표제지
ABSTRACT
목차
NOMENCLATURE 7
CHAPTER 1. INTRODUCTION 16
CHAPTER 2. Multi-scale, Two-phase PEMFC model 21
2.1. Model assumptions 21
2.2. Governing equations and source terms 22
2.3. Microscale CL model 22
2.4. Boundary conditions and numerical implementation 23
CHAPTER 3. Reliability Based Design Optimization 24
3.1. General formulation of RBDO 24
3.2. Sampling based RBDO 26
3.3. Metamodeling (Surrogate model) 27
CHAPTER 4. Results and Discussion 31
4.1. Experimental results and Construction the MLP through Validated 3D PEMFC model 31
4.2. Deterministic Design Optimization based on Particle Swarm Optimization 32
4.3. Design optimization considering the tolerance through RBDO 34
CHAPTER 5. Conclusion 38
REFERENCES 68
Table 1. PEM fuel cell model: Governing equations 40
Table 2. PEM fuel cell model: Source/sink terms. 41
Table 3. Kinetic, transport, and physiochemical properties of the fuel cell. 43
Table 4. Expressions used in the two-phase mixture model. 45
Table 5. Transport properties in the electrolyte. 47
Table 6. Correlations of the microscale CL model. 48
Table 7. Boundary and operating conditions. 51
Table 8. Marginal PDF, CDF, and its parameters. 52
Table 9. Variables, Power density, Reliability. 53
Figure 1. Microscale and macroscale computational domain and coupling variables for multi-scale PEMFC simulations 54
Figure 2. (a) Deterministic optimum point that violates target reliability, (b) Reliability based optimum point moved to feasible region and satisfies the... 55
Figure 3. The general flow chart of Sampling based RBDO using surrogate model 56
Figure 4. Schematic of (a) MLP and weight update vectors, (b) weight update vector between hidden layer - output layer, (c) weight update... 57
Figure 5. Experimentally fabricated 3D cathode Parallel flow field designs and computational domain of PEMFC single channel 58
Figure 6. Comparison of simulated polarization curves with experimental data 59
Figure 7. Scatter plots for the training data and test data that depict the correlation between simulation results by 3D PEMFC model and prediction... 60
Figure 8. Comparison of the pareto optimal front of two objectives (Vcell and Pressure drop) with single-objective optimal point by PSO and the non-...[이미지참조] 61
Figure 9. 2D surface including deterministic optimal point(DO) and constraints optimal point(CO) with feasible region and infeasible region 1,2... 63
Figure 10. The 3D scattered plot of Deterministic optimal point(DO), Constraint optimal point(CO) and RBDO optimal point(Target reliability >... 64
Figure 11. The 3D scattered plot of (a) RBDO optimal point(Target reliability > 99 %) and RBDO optimal point(Target reliability > 90 %), (b) RBDO... 65
Figure 12. The comparison of (a) RBDO optimal point(Target reliability > 99 %) and RBDO optimal point(Target reliability > 90 %), (b) RBDO optimal... 66