This paper presents the identiflcation of the rotordynamic force coefficients for direct lubrication flve-pad and four-pad tilting pad bearings. The bearing is ¡10 mm in diameter with a UD of 0.4 pad axial length (44 mm). The e.Kperiments include load-on-pad and load-between-pad conflgurations, with 0.5 and 0.6 pivot offsets, for rotor speeds ranging from 7500 rpm to 15,000 rpm. The bearing force coefficients are identifled from multiple frequency excitations (20-300 Hz) exerted on the bearing housing by a pair of hydraulic shakers and are presented as a function of the excitation frequency and rotor speed for a 300 kPa unit load. The experimental results also include temperatures at the trailing edge of three pads. The direct force coefficients, identifled from curve-fits of the complex dynamic stiffness, are frequency independent if considering an added mass term much smaller than the test device modal mass. The force coefficients from the four-pad bearing load-between-pad conflguration show similar coefficients in the loaded and orthogonal directions. On the other hand, as expected, the flve-pad bearing load-on-pad shows larger coefficients (-25%) in the loaded direction. The maximum pad temperature recorded for the 0.5 pivot offset conflgurations is up to 20° C higher than those associated to the 0.6 offset conflguration. Results from a predictive code are within 50% of the experimental results for the direct stiffness coefficients and within 30% for the direct damping coefficients.

The following paper presents and compares rotordynamic force coefficients for three types of non-contact annular gas seals, which include a labyrinth (LABY), honeycomb (HC), and a fully partitioned damper seal (EPDS). These three annular seals represent the typical seal types used in process gas centrifugal compressors at the balance piston location or center seal location to limit internal leakage and ensure a robust rotordynamic design. Tests were conducted on 170.6 mm (6.716 in) diameter seals for rotor speeds up to 15 kprm, inlet air pressure of 6.9 bar (100psi), ambient back pressure, and with inlet gas preswiri. The three seals were designed to have the same nominal clearance and similar axial lengths. Testing was conducted on a controlled motion test rig possessing non-synchronous excitation capability up to 250 Hz. Three different test methods were employed to give confldence in the rotordynamic coefflcients, which include static force deflection tests, mechanical impedance tests, and dynamic cavity pressure tests. Results from experiments compare force coefflcients for all seal conflgurations while paying special attention to the crossover frequencies of the effective damping term. All seals possessed negative effective damping at lower excitation frequencies with inlet preswiri, where the straight-through EPDS possessed the lowest cross over frequency of 40 Hz at 15 krpm. The testing also revealed that the preswiri parameter had significantly more influence on effective damping levels and crossover frequencies when compared to rotor speed.

The following paper presents and compares rotordynamic force coefficients for three types of non-contact annular gas seals, which include a labyrinth (LABY), honeycomb (HC), and a fully partitioned damper seal (FPDS). These three annular seals represent the typical seal types used in process gas centrifugal compressors at the balance piston location or center seal location to limit internal leakage and ensure a robust rotordynamic design. Tests were conducted on 170.6mm (6.716 in) diameter seals for rotor speeds up to 15kprm, inlet air pressure of 6.9 bar (100 psi), ambient back pressure, and with inlet gas preswirl. The three seals were designed to have the same nominal clearance and similar axial lengths. Testing was conducted on a controlled motion test rig possessing non-synchronous excitation capability up to 250Hz. Three different test methods were employed to give confidence in the rotordynamic coefficients, which include static force deflection tests, mechanical impedance tests, and dynamic cavity pressure tests. Results from experiments compare force coefficients for all seal configurations while paying special attention to the cross-over frequencies of the effective damping term. All seals possessed negative effective damping at lower excitation frequencies with inlet preswirl, where the straight-through FPDS possessed the lowest cross over frequency of 40Hz at 15krpm. The testing also revealed that the preswirl parameter had significantly more influence on effective damping levels and cross-over frequencies when compared to rotor speed.

In the current industrial research on centrifugal compressors, manufacturers are showing increasing interest in the extension of the minimum stable flow limit in order to improve the operability of each unit. The aerodynamic performance of a compressor stage is indeed often limited before surge by the occurrence of diffuser rotating stall. This phenomenon generally causes an increase of the radial vibrations, which, however, is not always connected with a remarkable performance detriment. In case the operating curve has been limited by a mechanical criterion, i.e., based on the onset of induced vibrations, an investigation on the evolution of the aerodynamic phenomenon when the flow rate is further reduced can provide some useful information. In particular, the identification of the real thermodynamic limit of the system could allow one to verify if the new load condition could be tolerated by the rotordynamic system in terms of radial vibrations. Within this context, recent works showed that the aerodynamic loads due to a vaneless diffuser rotating stall can be estimated by means of test-rig experimental data of the most critical stage. Moreover, by including these data into a rotordynamic model of the whole machine, the expected vibration levels in real operating conditions can be satisfactorily predicted. To this purpose, a wide-range analysis was carried out on a large industrial database of impellers operating in presence of diffuser rotating stall; the analysis highlighted specific ranges for the resultant rotating force in terms of intensity and excitation frequency. Moving from these results, rotordynamic analyses have been performed on a specific case study to assess the final impact of these aerodynamic excitations.

One of the main challenges of the present industrial research on centrifugal compressors is the need for extending the left margin of the operating range of ihe machines. As a result, interest is being paid to accurately evaluating the amplitude of the pressure fluctuations caused by rotating stall, which usually occurs prior to surge. The related aerodynamic force acting on the rotor can produce subsynchronous vibrations, which can prevent the machine's further operation, in case their amplitude is too high. These vibrations are often contained due to the stiffness of the oil journals. Centrifugal compressor design is, however, going towards alternative journal solutions having lower stiffness levels (e.g., active magnetic bearings or squeeze fllm dampers), which will be more sensitive to this kind of excitation: consequently, a more accurate estimation of the expected forces in the presence of dynamic external forces such as those connected to an aerodynamically unstable condition is needed to predict the vibration level and the compressor operability in similar conditions. Within this scenario, experimental tests were carried out on industrial impellers operating at high peripheral Mach numbers. The dedicated test rig was equipped with several dynamic pressure probes that were inserted in the gas flow path; moreover, the rotor vibrations were constantly monitored with typical vibration probes located near the journal bearings. The pressure fleld induced by the rotating stall in the vaneless diffuser was reconstructed by means of an ensemble average approach, thus deflning the amplitude and frequency of the external force acting on the impeller. The calculated force value was then included in the rotordynamic model of the test rig: the predicted vibrations on the bearings were compared with the measurements, showing satisfactory agreement. Moreover, the procedure was applied to two real multistage compressors, showing notable prediction capabilities in the description of rotating stall effects on the machine rotordynamics. Finally, the prospects of the proposed approach are discussed by investigating the response of a real machine in high-pressure functioning when different choices of journal bearings are made.