AIRCRAFT STRUCTURE AND DESIGN

Aircraft Structure and Design



Introduction

A significant level of interest has developed in recent years in the possibility of developing analysis capabilities with sufficient fidelity that they could be evolved into the primary source of information leading to the qualification or certification of a new aircraft structure for flight. In this design and acceptance environment, structural testing would evolve to a role of assuring correct analytical evaluations, rather than providing final clearance for flight using the traditional full-scale structural article limit load/ultimate load testing approach, especially of large final articles.

In this environment, testing would evolve into a new role of developing a sufficient database to model the way structure could vary over a production run. Testing would be used to fine-tune the analysis methods and models. These analyses would then provide the final check for clearance of a new design. The key to success is developing various procedures to raise analysis precision to this level. This effort explores the use of probabilistic methods in aircraft structural design to achieve the needed analysis precision. Examination is made of how test programs supporting the analyses would be restructured and, hopefully, reduced in cost when restructured to support analyses-based clearances Weimer 2003, Pp. 11-15.

Use of probabilistic methods in design has long been considered an intriguing way of introducing the unavoidable uncertainties in the characterizations of all our physical descriptors of aircraft structure. They allow the characterization of the uncertainties in modeling, materials, structural geometry, loads, etc. in the analyses such that the uncertainty of the analyses can be observed and addressed. The primary technical drawback to the use of probabilistic methods, for a long time, was the Monte Carlo methods traditionally used for calculating the probabilities of occurrences.

By their very nature, aircraft structures must have very low probabilities of failure in order to be acceptable for flight. Thus, the use of traditional Monte Carlo methods demanded a large number of analysis evaluations corresponding to all the variations needed for the design variables. Many evaluations were needed to generate enough failure cases to adequately define the "tail" of the probability of failure curve where acceptable values of aircraft structural failures reside. This would now be further aggravated by the large increases expected in analysis model complexity such as structural finite element analysis (FEA) models as the fidelity is increased to meet the objective of analysis carrying the burden of clearance for flight.

Developments in probabilistic methods have changed the analytical environment needed to calculate probabilities of failure of very reliable structure such as that found on aircraft. These methods, outlined in this paper, have the ability to project functional representations of the probability of failure with few analytical evaluations, as compared to that required by the older Monte Carlo methods. These methods, as shown in this paper, make the concept of characterizing structural failure in terms of probability of failure a practical reality.

The New Design Process

As the concept of probabilistic methods-based design is now computationally practical, the whole structural design ...