Jurriaan Boon

This work integrates two intensified processes, the sorption-enhanced steam reforming (SESR) of biogas and sorption-enhanced dimethyl ether synthesis (SEDMES), offering a route for the efficient production of bio-dimethyl-ether (bio-DME). A full process simulation was developed in Aspen Plus, integrating a 40 MW SESR unit with a SEDMES system. Optimisation of the bio-DME plant increased the cold gas efficiency from 64% to 74% and the overall DME yield from 39% to 45%. A base-case configuration was established and improved through three integration strategies: (i) heat recovery, (ii) recycling of unconverted gases, and (iii) steam integration through the regeneration of the SEDMES adsorbent. The levelised cost of DME (LCODME) for the optimised system is estimated at ∼1 €/kg DME, which is lower than costs reported for alternative power-to-X DME routes. The net present value analysis indicates that plant profitability becomes positive for DME selling prices above ∼1000 €/t. This study demonstrates that intensifying both hydrogen production and DME synthesis enables a technically and economically attractive pathway for bio-DME production from biogas.

Ammonia, widely regarded as the “hydrogen carrier of the future,” offers high hydrogen content, ease of production, and a well-established infrastructure for handling and transportation globally. Meanwhile, ammonia cracking requires a heat supply at high temperatures, and induction heating provides efficient, precise, and rapid heating to conductive materials of different shapes and sizes. Therefore, this work presents a proof of concept for ammonia cracking using induction heating with three different reactor configurations: (1) a 3D metal workpiece; (2) a 3D metal workpiece and Ni/Al₂O₃ catalyst; and (3) only Ni/Al₂O₃ catalyst. The performance of the inductively heated reactor is also compared to an electric furnace. The results showed that the reactor with the workpiece and the catalyst required 97 W to reach 650 °C, being the most efficient in terms of power usage when compared to the workpiece alone and the electric tube furnace, which required 39% and 132% more, respectively; the least efficient configuration is with just the catalyst, needing 138 W to reach just 116 °C. Overall, the introduction of the 3D workpiece allowed for fast and uniform conversion and heating within the reactor, enabling efficient and dynamic process control when applying induction heating to chemical reactors.

Ammonia, widely regarded as the "hydrogen carrier of the future," offers high hydrogen content, ease of production, and a well-established infrastructure for handling and transportation on a global scale. Meanwhile, ammonia cracking requires heat supply at high temperatures and induction heating provides efficient, precise, and rapid heating to conductive materials of different shapes and sizes. Therefore, this work presents a proof of concept for ammonia cracking using induction heating with 3 different reactor configurations: (1) a 3D metal workpiece; (2) a 3D metal workpiece and Ni/Al2O3 catalyst; and (3) only Ni/Al2O3 catalyst. The performance of the inductively heated reactor is also compared to the one using an electric furnace. The results showed that the reactor configuration containing both the workpiece and the catalyst was the most efficient in terms of electric power usage to achieve high temperatures quickly; the least efficient configuration is the one with just the catalyst. While the workpiece surface showed minor structural changes after time on stream, the system’s performance was not affected. Overall, the introduction of the 3D workpiece allowed for fast and uniform conversion and heating within the reactor enabling efficient and dynamic process control when applying induction heating to chemical reactors.

Physics-informed neural networks (PINNs) have emerged as a promising alternative to purely data-driven neural networks (NNs) for surrogate modeling, particularly in data-scarce scenarios. This study evaluates the performance of hybrid-PINNs against traditional NNs for modeling the adsorption step of a Direct Air Capture (DAC) process. As the complexity of the modeled system increases, larger datasets and longer computational times are required for numerical methods. Therefore, the study aims to develop approaches that minimize data requirements while maintaining accuracy, which is crucial for efficient modeling of complex physical systems. While both AI models can achieve high accuracy with abundant data, the advantages of hybrid-PINNs become more evident as data becomes scarce. In the intermediate and low-data regimes, the physics constraints embedded in hybrid-PINNs significantly improve generalization and predictive accuracy. For extreme low-data conditions, a curriculum learning strategy is implemented, progressively enforcing physics constraints to mitigate underfitting and enhance model stability. Despite these benefits, hybrid-PINNs exhibit a computational cost approximately one order of magnitude higher than traditional NNs as enforcing physics constraints increases training complexity. The results suggest that PINNs hold potential for modeling complex multi-physics problems in DAC and beyond, provided challenges related to gradient balancing and computational efficiency are addressed.

Physics-Informed Neural Networks (PINNs) have emerged as a promising alternative to purely data-driven neural networks (NNs) for surrogate modeling, particularly in data-scarce scenarios. This study evaluates the performance of hybrid-PINNs against traditional NNs for modeling the adsorption step of a Direct Air Capture (DAC) process. As the complexity of the modeled system increases, larger datasets and longer computational times are required for numerical methods. Therefore, the study aims to develop approaches that minimize data requirements while maintaining accuracy, which is crucial for efficient modeling of complex physical systems. While both AI models can achieve high accuracy with abundant data, the advantages of hybrid-PINNs become more evident as data becomes scarce. In the intermediate and low-data regimes, the physics constraints embedded in hybrid-PINNs significantly improves generalization and predictive accuracy. For extreme low-data conditions, a curriculum learning strategy is implemented, progressively enforcing physics constraints to mitigate underfitting and enhance model stability. Despite these benefits, hybrid-PINNs exhibit a computational cost approximately one order of magnitude higher than traditional NNs, as enforcing physics constraints increases training complexity. The results suggest that PINNs hold potential for modeling complex multi-physics problems in DAC and beyond, provided challenges related to gradient balancing and computational efficiency are addressed.

Sorption-enhanced reverse water–gas shift (SE-RWGS), designated as COMAX, was studied using a Pt4A bifunctional catalyst (reactive adsorbent). The bifunctional Pt4A catalyst integrates CO₂ activation and reaction with water adsorption functionality, where the active phase is loaded onto a carrier that provides a surface area for Pt dispersion as well as H₂O adsorption capacity. The 0.3 wt% Pt-4A molecular sieve reactive sorbent was tested at a kg scale in a pressure swing (reactive) adsorption–regeneration process. More than 400 cycles over 50 days of operation were successfully demonstrated without significant decay. Cyclic stability was achieved, provided that the regeneration temperature was sufficiently high to ensure near-complete dehydration. The single-bead structure withstood the pressure swing operation effectively, with only a maximum of 2% of the total recovered reactive sorbent turning to fines (2 conversion.