Abstract
Green hydrogen and ammonia are critical technologies in our decarbonisation toolkit, but at present remain more expensive than traditional energy vectors. In order to increase their competitiveness with respect to fossil fuels, many authors have optimised production systems based on hourly solar and wind profiles and a range of technologies to maximise production and minimise costs.
This optimisation problem, however, is enormous in scale: it requires consideration of a large number of possible production sites and their performance over many years. Failure to consider both spatial and temporal variation in green ammonia production costs may exclude excellent locations, or include sites that are unreliable due to interannual variation.
In this work, we examine three techniques which can reduce the complexity of input data: time aggregation, hierarchical clustering, and K-means clustering. We compare the suitability of each of these approaches based on the extent to which they accelerate the solution of a green ammonia plant design optimisation problem, and the error between the simplified and actual solutions to the problem. Using these simplification approaches, we demonstrate the importance of considering interannual variation in green ammonia plant design.