This study compares the energy efficiency of a real battery pack and a simulated battery pack using a hardware-in-the-loop battery emulator, through experimental testing on a dedicated inertia dynamometer for light quadricycles. The investigated LiFePo4 battery pack has a nominal voltage of 48 V and a nominal capacity of 100 Ah. Initial characterization identified a reduced State of Health based on capacity (SOHC), with a measured usable capacity of 48 Ah (from the nominal 100 Ah) and a corresponding reduction in charge acceptance capability. The emulator was configured to replicate the degraded battery characteristics, including the open-circuit voltage (OCV)-SOC relationship, internal resistance, and current limitations, enabling a direct comparison between simulated and experimental dynamic behavior. The experimental setup was designed to overcome the limitations of conventional chassis dynamometers for low-mass, independent four-wheel-drive quadricycles operating without mechanical friction braking systems. Multiple driving cycles were reproduced using a PLC-based closed-loop control system, with data acquisition performed via CAN communication. The comparative analysis highlights significant differences during regenerative braking events. While the emulator accurately reproduces baseline electrical behavior under mild operating conditions, the aged battery exhibits strong limitations during both high-power acceleration and severe deceleration phases. In particular, increased internal resistance and transient electrochemical polarization lead to premature saturation of charge acceptance, resulting in rejection of high-frequency current transients. Consequently, the experimentally observed energy recovery is significantly lower than the theoretical values predicted by the emulator. In addition, due to the absence of mechanical braking, the reduced regenerative capability directly leads to speed tracking deviations, as the required braking torque cannot be fully achieved. These results identify battery degradation as a key physical constraint affecting both energy efficiency and dynamic braking performance, highlighting the importance of improved electro-thermal and aging-aware model calibration for realistic system-level simulations.