Statistical analysis of CFD solutions from the drag prediction workshop

TLDR

The study uses CFD solutions from the June 2001 Drag Prediction Workshop, where multiple developers and users ran RANS codes on the DLR‑F4 wing‑body configuration. The authors present a simple, graphical framework for robust statistical evaluation of N‑Version testing results of RANS CFD codes. The framework evaluates N‑Version test results for a single cruise design point, drag polars, and drag rise using the DLR‑F4 wing‑body configuration previously tested in wind tunnels. The authors conclude that the statistical framework offers insights into the predictability, validation, and reporting of CFD solutions.

Abstract

A simple, graphical framework is presented for robust statistical evaluation of results obtained from N-Version testing of a series of RANS CFD codes. The solutions were obtained by a variety of code developers and users for the June 2001 Drag Prediction Workshop sponsored by the AIAA Applied Aerodynamics Technical Committee. The aerodynamic configuration used for the computational tests is the DLR-F4 wing-body combination previously tested in several European wind tunnels and for which a previous N-Version test had been conducted. The statistical framework is used to evaluate code results for (1) a single cruise design point, (2) drag polars and (3) drag rise. The paper concludes with a discussion of the meaning of the results, especially with respect to predictability, Validation, and reporting of solutions.