Abstract
The experimental setup for a hole-pattern seal is modeled using computational fluid dynamics (CFD) and results compared with measured test data and bulk flow model (ISOTSEAL) predictions. The inlet swirl boundary condition for prior CFD analyses of this test case have either been assumed or based on pitot-tube measurements. In this paper, the validity of each is investigated by including radial inlet nozzles with the inlet plenum in the model geometry. A transient mesh deformation technique with multiple frequency journal excitations is used to determine frequency-dependent rotordynamic coefficients. This multifrequency excitation method is validated against single frequency sinusoidal journal excitation. An empirical limit on the number of frequencies that can be packed in a multifrequency excitation signal to provide a reasonable estimate of rotordynamic coefficients is provided. Rotordynamic coefficients estimated using CFD compare well with measured rotordynamic coefficients. For the given test data, the ISOTSEAL bulk flow model does not provide good correlation for cross-coupled stiffness if the measured swirl ratio at the inlet of the seal is used in the prediction. However, improvement in correlation for cross-coupled stiffness is obtained if the swirl ratio found from CFD analysis is used in the bulk flow model, indicating that pitot-tube measurements of swirl may not be accurate.
