Parallel Unstructured Three-dimensional Turbulent Flow Analyses Using Efficiently Preconditioned Newton-Krylov Solver

A recently developed parallel three-dimensional flow solver of the turbulent Navier-Stokes equations is presented. The implicit integration process of the discrete equations is based on an effectively preconditioned Newton-Krylov subspace algorithm. The space discretization of the continuous equations is performed by means of the Fluctuation Splitting algorithm. This method can be viewed as a mixed finite volume/finite element discretisation of the system of conservation laws. The presented Computational Fluid Dynamics solver has a very high computational performance. Such a performance as well as the accuracy of the code are highlighted by considering a suite of two- and three-dimensional test cases, among which the calculation of the flow past the DPW-W1 wing configuration of the 3rd AIAA Drag Prediction Workshop. This analysis is carried out with a grid featuring about 1 million nodes. The use of Newton-Krylov subspace solvers in implicit CFD codes has a great potential for dramatically reducing the computational cost associated with the analysis of complex aerodynamic problems. Included results also highlight the remarkable computational effectiveness of Newton's solution procedure in the case of unsteady flows. The theoretical and numerical results reported herein will contribute to highlight this potential and enrich the database of this technology.