Tony Levand builds a streamliner HPV
LevaLiner 1.0
By Tony Levand

Tony Levand has been building and racing recumbent bikes for over 10 years. After racing his practical LWB Coroplast bodied streamliner for the past several years, and commuting on it almost daily, Tony has decided to build a real racing streamliner. This page documents his journey.

Streamliner Parameters
To determine the wheelbase configuration of the streamliner the following parameter study was made.

Long wheel base
+Most of my latest bikes have been long wheel base, so I am somewhat preferential.
- Makes the bike long:
-Doesn't fit in car.
-Increased side wind loading.
-Requires more steering input.
-Front tire is lightly loaded, reduced handling.
-Increased turning radius.
+Uses standard fork and front wheel.
+Simplifies frame.
+Straight, stiff frame between seat and cranks.
+Chain runs along frame.
+Front wheel can be isolated from cockpit, reducing water and debris spray.
-Small front wheel more affected by road bumps.
+Long wheel base rides smoother.
+ Better control on gravel and stability at high speed.
+Absorbs frontal impact better.
+/- 30 degree steering input and 82 inch wheel bas makes a 183 inch turning radius.
This requires the fairing to be 11 inches wide at the front of the wheel.

Short wheel base
+Bike is significantly shorter, wheel base if half of long bike.
-More than half weight is on the small wheel.
-More side load is transfer to front wheel.
-Wheel is between legs.
-Needs narrowed hub, and fork.
+Half steering input needed.
-Frame is more complex.
-Small wheel affected by road bumps.
+Best aerodynamics due to delayed wheel turbulence.
+/- 20 degree steering input makes 122 inch turning radius.

Medium wheel base
+Medium wheel bas has ideal weight , 50/50 front rear.
+Uses 700c front wheel.
-Front wheel is disk to prevent spoke damage.
-Requires narrow hub and fork to reduce heel strike.
-Needs special split crank.
-Increased complexity.
-Additional chain losses.
+Could be front wheel drive with cassette on jackshaft.
+Split crank could have very low Q.
+ Slightly toed in to follow natural leg extension.
+ Off center drive sprockets could reduce dead spots like the rotor cranks motion.
+Split cranks BB bearings could be incorporated into tub.
+/- 30 degree steering input and 60 inch wheel base makes 134 inch turning radius.
Fairing length is same as short wheel base, since the front of the wheel is about the same as the extended foot. Tighter turns are make by extending the outside leg to 10:30 position for wheel clearance around knee.

2/08/2008
Here are streamliner sketches of each wheelbase configuration outlined above.

All in all I am leaning toward the simplest construction and rear wheel drive.

Aerodynamics
I have chosen a NACA 66-021 wing profile to model in 3d, which I'll discuss later. I want to see if this is the best profile for an all around HPV. The section is a low drag laminar flow type, which are sensitive to surface roughness and have low stall angles, which means the air flow becomes detached in low side winds.

I was reading and article in BHPV, I forget the issue, about wind tunnel testing and the streamliner had twice the drag at an angle of attack of only 10 degrees. It had a traditional section with maximum width at about 30% cord. This is a 4.4 mph side wind when traveling at 25 mph.

 I have noticed from my long commutes in the Carp that the best days were with low or no wind. Even a tail wind wasn't as good because the general direction home was northwest was never directly with the wind. I want to investigate the NACA 1 (16) series used for propellers and wind mills. They have low lift and low drag. I'll try to find some specific lift curves for various sections.

This is a model of a 100 inch long by 20 inch wide shell.
I am using K3d, an open source 3d modeling and animation program that downloaded and installed itself onto my Ubuntu Linux system. The program is a beta version with some quirks. I haven't figured out how to accurately reposition objects except by dragging on the screen. What is cool is that you can morph objects, squeeze , stretch bulge. It has a library of standard geometrical shapes. It also has an export feature that writes a file with all the point coordinates of the mesh. it's really not a CAD program per se, but I think the point data can be converted for NC.

I used an OpenOffice spreadsheet and pasted in the profile, thickness vs cord in percent of a NACA 66-21 wing section that I got off the web, UIUC I think. I then wrote a basic macro that takes the points and generates an elliptical solid by multiplying the x.y coordinates by the thickness.

To make the tail flat, I used a high order polynomial function of the cord position to expand the y dimension. I'll include a list of the source file once I get the interpolation routine working. As the profile has only 26 cord points, the thickness and slope are given, so I need to interpolate a curve to generate the in between points. There are ways to do this using b-splines but I haven't used them before. I think the code will paste directly into MS excel visual basic macro or visual studio and work.

I also found on the web a profile generator for NACA 4 series sections and gives a list of points with user parameters.
 

2/14/2008
I tried curve fitting a 10th order polynomial to the NACA profile , but it didn't look good, so I found a program that does a cubic spline interpolation. A cubic spline fits to points with a piecewise curve that matches second derivatives at the junctions. I used the Euler program, There was a bit of a learning curve but I got it to write a file of 106 points along the NACA profile. Unfortunately the file input in the basic macro only reads the data file as one long string and not as separate variables, so I have to parse it. The cubic spline looked so good I could not see a difference when I plotted the original 26 points and the 106 points. I am wondering now if using NERBS, a 2d surface fitting algorithm would be better. I don't know what NC machine uses.

I have decided to go with the long wheel base. This way I don't have to build a special wheel and forks. The wheel base will be between 82 and 84 inches. I have a BMX 451 front wheel. It has a Alex DA22 rim, which is an aero profile. I'll go with a 1 inch or 1 1/8 inch tire. The seat height is 6.5 to 7 inches, reclined 44 degrees and the BB height is 14 to 16 inches with a narrow Q and triple rings. Crank length is 150mm to 155 mm. The rear wheel is 700C with no offset and 8 or 9 speeds.

Steering will be remote control rod with +/- 30 to 45 degree movement, depending how much room is in the fairing. The steering inclination will be steep with 3 to 4 inches trail to reduce the swing of the front of the wheel. The handle bars may be U type, with hands to the side of the knee, I like this position on the Winter bike. I can pull on the bars and sit up when in snow or rough conditions or at intersections for better balance.

I want a suspension because I found while riding the Carp that what limits speed on bumpy roads are the bumps. I need to design the rear suspension and seat. On my past bikes I had a monotube frame that stuck out 3 inches below the seat. I want to incorporate the frame into the seat for a smooth bottom without making a tub bike. The bike may end up looking like a Moby, but with a more rounded and symmetrical nose and more rounded and less abrupt tail. I want to use glass with a foam core for the fairing. I would like to vacuum bag it in a female mold.
 

2/20/08
I was able to get the cubic splined interpolated 118 cord points and 64 circumferential points written. Now the hard part, how to make the mold. I decided for time sake to use the race bike even though the mono tube frame sticks down 3 inches below the seat. I can build a new frame later.

 
Other views:

Two CNC shops were asking for a STEP or IGES file, they couldn't read the text file with point coordinates. There is a 4 axis wire cutter for sale on E-bay for $3k. 4 axis is each end if the wire can translate in x and y separately.
 

More soon...  

Back to the recumbent projects