HUMAN POWER No. 53

8/12/2019 HUMAN POWER No. 53.

1/13

NUMBER 53, SPRING 2002

From the editor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2

A brief introduction, Theo Schmidt . . . . . . . . . . . . . . . . . . 2

Contributions toHuman Power . . . . . . . . . . . . . . . . . . . . . 2

ArticlesComparison of measurements of bicycle spoke tension

using a mechanical tensiometer and musical pitchJohn S. Allen . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3

Adding arm power to a recumbentDaniel Kirshner . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7

Bicycle design, safety, and product-liability litigationDavid Gordon Wilson. . . . . . . . . . . . . . . . . . . . . . . . . . . . 10

LettersHuman-powered trackway systems

John Barber . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6

Demise of rickshaws . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19

Independent feelings of water moleculesGeorge Tatum . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19

Arm-powered tricycle,Mike Eliason . . . . . . . . . . . . . . . . 20Theos mini ice-scooter, Theo Schmidt . . . . . . . . . . . . . . . 21

Future of the HPVA?,John Snyder. . . . . . . . . . . . . . . . . . 23

The future ofHuman Power,Jean E. Anderson. . . . . . . 23

Book reviewsVon Null auf 140 mit 93 Zhnen: Aerodynamik

von Pedalfahrzeugeby Andreas Pooch

reviewed by Theo Schmidt . . . . . . . . . . . . . . . . . . . . . . . 22

The Recumbent Bicycleby Gunnar Fehlau; translated

by Jasmin Fischer;reviewed by Dave Wilson . . . . . . . . 22Number 53

Spring 2002 $5.50

TECHNICAL JOURNAL OF THE IHPVA

HUMANPOWER


2/13

Human Power Number 53, Spring 2002

Abstract

The tension of 140 spokes in fourbicycle wheels was m easuredusing a tensiometer a

mechanical device designed to measurespoke tension and by pluckingthe spokes like harp strings anddetermining their tension using thestandard formula for the fundamentalfrequency of vibration of stretchedwires. The measurements by the twomethods were compared. When spokeswere laced such that they touched eachother, the musical measurement wasof the average tension of a laced pair,and so there were 93 rather than 140musical measurements. Tensiometer

measurements of laced spokes wereaveraged for comparison with themusical measurements.

Both types of m easurements provedaccurate enough to use in establishingthe correct tension of spokes whenbuilding bicycle wheels. The functionrelating the measurements conductedusing the two methods is linear andconsistent, though there was somediscrepancy between the results. Eachmethod proved to have strengths andweaknesses related to convenience,which will be discussed.

The wheels usedin the investigation

The tension of the 140 spokes offour bicycle wheels was measured. Thewheels were: A 622 mm, 40-spoke dished rear wheel

with 14 gauge (~2.0 mm diameter)plain gauge spokes on the right sideand 14/16 gauge butted spokes onthe left side (~1.6 mm diameter overmost of the length, ~2.0 mm nearboth ends); both sides quadruple-crossed, spoke length 296 mm.

A 630 mm, 36-spoke wheel with 14/15gauge (~1.8 mm diameter) buttedspokes on both sides. One side islaced triple-crossed and the otheris laced radially. Spokes on bothsides are 296 mm long. (Same lengthdespite the difference in patternbecause the hub flanges are of differ-ent sizes).

A 451 mm, 28-spoke wheel with 14-gauge (~2.0 mm diameter) spokes,

double-crossed, not laced, 215 mm.This wheel had a steel rim and a verywide range of spoke tensions.

A 406 mm 36-spoke wheel with 14/15gauge (~1.8 mm diameter) spokes,triple-crossed, 183 mm.Using wheels with different spoke

lengths, patterns and tensions made itpossible to determine how these fac-tors affected the relationship betweenthe tensiometer measurements andmusical measurements.

The musical methodThe fundamental frequency of vibra-

tion of a stretched string or w ire variesaccording to the following equation,

assuming small amplitudes [1]:where

F

FL

T

m

L

T

1

1

1

2

=

=

=

=

the fundamental frequency [Hz]

the length of the string

tension of the string

m= mass per unit of length

This resolves to T=4F12L2mwhich is used to calculate the tensionfrom the measured frequency.

The cross-sectional area of the spokeand the mass per unit lengthmareexactly proportional to each other.Therefore, for two different strings orwires of equal length, one thick andanother thin, the frequency is the sameif the tension per unit of cross-sectionalarea is the same. One way to think ofthis is to imagine two identical spokesside by side, both of the same gaugeand at the same tension. They vibrateat the same frequency. Now imaginelightly connecting them together allalong their length. They still vibrate atthe same frequency. Finally, imagine

merging them into one, thicker spoke.It still vibrates at the same frequency.These facts greatly simplify the

measuring of spoke tension for wheelbuilders. To determine whether a spokeis optimally tensioned, we dont have tomeasure the thickness or, what is moredifficult, the tension, since the musical

pitch translates directly into the ten-sion per unit of cross-sectional area.

Note that the fundamental frequencyof a spoke increases only as the square

root of tension. Therefore, everydoubling of frequency one muoctave raises the tension by ator of 4. A spoke whose fundamfrequency is only 1.2 times as hthe value given in the table a minor third higher is already more than 1.4 times as much tenand is likely to fail quickly.

Bicycle spokes rarely break dexcessive tension; but the rim mwithstand it, and when the rim around the spoke holes, the whWeight loading on a wheel decrtension of the few spokes at theof the wheel greatly, and raises sion of the remaining spokes on

slightly, but lateral loading whiling out of the saddle causes sigincreases in spoke tension and to rapid failure of an over-tensiowheel. On the other hand a comerror in wheel building is to leaspoke tension too low, resultingweak wheel, since spokes go slaunder smaller loads, and fail to the rim steady. The excess motioslack wheel is what breaks the and allows the nipples to unscr

As thethicknessof a wireincreases, ourequation formusical pitchbecomesslightlyinaccuratebecause thegreater bend-ing stiffnessadds its con-tribution tothe stiffnessgenerated

by the wirestension. Thediscrepancyis not largeand onlyamounts toa few per-cent. Table 1includesa correc-tion which I

Human Power Number 53, Spring 2002 2 Number 53, Spring 2002 Human Power

H U M A N P O W E Ris the technical journal of the

International Human Powered

Vehicle Association

Number 53, Spring 2002

Editor

Theodor Schmidt

Ortbhlweg 44

CH-3612 Steffisburg, Switzerland

[email protected]

Associate editors

David Gordon Wilson

21 Winthrop Street

Winchester, MA 01890-2851 USA

[email protected]

Toshio Kataoka, Japan

1-7-2-818 Hiranomiya-Machi

Hirano-ku, Osaka-shi, Japan 547-0046

[email protected]

Philip Thiel, Watercraft

4720 - 7th Avenue, NE

Seattle, WA 98105 USA

Production

JS Design & JW Stephens

IHPVA

Paul MacCready, Honorary president

Paul Gracey, USA, Chair,

Open, Secretary/treasurer

Publisher

IHPVA

PO Box 1307

San Luis Obispo, CA 93406-1307 USA

[email protected]

Human Power(ISSN 0898-6908) is

published irregularly for the Inter-

national Human Powered Vehicle

Association, a non-profit organization

dedicated to promoting improvement,

innovation and creativity in the use of

human power generally, and especially

in the design and development ofhuman-powered vehicles.

Material in Human Power is copyrighted

by the IHPVA. Unless copyrighted also by

the author(s), complete articles or repre-

sentative excerpts may be published else-

where if full credit is given prominently

to the author(s) and the IHPVA. Individual

subscriptions and individual issues are

available to non-IHPVA and non-HPVA

members.

FROM THE EDITOR

This issue ofHuman Poweris a bitdifferent. David Gordon Wilson hasasked me to take over as editor at leastfor a while, as he is involved in startinga company building small gas turbinesand in completing the third edition of

Bicycling Science. Dave Wilson hasmadeHuman Powerinto the worlds

premier technical journalon HPVs, (including HPBs,HPAs) and also some sta-tionary applications ofhuman power. If there werea knighthood for human

power, he would deserve it!

I would like to broadenour scope to includearticles to do with the phi-losophy of human power.There are two reasons forthis. Firstly, we are gettingless material than beforeas HPV physics becomemore and more specialised;and secondly, the status ofhuman power in society is a

precarious one and we needto give it all the supportwe can in order to improvethe quality of life in presentand future societies. Please send yourarticles and letters.

Human Power continues to beproduced by Jean Anderson and John(Elrey) Stephens of the HPVA. It is sentto HPVA members together withHPV

News(now edited and produced byPeter Eland of VeloVision) and is avail-able for subscription to all other IHPVAmembers at reduced rates.

A brief introduction

I am 48 and live with my wife in anold house in the foothills of the Bernese

Alps near Lake Thun. Its full of partsfrom human- and solar-powered vehicleand boat projects. I grew up nearMount Tamalpais in California, wheremountain biking started, and studied inBasel and in hilly Wales and England,where I was introduced to HPVs.

My own projects have mostly beenhuman-solar hybrid vehicles and boatsand formerly racing these in the Tourde Sol. These have always been eithereasily transportable or semi-amphibi-ous in order to tour without being car-dependent. Im vice president of FutureBike Switzerland, where one of ourspecialties is organising races for HPrail vehicles.

Theo Schmidt

Stefsburg, Switzerland

Contributions to Human Power

The editor and associate editors (you may choose with whom to correspond) welcomecontributions toHuman Power. They should be of long-term technical interest. Newsand similar items should go toHPV Newsor to your local equivalent. Contributionsshould be understandable by any English-speaker in any part of the world: units shouldbe in S.I. (with local units optional), and the use of local expressions such as two-by-fours should be either avoided or explained. Ask the editor for the contributors guide(available in paper, e-mail and pdf formats). Many contributions are sent out for reviewby specialists. We cannot pay for contributions. Contributions include papers, articles,technical notes, reviews and letters. We welcome all types of contributions from IHPVAmembers as well as from nonmembers.

Comparison of measurements of bicycle spoke tensiousing a mechanical tensiometer and musical pitchby John S. Allen

H U M A N P O W E RNumber 53 Spring 2002 $5.50/IHPVA members, $4.50

Length

(mm),

plain

Length

(mm),

butted

308

292

276 308

262 292

248 276

236 262

224 248

212 236

201 224

191 212

181 201

172 191

163 181

156 172

147 163

156

Table 1.

JasonPatient


3/13

for bending stiffness. However, tensionand bending stiffness affect the read-ing of the tensiometer similarly, and soadjustments in the parameters of theformula should be able to compensatefor differences in bending stiffnessquite well. It is to be expected that the

values of band cwill vary with spokegauge, reflecting the differences inspoke thickness and bending stiffness.

The parameters a, band cof theequation above were adjusted by eye to

produce smoothed curves for14-, 15- and 16-gauge (~2.0, ~1.8, and~1.6 mm diameter) spoke shafts, whichconformed as well as possible to thetensiometer calibration readings. Thecomparison between the calibrationreadings and the smoothed curves areshown in the calibration graph (fig. 4).The vertical scale of the graph is loga-rithmic, reflecting the fact that the ratiorather than the difference between the

calibration reading and the height ofthe smoothed curve is to be kept assmall as possible.

As can be seen, the fit of thesmoothed curves for all three spokegauges is quite good. It does appearthat the smoothed curves for the lightergauge spokes may be a bit low near thelow end of the tension range, and highnear the high end. However, given thenoisiness of the calibration readings,no firm conclusion can be reached onthis issue. There is, however, a wide

variation in the constant a between thedifferent spoke gauges.

The goal of the modeling is in anycase only to derive a formula w hose

form reasonably well reflects the ten-siometers response to spoke tension,so that the parameters of the smoothedcurves may be used to calculate aspoke tension that corresponds to anytensiometer reading. The tension is cal-culated by inserting the values of a, band c for the appropriate spoke gaugeinto the equation for tension, alongwith the tensiometer reading y. Thisresults in a calculated tension for eachspoke examined.

Using the musical method, thetension can be calculated for indi-

vidual spokes in a wheel with radialor unlaced spokes, but in a wheel withlaced and touching spokes, it can becalculated only for each laced pair,which resonates as a unit.

Though a tensiometer measurementis available for each spoke, a tensionreading for each laced pair of spokesmust be derived in order to make a

comparison against the musical read-ings. The combined tension reading isthe average of the tensions of the twospokes.

The values of tension using the musi-cal and tensiometer methods may nowbe compared. The results of the com-

parison are shown in figure 5.

Conclusions1. As can be seen in the graph, there

are a few outliers, which most likelyresult from recording errors ratherthan measurement errors, consideringtheir extreme values and small num-ber. Aside from these, the correlationbetween the musical and tensiometerreadings is tightly grouped and nicely

linear over a 10 to 1 range of sptension, for spokes of differing of all three spoke gauges, lacedunlaced.

2. A perfect correspondence bthe musical and tensiometer reawould produce the function

y x=

where yis the tensiometer meament andxis the musical tensiosurement.

The actual function (in Newtapproximately

y x= 1 2 200.

The causes of this difference measurements could be in eithetensiometer or the musical methboth, and can not be determineout measuring the tension of actensioned spokes, for example, supporting a hub from which ware suspended. It is, however, m

likely that the offset in the crosthe yaxis is due to an error in tsetting of the tensiometer calibrbecause frequencies of vibrationently correspond accurately to tratios.

3. Both the musical method atensiometer are accurate enouguse in establishing the correct tsion level of spokes in wheel buHowever, the accuracy of both msurement means could be improby accurate calibration. In the cthe musical method, calibrationserve to establish accurately thecompensation which is necessaaccount for bending stiffness fodifferent spoke gauges.

Human Power Number 53, Spring 2002

determined empirically by measuringhe musical pitch of tensioned spokes

clamped off at different lengths.Part of a spoke at the outer end is

nside the spoke nipple, and part athe inner end is in contact with the

hub. These parts do not contributeo the vibrating length. The table also

accounts for this. The ends of a buttedspoke are sufficiently thicker than theshaft of the spoke so that the endscontribute only slightly to its effectivevibrating length. This, and the greaterstrength at the threads and head of abutted spoke, account for the highermusical pitch recommended for buttedspokes.

The yield strength of good steel isabout 150 000 pounds per square inchor 1 040 N/mm2, and the tension recom-mended in the table is 1/3this aboutas high as you can take the tension andstill leave an adequate margin of safety.

In the following comparison with theensiometer measurements, I used aength adjustment of 3 cm for all spokesn this investigation (reduced vibrat-ng length to account for the effects of

bending stiffness), and used the musi-cal method to measure existing spokeensions rather than to adjust a wheel

for optimum working tension.

The tensiometer methodThe tensiometer (fig. 1, 2) is a

mechanical device which contacts thespoke at three points. The middle oneof these points is the tip of a probewhich is pressed against the spoke bya spring. The higher the tension of thespoke, the less it is flexed out of line byhe probe. A reading of the probe posi-ion is taken from a dial gauge which

registers the probe position.

The response of the tensiometerdiffers with spoke thickness, becausehe increased thickness displaces theensiometer probe more, and becausehicker spokes also have greater bend-

ing stiffness. The tensiometer used forthis investigation was supplied with acalibration chart, with a different set

of readings for different spoke gauges.No information was provided as to howthe chart was developed, but appar-ently one or more readings were takenfrom spokes whose tension had been

premeasured, at each of several ten-sions.

Because of the effect of bendingstiffness and the geometry of the ten-siometer mechanism, it is clear thatthe tensiometer readings should forma nonlinear function of spoke tension,and in fact, they do. But a graph of thecalibration readings shows clearly thatthe calibration measurements are alsosomewhat noisy they do not forma smooth curve. The inconsistenciesin measurement might be due to fric-tion in the tensiometers mechanism,friction between the spoke and thetensiometer probe; inconsistenciesin thickness of spokes that are nomi-nally of the same gauge; variation the

position of the tensiometer along thespoke; nonlinearity in the tensiometersresponse, or perhaps other factors.

Comparing the methods

In order to compare the tensiometerreading of spoke tension with the valuecalculated using the musical method, itis necessary to interpolate between the

points in the tensiometer calibrationchart. If it is assumed that the noisein the tensiometer calibration resultsfrom measurement error, then the bestinterpolation is a smooth curve whoseform reflects on the tensiometersgeometry.

Both the musical method and thetensiometer method as used in buildingwheels tend inherently to have a quan-tization error. In the musical method,it is the tendency to assign a musical

pitch to the nearest musical semi-tone.With the tensiometer, it is the tendencyto assign the tension to the nearestcalibration table entry. Interpolationis possible, but requires higher musi-cal skill with the musical method, andmathematical calculation when usingthe tensiometer.

When curve fits are used in derivinga tension measurement from the ten-siometer reading, quantization error isno longer an issue, and the noise ofthe calibration tables is smoothed out,though errors in curve fit may occur.

In order to generate the curve fits,a geometric model of the tensiometerwas developed (fig. 3).

As apparent from figure 3, the ten-sion on the spoke relates to the forceon the probe as

F T

FTy

y a

=

=

+

2

22 2

sin

or

where Tis the tension, yis the posi-tion of the tensiometer probe,Fis theforce on the probe and a represents thelength of the spoke between the probeand the support at each side of thetensiometers arch.The force on the tensiometer probe varies

as

F b y c= +( )where bis a constant which representsthe spring rate of the tensiometer

probe and c is a constant represent-ing the position of the probe when thespring is at rest.

Now, equating the force,Fin theequations above, and solving for T, weobtain:

Tb y c y a

y=

+( ) +2 2

2

The geometric model accounts forspoke tension and thickness, but not

4 Number 53, Spring 2002 Human Power

Figure 1. The Hozan Tool Industry, Inc.spoke tension meter C-737 used inthis investigation

Figure 2. In use, one arm of thetensiometer is hooked under the spokeand the other is pressed down until it

just touches the spoke, depressing thecentral plunger against spring restoringforce and producing a dial reading. Thetension must be read from a calibrationtable, as the reading is a nonlinearfunction of tension, spoke thicknessand spoke bending stiffness.

Figure 3. Geometric model of thetensiometer

Figure 5. Musical vs. tensiometer measure [N]

Figure 4. Tensiometer readings vs. tension [N]


4/13

6 Number 53, Spring 2002 Human Power Human Power Number 53, Spring 2002 Human Power Number 53, Spring 2002

4. There are strengths and weak-nesses to both methods. The musicalmethod is much faster than using theensiometer, and the musical pitch

relates directly to the optimum workingension of spokes based on their stress

per unit of cross-sectional area, regard-ess of the spoke gauge. However, the

musical method requires musical train-ng. When spokes are laced and touch-ng, the musical method as used in thisnvestigation does not measure theension of individual spokes, but rather,

gives an average tension reading foreach pair of laced spokes sufficiento measure the tension level of a wheel

as a whole, but not as useful for iden-ifying individual spokes which are tooight or loose. (It is possible, though

more difficult, to measure the tensionof individual spokes by listening forhe musical note produced by the part

between the lacing and the rim).

Suggestions for further research1. Check the calibration of the ten-

siometer used in this investigation, sohat the causes of the discrepancies

between the tensiometer and musicalmeasurements might be identified.

2. Providing a more sophisticatedmodeling based on a larger sampleof calibration readings, accountingfor bending stiffness, and using acurve-fitting algorithm rather than aneyeball comparison would certainly

produce a more accurate correspon-dence between the readings and thesmoothed curves.

3. Perform musical measurements onaccurately measured spokes (for exam-

ple, using a hub from which weightsare suspended) so as to identify anaccurate function of musical pitch as itrelates to tension and spoke length.

4. Provide a statistical analysis ofresults.

5. Measure a number of differenttensiometers, to determine the levelof accuracy which can be expected ofthem.

AcknowledgementsThanks to Calvin Jones of Park Tool

Company [http://www.parktool.com/]for motivating me to conduct this

investigation, and for the loan of atensiometer; David Gordon Wilson andSheldon Brown for their encourage-ment, and Jobst Brandt for the back-ground information in his book The

Bicycle Wheel[2] and for his doubtsabout the musical method, which fur-ther motivated me to investigate it.

References[1] The derivation of this formula is

given, for example, in Alonso andFinn,Fundamental university phys-ics, vol. 2, Fields and waves (1967,

Addison Wesley), section 18-7.[2] Brandt, Jobst. 1993. The bicycle

wheel, 3rd ed. Menlo Park, CA:Avocet. (ISBN #0-9607236-6-8)

About the authorJohn S. Allen ,[http://www.bikexprt.com] has had along career as a writer about bicycling,and is author or co-author of a numberof publications, including Sutherlands

Handbook for bicycle mechanicsandBicycling street smarts. He lives nearBoston, Massachusetts, USA.


Adisadvantage of recum-bents the rider cant rise outof the saddle in a sprint can be

turned to advantage. The recumbentriding position allows use of thearms to add power to the bike if a

practical way to do so can be found. Ihave developed a working prototypethat allows this.

The most surprising aspect of thedesign is how easy it is to both steerand power the bike at the at same time.The arrangement however also appearsto be surprisingly effective in allowingme to put more of my human powerto use.

Here is a description of the armpower mechanism, its development,and its effects. I also describe my plan

to make the mechanism a simple add-on to just about any recumbent.

How the arm power mechanismworks

Figure 1 shows the arm powermechanism on my custom recumbentbike. A professional frame builderbuilt the bike to my design (withoutarm power!) about 17 years ago. Two

vertical handles on either side of theseat are part of a single handlebar unit.Power is supplied through a rowing,back-and-forth motion of the handlebarunit, which pivots about a horizontalaxle, transverse to the bike, under theseat. Power can be applied both on theforward (push) stroke, and on the rear(pull) stroke.

Steering is accomplished by differen-tial motion of the two vertical handles,which also pivot about a generally

vertical axle that itself rotates aboutthe transverse, horizontal axle. Figure 2

provides a close-up of the rowing/steering mechanism. The floatingchainring serves as a chain tensioneras there is no m echanism to tension

the primary chain.Figure 3 illustrates the main features

of the mechanism.

Rowing and steering axles

The rowing axle is a transversehorizontal axle. A short tongueextends rearward from the center ofthe axle. A vertical hole through thetongue is used to attach the verticalsteering axle (although, of course, thisaxle tilts backwards and forwards from

vertical dur-ing the rowingstroke). Thehandlebar clampholds bearingsmounted on thesteering axle.

Power take-off

Arm poweris transmittedthrough the hor-izontal axle toa short crankarm attachedto the end ofthe axle on theleft side of thebike. From theend of the crank

arm a shortspindle-axle holds a bearing attachedto the connecting rod that trans-mits the back-and-forth motion ofthe crank arm to the rotary motion ofanother bearing/spindle-axle on theintermediate/crossover drive.

An important feature of the powermechanism is that the handlebarshave a fixed connection to theintermediate/crossover drive, and thusto the pedals. Because there is no free-wheel mechanism between the pedalsand the handlebars, when your feetmove, the handlebars move. This lets

your feet carry the arm levers throughthe dead spot at the end of each

stroke.Steering take-off

Steering motion is transmittelink with rod-end bearings at ea

At the front of the link, a short arm is attached to the steererof the forks (where the handlebwould go on an upright bike). Arear of the link, a short arm attato the horizontal portion of the bars positions the rear rod-end bing a couple of inches in front ohandlebars.

This positioning of the rear rbearing is important: it is locatethe axis of the horizontal axle. T

Adding arm power to a recumbentby Daniel Kirshner

Figure 1. Dan Kirschner on his custom recumbent bicycle

Figure 2. Close-up of the arm powered rowing/steering mechanism

LETTER

Human-powered trackway systemsontinue to fascinate. A new variation is

described in the excerpt of two lettersby John Barber, whose company hasdeveloped a magnetic suspension unitneeding no energynput or electronicontrol system (butequiring separate

mechanical guidingand drive systems).t is interesting that

he sees his sys-em as a low-costolution for third

world problems,with the lifting unitsbeing part of indi-

vidual vehicles freeo join and leave anoverhead track attations. So far theompany has builteveral models ofhe maglev unit, but

not of the overheadrack. John Barber

writes: In manyareas small vehi-

les, either human

powered or propelled by internal com-bustion or electric motors, are wellsuited for providing a significant por-tion of local transport needs. However,their effectiveness is frequently com-

promised by problems of local roadwayconditions, terrain, inability to travel

longer distances andlimited endurance.It is true, for highlyefficient bicycles ongood roads, that therunning friction is nothigh. But that is oftennot the case. Balloontires on primitiveroads are pretty com-mon in large areas ofthe world. Resistancehere is sizable. I have

read commentaries onthe problems faced byrickshaw drivers, of asimilar nature.

MTSC, of WestlakeVillage, CA, USA,has developed and

patented a particularmagnetic supporttechnology for trans-

port systems that uses

permanent magnets mounted on thevehicle for generating lift. The configu-ration we use is such that no vehiclemotion, nor input of energy, is requiredfor the lift. The lift is inherently stable,although the vehicles do need to besteered. Additional information on thetechnology may be found in our website: http://www.magsupport.com. The

patent is # 5825105 (US).The concept envisions the con-

struction of a network of independentelevated guideway segments, on whichthe vehicles, levitated by the MTSCmagnetic support system, would oper-ate. This could provide, with modestexpenditure, a grade-separated, high-quality travel way, generally immuneto weather, offering a smooth ride, andrequiring relatively little energy input

for propulsion.The cost of a lifting unit: in mass

production where the magnets arebeing purchased in quantity, and the liftunit parts are likewise being fabricatedin quantity, we estimate a cost on theorder of [US]$1.00$1.50 per kg of masslifted.

John Barber

President, MTSCSegment of an elevated trackway with

magnetic suspension unit supportinga vehicle.


5/13

8 Number 53, Spring 2002 Human Power Human Power Number 53, Spring 2002 Human Power Number 53, Spring 2002 8 Number 53, Spring 2002 Human Power

back-and-forth power strokeof the handlebars thus hasno effect on the steering (ora negligible effect when thesteering motion moves thebearing position slightly offhe axis).

You might be able to see inhe photographs that the proto-ype handlebars are construct-

ed from modified aero-barsclamped to a horizontal tube.Weight

In its current form thearm-power mechanism addsapproximately 1.4 kg (3 lb)o the bike, not counting thentermediate/crossover drive, whichtself adds about 0.5 kg (1 lb). A refined

design (eliminating the aero-barclamps, for example), could probablysave 0.5 kg, and under Future develop-ments, below, I discuss plans to elimi-

nate the crossover drive. In this casehe net weight addition should be about

1 kg (2.2 lb).

Arm power backgroundI was intrigued by the notion of add-

ng arm power to bicycles by ergometerest results summarized inBicycling

Science.[1] These results showed thatwith a forced rowing mechanismusing both arms and legs, about 12.5percent more power than with normalpedaling was obtained throughout theime period for all subjects. Forced

rowing is a mechanism that defines theend of the stroke and thus conserveshe kinetic energy of the moving mass-

es. This is unlike typical rowing in aboat, where the rowers must decelerateand reverse their motions without helpfrom the mechanism. [2]

While the result showing additionalpower available from the arm and legpower mechanism is indeed intriguing,t should be noted that the test period

extended only as long as five minutes.With respect to creating arm and

eg power mechanisms for human

powered vehicles, references on thenternet indicate that there has been

quite a bit of activity. (See, for exam-ple, www.geocities.com/rcgilmore3/and_rowers.htm.) Im aware of onlywo bikes currently in production: the

Thys Rowingbike is built in the Neth-erlands (see www.rowingbike.com);Scott Olsons Rowbike is built in theUSA (see w ww.rowbike.com). Thesebikes use a free rowing motion (as

opposed to the forced rowing mecha-nism discussed above). The linearrowing motion is transmitted directlyto the hub, with a ratcheting/freewheelmechanism for the return stroke. Sucha mechanism does not decelerate therowers feet or arms at the end of the

stroke. I have not seen information thatcompares the performance of thesebikes with other legs-only machines.

Gardner Martin has built severalmodified Easy Racers [Tour Easys] that

put hand cranks in place of the handle-bars. The hand cranks are connected

via chain, idlers and freewheel to thebottom bracket chainrings. The chaintwists a bit during steering. Gardnersays that the rider does have to learn tocounter some of the torques introducedby arm power, and indicates that thearm-powered bike lets a rider producemore power, and use higher gears onhills, for example. Gardners ergometertests showed a higher heart rate assoon as the rider starts using his/her-arms, so it may be that the arm-and-leg-power combination is less efficientthan a legs-only machine. My tests,however, do not show this result, as Idescribe below.

DevelopmentI gave active thought to adding arm

power to my recumbent for at least fiveyears, and ran through many possibili-

ties in my mind. First, however, I need-ed someone who could help with proto-type work. A short search of local bikeshops turned up Stephan Long. He builta stationary bike/trainer that includedhandlebars much like those on the bikedescribed in this article, linked to thecrank chainwheels in much the sameway. I didnt do any scientific tests, butit was clear that the mechanism wascomfortable to operate, and appeared

to allow me to increase mypower output.

I knew that I would wantto experiment with differentratios between the arm row-ing speed and the leg rotationspeed. The stationary trainerconvinced me that I wantedmy arms going half as fastas my feet. I also thought thatany faster movement of myarms would make it that muchharder to steer. Of course, thestationary bike told me noth-ing about whether it would be

possible to both power andsteer a bike at the same time.

Designs

The first design I chose to build wassimilar to the mechanism describedabove, except that the handlebar unitdid not have a vertical axle. Instead,both vertical handles could twist about

their own axes. A steering linkagemuch like a cars with link rods toeach side transmitted the twistingmotion through an idler to a fore-aftlink to the front of the bike.

This prototype proved to be unride-able even though it did not have arm

power motion at all. I could not pro-duce enough torque to control the bikemerely by gripping the vertical handles.The quick addition of horizontal exten-sion handles (bar ends) to each verticalhandle produced an easily-controlledbike but it was now much wider thanI desired.

The next (and present) prototypeinvolved modification of my bikesexisting handlebar clamp/bearing unit(see fig. 3) and the addition of the hori-zontal axle, held by bearings placed inmodified handlebar clamps. I first triedthe bike without the arm-power con-necting link the handlebars workedfine for steering and then hooked upthe arm power link. As I said, it wassurprisingly easy to ride.

Over the next several months nearlyall the components were replaced asthey either broke or proved too flex-ible. Also, I did not feel that my armswere making an adequate contributionto powering the bike my legs wouldfeel fatigued while my arms didnt seemto be doing much work. So I experi-mented with different ratios betweenthe arms and legs. The original ratiowas 2:1 the arms going half the speedof the legs. I then tried a 1.5:1 ratio.While this sounds odd, it was still com-

fortable to power. Nevertheless, I stilldidnt feel that my arms were contribut-ing enough. I have kept a 1:1 ratio sincethen.

How well does it work?The bike seems to work very well:

you definitely feel like you are add-ing power with your arms, and I canuse higher gears on hills. But is therean advantage? How big is the effect,if any? Finally, over the long run, youwould expect to be limited by youraerobic capabilities, so you might notexpect any advantage except in short-term sprints.

Ive been the only rider so far, so thetests are limited. I have three differ-ent results to report: (1) comparisonsamong my recumbent with arm power,without arm power, and upright bicy-cles on a half-hour uphill ride;(2) similar comparisons on a very brief

uphill sprint; and (3) heart rate com-parisons between using and not usingarm power on a trainer.

Table 1 shows comparisons amongmy recumbent with arm power, withoutarm power, and an upright bicycle on ahalf hour uphill ride. Its a challengingride; the one time I rode with a heart-rate monitor it showed a maximum of186 beats per minute. My wife tells methe charts show that at my age (47)that should have killed me! Table 1shows the times for three parts of theride in certain cases I did not com-

plete the ride, or did not get a time forthe final part (stopwatch error!).

The comparisons are only rough. Astable 1 indicates, I used a Bromptonfolding bicycle as the upright. TheBrompton has 16-inch wheels and afive-speed hub, so may be less efficientthan the recumbent. Then again, theBrompton has high-pressure (85 lbs)tires, and has a weight advantage overmy recumbent 12.3 kg (27 lbs) versus15.5 kg (34 lbs).

While I came close on the upright inone case (trial 4 compared to trial 1)

nevertheless, the best times went to thearm-powered recumbent. Without

arm power the recumbent was 3.5minutes or so behind the same bikewith arm power.

Table 2 shows comparisons for abrief, approximately 20-second, uphillsprint. I did the sprint about three orfour times on each bike/configuration.Table 2 reports the best times, and alsosome statistics on the percentage com-

parisons of the times and weights ofthe bikes and rider (who was approxi-mately 63.9 kg (141 lbs) in each case).

In this case, the upright bicyclesare definitely ahead of the recumbent.The percentage comparisons supportthe advantage of uprights in the shortsprint. While my heavy old Schwinnmakes that configuration 10% heavierthan the lightest, fastest bike theBrompton its only 3.2% slower, whilethe recumbent configuration, which isonly 4% heavier, is 7.7% slower (witharms) and fully 12.2% slower withoutarms.

I should note that the times shownin table 2 were taken early in the devel-opment of the arm power recumbent.Perhaps additional conditioning wouldmake a difference.

Finally, I also compared my heartrate with and without arm power on atrainer. I used my old Houdaille RoadMachine trainer: this trainer usesa flywheel/fan to provide both windresistance and realistic simulation ofthe momentum of the bike and rider.I established a steady speed (as mea-sured by a typical cycle-computer) andnoticed that my heart rate was stable atthat speed (within about plus or minus1 beat per minute). I then stoppedusing my arms, and used my legs

alone to maintain that speed. At everyspeed-heart rate combination that Itried from a sedate 12 miles per hour

at about 130 beats per minute, tdifficult to maintain 29 miles peat about 182 beats per minute use or non-use of arm power mdifference. I conclude that my hrate, at least, closely reflects therequirement, however it is achie

Combining these results withsubjective impressions, the armappears to allow me to exercisehigher aerobic level, less limitedcapability of my leg muscles oveger periods. Certainly, when I mhalf-hour hill climbing comparislegs ached a great deal more withe arm power. It remains to bewhether this will be true for others, and under different conditiexample, a longer exercise perioNevertheless, in short sprints, thity to move around on the bike to generate more power for at lshort period.

General observationsHow does it feel to ride? Goo

when you are using a great dealforce to push and pull you caboth strokes for power you arable to make fine steering adjus

Apparently your body is well atto controlling small differencesmotion of your arms, even whenare moving quite a bit.

One thing you cannot do is ridhanded or, you can, but only istop pedaling. If you stop pedalcan use your feet to hold the pomechanism steady. Then pushin

pulling on one handlebar gives conventional steering. But steerwith one hand while the handlemoves back and forth with the pis nearly impossible. This is a sedrawback that I shall try to fix, describe in Future developmenbelow.

As I mentioned earlier, there freewheel between the pedals aarms, so that your foot motion c

your arms through the dead spothe end of each stroke. In fact, ialmost impossible to use arm poonly you tend to get stuck at

Table 1. Times for uphill ride (min:sec)

Trial Bike Part 1 Part 2 Part 3 1+2 1+2+3

1. Recumbent - with arms 05:06 14:07 11:17 19:13 30:30

2. Recumbent - no arms 05:37 15:43 12:41 21:20 34:01

3. Recumbent - with arms 05:15 13:45 11:12 19:00 30:12

4. Upright - Brompton 04:58 14:16 19:14

5. Upright - Brompton 05:23 14:42 12:15 20:05 32:20

6. Upright - Brompton 05:24 14:23 19:47

Table 2. Comparisons: short uphill sprint

kg kg sec. % kg

seconds bike total slower slower heavier h

Best Schwinn upright 21.34 19.5 83.4 0.66 3.2% 7.3

Best Brompton upright 20.68 12.2 76.2

Best recumbent arm & legs 22.28 15.4 79.4 1.60 7.7% 3.2

Best recumbent legs only 23.21 15.4 79.4 2.53 12.2% 3.2

Figure 3. Diagram showing the main feaures of the mechanism


6/13

10 Number 53, Spring 2002 Human Power Human Power Number 53, Spring 2002

AbstractThe United States is a litigious soci-

ety, and product-liability litigation isof considerable concern to companieshat export to the U.S. Much publicitys given to horror stories of seemingly

excessive judgments against apparentlyethical manufacturers after they havebeen sued by unscrupulous peoplepretending to be victims of what isclaimed to be deficient design. How-ever, these reports are far from repre-sentative of the actual situation. Theother side of this story is that product-iability litigation is decreasing quite

markedly in the U.S.; that this form ofitigation brings about major improve-

ments in product design and in thesafety of the public; and that it is pos-

sible to avoid most negative impacts ofsuch litigation by striving for, and docu-menting, excellence in the design andmanufacturing of products, by clearlywarning users of dangerous situations,and by putting trust in insurance that isstandard for the industry. Some areasin which improvements in design andmanufacture of bicycles are needed arediscussed as examples.

IntroductionBackground

Product-liability litigation in theU.S. has been governed by state laws,despite frequent attempts by business-friendly legislators to get unified fed-eral laws passed. Business people musttherefore be concerned about their

products being used in states where

requirements may be particularly oner-ous. In practice, state legislators arequick to copy laws that have workedwell elsewhere, so that the differencesamong regulations in different statesare not as large as might be imagined.However, in many areas of modernlife we are driven by what we know ofextreme cases: only these are reportedby news organizations. Here is a recentexample (disguised so that your author

does not get sued, because I was anexpert witness for one of the manu-facturers involved) of what seems tome to be an unfortunate extreme situ-ation.

An extreme case?Bill, a young and energetic U.S.

physician, bought a regular roadbicycle for recreation. He found that heliked biking, and hearing that sew-uptires are used by racing cyclists andwould enable him to go faster, boughtnew wheels and tubular tires and hadthem installed on his bike. One day hewent with a group of fellow physicianson a ride that included the summit ofa small mountain. While pausing atthe top he joked to his friends that hehad bad brakes, showing that with the

brake levers fully squeezed against thehandlebars he could move his bikeeasily back and forth. He then saidLast man down the mountain buys thebeers! and rode off down the steep,rough, bumpy, asphalt road with theothers in hot pursuit. The road hadsigns showing a speed limit of 35 km/h,and, after about a kilometer, a warn-ing of a sharp S-bend. The person whowas closest behind Bill said that as he

of the stroke or the other, or else youpush or pull a moment too soon andend up freewheeling backwards.

I worried about play in the mecha-nism between the arms and the pedals.

worried about wrist strain, since thepivoting handlebars would appear tomove your wrists in a way that naturedid not intend. So far, however, thathas not been a problem.

Future developmentsI am working in two areas. Firstly,

a long connecting rod can be used toransmit arm power to the pedal crank-

set. This would eliminate the need foran intermediate/crossover drive, andmake the mechanism simple to add toust about any recumbent. Secondly,he arm power mechanism needs the

ability to disengage from the pedals, sohat you can continue pedaling while

riding one-handed.

Finally, I have not yet decided wheth-er to patent the arm power mechanism.My understanding is that U.S. lawallows me to file within one year of thedisclosure marked by this publication(while I have now forfeited Europeanrights). No one that I have consultedwho has expertise in this area, how-ever, has recommended pursuing a pat-ent the recumbent market is small,and the number of potential arm powerconverts smaller still. One is unlikelyto get ones money back, which mightbe better invested in developing the

product. I will be interested to hear ifthis publications readers support thisadvice!

References[1] Whitt, F. R. & Wilson, D. G.

1982.Bicycling Science, 2nd ed.Cambridge, MA: MassachusettsInstitute of Technology.

[2] The ergometer tests were conductedby J.Y. Harrison, and reported inMaximizing human power outputby suitable selection of motion cycleand load,Human Factors12:3,1970, pp. 315329.

The authorDan Kirschner lives

in Berkeley, California, with his wifeand two daughters. Most days he ridesthe bicycle described in this article5 miles (8 km) each way to and fromhis work as an economic analyst inthe regional office of EnvironmentalDefense. He also enjoys rides in theBerkeley and Oakland hills. He previ-ously contributed an article, Aerody-namics vs. weight, to the Second Inter-national Human Powered Vehicle Sci-entific Symposium in 1983 (reprinted in

Human Powered Vehicles, Abbott andWilson, eds., Human Kinetics, 1995).


Bicycle design, safety, and product-liability litigation*with case studies of wire-ropes, brakes, rims, and tire-fit)

by David Gordon Wilson

* This is an adaptation of a paperiven at the sixth annual bicycle-designompetition in the Taiwan Bicyclendustry R&D Center, 29 August 2001,

which is a considerably updated andxpanded version of a paper first given onAugust 1998, under the title The designf advanced human-powered vehicles/elomobiles and product-liability litigation:an they co-exist in the light of apparentlyutrageous U.S. cases?Proceedings of the

Third European Seminar on VelomobileDesign, Roskilde, Denmark.

approached the bend his cycle comput-er was registering about 75 km/h andthat Bill was out of sight ahead of him.He braked to get around the bend andsaw that Bill had hit a stone wall andwas lying on his back some distancefrom his bicycle.

Bill had severed his spinal cord andwas, tragically, a quadriplegic fromthen on. He gave his bike to a familymember, who, after having the frontwheel and fork replaced, used it regu-larly. Bill confessed at some point thatthe accident was his fault. However,after over a year he (or possibly hisinsurance company) decided to try toget some money through the courts,and his lawyer sued the bicycle shopthat sold him the bike, the bicyclemanufacturers, and all possible manu-facturers of the rims and the tires (theactual front wheel and tire had beendisposed of). One would have thought

that these companies would have hada very strong case. Yet one by one they,or rather their insurance companies,all settled out of court, meaning thatthey agreed to pay large sums to the

plaintiff to avoid the far-larger costs ofgoing to trial. They also may have feltthat, however strong their case, thesight of this young man sitting para-lyzed in a wheelchair, with his wife andchild, would be enough to make an

American jury decide that these insur-ance companies were rich and Bill andhis family had already been punishedterribly. To award him a large settle-ment even though he was at fault couldbe possibly some form of jury-adminis-tered social justice.

The present status of product-liability

litigation in the U.S.

Cases like this seem to be typicallyAmerican. In what is considered to bea free-enterprise system (but is in factincreasingly regulated) the absence ofa national health-care and welfare sys-tem seems to give credence to reportsof juries leaning to the left. They are

drawn largely from the lower end ofthe economic spectrum because pro-fessional people try to find reasons tobe excused from jury service. However,contrary to popular belief, jurors donot overwhelmingly sympathize withindividual plaintiffs at the expense ofcompanies. According to Jury verdictresearch reported inBusiness Weekon November 8, 1993, defendants (usu-ally manufacturers) won 57 percentof the products-liability suits in 1992.

This proportion had been 45% in 1989.Popular opinion also paints a pictureof a flood of products-liability litiga-tion. In fact, products-liability lawsuitswere less than 1 percent of the totalstate and federal caseload in 1994 [1]and less than 0.4% of the civil casesin state courts. (There is a huge back-log of lawsuits awaiting trial in mostU.S. jurisdictions, but most cases aresuits between businesses and betweenfamily members, particularly divorcecases.) The number of product-liabilitylawsuits is also in sharp decline, hav-ing dropped 40 percent between 1985and 1991. Insurance premiums covering

product liability dropped 45 percentbetween 1987 and 1993. [2]

There is also concern regarding so-called punitive damages awardedby some courts. These are imposedfor particularly egregious cases insome states (punitive damages are

not allowed in many states, includingMassachusetts) and are derived fromancient Roman and English law. In fact,apart from the special and shockingcase of asbestos liability, the award-ing of punitive damages is very rare inthe U.S. Michael Rustad, of the SuffolkLaw School faculty, performed a studyshowing that between 1965 and 1990,only 355 product-liability cases resultedin punitive-damage awards in state andfederal courts, an average of fourteen

per year for the whole U.S. [3] A manu-facturer of bicycles or componentswould have to be very delinquent, orexceedingly unlucky, to be included inthis number.The remaining fear of liability lawsuits

So far I have given some details ofthe type of case that strikes fear inthe heart of small manufacturers whoare concerned that one such lawsuitcould put them out of business; andI have also tried to show that much ofthe concern is exaggerated. However,I should describe how lawsuits comeabout and are adjudicated or settled inorder to give bicycle manufacturers,

particularly those outside the U.S., anunderstanding of the risks and rewardsof exporting to the United States.

The U.S. is a country where even apoor person can sue the worlds largestcorporation. To do so she/he needs to

persuade a lawyer who specializes inthis type of case that her/his injuriesor other harms are sufficiently seriousto justify taking action. The lawyerwill generally do this on a contingent-

fee basis: that is, she/he will chthe client nothing for his/her sebut will take 2533% of any monaward. This has the socially desconsequence that people of limiwealth are given full access to tcourts in cases where they haveharmed. Although occasional laawards receive a great deal of p

juries are generally hard-headedreasonable in awarding damage

Most cases, however, do not gtrial. The early stages of a lawsutaken with discovery, a procewhich each side is required to mavailable all relevant written recand all relevant people to give dtions. So-called expert witnessehired by both sides to add weigtestimony and to act as engineescientific detectives. The discov

process can be a time-consuminruptive and costly period for a m

facturer, although the attorneysexperts costs are usually handlthe insurance company. The opping lawyers can demand, howevdrawings, sketches, notes and orecords that have any possible cnection with the injury to the plEach item considered actually ris labeled as Exhibit A, B, etc.this period the attorneys for eacare assessing their situations anlikelihood of winning or losing itrial. At some point the lead attoon one side will contact the leaney on the other side and say sothing like the following. As a rediscovery and depositions we hoverwhelming case, and your silikely to have to pay large sumsgo to trial. My client has expreswillingness to settle out of cour

payment ofX dollars. Sometimother attorney accepts the offeralacrity. More often there is a pnegotiation, as in a market anywIn under ten percent of cases agment is not reached, and a trial set. This may be several years asuit is filed.

I believe that this procedure i(with the exception of the effecinordinate delay between the coand any eventual resolution) anto social justice in the large majcases. It is difficult to be fair in where a life has been lost or ser

permanent injury has resulted fproduct defect. Suppose, for insa promising young person, just m


7/13


and just launched on a promisingareer, is permanently confinedo a wheelchair because the fork

of a new bicycle snapped in nor-mal use. No amount of money

ould compensate this person andher/his spouse and family for theerrible change in the quality ofheir lives for perhaps the nextifty years. The medical-care costs

alone could amount to a huge sum.Such cases could be regarded as thenorm in m alpractice lawsuits againsthe U.S. medical profession, whichakes extraordinary steps to prove thatvery decision and procedure taken has

been for the best. A whole battery ofvery expensive tests will often be speci-ied for a minor ailment, purely to ward

off a suit against supposed malpracticen the event that a patients recoverys not all that might be expected. A

bicycle manufacturer does not need to

o to these extremes. However, she/hemust likewise take very conscientiously,and document in some way, the designand manufacture of any componenthe failure of which could cause, witheasonable probability, serious injury or

death.Manufacturers in countries where

iability litigation is rare might welleact with some alarm at having to take

major precautions to avoid being sued,and to face unwelcome prying into theirdesign, manufacturing and businesspractices if they are sued. These riskseem to be the price we pay to have

markedly safer products in the U.S.and increasingly the safety advances

achieved in the U.S. have been adoptedn Europe and elsewhere. The European

Commission has in fact recently pub-ished The Green Paper recommend-ng changes substantially in the direc-ion of U.S. practice. [4])

I believe that the quality of designand manufacture is enhanced by thepossibility of liability litigation. Theres, however, some question about the

benefits that occur if a case is settledout of court, because of the secrecyhat is more marked in the U.S. than in,

at least, Britain. (My professional fields turbine design, and the catastrophicailure of a turbine in Britain is followed

by a full exposure of the causes, andhe steps taken to cure the problem, in

papers presented to the Institution ofMechanical Engineers.) This public air-ng seldom occurs in the U.S., exceptn the case of airline crashes. However,

I believe that the message does getbroadcast.

A sample case:fatigue in wire ropes

An example is a case in which Iserved as an expert witness. A lines-

man working on overhead wires whileon a truck-mounted aerial ladder wasseverely injured when the ladder sud-denly collapsed, dropping him to the

pavement. The cause was relevant tosafety in bicycles: the ladder was oper-ated by wire ropes that passed aroundseveral sheaves (pulleys).

The sheave diameter was only seventimes the wire diameter. The standardsset by the wire-rope manufacturers are

that the sheave/rope diameter ratioshould be 72 for long life, with 42being an absolute minimum. At aratio of 7, the rope was bound tohave a very short life before metalfatigue caused it to fail withoutwarning. When the lawyers forthe two sides agreed on an out-of-court settlement, I became verydisturbed that workers wouldbe killed or injured because, it

seemed, the information about theextreme hazard that these ladders andbooms posed would not be made pub-lic. The attorneys agreed with me thatmy professional-engineering ethics out-weighed my expert-witness responsibili-ties, and allowed me to send warningsof the extreme danger of these laddersto unions and other places. However, Ibelieve that the manufacturer recalledthe trucks faster than did any actionsresulting from my warnings: the com-

pany did not want to face the rash oflawsuits that it now knew would becertain to come. Liability litigation hadworked!

Does it work for bicycle design andmanufacture, or do we need more-strin-gent government standards? Below Idiscuss three areas in which I believethat bicycle manufacturers have notbeen as diligent as is required by theneed to protect the public.

Three examples ofavoidable defects in bicyclesFatigue failures in cables

The parallel for our industry to thewire-rope failures in the aerial-laddercase is that bicycle brake and gear-shift cables are taken around pulleysand bends with a diameter ratio of farless than 42, and also fail periodicallywithout, usually, any warning. I havehad many cables (and handlebars andcranks) break suddenly, but fortuitouslynever at a critical time. If I had, therewould have been a strong probabilityof a fatal accident, and, because bicycleaccidents are usually not investigated

with any degree of seriousness, thecause would not have become known.

A recent example illustrates the prob-lem: I normally ride recumbent bicycles.However, when snow or ice covers theroads I switch to an upright bicycle,and on such a day recently I borrowedmy wifes town bike to go to work. Welive on a steep hill, and I needed to rideup it. I applied the brakes gently toget ready to mount the bike. The front

Human Power Number 53, Spring 2002 12 Number 53, Spring 2002 Human Power

straddle cable broke. When I exam-ined the rear brake I found that it, too,had a frayed straddle cable and thatit, too, was about to break. This horri-fied me, because the next trip taken bymy wife would almost certainly havebeen downhill with our small daughterin a child seat, and the cable wouldhave certainly failed when she tried tostop at the major intersection at thebottom of the hill.

On examination, I found that thestraddle cable was attached to theright cantilever-brake arm by a pivoting

joint, where there was no sign of incipi-ent failure, but it was bolted to thetop of the left arm (fig. 1). The anglethrough which the average cantileverbrake rotates guarantees that the cablewill fail in fatigue, just as surely as willa paper clip that is bent back and forth.I put on a website an account of thisfailure, something about the design of

the brake, and details of a very-low-cost addition that would give virtuallyinfinite life to the cable (fig. 2, 3), and Ialso sent it to the manufacturers. Thisaccount is given here as appendix I.

Brake and wheel-rim failures

When I started bicycling most brakesacted on the wheel rims, and most rimswere steel. Braking was adequate indry conditions, but appallingly inad-equate when wet. When we studied the

problem, we found that the then-avail-able brake-pad materials suffered adrop of well over 90% of their frictioncoefficient when wet. [5] We also foundan aircraft-brake-pad material that hadalmost the same coefficient of frictionwet and dry. However, the level wastoo low for direct substitution in thecaliper brakes of the time. A simpleincrease of leverage to give a greaterbraking force was also impracticable,because such a brake would not givethe clearance needed for free-runningof naturally wobbling rims. We thendevised and patented a brake mecha-nism that had two sequential leverages:a low leverage that would bring the

pads up to the wheel rims with littlemovement of the hand-brake lever, andthen a high leverage that produced therequired braking force. We demonstrat-ed the brakes outstanding performanceto many U.S., Japanese and Europeancompanies, and lent them prototypes.

A description is included in appendixII, derived from an article published inVelo Vision[6].

To our consternation and disappoint-

ment, no company was interested inthe brake. After a long period therewas a fairly sudden switch to alumi-num rims. These do give much-bet-ter wet-braking performance than dosteel rims. Unfortunately, they alsowear very rapidly and are then liableto explode without warning under thehigh forces produced by tire pressures.When the front-wheel rim explodes, thewheel is likely to lock up suddenly, and

very serious injuries can result to therider [7].

This defect is similar to the first,above, in that the failures do not occuruntil the bicycle has been ridden forsome time. Most bicycles sold in theU.S. nowadays are ridden relativelylittle. Serious adult bicyclists (I amone!) are the people who bear thebrunt of these fatigue failures in cables,handlebars and cranks, and in the wearof aluminum rims. We are, alas, of little

concern to most bicycle manufacturers,or to government regulators.Dangerous run-flat

performance of bicycle tires

When a bicycle tire/tube deflates, thetire either provides directional stabilityor, more likely, produces such instabil-ity (through flopping from one side tothe other, fig. 4) that the bicycle rider isthrown off. This is particularly the caseif the tire is on the front wheel. Manyinjuries and some deaths have undoubt-

edly resulted.This problem, and an investig

into it, are described in appendiIt was found that a seriously untire (when flat) could be converto one that gave stable, controllconditions simply by improvingbetween the tire and the wheel seat (fig. 5). Further, we found tthe U.S., Japanese, and Internatstandards (ISO) there were stanfor rim diameters but there wasfor tire beads.

In my editorial forHuman Punder the heading Tiresome [8trasted the public concern overfailure that caused the fatal crasthe Concorde airliner and over tfailures on Ford sports-utility vewith the total lack of concern o

performance of bicycle tires, capossibly, a similar loss of life. Redies for bicyclists have the sam

tus as so-called orphan drugs. drugs are not developed for fatarelatively rare diseases becausecompanies see insufficient profbicycle-tire-rim case a situationindustry is not being sued enougThe m uch-maligned product-lialawyers can correct serious deficies in industry responses, or laresponses, to shoddy practice.

My sad conclusion from theseareas is that bicycle and compomanufacturers do not exercise thighest engineering capabilitiesbicycle design, and that improvare needed.

Impact of liability lawson bicycle design

The perceived impact of liabilaws in the late 1970s on the desof the Avatar 2000 which we belto be the first recumbent bicyclbe produced for general sale sinthe 1930s, was the following. Thinitial impetus for the design waconcern for safety, [9] because Iseen many reports of riders of r

road bicycles being severely inor killed after going head-first othe handlebars on applying the brakes too hard, or riding into aing or hole in the pavement, or hbaggage or a stick get caught infront wheel, for examples. It seeto me to be safer to go feet firsteasy to list, in addition, other vithat would improve safety: the nimpossibility of catching ones p

Figure 1. Cantilever brake showing cableclamped to the top of left cantilever,and typical angle through which thecable bends at each operation of thebrake.

Figure 2. Curved washer(above) made to fitunder the clamp bolton the cantilever arm,providing a large-radiusbend for the cable.

Figure 4. Tire with beads having slippedoff the bead set flops over first on oneside, and then on the other.

Figure 5. Tire with beads retained on therim bead seats runs symmetrically,giving good controllability.

Figure 3. Cantilever brake with curvedwasher installed, showing gentle curvethrough which the cable is now bent.


8/13


on the road; the great improvement inhe ability of the rider to see forward

and to the side; the improved brakingcapability on both wheels; the shorterreaction time resulting from the handsbeing on or close to the brake leversat all times; and the lessening of inju-ries because riders are closer to theground than when on road bikes. Therewere, and are, a few negative aspectso recumbents: the view to the rear is

more circumscribed unless one usesa rearview mirror; and it is difficulto recover from a skid because of theow center of gravity and the attendant

rapidity with which one is dumpedon the ground. The safety balances clearly in favor of the recumbent.

However, we knew that we would notreceive large cheques from grateful rid-ers who felt that our bicycles had savedhem from serious injury. We would

be more likely to be sued for larger

amounts in those few areas in whichour design might be worse than that ofupright bicycles.

We responded to this dilemma inhree ways:

1. We made the bicycle as safe aspracticable;

2. We gave prolific warnings aboutpossibilities of danger; and

3. We took out an insurance policyhat was standard for (small?) bicycle

manufacturers.We discussed the positive and nega-

ive features of the bicycle design withhe insurance representative, who felt

comfortable in giving Fomac, manu-facturers of theAvatar, a policy thatwould apply to manufacturers of regu-ar bicycles. There was an indicationhat, if theAvatar turned out to be as

much of an improvement in safety aswe claimed, our rates might even bereduced. This gave an added incentive,f one was needed, to increase safetyn our design wherever possible. As

mentioned above, insurance rates foriability have in fact dropped markedly

since that time.

The insurance industryInsurers are therefore major players

n liability litigation, frequently almostaking the place of the defendants in

pretrial organization of the defenseand in the trial itself. Their role is thatof insuring against risks to businesses,and of doing it in a way that is leastcostly to manufacturers (otherwisehey would go to other insurers) while

making a profit themselves. Insurershave a major stake in litigation, andhave an obligation to ensure that anysettlement is not greater than the limitsof the insurance that has been pur-chased by the manufacturer. An insurer(meaning an individual agent or thefirm she or he represents) may decideto settle out of court even though manymay believe the case to be defensible,as in the example quoted above, simplyto avoid the continuing high costs ofattorneys and expert witnesses and thelarge amount of time that its own per-sonnel will be spending on defendingthe suit.

Education and litigationDesign education has been helped

by liability litigation. At M .I.T., andIm sure at most universities, concernabout the impact of litigation on engi-neering has led to an m uch-increased

emphasis on engineering ethics andon our responsibilities to society.The disaster to the Challenger spaceshuttle was a shock that brought aboutchanges, particularly after it was foundthat engineers who had been fightingto have the launch put off because ofwhat seemed to them obvious flaws inthe low-temperature performance ofsome seals had been overruled by pol-icy-makers, some of whom were alsoengineers. Our students are shown a

videotape of a talk by one of the whis-tle-blowing engineers involved in theChallenger case, and many are movedto tears. We examine other case-studiesfor lessons to be learned. For instance,one of the first skyscraper fires in his-tory occurred in a New York buildingon the 37th floor, far too high to bereached by ladders. The first group offiremen decided to take the elevator tothe 38th floor, break through the ceilingand spray water on the fire. However,the elevator stopped on the 37th floor,the doors opened automatically, andall were killed. The elevator was oneof the first to be operated by heat-

sensitive buttons, and these naturallystopped it where the fire was blazing.We ask our students how it was thatin the several years required to invent,develop and manufacture this elevator-control system, no one in the companymaking them, nor in the architecturalengineering offices specifying the useof the buttons, ever considered whatwould happen in the case of a fire. Itseems likely that one or more people

did think of this possibility, but wereoverruled. One obvious conclusion isthat no one was concerned about beingsued for malpractice. Yet it is surelymalpractice to design and install adevice that, although it works wonder-fully for every expected use, will killor injure in an unexpected, but notunlikely, situation.

Can concern for safety go too far?Designs analogous to heat-sensitive

buttons for elevators can be found inmany areas. Only a few years ago wedrove cars that had rigid steering col-umns ready to pierce drivers chestseven in a low-speed collision. Nowwe have cars in which the driver andoccupants are surrounded by air bagsand restrained by belts and protectedby a passenger compartment that willallow people to walk away from afrontal collision at 60 km/h and higher.

Some research has found that somedrivers like to operate their vehicles atan exciting level, a level at which they

perceive a certain degree of danger.Give them seat belts and airbags andtheir average speed increases so thatthey feel the same degree of safety ordanger. On the other hand, there isin the U.S. at present an enthusiasmfor huge sports-utility vehicles, partlybecause they are much more likely tosurvive, along with their drivers and

passengers, in collisions with regularautomobiles. The safety of others,including pedestrians and riders ofbicycles, has thereby decreased. Thereis, therefore, an optimum level of safetyengineering. This level should be foundby estimating the benefit-cost ratio ofany proposed change, evaluated overthe whole affected population, not justthe users of the new system. [10] Thebenefit side of such analyses requirethe invidious decision on what value to

put on human lives saved. Perhaps itis justifiable to avoid this thorny ques-tion by using, instead, the expendituresthat could be predicted as having been

avoided in litigation lawsuits. In eithercase, benefit-cost analyses wouldindicate that some proposed safetymeasures had gone too far. It is alsocertain that safety aspects of bicycles,regular and recumbent, would be foundto have not received enough attention.We cheerfully ride bicycles with brakesthat wear fast and dont stop us safely,on rims and tires that can explode at atleast a thousand times the frequency of

those on motor vehicles, and so forth.There are several ways (research anddevelopment, industry standards andgovernment regulation being three)whereby improvements in bicycles canbe attained. We may have to depend ona fourth way: liability litigation.

AcknowledgmentsAttorneys Neil Sugarman and Phillip

M. Davis and my spouse Ellen Wilsonwere kind enough to read early draftsof this paper and to give valuableadvice and documents that I haveincorporated in the final version.

Appendix I: Serious, inevitable fail-ure on Shimano cantilever brakes

The brake cable on several types(e.g., BR-CT20) of Shimano cantileverbicycle brakes (and possibly those ofsome other manufacturers) will inevi-tably fail after a period of use. Brakesof this type have at least one of thestranded brake cables bolted to theends of the brake arms (fig. 1).

Therefore, when the brake is oper-ated, the cable is bent sharply at the

point of attachment and then bent backas the brake is released. This processis absolutely certain to fail the brakecable after a certain number of brakeapplications, just as the wire of a paperclip will fail after a small number oftimes being bent back and forth. This isa very serious danger.

Widespread standards for strand-ed-steel-wire cables (e.g., as givenin MarksStandard Handbook for

Mechanical Engineers, 7th. edition,McGraw-Hill, 1958, p. 8114) givethe minimum diameter of drums ortreads around which a cable shouldbe bent as 72 times that of the cablediameter, with 42 as an absoluteminimum in certain cases. A bicyclebrake cable is typically about 1.8 mm(0.07") diameter. Therefore the radiusof a curve through which the cableshould be forced to bend should be atleast 38 mm (1.5"). The incorporation

of a sharp bend for the cable in thesedesigns of brakes, relied upon by bicy-clists to save their lives at intersectionsand down hills, betrays a shockingignorance of standard practice.Steps that need to be taken

These brakes and cables should berecalled and replaced on an emergencybasis, either voluntarily by the manu-facturers or by government mandate.

As a low-cost alternative, curved exten-sion washers (fig. 2) could be supplied

by the manufacturers to individualsand to bicycle-repair shops, whoshould be paid for getting in touch w ithowners and for installing the washersand replacement cables. (I donate thedesign for this purpose.) That a washerof this type (a slightly different designwould be required for each type ofbrake) completely solves the cable-bending problem can be seen from fig-ure 7, showing the modified brake (inthe brake-on position) on a Special-ized bicycle.

Appendix II: The Positech Brake(Excerpt from: The Brake That Got

Away: The Positech mechanism was allset to revolutionise cycle brakingbutit never happened). [6]

In the 1960sbrake blocks were ofblack or red rubber, sometimes incor-

porating fibres. Braking in dry weatherwas superb, but in wet weather it was

abysmal and extremely dangerous. Thisseemed to me, a mechanical engineer,a crazy state of affairs. I put the topicof wet-weather braking on my projectlist for students at MIT in around 1968,and that year the first of three excellentstudents chose to work on the prob-lem. David Asbell measured the coef-ficients of friction of commercial brakeblocks on chromed-steel bicycle rimsin wet and dry conditions, and foundthat the standard black-rubber blocksuffered a loss of well over 90% of itsfriction capability when w et clearlyunacceptable for a road vehicles mainbraking system. He also tested someautomotive friction materials, andfound one that had only one-quarter ofthe dry friction that rubber could gen-erate but about three times the wetfriction. We later found a material usedin aircraft brake pads that also hadabout a quarter of the black-rubber dryfriction, but virtually identical friction

performance wet or dry. The followingyear, two students enlisted to work onthe topic, and we discussed how wecould use the new material. We could

not simply increase the leverage of thebrake operating mechanism, becausethen the pads would move only a quar-ter as far. Bicycle wheels cannot be

produced and maintained true enoughto have a pad such a short distancefrom the rim without it rubbing. Wethen hit on the breakthrough concept:a mechanism that would bring the padsrapidly up to the rim, moving with littleforce, and then when they hit the rim,

automatically switch over to a hmechanical advantage. In otheronce the pads hit the rim, insteaa hand movement on the brake moving the brake pads rapidly alittle force, further hand movemwould move them just a small dbut with massively more force. system would produce a sufficiforceful squeeze to take advantathe new material, but still give pclearance between pads and rimthe brakes were off. John Malworked on a nice design to do tusing hydraulic brakes. Howevefound that it had previously beeented for automobiles. Brian Hafor his bachelors thesis, measurmore precisely the friction behaof the new material, and subseqfor his masters thesis, worked won a mechanical braking systemachieved his objective: the inno

brake worked, although appearone would expect from an acad

project, rather clunky. MIT wainterested in patenting it, and wourselves.

Allen Armstrong of Positech duced a beautiful new design ofdouble-leverage brake (fig. 6). Hthe same locking-slider system fchanging to the higher leverageadded a feature to decrease theage during the pad-approach stathe braking action....

I also demonstrated the braketo the front wheel of a Raleigh GSport with steel rims (standard time) to Raleigh management atUS headquarters in Boston. I coshow exactly the same emergen

ping distance with the wheel wwhen it was dry.

The brake had additional advit was self-adjusting, and the paseemed to last for ever: over twfor me at a time when I w as bover 15,000km per year. It requimodifications to the bike or thelever. It could be made much ligthan was our prototype. We thothat the brake would be irresistthe companies that carried out tobtained the same or better resu...but not a single company wanto take out a license to manufacthem....

I even visited the Raleigh heaquarters in Nottingham, UK, andentertained to an impressive lunthe panelled boardroom with th


9/13


people in the company. None still rodea bicycle, and no one wanted to discussour brake. Someone stated that theywere working on another solution tohe wet-braking problem....

Within a few months the new solu-ion was revealed: the whole bicyclendustry switched to using aluminium-

alloy rims. They are much better thansteel rims in wet weather. They providea reasonable solution for people whowill travel less than 2000 km on theirbicycles. Those of us who use a bicyclefor everyday use are less well served byaluminium rims. The braking surfacewears very fast. Also, the pads pick

up pieces of grit, which cut groovesaround the rims. The rider has no indi-cation of how much wear has taken

place until the rim explodes under thehuge sideways force of the tyre pres-

sure. A rim exploding on the rear wheeljust stops the bike unexpectedly. Whenit happens on the front wheel it canbe fatal. How can this be a good solu-tion?...

While I sometimes yearn for the dayswhen I used a steel rim and a Positechbrake on the front wheel with almostno concern about any aspect of stop-

ping ability, wet or dry, I must confessthat there was always one worry. All

rim brakes heat the rims in long high-speed descents, and the heat can burstor deflate tyres, which, on the frontwheel, can lead to nasty injuries. Thereis a brake now that avoids tyre bursts,

rim explosions, and lost wet-weatherbraking: the disk brake.

How the Positech brake works

The left-hand arm L pivots aroundA, and the right-hand arm R aroundB. This R arm has a strong torsionalspring holding it open against a stop(fig. 7; neither the spring nor the stopcan be seen). The chain link on the Larm is attached to a slider which movesup and down a steel tube inside the

protective barrel. A relativelyweak coil spring pushes theslider to the top of this tube.When the brake lever on thehandlebar is pulled, at firstit does not overcome thetorsional spring, so that thebrake cable does not pull upon the R arm. Instead, theslider is moved down overthe tube, and the link pushesthe L arm quickly against therim with a low force level,because of the small lever-agerepresented by the distance l.

Further pulling on the lever causes thebrake to rotate on its pivot (unseen) sothat the R arm also contacts the rim.Further pulling of the brake cable cantmove the slider further, so it now over-comes the torsional spring on the Rarm, and presses the blocks to the rimwith the large force represented by thedistance r. In use one does not noticeany of these actions: the brake seemsto operate with a smooth m otion thatgives almost instant braking even withlarge gaps between the pads and rim.It thus automatically compensates for

pad wear.

APPENDIX III: BI CYCLE STABILITYAFTER FRO NT-TIRE DEFLATION

The problemOn three occasions I have had front-

tire blowouts, or at least rapid lpressure, that have resulted in ming been thrown off my bicyclesome violence. One was when r

Moulton road bike as a bus wasto pass; one was on anAvatar 2recumbent; and one on a new GViento recumbent, when I narroavoided being hit by a large trucfriend told me about someone wwas, in fact, killed after his fronburst, causing him to be propellthe path of a car.

The reporting from dead bicyis zero, and the reporting of andexamination of bicycle accident

perfunctory that it is highly prothat a considerable number of dand serious injuries are the resuinstability following front-tire dTherefore this has to be regardeserious problem.

Our study of the problemIn the summer of 1998 I wrotflat-tire instability to an e-mail lthen called HBS, for Hardcorescience. No one reported previstudies of this problem apart froone described by Doug Millikenwrote a letter Flat-tire directioformance toHuman Powerin 1991 (9:1, p. 17). He tested a mocycle fitted with proprietary runtires on the rear wheel. The tirea flap of rubber on the outside otire that fitted tightly over the riacted as a bead-retention systemwith a small flap did not in fact the bead when the tire was flat,bead fell into the well in the rtire flopped around, causing thebike to go unstable, even thoughtire was on the rear wheel. The tire with a wider flap held the b

place. With this tire, Milliken fohe could run the bike at high sp(80 km/h) and could perform vaous maneuvers without problemthought that good run-flat bicycwould probably be tubeless.

I also wrote to other e-mail liseveral writers contributed valuexperiences and suggestions. Soreported similar occurrences toincluding Dave Larrington of theish Human Power Club, who hainstant crashes from front-tiron regular bikes and on recumband Joshua Putnam, who consithe problem serious enough to ithe practice of letting the air co

Figure 6. (Left) AllenAmstrongs Positech brake.

Figure 7. (Below) How thePositech brake works.


10/13


y out of the front tire when trying outa new bike. Bill Volk wrote, I too findhe situation to be unacceptable. I run

heavy, inefficient thorn tubes becauseof my fear that a blow out at high speedwould be a disaster. Why cant we haveims that retain the tires even at nonflation? And perhaps a rubber striphat is placed around the rim, under theube, that supports the bike on loss of

air pressure. I had Performance semi-lick 26" tires that fit so snugly that

you could safely ride no-inflation. Thathould be the standard.

Presumably because of a tight-fittingire, Ed Deaton of Fools Crow Cycles,aced with difficult choices, rode 8 km

on a flat front tire: he had IRC Road-ites with Sun M14 rims. Similarly Andy

Milstein of Princeton had no troubleiding with a flat front tire. It was a

Tioga Comp Pool, measured by Mark B.of Wheel Life Cycles to be 46 mm wide,

on a Sun CR-18 20 1.75" rim of about7-mm width. (That was significant

because one of my early suspicions,and a concern of Larry Black, was thata wide tire on a narrow rim might havea greater tendency to flop alternatelyeft and right. This suspicion was thushown to be unfounded.)

Bill Volk mentioned that SutherlandsHandbook for Bicycle Mechanicshada good section on fits between differ-

nt brands of rims and tires, but mydition did not have this section, andcould not get an answer to my lettero Sutherland asking about standards

of fit. John Allen, prominent bicyclexpert and author, sent me a copy of his

Japanese Industrial Standards D 9421,Rims for bicycles, giving a tolerance

of 3 mm for rim circumference, andof standard K 6302 Rubber pneumaticires for bicycles, which, he pointed

out, gave neither tolerances nor dimen-ions of tire beads. (Later, Andy Ouryound that the International Standards

Organization ISO 5775/1 Standards forbicycle tires and rims also had toler-ances for rim diameters but not, as faras he could determine, for tire beads).

My instinct tells me that the old inchizes had some specified or custom-

ary standards because my old 27 1"and other inch tires were all at-leastgood fits on the rims. Now, it seemsrom our experience and that of many

people who wrote to me, it is entirelyby chance that one gets a tire that is aight fit on a rim and that will therefore

provide a substantial degree of safety

in the event of a front-tire blowout.However, Doug Milliken, a long-timeconsultant to Alex Moulton, wrote thatMoulton controls both the rim diameterand the bead size of his 17" tires.

In September 1998 I added the prob-lem statement on flat-tire stability to mylist of undergraduate-thesis topics atMIT. Andy Oury, then a senior, respond-ed enthusiastically, carried out several

valuable experiments, and has allowedme to report some of his results here.We drew up a too-ambitious programin which we wanted to look not only atbead retention but also at the effect ofthe ratio of tire width to rim width (ATBtires in particular are usually bulbous,having a pear-shaped cross-section onwhat seems like a small rim) and of tire-sidewall stiffness. Andy Oury workedon what the correspondents just quotedthought was the most important factor,bead retention.

The experimentsWe first thought that we could do a

highly controlled experiment by havingmy troublesome bicycle wheel and tire,held in a frame, running on the surfaceof an inverted portable belt sander.However, the tire did not display theextraordinary alternating flops, left andright (fig. 4), that had thrown me off mybike, and that had prevented me evenfrom pushing the bike subsequently(fig. 5). Oury found that, for the flop-

ping behavior to occur, he had to rig upa bike to run along a simulated road-way with a similar number of degreesof freedom as has a bicycle when it isbeing ridden (or pushed).

The simplest way of producing beadretention on the shoulders

Documents

HUMAN POWER No. 53