Measured Static and Rotordynamic Characteristics of a Circumferentially-grooved-stator/smooth-rotor Liquid Annular Seal: Influence of Viscosity, Eccentricity, Pressure, and Speed

Circumferential grooves are often machined into annular seals to reduce leakage rates (???). For this study, a circumferentially-grooved annular seal (CGS) was tested with three different viscosity (?) test fluids. In addition to varying ?; differential pressure (?P), rotor speed (?), and static eccentricity (?o) were varied. Tests are conducted to determine how static and rotordynamic characteristics of a CGS are affected by ? changes. The measured results are compared to predictions, and a smooth seal when applicable. Testing was conducted with three test fluids, ISO VG 2 (turbulent flow regime), VG 46 (laminar and turbulent flow), and VG 100 (laminar flow) oils. Speed was varied from 2-8 krpm, ?P across the seal 2.07-8.27 bar, and ?o from centered (0.00) to 0.80. Geometry of the CGS has a radial clearance (Cr) of 0.1905mm (7.5mils) with 15 equally spaced square grooves with a groove length and groove depth of 1.52 mm. Length to diameter ratio (L/D) is 0.5. Test fluid is supplied circumferentially and in the direction of rotor rotation to introduce preswirl at the seal inlet. Increasing ? produced the following results: (1) decreasing ??? , (2) increasing direct and effective damping, and (3) increased direct virtual-mass. Direct stiffness was often negative. Cross-coupled stiffness was opposite in sign (destabilizing). For an ESP with CGSs, increasing ? is expected to lower the natural frequencies of the rotor system. Additionally, increasing ? is not expected to result in a rotor instability. The VG 46 test results were significantly different from those of VG 2 and VG 100. Results for VG 46 at low ?, consisted of a low phase angle (angle between the applied static load vector and ?o), small cross-coupled stiffness, and small whirl frequency ratio. While these VG 46 results are consistent with themselves, this behavior was not observed in the VG 2 and VG 100 test results. For the turbulent flow seal, grooves have the effect of reducing the rotordynamic coefficients, and reducing the dependence of the rotordynamic coefficients on ?o (changes were generally small with changing ?o) when compared to a smooth seal. Static and rotordynamic characteristics were poorly predicated by the available turbulentflow seal analysis code. For the laminar flow seal code used, ??? , cross-coupled stiffness (k), and direct damping (C) were reasonably predicted. Direct stiffness (K), cross-coupled damping (c), and static load were poorly predicted