0.2755

I was recently asked by a few people around the likelihood of claims made by Windrose Technology 's upcoming Class 8 BEV. In particular, whether or not a claim of a drag coefficient of 0.2755 (live link here) appeared valid. This is important, because:

• Cd<0.3 for this class of vehicle is an incredible (but not impossible) claim,
• Traditionally, aerodynamic forces dominate resistive forces at highway speeds (and so efficiency and aerodynamic claims go hand-in-hand, though they diminish in relevance as aerodynamics load reduces); Windrose has shared some total trip energy efficiencies which are highly competitive and may suggest one or more leading developments, and
• The claim wasn't supported by methods and or context, and the pics in the article (showing some non-ideal tractor-trailer gap) didn't help.

So in being unable to find any further data on the drag performance claims, I put it out there:

And whilst the collective Internet didn't respond with 'over here mate, here's some a more substantiative explanation', this was picked up by Windrose's CEO, Wen Han :

To which I responded:

As described I did send as much as an email too. To be fair I could have been more descriptive around my intent in my initial post. At the time of writing this article Windrose's CEO hasn't written back (he has commented adjacently on the post several times), though he's got a company to run so let's keep an open mind there.

Per my usual rules three or more parties asking the same question pro-bono in under a fortnight triggers a LinkedIn article to direct remaining interest to, so let's unpack why this is important, and more importantly what it means for industry.

A bit about me

Whilst some of Windrose's more visceral supporters have suggested I believe the claims impossible/am an armchair critic/have not achieved in the space, I've previously worked (quite happily) as a vehicle aerodynamicist for a range of clients and organisations spanning passenger cars, race cars, commercial vehicles and other goods for clients and employers as diverse as established organisations, startups, race teams and what not.

I'd actually done a fair bit on commercial vehicles - they're highly interesting and relevant to most people in vehicle aerodynamics because their impact is traditionally outsized, and for which positive changes make for significant impact.

Quick thoughts on Windrose

Windrose is a company founded in 2022, and - per their website - develops globally, appearing to leverage supply and value chains spanning China, Europe and the US at least. Whilst it could be argued that BEV trucks will come together faster in the current age by virtue of EV technology supply chains being relatively mature and unique development requirements from passenger cars - Windrose is not building a Tesla Model S in 2010; market conditions and an ability to assemble relevant, competitive domain knowledge (it's fair to say there's already been a first generation of attempts to learn from in this space) give them fair potential to go faster and achieve better results - from the outside what Windrose represents in a corporate sense is a very serious prospect for interested investors: a genuine, (for those understanding the moment in the automotive sector) timely and well-considered disruptor in a market that very much needs as much.

Thoughts on aero claims

Let's be clear: I've zero beef with anyone at Windrose, I've not met anyone there personally, and I'm neither a shareholder in Windrose or any competing organisation. I'm ex-Tesla, had no involvement in the Semi (I left long before that started) and hold no shares or interest in as much.

I've seen Windrose's truck up close at All Energy Australia . It looks impressive, though a quick walk around won't tell everything about efficiency. I didn't get to talk to anyone there, I didn't get to crawl under it either. I'd suggest from the outside that aerodynamic development has received good attention beyond the obvious lack of side mirrors - swapped for cameras on pylons which in of themselves will reduce drag usefully.

Though it's outwardly similar to the Tesla Semi it's completely possible that it could have a lower drag coefficient than the Tesla, which claims Cd=0.36.

(Noting carefully that Tesla's drag claims - which are also unverified - were made against a backdrop of trucks and images thereof with very closely aligned trailers, minimised gaps, side skirts, no side mirrors and an image that may infer a straight-ahead condition).

At Cd=0.2755 Windrose is claiming >23% less drag, which is... huge. A 5% advantage in aerodynamic efficiency come production against established, detailed design efforts to the same end is incredible, but 23% is huge. If it's something that plays out reliably in a production context... then it'll go further for same or less battery. Less battery increases potential for more payload, opening up more revenue potential for operators on a lower cost base. It can means less reliance on charging networks, and lower adjacent infrastructure and service costs. Run the numbers: if we've two trucks with same frontal areas, same loads, same mechanical losses running the same route under same conditions and same speeds (you get the drill) and one with Cd=0.36 consumes e.g., 1.1kWh/km, then if the other has Cd=0.2755 it could run approximately 0.84kWh/km. As said, huge.

Trucks not only present with larger cross-sectional areas relevant to drag force, they typically run long distances regularly: transformationally-lower drag in the trucking industry creates opportunities for operators that simply don't exist today.

It should be stressed that getting a Class 8 rig under Cd=0.3 isn't impossible - it's been done before in research programs - but a real truck, with real loads and respecting real usage practicalities? It's the aero equivalent of running a four-minute mile in athletics: not impossible, requires a very dedicated effort, significantly difficult, and a watershed moment for the industry that'd redefine performance expectations in the space. If Windrose has delivered a realistically Cd=0.2755 truck with no obvious caveats then US taxpayers would have good reason to ask why a company that didn't exist 5 years ago just delivered what three generations of the DoE Supertruck project hasn't yet achieved.

If that number plays out reliably - even to being half as far ahead as Tesla's own numbers, which are already very good - then today there's not much reason to buy any other electric truck. Hence the interest in understanding that number contextually.

BEV trucks are supposed to be slick

That Windrose's product - like the Tesla Semi - is a BEV truck can raise expectations around aero performance usefully. There's a lot that being a pure BEV can do towards increased design efficiency on many fronts, not just aero - e.g., cooling requirements and engine placement vis-a-vis ICE trucks allow BEVs to be far more flexible in packaging towards aero-optimised outputs, and have less inherent packaging and operational compromises than an ICE might with a larger, hotter lump in the way. The potential of this paradigm shift can't be understated. Similarly, ICE designs have been improving aerodynamic performance a good bit now - operators generally want improved fuel consumption and so fewer rigs on road are chromed, bluff parallelograms with style over aerodynamic substance - but whilst there's many learnings, parts and systems that are transferable to a BEV make no mistake: BEV trucks should offer lower drag.

There are some things that BEV-ness doesn't automatically solve either e.g., ground clearance and trailer packaging broadly is what it's allowed to be in your region irrespective of what's towing it, and afterbody geometries is (generally) exceptionally bluff (e.g. many trailers, for many reasons, terminate in the back end of a rectangular prism creating a local, large region of low pressure that contributes significantly to overall drag (some fleets optionally fit afterbody treatments here - I've worked on a few - though not all do and some practical issues create a general reluctance for many fleets). So whilst Cd=0.2755 is reasonable for a new, aero-optimised-with-lots-of-attention-to-afterbody-flows passenger vehicle BEV, the ask on a traditional tractor-trailer configuration without as much terminating in a large rectangular prism is considerably harder.

But finding apples to compare with - difficult

The first difficulty with any truck OEM's claim - what configuration was it evaluated in? Lots could change it outside of a basic cab configuration - any optional tractor fairings and seals, tractor-trailer gap, any aerodynamic aids (side skirts, boat tails, undercarriage fairings, gap fillings etc), wheel configurations, trailer height and loading, trailer type and configuration... those from the industry can talk significantly to what's possible here. To my original point about the two photos in Windrose's original article (one of which was replicated in my LinkedIn post) are unlikely to be that which originates the drag coefficient claim. A tractor-trailer with Cd <0.3 is likely to have some pretty slick/sealed alignment. Beyond which its worth understanding just what configuration originates the drag performance claim, as it'll contribute much towards understanding how replicable the claim is in real-world conditions.

There's fair questions to aerodynamic configuration too; was it straight ahead, wind-averaged, headwind or tailwind... lots to consider. Whilst zero-yaw+zero-sidewind configurations will give best drag performance, any meteorologist will tell you they're not completely realistic.

With so much possibility there's real questions as to how two different trucks might really compare: e.g., does Windrose's current effort really beat Tesla's for the same load and same aerodynamic conditions, or is the delta just in configuration differences? Is Tesla's 0.36 even able to be considered 'real'? There's not enough information in the wild yet to substantiate as much (not least no public A/B tests) and I've not reviewed design details for either in detail, though I'd bet Windrose's side-vision cameras give a material advantage and Windrose's efficiency claims from trips present some compelling data (if not as detailed towards objectivity as public data for trips with the Tesla Semi yet - though I'd suggest neither dataset is perfect and that interested parties do review Windrose's disclosures here, as the inferred efficiency performance is strong).

Can't just stick it in a wind tunnel

The second difficulty comes with how to accurately measure drag performance at all.

Wind tunnel testing generally won't cut it; trucks of this size are rather large, best wind tunnels are larger still - replicating the real world in hyper-accurate ways would involve extremely large facilities adept in simulation of all ancillary factors (e.g., moving grounds, wheel rotation, yaw, freestream turbulence, etc). As (to put it mildly) wind tunnels and accountants tend not to like each other, the returns in capital and operating costs of most any test facility that can drive development well far exceed those of mythical facilities that might make for test with absolute perfection. Accordingly wind tunnels generally do not replicate the practical environment: they provide controlled, accurate, (often) low-noise and repeatable environments in which to examine aerodynamic performance. A good aerodynamicist will understand the limitations of whatever facilities at their disposal to drive development.

With respect to a truck the size of Windrose's efforts the number of facilities that could accommodate a full-size tractor-trailer combination are well-known, and each have practical limitations that would limit any ability to resolve truck aerodynamics accurately relative to the practical domain. Scale testing would be more likely, which would present with many factors that could each alter the flow field around a scale test article in ways that differ from the real thing. The limitations of any particular wind tunnel facility also play a part - open or closed jet, tendency to under or overestimate drag forces? We don't know.

Quoting four significant figures is unusual for a wind-tunnel-derived drag coefficient (it's highly resolute though wind tunnels are rarely so accurate); numbers with that sort of resolution usually come out of computational fluid dynamics (CFD). CFD tests can simulate practical domains more readily given sufficient compute resources, though the key word there's 'sufficient' - despite some simulations requiring supercompute-class resources, in a general sense CFD methods are neither exact; they approximate the flow field rather than calculate it exactly. While the governing equations (the Navier-Stokes equations) in principle describe fluid flow exactly, any numerical solution involves approximations in discretisation (breaking a continuous domains into a finite grid/mesh, numerical schemes (algorithms approximating key derivatives and integrals) and modelling assumptions (introducing turbulence models to approximate complex, unsteady flows). There's a very complete and ever-evolving field of science (one sufficiently prominent to have a significant portion of the world's supercompute resources dedicated to some form of CFD at any time): in short, any CFD assessments of drag performance are as good as the methods used; they are not (nor is their accuracy) simply or consistently expressed.

On the other hand practical domain tests (e.g., coastdown) give results that typically aren't extremely resolute (typically two significant figures) but can be highly accurate.

So to this 0.2755 - hence the interest in what it means, how it came to be and how relevant it is in a competitive sense - at this stage lots of possibility. Which is a nice way of suggesting that methods between OEMs can vary, and as such the absolute value of any individual claim is relatively meaningless without context.

Conclusions

OEMs have been publishing drag coefficients with some degree of creativity towards optimisation for some time now, that's nothing new among a slew of vehicle performance statistics that often require some creative interpretation to replicate in the real world.

There is little to gain in claiming a very explicit, industry-transformational level of key performance if the claim is not credible; whilst Cd=0.2755 is going to raise (at least) one eyebrow from most aerodynamicists, the bigger question's going to be around how Windrose's efforts compare against contemporary efforts in a relative sense and, if in any way favourable, just how a firm that didn't exist not so long ago managed to approach or even reset a key performance bar in an industry very much in need of as much (it's not as though incumbent firms haven't had a crack at lowering drag or BEV concepts/studies).

Ultimately someone, somewhere is going to A/B Windrose's efforts with Tesla's Semi or any other prime mover under tightly-known-or-controlled conditions. Right now both trucks aren't completely thick on the ground, are both somewhat in later stages of development and are unlikely to be released by their respective OEMs for a public group test. It'll likely happen eventually, but not today. So specific interest to these ends relies on disclosures detailing the reliability of the whatever configurations and methods underpin the claim, which Windrose may or may not provide publicly. This is in part understandable; aero performance won't be the only reason a firm does or does not buy a given truck (it's neither the only way to achieve leading efficiency numbers), though where industry-transformational numbers are being quoted openly by leading players there is (at least) significant investor and policymaker interest in leveraging these developments in productive ways.

Because if their claims are anywhere near ultimately verified it will do much to force a wider discussion around fair efficiency expectations in the sector - one that's had work to these ends percolating for decades, though which threatens real and needed relevance in an age where the efficiency performance of delivery factors directly in the carbon disclosures of the firms requesting relevant services. At a time where passenger vehicle BEV sales are overrun in some segments by cheaper new entrants with larger batteries to offset lesser intrinsic vehicle efficiencies, BEV trucking might hold us all to account.

So then it's possibly more aerodynamically efficient than the Other Big New Dog in electric prime movers, and I'd love to see some redefinition of expectations in the segment to these ends; many would. It just might happen later than sooner.

But above all it kinda makes you wonder what Windrose might do next. The industry they operate in very much needs organisations just like them to succeed - to which ends I wish them a receptive market, corporate longevity, good product... And a positive, credible contextualisation of this particular claim.