Publications

Detailed Information

앙상블과 포텐셜 변수들의 불확실성을 고려한 나노재료의 설계민감도 해석 : Design Sensitivity Analysis of Nanomaterials considering Uncertainties of Ensemble and Potential Parameters

DC Field Value Language
dc.contributor.advisor조선호-
dc.contributor.author김현석-
dc.date.accessioned2017-07-13T06:06:30Z-
dc.date.available2017-07-13T06:06:30Z-
dc.date.issued2015-08-
dc.identifier.other000000067356-
dc.identifier.urihttps://hdl.handle.net/10371/118275-
dc.description학위논문 (박사)-- 서울대학교 대학원 : 산업·조선공학부, 2015. 8. 조선호.-
dc.description.abstract본 연구에서는 앙상블과 포텐셜 변수들의 불확실성을 고려하여 분자동역학 (molecular dynamics)을 이용한 나노재료의 설계민감도 해석 방법론을 개발하였다. 불확실성을 고려한 설계민감도를 바탕으로 하는 1차 신뢰성 방법 (First-Order Reliability Method)의 개념을 도입하여 불확실성이 분자동역학 해석과 그 설계민감도 해석에 미치는 영향을 효율적이고 정확하게 분석하였다. 실제적인 나노스케일에서의 문제에의 적용을 위해 노제-후버 온도 조절 장치 (Nose-Hoover thermostat)을 이용한 항온 분자동역학 전산모사를 수행하였다.
최근 조선해양공학 분야의 전반에서는 기존 연속체 기반 해석법의 한계를 뛰어넘고자 분자 레벨에서의 해석과 설계에 대한 연구가 활발하게 진행되고 있다. 나노스케일에서의 동적 그리고 열역학적 특성을 실험을 통해 관측하는 것은 매우 어렵거나 불가능하다. 하지만, 분자동역학 전산모사는 분자 레벨에서 물리적으로 복잡한 현상들의 이해에 있어 적합한 프레임워크 (framework)를 제공한다. 분자동역학 전산모사에서 불확실성의 영향은 잘 알려져 있다. 이러한 불확실성은 앙상블 (ensemble) 과 포텐셜 (potential) 변수들에서 기인한다. 앙상블 변수로부터의 불확실성은 임의의 초기 속도에 따라 발생한다. 포텐셜 변수에 의한 불확실성은 분자동역학 전산모사에서 분자 간 포텐셜이 알려져 있지 않음에 따라 발생한다. 기존에는 이러한 불확실성에 따른 영향을 초기 속도 조건 또는 포텐셜 변수들을 바꿔가면서 반복적인 분자동역학 해석을 통해 분석하였다.
이러한 불확실성이 분자동역학 전산모사와 설계민감도 해석에서 미치는 영향을 개발한 설계민감도 바탕의 일차신뢰성방법을 활용하여 효율적이고 정확하게 분석하였다. 분자동역학 전산모사에서의 설계민감도 해석에는 효율성과 정확성을 위해 분석적인(analytical) 설계민감도 해석기법을 적용하였다. 유한차분법과 같은 근사화된 설계민감도 해석법은 비선형성이 높은 설계 변수들을 포함하는 분자동역학 전산모사에서 그 정확성과 효율성을 보장하기 힘들다.
나노 재료는 조선해양분야뿐만 아니라 다양한 분야에서 그 활용도가 매우 높다. 본문에 수록한 예제들을 통해 다양한 나노 재료들에 대하여 개발한 불확실성을 고려한 설계민감도 해석기법을 적용하였고, 분자동역학 전산모사와 그 설계민감도에서 불확실성의 영향을 분석하였다. 또한 나노입자와 나노구조물 사이의 알려지지 않은 포텐셜 변수들을 찾기 위해 설계민감도를 바탕으로 한 최적 설계를 수행하였고 의미 있는 결과를 얻을 수 있었다.
-
dc.description.abstractA design sensitivity analysis (DSA) method considering uncertainty from ensemble and potential parameters in molecular dynamics (MD) is developed. By utilizing the first-order reliability method (FORM) concept based on the developed design sensitivity, impact of the uncertainty to both MD simulation and DSA of MD is evaluated efficiently and accurately. For practical application to nanoscale design problems, constant temperature MD simulation is considered under Nose-Hoover thermostat.
Recently, in both academic and industrial environment of naval architecture and ocean engineering, atomistic level simulation and design is essential to overcome the limits in conventional continuum based approach. It is well known that to study dynamical and thermo-dynamical properties of a system in real time by experimental approach at nanoscales is significantly difficult or impossible. However, the molecular dynamics simulations provide a suitable framework for elucidating physically complex phenomena at the atomic level with femtosecond resolution. Generally, uncertainty have a significant impact on observations and predictions in both simulation and design of nanomaterials. We identify the origin of uncertainty as ensemble and potential parameters. The uncertainty from ensemble parameters is inherently present in atomistic systems due to the random initial velocities. The potential parameter uncertainty arises when some parameters defining the interatomic interactions in MD simulations are unknown. Normally, the influences of uncertainty to the MD simulation is evaluated through a tedious repetition of MD simulations with different initial conditions or by fitting the potential parameters to the desired bulk properties.
In this study, we perform the FORM analysis based on the developed analytical first and second order sensitivities with respect to random and design variables to evaluate the effect of uncertainty in MD simulation and DSA of MD. For accuracy and efficiency in the DSA of MD system, analytical sensitivity methods such as direct differential method (DDM) and adjoint variable method (AVM) are employed. The approximated DSA methods such as finite difference method (FDM) are impractical in nanoscale problems since MD simulations are highly nonlinear to design variables and computationally costly.
Various nanomaterial structures are applied in broad fields. For instance, nanoparticles are utilized for antifouling of marine structures. Through some numerical examples, the accuracy and efficiency of the developed DSA method considering the uncertainty is demonstrated for various nanoscale problems. The variances of performance measure in MD simulation and sensitivity due to the uncertainty is evaluated trough FORM analysis based on the developed sensitivities. We further extend to a design optimization problem based on sensitivity to find unknown potential parameters between nanoparticle and nanostructure.
-
dc.description.tableofcontentsAbstract i
Contents iii
List of Tables vii
List of Figures x
Nomenclature xiv
Chapter 1. Introduction 1
1.1 Motivation 1
1.1.1 Various applications of versatile nanomaterials 1
1.1.2 Uncertainty in molecular dynamics (MD) and nanomaterial design 5
1.1.3 Design sensitivity of molecular dynamics considering uncertainty 8
1.2 Purpose and Scope 11
1.3 Literature survey 15
1.3.1 Molecular dynamic simulations 15
1.3.2 DSA of transient dynamics 18
1.3.3 Reliability analysis 20
1.3.4 Applications of nanomaterials 21
1.3.5 Organizations of thesis 24
Chapter 2. Molecular Dynamics Simulations 25
2.1 Equation of motion 26
2.1.1 Lagrangian and Hamiltonian equations of motion 26
2.1.2 NVT ensemble 29
2.2 Interatomic potentials 32
2.2.1 Lennard-Jones (LJ) potentials 33
2.2.2 Embedded Atom Method (EAM) for metallic systems 34
2.2.3 Tersoff bond-order potential for covalent bond 38
2.2.4 Cut-off radius of potentials 39
2.3 Periodic boundary conditions 40
Chapter 3. Reliability Analysis 41
3.1 Descriptions of Random Variables 41
3.1.1 Mean Value 42
3.1.2 Variance and Standard Deviation 42
3.2 Probability Distributions 43
3.2.1 Normal Distribution 43
3.2.2 Lognormal Distribution 45
3.3 Limit State Function 47
3.4 Reliability Index 47
3.5 Second-Moment Theory of Reliability Analysis 47
3.5.1 Geometric measure of reliability 48
3.5.2 First-Order Reliability Moment (FORM) Method 54
Chapter 4. Design Sensitivity Analysis 57
4.1 Design sensitivity analysis of MD 57
4.1.1 Adjoint variable method for NVE ensemble 57
4.1.2 Adjoint variable method for NVT ensemble 60
4.1.3 Discontinuity problems introduced by cut-off 62
4.2 Design Sensitivity Analysis considering Uncertainty 71
4.2.1 FORM based approach 71
4.2.2 Design sensitivity considering uncertainty in MD simulation 72
4.2.3 Design sensitivity considering uncertainty in DSA of MD 73
4.3 Design sensitivity analysis for reliability analysis 75
4.3.1 Reliability based design sensitivity analysis 75
Chapter 5. Numerical Examples 78
5.1 Uncertainty effect in MD simulation 79
5.1.1 MD simulation of nanoparticle synthesis 80
5.1.2 Uncertainty effect to MD simulation of nanoparticle synthesis 86
5.1.3 Experiment of nanoparticle synthesis 93
5.2 Design optimization: unknown potential parameters 102
5.2.1 Optimization of nanoparticle-substrate force field parameters 103
5.2.2 Verification of developed design sensitivity 122
5.2.3 Uncertainties from random initial velocity 128
5.3 Atomic Mass Uncertainty: 3-D Gas Diffusion Problem 132
5.3.1 Verification of developed design sensitivity 133
5.3.2 Uncertainties from random atomic mass 136
5.4 1-Dimensional Harmonic Oscillator Problem 138
5.4.1 Verification of developed design sensitivity 141
5.4.2 Uncertainties from random initial velocity 144
5.4.3 Reliability based sensitivity analysis 148
5.5 Agglomeration of nanoparticles on a substrate 151
5.5.1 Experimental observation of nanoparticle agglomeration 151
5.5.2 Uncertainties from random initial velocity 156
5.5.3 Reliability based design sensitivity analysis 169
Chapter 6. Conclusions and Future Works 170
6.1 Conclusions 170
6.2 Future works 171
Bibilography 172
초 록 180
-
dc.formatapplication/pdf-
dc.format.extent6606989 bytes-
dc.format.mediumapplication/pdf-
dc.language.isoen-
dc.publisher서울대학교 대학원-
dc.subject분자동역학-
dc.subject앙상블 불확실성-
dc.subject포텐셜 불확실성-
dc.subject나노재료-
dc.subject설계민감도 해석-
dc.subject최적설계-
dc.subject.ddc623-
dc.title앙상블과 포텐셜 변수들의 불확실성을 고려한 나노재료의 설계민감도 해석-
dc.title.alternativeDesign Sensitivity Analysis of Nanomaterials considering Uncertainties of Ensemble and Potential Parameters-
dc.typeThesis-
dc.contributor.AlternativeAuthorKim, Hyun-seok-
dc.description.degreeDoctor-
dc.citation.pages181-
dc.contributor.affiliation공과대학 산업·조선공학부-
dc.date.awarded2015-08-
Appears in Collections:
Files in This Item:

Altmetrics

Item View & Download Count

  • mendeley

Items in S-Space are protected by copyright, with all rights reserved, unless otherwise indicated.

Share