Effect of methane/water flowrate on waste heat recovery and hydrogen production by steam methane reforming process and predicted by artificial neural network (ANN)

Hydrogen production is essential for the development of clean energy technologies, particularly in transportation and power generation. However, traditional hydrogen production methods are often limited by high energy demands and inefficiencies. Steam methane reforming is a widely adopted method for hydrogen production, with its efficiency largely dependent on optimizing key operating parameters such as temperature, pressure, and reactant flowrates. This study investigates the waste heat recovery and hydrogen production using exhaust heat under operating conditions of a 20 % hydrogen fraction, 18 % EGR (exhaust gas recirculation) ratio, 75 % engine load, and 1200 rpm engine speed in a stoichiometric compressed natural gas spark ignition engine. ASPEN Plus® software was used to simulate exhaust heat utilization with different methane/water flowrates (1–4 kmol/h). The findings reveal that increasing methane and water flowrates from 1 to 4 kmol/h leads to a proportional rise in heat duty, with heat exchanger (HX-A) ranging from 2.8 to 12.59 kW and heat exchanger (HX-B) from 16.41 to 63.59 kW. Temperature significantly influences reforming efficiency, as hydrogen production increases from 2.2706 kmol/h at 973 K to 2.95811 kmol/h at 1273 K, while unconverted methane declines from 0.27175 kmol/h to 0.01473 kmol/h, indicating improved methane conversion at elevated temperatures. In contrast, raising reformer pressure from 1 to 30 bar decreases hydrogen production from 5.79688 to 3.49486 kmol/h and raises unreacted methane levels from 0.07263 to 0.88786 kmol/h, demonstrating the equilibrium constraints that suppress reforming reactions at higher pressures. Accurate prediction of hydrogen production in steam methane reforming is essential for enhancing efficiency and sustainability. This study also investigates the use of artificial neural networks to predict hydrogen yield based on key operational parameters, including methane and water flowrates, reformer pressure, temperature, and heat duty. The best performance was achieved by combining all input parameters C-15 (methane/water flowrate, pressure, temperature & heat duty) with a mean square error (MSE) = 0.0001 and a regression coefficient (R2) is 0.999. The findings contribute to the development of AI-driven strategies for optimizing hydrogen production.