quote:

---

Original post by Xgkkp
You could have an algorithm that calculates a faces angle relative to a plane in front of the object, the more parallel, the more it adds to a resistance factor. then do this for each polygon. ( ignore back facing polygons).

Presumably a resistance would be calculated before run-time.

modification: this could allow for material resistances, but there are limitations. Fruny is right, the only way to truly know accurate values for air resistance is by sticking it in a wind chamber.

Edited by - Xgkkp on February 8, 2002 7:19:52 PM

---

Actually, that''s a fairly decent quick-and-dirty calculation! The "Newtonian" method of calculating the drag on an object flying at hypersonic speeds is almost exactly this, and its used for initial concept calculations in the engineering world! See this page:

http://www.aerospaceweb.org/design/waverider/theory.shtml

But it isn''t really a great estimate for objects not flying at hypersonic speeds (> 5 times speed of sound). It doesn''t really work if you have a streamlined object, though, since there will be pressures on the back faces as well that partially cancel the pressures on the front faces, .


Graham Rhodes
Senior Scientist
Applied Research Associates, Inc.

Here is a page that has a good description of the various sources of drag:

http://aerodyn.org/Drag/

Graham Rhodes
Senior Scientist
Applied Research Associates, Inc.

Anyone knows anything about something called "blade element theory"?.. Apparently the x-plane flight sim is using that. A breif description is on the homepage:

http://www.x-plane.com/descrip.html

Not sure though if thats is used to calculate the drag/lift forces on the fuselage, wich is made up from a mesh.

quote:

---

Actually, that''s a fairly decent quick-and-dirty calculation! The "Newtonian" method of calculating the drag on an object....

---

Thanks!

quote:

---

Original post by AndersO
Anyone knows anything about something called "blade element theory"?.. Apparently the x-plane flight sim is using that. A breif description is on the homepage:

http://www.x-plane.com/descrip.html

Not sure though if thats is used to calculate the drag/lift forces on the fuselage, wich is made up from a mesh.

---

Its similar to the method Xgkkp proposed, but does not ignore backfacing polygons. Essentially, you calculate a lift and drag force individually on each polygon (usually X-Plane does this for a simpler LOD mesh---the aerodynamics mesh), ignoring the effects (or "interference effects" as they are called in the aerospace engineering world) of all the other polygons or bodies. It is a good quick-and-dirty estimate, but there are better quick-and-dirty estimates as well (so-called "doublet-source panel" methods for example are better, and also run quickly). For example, the interference effects are not negligible and so the lift and drag calculated for each polygon are wrong enough to be statistically significant. Of course, in the end this method does produce results that are realistic enough for a flight simulator such as X-Plane. The results may end up being technically wrong by 20%, but that isn''t enough to make a massively huge difference. It looks good, feels right, and is quick enough. And that makes it perfect for a game.

Blade element theory is basically the easiest of the fast CFD methods for airplanes. It does NOT work well for objects that are not streamlined like an airplane. (Actually, Xgkkp''s method would generate more realistic drag than blade-element theory for bluff bodies such as balls and teapots.)


Graham Rhodes
Senior Scientist
Applied Research Associates, Inc.

Oh about the person asking what method i use, it''s a lattice boltzmann method, maybe my phrasing wasn''t entirely correct, calculating a single timestep takes 2-3 seconds but to get usefull results (ie when doing advection diffusion in a non stationary flow) i need a lot more timesteps than just 1. I guess you could get a good flow pattern in just a few iterations taking about a minute . But hey i''m a student, not a master so forgive me my mistakes .
And is it correct to assume that drag is directly related to momentum loss in the system? Since that''s what i thought one should use to calculate drag.

quote:

---

Original post by Anonymous Poster
Oh about the person asking what method i use, it''s a lattice boltzmann method, maybe my phrasing wasn''t entirely correct, calculating a single timestep takes 2-3 seconds but to get usefull results (ie when doing advection diffusion in a non stationary flow) i need a lot more timesteps than just 1. I guess you could get a good flow pattern in just a few iterations taking about a minute . But hey i''m a student, not a master so forgive me my mistakes .

---

Cool! I''m not familiar with the lattice boltzmann method, but the fact that it includes the word "Boltzmann" makes me think its a bit more than just Navier-Stokes. Doing a quick web search, I see that it seems to be a way of modeling flows through porous materials---now I''m not surprised that you don''t use a traditional pure-fluids method! And I''m not surprised that it takes longer to solve! Some of the very fast CFD methods (panel methods and the aforementioned blade element theory---along with lifting line theory and others) can take a fraction of a second per time step, and this can be sufficient for games, .

quote:

---

Original post by Anonymous Poster
And is it correct to assume that drag is directly related to momentum loss in the system? Since that''s what i thought one should use to calculate drag.

---

Absolutely! If you have far field flow velocities, densities, and pressures, you can calculate drag by integrating momentum in the Trefftz plane behind the body to get the momentum loss in the fluid that passes by the body. And that maps naturally into the drag force. Actually, this method gives more accurate drag calculations than simply integrating surface pressures and skin friction directly. There are some simpler methods that just do not produce far field data and so it isn''t straightforward to calculate drag by the momentum loss method...

Graham Rhodes
Senior Scientist
Applied Research Associates, Inc.

thanks for the interest, actually i model flow around corals . And lattice boltzmann methods are mostly used in cases where you have to deal with very irregular geometry (since you just give each node a tag solid/fluid and the algorythm does the rest) and want to run it fast (parallelisation of the algorythm is almost trivial due to the locality of your explicit update scheme). Also it''s a very natural way of describing fluids unlike "traditional" CFD where you start with linearizing systems of differential equations and using very complicated math untill you finally end up with some horrible expressions that should represent the same system you started with .
Oh and it might have some applications in games, since coding an LB system for a single processor is rather easy the person wanting to calculate drag could maybe try a very small lbe grid (say 16^3) and use very few iterations to get a result which might nowhere be correct, but at least has the right order of magnitude .

btw i was too lazy to sign up, the anonymous post was me..