Aero-Structural Optimization of a Transonic Compressor Rotor
Yongsheng Lian · 2005
This paper presents a framework for multi-objective and multidisciplinary design optimization using highfidelity analysis tools.In this framework the aerodynamic performance is evaluated based on a Navier-Stokes equation solver, and the structure dynamics is computed using commercially available finite element software.We employ a genetic algorithm as a robust design optimization tool to facilitate the multi-objective optimization.We also use the response surface approach to tackle the difficulties associated with the organizational complexity and computational burden inherent in the multidisciplinary optimization.The coupling between the fluid solver and structural solver is realized through a thin-plate spline interpolation algorithm.The proposed approach is then used to perform aerostructural optimization of a three-dimensional transonic compressor blade.Our numerical results show that this method can improve the existing design and reduce the required computational time by orders of magnitude. I. IntroductionW ITH the advancement of computational power and compu- tational methods, researchers have used optimization techniques to improve the performance of complex system, such as aircraft engine.In this instance, Oyama et al. minimized the entropy generation of the NASA rotor67 blade, 1 Benini improved the total pressure ratio and the adiabatic efficiency of the NASA rotor37 blade, 2 Mengistu and Ghaly 3 performed multipoint design of different compressor rotors to improve their aerodynamic performance, and Lian and Liou 4 carried out multi-objective optimization of the NASA rotor67 blade.These analyses were focused on a singlediscipline response, namely, the aerodynamic aspect.However, compressor design is inherently multidisciplinary, and a successful design should involve a combination of a variety of disciplines including aerodynamics, structure dynamics, acoustics, and control theory. 5In addition, the present design procedures are usually based on sequential discipline optimization, which might be insufficient to provide a satisfactory result.The resulting solution might satisfy some, but not all of the requirements.In that case, the coupled multidisciplinary optimization (MDO) design technique is required.The applications of MDO to compressor designs give rise to considerable challenges.First, the computational expense associated with MDO is usually much higher than the sum of the costs associated with each single-discipline optimization.Second, organizational complexity imposes another challenge. 6For example, different analysis codes can run on different machines at different sites.For these two reasons, a direct coupling of an optimizer with multidisciplinary analysis tools might be practically difficult, especially when a large number of design variables and computationally intensive tools are involved.Third, noisy or jagged response from some disciplines deteriorates the coupling effect and can lead to local