Electrical engines are becoming more common than thermal ones. Therefore, there is an increasing interest in the characterization of batteries and in measuring their state of charge, as an overestimation would cause the vehicle to run out of energy and an underestimation means that the vehicle is running in suboptimal conditions. This is of paramount importance for flying vehicles, as their endurance decreases with the increase in weight. This work aims at finding a novel empirical model for the discharge curve of an arbitrary number of battery pack cells, that uses as few tunable parameters as possible and hence is easy to adapt for every single battery pack needed by the operator. A suitable measurement setup for battery tests, which includes voltage and current sensors, has been developed and described. Tests are performed on both constant and variable power loads to investigate different real-world scenarios that are easy to reproduce. The main achievement of this novel model is indeed the ability to predict discharges at variable power based on a preliminary characterization performed at constant power. This leads to the possibility of rapidly tuning the model for each battery with promising accuracy. The results will show that the predicted discharged capacities of the model have a normalized error below 0.7%.