PDF e-Pub

## Section: New Results

Mesh adaptation is a powerful tool to obtain accurate aerodynamic simulations at limited cost. When the simulation is aimed at the accurate calculation of aerodynamic outputs (forces, moments) goal-oriented methods based on the adjoint vector of the output of interest are often advocated. The calculation of the total derivative $dJ/dX$ of the aerodynamic function of interest, $J$, with respect to the volume mesh coordinates, $X$, has been extended to the case of an unstructured grid. The software developments have been validated for inviscid and laminar viscous flows, and implemented in the ONERA code (elsA). Then a local sensor $\theta$ based on $dJ/dX$ was devised to identify areas where the location of the volume mesh nodes has a strong impact on the evaluation of the output $J$. The sensor has been shown to be adequate in different flow regimes (subsonic, transonic, supersonic), for internal (blade and nozzle) and external (airfoils, wings) aerodynamic configurations. The proposed method has been compared to a well-known goal-oriented method (Darmofal and Venditti, 2001) and to a feature-based method; it yields comparable results at lower cost in simple configurations. A publication is currently subject to minor revisions.