Towards an agile Power-to-Ammonia pathway : optimization of a dynamic ammonia production process

Details

Supervisors

Contino, Francesco

Faculty

Ecole polytechnique de Louvain

Degree label

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

Abstract

The intermittent nature of renewable energies had made its high penetration (≥15%) in the grid difficult unless upgrades are made to increase the grid's flexibility. However, this need can be circumvented by storing the excess energy in ammonia and re-injecting this energy back to the grid during shortage. This storage concept, called power-to-ammonia, relies on renewable energies to power the electrolyzer which produces the H2 needed for the Haber-Bosch synthesis process. Nevertheless, the variability of the electric supply causes fluctuations in the H2 supply to the synthesis loop, which is not optimized for flexible operation. These fluctuations can cause temperature to vary inside the reactor such that limit cycle behavior or temperature extinction is achieved, resulting to catalyst disintegration. Therefore, a reactor system designed to have high temperature resilience is of great interest. Using Non-dominated Sorting Genetic Algorithm-II (NSGA-II), deterministic design optimization was performed with the goal of maximizing temperature resilience and NH3 output flow rate. The Polynomial Chaos Expansion (PCE) was then used to propagate the uncertainties in the model, where robust design optimization was performed on the most resilient deterministic design, to have reliable net NH3 output flow rate. The effect of cooling system configuration and reactor volume on temperature resilience was determined by considering a non-optimized volume of the system for the former, while optimizing the volume for maximum resilience and flow rate, for the later. Results showed that introduction of the feed in smaller amounts as quench streams into the reactor system, as in direct quenching, led to slightly higher resilience (≥1.01 times) compared to when introducing a single, larger feed, as in indirect quenching. On the other hand, the effect of reactor volume depended on the cooling configuration adopted by the reactor system. The presence of uncertainties led to reactor extinction when using the lowest uncertainty values, while higher steady-state temperature values were obtained under operation at the highest values of uncertainties. Finally, the RDO showed that for the adopted lowest H2/N2 ratio during ramping, optimization was unsuccessful. Therefore, improvements need to be made to account for ramping duration during robust optimization.