02/23/04 Got more pictures of the plug progress. It's almost done, should be ready Wednesday. Have to plan how/when I'm going to pick it up. Will most likely rent the very same truck that brought the Westie kit home from the trucking terminal. Of course the drive is 250 miles each way but should still be doable in a day, if I get an early start.
I'm sure glad I didn't try to build this by hand!
Tried to go pick up the plug. Plan A was to use my trailer and borrow the same Ford Explorer that made my Elise trip to Thunderhill so 'interesting'. The Ford really doesn't like me though. When I went to get it the battery was dead despite having been recently replaced. After a lot of effort to jump start it and let it recharge, we hooked up the trailer (in the dark) and I set off on my way. Only made it halfway down my friend's driveway when the trailer popped off the hitch and ripped out the wiring. Some two hours later after trying to figure out how to put the whole thing back together we turned off the Explorer's engine, quickly to discover it won't start again. I gave up and decided to go with Plan B, using a rental truck. Next day I went to rent a truck. First place where I reserved hadn't gotten it back from the previous renters. So I went to another one but they wanted the truck back the same day which wasn't going to happen at that point. So I gave up again and on the way home a rock broke the Mini's windshield. Fate clearly did not want me on the road at this particular time so I took the hint and decided to get the plug shipped directly to Composites Unlimited. They won't have room for it until March 8 so that is when it's scheduled to arrive. I'll go meet it there.
03/03/04 In my quest to figure out the underbody design I've obtained another CFD run, this time using a different program (FloWorks instead of CFDRC). Many thanks to Chuck Boulware for all the work. The new results are VERY different from the first set, unfortunately in a disappointing direction. The original results sounded a bit too good to be true, these are a good deal worse than what I'd hope for. The numbers came in at 433 lbs drag and 62 lbs LIFT at 180 mph, vs 123 lbs of drag and 638 lbs DOWNFORCE the first time around. Well, if nothing else this demonstrates that CFD is not a simple endeavor. In fact it seems to be a bit of an art and very dependent on how the simulation is set up and how the data is processed and interpreted. According to MulsannesCorner database, a 1969 Porsche 908LH generated 124 lbs downforce with 423 lbs of drag at 180. That is a considerably larger car than mine, with no underbody diffusers (that I know of) and a wing hanging out in the breeze for drag. I would hope to be able to do a bit better than that. My goal is under 300 lbs drag and over 300 lbs downforce at that speed. This would allow me to actually reach 180 mph with a stock normally aspirated motor (if drag is 433 lbs it would require 250 hp to go that fast). So first order of business is to understand what the CFD results really mean and what the constraints are, then to make whatever tweaks necessary to the underbody to come as close to the goal as possible. One issue is that in the FloWorks simulation the ground is not moving but 'frictionless', which is similar but not quite the same. This may or may not account for some of the differences. In the meantime I have a few pretty pictures :)
The last picture is particularly puzzling, with air actually moving forward, i.e. FASTER than the car on the side and in the back. Considering that the velocity is relative to the car, if taken relative to the ground it would translate to a roughly 200 mph forward breeze. While I can almost understand that at the back if the air tumbles to fill the low pressure area behind the car, at the side it doesn't make much sense to me. Even so I'd think it would take a heck of a tumble to accelerate air that is moving forward at about 20 mph relative to ground on top of rear deck to 200 mph only two feet later. That would certainly account for a lot of drag though. Compare this with the CFDRC version:
Hmmm..... This may explain why top racing teams still spend hundreds of hours in the wind tunnel despite having almost unlimited CFD capabilities. At any rate, it seems that the aerodynamics are not too horrible either way and hopefully the reality is somewhere between the two sets of numbers. Time will tell.
And speaking of time, just learned today that the delivery of the plug will have to wait until March 15 because Composites Unlimited are not ready for it yet. Bummer.
03/04/04Got some pictures of the finished plug (click on pics for larger version).
Also I've been studying Mike Arnold's tape on making molds. Certainly clears up some mysteries and gives a good idea of what's involved. Highly recommended.
03/06/04 Got some revised results from the CFD run. I changed the underbody slightly and added some basic internal airflow. Also, the mesh was refined for the simulation. As a result drag went up from 433 lbs to 495 lbs but the previous 62 lbs of lift was replaced with 200+ lbs of downforce. I have a hard time accepting the drag number since that would mean a CD of 0.50 which would be very high for a car with no wings - it's similar to a pickup truck with tailgate up and much worse than most production cars and even some minivans, which are in the 0.30-0.45 range. For example Chrysler PT Cruiser has a CD of 0.38 and Jeep Cherokee 0.44. So perhaps the absolute numbers need to be taken with some scepticism, but there are a few interesting things that can be learned from the plots. First, it looks like I tapered the cockpit too much in the back and the air is slowing down more than it should. I can offset that to some degree by channeling more of the radiator, front diffuser and side intake air behind the car. You can see that the severe tumble behind the cockpit is not present on the other side and if that stagnant area is filled in I think the tumble would go away (assuming that it's really there and not a figment of the computer's imagination). At any rate, filling the wake behind the car can only help.
Another thing that is apparent is that the rearmost diffuser is too sharply inclined, causing flow separation about halfway along. I already have a solution to this which is in essence making this final diffuser stage into a dual-element profile. I think it should work well. Also, the front diffuser air that is shown above exiting to the side will be routed out the back instead, helping fill the wake and reduce turbulence. Since the area behind the car is low pressure it will help 'pull' the air through the diffuser and increase efficiency of the underbody as a whole.
The new pressure distribution plots below illustrate the effectiveness of the slightly raised lip on the front splitter (witness the two low pressure areas under the nose). The plots also show why putting louvers on top of front fenders increases downforce - not because of high pressure below the fender but rather because of LOW pressure above it which the louvers help fill in. The louvers on my car were omitted in the CFD model for simplicity.
Thanks to all the CFD work I now have a reasonable idea of what I want to do with the floor, so I can proceed to finish it. A few questions do remain and hopefully I'll be able to get a few more CFD runs to answer some of them. Of course much testing will have to take place when the car is finally driveable and the new floor configuration will facilitate making changes with minimal pain and effort.
03/10/04 Making some progress, although the detailed nature of the work means a lot of effort and little visible result. I'm still struggling with engine mounting solution. My latest idea is to use four BMW control arm bushings and hang the engine from the main aft mounts. The bushing is relatively cheap at $32 from a dealer and has varied stiffness depending on orientation. This would let it be relatively stiff in counteracting the considerable torque of the output shaft (up to 420 lbs-ft) and relatively soft in up/down motion of the 180 lb engine/gearbox assembly. Each bushing weighs just six ounces.
I have a few concerns about lateral motion of the engine but could always use a Panhard-rod type lateral link if it proves to be a problem. This solution is not final yet but I'm tempted to just go with it and debug later if need be. Other items that have seen progress are suspension arms and uprights.
Trying to design everything to carry the loads, be light and manufacturable at the same time is a challenge. I'm pretty happy with the current iteration, with uprights projected to weigh only 1.3 lbs each. They are identical all the way around, as are upper and lower suspension arms. Saves on tooling, inventory and spares. Still need a few dimension and tolerance tweaks but it's close. The Volk wheels I was going to use turned out to be unobtainable (unless I want to buy 20) so I am looking at alternatives there too.
Also settled on an underbody design, using the results from all the CFD work. I've decided to route the diffuser and radiator air inside the bodywork and use it to fill the wake and improve the efficiency of rear diffuser (hopefully). The rear diffuser itself is now dual-element to prevent flow separation and better control the wake. The front diffuser is shallower and not as wide.
The fact that the project is taking this long (it will be two years soon) is actually proving to be an advantage, as I have plenty of time to examine each solution and look at a number of alternatives. This being my first car I don't have the experience to leverage or proven designs to re-use - but I certainly will for my next design :)
03/16/04 Finally got to see the plug in person. Cool. It doesn't look all that small, in part because of rather significant width (the car is only 1" narrower than the current Corvette), and in part because the plug sits high on a base which in turn sits on a trailer at the moment. Still, it was exciting to see it in the flesh. I imagine this is what it feels like for parents-to-be to see the first ultrasound of a new baby. Naturally I took a few pics.
Overall the finish on the plug is pretty good but it will take quite a bit of sanding and several coats of primer before it's ready for making molds. The guesstimate is 50-100 hours of labor. I've arranged with Composites Unlimited to do most of that work myself at their facility, both to save money and to save time as their bodywork guru is pretty busy these days. They'll do all the really critical stuff and will provide advice. I'll supply the elbow grease. The goal is to have molds done by my birthday in early June and first article bodywork by end of June.
While there I also snapped a few pics of the Viper Jets under construction. Yes, this is still on the list too, in due time ;) Just as well to let them get the bugs sorted out in the meantime.
03/18/04 Spent six hours today sanding the plug. Janicki did a great job and it seems to be very accurate. There are however some minor issues that are due to the limitations of SolidWorks. I've had to use smaller fillet radii in places just because it would break down with larger ones. Also the intersection of the compound curves of the fender and the wheel openings is close but not perfect. Perhaps most people wouldn't notice but I do :) Fortunately the putty material sands really well and it was actually quite fun in a repetitive task kind of way. I'm quickly developing a feel for where the low and high spots are (luckily I'm fairly good with my hands). It's a very tactile kind of work. I got a quick lesson in using Bondo and picked it up pretty well. Probably another day getting the shape finalized, then it's a lot of primer and finish sanding with progressively finer paper. Right now I'm still working in 36 and 80 grit but will eventually go down to 400. So CNC does save time but it's not an end-all solution. A good deal of hand labor is still required.