In modern industrial practice, the quality of a new product is achieved by identifying characteristics critical to quality (CTQs) and minimizing deviation from targets, rather than merely optimizing CTQs in the absence of variation. Estimating the variation of CTQs is thus critical to understand and correctly manage risk caused by different interacting sources of uncertainty.

We have developed a method to estimate performance variability for centrifugal compressor stages. In this paper, we quantify the performance variation due to impeller manufacturing variability of two stage families. The stages studied are 2D stages designed for multi-purpose applications and 2D stages for high head applications. In a related paper, the stage performance variability here determined is considered together with other sources of variation to compute the variation in flange-to-flange performance for a full compressor.

The proposed approach propagates the uncertainty of the design parameters to the stage aerodynamic performance through a Monte Carlo method. In order to keep a low computational budget, calibrated 1D/2D aerodynamic models have been run in parallel to compute the performance of the stages with randomly modified geometry. The results allow the quantification of the stage performance variability and the identification of the main sources of variation.

As a step towards the corresponding analysis for a complete centrifugal compressor, results from this study have been used as input for a method where all factors affecting the flange-to-flange performance are considered. The method and the results are discussed in a companion paper.

This content is only available via PDF.
You do not currently have access to this content.