Natural hydrogen is a promising zero-carbon resource, but its small molecular size and high diffusivity make nuclear magnetic resonance (NMR) logging interpretation difficult. This study develops a pore-network and random-walk simulation framework to analyze hydrogen and methane responses in hydrogen-rich sandstones. The model couples pore structure, fluid properties, magnetic-field gradients, and acquisition parameters. Transverse relaxation time (T2) spectra are inverted with Tikhonov regularization and generalized cross-validation. The effects of waiting time (TW), echo time (TE), pore size, gradient strength, and burial depth are quantified. Results show clear T2 separation among water, oil, methane, and hydrogen. Dual-TW difference spectra improve fluid discrimination, and hydrogen shows the strongest TW sensitivity. TW mainly controls amplitude, while TE and gradient mainly control diffusion attenuation. Because hydrogen diffuses faster, it is more sensitive than methane to both factors. Greater depth and larger pores shift peaks to longer T2, but strong diffusion attenuation weakens hydrogen’s pore-size sensitivity. These results clarify the distinct NMR response mechanism of hydrogen. These findings can directly support NMR logging workflows for natural hydrogen reservoir identification by guiding TW–TE parameter design and gradient control, and by improving H2-CH4 discrimination in mixed-gas intervals. They can also reduce interpretation uncertainty in deep, heterogeneous formations, thereby improving confidence in identifying and evaluating natural hydrogen reservoirs.
Le et al. (Wed,) studied this question.