Parity-Based methodologies for fault diagnosis in UAVs often result in nonlinear residual generators. Still, a systematic framework to perform detectability and robustness analyses of residual generators does not exist. In this work, detectability and robustness metrics for static and dynamic residuals are presented, while numerical methods, specifically Particle Swarm Optimization, are employed to calculate them. The results are used to characterize the performance of a fault detection system. An application on a UAV model is shown, based on real flight data.