Description
Key Learnings
- Discover the capabilities and benefits of the project and how it addresses traditional challenges with aero simulation.
- Compare the Studio Wind Tunnel Project to traditional methods intended for simulation engineers and analysts.
- Learn about validating aerodynamic performance at any stage of the design pipeline.
- Learn how to integrate optimization insights into automotive surfacing to improve efficiency and performance.
Speakers
- James NevilleJames Neville is a simulation expert with a specialty in CFD and Generative Design. He began working in the simulation field in 2003 and has experience across a wide range of industries. A mechanical engineering graduate from Virginia Tech, he began his career at Blue Ridge Numerics, where he focused on customer success through consulting services and now serves as a global subject matter expert at Autodesk. James lives in Pittsburgh, Pennsylvania.
- James CroninJames Cronin is a Subject Matter Expert for Automotive Design at Autodesk Inc. James joined Autodesk after spending 11 years at Nissan Design America as an Alias Modeler and most recently as their Visualization Lead. Prior to Nissan he worked for Alias|wavefront as a design consultant. He graduated with honors from College for Creative Studies with a BA in Industrial Design.
JAMES NEVILLE: Welcome, everyone, to the Autodesk University session on the Studio Wind Tunnel Project. Here is a safe harbor statement. You've probably seen this before, but feel free to take some time and read it if you need to.
So my name is James Neville, and I'm a Simulation Subject Matter Expert for Autodesk. I've been with Autodesk for about 12 years now, and I have almost 20 years of experience in the simulation field. I'm looking forward to sharing this project with everyone today. And I'll pass it over to James for his intro.
JAMES CRONIN: I'm James Cronin. I'm the Subject Matter Expert in Automotive Design at Autodesk. I've also been here for about 12 years. And before that, I was in the Design Studio doing visualization and Alias surfacing.
JAMES NEVILLE: All right. For the session today, we're going to start with a very high-level overview of automotive aerodynamics. And then, we'll discuss the various challenges that are associated with evaluating aerodynamics with conventional tools and processes. And then, that will lead into an overview of the Studio Wind Tunnel Project and how it aims to solve some of those challenges.
So aerodynamics is important-- it's an important science of vehicle design. The influence of aerodynamics on drag, range, handling, cooling, noise, they're all important. And automakers evaluate designs in a number of different ways. There is virtual wind tunnel testing or CFD, physical wind tunnel testing, and ultimately physical testing of on-road prototypes as well. For our talk today, we're going to focus on virtual wind tunnel testing and how it can be used to help automotive designers.
Let's talk a little bit about drag. So this is the only part of the presentation with an equation, but I think it's an important one. First, drag is simply the force on an object from a fluid as it travels through that fluid. And in this case, air can be considered fluid. Every object, whether a baseball, a vehicle, or a rocket, all of those objects will experience a drag force as it travels through the air. Every object can be described with a drag coefficient.
The more aerodynamic or streamlined the shape, the lower the drag coefficient. In the equation here, it simply shows that the drag coefficient goes up or down. As it goes up or down, the power required to keep that object traveling at that same speed will also go up or down. So in a nutshell, lower coefficient equals lower drag. So if you learn one thing from this slide, lower coefficient equals lower drag.
The illustration on the right provides a little list and the drag coefficient for a few simple geometric shapes. So no surprise here that the streamlined or the teardrop shape at the bottom, you know, that body has a significantly lower drag coefficient compared to the rest of the shapes. For a bit of fun, in case anyone was wondering, my esteemed colleague and co-presenter, Mr. Cronin has a CD of 1.0, more or less making him as aerodynamic as a brick. Literally, as aerodynamic as a brick.
I don't know how I would fare in a wind tunnel, since I have a lot more hair up top, a lot less beard. Maybe 1.1. I'll have to test one day.
So back to the automotive industry, we're going to take a look at a few examples to see a representative range of automotive drag coefficients. So the scrolling list on the right, that shows the CD from some of the top 100 most aerodynamic vehicles. And it's no surprise that most of them were designed in the last three to five years. It's also interesting that many of them are electric. We'll get to that in a little bit.
If you look at the average drag coefficient of vehicles on the road today, you'll find that most of them have a CD of between 0.25 and 0.35. SUVs, trucks, they're not as aerodynamically efficient, and their CD can range anywhere from the mid 0.3s all the way up to almost 0.5. The Lucid Air, the car pictured on the left here, is one of the most "slippery," quote, unquote, vehicles today, with an extremely low 0.21 drag coefficient. The Tesla Model 3, another very aerodynamic car, the low CD was one of the design hallmarks of this vehicle when it was first introduced. And I believe they recently improved it slightly with the facelift for even further aerodynamic efficiency.
The C8 Corvette is a little bit different. It sits in the 0.32 range, but the design doesn't really aim for all out aerodynamic efficiency. You can see, it's got scoops, and vents, and louvers, and wings. And all of those elements have an important role in managing the cooling and producing downforce for maximum performance.
The latest generation Grand Cherokee is actually a much more aerodynamic design than the previous models, but it's still an SUV. Right? It's still a multibox shape, and it it manages just a 0.357.
And then, for giggles, here is a modern F1 car. So depending on the track requirements, teams will use a variety of different wings, different configurations. They'll use skinny ones. They'll use fat ones like you can see here. And the CD can vary wildly. But I just think it's an interesting data point that road cars are about three times more aerodynamically efficient than F1 cars. Of course, they're designed to do very different things, but just an interesting piece of information on the CD chart so you can sort of get a bracket of where vehicles fall on that number.
So let's take a look at the industrial shift away from internal combustion and towards battery electric vehicles. We're going to compare two vehicles which are essentially identical in every way except for their source of power. They'll have the same exterior design, the same weight, same wheels, just one has an engine and the other has batteries and motors. So both of these vehicles generate power to move forwards. The power, or the motive power, is the same for each car.
So a portion of it goes towards fighting rolling resistance-- so the wheels and the tires-- and at highway speeds, a slightly greater portion of that fights against drag or wind resistance. But not all of that power that's created makes it all the way to the wheels. So there are inefficiencies. There are thermal losses, friction, et cetera.
When we look at the total power generated, it shows how much goes towards motive power and how much is lost before it even makes it to the wheels. And this is one area where it's a huge difference between internal combustion and electric vehicles. So as much as 80% of the total power generated by an engine is lost to friction. Ultimately, only about 20% of that power goes towards actually pushing the car forwards. So comparing that to electric vehicles, they simply don't have as many losses. They are far, far more efficient, with only about 15% of their total power lost and the vast majority of their power goes to the wheels.
So this chart here really illustrates how gasoline or diesel-powered vehicles must make a ton more power-- total power-- to go the same speed as an electric vehicle. Just a lot of that power is lost to waste. So let's just focus on the aerodynamic drag for a moment.
Of the total power produced, for a gasoline car about 10% of that power will fight aerodynamic drag. For an electric car, over 50%. More than half of the total power goes towards fighting drag.
So if we're able to improve the designs of both of these cars to reduce drag by 10%, then the internal combustion car would only see a 1% improvement in fuel economy compared to a 5% improvement to the electric vehicle. So what this shows is that electric vehicles have much more to gain from improvements in aerodynamic efficiency. And the industrial shift away from internal combustion and towards electric power makes aerodynamics far more important now than it was in the past.
JAMES CRONIN: So now, let's talk about the current challenges in the automotive design pipeline. Typical-- this is an illustration showing traditional design pipeline from start to production. So many sketches are drawn. From those sketches, 3D models are started, whether in clay and SubD surfacing. Those designs are then-- from those there's chosen maybe two, and then from those two are chosen down to one.
Now, if we look at where engineering lands in this chart, traditional simulation tends to happen farther down that stream. So designs have already been chosen, some selections have been made, then simulation is involved. Very little simulation actually happens early in that process.
So the simulation challenges for design start with geometry preparation. This can take many days. The simulation expertise required is outside of the Design Studio. The tools and software that are used to simulate the aerodynamics require an engineer and analyst to run.
And there's the hardware and software accessibility. And then, the number one issue with the Design Studio with simulation is the turnaround frame. It's not fast enough for design.
So if we look at the typical process today, it starts with prepping geometry. So as an Alias modeler, I would have to stop working on the design and prepare the model to hand off to simulation. This might require building closeouts, stitching the model, removing gaps, making sure that it's a solid, watertight volume to be able to then hand off to the simulation team.
Once the simulation team has it, they have to do their own prep work. So they have to set up the model, prepare it for their software, and then run the simulation. And that requires first to generate the mesh, and then they're sending the mesh and the geometry up to the compute platform, and then the simulation is run.
Once the simulation is run, the results are in interpreted, and then the feedback is given back to design, at which point the cycle starts over. And this can take up to one to three weeks. And if this is happening for every single design that's in the pipeline, this can add a huge amount of time to the design workflow.
JAMES NEVILLE: So it seems like a pretty good problem to solve, right? So frankly, companies have been trying to solve it for many years. And I think the timing just wasn't quite right yet. So you might ask, so why now?
First, the fluid solver technology has improved a lot in the last couple of years. There are new methods. There are refinements to older methods. And various optimizations simply make the new tools way more capable than they were in the past.
Hardware resources to run these new tools is far more accessible than it used to be. We have on-demand access to seriously massively parallel multi-GPU cloud instances that we didn't have even a couple of years ago. We sort of already discussed this, but aerodynamic performance is even more critical today to modern electric vehicles. And then, finally, designers, they've been asking for this for a long time. They have a desire for more feedback earlier in the design pipeline. And yeah, these are some of the reasons why we're looking into this now.
And this is where we introduced the Studio Wind Tunnel Project. It's really a collaboration between several teams here at Autodesk. It has three key pillars. So first and foremost, the project is aimed primarily and directly at automotive designers-- so users who are familiar with Alias or surfacing tools, and they want to get rapid aerodynamic feedback to let them know if their design is essentially going in the right direction.
Second, automation. So if it can be automated, it will be automated. So the project really aims to provide aerodynamic feedback in the most streamlined and hands-off way possible. And finally, third, the project relies heavily on a next-generation CFD platform. So this unlocks new potential for us and our automotive customers.
So starting off with a focus on the designer, the project directly integrates into the Alias user interface. Once a user has a model on which they want to gain aerodynamic insight, they will simply select the components that make up the body and the four wheels in the Studio or the Studio Wind Tunnel Plugin. And want to stress here that there's very little geometry preparation required, as James mentioned a couple minutes ago.
In many cases, the geometry preparation is the same as what is already being done to prepare for visualization needs, like adding in blackouts or maybe a simple underfloor. So it's worth pointing out that the user is not required to make a watertight model. They don't have to remove excessive detail. They don't have to close panel gaps.
So once a user has assigned the various body and wheel groups, the tool will automatically orient the model correctly for the simulation. This ensures that all of the designs are placed in the same space and they can be accurately compared. So the last step is for the user to simply select a wind tunnel speed and click Solve.
Alias users will often work on a wide variety of model types. Early conceptual SubD geometry, laser scan data, fully detailed NURBS surfacing, and sometimes a particular model may include a hodgepodge of all three types of geometry. And the Studio Wind Tunnel Project is designed to cater to all of these workflows and to provide feedback to a designer regardless of the state of their underlying model.
Like I said before, if it can be automated, it will be automated. And full automation has been a critical focus for the Studio Wind Tunnel Project. A user starts in their native Alias environment, kicks off the wind tunnel through a simple and fast plugin, and then every detail of the analysis is handled behind the scenes. So ultimately, this ensures that any user, regardless of skill level or knowledge of simulation, will be guaranteed the same level of accuracy and consistency.
All of the solves are handled on the Autodesk compute platform. This is a robust and scalable cloud-centric platform that provides token-based access to virtually unlimited amount of computing resources. And then design insight from the studio wind tunnel is brought directly back into the Alias user interface-- so velocity planes, pressure maps, drag coefficients, and a bunch of other design insights are automatically created for review. So this entire process, from start to finish, will take less than one hour.
So what exactly is going on under the hood, you may ask? The Studio Wind Tunnel Project leverages a state of the art lattice Boltzmann fluid solver. It's a multi-GPU, multi-resolution, voxel-based solver. And what that allows it to do is to be very fast, very efficient, and very geometry tolerant. It's built on the Autodesk compute platform, so it's going to provide easy customer access through existing enterprise agreements.
In addition to bringing simulation insight to automotive designers in a way that hasn't really been done before, we see this next-generation CFD solver as a foundation for future technology as well. There are a lot of really interesting things that the Studio Wind Tunnel Project would like to bring to customers in the future, like automatic optimization and near real-time simulation. So I think the next few years are really going to be an interesting and exciting place for automotive design.
All right, let's take a look at it in action. This is the next-gen CDF solver-- some of the raw data. Caravaggio Corvettes was kind enough to lend us a model for testing, and here you can see the studio wind tunnel simulating high speed airflow around the vehicle. So this method of simulation is super accurate, especially for external aerodynamics, as shown here, and it scales extraordinarily well on multi-GPU architectures. This allows an entire car to be simulated with a really high level of detail in a very short amount of time.
Results from the studio wind tunnel, the simulation there brought directly back into the Alias user interface. So designers can quickly assess raw numbers, like drag coefficient, or they can dig in a bit further and visualize which areas of the car are contributing most to drag. They will be able to choose from surface pressure visualizations or maybe maps of flow separation that indicate where the surfaces are causing inefficiencies.
Results-- visualization really brings the simulation to life within Alias, and it allows a user to dissect the flow around a vehicle. In addition to creating compelling imagery for downstream visualization pipelines, showing the flow can really help a designer better understand how their surfacing interacts with the air and what impact their design changes have on overall vehicle performance. So a quick recap to compare and contrast between traditional simulation methods and the studio ones on a project.
So the studio wind tunnel is laser focused for the automotive designer, not an engineering expert. No prior expertise, when it comes to simulation, is required to use the tool. Feedback is returned in a time frame that's much more in line with the pace of automotive design. And there's an emphasis on automation and ease of use over full customization.
JAMES CRONIN: Now, let's look back at the pipeline collaboration and the current state of the Design Studio. Design works on their data for a span of time and then hands it off to engineering. And as we mentioned earlier, it can take one to three weeks for that engineering feedback to come back into the Design Studio. But the issue is that the Design Studio doesn't stop. They continue to work on their data during this time span, and this causes a impact point where the feedback-- the information that they receive from engineering-- might differ from all the changes they have applied to their model.
So the feedback and the results would have been very useful to receive earlier in the design pipeline because of the massive impact or the changes to the general proportion or shape that's required to solve the aero issues that may have come up through the shape or the design that the designer has chosen. So if we look at the studio wind tunnel solution, what we have is design making smaller design cycles, with engineering in an over the shoulder capacity that is able to join a meeting, look at the results inside of Alias, give feedback on changes they would recommend to the surfaces, so that the designer can quickly reiterate, make adjustments as needed, and continue to work on the proportion and shape of the design without having this huge impact from late-stage feedback. Now, when the design data is handed off to engineering, engineering is getting a much more accurate model that's designed with some intelligence behind it. This will allow engineering to spend less time giving more accurate feedback to then hand back to the design staff.
JAMES NEVILLE: So in summary, aerodynamic insight for automotive design, it's super important. And it's probably more important today than it's ever been. The shift to electric power vehicles really puts a much greater emphasis on aerodynamic efficiency. Current methods to obtain aerodynamic insight, they just don't work well for designers. They're largely intended for a different audience, and the turnaround time, it's just not fast enough to keep up with the pace of design.
The Studio Wind Tunnel Project aims to tackle this problem and give our automotive design customers really a better way forward. The intention is to create new workflows within the existing design pipeline and not to replace traditional downstream engineering simulations. There's still a place for downstream engineering, but this is really a new workflow within the design pipeline.
So I want to thank everyone for attending the session today. If you have any specific questions about the project or if you want to chat about automotive design or simulation, please feel free to reach out to me via email. Also, go ahead and add a comment to the class page if you found the session interesting and click Recommend. Thanks again.