Reflections on the potential of human power for transportation

Saturday, March 31, 2012

Pedaling Along the Skyway

I admit to being a futurist. I have this fascination for the overlap between science fiction and technology. Some of the concepts that fall into this category keep resurfacing, but never become reality. In some cases the timing has not been right. In others the technology is not adequate to make the concept viable. And in some, there are social factors that prevent realization. Take jetpacks or flying cars. How about human-powered monorails? Admittedly, HPMs do not surface as often as the former two concepts, but I have read about at least six systems built before 1900 (see Wilson below), and at least five proposed since 1990. I consider these offshoots of the modern Human Powered Vehicle movement. Is there a system coming to a neighborhood near you?

The people at Google with their 10^100 Prize would have you think so. They have awarded one-million dollars to the makers of a human-powered monorail called Shweeb, the winner in their “Drive Innovation in Public Transportation” category.
We’ll get back to the Shweeb’s practicality after reviewing a bit of HPM history
Prof. David Wilson discusses the pre-1900 HPMs in his article “Transportation Systems Based on HPVs, Past, Present and Perspective”, presented to the “Proceedings of the Fourth International Human Powered Vehicle Scientific Symposium” from August 1992 (parts of which are also in “Bicycle Science” 3rd ed.).
The contemporary interest in HPMs may have begun with a 1990 article in “Bicycle Guide” magazine from June of that year. In that article, entitled “Original Thinking”, seven leading bicycle builders were asked “to tell us about the bike they would build if the prosaic constraints of cost, customer whims and current technology could all be magically lifted.” One of the builders, Gary Fisher ( of mountain  bike fame), responses was “I’d love to see elevated bikeways built like the wooden velodromes, where you’re totally covered in a pipeline that would go in one direction, and you’d have air piped in , in the direction you’re going, so you’d have a perpetual tailwind. Its quiet, its lightweight, and you could elevate it 20, 40, 60 feet in the air to eliminate hills.” The water color painting at the beginning of this post was David Graves’ artistic interpretation of Fisher’s idea.
Jim Kor (of Urbee fame, see the post “Bucky and the Urbee”, below) borrowed heavily on Fisher’s idea with he proposed his Skyway system from 1992. Below are two illustrations from a “Bicycling” magazine article on the Skyway system, from March of 1992, entitled “Cycle City”. In this article the Skyway system has been installed in Seattle using much the same route that is being considered for light-rail. Wouldn’t you know it? The system doesn’t extend far enough North along the Eastside of Lake Washington for me to have easy access to it!

Kor embellishes Fisher’s vision adding specifics like hard rubber tires running in tracks, multiple lanes for vehicles of differing speeds, a double-decker track arrangement  and the suggestion that  the system could support electric vehicles as well. Notice in the Skyway car illustration, the absence of steerable wheels. This would restrict it to very-large-radius curves. Kor does manage to get a lot more press for his concept, including an article in the January 1994 issue of “Popular Mechanics”.

And Bauke Muntz, the talented industrial designer from the Netherlands, proposed a HPM for his Velo Nova HPV theme park in the early 2000s.

Speaking of theme parks, that brings us back to the Shweeb, which has a working prototype in Agroventures theme park in Rotorua, New Zealand.
Before discussing the Shweeb’s transition to public transportation, there is another very Shweeb-like HPM, the Skyride, being championed by Scott Olsen, the inventor of Rollerblades.

While Olsen’s concept of a suspended gondola system may have originated in the mid 1990’s, the building of actual prototypes appears to be relatively recent. The current iteration of the Skyride gondolas can be propelled by rowing as an alternative to pedaling. It is clear that Olsen’s current efforts to commercialize his system are riding the wave from Shweeb’s winning the Google 10^100 Prize.
I am sure there are other HPM concepts I have overlooked, but for the purposes of the subsequent discussion, the Shweeb makes a sufficient point of departure.

The Shweeb is the brainchild of Geoffrey Barnett. The bullet shaped pods are coupled to the overhead rails by a boogie that allows the pod to swing +/-70deg. from side-to-side, like the inverse leaning of a bicycle. The rails are made of an 8” square C-section, and though I don’t have access to the design details, it appears that the drive force comes from a small wheel (less that 8” in dia.) rolling on the inside of the C-section. Pedals are connected to the drive wheel through a seven-speed derailleur system and universal-joint that allows for simultaneous driving and swinging. The installation at Agroventures features two 200m tracks arranged in what appears to be a double-D-oval pattern.
Barnett claims the Shweeb is the most efficient human-powered vehicle on earth and that speeds of in excess of 50mph. are possible. The current speed record for three laps is 58.2 seconds (23mph.) To put this speed in perspective, the world record for running 600m is about 72 seconds, the hour record for an HPV (Varna Tempest) is 56 miles and the flying 200m speed is over 82mph(Varna Tempest again).  Since the pods do have a small cross-section and have an aerodynamic shape, I suspect that the discrepancy between Barnett’s claims and reality is due to the rolling resistance of the small wheels in the track. If the rolling resistance is that substantial, then claims that two pods in tandem will travel faster that one are not valid.
The Shweeb is, in fact, a very expensive version of the bike lane. Granted, being elevated, it can make use of space unavailable to regular bike lanes, but I believe the high cost and very limited user group will prohibit its adoption as a transportation system. I also believe there are two broad changes  
that can be incorporated into a HPM which may bring implementation closer to reality. The ideas I am proposing have been suggested by others, I am only grouping them together as a holistic system.
First, minimize the cost of the system by doing three things. Limit the system to two lanes only, one in each direction. This is simple enough and in keeping with the Shweeb concept for public transportation. Reduce the height of the tracks. Shweeb advocates tracks at 19’ height. Reduce this so the bottom of the system is about 8’ high, enough to safely clear a walking person. An approach for dealing with elevation changes is discussed below. And, most significantly, don’t supply the vehicles. Put all the money into the infrastructure and let the users provide the vehicles. Vehicles could be supplied for rental fee in addition to a system-use fee.
Second, design the lanes so they can be used by multi-modal vehicles. Many communities have transit buses that will carry bicycles, allowing riders to pedal between home and the bus stop and again from the bus stop to the destination. So design the HPM cars so they can be ridden on the street from home to the nearest skyway access node. Kor suggested four-wheel cars with hard rubber (urethane?) tires that  
would run in metal tracks. Provide adequate suspension so the ride on roads is acceptable and keep the spacing between the wheels the same so they can share the same track.
 Miles Kingsbury’s Quattro with a bigger trunk?

Since there is only one lane in each direction, there is the problem of slower cars being over taken by faster cars. This can be prevented by regulating car speed and keeping it the same for all vehicles. As each car enters the skyway from the street an electrically-motorized tow-hook engages the vehicle. The tow-hook, or traction rabbit (see Wilson in Proceedings above), runs between the tracks below the vehicle and accelerates the car up to a speed that is slightly faster that a very-fit rider can propel it. The rabbit has force measuring sensors within it that monitor how much force is needed to pull the car along. If the rider chooses not to pedal then, in effect, his car becomes a purely electric vehicle and the user fee reflects that fact. If the rider chooses to pedal, the force sensor records the reduction in required force, and the user fee is reduced up until the level where the rabbit is only limiting the car speed but the rider is providing all the propulsive energy. The rabbit also can provide extra propulsive force to overcome inclines so the tracks can follow the landscape instead of being high enough to pass over all obstacles and remain nearly horizontal.
The cars could be purely electric, with no provision for pedaling. The rider could use the onboard battery energy to get to the skyway, use the rabbit for the skyway transit to conserve battery energy and use the batteries again to go from the skyway to the destination.
The multi-use skyway cars and the multi-use skyway would maximize ridership and make what could have only been an expensive bikeway into environmental friendly commuter system.

Thursday, March 15, 2012

Recumbents and Convergent Evolution

By the late 1870’s bicycle innovators realized that there needed to be a safer alternative to the high-wheel bicycle that had become so ubiquitous. All manner of design approaches to lower the rider and move him backward prevailed, each with the intent of preventing headers, the rider pitching forward over the handlebars. These included gearing mechanisms so the size of the drive wheel could be reduced while keeping the ratio of pedaling speed to vehicle speed the same. Four bar linkages were used that moved the rider rearward as well as levers with overrunning clutches. Separate crank arms being coupled to the wheel by separate chains allowed the pedal location to be lowered, and there was the novel approach to use a single chain to drive the rear wheel. This of course was the safety bicycle and not long after its appearance, the other approaches faded away. A common design was converged upon and human-powered vehicle technology moved forward.
Now if someone were to have asked me, I would have told them that recumbents haven’t converged on a common design that satisfactorily addresses what I will call "the recumbent packaging problem"; that being, that for a proper weight distribution, the pedals want to occupy the same volume as the front wheel. But after a bit more reflection and staring at pictures of the participants in last summer’s “Roll Over America” velomobile tour, I realized that recumbents have converged on a common design, and a pretty good one at that.
I saw a post on about an Italian designer who was developing a rear-steering-recumbent bicycle.!/2012/01/rws-recumbent.html .  He was trying to solve the recumbent packaging problem.

 Now there a number of solutions to this problem. It turns out that rear-steering is the least feasible of those solutions. I should know, I spent 10 years building rear-steerer prototypes.

The Plywood Flyer from 1978

 The VelAero from 1987

In the end, the VelAero, could be balanced at moderate to high speeds, but could not be ridden slowly without weaving drastically. So I wrote up my results, published them in “Human Power” and moved on.

One place to put the cranks is behind the front wheel, like the Avatar and many other long wheelbase recumbents. The upside is a very comfortable riding posture and ease of balancing. The downside is a very long vehicle that is difficult to transport and a lightly loaded front wheel that can slip laterally on all but the driest of surfaces. Detractors to this layout claim the chain is too long. I disagree. It minimizes the angle the chain makes with the sprockets in the extreme gears and it extends chain life. 22 years of riding an Avatar, breaking the seat twice, sawing through a front derailleur, breaking a crank and breaking the frame but never having to replace a chain.

The other pole is to place the cranks far ahead of the front wheel. The vehicle shown is the aptly named Hypercycle. It is very compact but also suffers from poor weight distribution with a heavily loaded front wheel. The heavily loaded front wheel also resulted in very skittish handling.

The cranks could be located well above the front wheel like the classic racing Velocar of the 1930’s. The package is compact, and the front wheel loading is similar to a long-wheel base design. The downside is that the rider’s feet end up a significant distance above the ground, especially if the front wheel becomes large. This makes it difficult to get started from a stopped condition.

The cranks could overlap the front wheel with the bottom bracket near the steering axis. This is a picture of the final crank location of the EcoVia leaning tricycle. This limits the steering angle, but surprisingly, for medium wheelbase layouts (40-48”), you don’t need much more than +/-20deg. of steering lock for all but the tightest of turns. George Georgiev has made good use of this fact with his Varna speed demons, where the rider’s legs and cranks overlap the front wheel to the extent that only a few degrees of steering lock are available.

Another way of having the cranks overlap the front wheel, without limiting steering lock to the degree above, is to allow the cranks to pivot with the steered wheel. The worked satisfactorily for high-wheel bicycles, where the direction of the pedal force was roughly parallel with the steering axis, but with recumbent posture the pedal-force direction is closer to being perpendicular to the steering axis. Thus pedaling forces will perturb the steering and make the vehicle difficult to ride. (See the post “Arm-Power and the Avatar” below.)

A more complex version of the cranks moving with the steered wheel is having the cranks pass through the front hub. In the Evolution Bike above, the cranks drive an infinitely-variable transmission in the hub. An interesting note, the concept recumbent shown above is nearly identical to a Polish entry in Prof. David Wilson’s “Human –Powered-Land-Vehicle Design Contest held in 1969. That vehicle used a hub transmission as well but most interesting, the laid-back steering axis passing through the contact patch of the front wheel was the same. Convergent evolution? In both designs the pedal force is closer to being parallel to the steering axis than perpendicular, so pedaling perturbations to the steering might be manageable. A most attractive package, but again the driving wheel is seeing significantly less than half of the vehicle’s weight. Another issue with hub-centered pedals is that the bottom bracket height is half the wheel diameter, so the wheel must be large enough to prevent the rider’s heels from hitting the ground when pedaling.
And yet another variation of hub-center pedals it to fix the cranks and pivot the wheel using a hub-center steering approach.  The steering lock is again restricted by the rider’s legs but the posture can be comfortable and steering is not perturbed by pedaling. Of course you must couple the cranks to the rear wheel (or front wheel for that matter). This approach requires a lot of non-standard components and has rarely been used.  

A pedal path that is elongated horizontally to minimize pedal/wheel overlap can be used. The pseudo-linear linkage above was tried on the EcoVia. Notice the two-sided pedal where the rocker link was outboard of the pedal and the coupler link was inboard of the pedal. It packaged very compactly. Unfortunately, the harmonic-velocity linear motion had pedals that came to a stop at the ends of the stroke. This significantly reduced the maximum pedaling cadence and made climbing of any type of hill prohibitive.

To eliminate the extreme velocity fluctuations associated with linear-type drives, but to preserve the squashed pedal path, I played with an egg-drive concept. A triangular link holds the pedal. This link is supported by a rocker link at the rear and a rotary crank at the front. The big circle below the egg is a 20”dia. wheel. The smaller circle behind the egg is a conventional crank for comparison. This approach is obviously a bit over the top in complexity.

And then there is the most unconventional means of addressing the packaging problem, ride the recumbent backwards. The steered wheel is in front and so is the riders head. The drive wheel is in back straddled by the rider’s legs and the vehicle conforms nicely to a teardrop cross-section with rider’s shoulders in front tapering to the rider’s feet at the rear. The idea has been used at least three times. By the inventor Milt Raymond in the mid 1980’s, by Ohio State with their “Buckeye Bullet” at one of the human-powered speed championships and most recently and most successfully with the Eiviestetto, which just wrested the hour record from the Varna Tempest (see the post “Back to the Future” below) covering almost 57miles in an hour.

Of course, there is an “out of the box” solution to the pedal/steered wheel interference problem, but purist may feel the original design intent is violated. Use two front steering wheels, place the pedals between them and drive the single back wheel.

This layout dates back before WW2 with the Fantom from Sweden. Supposedly 100,000 copies of the plans for these were sold. The layout was resurrected by Mike Burrows with his Windchetah in the late 1970’s, again used by the Vector team to break the break HPV speed records in 1980, by Carl-Georg Rasmussen with the Leitra and it has become the default layout for most velomobiles. It was while staring at a picture of a line of recumbents winding through a Portland OR. park at the beginning of the “Roll Over America” velomobile tour that I realized that seven of the nine visible vehicles looked essentially the same. They were based on the Fantom layout. The two outliers were a long-wheel base recumbent bike with a partial-fabric faring and Miles Kingsbury’s Quattro.

Sure, you could argue that a velomobile, by default, is an enclosed trike of the Fantom layout, but since velomobile is a generic term for an enclosed recumbent bicycle or tricycle, this only enforces my contention. The recumbent HPV has converged on this design approach, which I will refer to as nufantom.  
The move toward commercial recumbent trikes appears to have been going on for some time, and now that I think about it, I have seen more Nufantom trikes than recumbent bikes on the local bike trail. It is not surprising. A new rider can adapt to riding a tricycle a lot faster than becoming confident on a recumbent bike. I have only seen a few enclosed trikes, but was recently very impressed with a Quest from Blue Velo.
Converging on a common solution allows designers to concentrate on refinements, since they don’t have to worry about the overall vehicle layout being feasible. The Quest exhibited a lot of refinements to the nufantom approach. These included enclosed wheels for improved aerodynamics, cantilevered wheels all around for ease of flat repair, without removing the wheels, a monocoque chassis, a well thought-out suspension, Burrows-style u-joint steering controls and a very trick cover for the rear derailleur.
 As impressed as I am with the Quest, it doesn’t meet two of my criteria for a Human Powered Commuter Vehicle.  Most significantly it is too low to provide adequate visibility in traffic, and, case in point, I have only seen these vehicles on the bike trails. The other is ease-of-entry and riding posture. I suppose I could get used to the laid back seating, but my old back wouldn’t be too happy without a lumbar support. Entry and exit, however, appear to be extremely difficult, very similar to getting into a formula one car. The upside is the vehicle is statically stable for this process.
I have long felt that the recumbent bicycle is a transient form, evolving toward another vehicle. Streamlined recumbent bicycles will probably always hold the speed records, but for a utilitarian vehicle the Nufantom-style trike seems to be the most popular option.

Saturday, March 10, 2012

Arm-Power and the Avatar Recumbent

For short term, anaerobic, activities the energy available for physical activities is related to how much muscle mass can be recruited for the activity. The limitations imposed by the cardio-pulmonary system do not apply. During long term, aerobic, activities the rate of oxygen delivery to the muscles is the limiting factor in how much power an athlete can produce.
Back in the day, (prior to the HPV movement in the mid-1970s) the reining wisdom was that a cyclists employed enough muscle mass with their legs while pedaling to utilize all the oxygen the cardio-pulmonary system could deliver. Nothing would be gained by adding more muscle groups in propelling the bicycle. I strongly suspect that this idea came out of the ban of unconventional bicycles that was initiated by the Union Cyclist International in the 1930s and was not based on empirical data.
In the second edition of their book, “Textbook of Work Physiology”, Astrand & Rodahl do a literature study comparing the maximum oxygen consumption (Vo2 max) of various activities. If running uphill is scaled at 100%, horizontal running is between 95-100%, upright cycling is 92-96%, supine cycling is 82-85% and arm and leg cranking (where the load on the arms is 10-20% of the total load) is 100%. The above data must be taken with a grain of salt since it is a compilation of different tests. It does serve my purpose though, and that is to indicate that more power can be obtained by adding arm power to leg power, especially for pedaling in the recumbent position.
A major advantage to adding arm power to a recumbent is that the arms are not as involved in postural support as with an upright bike, and therefore can perform the task of propulsion while the rider remains stably mounted on the machine.
In the post “Rx for a Healthy Commute” I extol the comfort I experienced in riding an Avatar 2000 recumbent over a period of 22 years. What I didn’t mention was I never obtained upright-bicycle speeds while riding it. I attributed my lack of power to coming from a runner’s background, where more of the leg force came from the gluteus and hamstring muscles as opposed to the quadriceps muscles. I suspected that sitting on the major force generators significantly reduced my supine power output, but again I have no empirical data to substantiate this notion. On the other hand, I estimate that my recumbent speed was only about 80% that of my upright speed. This would equate to a drastic loss of power!
The bottom line, though, was I always felt I had cardio-pulmonary capacity I wasn’t using and the only way to remedy that would be to add more muscle mass to the job of propulsion.
So the question was what type of arm motion should I use?
There was an arm and leg cranked recumbent from the mid 1970s called the Manuped, invented by John Carl Thomas. It had leg cranks passing through the front wheel and hand cranks mounted on a fork above that wheel. The entire assembly pivoted to steer, and the seat and rear wheel comprised the rear portion of the frame. That’s Fred Tach (sp.?) riding a Manuped during the 1978 International Human Powered Speed Championships. I talked to Gary Hale who was riding a Manuped at the 1990 Speed Championships in Portland. Gary had made some of the frames for the Manupeds. When I asked him about the bike, he confided that it was very difficult to ride because of the arm and leg force inputs perturbing the steering. So I had a reason to not use the hand-crank-and-steer approach.
If one wants to avoid the differential force application of propelling and steering the alternative to hand cranking is a rowing motion. While rowing does not make as good of use of the legs as cycling, it involves the muscle mass of the back in addition to the arms. The decision was made to use a rowing motion for the arms and conventional pedaling for the legs. The two motions were not at all incompatible and worked quite well.
The basic approach is to have a rowing frame that pivots at the recumbent’s top tube and during its rotation pull a chain over a 16-tooth single-speed freewheel  attached to the left crank arm. The chain must pass under the freewheel to move the cranks in the correct direction, and therefore an idler is required to route the chain backward where it is kept taut by a screen-door spring attached to the bike frame.

 The handlebars are pivotally mounted within the rowing frame and the steering axis is perpendicular to the  rowing axis. The lever length of the rowing frame is 18.5” and it can rotate through 90deg.

The lever length for the drive chain attachment is adjustable from 3.5” to 8” in six, 0.9” increments. The chain is attached to the rowing frame by an old-style master link so it can be easily moved from one position to another. To determine the most-natural feeling position, I started riding down the street at 3.5”. After about 100’, I stopped because that was too easy and moved it to 4.4”. I rode about a quarter mile and moved it to 5.3”. By the time I was a mile away from home I was moving it back and forth between the 6.2” and 7.1”. I settled on the 6.2” position.  The resulting arm to leg motion during the power stroke (pulling the rowing lever from its most forward to most rearward position) equates to slightly less than one pedal rotation. Since the return stroke is faster than the power stroke, two pedal strokes are approximately equal to about 1.5 arm strokes.
Since less than 150 Avatars were build (mine is #085) I wanted to preserve its value, so the arm-power mechanism and the idler clamp around the frame tubes. The arm power mechanism includes a diagonal brace to stiffen the top tube against the bending moment imposed by the rowing. This approach could easily be adapted to any recumbent with the only major modification being the mounting of a single-speed freewheel on the left-hand-crank arm.

The handlebars are attached to a tube that pivots on bronze bushings that are in turn held by the rowing frame. The bottom of the handlebar tube is attached to a steering knuckle. The steering knuckle is connected to the fork by a tie-rod with spherical joints at both ends.
The key to why this design works so well is that the steering and rowing motions are decoupled. Rowing does not perturb the steering and steering does not require rowing. This occurs because the axis of the tie-rod bearing at the rowing link end passed through the rowing axis. As a result, the rowing motion does not move the tie-rod and does not perturb the steering. In fact the steering appears to be more stable while rowing. I surmise that this is because the rowing forces applied to the handlebars dampen out any unintended inputs.

Cable routing details, at handlebars and at the steering knuckle.

My seat was getting a bit wobbly so I took the opportunity to attach struts between the lower end of the diagonal brace and the ends of the seat tubes. I did loose seat adjustability by doing this. 

The only downside I can see with this approach is the added weight, added friction and the loss of the under-seat steering which I find so comfortable. On the upside the speed is significantly improved, especially on hills and it makes the bike a lot more fun to ride. It is especially fun to race along next to someone without using the arm power and then kick it in. You should see their faces! And because of the front wheel is lightly loaded, rapid application of rowing power has resulted in wheelies on a number of occasions; quite unheard of for a recumbent bicycle!