A Residual Smoothing Strategy for Accelerating Newton Method Continuation

Abstract

A technique for accelerating global convergence of pseudo-transient continuation Newton methods is proposed based on residual smoothing. The technique is motivated by the effectiveness of local nonlinear smoothers at overcoming strong nonlinear transients. In the limit of a small pseudo-time step, the method reduces to a local nonlinear smoothing technique, while in the limit of large pseudo-time steps, an exact Newton method is recovered along with its quadratic convergence properties. The formulation relies on the addition of a smoothing source term while leaving the Newton Jacobian matrix unchanged, thus simplifying implementations for existing Newton solvers. The proposed technique is demonstrated on a steady-state and an implicit time-dependent computational fluid dynamics problem, showing significant gains in overall solution efficiency.