A PRECONDITIONED ALGORITHM FOR TURBOMACHINERY VISCOUS FLOW SIMULATION
Name: Xiao Wang
Date of Degree: December 9, 2005
Institution: Mississippi State University
Major Field: Computational Engineering
Major Professor: Dr. David L. Whitfield
Title of Study: A PRECONDITIONED ALGORITHM FOR TURBOMACHINERY VISCOUS FLOW
Pages in Study: 106
Candidate for Degree of Doctor of Philosophy
The MSU TURBO code, distributed to U.S. engine companies by NASA Glenn, is a heavily used parallel compressible Reynolds-averaged Navier-Stokes flow solver for multistage turbomachinery flows, primarily for compressible flows at subsonic and transonic speeds. Many low speed turbomachinery flows in aerial vehicles or marine propulsion systems can not be effectively addressed by compressible flow solvers. It is well known that compressible flow equations face difficulties at low Mach number due to the large disparity of the acoustic and convective wave speeds. The current study is to develop and implement the computational capability for flow simulations at low Mach number, or incompressible Mach regime, under the framework of the MSU TURBO code. This is accomplished by applying a global preconditioning scheme to the unsteady term of the compressible governing equations and solving the conservative Riemann flux based on primitive variables. The preconditioning scheme is a single parameter diagonal matrix depending on a reference Mach number which represents the global flow properties in the flow simulation. For flows in rotating machines where speed varies along the radial direction from the axis of the rotation, it is found that a modified preconditioning parameter is necessary to assure numerical stability in simulating low Mach number rotating flows. The effectiveness of the modified preconditioning scheme has been analyzed, under various flow conditions, through Fourier footprints and validated by numerical investigations.
The development of a preconditioned structured turbomachinery flow solver was accomplished in this dissertation. The conservative form of the governing equations were cast in the non-inertial relative rotating frame in terms of primitive variables and absolute velocity vectors. Characteristic-based boundary conditions, with implicit treatment of the source term resulting from the rotating relative frame, are derived for internal and external flows. The implicit finite volume scheme is developed for the preconditioned scheme with the flux Jacobians evaluated by either a flux approximate method or flux-vector-splitting. The viscous flux is also treated implicitly, and an analytic form of viscous flux Jacobians was developed in the preconditioned flow solver to reduce numerical uncertainties, and computing time.
A series of flow simulations have been carried out by this preconditioned unsteady turbomachinery flow solver. The simulations of viscous boundary layer development over flat plates at very low Mach numbers demonstrate the effectiveness of the preconditioning algorithm. Computations of compressor rotor, and rotor/stator at subsonic, and transonic flow regions with acceptable results indicate that the preconditioned TURBO solver is compatible with the compressible version of the TURBO solver for subsonic and transonic flows. Moderate improvement in numerical convergence for flows in a rotating frame with mixed flow speeds is observed in the case of a tiltrotor blade at hover. The marine propeller simulation demonstrates the accomplishment of the preconditioned TURBO solver for an incompressible flow simulation. In the simulation of a low speed centrifugal compressor, the preconditioned TURBO is able to predicate the wake locations accurately.
Advisor:David L. Whitfield; Chunhua Sheng; W. Roger Briley; Lafayette K. Taylor; Jen Ping Chen
School:Mississippi State University
School Location:USA - Mississippi
Source Type:Master's Thesis
Date of Publication:11/28/2005