A Python-Based Framework for Computationally Efficient Trim and Real-Time Simulation Using Comprehensive Analysis

摘要

This paper describes an open-source framework to use a rotorcraft comprehensive analysis with a geometrically exact beam model of elastic blades and free wake for real-time simulation. No simplification is performed for the rotor dynamics or flight dynamics to achieve real-time execution. Instead, several multicore acceleration strategies are identified and employed with load-balanced parallelization algorithms to achieve this goal. Up to 24 times speed-up for trim and 90 times speed-up for time marching were demonstrated for a single-rotor system with four blades, allowing for 5 deg azimuthal time steps. Heterogeneous computing for accelerated analysis with free wake was also explored as a preliminary step toward real-time wake modeling. Time calculation speed-ups of 23 and 29 were obtained with a graphics processing unit (GPU) for a single rotor and coaxial rotor, respectively, compared to serial execution on CPUs. Lag-free communication between the analysis and a pilot interface is provided through a Python framework.