Flow analysis and geometry optimization of an integrated ammonia decomposition-combustion reactor through CFD Modeling

EFCF 2025: Low-Temp. Fuel Cells, Electrolysers & H2 Processing 1 – 4 July 2025, Lucerne Switzerland https://doi.org/10.5281/zenodo.17244117 A0717 Page 1/6 A0717 Flow analysis and geometry optimization of an integrated ammonia decomposition-combustion reactor through CFD Modeling Soomin Lee (1), Jihyeon Son (1), Sangho Lee (2), Sanghun Lee* (1) (1) Department of Climate and Energy System Engineering, Ewha Womans University; (2) Department of Mobility Power Research, Eco-Friendly Energy Conversion Research Division, KIMM Institute of Carbon Neutral Energy Machinery; *Contact corresponding authors: www.EFCF.com/ContactRequest Abstract Ammonia, one of the hydrogen compounds, has gained attention as a hydrogen carrier due to its ease of liquefaction, high storage capacity, and environmental friendliness. Ammonia can also be utilized as a direct fuel in some fuel cells. However, it is necessary to consider a system with an ammonia reformer (cracker) because the performance of the fuel cell is significantly improved when the ammonia is pretreated (cracked) rather than directly injected. In order to improve the efficiency of the ammonia reformer, this study analyzes a system in which the off-gas from the fuel cell is reused in the combustor and the heat generated through combustion is utilized for reforming. The internal flow and combustion reactions are simulated by modeling the integrated ammonia reformer and combustor using Computational Fluid Dynamics. Based on this, the optimal operating conditions and reactor geometry are derived to improve ammonia reforming efficiency, achieve a uniform internal temperature, and reduce NOx emissions. The reformer system developed through the results of this study is expected to provide insights for ammonia-based fuel cell system design and operation strategies. EFCF 2025: Low-Temp. Fuel Cells, Electrolysers & H2 Processing 1 – 4 July 2025, Lucerne Switzerland https://doi.org/10.5281/zenodo.17244117 A0717 Page 2/6 Introduction As the demand for an energy transition toward low-carbon energy sources increases, ammonia is gaining attention as an efficient hydrogen carrier. Ammonia, a hydrogencontaining compound, has long been widely used around the world, particularly in the fertilizer industry. A well-established global infrastructure for ammonia production, storage, and transportation further highlights its advantages for a low-carbon energy society. Ammonia can serve not only as a hydrogen carrier but also as a direct fuel for power generation through solid oxide fuel cells (SOFCs). SOFCs operate at high temperatures and offer fuel flexibility, making it possible to utilize various fuels such as hydrogen, natural gas, ammonia, and biogas. However, a 1 kW SOFC stack demonstration conducted at Kyoto University confirmed performance differences between direct ammonia injection and injection after pre-reforming [1]. Improved fuel cell performance was observed when ammonia was decomposed prior to injection, and supplying pre-cracked ammonia also helped mitigate degradation reactions at the anode [2]. Therefore, to achieve stable and efficient fuel cell operation, it is necessary to pre-treat ammonia before injection. The Korea Institute of Machinery and Materials (KIMM) has developed an ammonia pre-treatment system that integrates an ammonia decomposition reactor with a combustor [3]. This system is designed to reuse the off-gas from the fuel cell for combustion, and the heat generated during combustion is then utilized to drive the reforming reaction. A schematic of the system is shown in Figure 1. This study aims to optimize the system to improve reforming efficiency, achieve temperature uniformity, and reduce NOx emissions in the developed ammonia pre-treatment unit. Since the fabrication of a physical prototype requires significant costs and materials, this study adopts computational fluid dynamics (CFD) simulations to optimize the geometry of the ammonia pre-treatment system. CFD is a powerful tool that numerically solves fluid motion equations to simulate flow, temperature, and pressure fields in a virtual environment. It is widely used across industries such as automotive, energy, and chemical engineering to evaluate system performance and reduce development time through simulation. In this study, reforming efficiency is defined as the optimization target, and key variables include the number and volume of catalyst layers, operating temperature, and the direction of ammonia injection. Prior to analyzing the full-scale pre-treatment unit, a simplified base model using a 3/4’’ stainless steel (STS) tube was analyzed. Based on validation with experimental data, the study will be extended to the actual reactor geometry. Figure 1. The configuration of the ammonia pre-treatment system EFCF 2025: Low-Temp. Fuel Cells, Electrolysers & H2 Processing 1 – 4 July 2025, Lucerne Switzerland https://doi.org/10.5281/zenodo.17244117 A0717 Page 3/6 1. Modeling Setup Before conducting CFD modeling of the ammonia pre-treatment unit, a base model was first developed and analyzed. The overall and elementary reactions for ammonia decomposition are shown below [4]. In the base model, the decomposition was modeled based on the overall reaction. 𝑶𝒗𝒆𝒓𝒂𝒍𝒍 𝒓𝒆𝒂𝒄𝒕𝒊𝒐𝒏 2𝑁𝐻3→𝑁2+3𝐻2 𝑬𝒍𝒆𝒎𝒆𝒏𝒕𝒂𝒓𝒚 𝒓𝒆𝒂𝒄𝒕𝒊𝒐𝒏𝒔 (𝑅1) 𝑁𝐻3+ ∗ ↔𝑁𝐻3 ∗ (𝑅2) 𝑁𝐻3 ∗+ ∗ ↔𝑁𝐻2∗+𝐻∗ (𝑅3) 𝑁𝐻2 ∗+ ∗ ↔𝑁𝐻∗+𝐻∗ (𝑅4) 𝑁𝐻∗+ ∗ ↔𝑁∗+𝐻∗ (𝑅5) 2𝑁∗↔𝑁2+2∗ (𝑅6) 2𝐻∗↔𝐻2+2∗ ∗:𝑉𝑎𝑐𝑎𝑛𝑡 𝑠𝑖𝑡𝑒 For the kinetics of the ammonia decomposition reaction, the Arrhenius constant and activation energy were obtained using the built-in Parameter Study function in COMSOL Multiphysics. The calculated values were applied to the chemical reaction rate equations (Eq. 1 and 2). The ammonia decomposition reaction was conducted in a plug flow-type reactor at 973.15 K. 𝑟=𝑘𝑓∏ 𝑐𝑖−𝑣𝑖 𝑖 ∈ 𝑟𝑒𝑎𝑐𝑡 −𝑘𝑟∏ 𝑐𝑖𝑣𝑖 𝑖 ∈ 𝑝𝑟𝑜𝑑 (𝑒𝑞.1) 𝑘=𝐴( 𝑇 𝑇𝑟𝑒𝑓)𝑛𝑒𝑥𝑝(−𝐸𝑎 𝑅𝑇) , 𝑇𝑟𝑒𝑓 =1𝐾 (𝑒𝑞.2) The geometry of the ammonia base model was simplified, as shown in Figure 2. The reactor and catalyst settings, assumptions, and boundary conditions were defined as presented in Tables 1–4. These values were set to match the experimental conditions provided by the Korea Institute of Machinery and Materials (KIMM). The overall reaction considered multicomponent species transport. As for boundary conditions, a GHSV of 2000/h and a flow rate of 167 mL/min were applied, and the wall was set to a no-slip condition. Table 1. Reactor parameters used in the CFD simulation of the base model Reactor 3/4‘’ STS tube Outside diameter 27.2 mm (KS Standard) Wall thickness 2.1 mm (KS Standard) Reactor height 81.6 mm Catalyst layer height 10.10 mm Table 2. Catalyst parameters used in the CFD simulation of the base model Catalyst Ru/Al2O3 (Ru 2.00 wt.%, 24.5 g/L) Pellet size Φ1.4mm~3.0mm EFCF 2025: Low-Temp. Fuel Cells, Electrolysers & H2 Processing 1 – 4 July 2025, Lucerne Switzerland https://doi.org/10.5281/zenodo.17244117 A0717 Page 4/6 Catalyst volume 5 mL Porosity 0.7 [5] Table 3. Assumptions and boundary conditions in the CFD simulation of the base model Assumptions Boundary conditions - Steady-state - Plug flow type reaction - Transport of concentrated species in porous catalysts - Outer walls: adiabatic - GHSV = 2000/h - CNH3 = C0, CN2 = 0, CH2 = 0 - NH3 flow rate = 167 mL/min - Treact = 973.15 K - No slip at wall Figure 2. Geometry of the base model for the ammonia decomposition reactor For fluid flow in porous media, the Brinkman equation (Eq. 3) was used as the governing equation. This equation was employed to account for mass and momentum conservation, as well as porous media resistance and fluid viscosity effects. The Brinkman equation is commonly used in complex porous flow problems. Next, for species conservation, the Maxwell–Stefan equation (Eq. 4) was applied. Unlike Fick’s law—which is typically used for ideal gases or single component systems and does not consider multicomponent interactions or pressure gradients—the Maxwell–Stefan equation provides a more detailed multicomponent diffusion model that incorporates species interactions, pressure gradients, and convective effects. 0=𝛻[−𝑝𝐼+1 𝜖𝑝{𝜇(𝛻𝑢+(𝛻𝑢)⊤) −2 3𝜇(𝛻∙𝑢)𝐼}]−(𝜇 𝜅+𝛽𝜌|𝑢|+𝑄𝑚 𝜖𝑝 2)𝑢 (𝑒𝑞.3) 𝛻∙−[𝜌𝜔𝑖∑𝐷 𝑖𝑘 𝑘𝜔𝑘 𝑀𝑘𝑀𝑛+1 𝑝𝐴{(𝜔𝑘 𝑀𝑘𝑀𝑛−𝑤𝑘)𝛻𝑝𝐴}+𝐷𝑖⊤𝛻𝑇 𝑇]+𝜌(𝑢∙𝛻)𝜔𝑖=𝑅𝑖 (𝑒𝑞.4) EFCF 2025: Low-Temp. Fuel Cells, Electrolysers & H2 Processing 1 – 4 July 2025, Lucerne Switzerland https://doi.org/10.5281/zenodo.17244117 A0717 Page 5/6 2. Results The results of the PFR reaction modeled using a 0D approach are shown in Figure 3. The figure presents the molar flow rates of each species with respect to the reactor volume. Initially, the reaction proceeds vigorously, resulting in a rapid decrease in ammonia and an increase in hydrogen and nitrogen. Around a reactor volume of 0.8 mL, the reaction reaches equilibrium, as observed in the plot. Figure 3. Molar flow rate of ammonia decomposition in PFR After confirming that the ammonia decomposition reaction was properly represented in the 0D component, modeling of the designed 3D base model was conducted. Ammonia is injected from the top (inlet), resulting in a high ammonia concentration near the top region. As the reaction occurs at the catalyst zone, the ammonia concentration decreases along the reactor length. In the case of hydrogen and nitrogen, since they are not present at the inlet, their concentrations are initially zero. However, as the reaction progresses toward the lower part of the reactor due to thermal effects, hydrogen and nitrogen are rapidly generated in the catalyst zone, leading to increased molar concentrations. The molar concentration of hydrogen reaches approximately three times that of ammonia, indicating that the 0D chemical reaction model has been successfully incorporated into the 3D simulation. Figure 4. CFD modelling results of ammonia decomposition in the base reactor EFCF 2025: Low-Temp. Fuel Cells, Electrolysers & H2 Processing 1 – 4 July 2025, Lucerne Switzerland https://doi.org/10.5281/zenodo.17244117 A0717 Page 6/6 3. Conclusion In this study, CFD modeling of a base model was conducted as a preliminary step for simulating the ammonia pre-treatment unit. The modeling was performed by applying the chemical reaction rate equation, the Brinkman equation, and the Maxwell–Stefan equation. As a result, it was confirmed that the ammonia decomposition reaction was appropriately represented in the simulation. The reliability of the modeling results will be validated through comparison with experimental data in future work. After validation, the geometry will be further refined to perform CFD modeling of the fully developed pre-treatment system. References [1] Kishimoto, M., et al, Development of 1 kW‐class ammonia‐fueled solid oxide fuel cell stack, Fuel Cells 20.1, 2020 [2] Wan, Zhijian, et al, Ammonia as an effective hydrogen carrier and a clean fuel for solid oxide fuel cells, Energy Conversion and Management 228, 2021 [3] Lee, Sangho, et al, Development of an ammonia decomposition reactor, afterburner and post-decomposition reactor for 1 kW solid oxide fuel cells using ammonia, Energy Conversion and Management 314, 2024 [4] Prasad, V., et al, Assessment of overall rate expressions and multiscale, microkinetic model uniqueness via experimental data injection: Ammonia decomposition on Ru/γAl2O3 for hydrogen production, Industrial & engineering chemistry research 48.11, 2009 [5] Wu, Gaowei, et al, Continuous flow aerobic oxidation of benzyl alcohol on Ru/Al2O3 catalyst in a flat membrane microchannel reactor: An experimental and modelling study, Chemical Engineering Science 201, 2019 Keywords: EFCF2025, H2, LowTemp. Fuel Cells & Electrolysers, Ammonia pre-treatment, Computational Fluid Dynamics, Modeling & Simulation, Optimization Remark: This work is licensed under Creative Commons Attribution 4.0 International