To achieve higher integrated nucleon-nucleon luminosity at the Large Hadron Collider (LHC) with lighter ions, as proposed by the ALICE detector collaboration, the CERN ion injector chain must deliver significantly higher beam intensities than for present lead (Pb) beams. Currently, the performance of Pb beams is limited by losses, the origin of which are not yet fully understood. In addition, the limitations for future lighter ions are largely unknown. This thesis addresses these challenges by quantifying key beam dynamics effects, developing predictive simulation models, and identifying optimal production schemes to maximize LHC luminosity production. The modelling efforts focus on understanding the main performance bottlenecks. First, a semi-empirical model for beam lifetime reduction caused by residual gas interactions was benchmarked via dedicated neutral gas injection experiments in the Proton Synchrotron (PS). Second, a detailed tracking model for the Super Proton Synchrotron (SPS) was developed to investigate the combined effect of space charge, intra-beam scattering (IBS), and power converter ripple inducing tune modulation. This model, validated against dedicated experiments, correctly predicted that a combination of these effects drives losses. These experiments, coupled with a newly implemented 50 Hz tune ripple compensation scheme, resulted in extracted Pb beam intensity exceeding the present operational goals by more than 40%. These findings were integrated into a semi-empirical Injector Model that propagates bunch intensities for any ion species through the accelerator complex. The model reveals that different ion species are limited by different effects in each machine. In an optimized production scenario, these studies predict that oxygen beams have the potential to increase the one-month integrated nucleon-nucleon luminosity by up to a factor of 4 compared to the baseline Pb case.