• 沒有找到結果。

4-1 Summary

In the chapter three, we deal with the two cases, one is two free identical particles, and other is two identical particles in a cylindrical pore. In order to make sure code validity, we use the same parameter in previously literature. All of the previously works are assumes the numerical model is symmetry and reduce to two-dimensional problem. But the calculation of the interaction between two spheres closed to a planar wall [4] requires a full three-dimensional problem. Assume the problem to two-dimensional may have some uncertain. We develop this three-dimensional Poisson-Boltzmann equation just can to deal with this troublesome problem.

In the propose study, we can obtain the parallel Poisson-Boltzmann equation coupled with PAMR is very useful and convenience. It can be more closed to real situation at colloidal systems.

4-2 Recommendations for Future Work

In this propose study, we meet two difficulties, one is the convergence rate too slow, the other is refinement level can not over than five. We can use fine initial guess to overcome the convergence rate problem, but the cost of the refinement computation, we can not overcome directly. But there still have some way to overcome this situation, we can use the high order mesh and coupled different mesh such as hexahedral cell and pyramid. If we can use high order mesh to capture the EDL potential distribution, we can reduce the inaccuracy without large computationally cost.

References

[1] D.J. Shaw, Electrophoresis, Academic Press, New York, (1969)

[2] W.R. Bowen, A.O. Sharif, “Adaptive Finite-Element Solution of Nonlinear Poisson-Boltzmann Equation: A Charged Apherical Particle at Various Distances from a Charged Cylindrical Pore in a Charged Planar Surface” J. Colloid Interface Sci. 187(1997)

[3] W.R. Bowen, A.O. Sharif, “ Long-range electrostatic attraction between like-charge spheres in a charged pore” Nature 385 (1998) 663

[4] A.E. Larsen, D.G. Grier, “Like-charge attractions in metastable colloidal crystallites” Nature 385 (1997) 230

[5] P.E. Dyshlovenko, “Adaptive Mesh Enrichment for the Poisson-Boltzmann Equation” J. Comp. Phs 172 (2001) 198

[6] P.E. Dyshlovenko, “Adaptive numerical method for Poisson-Boltzmann equation and its application” CPC. 147 (2002) 335-338

[7] R. Tuinier, “ Approximate solutions to Poisson-Boltzmann equation in spherical and cylindrical geometry” J. Colloid Interface Sci 258 (2003) 45-49

[8] Prodip K. Das, Subir Bhattacharjee, “Finite element estimation of electrostatic double layer interaction between colloidal particles inside a rough cylindrical capillary: effect of charging behavior” Colloids and Surfaces A 256 (2005) 91-103 [9] R. –C. Chen, Jinn-Liang Liu. “Monotone iterative methods for the adaptive finite

solution of semiconductor equation” J. of Computational and Applied Mathematics 159 (2003) 341-364

[10] Almasi, G. and A., G., “Highly Parallel Computing, Benjamin/Cummings, Red-Wood city, CA,1989

[11] Shivaraju B. Gowda “A Comparison of Sparse & Element-by-Element Storage

Schemes on The Efficiency of Parallel Conjugate Gradient Iterative Methods for Finite Element Analysis” (2002) Graduate School of Clemeson University

[12] Yuan-Wen Jeng , NCTU MUST, ‘’Parallel Implementation of Three-dimensional Unstructured Mesh Refinement in the Direct Simulation Monte Carlo Method’

[13] Kuo, Chia-Hao, "The Direct Simulation Monte Carlo Method Using

Unstructured Adaptive Mesh and Its Applications,", MS Thesis, NCTU, Hsinchu, Taiwan 29, Jun., 2000.

[14] Wu, Fu-Yuan, "The Three-Dimensional Direct Simulation Monte Carlo Method Using Unstructured Adaptive Mesh and It Applications", MS Thesis, NCTU, Hsinchu, Taiwan , July., 2002.

[15] S. Levine, J.R. Marriot, G. Neale, N. Epstein. J. Colloid Interface Sci. 52 (1) (1975) 136-149

[16] S. Levine, G.M. Bell, Discuss. Faraday Soc. 42(1966) 69 [17] E. Bresler. J. Colloid Interface Sci. 118 (2) (1987) 326 [18] P.L. Levine, J. Colloid Interface Sci. 51(1975) 72

[19] T.A.R. Strauss, H.K. Bowen, J. Colloid Interface Sci. 118(2) (1987) 326 [20] H.K. Bowen, J. Colloid Interface Sci. 173 (1995) 388-395

[21] G.M. Mala, D. Li, C. Yang, ‘’Electrical double layer potential distribution in a rectangular micro channel,’’ Int. J. Heat Fluid Flow, (1997).

[22] Patankar, N. A., and Hu, H. H., ‘’Numerical Simulation of Electroosmotic Flow,’’ Anal. Chem. 70, 1870 (1998)

[23] Arulanandam, S., and Li, D., ‘’Liquid transport in rectangular microchannels by electroosmotic pumping,’’ Colloids and Sufaces A: Physicochemical and Engineering Aspects, 161, 89-102 (2000).

[24] J. Yang, L. M. Fu, and C. C. Hwang, “Electroosmotic Entry Flow in a

Microchannel,”J. Colloid Interface Sci., 244, 173-179, (2001).

[25] J. Yang, L. M. Fu, and Y. -C. Lin ‘’Electroosmotic Flow in Microchannels,’’ J.

Colloid Interface Sci. 239, 98-105 (2001).

[26] H.A. Stone, A.D. Stroock, and A. Ajdari, Annu. ‘’Engineering Flows in small devices,’’ Rev. Fluid Mech. 36:381-411 (2004)

[27] R. -C. Chen, Jinn-Liang Liu. ‘’Monotone iterative methods for the adaptive finite element solution of semiconductor equations,’’ J. of Computational and Applied Mathematics 159 (2003) 341–364

[28] Yiming Li, ‘’A parallel monotone iterative method for the numerical solution of multi-dimensional semiconductor Poisson equation,’’ Computer Physics Communications 153 (2003) 359–372

[29] R.J. Hunter Zeta Potential in Colloid Science: Principles and Applications, Academic Press, New York, (1981)

[30] H. J. Keh, and H. C. Tseng, J. Colloid Interface Sci. 242, 450-459, (2001) [31] J. Yang, and D. Y. Kwok, J. colloid Interface Sci., 260, 225-233, (2003) [32] C. C. Chang, and R. J. Yang, J. Micromech. Microeng., 14, 550-558, (2004) [33] Kuo-Hsien Hsu, NCTU MUST ‘’Three-dimensional parallelize Poisson slover’’

[34] W.B. Russel, D.A. Saville, W.R. Schowalter, Colloidal Dispersions, Cambridge University Press, New York, 1989.

Table

Table3-1: Steps of the adaptive process:

Refinement level

Number of nodes Number of elements Force of interaction(Fs)

0 924 3821 20.9924

1 3045 14305 34.6570

2 18253 96638 40.4853

3 132477 747048 44.0112

4 187732 1049734 48.0722

5 213224 1190741 48.4066

PS: These cases are runed by six parallel processors.

Table3-2: Steps of the adaptive process for previously literature [6]

Refinement level

Number of elements Force of interaction(Fs)

0 946 36.518

1 1992 43.716

2 4466 46.596

3 7642 47.597

4 11314 48.472

5 13525 48.630

6 15187 48.862

7 15477 48.826

8 15541 48.840

9 15555 48.843

10 15578 48.841

11 15588 48.840

Table3-3: The interaction forces of two particles in a cylinder pore at different separation distance.

Separation Distance(h) 0.5 1 3.5 6 8

Force of interaction(Fs) 26 13 -0.7 -4.7 0.9

PS: These cases are runed by six parallel processors.

Figure

Fig 1.1 The illustration of like-charge attractions phenomenon

Fig 1.2 The Electric Double Layer distribution

No

Yes

Read in mesh generation

Start

Com pute the coefficients of shape

function

Setup initial &

boundary conditions

Com pute coefficient m atrix

Com pute loading m atrix

Solve matrix

Converge?

Output

Stop

Fig 2.1 Simplified flow chart of the three-dimensional nonlinear Poisson-Boltzmann solver

Fig 2.2 Distributed memory architecture

Fig 2.3 Shared memory architecture

Fig 2.4 The flow chart of parallel Poisson-Boltzmann solver

Fig 2.5 Isotropic mesh refinement of tetrahedral mesh (T: Tetrahedron)

isotropic ( 1st stage) isotropic ( 2nd stage)

anisotropic ( 2nd stage) initial grid

initial grid

initial grid

2 hanging node anisotropic (2nd stage)

3 hanging node isotropic (2nd stage)

1 hanging node anisotropic (2nd stage)

2 hanging node

3 hanging node 1 hanging node

Fig 2.6 Mesh refinement rules for two-dimensional triangular cell

refined 4 cells

refined 8 cells refined 2 cells

(+1) (+2) (+3) (+1) (+4)

non-coplanar coplanar

6 5 4 3 2 1

coplanar

: n hanging node (+n) : adding n nodes

n

Fig 2.7 Schematic diagram for mesh refinement rules of tetrahedron

Fig 2.8 Schematic diagram of the proposed cell quality control

Fig 2.9 Schematic diagram of typical cell quality control

Fig 2.10 Schematic diagram of simple cell quality control

Fig 2.11 A case that the proposed cell-quality-control would not affect to it

CPU0 gather output data

flag the cells based on refinement criteria

renumber added nodes

synchronize

stop start

add nodes on cell edge refinement required

build neighbor identifier array CPU K (K=0,…,np-1)

N=N+1

distribute the data read grid data and

relative cell data

Fig 2.12 Flow chart of parallel mesh refinement module

Fig 2-13 Flow chart of moduleⅠ

remove hanging nodes

stop start

add nodes on isotropic cells

communicate hanging node information

hanging nodes on IPB

?

yes

no

add nodes on anisotropic cells

Fig 2-14 Coupled PPBS-PAMR method

visual preprocessor

stop PAMR

mesh partition

(weighting from parallel Poisson-Boltzmann solver)

PPB mesh partition

(weighting based on cell volume)

start

(N=0)

N< NMAX

Refinement criteria satisfied ?

no yes

yes

N=N+1

Fig 3.1 Geometry of the potential around a sphere case

Fig 3.2 The comparison of different result at Ψ=1(sphere)

Fig 3.3 The comparison of different result at Ψ =5(sphere)

Fig 3.4 The relationship between iteration and residual with different initial guess (Ψ =5)

Fig 3.5 Geometry of the potential around a cylinder case

Fig 3.6 The comparison of different result at Ψ =1(cylinder)

Fig 3.7 The comparison of different result at Ψ =5(cylinder)

Fig 3.8 The illustration the case of two interacting identical particles

Fig 3.9 Level 0 initial mesh of two interacting identical particles case

Fig 3.10 Level 1 refinement mesh of two interacting identical particles case

Fig 3.11 Level 2 refinement mesh of two interacting identical particles case

Fig 3.12 Level 3 refinement mesh of two interacting identical particles case

Fig 3.13 Level 4 refinement mesh of two interacting identical particles case

Fig 3.14 Level 5 refinement mesh of two interacting identical particles case

Fig 3.15 The potential distribution of two identical particles

Fig 3.16 The potential distribution for the sphere confined in a pore at separation distances h=0.5

Fig 3.17 The potential distribution for the sphere confined in a pore at separation distances h=6

Fig 3.18 Potential along the midplane between two spheres

Fig 3.19 Potential along midplane between two spheres (in local)

相關文件