Capture of manufacturing uncertainty in turbine blades through probabilistic techniques

Description/Abstract

Efficient designing of the turbine blades is critical to the performance of an aircraft engine.
An area of significant research interest is the capture of manufacturing uncertainty in the
shapes of these turbine blades. The available data used for estimation of this manufacturing
uncertainty inevitably contains the effects of measurement error/noise. In the present work,
we propose the application of Principal Component Analysis (PCA) for de-noising the
measurement data and quantifying the underlying manufacturing uncertainty. Once the
PCA is performed, a method for dimensionality reduction has been proposed which utilizes
prior information available on the variance of measurement error for different
measurement types. Numerical studies indicate that approximately 82% of the variation in
the measurements from their design values is accounted for by the manufacturing
uncertainty, while the remaining 18% variation is filtered out as measurement error.