01/06/21 Some exciting news on the D8 front. We have a committed first customer, and the plans are ambitious - not ready to reveal the full extent yet, but suffice it to say it will be in keeping with our record of pushing the envelope. Some things I can talk about now though.

First, the frame refinement has started. This is really the evolution and culmination of all the experience gained so far. And that's saying a bit. The pictures below reference a GT40 layout - remember that part of the D8 mission is to be a modern platform for CAN-AM era replicas. The emotion and the raw connection, but leveraging all we've learned in the ~50 years since.

First, the pictures:

Yes, this is a work in progress. Not final by any stretch, but gets the fundamentals down. The D8 platform will be able to take a variety of powertrains - gas, electric and hybrid. It will have RWD and AWD options. Many replica bodies will be possible, as well as many originals. My triple-shock suspension, chromoly frame, air skirt aero and other bits will come into play. And yes it will be expensive. Sorry, this is just the reality of it.

The year is off to a good start as far as ideas and progress go. Let's see what we can make of it, despite the craziness. If all goes as planned, this will be epic.

03/21/21 Here is another aero post. I've been running CFD almost daily for several months now, testing one thing at a time, with the goal to learn what does what.

As before, all the simulations are carried out using Solidworks Flow Simulation 2012, on a AMD Ryzen 9 with 64G RAM. The runs are at 100mph, sea level, 20C, with moving ground and rotating tires.

Of the two configurations below, one makes 348 lb downforce at 325 lb drag (1.07 L/D). The other makes 879 lb downforce at 324 lb drag (2.71 L/D). Which is which?

The answer is that the first car, with the messy cockpit and no wings, is the better performing one. And not just a little bit better. 2.5 times the downforce at identical drag.

On page 1 I went into some of the steps of how that is possible, getting to about 700 lb downforce. Below is some of the work of getting from 700 to 879 with no drag increase.

One of the things that makes such rapid (which means low cost) iteration of runs is the SolidWorks meshing algorithm. It automatically creates, and then refines, a high quality mesh in a few minutes whereas a manual process would take many hours each time. Yes, when expertly done, the manual approach *may* yield more accurate results in some cases. But when all you're after is apples-apples comparison, it's hard to beat click-and-go efficiency of the SolidWorks setup. Below are some examples of the mesh, and I'll discuss them further.

The blue cells are full fluid, green ones are partial. You can readily see the refinement in areas of high detail where it's necessary, and much coarser mesh in free air to reduce computational burden.

Yes, the cells are prismatic only - which some may argue is 'inferior' to hex, but it works well when sized properly for local conditions. Which is what the refinement algorithm does.

Another part of overhead is model prep. I discussed running my model as a reality check with someone who does this professionally. We got into model prep - and things like eliminating sharp inside corners, having only a single solid, etc. It became clear it would take many hours just to do that. Multiply it by 200 runs and it really starts to add up. Whereas Solidworks takes it all in stride and is happy with assemblies, interfering solids (most of the time), sharp inside corners and all. A good example is where the rubber meets the road, literally.

It handles the wheel/road interface gracefully without having to add artificial fillets or over-simplify things.

That isn't to say the software is bulletproof. You still have to do reality checks. For example, I noticed airflow that didn't look right. It may not be obvious in the first pic but look closely at the zoomed-in one.

The air goes THROUGH the side wall. A bit of poking around showed why - it's a thin wall (0.3") at 0.25 degree angle to the airflow. The meshing algorithm didn't see it necessary to make the mesh small enough in this area so it made a hole where there isn't one. Increasing wall thickness to 0.5" solved the issue. Being aware of the limitations of the tools is a key part of engineering, and is one of the reasons that at some levels it is very much an art.

In earlier posts I talked about 'hollow body' vs 'wing' downforce, mentioning that this car takes primarily the former approach. However the real world is seldom clear cut, and in fact all of the downforce gains from earlier versions were made by utilizing some of the 'wing' aspects. It was done primarily by raising the front lip of the floor and letting more air in under the car, as opposed to keeping it out.

The resulting low pressure areas under the nose and the start of diffuser can be clearly seen in the first picture, as compared to the second. As mentioned earlier, the 'wing' approach does have the characteristic of being very sensitive to shapes and interferences. The following illustrates. All I did was add a very subtle, large radius 'crown' to the floor. Only 0.75" max deviation over a 100" span. You can barely see it below (the crown doesn't go full width).

I then did some runs with the peak in various positions along the wheelbase. The pix below are no bump, bump forward, and bump aft. It should be obvious which is which.

It does not significantly change the overall downforce or drag, but it does change front/rear balance quite a bit. Which makes this knowledge another useful tuning tool.

Next I experimented with wing effects, starting by just doing half a wing for a clear a/b comparison.

I was expecting more effect on the bottom side airflow than I got. The topside and wake differences are interesting and informative. The half wing resulted in 1,013 lb downforce and 382 lb drag (2.65 L/D). The next test was a reduced span wing, followed by full span. In the first pic below the grey portion of the wing is the difference between the two. the colored portion is what was actually run to produce the flow lines. The differences are interesting, with partial wing getting 1,035 lb downforce at 387 lb drag (2.67 L/D) and full wing getting 1,084/394 (2.75 L/D). So the near-full span is only a 22 lb downforce gain over half wing, but the extra bit of span adds another 50 lbs. Something to explore in more detail later perhaps, as well as the use of endplates which are absent in these runs.

This is apples-apples with the wing being in the same position and AOA. Significantly higher downforce configurations are possible but the point here is to see the relative effects of a specific change. This is why being able to do a couple hundred runs with minimal pain is an advantage. Much of this was done literally in my sleep, with the simulation running overnight.

I also experimented with longitudinal/vertical placement of the wing and different angles. The effects were largely as expected, if one keeps in mind the wing/body interactions discussed in earlier posts. The overall downforce goes up, as does drag (keeping L/D fairly constant), and the balance shifts rearward as more wing is added.

One final tweak on this conceptual bodywork is turning the spoiler into a slotted flap (some could argue low wing) by putting a slot in front of it. This invokes more of wing-like behavior to the overall shape. It is not any taller than the previous solid spoiler approaches. but it now interacts a lot more with the underbody flow and is more sensitive to the interference from the cockpit.

The flow separation on top shows there's more to be gained with shape fine-tuning, but the numbers went from a best of 879 lb downforce/324 drag (distributed 486f 393r) with solid spoiler to 930 lb downforce/326 drag (distributed 480f 450r) with slotted one, bumping L/D to 2.85.
Could likely get to 3.0 with tinkering, but I'll save that for the styled version. This tells me what the parameters are and what I can change to adjust performance.

Next comes the fun part of actual styling. It will be an iterative process. Once the basic shape is in CAD I'll run some CFD to see what effects it has, then will use what I've learned to make it do what I want while keeping appearance in line with goals. More to come.

06/05/2021 A quick but significant update. I have recently launched the development of the D9 (see the separate blog), and perhaps more importantly, a separate company called Modular Battery Technologies Inc. that is developing the fast swap battery module technology to power the car, and much more.

There are many reasons and benefits to the ModBatt tech (6 patent apps files in last 5 months, 4 more in progress), with wide range of applications. The goal is to change the world. But when applied specifically to motorsport, the modules enable an EV to participate in endurance racing. Each module is 3KWh, weighs about 35 lbs, is 800V and can deliver a peak of about 45KW of energy (60hp). Yes, 15C, proven with D2EV on the dyno, on the Mountain and on the salt. This is buildable today and we are designing and building them starting right now.

The modules are sized to be easily handled by a single crew member and are completely safe when removed from a vehicle (this is part of the patent-pending stuff). Each module can be swapped out in about 15 seconds. A light car like D9 would use 4-10 modules. A larger D8 would use 10-20. A superbike would use 3-4. This means that a 10-pack of modules can be swapped out in under 3 minutes. Depending on particulars, that would get about 30 minutes of run time at track speed, during which time the other set of modules can be recharged. So still a pit time penalty vs a gas powered car, but no longer impractical.

Not all modules have to be swapped, and not all modules need to be present. Fast charge capability is there too if that's preferred. This gives the most flexibility - even in shifting weight around as needed.

When starting with a blank sheet of paper, it can be anything. You need constraints to make it real. So what better constraints than several different applications at once? The goal is to use the same modules in D9, D8 and the next D47EV (the current one already has old-tech battery made).

In the D47 picture the tubes will obviously have to be reconfigured - this is just a quick reality check. SolidWorks is awesome for this. Also notice a first pass at the 3-shock configuration on the D9.

The emphasis in these designs is access for swapping of the modules rather than CG height optimization. Yes, the modules can be oriented horizontally and stacked flat if that is the priority. And in future designs they may well be. This is still slightly lower CG than an equivalent gas car.

All the cars will be using the new Cascadia integrated drive modules (motor/controller combo), with over 350KW (470+hp). D9 and D47 will use one motor unit and up to 10 modules, D8 will use two motor units (with front being optional) and up to 20 modules.

It may seem crazy to be working on all-new modules and three car designs at once, but if you consider all the above, it's actually the fastest and most efficient way to arrive at a design that works in the real world.

This is fun :) I'm excited.

07/03/21 Good progress on many fronts. Finished the D47EV2 chassis redesign, updated the ModBatt module configuration, and ordered the chassis tubes. The frame is already 90% done. When the tubes come in, it will be finish welded and off to powdercoat.

Having figured that out, it was the D8's turn. Now the basic chassis layout is done and fundamental interference checks are complete. Still need to detail out the tube kit for trim, fit, weld vent holes, etc.

The entire frame will weigh about 250 lbs. The main rollhoop structure (highlighted) is 1.75" OD, 0.120" wall. The rest is a mix of 1.5" and 1.0", 0.065" wall tube. All annealed chromoly of course, TIG welded as all of our chassis are.

Since the D8 is an evolution of the D2, it is a fun comparison. The new car is 9" shorter in length (95" wheelbase vs 104"), 6" shorter in height (40" vs 46"), and just a tad wider at 82" vs 81.5".

This is the EV version. Yes, there will still be a gas-powered (or hybrid) version using the LT engines, which will have a different rear tubing structure. The main packaging difference vs the D2 is a much more reclined, feet-raised driving position. The feet are raised to allow for the airskirt. The height is reduced to be able to fit GT40 bodywork, which is still a planned option. We even have one onhand.

With the chassis tubes defined, I ran some more CFD to confirm that the ariskirt still works. It does. Also added a rough cut of the enclosed bodywork.

Styling work is in progress as well, but too early to post yet :) Once that is farther along, I'll rerun CFD. It's a process. Everything is connected to everything else. It's about a 20-dimensional puzzle. Which is what I enjoy the most. Keeps the mind busy. I haven't watched TV in over two decades, and haven't watched a movie or a show since Westworld ended. Too busy spinning the various car designs in my head. Now comes the fun part of making them real.

07/20/21 Lots of fun progress. First, we are finally at a stage with the styling where it can be shared. Keep in mind it is not final and will continue to evolve. I'm collaborating with Michael Young on this design. He did the D5 concept earlier and we work well together. Rather than commissioning a from-scratch design, the goal is for him to help shape my vision and come up with a car that fits into the Palatov family, as well as accomplishes the packaging, aero and other goals.

Here is the most recent set of renderings, prior to some additional CFD-driven revisions. It's getting there, although many areas still need adjustment and refinement.

The design is an iterative process and we communicate by a combination of CAD, photoshop and sometimes hasty hand-scrawled notes like this:

With the body this far along I plugged it into the CFD model. First runs were without the airskirt or any other refinements, resulting in 189 lb downforce and 220 lb drag at 100mph. Now, after more than 20 revisions, I'm up to 581 lb downforce and 240 lb drag. Good progress but not done yet.

To correlate some of the CFD concepts we track tested a D2, with the objective of confirming the effects of a spoiler on a wing at high angle of attack. The first picture is 'low' angle (still very high actually) without the spoiler. You can clearly see the wing is stalled and airflow is detached. Second is 'high' angle with the spoiler. Despite the much higher angle the airflow remains attached because the spoiler changes direction of the air, and the wing ends up at a much lower AOA relative to local airflow.

This confirms CFD predictions and allows me to better optimise the 'base' D8 wing. This will be the standard configuration with a target of about 600 lb downforce at 100 mph. With everything I've learned, I can easily create a high downforce version with the addition of a second wing element and some stuff upfront that will more than double those numbers when needed (like for Pikes Peak).

This is what the latest CFD-updated version is looking like. More changes to come.

If you find this site useful or interesting, consider supporting it with a paypal donation in the amount of your choosing to dp@dpcars.net