NASA/CR—2012-217664

Vibration Propagation of Gear Dynamics in a

Gear-Bearing-Housing System Using
Mathematical Modeling and Finite
Element Analysis

Robert G. Parker, Yi Guo, Tugan Eritenel, and Tristan M. Ericson

The Ohio State University, Columbus, Ohio

August 2012
 NASA STI Program . . . in Profile

Since its founding, NASA has been dedicated to the • CONFERENCE PUBLICATION. Collected
advancement of aeronautics and space science. The papers from scientific and technical
NASA Scientific and Technical Information (STI) conferences, symposia, seminars, or other
program plays a key part in helping NASA maintain meetings sponsored or cosponsored by NASA.
this important role.
• SPECIAL PUBLICATION. Scientific,
The NASA STI Program operates under the auspices technical, or historical information from
of the Agency Chief Information Officer. It collects, NASA programs, projects, and missions, often
organizes, provides for archiving, and disseminates concerned with subjects having substantial
NASA’s STI. The NASA STI program provides access public interest.
to the NASA Aeronautics and Space Database and
its public interface, the NASA Technical Reports • TECHNICAL TRANSLATION. English-
Server, thus providing one of the largest collections language translations of foreign scientific and
of aeronautical and space science STI in the world. technical material pertinent to NASA’s mission.
Results are published in both non-NASA channels
and by NASA in the NASA STI Report Series, which Specialized services also include creating custom
includes the following report types: thesauri, building customized databases, organizing
and publishing research results.
• TECHNICAL PUBLICATION. Reports of
completed research or a major significant phase For more information about the NASA STI
of research that present the results of NASA program, see the following:
programs and include extensive data or theoretical
analysis. Includes compilations of significant • Access the NASA STI program home page at
scientific and technical data and information http://www.sti.nasa.gov
deemed to be of continuing reference value.
NASA counterpart of peer-reviewed formal • E-mail your question to [email protected]
professional papers but has less stringent
limitations on manuscript length and extent of • Fax your question to the NASA STI
graphic presentations. Information Desk at 443–757–5803

• TECHNICAL MEMORANDUM. Scientific • Phone the NASA STI Information Desk at

and technical findings that are preliminary or 443–757–5802
of specialized interest, e.g., quick release
reports, working papers, and bibliographies that • Write to:
contain minimal annotation. Does not contain STI Information Desk
extensive analysis. NASA Center for AeroSpace Information
7115 Standard Drive
• CONTRACTOR REPORT. Scientific and Hanover, MD 21076–1320
technical findings by NASA-sponsored
contractors and grantees.
NASA/CR—2012-217664

Vibration Propagation of Gear Dynamics in a

Gear-Bearing-Housing System Using
Mathematical Modeling and Finite
Element Analysis

Robert G. Parker, Yi Guo, Tugan Eritenel, and Tristan M. Ericson

The Ohio State University, Columbus, Ohio

Prepared under Contract NNC08CB03C

National Aeronautics and

Space Administration

Glenn Research Center

Cleveland, Ohio 44135

August 2012
 Trade names and trademarks are used in this report for identification
only. Their usage does not constitute an official endorsement,
either expressed or implied, by the National Aeronautics and
Space Administration.

This work was sponsored by the Fundamental Aeronautics Program

at the NASA Glenn Research Center.

Level of Review: This material has been technically reviewed by NASA technical management OR expert reviewer(s).

Available from
NASA Center for Aerospace Information National Technical Information Service
7115 Standard Drive 5301 Shawnee Road
Hanover, MD 21076–1320 Alexandria, VA 22312

Available electronically at http://www.sti.nasa.gov

 Vibration Propagation of Gear Dynamics in a
Gear-Bearing-Housing System Using
Mathematical Modeling and
Finite Element Analysis

Robert G. Parker, Yi Guo, Tugan Eritenel, and Tristan M. Ericson

The Ohio State University
Columbus, Ohio 43210

NASA/CR—2012-217664 1
Contents
1 Introduction 9

2 Preliminary Step: Experimental Modal Test 11

2.1 Translational Impact to Gear Tooth . . . . . . . . . . . . . . . . . . . . . . . . . 12
2.2 Translational Impact to Gear Shaft . . . . . . . . . . . . . . . . . . . . . . . . . 13
2.3 Impact to Gearbox Input Shaft . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
2.4 Impact to Rear of Gearbox Housing . . . . . . . . . . . . . . . . . . . . . . . . . 15
2.5 Impact Test Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16

3 Gearbox Vibro-Acoustic Propagation Analysis Method Overview 17

4 Finite Element/Contact Mechanics Modeling and Analysis of the Gearbox 19

4.1 Gearbox Modeling Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
4.1.1 Contact Solver . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
4.2 Fluid Film Wave Bearing Modeling . . . . . . . . . . . . . . . . . . . . . . . . . 23
4.3 Rolling Element Bearing Modeling and Analysis . . . . . . . . . . . . . . . . . . 23
4.3.1 Cross-Coupling Bearing Stiffnesses . . . . . . . . . . . . . . . . . . . . . 25
4.4 Gear Transmission Error . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
4.5 Shaft Modeling and Validation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
4.6 Dynamic Analysis and Correlation with Experiments . . . . . . . . . . . . . . . 30

5 Analytical Gearbox Dynamic Modeling and Analysis 32

5.1 The Gear Pair . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
5.2 Incorporating Housing Compliance with the Analytical Gear/Shaft Model . . . . 39
5.2.1 Including the Housing by Using Influence Coefficients . . . . . . . . . . . 41
5.3 Dynamic Analysis Using the Analytical Model . . . . . . . . . . . . . . . . . . . 41

6 Acoustic Radiation Gearbox Modeling and Analysis 43

6.1 Model Validation and Mesh Convergence Study . . . . . . . . . . . . . . . . . . 43
6.2 Sound Pressure and Power Computation Using Transfer Functions . . . . . . . . 45
6.3 Radiated Noise Correlation with Measurements . . . . . . . . . . . . . . . . . . 47
6.4 Noise Radiation Properties with Different Bearings . . . . . . . . . . . . . . . . 51

NASA/CR—2012-217664 2
7 Summary and Conclusions 52

NASA/CR—2012-217664 3
List of Figures
1 Fluid film wave bearing. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
2 Nine mounting locations for additional OSU accelerometers in and around gearbox. 12
3 Impact locations for relevant tests: (a) translational impact to gear tooth with
accelerometers A5 and A6 mounted to the pinion shaft, (b) impact to output gear
shaft with accelerometers A5 and A6 mounted to the pinion shaft, (c) impact to
rig input shaft with accelerometers A8 and A9 mounted to the input shaft, (d)
impact to rear of gearbox housing with accelerometers A3, A4 and A7 mounted
to the gearbox housing. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
4 Frequency response functions for impact test to output gear tooth with accelerom-
eters A5 and A6 mounted to the pinion shaft and accelerometers A1 and A2
mounted to the input shaft pillow block for (a) mean NASA accelerometers and
(b) select OSU accelerometers. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
5 Frequency response comparison for (a) impact to pinion gear tooth, and (b) im-
pact to gear shaft. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
6 Frequency response functions for impact test to gearbox input shaft with ac-
celerometers A8 and A9 mounted to the input shaft and accelerometers A1 and
A2 mounted to the input shaft pillow block for (a) four NASA accelerometers
and (b) select OSU accelerometers. . . . . . . . . . . . . . . . . . . . . . . . . . 15
7 Frequency response functions for impact test to rear of gearbox housing with
accelerometers A3 and A4 mounted to the front and rear of gearbox respectively
and accelerometers A1 and A2 mounted to the input shaft pillow block for (a)
mean NASA accelerometers and (b) select OSU accelerometers. . . . . . . . . . 16
8 (a) Outside and (b) inside of the gearbox at NASA Glenn Research Center. . . . 18
9 Key steps to perform the gearbox acoustic analysis. . . . . . . . . . . . . . . . . 19
10 Cut-away finite element mesh of the radial ball bearing used in the gearbox (di-
mensions detailed in Table 2) and a double-row cylindrical rolling element bearing
(from a helicopter application). . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
11 Assembly of the gear-bearing-shaft-housing model. . . . . . . . . . . . . . . . . . 22
12 (a) Mating gears in the NASA GRC gearbox; (b) contact pressure on gear teeth
over one mesh cycle; . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23

NASA/CR—2012-217664 4
13 (a) Finite element model of a double row cylindrical bearing (outer race is re-
moved); (b) contact pattern on one of the loaded cylinders. . . . . . . . . . . . . 24
14 Radial stiffness of examined cylindrical bearing and radial ball bearing vs. applied
radial loads calculated by the Harris (−) [1], Gargiulo (·−·−) [2], and While (−−)
ΔF
[3] models. The While [3] model is modified to use Δq
to calculate the stiffness
F
instead of q
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
15 Comparison between the proposed method with zero (· · ·  · · · ) and 0.01 mm
(· −  · −) radial clearances and Kraus et al.’s [4] experiment (− ◦ −) for radial
and axial stiffness of the ball bearing in [4] under axial preloads. . . . . . . . . . 26
16 Comparison between the proposed method (· · ·  · · · ) and Royston and Basdogan
[5] experiment (− ◦ −) for radial and axial stiffness of self-aligning ball bearing
in [5] under radial and axial preloads, respectively. . . . . . . . . . . . . . . . . . 27
17 (a) Radial and (b) tilting stiffness of the cylindrical bearing over a ball pass
period. The bottom figures show the number of rolling elements in contact over a
ball pass period. The applied load and moment are 1000 N and 1 N m, respectively. 28
18 (a) Radial and (b) tilting stiffness of the ball bearing over a ball pass period. The
bottom figures show the number of rolling elements in contact over a ball pass
period. The applied load and moment are 1000 N and 1 N m, respectively. . . . . 29
19 Numerical torque impulse response of gear dynamic transmission error with fully-
populated (−) and diagonal (−−) stiffness matrices of the rolling element bearings
mounted in the examined gearbox based on [6]. The input torque equals 84.74 N m. 30
20 Coordinates of the examined gear pair. . . . . . . . . . . . . . . . . . . . . . . . 30
21 Static transmission error of the gear pair without the shaft and bearing compli-
ance. The results are calculated by finite element (−), Program X (− · −), NASA
DANST (· · · ), and Load Distribution Program (−−). The torque equals 79.09 N m. 31
22 Peak to peak amplitude of static transmission error of the gear pair with (− × −)
and without (− ◦ −) shafts at different torques. . . . . . . . . . . . . . . . . . . 32

NASA/CR—2012-217664 5
23 (a) Input shaft bending deformation calculated by analytical beam theory (solid
line) and finite element method (square marker) with simply-supported boundary
conditions under various input torques; (b) Input shaft torsional deformation cal-
culated by analytical beam theory (solid line) and finite element method (square
marker) with clamped-free boundary conditions under various input torques. The
shaft has uniform outer diameter (30.23 mm). . . . . . . . . . . . . . . . . . . . 33
24 Shaft bending deformation under various input torques. . . . . . . . . . . . . . . 34
25 Numerical impulse test results of (a) dynamic transmission error and (b) the
input shaft horizontal displacement of the gear-bearing-housing system within
speed range from 0 Hz to 7000 Hz. The applied torque is 79.09 N m. . . . . . . . 35
26 Analytical model of the gear pair. The parameters are defined in [7]. The dashed
line is at the center of the active facewidth. . . . . . . . . . . . . . . . . . . . . . 36
27 (a) Distributed spring network over a contact line with the local and bulk stiff-
nesses, kc l(t) and kb . (b) Local kc l(t) and bulk kb stiffnesses are combined into
contact stiffness ki (q, t) by Eq. (14). . . . . . . . . . . . . . . . . . . . . . . . . . 37
28 Static transmission error from the analytical (solid line) and finite element (cir-
cles) model. (a) A helical gear pair. Quadratic tip relief starting at α = 28 deg
and root relief at α = 27 deg. Tip relief, root relief, and circular lead crown are
10 μm. The applied torque is 200 N-m. (b) Spur gear pair in [8]. Linear tip relief
starting at α = 23.6 deg with amplitude 10 μm. Circular lead crown is 5 μm.
The applied torque is 340 N-m. . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
29 Description of the connection between the gears, shafts and bearings to the hous-
ing model. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
30 Graphical representation of the 12th mode at 3893 Hz (mesh deflection mode)
from the system with ball/cylindrical bearings with housing. . . . . . . . . . . . 43
31 Peak-to-peak dynamic transmission error of four systems. Analysis with roller
element bearings are marked by B, analysis with wave bearings are marked by
W, analysis including the housing flexibility is marked by H, analysis without the
housing is marked by nH. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
32 Dynamic bearing forces. Analysis with roller element bearings are marked by
B, analysis with wave bearings are marked by W, analysis including the housing
flexibility is marked by H, analysis without the housing is marked by nH. . . . . 45

NASA/CR—2012-217664 6
33 Boundary element model of the gearbox established in Coustyx. . . . . . . . . . 46
34 Theoretical solutions used to validate the boundary element housing model. Noise
radiated from the housing with monopole velocity field at the gearbox surface as
the boundary condition equals to that with monopole in the free space. . . . . . 47
35 Effects of boundary element length on the relative error of calculated sound pres-
sure compared to the theoretical solution. . . . . . . . . . . . . . . . . . . . . . 48
36 Sound pressure at the NASA microphone 1 location calculated by Coustyx ()
and theoretical models (−). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
37 Sound pressure transfer functions when unit dynamic loads are applied at bear-
ings. Six transfer functions are generated per each bearing along x, y, z, θx , θy , θz
directions. The input torque is 79.09 N m. . . . . . . . . . . . . . . . . . . . . . 49
38 Microphones (1 and 2) mounted above the NASA GRC gearbox. . . . . . . . . . 50
39 The time average of the experimental (−−) and calculated (−) mean squared
sound pressure at the microphone 1 location within mesh frequency range from
500 to 3000 Hz. The applied torque is 79.09 N m. . . . . . . . . . . . . . . . . . 50
40 Frequency spectrum of the measured (left) and simulated (right) sound pressure
at the microphone 1 location when mesh frequency is 2000 Hz. The applied
torque is 79.09 N m. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
41 The frequency spectrum of the measured (left) and computed (right) sound pres-
sure at the microphone 1 location when mesh frequency is between 1000 Hz and
2500 Hz. The applied torque is 79.09 N m. . . . . . . . . . . . . . . . . . . . . . 52
42 Sound power of radiated gearbox noise excited by certain mesh frequency har-
monics of bearing forces with rolling element (−) and fluid film wave bearings
(−−) at 79.09 N m input torque. The excitation from 1st to 6th mesh frequency
harmonics of bearing forces are considered during the computation. A weighing
filter (ISO standard) is used to adjust sound pressure levels. . . . . . . . . . . . 53

NASA/CR—2012-217664 7
List of Tables
1 Natural frequencies observed in the NASA GRC gear test rig by impact testing. 17
2 Single row cylindrical roller and ball bearing parameters. . . . . . . . . . . . . . 20
3 Dimensions of the spur gear pair . . . . . . . . . . . . . . . . . . . . . . . . . . 21
4 Natural frequencies predicted by numerical impulse tests and measurements (Hz) 31
5 Natural frequencies of the NASA gearbox with cylindrical roller/ball bearings,
wave bearings, with housing flexibility and without housing flexibility. The mesh
deflection mode natural frequencies are highlighted. . . . . . . . . . . . . . . . . 42

NASA/CR—2012-217664 8
 Abstract
Vibration and noise caused by gear dynamics at the meshing teeth propagate through
power transmission components to the surrounding environment. The purpose of this
work is to develop computational tools to investigate the vibro-acoustic propagation of
gear vibration and to investigate the effect of different bearing types on noise radia-
tion. Detailed finite element/contact mechanics and boundary element models of the
gear/bearing/housing system are established to compute the system vibration and noise
propagation. Both vibration and acoustic models are validated by experiments including
vibration modal testing and sound field measurements.
Bearings are critical components in drivetrains. Accurate modeling of rolling element
bearings is essential to assess vibration and noise of drivetrain systems. This study also
seeks to fully describe the vibro-acoustic propagation of gear dynamics through a power-
transmission system using rolling element and fluid film wave bearings. Fluid film wave
bearings have higher damping than rolling element bearings and so could offer an energy
dissipation mechanism that reduces the gearbox noise. The effectiveness of each bearing
type to disrupt vibration propagation is explored using multi-body computational models.
These models take into account gears, shafts, rolling element and fluid film wave bearings,
and the housing. Radiated noise is mapped from the gearbox surface to surrounding
environment. The effectiveness of each bearing type to disrupt vibration propagation is
speed-dependent. Housing plays an important role in noise radiation. It, however, has
limited effects on gear dynamics.

1 Introduction
Gearbox vibration contributes the structural-borne noise in helicopters [9]. The gearbox
noise consists of a wide range of gear mesh, shaft, and bearing frequencies within relative
lower audio frequency range than the jet engine noise, which is another source of helicopter
structural-borne noise. Systematic studies of the noise and vibration behavior are essential
to design quiet helicopters and to identify noise and vibration sources. Limited work has,
however, investigated the relationship between the gearbox noise and vibration.
Experimental vibro-acoustic analysis of helicopters requires effort [9, 10, 11, 12, 13]
because advanced signal processing techniques are needed to separate the gear, shaft, and
bearing signals [14]. The analytical or computational vibro-acoustic analysis of geared sys-
tems is also sparse [15] in the literature due to the complexity of these problems. Structural-
borne noise calculations of geared systems are, furthermore, semi-empirical [16, 17]. These

NASA/CR—2012-217664 9
acoustic models are not able to capture realistic dimensions of gearboxes which often
have complicated structures, which affect the noise estimate. This work defines the vibro-
acoustic behavior of gear dynamics through power-transmission systems using multi-body
dynamics gearbox models established in [18, 19] and an in-house program.
Dynamic forces at the meshing teeth drive a system vibration through power trans-
mission components to the fuselage. The bearings linking the gear shafts to the housing
are a primary factor in this noise path. Fluid film wave bearings are a special type of
journal bearings, which have waved inner diameters of the stationary bearing sides shown
in Figure 1. The wave profile amplitude is a few micrometer so that it is not obvious
in this figure. The wave bearing technology used for gas turbine lubrication [20, 21] is
recently applied to the planet bearings used in aviation planetary gears [22, 23, 20]. This
technology provides higher stiffness and better lubrication for the bearings. Experiments
on an aviation gearbox [22] with wave bearings demonstrate 25% higher load capacity com-
pared to plain journal bearings. Dimofte [20] compared the load capacity between wave
and journal bearings through an analytical formulation. He concluded that wave bearings
are more stable than plain journal bearings under light-load or unloaded conditions in any
operating regime. Furthermore, wave amplitude and starting positions of the wave profiles
are important parameters affecting wave bearing performance.
Machinery applications use rolling element bearings that do not create meaningful
damping to reduce the transmitted structural-borne noise. Wave bearings have higher
damping [24] and could offer energy dissipation. If fluid film wave bearings are shown to
withstand harsh operating conditions and provide better vibration characteristics, they
may prove an attractive alternative to standard rolling element bearings.
The major objectives of this study are to: 1) develop a finite element gearbox model
which includes the detailed contact analysis of the gear tooth mesh and individual bearing
rolling elements; 2) build up analytical (lumped parameter) model of the gear/bearing/housing
system, which provides efficient dynamic analysis; 3) establish a boundary element model
of the gearbox housing and map the radiated noise from the gearbox surface through
acoustic analysis; 4) validate the vibration and acoustic models of the examined gearbox
against measurements and theoretical solutions in the literature; and 5) understand the
effectiveness of fluid film wave bearings in breaking the vibro-acoustic propagation path
from the gears to the housing.

NASA/CR—2012-217664 10
 Figure 1: Fluid film wave bearing.

2 Preliminary Step: Experimental Modal Test

Frequency response measurements provide vital system information: natural frequencies,
mode shapes, and damping characteristics. Modal analysis by impact testing establishes
a baseline of comparison for mathematical simulations. Computational and analytical
tools are used to fully characterize the performance of rolling element and fluid film wave
bearings. Experimental measurements provide benchmarks, against which mathematical
models are validated.
Impulse test data was collected on the NASA GRC gear test rig by Ohio State re-
searchers in November, 2008. Data was collected to provide frequency response curves of
the gear test rig for multiple inputs and outputs. This data provides transfer functions
across system components and can be used to extract vibration mode shapes. A stan-
dard impact hammer with built-in force transducer was used to provide a known input
excitation. Accelerometers were used to obtain dynamic response. The known input and
measured output yield system frequency response characteristics that are independent of
the testing method. These results help predict the system behavior in response to gear
vibration, establishing benchmarks that validate the modeling effort. The experimental
study also helps defining the system for computational/analytical study.
At the time of the OSU experiments, four accelerometers were already installed by
NASA researchers. One accelerometer was mounted vertically and one horizontally on
each of the pinion shaft bearings. Nine additional accelerometers were mounted and used
by OSU to measure the frequency response around the test gearbox. Figure 2 shows the
approximate locations of these accelerometers mounted in and around the gearbox. Impact
tests were performed throughout the system. This report contains findings from four of
the most relevant impact tests. Figure 3 identifies the impact locations for these four tests.
Three impacts were averaged for each experiment. Data was taken with the gearbox cover

NASA/CR—2012-217664 11
 off to measure vibrations inside the gearbox. The subsequent results are shown on a dB
scale.

Figure 2: Nine mounting locations for additional OSU accelerometers in and around gearbox.

2.1 Translational Impact to Gear Tooth

In this test an impact was made to a gear tooth to simulate excitation near the meshing
gears. Figure 3(a) shows the impact point and two relevant accelerometers mounted on
the pinion shaft (A5 and A6). The averaged frequency response function from the NASA
accelerometers is shown in Figure 4(a). The frequency response functions from the relevant
OSU accelerometers for this test are shown in Figure 4(b).
The data in Figure 4 shows distinct natural frequencies at 750 Hz and 2500 Hz in the
NASA bearing accelerometers (a) and the additional accelerometers (A5 and A6) mounted
to the pinion shaft (b). The pinion shaft also shows a smaller peak around 1800 Hz. Both
graphs show less-defined dynamic behavior between 3300 Hz and 3800 Hz. Accelerometers
A1 and A2, which are mounted to the input shaft pillow block outside the gearbox (Figure
2), show a reduction of vibration amplitudes by about 20 dB. This shows that vibration
propagation is well-contained within the gearbox. Most importantly, this test shows two
natural frequencies associated with the gears and/or shafts at 750 Hz and 2500 Hz which
appear to be contained within the gearbox.

NASA/CR—2012-217664 12
 (a) (b)

(c) (d)

Figure 3: Impact locations for relevant tests: (a) translational impact to gear tooth with ac-
celerometers A5 and A6 mounted to the pinion shaft, (b) impact to output gear shaft with
accelerometers A5 and A6 mounted to the pinion shaft, (c) impact to rig input shaft with ac-
celerometers A8 and A9 mounted to the input shaft, (d) impact to rear of gearbox housing with
accelerometers A3, A4 and A7 mounted to the gearbox housing.

2.2 Translational Impact to Gear Shaft

In this test an impact was made to the output gear shaft, between the gear and the
supporting bearings. This impact location is near the previous test, and it was performed
to determine if the dominant 750 Hz and 2500 Hz peaks are predominantly gear or shaft
modes. Figure 3(b) shows the impact point and two relevant accelerometers mounted on
the pinion shaft (A5 and A6). The averaged frequency response function from the NASA
accelerometers and the frequency response functions from the relevant OSU accelerometers
for this test are shown in Figure 5(b) and compared to the results from the previous test
to the gear tooth in Figure 5(a).
The results from the test on the gear shaft show the same two dominant peaks at
750 Hz and 2500 Hz. The amplitudes, however, have changed significantly. The 750 Hz
peak is about 5 dB higher in the impact to the gear tooth, but the 2500 Hz peak is about
10 dB higher in the impact to the gear shaft. This suggests that the 2500 Hz mode is

NASA/CR—2012-217664 13
 Mean NASA Accelerometers FRF OSU Accelerometers FRFs
10 10
A1
0 0 A2
A5
-10 -10 A6

-20 -20

dB (g/lbf)
dB (g/lbf)
-30 -30

-40 -40

-50 -50

-60 -60
500 1000 1500 2000 2500 3000 3500 4000 4500 5000 500 1000 1500 2000 2500 3000 3500 4000 4500 5000
Frequency (Hz) Frequency (Hz)

(a) (b)

Figure 4: Frequency response functions for impact test to output gear tooth with accelerometers
A5 and A6 mounted to the pinion shaft and accelerometers A1 and A2 mounted to the input
shaft pillow block for (a) mean NASA accelerometers and (b) select OSU accelerometers.

dominated by shaft vibration and the 750 Hz mode has more gear body translation.
NASA and OSU Accelerometers FRFs NASA and OSU Accelerometers FRFs
5 20
NASA NASA
0
A5 10 A5
-10 A6 A6
0
-20
dB (g/lbf)

dB (g/lbf)

-10
-30 -20
-40 -30
-50 -40

-60 -50
500 1000 1500 2000 2500 3000 3500 4000 4500 5000 500 1000 1500 2000 2500 3000 3500 4000 4500 5000
Frequency (Hz) Frequency (Hz)

(a) (b)

Figure 5: Frequency response comparison for (a) impact to pinion gear tooth, and (b) impact
to gear shaft.

2.3 Impact to Gearbox Input Shaft

In this test an impact was made to the large, hollow input shaft of the test rig. This test was
performed to measure vibration transmissibility into and out of the gearbox. Knowing that
this shaft (and the output shaft like it) will have bending and torsional modes within the
frequency range of interest, it was necessary to determine if these accessory components
were coupled to the gearbox system under study. Figure 3(c) shows the impact point
and two relevant accelerometers mounted on the input shaft itself (A8 and A9). All four
frequency response functions from the NASA accelerometers are shown in Figure 6(a). The
frequency response functions from the relevant OSU accelerometers for this test are shown
in Figure 6(b).

NASA/CR—2012-217664 14
 The data in Figure 6(b) shows five distinct natural frequencies of the input shaft be-
tween 3050 Hz and 4500 Hz, measured by A8 and A9. This is within the frequency range of
expected gear dynamics. Therefore, if the input shaft is coupled with the gearbox system,
it would need to be modeled. Accelerometers A1 and A2, which are mounted to the input
shaft pillow block outside the gearbox (Figure 2), show a reduction of vibration amplitudes
by at least 20 dB and do not pick up any of the input shaft modes. In addition, all four
NASA bearing accelerometers do not pick up any vibration from this impact, not even in
trace vibrations between 3050 Hz and 4500 Hz. This suggests that the gearbox system
under study is not coupled with the accessory drive components and permits confident
modeling of the chosen system.
NASA Accelerometers FRFs OSU Accelerometers FRFs
40 40
30 2V 30 A1
2H A2
20 4V
20 A8
10 4H 10 A9
dB (g/lbf)

0 0

dB (g/lbf)
-10 -10
-20 -20
-30 -30
-40 -40
-50 -50
-60 -60
500 1000 1500 2000 2500 3000 3500 4000 4500 5000 500 1000 1500 2000 2500 3000 3500 4000 4500 5000
Frequency (Hz) Frequency (Hz)

(a) (b)

Figure 6: Frequency response functions for impact test to gearbox input shaft with accelerome-
ters A8 and A9 mounted to the input shaft and accelerometers A1 and A2 mounted to the input
shaft pillow block for (a) four NASA accelerometers and (b) select OSU accelerometers.

2.4 Impact to Rear of Gearbox Housing

In this test an impact was made to the rear of the gearbox housing to estimate the dominant
gearbox structural modes. Figure 3(d) shows the impact point and one of the accelerome-
ters mounted to the gearbox (A4). The mean frequency response function from the NASA
accelerometers is shown in Figure 7(a), and the frequency response functions from the
relevant OSU accelerometers for this test are shown in Figure 7(b).
Figure 7(b) shows multiple natural frequencies picked up by accelerometers mounted
to the gearbox housing: 550 Hz, 1000 Hz, 2000 Hz, and 2800 Hz. Two of these peaks
(2000 Hz and 2800 Hz), are evident in the NASA bearing accelerometers of Figure 7(a),
but the two lower-frequency peaks are not apparent. This data provides an estimate of the
primary gearbox modes and shows which of these includes significant bearing dynamics.

NASA/CR—2012-217664 15
 As seen before, accelerometers A1 and A2, which are mounted to the input shaft pillow
block outside the gearbox, show a reduction of vibration amplitudes by about 20 dB and
hardly pick up the natural frequencies at 2000 Hz and 2800 Hz. This adds confidence to
the assumption that the gearbox system under study is not significantly coupled with the
accessory drive components, validating the modeling approach.
Mean NASA Accelerometers FRF OSU Accelerometer FRFs
20 20
10 10
0 0
-10 -10

dB (g/lbf)
dB (g/lbf)

-20 -20
-30 -30
A1
-40 -40 A2
-50 -50 A3
A4
-60 -60
500 1000 1500 2000 2500 3000 3500 4000 4500 5000 500 1000 1500 2000 2500 3000 3500 4000 4500 5000
Frequency (Hz) Frequency (Hz)

(a) (b)

Figure 7: Frequency response functions for impact test to rear of gearbox housing with ac-
celerometers A3 and A4 mounted to the front and rear of gearbox respectively and accelerome-
ters A1 and A2 mounted to the input shaft pillow block for (a) mean NASA accelerometers and
(b) select OSU accelerometers.

2.5 Impact Test Summary

Important information was learned from these brief impulse tests. General system behav-
ior was characterized by identifying several important natural frequencies. With thirteen
total accelerometers, it was easy to identify the components of principal vibration at these
natural frequencies. These tests also helped establish boundaries of the system for compu-
tational and analytical studies. While the gears, shafts, bearings, and housing are clearly
coupled, it is apparent that connecting input/output shafts, pillow blocks, the drive motor,
and dynamometer structure do not contribute significantly to the dynamic behavior of the
system. Table 1 summarizes the modes identified in this study and the most significant
components of vibration.

NASA/CR—2012-217664 16
Table 1: Natural frequencies observed in the NASA GRC gear test rig by impact testing.

Number Frequency Primary component

1 500 Hz Gearbox
2 750 Hz Gears/shafts
3 1000 Hz Gearbox
4 1800 Hz Shafts
5 2000 Hz Gearbox/bearings
6 2500 Hz Gears/shafts
7 2800 Hz Gearbox/bearings
8-12 3050 Hz − 4500 Hz Input shaft (accessory)

3 Gearbox Vibro-Acoustic Propagation Analysis

Method Overview
The vibro-acoustic gearbox modeling consists of four parts. They are: 1) full-fidelity finite
element/contact mechanics model of the entire gearbox including gears, shafts, bearings,
and the housing [18]; 2) mathematical lumped-parameter gearbox model; 3) finite element
housing model; and 4) boundary element model of the housing [19].
The finite element model of the analyzed gearbox (shown in Figures 8(a) and 8(b)) in-
cludes realistic rolling element bearings, shafts, and the housing. Other models may model
these components analytically because of the modeling complexity. The unique contact
algorithm seeks the gear and rolling element contact in three-dimensional space. This al-
gorithm has been validated against experiments [25, 26, 27, 28, 29, 30, 31]. Output of this
detailed finite element model includes dynamic bearing force, bearing stiffness, and trans-
mission error, which are crucial information for the acoustic analysis and mathematical
modeling.
The analytical model of the gearbox provides fast dynamic analysis of the gearbox.
Bearings are modeled as linear stiffnesses using 6 × 6 matrices computed by the finite
element/contact analysis. Housing is included through its static condensation compliance
matrix, computed by finite element analysis. The gear contact is modeled by a network of
springs along lines of contact. The gear model is nonlinear and captures the non-nominal
load distributions due to shaft tilting, misalignment, and profile/lead modification.

NASA/CR—2012-217664 17
 Isolated finite element housing model is established in ANSYS. The housing excitation
source is dynamic bearing forces calculated using the finite element/contact or analytical
model. The surface velocity of the bearing is obtained by the dynamic responses. This
step can not be omitted because it connects the vibration analysis of the entire gearbox
and the housing noise radiation computation in the following step.
The full-fidelity boundary element model of the gearbox housing is established in the
software Coustyx [32, 19]. The acoustic model employees a multipole method ([32]) to
provide noise calculation with realistic housing dimensions. This housing surface velocity
calculated by ANSYS is inputted into the acoustic model as its boundary condition. This
acoustic model computes the radiated noise from the housing.
Major steps of the gearbox vibro-acoustic analysis is depicted in Figure 9.

(a) (b)

Figure 8: (a) Outside and (b) inside of the gearbox at NASA Glenn Research Center.

NASA/CR—2012-217664 18
Finite Element Analysis

Housing Vibration Analysis Acoustic Analysis

Housing Surface
Velocity
Transmission Error, Dynamic Bearing
Bearing Stiffness, Forces
Housing Compliance

1
.
.. .....

.. ....
..
....

....

Analytical Analysis

Figure 9: Key steps to perform the gearbox acoustic analysis.

4 Finite Element/Contact Mechanics Modeling and

Analysis of the Gearbox
4.1 Gearbox Modeling Overview
The examined NASA GRC gearbox (shown in Figures 8(a) and 8(b)) includes a pair of
spur gears with webbed rims, two staged shafts on which the gears are mounted, four
rolling element bearings that support the shafts, and the housing. The gear parameters
are listed in Table 3. Two types of rolling element bearings are used: cylindrical rolling
bearings and deep groove ball bearings (described in Table 2). Product designations are
SKF N205ECP and SKF 6205, respectively. The height and width of the gearbox housing
is 279.4mm and 254.0mm with 330.2mm length. The housing wall and lid thickness is
6.350mm. Details of the gearbox are described in [33, 24].
The realistic bearing model captures detailed bearing mechanics as shown in Figure 10,
including rolling elements, races, and the cage (not shown). This detailed bearing model
is used to determine the full 6×6 stiffness matrix between the shafts and housing. That
matrix is used in the mathematical gearbox model to perform fast dynamic simulations,

NASA/CR—2012-217664 19
 Parameters (mm,degree) Cylindrical Roller Ball Bearing
Number of rows 1 1
Number of rolling elements 13 9
Contact angle 0 0
Pitch diameter 39.00 38.50
Bore diameter 25.00 25.00
Roller length 8.600 7.900
Roller diameter 7.500 7.900
Bearing width 15.00 15.00
Outer diameter 52.00 52.00
Outer diameter of inner raceway 31.50 34.40
Inner diameter of outer raceway 46.40 46.30
Radial clearance 40.00 × 10−3 20.00 × 10−3
Inner race crown curvature 10−7 0.520
−7
Outer race crown curvature 10 0.520

Table 2: Single row cylindrical roller and ball bearing parameters.

as discussed later.
The gearbox housing is modeled by importing the full fidelity mesh established in
commercial finite element software, PATRAN. The housing is then assembled into the
gear/bearing/shaft system as shown in Figure 11.

4.1.1 Contact Solver

The contact solver of [18] seeks contact between gear teeth and bearing rolling elements
and raceways. Mesh stiffness variation, transmission error, tooth separation, and bearing
stiffness variation are inherently included; they are outputs rather than inputs. Transmis-
sion error is the major excitation source in geared systems [34] so an accurate transmission
error estimate is crucial. The finite element/contact analysis of the gearbox provides the
required reliable transmission error estimate.
The geometric surface descriptions of the contacting bodies must be precise to fully
address the contact characteristics. Additionally, the contact area is narrow and travels
over the entire body surface. Conventional finite element analysis requires a prohibitively

NASA/CR—2012-217664 20
 (a) (b)

Figure 10: Cut-away finite element mesh of the radial ball bearing used in the gearbox (dimen-
sions detailed in Table 2) and a double-row cylindrical rolling element bearing (from a helicopter
application).

Parameters Values (mm, degree)

Number of teeth 28
Outer diameter 95.25
Root diameter 79.73
Facewidth 6.350
Module 3.175
Pressure angle 20
Center distance 88.90
Tooth thickness 4.851
Cutter edge radius 1.270
Linear tip relief 0.1778 starting at 24 degrees

Table 3: Dimensions of the spur gear pair

refined mesh to address these problems; a complete dynamic response analysis becomes
impossible in that case.
The finite element/contact mechanics model used here addresses these issues by using
a combination of the Boussinesq solution near the contacting surfaces and traditional finite
element analysis far away from the contact zones to exploit the advantages of each. The
details about this contact solver can be found in [35].
To accurately describe the contact area and pressure, the contact zone is discretized
into many small patches (grid cells). Sufficient number of grid cells within the contact

NASA/CR—2012-217664 21
 Figure 11: Assembly of the gear-bearing-shaft-housing model.

zone is essential to obtain the correct contact pressure and load distribution. The finite
element model of the gear pair and contact pressure on individual tooth over a mesh cycle
are shown in Figure 12(a) and Figure 12(b). Figure 13(a) shows the finite element model
of a double-row cylindrical rolling element bearing. Contact patches on one of the radially
loaded cylinders are shown in Figure 13(b).
When the gears and rolling elements rotate, the number of teeth and rolling elements
in contact change, as do the location, size, and shape of the contact areas. These changes
are important as they affect bearing forces, gear tooth loads, and transmission error cal-
culations. The contact solver addresses these issues by determining and analyzing the
instantaneous gear and bearing contact conditions at every time instant.
This specialized finite element/contact mechanics software allows dynamic simulations
with greater modeling fidelity than conventional finite element tools. It is validated against
benchmark studies of complex gear dynamics problems [25, 26, 27, 28, 29, 30]. In exper-
imental comparisons, it has proven accurate in capturing the complex tooth mesh forces
leading to strong nonlinearity in the dynamics of single gear pairs [26], idler [29, 36], and
planetary gears [25, 27, 28, 37]. The rolling element contact in multiple bearings has been
validated against experiments in [31]. The shafts in the gearbox introduce system compli-
ance, could cause misalignment, and eventually affect transmission error. The accuracy of
the shaft models have been validated against classical beam theories as discussed later.

NASA/CR—2012-217664 22
 (a) (b)

Figure 12: (a) Mating gears in the NASA GRC gearbox; (b) contact pressure on gear teeth over
one mesh cycle;

4.2 Fluid Film Wave Bearing Modeling

Bearings are critical components in geared systems. The fluid film wave bearings are
included in the gearbox model through the stiffness and damping matrices calculated by
the program developed by Hanford and Campbell [38, 39]. This wave bearing model uses
a perturbation method based on the Reynold equation to calculate the stiffness, damping,
pressure distribution, and load capacity of the fluid film. The program, however, is limited
to be two-dimensional by excluding the bearing tilting motion.

4.3 Rolling Element Bearing Modeling and Analysis

Theoretical bearing models [1, 40, 41, 42] make different assumptions to formulate the load-
deflection relation. These assumptions include different race elasticity, different property of
the rolling element contact, and ignoring microgeometry dimensions of bearing raceways.
Figure 14 shows the nonlinear stiffness-load relations of the cylindrical and radial ball
bearings calculated using the Harris [1], Gargiulo [2], and While [3] models. Significant
discrepancy is present among them.
The method developed to determine bearing stiffness does not make any assumptions
about the load-deflection relation. Instead, it calculates the 6 × 6 bearing stiffness matrix
by partial derivatives of applied forces and moments related to six degrees of freedom. This

NASA/CR—2012-217664 23
 (a) (b)

Figure 13: (a) Finite element model of a double row cylindrical bearing (outer race is removed);
(b) contact pattern on one of the loaded cylinders.

stiffness determination approach is detailed in [31].

Radial and axial stiffness of radial ball bearing in [4] calculated by the proposed method
under axial preloads is compared against the experiment conducted by Kraus et. al. in [4]
as shown in Figure 15. Because of the lack of information on internal radial clearance, two
clearance values of 0 and 0.01 mm are used to compute stiffness. Radial and axial stiffness of
self-aligning ball bearing in [5] calculated by the proposed method is compared against the
Royston and Basdogan [5] experiment under radial and axial preloads, respectively (Figure
16). Good agreement is evident between the proposed method and the measurements for
both bearings.
The number of rolling elements in contact oscillates when they rotate, which indicates
bearing stiffness is periodic about the ball pass frequency. The ball pass period T =
2
De is the amount of time between one rolling element leaving a reference point and
Zω(1− D )
p
the next rolling element arriving. Here, z is the number of rolling elements, ω denotes the
shaft speed, and quantities De and Dp denote the diameter of rolling elements and pitch
diameter of the bearing, respectively.
Figures 17(a) and 18(a) show radial stiffness of the cylindrical and radial ball bearings
in a ball pass period with 1000 N applied load. Radial stiffness of the cylindrical bearing
fluctuates with 16% deviation from the mean stiffness, while the number of rolling elements
in contact alternates between four and five over the ball pass period. Figures 17(b) and
18(b) show tilting stiffness of the cylindrical and ball bearings when 1 N m moment is
applied. These stiffness fluctuations can excite gearbox vibration.

NASA/CR—2012-217664 24
 600

Cylindrical Bearing
500

Radial bearing stiffness (kN/mm)

400

300

200
Ball Bearing

100

0
0 2 4 6 8 10
Radial load (kN)

Figure 14: Radial stiffness of examined cylindrical bearing and radial ball bearing vs. applied
radial loads calculated by the Harris (−) [1], Gargiulo (· − ·−) [2], and While (−−) [3] models.
ΔF F
The While [3] model is modified to use Δq
to calculate the stiffness instead of q
.

4.3.1 Cross-Coupling Bearing Stiffnesses

Traditionally, diagonal stiffness matrices are used to represent rolling element bearings.
These stiffness matrices, however, include both diagonal and off-diagonal terms. Equation
1 demonstrates the stiffness matrix structure. The quantities kxx , kyy denote radial stiff-
ness. kzz denotes axial stiffness. The quantities kθx θx , kθy θy denote tilting stiffness, which
prevents the tilting motion of the shafts. The off-diagonal stiffness falls into four cate-
gories: the coupling between radial and rotational displacements (kxθx , kxθy , kyθx , kyθy ),
the coupling between radial and axial displacements (kxz , kyz ), the coupling between axial
and rotational displacements (kzθx , kzθy ), and other coupling terms (kxy , kθx θy ). Rolling
elements are free to rotate in θz direction so that stiffness in θz direction and other related
matrix components are zeros. The stiffness matrix is symmetric because rolling element
bearings are conservative systems.
⎡ ⎤
kxx kxy kxz kxθx kxθy 0
⎢ ⎥
⎢ kyy kyz kyθx kyθy 0 ⎥
⎢ ⎥
⎢ ⎥
⎢ kzz kzθx kzθy 0 ⎥
K=⎢ ⎢
⎥
⎥ (1)
⎢ kθx θx k θx θy 0 ⎥
⎢ ⎥
⎢ Symmetric k θy θy 0 ⎥
⎣ ⎦
0

Cross-coupling terms in the stiffness matrix indicate interactions between radial, axial,

NASA/CR—2012-217664 25
 100

90

Radial/axial stiffness (kN/mm)

80 Radial stiffness

70

60

50

40

30
Axial stiffness
20

10

0
50 100 150 200 250 300 350 400
Axial load (N)

Figure 15: Comparison between the proposed method with zero (· · ·  · · · ) and 0.01 mm (· −
 · −) radial clearances and Kraus et al.’s [4] experiment (− ◦ −) for radial and axial stiffness
of the ball bearing in [4] under axial preloads.

and tilting motions of rolling element bearings. They demonstrate the coupling between
the shaft tilting motion, the flexural motion of the structure connected to the outer race,
and the shaft radial and axial motions. The effects of cross-coupling terms on the gearbox
vibration transmissibility through rolling element bearings are investigated. As the primary
excitation source, gear transmission error is an important measure of gearbox vibration.
Figure 19 shows the spectra of dynamic transmission error in the frequency range from 1500
to 4000 Hz from numerical torque impulse cases. Bearing models with fully-populated and
diagonal stiffness matrices are compared as shown in Figure 19. Differences in resonant
frequencies and amplitudes are evident between these two bearing models. This stresses
the significance of the cross-coupling stiffnesses.

4.4 Gear Transmission Error

Transmission error is the major excitation source in geared systems. Accurate transmis-
sion error estimate is crucial. With the precise contact solver, the finite element/contact
analysis of the gearbox provides reliable transmission error estimate.
Transmission error is computed according to the tooth mesh deflection T E = (x1 −
x2 )sin(α) + (y1 − y2 )cos(α) + r1 θ1 + r2 θ2 , where xi , yi , θi , i = 1, 2 are the coordinates of
the mating gears as shown in Figure 20. α is the pressure angle and ri , i = 1, 2 are the
radii of the meshing gears. This formulation includes the shaft and bearing compliance.

NASA/CR—2012-217664 26
 100

90 Radial stiffness, zero axial load

80

Radial/axial stiffness (kN/mm)

70

60

50

40

30
Axial stiffness, zero radial load
20

10

0
100 150 200 250 300 350 400 450
Radial/axial load (N)

Figure 16: Comparison between the proposed method (· · ·  · · · ) and Royston and Basdogan [5]
experiment (− ◦ −) for radial and axial stiffness of self-aligning ball bearing in [5] under radial
and axial preloads, respectively.

Transmission error of this gear pair without shaft or bearing compliance has been com-
pared among Program X, Load Distribution Program, NASA DANST [43], and the current
approach in Figure 21. Program X is multi-body dynamics software that is used by in-
dustries worldwide. We are not free to state its name because of license restrictions for
academic use. The agreement on the peak to peak amplitude is reasonable. Differences
among the mean amplitudes are present. These differences are mainly caused by different
rim models these programs have used. Gear rims introduce compliance into the system,
leading to high amplitude of the mean transmission error. The current model includes the
realistic rim (as shown in Figure 12(a)). Others model the rims differently by excluding
the rim shoulders. Thus, their estimates of transmission error are lower. The Harris map
of transmission error at various torques is shown in Figure 22. The minimum transmis-
sion error without including the shaft compliance is at 67.79 N m torque, which matches
the torque the gear teeth are modified at. This further validates the transmission error
estimate.
In addition, the minimum of the peak to peak value of transmission error is at lower
torque (57.62 N m) when flexible shafts are included as shown in Figure 22. This suggests
shaft compliance needs to be considered to estimate transmission error and modify gear

NASA/CR—2012-217664 27
 240 0.93

230 0.92

Tilting Bearing Stiffness (kNm/rad)

Radial Bearing Stiffness (kN/mm)
0.91
220
0.90
210
Mean Stiffness 0.89
Mean Stiffness
200 0.88

190 0.87

0.86
180
0.85
170
0.84
160
# of Rollers in Contact

# of Rollers in Contact
0 0.2 0.4 0.6 0.8 1 0.830 0.6 0.8 1
0.2 0.4

5 5 7 7
6 6 6
4

0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1

Ball Pass Period Ball Pass Period

(a) (b)

Figure 17: (a) Radial and (b) tilting stiffness of the cylindrical bearing over a ball pass period.
The bottom figures show the number of rolling elements in contact over a ball pass period. The
applied load and moment are 1000 N and 1 N m, respectively.

teeth.

4.5 Shaft Modeling and Validation

The long shafts in the gearbox introduce system compliance, could cause misalignment,
and eventually affect transmission error. The accuracy of the shaft models are important.
The shaft bending and torsional deformations are compared against classical beam
theories. The shaft bending model is considered as the elastic beam with a concentrated
load applied at the gear location. The beam bending boundary conditions are chosen as
simply-supported at each end of the shaft. The transverse deflection y along the shaft is
calculated as
P b[x3 −(L2 −b2 )x]3/2
y= 6EIL ,x <a
(2)
π( 4 −R4
Rout )
I= in
4

where the parameter P is the applied concentrated force. The parameters a, b are the
distances between one shaft end and the location where P is applied. The quantities
E, I, L denote the Young’s modulus, moment of inertia, and shaft length. The quantities
Rout , Rin denote the shaft outer and inner radii.
The analytical model to calculate the shaft torsional deflection has the clamped-free

NASA/CR—2012-217664 28
 88 5.25

5.20

Tilting Bearing Stiffness (KNm/rad)

86
5.15
Radial Bearing Stiffness (MN/m) 5.10
84
5.05
Mean Stiffness Mean Stiffness
82 5.00

4.95
80
4.90

78 4.85

4.80
76
# of Rollers in Contact

4.75

# of Rollers in Contact
0 0.2 0.4 0.6 0.8 1 0 0.6 0.8 1
0.2 0.4
5 5 4 4 4
4
3 3

0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1

Ball Pass Period Ball Pass Period

(a) (b)

Figure 18: (a) Radial and (b) tilting stiffness of the ball bearing over a ball pass period. The
bottom figures show the number of rolling elements in contact over a ball pass period. The
applied load and moment are 1000 N and 1 N m, respectively.

boundary condition. A torque T is applied at the free end of the shaft. The torsional
deflection θ is calculated as
θ = T x/GJ

G= E
2(1+ν)
(3)

 4 − R4

J= π
2 Rout in

where the quantities ν, J denote the Possion’s ratio and the second moment of inertia.
Figures 23(a) and 23(b) show the shaft bending and torsional deformations calculated
by analytical solutions (solid line) and finite element results (square marker) at various
torques/forces. The finite element results of shaft bending and torsional deformations
agree with the analytical predictions.
In addition, as shown in Figure 4.5, the bending deformation is the same order of
magnitude of transmission error near the operating torque. Shaft deformation is significant.
Thus, including shafts would increase the overall accuracy of gearbox modeling.

NASA/CR—2012-217664 29
 ) (μm/Nm)
0
10

Dynamic transmission error

-1
10

Input torque
-2
Log ( 10

1500 1700 1900 3300 3500 3700 3900

Frequency (Hz)

Figure 19: Numerical torque impulse response of gear dynamic transmission error with fully-
populated (−) and diagonal (−−) stiffness matrices of the rolling element bearings mounted in
the examined gearbox based on [6]. The input torque equals 84.74 N m.

y1
y2

o1 x1 α o2 x2
α
r1 r2
Line of Action
θ2
θ1

Figure 20: Coordinates of the examined gear pair.

4.6 Dynamic Analysis and Correlation with Experiments

Modal analysis is performed numerically on the gearbox model by applying a torque im-
pulse at the input shaft and measuring the dynamic response. Numerical impulse test
provides natural frequencies and mode shapes of the gearbox. Results of the impulse test
are correlated with experimental measurements conducted by NASA personnel [44] and
the Ohio State research group.
The frequency spectrum of dynamic transmission error and shaft displacement for the
speed range from 0 to 8000 Hz are shown in Figures 25(a) and 25(b). The computed
natural frequencies agree with experiments as compared in Table 4. NASA accelerometers
identify natural frequencies near 3000 to 4500 Hz, but the coherence is relatively low at

NASA/CR—2012-217664 30
 30

28

26

Static Transmission Error (μm)

24

22

20

18

16

14

12

10
0 0.2 0.4 0.6 0.8 1
Mesh Cycle

Figure 21: Static transmission error of the gear pair without the shaft and bearing compliance.
The results are calculated by finite element (−), Program X (− · −), NASA DANST (· · · ), and
Load Distribution Program (−−). The torque equals 79.09 N m.

these higher frequencies. Measurements show several modes near 6500 to 7500 Hz that
are difficult to resolve, while simulations predict only one natural frequency at 6856 Hz.
The mode shapes of these natural frequencies include mesh deflection modes, shaft modes,
housing modes, and coupled modes.

Finite Element Experiment

759 750
1942 2000
2195, 2283 2500
3803, 4449, 4654 3000-4500
6856 6500-7500

Table 4: Natural frequencies predicted by numerical impulse tests and measurements (Hz)

NASA/CR—2012-217664 31
 7

Peak to Peak Transmission Error (μm)

6

1
Designed Torque
0
10 20 30 40 50 60 70 80 90 100 110
Torque (Nm)

Figure 22: Peak to peak amplitude of static transmission error of the gear pair with (− × −)
and without (− ◦ −) shafts at different torques.

5 Analytical Gearbox Dynamic Modeling and Anal-

ysis
Mathematical modeling of the gearbox provides fast dynamic analysis of the gearbox.
Dynamic analysis, especially at multiple operating speeds, requires much computational
effort using the finite element model. Analytical model gives the natural frequencies and
mode shapes, and dynamic response at various operating speeds.
This analytical model employees crucial information on transmission error, mesh stiff-
ness, bearing stiffness, and housing compliance from the finite element analysis.

5.1 The Gear Pair

The gear pair model is constructed following the literature [45, 46, 47, 48, 49, 50, 51], in
which detail discussions into the validity of the modeling can be found. The model consists
of two gears mounted on shafts. Each gear body is combined with its supporting shaft
into a single rigid body. These gear-shaft bodies are each mounted on up two bearings
placed at arbitrary axial locations. Figure 26 shows the gear model and the bases. A fixed,
right-handed, orthonormal basis {E} = {E1 , E2 , E3 } is oriented such that E1 is parallel to
the line of action of the gear mesh. The origin is on the rotation axis of the pinion body,
midway in the active facewidth. This dimensioning allows arbitrary axial positioning of
meshing gears with different facewidth because any inactive facewidth is treated as part of

NASA/CR—2012-217664 32
 90 4

101.7 Nm

Shaft Torsional Deformation (10-3 radians)

80 101.7 Nm 3.5

70
Shaft Bending Deformation (μm)

3
84.74 Nm
60
2.5 84.74 Nm
50
2
40
1.5
33.90 Nm 33.90 Nm
30

1
20

10 0.5

0 0
50 100 150 200 250 300 350 50 100 150 200 250 300 350
Shaft Length (mm) Shaft Length (mm)

(a) (b)

Figure 23: (a) Input shaft bending deformation calculated by analytical beam theory (solid line)
and finite element method (square marker) with simply-supported boundary conditions under
various input torques; (b) Input shaft torsional deformation calculated by analytical beam theory
(solid line) and finite element method (square marker) with clamped-free boundary conditions
under various input torques. The shaft has uniform outer diameter (30.23 mm).

the shaft. The translational (xp , yp , zp ) and angular (φp , θp , βp ) coordinates of the pinion
body are assigned to translations along and rotations about E1 , E2 , and E3 , respectively.
The translational and angular coordinates of the gear body follow similarly with subscript
g. Body-fixed bases {ep } = {ep1 , ep2 , ep3 } and {eg } = {eg1 , eg2 , eg3 } for the pinion and gear are
adopted. Positive axial quantities are measured along E3 from the dashed line in Figure
26.
The pinion translational and angular velocity vectors are

ṙp = ẋp E1 + ẏp E2 + żp E3 ,

 
 
(4)
ω p = φ̇p − θp β̇p + Ωp ep1 + θ̇p + φp β̇p + Ωp ep2 + β̇p + Ωp − φp θ̇p ep3 ,

where Ωp is the constant angular rotational speed of the pinion. The velocity vectors for
the gear are identical except with components for the gear.
The pinion body is supported by two bearings at points Ap and Bp . The axial positions
of these bearings measured along E3 are LA
p and Lp . The pinion bearing deflection vectors
B

at point Ap and Bp are the relative deflections of points Ap and Bp with respect to ground,

NASA/CR—2012-217664 33
 90

80

70

Magnitude of Shaft Bending (μm)

60 101.7 Nm

50
79.09 Nm
40

30
33.90 Nm

20

10

0
100 150 200 250 300
Shaft Length (mm)

Figure 24: Shaft bending deformation under various input torques.

giving
  A    
p = θp Lp − ep + xp E1 + φp ep − Lp + yp E2 + zp E3 ,
dA A

  B     (5)
p = θp Lp − ep + xp E1 + φp ep − Lp + yp E2 + zp E3 .
dB B

The bearing deflections for the gear follow similarly. The bearings resist tilting as well.
The angular deflection of the pinion body bearing at Ap is

p = φ p E1 + θp E2 + β p E3 .
γA (6)

The angular bearing deflection at point Bp is identical to Eq. (6) for rigid shafts. The
bearings are isotropic in the E1 − E2 plane. At point Ap , the bearing stiffness matrix KA
p
as given in Eq. (1) is fully-populated, where the equality of stiffness in the two translation
directions is evident. The bearing translational and angular displacements combined are
{d
Γ= }. Similar definitions follow for point Bp and for the gear body.
γ
The gear mesh interface is modeled by a series of springs along the nominal lines of
contact for no mesh deflection. These lines change as the gears rotate. Each spring acts
at a point denoted by Ci . When the gear bodies deflect, the contact points on the pinion
separate or compress against the contact points on the gear. The difference between
the position vectors of the contact points on the pinion and gear gives the relative mesh
deflection vector at Ci . The projection of the relative mesh deflection vector on the tooth
surface normal gives the relative compressive deflection at the ith contact point. The

NASA/CR—2012-217664 34
 4.5
0.7
4449Hz 759Hz
4
0.6
(μm/Nm)

(μm/Nm)
3.5
0.5
3
Dynamic transmission error

5806Hz

Shaft Vertical Displacement

0.4 2.5
3803Hz
Input torque

0.3 6856Hz 2

Input torque
2195Hz 1942Hz
1.5
2283Hz 4654Hz
0.2
1
3865Hz
0.1
0.5

0 0
1000 2000 3000 4000 5000 6000 7000 8000 1000 2000 3000 4000 5000 6000 7000 8000
Frequency (Hz) Frequency (Hz)

(a) (b)

Figure 25: Numerical impulse test results of (a) dynamic transmission error and (b) the input
shaft horizontal displacement of the gear-bearing-housing system within speed range from 0 Hz
to 7000 Hz. The applied torque is 79.09 N m.

relative compressive deflection is

 
δi (q, t) = [ep − ci (t)] θp + [ci (t) − eg ] θg − xp + xg + hi + βp rp + βg rg cos ψ
  (7)
− [bi (t) + hi ] θp + [(rp + rg ) θg tan Φ − bi (t)] + zp − zg + φp rp + φg rg sin ψ,

where rp and rg are the base radii, Φ is the transverse operating pressure angle, and ψ is
the base helix angle. The vector q comprises generalized coordinates
 
q= φp , θp , βp , xp , yp , zp , φg , θg , βg , xg , yg , zg (8)
     
pinion gear

The axial position of a contact point is ci (t) measured from the origin along E3 , and the
transverse position of a contact point is bi (t) measured from the origin along −E1 . They
are known functions of time determined by the contact line progressions as the gears rotate.
Micron-level deviations of the tooth surface from an involute at any contact point i, such
as from gear tooth surface modifications and manufacturing errors, are denoted by hi .
Figures 26 and 27(b) depict these quantities.

NASA/CR—2012-217664 35
 Gear
g
LAg LBg e2
yg
Ag Bg Gear g
e1
xg
eg rg bi
Φ

Contact
points Ci E2
ci Pinion rp
Contact
Ap E3 Bp points Ci Φ E1
xp
ep
Pinion
LAp LBp

Figure 26: Analytical model of the gear pair. The parameters are defined in [7]. The dashed
line is at the center of the active facewidth.

The kinetic and potential energies are

1 T
T = ω p Jp ω p + ω Tg Jg ω g + ṙTp mp ṙp + ṙTg mg ṙg ,
2
1
AT A A T B B AT BT

V = Γp Kp Γp + ΓB p K p Γ p + Γ p K G Γ p + Γ p K G Γp
A A B B
2 (9)
1
n(t)
+ ki (q, t)δi (q, t)2 ,
2
i=1

where ki (q, t) is the ith contact stiffness, and n(t) is the number of contact segments at
an instant t. These quantities change as the gears rotate, hence the time dependence.
 
The inertia tensor of the axisymmetric pinion body is Jp = diag Jpx , Jpx , Jpz with similar
definition for the axisymmetric gear body.
Lagrange’s equations of motion for unconstrained generalized coordinates follow after
substitution of equations Eqs. (4) through (7) into the energy expressions Eq. (9). In
matrix form they are
 
Ms q̈s + Dq̇s + Ωp Gs q̇s + Ks (q, t) − Ω2p Cs q = F(q, t)s , (10)

The vector F includes external loading; the driving and absorbing torques and tooth surface
modifications hi appear here. The matrix K represents the system elasticity with losses
contained in the modal damping matrix D. Tooth surface modifications hi are neglected
in K because hi  bi (t), and the hi appear as additions to bi (t). The terms that arise from

NASA/CR—2012-217664 36
 Contact
point Ci
Gear Gear
{
hi
kc hi{ Contact
kb ki point Ci
l(t)

{
ψ ci bi
Pinion Pinion

(a) (b)

Figure 27: (a) Distributed spring network over a contact line with the local and bulk stiffnesses,
kc l(t) and kb . (b) Local kc l(t) and bulk kb stiffnesses are combined into contact stiffness ki (q, t)
by Eq. (14).

the constant rotation speed are contained in the gyroscopic matrix G and the centripetal
acceleration matrix C. Individual elements of M, K, G, C, and f are given in [7].
Following [52, 50] the nominal contact lines are discretized into n(t) segments of equal
length l(t), as shown in Figure 27(a). Each contact point Ci is positioned at the center of
its segment. As the contact lines progress with gear rotation, the total number of segments
n(t) and the length of a segment l(t) change. Each contact line has a specified number
of segments. This discretization is based on the nominal lines of contact with no gear
deflections.
Each contact spring is attached to its contact point Ci . The stiffness ki (q, t) of contact
springs are obtained by considering two separate categories of tooth deflection: local (i )
and bulk (δb ). Discussion of this categorization can be found in [52, 49, 53]. The local
deflection represents the Hertz contact deflections. The associated local stiffness is kc l(t),
where the constant kc is the local stiffness per unit contact length. The bulk deflection
represents all deflections except local deflection, and those include gear blank deflection,
tooth bending, shear, etc. Because the Hertz contact deflections are localized and far
enough from the bulk deflections, the bulk deflection is assumed to be the same for all
contact segments. The bulk stiffness kb is assumed constant. The bulk spring is in series
with the local springs, so the total deflection at the ith contact point Ci is

δ i = i + δ b (11)

The mesh force F equals the sum of all forces carried by the local springs and also the

NASA/CR—2012-217664 37
force carried by the bulk spring due to the series connection. The mesh force is

n 
n
F = Fi = kc l(t) i H(i ) = kb δb , (12)
i=1 i=1

1; i ≥ 0
H(i ) = (13)
0; i < 0
is the Heaviside function that represents the contact condition at each contact spring. Use
of Eqs. (11) and (12) reduce the network of local and bulk springs into n(t) contact springs
(ki , i = 1, 2, . . . , n) in parallel, as shown in Figure 27(b). The ith contact stiffness is given
by
kb kc l(t)H(i )
ki (q, t) = . (14)

n(t)
kb + kc l(t) H(i )
i=1

One can then calculate the ith potential energy ki δi2 /2 stored in the gear mesh in Eq. (9)
from the contact force Fi = ki δi .
The local stiffness per unit length kc and the bulk stiffness kb are parameters of the
gear pair determined by the contact mechanics and elasticity of the gears. These constants
can be approximated analytically [52] or semi-analytically [50] by assigning certain types
of stiffnesses such as Hertz contact, tooth bending, shear, etc. to kc and kb . A different
approach is used in this work, where kc and kb are solved for from the deflections obtained
from an external analysis tool for computational static analyses. In this case, kc and kb are
numerical values that best fit the deflection obtained from finite element contact analysis
of gears.
The following stipulations simplify the algebra to find kc and kb . 1) The tooth surface
is perfectly involute, that is, hi = 0 for all i; 2) All degrees of freedom are constrained
to be zero except the pinion rotation βp ; and 3) A specified moment about E3 is applied
to the pinion. With these stipulations, the deflections at all contact points are identical,
that is, δi = δ for all i in Eq. (7). Consequently, all points are in contact; Hi = 1 for
all i. The subscript i of ki is unnecessary because when all segments are in contact,
k1 = k2 = . . . = kn . Use of static equilibrium, Eq. (11), and Eq. (14) gives
 
1 1
δ(t) = F + , (15)
kb kc L(t)
where δ(t) is the static transmission error, L(t) = n(t)l(t) is the total contact line length
at an instant t, and F is the constant mesh force obtained from the known applied torque.
The two unknowns (kc , kb ) are solved using the data from finite element analysis results

NASA/CR—2012-217664 38
 at two instances {δ(t1 ), L(t1 )} and {δ(t2 ), L(t2 )} within a mesh cycle. To increase accu-
racy, values at these two instances are calculated from averages of the four points where
transmission error is highest (giving the values for the first instance) and the four points
where transmission error is lowest (giving the values for the second instance).
The gear mesh model is validated by finite element analysis. Figure 28 shows trans-
mission error comparison between the finite element and the analytical model for a helical
and a spur gear pair.

7.5
Transmission Error [μm]

Transmission Error [μm]

7 Analytical 24 Analytical

FEA
6.5 22

FEA
20
6
0 T/4 T/2 3T/4 T 0 T/4 T/2 3T/4 T
Gear Mesh Cycle Gear Mesh Cycle
(a) (b)

Figure 28: Static transmission error from the analytical (solid line) and finite element (circles)
model. (a) A helical gear pair. Quadratic tip relief starting at α = 28 deg and root relief at
α = 27 deg. Tip relief, root relief, and circular lead crown are 10 μm. The applied torque is
200 N-m. (b) Spur gear pair in [8]. Linear tip relief starting at α = 23.6 deg with amplitude 10
μm. Circular lead crown is 5 μm. The applied torque is 340 N-m.

5.2 Incorporating Housing Compliance with the Analytical

Gear/Shaft Model
The equations of motion for the housing is given by

MH q̈H + KH qH = FH (16)

where MH and KH are the mass and the stiffness matrices with degrees of freedom qH
and the force vector fH . Figure 29 helps explain the approach. The gear/shaft equations
of motion are expanded by the addition of bearing degrees of freedom qb and the housing
equation of motion by Partitioning of (16) to separate the degrees of freedom where the

NASA/CR—2012-217664 39
bearings are attached from the rest of the housing giving
       
Mc 0 q̈c Kc Kch qc fc
+ = (17)
0 Mh q̈h KTch Kh qh fh

The new equations of motion are given by

       
Ms 0 q̈s Ks Ksb qs fs
+ = (18)
0 Mb q̈b KTsb Kb qb fb

where, the matrix Ksb includes the coupling between the bearing degrees of freedom and
the system. Mc and Kc are the mass and stiffness matrices, qc is the degrees of freedom,
and fc is the force at the connection points. The bearing ends are massless, Mb = 0.
The rest of the housing mass, stiffness, degrees of freedom and forcing are contained in
Mh , Kh , qh , and fh . The connection between the bearings and the housing requires that
at the interface displacements are equal, qb = qc , and the forces are transmitted fb = −fc .
Eliminating qc using these connection requirements gives
⎡ ⎤⎛ ⎞ ⎡ ⎤⎛ ⎞ ⎛ ⎞
Ms 0 0 q̈s Ks Ksb 0 qs fs
⎢ ⎥⎜ ⎟ ⎢ ⎥⎜ ⎟ ⎜ ⎟
⎢ 0 Mb 0 ⎥ ⎜ q̈b ⎟ ⎢ T ⎥ ⎜ qb ⎟ = ⎜ 0 ⎟
⎣ ⎦⎝ ⎠ + ⎣ Ksb Kc + Kb Kch ⎦⎝ ⎠ ⎝ ⎠ (19)
0 0 Mh q̈h 0 KTch Kh qh fh

This is the general form that connects the gears, shafts, and bearings to the full housing
model which includes the flexibility as well as the inertia. If the housing inertia is neglected,
i.e. static condensation, Mh = Mb = 0 and there are no external forces on the housing
fh = 0, the last row of (19) becomes qh = −K−1 T
h Kch qb . Substitution of qh into the second
 
T −1 KT q . Substitution of q into the
row of (19) gives qb = − Kb + Kc − Kch K−1 h K ch sb s b
first row of (19) gives
   
T −1 T
Ms q̈s + Ks − Ksb Kb + Kc − Kch K−1 K K sb qs = 0 (20)
  h ch
Kf

This equation incorporates the housing flexibility, without added degrees of freedom and
the housing inertia, into the equations of motion of the gears, shafts, and bearings. To this
end, the calculation of Kc , Kch , and Kh separately is unnecessary. A shorter path can be
taken by evaluating the housing stiffness Kf at the connection interface using influence
coefficients of the housing (16). This is discussed in the following subsection.

NASA/CR—2012-217664 40
 Gears & shafts
Ms,Ks
Bearings Connection
Kb Mc, Kc
Housing
Mh,Kh
xs xb xc

Figure 29: Description of the connection between the gears, shafts and bearings to the housing
model.

5.2.1 Including the Housing by Using Influence Coefficients

The influence coefficient matrix Cf can be solved from (16) Cf fc = qc Solving for qh
from (17) gives
C−1 −1 T
f = Kf = Kc − Kch Kh Kch (21)

which allows use of the influence coefficient matrix in (20).

The procedure to obtain elements of Cn×n
f consists of applying a unit force to each
degree of freedom at the connection points fc one at a time, and measuring the deflection
of all the degrees of freedom at the connection points qc . Explicitly, elements in the ith
column, C<i><j>
f , are given by the displacements q<j>
c , calculated by (16) when there is
a unit force at fc<i> = 1, where j = 1, . . . , n.

5.3 Dynamic Analysis Using the Analytical Model

Table 5 shows the natural frequencies of four systems: 1) cylindrical/ball bearings with
housing flexibility, 2) wave bearings with housing flexibility, 3) cylindrical/ball bearings
without housing flexibility, and 4) wave bearings without housing flexibility. Figure 30
shows the mesh deflection mode from the analysis with cylindrical/ball bearings with
housing flexibility. The mesh deflection mode is an important mode because the tooth
pass excitation excites primarily the mesh deflection mode. This mesh deflection mode is
demonstrated in experiments, which is strongly excited by the 2nd harmonic of tooth pass
frequency at 4200rpm. This mode shows up strongly in our vibration speed maps. It is

NASA/CR—2012-217664 41
 highlighted in Table 5. The mesh deflection mode persists when bearing type is changed,
and when the housing flexibility is included.
Linear dynamic analysis is performed using the mean system stiffness matrix obtained
by averaging the quasi-static time-dependent system stiffness over a mesh period. The
harmonic excitation is approximated from the first ten harmonics of the quasi-static dis-
placement vector. Figure 31 shows the dynamic peak-to-peak transmission error from
the four systems mentioned above. All systems have similar dynamic transmission error.
Figure 32 shows the dynamic bearing reactions of the four systems. Wave bearings seem
to produce higher dynamic bearing loads. Use of the housing flexibility moves the natu-
ral frequencies and thus the resonant peaks, but does not alter the peak amplitudes in a
significant manner.

Cylindrical Roller Ball Bearing

With Housing Without Housing With Housing Without Housing
576.6 975.5 875.2 2279
954.6 1000 1988 2535
987.7 1597 2246 3173
1515 1711 2387 3209
1599 1868 2749 3480
1793 1977 3003 3549
1915 2736 3150 3982
2075 2848 3373 4383
2356 3910 3514 4473
2722 - 3957 -
3893 - 4392 -

Table 5: Natural frequencies of the NASA gearbox with cylindrical roller/ball bearings, wave
bearings, with housing flexibility and without housing flexibility. The mesh deflection mode
natural frequencies are highlighted.

NASA/CR—2012-217664 42
Figure 30: Graphical representation of the 12th mode at 3893 Hz (mesh deflection mode) from
the system with ball/cylindrical bearings with housing.

6 Acoustic Radiation Gearbox Modeling and Anal-

ysis
The boundary element model of the gearbox housing is shown in Figure 33. This model
includes 50,401 tetrahedron elements. Each element has nearly uniform length in order
to have good aspect ratio. This model uses the fast multipole method (detailed in [32])
to calculate the gearbox sound radiation. This method provides fast acoustic calculation
with a realistic system modeling. Given the complexity of housings, the great increase in
the number of degrees of freedom admits solutions for realistic models not possible with
conventional boundary element software. This method is implemented in Coustyx [19].

6.1 Model Validation and Mesh Convergence Study

This boundary element model of the gearbox has been validated against theoretical solu-
tions of the sound field with a monopole. A monopole is a source which radiates sound
equally well in all directions. The simplest example of a monopole source would be a
sphere whose radius alternately expands and contracts sinusoidally. The far-field pressure
radiated by a monopole may be written as
iQρck i(ωt−kr)
p(r, θ, t) = e (22)
4πr
where ρ is the fluid density, c is the speed of sound, k is the wave number, ω is the
frequency, and r is the local distance from source. Q is the complex source strength, which

NASA/CR—2012-217664 43
 30
B/nH
25 B/H
W/nH
20 W/H

PP TE [µm]
15

10

0
0 1000 2000 3000 4000 5000
Frequency [Hz]

Figure 31: Peak-to-peak dynamic transmission error of four systems. Analysis with roller ele-
ment bearings are marked by B, analysis with wave bearings are marked by W, analysis including
the housing flexibility is marked by H, analysis without the housing is marked by nH.

is a constant number defined in [54].

The sound field excited by a monopole is not affected by the gearbox geometry, which
allows simulation results to be compared against the theoretical solution of a monopole
for an arbitrary geometry. In other words, the noise radiation from the gearbox surface
equals to the theoretical solution of a monopole in the free space (shown in Figure 34).
The velocity field caused by the point source (monopole) in free space at gearbox surface is
calculated theoretically and used as the boundary condition for the actual housing model.
Optimal element length has been carefully chosen through mesh convergence study.
The relative error of the computed pressure compared to the theoretical value in (22)
with various element length is shown in Figure 35. Result accuracy is improved when
element length is small. However, when element length is smaller than 8.636 mm, the
result accuracy is disrupted as round off error becomes significant. The optimal element
length is between 8.890 mm and 10.16 mm with less than 2% relative error. The mesh
convergence study is important as the element length affects the solution accuracy.
Sound pressure of the radiated noise calculated by Coustyx and the theoretical solution
is compared in Figure 36 for the speed range from 500 to 3500 Hz using 9.000 mm element
length. Excellent agreement is present.

NASA/CR—2012-217664 44
 Gear bearing A Gear bearing B
1000 1000
B/nH
800 B/H 800
W/nH

FB gear [N]
FA gear [N]
600 W/H 600

400 400

200 200

0 0
0 1000 2000 3000 4000 5000 0 1000 2000 3000 4000 5000

Pinion bearing A Pinion bearing B

1000 800

800
600

FB pinion [N]
FA pinion [N]

600
400
400
200
200

0 0
0 1000 2000 3000 4000 5000 0 1000 2000 3000 4000 5000
Frequency [Hz] Frequency [Hz]

Figure 32: Dynamic bearing forces. Analysis with roller element bearings are marked by B,
analysis with wave bearings are marked by W, analysis including the housing flexibility is marked
by H, analysis without the housing is marked by nH.

6.2 Sound Pressure and Power Computation Using Transfer

Functions
The computation of sound pressure and power of the radiated gearbox noise uses a network
of transfer functions from unit bearing loads to sound pressure. Transfer functions describe
the noise radiation property of the gearbox housing. The computation of these transfer
functions is independent of the vibration analysis of the bearing reaction forces. Thus,
these transfer functions are applicable for any type of bearings or gears.
The procedure to calculate transfer functions consists of applying a unit dynamic force
to each degree of freedom (x, y, z, θx , θy , θz ) at every bearing one at a time in the finite
element model, measuring the velocity at the gearbox surface for various speeds in the
finite element housing model, and computing the sound pressure of the radiated noise at
specified locations in the gearbox environment using the computed velocity as the boundary
condition in the boundary element model. Figure 37 shows a few calculated transfer
functions at four different bearings. Sound pressure at arbitrary locations outside the
gearbox is determined by the superposition of these transfer functions and multiplying the

NASA/CR—2012-217664 45
 Figure 33: Boundary element model of the gearbox established in Coustyx.

fluctuating amplitudes of dynamic bearing forces at each degree of freedom of individual

bearings.
Sound power is a better measure of the gearbox noise than sound pressure as it is inde-
pendent of the geometric locations where noise is measured. Sound power measurements
in experiments, however, require much effort. This study calculates the sound power using
the aforementioned transfer functions.
To obtain accurate estimate of sound power, many numerical microphones are outside
the gearbox. Those microphones are mounted on a measurement surface enveloping the
gearbox, in this study, a sphere. According to the international standard ISO 3745, 20
microphones are used. The sound power is computed by integrating the sound pressure at
these “microphones” in the sphere.
The sound power level, Lw is computed as:
s
Lw = Lp + 10log10 ( )(in dB) + c1 + c2
s0
B 313.15 0.5
c1 = −10log10 [ ( ) ](in dB) (23)
B0 273.15 + θ
B 296.15 0.5
c2 = −15log10 [ ( ) ](in dB)
B0 273.15 + θ

where s is the area of the measurement surface and s0 = 1 m2 . B is the barometric air
pressure during measurements, in Pascals; B0 is the reference barometric pressure. θ is
the air temperature during measurement. Lp is the weighed surface sound pressure level

NASA/CR—2012-217664 46
 Microphone

Monopole

Theoretical Boundary Condition Excited by Monopole

Figure 34: Theoretical solutions used to validate the boundary element housing model. Noise
radiated from the housing with monopole velocity field at the gearbox surface as the boundary
condition equals to that with monopole in the free space.

over the measurement surface in decibels.

1  0.1(Lp,i +W )
Lp = 10log10 {( [ 10 ])}in dB
N
i (24)
pi
Lp,i = 20log10 ( )
p0
where N is the number of microphone positions. W is the weighting function applied
by the filter at the frequency of analysis. pi is the root mean square pressure in Pa and
p0 = 2 × 10−5 P a.

6.3 Radiated Noise Correlation with Measurements

The calculated noise is compared against the measurements taken at NASA GRC gearbox.
Two microphones are mounted to measure sound pressure in experiments at 790 mm
directly above the gearbox, and are separated horizontally by 180 mm as shown in Figure
38.
Here, only the noise transmitted through rolling element bearings are compared against
measurements because experimental data on sound pressure with wave bearings is not
available. Figure 39 shows the time averages of the mean squared sound pressure (p)2av =
$ 2 2
b (pb )av at microphone 1 estimated by measurements and computed, where (pb )av is
the addictive measure of sound pressure associated with the frequency band b from 0 to

NASA/CR—2012-217664 47
 6

Percentage of Relative Error, Sound Pressure

5

0
7 8 9 10 11 12 13
Boundary Element Length (mm)

Figure 35: Effects of boundary element length on the relative error of calculated sound pressure
compared to the theoretical solution.

8000 Hz. For each gear mesh frequency, two different experimental results are shown:
the minimum sound pressure including only mesh frequency harmonics; and the maxi-
mum sound pressure including mesh frequency harmonics and other frequency components.
These frequencies are the side bands caused by the shaft rotation frequency, belt splice
pass frequency, etc. The simulated sound pressure is within the experimental data range.
Figure 40 shows the frequency spectrum of the measured and simulated sound pressure
when mesh frequency equals 2000 Hz. Good agreement is evident between the measure-
ments and simulations at mesh frequency harmonics. Sidebands around the mesh frequency
harmonics are present in the measured data. These sidebands are not included in the com-
putational model because the causes are not known. Including these sidebands in the
vibration/acoustics model would provide better agreement with the measurements.

NASA/CR—2012-217664 48
 20

15

Sound pressure, real part (Pa)

10

-5

-10

-15

-20
500 1000 1500 2000 2500 3000 3500
Frequency (kHz)

Figure 36: Sound pressure at the NASA microphone 1 location calculated by Coustyx () and
theoretical models (−).

0.012

y
y 0.01

θ_y z
θ_z
0.008
x
0.006
θ_x
x 0.004

θx
0.002

θy
0
0 2 4 6 8 10
Sound Pressure (Pa)~Mesh Freq.(kHz)

Figure 37: Sound pressure transfer functions when unit dynamic loads are applied at bearings.
Six transfer functions are generated per each bearing along x, y, z, θx , θy , θz directions. The input
torque is 79.09 N m.

NASA/CR—2012-217664 49
 Figure 38: Microphones (1 and 2) mounted above the NASA GRC gearbox.

100

95
Experiment, Max
90

85
Mean Squared Pressure, dB

80

75

70

65
Simulation Experiment, Min
60

55

50
500 1000 1500 2000 2500 3000
Mesh Frequency (Hz)

Figure 39: The time average of the experimental (−−) and calculated (−) mean squared sound
pressure at the microphone 1 location within mesh frequency range from 500 to 3000 Hz. The
applied torque is 79.09 N m.

NASA/CR—2012-217664 50
 Measurement Simulation
0.18 0.18

0.16 0.16

0.14 0.14

Sound Pressure, Pa

Sound Pressure, Pa
0.12 0.12

0.1 0.1

0.08 0.08

0.06 0.06

0.04 0.04

0.02 0.02

0 0
0 2000 4000 6000 8000 10000 0 2000 4000 6000 8000 10000
Mesh Freq (Hz) Mesh Freq (Hz)

Figure 40: Frequency spectrum of the measured (left) and simulated (right) sound pressure at
the microphone 1 location when mesh frequency is 2000 Hz. The applied torque is 79.09 N m.

6.4 Noise Radiation Properties with Different Bearings

This section investigates the effectiveness of fluid film wave bearings in reducing noise. To
that end, sound pressure and power of the radiated noise are compared between fluid film
wave and rolling element bearings.
Sound pressure at the microphone 1 location excited by the first (left) and second
(right) mesh frequency harmonics is compared between rolling element and wave bearings
in Figure 41. The sound pressure of wave bearings is generally higher than with rolling
element bearings. The difference is frequency dependent. When the second mesh frequency
harmonic is near 3803 Hz (dominant mesh deflection mode), sound pressure with rolling
element bearings is similar in shape and amplitude to that with wave bearings. The noise
radiation is independent of bearing types at this frequency.
The sound power radiated from the gearbox is also compared between wave and rolling
element bearings in Figure 42. The sound power level of the gearbox with rolling element
and wave bearings is also speed-dependent. The radiated gearbox sound power with rolling
element bearings is lower than that with fluid film wave bearings in general within the speed
range of interest. There is no obvious benefit to replace rolling element bearings with wave
bearings to reduce the gearbox noise within the speed range of interest, given the two
dimensional wave bearing model used [38, 39].

NASA/CR—2012-217664 51
 90 Wave Bearing 90 Wave Bearing

Rolling Element Bearing Rolling Element Bearing

80 80

Root Mean Square Pressure, dB

70 70

60 60

50 50

40 40
1st Mesh Freq Harmonic 2nd Mesh Freq Harmonic
1000 1500 2000 2500 2000 3000 4000 5000
Mesh Freq (Hz) Mesh Freq (Hz)

Figure 41: The frequency spectrum of the measured (left) and computed (right) sound pressure
at the microphone 1 location when mesh frequency is between 1000 Hz and 2500 Hz. The
applied torque is 79.09 N m.

7 Summary and Conclusions

This work investigates noise radiation from a gearbox with different types of bearings. A
multi-dynamics model of the examined gearbox is established to address the vibro-acoustic
propagation of the gear dynamics. It includes: 1). full fidelity finite element gearbox
model; 2). lumped-parameter gearbox model; 3) finite element housing model; and 4).
boundary element housing model. The procedure of the noise radiation analysis consists
of: a). dynamic bearing force computation using the finite element and analytical gearbox
models; b). gearbox surface velocity estimate with excitations from dynamic bearing forces
obtained from a); and c). acoustic analysis of the gearbox with the boundary condition of
the surface velocity calculated from b).
The full fidelity finite element/contact mechanics model of the NASA gearbox has been
validated. Gear transmission error is compared among commercial software and matches
the design objective. The rolling element and gear contact properties have been validated
against experiments in the literature. Modal analysis has been performed on the entire
gearbox model and results agree with measurements. The simplified lumped-parameter
model of the gearbox is suitable for parametric studies on the dynamic response. The
analytical model has been validated against the finite element analysis.
Cross-coupling stiffnesses are significant as they affect the gearbox dynamics. A fully-

NASA/CR—2012-217664 52
 80 80 80

70
70 70
60
60 60
50
Sound Power (dB) 50 50
40
1st Harmonic 2nd Harmonic 3rd Harmonic
40 40 30
0.5 1.0 1.5 2.0 2.5 0.5 1.0 1.5 2.0 2.5 0.5 1.0 1.5 2.0 2.5

80 80 80

60 60 60

40 40 40

20 20 20
4th Harmonic 5th Harmonic 6th Harmonic
0 0 0
0.5 1.0 1.5 2.0 2.5 0.5 1.0 1.5 2.0 2.5 0.5 1.0 1.5 2.0 2.5
Mesh Frequency, kHz

Figure 42: Sound power of radiated gearbox noise excited by certain mesh frequency harmonics
of bearing forces with rolling element (−) and fluid film wave bearings (−−) at 79.09 N m input
torque. The excitation from 1st to 6th mesh frequency harmonics of bearing forces are considered
during the computation. A weighing filter (ISO standard) is used to adjust sound pressure levels.

populated stiffness matrix is needed to fully represent rolling element bearings. Bearing
stiffnesses fluctuate periodically as rolling elements enter and leave the region of loaded
rollers. These stiffness variations could excite gearbox vibration.
This acoustic model has been validated by comparing against theoretical solutions and
measurements taken at the NASA GRC gearbox. The transfer function method determines
sound pressure and power for arbitrary bearings, which is exceptionally suitable for the
gearbox optimization study to reduce noise. Noise radiation property of fluid film wave
bearings is speed-dependent, and wave bearings are not superior to rolling element bearing
in reducing gearbox noise. An one-way vibro-acoustic pass of the gearbox is present at the
mesh deflection mode. Gear dynamics significantly affects the housing vibration and noise
radiation. Housing and bearings, however, have limited effects on the gear vibration.

NASA/CR—2012-217664 53
References
[1] Harris, T. A., 1990. Rolling bearing analysis, 3rd ed. John Wiley and Sons, Inc.

[2] Gargiulo, Jr., E. P., 1980. “A simple way to estimate bearing stiffness”. Mechine
Design, pp. 107–110.

[3] While, M. F., 1979. “Rolling element bearing vibration transfer characteristics: Effect
of stiffness”. Journal of Applied Mechanics, 46, pp. 677–684.

[4] Royston, T. J., and Basdogan, I., 1998. “Vibration transmission through self-aligning
(spherical) rolling element bearings: Theory and experiment”. Journal of Sound and
Vibration, 215(5), pp. 997–1014.

[5] Kraus, J., Blech, J. J., and Braun, S. G., 1987. “In situ determination of rolling
bearing stiffness and damping by modal analysis”. Journal of Vibration, Acoustics,
Stress, and Reliability in Design, 109, pp. 235–240.

[6] Seybert, A. F., Wu, T. W., and Atherton, W., 1992. “Comparison of analysis and
experiment for gearbox noise”. NASA Technical Memorandum 251523. National
Aeronautic and Space Administration, Cleveland Lewis Reseach Center.

[7] Eritenel, T., and Parker, R. G., 2011. “An investigation of tooth mesh nonlinearity
and partial contact loss in gear pairs using a lumped-parameter model”. Mechanism
and Machine Theory. Submitted.

[8] Kahraman, A., and Blankenship, G. W., 1999. “Effect of involute tip relief on dynamic
response of spur gear pairs”. Journal of Mechanical Design, 121(1), Mar., pp. 112–
118.

[9] Mucchi, E., and Vecchio, A., 2010. “Acoustical signature analysis of a helicopter cabin
in steady-state and run up operational conditions”. Measurement, 43, p. 283293.

[10] Hoffmann, F., Maier, R., Janker, P., Hermle, F., and Berthe, A., 2006. “Helicopter
interior noise reduction by using active gearbox structs”. in: Proceedings of the 12th
AIAA/CEAS Aerospace Conference, May. Massachusetts, Cambridge.

[11] Dozio, L., Corbetta, W., Vigono, E., Forghieri, A., Ghiringhelli, G., and Cenedese,
F., 2006. “Active solutions for the reduction of the transmission of structure-borne
noise to the cabin structure on a a109 mockup”. Proceedings of ISMA, September.
Leuven, Belgium.

NASA/CR—2012-217664 54
[12] Viswamurthy, S., and Ganguli, R., 2004. “An optimization approach to vibration
reduction in helicopter rotors with multiple active trailing edge flaps,”. Aerospace
Science and Technology, 8(3), p. 185194.
[13] Hirsch, S., Jayachandran, V., and Sun, J. Q., Composite Structures. “Structural-
acoustic control for quieter aircraft interior-smart trim technology”. 1998, 42(2),
p. 189202.
[14] Randall, R., 2004. “Detection and diagnosis of incipient bearing failure in helicopter
gearboxes”. Engineering Failure Analysis II, p. 177 190.
[15] He, S., Singh, R., and Pavic, G., 2008. “Effect of sliding friction on gear noise based on
a refined vibro-acoustic formulation”. Journal of Noise Control Engineering, 56(3),
pp. 164–175.
[16] Lim, T. C., and Singh, R., 1991. “Statistical energy analysis of a gearbox with
emphasis on the bearing path”. Journal of Noise Control Engineering, 37(2), pp. 63–
69.
[17] Lim, T. C., and Singh, R., 1989. “A review of gear housing dynamics and acoustic
literature”. NASA Contractor Report 185148.
[18] Vijayakar, S. M. Calyx User’s Manual. Advanced Numerical Solutions LLC, Hilliard,
Ohio, 43026 USA. http://www.ansol.us/Manuals/Helical3DManual.pdf.
[19] http://www.ansol.us/Products/Coustyx.
[20] Dimofte, F., 1995. “Wave journal bearing with compressible lubricant; part i: The
wave bearing concept and a comparison to the plain circular bearing”. STLE Tribology
Transaction, 38(1), p. 153160.
[21] Dimofte, F., and Addy, Jr., H. E., 1994. “Preliminary experimental results of a
three wave journal air bearing”. Proceedings of Advanced Earth-to-Orbit Propulsion
Technology Conference, 11, May, p. 375384.
[22] Dimofte, F., Proctor, M. P., Fleming, D. P., and Keith, Jr, T. G., 2000. “Wave
fluid film bearing tests for an aviation gearbox”. NASA Techanical Report,
NASA/TM2000-209766, Jan.
[23] Dimofte, F., 1995. “Wave journal bearing; part i: Analysis”. NASA Contractor Report
195431.
[24] Hambric, S. A., Hanford, A. D., Shepherd, M. R., Campbell, R. L., and Smith, E. C.,
2010. “Rotorcraft transmission noise path model, including distributed fluid film
bearing impedance modeling”. NASA Technical Report. Contract No. NNC08CB07C.

NASA/CR—2012-217664 55
[25] Parker, R. G., Agashe, V., and Vijayakar, S. M., 2000. “Dynamic response of a
planetary gear system using a finite element/contact mechanics model”. Journal of
Mechanical Design, 122(3), Sept., pp. 304–310.

[26] Parker, R. G., Vijayakar, S. M., and Imajo, T., 2000. “Non-linear dynamic response
of a spur gear pair: Modelling and experimental comparisons”. Journal of Sound and
Vibration, 237(3), Oct., pp. 435–455.

[27] Ambarisha, V. K., and Parker, R. G., 2007. “Nonlinear dynamics of planetary gears
using analytical and finite element models”. Journal of Sound and Vibration, 302,
pp. 577–595.

[28] Bahk, C.-J., and Parker, R., 2011. “Analytical solution for the nonlinear dynamics of
planetary gears”. ASME Journal of Computational and Nonlinear Dynamics, 6(2),
p. 021007.

[29] Liu, G., and Parker, R. G., 2008. “Dynamic modeling and analysis of tooth pro-
file modification of multi-mesh gears”. ASME Journal of Mechanical Design, 130,
pp. 121402–1.

[30] Liu, G., and Parker, R. G., 2008. “Impact of tooth friction and its bending effect on
gear dynamics”. Journal of Sound and Vibration, 320(4-5), Mar., pp. 1039–1063.

[31] Guo, Y., and Parker, R. G., 2010. “Stiffness matrix calculation of rolling element
bearings using a finite element/contact mechanics model”. Mechanism and Machine
Theory. Submitted.

[32] Gunda, R., 2008. “Boundary element acoustics and the fast multipole method (fmm)”.
Sound and Vibration, March.

[33] Oswald, F. B., Townsend, D. P., Valco, M. J., Spencer, R. H., Drago, R. J., and
Lenski, J. W., 1994. Influence of gear design parameters on gearbox radiated noise.
Tech. Rep. ARL TR-381, NASA.

[34] Greogory, R. W., Harris, S. L., and Munro, R. G., 1963. “Dynamic behavior of spur
gears”. Proceedings of the institution of mechanical engineers, 178, pp. 261–266.

[35] Vijayakar, S. M., 1991. “A combined surface integral and finite element solution for
a three-dimensional contact problem”. International Journal for Numerical Methods
in Engineering, 31, pp. 524–546.

[36] Liu, G., and Parker, R., 2008. “Nonlinear dynamics of idler gearsets”. Nonlinear
Dynamics, 53, pp. 345–367.

NASA/CR—2012-217664 56
[37] Guo, Y., and Parker, R. G., 2011. “Dynamic analysis of planetary gears with bearing
clearance”. ASME Journal of Computational and Nonlinear Dynamics. Submitted.
[38] Hanford, A., 2009. Wave Journal Bearing Dynamic Coefficients Users Manual.
ARL/Penn State.
[39] Campbell, R. L. a., 2003. “Distributed journal bearing dynamic coefficients for struc-
tural finite element models”. Proceedings of ASME IMECE. NCA-43770.
[40] Jones, A. B., 1946. “Analysis of stresses and deflections”. New Departure Engineering
Data. General Motors.
[41] Palmgren, A., 1959. Ball and roller bearing engineering. S. H. Burbank, Philadelphia,.
[42] Brandlein, J., Eschmann, P., Hasbargen, L., and Weigand, K., 1999. Ball and roller
bearings: theory, design and application. John Wiley and Sons.
[43] Oswald, F. B., Lin, H. H., and Delgado, I. R., 1996. “Dynamic analysis of spur gear
transmissions (danst)”. NASA Technical Memorandum 107291. Technical Report
ARLTR1189.
[44] Guo, Y., Parker, R. G., Eritenel, T., and Ericson, T. M., 2011. Vibration propagation
of gear dynamics in a gear-bearing-housing system using mathematical modeling and
finite element analysis. Tech. rep., NASA Contract NNC08CB03C.
[45] Blankenship, G. W., and Singh, R., 1995. “Dynamic force transmissibility in helical
gear pairs”. Mechanism and Machine Theory, 30(3), Apr., pp. 323–339.
[46] Blankenship, G. W., and Singh, R., 1995. “A new gear mesh interface dynamic model
to predict multi dimensional force coupling and excitation”. Mechanism and Machine
Theory, 30(1), Jan., pp. 43–57.
[47] Salzer, M., Smith, J., and Welbourn, D., 1977. “Simulation of noise from gears when
varying design and manufacturing parameters”. Proceedings of the World Congress
on Gearing, 1, pp. 297–308.
[48] Kahraman, A., 1993. “Effect of axial vibrations on the dynamics of a helical gear
pair”. Journal of Vibration and Acoustics, 115(1), Jan., pp. 33–39.
[49] Andersson, A., and Vedmar, L., 2003. “A dynamic model to determine vibrations in
involute helical gears”. Journal of Sound and Vibration, 260(2), Feb., pp. 195–212.
[50] Velex, P., and Maatar, M., 1996. “A mathematical model for analyzing the influence
of shape deviations and mounting errors on gear dynamic behavior”. Journal of Sound
and Vibration, 191(5), Apr., pp. 629–660.

NASA/CR—2012-217664 57
[51] He, S., Gunda, R., and Singh, R., 2007. “Inclusion of sliding friction in contact
dynamics model for helical gears”. Journal of Mechanical Design, 129, Jan., pp. 48–
57.

[52] Conry, T. F., and Seireg, A., 1973. “Mathematical programming technique for eval-
uation of load distribution and optimal modification for gear systems”. Journal of
Engineering for Industry, 95(4), pp. 1115–1122.

[53] Ogawa, Y., Matsumura, S., Houjoh, H., Sato, T., and Umezawa, K., 2000. “Rota-
tional vibration of a spur gear pair considering tooth helix deviation: Development of
simulator and verification”. JSME International Journal, Series C, 43(2), pp. 423–
431.

[54] Kinsler, L. E., Frey, A. R., Coppens, A. B., and Sanders, J. V., 1982. Fundamentals
of Acoustics. Wiley, New York.

NASA/CR—2012-217664 58
 Form Approved
REPORT DOCUMENTATION PAGE
OMB No. 0704-0188
The public reporting burden for this collection of information is estimated to average 1 hour per response, including the time for reviewing instructions, searching existing data sources, gathering and maintaining the
data needed, and completing and reviewing the collection of information. Send comments regarding this burden estimate or any other aspect of this collection of information, including suggestions for reducing this
burden, to Department of Defense, Washington Headquarters Services, Directorate for Information Operations and Reports (0704-0188), 1215 Jefferson Davis Highway, Suite 1204, Arlington, VA 22202-4302.
Respondents should be aware that notwithstanding any other provision of law, no person shall be subject to any penalty for failing to comply with a collection of information if it does not display a currently valid OMB
control number.
PLEASE DO NOT RETURN YOUR FORM TO THE ABOVE ADDRESS.
1. REPORT DATE (DD-MM-YYYY) 2. REPORT TYPE 3. DATES COVERED (From - To)
01-08-2012 Final Contractor Report
4. TITLE AND SUBTITLE 5a. CONTRACT NUMBER
Vibration Propagation of Gear Dynamics in a Gear-Bearing Housing System Using NNC08CB03C
Mathematical Modeling and Finite Element Analysis
5b. GRANT NUMBER

5c. PROGRAM ELEMENT NUMBER

6. AUTHOR(S) 5d. PROJECT NUMBER

Parker, Robert, G.; Guo, Yi; Eritenel, Tugan; Ericson, Tristan, M.
5e. TASK NUMBER

5f. WORK UNIT NUMBER

WBS 877868.02.07.03.04.01
7. PERFORMING ORGANIZATION NAME(S) AND ADDRESS(ES) 8. PERFORMING ORGANIZATION
The Ohio State University REPORT NUMBER
E-18341

9. SPONSORING/MONITORING AGENCY NAME(S) AND ADDRESS(ES) 10. SPONSORING/MONITOR'S

National Aeronautics and Space Administration ACRONYM(S)
Washington, DC 20546-0001 NASA
11. SPONSORING/MONITORING
REPORT NUMBER
NASA/CR-2012-217664
12. DISTRIBUTION/AVAILABILITY STATEMENT
Unclassified-Unlimited
Subject Category: 37
Available electronically at http://www.sti.nasa.gov
This publication is available from the NASA Center for AeroSpace Information, 443-757-5802

13. SUPPLEMENTARY NOTES

14. ABSTRACT
Vibration and noise caused by gear dynamics at the meshing teeth propagate through power transmission components to the surrounding
environment. This study is devoted to developing computational tools to investigate the vibro-acoustic propagation of gear dynamics
through a gearbox using different bearings. Detailed finite element/contact mechanics and boundary element models of the
gear/bearing/housing system are established to compute the system vibration and noise propagation. Both vibration and acoustic models are
validated by experiments including the vibration modal testing and sound field measurements. The effectiveness of each bearing type to
disrupt vibration propagation is speed-dependent. Housing plays an important role in noise radiation .It, however, has limited effects on gear
dynamics. Bearings are critical components in drivetrains. Accurate modeling of rolling element bearings is essential to assess vibration and
noise of drivetrain systems. This study also seeks to fully describe the vibro-acoustic propagation of gear dynamics through a power-
transmission system using rolling element and fluid film wave bearings. Fluid film wave bearings, which have higher damping than rolling
element bearings, could offer an energy dissipation mechanism that reduces the gearbox noise. The effectiveness of each bearing type to
disrupt vibration propagation in explored using multi-body computational models. These models include gears, shafts, rolling element and
fluid film wave bearings, and the housing. Radiated noise is mapped from the gearbox surface to surrounding environment. The
effectiveness of rolling element and fluid film wave bearings in breaking the vibro-acoustic propagation path from the gear to the housing is
investigated.
15. SUBJECT TERMS
Gear noise; Noise transmission; Vibration; Bearings; Rolling element bearings; Journal bearings; Wave journal bearings

16. SECURITY CLASSIFICATION OF: 17. LIMITATION OF 18. NUMBER 19a. NAME OF RESPONSIBLE PERSON
ABSTRACT OF STI Help Desk (email:[email protected])
a. REPORT b. ABSTRACT c. THIS PAGES 19b. TELEPHONE NUMBER (include area code)
U U PAGE UU 64 443-757-5802
U
Standard Form 298 (Rev. 8-98)
Prescribed by ANSI Std. Z39-18