This paper presents a complexity-based methodology for the design of aero engine components. Upon a rigorous definition of complex system, a metric for the complexity is introduced as a function of system’s topology and entropy. As a consequence, complexity becomes a measurable and manageable property of systems. Furthermore, a novel definition of robustness is provided, based on the shape of the probability density functions (PDF) of the performances. Complexity and robustness are related together by a simple, qualitative law. Based on these premises, two algorithms are introduced, namely the Stochastic Design Improvement (SDI) and the Complex Systems Analyzer (CSA). The former searches the design space seeking for solutions which meet the design requirements. The latter extracts the fundamental features of the design, previously perturbed by means of Monte Carlo Simulation (MCS). The SDI is proposed as a competitor of the practice of optimization. Though both can be used separately, the combination of SDI and CSA provides a powerful novel method for design. The capabilities of the algorithms are illustrated on three test-cases, namely an LPT Casing, a Turbo-prop bearing retainer and an LPT disk. It is important to point out that response surfaces or other surrogates have never been used.

No abstract