Reliability-Based Multi-Discipli nary Optimization for Aerospace Systems

This paper presents approaches to formulate and solve multi-disciplinary optimization (MDO) problems wherein the system parameters (e.g., material properties, boundary conditions, loads, model prediction errors, etc.) are not necessarily deterministic and are described by probability distributions. For these problems the objective is to maximize system performance (e.g., payload, aerodynamic efficiency, etc.) while satisfying constraints that ensure reliable operation. Since system parameters are not necessarily deterministic, the objective function and constraints must be stated probabilistically. This class of problems is called reliability-based multi-disciplinary optimization (RBMDO) problems. The formulation used herein allows designers and analysts to solve such RBMDO problems in much the same way that modern MDO problems are posed, thereby enabling all the benefits of modern MDO but achieving more robust designs. The designs are optimal over the range of operating conditions that a system may be subjected to while providing a desired level of reliability. We have already designed and demonstrated a framework that packages the necessary RBMDO tools in a way that engineers and designers can use the CAE tools that they are already familiar with and do not need to be experts in probabilistic methods. This paper presents the results of several analytic example problems that test, verify, and show the advantages and robustness of our methodology and the framework. Further, we detail two full-scale wing design applications along with preliminary analysis results. This first problem is an RBMDO of a smart wing assembly and the second problem is a reliability-based structural optimization of a transport aircraft wing.