Prev: .M file to .Mdl conversion
Next: adult
From: Supriyo Ghosh on 26 May 2005 17:03
Hi,
I am trying to write a code in Matlab (for my PhD research) that
would simulate random packing of spherical grains obeying a user
defined specific Gaussian distribution, where one spherical particle
falls vertically each time onto the surface of an existing particle
cluster, which grows upwards.
Has anybody tried anything like this before (preferably in Matlab
environment)? Any kind of help in this regard will be greatly
appreciated.
Thanks!
-Supriyo
From: Ben Barrowes on 26 May 2005 17:20
I am not exactly sure what you mean. It sounds like the surface is
binned (with bin widths = the diameter of the sphere?) and each sphere
has to fall within one of the bins. Is that right? You simply want to
count how many spheres have fallen into that each bin? Or do they fall
onto each ohter? Do they slip? Do they stick to each other?



Supriyo Ghosh wrote:
> Hi,
> I am trying to write a code in Matlab (for my PhD research) that
> would simulate random packing of spherical grains obeying a user
> defined specific Gaussian distribution, where one spherical particle
> falls vertically each time onto the surface of an existing particle
> cluster, which grows upwards.
> Has anybody tried anything like this before (preferably in Matlab
> environment)? Any kind of help in this regard will be greatly
> appreciated.
> Thanks!
> -Supriyo
From: Supriyo Ghosh on 26 May 2005 17:36
Hi,
I appreciate your reply !
Suppose I have a hollow cylinder with flat closed bottom, and I
specify its dimensions(which is many order of magnitude higher than
the mean spherical particle dia.). Now I am dropping one sphere at a
time inside this cylinder. The dia. of the falling sphere has to obey
a specific Guassian distribution. As the spheres are dropped, each
one has ultimately to take the position that minimizes its potential
energy.The position has to be chosen from among all the positions
available, based on the locations of the spheres those are dropped
prior to the current sphere. The process has to go on until the
sphere pack reaches the brim of the cylinder.
This is what I am trying to simulate.
Any help from you in this regard will be greatly appreciated!
Thanks,
-Supriyo

Ben Barrowes wrote:
>
>
> I am not exactly sure what you mean. It sounds like the surface is
> binned (with bin widths = the diameter of the sphere?) and each
> sphere
> has to fall within one of the bins. Is that right? You simply want
> to
> count how many spheres have fallen into that each bin? Or do they
> fall
> onto each ohter? Do they slip? Do they stick to each other?
>
>
>
> Supriyo Ghosh wrote:
>> Hi,
>> I am trying to write a code in Matlab (for my PhD research)
that
>> would simulate random packing of spherical grains obeying a
user
>> defined specific Gaussian distribution, where one spherical
> particle
>> falls vertically each time onto the surface of an existing
> particle
>> cluster, which grows upwards.
>> Has anybody tried anything like this before (preferably in
Matlab
>> environment)? Any kind of help in this regard will be greatly
>> appreciated.
>> Thanks!
>> -Supriyo
>
From: Roger Stafford on 26 May 2005 18:20
In article <ef07394.1(a)webx.raydaftYaTP>, "Supriyo Ghosh" <sghosh(a)ncsu.edu>
wrote:

> Hi,
> I appreciate your reply !
> Suppose I have a hollow cylinder with flat closed bottom, and I
> specify its dimensions(which is many order of magnitude higher than
> the mean spherical particle dia.). Now I am dropping one sphere at a
> time inside this cylinder. The dia. of the falling sphere has to obey
> a specific Guassian distribution. As the spheres are dropped, each
> one has ultimately to take the position that minimizes its potential
> energy.The position has to be chosen from among all the positions
> available, based on the locations of the spheres those are dropped
> prior to the current sphere. The process has to go on until the
> sphere pack reaches the brim of the cylinder.
> This is what I am trying to simulate.
> Any help from you in this regard will be greatly appreciated!
> Thanks,
> -Supriyo
--------
Hello Supriyo,

You have posed a very difficult computational task for a computer
algorithm to perform. I gather from your description that smaller spheres
are expected to drop down through spaces between larger spheres provided
there is a proper route for them to follow. That involves more than just
searching for lowest (minimum potential energy) positions where they might
fit. The task of finding possible routes is, I think, the most difficult
part of that task - analogous to spelunkers finding their way through
caves with extremely intricate tunnels. Also the problem arises of
whether such a route is dynamically possible, or whether a sphere might
reach a point with a local minimum in energy and decide to park there.

I seriously doubt if you will find very much help on a project of such
magnitude in this newsgroup. Most of the people giving assistance prefer
questions involving comparatively short answers. If the algorithm you
develop contains specific problems which are brief and can be expressed in
more or less abstract terms, help might be available here for these.

By the way, a Gaussian distribution involves negative as well as
positive quantities. You need to make whatever distribution is used be
compatible with spheres whose radii must all be positive.

(Remove "xyzzy" and ".invalid" to send me email.)
Roger Stafford
 | 
Pages: 1
Prev: .M file to .Mdl conversion
Next: adult