Experimental Determination of Mechanical Stress Induced by Rotating Stall in Unshrouded Impellers of Centrifugal Compressors

Abstract

Unshrouded centrifugal compressor impellers typically operate at high rotational speeds and volume flow rates. The resulting high mean stress levels leave little margin for dynamic excitations that can cause high-cycle fatigue. In addition to the well-established high-frequency impeller blade excitations of centrifugal compressors caused by the stationary parts, such as vaned diffusers or inlet guide vanes (IGVs), the presented study addresses an unsteady rotating flow feature (rotating stall) which should be taken into account when addressing the high-cycle fatigue during the design phase. The unsteady fluid–structure interaction between rotating stall and unshrouded impellers was experimentally described and quantified during two different measurement campaigns with two full-size compression units operating under real conditions. In both campaigns, dynamic strain gauges and pressure transducers were mounted at various locations on the impeller of the first compression stage. The casing was also equipped with a set of dynamic pressure transducers to complement the study. Rotating pressure fluctuations were found to form an additional impeller excitation at a frequency that is not a multiple of the shaft speed. The measurements show that the excitation amplitude and frequency caused by the rotating pressure fluctuations depend on the operating conditions and are therefore challenging to predict and consider during the design phase. Furthermore, the excitation mechanism presented was found to cause resonant impeller blade response under specific operating conditions. For the experimentally investigated impeller geometries, a rotating pressure fluctuation caused approximately 1.5 MPa of additional dynamic stress in the structure per 1 mbar of dynamic pressure amplitude when exciting the first bending mode of the impeller. The induced dynamic mechanical stresses due to rotating stall are in the order of 10% of the endurance limit of the material for the tested impeller geometries; therefore, they are not critical and confirm a robust and reliable design.