output to keep the battery fully rapid, sensitive control over drive its changes in rpm. The high average charged as power demand increases. torque!) rotational speed does not contribute to A flywheel accomplishes two pur- poses: to accommodate varying energy p. 10 Human Power, fall & winter, 1991-2, vol. 9/3 & 9/4 input, and to smooth energy output; but the pedals, we make the simplifying as- Mechanical solutions other devices may achieve one of these- sumption that the capacitor must store If voltage fluctuations are unaccept- goals without the other. The conven- energy for 1/4 crank rotation and release able, if more energy is needed than a ca- tional voltage regulator and battery it for the next 1/4. Then, the capacitor pacitor can conveniently store, or if we smooth the output but do not accommo- must store 375 x 60/400, or 56.25 watt- are powering a non-electric device, then date to the input. The cadence-sensing seconds (joules) of energy. we must look elsewhere, toward a me- chanical solution. voltage regulator, on the other hand, ac- Let us assume that we are using a 12-volt battery. This charges at 14.4 One possible device is a mechanical commodates the varying input from the volts and discharges at 13.2 volts. The input-power regulator that operates as human engine by varying the output to difference between these voltages repre- does our tachometer sensor on the elec- the battery. sents a power loss -- an efficiency of trical generating system, by somehow Despite this difference, the "feel" of only about 92 percent. To avoid this varying mechanical loading or mechani- a cadence-sensing voltage regulator will loss, the capacitor must prevent the volt- cal advantage. Achieving such control be similar to that of a bicycle with a age from going below 13.2 volts. is easy in a fluid-pumping system such freewheel: very little mechanical re- Taking the equation P = V2/R or R = as a water pump, air pump or hydraulic sistance up to the cadence at which the V2/P, our 375-watt load at 14.4 volts is press. generator cuts in; though, unlike with equivalent to an electrical resistance of In fact, there are classic examples of the bicycle, the "wall" at that cadence 0.552 ohm. The current is 26.06 am- regulated pumping systems: the player will be "soft" if the the rider pushes peres. piano and reed organ use the pneumatic down hard: in this case, conventional A one-farad capacitor, approximately equivalent of a dual mechanical ratchet voltage regulation will set in and reduce the largest commonly available, will de- drive with spring-loaded return stroke. the load at the pedals. This problem is liver one ampere-second when charged They limit pedal force and regulate out- relatively unimportant compared with to a potential of one volt. In our exam- put air pressure by dumping excess air the dead-center problem. A more con- ple, when charged to 14.4 volts, the ca- volume through a relief valve. The per- ventional bicycle-like feel might be pacitor holds a total of 14.4 ampere- son pumping the bellows quickly learns achieved by sending excess power to an seconds. In the 0.15 second correspond- that there is no advantage in pumping auxiliary load, such as a second battery ing to one quarter of a crank rotation, it more air than the air chest can store, and in need of charging. must deliver current at the rate of 26.06 that steady airflow and pressure can be The control device might be de- amperes, for a product of 3.909 ampere- maintained by using a quick return signed to auto-adapt to changes in pre- seconds. stroke, so the power phases for the two ferred cadence by slowly raising and The actual behavior of a capacitor is feet overlap. lowering the cut-in cadence in response not to deliver a constant current until ex- A more efficient arrangement to to changes in power input. Such a fea- hausted, and then cut off; actually, the pump gases or liquids -- though it would ture would be especially useful when the current and voltage together decay expo- not maintain steady flow -- would be a generator is mounted on a bicycle like nentially. Our 3.909 ampere-seconds dual-piston pump operated by two crank mine, intended to be ridden while gener- are more than a quarter of the total throws between the pedal cranks, driving ating electrical power. Control circuitry charge in the capacitor, and extracting two connecting rods 180 degrees out of might also be designed to mimic the ef- them will reduce its voltage by more phase so the maximum power demand fects of different chainwheel shapes, than a quarter -- well below our would be at the part of the pedal stroke since opinions differ as to which is pre- 12.2-volt minimum. when the most power is available. For ferable. Another way to put this is that the greater efficiency, our power regulator 1.2 ampere-seconds to discharge the ca- could use mechanical-advantage change Capacitors for short-term storage pacitor from 14.4 to 13.2 V corresponds rather than power-dumping. For exam- If some electrical output is to be used to 1.2 amperes times 13.8 (average) ple, we could simply place a spring in while pedaling, a large capacitor (in ef- volts for I second, or about 16.6 joules. each connecting rod between crankshaft fect, an electrical flywheel) in parallel (In this calculation, we have ignored the and pump piston. The spring would be with the battery is desirable, to avoid a small difference between the arithmetic equipped with a limit stop so it would battery discharge and loss in efficiency average and the geometric mean of volt- deflect only when the pressure on the between pedal thrusts. If we do not need age). piston exceeded a predetermined limit. to store up energy for use later, we can The one-farad capacitor is not large As with the pump organ, there is no dispense with the battery and use only enough to store all the energy needed, speed regulation with this arrangement, the capacitor, which is essentially though it fails by factor of only about 8. only an adaptation of the force require- 100-percent efficient. Let us examine As long as fluctuations of a fraction of a ment to the characteristics of the human whether we can easily obtain a large- volt are acceptable to the power-using power source. Speed regulation would enough capacitor to even out power de- device, the one-farad capacitor would be require a mechanical governor which livery through one pedal stroke. just about large enough with a current would, for example, close a valve, al- Let us assume that our human power demand of 50 watts -- a low-to-average lowing pressure to build up in the pump source is delivering 375 watts of power bicyclist's output. Also, recall that the once the desired cadence was reached. (about 1/2 horsepower) at a cadence of capacitor need be large enough only to A crankshaft-and-connecting-rod ar- 100 rpm. This is more or less a worst- store power which is not going into the rangement like that of the proposed case situation, representing the highest battery. For example, if we are running pump can also be connected through a power output which a top-rate human a radio drawing 50 watts and charging a pair of ratchets to a mechanical load. engine can maintain for a sustained large battery which can absorb 500 This arrangement is suitable for a winch time. Since there are two power phases watts, we need a capacitor capable of or a jack, working against weight or fric- and two coasting phases per rotation of handling only the 50-watt load. tion and whose power requirement is Human Power, fall & winter, 1991-2, vol. 9/3 & 9/4, p.11 proportional to drive speed. The power most riders would not produce as much leave out the ratchet drive and connect input required at the pedals then varies power. one end of the torsion spring to the fly- sinusoidally between a low and high val- The spring device would work equal- wheel and the other end to a rotating ue, since the load advances at a rate that ly well at any cadence, but it would not shaft of the drivetrain. Actually, two also varies sinusoidally. The peak pow- adjust itself to different levels of force equal torsion springs attached to the fly- er demand, as in our other examples, input -- probably not a problem with a wheel and driveshaft 180 degrees apart should be at the crank angle at which turbine or propeller, which typically op- would be preferable, in order to cancel maximum rider power is available. timizes in a narrow range of torque and radial loads on the bearings supporting Another approach for a winch is to rotational speed. A variable-rate spring the flywheel. wind the cable on a rectangular plate, as (with coils closer together at one end, so We select the mass and spring to is commonly done with children's kite- they are fully compressed before those at resonate somewhere between 150 and string holders. As the plate turns around the other end) could be made stiffer for 220 Hz -- twice the expected pedaling a central axis, the rope winds on at an heavier power applications by manually cadence. This is the predominant com- approximately sinusoidally varying rate. applying a preload. ponent of unevenness in pedaling output. Such a winch must use flexible rope or Placing the power-evening spring de- Any unevenness in shaft speed at this string rather than stiff cable, to flex over vice on the output shaft would avoid any rate will drive the resonating mass and the edges of the plate. A bicycle chain problem with backlash, but would re- spring system so it stores energy (winds on a hyperelliptical chainwheel could quire the output shaft to turn at twice the up) during the powered part of the stroke achieve similar results while allowing pedaling cadence -- actually, within the when the shaft is turning faster, then the use of conventional derailleur gear- range of common practice with the pro- overshoots and releases energy ing to vary the output ratio. pellers of human-powered boats and air- (unwinds) at dead center, when the shaft In many applications, however, the planes. Another option to avoid is turning more slowly. The oscillation connection between the human engine backlash would be to place the device on of the system have to store a maximum and the load is lossy, making the load an intermediate shaft on the way to the of about one quarter the energy deliv- itself into a poor flywheel, but the load output. A third option would be to drive ered in a full turn of the cranks. must advance steadily. A steady load on the spring from a two-lobed cam on the For these reasons, the flywheel can an electrical generator is one good ex- pedal crankshaft, or to use a pair of be lighter than a conventional, brute- ample. A human-powered boat or air- carefully-designed variable-rate springs force flywheel and/or can turn more plane is another: its propeller is nearly actuated by two connecting rods oppo- slowly. We must still bring the flywheel idling when the pedals are at dead cen- site one another on a crank throw on the up to speed when starting out, and as we ter. The propeller's rated efficiency de- pedal crankshaft itself. All of these op- do, it will require an energy input; but pends on fluid-dynamic characteristics tions could provide the appropriate this will not be as great as with a con- calculated for a given power input, and twice-per-crank-rotation energy-storage ventional flywheel. serious losses in efficiency are unavoid- and -release characteristic without back- Lag and overshoot when starting and able when the power input is uneven. lash. stopping can be minimized by rotational Almost needless to say, the energy- limit stops, also desirable to prevent Spring-mass energy storage storage device must be designed for the overstressing the spring. We might also One solution to this problem is a hu- lowest possible frictional loss, or it can consider pre-winding and latching the man engine consisting of two or more swallow up any efficiency gain that it spring so we can release the stored ener- well-matched riders, with the cranks 90 might realize. The spring should run gy as we start. This would allow the fly- degrees out of phase. But for a single- free, without a sliding guide, and all wheel to assist rather than hinder startup rider device, we can take our inspiration connecting-rods or cam followers should in a critical application like bringing a from the bicyclist's slow-speed "pigeon use rolling-contact bearings. Fortunate- human-powered boat up to planing walk" discussed earlier. If we include in ly, the low rpm of bicycle power produc- speed. the drivetrain a device that in effect tion favor such solutions. If our device uses a constant-rate rocks backward and forward, we can ef- spring, it would work best at only one fectively even out the power input. Mechanical resonator cadence, though, as mentioned above, it This device could be a spring, de- A second type of device that can would self-adapt to different power in- flected by a connecting rod on a crank- smooth power input when the output puts at that cadence. A variable-rate shaft turning twice as fast as the pedal speed must be steady is a mechanical spring could make the device adapt to cranks. The crankshaft should be phased resonator. Unlike the spring device de- different cadences, at the expense of so that the spring stores energy at the scribed above, this adjusts automatically some harmonic generation, perceptible strongest parts of the pedal stroke and to different levels of power input. It is as "jiggle" at the pedals, due to nonlin- releases it at the dead centers. Given the especially suitable to a stationary device ear operation. For example, a spiral same assumptions as in our example of such as a human-powered lathe, in spring can be designed so that some of electrical power storage -- 375 watts which extra mass is tolerable and load- the coils wind tightly onto each other power output by the human engine at a ing may vary considerably, though the near the limits of travel. The spring cadence of 100 rpm -- the spring must pedaling cadence remains relatively con- constant then increases with increasing store the same 56.25 joules (newton me- stant. displacement, raising the resonant fre- ters) of energy. For example, we could I propose a device like the balance quency of the system. Another way to use a spring with a constant of 225 new- wheel of a spring-wound watch. The broaden the cadence range is to use two tons/meter (166.0 lbf/ft) operating balance wheel consists of a flywheel or more coupled resonators. through a distance of 0.25 meter (0.82 ft driven by a ratchet escapement and con- Here's a rough-and-ready calculation or about 10 inches). A more compliant nected to the body of the watch by a spi- for the simplest case of an oscillating spring would be more typical, since ral torsion spring. In our device, we flywheel: for the same rider as before, p. 12 Human Power, fall & winter, 1991-2, vol. 9/3 & 9/4 delivering 510 watts (375 foot pounds A 2.7-kg (5-pound) flywheel, with dramatically increase speeds of athletes. per second) at a cadence of 100 rpm, the perhaps as much mass again for the But there is something about a running (2.3-kg) 5-pound rim of a flywheel will spring and supporting framework, would shoe that does not fit the accepted mean- have to oscillate through a linear dis- certainly be tolerable in a human- ing of a human-powered vehicle. tance of 3.43m (11.25 feet) to store and powered boat, though the devices dis- Impedance matching is an important release the energy that will take the cussed earlier that use only springs are concept in electrical engineering but is cranks through one quarter-turn. If, for probably more suitable for a human- also used by mechanical engineers, and example, the flywheel is 762 mm (2.5 powered airplane. is especially important when working feet) in diameter, it has to oscillate with the limited human power outputs through 756 degrees, alittle more than Closure used to drive HPVs. two turns, with respect to the shaft to I've described a few possible devices, The driver of a car must shift gears, which it is attached. This appears realiz- some well-known and some as yet un- or the transmission may do so automati- able, though I have not yet built a mod- built, to adapt the uneven power output cally, when the engine is running either el. of the human engine to various loads. too fast or too slow to produce power The rotational inertia and spring con- All of these devices achieve a flywheel efficiently. The transmission provides stant are noncritical as long as they effect or something like it without a con- an impedance match between the en- combine to produce the correct resonant ventional, brute-force flywheel. I've gine's and car's speed and power require- frequency: at that frequency, the sys- seen quite a number of human-powered ments. tem's mechanical impedance goes to a devices that could be more useful or ef- Humans can produce widely varying very high value. In order to optimize ficient if they incorporated such devices, power outputs over a wide range of mus- the oscillatory travel for a flywheel of a and I think that developing such devices cular rates and displacements, but most given rotational inertia, however, the is an interesting and fruitful object for outputs, rates and ranges are very ineffi- rotational rate of the shaft driving the human-power research. I hope that I cient, which can lead to early exhaustion flywheel must be chosen so that the have provided some food for thought, and even injury. To produce power with torque that the spring delivers to and and I am very interested in hearing from maximum efficiency humans need to from the shaft at its working limits of anyone who tries out one of the ideas operate over a narrow range of muscular deflection equals the peak torque that is I've described. rates and displacements. If a bicycle is required to even out the power input of geared too low, where the rider has more the drive system. John S. Allen is an engineer, writer and energy available than is required to pro- If the shaft turns too slowly, the consultant on bicycling. He is authorof pel the vehicle at its maximum attain- spring cannot deliver enough torque un- The Complete Book of Bicycle Commut- able speed, then the impedance match less it winds and unwinds too far, and ing and co-author of Sutherland'sHand- for that rider is not optimum. In this the flywheel hits the limit stops. If the bookfor Bicycle Mechanics. case the rider runs out of pedalling rate shaft turns too fast, the spring does not (rpm) before he runs out of energy to wind and unwind to the full extent possi- John S. Allen, 7 University Park, Wal- push the pedals harder. The opposite is ble; a softer spring and lighter flywheel tham, MA 02154-1523; 617-891-9307 also true where the gear is too high, so could achieve the same results. the rider is operating below an efficient Firstserial rights (c) 1991 John S. Allen speed and runs the risk of muscle and joint injury. WHAT IS AND WHAT IS NOT A fixed-gear track bike can have the gear ratio altered to suit rider and track A HUMAN-POWERED VEHICLE conditions. A derailleur-equipped bi- AND WHY cycle's gears can be changed while being by Rob Price ridden to optimize rider output over va- rying riding conditions. To be rated a human-powered ve- DEFINING SOME TERMS. hicle, there must be a way that the ma- chine provides for a reasonable machine-to-rider impedance match. It is always a problem to decide "Human-powered vehicle" is a term This is usually provided on wheeled ma- which of the many vehicles and "people with several sub-meanings. "Vehicle" chines by a drive mechanism. accessories" that abound in the world indicates that something is supported. "Powered" infers that the vehicle moves. Coupling speed is the speed at which might be considered to be human- a machine's drive train is accelerated to powered vehicles. This article attempts "Human-powered" indicates that one or equal the speed of the machine and to categorize these devices and decide more person(s) is or are providing the couples to allow power transfer. Scarce some are, and some are not, HPVs. energy required to move the vehicle. why energy and much of the power stroke are Guidelines are also developed that may Though not stated, it can be inferred that to bring the rider's power- in general whether the human power source is aboard the used be used to determine producing limbs and the drive mecha- a machine can be considered to be an vehicle, that is, the vehicle supports the to coupling speed, when a machine HPV. The International Human Pow- human, and that the human is able to nism is traveling at high speed, before the driver control where the vehicle goes, and ered Vehicle Association requires limbs can be used to input power. on vehicles used in where and when it stops. control of and brakes A standard bicycle requires most of a their land competitions, so methods of The foregoing definition could be pedal stroke to engage the drive when devices are also interpreted to include running shoes, guiding and stopping beginning to pedal again after coasting. discussed. which have been developed over time to

Human Power, fall & winter, 1991-2, vol. 9/3 & 9/4, p.13 Bicycles are constant-drive machines, so the coupling process is done only when the rider begins to input power after coasting. Devices with intermittent power input, such as ice skates, Nordic PEDe. Fo skis, skateboards, scooters, and some HPVs, need to achieve coupling speed on each power stroke. This is indicative of poor efficiency, hence a poor im- pedance match. Poor impedance match- ing is inherently inefficient. The coupling-speed and constant- versus intermittent-drive concepts are

discussed in detail in the HPV Drive SCt Trains chapters of the forthcoming HPV AK5 Handbook being edited by Allan Abbot t I;P and David Gordon Wilson. \ r ~~~-- II V-----

WHEELED AND NON-WHEELED L-LDi---w S- I MACHINES. - -I 15uLLD1W SEAT There are many devices that can sup- port a person, many by attaching to the user's feet, which are designed to aug- ment movement and be controlled by Figure I Rowing-scull design fatures

. their riders. The list is long, but divides . into the water and push against the re- length and into just two groups: those with wheels to use the trunk and leg sistance of the water to move the boat in and those without wheels. The wheeled muscles to assist the arms. [Rowers may the opposite direction. The human de- also use one long oar, called a "sweep", category further divides into wheel- livers power to the paddle or oars via the in eight-oared shells and similar boats]. driven and wheel-idling sub-groups, dis- arms. On paddled boats, such as kayaks Turning of paddled and rowed cussed later. In the unwheeled category boats and canoes, one arm acts as the fulcrum is accomplished by working only one there are three sub-groups: aircraft, wa- and the other dips the paddle into and tercraft and land machines. paddle or oar, dragging one oar while pulls through the water. Kayaks use a working the other, working the paddles double-ended paddle and the HPAs AND HPBs. paddler's or oars in opposite directions, or, for ca- two hands are stationed a few feet apart Air is a gaseous medium character- noe paddles, setting one end in the water near the center. Canoes use a single- ized by low friction. There are two air- behind the boat and pushing sideways, ended paddle with one hand gripping the like a rudder. Prop and paddlewheel borne HPV possibilities: gliders and end and the other about a third of the boats human-powered aircraft, or HPA. generally use rudders for direction- Glid- way out the length of the paddle. er lift is provided by updrafts, caused by al control. Since water is a high- The paddler dips one end of the oar viscosity fluid, terrain or thermals, once the airplane is boat speed decays quick- then paddles a few strokes on one side of ly, eliminating the need towed to flying speed. Gliders are hu- for brakes, al- the boat while providing a fulcrum for though man controlled, most drives can be reversed to but not human powered. the paddle end with the other hand. The Current HPA designs are brought to fly- stop quickly. paddle then changes hands so the action Paddled boats cannot ing speed, maintained there and con- be considered is moved to the opposite arm for the to be HPVs trolled by human power, via an efficient because impedance- next series of power strokes. In this way matching options are limited impedance-matching drive system to a to changing power is balanced on both sides of the hand position on the paddles. Kayaks propeller, so they are HPVs. Current boat and both arms, and the boat goes in HPA designs take off and land at very aredesigned for use in running water and a fairly straight line atop the water. the paddles are used more for directional low speeds, and tend to roll on small 50- Rowed boats use two oars for each control to 150-mm- (2 to 6") -diameter wheels and for fending off obstructions rower, one for each hand Each oar has than making headway. for their short takeoff and landing runs. [But the Aleuts a pintle, or pivot, mounted into each Once power input is stopped, HPAs land wouldn't agree - ed]. Row-boat oar gunwale, which is the top of the side of lengths are determined by boat width, and stop quickly without the need for the boat. Pintles allow the outboard brakes. which determines distance between the ends of the oars to be lowered into and Water is available in two forms use- pintles. The inboard oar ends must not raised out of the water and moved for- ful to vehicles, liquid and solid, or ice. overlap so they can be worked simulta- ward and aft, but not revolved. The Rowboats, racing sculls, canoes and kay- neously, and optimum outboard- to rower pulls on the inboard end of the aks are vehicles that utilize human inboard- length ratio is less than 2: 1. pow- oars. Putting the pivots outboard of the er via arms working oars or Rowboats do have sufficient impedance paddles rower reverses the direction of oar mo- through liquid water. Human-powered matching to be HPVs. Racing shells, for tion so the rower faces aft instead of for- solo or multiple rowers, mount the boats, HPBs, have been built that utilize ward, where paddlers face. Rowers may a drive train to turn a screw propeller or pintles well outboard of the gunwales on brace their feet against the boat hull and paddlewheel for motive power. brackets to greatly increase oar length to move their torsos forward and Paddles aft, bend- make a higher drive ratio. Sculls are and oars have large flat areas ing at the waist, to increase the stroke on their outboard ends that are dipped speed-limited by the rate at which the p. 14 Human Power, fall & winter, 1991-2, vol. 9/3 & 9/4 rowers can stroke, rather than water fric- Snowshoes are simple devices de- seat area as a fulcrum, a favorite of men tion. At high speeds the effective stroke signed to increase the surface area of the contemplating castration. As with Al- is limited by the need of the rowers to foot of the user to prevent falling pine skis, the quickest way to stop is to synchronize the oars with the relative through the crust atop deep snow. Nor- fall over. Though Nordic racers can water flow, coupling speed, before dip- dic skis, even after they have broken move very fast, long arm-with-pole and ping the oars and applying power. Some through the crust, are much more effi- leg strokes are necessary to reach coupl- shells, Figure 1, use sliding seats to al- cient. Snowshoes are designed only to ing speed before the limbs can input low the legs to assist the arms and to in- sap the strength of the best human en- power. Despite the bottom surface crease oar stroke length more than can gine, thus are not HPVs. Sleds and to- clutches, there is no impedance- be done in a fixed-seat rowboat. The boggans are purely gravitypowered, are matching mechanism, so Nordic skis impedance match is improved by the use unsteerable and have no brakes, so can- cannot be considered HPVs. of the seat-slide mechanism, and sculls not possibly be HPVs. Snowboards are a wheel-less skate- are good examples of HPVs. Propeller Skis come in two major varieties: al- board on which both feet rest, which and paddlewheel-driven HPBs are also pine, or downhill; and Nordic, cross- prevents the problem of the two skis good examples of HPVs because the country or X-C. Alpine skis may be moving in different directions. Like Al- speed-changing or crank drive provides turned by rolling the ankles in the de- pine skis these are entirely gravity- a good impedance match. sired direction, which pulls the outer powered so are not HPVs. Ice skates attach to the feet and pro- edge off the snow and digs the inside Concluding this part, the only non- vide a low-friction way to glide over sol- edge into it. The skis are narrower in wheeled vehicles considered to be HPVs id water. Ice is characterized by a very the center which presents a curved edge are propeller-driven HPAs, propeller- or low coefficient of friction because the to the snow, causing the turn. Alpine paddlewheel-driven HPBs, rowboats and skate blade is normally lubricated with a skis may be stopped by snowplowing, rowed racing shells. Paddled boats did thin film of liquid water. This is accom- which is rolling both ankles inward, not qualify. Air and water propellers plished by using blade edges that are pointing the fronts of the skis toward could also be considered to be wheels, very narrow, resulting in a tiny contact each other, then forcing the feet apart, though the direction of vehicle travel area on the ice. The high pressure melts plowing up a vee-shaped ridge of snow. relative to the axis of wheel rotation is at the ice as the skate blade slips along. The skis may also be slid sideways the right angles to that of a conventional Foot placement in yaw, pointing left or same way as ice skates. Alpine skis are land wheeled vehicle. No wheel-less right, and roll, blade directly under or entirely gravity-powered so cannot be land, that is ice or snow, vehicles met off to either side of the centerline of the called HPVs, although potential stored the criteria. foot, purportedly determines the direc- energy of gravity may come from human WHEELED VEHICLES. tion of travel. The forward tips of the power, so if you walk up the hill you can Because earth and pavement are skate blades often have teeth, so accel- call your skis HPVs. characterized by high coefficients of eration and braking are done by raising Nordic skis are able to ascend gradu- friction, wheel-less vehicles will not the heel and digging the blade toe into al slopes due to the fishscale-shaped or slide easily on them as they will through the ice. When accelerating at higher stepped bottom surfaceconstruction. air, or on water, ice or snow, making speeds the skate blades may mark the ice This provides a one-way clutch against wheels necessary. Wheeled vehicles can in a narrow herringbone pattern, where the snow, so that they do not slide back- be divided into two categories: those the rider shifts weight from one foot to ward down gentle grades. Skiers use in- with power transmitted through one or the other near the end of the stroke, and ertia and poles to augment the uphill more of the wheels and those where all the forward component of the slight struggle until the grade gets too steep; the wheels idle, or are not driven. Un- angle serves to accelerate. An alterna- then they may ascend making marks in a driven wheeled vehicles are reviewed tive method of braking is toslide the herringbone pattern similar to that ex- first. skates sideways so that the blades shave plained in the ice-skating section, or Rollerblades use several round-edged off a thin layer of ice, quickly absorbing may step sideways, skis parallel, making wheels in tandem of about 50-mm (2") the rider's kinetic energy. Speed skaters marks like a caterpillar-tractor tread up diameter to give the rider the feel of ice can achieve high speeds, but long the slope. The skis may also be re- skates. Rollerblades are built onto boots strokes to bring the legs to coupling moved and carried on grades over about worn on the rider's feet. The path rol- speed with the ice at high velocities in- 0.5%. It is possible to steer these skis, lerblades make when accelerating is a dicate a poor impedance match, prevent- which are longer, narrower and less ta- narrow herringbone pattern as explained ing ice skates from being considered pered than Alpine skis, but most Nordic in the section on ice skates. There is no HPVs. skiing is done in snowmobile-prepared Cn rnntral is hvl i1tIiPqPtn l[ m^hsnizm~11,Lt1i)111, OJ ;ILtll 13 Uy tracks, grooves in the snow which func- picking, aiming and placing each foot in WHEEL-LESS LAND VEHICLES. tion like railroad tracks, eliminating the the desired direction of travel. Stopping Earth is generally rough and charac- need to learn the elusive steering tech- rollerblades is accomplished by rotating terized by a high coefficient of friction, nique. the ankles as with ice skates, but by rais- so most land vehicles include wheels to Like Alpine skis, stopping on Nordic ing the toes instead of the heels, the smooth the ride and change friction from skis may be accomplished by snowplow- brake pads being mounted in a safer sliding to rolling. An exception occurs ing, and sometimes by lifting and plow- location beneath the heels. A version of when the ground is covered with snow. ing only one ski out of the track, to rollerblades for summer use on ski After a little packing the coefficient of maintain directional control. An alterna- slopes uses a flat belt like a caterpillar friction of snow trends toward that of tive is to drag the baskets of the poles in tractor tread to offer more contact area ice. Snowshoes, toboggans, sleds, snow- the soft snow next to the tracks. Basket over the uneven surface. boards, and skis are favorite wintertime brake force may be increased by placing human accessories. the poles between the legs and using the

Human Power, fall & winter, 1991-2, vol. 9/3 & 9/4, p.15 vary from 120 to 300 mm (5 to 12") in diameter. Older machines had no brakes t . . d . t ... _ ...... _ but newer models use bicycle-type rim ,1M5>Y _ brakes on the rear and sometimes front wheels. I) o cPPoSt. N!D> Rollerblades, roller skates, skate- boards and sidewalk scooters are all characterized by a lack of drive to the

_ _ _ wheels, and because they have the same poor impedance match at speed, where coupling speed is a problem, they cannot be called HPVs.

WHEEL-DRIVEN VEHICLES. AWC LE Conventional wheelchairs use the rider's hands to drive two rear wheels of approximately 600-mm (24") diameter. The wheels are driven independently via drive rings slightly smaller than the tire diameter and mounted to the wheels.

TICOtV l I /A AaJ& I A ,'lg The wheels may be powered separately - I I _ _ - I - I _c1~T yluWI rl- Y \ to negotiate corners or in opposite direc- tions to turn the chair in place. Small Figure 2 Skateboard steering mechanism front wheels which caster freely are used to prevent forward tipover, but the rear Roller skates are attached to or Suspension Design," HP 7-3, (Spring wheels are set close to the center of mounted on boots or shoes and have two 1989). gravity of the rider to allow the chair to wheels each at the front and rear ends, In the 1960s Mickey Thompson be tipped back onto its rear wheels to each pair on a common axle. These run brought a four-wheel-steered race car to negotiate steps and curbs. Wheelchairs 30 to 50 mm (1 to 2") in diameter and Indianapolis, where the rear wheels are a fixed-gear arrangement and stop- are placed about 50-mm apart. In sim- steered outward as on skates. As prac- ping is done by gripping the drive rings ple designs there is no steering mecha- tice went on the rear steering was ad- or running the palms against them. In nism. The side-by-side wheel justed down to the point that it was addition crude brakes can be set once the configuration is designed to aid a young eliminated by race day. Only slow machine is stopped to prevent drifting on rider's balance. On more sophisticated earth-moving equipment use the concept slight inclines. designs the wheel pairs turn as the rider's now. Some modem cars have reintro- Racing wheelchairs use a smaller ankle is rolled in the direction of the duced the all-wheel-steer concept but drive ring to increase the effective gear turn, as illustrated in Figure 2. The for- they steer the rear wheels in the same and some use a single non-swiveling ward axle turns in the direction of the direction as the fronts [in some cars the front wheel which rises off the ground turn and the rear axle opposite, allowing rear wheels steer in the contrary direc- under power and is placed in the new for very-short-radius turning circles. tion at certain speeds - ed], and steer the desired direction of travel by careful use Braking is similar to ice skates, where rear wheels only a few degrees. of differential power input, steering the friction pads rest under the toes, ac- Braking a skateboard is accom- chair while the front wheel is airborne. cessed by raising the heels. plished by rotating about the rear axle, Wheelchairs, even the racing variety, do Skateboards may be gravity powered, raising the front axle off the road, until a not have an efficient impedance- where speeds over 25 m/s (60 mph) have brake pad on the rear edge of the board matching drive and have the coupling- been recorded, or one foot may control scrubs along the ground, or by rotating speed problem at high speeds, so cannot the board while the other pushes against the board in the same way but flipping be considered to be HPVs. the ground to incite motion. the board into the air and jumping off A type of Chinese tricycle uses an Skateboards use the roller-skate simultaneously, catching the board as interesting hand-crank drive. Three steering design, illustrated in Figure 2, the rider decelerates on foot. 700-mm- (27") -diameter wheels are ar- but with a wider track, or distance be- Sidewalk scooters are an old device ranged two at rear, both driving through tween the wheels, of 100 to 200 mm (4 which have seen a resurgence in recent a differential, and one forward, steered to 8"). Skateboard wheels are about 50 years. They help children ease into via a tiller. The frame supporting the mm in diameter. The steering mecha- skateboarding, and the balance learned front end is built around one side of the nism uses a high negative caster on the on scooters makes learning bicycle bal- chair while the other side is open for forward truck, where the steering pivot ance a very easy task. Scooters are com- mounting and dismounting. intersects the ground well behind the prised of two wheels with a platform A vertical shaft is mounted to the axle centerline, and a high positive cast- between for resting the feet. Power is frame near the front of the seat which er, intersect forward, on the aft truck. input by pushing off the ground as on a has an armnn and a handle thatis cranked This turns the axles inward at the front skateboard. The front wheel is turned by to make the machine go. The other hand and outward at the rear as the board is a fork fitted into bearings as on a con- steers the tiller and actuates a standard leaned into a turn. These terms are ex- ventional bicycle, including a handlebar bicycle brake lever. As shown in Figure plained in "Human-Powered Vehicle and on a long stem so the rider may steer 3, the vertical crankshaft has a bevel while standing. Scooter wheel sizes gear at the bottom that engages a similar p. 16 Human Power, fall & winter, 1991-2, vol. 9/3 & 9/4 -diameter rear wheel through a roller clutch 12-mm (1/2") inside diameter x CRPIX AA 16-mm (5/8") long. Each lever stroke advances the scooter 1.3 m (51") making for a gear number of 410 mm (16"), just right for a child. The Kick'N Go utilizes CRA ItK standard scooter steering for the 230-mm- (9") -diameter front wheel and a hand brake that pushes a metal pad &L 4C-S against the rear tire face. Sidewalk pedal cars are designed for small children and are safer than the E E-sLMEELS single-wheel-forward tricycles they graduate onto later. A pendant lever- crank mechanism is used to propel these popular cars. Adult versions using stan- 0c(U-s wr4 dard circular cranks or lever pedalling D PF~RL4l A L s FR4u-mECE have also been produced. A hand- powered variation was owned by the au- thor decades ago, but these have not - r 4 C_(AEAL = LAD been seen since. Because they have impedance-matching mechanisms, they Figure 3 Chinese hand-crank tricycle drive are HPVs. Sidewalk tricycles and the large-

- - bevel gear on a jackshaft that runs cross- 250-mm (10") long and connects to an wheeled "ordinaries" of the 19th century wise under the plate where the rider's idler sprocket 90 mm (3.5") from the have no discernable drive mechanism, feet rest. Two freewheels engage fulcrum. The drive is taken through 1/4" but the diameter of the drive wheel is sprockets on the jackshaft near the cen- (6.2 mm) pitch roller chain. One end of tailored to the leg length of the rider and ter of the trike. The freewheels are ar- the chain is fixed to the frame; it wraps the crank length is tailored to allow a ranged for opposite engagement, so that around the lever idler, runs over a comfortable leg stroke. The effective one engages and the other freewheels 14-tooth drive sprocket on the wheel gear number of the machine is the wheel when the crank is turned in either direc- axle, around another idler, then around a diameter. Children's tricycle wheels run tion. Two chain drives run from the third idler, and the other end is fixed to in a range of 220 to 300 mm (8 to 12"). jackshaft sprockets rearward to the dif- the frame, as illustrated in Figure 4. The Though there is no mechanism, there is ferential on the rear-wheel axle. One first idler acts to double the effective le- an effort to make the drive-wheel diame- chain runs in a simple loop to a sprocket ver arm, the second increases the chain ter match the power capability of the on the differential mechanism case. The wrap around the drive sprocket and the rider, so both may be considered HPVs. other chain runs to a different-diameter third reduces the extension of the spring, Track bicycles use a single-ratio sprocket on the differential case but the which stretches on the power stroke and step-up sprocket-and-chain drive to al- low use of a smaller drive wheel than chain runs from the bottom of the jack- returns the lever on the retract stroke. shaft sprocket over the top of the differ- The drive is coupled to a 180-mm- (7") the ordinary, making its impedance- ential sprocket, around an idler sprocket and back to the jackshaft sprocket. This reverses the drive direction, so that At4 w '4 cranking the handle in one direction en- C- gages one drive ratio and reversing the \\ ;ooT crank direction engages a different ratio. FooT Because the bevel-gear pair and chain drives may be arranged to produce a va- riety of ratios and it is a constant-drive machine, this Chinese tricycle is a good example of an HPV. A basic wheel-driven HPV is the Honda "Kick'N Go," which adds a drive mechanism to a conventional sidewalk ILEa scooter. This single-speed, lever- actuated, intermittent-drive machine was marketed by Honda motorcycles around E. SPC4>cKbr 1980. The drive lever, which may be pumped by either foot, is pivoted just ~E.-LiJfru CwLU4 forward of the rear wheel. The lever arc spfItw9^ TO rFly- is rearward from 15 degrees aft of verti- cal to 5 degrees above horizontal, for a Ft 'Rsr tDL.EJ sweep of 70 degrees. The lever is ,, ., _,, Figure 4 Honda "Kick'n-go" drive Human Power, fall & winter, 1991-2, vol. 9/3 & 9/4, p.17 front and rear trucks, which are identical on current designs. A skateboard could be modified to use the Kick'N Go drive, using three sets of gears. The rear axle could still lean and both wheels would be driven when going straight and a positraction-type drive would result.

Rob Price, 7378 S. Zephyr Way, Little- ton, Colorado, 80123 USA 303-973-6105

Rob is an airbornestructures staff en- gineer in the NASA Space Systems Group at Martin MariettaAstronautics Corporationin Denver. He designs in- CANOES PADDLEWiLEEPADDLEWYIIEEL L IMPEDANCE-MATCHIING KAYAKS PROPELLER lUMAN-POWERED stallationsof equipment in the Shuttle ~~~ROWBOATS~VEIIICLES ARE cargo bay and is mission integration sys- SCULLSR TSCULLS INDICATED IN ITALICS tems engineerfor the tethered satellite program. He has a B.S. in mechanical Figure 5 Human-propelledmachinery versus human-powered vehicles engineeringand is a member of the American Institute ofAeronautics and matching system more visible. Track Astronautics and, of course, the IHP VA. bikes also share with ordinaries and CONCLUSIONS. He has been designing and building sidewalk tricycles the ability to stop the This article has attempted to describe HPVs that utilize aluminum monocoque machine by reversing the pressure on the many of the devices people use as ve- constructionfor 12 years. pedals, so these machines are seldom hicles, to decide which are human- fitted with separate brakes. powered vehicles and to generalize the Standard bicycles are often fitted characteristics necessary for consider- CYCLING SCIENCE 3/3-4 with multiple-ratio gearing to allow opti- ation as an HPV. The vehicles re- mization of pedalling speed and power viewed, the categories they fit into and This double issue, dated September output over a very wide range of operat- whether they fit the impedance- and December, 1991, was the last under ing speeds and conditions. This is the matching drive definition are tabulated the editorship of Chester R. Kyle, co- ultimate in impedance matching but of- in Figure 5. founder of the IHPVA. (Cycling Sci- ten requires a gear chart to discern Examples of HPVs include propeller- ence will continue under another pub- which of up to 21 ratios is the next high- driven aircraft, prop- and paddlewheel- lisher, with Chet's continued er orlower. And finally, virtually all the driven boats, rowboats and racing shells. association). This issue has 64 pages, human-powered vehicles Human Power Interestingly, no winter accessories and included three articles by Chet Kyle readers build use complex impedance- passed the test. Among wheeled ve- himself (on alternative bicycle transmis- hicles the Honda Kick'N Go scooter, matching drive trains, so will be true sions, the effects of cross winds upon HPVs. pedal cars, sidewalk tricycles, the hand- time trials, and wind-tunnel tests of aero cranked wheelchair, ordinaries and con- ventional bicycles are included, but none bicycles). Matt Weaver has an article on of those wheeled vehicles where the his "Cutting Edge". There's a useful ar- wheels are no driven. ticle by Kohi Danh et al on frictional re- A final definition of an HPV is a ve- sistance in bicycle-wheel bearings, and hicle that carries its human power source another on the aerodynamics of bicycl- and moves, maneuvers and stops in a ing clothing by Brownlie et al. John controlled manner. Further, a true HPV Stegmann has four pages to himself to must be impedance matched to utilize present his view of front-wheel-drive human power efficiently, which is usual- recumbents. He thus competed directly ly accomplished via wheels and mecha- with Mike Eliasohn's long-announced nisms. article on the same topic in the last issue of Human Power (9/2), to which John EXTRAS OR THINGS TO TRY. contributed. I am not an unbiassed re- An easier-steering skateboard could viewer when a rival journal tries to be made by eliminating the steering ac- Eliasohn collected ar- tion of the rear truck while allowing it to scoop us. Mike lean relative to the board. This would ticles on many different FWD recum- be easier to learn on, because the steer- bents and reported on various points of ing effect would be about halved, but view, and I thought that his was the bet- Chinese tricicyle and customer being would require different tooling for the ter piece. weighed. Photo: Rob Price p. 18 Human Power, fall & winter, 1991-2, vol. 9/3 & 9/4

B=480 mm (19 inches), L=5.8m (19 from the Salish native dialect, meaning HPV-builders' workshop in Cambridge feet) (say). A cyclist, without paddle at "to fly" or "it flies". Pronunciation MA on the subject about six years ago). ready, may not employ the braces which stresses the second syllable, roughly would assist a conventional kayaker in a "kwok".] Very large reductions in drag coeffi- developed sea. The answer to this ap- Kawak most resembles an open- cient can be obtained by shaping a body pears to be outriggers. But these are not cockpit kayak including a daggerboard- so that it produces laminar, instead of outriggers that run in the water, generat- like trunk through which one inserts a turbulent, flow in the boundary layer - ing wetted area and drag. Rather, we drive unit for pedal drive. The feature the air right against the surface. Here is use low-volume (10-litre) outriggers, which defines the kawak (as a type of an example taken from the paper show- supported above the waterline, well aft boat) is this quick conversion (a few sec- ing the very large drag penalty from tur- of the cockpit. In the kawak design (see onds) between being pedal-powered and bulent flow. figure), this location keeps the outriggers being a more conventional kayak. In While one can use advanced com- out of the way while paddling with a shallows, in weed, or in uncertain wa- putational fluid mechanics to design en- conventional kayak paddle. The loca- ters, the kawak has all the versatility of a closures to produce laminar flow, in tion also minimizes occasions in sea kayak. Then, with a stretch of open wa- practice there are many differences be- states when splash from an outrigger is ter to cross, the kawak converts (in sec- tween the theoretical and actual condi- thrown onto the pilot. Ten-litre outrig- onds) to pedal drive, opening up greater tions. Thus being able to see where gers have prevented capsize in any sea speed and -- especially -- greater range transition to turbulence occurs gives the state yet encountered. However, the low possibilities. fairing builder the knowledge of where volume has proven useful in deliberate In proof-of-concept prototype, a ka- to change the shape or the roughness to capsizes when it is found that the time wak was assembled from a Valhalla preserve laminar flow. from full-over capsize until the boat is Surfski hull, using a SeaCycle drive unit. The advantages of liquid-crystal righted and re-entered is less than 20 Ongoing developments include both the coatings vs the traditional sublimating seconds (an important consideration in use of a more conventional kayak (nicer chemicals, china clay and oil coatings colder water.) With such added safety long-range touring features) as well as are the reversibility of the indications, margin, one feels assured in building a effort toward a boat built purely for rac- the greater clarity of the display, and the main hull with characteristics similar to ing boat. New drive units are being de- low toxicity. The surface is first painted a catamaran hull. veloped for the newer boats. matte black, and then a liquid-crystal Point (4) concerns length. As we've Experience to date with the prototype film is sprayed on. Colors indicate skin seen, A tends to increase as J. A kawak has been encouraging. We have friction. Refer to the full paper for de- crystals, monohull (with outriggers) may be 20% not made an athletic contest out of this. tails of where to get the liquid longer than a corresponding catamaran Rather, we have found that ordinary etc. (Reviewed by Dave Wilson). (5.8m vs 4.9m, 19 ' vs 16', say). Greater people can expect to sustain cruising length of a monohull implies about 10% speeds of 3 m/s while my assistant (46- area. However, at the years old, with no particular athletic penalty in wetted -. 8 Upper speeds that pedal craft operate, wave ability) peaks to 5 m/s. There is no -. 4 drag becomes increasingly important doubt even at prototype stage that stron- ger athletes would turn in higher per- relative to friction (wetted area). Wave C 0 Lower drag is a strongly increasing function of formances. But kawaks are for everyone Cp .4 from children through seniors, offering S/ L, where S is speed. Increased .8 comfortable day trips of 30 km (20 1.2 p I I I wetted area on account of L is more than miles) and more for people of very ordi- wave drag. Thus, Turbulent offset by decreased nary ability. (Athletic users can plan .006 lower drag the monohull may achieve daily trips in excess of 60 km, 40 miles.) than a catamaran at the higher speeds for Cf .004 Laminar which wave drag is signficant. Augustus Gast 962 Lovat Ave Victoria .002 A suggestion of the difference in V8X 1V3 Canada drag between catamaran and kayak-type 0 ~-t I I/ I I hulls can be seen in drag tests on a Sea- (Azugustus Gast is a consultantphysi- Cycle catamaran (Human Power 8 (4), cist). winter 90) and tests on six kayaks re-

ported in Sea Kayaker 3 (3, winter 86) I- and 3 (4, spring 87). Advances in flow visualization 0 .2 .4 .6 .8 1.0 Finally, it must be said that there are using liquid-crystal coatings X/C many reasons to build pedal-powered by Bruce J. Holmes boats. Considerations of speed and drag, and Clifford J. Obara though important, are only a few among (The following is from a NASA "techni- many considerations. Various other ar- cal supportpackage, LAR-14342 sent in ,kin-friction and pressure distributions guments may favor either the catamaran by Mark Bruce, New Canton, OH, USA. for a NACA 671-215 airfoil or the monohull. Let me close with a The full paper is SAE no. 8 7101 7. Bruce brief account of "kawak". [The word is Holmes of NASA Langley addressed an

------ p. 20 Human Power, fall & winter, 1991-2, vol. 9/3 & 9/4 Design for a wind+human-powered quadracycle by Wally Flint

- I This vehicle uses a pedal drive parent wind angle (awA) is large enough mechanism similar to a bicycle but aug- and if the lift-to-drag ratio is large ments the power with symmetrical air- enough then the net force produced by foils extending vertically from the the airfoil will have a forward compo- vehicle (with the leading edges pointing nent (although it will also have a much forward). Airfoils are more efficient than larger side component). Thus, the forces sails, permitting higher speeds, but sails on this vehicle are exactly analogous to provide greater power at low speeds. For the forces on a sailboat - the only differ- Airfoil-vehicle arrangement this application airfoils are preferable ences being that airfoils are used to gen- because we can use human power to erate the forces instead of sails and mm (0.6 feet) and an apparent wind reach speeds at which the airfoils be- wheel friction is used to resist the side speed of about 13.5 m/s (30 mph) then come effective. force instead of a centerboard. the Reynolds number Re= It should be possible to rotate the air- p[aw](chord)/g =187,000 where p is foils as a unit, so that the leading edges the density of air (1.24 kg/cu.m.) and .t can point to the right or left as in figure 5 2. Thus when awA becomes larger than is the absolute viscosity of air (1.79x10 2 the stall angle for the airfoils, one can Ns/m ). simply point the airfoils more in the The lift of a single airfoil is 2 direction of the apparent wind. L = C, 0.5 p area lawl where Cl is the For a given magnitude of apparent- lift coefficient and "area" is the area of wind, the smaller the awA the smaller the wing. From figure 3 the lift coeffi- vhick moon the forward component of lift. In the fol- cient is equal to 0.1 awA (awA in de- 2 iev of Airfoil lowing equations I will assume that awA grees). Then L = 0.062 awA area lawl . is less than the stall angle and therefore The actual lift will be less than that the airfoils will be pointed directly for- described by the above equation because Figure 1 Definitions ward. In this case the airfoil angle of at- of the close proximity of the other air- tack is equal to awA. The following foils. If two wings are separated by a In this article the standard equations formulas give the awA and the magni- distance of 1.5 times the chord length, that describe the lift and drag on the air- tude of the apparent wind vector (lawl): then the resulting lift generated by either foils can be found in any textbook on wing is equal to 0.92 times the lift pre- aerodynamics or wings. Figure 1 defines tw sin(twA) (2 ) awA=arc tan twcos(twA) (1) dicted by the above equation . This ap- the terms and angles used in computing iw+ttw cos(twA)] plies to two wings. I don't know what the propulsive force of the airfoils. "Ap- the correction factor is for three or four parent wind" (aw) is the vector sum of [aw] = twsin(twA) wings. These additional wings are a dis- [aw] sin awA (2) "true wind" (tw) and "induced wind" advantage in that they no doubt make (iw). "True wind" is the wind you feel the 0.92 correction factor even smaller. when the vehicle is at rest. "Induced Figure 3 gives data at low Reynolds numbers for the E-169 airfoil') and fig- There are, however, some advantages to wind" is the wind produced by the mo- having more than one wing. For one tion of the vehicle. Note that the ure 4 gives the airfoil coordinates. In our case if we assume a chord of about 200 thing the vortices from adjacent wing induced-wind vector always has the tips tend to cancel each other which same magnitude as the vehicle-velocity tends to lower the induced drag. Another vector, but is pointed in the opposite advantage is that the equations we will direction. "awA" (apparent wind angle) be using to calculate drag assume that denotes the angle between the apparent both wing tips are away from any ob- wind and the vehicle's path of motion; structions, but in our case the lower twA (true wind angle) the angle between wingtip is close to the ground and, in the true wind and the path of motion. addition, it may terminate into the body "L" is lift and "D" is drag. of the vehicle. Both of these factors tend To understand where the propulsive to further reduce the effect of wing-tip force comes from note that lift is by vortices. Thus I will assume that these definition the force which is at right effects will make up for the fact that the angles to the oncoming airstream 0.92 correction factor may be too large (apparent wind). Similarly, drag is by when using more than two wings. In definition the force which is in the same Figure 2 Rotatable airfoils addition, we will find that there is direction as the apparent wind. If the ap- enough room to separate the airfoils by a

Human Power, fall & winter, 1991-2, vol. 91/3 & 9/4, p.21

our equations are valid is that speed YU/T NR X/T YO/T which reduces the awA to 9 degrees. A 1 1.00000 0.00000 0. 00000 further increase in speed will result in 2 0.99893 0 00006 -0.00006 awA < 9. Beyond this speed we cannot 3 0.99572 0.00027 -0 00027 4 0.99039 0.00073 -0.00073 expect the forward force of the airfoils 5 0.98296 0.00147 -0.00147 to balance the backward force of the 6 0.97347 0.00248 -0.00248 wind resistance. 7 0.96194 0.00369 -0.00369 How often will we encounter "good 8 0.94844 0.00503 -0 00503 9 0.93301 0.00649 -0. 00649 wind conditions"? Look at figure 1. 10 0.91573 0.00808 -0.00808 Hold the wind speed (tw) and vehicle 11 0.89668 0 00986 -0. 00986 speed (iw) constant and 12 0.87592 0. 01187 -0.01187 then rotate the 13 0.85355 0.01411 -0.01411 tw vector through 360 degrees. At what 14 0.82967 0.01660 -0.01660 values of twA is awA < 9? Equations 9 15 0 80438 0 .01934 -0 01934 16 0 77779 0. 02231 -0.02231 and 10 (which were derived by applying 17 0. 75000 0.02552 -0 02552 the law of sines to figure 1) help us to 18 0 72114 0. 02893 -0.02893 answer this question. The answer de- 19 0 69134 0.03252 -0. 03252 pends on the relative magnitudes 20 0.66072 0.03626 -0.03626 of tw 21 0 62941 0. 04011 -0.04011 and iw. There are three situations which 22 0.59755 0. 04401 -0. 04401 occur. 23 0 56526 0.04793 -0. 04793 24 0 53270 0.05179 -0. 05179 25 0.50000 0. 05553 -0.05553 l.iw< tw 26 0.46730 0. 05908 -0 05908 When iw < tw, there will be a range 27 0.43474 0 06238 -0. 06238 of values of twA, symmetrical about 2G 0.40245 0.06533 -0.06533 29 0.37059 0. 06786 -0.06786 twA = 0, such that whenever twA is 30 0.33928 0.06967 -0.06987 within this range the magnitude of awA 31 0.30866 0. 07127 -0.07127 will be less than 9 degrees. This range of 32 0.27886 0 07196 -0.07196 33 0.25000 0 07189 -0.07189 values of twA (in degrees) is given by 34 0.22221 0 07107 -0.07107 equation 9 with awA,,) = 9 degrees. 35 0.19562 0. 06952 -0.06952 Equation 10 36 0.17033 0 06723 -0.06723 (also using awA,) = 9) 37 0.14645 0. 06422 -0 .06422 gives a negative answer when iw iw > tw. 43 0.03806 0 03441 -0.03441 In this case there will two ranges, 44 0.02653 0.02821 -0. 02821 45 0.01704 0 02250 -0.02250 one symmetrical about twA = 0 as be- 46 0.00961 0.01604 -0.01604 fore, and another symmetrical about twA 47 0.00428 0.01011 -0. 01011 = 180. Equation 9 gives the range sym- 48 0.00107 0 00500 -0.00500 49 -0.00000 0 00000 0.00000 metrical about twA = 0. Equation 10 gives the range symmetrical about DICKE/T...= 0.144 RUECKLAGE/T= 0.279 twA = 180. WOELBUNG/T= 0.000 RUECKLAGE/T= 0.001 PROFILTIEFE...= T 3.iw>(tw/sin(9)) In this situation awA < 9 for all val- ues of twA; the two ranges have merged Figure 4 Profile E 169 coordinatesfromref: 1 with each other and equations 9 and 10 have no solution. wide. This gives us room to space the twA range = 2 arc sin(iwsin(awA)/tw) Later I will show that the top speed airfoils about 600 mm apart. The 0.92 + 2 awA (9) of the vehicle is maximized when span = correction factor included in equation 3 twA range = 2 arc sin(iw 2.18 m and chord = 178 mm. For now assumed the airfoils were separated by a sin(awAm.)/tw) - 2 awAm. (10) simply note that these values satisfy distance equal to 1.5 x chord, but here equation 8. At the local U-haul rental we have separated them by a distance of Let us consider a wind speed of 6.7 place I found out that a small U-haul 3.4 x chord. m/s (15 mph) and a vehicle speed of truck requires 2.64 m clearance. If we Equation 8 seems to state that the 13.4 m/s (30 mph). Equation 9 states assume that the lower tips of the airfoils airfoils will supply a forward force equal that the range symmetrical about twA = are going to be about 300 mm off the to the backward force of wind resistance 2 0 is 54.5 degrees. Equation 10 states that ground, then a span of 2.18 m would at any vehicle speed (any value of lawl ). the range symmetrical about twA = 180 give us a vehicle that is not quite as high But remember that as the speed of the is 18.5 degrees. If the direction of the as the small truck. What about the vehicle increases, the awA decreases. road is random then we will encounter a width? My Ford F-150 pickup is 1.93 m The maximum vehicle speed for which twA ____fallino __into _ one___ of__ these regions Human Power, fall & winter, 1991-2, vol. 9/3 & 9/4, p.23 ((54.5+18.5)/360) = 20% of the time. Perhaps it would be better to say how OF TIME PROPULSIVE FORCE often we will encounter a value of twA PERCENTAGE giving awA greater than 9 instead of less BALANCES WIND RESISTANCE than 9. This would give us a rough evaluation of how often we can expect to encounter "good wind conditions". vehicle speed (mph) The accompanying table presents this tw 15 20 25 30 40 figure at various wind speeds and ve- (mph) hicle speeds. The table should be read as 5 69% 57% 43% 22% 0% in the following example (using the 10 85% 80% 74% 69% 57% wind speed = 15 mph row and the ve- hicle speed = 40 mph column): when the 15 90% 87% 83% 80% 73% wind is 15 mph, we should be able to 20 91% 90% 87% 85% 80% drive at 40 mph 73% of the time. (This assumes that the driver provides only 25 92% 91% 90% 88% 84% enough power to overcome the rolling resistance of the vehicle.) At awA = 0 the airfoils do nothing wind resistance at any vehicle speed and foil, the component of lift that points di- but increase the drag. It's easy to figure rectly to the side is L cos(awA) (see out how often the airfoils contribute any wind speed as long as awA 9. figure 1). If the lower tips of the airfoils some forward force (NFF > 0). Recalling Thus, given a fixed true wind speed and are 300 mm (1 foot) off the ground then equation 6 for the NFF of a single air- direction, the vehicle speed is limited to an integration of the torque at each foil: the value which causes awA to equal 9. Applying the law of sines to figure 1 height will show that the net torque pro- duced by all four airfoils is 5.578L NFF =[(0.000998 awA2 span chord) - gives equation 11, which tells us what cos(awA). We write 5.578L cos(awA) = (0.00217 awA2 chord2)-0.0155 span the vehicle speed will be when awA 858.7 to get the maximum allowable L. chord)] [aw2 ] (6) reaches 9 degrees. speed limit = iw = tw (sin(twA-9)/sin(9)) In the equation 2 (1) L = 0.057 awA span chord awl , NFF is positive when the expression The speed is also limited to the value the value of awA will always be 9 de- in the brackets is positive. Putting span which causes the lift to become so large grees. This is because awA is really = 2.18m and chord = 0.178m into that that it causes the vehicle to topple over. angle of attack in this equation. Since expression gives: awA will be greater than 9, the airfoils 2 The vehicle will begin to topple over 0.0003185 awA - 0.006015 . when the torque of the airfoils about the will be turned into the wind until the The minimum value of awA for which wheels on one side is equal to the torque angle of attack is equal to 9. (If awA this expression is positive is awA = 4.35 due to the weight of the rider and ve- were less than 9 then the speed would degrees. Consider a typical bicycle hicle (see figure 5). have already been limited by equation speed of 6.7 m/s (15 mph) and a wind If the vehicle and rider weigh 890N 11.) But we cannot use awA = 9 when it speed of 2.2 m/s (5 mph). Equation 9 (200 pounds) the torque will be shows up in other places, such as with awA = 4.35 gives 35 degrees; 3.0.963x890 = 859 N-m. For a given air- equation 10 gives 17.6 degrees. We ex- pect to encounter awA<4.35 about torque torque component ((35+17.6)/360) = 15% of the time. f airfoil Thus, with a wind of 2.2 m/s (5 mph), of veight W cos( theta ) we can expect to drive 6.7 m/s (15 mph) veight 85% of the time without the airfoils add- theta ing additional drag to the vehicle. Suppose we are going directly into (distance from the wind (awA = 0). How much is the y 0.3 ground to driver) drag increased by the airfoils? Putting awA = 0 into equation 6 (and remember- 0.965 (distance from point vhere ing to multiply by 4) we get: wheel touches road to driver) 4 NFF= -0.00222 Iawl2. Dividing this by the equation for wind resistance x (equation 7) shows that the wind re- f2.48 sistance will be increased by 30% at torque for airfoil located - L cos(avA) awA = 0. a distance of x feet from wheel 2 18 y dy Now let's address the issue of speed. The above analysis shows that the torque exerted about a Vheel by an airfoil is 0.3 We know from equation 8 that the air- not a function of the distance between the airfoil and the vheel foils will at least counterbalance the Figure 5 Topplingforces on wind-powered vehicle p. 24 Human Power, fall & winter, 1991-2, vol. 9/3 & 9/4

TABLE 1: VEHICLE ROAD-LOAD DATA VEHICLE F'RONTAL ROLLING AERODYNAMIC TOTAL CONFIGURA ,TION A\REA RESISTANCE RESISTANCE MASS achievable on the Mk 4 machine de- 2 \ m CR CD M kg scribed above. The bare-HPV drag coef- ficient of 0.92 for the Mk 3 prototype 1 Bare HPV 0.45 0.006 0.92 98 deter- agreed well with the value of 0.91 2 HPV + Front 0.54 0.006 0.51 107 mined in 1/5th-scale wind-tunnel tests Fairing only (Amor, 1989) of a geometrically similar model. 3 HPV + Full 0.55 0.006 0.67 111 THE VEHICLE ENERGY- Fairing CONSUMPTION SIMULATION 4 Touring Cyclist 0.53 0.004 1.00 89 Driving-cycle energy consumption for the HPV was simulated using a ve- 5 HPV - Mk 4 0.54 0.006 0.40 105 hicle energy-consumption model (Raine Development and Epps, 1988) developed for use in Note: 1. In each case the rider was a male of 188 cm height and 78 kg weight. conjunction with the University of Can- 2. To the 20 kg bare mass, the frontal half fairing added 7 kg and the full terbury methanol-fuel research pro- fairing 13 kg. 2 kg extra mass is present as rear panniers on HPV 2. gramme. 3. The future Tricanter, with full aerodynamic fairing, is projected to The computer model, written in Wa- achieve a drag coefficient of 0.4, with 17 kg bare vehicle mass and terloo Fortran 87 and run on an IBM 10 kg aerodynamic fairing. PC/AT, steps through standard fuel- - I consumption and emissions driving cal journey from Amor's residence to the sumption for the bare HPV is the datum cycles, calculating the required road- University. The cycle is over almost against which other vehicle configura- load power at one-second intervals. The flat terrain, about 20% of the distance tions are compared. road-load simulation can include con- being through city streets (office Table 3 shows that the HPV with stant, V- and V2 -dependent coefficients blocks), and the remainder through sub- only the front half fairing gives a sub- of rolling resistance, and builds in head urban areas (low-density low-rise resi- stantial reduction in CE for all three or tail wind. It can also model the ef- dential buildings). Ten sets of traffic driving cycles. It is also clear that the fects of driver error in following the lights are negotiated. The cycle was re- extra mass of the full aerodynamic body demand-velocity-versus-time path. corded using a video camera mounted on is unacceptable unless accompanied by a For the purposes of the present work the Tricanter HPV to log speed and time large drop in drag coefficient. The bi- the computer model was run using the from on-board instruments. cycle is seen to consume more energy simplified road-load power equation 2, Our objective in the HPV energy- than the bare HPV, but ridden in a full and assumed zero driver error in the fol- consumption modelling was to compare racing crouch, with reduced CD and fron- lowing of driving cycles. The total ener- CE for different configurations of the tal area, A, it would have a slightly low- gy consumption over a driving cycle, the Tricanter and for a touring bicycle. A er CE than the bare HPV. The projected cycle energy, CE, is determined by tak- further aim was to look at the effect of Mk 4 Tricanter, 15% lighter, with a 13% ing the following sum: road gradient and head wind. lighter full aerodynamic body (CD =0.4), achieves better than 22% lower CE than CE = cyceP.t (3) RESULTS AND DISCUSSION the bare HPV over the Christchurch where t = 1-second interval. Results are given in table 3 for the Cycle. Note: If P becomes less than zero at computed reference cycle energy, CE, The results in table 3 also reflect the any time under deceleration, it is set to for each of the five vehicles, on a flat average speed of the respective cycles. zero for the calculation in equation 3. road with no ambient wind. Energy con- Inertial energy consumption is relatively Using simple menu commands, the operator may change any of the vari- ables in the road-load equation, and may TABLE 2: DRIVING-CYCLE DATA also do perturbation studies in which the sensitivity of the cycle energy consump- DRIVING CYCLE tion (CE) to changes in individual pa- rameters is evaluated. The model also PARAMETERCHRISTCHURCH MODIFIED MODIFIED gives a breakdown of the percentage of BICYCLE AS2077 AS2077 CE used in overcoming inertia, rolling URBAN HIGHWAY resistance, aerodynamic resistance and Distance km 6.67 5.23 5.61 gradient. The model was run with three driv- Cycle Time sec 1200 1256 766 ing cycles, for which data are given in Max Speed km/hr 30.0 31.4 32.7 table 2. The modified AS 2077 (1979) urban and highway motor-vehicle driv- Av. Speed km/hr 20.0 15.0 26.4 ing cycles, and the Christchurch bicycle- Note: Speeds in AS 2077 Urban Cycle are reduced by a factor of 0.56, . urban-commuter driving cycle are illus- and in AS 2077 Highway by a factor of 0.34 to achieve the modified trated in figure 6. The most relevant is form. AS 2077 Urban also has the period of the cycle from t = 198 the last of these, which represents a typi- sec to t = 316 sec removed to avoid unrealistically high speeds.

Human Power, fall & winter, 1991-2, vol. 9/3 & 9/4, p.29 greater in the two urban cycles, whilst aerodynamic drag is more important in MODIFIED AS2077 URBAN CYCLE the AS 2077 highway cycle which is representative of energy consumption for cycle touring on the open road. The proportions of cycle energy con- sumed in inertial acceleration, rolling, E and aerodynamic resistances are shown for the flat road/still air case in table 4. We were particularly interested in study- ing the effect of gradient and head-wind perturbations to the model when decid- Ca ing whether to include an aerodynamic body on the Tricanter for general use. To explore gradient, the model was run with constant gradients ranging from 0% to 10%. The gradient effect is very 0 200 400 600 800 1000 1200 strong. CE values relative to CE I are TIME t seconds plotted against gradient in figure 7. CE 1 is the Cycle Energy for vehicle 1, at gra- dients G%, given by 2 2 MODIFIED AS2077 HIGHWAY CYCLE CEI = 138.7{1 +46.2G/(100 +G )' kJ (4) It is seen that the CE advantage held by the (necessarily heavier) vehicles with aerodynamic fairings is eroded as gradient increases, due to the extra work to be done against gravity. At very steep gradients, the gravitational load domi- nates, so that CE relates quite closely to all-up mass. At a gradient of about 2.8% the bi- LX InW 0- cycle, vehicle 4, starts to undercut the bare HPV, vehicle 1, which in turn achieves a lower CE than vehicle 2 with the front half fairing for gradients over

0 200 400 600 800 1000 1200 3.4%. Vehicle 5, the future develop-

TIME t seconds ment of the HPV, maintains an advan- tage in CE until a gradient of about 5% is reached, where the bicycle becomes the most energy-efficient machine. The CHRISTCHURCH BICYCLE CYCLE S;R bicycle, being lightest, comes into its own on climbing steep hills. For each of the five vehicles, the 30 computer model was run with head

25 winds, Vw, ranging from 1 to 20 km/hr E (0.278 to 5.56 m/s). CE values are

20 plotted against head-wind speed for the Christchurch bicycle cycle in figure 8,

15 relative to CE for the bare HPV in still Cx air. Figure 8 shows that the advantage LU - 10 enjoyed by the HPV with the front half Ln fairing gradually increases as the head 5 wind increases. The cyclist riding with straight arms on a sports bicycle is at an 0 increasing disadvantage. Predictably, 0 200 400 600 800 1000 1200 the vehicle with best aerodynamic-drag TIME t seconds characteristics copes best with the head wind. The comparisons made in figures 7 Figure 6 Vehicle energy consumption in driving cycles and 8 make no reference to duration of p. 30 Human Power, fall & winter, 1991-2, vol. 9/3 & 9/4 TABLE 3: PERCENTAGE CHANGE IN ENERGY human power output at various power CONSUMPTION, CE, COMPARED WITH BARE HPV levels, or to vehicle drive-train losses. Kyle and Caiozzo (1981) give a table of transmission efficiencies from which ENERGY CONSUMPTION DRIVING CYCLE 98% may be taken as an average figure VEHICLE to apply here. Kyle and Caiozzo (1981) CONFIGURATION CHRISTCHURCH MODIFIED MODIFIED also graph duration of human power out- BICYCLE AS 2077 AS2077 put for tests on a number of average rec- URBAN HIGHWAY reational cyclists ranging in age from 20 to 47 years old. Standard and modified Reference CE for bicycle ergometers were used. The dura- Bare HPV 138.70 kJ 120.82 kJ 120.70 kJ tion curve for an average of five cyclists 1 Bare HPV 0% 0% 0% is reproduced in figure 9. The average cyclist can maintain about 0.27 kW over 2 HPV + Front -14.37% -8.14% -21.20% the 20-minute duration of the Christ- Fairing only church bicycle cycle, ignoring speed 3 HPV + Full -0.87% +2.89% -4.66% variations which will make the cycle Fairing more arduous in reality. For the bare HPV driven over the Christchurch cycle 4 Touring Cyclist + 10.03% +5.12% + 15.59% on a flat road in still air, the CE of 138.7 5 HPV- Future -22.40% -14.76% -31.59% kJ corresponds to an average power dis- Development sipation at the road of 0.1 16 kW, or 0.1 18 kW allowing for drive-train losses. In figure 7, an average power of 0.27 kW over the Christchurch bicycle cycle TABLE 4: PROPORTIONS OF BASIC CYCLE ENERGY corresponds to a gradient of 2.8% (at CONSUMPTION, CE, CONSUMED BY VARIOUS which peak power from rider = 0.43 kW COMPONENTS OF VEHICLE ROAD LOAD, percent. at 30 km/hr) or a relative CE of 2.29 for the bare HPV. Other test data (Whitt & Wilson, 1982) indicate that a first-class athlete might average about 0.38 kW over the same 20-minute cycle. RESISTANCE: ENERGY CONSUMPTION Limits on duration of human power I = INERTIAL DRIVING CYCLE output mean that comparisons of CE for R = ROLLINGR ROLLING CHRIST- MODIFIED MODIFIED different vehicles are probably relevant A = AERODYNAMIC CHURCH AS 2077 AS2077 only for gradients less than about 3.5% BICYCLE URBAN HIGHWAY for the conditioned cyclist. Whilst gra- dients up to 3.5% tend to level the per- VEHICLE % % % formance of vehicles 1, 2 and 4, a light CONFIGURATION CE CE CE vehicle with good aerodynamics such as 1 Bare HPV I 19.71 39.59 4.01 vehicle 5 will still be more energy effi- R 26.69 21.26 27.08 cient. The light bicycle remains the ulti- A 53.60 39.15 68.91 mate steep-road HPV. The maximum head wind of 20 2 HPV + Front I 30.69 49.17 6.95 km/hr corresponds to an average power Fairing only R 30.43 23.48 35.49 output from the rider of the bare HPV of A 38.88 27.35 57.55 just under 0.27 kW. It would therefore 3 HPV + Full I 23.96 44.72 5.27 appear that all vehicles except the bi- Fairing R 28.69 22.09 30.67 cycle, vehicle 4, could complete the A 47.35 33.19 64.06 Christchurch cycle into a 20 km/hr head wind, but with a maximum road speed 4 Touring Cyclist I 16.29 33.72 2.84 touching 30 km/hr, peak power output R 21.24 17.94 20.56 from the rider of the bare HPV would be A 62.47 48.34 76.60 about 0.47 kW. i.e. higher speeds during 5 HPV- Future I 35.51 52.91 8.74 the cycle would test the stamina of the Development R 32.02 24.41 39.72 rider. A 32.47 22.68 51.54 Conclusions above regarding per- formance of the different vehicles ex- ecuting the driving cycle in a head wind or on an incline must be treated as in- dicative only, in view of speed varia-

Human Power, fall & winter, 1991-2, vol. 9/3 & 9/4, p.31 tions during the cycle and limits on I I.e duration of human power output.

A MOTORISED TRICANTER? L U The Mk 1 Tricanter prototype was I run with a 49-cc Velo Solex moped en- U 0:(2 gine mounted over the 27-inch rear z wheel. The high position of the engine W resulted in some sway on corners, but U U good straight-line performance. Addi- L.J tion of a similar wheel-friction-drive UJ motor over a larger-section 20-inch tyre ck on the rear wheel of the Mk 3 Tricanter could give a very workable vehicle with good handling and braking combined

with a cruising speed of 30 kmn/hr. A ]) 49cc-engined version of the vehicle GRADIENT % could afford the bulk of a more refined aerodynamic body, and enjoy the sim- pler vehicle licensing and design rules attaching to under-50cc vehicles.

CONCLUSIONS The Tricanter commuter HPV devel- opment project has led to a compact and robust recumbent tricycle with excellent cornering and braking characteristics. Road testing and computer simulation of the vehicle in different aerodynamic configurations have shown that the bare vehicle should have slightly lower ener- gy consumption than a low-handlebar sports bicycle ridden in a straight-arm U 2 4 6 8 I0 12 14 16 18 20 HEAD WINrD SPEED

i- - - l l i l - reduce energy consumption over typical 1.2 I [ -- T driving cycles provided that the aerody- namic drag coefficient is low enough to offset the effect of added body mass. Vehicles with aerodynamic bodies show a predictable advantage in head winds. 0: On inclines, the heavier more- aerodynamic vehicles suffer, but can re- 0.8 _...... ,...... tain an advantage over the bare vehicle D(- if they are low enough in CDA. For ex- 0 ample, a vehicle of total mass 29 kg, 0.6 LUJ with 78-kg rider and C DA less than 0.28 L 2 m would consume less energy than an 0

11-kg bicycle for the Christchurch bi- zD 0.4 _ ...... ,...... ---...A>- · cycle driving cycle executed on gradi- :

ents up to about 3.5%. I

ACKNOWLEDGEMENTS 0.2 The authors gratefully acknowledge the contribution of the following stu-

dents who carried out prototype develop- n l I I ii -- I I i i l Ii I . I ment and testing: K.C. Brand, L.B. 10--1 100 101 Crockett, S.J. Howman, G.D. Irving, P.M. Kelly, D.N. Lovegrove, P.J. DURATION OF POWER OUTPUT - MINUTES McMillan, P.M. Wilkinson. Figure 9 DuIration of human power output (after Kyle & Caio-zo. 1981) p. 32 Human Power, fall & winter, 1991-2, vol. 9/3 & 9/4 REFERENCES MODELLING, Proceedings of IPENZ most all the races they entered, first in Amor, M.R. (1989) ENERGY CON- Annual Conference, New Plymouth, Europe and then at the IHPSC in the SUMPTION MODELLING FOR HPV, N.Z., February 1988, vol. 2, pp 323 - USA. Ballantine's Avatar thereby ended Bachelor of Engineering Project Report 334. the supremacy of the Vector recumbent No.2, Department of Mechanical Engi- SAE J1263 (1980) ROAD LOAD tricycles in the 200-m speed champion- neering, University of Canterbury, MEASUREMENT AND DYNAMOME- ships and in other events. In his maga- Christchurch, N.Z. TER SIMULATION USING COAST- zine and his books, Richard Ballantine AS2077 (1979) METHODS OF TEST DOWN TECHNIQUES", SAE has spread his enthusiasm for HPVs as FOR FUEL CONSUMPTION OF PAS- Recommended Practice, Society of Au- well as for bicycles, and to him must be SENGER CARS AND THEIR DERIV- tomotive Engineers, Inc., 400 Common- given a large part of the credit for the ATIVES, Standards Assn. of Australia. wealth Drive, Warrendale PA 15096, new sport of HPV racing in Europe and Damam, Stoudt, McFarland (1966) THE USA. elsewhere. HUMAN BODY IN EQUIPMENT DE- Whitt, F.R. and Wilson, D.G. (1982) BI- In the "Ultimate Bicycle Book" SIGN, Harvard University Press. CYCLING SCIENCE, Cambridge, Mas- HPVs are an integral part from the cover Gross, A.C., Kyle, C.R., and Malewicki, sachusetts, The MIT Press. flap on. It is the most beautifully illus- D.J. (1983) THE AERODYNAMICS trated bicycle book on the market, with OF HUMAN-POWERED LAND VE- John K Raine, School of Engineering, photographs that have to be described as HICLES, Scientific American, vol. 249, University of Canterbury, Christchurvh, stunning (the photography is by Philip no. 6. New Zealand, FAX 64 3 642- 705 phone Gatward, and appears to be almost en- Hindin, P.J. and Raine J.K.(1986) VE- 667-001 tirely specially done for the book). The HICLE TESTING WITH ALTERNA- treatment of the various types of bi- TIVE FUELS - HOW AND WITH RICHARDS' ULTIMATE cycles for different types of racing and WHAT?, Proceedings of 1986 Interna- BICYCLE BOOK recreation is also broader than I have tional Conference of the Society of Au- A book review by Dave Wilson ever seen: it is almost encyclopaedic. I tomotive Engineers - Australasia Inc., have never been able to find a good defi- Auckland, N.Z., 20 - 26 March 1986, pp Richard Ballantine has been joined nition of criterium racing, for instance, 352 - 380. by Richard Grant in the lavish produc- but it is given a two-page spread here, IHPVA (1986), Proceedings of the 3rd tion of the latest edition of his book on with more on time-trial bikes, on the International Human-Powered Vehicle bicycles and bicycling, thereby changing Tour-de-France bikes, and much on Scientific Symposium and 12th Annual the position of the apostrophe in the mountain bikes. The material is up to Human-Powered Speed Championships, title. Richard Ballantine produced his date, with good authoritative informa- IHPVA, Expo 86, Vancouver, Canada, "Richard's Bicycle Book" in 1972, and, tion on frame and wheel design and ma- 28 - 29 August 1986. according to a note on the cover, it has terials. HPVs have far less than Lovegrove (1991) OPTIMISATION since sold over one-million copies exhaustive treatment, but there are three AND DEVELOPMENT OF THE HU- world-wide. That is strong testimony short sections. There is more than in MAN POWERED VEHICLE, Bachelor both to the quality of his writing and to any other general book on bicycles. of Engineering Project Report No.66, the interest in bicycling. The only sour note to mar this paeon Department of Mechanical Engineering, Ballantine has also done much for of praise is that I wish the authors had University of Canterbury, Christchurch, the HPV movement. I first met him asked a bicycle historian to check the N.Z. over ten years ago in Germany where we section on bicycle evolution. They re- Kyle, Chester R., and Caiozzo, Vincent were attending a bicycling conference in peat the oft-debunked myth about the J. (1981) EXPERIMENTS IN HUMAN Bremen. We appeared on German TV. unsteerable two-wheeled "celerifere" ERGOMETRY, Proceedings of the Ist I had brought with me an Avatar 2000 supposedly invented by de Sivrac. International Scientific Symposium on which I demonstrated to, apparently, There was no such vehicle. (I confess Human-Powered Vehicles, Anaheim, good effect. Several were later bought that I have helped in some measure to California, USA., November 1981. by various people in Europe, and small perpetuate the myth of the existence of McMillan, P.J. (1988) THE DESIGN, companies were started to produce re- any form of unsteerable "bicycle" in Bi- CONSTRUCTION & TESTING OF AN cumbents with strong similarities. Ri- cycling Science. And I did have my AERODYNAMIC FAIRING FOR THE chard Ballantine bought one for himself. chapter checked by a bicycle historian). HPV, Bachelor of Engineering Project He had it reviewed for his new magazine Also, the name of the rider, Faure, of the Report No.56, Department of Mechani- BICYCLE! This review was, I believe, record-breaking Velocar recumbent of cal Engineering, University of Canterbu- the longest, most thorough, and most en- the thirties is misspelled. ry, Christchurch, N.Z. thusiastic ever devoted to a new bicycle. In other respects, this beautiful book Raine J.K., Crockett, L.B., Irving, G.D. The entire color cover of the issue was is a delight. It is informative and enter- and Wilkinson, P.M. (1986) DESIGN given to the Avatar and a very striking taining. At $29.95 it could be regarded TO ENJOY - A HUMAN-POWERED model. Ballantine's Avatar was bor- as a bargain. It is bound to lead to great- VEHICLE, Paper presented at the rowed by Derek Henden in London who er acceptance of HPVs. For this alone IPENZ Annual Conference, Auckland, designed a fairing for it, renamed it the we owe Richard Ballantine and his co- N.Z., February, 1986. Nosey Ferret and, later, Bluebell, and, author Richard Grant (a rival bicycle- Raine, J.K. and Epps, I.K. (1988) VE- with a vacationing Australian lawyer, magazine publisher) a debt of gratitude. HICLE ENERGY CONSUMPTION Tim Gartside, riding began winning al- Dave Wilson

Human Power, fall & winter, 1991-2, vol. 9/3 & 9/4, p.33 THE SECOND INTERNATIONAL HUMAN- these rules the surface buoys created the POWERED-SUBMARINE RACE most problems. FROM THE PERSPECTIVE OF A SeaDAWG Description of entries There were many different hydrody- by Cory Brandt namic theories put to use. Some, like Two years ago seventeen teams (of course, as well as having a separate tank the Borti I from Berlin, attempted to the nineteen teams that registered) got for ballast. The air pressure was mea- make their submarine as small as possi- together to race submarines around a sured just before and just after running ble. By making the submarine small 1,000-meter kidney-bean-shaped course. the course. If there was not enough ex- they decreased the wetted-surface area Due to logistics, foul weather, and tech- cess the competitors were to be disquali- and the frontal area, which reduce the nical problems, only a 100-meter fied. drag. Others made their hulls a little straight-line sprint was run. Only nine Each submarine was allowed one larger but strove to eliminate boundary- of the seventeen teams arrived able to practice run down a 100-meter straight- layer separation (such as the SeaDAWG) finish. line course. A timed 100-meter sprint or to have a mostly laminar-flow body This year the weather cooperated. was run and the time achieved deter- (such as Team Wahoo). Thirty-four of the 36 submarines that mined the seeding. The fastest eight Most teams strove for efficiency in entered were in West Palm Beach, Flori- submarines were given a by for the first the propulsion systems. The majority of

da ready to race. - - The races are organized by the H.A. Perry Foundation and are intended to spark an interest in ocean engineering and to foster advances in hydrodynam- ics, propulsion, and life support for sub- sea vehicles. The entries came from large corporations, small businesses, universities, and even garage tinkerers. The teams flew from as far as Berlin, Germany and drove from as far as Van- couver, British Columbia. In the fall of 1990, I and thirteen oth- er University-of-Washington mechanical-engineering students got to- gether and decided to work on an entry for the 1991 race. The result was the SeaDAWG (figure ). It was quickly discovered that designing and building a submarine is a very time-consuming pro- cess. Like many of the teams down in Figure 1 The SeaDA WG Florida we had the occasional all-night round. If a submarine was not able to competitors used a standard bicycle work party. complete the 100-meter sprint in under crank, but some used linear drive sys- ten minutes, it was disqualified and tems. Team Wahoo was the only team General rules and course description thereby eliminated from all further rac- to use hand-and-foot cranks. The submarines were required to ing. The majority of the entries used two- have two occupants: a pilot and a pro- Initially two submarines were to race bladed propellers. Team Effort used a pulsor (the "stoker"). The pilot was not side-by-side going around a 400-meter 12-bladed propeller. Using hydraulics permitted to input any power that would oval course twice. However, due in part their pilot could control the blades and contribute to the forward momentum of to fear for diver safety, the 800-meter steer the submarine. The SubHuman II the submarine. He (or in a couple of race was shortened to 475 meters for all team was one of about a half-dozen cases she) was in charge of steering, except the final race. teams to use counter-rotating propellers. navigation, trim and ballast, and life Rules for safety were fairly stringent. support. The stoker's job was to be the Each submarine had to have a surface Non-standard propulsion engine: he was not permitted to engage buoy, an emergency-release buoy with a Five of the submarines broke away in any other activities. deadman switch for both occupants (thus from using propellers and went with Both occupants had to be fully en- if the switch were released the deadman something completely different. closed, and the submarines were re- buoy would surface alerting surface The U.S. Navy's Subdue used a ra- quired to be fully flooded. This meant crews of a problem), strobe light dially extended linear impeller (paddle that the occupants had to use SCUBA mounted on the submarine, and a spare wheel). It was not very reliable and ac- gear. Each submarine had to have 150% supply of air on the person of both occu- cording to a member of the Navy's team of the amount of air necessary to run the pants (most teams used Spare Aires). Of it couldn't keep up with the Squid (the p. 34 Human Power, fall & winter, 1991-2, vol. 9/3 & 9/4

Life support, trim and ballast Adjusting the ballast during the race is not a major concern. An 800-meter race will last less than ten minutes for a competitive boat. In that time the team should not need to make major buoyan- cy changes. We used a single ballast tank in conjunction with a sliding weight system (to adjust the trim). This worked well. There were many different life- support systems. Some had as little as 4.5 cu.m. (160 cu.ft.) and some had over 14 cu.m. (500 cu.ft.). There were advantages to both options. The less air in the submarine the less room was tak- en up by the SCUBA tanks and the smaller the submarine could be. But the Figure4 Linearpedal-drive system submarines with excess air could stay in the water longer and thereby increase stroke and the reduction in the amount by the increased drag. Having an open- their in-water time for testing. Since the water propeller is also much easier from of room required inside of the subma- tanks will become lighter as the air is a manufacturing standpoint. The rine. When using a linear system the leg used it may be worth considering plac- propeller did increase sweeps out 30 percent less water than counter-rotating ing the SCUBA tanks at the center of the overall propeller efficiency but in with a rotary system. Also the ideal ca- buoyancy of the submarine. our opinion it was not enough to over- dence for a linear system is approxi- Although not yet approved for use, mately 40 strokes per minute while the come the bearing and mechanical losses rebreathers are available. These provide ideal is approximately 60 strokes per or the increased complexity and reduced more than enough air and take up much reliability. minute for a rotary system. Thus a per- less volume than the traditional SCUBA son moves about twice the amount of systems. However, they are extremely Maneuverability water with a rotary drive system than expensive. with a linear system: this results in Many entries had the dive planes on the front of the submarine and the rudder wasted energy. Hull shape on the rear. Some entries adjusted their The disadvantages are that they are Designing a low-drag hull is not as propeller pitch to steer, others angled harder to make as mechanically effi- easy as it may seem. Most fluid- their propeller. A couple of teams used cient, and parts are harder to attain in dynamics courses cover the coefficient hydraulics instead of a direct linkage. most designs. of drag over simple two- and three- One consideration when choosing a Stroke length is also important. dimensional shapes. In order to success- Many of the teams had stroke lengths of steering system is its effectiveness when fully find a low-drag three-dimensional the submarine is not moving or is mov- 150-200 mm (six to eight inches). So shape (within a reasonable time) one ing very slowly. small a stroke length is very tiring un- must either rely on the work of others or derwater (or above water). It is benefi- use computational fluid dynamics (CFD) cial to have a setup such that the stoker to model different alternatives. can choose the desired stroke length I while racing. This is possible with a 3.5 well-designed linear drive system. 6 According to one of the Navy people 3

a study on human performance under- 12 water showed that an above-average ath- 2.5 21 lete can sustain 370 watts (0.5 hp) for a 0 period of ten minutes. The propeller r 2 31 should therefore be designed for that 30 18' power input (ours was designed using a u) 26' 29 previous study that recommended 260 1 watts (0.35 hp). Lastly, we extensively researched 0.5 both the counter-rotating propeller and a ducted propeller and decided to use nei- 0 I · · 1 I 1 ther. Although the duct would be nice 0 10 20 30 40 50 60 70 80 as a means of protection for the propel- Displaced Volume (ft^3) ler, the increase in efficiency was offset Figure5 Relationship between displaced volume and averagespeed p. 36 Human Power, fall & winter, 1991-2, vol. 9/3 & 9/4 We used CFD to model our subma- area and less frontal area for the same Questions about future races should rine. We started by using a rotated cross internal volume. be directed to the H.A. Perry Foundation section of different proven low-drag or Florida Atlantic University. cable fairings as a basis. Different pack- Advice to future competitors aging arrangements of the occupants and Once again all the teams that partici- References gear were evaluated using Intergraph (a pated learned a lot. Most of the teams 1. J. Osse, "The human-powered sub- computer drafting program) and a sim- were saying they would be back in two mersible race: A review from down un- ple adjustable plywood mock-up. After years with even smaller submarines. I der" Human Power Vol. 8 No. 1, a few iterations of modifying the inter- plan to assemble a team to compete in Summer 1990 pp. 1, 16-20 nal layout and the hull shape (using a two years. 2. M.L. Nuclols, P.K. Poole, R.M. program called GEODES) we arrived at Testing is crucial. The teams with Price, and J. Mandaichak, 1989: "Project our present shape. When it was done we the most in-water testing time performed SQUID, a lesson in design simplicity," had a low-drag, space-efficient, and better in Florida. This is the predomi- OCEANS '89 Proceedings, Vol. 6, Hu- somewhat cramped submarine. nant lesson. man Powered Submersibles, IEEE Pub. Figure 5 shows how the displaced Reliability is also very important. Of 89CH2780-5, pp.38-42 volume affected the speed of the subma- the 34 teams that made it to the competi- 3. S.L. Merry, S.L.Sendlein, and A.P. rine. The plotted speed of the Sea- tion, eleven did not qualify for the elimi- Jenkin, 1988: "Human power generation DAWG (number 21) does not reflect the nation races. This was usually due to a in the underwater environment," actual speed it could have run the lack of testing and poor reliability. OCEANS '88 proceedings, Vol. 4, IEEE course. A competitor's surface buoy be- The pilots can't race around a course Pub. 88CH2585.8, pp. 1355-1320. came tangled around the prop shaft at they can't see, so make sure the pilot has 4. The 1991 2nd International Subma- the beginning of the race and the boat enough windows. Many of the entries rine Races program. proceeded to tow both buoys around the had a hard time seeing the starting lights 5. The final race results handout (four remainder of the course. The actual top which meant that the other team got pages). speed would have been well over a half ahead at the start. It is important that knot faster (probably about 0.5-0.6 m/s - the pilot be able to see in front, side to 1 to 1.2 knots - faster). side, and down, It is also important that Coty Brandt is a recent graduateof the rescue diver can see at least one of the University of Washington where he Hull construction the occupants. received a B.S.M.E. degree. He present- Most teams used fiberglass. Some I strongly recommend a communica- Iy works at the Boeing Company. used polycarbonate and Lexan, one used tion system between the support crew Kevlar, and we used carbon fiber. Our and the submarine occupants, even if it Cory Brandt goal was to minimize the thickness but is only a one-way system. This is espe- 1151 N 82nd St. Seattle, WA 98103, still retain the strength. Carbon fiber cially important if you are testing in an USA (206) 522-7461 fulfills this goal very well. It is also al- area with low visibility. One option most as easy to work with as fiberglass. when testing is to use an underwater FIETS HPV PRIZE The majority of the boats were con- speaker. The Dutch bicycling magazing Fiets structed by first making a plug (or male Use a linear drive system if it is is offering a prize of 25,000 Dutch guil- mold). From the plug a female mold within your budget and you have the ders for an HPV that is superior to a was made. Then two splashes (or parts) mechanical aptitude. regular bicycle and that can be ridden were made using the female mold. Invest a fair amount of time picking 365 days of the year. Specifically men- These two halves were joined to form an arrangement that minimizes the vol- tioned are rain at 3C and a wind of force the submarine hull. The advantage of ume, and spend some time choosing a 7. It must have a large carrying capacity using a female mold is that one can at- low-drag shape. (minimum 80 litres and 15 kg) and re- tain a good surface finish on what will Lastly I recommend a launch and re- quire little or no maintenance. The become the outside of the submarine. covery system that can either be raised minimum average speed is 35 kph. En- The choice of material will affect the and lowered in the water or can raise the tries are open until January 31, 1993, buoyancy and the ease of testing the submarine out of the water. The ideal and judging, including trials of the ve- submarine. One of the key advantages launch and recovery system will have hicle, will take place in Holland in Feb- of carbon fiber is its high strength-to- these features and will have steerable ruary 1993. weight ratio. This enabled us to have a wheels. hull that was about two-millimeters For details, write thick in most areas. It was very light Summary Fiets B.V. Publishing Company (two people could carry it comfortably), This article is meant to provide some Valkenburgerstraat 188 extremely rugged (we survived a head- insight into what happened at this year's 1011 NC Amsterdam Netherlands. on crash into a piling at over two knots race and what must be done to make a with no damage), and the hull was close competitive submarine for the next race Alternatively, send a stamped self- to being neutrally buoyant (hence we in two years. I am open to questions and addressed envelope plus an additional didn't need a lot of lead or foam as did insight into human-powered submarine 29-cent stamp to Dave Wilson, MIT rm many of our competitors). The reduced component design. 3-455 Cambridge MA 02139 and I will thickness also meant less wetted surface ask Carolyn Stitson to send you a copy.

Human Power, fall & winter, 1991-2, vol. 9/3 & 914, p.37

I~~~~~~ ( ,~

c a) c, C , .n . c, ~o~~~~~~~~~~~~o 4~~~~~~~~~~~C 04C cD

- ~ . o

3~~~~~~~~~~~~~~~o 8 C~~~~~~~ooo C ,h~~N a 5 -- :~:-~= O~~~~~~~~~. -- C \A ~~~~~~~~~~~~~~EA

0 BP8 c ,'>,4; U Ut " C ~~~~~·3 ~~~cO~ C' S -O~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~. 4) coca "O Co O.,a,o o Q~r, _ Qco~o 6 Er. .0 .

, .c = , m

.1 ,, - $. $3.

It r-i O

I .S , s II "-*a t3 >E ;tsSW s ffi :: o Z l O o C.)Q D0 s EJ

.e. a i teq Es 9 Z ,Z o ° O ,O E a

)E 3 E en "` ' S _ -_ ._ _ t . International Human Powered Vehicle Association P.O. Box 51255 Indianapolis, Indiana 46251-0255 U.S.A. (31 7) 876-9478