01/20/13 I had mentioned CFD earlier in this and other blogs. The SolidWorks Flow Simulation package that I'm using is quite impressive. One of its features is solution-adaptive meshing. Put simply it can run a number of iterations with the existing mesh, then examine results and see if there are areas where refining the mesh would improve accuracy and where relaxing it would improve computation without a penalty. The catch is that if it did this on every iteration it would take forever. How often it should be done really depends on the nature of the object being simulated and a number of other factors so SolidWorks gives you a way to set it up so that it's consistent from run to run and therefore lets you make valid comparisons.
This is all just part of the learning curve and really the only way to figure out what works for my particular case is by experimentation. Since every experiment takes several hours it's somewhat slow going. The upside is once I start a run I can go and do something else, like run the CNC, build bodywork panels, or even sleep. So in terms of real time it's not too bad of a burden but calendar-wise it takes a while.
One part of making sure that the results I'm getting are useful for decision making is to compare them to real-world data. Fortunately we did a good deal of aero testing on the Pikes Peak car last year and have both flow visualization (from wool strings) and lift/downforce data (from pushrod strain gages). To make the comparisons I had to create a CFD-friendly model of the D4 PPS and then run some simulations. After some tweaking (not quite done yet but getting there) I got results that generally agree with test data, within a few pounds.
Note that the simulation even picks up the flow reversal on top of the body, just ahead of the rear wing. It was indicated by wool strings captured in our test videos so it's encouraging to see same pattern in the virtual world.
Now that I have a bit more confidence in the setup and the results I've been able to test some simple tweaks on the D2. Below are three pictures - first is baseline, second is with cutting an opening in the rear bodywork around the rear license plate, and third is adding a panel to close the gap between the nose and the main floor.
The first change resulted in a 30 lb reduction in lift and a 20 lb reduction in drag. The second, 30 and 10 respectively. You can see the differences in flow patterns that explain the gains. Look closely behind the license plate in pictures 2 and 3 and you can see a difference in flow that was caused by a small change made in a completely different part of the car.
Of course the numbers have to be treated as 'ballpark' but it's an apples-to-apples comparison and valid as such. There are tons of other tweaks that I want to test. In real life this would be very time consuming and expensive. While in CFD it's not exactly instantaneous, I can still do it much faster, cheaper, with a lot more insight for analysis and in the middle of winter, too :)
Like any tool, CFD doesn't do the job for you but it lets you be much more efficient and productive as long as you know what you're doing. I'm not an expert yet but learning fast.
01/23/13 A bit of a milestone - the frame tube kits are here and the construction has begun!
This is a real growing stage for the company. A lot of stuff is happening in the background to make sure all the processes and procedures are in place and the build goes smoothly. Jay, our new intern, is working hard to transition resource management and planning from the combination of dry erase on windows, instant messages and emails into something far more organzied and formal. Concurrently we now have internal purchase orders for CNC and fabrication work, driven by M generating requirements from Quickbooks. All this is challenging but necessary and we're going to get better at it as time goes on.
Meanwhile I'm doing design work, running CFD simulations and machining parts on the mill. Often at the same time. The D2 design in the virtual world has gone from over 200 lbs of lift at 100 mph to less than 30 and more progress is being made. Of course this is not real life measurements and I have to constantly keep in mind the limitations of the process but the results are encouraging. The goal is no lift and slight downforce in base configuration with a substantial increase from the optional aero package. I believe it's very doable.
02/04/13 Our life these days is a continuous and somewhat extreme exercise in resource management. Key among these resources is time but all the others matter too. So much happening at once. There are many processes and systems we're putting in place, many existing ones are getting refined and adjusted. We are growing up fast as a company.
As for visible results... Well, we've had the first production D2 frame tacked together for a couple weeks now.
In the past, the next step would be to pop it out of the fixture, finish weld, add brackets and ship it off to powdercoat. However we want the subsequent frames to go together quicker and with much less effort needed to achieve the necessary precision. To that end Tristan has been designing and making additional fixturing ranging from laser-cut plates for locating various brackets to a 'rotisserie' made out of two engine stands and some custom fabricated pieces that will allow much better access for finish welding.
At the same time the CNC machines have been busy as well. Tristan has been programming the lathe and I've been running both the lathe and the mill (sometimes at the same time) to crank parts out. The CNC queue (being organized by efforts of M and Jay) is rather lengthy but it's getting done. This particluar aspect of our operations requires coordination of material orders, tooling (cutters, softjaws, fixtures), programming, setup and actually running the machines.
Composites work is also being taken care of. We have received the MK3 D4 bodywork (the best out-of-the-mold piece we've gotten yet) and are making progress on a number of other bits.
The new Pilot Pod design is coming along as well with Jay working on the prototype that willl become the master from which molds are pulled.
And yes, we do sleep (sometimes uneasily ;). But while that's happening a computer is running CFD to help nail down the D2 bodywork and aero options. A 5-6 hour run is perfect for overnight - set it up the last thing before going to bed and the results are ready in the morning. Having such a tool at my disposal is a great help but it can only tell me how my designs are likely to perform. It doesn't create the designs or even make suggestions. In some way CFD work feels like I'm constantly being judged - I come up with a new thing to try and then ask 'is this any good?'. The process of 'asking' consists of visualizing a potential solution and then coming up with a complete and detailed 3D model of it. Quite often it's a step backwards and the answer is NO. For all the science and technology there's an amazing amount of art to this. Explains why F1 teams continue to spend tens of millions of dollars on aero every year. If it were a simple matter of pushing a button and getting a definitive optimal result, they'd certainly be doing that by now.
Anyway, lots of momentum happening here. The results are at times not instantly visible but progress is being made every day.
02/16/13 More progress on things. The rotisserie and other fixtures are now done and production of the frames is moving forward....
We've received lots of materials for machinining and both mill and lathe have been running pretty much continuously cranking parts out. The CFD runs have continued as well. All of this will come together shortly.
02/26/13 The result of all the CFD is that I've finally arrived at the baseline bodywork setup. I may have mentioned that the goal was zero lift - I ended up with just under 200 lbs downforce at 100 mph. I actually think this is the minimum safe base setup for a car of this capability. I've learned a lot about what works and what doesn't (and why) in the process.
So now that the styling is final (almost) it's time for some higher-quality renders.
A more aggressive aero package will be optional, with the goal of 500+ lbs downforce at 100. But that's later. On the standard setup, the front 'winglet' is actually a full 6" off the ground (because of how it interacts with the diffuser) so it should clear most curbs and be pretty streetable.
Well, the next set of pictures better be of the actual bodywork :) Meanwhile the CNCs are still running (how did we ever do without?),, there's welding going on and design too. It's just a steady, relentless push to get everything done. Production frame #1 is getting finish welded in the rotisserie.
While frame #2 is taking shape in the jig.
Spring is coming up fast. I am feeling the same kind of pressure and urgency as I did last year for Pikes Peak. We made that happen. The goal is to make all of this happen too. And much more.
03/20/13 It's been a while since the last update and I've gotten many question about what is going on and when we'll see finished cars. The answer is that we've been working nonstop to crank things out. They are not glamorous or even readily visible things - ordering materials, programming making parts, building fixtures, welding, etc. As an example, the D2 frame has 122 individual tubes each of which has to be precisely positioned and TIG welded at both ends.
On top of that, there are 129 brackets/tabs/inserts each of which has to be made, positioned and welded. Times 4 cars. And of course before any of this can be done repeatably and accurately, fixtures have to be designed and made. And that's just the frame - suspension and various subsystems are equally as complex and all of it is custom.
For the uprights we have to hold 0.0005" tolerance on the bore so I first made a test part and installed a bearing in it, then pressed it out to check the retention force (2 tons, backed up with a snap-ring of course).
Now these and a lot of other parts can go off to anodizing. Yes, there are more than 4 cars' worth of uprights in the picture below - considering what it takes to get the material in, set up the CNC, verify tolerances and then run the parts it makes sense to do a few spares! This way the next car we build after the current batch won't be nearly as much of a mad rush and we'll be able to deliver it much faster.
Among other items the wheels and tires are here, so we threw the black ones on the proto to see how they look. I like it.
I've also been rethinking the approach to making the bodywork. Making a full-scale plug is not the best strategy for this type of design since there are many separate panels. The initial plan was therefore to hand-build the body on the prototype directly, then take the molds off it. But to get accurate results would take way too much hand labor, meaning time which we don't have. So now after evaluating a lot of options I think I've come up with an approach that will work and I'm creating the necessary parts in SolidWorks. There will still be a good deal of hand finishing but on a much more reasonable scale and I think the result would be much more accurate than what I had in mind previously. We'll find out very soon!