Download as pdf or txt
Download as pdf or txt
You are on page 1of 9

Tribology International 88 (2015) 170–178

Contents lists available at ScienceDirect

Tribology International
journal homepage: www.elsevier.com/locate/triboint

A thermal tribo-dynamic mechanical power loss model for


spur gear Pairs
Sheng Li n
Wright State University, 3640 Colonel Glenn Highway, Dayton, OH 45435, USA

art ic l e i nf o a b s t r a c t

Article history: This study proposes a formulation for the description of the gear mesh mechanical power loss under the
Received 11 December 2014 thermal tribo-dynamic condition. A six degree-of-freedom motion equation set and the thermal mixed
Accepted 16 March 2015 elastohydrodynamic lubrication governing equations are coupled to model the mechanical power loss
Available online 24 March 2015
under the condition where the gear dynamics and tribology disciplines interact. The important role of
Keywords: the gear thermal tribo-dynamics in power loss is demonstrated by comparing the predictions of the
Thermal proposed model to those under the thermal quasi-static condition, and those under the iso-thermal
Tribo-dynamics tribo-dynamic condition. Considering an example spur gear pair, the impacts of the lubricant inlet
Gear temperature, input torque, and surface roughness on gear mesh mechanical power loss under the
Power Loss
thermal tribo-dynamic condition are also investigated.
& 2015 Elsevier Ltd. All rights reserved.

1. Introduction tribology field. A number of the early studies [4,5] used assumed
constant friction coefficients, μ, to exclude the complex descrip-
Gear mesh mechanical power loss and efficiency has become tion of the gear tribological behavior in the friction evaluation. To
an important research topic, mainly because of the high demand avoid the subjective selection of μ, and include the μ variation
for energy efficiency in both ground and aerospace vehicles, as along the LOA (that is due to the mesh-position-dependent tooth
well as the environmental concerns such as the emission of force, surface velocities, sliding, etc.), the studies such as Refs. [6–
pollutant gases and particulates. Gear pairs often operate at 8] adopted the friction coefficient formulae that were arrived from
relatively high speeds, under which condition, the gear dynamic experimental measurements. The validity of this approach, how-
behavior is often inevitable. Gear dynamics not only dictates the ever, is confined by the narrow ranges of the operating conditions,
noise and vibration levels, but also plays an important role in the surface roughness conditions, and lubricant type considered in
gear lubrication performance. Incorporating the dynamic tooth those experiments. To break this limitation, a group of researchers
force in an elastohydrodynamic lubrication (EHL) model, Li and [9–11] incorporated elastohydrodynamic lubrication models in the
Kahraman [1] showed significant impacts of the gear dynamics on gear mechanical power loss evaluation. In which way, the friction
the contact pressure and film thickness distributions under both and power loss were directly determined by solving the EHL
smooth and rough surface EHL conditions. On the other hand, the governing equations, instead of relying on any empirical formula.
sliding and rolling frictions at the gear mesh interfaces form the Although the accuracy and applicability of this much more
main excitations for the gear vibration along the off-line-of-action sophisticated method are superior, the shortcoming is the
(OLOA), and produce friction moments to couple the OLOA and increased computational time that can be cumbersome for gear
line-off-action (LOA) motions. Furthermore, the viscous shearing system design. To achieve the high numerical efficiency, high
in the lubrication film provides a power dissipation mechanism for accuracy, and the ability to be applied to wide operating ranges,
the gear mesh damping [2], which is one of the critical compo- the studies of Refs. [12–15] devised the friction formulae by
nents influencing the gear dynamic responses. These interactions regressing the massive friction data points generated by numerical
between the gear dynamics and gear tribology disciplines are EHL simulations that cover wide operating ranges of gear contacts
referred as the gear tribo-dynamics [3]. (instead of the limited experimental data). In all these gear mesh
Over the past decades, the approach for the estimation of the power loss investigations, however, the gear dynamic behavior
gear mechanical power loss has been greatly advanced in the was neglected.
In the gear dynamics field, most of the works have been
focusing on the noise generated by gear systems [16,17]. Gear
n
Tel.: þ 1 937 775 3211; fax: þ 1 937 775 5082. dynamics models of single or multi-degree-of-freedom have been
E-mail address: sheng.li@wright.edu proposed using either the discrete lumped-parameter description

http://dx.doi.org/10.1016/j.triboint.2015.03.022
0301-679X/& 2015 Elsevier Ltd. All rights reserved.
S. Li / Tribology International 88 (2015) 170–178 171

or the deformable finite element description. The main challenges


for these models are the modeling of the gear mesh damping and
the friction forces at the mesh interfaces. A number of nonlinear-
time-varying models that used assumed constant gear mesh
damping and neglected the mesh friction [18–23] were shown to
correlate well with the experimental measurements in terms of
the torsional or LOA motion only if the viscous mesh damping is
adjusted to allow such agreement. Another group of gear
dynamics studies examined the impacts of the gear mesh friction
forces on the gear dynamic responses [24–27], assuming constant
friction coefficients or using empirical friction formulae. Recogniz-
ing the interactions between the gear tribology and the gear
dynamics [1,2], Li and Kahraman [3] recently coupled a gear
dynamics model with a gear EHL model [28] in an iterative
manner to determine the converged friction forces, gear mesh
damping, and the resultant dynamic responses in terms of the
dynamic transmission error and bearing forces. In all these gear
dynamics or tribo-dynamics studies, however, the performance of
the mechanical power loss is missing.
In view of the literature, it is evident that neither the gear
tribology field nor the gear dynamics field has put any significant
effort investigating the gear mechanical power loss under the
dynamic condition. Although the efforts have been taken to bridge
the two disciplines [1–3], it is necessary to move further, studying
how gear mesh power loss behaves under the tribo-dynamics
Fig. 1. The schematic of a discrete lumped-parameter description of the spur gear
condition. In addition, the tribo-dynamics model of Li and Kahra-
dynamics.
man [3] adopted the iso-thermal assumption, neglecting the
lubricant temperature rise which will reduce the lubricant viscos-
ity and consequently viscous friction and power loss. This assump- The former is modeled by two sets of spring-damping elements
tion is not appropriate under the high sliding condition where the (one in the LOA direction and being denoted as kyj and cyj ; one in
thermal effect is evident, demanding a thermal lubrication model the OLOA direction and being denoted as kxj and cxj , where j ¼ 1; 2)
to be in the place of the iso-thermal one. as shown in Fig. 1(a); and the latter is represented by the torsional
In this work, a six degree-of-freedom dynamics formulation is damping of ctj (j ¼ 1; 2) to take into account the bearing viscous loss
combined with the governing equations of thermal mixed elasto- [29] (ctj is not shown in Fig. 1).
hydrodynamic lubrication for spur gear pairs. The iterative numer- As illustrated in Fig. 1(b), the tooth surface velocities of u1 and
ical procedure proposed by Li and Kahraman [3] is used to u2 point in the OLOA direction of x. The difference between u1 and
determine the contact pressure, film thickness, lubricant and u2, i.e. the sliding velocity of us ¼ u1  u2 , not only introduces the
surface temperature rises, and power loss distributions within friction forces of F1(t) and F2(t), but also leads to the viscous
the contact. This study is kept limited to spur gears where the damping of cm ðtÞ through the power dissipation mechanism of
contact lines are parallel to the gear rotation axes and the load shear heating within the lubrication film. Both of the friction
distributions are quite uniform along the axial direction, such that forces and the viscous damping act in the OLOA direction of x,
a line contact can be assumed. pointing to the insufficiency of most gear dynamics models in
literature [18–27] that assumed LOA direction mesh damping. In
addition, cm ðtÞ was theoretically derived to be a time-varying
2. Formulations parameter, depending directly on the instantaneous lubrication
film thickness and the lubricant viscosity [2,3], which has also
2.1. Dynamics of a spur gear pair been missing in most gear dynamics publications.
The equations of motion of the spur gear dynamic model of
The dynamic behavior of a general spur gear pair that is Fig. 1 are summarized as [3]
composed of the driving gear (gear 1), whose base radius is r 1 ,
X
N
and the driven gear (gear 2), whose base radius isr2, is modeled J 1 θ€ 1 ðtÞ þ ct1 θ_ 1 ðtÞ þ r 1 km ðtÞδðtÞ ¼ T 1 þ ½F 1 ðtÞR1 ðtÞn ; ð1aÞ
using a six-degree-of-freedom description as proposed in Ref. [3]. In n¼1
this lumped-parameter description, each gear body j ðj ¼ 1; 2Þ is
modeled as a rigid disk with the radius of r j , whose mass (denoted m1 y€ 1 ðtÞ þ cy1 y_ 1 ðtÞ þ ky1 y1 ðtÞ þ km ðtÞδðtÞ ¼ 0; ð1bÞ
as mj ) and polar mass moment of inertia (denoted as J j ) are the
same as those of gear j. The schematic of this discrete dynamic X
N
m1 x€ 1 ðtÞ þcx1 x_ 1 ðtÞ þkx1 x1 ðtÞ ¼ ½F 1 ðtÞn ; ð1cÞ
model is shown in Fig. 1. Along the LOA direction of y, the gears are n¼1
coupled through a spring element km ðtÞ that varies periodically with
the gear mesh position to simulate the mesh stiffness. The clearance X
N

of magnitude 2ℓ is included to model the backlash of the gear teeth. J 2 θ€ 2 ðtÞ þ ct2 θ_ 2 ðtÞ  r 2 km ðtÞδðtÞ ¼  T 2  ½F 2 ðtÞR2 ðtÞn ; ð1dÞ
n¼1
In addition, the static motion transmission error of εs is incorpo-
rated for the purpose of describing any deviation of the tooth
m2 y€ 2 ðtÞ þ cy2 y_ 2 ðtÞ þ ky2 y2 ðtÞ  km ðtÞδðtÞ ¼ 0; ð1eÞ
geometry from the involute shape resulted from manufacturing
errors, intentional modifications, as well as any deformation under X
N
load. It is noticed that the supporting shaft and bearing of gear j m2 x€ 2 ðtÞ þcx2 x_ 2 ðtÞ þkx2 x2 ðtÞ ¼  ½F 2 ðtÞn ð1fÞ
introduce both the translational and the rotational flexibilities [3]. n¼1
172 S. Li / Tribology International 88 (2015) 170–178

where θj is the angular displacement and Rj is the contact radius of distribution in Eq. (4) is written as
gear j (j ¼ 1; 2), n denotes the nth tooth pair in contact, and N
hðx; tÞ ¼ h0 ðt Þ þ g 0 ðx; tÞ þ Vðx; tÞ  S1 ðx; tÞ  S2 ðx; tÞ ð5Þ
represents the total number of tooth pairs in contact. The dis-
placement δðtÞ in Eq. (1) is a non-linear function [3], describing the where h0 ðtÞ is a parameter that is adjusted in the numerical
tooth-in-contact and tooth-separation circumstances. solution of the EHL governing equations to enforce the equilibrium
When the rotational speed is high, the dynamic gear mesh between the applied normal load and the predicted contact
 
loading can differ greatly from the quasi-static counterpart. To pressure [28], g 0 ðx; tÞ ¼ 1=R1 ðtÞ þ 1=R2 ðtÞ x2 =2 is the surface cur-
take into account the impact of this dynamic behavior on the gear vature gap before loading, and Sj ðx; tÞ represents the surface
mesh friction forces and damping, the dynamic tooth force that is roughness height distribution of gear j (j ¼ 1; 2).
approximated as [1,3] Due to the frictional heat, the lubricant temperature rises
  within the contact. Although such thermal response doesn’t alter
r1
W T ðtÞ ¼ W sT ðtÞ km ðtÞδðtÞ ð2Þ the film thickness, which is mainly controlled by the lubricant
T1
inlet temperature, it does impact the fluid viscosity and thus
instead of the quasi-static tooth force of W sT ðtÞ, is used in the viscous shearing power loss within the contact. The fluid energy
elastohydrodynamic lubrication analysis described below. In equation [30,31] of
Eq. (2), W sT ðtÞ is evaluated using the same gear load distribution !
∂ 2 ϕf ∂ ϕf ∂ ϕf
program as that used in Ref. [3]. kf þ τ γ
_ ¼ ρc f u þ ð6Þ
∂z2 ∂x ∂t

2.2. Friction and viscous damping at gear mesh interface under is, thereby, used to describe the distribution of the lubricant
thermal condition temperature ϕf within the EHL conjunction. In Eq. (6), kf and
cf represent the lubricant thermal conductivity and specific heat,
Under the dynamic condition, the tooth surface velocity [uj and τ and γ_ are the shear and shear strain rate of the lubricant.
(j ¼ 1; 2)] is composed of the kinematic velocity uj (produced by When the shear flow dominates and the no-slip boundary condi-
the gear nominal rotation) and the alternating velocity (resulted tion applies, the fluid velocity u varies linearly along the film
from the vibratory motions). Therefore, the tangential surface thickness direction z as
velocity of gear j (j ¼ 1; 2), which rotates at the angular velocity hz z
of ωj , reads [3] uðz; t Þ ¼ u1 ðtÞ þ u2 ðtÞ ð7Þ
h h
uj ðtÞ ¼ uj ðtÞ þ Rj ðtÞθ_ j ðtÞ  x_ j ðtÞ ð3Þ where u1 and u2 are defined in Eq. (3). In order to reduce the
h solution dimension of Eq. (7), the fluid temperature variation
where uj ðtÞ ¼ Rj ðtÞωj is the kinematic component and j R ðtÞθ_
j along the z axis is approximated by [30,31]
ðtÞ  x_ j ðtÞ is the alternating component. These tooth surface
 z 2 z
velocities entrain the lubricant into the contact, establishing an ϕf ¼ ð3ϕ1 þ 3ϕ2  6ϕm Þ  ð4ϕ1 þ 2ϕ2 6ϕm Þ þ ϕ1 ð8Þ
elastohydrodynamic lubrication film. However, the existence of h h
the surface roughness profiles on the gear tooth surfaces (tool where ϕm is the mean temperature of the fluid along the z axis,
marks caused by the finishing processes such as shaving and and ϕ1 and ϕ2 represent the tooth surface temperatures.
grinding) may introduce asperity interactions under certain oper- The temperature rise of tooth surface j at position x and time t
ating conditions such as the low speed and heavy loading condi- is governed by the energy equation of [33]
tion. This study, thus, targets the mixed lubrication regime, where Z Z ( )
the EHL fluid film and the asperity contacts support the normal ½ðx  x0 Þ uj ðt t 0 Þ2 Q j ðx0 ; t 0 Þdx0
Δϕj ðx; tÞ ¼ dt exp 
0
ð9Þ
load together. Considering a line contact, the fluid flow between t Γ 4κ s ðt  t 0 Þ 2π ks ðt  t 0 Þ
the meshing spur gear tooth surfaces is described by the Reynolds
equation of [28] where κs and ks are the thermal diffusivity and conductivity of the
      gears, and Q j is the instantaneous local heat flux going into tooth
∂ ∂pðx; tÞ ∂ ρðx; tÞhðx; tÞ ∂ ρðx; tÞhðx; tÞ surface j (j ¼ 1; 2) and
f ðx; tÞ ¼ ur ðtÞ þ ð4aÞ
∂x ∂x ∂x ∂t
Q 1 ¼ ϑQ ; Q 2 ¼ ð1  ϑÞQ ð10a; bÞ
where pðx; tÞ is the contact pressure, hðx; tÞ is the film thickness,
where Q is the total instantaneous local frictional heat flux, and ϑ
and ρðx; tÞ is the lubricant density at the local position x and time t.
is the heat partition coefficient determined according to [30,31]
The rolling velocity ur ðtÞ in Eq. (4a) is the average of the transient
surface velocities, i.e. ur ðtÞ ¼ 12 ½u1 ðtÞ þ u2 ðtÞ. In this study, the h
ϕ1  ϕ2 ¼ ð1  2ϑÞQ ð11Þ
Eyring fluid is assumed and the corresponding flow coefficient 2kf
f ðx; tÞ is approximated the same way as that in Ref. [28]. Within the
The frictional heat flux Q is composed of the sliding and rolling
asperity contact areas, where the fluid hydrodynamic behavior
components for any hydrodynamic fluid area as [11,14]
vanishes, the following contact equation is implemented [30–32]
3  
∂hðx; tÞ u2 ðtÞ h ðx; tÞ ∂pðx; tÞ 2
¼0 ð4bÞ Q ðx; tÞ ¼ ηn ðx; tÞ s þ ð12Þ
∂x hðx; tÞ 12η ðx; tÞ
n ∂x

At the bounds that divide the hydrodynamic areas and the where the effective viscosity ηn ¼ η=coshðτm =τ0 Þ for an Eyring
asperity interaction areas, the local film thickness is considered to fluid. The first and second terms on the right hand side of Eq.
maintain its shape and travel through the EHL channel at the (12) represent the sliding and rolling loss components, respec-
velocity of ur ðtÞ, as such [30–32] tively. It is noted that the rolling loss can constitute a significant
portion of the total frictional heat especially when the roughness
∂hðx; tÞ ∂hðx; tÞ
ur ðtÞ þ ¼0 ð4cÞ amplitude is high, where the surface roughness introduces large
∂x ∂t
pressure fluctuations within the contact and results in large
Assuming the surface hardness is sufficiently high and only the pressure gradients that dictate the rolling
 heat flux [11]. For areas
elastic deformation Vðx; tÞ takes place, the lubricant film thickness where h ¼ 0, Q ðx; tÞ ¼ μb pðx; tÞus ðtÞ [11], where the boundary
S. Li / Tribology International 88 (2015) 170–178 173

lubrication friction coefficient μb is assumed to be 0.1 [11,14] due


to the lack of the μb measurement.
To ensure the total contact force density (per unit face width)
due to the predicted pressure over the entire contact zone equals
the transient dynamic tooth force density W 0T , which is the ratio of
the dynamic tooth force defined by Eq. (2) to the face width of the
R
gear, the load balance equation of W 0T ðtÞ ¼ Γ pðx; tÞdx is enforced.
For the description of the lubricant density and viscosity depen-
dences on pressure and temperature, the relationships used in
Refs. [30,31] are employed.
Applying a mesh grid with I elements, whose individual area is
denoted by A, the gear mesh viscous damping cm ðtÞ can be numeri-
cally evaluated using the converged EHL solutions through [3]
XI  n 
η ðxi ; tÞ
cm ðtÞ ¼ A ð13Þ
i¼1
hðxi ; tÞ

Eq. (13) states the gear mesh viscous damping is a function of


the lubricant viscosity and the film thickness. Since the lubricant
temperature impacts the lubricant viscosity greatly, it shall play an Fig. 2. The surface roughness profiles along the tooth profile direction for
important role in the magnitude of the gear mesh damping. The (a) ground, and (b) polished surface finishes.

thermal governing Eqs. (6–12), thus, are necessary to be included


under the high sliding condition.
75
The friction force excitations in the equations of motion of Eq. (1)
were also derived in [3] as 60
h i
F 1 ðtÞ ¼ cm ðtÞ R2 ðtÞθ_ 2 ðtÞ  x_ 2 ðtÞ  R1 ðtÞθ_ 1 ðtÞ þ x_ 1 ðtÞ þ F s ðtÞ  F r ðtÞ 45

30
ð14aÞ
h i 15
F 2 ðtÞ ¼ cm ðtÞ R2 ðtÞθ_ 2 ðtÞ  x_ 2 ðtÞ  R1 ðtÞθ_ 1 ðtÞ þ x_ 1 ðtÞ þ F s ðtÞ þ F r ðtÞ
0
0 1000 2000 3000 4000 5000
ð14bÞ
Rotational Speed (RPM)
where F s ðtÞ and F r ðtÞ are the sliding and rolling friction forces
Fig. 3. The thermal tribo-dynamic prediction of the RMS amplitude of the dynamic
defined by [3]
motion transmission error under the baseline condition.
X
I X
I
ηn ðxi ; tÞ
F s ðtÞ ¼ A μb pðxi ; tÞ þA ½u2 ðtÞ  u1 ðtÞ ð15aÞ
i¼1 i¼1
hðxi ; tÞ
1

I 
X  0.8
∂pðxi ; tÞ
F r ðtÞ ¼ A 2 hðxi ; tÞ
1
ð15bÞ
i¼1
∂x 0.6

0.4
It is noted the summations in Eqs. (13) and (15b) are only
0.2
performed for the elements where hydrodynamic fluid film exists.
In Eq. (15a), the first summation is carried out for the asperity 0
99.85
interaction elements, while the second summation is performed for
hydrodynamic fluid elements. The iterative computational method 99.8
used in this study to couple these gear mixed thermal elastohy- 99.75
drodynamic lubrication (TEHL) formulation with the spur gear
99.7
dynamics formulation of Section 2.1 is the same as that in Ref. [3].
99.65

99.6
2.3. Gear mesh mechanical power loss 0 1000 2000 3000 4000 5000

The frictional heat flux of Eq. (12) is the direct cause of gear
Fig. 4. The comparisons of the (a) average mechanical power loss and (b) average
mesh mechanical power loss. The total loss over the entire contact
mechanical efficiency between the thermal tribo-dynamic and thermal quasi-static
predictions under the baseline condition. (For interpretation of the references to
Table 1 color in this figure legend, the reader is referred to the web version of this article.)
Design parameters of the unity-ratio spur gear pair
used in this study.
zone of nth loaded tooth pair at the time instant t ζ (index ζ
Number of teeth 50 denotes the rotational mesh position within one mesh cycle) is the
Module [mm] 3.0
sum of these instantaneous local heat flux distribution, given as
Pressure angle [deg] 20.0
Outside diameter [mm] 156.0
Z
Pitch diameter [mm] 150.0 Φn ðt ζ Þ ¼ L Q ðx; t ζ Þdx ð16aÞ
Root diameter [mm] 140.0 Γ
Center distance [mm] 150.0 where L represents the tooth face width. A spur gear pair with a
Face width [mm] 20.0
Backlash [mm] 0.14
profile contact ratio of ς (χ o ς o χ þ 1 where χ is an integer and
typically χ ¼ 1) has N ¼ ðχ þ 1Þ tooth pairs in contact for ðς  χ Þ of
174 S. Li / Tribology International 88 (2015) 170–178

Fig. 5. The comparisons of the mechanical power loss (left column) and maximum the Hertzian pressure (right column) distributions along the gear 1 roll angle of a meshing
tooth pair between the thermal tribo-dynamic and thermal quasi-static predictions at (a) A, (b) B, (c) C, (d) D, and (e) E as defined in Figs. 4 and 5(b) under the baseline
condition.

mesh power loss is the sum of the losses occurring at all the N
contacting tooth pairs as
X
N
Φðt ζ Þ ¼ Φn ðt ζ Þ ð16bÞ
n¼1

For a continuous gear EHL analysis performed for M number of


rotational gear mesh cycle increments, the average gear mesh
mechanical power loss is thus arrived as

1 X
M
Φ¼ Φðt ζ Þ ð17Þ
Mζ¼1

3. Results and discussion

The unity-ratio spur gear pair, whose design parameters are


listed in Table 1, is considered in this study. On both gears, the tip
relief of 10 μm starting at the roll angle of 20.91 is applied. This gear
Fig. 6. The comparisons of the (a) average mechanical power loss and (b) average
pair has been used in a number of experimental works on gear
mechanical efficiency between the thermal tribo-dynamic and iso-thermal tribo-
dynamic predictions under the baseline condition. (For interpretation of the
dynamics [18,19]. The equivalent bearing stiffness and damping in
references to color in this figure legend, the reader is referred to the web version Fig. 1 are taken from [3] to be ky ¼ 1:15ð10Þ9 N/m and cy ¼ 5360 Ns/
of this article.) m in the LOA direction, and kx ¼ 8:0ð10Þ8 N=m and cx ¼ 2980 Ns/m
in the OLOA direction for both gears. ct ¼ 10 N m s=rad is assumed
for the torsional bearing damping [3]. The lubricant used here is the
turbine fluid Mil-L23699, whose density and viscosity properties
the entire gear mesh cycle, and N ¼ χ loaded tooth pairs during the are referred to Ref. [30]. Two surface finishes, i.e. ground and
remaining ðχ þ 1  ςÞ of the mesh cycle. At the gear rotational mesh polished tooth surfaces, are implemented to assess the surface
position of time instant t ζ (ζ A ½1; M), the total instantaneous roughness effect on the mechanical power loss under the thermal
S. Li / Tribology International 88 (2015) 170–178 175

Fig. 7. The thermal tribo-dynamic predictions of p, h, ϕf , ϕ1 , and ϕ2 at an example mesh position where the gear 1 roll angle equals 251 at the rotational speeds of (a) B and
(b) E as defined in Figs. 4 and 5(b) under the baseline condition.

curves). To demonstrate the tribo-dynamic effect on the power loss,


the counterparts under the thermal quasi-static condition (red curves)
are also included in the figure to allow direct comparison. A similar
comparison with respect to the average efficiency, Ψ ¼ Φ=T 1 ω1 , is
performed in Fig. 4(b). It is seen, the deviations of the mechanical
power loss and efficiency between the tribo-dynamic and quasi-static
conditions are evident in the vicinities of the resonance peaks. For
instance, at point A (before the 1st resonance peak), point B (after the
1st resonance peak), point D (before the 2nd resonance peak), and
point E (at the 2nd resonance peak) as defined in Fig. 4(b) as well as in
Fig. 3, Φ ¼ 257, 343, 431, and 948 W under the tribo-dynamic
condition and Φ ¼ 276, 302, 526, and 532 W under the quasi-static
condition, pointing to the relative errors of 7%, 12%, 22%, and 44%,
respectively. At the rotational speed that is away from the resonance
peaks, the mechanical power loss under the tribo-dynamic condition
is seen to vary limitedly from that under the quasi-static condition, for
instance at point C of Fig. 4(b).
An interesting observation in Fig. 4 is that the tribo-dynamic
Fig. 8. The comparisons of the thermal tribo-dynamic (a) average mechanical behavior of a gear pair not necessarily increases the mechanical
power loss and (b) average mechanical efficiency between the predictions of the
baseline and those when the lubricant inlet temperature is reduced from 90 1C to
power loss, such as at the rotational speeds of points A and D where
60 1C. Φ is decreased to increase the mechanical efficiency. In Fig. 5, the
mechanical power loss distributions along the gear 1 roll angle for a
meshing tooth pair, n, under the thermal tribo-dynamic and
tribo-dynamic condition. The surface roughness profiles measured thermal quasi-static conditions are compared at the five rotational
along the profile direction from the root to the tip for these two speeds of A E as defined in Figs. 3 and 4(b). In this figure, the
surface finishes are compared in Fig. 2. corresponding maximum Hertzian pressure distributions, ph , are
The baseline condition of this study is set to have the input torque also shown. At the rotational speeds of B and E [Fig. 5(b) and (e)],
of 700 Nm and the lubricant inlet temperature of 90 1C with the the tribo-dynamic response in terms of ph largely differs from that
ground surface finish. For a complete simulation, the rotational speed under the quasi-static condition. It is seen the tribo-dynamic ph
of the gears, Ω, is set to increase stepwise from 500 RPM to 4200 RPM becomes larger in comparison to the quasi-static one during most of
with the increment of 25 RPM. The RMS dynamic motion transmission
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi the double-tooth-contact (DTC) regions (gear 1 roll angle less than
P10
error defined as ðεd Þrms ¼ q ¼ 1 εdq , where εdq represents the q
2 th
19:98 3 , and gear 1 roll angle larger than 21:73 3 ) where sliding is
harmonic amplitude of εd ðtÞ [εd ðtÞ ¼ r 1 θ1 ðtÞ þ y1 ðtÞ r 2 θ2 ðtÞ  y2 ðtÞ], high. Thereby, Φ is well increased within the DTC regions as
n

is shown in Fig. 3 for the baseline condition. Two resonance peaks at shown in the left column of Fig. 5(b) and (e). In the single-tooth-
Ω ¼ 1750 and 3500 RPM are observed. The former is associated with contact (STC) region (gear 1 roll angle between 19:98 3 and 21:73 3 )
the mode that the LOA transverse motion and the torsional motion of where the power loss is usually low due to low sliding, the
the rigid disks in Fig. 1 are out of phase and offset each other; the latter decreased tribo-dynamic ph doesn’t introduce significant power
is governed by the mode that these two motions are in phase and add loss reduction. As a result, the total average mechanical power loss
to each other to produce a much larger peak at the 3500 RPM under the tribo-dynamic condition is elevated at B and E in
rotational speed. Under this thermal tribo-dynamic condition, the comparison to the quasi-static condition in Fig. 4. On the contrary,
average gear mesh mechanical power losses (the rolling component, the tribo-dynamic ph at the rotational speeds of A and D becomes
the sliding component, and the total) are plotted in Fig. 4(a) (black smaller during most of the DTC regions and larger within the STC
176 S. Li / Tribology International 88 (2015) 170–178

Fig. 9. The thermal tribo-dynamic predictions of p, h, and Q at an example mesh position where the gear 1 roll angle equals 201 at the rotational speed of D as defined in
Figs. 4 and 5(b) under the (a) baseline condition (90 1C lubricant inlet temperature) and (b) lower lubricant inlet temperature of 60 1C.

Fig. 10. The comparisons of the thermal tribo-dynamic (a) average mechanical Fig. 11. The comparisons of the thermal tribo-dynamic (a) average mechanical
power loss and (b) average mechanical efficiency between the predictions of the power loss and (b) average mechanical efficiency between the predictions of the
baseline and those when the input torque is increased from 700 Nm to 1500 Nm. baseline and those when the ground surface roughness profile is replaced by the
polished surface roughness profile.

region in comparison to the quasi-static ph . Such a tribo-dynamic viscosity and the viscous shear power loss of Eq. (12). In Fig. 7, the
loading condition effectively reduces the power loss in the DTC numerical solutions of pressure p, film thickness h, and temperature
region without a large increase of the power loss in the STC region. rises of the lubricant ϕf as well as the tooth surfaces ϕj (j ¼ 1; 2) are
Therefore, the mechanical efficiency are increased at A and D as depicted at the rotational speeds of B and E that are defined in Figs. 3
shown in Fig. 4(b). At the rotational speed of point C that is away and 4(b) for an example mesh position where the gear 1 roll angle
from the resonances, the tribo-dynamic ph varies limitedly from the equals 25 3 . It is seen the contact pressure fluctuates within the contact
quasi-static one in Fig. 5(c). As a result, the difference of Φ
n
zone and peaks where asperity contacts occur (film thickness
between the tribo-dynamic and quasi-static conditions are negli- approaches zero). Although, at this mesh position, the tribo-dynamic
gible, and the resultant mechanical efficiency under the tribo- tooth forces of these two speeds have the same magnitude
dynamic condition is barely changed from that under the quasi- [ph ¼ 0:8 GPa as shown in Fig. 5(b) and (e)], the temperature rises
static condition. under the higher speed condition (point E) in Fig. 7(b) are observed to
Fig. 6 shows the comparisons of the average mechanical power loss be much higher than those under the lower speed condition (point B)
and efficiency between the thermal and iso-thermal simulation results in Fig. 7(a). This is the direct results of the elevated frictional heat at
considering the baseline operating and surface conditions. The tribo- the higher rotational speed at point E. The resultant lubricant
dynamic behavior is included in both cases. It is observed when the temperature rise at the higher speed, thus, reduces the lubricant
iso-thermal assumption is used (red curves), the predicted power viscosity and leads to the larger deviation in the power loss between
losses become larger in comparison to those predicted according to the thermal and iso-thermal assumptions.
the thermal EHL formulations of Section 2.2. This deviation increases The impact of the lubricant inlet temperature on the gear mesh
with the rotational speed and is mainly due to the exclusion of the mechanical power loss and efficiency under the thermal tribo-
lubricant temperature description, which dictates the lubricant dynamic condition is investigated in Fig. 8, where the simulation
S. Li / Tribology International 88 (2015) 170–178 177

results with the lubricant inlet temperature reduced from the In view of the magnitudes of the rolling and sliding compo-
baseline of 90  60 1C are compared to those of the baseline. The nents that constitute the total power loss in Figs. 4(a), 6(a), 8(a),
reduction in lubricant inlet temperature leads to the lubricant 10(a) and 11(a), the sliding loss is found to be the dominant
viscosity increase and elevates the film thickness within the component when the rotational speed is relatively low. As Ω
contact zone. For the very rough ground surface finish of Fig. 2(a), increases, however, the rolling component becomes more impor-
such a film thickness increase can decrease the asperity contact tant and constitutes a significant portion of the total loss, and is
activities. Fig. 9 compares the p, h, and frictional heat flux Q therefore required to be included in any mechanical power loss
distributions at an example mesh position where the gear 1 roll and efficiency prediction.
angle is 20 3 under the rotational speed of D defined in Figs. 3 and
4(b) between 90 1C [Fig. 9(a)] and 60 1C [Fig. 9(b)] inlet tempera-
tures. It is observed the film thickness is thinner and the contact 4. Conclusions
pressure is higher under the higher inlet temperature condition
due to increased asperity interaction activities. When the inlet A thermal tribo-dynamics model that combines the governing
temperature is reduced, Q is seen to become smaller in the areas equations of the vibratory motion and the thermal mixed EHL for
where asperity contacts occur (h ¼ 0), and become larger in the spur gear pairs is developed in this work. By comparing the gear
areas where the surfaces are separated by the hydrodynamic film mesh mechanical power loss between the thermal tribo-dynamic
(h 4 0). Since the asperity contact frictional heat constitutes a and thermal quasi-static predictions, large deviations are observed
more significant portion of the total mechanical power loss for the in the vicinities of the resonance peaks. Therefore, the quasi-static
ground surface finish, the decrease in lubricant inlet temperature assumption is valid only when the rotational speed is far away
results in the decrease in Φ and increase in Ψ as shown in Fig. 8. from the resonances. Through the same comparison, an interesting
The influence of the input torque on gear mesh mechanical finding is that the gear dynamic responses not necessarily elevate
power loss under the thermal tribo-dynamic condition is shown in the mechanical power loss. When the dynamic response leads to
Fig. 10. After the input torque is increased from the baseline of reduced tooth force within the double-tooth-contact region where
700 Nm (corresponding to the quasi-static pitch line ph of 1.2 GPa) sliding is high, the frictional power loss can be effectively
to 1500 Nm (corresponding to the quasi-static pitch line ph of decreased. A similar comparison between the thermal tribo-
1.7 GPa), the mechanical power loss is significantly increased dynamic and iso-thermal tribo-dynamic predictions is also per-
owing to the increase in the contact pressure and consequently formed, pointing to the necessity of the inclusion of the lubricant
the frictional power loss. However, the mechanical efficiency temperature description for the accurate modeling of the gear
decreases only limitedly, which is due to the large increase of mesh mechanical power loss when the rotational speed is high.
the input power. The last comparison of this study is made In addition, a parametric investigation is implemented in
focusing on the surface finish effect on Φ under the thermal this study to show the impacts of the lubricant inlet tempera-
tribo-dynamic condition as shown in Fig. 11. The polished surface ture, input torque, and surface roughness on gear mesh
finish is found to produce significant reductions in both Φ and Ψ . mechanical power loss under the thermal tribo-dynamic con-
This is because the decrease in asperity contact activities when the dition. It is found, the lubricant inlet temperature reduction can
polished surfaces are implemented as illustrated in Fig. 12 at an increase the mechanical efficiency though increasing the fluid
example mesh position where the gear 1 roll angle is 17 3 at the film thickness and decreasing the asperity interaction activities.
rotational speed of E defined in Figs. 3 and 4(b). In Fig. 12(a), local Although the increase in the input torque elevates the power
asperity contacts are observed where p shoots up, with the loss significantly by increasing the contact pressure, the corre-
maximum contact pressure exceeding 3 GPa. When the roughness sponding efficiency is seen to be impacted limitedly since the
amplitude is reduced in Fig. 12(b), full film lubrication is achieved input power is increased at the same time. Comparing the
and the maximum contact pressure is well below 2 GPa. As a ground and polished surface finishes, the surface roughness
result, the frictional heat is greatly decreased and the mechanical reduction is shown to be able to largely decrease the power loss
efficiency is well improved. and at the meantime increase the efficiency.

Fig. 12. The thermal tribo-dynamic predictions of p, h, and Q at an example mesh position where the gear 1 roll angle equals 171 at the rotational speed of E as defined in
Figs. 4 and 5(b) under the (a) baseline condition (ground surface finish) and (b) polished surface finish.
178 S. Li / Tribology International 88 (2015) 170–178

References [17] Wang J, Li R, Peng X. Survey of nonlinear vibration of gear transmission


systems. Appl Mech Rev 2003;56(3):309–29.
[18] Blankenship GW, Kahraman A. Steady state forced response of a mechanical
[1] Li S, Kahraman A. Influence of dynamic behaviour on elastohydrodynamic
oscillator with combined parametric excitation and clearance type nonlinear-
lubrication of spur gears. J Eng Tribol 2011;225:740–53.
ity. J Sound Vib 1995;185(5):743–65.
[2] Li S, Kahraman A. A spur gear mesh interface damping model based on
[19] Kahraman A, Blankenship GW. Interactions between commensurate para-
elastohydrodynamic contact behavior. Int J Powertrains 2011;1(1):4–21.
metric and forcing excitations in a system with clearance. J Sound Vib
[3] Li S, Kahraman A. A tribo-dynamic model of a spur gear pair. J Sound Vib
2013;332:4963–78. 1996;194(3):317–36.
[4] Pedrero JI. Determination of the efficiency of cylindrical gear sets. Paris, [20] Kahraman A, Singh R. Non-linear dynamics of a spur gear pair. J Sound Vib
France: Proceedings of the 4th World congress on gearing and power 1990;142(1):49–75.
transmission; 1999. [21] Ozguven HN, Houser DR. Dynamic analysis of high speed gears by using
[5] Michlin Y, Myunster V. Determination of power losses in gear transmissions loaded static transmission error. J Sound Vib 1988;125(1):71–83.
with rolling and sliding friction incorporated. Mech Mach Theory [22] Theodossiades S, Natsiavas S. Non-linear dynamics of gear pair systems with
2002;37:167–74. periodic stiffness and backlash. J Sound Vib 2000;229(2):287–310.
[6] Heingartner P, Mba D. Determining power losses in the helical gear mesh: case [23] Tamminana VK, Kahraman A, Vijayakar S. A study of the relationship between
study. In: Proceedings of the 9th international power transmission and the dynamic factors and the dynamic transmission error of spur gear pairs.
gearing conference, Chicago, Illinois, USA; 2003. J Mech Des 2007;129(1):75–84.
[7] Anderson NE, Loewenthal SH. Efficiency of nonstandard and high contact ratio [24] Vaishya M, Singh R. Analysis of periodically varying gear mesh systems with
involute spur gears. J Mech, Trans Autom Des 1986;108:119–26. Coulomb friction using Floquet theory. J Sound Vib 2001;243(3):525–45.
[8] Anderson NE, Loewenthal SH. Design of spur gears for improved efficiency. [25] Lundvall O, Stromberg N, Klarbring A. A flexible multi-body approach for
J. Mech Des 1982;104:767–74. frictional contact in spur gears. J Sound Vib 2004;278(3):479–99.
[9] Martin KF. The efficiency of involute spur gears. ASME J Mech Des [26] Kahraman A, Lim J, Ding H. A dynamic model of a spur gear pair with friction.
1981;103:160–9. In: Proceedings of the 12th IFToMM world congress, Besancon, France; 2007.
[10] Wu S, Cheng HS. A friction model of partial-EHL contacts and its application to [27] He S, Gunda R, Singh R. Effect of sliding friction on the dynamics of spur gear
power loss in spur gears. Tribol Trans 1991;34(3):398–407. pair with realistic time-varying stiffness. J Sound Vib 2007;301(3):927–49.
[11] Li S, Kahraman A. Prediction of spur gear mechanical power losses using a [28] Li S, Kahraman A. A transient mixed elastohydrodynamic lubrication model
transient elastohydrodynamic lubrication model. Tribol Trans 2010;53 for spur gear pairs. ASME J Tribol 2010;132(1) 011501 (9 pp).
(4):554–63. [29] Harris TA, Kotzalas MN. Essential concepts of bearing technology. 5th edition.
[12] Xu H, Kahraman A, Anderson NE, Maddock DG, D.G.. Prediction of mechanical Boca Raton, London, New York: Taylor & Francis Group; 2007.
efficiency of parallel-axis gear pairs. ASME J Mech Des 2007;129(1):58–68. [30] Li S, Kahraman A, Anderson N, Wedeven LD. A model to predict scuffing
[13] Li S, Vaidyanathan A, Harianto J, Kahraman A. Influence of design parameters failures of a ball-on-disk contact. Tribol Int 2013;60:233–45.
on mechanical power losses of helical gear pairs. J Adv Mech Des, Syst, Manuf [31] Li S. Influence of surface roughness lay directionality on scuffing failure of
2009;3(2):146–58.
lubricated point contacts. ASME J Tribol 2013;135(4) 041502 (10 pp).
[14] Li S, Kahraman A. A method to derive friction and rolling power loss formulae
[32] Li S. A computational study on the influence of surface roughness lay
for mixed EHL contacts. J Adv Mech Des, Syst, Manuf 2011;5(4):252–63.
directionality on micro-pitting of lubricated point contacts. ASME J Tribol
[15] Kolivand M, Li S, Kahraman A. Prediction of mechanical gear mesh efficiency of
2015;137(2) 021401 (10 pp).
hypoid gear pairs. Mech Mach Theory 2010;45:1568–82.
[33] Carslaw HS, Jaeger JC. Conduction of heat in solids. New York: Oxford Press;
[16] Ozguven HN, Houser DR. Mathematical models used in gear dynamics – a
review. J Sound Vib 1988;121(3):383–411. 1959.

You might also like