Kostandin Gjika Engineering& Technology Fellow Honeywell Turbo Technologies

1. ASME TURBO EXPO 2009,Power for Land, Sea and Air Turbocharger Nonlinear Response with Engine-Induced Excitations: Predictions and Test Data Luis San Andrés Mast-Childs Professor Fellow ASME Kostandin Gjika Engineering& Technology Fellow Honeywell Turbo Technologies Sherry Xia Rotordynamics Manager Honeywell Turbo Technologies Ash Maruyama Research Assistant (05-07) Texas A&M University ASME Paper GT 2009-59108 Accepted for journal publication Supported by Honeywell Turbocharger Technologies (HTT)

2. Increase internal combustion (IC) engine power output by forcing more air into cylinder Aid in producing smaller, more fuel-efficient engines with larger power outputs Turbochargers:

3. RBS: TC Rotor Bearing System(s) RBS With Semi Floating Bearing RBS With Ball Bearing RBS With Fully Floating Bearing Desire for increased IC engine performance & efficiency leads to technologies that rely on robust & turbocharging solutions

4. Low shaft motion Relatively expensive Limited lifespan Locking Pin Floating Ring Outer Film Inner Film Shaft Oil Feed Hole Semi-Floating Ring Bearing (SFRB) Floating Ring Bearing (FRB) Bearing Types: Locking Pin Squeeze Film Ball Bearing Inner Race Outer Race Shaft Ball-Bearing • Economic • Longer life span • Prone to subsynchronous whirl

5. Shaw & Nussdorfer (1949): Test results show superior performance of FRBs over plain journal bearings Tatara (1970): Initially unstable FRB-supported test rotor becomes stable at high speeds, ring speed reaches constant speed Li & Rohde (1981): Numerically show FRB-supported rotors whirl in stable limit cycles Trippett & Li (1984): Shows lubricant viscosity changes cause unusual floating-ring speed behavior, isothermal analysis is incorrect ENGINE INDUCED Vibrations: Literature Review Kirk et al. (2008): Measure shaft motions of TC on FRB attached to diesel ICE. Engine-attributed low frequency amplitudes comparable to TC subsynchronous amplitudes. Little to no insight on RBS analysis Ying et al. (2008): TC-RBS NL analysis with engine foundation excitation. Rotor response is quite complicated showing chaos at the lowest shaft speed. Little to no insight on test data

6. TAMU-HTT VIRTUAL TOOL for Turbocharger NL Shaft Motion Predictions XLTRC2® based with a demonstrated 70% cycle time reduction in the development of new CV TCs. Since 2006, code aids to developing PV TCs with savings up to $150k/year in qualification test time Predicted shaft motion Measured shaft motion ASME DETC2007-34136

7. Literature Review: San Andres and students • TC linear and nonlinear rotordynamic codes – GUI based • Measure ring speeds with fiber optic sensors • Realistic thermohydrodynamic bearing models • Novel methods to estimate imbalance distribution and shaft temperatures Tools for shaft motion prediction with effect of engine excitation needed –benchmarked by tests data

8. Refine rotordynamics model by including engine-induced housing excitations Deliver predictive tools validated by test data to reduce the need for costly engine test stand qualification Further understanding of complex TC behavior Objectives: quantification TAMU-HTT publications show unique -one to one- comparisons between test data and nonlinear predictions

9. TC rotor & bearing system 2 shaft model RBS with Semi Floating Bearing

10. u C T Rotor finite element model: 2 shaft model Validate rotor model with measurements of free-fee modes (room Temp) Shaft measurements (STN 3) & predictions Thrust Collar SFRB Turbine Compressor Rotor: 6Y gram SFRB: Y gram Static gravity load distribution Compressor Side: Z Turbine Side: 5Z

11. Free-free mode natural frequency & shapes: Measured and predicted free-free natural frequencies and mode shapes agree: rotor model validation

12. (Semi) Floating Bearing Ring : • Actual geometry (length, diameter, clearance) of inner and outer films, holes size and distribution • Supply conditions: temperature & pressure • Lubricant viscosity varies with temperature and shear rate (commercial oil) • Side hydrostatic load due to feed pressure • Temperature of casing • Temperature of rotor at turbine & compressor sides derived from semi-empirical model: temperature defect model XLBRG® thermohydrodynamic fluid film bearing model predicts operating clearance and oil viscosity (inner and outer films) and eccentricities (static and dynamic) as a function of shaft & ring speeds and applied (static & dynamic) loads.

13. Operating conditions from test data: • TC speed ranges from 48krpm – 158 krpm • Engine speed ranges from 1,000 rpm – 3,600 rpm • 25%, 50%, 100% of full engine load • Nominal oil feed pressure & temperature: 2 bar, 100°C TC Engine Test Facility Stand Engine Compressor Housing Proximity Probes (X, Y) Air Inlet

14. (S)FRB Predictions : Peak film temperatures Inner film Outer film Supply temperature Increase in power losses (with speed) lead to raise in inner film & ring temperatures. No effect of engine load

15. (S)FRB Predictions : Oil effective viscosity Lubricant type:SAE 15W - 40 outer film Inner film Supply Viscosity: 8.4 cP Increased film temperatures determine lower lubricant viscosities. Operation parameters independent of engine load LUB: SAE 15W-40

16. (S)FRB Predictions : Film clearances Inner film nominal clearance outer film Clearance thermal growth relative to nominal inner or outer cold radial clearance Inner film clearance grows and outer film clearance decreases – RING grows more than SHAFT and less than CASING. Material parameters are important

17. TC housing acceleration measurements: TC center housing and compressor housing accelerations measured with 3-axes accelerometers for three engine loads: 25%, 50%, 100% of full engine load accelerometers

18. TC housing acceleration analysis: 100% engine load Center Housing m/s2 ~570 Hz Last 2,048 (out of 15,000) time data points converted to frequency spectrum via Fast Fourier Transformations (FFTs) Combined manifold & TC system natural frequencies ~300 Hz Comp. Housing m/s2 3600 rpm 1000 rpm

19. TC housing acceleration analysis: 100% engine load Center Housing m/s2 ~570 Hz 2, 4, and 6 times engine (e) main frequency contribute significantly 1e order frequency does not appear Combined manifold & TC system natural frequencies ~300 Hz Comp. Housing m/s2 3600 rpm 1000 rpm

20. TC housing total acceleration 100% engine load Compressor housing m/s2 Center housing Center and compressor housings donot vibrate as a rigid body

21. Integration of housing accelerations into rotordynamics model Displacement transducers record shaft motion relative to compressor housing Rotordynamics model outputs absolute shaft motion shaft motion relative to compressor housing needs of casing motion Displacement transducers Note: TC Housing accelerations and TC shaft motions NOT recorded simultaneously

22. Housing accelerations into model Basic assumptions • TC housings move as a rigid body • TC housing vibrations transmitted through bearing connections • Each bearing transmits identical housing vibrations

23. Rotordynamics model Z Vector of rotor & ring displacements & rotations along (X, Y) at the DOFs of interest M, K, D, G() – Matrices of rotor & ring inertias, stiffness, damping & gyroscopics at the rated rotor speed () Fext(t) Imposed time varying forces acting on the rotor & ring, such as imbalances, aerodynamics, side loads FB(t)Vector of bearing reactions forces including engine vibration excitation Shaft motion (ring motion – base motion)

24. 0 Housing accelerations into model Fourier coefficient decomposition of housing acceleration time data Double time integration Procedure: Find first 10 Fourier coefficients (amplitude and phase) of center housing acceleration and input into rotordynamics model. Run nonlinear time transient analysis and find absolute shaft motion response. Subtract compressor housing displacements to obtain shaft motion relative to compressor

25. RBS damped natural frequencies 100% engine load 1st elastic mode cyl. turb. bear. ringing mode 1X Critical speed cyl. comp. ringing mode conical mode

26. u C T RBS response to imbalance 100% engine load Test data Differences between predictions and test data attributed to inaccurate knowledge of imbalance distribution NL pred.

27. Transient time NL rotor response • XLTRC2® Nonlinear numerical integration of equation of motion (time-marching ) with bearing forces evaluated at each time step. • Gear stiff method • Component mode synthesis • Post processing in frequency domain (Virtual Tools) • Integration parameters CPU ~ 30’ per shaft speed • Results (amplitudes at) compressor nose vertical direction • shown relative to maximum conical motion at the compressor shaft end

28. Waterfalls of shaft motion at compressor end 100% engine load Housing accelerations induce broad range, low frequency shaft whirl motions Test data shows broad frequency response at low frequencies (engine speeds) 3600 rpm 1000 rpm

29. Total shaft motion at compressor end (amplitude) Test data Good correlation with test data for all shaft speeds Amplitude pk-pk (-) NL pred. Rotor speed (RPM) 100% engine load

30. Subsynchronous amplitudes vs engine speed NL pred. Good agreement b/w predictions and test data from 1750 – 2750 rpm Amplitude 0-pk (-) Test data Engine speed (RPM) 100% engine load

31. Subsynchronous amplitudes vs engine orders NL pred. 2e and 4e orders engine frequency contribute the most to shaft motions 14e order is due to shaft self-excited vibration (whirl from bearings) Amplitude 0-pk (-) Test data Orders of main engine speed 100% engine load

32. Subsynchronous frequency vs. rotor speed 2e frequency shown in test data and preds 4e frequency tracks rotor conical mode Subsynchronous frequencies ~ super-harmonics of conical mode Test 1X Group 3 (4C) ~570 Hz Subsynchronous frequency (Hz) System (manifold & TC) natural frequency ranges Group 2 (2C) ~300 Hz 4e order freq. Group 1 (0.5 C) 2e order freq. Rotor speed (RPM)

33. Subsynchronous freq. vs. IC engine speed Subsynch. freqs. are multiples of IC engine frequency Higher engine order frequencies not predicted Test Subsynchronous frequency (Hz) NL Engine speed (RPM) TC manifold nat freq. 100% engine load

34. Engines induce significant and complex, low frequency subsynchronous whirl in turbochargers 2e and 4e order frequencies contribute significantly to housing acceleration Center housing and compressor housing do not vibrate as a single rigid body Engine super-harmonics excite TC rotor damped natural frequencies. Whirl frequencies are multiples of engine speed Conclusions Good agreement between predictions and test data validates the nonlinear rotordynamics model!

35. Validation against test data from different TCs is needed Housing accelerations and TC shaft motion must be recorded simultaneouslyand for longer periods of time (smaller frequency step size)Work completed in 2008 Understand why higher order subsynchronous frequencies are not predicted Update model to account for unequal housing excitations at each bearing location Recommendations

36. Honeywell Turbocharging Technologies (2000-2008) TAMU Turbomachinery Laboratory Turbomachinery Research Consortium (XLTRC2®) Acknowledgments Learn more at http://phn.tamu/edu/TRIBgroup Questions?