Wind Tunnel Vs CFD Simulation

[mycadcae]

Which is the most accurate 99% of the result.

Actuarial Model Vs 3D Modeling?...please explain more details on this, for the experts in this.

Tq,

[andylaurence]

The latest Bernoulli magazine has a nice little primer on wind tunnels vs CFD vs track testing.

[Kiril Varbanov]

BTW, the Bernoulli magazine requires subscription and money.

There's probably no straightforward or correct answer to that question. Therefore, I can offer you four good readings:
1. The Role of Wind Tunnel Experiments in CFD Validation - http://onlinelibrary.wiley.com/doi/10.1 ... eae034/pdf
2. Computational Fluid Dynamics or Wind Tunnel Modeling? - http://www.envirometrics.com/abstracts/CFDvsWT.pdf
3. http://www.cd-adapco.com/sites/default/ ... ighton.pdf
4. http://www.symscape.com/blog/wind-tunnels-and-cfd

[marcush.]

None really . As always it´s causes for horses and depends on what you are looking for and recources available.
with Computer power available and practically every race car designed todays you are ina Position to test your design without touching any material apart from your Keyboard.
so CFD is as a logic Extension of CAd design and a cost effctive way to evaluate your designs .Only a few years such was not even remotely possible so Teams simply had no choice but had to Resort to scale models as there was no full size tunnel available where they could generate meaningful data.
Now there is huge knowledge /expertise of decades of scale windtunneltesting available challenged by cfd restricted by computing power available when you run a full vehicle .
Having both is of cause the best -as you can cross check results and trends it does pay off if you can demonstrate the soundness of aero concept s using cfd ,windtunnel and maybe even road coastdown/straightline testing .Omit one of them and chances are rapidly increasing you will fall flat on your face -but some effects are not easy to model in CFD and others a bitch to implement into your windtunnel- How do you run a 50% model with exhaust blowing in a tunnel? ,how do you replicate tyre Deformation ....it´s a simplification always and there is a big Chance to oversimplify your Experiment to make it even possible to handle -but then the effects you see in your modified Experiment may have no relevance to what actually happens on a road and you chased a Phantom ....life can be cruel.

[riff_raff]

CFD model results are only as accurate/reliable as the capabilities of the engineers/analysts that created it. Wind tunnel testing is mostly used to validate and refine computational models. Once a computational model has been validated, it is far more efficient to refine and optimize the design using the model.

One area that computational models are getting better at is coupled analysis, involving structures, aero and dynamics.

[DaveW]

One area that computational models are getting better at is coupled analysis, involving structures, aero and dynamics.

Interesting. Could you tell us more?

[marcush.]

you can with todays Tools simulate structure(FEM) and CFD in ONE simulation .this can be done in a multiphysics "Switched" mode...so every so and so time stamp the Simulation Switches to FEM takes a snapshot of the current load effects and Returns to CFD using the new geometry.
Obviously this is has potential to increase accuracy of the cfd sim.similar things are done with FEM and CFD simulations to predict temperature behaviour.needless to say if you look into a lot of variables your calculating demands go ballistic.

[Jersey Tom]

Wind tunnel testing is mostly used to validate and refine computational models.

Mostly?

[marcush.]

I dare to say a Validation can only live in the real world -where is the logic to call a correlation of cfd and windtunneltesting a validation?
a Validation is the proof of a certain methodology(simulation) to give results that correlate with the real product .Validating a cfd calculation with windtunnel testing is not proving anything of real world Performance..As we have seen again and again.

With limited track Access -Teams have failed to react to this in a proper way .a blotch of flowviz or funny aero Arrays is not really going to tell you much more than you already know from staring at your stopwatch if the times don´t come.
windtunnel ,CFD and tracktesting has to be conducted orchestrated and integrated -the addons simply do not look like they were designed into the actual car and that limits the data you can generate ,especially considering you are looking at Minute effects that could be easily drowned by your testing Equipments interaction with the car.

This is not saying component Validation in virtual and loboratory Environments has no place far from it .It is the base of all to properly understand what parts are capable of and characterise them .
Same goes for Systems Validation -but as Renault had to realise recently realistic Environments can be very hard to model
and so you Need to have a clear idea experimental and calculating wise where the outer rim of your Performance envelope lives- you have to perform sensivity tests for Change of conditions that is .
It can be very tricky to understand all Parameters influecing track Performance of your new development -as seen again and again when new developments do not translate into laptime gains.

Teams tend to reduce These to correlation issues as a matter of calibration of tunnels ,but the Problem most likely is running deeper than this( what are your development Goals and how do you measure improvement)

[DaveW]

you can with todays Tools simulate structure(FEM) and CFD in ONE simulation .this can be done in a multiphysics "Switched" mode...so every so and so time stamp the Simulation Switches to FEM takes a snapshot of the current load effects and Returns to CFD using the new geometry.

Has the strategy been used to simulate a porpoising event? If so, how reliable was the estimated critical airspeed?

...windtunnel ,CFD and tracktesting has to be conducted orchestrated and integrated -the addons simply do not look like they were designed into the actual car and that limits the data you can generate ,especially considering you are looking at Minute effects that could be easily drowned by your testing Equipments interaction with the car.

An example of Schrödinger's cat, I suppose.

[riff_raff]

Wind tunnel testing is mostly used to validate and refine computational models.

Mostly?

With modern CFD analysis tools, once you have a valid baseline model you can perform numerous design iterations quickly and cheaply, and have a high level of confidence in the accuracy of the qualitative results (ie. did the change make things better or worse). However, validating the baseline model usually still requires some physical test work, such as a wind tunnel test.

[mycadcae]

Wind tunnel testing is mostly used to validate and refine computational models.

Mostly?

With modern CFD analysis tools, once you have a valid baseline model you can perform numerous design iterations quickly and cheaply, and have a high level of confidence in the accuracy of the qualitative results (ie. did the change make things better or worse). However, validating the baseline model usually still requires some physical test work, such as a wind tunnel test.

quickly and cheaply,...agree..

[Lycoming]

If your primary purpose for having a wind tunnel is just to double check against your CFD, why do the teams spend so much time in them and cycle so many parts through them?

[autogyro]

If your primary purpose for having a wind tunnel is just to double check against your CFD, why do the teams spend so much time in them and cycle so many parts through them?

Mainly because they have little idea on how the component will work until it is actually on the car on track.
They need to further qualify their guess work.

[Greg Locock]

Your baseline calibration and correlation may be accurate, but testing combinations of parts in the WT gives you a much better idea of whether the CFD is working properly. Interactions are far more difficult to calculate than baselines.