Professional Documents
Culture Documents
Department of Mechanical
and Aerospace Engineering,
A Helical Gear Pair Pocketing
The Ohio State University,
Columbus, OH 43210
e-mail: talbot.11@osu.edu
Power Loss Model
A new fluid dynamics model is proposed to predict the power losses due to pocketing of
Ahmet Kahraman air, oil, or an air-oil mixture in the helical gear meshes. The proposed computational
Department of Mechanical
procedure treats a helical gear pair as a combination of a number of narrow face width
and Aerospace Engineering,
spur gear segments staggered according to the helix angle and forms a discrete fluid dy-
The Ohio State University,
namics model of the medium being pocketed in the gear mesh. Continuity and conserva-
Columbus, OH 43210
tion of momentum equations are applied to each coupled control volume filled with a
compressible fluid mixture to predict fluid pressure and velocity distributions from which
Satya Seetharaman the instantaneous pocketing power loss is calculated. The proposed model is exercised in
Department of Mechanical order to investigate the fluid pressure and velocity distributions in time along with the
and Aerospace Engineering, pocketing power loss as a function of the speed, helix angle, and oil-to-air ratio.
The Ohio State University, [DOI: 10.1115/1.4026502]
Columbus, OH 43210
1 Introduction of fluid around the gear set, the main components of the spin
power losses of a gear set are: (i) the drag losses of the of the
The efficiency of geared transmission systems has become
rotating components, (ii) losses due to pocketing of the fluid at the
increasingly important due to the demands imposed on automotive
gear mesh interfaces, and (iii) viscous losses of rolling element
and aerospace products in terms of fuel economy. As a result of
bearings, in addition to other secondary sources such as synchron-
the market push and environmental and energy related regulations
izers and oil seals [10].
on vehicles, current and future drive trains and transmissions must
There are various published experimental studies on the power
be designed to meet stringent efficiency requirements. Thus,
loss of a gear or a disk rotating fully, or partially, submerged in
power losses associated with gears have become a major research
lubricant [11–13]. Empirical relations for dimensionless churning
topic in recent years.
torque were provided from experiments for disks [14] and for
As is evident from various published experimental studies
gears in mesh [15,16]. Ariura et al. [17] presented experimental
[1–5], power losses in any geared transmission can be broken into
load independent power loss measurements on jet lubricated spur
two main categories: load-dependent (mechanical) power losses
gears. The experiments by Petry-Johnson et al. [1] provided data
and load-independent (spin) power losses. Load-dependent power
on the influence of the face width and module of gears on spin
losses are attributed to loaded lubricated contacts of the mating
losses under jet lubricated (windage) conditions.
surfaces at the gear meshes and rolling element bearings. They
A theoretical approach to load independent power loss due to
originate from the relative sliding and rolling action of the elasto-
oil fling-off was provided by Akin et al. [18,19]. Models on the
hydrodynamic lubrication (EHL) film at the contact interface that
windage losses of gears rotating only in air have been limited to
can be loosely defined as friction. The state-of-the-art in the mod-
empirical studies [20,21] or computational fluid dynamics models
eling of gear mesh mechanical power losses is represented by a
[22,23]. As stated by the review paper of Eastwick and Johnson
number of recent studies on spur gears (e.g. Refs. [6,7]) where the
[24], most of these studies excluded losses associated with fluid
EHL theory was employed to predict friction at gear contacts.
flow at the gear mesh interface. Pechersky and Wittbrodt [25] pro-
These studies either relied on a real-time transient analysis of the
posed a theoretical analysis to compute the pressure and velocities
EHL conditions of the gear contacts as they move along the tooth
of oil trapped in the meshing zone of spur gears without comput-
surface or used friction models based on the regression analysis of
ing the resultant power loss. Houjoh et al. experimentally investi-
the EHL results covering wide ranges of key contact parameters,
gated the fluid pressure and velocity distributions in spur [26] and
rather than employing any empirical friction formula. Some of
helical [27] gears from the standpoint of improving lubrication
these models included only the sliding power losses under full-
techniques. Diab et al. [21,28] introduced a modeling effort simi-
film lubrication conditions [6], while others captured both sliding
lar to the one proposed here using conservation of mass and con-
and rolling losses that take place at the gear contacts [7] via mixed
servation of energy in order to model the trapping of air only in
(or boundary) EHL formulations [8] to capture actual asperity
the meshing zone, while the use of the conservation of energy to
(metal-to-metal) contacts common to gear interfaces. This EHL-
predict air pressure and velocity for determining power loss is de-
based method was also applied to helical gears by simply treating
batable. Most recently, Seetharaman and Kahraman [10] proposed
them as a collection of narrow face width spur gear slices stag-
a fluid-mechanics based formulation to compute drag power
gered according to the helix angle [9].
losses along the sides and periphery of rotating spur gears and
The mechanisms for spin power losses of gearboxes are entirely
pocketing losses at the gear mesh interface under dip lubrication
different from those of the mechanical losses. Since contact fric-
conditions. This incompressible flow formulation for oil churning
tion is negligible under no-load conditions, interactions of the
losses was later complemented by a compressible flow formula-
gear set components with the surrounding medium (oil, air, or any
tion to predict the same for windage losses under jet lubrication
mixture of oil and air) are responsible for such losses. They have
conditions [29]. The pocketing losses were predicted to represent
often been referred to as churning (in the case of oil) and windage
a large portion of the total windage losses. The same investigators
(in the case of air or an air/oil mix) losses. Regardless of the type
[30] provided comparisons between their predictions and
1
experiments.
Corresponding author.
Contributed by the Tribology Division of ASME for publication in the JOURNAL
Gear pocketing models proposed in Refs. [10,28–30] are either
OF TRIBOLOGY. Manuscript received August 3, 2013; final manuscript received only for spur gears where the pockets formed between the teeth of
January 8, 2014; published online February 24, 2014. Assoc. Editor: Xiaolan Ai. mating gears at any given instant remain constant along the face
K
ðxði;j;1;kÞ xði;j;2;k1Þ Þ
ðnÞ 1X
Aeði;jÞ ¼ (1)
2 k¼2
ðx x Þ
ði;j;2;kÞ ði;j;1;k1Þ
the speed of sound) at the exits of the control volume are given, ðnÞ ðnÞ ðnÞ ðnÞ
pði;jÞ pði1;jÞ xrcði;jÞ xcði1;jÞ
respectively, as ðnÞ ðnÞ
prði;jÞ ¼ pði1;jÞ þ
ðnÞ
(11)
ðnÞ ðnÞ ðnÞ
ð ð xrcði;jÞ xcði;jÞ þ xrcði;jÞ xcði1;jÞ
@
qdV þ qudA ¼ 0 (8)
@t CV CS With the density at the axial connections qeði;jÞ and circumferential
ðnÞ
@ @p ðnÞ
connections qrði;jÞ defined, the discrete form of Eq. (8) is given as
ðquÞ þ ¼0 (9)
@t @x 0 1 0 1
ðnÞ ðn1Þ ðnÞ ðn1Þ
V Vði;jÞ
ðnÞ @ ði;jÞ ðnÞ
qði;jÞ qði;jÞ
where q is the fluid density, V is the pocket volume, A is the exit qði;jÞ AþV @ A
Dt ði;jÞ Dt
area, u is the exit velocity and @p=@x is the pressure gradient
across an exit. In these equations, CV denotes integration over the
X
1 X
1
control volume and CS denotes integration over the exit surfaces þ
ðnÞ
1k qeðiþk;jÞ Aeðiþk;jÞ ueðiþk;jÞ þ
ðnÞ
1k qrði;jþkÞ Arði;jþkÞ urði;jþkÞ ¼ 0
of the control volume. Considering the calculated geometrical k¼0 k¼0
inputs and the index notation presented in Sec. 2.1 for a discrete
(12)
number of IJ control volumes shown in Fig. 5, while also assum-
ing pressure and velocity gradients between the pockets to be lin-
ear, Eqs. (8) and (9) can be applied in discretized form for both where Dt is the time step that is dependent on the rotational speed
space and time. and the number of teeth of the driving gear (gear 1). Here,
1k ¼ 1 for k ¼ 0 and 1k ¼ 1 for k ¼ 1. Discretized conservation
Consider the jth control volume of the i-th circumferential
ðnÞ ðnÞ of momentum (see Eq. (9)) equations in the circumferential and
pocket with pressure pði;jÞ and density qði;jÞ . The circumferential axial directions are given, respectively, by
ðnÞ
fluid velocity urði;jÞ is the fluid velocity between the axial control 0 1 0 1
ðnÞ ðn1Þ ðnÞ ðn1Þ
volumes j of pockets i and i 1. Axial fluid velocity ueði;jÞ is the
ðnÞ
ðnÞ
urði;jÞ urði;jÞ ðnÞ
qrði;jÞ qrði;jÞ
qrði;jÞ @ Aþ urði;jÞ @ A
fluid velocity between pocket i of the axial control volumes j and Dt Dt
j 1. Assuming a linear pressure gradient between control vol- 0 1
ðnÞ ðnÞ ðnÞ
umes, the pressure peði;jÞ at the axial connections and the pressure ðnÞ ðnÞ B
urðiþ1;jÞ urði1;jÞ C
ðnÞ þ
qrði;jÞ urði;jÞ @ A
prði;jÞ at the circumferential connections are defined as ðnÞ ðnÞ
xrcðiþ1;jÞ xrcði1;jÞ
0 1
ðnÞ ðnÞ ðnÞ ðnÞ ðnÞ ðnÞ
pði;jÞ pði;j1Þ xecði;jÞ xcði;j1Þ p p
ðnÞ ðnÞ
B ði;jÞ ði1;jÞ
C
peði;jÞ ¼ pði;j1Þ þ (10) þ @
A ¼ 0; (13)
ðnÞ ðnÞ ðnÞ ðnÞ ðnÞ ðnÞ ðnÞ ðnÞ
xecði;jÞ xcði;jÞ þ xecði;jÞ xcði;j1Þ xrcði;jÞ xcði;jÞ þ xrcði;jÞ xcði1;jÞ
Fig. 6 (a) Pressure, (b) end velocity, (c) contact-side velocity, and (d) backlash-side velocity
time histories for the example spur gear pair (b 5 0 deg) at X 5 3000 rpm and n 5 0:05
Fig. 7 Side views of the spur gear pair of Fig. 6 at representative mesh positions. The pocket i
of interest is marked with .
Fig. 9 End exit velocity time histories of gear pairs with helix angles of (a) b 5 0 deg, (b)
b 5 5 deg, (c) b 5 15 deg, and (d) b 5 30 deg at X 5 3000 rpm and n 5 0:05
come into contact to block the passage (Arði;jÞ ¼ 0 in Fig. 4(c)). After # ¼ 1:40, the connected pocket ahead begins expanding
The fluid flow through the contact-side exit area is abruptly inter- while the pocket i being analyzed is compressed. Thus, a rapid
rupted at # ¼ 1:40 by the tooth pair coming into contact. flow forward out of pocket i is observed within # 2 ½1:40; 1:90 in
urði; jÞ ¼ 0 at b ¼ 0:5 within # 2 ½1:40; 3:35 represents the period Fig. 6(d). Beyond # ¼ 1:90, pocket i expands more rapidly to
when the contact is maintained. Once the contact is lost at cause the fluid to backflow again into the pocket.
# ¼ 3:35, a reverse flow from pocket i 1 is predicted as pocket i The influence of the oil-to-air ratio on the fluid pressures and
is rapidly expanding. Meanwhile, as shown in Fig. 6(d), the fluid end exit velocities are shown in Fig. 8 using the same spur gear
initially flows backward into the pocket i through the backlash pair as in Fig. 6. In Fig. 8(a), the pði;jÞ and ueði;jÞ time histories are
area from the pocket ahead of it that is being rapidly compressed. shown for pocket i at different b for n ¼ 0:01 when the gear
Fig. 11 End exit velocity distributions of gear pairs with helix angles of (a) b 5 0 deg, (b)
b 5 5 deg, (c) b 5 15 deg, and (d) b 5 30 deg at X 5 3000 rpm and n 5 0:05
operates at X ¼ 4000 rpm. Figure 8(b) shows the same for for the significant differences in the pði;jÞ time histories. The flow
n ¼ 0:05. Comparing Figs. 8(a) and 8(b), only slight differences of fluid at n ¼ 0:05 carries much more momentum than that of the
in the ueði;jÞ profiles are observed between the cases of n ¼ 0:01 flow at n ¼ 0:01, requiring a much larger pressure gradient in
and n ¼ 0:05, while the same cannot be said for pði;jÞ . Since the order to accelerate and decelerate the fluid. This difference in
geometry and rotational speed of the system is the same between behavior points to the need to include the lubricant density in the
the two analyses, the volumetric flow rates between the two differ analysis.
only slightly due to compressibility. This is evident in the fluid ve- Since the main focus of this study is on the pocketing losses of
locity time histories. On the contrary, the difference in the density helical gears, the pocket pressure and velocity distributions of
of the medium between n ¼ 0:01 and n ¼ 0:05 is the main reason gears having helix angles b ¼ 0 deg; 5 deg; 15 deg, and 30 deg will
4 Conclusions
Pocketing power losses can contribute a significant amount to
the overall power loss of a power transmission system, especially
under lightly loaded and high speed applications. Wide face
widths and small helix angles show larger pocketing power loss
effects. In order to quantify this effect, a new fluid dynamics
model was proposed in this study in order to predict the power
losses of helical gear meshes due to the pocketing of air, oil, or an
air-oil mixture in helical gear meshes. This model treated a helical
gear pair as a combination of a number of narrow face width spur
gear slices staggered according to the helix angle. With each cir-
cumferential pocket discretized in this manner, the time variation
of the volumes and escape areas associated with each control vol-
ume were defined to form a discrete fluid dynamics model of a
Fig. 12 Variation of pocketing power loss with the (a) helix
angle at n 5 0:05, and (b) oil-to-air ratio at b 5 30 deg medium being pocketed in the gear mesh. Continuity and conser-
vation of momentum equations were applied to each coupled con-
trol volume filled will a compressible fluid mixture to predict the
be compared next in Figs. 9 and 10 at X ¼ 3000 rpm and fluid pressure and velocity distributions from which the instanta-
n ¼ 0:05. As stated earlier, the transverse geometries of these neous pocketing power loss was calculated. This proposed model
gears (as listed in Table 1) are identical such that b would be the was used to investigate the mechanisms for the helix angle and
sole reason for the differences in the results. The dominant influ- the air-to-oil ratio to influence the pocket pressure and velocity
ence of b on the axial fluid velocities is demonstrated in Fig. 9. As distributions. The impact of the same parameters on the helical
b increases, fluid is driven more and more out of one side of the gear pocketing power losses was also quantified within the ranges
gear pocket. The spur gear pair shows symmetric exiting veloc- of the operating speed.
ities with respect to the midplane of the gears at b ¼ 0:5. With an The complexity of the problem, including rapidly changing
increase in b, this becomes more skewed to drive all of the fluid geometries, has limited most recent computational fluid dynamics
out one side of the gear pair. Larger helix angles allow for pro- studies of gears to single gear analyses [22,23], completely avoid-
gressively larger end and circumferential exit areas across the face ing pocketing losses. The computational model presented here
width, decreasing the magnitudes of the exit velocities in the pro- provides a relatively fast method of predicting power loss due to
cess. Because the maximum exit velocities are decreased, the nec- pocketing of an air/oil mixture from a helical gear mesh. The anal-
essary change in the momentum of the fluid is decreased. yses presented here took approximately 1–3 min using one
Therefore, significantly lower pocket pressure deviations are 3.4 GHz processor, depending upon the considered operation and
experienced in helical gear pairs with larger b, as shown in fluid conditions. Thus, the proposed model can also be used in
Fig. 10. These pressure distribution results also qualitatively agree design evaluations. Finally, the computational methodology pre-
well with the measurements of Diab et al. [28]. sented in this work is also suitable for other gear types. The
Figures 9 and 10 indicate that helical gears cannot be assumed authors have recently applied this formulation to spiral bevel and
to behave like spur gears in terms of their pocketing power losses. hypoid gears, which will be presented in a separate paper.
This variance in the face width direction can be shown more
clearly by plotting the data of Fig. 9 as a function of b at various
discrete rotational positions #, as shown in Fig. 11. Here, the
Acknowledgment
effect of b in driving the flow in one direction is evident. The The authors would like to thank General Motors Powertrain
symmetric flow (squeezing in both directions) seen in Fig. 11(a) is and, specifically, Dr. Avinash Singh for motivating and supporting
progressively changed to mainly flow in one direction denoted by this research activity.
the positive fluid velocity across the face width in
Figs. 11(b)–11(d). References
Figure 12(a) shows the variation of the resultant pocketing [1] Petry-Johnson, T., Kahraman, A., Anderson, N. E., and Chase, D. R., 2008,
power loss Pp of gear pairs having b ¼ 0 deg; 15 deg, and 30 deg “An Experimental Investigation of Spur Gear Efficiency,” ASME J. Mech.
with the speed at n ¼ 0:05. Here, the spur gear (b ¼ 0 deg) is pre- Des., 130, p. 062601.
dicted to have Pp values that are almost three times those for the [2] Yada, T., 1972, “The Measurement of Gear Mesh Friction Losses,” ASME
Paper No. 72-PTG-35, pp. 8–12.
helical gear pair with b ¼ 30 deg at X ¼ 10; 000 rpm while the [3] Naruse, C., Haizuka, S., Nemoto, R., and Kurokawa, K., 1986, “Studies on Fric-
difference between gear pairs having b ¼ 0 deg and 15 deg is tional Loss, Temperature Rise and Limiting Load for Scoring of Spur Gear,”
about 10%. This demonstrates the complex and combined Bull. JSME, 29(248), pp. 600–608.