Optimal nozzle design using finite element
conjugate gradient based software
Z. Mrsa, G. Medic
Technical Faculty, University ofRijeka, Vukovarska 58,
Abstract
The finite element incompressible viscous flow solver, with discretization of
total derivative, has been originally developed. Optimization with conjugate
gradient method, with finite difference approximation of gradient of objective
function and specially tailored mesh generator for chosen shape parametrization
in the form of Bezier curve segments, has been used for optimal shape design
of the nozzle. Almost identical optimal shape was achieved using different
polynomial curve basis and different initial parameter values, qualifying the
proposed methodology as appropriate for shape optimization in viscous flow
problems.
1 Introduction
The shape optimization technique is today accepted engineering tool. It is a
promising method for the engineer designer because it can supplement his
experience and guide him when designing new geometry or trying to improve
the efficiency of the existing one. For the simple geometry and simple flow
models, such as 2D airfoils or wing profiles, the shape optimization methods
are widely used, but they remain quite unused for complex configurations or
complex flow models. For 2D domains and potential flow models a good
theory exists for a shape optimization, Pironneau [1] and a lot of papers can be
found in literature, e.g. Weeber & Hoole [2], Mrsa [3], Bugeda et al. [4], Mrsa
& Sopta [5] and Sopta & Mrsa [6]. Approach of using a cheap direct flow
solver and very general robust constraint methods is accepted in 3D industrial
aerodynamics design, Fol et al [8] A review of recent advances for modeling of
aeronautical flows can be found in Dervieux [7], where genetic algorithms have
Transactions on Ecology and the Environment vol 12, © 1996 WIT Press, www.witpress.com, ISSN 1743-3541