The path from Ammonia to Power

One of the major drawbacks of renewable energies is the instability of the power generation, varying with the climate conditions. Hydrogen is being studied as a potential energy vector stabilizing this power production, by using the generated renewable electricity to produce hydrogen via electrolysis and converting it back into electricity following the power demand. As it has very poor natural storage properties, a better way to store and transport this hydrogen is by converting it to ammonia. The conversion of this ammonia back into electricity is the subject of this work. Multiples techniques exist that can convert ammonia into electricity, namely fuel cells and Combined Cycle Gas Turbines (CCGT). In this study, three power plants based in Belgium had to meet the power demand of a typical Brussels neighbourhood, using only the electricity produced via their ammonia to power conversion method or electricity bought from the grid. The three methods studied are a Proton Exchange Membrane fuel cell (PEMFC) requiring transformation of the ammonia into hydrogen beforehand (i.e. "cracking"), a Solid Oxide fuel cell (SOFC) working directly with ammonia, and a CCGT working directly with ammonia. The power plants could adapt their total power capacity in order to better perform economically or environmentally. First, the three methods were modelled in a Python environment. Then a multi-objective deterministic optimization using a Non-dominated Sorting Genetic Algorithm (NSGA-II) was performed on the models for three different scenarios, finding out the best parameters that minimize three objectives: the Levelized Cost of Electricity (LCOE), the Carbon Dioxide Equivalent Emissions (CDEE), and the Self-Sufficiency Ratio (SSR). A robust design optimization using the same NSGA-II algorithm was then performed on the models to study their robustness in regard to variation of the parameters. The uncertainties are propagated in the models via the Polynomial Expansion Chaos method (PCE). After that, an uncertainty quantification (UQ) was performed to find out which parameters influenced the most the LCOE. The results obtained show that for current scenarios, trade-offs between low LCOE and low emissions exist for the PEMFC model and the CCGT model, but not for the SOFC model which optimizes both objectives by buying all the demand from the electricity grid. Tradeoffs between SSR and LCOE exist for all three methods. Horizon-2030 scenarios show a clear amelioration of the LCOE forecast for the PEMFC method (-31%), while both SOFC and CCGT models will be more expensive and have higher CDEE than grid electricity. The cheapest method to produce electricity from ammonia currently is the CCGT with a LCOE of 187.42 [AC/MWh], followed by the PEMFC (+96.4%) and the SOFC(+98.5%). For the Horizon-2030, the cheapest will be the CCGT with a LCOE of 113.07 [AC/MWh], followed by the PEMFC (+131.28%) and the SOFC (+146.32%). The cleanest method currently is the PEMFC with a CDEE of 45.77 [g CO2,eq./kWh], followed by the CCGT (+176.7%) and the SOFC (+239.7%). For the Horizon-2030, the cleanest method will be the PEMFC method with a CDEE of 43.89 [g CO2,eq./kWh], followed by the CCGT (+184.3%) and the SOFC (+250.0%)