Experimental Force Coefficients for a Fully-Partitioned Pocket Damper Seal and Comparison to Other Two Seal Types

Abstract

Labyrinth seals (LSs), honeycomb seals (HCSs) and hole-pattern seals, pocket damper seals (PDSs), and fully-partitioned damper seals (FPDSs) serve as balance pistons on the discharge side of barrel-type centrifugal compressors. These seals, facing large pressure differentials along with significant changes in gas density, produce sizeable lateral forces that impact compressor stability and rotordynamic performance. There is extensive archival data on the dynamic forced performance of LS and textured surface seals. However, the experimental database for a FPDS is insufficient, thus a comprehensive comparison of their dynamic force performance against that of textured surface seals is missing. The paper presents experimentally derived rotordynamic force coefficients for a FPDS along with a direct comparison to published data for a HCS and a LS, both similar in size and in operating conditions. The dynamic load tests with the FPDS include operation at a shaft speed (Ω) of 10 krpm (rotor surface speed of 60 m/s) while supplied with air at Pin = 70 bar and Tin =10 °C, and a discharge at Pout = 0.25, 0.5, and 0.65 of Pin. The preswirl circumferential velocity at the seal inlet is low, approximately 6% of rotor surface speed. The FPDS having eight pockets × eight axial blades produces a direct stiffness (K) increasing with excitation frequency (ω), though K < 0 for the lowest pressure ratio PR = (Pout/Pin) = 0.25. The cross-coupled stiffness (k) is invariant to frequency whereas the direct damping (C) steadily decreases as ω/Ω → 1, and then increases for larger frequencies. Hence, the effective damping coefficient (Ceff = C − k/ω) is approximately constant for ω < Ω, while increasing for higher frequencies. For the three PRs, the FPDS leakage is roughly up to 26% larger than the HCS leakage, while the 20-blade LS leaks in between both. The comparison of results shows the LS has insignificant forces compared to those from the HCS and FPDS, both producing a comparable Ceff and similar crossover frequencies (Ceff > 0). The HCS produces a large direct stiffness (K), three to four times that of the FPDS. Comparison of computational fluid dynamics (CFD) predictions for the FPDS against the experimental force coefficients shows significant differences, in particular an overestimation of direct stiffness as the frequency grows along with a lower direct damping for frequencies equal and above the shaft speed. Besides manufacturing considerations, a FPDS is a better alternative to a HCS whose large centering stiffness (K) may affect the compressor's critical speed, hence shortening the separation margin with respect to the operating speed.