A surrogate-based approach for uncertainty analysis of the ONERA 7A rotor

Abstract

An uncertainty quantification framework for rotorcraft applications is introduced. The capabilities of the developed approach are demonstrated using the ONERA 7A rotor at high-speed to model the uncertainties in blade properties including torsion, flap, and lag stiffness on rotor power and loads including torsion, flap bending, and chord bending moments. To support large-scale simulations which are needed to establish statistical convergence, a surrogate-based framework is introduced using an artificial neural network that is conceptualized, trained, and validated using data derived from rotorcraft comprehensive analysis code. The analysis characterizes the input uncertainties as aleatory, hence are normally distributed. Through propagation it is established that the uncertainties in rotor power are limited; moderate for peak torsion moment; and significant for peak flap bending and chord bending moments. The analysis further quantified that uncertainties in spanwise flap bending moment are present and are influenced by the variability in flap and torsion stiffness. The results demonstrate the integration of a probabilistic-based framework to a surrogate-based approach for the quantification of system uncertainties to facilitate informed decision making based on model-based predictions.