Convergence of Methods for Nonlinear Eigenvalue Problems (Q4363489)

First, the authors recall the classical numerical method for computation of aircrafts in dynamic aeroelasticity in the flutter-safe range. This method, called the ``\(p\)-\(k\) method'', consists in injecting in the linear structural equation of aerodynamics an assumed value of the reduced frequency \(k\) of unsteady aeroelasticity terms, and in computing the eigenvalue solutions \(p\). Then, comparing the imaginary part results with the assumed value, one iterates by sweeping the latter value until the equality is reached. In nonlinear aeroelasticity, the unsteady aeroelastic terms are multiplied by a parameter \(q\), to take account for any flight conditions. It appears that convergence is very slow, and that the radii of convergence for all the eigenvalues are very close to zero. The proposed method aims to accelerate the convergence by the use of an iteration of the derivative of the imaginary part of \(p\) with respect to \(k\). Some examples show the benefit of the method. It must be noted, a correct introduction of nonlinear aeroelastic terms requires to iterate values linearized from correct flight conditions, and not only to take into account the increase in the parameter \(q\).