| dc.description.abstract | Thrust collars (TCs) are bearing elements used in geared machinery that transmit axial loads from one shaft to another. TCs are primarily used in integrally geared compressors (IGCs), but have applications in gearboxes, and marine propulsion. Accurate modeling requires knowledge of the film characteristics such as lubricant cavitation, flow turbulence, and air ingestion, all of which could potentially reduce load capacity. Current models in the literature do not properly account for lubricant cavitation or flow turbulence within the TC. The following dissertation introduces a new experimental set-up that optically characterizes the oil film in a thrust collar, along with two predictive models.
The test rig geometries, speeds, and loads match those seen in a typical industrial IGC. The test rig utilizes a transparent acrylic window in conjunction with a high speed camera to capture images of the oil film. Images are filtered and averaged to obtain areas of interest in the oil film. Lubricant cavitation and flow turbulence areas are measured for pinion speeds of 2.5, 5, and 7.5 krpm, and axial loads of 0.5, 1, and 1.5 kN. Cavitation occurs in the diverging (upper) region of the TC and appears at pinion speeds over 5,000 rpm, but does not change in shape after that speed. Turbulence at the inlet region (bottom) occurs at all speeds, but increases to almost 35% of the total area at the highest speed. The turbulence area decreases as the axial load increases.
The dissertation advances a finite element (FE) model that accounts for elastic deformation of the disks and uses a mass conserving function to model lubricant cavitation. Elastic deformation only becomes important to predicting load capacity at high loads (above 2 kN for a TC made of acrylic). At higher loads (>2kN, 23.2 bar specific load), the predicted load capacity can be significantly lower when considering elastic deformation, even for a TC made of steel.
The dissertation also includes the first computation fluid dynamics (CFD) model predictions on a TC. The CFD model predicts a larger cavitation area than the FE model, or the experimental results (35-43% for CFD and 26% - 33% for FE). The CFD model predicts flow turbulence in the lower region that increases for increasing spin speed, which matches the experimental results, but tends to overpredict the total area.
Experiments include testing at various oil inlet velocities. As the oil inlet velocity approaches the surface speed of the TC, the turbulence area increases, until the oil velocity matches or exceeds the surface speed, at which point the flow turbulence drops off. CFD modeling matches this trend, but also tends to overpredict the total turbulence area. As IGCs move into new application areas to satisfy new needs, the increase in efficiency and capacity comes at a cost of more load and higher speed requirements on the TCs. This work will help original equipment manufacturers model TCs more accurately to ensure safe and efficient operating conditions. | en |
