Reflections on the potential of human power for transportation

Thursday, November 29, 2018

Aerion: An Optimized All-weather Ped-electric Trycicle

Those of you that have followed my blog at all know my opinion on what constitutes an optimal ped-electric commuter vehicle.

1. The vehicle should have three wheels for stability on slippery surfaces.
2. It should place the rider's head at the height of a typical automobile.
3. It should be narrow enough to comfortably fit in a bike lane. I can be more specific on this. It should fit between barriers spaced 36" apart, so assume a maximal width of 34".
4. It should not overturn when cornering. Assume a tipping resistance of 1 gee.
4. It should be enclosed to protect the rider from the elements.
5. And for traction on slippery surfaces, it should drive two wheels.

Now if the vehicle is a static tricycle point 2 and 3 are at odds with each other. High enough rider and wide enough to not tip over in turns results in a vehicle that has a width of about 48".

The solution to this contradiction is a leaning tricycle.

When correctly designed, leaning trikes can be balanced just like bicycles. Like bicycles, the most difficult task is getting started from a stop. One has to develop balancing speed often with one pedal thrust to avoid tipping over or having to put a foot down and start again.

Problems appear when the leaning trike is enclose in a body that interferes with the action of starting.
To remedy this potential problem the vehicle need a mechanism to lock-up the leaning for starting, and stopping. The locked-up mode could also be used in slippery conditions when balancing is not practical. You now have two modes of operation, balancing and being stable. The rider has to switch modes and decide when to do so.
Refer to the Velotilt design where the leaning needs to be locked up when the rider gets into the vehicle and starts moving or when the vehicle is stopping.

I discussed the problems I had starting and stopping my EcoVia trike in the post below.

When a fellow HPV-er tried to ride the EcoVia he pointed out the average rider would have trouble mentally switching from one mode to the other. I could not disagree, having once forgotten that I was in the balancing mode and thinking I was in the static mode. This resulted in a swerve that may have ended my HPV career had there been any other cars near me at the time.

So how to address this problem?

I believe you must have the static mode for starting, stopping and slippery conditions. The balancing mode is the most problematic so eliminate that. Since you still need to lean the trike to prevent tipping over in corners, make it controlled by the rider by means other than balancing. 

Now since accelerators, controllers and stepper motors are inexpensive, the logical approach for controlling lean for an electric vehicle is computer control, with it being transparent to the rider. Since lean angle is a function of radial acceleration, which in turn is a function of steering angle and vehicle speed, the necessary calculations are relatively simple.

Lean control for non-electric vehicles is more problematic.

GM's Lean Machine from the 1980's used foot pedals, since the vehicle as gas-powered.

A human-powered vehicle more likely than not uses foot pedals for propulsion, so that is not an option.

One could combine lean control and steering into one action. I tried that with the first version of the EcoVia, but the approach introduced bump-steer with disastrous consequences.

Or one could use dual-control that used two hand motions, one for steering and one for leaning. The Tripendo uses this approach with apparent success. It employs two long levers. One does the steering and one does the leaning. The rider continually adjusts both to keep the trike stable.

Dual control could be used without requiring continuous lean adjustment. For example, the MK3 version of my EcoVia has a c.g. height of approx. 23 in. an a track of 30". This allows for a .43 gee turn without needing to lean.


One can calculate the required geometry for a fixed lean that can resist one gee just at the limit of tipping over.

For a vehicle that leans by having its wheels move essentially vertically, staying parallel to the vehicle, like the Toyota iRoad above, the relations below apply:

The first equation calculates the relation between width, height and lean angle at the limit of tipping. The second equation calculates the relation between the width, height and lean angle that will resist one gee laterally without tipping.

Interestingly, the lean angle for this condition is not a function of w (track) or h (cg height). It ends up being 26.6deg. Not surprisingly,  The lean angle of the iRoad is about 26deg. And the iRoad width is about 34"

The cg. height for the above conditions is a function of the lean angle and the effective track, w. If the actual track is 30", the effective track is approx 2/3 of that for an equal-wheel-load configuration or 20". For a 20 track the required c.g. height is 25".

 Let us compare the gee limits for several configurations with this geometry.

1. Statically, without any lean, it would take .4 gees to tip over.
2. For balanced leaning, it could withstand .5 gees. The vehicle would have to lean 45deg to resist 1 gee.
3. If the vehicle had four wheels where the front and back track was equal, it could withstand .6 gees. The track would have to be 50" to resist 1 gee statically.
4. And with the lean locked at its 26.6 deg limit, it could withstand 1 gee.

Clearly, for a given track width, a tricycle with controlled leaning gives the greatest resistance to tipping.

So our optimal all-weather ped-electric trike will have controlled leaning where the leaning is determined by computer.

What about the wheel layout? Should it be a delta trike or a tadpole trike. Due to drive-train complexity I will eliminate the tadpole configuration where the front wheels drive and steer. That means that if the two front wheels drive, the single rear-wheel steers. (Like the iRoad).

Below is the EcoVia Mk3. The long lever controls the leaning while steering is located below the seat.

The biggest advantage of the delta configuration is the pedals can overlapped with the front wheel to produce a very short vehicle. The vehicle can be lighter weight because the structure supporting the front wheel can also support the pedals. Along with these advantages comes three disadvantages. The steering lock is limited. Not enough to be a problem for normal turns but enough to prevent tight turns. Only one wheel is available up front for braking where the load is transferred to during hard braking. Lastly, the steered wheel is the first wheel to encounter obstacles like railroad tracks. The obstacles can perturb the steering and unexpectedly change vehicle direction.

Below is the EcoVia Mk1

The biggest advantage of the tadpole configuration is it doesn't suffer from the three issues plaguing the delta configuration. On the other hand, the package is longer that the delta configuration because the pedals are on one end and the steered wheel is on the other. The space between the seat and the rear wheel could be used as a trunk. And there is the issue with rear-wheel steering. Since the controlled leaning approach does not involve balancing, this should not be a show-stopper. After several communications with Jim Kor of Urbee fame, he described his approach to make his rear steering stable. The Urbee without body is shown below.

I believe despite its problems, the tadpole configuration wins out.

Below is a drawing and photo of the Cyclodyne, a tadpole trike from the early 1980s that was capable of cruising at over 30mph. The Cyclodyne, of all the commercial human powered vehicles, came the closest to harnessing the potential of human power as a transportation alternative.

To resist tipping rider's head height was about 39" and the vehicle width was about 48". Roll-over resistance as at least 1gee. While the rider's head height was acceptable, being comparable to a sports car, the width was too wide to fit comfortably on a road shoulder.

Had it been made a quad by adding a fourth wheel, the width could have been reduced to the target 34" and speed would have been increased as well, making it a very practical commuter vehicle. But the fourth wheel would prevent it from being considered an electric bike if a motor was added.
Adding controlled leaning could have reduced the vehicle width, increased visibility by raising the rider's head height and allowed it to be electrified. An optimal all-weather tricycle configuration.

Even though it only exists as a few equations and a few crude sketches, I have a need to give a name to this all-weather, ped-electric, fixed-leaning, front-driving, rear-steering tricycle. I will call it the Aerion. (Aerion is also the name of an aviation company developing a supersonic business jet).

I have a bias for the Toyota iRoad. My first configuration in 1990 for the EcoVia was a front-driven, rear-steered leaning configuration. I was not using controlled leaning but was balancing the vehicle and I could not make the rear-steering stable. More than two decades later, electronic controls allow the iRoad to do what I couldn't and reap the benefits of an optimal configuration.


Wednesday, December 23, 2015

The Velotilt: a Pedal-Powered Commuter Trike for a Bike-Friendly World

In 1980, after the Vector trikes shattered the Human Powered Vehicle Championship speed records, people began to believe that human-powered vehicles could travel fast enough to be viable transportation vehicles. The ultimate nod to HPV recognition was an illustration of a Vector trike on the cover of Scientific American magazine. The age of pedal-power had arrived.

Well, not really. Vehicles like the Vector were low to the ground, and, as a result, not very visible in traffic. Their extreme streamlining kept the wheel-track narrow so high-gee turns were not possible.
One vehicle, the Cyclodyne did come close to being a practical commuter vehicle. A tadpole-layout trike with front-wheel drive and steering, the Cyclodyne could be pedaled over 30mph by a fit rider. Its practicality was compromised, however, by its width. It was too wide to fit in a 36” bike lane next to a car lane, but not fast enough to keep up with 35mph traffic in the car lane. A noble effort, but a dead-end.

Now to some extent, the lack of a fast commuter vehicle may be as much a problem with a lack of suitable places to ride them as the vehicles themselves. The really fast vehicles (>30mph) are low to the ground for minimal aerodynamic drag. Such vehicles are difficult to see in traffic. On the other hand, dedicated bike paths, where vehicle visibility may not be an issue, have 15mph speed limits that prohibit high-speed vehicles. (Riders on the Burke-Gillman trail in Seattle have received tickets for exceeding 15mph.)

If there was a bike-commuter infrastructure that separated bicycles from car traffic and allowed for very-fast pedaled vehicles, Wim Schermer’s Velotilt would be the vehicle to ride.


The Velotilt is a delta trike where the front wheel is driven and the rear wheels are attached to beams that can pivot. The act of pivoting allows the Velotilt to lean into corners like a bicycle.


 The degree of leaning is limited so the vehicle cannot tip over statically. The beams are interconnected by a mechanism that causes one beam to move down while the other moves up. The middle link in the mechanism is composed of a sector of an arc. As the mechanism moves the sector rotates. This sector can be clamped in any portion of its extremes of travel and thereby lock the leaning.


In addition the entire mechanism can translate, allowing both beams to pivot together. This motion is resisted by a mountain-bike shock and provides suspension for the rear wheels.
The rider’s legs wrap around the front-steered wheel. The cranks are located in front of that wheel and the wheel is driven through a very pricey Rolloff 14 speed internal-geared hub.

All three wheels have disc brakes but in all the photos, the rear disc are lacking calipers and actuation cables.

Commercial vehicle weight is expected to be 55lb. The vehicle height is approx. 40”.

The Velotilt body and the rear-wheel pants make the vehicle extremely aerodynamic. Schermer calculates that it will take 150W to do 40mph. Bicycling Science 3rd Ed (D.W. Wilson) estimates that an ultimate HPV would require 200W to do 40mph, so the efficiency of the Velotilt is indeed impressive.
An optional 750W motor could help propel the Velotilt to 60mph for a duration of over 2.5 hours, but where would you ride it? Too low to mingle with car traffic and too fast to mingle with regular bicycles.

Possibly, one could transform the high efficiency into something other than raw speed.
For example, the Velotilt might be able to sustain a constant 15mph on solar power alone. Or the vehicle could be made higher than its current 40"to be more visible in traffic. As it stands, I believe some evolution remains in the Velotilt’s future.


Some of the pictures in this post came from Adam Ruggiero’s article, “The Future of Human-Powered Transport is a Trike” in the March 18 issue of GearJunkie.

Monday, December 21, 2015

The Sinclair C5: The Worlds Worst Invention or the Template for an All-weather E-bike

I confess to being overly fascinated by the Pedicar. Maybe because it became available when I was discovering recumbents and realizing there were more utilitarian human-powered vehicles than the safety bicycle. It made a significant public relations splash because it was aimed at replacing the automobile for short trips. And at least one auto maker, Chrysler, took enough notice to lampoon the vehicle.

Now if available today, the Pedicar would make a worthy commuter vehicle. Placing the rider’s head at the same height as an auto driver, it had good visibility on the road, a feature I feel is a must if you are going to ride next to 35mph traffic in a 36” wide bike lane. Add wheel pants and a more aerodynamic body along with rotary pedals and a derailleur shifting system and you would have a faster, lighter and less expensive vehicle. (It probably would not climb hills as well.)
Drive the wheel (wheels) with an electric motor and you would have a very functional all-weather E-bike. (I use the word bike in this sense to refer to a pedal-powered vehicle that can have more than two wheels.)

But not so fast! In the US and other countries, and E-bike can have no more than three wheels. Add a fourth wheel and you become a moped, losing the legal status that lets you use bike lanes and bikeways.

And removing that fourth wheel causes problems with the vehicles ability to corner without overturning. The Pedicar, with a 36” track had a rollover resistance of in excess on one gee. If you went to a three-wheel layout and located the center of gravity (c.g.) such that each wheel was equally loaded, (1/3 of the wheelbase from the paired wheels,) the rollover resistance would be reduced to 66% that of a four-wheel vehicle.

Let’s look at some contemporary all-weather E-bikes.
The ELF is manufactured by Organic Transit and uses the tadpole wheel configuration (two front steered wheels and one rear driven wheel). I would consider it a laudable effort at producing an all-weather E-bike, (AWEB) but it has some issues that would prevent me from using in in my commute to work, namely it has a 48” width. 

The width is necessary to prevent rollovers when cornering. I will assume the track is 44” and there is 2/3rds of the weight on the front wheels. Even with the added width, compared to the Pedicar the ELF’s roll over resistance is only 80%.

In addition to the 48” with causing the ELF to hang out into the auto traffic lane on some bike lanes, it results in poor aerodynamics. This may not be an issue because E-bikes in the US are limited to 20mph, in some cases with pedal assist and in others without.

 The ELF uses a 750W electric motor to achieve its 20mph, which is the maximum power for the E-bike regulations the Organic Transit people are citing. With respect to the power a human can generate, 750W is a lot of power. World-class cyclists can sustain half that amount for about an hour. A racing cyclist requires about 155W to go 20mph (Bicycling Science 3ed-D.G.Wilson). This source states that a commuter-human-powered vehicle should require on the order of 75W to go 20mph. (The Organic Transit site states that the ELF has the same drag coefficient, Cd as a bike rider. We know that air drag is proportional to the product of Cd and cross-sectional area, so we can extrapolate that the ELF has about four times the frontal area of a bike rider.)

Vehicle weight is between 170 and 190lb
So as a pedal-powered-only vehicle, the ELF is not very efficient. As long as the battery is at full charge, this is probably not an issue. Run the battery down, however, and you will be working very hard to hit 15mph and with a topped off battery it will require a lot of effort to reach 25mph.

The prototype Velocar from VeloMetro has a layout similar to the ELF and, as a result will be subject to the same issues.
Now the 33% loss of rollover resistance when going from a quad to a trike is based on locating the c.g. so all the wheels are equally loaded. There is no requirement to do this. If the c.g. were located near the paired wheels the rollover resistance of a trike would be minimally compromised. With a typical tadpole layout, this would result in a lightly loaded drive wheel and potential tire slippage. Unless, of course you are driving and steering the front wheels (which, while technically interesting is more complex an approach than driven the rear wheel only).

The alternative trike layout is the delta, with one wheel forward and two wheels back. Biasing the weight toward the rear wheels aids traction but results in the front tire being lightly loaded and more prone to lateral slippage in slippery conditions. A delta trike, with an enclosing body, is inherently more streamlined than the tadpole trike, because the steered-front wheel is easily enclosed within the body.

The Sinclair C5 was a delta trike. A sectioned view is shown below.


The C5, the brainchild of the British industrialist Sir Clive Sinclair, had been called one of the world’s worst inventions. Even so, until the sale of the Nissan Lear, It held the record for number of electric vehicles sold at over 7500 units. The separation between genius and madness is often small.
The C5 used a 250W motor to reach 15mph, the legal limit for vehicles of the type in Great Britain. IMO the vehicle was quite attractive and relatively aerodynamic. Bicycling Science gives a racing cyclist’s power to travel at 15mph as approx. 80W, but unlike the ELF the C5’s inefficiencies were a result of rolling resistance and mechanical losses.

The front tire was 12” in diameter, the rear tires were 16” in diameter. There was only one pedal speed and no provision to adjust the distance between the seat and pedals. The crank length was about 130mm (normal is closer to 170mm) and the gear was about 44”.

Wheelbase was 51” (not too long for a recumbent bicycle), width was 29” and the seat height appeared to be on the order of 12 to 14”. Vehicle weight with a battery was about 99lb.
A custom geared motor drove the left-rear wheel through a belt drive and the pedals drove the right-rear wheel through a freewheel on the axle.

An attempt was made to allow the wheel-support outriggers on the rear chassis to flex and result in some suspension. (Photos of used vehicles show that the chassis developed cracks between the outriggers, suggesting that the chassis designer, Lotus Sports Cars, underestimated the stresses the flexing would cause.)

The video below is quite lengthy. If nothing else, skip to the end to the C5 on the road. It doesn’t seem to roll very smoothly. I don’t know if it is due to wheel size or lack of rear suspension. The original front wheel rims were plastic with caliper brakes acting on them. The brakes resulted in the rims melting so restorers often used pram wheels in the front. These could be as small as 8” in diameter.

From a human-powered assist perspective, the biggest design flaw was that the pedals seemed to be added after fact as opposed to being integrated into the design. Restorers often insert a 3-speed hub ahead of the rear axle to get the extra speeds. Making the seat to pedal distance adjustable is not something that could be easily retrofit-able because of one-piece chassis and one-piece lower body pan.

Besides the pedal-ergonomic issues, other concerns were that the rider sat too low for good visibility in traffic and inadequate weather protection. The former concern was addressed by adding a framework behind and above the rider that held reflectors. The latter was addressed by optional side curtains and a rain suit for the rider.

Three factors probably resulted in adequate rollover resistance. The seat height was relatively low compared to the track of the trike. The c.g. was biased strongly over the rear wheels and the maximum vehicle speed was only 15mph.

In 2010 Sinclair was working on resurrecting the concept of an all-weather E-bike in the form of the X1, which is a true bicycle, having only two wheels.

The latter part of the video below deals with the X1.

As the comparison illustration states, the X1 had better weather protection and a higher rider height Than the C5. The top of the vehicle is 55”, which in the photo next to an auto, looks like it should be even higher for good visibility. On the other hand, the roof of my Honda Civic is about that height. The width at 27” is 2” narrower than the C5. Both wheels are now 16” in diameter. Although it is not stated, the fact that, at 83”, the X1 is approx. 12” longer than the C5 may indicate that the seat to pedal distance may be adjustable by sliding the seat. There is no indication that the vehicle had more than one pedal speed. The electric motor drives the rear wheel and produces 190W and the overall vehicle weight with battery is a light 66lb. Like the C5, the design of the body is quite attractive. The vehicle seems to roll with less vibration than the C5. Partly because of the longer wheelbase, (about 73” vs the 51” for the C5) but mostly because both wheels appear to be suspended.
The fact that five years have passed and there is still no commercial X1 may indicate that the design had issues that prevented it from being commercially viable. I can speculate about what one major issue might have been.

Going from a tricycle to a bicycle, one loses the static stability necessary to get into a semi-enclosed vehicle. The body can interfere with lifting the support foot up when starting and putting a foot down when stopping. Even though there are cutouts in the lower body so the feet can touch the ground, the cutouts prevent the feet being paced laterally very far from the vehicle. So one must put their foot down before the vehicle has tipped very far. In the video above, we see the rider stopping but not restarting, the more difficult of the operations. See below.

So the Sinclair team apparently went from three to two wheels as a weight saving and lost the static stability in the process.

So as is my want, l would like to speculate about the changes required to the C5 to make is a viable all-weather E-bike for the US market where the top speed could be 20mph. 
I would use larger wheels, 16” in front for better rolling resistance and 20” in the rear for the ability to obtain higher gears.

I would make the seat to pedal distance adjustable by making the structural frame in two parts. The rear part would hold the rear wheels, the seat, batteries and any cargo. The front part would hold the front wheel and the pedals. The front part could be slid into the rear part to adjust seat to pedal distance. A similar approach is used in the Kettwiesel tricycle below.

I would have the seat height about 18”, placing the riders head about 48” off the ground wearing a helmet. I would keep the bottom bracket at about 15”, as low as I could get it without having the rider’s heals scrape the ground.

I would use a single chainring in front and attach a cassette to the rear axle. I would connect both rear wheels to that axle through freewheels, thus insuring a posi-traction-type drive.
I would use a hub-motor in the front wheel for the electric drive.

I would suspend the rear wheels and load the front wheel lightly enough so it didn’t need suspension, say 25% of the vehicle weight.

I would divide the body into two parts. The front cowl would completely enclose the front wheel and cover the rider up over the legs. This cowl would move with the crank and front wheel when adjusting the seat-to-bottom-bracket distance. The rear portion of the body would cover the riders head and the rest of the body. The rear portion would rotate upward about a pivot in the back of the vehicle to allow the rider to enter and exit.

I call the vehicle below the Avacar because it has a similar rider layout to the Avatar recumbent. The rider’s head height with helmet is 48”. The wheelbase is 53” and the vehicle track is 26”.

To this point I haven’t discussed turn-induced rollover issues with my redesigned C5. One advantage of making the Sinclair X1 a bicycle is that it has dynamic stability and can lean into the turns.
Let us assume that the original C5 had no rollover issues. I have made two changes that will reduce its rollover resistance, (ROR). I have raised the c.g. of the vehicle and I have increased its top speed. The 4” increase in seat height reduces the ROR by about 10%, assuming a c.g. height of about 17” for the original C5. The 5mph increase in speed reduces the ROR by about 44%. Together I have reduced the ROR by about 50%
The track of the old C5 was about 27”. To maintain the ROR of the original, the track would have to increase to 40.5” and the width to almost 43”. The width of the redesign is no better than the ELF or the Velocar.

That brings me to the question,” Should the AWEB be a leaning tricycle?” There is an all-weather E-bike on the market, the Drymer. It utilizes the tadpole layout.

Now I have discussed the design of leaning trikes ad nauseam. However, I have two new observations about the design of leaners when they have bodies that prevent riders from easily putting their feet on the ground to support the vehicle.

The first observation is related to the lean-lock mechanism that prevents leaning at very low speeds and when entering and exiting the vehicle. The default state for the lean-lock should be engaged and the rider should actively disengage it. For electric vehicles, engagement could occur when power is shut off or when vehicle speed falls below a predetermined low level. Engagement must be quick and positive, where the engagement force is not provided by the rider but inherent in the mechanism. The rider merely overcomes the engagement. This approach insures that there is always enough engagement force for complete prevention of leaning. Engagement mechanisms that are position dependent as opposed to force dependent are preferred.

The second observation is that, under no circumstances, should the vehicle be allowed to lean far enough to tip over statically. Now, an observant reader might point out that under these circumstances, the dynamic ROR is only slightly better than the static ROR. But that is not true if the lean-lock is engaged when the vehicle is completely leaned over. (The vehicle is leaned into the turn and locked in that position.) With the leaning locked in that position, the acceleration is acting against the full track of the vehicle instead of just one-half the track as in the static case.

Now I decided I wanted my design to be able to resist one gee acceleration when locked fully leaned over at the same time it was at its static tipping limit. It turns out the lean angle for this condition is 24.5deg. For a given c.g. height the vehicle track at the c.g. location is .829 times the c.g. height. The actual track is the c.g. track divided by the % of the weight on the rear wheels.

For example, if my CG height is 20.5”, the c.g. track is 17”. If 71% of the weight is on the rear wheels, the actual track would be 24”.

The Avacar drawing above is based on these numbers with 2” added to the track for a little safety margin. The vehicle width is 4” wider than this or 30”.