Achieving zero emissions across transportation is a tremendous challenge. The upcoming Euro 7/VII standards, set to be enforced in 2025, will mandate further reduction in ICEs exhaust emissions. Thus, additional improvements and potential new technologies and fuels are needed to design ultra-low emissions vehicles.
Hydrogen seems to be a very attractive fuel, thanks to its high lower heating value, clean combustion, and extremely low pollutant emissions, due to the zero-carbon content. Nevertheless, NOx emissions are still an issue in hydrogen fueled engines and optimized lean-burn combustion and suitable after-treatment NOx reduction are mandatory to reach high specific power and efficiency and near zero NOx emissions, thus enabling H2-ICE powered vehicles to be zero-impact emitting technology solution.
Selective Catalytic Reduction by using NH3 as the reducing agent is the most effective control technology for NOx abatement. Nevertheless, ongoing research and innovation are critical in developing new strategies for reducing NOx emissions, to overcome the NH3-SCR system main critical issues (extra equipment for urea storage and dosing, ammonia-slip, deactivation and fouling, low efficiency at low temperature).
The SCR of NOx by hydrogen is considered a promising alternative to traditional ammonia-based deNOx technology. The H2-SCR catalysts are suitable for engine exhaust after-treatment during cold-start and urban-driving operations, exhibiting higher catalytic activity for low-temperature NOx emissions conversion (180 – 200 °C). Furthermore, in H2-ICE powered vehicles, the additional tank to store the reducing agent is not needed with a significant simplification of the engine exhaust lay-out.
Experimental investigations on different H2-SCR catalysts have demonstrated that the conversion activity is greatly affected by the catalytic system configuration, the adsorption properties of the support material, the H2 spillover, that in turn are affected by H2/NO ratio and exhaust gas composition, temperature and flow rate.
In the present paper, a feed-forward neural network model of H2-SCR is presented, with the aim of performing real-time estimation of reduction efficiency and supporting the design of suitable H2 management to achieve maximum NOx reduction and minimum hydrogen consumption. Model training and identification are carried out against a large set of experimental data measured on a H2-SCR small scale prototype at the Synthetic Gas Bench. The experimental tests were designed to reproduce the real conditions expected at the exhaust of a H2-fueled engine.