Improving the convergence rate of rotorcraft loose coupling algorithms using interaction laws based on autoregressive modelling

Problem area In many dynamical systems, the physics of the overall system can be modelled as the interaction between several subsystems, each governed by its own laws of physics. A prime example of this is rotorcraft ight, where aerodynamics and structural mechanics are closely coupled. In order to simulate the dynamics of rotorcraft ight, a widely used approach is a Loose Coupling, in which a structural dynamics model and a (computationally expensive) uid dynamics model are combined with a simplified aerodynamics model. However, even with a loose coupling approach the uid dynamics model needs to be solved many times before a coupled solution is found. The goal of the present work is to reduce the amount of computational work needed for accurate rotorcraft aeroelastic simulation. Description of work A one-dimensional model problem is constructed that resembles the rotorcraft simulation in a very simple way. It is expected that established techniques from time series analysis can be used to improve the convergence of the coupled models. Using the model problem various methods for improving the loose coupling are evaluated. Results and conclusions The time series resulting from the convergence of the loosely coupled procedure can accurately be modelled as an autoregressive process. This model is subsequently used to estimate the final solution of the iterative procedure, improving the overall convergence rate. Applicability The described methods can be applied to rotorcraft simulations in order to reduce the amount of computational work required or, alternatively, use more accurate models for the same amount of computational work. This Master's Thesis provides a basis for further research into the applicability of time series analysis on rotorcraft loose coupling in particular and partitioned procedures in general.