[Anna Tian]

Hi,

I currently read articles telling that CFD still has very bad accuracy for vehicle aerodynamics design. So the vehicle aerodynamics design is still driven by experience & experiment. In the near future (next 10 years), the accuracy of CFD on that won't be improved a lot.

I'm wondering what are the origins of the inaccuracy? Is that because vehicle always has bluff rear part so that the BL separation can not be avoided. And if there's BL separation at the rear part, the pressure distribution at the rear part can't be predicted nicely so the drag prediction will be very bad?

I'm also wondering what is the role of CFD in vehicle aerodynamics design in the car industry nowadays? What do you think about what its role will be in the near future?

[Anna Tian]

Originally Posted by Anna Tian

Hi,

I currently read articles telling that CFD still has very bad accuracy for vehicle aerodynamics design. So the vehicle aerodynamics design is still driven by experience & experiment. In the near future (next 10 years), the accuracy of CFD on that won't be improved a lot.

I'm wondering what are the origins of the inaccuracy? Is that because vehicle always has bluff rear part so that the BL separation can not be avoided. And if there's BL separation at the rear part, the pressure distribution at the rear part can't be predicted nicely so the drag prediction will be very bad?

I'm also wondering what is the role of CFD in vehicle aerodynamics design in the car industry nowadays? What do you think about what its role will be in the near future?

Could anyone answer this question?

[som87]

1.The "funding" needed for a "highly accurate" cfd analysis will raise the production cost resulting in raise in the sell off prize of the product. Consider a space vehicle or defence-aircraft project..in that case the aim is not commertial success, hence more effort>high accuracy whereas for our normal industry the "economic" (not computational) aspect will always be a burden for good quality research.


2. On future of CFD's application on vehicle aerodynamic design, from my view there'll be little effort on "improving" the accuracy. Instead topics like Automatic Differentiation (eg tapenade) will have more funding.

[Anna Tian]

Originally Posted by som87

1.The "funding" needed for a "highly accurate" cfd analysis will raise the production cost resulting in raise in the sell off prize of the product. Consider a space vehicle or defence-aircraft project..in that case the aim is not commertial success, hence more effort>high accuracy whereas for our normal industry the "economic" (not computational) aspect will always be a burden for good quality research.


2. On future of CFD's application on vehicle aerodynamic design, from my view there'll be little effort on "improving" the accuracy. Instead topics like Automatic Differentiation (eg tapenade) will have more funding.

What is the current role of CFD in vehicle aerodynamics design? Is it still only used for very rough estimation (I heard about that drag coefficient is still not easy to be predicted by CFD) or it is already widely used for geometry optimization like the CFD for the turbomachinery design industry?

[som87]

Originally Posted by Anna Tian

What is the current role of CFD in vehicle aerodynamics design? Is it still only used for very rough estimation (I heard about that drag coefficient is still not easy to be predicted by CFD) or it is already widely used for geometry optimization like the CFD for the turbomachinery design industry?

Vehicle industry in more a CAE dominated industry and honestly, the buyers wont have any headache on the accurate location of flow seperation over a car (say). I would not go that far to say "very rough", but yes, the "required" accuracy is low for the commertial use (unless we are talking about batmobile).

On other note, turbomachinary has a much wider place and there are still enough openings to do quality research.So, the accuracy level of cfd analysis in turbomachinary is often high enough to ensure a quality work.
(signing off for now)

[Anna Tian]

Originally Posted by som87

Vehicle industry in more a CAE dominated industry and honestly, the buyers wont have any headache on the accurate location of flow seperation over a car (say). I would not go that far to say "very rough", but yes, the "required" accuracy is low for the commertial use (unless we are talking about batmobile).

On other note, turbomachinary has a much wider place and there are still enough openings to do quality research.So, the accuracy level of cfd analysis in turbomachinary is often high enough to ensure a quality work.
(signing off for now)

How does CFD corporate with experiment in vehicle external aerodynamics design? In some industries, CFD is only used for very rough estimation and the design mainly rely on the experimental measurement. In some other industries, experiment is only used to verify the CFD and the aerodynamics design is driven by CFD. It looks like the former case become fewer and fewer and but later case is more and more. We see a clear shifting trend. But what is the current status of the vehicle aerodynamics design industry?

[Totalsim]

CFD is already being extensively used in industry and the accuracy can easily match that of the wind tunnel.

The hardest thing is getting the geometry and testing conditions correct. Even modelling things such as tyre squash can have an influence on the prediction.

Race cars quite often go straight from CFD to the track with good correlation.

[som87]

Originally Posted by Totalsim

CFD is already being extensively used in industry and the accuracy can easily match that of the wind tunnel.

The hardest thing is getting the geometry and testing conditions correct. Even modelling things such as tyre squash can have an influence on the prediction.

Race cars quite often go straight from CFD to the track with good correlation.

This is exactly what I wanted to say.
The need of accuracy dominates the quality of work.
For top notch racing cars (after all, they are our batmobile) the accuracy criterea should be of highest standard.
But apart from that, if we focus on the mre generalized part of vehicle industry, its more CAE than CFD.

[t.teschner]

I am actually surprised to hear that. My experience is more that, especially in the last ten years, CFD has become a major role in many areas and even in vehicle aerodynamics. i agree that bluff body aerodynamics is quite challaging, especially if you are interested in drag prediction. although if you look at drag and friction prediction on say, the ahmed body, current RANS model can predict the value of the drag coefficient quite well and also reconstruct the flow field in the wake to a satisfactory degree. (well ok, satisfactory degree is probably a matter of definition but you get as close as 5% if I remember from my simulations.

I was talking about exactly that (accuracy of CFD simulation) with the head of CFD from jaguar landrover, UK, and he told me that CFD is widely used in his company and that they use a lattice boltzman based approach because it presumably gives you better results in terms of accuracy then a navier stokes based solver. and that is not only him, the rest of the industry is using the LBM for their simulation.

furthermore, you have to consider that computational resources are becoming more and more available so we are currently at the edge where industry is starting to use large eddy simulations. this supposedly should also increase your accuracy.

one last point is, although it sounds obvious, but it all comes down to who is doing the simulation. from meshing to solver settings to post processing, there is a lot that can be done wrong.

[t.teschner]

Originally Posted by som87

For top notch racing cars (after all, they are our batmobile) the accuracy criterea should be of highest standard.

That's what I've thought as well ... but after having to learn how CFD is incorporated into formula 1, I was quite surprised to see how big the tolerances were (however, as long as they are able to corelate their data to track and wind tunnel test, it seems to give them still good results, if however more in an engineering sense

[som87]

Originally Posted by t.teschner

it presumably gives you better results in terms of accuracy then a navier stokes based solver.

I am sooo tempted to start an arguement XD (jk)

[Anna Tian]

Originally Posted by som87

1.The "funding" needed for a "highly accurate" cfd analysis will raise the production cost resulting in raise in the sell off prize of the product. Consider a space vehicle or defence-aircraft project..in that case the aim is not commertial success, hence more effort>high accuracy whereas for our normal industry the "economic" (not computational) aspect will always be a burden for good quality research.


2. On future of CFD's application on vehicle aerodynamic design, from my view there'll be little effort on "improving" the accuracy. Instead topics like Automatic Differentiation (eg tapenade) will have more funding.

What do you mean by 'Automatic differentiation'? Is that kind of automatic grids generation?