11/09/03 Did some more rendering on the design. It's looking almost tangible... Which is frustrating in a way because I'm still many months and tens of thousands of dollars away from seeing it in the flesh. But that's what I signed up for, right?... :)
Of course I had to make a helmet for the driver too (should have done it sooner actually). Looks like I'll need to raise the rollbar an inch or so...
11/19/03 Got the initial results from the CFD work being done on my car (many thanks to Tudor Miron!). We ran it at 3" streetable ride height and zero rake. Even so, at 170 mph the downforce comes out to something around 570 lbs. So my eyeball aerodynamics are not too bad, although some of the features are not doing quite what I expected them to. For comparison, according to Mulsanne's Corner aero database the new Posche Carrera GT produces 493 lbs downforce at 180 mph and Ferrari F360 produces 424 lbs at the same speed. Older street designs generate lift at speed, the 2000 Porsche 911 generating 600 lbs of it at 150 mph. This kind of lift would nearly send my 700 lb car flying! However in my application I might actually want to reduce downforce a bit in order to be able to use soft springs. The soft springs, besides providing a more livable ride, help generate mechanical grip and aid both traction and cornering at slower speeds. Since on most tracks I've run the speeds are in the 50-130 mph range, optimizing low-speed performance would have a great impact on lap times. It's all a huge learning process.
The pressure and velocity distributions help visualize what's happening with airflow:
Some 2-dimensional crossections provide further detail. On the first illustration you can see the effect of my 'area rule' design, where the cockpit protrusion pushes air sideways just when the front fender drops down. This keeps the flow attached and pressure positive in that area - otherwise it would separate just as it does on the back fenders. Another item of note is the spiral flow off the rear fender. This is fairly normal and typically one would expect to see the same off the front fender too. Except my diffuser side exhaust tunnels fill the low pressure areas on the sides of the car and and keep the air traveling horizontally instead of downward (downward momentum on air produces lift). At least this is how I'm interpreting what I'm seeing. I'd be interested in hearing comments from any aero experts out there...
Of course I'll need to stare at all this for a while to better understand what is happening and what works and what doesn't. Additional results will help shed more light on the design, too. Of particular interest are drag figures since my car will, at least initially, 'only' have 180 hp.
11/21/03 Got the drag numbers from CFD and they are a lot better than I had hoped for. It seems at 170 mph there is about 110 lbs of aero drag. This means I only need a total of around 60 hp to go that fast and with the 'Busa engine the car will have a theoretical top speed of around 235 mph. This is about 30 mph better than the bike the engine came from :) The drag coefficient works out to roughly 0.12 and downforce/drag ratio to around 5 which is on par with the best LMP designs such as the Lister Storm. Now of course this is a bit optimistic because there were no radiators in the airflow which would affect things somewhat. Also rollbar and mirrors were not included and would certainly add a good deal of drag. Still, for just eyeballing it I'm quite happy - even if real-life performance is only half as good it will still work pretty well. I did some comparisons for a reality check: the most efficient single-seat airplane, the AR-5, went over 210 mph on 65 hp. The dp1 would need approximately double the power to go that fast so OK it's not THAT efficient ;) A sit-up 125cc shifter kart can go 125 mph on 40 hp and would need around 100 hp to go 170. Considering that dp1 has much better aerodynamics and frontal area which is only a little bit bigger this is actually believable.
Extrapolating the numbers to 180 mph it comes out to 123 lbs drag and 638 lbs downforce and L/D ratio of 5.2 - this can be directly compared to the data in Mulsanne's Corner aero database. A further comparison - I believe the Stohr car, being larger in size and with a smaller 1.0L motor (and arguably worse aero at least due to having a wing) has done 1:12 laps at the two-mile Portland International Raceway. So with bigger motor and less drag one might conclude that dp1 has 1:10 potential in the right hands (probably not mine). This is put in perspective by the fact that the ALMS Audi R8 qualified on the pole at 1:05 and CART qualifying times are around 0:59. And those cars can't be driven to/from the track, either, except in Hollywood ;) Well, it's fun to speculate anyway.
In other news I'm dealing with sticker shock on initial quotes for body plug and moulds, about double what I expected (think brand-new Porsche 911 territory). Trying to look for ways to make this cheaper. I will probably end up doing quite a bit of hand labor myself, using the vendor's facilities. Sigh... it's never easy :(
11/23/03 I've been thinking a lot about the underbody aero based on what I've learned from the CFD results. One thing I realized is that it should be advantageous to start tunnels wide and then narrow them down as they get deeper. This would maintain air velocity and prevent unwanted pressure rise below the car (it seems Ferrari is doing a similar thing with their 360 and Enzo). The other item I came up with is what I call a 'dual-phase' (yes, 'dp' ;) rear diffuser. The idea being to start first phase well ahead of the rear wheels and as wide as possible, narrow it down just ahead of rear suspension and then put the air on top of second phase. The concept is vaguely similar to a dual-element wing and is illustrated below:
Chances are that in a car as wide and short as the dp1 a couple of strakes will be required in each tunnel - something to tinker with in development. I probably won't get a chance to CFD the new configuration, but might be able to get results for a similar configuration later on.
12/08/03 Now that I have the cash from the sale of the M Coupe, it's just about time to place the order for the bodywork plug. With that in mind I'm going over some final details and tweaks. It's a pretty big commitment so I don't want to have to redo it. Transparency is a great tool in SolidWorks to check for component fit and make sure there's no interference.
Looks like this ought to be pretty close. The packaging is tight but that's what is required to achieve minimum weight and best aerodynamics. Plus I enjoy a challenge ;)
12/09/03 The big step. Ordered the work on the body plug. Of course there were several last-minute tweaks and second-guesses, mostly in the cockpit and rear diffuser area. Basically I could just tweak this forever so the time has come to put the money where the mouse is and go for it...
The company doing the work is Janicki Industries about 70 miles north of Seattle. Their 5-axis CNC machine is 88 feet long! Much of their business is making plugs for fiberglass boat moulds. When the dp1 plug is done I'll go pick it up with a U-Haul and hopefully get a tour of the facility then. It's gotta be pretty cool. Maybe I'll get lucky and get to see the machine in action, too...
Of course I couldn't resist some last-minute tweaks and ended up sending no fewer than six 'final' versions to Janicki. I guess the deadline (Monday) really forced me to take a very critical look at everything and tackle any and all issues I felt the least bit iffy about. Some may note a few differences between the full-car renderings above and the plug renderings that follow. Quite happy with it now though.
12/16/03 While waiting for machine time to be scheduled on body plug production, I used the chance to tweak things some more. I guess there's no such thing as fully satisfied :) I don't know but certainly hope that THIS version is good enough to invest the big money in. Below is a comparison of a week-ago 'best effort' and the current version. The changes are subtle but noticeable (to me anyway) and I think result in a car that looks sleeker and more coherent visually.
Getting lines and shapes to 'flow' from all angles is tough. Should be fun to see it in the flesh.
I've also been doing some work on someone else's two-seater project. Below is a comparison of the two side-by-side. The two-seat car is about the size of a Radical SR3... Yes dp1 is tiny :)
12/18/03 Seeing how the exterior is now final I came up with an official dp1 desktop 'wallpaper'.
It's been quite an endeavour to even get this far, having started the project some twenty months ago. To get a better view of the progress made I've put together a single page with pictures of the design's evolution - it can be viewed here (warning - the page is a bit lengthy).
12/23/03 Now that the upper bodywork is finalized I've turned my attention to the underbody. After staring at what I had before and the various plots of the CFD results, a couple new ideas emerged. This led to a pretty radical redesign of the underfloor...
The design is quite different from traditional tunnels and diffusers you'd find on a race car. A big reason for this is that I'm trying to come up with something that generates moderate but consistent downforce at 3" ride height and is not pitch sensitive. I probably won't get a chance to CFD this configuration so this is all eyeballing it again. The floor is now one piece. An approach similar to the one used on the upper body is employed to keep cross-sectional area variations down. The main focus of the underfloor now is to route as much air as possible on top of the second phase of the diffuser. This should shift the center of pressure rearward and will hopefully further cut drag. Generally, CFD shows that it's pretty difficult to get air to follow the underside of an upward-sweeping tunnel but it's quite easy to 'shovel' it upwards with the top of an angled surface. Compare the relative effectiveness of the earlier version of front diffuser where both top and bottom were used and the former rear diffuser where only the bottom surface was exposed to airflow. The 'dp' rear diffuser is intended to put the top surface to work in the back.
The front downforce comes primarily from the top of the front bodywork and rear downforce comes primarily from the diffuser second phase. The cooling air for the front brake now exhausts under the floor and is integrated into the overall flow.
There are other benefits as well - the floor is now much stronger and can be used as a structural member to increase chassis stiffness. Only one piece needs to be manufactured instead of five. I also like the aesthetics of the slender look created by having higher floor at the sides (and it's less likely to get damaged on track curbs). This is a very short and wide car so there are quite a few styling tricks here to make it look slimmer and the floor change helps in that regard. How well all this is going to work is anybody's guess but it certainly does look neat. Well, to me at least :)
12/25/03 While waiting for a quote on fabricating the floor tooling I've been tinkering with it some more. A quick check for clearance revealed a need for some adjustment...
With the inside wheel turned 25 degrees the outside one is turned 19 degrees (lots of Ackerman - a result of very short wheelbase and very wide track). This gives me a 20-foot turning radius which is decent. Also checked for radiator fit. I can squeeze in two 15"x9" units. In comparison the Westie uses a single 18"x12" radiator which is adequate but marginal. My setup ends up with 25% more area so that should help. I might also consider a triple-row design.
After the bodywork is finally taken care of I'll be moving on to the chassis. May have to once again rethink how I'm building the frame and the suspension. It's amazing how much work still remains. The design process is iterative. First came the overall specs, then rough sketches, then initial chassis design, then initial bodywork. After that it's a matter of refining a subsystem then propagating all the changes to the rest of the car. Occasionally this requires totally redoing a few things. At least in CAD it's relatively easy to do but even so I've already had to scrap quite a few components.
On the subject of component suitability, a question came up about the steering rack I've selected. Apparently some FSAE teams have had failures with these racks. At the time of purchase I did some research and determined that Stiletto makes two models, one steel and one aluminum. I made sure to order the steel one, and it seems that the failures had been in aluminum racks. Having learned on the earlier ATV balljoint not to trust what's under a rubber boot I decided to take a look inside the rack. Thankfully I do indeed have the steel version and it's actually fairly stout.
There also seems to be a brass insert which acts as a bearing for the rack itself. A pretty nicely done unit, really. Very light and very low friction. It is also very compact being only 11.25" between end bearing centers which is important for my design. The ends may appear to not be well supported but they are really no worse than any standard rack, it's necessary to provide the travel. The only potential issue is greater load on the brass sleeve and so some possibly faster wear in that area. I certainly don't see anything that would cause me concern though.
01/03/04 The tires came in. I'm initially going with Hoosier A compound in 225/45-13 size, in their 'older' glass-belted model (it's 1.5 lbs per tire lighter than the new steel-belted kind). These things are little steam rollers! A fact especially put in perspective by stacking them next to the optional, wide Elise rubber (I even had to update CAD data, but not enough to change bodywork).
So let's see, the car is going to weigh some 1,000 lbs less than the Lotus (700 vs 1,700) and it'll have more rubber on the road :) Of course it will also have nearly identical power and a lot more downforce... The sidewalls on the Hoosiers are extremely stiff - much more so than the Yoko 048s, not to mention the Neovas. So much so in fact that I suspect I'll have runflat capability given my very low corner weights. The tires weigh 15.5 lbs each. Not bad. I have previously calculated that the total unsprung weight on my car will be around 38 lbs per corner. That's 3 lbs lighter than just the rear wheel and tire weight on the Elise. The overall sprung/unsprung weight ratio should be similar for the dp1 and the Lotus.
Also, this being the start of a new year, I took some time to reflect on the last eight months' worth of progress on the design, from the first CAD that I had considered 'acceptable' to the final version that I considered good enough to actually go forward with manufacturing. A bit of a difference :)
01/27/04 I've been trying to figure out how to mount the motor so that vibration is isolated. In the Westie, despite the motor being mounted in rubber bushings, the vibration is enough to be problematic. For example the car eats lightbulbs because filaments fail. Before I padded the seats in that car it was also quite unpleasant on the kidneys. So I've decided to solve the problem in the dp1, in the interests of reliability, comfort and overall refinement. After searching around I discovered that Subaru uses lightweight hydraulic engine mounts with composite housings on the new Legacy (I think 2002 and on). A hydraulic mount is essentially a rubber cell filled with antifreeze, the rubber acting as a spring and the fluid acting as a shock absorber. Supposedly these are five times more effective at isolating vibration than pure rubber mounts (this according to a Subaru whitepaper on the subject). Now of course properly designing such a mount requires figuring out weight, frequencies, etc. But I guessed that since my motor weighs about half of the Legacy one but spins at roughly twice the RPM the overall damping should be pretty close. So I ordered one mount to see what it's like (at $76 it's not exactly cheap but not prohibitive either). It's a bit bigger than I anticipated....
Still, due to the composite frame the mount only weighs 2 lbs. I'd need three of them. Perhaps a 4.5 lb weight penalty would be acceptable (guessing that the lightest acceptable rubber bushings would weigh 0.5 lbs each). There is a complication in that the mounting tabs are not parallel to each other and are at some weird angles, so I'd have to think about the installation for a while. Maybe I'll look at some alternatives in the meantime.
02/05/04 Decided that the Subaru engine mount was not the right part and returned it. In response to my earlier engine mount post I got an e-mail recommending a vendor in England. It took a bit to get a hold of them but they're looking at the application and will come back with a recommendation shortly.
In the meantime, got the first photo from Janicki, the outfit manufacturing the body plug. It is of the raw foam prior to machining.
You can sort of make out the general contours. Apparently the part is already on the machine and I'll be getting progress photos soon. I'll post them as I get them. Rather exciting, for this will be the first part that actually resembles the finished car, and is all 'mine'. Probably another week or two before the plug is complete.
Got a new in-progress picture. The foam has been rough-cut and is just about ready to be fiberglassed, after which it will be machined to final shape and sanded. Then I'll bring it here, have it primed, polished, and ready for mold construction... Fun :)
This is where it starts to sink in that hey, I'm really doing this... Better spend some time on the frame.
02/22/04 The plug should be done tomorrow. Can't wait! Wish they'd send more pics...
In the meantime I've decided that my most recent approach to the underbody is way too complex and too iffy aerodynamically. I was hoping to get another CFD run done but it's not clear when or if that will happen. One issue is that I'd be asking the air to move laterally quite a bit, but the channels are really shallow at 2.5" relative to 3" ride height. After thinking about it for a while I decided that it's bound to just create a bunch of turbulence with little downforce. Also the one-piece construction would be expensive to repair or modify. So I'm now working on an improved 3-piece floor solution that allows the underbody air to flow straight, is more manufacturable and more repairable. It is also a refinement of my dual-phase diffuser concept. I've got the nose and tail done, will post pictures when I have the middle as well.
Also spent some time thinking about the frame and engine mounting. I've looked at what must be a dozen different ways to build the frame but keep coming back to the tubes-and-brackets concept I had come up with back in July. Similar process to the bodywork design, really. A lot of work where the most important outcome is being more confident in the original concept. Which is worthwhile. So this is the current architecture (click on the transparency pic for large version):
Before I get any more e-mails on the subject, yes there WILL be protection around the driver's feet and a lot of triangulation in the frame! Some portions of the frame will be skinned with aluminum panels, and chains will be 'managed' in the event they should break. I just haven't gotten around to translating it from my in-head 3D simulator to CAD yet :) The same 'simulator' has what it thinks is a good solution to engine mounting. Now I need to make a trip to my local BMW dealer for some differential mounts and a driveshaft 'guibo' flex coupler. One interesting thing about the chains is that due to small sprocket diameter, at 160 mph car velocity the speed of the chain relative to the chassis is only 40 mph. Which of course is still a nine-pound bundle of steel traveling at 40 mph and so deserves due respect.
I will finalize the underbody first, since mold construction is a long-leadtime item, then go for the final push on the chassis.