A computational study of wingtip vortex flowfield

On numerical errors and turbulence modeling in tip vortex flow prediction

“…the vortex is diffused faster compared with experiment. The authors found from previous analyses [3][4][5] that the first and foremost requirement to reduce the numerical diffusion was to modify the turbulence model production terms such that the eddy viscosity at the vortex core is a minimum instead of a maximum value. The improvement of the core axial velocity and static pressure comparisons was readily apparent (i.e an excess core axial velocity is obtained, as opposed to a deficit [4]).…”

“…Higher-order-accurate differencing schemes have been shown to reduce numerical diffusion errors considerably in inviscid vortex preservation tests [17]. Early calculations [3] showed that a third-order-accurate scheme is too diffusive. This was evident from a grid refinement study of the subset problems, which are discussed in the Section 4.…”

“…This shortcoming can be overcome by modification of the production term, suggested by Spalart [20], and implemented in both of these models [3,4]. The production term P is approximated by P= C l wS.…”

“…This paper is part of a comprehensive computational and experimental study to accurately compute the formation of a turbulent wingtip vortex [3][4][5][7][8][9][10]. In this paper, the numerical errors and numerical diffusion from turbulence models in tip vortex predictions are discussed.…”

“…The flow field generated by a wingtip is often turbulent and highly three-dimensional, with existence of 'local separation' and secondary flows [3][4][5]. Thus, accurate prediction of the flow presents some challenges numerically and experimentally.…”