Quattro(A Beano Beater??)
By Miles Kingsbury

CAD Picture Quattro Initial Design

I have had a very enjoyable season racing my Bubble & Squeak faired trike but the winter is nearly upon us so it’s time to start thinking about a 2010 machine. I could of course use the same trike for the 2010 season but I do love designing and building but more importantly I am desperate to beat Slash.

Firstly I would like to recommend two books that have helped me a great deal.

  1. The Leading Edge by Goro Tamai which is mainly about solar cars.
  2. The World’s Most Fuel Efficient Vehicle by various authors which is about designing and building a fuel cell car.

Before going into details of a new design let’s get the maths out of the way first.

Most of our HPV circuit races are won (by Slash) with an average speed of about 35mph (15.5m/s).

If I want to beat the little s###, I will have to go a little faster or saw up his Beano!

The power required to maintain 35mph is made up of two main elements.

Air Resistance Power = CdxA x½ρV³

Where Cd is the drag coefficient, A is the frontal area in m², ρ is air density (1.2kg/m³ at 20°C) and V is the speed in m/s.

Rolling Resistance Power =CrxVxN

Where Cr is the rolling resistance coefficient, V is the speed in m/s and N is the normal force in Newtons (kgx9.81) on the tyre.

There are other mechanical losses such as bearings and chain friction but these are small and I will ignore them in the following examples.
35mph Beano Power

The Cd for the Beano is probably about 0.12 taking into account gaps, wheels etc. The frontal area is about 0.35m².

Air Resistance Power = CdxA x½ρV³ gives 0.12x0.35x0.5x1.2x15.5³ =94W

The Cr is about 0.006 for good cycle tyres, Slash weighs about 65kg and the Beano about 17kg giving a total weight of 82kg.

Rolling Resistance Power =CrxVxN gives 0.006x15.5x82x9.81 =75W

Therefore Slash needs to put out about 169W to maintain 35mph in the Beano. So why is he so sweaty when he gets out after a race?
This is because it is his cruising power and doesn’t take into account accelerating out of corners and climbing hills.

35mph Quattro Power
The frontal area of Quattro is about 0.6m². Let us hope to get the drag coefficient down to the same value of the Beano at 0.12.

Air Resistance Power = CdxA x½ρV³ gives 0.12x0.6x0.5x1.2x15.5³ =160W

Let’s show Mike Burrows some respect and assume the smaller wheels have a slightly higher Cr of about 0.0065, I weigh about 73kg and guess the

Quattro will weigh about 20kg giving a total weight of 93kg.

Rolling Resistance Power =CrxVxN gives 0.0065x15.5x93x9.81 =92W

This makes a total of 252W which is an attainable figure but does give us an idea of what we are up against.
35mph “Upright” Racing Bike

Putting things in to perspective, a racing cyclist in a crouch position has a frontal area of about 0.3m² and a Cd of 0.9.

Rolling Resistance Power =CrxVxN gives 0.006x15.5x82x9.81 =75W

Air Resistance Power = CdxA x½ρV³ gives 0.9x0.3x0.5x1.2x15.5³ =603W

This makes a whopping 678W!  I could probably get a Knighthood if I was able to generate this sort of power output.

New Design
Bubble & Squeak is great fun to drive and is quite competitive, particularly on twisty circuits. I can throw it round corners with gay abandon and if I get it wrong I can just jam on the brakes or take to the grass. However it does have a few faults.

Ground Clearance
With only 25mm of ground clearance and a flat floor, the Trike bottoms out on all but the smoothest circuits, I have also worn away the heels of my expensive Sidi cycling shoes.

So when we went to the world championships in Holland this year, I didn’t take the trike as I could have only competed in three of the six events because of rough and uneven surfaces.

A ground clearance of 75mm to cope with speed bumps would be lovely.

Turning Circle
Curborough was a nightmare for me this year. After quite a bit of filing and sanding I was still unable to get round the hairpin without severe rubbing of the front tyres. The poor turning circle also makes general manoeuvring tiresome.

 A turning circle diameter of about 7m would allow me to turn in most roads.

B&S weighs about 27kg which is less than a Quest but still too much compared with most other recumbents. My sprint results this season have shown that I am carrying too much weight, not to mention the dreaded Hog Hill.

A total weight of less than 20kg should be achievable.

The combination of a large bubble screen and poor internal ventilation make B&S very hot, particularly on sunny days. I found that my performance dropped off dramatically after 25-30mins at most races this year. The screen also fogged at some of the events.

Direct ventilation onto my face and screen would help considerably.

Initial Specification

Length 2300mm
Width  730mm
Height 930mm
Weight 18kg TBC
Frontal Area 0.6sqm TBC
Track  650mm
Wheelbase 850mm
Ground Clearance 75mm
Seat Height 100mm
Chainring 80T
Cassette  9 Speed 11-34T Shimano Driving Front Wheel Axle
Cranks 150mm Custom
Wheels 16” 349 Alloy Rims on Cellite Disks Custom Hubs
Tyres Schwalbe Kojak 349 x 32
Brakes   Avid 160mm Mechanical Disks on Front Wheels
Steering Four Wheel - Rear Progressive
Chassis Tubular Steel
Bodywork  GRP

My idea of going for four wheels is mainly because I want to try something different rather than going for a Quest type layout. The Quest is a very well sorted all rounder and is hard to beat on practicality and performance.

When I started thinking about the design, I wasn’t planning to compete in pedal car races. Their regulations had a length restriction of 2030mm which was too short for my design; this has now been increased to 3000mm.

The main advantage I can see with a four wheeler is maintaining the cornering stability of B&S with the extra ground clearance I would like to achieve. After all, how many three wheel cars do you see on the roads?

Four wheel steering gives a better turning circle for the same amount of wheel movement and maybe less scrubbing round the corners, but whether it is worth the extra complexity will have to be seen.

I have decided to go for 16” wheels for a number of reasons. There is now a good selection of tyres for this size, they are smaller and easier to package in particular when steering and they are lighter. I estimate a total weight saving of over 1kg on 5 wheels (including spare).

All four wheels will be the same with a plain hub. The bearings and brakes will stay with the chassis, which should keep the weight down and make it possible to carry a spare.

The wheels are leaning in at 80 to give maximum track width with minimum frontal area. I have based this angle on the research that was done for the PACII fuel efficiency car. They suggest that angles up to 80 have a very small rolling resistance penalty.

The leaning wheels have made it difficult to get the front drive sorted. I have toyed with various ideas including a lay shaft with short chains and universal joints.

The screen is much smaller than B&S and will have a ventilation duct at its base. It will hopefully hinge like a motorcycle visor to give forward. It is a wrap around screen so it will be easy to fit solar film to keep the heat out.

The basic aerodynamic shape of Quattro is a lot more complicated than B&S or a faired bike. The wheel spats, proximity to the ground and head fairing make it very difficult to imagine what is going on.

Graham Sparey-Taylor has been a great help in getting me started with CFD. He analysed my original model and pointed me in the right direction regarding tweaking the shape.

As can be seen from the initial CFD images, there is a lot going on, particularly around the wheel spats. The result of all this churning is an estimated aerodynamic drag of 18N at 35mph (15.5m/s) which equates to 290w. If the rolling resistance power of about 92W is added, you can see that an old git like me is going to struggle against another old git like Slash in the Beano. More work is needed!.........


Quattro Part 2 (A Beano Beater??)

By Miles Kingsbury


Old shape

New Shape

As you can see the final shape is very different to the original.

I learnt a lot about airflow in the many steps between the two. I must have done about 30 design changes and test runs using the CFD software to get to the new shape. Each CFD test takes about five hours on a reasonably fast computer.

The combination of the spats and upward sloping shape was causing a lot of drag on the original shape. The slope was encouraging the air to move round from the sides to the underneath but the spats were in the way. This resulted in the air flowing outside the front spats and then inside the rear spats causing a lot of turbulence.

I did the initial CFD tests on shapes without wheels and got some very low drag figures but as soon as the wheels were added the drag went up and a lot of downforce was generated. As any sort of lift causes extra drag, I was determined to reduce it as much as possible.


I have always believed that the bottom quarter of the wheel can stick out of the fairing with little drag penalty. This is because it is effectively going a lot slower than the vehicle. The contact point is instantaneously stationary relative to the road. Unfortunately this phenomenon doesn’t show up in the CFD analysis. For this reason, I have got rid of the spats and on the CAD model and have simulated this effect by tapering the wheels from full width at axle height down to a point at ground level.

The flat bottom of the new shape produced parallel air flow underneath the vehicle which did not try and get between the wheels. By having the nose 1/2deg up, this was improved further. The flat bottom also makes the construction and sealing around the wheels easier. I can always add wheel spats or half fairings later if I want.

General Shape

I had one of those eureka moments while in the bath with Goro Tamai (The Leading Edge) reading about down force when close to the ground. Having the nose and tail at the same height makes a lot of sense when you think about it. If an air particle hits the nose, travels along the body and then ends up 200mm higher at the tail then another particle has to fill the gap left. This is bound to cause turbulence with a horizontal tail. This is a problem that doesn’t exist with a trike or bike with a vertical tail.

Wind tunnels and CFD tests are back to front in the way they test a vehicle as normally the air is still and the vehicle is passing through it. Sometimes I find it helps to think about the vehicle parting the air and then letting it go back to where it was before you came along. This way of thinking was demonstrated nicely last autumn while cycling through leaves behind someone on an upright bike. Despite the bulk of the rider being a long way from the ground, the leaves moved outwards about 300mm and then back in a swirling movement as he passed.

Having said this, I was unable to get rid of the vortices in the wake of the vehicle but the drag figures were still coming out very much better than the original shape.

We did all the CFD test runs at 15m/s (Slash beating speed of 35mph) and the drag force was down from 18N to less than 6N which equates to 90W. Adding the rolling resistance power of 92W calculated earlier gives a total of 182W. Slash watch out!!


Flow lines over the rear of Quattro showing the trailing vortices.


Pressure distribution over Quattro (red high, blue low) showing good positions for ventilation points on nose and base of screen.

Fairing Mould Making

The fairing production method I am using is in three stages, firstly make a full size model (plug), make a mould from this and then make the part from that mould.

Head Fairing

I thought I would start with the head fairing plug first as in my experience this bit always takes longer than expected because of its tight curvatures. I can also make it separately from the main mould. The top is a revolved aerofoil profile simply constructed from an elliptical nose and curved tail. I did try some more sophisticated shapes but they didn’t give significantly better drag figures and were less practical for my purposes.


MDF Disk Dimensions

MDF Sawn and Turned Disks

Turning the ‘Kebab’

Because of the problems I have had in the past, I decided to try turning a full revolve in the lathe. I did this by splitting the shape into 25mm thick MDF discs and gluing them together on a 38mm steel tube. The result looked quite similar to one of my favourite local delicacies.

Sanding the head fairing revolve on the lathe was definitely quicker than doing it by hand and it produced a smoother finish. I then added a turned nylon tail and a mock screen and blended it in.


Filling Surface


Smoothed Filler.

Mounted Revolve

Gluing 1.5mm Plywood Screen


Screen Fitted and Sanded

Main Body

The main body plug was made by dividing the CAD shape into 25mm, 50mm and 100mm thick vertical slices and getting these sections plotted full size. I only used the thinner slices for the nose (2x25 and 3x50mm) where the curvature was tighter. The paper plots were then roughly cut out and stuck to Xtratherm insulation boards (I have used Celotex in the past) with double sided tape. I then roughly jig sawed out the shapes and then accurately finished them on the band saw.

I decided to use this type of insulation board as it is a polyurethane based material which meant I could use polyester resin and filler straight onto the surface without it melting.


Plotted Paper Section

All Sections

Plots Stuck to Insulation Board

The paper plots also had a vertical centre line and two 19mm holes marked out on them for alignment purposes. The holes were drilled through and a small saw cut was made at the top and bottom on the centre line.

I removed the silver foil backing from the cut pieces and assembled them onto two 19mm diameter steel tubes. I also fitted a straight edge into the lower centre line saw cut. I assembled the sections alternately facing forwards and backwards as I was not sure my CAD shape was perfectly symmetrical. The bench for the assembly is a large fire door which are reasonably light to handle and very flat.


Cut Out Sections


Trial Assembly


I decided to paint the whole thing with some left over pink vinyl emulsion (other colours will work) which acts well as an adhesive and also makes the intersection lines visible for rubbing down.


Painted Joints

Sanding Top

Sanding Bottom

It only took a few hours to get the rough shape finished but this is where I made my first mistake. I decided to use only car body filler to get a hard surface onto the plug.


Part Filled

Top Filled

Plywood Bottom

It is very difficult to get a uniform thickness of filler onto a surface and so I ended up introducing more waviness than I started with. The last plug I worked on was for a school Greenpower electric vehicle, which we covered in two layers of glass fibre. This was really hard to rub down but was quicker in the end to get to a descent finish.

My second mistake was to cover the flat and single curvature sections of the bottom with 1.5mm plywood. I thought this would save time but it didn’t stay flat and caused more work where it distorted.


Head Fairing in Position

Head Fairing Fillet

Finished Plug

I then sanded an angled face on the top of the plug and fitted the head fairing. It was fixed and filleted using car body filler. The plug was sanded to a smooth finish using 600 grit wet and dry paper but not polished. My moulding man prefers some texture when he waxes the surface.

Mould Split Faces

The mould needs to be split for such a shape in order to extract the finished part.

I decided to split the mould into quarters, although a single vertical split would suffice for extraction purposes. The main reason for this top / bottom split is to allow me to add some height in future if necessary. I have been known to make my vehicles a little tight in the past!


Marking Split Line

6mm MDF Board

Split Support

I found the split line at the widest point by rubbing a crayon along the edge of a large square and then sliding this along the length of the plug. A support strip was then cut to this shape and the split surface glued to it.


Gluing Split Surface onto Support

Split Surface

Parcel Tape as Release Layer

The split surfaces were then stiffened on one side and fixed into position with clamps to the fire door bench. Wood screws were used to on the ends and into the plug where necessary.

I used parcel tape along the centre lines to protect the plug and to act as a release layer while I filled the gaps with the same car body filler.


MDF Splits to Produce

Hole in Mould Sprayed Black

Finished Moulds

The bottom of Quattro has a large rectangular hole that will probably be filled with a flat sheet of Correx that will be easily removable for wheel changes and general maintenance. This large hole in the mould is also useful as it will allow the mouldings to be easily joined from the inside.

I tried spraying the plug gloss black to hide the patchiness of the filler which I hoped would help show up any waviness in the shape. This did not work and I ended up with a black patchy shape so I decided to leave it at that and work on the inside of the moulds when I got them back.


My brain went AWOL in February so I have been struggling with the internal details of Quattro but it is coming together now.

Things Still To Do

  • Internal Structural Mouldings Not Started
  • Suspension and Steering Not Started
  • Cranks and Chainring Not Started
  • Screen and Hatch Not Started
  • Running Gear (Wheels, Hubs & FWD) Being Manufactured

There does seem a lot of work to do still but I am hoping to have Quattro running well before the end of this season. Until then, it is up to someone else to thrash the Slash!!