Preliminary Wing Weight Estimation Under Probabilistic Loads for a Transport Aircraft

Abstract

In preliminary design of aircraft wings, a common practice is to set target loads and target weights as governing parameters for more detailed design phases. These values must be reliable in the presence of uncertainty because overly conservative target estimates could result in over-design of the wing while non-conservative estimates could lead to costly latephase redesign to satisfy constraints. The process of confidently identifying these targets is typically very computationally expensive because a large multivariate space of flight scenarios and vehicle configurations must be investigated. Therefore, a method has been implemented to reduce the computational expense of reliability-based aero-structural preliminary design. To efficiently identify critical loading conditions for structural sizing, this method adaptively samples the discretized multivariate space with a kriging-based sequential routine. Each new sample point is identified by an expected improvement function based on the probabilistic distributions of internal structural loads. These distributions are determined by propagating uncertainty using Monte Carlo simulation through an aerostructural analysis code. This process is applied to the wing of a reference aircraft derived from the NASA Common Research Model, and the effect of reliability indices is shown on preliminary sizing and weight estimation.

Update on the Development of a Flutter Analysis Capability for Unconventional Aircraft Concepts using HCDstruct

“…Researchers across government, academia, and industry are leveraging developments in these areas to build physics-based structural weight prediction and aeroelastic analysis tools for advanced aircraft, with several examples found in Refs. [2][3][4][5][6][7][8][9][10][11]. NASA is actively engaged in these efforts through the development of in-house software as well as through external research grants.…”

Parametric reduced order model approach for rapid dynamic loads prediction

“…It was shown that considerable savings in computational time can be made without sacrificing accuracy; however, the IQs were dealt with independently. Recent work [9], [10] have investigated approaches to determine reduced order models of correlated loads based upon the use of the Singular Value Decomposition (SVD), whereas a robust performance approach was considered in [11]. Little work has considered correlated gust responses.…”