22nd Congress of International Council of the Aeronautical Sciences, Harrogate, UK, 28 August - 1st September, 2000
Paper ICAS 2000-2.4.5
ACCURACY OF GRADIENT COMPUTATIONS FOR AERODYNAMIC SHAPE OPTIMISATION PROBLEMS
M. Chevalier, M. Berggren
The Aeronautical Research Institute of Sweden
Keywords: shape optimization, adjoint equations
Applying nonlinear optimization techniques such
as quasi-Newton methods to aerodynamic shape
optimization problems requires the calculation
of gradients of a given objective function. An
effective way of calculating such gradients is
through the use of the so-called adjoint equations.
To achieve fast convergence in the optimization
algorithm, accurately computed gradients
are needed. In the computation of such gradients
the discretization of the problem and the
choice of boundary conditions are two important
aspects. These issues are studied in the context
of shape optimization of a quasi-1D nozzle using
physically relevant boundary conditions. Isentropy
is enforced at the inlet boundary, and the
static pressure is specified at the outlet boundary
for subsonic flows. A cell-centered finitevolume
discretization with a standard implementation
of the boundary conditions is applied, and
the corresponding numerical scheme and numerical
boundary conditions for the adjoint equations
are derived in a fully discrete sense.
Numerical experiments at subsonic and transonic
speeds, show that the gradient evaluations
are accurate enough to obtain satisfactory convergence
of the quasi-Newton algorithm.
view full paper