Integrating Aeroacoustics into Development through Simulation
Automotive IQ recently spoke with Franck Perot and Michelle Murray-Ross of Exa Corporation about the evolution of aeroacoustics engineering in the auto industry.
"One of the challenges here is to be able to integrate acoustics into the development cycle as early as possible to give enough freedom to the engineers and to truly have the opportunity to change things to designbetter vehicles."
Could you please tell us a little bit of background about yourselves and the company and how you all got into the subject?
Franck Perot:I started working on computational aeroacoustics almost 15 years ago and this was the topic of my Ph.D. I then worked at PSA Peugeot-Citroìn as an aeroacoustics research engineer doing testing in wind tunnels but mainly developing and introducing the usage of transient flow and noise simulations in the development. My focus was vehicle exterior wind noise and internal HVAC systems noise problems. Seven years ago, I joined Exa Corporation with the intention to develop and further deploy this type of aeroacoustics solution for the ground transportation industries.
Michelle Murray-Ross:Since the mid-90's Exa has offered a software solution based on Lattice-Boltzmann physics developed by our software development and physics teams. Our flagship product, PowerFLOW, was quickly accepted by the automotive industry as it provided an accurate, fast and economical alternative to traditional testing methods that required clay models and wind tunnels. Since then, we have expanded our simulation capabilities to include thermal and aeroacoustics simulations.
The amount of data that you’re able to process must be infinitely more than it was in the early years.
M.M-R.:Yes, it’s amazing how much things have changed and how expensive hardware was when PowerFLOW was first introduced. At that time, it was only organizations that could actually afford enormous refrigerator-sized servers that could run high end simulations, but that’s all changed in the past 15 years and now some simulations can be run on desktop systems or Linux clusters which are much more viable and affordable for smaller companies as well as for the larger ones with which we initially worked. We now offer on demand servers as well, so companies can access our full site of simulation products on a CPU hour basis.
F.P.:Ten years ago, when I started using PowerFLOW, I was fighting to get my simulations running on twenty processors and was struggling at getting ten gigabytes of disk space. These days, simulations run on large clusters with hundreds or even thousands of cores and we have access to terabytes of disk space. We can run much bigger models and much faster. It’s really impressive.
In your view, what are the biggest challenges faced by designers and aerodynamics engineers with specific regard to acoustics?
F.P.: In the past, engineers mainly focused on crash, full efficiency and thermal management. Acoustics was certainly not the biggest issue engineers focused on during the development. Over the past twenty years, acoustics comfort and noise quality have however taken a growing importance. What we observe is that NVH engineers were able to work on primary sources such as engine, transmission and vibrations and were able to significantly reduce their contributions. One consequence is that noise sources previously considered as secondary are now emerging. A lot of them related to aeroacoustics have become really important like aerodynamics noise, flow-induced HVAC systems, blowers and cooling fans. One challenge for the designers and engineers is that the design process has not necessarily evolved accordingly and the way vehicles are designed is not necessarily optimal to consider these new problems. It can then be quite challenging to detect possible failures and to treat them early enough.
Historically, engineers have focused on adding mass to improve insulation and absorption and about finding innovative NVH solutions to reduce noise. Practically you can only do this relatively late in development when you have the first prototypes available; something reliable you can work on. By then you have already spent a lot of time and money and you have a limited bandwidth to act on the vehicle to reduce the noise. You frequently have to put in place late counter-measures which are extensive, tedious and result in many sleepless nights to fix some urgent problems. One of the challenges here is to be able to integrate acoustics into the development cycle as early as possible to give enough freedom to the engineers and to truly have the opportunity to change things to design better vehicles.
The goal is to be able to make more major changes if needed and not be constricted to minor ones due to the late stage of development.
F.P.: Let’s assume aerodynamic noise is not considered early enough, maybe after aerodynamics performances are assessed. Then, by the time aeroacoustics engineers can act, they more or less have a frozen design to work with and consequently little freedom on the geometry. As an example, it might be possible to only have the freedom to modify the shape of the A-pillar by 1 mm and the location of the side mirror by 1-2 centimeter up or down. The shape of the moldings can also be optimized but, as you said, these are minor changes. If a big problem on your vehicle appears, requiring ideally a slight change of the angle of the windshield or a modification of the upper shape of the hood, it is likely too late. If this happens, you have to do with what you have and limit yourself to minor changes. Certainly a laminated side glass or materials with better acoustic performances can be used but this has a negative impact on mass and has higher costs. This is not a preferred solution.
Can you discuss some of the particular challenges that you’ve seen with vehicles that run in a quiet mode?
F.P.: It is related to one of my previous points. In quiet mode, you have for instance no engine, no transmission and no engine cooling fan. The consequence is that passengers are more likely to hear noise from the ventilation systems, from the blower, from the compressor, from the doors displacement and from the air running in the system. Passengers are more likely to be annoyed by electronics’ cooling fans and are more likely to perceive aerodynamics noise from the side mirror, the side glass and the underbody. The perceived quality is drastically different in this type of vehicle and it poses real problems and challenges during the development since the architecture is completely different.
How can engineering teams best take advantage of the solutions you have to offer?
F.P.:Finding solutions at various stages of the development using simulations and diagnosis capabilities is of course one important aspect. Our solutions provide guidance to engineers facing very complex noise problems that are not straight forward to understand experimentally. Most of the time, handling aeroacoustics experimentally is a trial-and-error process requiring prototyping and a specific testing facility. This take a lot of time, requires a lot of expertise and all the associated costs are high. Since our solutions are based on CAD input data, broader design spaces can be investigated automatically for optimization. Our solutions accelerate the development process and increase the chances to come-up with a better vehicle since problems can be raised and discussed very early.
A lot of people talk about a combined approach using simulation and wind tunnels. Do you believe that will always be the case? Or are we starting the transition already?
F.P.:The transition has started, that’s certain. We can define here the ratio between the numbers of testing performed around the world divided by the number of simulations. What is happening, and will happen even more with time, is that this ratio will get smaller and smaller. Looking for instance at crash, this ratio is already very small and OEMs don’t crash that many vehicles as compared to 10-20 years ago. We observe this trend at Exa with more and more of our customers. They extensively adopt our simulation process to predict and assess aerodynamics, thermal and acoustics performances. Prototyping comes later and later in development since all OEMs want to minimize their usage. As a practical example, flow simulations can now officially be used for regulations purposes in the US with the EPA. Truck companies don’t necessarily need to certify their vehicles in a wind tunnel; they can do it digitally.
That must represent a major cost-savings because finding wind tunnels that are specifically large enough to accommodate heavy trucks is a challenge.
F.P.:Absolutely. Even talking about cost-savings, finding a wind tunnel in which you can fit a heavy truck is already quite challenging! This is even more relevant for aeroacoustics since I know only one facility around the world able to fit-in a truck…
In general, using an aeroacoustics wind tunnel is very expensive in terms of construction, maintenance and operation. Putting everything together and including the added value, there is no question in my mind that switching to a more digital process is what has to be done. As I already mentioned, we already observe this trend and I believe this is only the beginning.
M.M-R.: Many customers now do early stage simulation and optimize designs exclusively through simulation solutions. Initially, when PowerFLOW was first introduced, our engineers used to spend time trying to help customers create custom templates of particular wind tunnels because they’re all different yet many automotive companies still required some wind tunnel testing at one point in the design cycle. Looking back on it now, it does seem very counter-productive to match a man-made version of reality. PowerFLOW does a better job of simulating real world conditions.
Each Wind Tunnel has certain unique attributes and biases I suppose?
M.M-R.: Yes, they all have different capabilities. For example, some have rolling road capabilities, some do not. They are all built with different requirements based on the company’s request and historically our teams have had to emulate some of them. Now, often companies are just foregoing physical testing because the costs are so prohibitive of using the wind tunnel and/or building a prototype to put in the wind tunnel when product optimization can all be done upfront, earlier in the process using simulation.
F.P.: There is actually also another aspect to this question. What is measured in a wind tunnel on a prototype does not fully represent the reality because there are always uncertainties on the geometry and multiple possible sources for errors. I will not claim that simulation is accurate 100% of the time because we are not sure either about the quality of the input geometry and because physical mechanisms might sometimes be on the fringe of what can be simulated. Considering however all the experience gained at working with many customers I feel confident that this percentage is close to 90-95%. To my knowledge, this percentage is higher compared to the accuracy wind tunnels can bring.
Would you be willing to take an educated guess about the number of years it will take for this type of simulation to catch up to the way it’s used in some of the other disciplines?
F.P.: It will depend on the OEM and on the strategy they define and choose. We see some OEMs having already fully adopted our technology and they use it in production for many applications. Some customers are still converging and are still in a transient phases. These customers still evaluate our solutions because they need to better understand how to get more value out of them and because they need to gain confidence with this relatively new technology. We also have late adopters and I think it will only be a matter of years or months before they can fully appreciate the value.
Exa aeroacoustic solution applications include: Greenhouse wind noise; cooling fan noise; pass-by/community noise; sunroof/window buffeting; component noise; gap/seal noise; underbody wind noise.
For more information please contact Exa Corporation at www.exa.com