Electrical Machinery Windage Loss Reduction Overview As surface speeds of electrical machinery increase to meet ever more demanding application requirements, windage power losses due to shearing of air (or other process fluid) between the rotor and stator take on an increasingly significant role. Historically, these losses have not received a great deal of research attention. Common approaches include making the rotor and stator surfaces as smooth as possible, keeping the rotor-stator gap as large as practical without compromising electrical efficiency, and simply accepting whatever losses are present. Xdot Engineering and Analysis, along with Computer Aided Engineering Associates, have recently completed a two year Phase II SBIR effort funded by the US Air Force to look at these losses in more detail. This work included extensive Computational Fluid Dynamics (CFD) analysis and experiments in a new high speed windage power loss test rig. The outcomes of this effort included: Development of an easy to use Ansys/CFX windage power loss modeling and analysis module. Generation of a new database of analytically predicted rotor power losses for high speed electrical machinery. Development and commissioning of a windage loss test rig capable of accurately measuring rotor power loss at speeds of up to 60,000 rpm. Generation of a new database of experimentally measured rotor power loss at relevant surface speeds and test article scale for 50 kW to 200 kW, high speed electrical machinery. Xdot Xdot Engineering and Analysis, pLLC Xdot Engineering and Analysis 370C Greenbrier Drive, Charlottesville, VA 22901 www.XdotEA.com 1 CAE Associates, Inc. 1579 Straits Turnpike, Suite 2B Middlebury, CT 06762 www.caeai.com Identification of several strategies which are predicted to reduce the windage loses by as much as 30 percent in specific cases for high surface speed applications. These approaches also appear to have promise for higher viscosity gasses and fluids at lower surface speeds. Flow Physics in Motor Radial Air Gap Windage loss in a motor is generated by the friction between the rotor and air. The magnitude of windage loss varies widely from machine type, rotor shape and size, rotational speed, air-gap size, and rotor surface finish. Each parameter can present its own unique challenge to machine design when windage loss amounts to a substantial portion of the overall power loss. Since windage loss is generated by the shearing interactions between air, rotor surface, stator surface, and cavity, understanding the flow physics in the air-gap is important when assessing the influence of windage loss on machine performance. The commonly assumed fluid flow pattern in the air gap is Couette flow, with maximum air speed on the rotor surface equal to the machine rotational speed, linearly tapering to zero on the stator wall. This pattern is based on Couette flow theory applied to two concentric cylinders. In earlier electric motor designs and operation, motor speed was not as high, and the theory predicted reasonable results. However, with the increase of machine size and rotor speed, Couette flow theory is no longer valid for modeling high air-gap Reynolds numbers. The increasing Reynolds number pushes the air-gap flow over an instability threshold, resulting in the formation of Taylor vortices in the axial direction. These counter-rotating vortex pairs represent pressure and velocity oscillations and lead to additional power loss in the motor. A further increase in the air-gap Reynolds number will lead to turbulent transition of Taylor vortices to a very complicated near-wall turbulence structure. Data taken by Gorland and Kempke (1970) to address windage loss showed additional power loss was observed for high speed motors, which exceeded the traditional theory predictions. In some cases, the windage loss accounted for 35% of the total power loss. Therefore, windage loss in high speed motors is an important concern for motor design and must not be overlooked. An example of a classic high speed electric motor gap geometry is shown in Figure1. The machine is composed of a rotor and stator with cavities. The rotor diameter is 2.85" with an axial length of 7.5". There are 24 stator cavities. The air gap is 0.05". The rotor speed is 60,000 rpm. Figure 2 shows the rotor surface pressure contours. Of immediate interest is the highly non-uniform pressure distribution in the axial direction. Further examination of axial velocity contours and streamlines in the air-gap, shown in Figure 3, confirms the presence of counterrotating Taylor vortex pairs in the axial direction. These vortices create increased wall-shear, resulting in increased windage loss in the motor. Xdot Xdot Engineering and Analysis, pLLC Xdot Engineering and Analysis 370C Greenbrier Drive, Charlottesville, VA 22901 www.XdotEA.com 2 CAE Associates, Inc. 1579 Straits Turnpike, Suite 2B Middlebury, CT 06762 www.caeai.com Figure 1: A Classic Rotor-Stator Configuration Flow Domain Figure 2: Rotor Surface Pressure Contours Xdot Xdot Engineering and Analysis, pLLC Xdot Engineering and Analysis 370C Greenbrier Drive, Charlottesville, VA 22901 www.XdotEA.com 3 CAE Associates, Inc. 1579 Straits Turnpike, Suite 2B Middlebury, CT 06762 www.caeai.com Figure 3: Axial Velocity Contours and Streamlines in the Air Gap Another class of machine is the Switch Reluctance Motor (SRM) . The rotor is shaped with several poles on the rotating cylinder. This geometry results in a more complex mechanism of windage loss. These poles provide a pumping mechanism while rotating, acting like “blades” in pumps. Figure 4 shows a representative flow domain (air gap and stator poles) for an SRM. Figure 5 shows the stator surface pressure contours, which exhibit a non-uniform pressure distribution in the axial direction, as well as localized pressure gradients where the flow enters and leaves the stator pole region. Figure 6 shows the streamlines for the cross-section in the middle of a stator cavity. A recirculating flow region exists following the suction side of each pole. In addition, Taylor vortices also exist between the pole top and the stator housing in the axial direction, further compounding windage losses. Figure 4: Air Gap and the Stator Poles of a Switched Reluctant Motor Xdot Xdot Engineering and Analysis, pLLC Xdot Engineering and Analysis 370C Greenbrier Drive, Charlottesville, VA 22901 www.XdotEA.com 4 CAE Associates, Inc. 1579 Straits Turnpike, Suite 2B Middlebury, CT 06762 www.caeai.com Figure 5: Stator Surface Pressure Contours of a Switched Reluctant Motor Figure 6: Cross-Section Streamlines of a Switched Reluctant Motor Xdot Xdot Engineering and Analysis, pLLC Xdot Engineering and Analysis 370C Greenbrier Drive, Charlottesville, VA 22901 www.XdotEA.com 5 CAE Associates, Inc. 1579 Straits Turnpike, Suite 2B Middlebury, CT 06762 www.caeai.com Recent CFD Research With the development of computational fluid dynamics (CFD), numerical flow visualization has enabled detailed examination of rotor air-gap flow physics for different types of machines. CFD investigations focused on Taylor vortex formation in concentric cylinders and comparisons with flow visualization images were reported by Braun, et. al (2002). The evolution of the Taylor vortex pairs was clearly observed. The effect of end walls on Taylor vortices structures and the overall windage loss using CFD analyses were reported by Wild (1996). The presence of an axial flow on air-gap flow patterns and the windage loss prediction of a high speed motor were reported by Ren (2003). Numerical study of windage loss of eccentric rotor configurations was conducted by Scurtu, et. al (2008). Furthermore, Tong (2008) has adopted the versatility of CFD to analyze the stator end winding region of an entire motor to optimize the cooling flow path and its interactions with gap flow. This research has demonstrated that CFD can provide detailed flow analysis for complex machines, and is an essential part of recent high speed motor research. Overall however, in the existing literature, the application of CFD to address windage loss in high speed motors has been sparse, and has received little attention from major aerospace, power generation, and machinery industries. The common thread in the current literature is to conduct a CFD analysis for a particular machine for the purpose of examining the three-dimensional flow structure. Windage loss data is a validation quantity used to measure the accuracy of numerical results. Unfortunately, to date, there have been minimal research efforts to address the root cause of Taylor vortices in radial air-gaps, and how to reduce the windage loss caused by these vortices. Innovative Flow Technology There were limited attempts in the past to either "block" or "disrupt" the Taylor vortex pairs in the air-gap by means of geometry variation, such as inserting fin like structures on rotor or stator surfaces. Unfortunately, these efforts didn't diminish the vortices and instead increased the power loss. To date, the commonly used methods to reduce power loss include the use of smooth rotor and stator surfaces, and increased radial air-gaps when they don’t compromise efficiency. In many cases, the choice is simply to acknowledge and accept the presence of Taylor vortices in the gap, with the hope of minimizing their impact on power loss through traditional experience or trial-and-error. Our approach was to start the development effort through a detailed understanding of flow physics -why and how the Taylor vortices are formed in the radial air gap. We developed a CFD-based motor shaft and radial gap design software tool which allows us to analyze thousands of radial air-gap and flow configurations quickly and efficiently. The software tool is based on the ANSYS Workbench/CFX platform with automated geometry generation, editing, meshing, CFD setup, solve, post-processing, multiple design point iterations, etc. After gaining an understanding of the flow physics associated with Taylor vortices in the radial air-gap by analyzing many design geometries, we developed an innovative flow technology for radial gaps which reduces windage power loss. When the flow technology is applied to flow gaps, the Taylor vortices are entirely removed, resulting in a substantial power loss reduction in many representative classes of electric motors. Xdot Xdot Engineering and Analysis, pLLC Xdot Engineering and Analysis 370C Greenbrier Drive, Charlottesville, VA 22901 www.XdotEA.com 6 CAE Associates, Inc. 1579 Straits Turnpike, Suite 2B Middlebury, CT 06762 www.caeai.com We can use the classic electric motor example in Figure1 to demonstrate the effectiveness of the flow technology. Figure 7 shows the axial velocity contours while Figure 8 shows the streamlines. The "before" plots in these figures are shown in Figure 3, where Taylor vortices are clearly present. The "after" plots are the results when the flow technology is applied. The resultant axial flow in the gap is free from any Taylor vortices, and provides a significant windage loss reduction. Representative windage loss reduction results for machines of different classes are summarized in Table 1. This data shows the innovative flow technology provides windage loss reduction benefits, as high as 25% to 30%, for a wide range of machines. The innovative flow technology is non-intrusive and could potentially be implemented in many existing machines. This is particularly helpful for retrofitting existing systems in order to take advantage of this technology for reducing power loss and operational cost. The CAE Associates/Xdot team is ready to help you apply this technology in order to reduce power loss and realize the resulting cost savings in your high speed machines. Figure 7: Axial Velocity Contours - Before and After the Flow Technology is Applied Xdot Xdot Engineering and Analysis, pLLC Xdot Engineering and Analysis 370C Greenbrier Drive, Charlottesville, VA 22901 www.XdotEA.com 7 CAE Associates, Inc. 1579 Straits Turnpike, Suite 2B Middlebury, CT 06762 www.caeai.com Figure 8: Streamlines - Before and After the Flow Technology is Applied Table 1: Potential Power Loss Reduction for Representative Machines Xdot Xdot Engineering and Analysis, pLLC Machine Speed % Reduction Low (10 - 20 K RPM) 10 - 15 Mid (30 - 40 K RPM) 25 - 30 High (50 – 70 K RPM) 20 - 25 Xdot Engineering and Analysis 370C Greenbrier Drive, Charlottesville, VA 22901 www.XdotEA.com 8 CAE Associates, Inc. 1579 Straits Turnpike, Suite 2B Middlebury, CT 06762 www.caeai.com Test Rig The experimental portion of this research effort was conducted using a new high speed windage loss test rig design, produced and commissioned by Xdot. The rig, shown in Figure 9 below, was built around a 2.85" diameter, 7.5" long test section simulating a generator/motor. The rig's unique design allows the test section rotor torque to be directly measured with a resolution better than 0.0001 Nm, and an uncertainty on the order of 0.0025 Nm. The rig is driven by a 14 hp (10.5 kW), 60,000 rpm spindle. Measurement of the rotor torque, rather than stator reaction torque, is important in highly turbulent systems such as this one where there is potential for appreciable fluid inertia effects which could act differently on the rotor versus the stator. For example, in the case of combined axial cooling flow plus rotationally driven flow at full operating speed, the difference is predicted to be on the order of 10 to 15 percent. For more complex flow arrangements, the difference could be even larger. Baseline Experimental Effort Using the test rig, a set of six baseline datasets were generated for rotor-stator air gap (radial) clearances of 0.025, 0.050, and 0.075 inches (0.635, 1.270, and 1.905 mm) with both a smooth wall stator, and a simulated stator with 40 axial slots. Data were collected at 10, 20, 30, 40, 50 and 60 krpm. A typical dataset is shown in Figure 10. The averaged points for the smooth wall data are shown overlaid on the classic data from Bilgen and Boulos' 1973 windage loss paper in Figure 11. As can be seen, the agreement is quite good. Floating Test Shaft and Drive 14 Hp, 60,000 rpm drive spindle Fixed Test Section Base Hydrostatic Air Bearing Supports Figure 9: Windage Loss Test Rig Side View Xdot Xdot Engineering and Analysis, pLLC Xdot Engineering and Analysis 370C Greenbrier Drive, Charlottesville, VA 22901 www.XdotEA.com 9 CAE Associates, Inc. 1579 Straits Turnpike, Suite 2B Middlebury, CT 06762 www.caeai.com 0.635 mm Gap Smooth Wall 0.1 Torque (Nm) 0.08 95% Confidence Intervals 4 repetitions each 10-50 krpm 5 repetitions 60 krpm 0.06 0.04 0.02 0 0 10 20 30 Speed (krpm) 40 50 60 Figure 10: Raw Data and Error Bars, 0.025" Gap Smooth Wall Test Article Test Data and Data from Bilgen and Boulos (1973) 1 Modified Moment Coeff Cm(C/R) -0.3 10 0 10 From Bilgen and Boulos Test Data -1 10 -2 10 -3 10 1 10 2 10 3 4 10 10 Reynolds Number (Re = 5 10 RC/ ) 6 10 7 10 Figure 11: Comparison of Test Data to Bilgen and Boulos 1973 Paper Xdot Xdot Engineering and Analysis, pLLC Xdot Engineering and Analysis 370C Greenbrier Drive, Charlottesville, VA 22901 www.XdotEA.com 10 CAE Associates, Inc. 1579 Straits Turnpike, Suite 2B Middlebury, CT 06762 www.caeai.com Power Loss Reductions Using the CFD tool, several promising approaches to reducing windage power loss were identified and optimized for this geometry. For specific cases, predicted reductions of much as 30 percent relative to a baseline case were obtained. For reasons that are not yet fully understood, the measured reductions were on the order of 5 percent. Several of the approaches appear to be feasible for a wide range of machinery without extensive design changes. One of the approaches appears to be especially relevant for machines a standby or low power mode. Acknowledgements The work described was a Phase I/II SBIR effort sponsored by the Air Force Research Laboratory (AFRL). This support is gratefully acknowledged. References Gorlans, S., Kempke, E. and Lumannick, S., “Experimental Windage Loss for Close Clearance Rotating Cylinders in the Turbulent Flow Regime”, NASA TM X-52851, 1970. Braun, M., Kudriavtsev, V., and Corder, R., “Flow Visualization of The Evolution of Taylor Instability and Comparisons with Numerical Simulations”, Computational Technologies for Fluid/Thermal/Structural/Chemical Systems with Industrial Applications, Vol. 1, ASME, 2002. Wild,P.,Djilali, N., and Vickers, G., “Experimental and Computational Assessment of Windage Losses in Rotating Machinery”, Transactions of the ASME, Vol. 118, 1996. Ren, W., “Windage and Axial Friction Losses of High Speed Generator”, 2003 International Joint Power Generation Conference, Atlanta, GA, 2003. Scurtu, N., Stuecke, P., and Egbers, C., “A Numerical Study of the Three-Dimensional Structure of the TaylorCouette Flow in Eccentric Configuration with Superimposed Cross Flow”,15th International Couette-Taylor Workshop, J. of Physics Conf. Series 137, 2008. Tong, W., “Numerical Analysis of Flow Field in Generator End-Winding Regions”, Intl. J. of Rotating Machinery, Vol. 2008, 2008. Bilgen, E. and Boulos, R., “Functional Dependence of Torque Coefficient of Coaxial Cylinders on Gap Width and Reynolds Numbers,” Journal of Fluids Engineering-Transactions of the Asme, Vol. March, pp. 122–126, 1973. How We Might Be Able To Help You The Xdot/CAEA team has developed considerable insight into the problem of windage loss and ways to minimize it. We also have developed a unique toolset and a new experimental capability. If you have a high surface speed machine with a long close-clearance region, we have the tools and experience to efficiently look at the windage power loss and may be able to help you develop ways to reduce it. The test rig also gives us the capability to accurately characterize the power loss of a wide range of components and features at shaft speeds up to 60 krpm. Xdot Xdot Engineering and Analysis, pLLC Xdot Engineering and Analysis 370C Greenbrier Drive, Charlottesville, VA 22901 www.XdotEA.com 11 CAE Associates, Inc. 1579 Straits Turnpike, Suite 2B Middlebury, CT 06762 www.caeai.com