On Core and Curvature Corrections used in Straight-Line Vortex Filament Methods

TL;DR

This paper rigorously analyzes the convergence of curvature and core corrections in straight-line vortex filament methods, demonstrating that combining curvature correction with an improved core correction significantly enhances accuracy and convergence.

Contribution

The study introduces a high-order numerical filament method with rational B-splines and compares two core correction models, showing the improved correction extends accurate discretization.

Findings

Curvature corrections are essential for accuracy at coarse discretizations.

Combining curvature correction with the improved core correction improves convergence.

Original core correction without curvature correction fails to converge.

Abstract

The convergence characteristics of two viscous core corrections as used in straight-line segmentation methods are rigorously analysed. These are \emph{curvature corrections} that account for the induced velocity contribution at a point on a vortex filament due to local curvature and \emph{core corrections} that remove unrealistically high induced velocities near vortex segments. Two alternative versions of the latter correction are studied: the original as introduced by Scully and a recently introduced improved correction by the authors. The problem is analysed using a viscous vortex ring. A high-order numerical filament method is presented that uses rational B-spline curves to model the vortex ring geometry exactly. First, the separate influence of the two corrections on induced velocity values are studied using the numerical filament method. Afterwards, the corrections are combined in four separate cases that are studied using the segmentation method. It is found that in order to get accurate results at coarse discretisations, curvature corrections cannot be neglected. When used together with the original core correction model, results start to diverge from the reference values for discretisations smaller than approximately ten degrees. Combining the curvature correction with the improved core correction extends the discretisation region for which accurate results are found down by approximately one order. Finally, it is shown that segmentation methods that use the original core correction model without curvature correction completely fail to converge to the reference values.