Multidisciplinary Design Space Characterization of a Prediffuser, Strut, and Frame

Abstract

The multidisciplinary exploration and characterization of a design space challenges the boundaries of analysis capability and design process structure for many current industrial applications. In the present study, parallel independent structural and aerodynamic analysis of a prediffuser flow path, strut, and frame was performed in conjunction with design of experiments (DoE) techniques to efficiently characterize the multidisciplinary design space. The key interactions between geometric parameters and the sensitivity factors of each parameter in relation to stress output were calculated. Aerodynamic analysis was performed using full three-dimensional steady-state computational fluid dynamics. Key structural performance, aerodynamic performance, and weight drivers were identified through screening analysis using design of experiments. The use of common geometry definitions and an overlapping parameter set allowed for the multidisciplinary evaluation of parameters such as strut fillet radius and strut chord. In addition, it aided in the identification of trade-offs between structural performance and aerodynamic performance. Further design space characterization with respect to aerodynamic performance was performed through the use of a more detailed multilevel full factorial design of experiment. The sensitivity of the aerodynamic performance metrics to mesh resolution and turbulence models was statistically determined as well as the variation due to the strut design parameters using design of experiments techniques to address the uncertainty and variation due to the numerical analysis. Additional design space characterization and visualization were also achieved through the use of multidimensional response surfaces generated based on latin hypercube DoEs. All three levels of fidelity in DoE analysis (two-level screening, two-level full factorial, and latin hypercube) identified the same primary driver in aerodynamic performance. However, the advanced latin hypercube analysis was necessary to accurately characterize the secondary drivers of aerodynamic performance. Application of the advanced Design of Experiments techniques along with parallel aerodynamic and structural analysis has led to a better understanding of the multidisciplinary design space and helped to facilitate design decisions that impact both aerodynamic and structural performance.