Optimize systems for reliable performance.
Efficient optimization of coupled thermal-structural-fluid systems
Engineered systems such as aircraft, cars, buildings, and bridges are all subject to “nature’s changing course untrimm’d” and require a high degree of reliability, with potentially disastrous consequences otherwise. However, traditional approaches for designing reliable systems such as safety factors or conservative material properties are limited in scope and may lead to overly conservative designs with decreased performance as well as under conservative designs with insufficient reliability.
In contrast, reliability-based design optimization (RBDO) directly takes uncertainties into account to generate a design that achieves a desired reliability target. However, the solution of RBDO problems is challenging due to the need to estimate small failure probabilities while performing nonlinear optimization for computationally expensive quantities in high dimensions (10–100 variables).
xStar Research has extensive experience in developing and using methods for efficient reliability-based design optimization of complex engineering systems. Previous original research has included an adaptive kriging algorithm for efficient and accurate reliability analysis and a provably convergent optimization algorithm for solving the RBDO problem which takes advantage of multifidelity models. Both algorithms are state-of-the-art in performance and were used to design a reliable supersonic nozzle using a coupled fluid-thermo-structural analysis. For more information, see here.
Slice of an optimal and reliable supersonic nozzle.