Quantification of manufacturing uncertainties for an axial compressor blade in CFD


Any technical application is affected by a range of uncertainties. These can be variations of the environmental conditions, which impact the systems behaviour. But also, the inevitable tolerances of the manufacturing processes involved can have a large impact on the final performance. In turbomachinery CFD, deterministic evaluations are common since a long time, however the uncertainties involved have been neglected. In NUMECA’s FINE™/Turbo a dedicated module for UQ has been developed, which was recently extended to cope with geometrical uncertainties as well. The UQ method adopted uses a non-intrusive probabilistic collocation method, which means that no modifications to the flow solver itself are necessary. To allow for a larger number of concurrent uncertainties, a sparse grid formulation for the collocation points is included as well. Using this methodology, the rotor blade of a transonic axial compressor stage is analysed in detail. The impact of a variety of geometrical deviations on the performance of the compressor is evaluated, along multiple points of the speed line – these are near choke, best efficiency and near stall conditions. Furthermore, both the computational effort and numerical accuracy of the UQ module are assessed as well, showing that an industrial application of such UQ is feasible. Also, an inverse design problem is derived, which allows to calculate the necessary manufacturing tolerances for a required performance variability.

Thomas Hildebrandt, NUMECA Ingenieurbüro
Tino Biebrach, TU Dresden
Sven Albert, NUMECA Ingenieurbüro

The paper will be published at the 15th International Probabilistic Workshop.
Date: 27 – 29 September 2017
Location: TU Dresden, Germany