07/22/18 I keep meaning to do a post, finally found some time. It's been a month since Pikes Peak already - we didn't run this year but David Meyer ran his modified D2 and finished with a respectable 10:38.7.

The modifications to the nose are his own, as well as those to the dash.

While we did not have a factory effort this year, that will change for 2019. We are doing an electric D2 with the goal of going after the new overall record set by Dumas in the VW this year. It'll be fun! A lot of the initial design and engineering has already been done. This is a client project so exact details will remain under wraps for a while, but it has officially started and the work is moving forward. This car will also lay the foundation for possible production electric D2, as well as a hybrid version and eventually the D5.

In the meantime, regular D2 work is moving forward as well. The seat insert mold is completed and we've pulled a couple parts that will be used to mock up the new interior.

With the insert the seat is significantly more comfortable than without so it's a step in the right direction. Will be even more so once the inserts get upholstered.

Fitting the AC unit without impacting forward view too much is a challenge, but it looks like it'll work.

09/11/2018 Quite a few things are happening on the D2 front, some are too early to talk about but I'll provide a small peek. First, the stuff that I can discuss freely - interior work. And it's a pain. We are on a very steep learning curve, but we are learning.

A number of dash shapes have been tried and we're narrowing it down. The tradeoffs are fitting the AC unit, fitting the car, the windshield, not obscuring the view for a wide range of drivers, providing room for knees and instruments, not feeling claustrophobic and so on. A few of the countless iterations are shown below.

You get the idea. Nowhere near done but getting there.

Now the stuff I can only talk a bit about. We are working on the next generation D2 which will also serve as the platform for D5. It is as much about developing the technology as it is about any specific car design. The range of bits spans from bigger brakes (that will become carbon), very nice Forgeline wheels (18x13 rear), to the most complicated bellcrank design I've done yet. And of course a whole new chassis and suspension setup, drawing on all the accumulated experience and moving the game on a bit.

The focus point and motivation for all of this is the most ambitious project we've undertaken yet, and those who know our past might know that it's saying something :) Stay tuned.

10/25/18 Time for another update. Among the many things I've been doing is running a lot of CFD in the SolidWorks Flow Simulation package. The last time I posted about this I got a variety of responses, including some saying that if I'm not using FLUENT or Open FOAM then it's not even worth bothering. So, being an engineer that trusts data over opinion, I decided to do some research and conduct some experiments.

All this should be prefaced with the fact that I have previously carried out real-world testing and the correlation of the setup I use with real life strain-gage instrumented pushrods has been very good. This is why I keep using it. But could it be 'better'?

Mesh is a key part of CFD (though not the only one, by a long shot). SolidWorks has adaptive meshing and since I'm running this on a single ordinary PC (i7, 16G RAM and GPU that simulation doen't use) I've settled on parameters that result in about 1.5M cells for a half-car D2 model. It takes about 40 minutes to mesh and roughly 6 hours to run. This is what a close-up of the mesh looks like. All cells are orthogonal, unlike other packages that offer hex and other shapes, and the smallest is about 0.05".

Some opinions out there are that to get any accuracy on a car model you need 30M+ cells. I can't squeeze that into my PC, but after days of experimentation I managed to generate a 10.5M cell mesh. It took almost 4 hours to mesh and over 48 hours to run.

Unsurprisingly, the finer mesh does appear to produce better resolution of 'small' flows, as evidenced by improved vorticity plots - these are iso-surfaces at 1,500 cycles/sec. Basically small, fast moving stuff.

The 'bulk' flows don't look that much different from the coarse-mesh results. Sharp eyes will notice I'm now up to around 700lb total downforce at 100mph, which came in small and medium incremental steps from around 200 that we ran at Pikes Peak most recently. There's a practical side to all this work, first and foremost :)

The key thing that I care about in doing CFD is the ballpark numerical results for downforce and drag. This is what tells me whether a proposed change is worthwhile and also tells me what effects various pieces have. The objective is to determine what's worth building and testing in real life.

This is where the (pleasant) surprise came in: the difference in numerical results between 1.5M cell/6 hour mesh and 10.5M cell/ 48+ hour mesh is less than 1.5%, with the coarse results being more 'pessimistic'. In other words, for what I'm doing, the quick setup is perfectly fine. I also did an in-between test with 4M cells which, reassuringly, landed in-between the other two.

I recently bought my first DSLR camera and have been experimenting with it. Hugely capable piece of equipment, but in 90% of situations I still get 'better' (read more usable) pictures with my cellphone. That's because the DSLR has lots of sophisticated features that can be used by an expert or misused by someone without the knowledge or simply the time/energy to mess with it.

Based on what I've learned, this is the analogy - SolidWorks FS is the cellphone camera here. Some other packages are the DSLRs. For ordinary engineers in vast majority of cases, SW will give quick, usable results as long as you take basic care to set it up and, most importantly, correlate the results with testing. It is relatively accessible, usable, and accurate enough for most situations. But you do need to keep in mind the limitations. If designing an F1 car (majority vortical flow) this is probably not the right tool.

On the other hand, other packages offer potentially greater functionality but the effort and skill required are exponentially greater, and lacking those the results can actually be less useful. As I'm fond of saying, right and wrong are context-sensitive. All depends on the situation and the goals. I now have better confidence in what's right for the work that I'm doing so it was well worth the effort to look into.

Back to the physical world - many things are moving along at once. The D2 interior is one of those.

The special D2 project is moving forward as well. Part of the frame needs to be removable so we designed and made conical inserts that facilitate this.

Lots of tabs have been fabricated and materials arrived for the machined bits.

Also mounted the Hoosiers on the Forgeline wheels (those are 335 rears on 18x13 rims).

Halfshafts are here as well, custom-made by Driveshaft Shop.

If you just asked 'does that mean...' - yes, it does. In all, this is a logical but at the same time fairly radical evolution of the original design. We never stop improving :)

11/15/18 Lots of things happening. First, we got a new piece of equipment and a new (eventual) capability.

Yep, that's an industrial sewing machine and the purpose is to create our own custom interiors. This is one aspect that has been missing from our cars and as with all key pieces (design, welding, CNC machining, some composites) it is best to bring the capability inhouse if we can. This way we control the schedule, the cost, the design, the quality, and everything we learn is ours. Valerie, who is way past 'intern' status now, will take charge of the interior stuff. She has a lot of sewing experience in other applications, but car interiors will be a big learning curve for her and for us as a company. Judging from first day's experimentation it's going to be very good.

Along similar lines, Jay has been experimenting with 3D printing for some time. Since printers are cheap these days, they seem to be multiplying... It is actually odd in a way - closing up late at night I heard some odd whirring noises coming from upstairs so I went to investigate. This is what I found:

He even has a webcam watching them to see if anything went awry, and remote control to start/stop/program them. Kind of like discovering alien pods growing in your basement. The 'pod farm' has produced some useful stuff already, including this cover for an AIM dash.

Yes, we (for now) have given up on our own design due to not being able to dedicate the necessary time/resources to that project, and we are now an AIM dealer. But that dream is not dead, it just wasn't the right time/approach. But at least now we can make pods. Soon we'll be able to wrap them in leather.

And we're finally using the two offices upstairs that for the last 2.5 years have just been 'deep storage'.

Other random bits - more parts arriving for the special D2 build. Can't show most of them yet but here's a teaser.

And of course doing more CFD. If you noticed something odd about the numbers - especially for a configuration without wings - well they are at a different speed than before, with a different purpose :) Learning tons every day, much more to go.

11/21/18 More progress. The first set of suspension arms for the new generation chassis is back from powdercoat. They'll get assembled with spherical bearings soon.

Been machining some parts as well. These are the new bellcranks - by far the most complicated bellcranks I've made yet. We'll see how they work.

At the same time, running more CFD. Decided to add the wings back in for the high speed stuff. They don't actually add that much drag when set at low angle of attack. I chose the Eppler 420 airfoil when I designed the wings (it's a high-lift glider foil) and it's proving to be a very good choice. Even though the front wing is in close proximity to ground it does not appear to be overly rideheight sensitive. Some, of course, but not as bad as I would have expected. All this is great stuff to learn.

Next on the study list is yaw sensitivity, for which I will need a whole-car model since it's no longer symmetrical as far as airflow goes. The earlier experiments with mesh density give me confidence to use a setup that still fits into the machine, even though the overall cell count is doubled. Took a few tweaks but it's running now. Some interesting preliminary results already, I will post more when it's farther along.

Site Sponsors and Links:

- CMS Lap Timer - Advanced Track Timer App With Video Integration

Tire Rack - Tires, Wheels (use link when buying wheels/tires to support this site)