A CFD-based multidisciplinary optimization of aeroelastic systems with coupled reliability constraints

M. Nikbay, M. N. Kuru, and P. Acar (Istanbul Technical University)

CHALLENGE - Taking into account reliability in the design of aero-elastic systems is a challenging problem in multi-disciplinary optimization. Using the traditional deterministic approach may lead to over-optimized but unreliable or inefficient solutions when dealing with uncertain variables.This paper presents applications of a Reliability Based Design Optimization (RBDO) approach based on high-fidelity CAD, CFD and CSD software coupling and fluid-structure interaction. The aim is to obtain a fully automatic design framework for aero-elastic optimization problems using an AGARD 445.6 aircraft wing as test case.

SOLUTION - A reliability based multidisciplinary optimization workflow is built in modeFRONTIER by coupling commercial solvers for aero-elastic analysis and an in-house developed code for reliability analysis. The geometry wing is modeled parametrically using Catia V5, loads are calculated performing the structural analysis with Abaqus 6.7.1. Aerodynamic performances are obtained solving the inviscid 3D Euler flow equations using the finite volume based flow solver Fluent. MpCCI, mesh based parallel code coupling interface, is utilized to exchange the pressure and displacement information between Fluent and Abaqus to perform a loosely coupled aero-elastic analysis. The multi-objective optimization process is driven by modeFRONTIER software while an in-house developed Matlab code performs an inner loop reliability analysis for probabilistic constraints.



BENEFITS -  In this benchmark problem, modeFRONTIER has been successfully used to obtain a fully automated multi-disciplinary aero-elastic optimization framework for reliability optimization. Results obtained in a benchmark case and in the real application show how performances are enhanced considering the uncertain variables. From the Pareto frontier the design chosen as optimum design satisfying the target reliability index constraints and the L/D ratio gives an increase in mass of 11.7% and 5.9% of L/D.