This thesis presents the results of high supply (up to 84 bar) pressure testing of hole-pattern annular gas seals performed at the Texas A & M Turbomachinery Laboratory in College Station, TX. The test variables were chosen to determine the influence of pressure ratio, rotor speed, and negative preswirl on seal performance. Preswirl signifies the circumferential fluid flow entering a seal, and negative preswirl indicates a fluid swirl in the direction opposite of rotor rotation. Changes in pressure ratio had only small effects on most rotordynamic coefficients. Cross-coupled stiffness showed slightly different profiles through the mid-range of excitation frequencies. Pressure ratio showed some influence on direct and cross-coupled damping at low excitation frequencies. Rotor speed significantly affected both cross-coupled stiffness and cross-coupled damping. As rotor speed increased, the magnitude of cross-coupled rotordynamic coefficients increased due to the positive fluid swirl induced by rotor rotation. For the low rotor speed, negative inlet preswirl was able to overpower the positive rotor induced fluid rotation, producing a negative cross-coupled stiffness. This outcome showed that, for hole-pattern seals, positive fluid swirl does indeed produce positive cross-coupled stiffness coefficients and negative fluid swirl produces negative cross-coupled stiffness coefficients. The addition of negative preswirl greatly reduced cross-coupled rotordynamic coefficients, while direct rotordynamic terms were unaffected. Cross-over frequency signifies the excitation frequency where effective damping transitions from a negative value to a positive value with increasing excitation frequency. Peak effective damping was increased by 50 percent and cross-over frequency reduced by 50 percent for high-negative preswirl versus zero preswirl results. This led to the conclusion that a reverse swirl could greatly enhance the stability of hole-pattern balance piston seals. A two-control-volume model that uses the ideal gas law at constant temperature (ISOT) was used to predict rotordynamic coefficients and leakage. This model predicted direct rotordynamic coefficients well, but greatly under predicted cross-coupled rotordynamic coefficients especially at high negative preswirls. The model predicted seal leakage well at low pressure ratios, but showed increasing error as the pressure ratio was increased. These results showed that the prediction model could not adequately estimate cross-coupled rotordynamic coefficients for a hole-pattern seal with negative inlet preswirl and requires modification to do so.

In high-pressure centrifugal compressors, honeycomb seals are often used as replacements for labyrinth seals to enhance dynamic stability. A concern exists with the loss of this enhanced stability if the honeycomb cavities become clogged with debris over time. So, as a first objective, static and dynamic tests were conducted on a constant-clearance honeycomb and a constant-clearance smooth-bore seal under three inlet preswirl conditions to determine the effects of inlet preswirl. The resulting leakage flowrate and dynamic parameters, effective stiffness and damping of the seal, were measured for each seal and then compared, with the smooth-bore seal representing the honeycomb seal with completely clogged cells. The second objective was to evaluate a two-control volume theory by Kleynhans and Childs with the measured data under the influence of preswirl. Both seals have a 114.7mm bore with a radial clearance of 0.2mm from the test rotor. The honeycomb seal has a cell width of 0.79mm and cell depth of 3.2mm. The target test matrix for each preswirl setting consisted of three exit-to-inlet pressure ratios of 15%, 35%, and 50%, and three rotor speeds out to 20,200 rpm. The target inlet air pressure was 70 bar-a. Experimental results show that, for a clean honeycomb seal, preswirl has little effect on effective stiffness, Keff*, and decreases effective damping, Ceff*, by about 20% at the high inlet preswirl ratio (~0.6). However, comparing smooth and honeycomb seal results at higher inlet preswirl shows a potential reduction in Keff* by up to 68%, and a large drop and shift in positive Ceff* values, which could cause an instability in the lower frequency range. Measured leakage shows a potential increase of about 80%, regardless of test conditions. A swirl brake at the seal entrance would fix this loss in stability by significantly reducing inlet preswirl. The two-control-volume theory model by Kleynhans and Childs seems to follow the frequency-dependent experimental data well for the honeycomb seal. Theory predicts conservatively (under-predicts) for stability parameters such as k* and Ceff* and for leakage. Predictions for K and Keff may possibly be improved with better measured friction factor coefficients for each seal.