B.Sc. · Completed

Pedro Morel Rosa

Data-Driven Spalart-Allmaras Model Correction for Transitional Aeroelastic Instabilities

Profile

B.Sc.

Status

Completed

Date

01 Mar 2026

Length

83 pages

Institution

Universidade Federal do Rio de Janeiro

Advising

Advisor: Anjos, G. R.

Document

Navigation

Student work

Data-Driven Spalart-Allmaras Model Correction for Transitional Aeroelastic Instabilities

Abstract

This work comparatively investigates different adjustments to the SpalartAlmaras turbulence model through corrective control parameters obtained by datadriven techniques. From a physical perspective, the main objective is to obtain an accurate description of thelaminar separation flutterphenomenon, which does not present satisfactory results when simulated using classical turbulence methods and models. The complexity of the phenomenon and its highly localized character are used in order to promote a comparative study of configurations that differ in the type of modeling and nature of the data used for correction (parietal data or full field data). This investigation shows that it is possible to obtain a corrected model for Spalart-Almaras that accurately provides the aerodynamic forces intrinsic to the phenomenon. However, the data choices and the nature of the correction are not trivial and different advantages and shortcomings are explored until an optimal configuration is obtained. To this end, adjoint sensitivity analyzes are constantly used, in order to provide guidelines in terms of physical understanding of the correctness and effectiveness of the models obtained

Advising

Advisor: Anjos, G. R.