|Click HERE to go back to the Andy Douglas Streamliner page.
Andy Douglas' streamliner - the shell
In keeping with my overall philosophy of using the right tool for the job, my first task was to clearly define what I want the bike to do. I'm not terribly interested in racing or touring, but I do love to go fast on local day rides. I'm not going to be using the streamliner as a velomobile, so I don't need foul-weather capability or the ability to carry heavy loads.
Most of the club rides I go on require me to travel a couple of hours each way, so the shell has to fit in my car, a compact station wagon, without gutting the interior of the car. This dictated a compact size.
Weight is important since I do have to climb hills. A 50-lb. streamliner isn't worth the effort.
Relative ease of fabrication is important. I don't want to go the high-tech vacuum-bagged carbon-kevlar-Nomex honeycomb route necessary to get a really fast, light bike.
As it turns out, my needs coincide very closely with those of John Tetz. Therefore it made sense to follow his lead and build a foamshell streamliner (click HERE for a full article on how to build a foamshell).
In October 2001 John did me an enormous favor by giving me his first white foamshell:
This shell is somewhat the worse for wear, having seen a lot of use and one dramatic high-speed crash. But it's still intact. To our surprise and delight it fit the Pharobike almost perfectly, requiring only the cutting of a slot in the rear end to clear the edge of the back tire.
The fairing is unbelievably light (6.5 pounds... the bike WITH the fairing weighs only 33 pounds!!) works extremely well and validates the concept of the foamshell on the Pharobike. So now I have a target to aim at for my own efforts.
John has put a lot of work and thought into the aerodynamics of small streamliners, and his results speak for themselves. He's been a wonderful teacher and guide for me in my own pursuit of understanding about how it all works. So, the simplest way to proceed would be to just borrow the male mold he used for his orange foamshell:
Building a clone of John's fairing would work fine, but it wouldn't teach us anything. So I decided to go a slightly different route in terms of both mold construction (more on this later) and shape.
In my research, I came across this bike, built by a British fellow named Geoffrey Bird for the 2001 HPV championships in Brighton:
(To see more pictures of this bike, go to http://www.hpv-heaven.com/ and click on "A Message from the President") It's a hard-shell fairing with a more or less conventional SWB inside. It's notable for two important reasons: First, it's about the right size and shape for my needs, and second, it was designed using CAD, which is how I wanted to do the drawings for my own shell.
Geoff was kind enough to send me some CAD drawings derived from his original file. I asked for vertical sections every 4 inches (100mm), front to back. Here's what's in the file he gave me:
Geoff had already done what I was going to do anyway, basing the shape on simple airfoils and scaling it to fit around his body and the bike. So rather than reinvent the wheel I decided to use this design as the starting point for my shell.
I recorded the critical dimensions of the shell drawing at each vertical section (station), noting overall height and width at hip level and shoulder level. This gave me some numerical reference points for future use.
Next I had to determine whether or not I'd actually fit inside the shell. Using only measurements and paper/pencil this can become a complicated task because of the complex curved space into which the footbox fits, but with CAD it's easy. At first I thought of actually modeling my body in CAD, but that's overkill. What's actually necessary is to figure out the space the body occupies.
First I sat on the bike and, using a plumb bob, tape measure and square, took height and width measurements at several critical locations. These were:
- Most forward point
- Toe tip with crank at 45 degrees up
- Toe tip with crank vertical
- Heel bottom with crank vertical
(These four points describe the foot box.)
- Knuckles (with clearance above hands)
- Wide point of shoulders
- Wide point of hips.
This resulted in a wedge-shaped "clearance box" that defined the space I would occupy inside the shell. (Note that originally I was going to include the knees in the measurements, but because of the Pharobike's relatively low bottom bracket, knee height was not an issue.)
The clearance box was rendered in CAD and arbitrarily placed so that the front end of the foot box was placed at the station with enough width to accomodate it. Here's what that looked like:
With this accomplished, finding interferences was easy. All I had to do was rotate the model around and look for places where the clearance box stuck out beyond the surface of the shell. Sure enough, the lower outside corner of the foot box did:
It's obvious that I will fit inside the shell without difficulty. Fixing the clearance problem will be simple, and will probably involve simply mounting the shell a bit farther forward and couple of degrees nose-down, since there is much more room available above the toes than there is below the heels.
Reshaping the nose
For the most part, the fairing is a close enough fit that I need not scale it. However, Geoff Bird's original shell was built to fit a bike that has a much higher bottom bracket than the Pharobike. So the original nose shape has a big bulge above the feet that I don't need. In the CAD file I drew a plane parallel to the top front surface of the clearance box and about 2.25 inches above it:
This shows how much "extra" nose there is, even allowing for generous foot box clearances. (Note: the slope of this plane is about 17 degrees.)
I then drew a smooth lengthwise curve that would still provide adequate clearance (shown above running under the blue "arches" that show the upper surface of the nose), and rescaled the sections shown in blue to match it. Here's the result:
The first section at the tip of the nose now looks slightly oversized, but that won't matter much. The tip of the nose has to be shaped freehand anyway, so I'll just carve it so that it looks good. Whether or not the more steeply sloped nose will help or hinder me aerodynamically is an open question. I suspect that at the speeds I'm aiming for (cruising in the upper 20s) it won't make any noticeable difference. John Tetz's nose shape is different, with a flatter front deck and a vertical nose section. So I have a similar shell for comparison.
The fit exercise and nose reshaping shows the beauty of CAD: now that I have the model, everything else is trivial... I can run through many variations with very little time and effort.
Reshaping the headrest
The original shell was designed to be fully enclosed, so the bubble for the rider's head is much larger than it needs to be for my shell, which will be open. To get a good idea of the required size for the headrest, I did some additional rendering:
The blue plane represents the seat back. The red olive-shaped object represents my head, with helmet. The model does not show a windsheld. Note that in this image the headrest has already been reshaped as described below.
Tuft testing of other MARS aerodynamic aids (shells from John Tetz and Rich Sadler, and Dick Ludwig's tailbox) shows that the area behind the head is a problem. On each of these fairings, the area behind the head is about as wide as the head itself. We think that the air hitting the front of the rider's face (or the windshield in the case of an open-topped streamliner) is being deflected far enough to the side that it doesn't have a chance to reattach to the fairing surface, and remains fully detached all the way back.
On the supposition that this is the case, I'm going to try a slightly different approach. Make the front face of the headrest the usual size, but fatten it considerably a few inches aft of the rider's head before allowing it to taper. From above, the shape will be that of a fat teardrop. Hopefully, this will allow some of the detached flow coming off of the windscreen aft edge to reattach. This presents a larger frontal area, but the tradeoff could be lower-drag attached flow, resulting in a net gain.
It was easy to alter the shape with the CAD model. The front face of the original bubble behind my head was much too large, so I simply scaled it down to 80 percent of its original size. I then shortened the next section back by 10 percent, leaving at its maximum width. The remaining sections were left intact.
The final result looks like this:
It's now complete, and ready for plug construction.
Click HERE to go to the next page: Plug construction