Leakage and Dynamic Force Coefficients of a Pocket Damper Seal Operating Under a Wet Gas Condition: Tests Versus Predictions

摘要

High-performance centrifugal compressors presently favor pocket damper seals (PDSs) as a choice of secondary flow control element offering a large effective damping coefficient to mitigate rotor subsynchronous whirl motions. Current and upcoming multiple-phase compression systems in subsea production facilities must demonstrate long-term operation and continuous availability, free of harmful rotor instabilities. Plain annular seals and labyrinth (LABY) seals are notoriously bad choices, whereas a PDS, by stopping the circulation of trapped liquid, operates stably. This paper presents experimental and computational fluid dynamics (CFD) results for the leakage and dynamic force coefficients obtained in a dedicated test facility hosting a fully partitioned PDS (FPPDS), four ribbed and with eight pockets per cavity. The test PDS, operating at a rotor speed 5250rpm (surface speed 35m/s) and under a supply pressure/discharge pressure ratio up to 3.2, is supplied with a mixture of air and ISO VG 10 oil whose maximum liquid volume fraction (LVF) is 2.2%, equivalent to a liquid mass fraction of 84%. When supplied with just air (dry condition), the measured leakage increases nonlinearly with supply pressure. Under a wet gas condition, the recorded mass flow increases on account of the large difference in density between the liquid and the gas. CFD-derived mass flow rates for both dry and wet gas conditions agree with the measured ones. The test dry gas PDS produces a direct dynamic stiffness (H-R) increasing with frequency, whereas the direct damping (C) and cross-coupled dynamic stiffness (h(R)) coefficients remain relatively constant. The CFD-predicted damping agrees best with the test C albeit overpredicting H-R at low excitation frequencies and h(R) at all frequencies (<175Hz similar to twice rotor speed). Under a wet gas condition with LVF = 0.4%, the test force coefficients show great variability over the excitation frequency range; in particular, H-R<0, though growing with frequency due to the large liquid mass fraction. The CFD predictions, on the other hand, produce a dynamic direct stiffness H-R>0 for all frequencies. Both experimental h(R) and C for the wet gas PDS are larger than their counterparts for the dry gas seal. The CFD-predicted C and h(R), wet versus dry, show a modest growth, yet remaining lower than the test data. The CFD-derived flow field for a wet gas condition shows that the seal radial partition walls (ridges) reduce the circumferential flow velocity and liquid accumulation within a pocket. Both the test data and the CFD prediction show that the magnitude of the flexibility function for the PDS test system reduces when the two-component mixture flows through the seal, hence revealing the additional effective damping, more pronounced for the test data rather than that from the predictions. Further work, experimental and CFD based, will continue to advance the technology of wet gas seals while bridging the gap between test data and computational physics model simulations.