Search

Regions with abundantly available renewable energy are not necessarily the same as those with a high population density and energy consumption. Therefore, renewable energy can be produced in optimal climate conditions with a remote renewable hub and transported to these population-dense regions. To establish this energy transport to these regions, ammonia provides a flexible, easy-to-handle energy carrier. However, current literature rarely considers the impact of techno-economic uncertainty on the feasibility of this transport. Using those uncertainties, we performed a robust design optimization on the levelized cost of ammonia and the power-to-ammonia efficiency to compare the local (Belgium) and remote (Morocco) ammonia production and transport to Belgium. This paper provides the robust designs (i.e. least sensitive to uncertainty) for local and remote renewable ammonia production and the advantages of both approaches on the levelized cost and energy efficiency. The results confirm that ammonia production in regions with high solar irradiance followed by the transport of ammonia is cost-effective and robust (601 euro/tonneNH3in mean and 98 euro/tonneNH3in standard deviation) over local production (852 euro/tonneNH3in mean and 139 euro/tonneNH3in standard deviation). However, local ammonia production provides for more efficient (54.8% in mean) and less sensitive power-to-ammonia plant designs (0.16% in standard deviation), while the production in Morocco is less efficient (52.2% in mean) and more sensitive to uncertainties (0.39% in standard deviation). The capacity of the photovoltaic arrays and the electrolyzers highly influences both objectives. The sensitivity analysis shows that capital and operational expenses of the photovoltaics and electrolyzer stack dominate the designs with the lowest levelized cost in mean and standard deviation. However, the energy consumption uncertainty of the Haber–Bosch also impacts the cost of the lowest mean levelized cost. This uncertainty also dominates the designs with the highest energy efficiency in mean and lowest standard deviation.

High-fidelity thermodynamic simulation software is available to perform detailed simulations of power plants. However, these models depend on many operating parameters the user must characterize to assess the power plant performances. Unfortunately, most parameters are underdetermined by experience: the validation of the model using field measurements does not allow for the complete determination of all parameters, notwithstanding the unavoidable uncertainties of the measurements themselves. These limitations can result in a drastic mismatch between simulated and actual performance and lead to biased, suboptimal decision-making. To address these limitations, we performed Uncertainty Quantification on key furnace heat transfer parameters to predict the thermodynamic performance of coal-fired power plants retrofitted to biomass (co-)firing under uncertain operating conditions. We used a high-fidelity Thermoflex® model to simulate the thermodynamic performance of the power plant, and we adopted Polynomial Chaos Expansion to perform Uncertainty Quantification in a computationally-efficient manner. Finally, we evaluated the effect of various fractions of biomass in the fuel (from 0 to 100%) on the performance, which provides additional information in the decision-making process during the retrofit of the power plant. The results illustrate that the uncertainty on the non-uniform radiant flux factor dominates the uncertainty on the power, efficiency and flue gas temperature, meaning that efforts should aim at reducing the epistemic uncertainty on the radiative heat flux in the boiler. Increasing the biomass fraction results in a decrease in the gross power and gross efficiency. The mean Furnace Exit Gas Temperature remains relatively stable, but reaches a minimum value at 60% biomass co-firing. In conclusion, Polynomial Chaos Expansion allows for a computationally-efficient probabilistic assessment of non-validated operational conditions, such as a fuel switch, in high-fidelity models for thermal power plants. Future work will focus on extending the number of uncertain parameters.