Dean Billheimer, Applied Statistics, Boeing Phantom Works
"Statistical Methods for Shape Analysis."
Thursday, February 18, 1999
Communications 120 at 3:30 P.M.

Abstract

Boeing manufactures many parts and assemblies with complicated geometric shapes. These complex shapes are not easily described by a single or small number of critical dimensions. Consequently, it is difficult to evaluate manufacturing variability, and apply standard methods of statistical process control (SPC).

The Applied Statistics project Statistical Methods for Shape Analysis aims to develop statistical methods and visualization tools for evaluating variability in parts with complicated geometric shape. The long-term goals of this project are to:

• Characterize design shape with allowable tolerance.
• Evaluate manufacturing variability of parts with complex shape.
• Describe and predict part deformation in assembly and use.
• Develop tools for engineers and mechanics to apply this technology in design and manufacturing.

I will present an overview of the shape analysis project, and a case-study evaluating functional data analysis to assess variability in the 777 fuselage cross-section. If time allows, I will describe recent advances in incorporating part flexibility in statistical analysis.