The continued reliance on fossil fuels, particularly natural gas, has driven atmospheric CO2 levels higher, and intensified the urgency of decarbonising industrial energy use. Hydrogen and ammonia offer pathways to deliver net‑zero emissions while maintaining the reliability industries demand. However, the existing infrastructure and energy demands are not built to accommodate an abrupt, full-scale transition to hydrogen-powered systems. They must evolve gradually, using efficient, low-emission combustion systems that blend hydrogen with natural gas to balance performance and pollutant emissions. In this work, we present large-eddy simulations of a large-scale industrial burner (6m in length) operating with methane-hydrogen blends, using the Lattice Boltzmann method (LBM). The burner employs partially premixed combustion technology, and features a central fuel pilot and annular premixed fuel-air mixture injection through a perforated cylindrical holder into the combustion chamber. Simulations were carried out for three different fuel blends. Sub-grid combustion dynamics are modelled using a thickened flame approach, radiative heat transfer with the P1-WSGG model, and a low-Mach number approximation was used to reduce computational cost. A single-step chemistry is employed while accounting for intermediate species such as CO, while NOx formation is evaluated in post-processing using the frozen flow field. Results were compared with experimental in-flame and radiation measurements. Good agreement between experiments and simulations have been obtained, particularly for O2 and NOx evolutions, and radiative fluxes. Furthermore, the P1 radiation model implementation was verified and validated by showing that the radiative transfer balance closes (i.e., consistent divergence of radiative heat flux and boundary radiative fluxes). A significant over-prediction of CO mass fraction was noted, which is addressed by integrating CO oxidation into the reaction mechanism. Finally, by benchmarking against ANSYS Fluent results (RANS), we demonstrate that LES-LBM provides both the accuracy and computational feasibility needed to guide the design of next-generation, low-emission industrial burners.