|
|
ROMAC Rotating Machinery and Controls Laboratory |
|
|
|
|
Fluid Flows High-speed Centrifugal Compressor Student : Nathan Brown Advisor: Dr. Eric Maslen Project Start Date: July 2000 Project Objectives: The primary objective of this project is to design and construct a test rig for measuring impeller reaction forces in a high-speed centrifugal compressor. There is currently a design that employs active magnetic bearings in the motor / compressor assembly to measure these forces. One can determine the force vector imparted by the flow on the impeller by monitoring the individual currents drawn by each of the poles of the magnetic bearings during operation. The problem with the current design is that the 125 horsepower (@ 25,000 RPM) electric motor will have its own internal forces that will interfere with the forces from the compressor impeller, and the magnetic bearings will sense the combination of these forces instead of those only from the impeller. I plan to replace the magnetic bearings in the motor with conventional rolling contact bearings and separate the compressor and the motor into two different assemblies. The magnetic bearings will be dedicated exclusively to the compressor, and some type of flexible coupling will connect the motor to the new impeller shaft. The coupling will be such that it will only transmit torsional forces from the motor to the pump with no radial or axial components. This will allow the magnetic bearings to provide the correct readings of the impeller reaction forces. Progress in the Last Year: I have found suitable rolling contact bearings to replace the magnetic bearings in the motor and redesigned the motor shaft to accommodate the smaller bearings and the shortened length of the motor assembly. The motor requires a cooling system to expel its waste heat, and I created a spreadsheet to determine optimal dimensions for the motor water jacket using analytical heat-transfer methods. I was able to shorten the heat exchanger length significantly, which ill anslate to a greatly reduced fabrication cost. Finding a suitable coupling has proven to be a harder task than I had initially expected. Most conventional couplings are not rated for these high speeds, and those that can withstand the speed are extremely expensive. I have been in contact with a team of German graduate students working on a similar test rig and various coupling manufacturers and will have this issue resolved as quickly as possible.
Student: Arnaud Habsieger Advisor: Ron Flack Project Start Date: 1988 Sponsor: General Motors Corp. Research Objectives: Torque converters are common turbomachines used in all kind of transportation means like cars, buses, locomotives, and construction vehicles as a mean of enabling torque transmission between engines and gear-boxes. Also applied as soft clutches to engage gas turbines and for the propulsion of compressors, torque converters provide torque amplification during start-up conditions and they act as damper for engine torsional disturbances and shock loads. The highly complex and three-dimensional internal fluid flow characteristic is not fully understood. It is however very important to improve torque converter designs to increase their efficiency. Nowadays, the lack of knowledge of the internal flow field makes the numerical modeling and design of these turbomachines very difficult. The current state of theoretical analysis cannot accurately and reliably predict the internal flow field in torque converters. So the crucial design parameters cannot be determined solely with computational fluid dynamics results. Therefore, experimental three-dimensional flow data, including velocity distributions and vorticities, must be obtained to verify the computational results and the fundamental parameters of each new design. The non-intrusive technique of laser Doppler velocimetry is used to accurately measure fluid velocities in turbomachines. This measurement method is also flexible enough to measure velocities within all passages of the torque converter. Information about the internal velocity flow field can be used in several ways in torque converter design:
In addition, any improvement made in the laser velocimeter technique will aid in the analysis of other complex turbomachines. Within an automobile, the car engine is directly connected to the pump shaft. Torque converter pump acts like a centrifugal pump and energizes the working fluid within the torque converter itself. The turbine operates much like a radial inflow turbine converting the energy from the fluid into a rotational motion. The turbine shaft is connected the car’s transmission shaft. The fluid exiting the turbine is redirected towards the pump through the stator. The stator is designed to have the flow entering the pump with zero incidence angle at a specific design condition. Being a closed system, the same fluid is continuously moving throughout the entire system. The speed ratio, is defined as:
where w t
and w p are the turbine and pump rotational speeds, respectively.
When the car is stopped, Project Overview In 1988, General Motors Corporation, The University of Virginia and Pennsylvania State University Began a large-scale research project to study the internal flow field of a automotive torque converter. Later, Michigan Technological University became a partner in the project. The development of experimental and analytical tools to determine the internal complex flow field of torque converter and the effect of the torque converter on its performance parameters was the goal of the project. The same torque converter geometry was studied at similar operating conditions at all three research laboratories. Using laser velocimetry, the internal flow field were measured at the University of Virginia. At Pennsylvania State University, the presure in the flow was measured and the cavitation studies were performed at Michigan Technological University. General Motors developed at the same time a three-dimensional Navier-Stokes flow code to reproduce the experimental results. In 1990, Bahr, Flack, By, and Zhang [1] used a one-directional burst type laser velocimeter to measure the three-dimensional velocity flow field of an automotive torque converter stator. A torque converter machined entirely out of Plexiglas was employed for optical access. By aligning the laser velocimeter with the three Cartesian coordinates, all three velocity components were obtained in separate measurements. Measurements were taken at the inlet, quarter, mid, three-quarters, and exit planes on a 5-by-5 measurement grid at both the 0.065 and 0.800 speed ratios. These measurement showed separation regions at the mid plane pressure and core sides and at the exit plane suction and shell sides at the 0.800 speed ratio. Flow rate and torque distributions were found to be highly dependent on radial position. Undesirable torque distributions were presented in the core to shell direction. In 1992, Gruver [2] measured average and position dependent velocities in a torque converter pump using the same test facility. Measurements were made at the inlet, mid, and exit planes for speed ratios of 0.065 and 0.800 and three dimensional velocity distributions were obtained for all three planes. The pump inlet flow field was found to be influenced by the relative stator-pump position, while the relative pump-turbine position influenced the pump exit flow. Although secondary flows were seen in all three planes, the strongest occurrences were at the mid and exit planes at the 0.800 speed ratio. Large separation regions were reported in the mid and exit planes at the core and suction surfaces at both speed ratios. In 1993, Gruver, Flack, and Brun [3] analyzed the results taken by Gruver [38] of the pump velocity field. Slip factors, vorticities, and torque distributions were calculated using the data. The torque was found to be evenly distributed between the inlet and exit planes at both speed ratios. The slip factors calculated for pump mid-plane were approximately the same as for a conventional centrifugal pump. The exit plane slip factors, however, were significantly higher. Brun and Flack [4], in 1993, developed an analysis method to take measurements in unsteady, but periodic, flows typically found in multicomponent turbomachines using discrete sampling techniques such as laser velocimetry. The knowledge of the instantaneous angular position method is needed of all components at the instant the velocity is sampled. By correctly organizing the data, the velocity field resulting from the interaction of two or more components may be obtained. Brun [5], in 1993, measured the velocity flow fields in the turbine of the torque converter using the same experimental setup. Investigated were the turbine inlet, quarter, mid, and exit planes at the 0.065 and 0.800 speed ratios, and an average velocity field was obtained for each plane. Transient fields were found for the inlet, quarter, and mid planes. For both speed ratios, the turbine inlet was found to be highly dependent on the relative pump-turbine position. At the 0.065 speed ratio, large separation regions were seen in the quarter and mid planes, while significant secondary flows were seen in all planes. In 1994, Brun, Flack, and Gruver [6] measured the unsteady flow field in the pump of the Plexiglas torque converter. Strong periodic velocity fluctuations in the pump inlet plane were found to be due to the relative pump-stator position. Weaker fluctuations, although were present in the pump exit plane as a function of the pump-turbine relative position. No fluctuations were seen in the mid plane flow. The pump inlet incidence angle was reported to fluctuate more at the 0.065 speed ratio that at the 0.800 speed ratio and the pump exit plane slip factor fluctuated less than 1%. Brun and Flack [7] analyzed velocity results in the inlet, quarter, mid, and exit planes of the turbine at the 0.065 and 0.800 speed ratios in 1994. Mass flow rates, inlet incidence angles, exit flow angles, average vorticities, turbine output power, and viscous dissipation were calculated. The turbine exit flow was seen to follow the blade closely. Brun and Flack [8], in 1994, reported the unsteady velocity results in the turbine for the inlet, quarter, mid, and exit planes at the 0.065 and 0.800 speed ratios. Strong periodic fluctuations position, were seen and found to be function of the relative turbine-pump at the turbine inlet plane. Flow inlet angles, RMS unsteadiness, and output torque per blade passage were calculated. The through flow velocity component was reported to vary by approximately 3.6 m/s. Ainley [9] used the same test rig to measure the three-dimensional velocity flow field in the stator and pump of the torque converter in 1994. Data was taken at the 1mm upstream, inlet, quarter, mid, three-quarters, and exit planes on a 9-by-9 measurement grid of the stator. Average and unsteady measurements were taken at the inlet, mid, and exit planes at the 0.065 speed ratio and for all planes at the 0.800 speed ratio. Separation regions were observed in the quarter, mid, three-quarter, and exit planes at the 0.800 speed ratio. Average velocity measurements were taken at the pump mid and exit planes for five combinations of pump and turbine rotational speeds. At all operating conditions, high velocity regions were reported in the pressure-shell corner and low velocity regions were reported along the core side. In 1994, Brun, Flack, and Ainley [10], examined the secondary flow patterns at the pump mid and exit planes for several different pump and turbine rotational speeds using the same experimental setup. A counter-clockwise circulatory secondary flow in the mid plane and a clockwise circulatory secondary flow in the exit plane were noticed at all operating conditions. In both planes, average flow vorticity and secondary velocity magnitude decreased with decreasing pump speed. Whitehead [11] used the same setup to test the first of a series of six 245 mm automotive torque converters in 1995. The tested torque converter had the designation NW pump - LC turbine (names of the torque converter elements were assigned by General Motors Powertrain Division). Steady and unsteady measurements were taken for the 0.065, 0.600, and 0.800 speed ratios at the pump mid and exit planes and the turbine inlet and mid planes. Measurements were also taken in the pump inlet and turbine outlet for the 0.800 speed ratio. The average through flow velocity component was measured in the stator at the six planes at all three speed ratios. Comparing the results with those from the 230 mm torque converter test by Ainley [9], Whitehead [11] showed, less separation and weaker secondary flow was reported in all pump planes at the 0.065 and 0.800 speed ratios. More torque was generated between the mid and exit planes and less between the inlet and mid planes in the 245 mm converter. More pump exit plane unsteadiness induced by the turbine was found in the 245 mm converter, yet the pump caused less unsteadiness in the turbine inlet plane flow field at all speed ratios. Less separation was seen in the turbine inlet plane at the 0.800 speed ratio and in the mid plane at the 0.065 speed ratio. Mass flow rates, vorticities, torques, pump slip factors, and inlet incidence angles were calculated. Brun [12] measured average and unsteady velocity data of the NX-LC torque converter geometry, in 1995. He compared the results to the 230 mm torque converter tested by Ainley [9] and the NW-LC geometry test by Whitehead [11]. Data was taken for the 0.065, 0.600, and 0.800 speed ratios, in the pump, turbine, and stator, 13 planes in all. Results showed large flow separation regions in the turbine and stator and strong jet/wake and circulatory secondary flows in the pump. To determine the effect of viscosity, pump speed, and speed ratio variation on the internal flow field for a wide variety of operating conditions, parametric studies were accomplished. Mass flow rates, input and output torques, pump slip factors, and flow incidence angles were calculated. The analysis of the pump jet/wake flows and circulatory secondary flow phenomena and the estimation of the pump head flow losses were based on a two-dimensional Navier-Stokes and streamwise vorticity generation equations the for the development of a simplified analytical flow model. Parametric studies based on the models were performed to evaluate the effect that operating condition and pump geometry had on jet/wake flow phenomena, secondary velocities, and head losses. Analytical results agreed well with the experimental results. A full three-dimensional Navier-Stokes solution was obtained for a limited number of flow conditions using FlowPlusä . In 1997, Claudel [13] studied the remaining five 245 mm torque converters to analyze the effect of a change in geometry on the flow field of the pump and stator. The pump mid and exit planes for all geometries were tested at the speed ratios of 0.065, 0.600, and 0.800. The stator upstream, quarter, mid, three-quarter, and exit planes of the NX-LC geometry were tested. Mass flow rates, slip factors, vorticity, and torques were calculated for all measured planes. Results showed that variations in the pump exit blade angle and turbine inlet angle had significant effects on the mid and exit plane flow field of the pump. The flow field alignment improved and the slip factor increased as the exit blade angle of the pump decreased and as the inlet blade angle of the turbine decreased. There was an increase in throughflow velocity and a more evenly distributed mass flow across the plane. In 1998, Yermakov [14] analyzed the effect of a change in geometry on the flow field inside the turbine studied within the same five 245 mm torque converters as Claudel [13]. The turbine inlet and mid planes were measured at speed ratios of 0.065, 0.600, and 0.800. The turbine exit plane was measured in the NW-LC and NY-LC geometries at a speed ratio of 0.800 only. Mass flow rates, flow angles, vorticities, and torques were calculated. Yermakov [14] showed that the pump exit blade angle had significant influences on the turbine inlet plane. He also found that a moderate influence of the turbine inlet blade angle on the flow in the turbine inlet and mid plane. In 1998, Christen [15] measured the unsteady flow field of the NX-LC and NY-LC torque converter geometries at speed ratios of 0.065, 0.600, and 0.800. Both the pump exit and turbine inlet planes were analyzed. Results showed that the turbine has a moderate influence on the pump exit flow field but he turbine inlet plane flow field was seen to be highly unsteady due to the passing of the upstream pump blades. Mass flow rates, slip factors, and flow angles were calculated over the entire flow cycle for both planes. Also in 1998, Hotho [16] studied the unsteady pump exit and turbine inlet flow field of the NY-LC torque converter at nine different operating conditions. Speed ratios of 0.065, 0.600, and 0.800 were used with three different pump speeds for each speed ratio. Insignificant changes were seen in the flow fields when the pump speed was varied while a constant speed ratio was maintained. Pump speed only affected the mass flow rate and parameters dependent upon that quantity. In 1999, Gruber [17] measured the steady flow field of the JC-JD torque converter. This torque converter had the characteristic to have a shorter axial length than the six former geometries. The three-dimensional, time-averaged velocity flow field was obtained in the inlet, mid, and exit planes of the pump and turbine at the 0.065, 0.600, and 0.800 speed ratios. Only the throughflow velocity was measured at the stator quarter plane for all speed. The data collected was compared to a previously tested torque converter with a larger axial length to determine the effect of axial length on the internal flow field of the torque converter. The data collected indicates that changes in the axial length of the torque converter affect the pump flows much more than the turbine flows and that there was more secondary flow at the pump exit. Performance was much worse in the shorter axial length torque converter as defined by the pump slip factor. For the present phase of research, another 245 mm General Motors torque converter is tested. This geometry is the one which has the largest axial length. The three-dimensional, time-averaged velocity flow field is measured in the inlet, mid and exit plane of the pump, in the inlet, mid, and exit planes of the turbine. The velocity field in the stator is also measured. The operating conditions are 0.065, 0.600, 0.800, 0.875 speed ratio. As an addition, the unsteady flow at the pump exit / turbine inlet is also investigated. Experimental Facility A schema of the complete experimental test facility is shown in figure 1. The different sectors of the rig including the torque converter test fixture, the hydraulic system, the automatic traverse, the laser velocimeter, the signal processor, the shaft encoder, and the data acquisition system are described in detail below.
Figure 1 – Experimental Test Facility Test Rig A standard line of one-stage automotive torque converters entirely machined out of Plexiglas (shown in figure 2) is used for this research. The components are placed in a containment box also machined out of Plexiglas and entirely filled with oil. As it is shown in figure 3, the box has rectangular dimensions to provide flat laser entrance surfaces in the direction of the measurement.
Figure 2 – Plexiglas Torque Converter Both pump and turbine were made as a two-piece construction. A solid piece of Plexiglas was milled by using computer-controlled 3-axis milling machine so that the blade and the core were in one part. These blade nest/cores were glued to the shell using Dichloroethane as a solvent. The stator was also milled from a single piece of Plexiglas. The internal part of the torque converter (input and output shaft and stator shaft) are in stainless steel. The containment box was supported on a large aluminum plate securing the test fixture to the test stand.
Figure 3 – Plexiglas Torque Converter in the box Laser Doppler Velocimetry Laser velocimetry is the preferred method of internal flow visualization in turbomachines. It has been successfully used to measure steady and unsteady internal velocity and the method is considered as a reliable measurement technique. It is the only non-intrusive method and so allows measuring accurately the three components of the flow without disturbing it. Several methods of laser velocimetry exist but the Doppler velocimetry is was the method chosen for this study. The advantages of the Doppler velocimetry are fast response, fast signal processing and high spatial resolution.
Figure 4 – Laser Velocimetry system
Bibliography 13. Claudel, W.D., "A Comparison of the Internal Flow Fields of Three Automotive Torque Converter Pumps Using Laser Velocimetry", Masters Thesis, University of Virginia, May 1997
|
|
|
.
