| Topic: |
Science > Physics |
| User: |
"" |
| Date: |
28 Nov 2007 01:52:53 AM |
| Object: |
Gas Simulation Question |
Hi, a while ago I wrote a program that simulates a gas. It could
simulate 100,000 particles real easily without trouble. The problem
is, the volume it simulated was very small, on the order of a cube 5
micro-meters in length (5e-6 m).
Now my question is, if I simulate a surface moving in this gas
simulator (it did not support surfaces except the cylinder container
which particles recoiled off of), and the surface is a scaled-down
space craft, will drag be accurate?
I heard somewhere that drag at low speeds is proportional to velocity
for small objects, or velocity squared for large objects. Is this
right?
It seems a bit odd, why don't the laws of physics work the same no
matter what the scale is?
So how about hypersonic flight, e.g. simulating reentry (I will be
ignoring disassociation of molecules)? If I have a 3 micro-meter in
length scramjet and it works in my simulation, will it really work in
reality, scaled up to full-scale?
I heard somewhere that at hypersonic flight drag can be proportional
to higher powers of velocity than velocity squared. Can anyone here
corroborate this?
Is there some way to take a simulation that takes place in micrometer
scales, and "fix it up" so the data is still good at larger scales? I
thought maybe you could simulate a scramjet at 1 micrometer in length
and another 1.5 micrometers in length etc. to try to learn the scaling
laws like they did for nuclear bomb propulsion in Project Orion, but
I'm really not sure about the drag problem.
Lastly, does anyone here know if there's a way to simulate turbulence,
shockwaves, and so on without simulating each and every particle? It
doesn't appear to be as easy as integrating partial differential
equations, which can be Monte Carlo simulated.
My latest effort is to design a special-purpose high performance
computer for simulating gas particles. The computer would consist of
5000 cells x 5000 cells x 5000 cells. Each cell would be 2 micro-
meters in physical length (and would simulate 1 nm in virtual length).
Is it possible to fabricate a cube-sized computer as small as 2
micrometers in length, complete with RAM, CPU, floating point unit,
etc. and to make 5000^3 of them and hook them all up?
The computer is straight-forward enough to design and simulate a crude
version on a PC, where unanswered questions about its design could be
answered, but would it be useful for extrapolating how a full-scale
space plane would behave? I mean is pressure difference on a surface
going to be accurate? Can I just multiply to correct for its small
size? Or will drag really be proportional to velocity not velocity
squared at slow speeds?
Is there some way to fix it up???
Does anyone know how to simulate a full-scale scramjet?
Surely there is -some- use for small-scale simulations.
My space plane would have two configurations: one where it takes off
horizontally and flies to a suborbital trajectory and releases a
Command and Service Module (CSM) -- a second stage -- which boosts to
orbit. The CSM is throw-away. LH2-LOX is used, and LH2 for scramjet
propulsion. The other configuration, uses CH4-LOX in an external tank,
then throws away the external tank (which is not recovered). This
happens at high altitude. As it dives again it speeds up, reaching
Mach 7, where scramjets are activated and it rises again until Mach 14
is achieved, then switches to LH2-LOX propulsion. The main engine
would be capable of running on LH2 as well as CH4.
Any thoughts?
.
|
|
| User: "John Park" |
|
| Title: Re: Gas Simulation Question |
28 Nov 2007 03:46:57 AM |
|
|
(wrschlanger@gmail.com) writes:
Hi, a while ago I wrote a program that simulates a gas. It could
simulate 100,000 particles real easily without trouble. The problem
is, the volume it simulated was very small, on the order of a cube 5
micro-meters in length (5e-6 m).
Now my question is, if I simulate a surface moving in this gas
simulator (it did not support surfaces except the cylinder container
which particles recoiled off of), and the surface is a scaled-down
space craft, will drag be accurate?
I heard somewhere that drag at low speeds is proportional to velocity
for small objects, or velocity squared for large objects. Is this
right?
It seems a bit odd, why don't the laws of physics work the same no
matter what the scale is?
[...]
Any thoughts?
Yes (from a non-expert; others may have better ones).
You don't say what your boundary conditions are, but it sounds as
though you've put the gas in a smooth-walled box. Such a box could sustain
standing waves at micrometre wavelengths but not at larger wavelengths.
So:
How many real surfaces are smooth on scales of less than one micrometre?
(This is relevant for your aerodynamic models.)
What is the frequency of sound with a wavelength of a few micrometres? How
does that compare to the frequencies you might expect to deal with?
--John Park
.
|
|
|
|
| User: "Andy Resnick" |
|
| Title: Re: Gas Simulation Question |
28 Nov 2007 07:42:31 AM |
|
|
wrote:
<snip>
I heard somewhere that drag at low speeds is proportional to velocity
for small objects, or velocity squared for large objects. Is this
right?
It seems a bit odd, why don't the laws of physics work the same no
matter what the scale is?
The drag is a function of Reynolds number and shape. There is no
analytic solution to the general problem, but it does reduce to an
analytical solution in the limit of low Reynolds number for a sphere,
cylinder, and IIRC a range of airfoil shapes. An smooth solution is
also obtained in the high Reynolds number limit, but is based on
experimental data rather than an analytic solution.
<snip>
Any thoughts?
Yes- why are you simulating fluid flow by atomic-scale particle tracking?
--
Andrew Resnick, Ph.D.
Department of Physiology and Biophysics
Case Western Reserve University
.
|
|
|
|
| User: "Bruce Scott TOK" |
|
| Title: Re: Gas Simulation Question |
28 Nov 2007 05:13:44 AM |
|
|
Now my question is, if I simulate a surface moving in this gas
simulator (it did not support surfaces except the cylinder container
which particles recoiled off of), and the surface is a scaled-down
space craft, will drag be accurate?
Spacecraft drag is influenced by charging effects. I remember seeing
papers on this in the 1980s. They did particle in cell computations but
the effects of the electromagnetic field are central to the problem.
One of the authors was Phil Pritchett IIRC.
--
ciao,
Bruce
drift wave turbulence: http://www.rzg.mpg.de/~bds/
.
|
|
|
|
| User: "Sue..." |
|
| Title: Re: Gas Simulation Question |
28 Nov 2007 03:01:04 AM |
|
|
On Nov 28, 2:52 am, wrote:
Any thoughts?
There are indeed long-range induction forces that assert their
presence at about one micron, most noteably as the boiling
point of water. You may already be familiar with some of
the dedicated hardware for molecular dynamics:
http://atlas.riken.go.jp/mdm/research.html
http://www.research.ibm.com/grape/grape_ewald.htm
http://atlas.riken.go.jp/mdm/
Also those with acess to such machines seem to be quite
prolific writers so some work may be close enough to your
application that you can just glue the solution in where
it fits:
http://atlas.riken.go.jp/mdm/research.html
Sue...
.
|
|
|
|
|
Related Articles |
|
|
