(84ad) A Stochastic Optimization and Machine Learning-Based Framework for Evaluating Ammonia Utilization As a Hydrogen Carrier

Ammonia is a highly versatile chemical compound consisting of hydrogen and nitrogen. It has been widely used in various conventional industries such as agriculture, refrigeration, and cleaning agents. However, one of the most promising applications of ammonia is its use as a hydrogen carrier due to its relatively high hydrogen content of 17.7% by weight.¹ Using ammonia as a hydrogen carrier provides a safer and more efficient way to store and transport hydrogen compared to using compressed hydrogen gas that requires high pressure (>350 bar) and has some inherent risks such as high flammability and low volumetric energy density.^{2, 3}

In addition, ammonia can also be used as a feedstock for producing hydrogen through ammonia decomposition. During ammonia decomposition, hydrogen and nitrogen gas are produced using various methods including catalytic or thermal decomposition. Compared to other hydrogen production methods like steam reforming of methane, ammonia decomposition requires lower energy input making it a promising option for reducing carbon dioxide emissions.⁴ Overall, the versatility of ammonia as a hydrogen carrier and feedstock for hydrogen production makes it an attractive candidate for sustainable energy applications. Ongoing research efforts are exploring the potential of ammonia in various areas, such as fuel cells, transportation, and power generation. Advancements in this area may lead to significant breakthroughs in the transition towards a more sustainable and cleaner energy future.

Despite its potential as a hydrogen carrier and feedstock for hydrogen production, the current ammonia production process, the Haber-Bosch process, has some significant drawbacks. The process requires massive amounts of energy and emits enormous carbon dioxide because the Haber-Bosch process involves reacting hydrogen from fossil fuels with nitrogen from the air at high temperatures and pressure.⁵ Additionally, the high energy requirements of the process result in high production costs.

Therefore, the overall costs and emissions associated with ammonia utilization as a hydrogen carrier need to be carefully evaluated to ensure that it is a sustainable and cost-effective solution. While ammonia has several advantages over compressed hydrogen gas as a hydrogen carrier, it also has its own production costs and emissions associated with it. The hydrogen used in the ammonia production process comes from fossil fuels, so the cost of producing ammonia inherently includes the cost of producing hydrogen. Thus, it is essential to consider the life-cycle cost and emissions associated with the entire process, including the production, transportation, and use of ammonia as a hydrogen carrier. By evaluating these factors, we can ensure that ammonia utilization is a sustainable and cost-effective solution for transitioning towards a cleaner and more sustainable energy future.

This study aims to investigate the feasibility of utilizing ammonia as a hydrogen carrier and to propose an efficient evaluation framework for the overall cost and emissions associated with ammonia utilization strategies such as replacement of fossil fuel in on-site hydrogen refueling station or fuel cell power plants. Specifically, we develop an optimization framework that leverages the Monte Carlo method and machine learning to import ammonia into the Republic of Korea. To achieve this, we will begin by selecting several exporting countries based on historical import-export data and developing predictive models for ammonia production costs and associated emissions. With this information, we will establish trading routes between exporting countries and KOR taking into account the location of actual ammonia manufacturing complexes and export terminals. Additionally, we will construct a supply chain from KOR terminals to the actual ammonia demand.

To evaluate the entire cycle of ammonia, we will use a techno-economic analysis and optimize it in terms of life-cycle cost and carbon dioxide emissions using the Monte Carlo method. Based on the stochastic optimization results, we will develop several machine learning models and select the most suitable predictive model to make informed decisions regarding ammonia utilization. Our findings will provide insights into the feasibility of using ammonia as a hydrogen carrier and the potential benefits of using it to reduce carbon dioxide emissions.

With the stochastic optimization method, multiple scenarios can be established for ammonia utilization strategies such as determining the countries involved in trading and the method of ammonia production. Additionally, a system that users want to investigate can be optimized to minimize overall cost and environmental impacts. By utilizing machine learning-based regression models trained with the results of stochastic optimization, users can input specific parameters and obtain the total ammonia cost, unit product price (e.g., USD kWh^−1 or USD kgH₂^−1), and life-cycle emissions efficiently. Therefore, this study enables to efficiently analyze the feasibility of ammonia-based hydrogen facilities. In conclusion, this study provides a framework for efficiently evaluating the cost and emissions of different ammonia utilization strategies through optimizing ammonia import and training machine learning models for regression.

Reference

1. Li Y, Bi M, Li B, Zhou Y, Huang L, Gao W. Explosion hazard evaluation of renewable hydrogen/ammonia/air fuels. Energy. 2018;159:252-263.

2. Hua TQ, Ahluwalia RK, Peng JK, et al. Technical assessment of compressed hydrogen storage tank systems for automotive applications. Int J Hydrogen Energy. 2011;36(4):3037-3049. doi:10.1016/J.IJHYDENE.2010.11.090

3. Eberle U, Arnold G, Von Helmolt R. Hydrogen storage in metal–hydrogen systems and their derivatives. J Power Sources. 2006;154(2):456-460.

4. Cha J, Park Y, Brigljević B, et al. An efficient process for sustainable and scalable hydrogen production from green ammonia. Renew Sustain Energy Rev. 2021;152:111562.

5. Zhang L, Li R, Cui L, et al. Boosting photocatalytic ammonia synthesis performance over OVs-Rich Ru/W18O49: Insights into the roles of oxygen vacancies in enhanced hydrogen spillover effect. Chem Eng J. 2023;461:141892. doi:10.1016/J.CEJ.2023.141892