Ultra-Highspeed Mesoscopic Rotor On Foil Bearings

Page 1 of 9 © Copyright Mohawk Innovative Technology, Inc. 2003

A 700,000 rpm Feasibility Demonstration for Mesoscopic Scale Gas Turbine Engines Presenter:

James F. Walton II, M. Eng. Vice President of Program Development

Mohawk Innovative Technology, Inc. Albany, NY 12205

USA

Authors: Mohsen Salehi, Ph.D., Michael J. Tomaszewski, James F. Walton II,

and Hooshang Heshmat, Ph.D.

December 3-5, 2003

®Mohawk InnovativeTechnology, Inc.


ABSTRACT A new class of miniature weapons and war fighting systems is being developed that requires a mesoscopic sized turbojet engine for power and propulsion applications. These systems require a high power density, low cost, and fuel efficient turbojet engine that must operate reliably, even after being stored for an indefinite period of time. Due to the extremely high operating speeds expected (up to one million rpm), bearings have been shown to be a key limiting factor. A revolutionary design approach that integrates and uses oil-free, self-acting hydrodynamic compliant foil air bearings is therefore offered. This paper presents the preliminary design and testing of a simulator rotor bearing system that demonstrates the feasibility of developing such a system.

A nine-gram, single piece metallic rotor was fabricated and spun to 705,000 rpm. The rotor was sized to be representative of a turbojet engine with integral compressor and turbine stages capable of approximately 2 pounds of thrust or up to 144 watts of electrical power. The single piece rotor design, used to simulate the expected ceramic rotor construction, dictated that a unique split housing and split foil bearings be designed and fabricated. Testing was completed with the rotor subjected to a wide range of orientations, including rotor inverted (i.e., 180 degree roll) and rotor spin axis vertical. Changes in rotor orientation were accomplished while the rotor was spinning at speeds in excess of half a million rpm to demonstrate suitability for air vehicle application. Correlation between design predictions and measured response was good, indicating scalability of existing analysis tools and physical hardware to the mesoscopic scale. The success of these tests has demonstrated the feasibility of developing ultra high-speed mesoscopic rotating machinery systems for power and propulsion applications.

Key words: mesoscopic turbojet, mesoscopic generator, oil-free gas turbine engine, micro air vehicle, uninhabited air vehicle, compliant foil bearing, air bearing, gas bearing.

INTRODUCTION The predicted spectrum of conflict for the 21st Century has influenced and motivated the development of a wide range of weapon systems. The shift toward a more diverse array of military operations, often involving small teams operating in non-traditional

environments, has sparked interest in miniature or mesoscopic scale power and propulsion systems. Mesoscopic scale power generating (MPG) systems are needed to satisfy the warfighter’s high electrical power demands due to the increasingly power hungry electronic equipment needed to sustain the soldier in the field. Other needs include personalized cooling systems for the individual soldier; and vacuum pumps and blowers for battlefield chemical and biological warning systems. Commercially, MPGs with their high power density are also envisioned for use autonomous robotic and mobile applications. Propulsion systems for Microair vehicles (MAVs) and Small UAVs (SUAVs) are needed to support reconnaissance functions for on-demand information about surroundings and potential targets as well as providing a dense or autonomous swarm of sensor, decoy and/or intelligent de-mining or munitions carrying platforms. By providing unprecedented situational awareness, rapid response intelligent munitions, enhanced field capabilities and real time command and communication, greater effectiveness, battlefield superiority and fewer casualties will be realized.

Mesoscale machines (i.e., "sugar cube to fist sized" and typically weighing 50 grams or less) offer a unique opportunity to meet these power and propulsion needs efficiently, reliably and inexpensively. Mesoscopic scale turbine powered systems are currently attracting attention due to their high power density and output energy efficiency. Among the power sources using chemical fuels, the gas turbine generator is known to have the highest power to weight ratio [1]. Significant additional commercial applications are also expected in support of miniature fuel cells, diesel engines and other propulsion and power generating machinery. An analysis of a gas turbine-fuel cell hybrid micro generation system has shown power efficiency of over 65% in the best possible case [2]. Such a hybrid system would be based on a micro gas turbine (µGT) and a solid oxide fuel cell (SOFC) and is expected to achieve much higher efficiency than a traditional µGT. With progress in manufacturing techniques, it is expected that the cost of these mesoscopic scale machines will be reduced drastically, making the application of these devices more attractive [3]. However, many issues need to be addressed, including compressor, combustor and turbine efficiency, component fabrication and scaling.


In addition to these, the long overlooked rotor-bearing system integration issues are essential to the ultimate operation of such systems. For example, the reduced size of mesoscopic scale machines impacts internal clearances, balancing, manufacturing tolerances, damping and thermal management. To address the engine system development issues that will make these mesoscopic turbo machines a reality with high performance, reliability and long term storability, MiTi® has developed and demonstrated the supporting oil-free bearing technologies [4].

The importance of oil-free compliant foil bearings to the successful development of mesoscopic gas turbine engines is seen in the series of preliminary cycle analyses shown schematically in Figure 1. In the cases analyzed, the influence of compliant foil and rolling element bearing power loss on the mesoscopic engine was evaluated to assess their impact on available shaft output power. Figure 2 is a plot comparing power loss of each bearing as a function of speed. The bearing power loss calculations were performed assuming that each bearing was sized for the mesoscopic rotor. For a given operating speed of 870,000 rpm and varying inlet air temperatures, holding compressor, combustor and turbine efficiencies constant, the available output or generating power was determined.

Compressorc = 3.0c = 0.7W = 334 Watt = 870 krpm

•

Turbinet = 2.70t = 0.7W = 510 Watt•

Combustor = 0.92 P3 = 280 kPa

T3 = 1000C

P0 = 101.3 kPaT0 = 15Cm = 2 g/sec

P2 = 303.9 kPaT2 = 171Cm = 2 g/sec

Losscfb = 32 W144 Wattcfb

P4 = 101.3 kPaT3 = 840C

03-0031D

Lossreb = 180 W-4 Wattreb

Available Output Power(Watts)

T0 (C) Foil Bearing Ball Bearing

15 144 - 425 113 -3550 105 -43

Output

Figure 1. Mesoscopic Brayton cycle analysis comparison.

As seen in Figure 1, even at the lowest inlet temperature of 15°C, which results in the highest available output power, the ball bearing power loss would prevent full speed operation. In this first idealized case, the combined power needed to drive the compressor and overcome the ball bearing power loss exceeds turbine

output power by 4 watts. Conversely, the foil bearing supported system, with its substantially lower power loss, would make it possible to provide up to 144 watts of electrical generating power. The situation only gets worse as the air inlet temperature increases.

Mesoscale Turbine - Bearing Powerloss

0 50

100

150

200

250

0 200000 400000 600000 800000 1000000 1200000 Rotor Speed (rpm)

Ball Bearing Loss Foil Bearings

5.5 Times lowerlosses for foil

bearing

03-0030 Figure 2. Comparison of foil and ball bearing power loss versus speed.

The cycle analysis with increased inlet air temperatures also points out the need for system thermal management to limit the heat soak back from the turbine to the compressor during operation. Due to the close proximity of all components and the limited available pathways to remove heat, it is essential that materials with low thermal conductivity be used and that parasitic heat losses be minimized in these machines. For example, with higher physical temperatures of the compressor itself, additional turbine power is needed to maintain the compression ratio and mass flow rate. The extraction of additional work from the turbine reduces the available shaft output power to drive a generator. Thus, from this simple analysis, it is concluded that the feasibility of developing efficient mesoscopic power and propulsion turbine engines will require the use of ceramic rotors with their low thermal conductivity and foil bearings with their low power loss.

Having identified the need for ceramic rotors and foil bearings, one must then consider manufacturing related and cost issues, especially for mesoscopic sized rotor and bearing systems. For example typical clearance ratios (clearance/diameter) in gas bearings ranges from .0001 to .001, which for a 6 mm diameter shaft, would mean that the bearing diametral clearance would be between 0.6 and 6 m. Centrifugal growth of a ceramic shaft at 1 million rpm would be approximately 0.4 m, consuming from 15% to over 60% of the available bearing clearance, if a rigid pad


gas bearing design were used. With the addition of thermal growth, which for ceramics would be on the order of 1.5 m for a shaft-to-bearing differential temperature as small as 75°C, bearing clearance could be totally eliminated with certain gas bearing designs. The compliant foil bearing inherently is designed to readily accommodate such growth.

Figure 3. Mesoscopic turbojet engine with photos of simulator hardware

Besides shaft growth and manufacturing processes, balancing of the rotating group must also be considered, both from manufacturing and operating points of view. Balance specifications according to ISO 1940/1 and ANSI S2.19 for gas turbine and turbogenerator class rotors, would dictate that the balance limit for a nine gram rotor should be 9x10-5 gr-in when operating at 700,000 rpm and 1x10-5 gr-in when operating at one million rpm. At this balance level the expected rotor vibration would be approximately 2.8 m when operating at 700,000 rpm. Given that a bearing clearance of less than 6 m would be likely, the combined rotor growth of 2 m and rotor displacement of 2.8 m would make operation of the rotor system unlikely without a unique bearing system. Further, achieving such stringent balance levels would be nearly impossible with a multi-piece rotor assembly. It is also well known that joining and structural integrity are two issues that must be addressed when using ceramics. Thus when combined with the need for low cost, the only real logical manufacturing method is to make the rotating group as a single piece. The single piece rotor design choice dictates that split bearings be used in order for assembly to be possible. Figure 3 shows a conceptual turbojet engine cross section with a single piece ceramic rotor and

foil bearing cartridge along with photos of tested simulator hardware.

SIMULATOR DESIGN AND TEST METHOD Since bearings have been identified as one of the key limiting factors in the operation of high speed mesoscopic machines, it was felt that machine feasibility would best be demonstrated through rotor-bearing system simulator testing. As such a mesoscopic scale rotor-bearing system dynamic simulator was designed, built and tested to demonstrate rotor-bearing system stability, design analysis scalability and the potential of successfully operating and manufacturing split compliant foil bearings. In this section of the paper details of the experimental set up (simulator), rotordynamic analysis and experimental results are explained.

Turbine Simulator Test Rig The MiTi® mesoscopic turbine simulator consists of an impulse turbine driven single piece rotor supported by two compliant foil journal bearings. Two compliant foil thrust bearings were used to maintain rotor axial position. Compliant foil air bearings were selected for this demonstration because of their ability to operate at almost unrestricted speeds, their low power loss as discussed above and ability to operate at temperatures above 1100°F for extended durations. MiTi®’s Korolon high temperature, high durability coating will permit thousands of machine start/stop cycles without degradation. The newly developed split journal and thrust designs were also used.

The simulator housing included a split bearing shell, a turbine nozzle box, end caps and provision for fiber optic displacement sensors and speed pickup. Total simulator weight was 56 grams, including the 9-gram rotor. The simulator housing was made of aluminum with an overall length and diameter of 32.7 and 28.6 mm, respectively. The assembled simulator hardware is shown in Figure 4 and Figure 5. Two end caps with embedded holes were used on both ends of the simulator housing to secure to split housings and provide a location for the optical speed pick up and/or an axial displacement probe if desired and finally for directing the impulse turbine exhaust to the outside. The simulator shell or housing included holes for the impulse turbine drive air and optical displacement probes. The simulator housing and interior components such as the bearing shells and nozzle box were


made of two halves. The drive turbine nozzle box featured a ring with ID of 7.518 mm and OD of 11.43 mm on whose circumference 12 holes in six equally spaced circumferential positions were installed.

FiberopticSpeed Probe

DriveAir Inlet

Housing

FiberopticDisplacement

Probes

Figure 4. Assembled instrumented mesoscopic simulator test rig.

28.6 mm (1.125")

15 mm (0.59") Thrust Foil Bearings

SpeedPickup

Displacement Probes

Journal Foil Bearings

Impulse Turbine 6 mm

24 mm

03-0006-R3

32.7 mm(1.287")

Figure 5. Schematic diagram of mesoscopic simulator.

The PH 13-Mo metallic rotor shown in Figure 6 has an overall length of 24 mm, a shaft diameter of 6 mm and featured two 15 mm diameter, 1.664 mm long wheels at each end of the rotor. The 15 mm diameter wheels were used to simulate the mass and inertia properties of a turbine and compressor stage. The 6 mm diameter shaft center section was used as the foil bearing journal surface. The impulse turbine pockets used to drive the rotor were machined in the center of the single piece rotor. The split foil thrust bearings have an outside diameter of 15 mm and an inside diameter of 8 mm.

The PH-13 Mo bearing shells each had overall outside diameter and length of 21.26 and 5.08 mm, respectively as shown schematically in Figure 5 to accommodate the smooth top foil and supporting compliant bump foil assembly.

The simulator was equipped with two displacement probes and one speed pickup probe. The optical displacement probes were positioned to record simulator wheel motion. The speed pickup probe was installed on one of the end caps pointing at the axial face of the one of the two wheels. Probe locations are shown in Figure 4 and Figure 5.

Figure 6. Metallic simulator rotor.

BEARING SYSTEM DESIGN AND ANALYSIS Due to concerns regarding scalability of the physical hardware, detailed and iterative bearing and rotordynamic analyses were completed prior to hardware fabrication. First a finite element model of the rotor was prepared and critical speed analysis performed to establish required the nominal bearing stiffness range that would ensure adequate rotor-bearing critical speed margin. Figure 7 and Figure 8 present the results of the preliminary design analysis. As seen from Figure 8 bearing stiffness coefficients from 1.75 KN/m to 1750 KN/m would provide acceptable critical speed margins. Given this wide latitude, the 6 mm diameter compliant bearings were then designed to provide the most desired performance – namely high levels of damping for stability. The predicted bearing stiffness characteristics (non-dimensional values) are shown in Figure 9. Using the speed dependent bearing dynamic stiffness coefficients, rotor bearing whirl analysis was then conducted to assess system stability. During the whirl/stability analysis, the range of damping needed for a stable well controlled system was determined. Based on the required damping level, the bearing design was reviewed and further analyzed to ensure adequate damping would be provided. As an additional verification of the predicted bearing damping, data from MiTi®’s extensive empirically correlated database of bearing damping was mined and scaled to the mesoscopic scale bearings. A detailed technical background on


integration of structural and fluid film dynamic elements, frictional damping and scaling in foil bearings may be found in [5-10].

The results of the rotordynamic whirl and stability analysis are presented in Figure 10 and Figure 11. Figure 10 shows both the predicted and measured rigid body natural frequencies when operating at speeds up to 550,000 rpm. As seen, even with the small scale rotor and foil journal bearings, correlation between the measured and predicted natural frequencies is quite good. The logarithmic decrements reported on Figure 10 are those predicted for the two rigid body modes when operating at the corresponding spin speeds. Figure 11 shows that rotor system stability is expected over the entire operating speed range as noted by the positive logarithmic decrements for all speeds.

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21

File: P:\01025\Data\mg-msvh1.rot

1.5 mm

6.35 mm 17.7 mm

6 mm

01025-F024

ImpulseTurbine Mode 1

24,593 rpm

Mode 234,006 rpm

Mode 33,011,493 rpm

Figure 7. Rotor finitie element model and critical speed mode shapes

File: P:\01025\Data\mg-msvh1.rot

Critical Speed Map

Bearing Stiffness

103

104

105

106

107

101 102 103 104 105 106

175x101 175x102 175x103 175x104 175x105 175x106Lb/inN/m

01025-F023

Operating speed range

Figure 8. Preliminary critical speed map for mesoscale rotor.

TEST RESULTS In this section the results of tests with simulator are shown. The simulator was tested at various

speeds and at various rotor axis orientations through rolling or rotation of the simulator while it was spinning. These tests identify the performance of the bearing in maintaining a stable rotor system while transient dynamic conditions are imposed on the rotor system. The results are shown based on the rotor spectrum for various speeds and various orientations.

0 200 400 600 800 1000 1200 1400 1600

Speed (rpm in thousands)

-20

-10

-0

10

20

30

40

50

60

70

80

90

ShaftX

Y

01025-F025

Kxx

Kyy

Kxy

Kyx

Figure 9. Non-dimensional speed dependent bearing dynamic stiffness coefficients.

0 100000 200000 300000 400000 500000 600000Speed (RPM)

100002000030000400005000060000700008000090000

100000110000120000

Conical Predicted

Translatory PredictedConcical Measured

Translatory Measured

ConicalMode

TranslatoryMode

Mesoscopic Simulator Comparison

01025-F002C

Y

X

Z

Y

X

Z

Predicted Damped NaturalFrequency with Logarithmic

Decrement Between 0.43 and 0.52

Predictecd Damped NaturalFrequency with Logarithmic

Decrement Between 0.31 and 0.58

Analysis vs Measurement

Measured Damped NaturalFrequency

Figure 10. Whirl map comparing predicted and measured natural frequencies.

Figure 11. Stability map.

Figure 12, a Fast Fourier Transform (FFT) waterfall plot of rotor vibration versus frequency at discrete time steps, presents the results of the 702,000 rpm high-speed run of the


mesoscopic simulator. This plot shows rotor startup, acceleration to full speed and coastdown. The critical information gleaned from this plot besides the high speed operation is the very small amplitude levels for the two rigid body rotor natural frequencies when operating at high speed. The low amplitudes of these sub-synchronous vibrations, is indicative of good bearing damping characteristics. It is also instructive to note that the operating speed frequency is approximately 28 times greater than the rigid body natural frequencies. In early first generation air bearing designs circa 1980’s, rotor instability was often encountered when the spin speed frequency approached 5 or 7 times the rigid body natural frequencies. Through the improved fourth generation bearing designs developed by MiTi® that produce high levels of damping, this barrier has been successfully overcome.

Figure 12. Waterfall FFT plot showing 702,000 rpm

A more detailed examination of rotor vibration response was also conducted at a variety of speeds to assess overall vibration performance of the mesoscopic rotor-bearing system. The frequency spectrum obtained from two fiber optic displacement sensors with the rotor in the preferred horizontal position while spinning at 672,000 rpm is shown in Figure 13. As shown, a very small rotor displacement in the order of 0.25 μm (0.00001 inch) at the spin speed is observed.

Figure 14 and Figure 15 show rotor vibration spectrum with the rotor remaining horizontal, but the housing rolled or rotated about the spin axis 90° and 180° from its preferred orientation. Figure 16 shows the vibration spectrum when the rotor spin axis is rotated to the vertical orientation. The increase in rotor orbits when operating with the spin axis vertical is expected due to the reduced bearing stiffness experienced during unloaded operation [11].

Figure 13. Mesoscopic rotor instantaneous vibration spectrum when operating at 672 Krpm.

Figure 14. Vibration spectrum for operation with the simulator rolled 90° about the spin axis.

These tests verified the high-speed all attitude operation of the mesoscopic turbine rotor simulator. Of most importance is the lack of any real subsynchronous natural frequency vibration content in the measured spectrum. These clean spectra are indicative of a well-damped rotor bearing system. Further, the lack of observed subsynchronous vibrations throughout the test speeds indicates that operation to speeds of one-million rpm are possible. Based on previous high temperature testing of MiTi® foil bearings up to 1200°F, operation of mesoscopic turbojets appears highly feasible. For this initial test series, over 30 start stop cycles and more than 1 hour of run time were accumulated. The maximum operating speeds attained during testing


represent the first ever demonstration test of its kind.

Figure 15. Vibration spectrum for operation with simulator inverted (i.e., rolled 180° about spin axis).

Figure 16. Vibration spectrum for simulator operating with the spin axis vertical

CONCLUSIONS The successful operation of a mesoscopic turbojet simulator on compliant air foil journal and thrust bearing up to speeds as high as 705,000 rpm was demonstrated. The bearing DN values at this condition exceeds 4,200,000. The results of the experimental work showed that the air bearing performance scaled to the current mesoscopic level was successful. The performance of the newly developed technology of split bearing design was also

demonstrated. Based on this very successful demonstration of a mesoscopic sized rotor supported on compliant foil bearings, the potential to develop a 1 million rpm lubricant-free mesoscopic turbine rotor system appears highly feasible. Additional testing and developments are planned to further refine and quantify the benefits to be gained with foil bearings in mesoscopic sized applications. These data will then be used to assess the application of foil bearings to numerous other applications for both military and commercial systems. Specific potential applications include turbojets for MAV propulsion and MPG drives. Additional applications include mesoscopic turbochargers for miniature diesel engines, and motor driven compressors for high power density fuel cells.

ACKNOWLEDGMENT The work was supported by MiTi®, as an internal research and development program. The authors would like to thank MiTi® for supporting this project. Special thanks are reserved for Mr. David Slezak for his fabrication of the mesoscale foil bearings and Mr. Romano Sandomenico for machining critical parts used in the testing.

REFERENCES 1. Isomura, K. et al., 2002, “Development Of

Microturbocharger And Microcombustor For A Three-Dimensional Gas Turbine At Microscale”, Paper, GT-2002-3058, Proceedings of ASME Turbo EXPO.

2. Uechi, H., Kimijima, S., Kasagi, N., 2001, “Cycle analysis of gas turbine-fuel cell hybrid micro generation system”, proceedings of JPGC 01, New Orleans, LA, June 4-7.

3. Kang, S., Stampfl, J., Cooper, A., Prinz, F., 2001, “Application Of The Mold SDM Process To The Fabrication Of Ceramic Parts For A Mircro Gas Turbine Engine” from publication of Rapid Prototyping Laboratory, Stanford University, Stanford, CA 94305.

4. Mohawk Innovative Technology Inc.(MiTi®) Development Newsletter, 2003, “Mesoscopic Turbojet Simulator Tested at Speeds Above 700,000 rpm on Air Foil Bearings”, Vol. 17. January 2003.

5. Heshmat, H. and Ku, C.-P.R. "Structural Damping of Self-Acting Compliant Foil Journal Bearings." ASME Paper No. 93-


Trib-16, Presented at the STLE/ASME Tribology Conference, October 24-27, 1993 New Orleans, Louisiana, ASME Trans., J. of Trib., Vol. 116, No. 1 January 1994, PP 76-82.

6. Salehi, M., Heshmat, H., 2001, “Frictional Dampers Dynamic Characterization-Theory and Experiments” The 28th Leeds-Lyon Symposium on Boundary and Mixed Lubrication, Sep. 3rd to 7th , Vienna, Austria, Elsevier Science Publishers B.V., 2001, pp: 559-575.

7. Heshmat, C.A., Xu, D. and Heshmat, H., “ Analysis of Gas Lubricated Foil Thrust Bearings Using Coupled Finite Element and Finite Difference Methods,” presented at ASME/STLE Joint Tribology Conference Orlando, Fl October 10-13,1999, Transaction of ASME, Journal of Tribology, Vol. 122, pp: 199-204, January 2000.

8. Heshmat, H., “The Integration of Structural and Fluid Film Dynamic Elements in Foil Bearings- Part I: Past Approaches to the Problem,” Proceedings of the 1999 ASME

Design Engineering Technical Conferences, 17th Biennial Conference on Mechanical Vibration and Noise, ASME Paper No. DETC/VIB-8271, September 12-15, 1999 Las Vegas, Nevada

9. Salehi, M., Heshmat, H., Walton, J., 2003, “The Frictional Damping Characterization of Compliant Bump Foils”, Accepted for publication in ASME journal of Tribology, Vol. 125, No. 3.

10. Heshmat, H. “Operation of Foil Bearings Beyond the Bending Critical Mode,” presented at ASME/STLE Joint Tribology Conference Orlando, Fl October 10-13,1999, Transaction of ASME, Journal of Tribology, Vol. 122, pp: 192-198, January 2000.

11. Walton, J.F. and Heshmat, H., “Application of Foil Bearings to Turbomachinery Including Vertical Operation,” Presented at ASME Turbo Expo, Land, Sea and Air, June 1999, Indianapolis, IN. ASME Paper 99-GT-391

Documents

Ultra-Highspeed Mesoscopic Rotor On Foil Bearings