An optimization method based on the use of the derivatives of functional outputs with respect to (w.r.t.) solid body mesh nodes is presented. These derivatives are obtained by a discrete adjoint method that first computes the derivatives of functional outputs w.r.t. all volume mesh nodes. They are smoothed before being used in a numerical optimization algorithm. The procedure is demonstrated for a 2D flow governed by the compressible Reynolds-Averaged Navier-Stokes equations (RANS) completed by the Spalart-Allmaras turbulence model. Discrete derivatives are computed with or without making the frozen eddy-viscosity assumption. The design algorithm is compared with a more classical one using design variables related to B-splines on the four test cases introduced by Kim et al.1