Beyond Best-Fit: Spatial Compensation of Metrology Data

Abstract

Technological progress over the last decade has allowed metrology professionals to efficiently collect accurate and comprehensive datasets on large complex structures. With contemporary software, analysts distill meaningful results with unprecedented accuracy and speed. Specifically, automated feature extraction and cross sectional analysis allows analysts to report critical details efficiently. However, despite the increased speed and accuracy of the metrology technologies, the stiffness and stability of the structures limit our ability to achieve global tolerances without costly and time-consuming efforts.

To achieve the global best-fit tolerance requirements, these structures are typically inspected in either a manually adjusted support condition or on a costly holding fixture. Adjusting large, flexible structures to achieve global profile tolerances is notoriously expensive and time-consuming. When seeking to reduce this expense, two accepted metrology techniques are employed; profile at station lines and profile-per-unit-length (PPUL). Both techniques allow for verification in an indiscriminate support condition and can show an approximation of the optimally adjusted structure.

Profile at station lines locally fits cross sections extracted at incremental planes along the length of the structure. Unfortunately, these profiles are insensitive to potentially unacceptable deviations perpendicular to the station planes. These deviations can go undetected unless the global profile tolerance is controlled, which requires a return to costly global adjustment or fixturing. PPUL, where a structure is sectioned into locally-fit overlapping segments, can show lengthwise waviness and an approximation of the adjusted condition. However, the piecewise, rigid-body nature of the fit creates artificial discontinuities at the segment overlaps.

While the aforementioned techniques achieve acceptable results with mitigated cost and time commitment, they fail to explicitly address the flexibility of the structure. To solve this problem, our team has developed Spatial Compensation to produce a realistic simulation of a globally-adjusted structure independent of the global support condition. We have ensured validity by incorporating the structural characteristics (weight, area-moment of inertia and material stiffness) into the algorithm which compensates the measured data along the length of the structure.

The algorithm has degrees of freedom for fitting data in the local Y and Z axes and rotationally about the lengthwise X axis. Additional constraints can be applied to represent linear weight or clamping loads which, in combination with the physical properties of the structure, limit the solution to physical configurations achievable through manual adjustment. Empirical verification of the Spatial Compensation process was performed on a test article and the results are presented. The Spatial Compensation solution has been implemented through a user-interface within Spatial Analyzer.

Spatial Compensation successfully produces a simulation of the optimally adjusted support condition without masking deviations that would remain given realistic load limits. Spatially compensated metrology data is smooth and continuous while remaining sensitive to lengthwise waviness, resulting in higher quality results that expose subtle, often unidentified deviations. Spatial Compensation improves data quality while eliminating expensive holding fixtures or time-consuming adjustment operations, reducing cost and accelerating verifications.