CFD - Computational Fluid Dynamics, Motorsport, Formula 1

[Jersey Tom]

The simple fact is - there is no simple CFD. Just CFD with varying levels of assumptions where accuracy and assumptions are usually inversely proportional...

Similar could be said of vehicle dynamics simulations, tire models, etc etc.

Math models are just tools. Are there any perfect tools in anyone's toolbox? No. You use the right tool for the right job. Need to take some material off a desk drawer to make it fit better? If you can do it with a hand file - use a hand file, not a CNC machining center.

Just like giving someone a TIG machine to play with doesn't make them a welder, and doesn't make their welded joints any good... handing someone Fluent or ADAMS doesn't make them a CFD or simulation engineer, nor does it mean the results they get out of it are sensible and not garbage.

[Th0R]

The thing which makes cfd more complex than a lot of other simulations (in my opinion) is, that you can't run cfd simulations without lots of simplifications. Bascially you always run LES or RANS simulations. You always have mean values for the turbulence (In one model the mean over time in the other the mean over the space). Therefore, you need to model your trubulence and boundary layers according to that. The goal would be to calculate fine enough meshes that even the smallest turbulence scales can be discretised (google direct numerical simulation), but this won't happen in our lifetime. Computers are soooooooooooooooooo slow

So the only thing we can do is try to understand which implications our asumptions have on our simulation results. To do this not only the theoretical background is needed but also lots of experience. It is nice to know about different discretisation schemes, stabillity, order of convergence and the other numerical 'magic'. But in the end, it everything that matters is that you can interprete the results the right way.

[KeithYoung]

I've begun writing software/code to numerically analyze brake rotors. Eventually it will likely have internal and simple external flows. However, the main reason I'm posting this here is so you can see how many assumptions I made (alot) and what it takes to eliminate them. Following the posts will give you an understanding (and me, since I've never done anything much more complicated than this preview as an undergraduate) of what goes into getting good numerical results.

http://racingtech.wordpress.com/2012/03 ... s-preview/

Some of those assumptions will be easier than others to eliminate. For example, the thermal conductivity changes with temperature, so each iteration I would update the thermal conductivity according the a lookup table or least squares fit equation or some other method to increase accuracy at the cost of up front effort and computing power/time. Eliminating steady state however, will be a MUCH larger undertaking.




I forgot to list 2D as one of my assumptions in the article. I'm sure I'll think of more later.

So far its less than 100 lines of code

[Th0R]

You might also have forgotten incompressible. Otherwise solving this with under 100 lines of code would be quite good.

[KeithYoung]

Thanks, that is another one. Basically it will be modeled completely as an Ideal Gas. In fact it is, I just forgot to list that and probably other assumptions.

Appreciate the help.

[hardingfv32]

How do the CFD programs account for the various theories of lift generation? Those explanations based directly on Newton’s laws of motion and explanations based on Bernoulli’s principle.

Is it one or the other or are their values additive?

Brian

[hollus]

Both Newton's laws and Bernoulli's principle apply at the same time. There is only one set of physics laws, Bernoulli and Newton are more akin to two different languages to describe the same things happening to masses of air, although i suspect CFD code is written using something more closely resembling Newton's language.
It is a bit like polar coordinates and cartesian coordinates. They make the math look different but can be used to calculate the same things with the same outcome.
There are no different theories of lift, at least not properly supported ones. I can explain you how a plane flies using only Newton or using only Bernoulli. The newtonian explanation is longer and more convoluted in words, but it will be the same plane flying the same way wether the pilot believes on the conservation of momentum or in acceletated fluids.

[hardingfv32]

Well, that sure does not make the subject any easier to understand for me.

Brian

[hollus]

I guess another way to look at it is that CFD code is not trying to apply any theory of lift. It simply follows around small masses of fluid as they interact with each other and applies well known laws of physics to those finite elements.
How heat moves from one cell to the next is exactly understood.
How mass moves from one cell to the next is exactly understood.
How internal molecular energy is traded for external kinetic energy for the mass of air is exactly understood. And so on.
You apply all of the above to each bit of air, add Newton's laws, and follow the air around.

Lift (and drag) naturally follow, they are not trying to be explicitly described or explained by your CFD model, they just follow. The air just moves around doing what air does and in the end your object is subject to lift (or whatever). How you explain in words (using Newton's laws or using Bernoulli) how the air movements resulted in lift is up to you. I personally prefer Newton, it is more intuitive to me. But CFD models don't care what I find intuitive, and physics even less so.


Maybe these links help:

Stop abusing Bernoulli

...Although Bernoulli's law is sound and well proven, this popular explanation, world-wide, of wing operation is false.

...As indicated in the above figure, air approaching the plate, such as a sheet of plywood, accelerates into the reduced above-plate pressure with increasing velocity, while air approaching below is slowed in the increased pressure, in accordance with Bernoulli. Thus faster upper surface flow can be described as a result of pressure difference rather than the cause of it. Bernoulli's "principle, "law" or "effect" states that velocity varies in inverse relation to pressure but does not assign cause-and-effect relation. Unfortunately a great amount of confusion has been generated by abuse of Bernoulli's law in erroneous cause-and-effect explanations.

http://www.allstar.fiu.edu/aerojava/airflylvl3.htm


Note that the never say Bernoulli is wrong.
Also note haw many words and things pushing things that push other things it takes to explain it in Newtonian language.

[Pierce89]

I guess another way to look at it is that CFD code is not trying to apply any theory of lift. It simply follows around small masses of fluid as they interact with each other and applies well known laws of physics to those finite elements.
How heat moves from one cell to the next is exactly understood.
How mass moves from one cell to the next is exactly understood.
How internal molecular energy is traded for external kinetic energy for the mass of air is exactly understood. And so on.
You apply all of the above to each bit of air, add Newton's laws, and follow the air around.

Lift (and drag) naturally follow, they are not trying to be explicitly described or explained by your CFD model, they just follow. The air just moves around doing what air does and in the end your object is subject to lift (or whatever). How you explain in words (using Newton's laws or using Bernoulli) how the air movements resulted in lift is up to you. I personally prefer Newton, it is more intuitive to me. But CFD models don't care what I find intuitive, and physics even less so.

In less words, CFD works just like FEA, only it's applied to fluids.

[Th0R]

In less words, CFD works just like FEA, only it's applied to fluids.

That's not exactly true. In CFD mostly finite volume methods are used while FEA mostly uses finite element methods.

[n_anirudh]

Spectral element techniques used in DNS had their origins in FEA.

These methods use polynomial interpolation between elements and have exponential convergence properties..

[Th0R]

As far as i know there are no large scale DNS simulations, because they need too much memory. Please correct me if i'm wrong.

[Jersey Tom]

In less words, CFD works just like FEA, only it's applied to fluids.

That's not exactly true. In CFD mostly finite volume methods are used while FEA mostly uses finite element methods.

Well FEA is by definition.. the finite element way of doing things. And yes, some CFD packages do use FVM rather than FEM. In either event, to say that CFD works like FEA is not particularly accurate since you could do CFD by FE.

For the layman all of this is overboard level of detail. Let's say that generally speaking, finite methods are conceptually similar be they for structural analysis or fluid flow in that you are taking a complex problem (due to conditions, geometry, whatever) and breaking it down into many small pieces rather than some all-encompassing analytic solution like one would do to find the equation for the deformation of a cantilevered beam. Can take that general approach and choose to apply it a variety of engineering problems, be it structural or fluid flow.

[Jersey Tom]

In less words, CFD works just like FEA, only it's applied to fluids.

That's not exactly true. In CFD mostly finite volume methods are used while FEA mostly uses finite element methods.

Well FEA is by definition.. the finite element way of doing things. And yes, some CFD packages do use FVM rather than FEM. In either event, to say that CFD works like FEA is not particularly accurate since you could do CFD by FE.

For the layman all of this is overboard level of detail. Let's say that generally speaking, finite methods are conceptually similar be they for structural analysis or fluid flow in that you are taking a complex problem (due to conditions, geometry, whatever) and breaking it down into many small pieces rather than some all-encompassing analytic solution like one would do to find the equation for the deformation of a cantilevered beam. Can take that general approach and choose to apply it a variety of engineering problems, be it structural or fluid flow.