Unleash the renewable power within ammonia : optimization and uncertainty quantification of the Ammonia-to-Power pathway

Details

Supervisors

Contino, Francesco ; Verleysen, Kevin

Faculty

Ecole polytechnique de Louvain

Degree label

Master [120] : ingénieur civil en chimie et science des matériaux, à finalité spécialisée

Abstract

As the need for renewable energy to replace fossil fuels is getting more pressing, this transition is hindered as the erratic origins of renewables stand as a barrier to fully sustain the world's needs. To compensate for this natural imbalance, energy carriers such as hydrogen (H2) carried through green ammonia seem to fit the role of the missing link to a sustainable green society. To convert these carriers back to electricity, existing technologies such as Combined Cycle Gas Turbine (CCGT) have been studied. Novel technologies such as fuel cells, which allow to avoid exhaust fumes, are also alternatives. As the interest in the latter is rising, an optimisation and sensitivity analysis can be crucial to evaluate their capability in stationary applications. To assess this issue, I modelled a fuel cell power plant composed of an ammonia cracker, buffer tanks and a Proton Exchange Fuel Cell (PEMFC) stack. The model was tested to replace the imported load of a warehouse in Germany following their data in 2016. It was first submitted to a deterministic optimization through a Nondominated Sorting Genetic Algorithm II (NSGA-II) with 4 objectives: minimizing the PEMFC average over- and undershoots, the PEMFC transient response time, the buffer tanks mass content and power demand-production gap. Then, an uncertainty quantification was performed for several optimized designs with a Polynomial Chaos Expansion (PCE) algorithm. This report first presents a literature overview of both fuel cells and combined cycle gas turbine. It then exposes the model, its optimization and sensitivity analysis. It was found that minimizing the power over- and undershoot (8.63%) requires to increase the stack size (1499 cells) while minimizing both the transient response time (1.17 s) and the power demand-production gap (72 W/15min) is possible by decreasing the stack size (155 cells). The sensitivity analysis further revealed that the stack is highly sensitive to thermodynamical parameters defining its activation potential (76,1% to 93,4%), but also to its capacitance (83,4% to 91,6%) and to the power demand (64,1% to 92,1%). In conclusion, designs minimizing the PEMFC stack size are more flexible, while the system is still highly sensitive to thermodynamical parameters and to the power demand variation. Further researches including a dynamical description of the ammonia cracker and performing a robust optimization could help resolve the loopholes encountered in this study.