Multi-Sized Sphere Packing
Shuji Yamada1 Jinko Kanno2 Miki Miyauchi3
1 Department
of Computer Science Kyoto Sangyo University, Japan
2 Mathematics
and Statistics Program Louisiana Tech University, U.S.A.
3 Innovative Communication Lab NTT Communication Sciencce Labs, Japan
07/04/2009
Yamada, Kanno, and Miyauchi
Multi-Sized Sphere Packing
Motivation
Corrosion in bridge structures is a global problem.
Yamada, Kanno, and Miyauchi
Multi-Sized Sphere Packing
Motivation
Corrosion in bridge structures is a global problem. Filling the pores in concrete with solid materials or nanoparticles improves the strength signicantly.
Yamada, Kanno, and Miyauchi
Multi-Sized Sphere Packing
Motivation
Corrosion in bridge structures is a global problem. Filling the pores in concrete with solid materials or nanoparticles improves the strength signicantly. Which sizes of nanoparticles are most efcient or provide a highest density?
Yamada, Kanno, and Miyauchi
Multi-Sized Sphere Packing
Case 1: Mono-Sized Sphere Regular Packing.
For simplicity, we model nanoparticles as spheres.
Yamada, Kanno, and Miyauchi
Multi-Sized Sphere Packing
Case 1: Mono-Sized Sphere Regular Packing.
For simplicity, we model nanoparticles as spheres. Keplers Conjecture (1611), Proved by Hales (1998): The highest packing density of mono-sized spheres in the 3-dim space is 0.74048 18 .
Yamada, Kanno, and Miyauchi
Multi-Sized Sphere Packing
Case 1: Mono-Sized Sphere Regular Packing.
For simplicity, we model nanoparticles as spheres. Keplers Conjecture (1611), Proved by Hales (1998): The highest packing density of mono-sized spheres in the 3-dim space is 0.74048 18 . For example, the Face-Centered Cubic Crystal Structure in the Figure provides the density.
Yamada, Kanno, and Miyauchi Multi-Sized Sphere Packing
Case 2: Multi-Sized Sphere Regular Packing.
If we add smaller spheres to pack the vacancy of the previous packing, we can increase the density further.
If we increase the ratio of the two sizes of spheres, then we can reach to the following density. 1 1 18
2
0.93265
Yamada, Kanno, and Miyauchi
Multi-Sized Sphere Packing
Random Packings
In application, nanoparticles are randomly packed and the density will be lower than the regular packing. In literature, there are two distinct random packings when we pack objects in a container. Random Close PackingShake the container after a random packing. Random Loose PackingNo shaking is given to the container.
Yamada, Kanno, and Miyauchi
Multi-Sized Sphere Packing
Random Close Packing
In 1969, the approximate density 0.64 of random close packing of mono-sized spheres was obtained from the experiment packing steel spheres to a container and shaking the container.
Yamada, Kanno, and Miyauchi
Multi-Sized Sphere Packing
Random Loose Packing
There are different denitions for the random loose packing according to how you pack spheres randomly.
Yamada, Kanno, and Miyauchi
Multi-Sized Sphere Packing
Random Loose Packing
There are different denitions for the random loose packing according to how you pack spheres randomly.
Our Packing Method We place the center of a new sphere at the point (called pit) whose z-coordinate is the lowest among the points where new spheres touch old three spheres.
Yamada, Kanno, and Miyauchi
Multi-Sized Sphere Packing
Experiments for Random Loose Packing
We pack spheres of radius 0.02 to a 2 2 2 container C having a random shaped base.
Yamada, Kanno, and Miyauchi
Multi-Sized Sphere Packing
Experiments for Random Loose Packing
We pack spheres of radius 0.02 to a 2 2 2 container C having a random shaped base.
Yamada, Kanno, and Miyauchi
Multi-Sized Sphere Packing
Experiments for Random Loose Packing
We pack spheres of radius 0.02 to a 2 2 2 container C having a random shaped base.
The number of packed spheres: 126856 The average of the packing densities about the center of C: 0.604 The inuence from the boundary of C: Estimated about 30% of the radius of the spheres
Yamada, Kanno, and Miyauchi
Multi-Sized Sphere Packing
The Inuence from the boundary of C
The density distribution around C.
The horizontal scale indicates the distance from C (unit=the radius of sphere)
Yamada, Kanno, and Miyauchi
Multi-Sized Sphere Packing
The Inuence from the boundary of C
The density distribution around C. By assuming the average density to be 0.604,
The horizontal scale indicates the distance from C (unit=the radius of sphere)
Yamada, Kanno, and Miyauchi
Multi-Sized Sphere Packing
The Inuence from the boundary of C
The density distribution around C. By assuming the average density to be 0.604, we can calculate how much we should consider as a loss around C the distance range zero to 30% of the radius from C can be subtracted.
The horizontal scale indicates the distance from C (unit=the radius of sphere)
Yamada, Kanno, and Miyauchi
Multi-Sized Sphere Packing
Boundary Density
From now on, V is the volume of a container C and B is the surface area of C. We call the following the boundary density of C. (C) = B V
Yamada, Kanno, and Miyauchi
Multi-Sized Sphere Packing
Boundary Density
From now on, V is the volume of a container C and B is the surface area of C. We call the following the boundary density of C. (C) = B V
If C is a cube with the length of sides d , then 6 (Cube) = . d Therefore, can be any positive real number.
Yamada, Kanno, and Miyauchi
Multi-Sized Sphere Packing
Boundary Density
From now on, V is the volume of a container C and B is the surface area of C. We call the following the boundary density of C. (C) = B V
If C is a shpere with radius r , then (Sphere) = 3 . r
Yamada, Kanno, and Miyauchi
Multi-Sized Sphere Packing
Boundary Density
From now on, V is the volume of a container C and B is the surface area of C. We call the following the boundary density of C. (C) = B V
If C is a shpere with radius r , then (Sphere) = 3 . r
This is the important parameter when we approximate the packing density into a container C.
Yamada, Kanno, and Miyauchi Multi-Sized Sphere Packing
Approximate Density for Mono-Sized Sphere Packing
Suppose we pack a container C with identical spheres of radius r.
Yamada, Kanno, and Miyauchi
Multi-Sized Sphere Packing
Approximate Density for Mono-Sized Sphere Packing
Suppose we pack a container C with identical spheres of radius r. From the experiment, we can approximate the inuence of the C as a loss of 0.3r neighborhood or 0.3Br volume reduction and else use the average density 0.604.
Yamada, Kanno, and Miyauchi
Multi-Sized Sphere Packing
Approximate Density for Mono-Sized Sphere Packing
Suppose we pack a container C with identical spheres of radius r. From the experiment, we can approximate the inuence of the C as a loss of 0.3r neighborhood or 0.3Br volume reduction and else use the average density 0.604. Therefore, the total density after random loose packing of spheres with radius r can be approximated by 0.604(V 0.3Br ) = 0.604 0.1812r . V
Yamada, Kanno, and Miyauchi
Multi-Sized Sphere Packing
Two-Sized Sphere Random Loose Packing
We pack a container C with radii r1 > r2 two sized spheres.
Yamada, Kanno, and Miyauchi
Multi-Sized Sphere Packing
Two-Sized Sphere Random Loose Packing
We pack a container C with radii r1 > r2 two sized spheres. First, we pack C with large spheres having radius r1 . The total volume of large spheres: (0.604 0.1812r1 )V The total surface area of large spheres: 3 (0.604 0.1812r1)V r1
Yamada, Kanno, and Miyauchi
Multi-Sized Sphere Packing
Two-Sized Sphere Random Loose Packing
We pack a container C with radii r1 > r2 two sized spheres. First, we pack C with large spheres having radius r1 . The total volume of large spheres: (0.604 0.1812r1 )V The total surface area of large spheres: 3 (0.604 0.1812r1)V r1
Let us consider the remaining space C1 of the container. The volume: V1 = V (0.604 0.1812r1)V The surface area: B1 = B + (0.604 0.1812r1)V 3 r1
Yamada, Kanno, and Miyauchi
Multi-Sized Sphere Packing
Two-Sized Sphere Random Loose Packing
Let us consider the remaining space C1 of the container. The volume: V1 = V (0.604 0.1812r1)V The surface area: B1 = B + (0.604 0.1812r1)V 3 r1
Now if we pack small spheres of radius r2 into C1 , then the total volume is. 0.604(V1 0.3B1 r2 ).
Yamada, Kanno, and Miyauchi
Multi-Sized Sphere Packing
Two-Sized Sphere Random Loose Packing
Let us consider the remaining space C1 of the container. The volume: V1 = V (0.604 0.1812r1)V The surface area: B1 = B + (0.604 0.1812r1)V 3 r1
Now if we pack small spheres of radius r2 into C1 , then the total volume is. 0.604(V1 0.3B1 r2 ). The total density after randomly packing spheres of radii r1 > r2 is r 0.843184 0.0717552r1 0.0826997r2 0.328334 2 . r1
Yamada, Kanno, and Miyauchi Multi-Sized Sphere Packing
Two-Sized Particle Packings
The larger ratio of two sizes is, the higher density can be obtained.
Yamada, Kanno, and Miyauchi
Multi-Sized Sphere Packing
Two-Sized Particle Packings
The larger ratio of two sizes is, the higher density can be obtained. We may assume that the small size of radius r2 is xed. Then, the problem is how large radius r1 can be the best to fulll a highest density in the pores in concrete.
Yamada, Kanno, and Miyauchi
Multi-Sized Sphere Packing
Two-Sized Particle Packings
The larger ratio of two sizes is, the higher density can be obtained. We may assume that the small size of radius r2 is xed. Then, the problem is how large radius r1 can be the best to fulll a highest density in the pores in concrete. Theorem: r1 = 2.14
r2
Yamada, Kanno, and Miyauchi
Multi-Sized Sphere Packing
Computational Experiments
C: a very long cylinder with the square base 2 2 Spheres: 0.05 < r1 < 0.5,
=2
r2 = 0.02 (small xed)
Yamada, Kanno, and Miyauchi
Multi-Sized Sphere Packing
Computational Experiments
C: a very long cylinder with the square base 2 2 Spheres: 0.05 < r1 < 0.5, =2
r2 = 0.02 (small xed)
The total density is maximized when r1 = 2.14
0.02 2
= 0.214.
205 large spheres
Yamada, Kanno, and Miyauchi
Multi-Sized Sphere Packing
Computational Experiments
C: a very long cylinder with the square base 2 2 Spheres: 0.05 < r1 < 0.5, =2
r2 = 0.02 (small xed)
The total density is maximized when r1 = 2.14
0.02 2
= 0.214.
124393 small spheres
Yamada, Kanno, and Miyauchi
Multi-Sized Sphere Packing
Computational Experiments
C: a very long cylinder with the square base 2 2 Spheres: 0.05 < r1 < 0.5, =2
r2 = 0.02 (small xed)
density 0.80 0.78 0.76 0.74 0.72 0.70 r1
0.2
0.3
0.4
0.5
Yamada, Kanno, and Miyauchi
Multi-Sized Sphere Packing
Higher Digree Approximations
We may consider the average curvature and Gaussian curvature of the containers.
Yamada, Kanno, and Miyauchi
Multi-Sized Sphere Packing
Higher Digree Approximations
We may consider the average curvature and Gaussian curvature of the containers. 1 V 3 (C) = V (C) = 1 + 2 2 1 2
C
Yamada, Kanno, and Miyauchi
Multi-Sized Sphere Packing
Higher Digree Approximations
We may consider the average curvature and Gaussian curvature of the containers. 1 V 3 (C) = V (C) = 1 + 2 2 1 2
C
Mono-Sized Sphere Case 0.604 0.181r + 0.0544r 2 0.0163r 3
Yamada, Kanno, and Miyauchi
Multi-Sized Sphere Packing
High Degree Approximations
Two- Sized Sphere Case
r2 r r 0.843 0.328 r1 0.0985 r22 0.00985 r23
1 2 1 2 3
+(0.00646r1 3 + 0.00887r12 r2 + 0.00266r1r2 2 0.016r2 3 ) 3 +(0.0215r1 2 0.0296r1r2 + 0.0455r22 0.000887 rr21 )
r +(0.07176r1 0.0827r2 + 0.0296 rr21 + 0.00296 r22 )
1
Yamada, Kanno, and Miyauchi
Multi-Sized Sphere Packing
Higher digree Approximations
density 0.80 0.78 0.76 0.74 0.72 0.70 r1
0.2
0.3
0.4
0.5
blue: 1-degree approximation red: 2-degree and 3-degree approxiation
Yamada, Kanno, and Miyauchi Multi-Sized Sphere Packing
Summary and Future Problems
We found a simple approximate relation for the two sizes of spheres that provide a highest density in the multi-sized shere packing problem into a container C by using only the parameter B = V where B is the surface area of C and V is the volume of C.
Yamada, Kanno, and Miyauchi
Multi-Sized Sphere Packing
Summary and Future Problems
We found a simple approximate relation for the two sizes of spheres that provide a highest density in the multi-sized shere packing problem into a container C by using only the parameter B = V where B is the surface area of C and V is the volume of C. Find a relationship between the simulation results and compare with engineering experiments.
Yamada, Kanno, and Miyauchi
Multi-Sized Sphere Packing
Summary and Future Problems
We found a simple approximate relation for the two sizes of spheres that provide a highest density in the multi-sized shere packing problem into a container C by using only the parameter B = V where B is the surface area of C and V is the volume of C. Find a relationship between the simulation results and compare with engineering experiments. Find suitable models for medical applications, for example, repairing broken bones. For medical applications, the highest density is not always the best solution.
Yamada, Kanno, and Miyauchi
Multi-Sized Sphere Packing