Rotordynamic Performance of a Negative Swirl Brake for a Tooth on Stator Labyrinth Seal

Abstract

In the late 1970s, Benckert and Wachter (Technical University Stuttgart) tested labyrinth seals using air as the test media and measured direct and crosscoupled stiffness coefficients. They reported the following results: (1) fluid preswirl in the direction of shaft rotation creates destabilizing crosscoupled stiffness coefficients and (2) effective swirl brakes at the inlet to the seal can markedly reduce the crosscoupled stiffness coefficients, in many cases reducing them to zero. In recent years, “negativeswirlâ€‌ swirl brakes have been employed, which attempt to reverse the circumferential direction of inlet flow, changing the sign of the crosscoupled stiffness coefficients and creating stabilizing stiffness forces. This study presents test results for a 16tooth labyrinth seal with positive inlet preswirl (in the direction of shaft rotation) for the following inlet conditions: (1) no swirl brakes, (2) straight, conventional swirl brakes, and (3) negativeswirl swirl brakes. The negativeswirl swirlbrake designs were developed based on computational fluid dynamics (CFD) predictions. Tests were conducted at 10.2, 15.35, and 20.2 krpm with 70 bar of inlet pressure for pressure ratios of 0.3, 0.4, and 0.5. Test results include leakage and rotordynamic coefficients. In terms of leakage, the negativeswirl brake configuration leaked the least, followed by the conventional brake, followed by the nobrake design. Normalized to the negativeswirl brake configuration, the conventionalbrake and nobrake configurations mass flow rate was greater, respectively, by factors of 1.04 and 1.09. The directstiffness coefficients are negative but small, consistent with past experience. The conventional swirl brake drops the destabilizing crosscoupled stiffness coefficients k by a factor of about 0.8 as compared to the nobrake results. The negativeswirl brake produces a change in sign of k with an appreciable magnitude; hence, the stability of forward precessing modes would be enhanced. In descending order, the directdamping coefficients C are: noswirl, negativeswirl, and conventionalswirl. Normalized in terms of the noswirl case, C for the negative and conventional brake designs is, respectively, 0.7 and 0.6 smaller. The effective damping Ceff combines the effect of k and C. Ceff is large and positive for the negativeswirl configuration and near zero for the nobrake and conventionalbrake designs. The present results for a negativebrake design are very encouraging for both eyepacking seals (where conventional swirl brakes have been previously employed) and divisionwall and balancepiston seals, where negative shunt injection has been employed.