Contribution to the definition of non deterministic robust optimization in aeronautics accounting with variable uncertainties
- Autores: Jordi Pons Prats
- Directores de la Tesis: Gabriel Bugeda Castelltort (dir. tes.), Francisco Zárate (codir. tes.)
- Lectura: En la Universitat Politècnica de Catalunya (UPC) ( España ) en 2011
- Idioma: inglés
- Tribunal Calificador de la Tesis: Eugenio Oñate Ibáñez de Navarra (presid.), Roberto Maurice Flores Le Roux (secret.), Valentino Pediroda (voc.), Adel Abbas (voc.), Juan José Ródenas García (voc.)
- Materias:
- Texto completo no disponible (Saber más ...)
- Resumen
Shape optimization is a largely studied problem in aeronautics. It can be applied to many disciplines in this field, namely efficiency improvement of engine blades, noise reduction of engine nozzles, or reduction of the fuel consumption of aircraft. Optimization for general purposes is also of increasing interest in many other fields.
Traditionally, optimization procedures were based on deterministic methodologies as in Hamalainen et al (2000), where the optimal working point was fixed. However, not considering what happens in the vicinity of the defined working conditions can produce problems. That is, in many cases, if the real working point differs from the original, even a little distance, efficiency is reduced considerably as pointed out in Huyse and Lewis (2001).
Non deterministic methodologies have been applied to many fields (Papadrakakis, Lagaros and Tsompanakis, 1998; Plevris, Lagaros and Papadrakakis, 2005). One of the most extended non-deterministic methodologies is the stochastic analysis. The time consuming calculations required on Computational Fluid Dynamics (CFD) has avoided an extensive application of the stochastic analysis to shape optimization. Stochastic analysis was firstly developed in structural mechanics, several years ago. Uncertainty quantification and variability studies can help to deal with intrinsic errors of the processes or methods. The result is no longer a point, but a range of values that defines the area where, in average, optimal output values are obtained. The optimal value could be worse than other optima, but considering its vicinity, it is clearly the most robust regarding input variability.
Uncertainty quantification is a topic of interest from the last few years. It provides several techniques to evaluate uncertainty input parameters and their effects on the outcomes. This work presents a methodology to integrate evolutionary algorithms and stochastic analysis, in order to deal with uncertainty and to obtain robust solutions.
