Aerodynamic shape optimization of RAE-2822 airfoil by physics-informed parametric model embedding

The efficiency of optimization algorithms in engineering design is often hindered by the curse of dimensionality, where increasing problem complexity degrades al-gorithmic performance due to the high dimensionality of the design space. To address this challenge, dimensionality reduction techniques that minimize the number of variables or constrain their variability are essential. This study introduces physics-informed paramet-ric model embedding (PI-PME), a novel approach specifically designed for aerodynamic shape optimization of transonic airfoils. PI-PME extends the parametric model embedding (PME) framework by integrating physics-based insights, enhancing its capability to guide the optimization process toward physically relevant variations in the design space. By leveraging principal component analysis, PI-PME efficiently reduces dimensionality while maintaining the ability to reconstruct variables in the original space, avoiding the limitations of purely geometry-based formulations. To demonstrate its potential, PI-PME is applied to the optimization of the RAE-2822 airfoil at Mach 0.3, a benchmark case in aerodynamic design. Results are compared against PME and full-domain optimization , showing that PI-PME not only reduces computational complexity but also identifies superior configurations in the reduced design space. These findings underscore the practi-cality and robustness of PI-PME for industrial applications requiring efficient and reliable optimization strategies.