Abstract
The objective of this work is to study, in a first part, the impact of mesh adaptation on computational estimates of lift and drag coefficients. The convergence of these quantities with successive adaptation is demonstrated by comparing with experimental results. As a second part, optimization of relevant adaptation parameters to accelerate the convergence is investigated. A combination of adaptation variables is proposed to better capture some physical features that are poorly represented when a single variable is used. On the other hand, a new error metric is derived from both Hessian and gradient that are used separately in the framework of mesh adaptation in general. The impact of the proposed improvements is demonstrated through test cases.