Gear Windage: A Review: Carol N. Eastwick
Gear Windage: A Review: Carol N. Eastwick
”
Estimates on the percentage effect of windage vary as the value is
dependent on a number of different parameters. Townsend 关2兴
Carol N. Eastwick identified the variables as
e-mail: [email protected] • the diameter of the rotating elements
• the length of the rotating elements
Graham Johnson • the speed of rotation
e-mail: [email protected] • the web or gear-blank design
• the overall casing design
School of Mechanical, Materials and Manufacturing • the type of oil feed system
Engineering, • the operating temperature and viscosity of oil
• the pressurization of the casing.
University of Nottingham,
University Park, One of the critical parameters is the pitch line velocity; obvi-
Nottingham NG7 2RD, UK ously, high velocities 共Ref. 关2兴 suggests 51 m / s, and Ref. 关3兴 sug-
gests tangential speeds greater than 90– 120 m / s兲 produce greater
stirred motion, so large gears rotated at high rotational rates are
particularly vunerable. Additionally, the lubrication flow rate and
The efficiency of power and propulsive systems is increasingly scavenge design are critical as these directly affect the properties
being targeted as a means of reducing environmental impact. This of the fluid surrounding the gear 关2,4兴. So, in a case where you
has caused a renewed interest in industry in the losses associated have a high level of lubricant suspended around a gear with high
with meshing gears. Gearbox efficiency varies from 98% to 99% pitch velocity, as is the case in an aeroengine, windage becomes a
for the best designed high power applications, but that can still significant contributor to the power loss. It may only account for a
equate to losses in megawatts. There are different mechanisms for few percent, but this can be critical.
losses that have been identified within gearboxes; these are mesh- Traditionally, there have been two approaches to reducing
ing losses, bearing losses, windage losses, and churning losses. WPL; the first is to use a shroud or baffle to enclose gears and the
Depending on the application, the relative importance of these second is to positively pump the oil and air from the gearbox
mechanisms varies. This paper reviews information on windage casing. Pumping from the gearbox casing is known as evacuating
power loss. The motivation for this is that for some applications, the gearbox 关1兴, and allows a reduction of fluid density within the
this power loss can be a significant component, particularly casing. This can result in up to 1% improvement in efficiency but
lightly loaded high-speed applications. For instance, within some can only be used in a limited number of situations.
aeroengines, gears are mounted internally within bearing cham- This paper reviews issues that affect gear WPL, which is one
bers. The component of windage power loss becomes significant subset of problems that affect gearbox designers. The aim of the
in this case, and the flows associated with windage power loss review is to provide a comprehensive compilation of published
have a significant impact on the amount of heat transferred to the information on WPL to assist gearbox designers in identifying
oil within the chamber, which is a critical design consideration. relevant experimental and modeling information.
This paper provides a review of experimental investigations and To assist designers, some other issues that impact power loss
available models of gear windage power loss for spur, helical, and have been identified, and a few references have been supplied to
bevel gears. The aim of the review is to provide a comprehensive provide a starting point for interested readers. For example, there
compilation of published information on windage power loss to is the issue of the heat generated by gears. Some recent papers in
assist gearbox designers in identifying relevant experimental and the area provide a starting point; for example, Long et al. 关5兴 and
modeling information. While it is clear from the review of pub- Jia and Quinones 关6兴 described experimental and theoretical
lished work that the rotational speed, gear geometrical param- analyses for surface temperatures of gears, while Hohn et al. 关7兴
eters, degree of confinement, and density of the fluid surrounding provided a description of a whole gearbox analysis for heat gen-
the gear are important, the degree of effect and general solutions eration and dissipation. The effect of lubricant jet location on gear
for reducing power loss are less clear. 关DOI: 10.1115/1.2829983兴 operating temperatures is the subject of papers by Townsend and
Akin 关8兴 for spur gears and Handschuh 关9兴 for bevel gears.
Keywords: transmissions, gears, power loss, windage, gearbox
This review does not cover any issues regarding gear lubrica-
efficiency, empirical models, review
tion 共a recent review by Olver 关10兴 provides an excellent starting
point on this topic兲. Nor does this review cover dipped or im-
1 Introduction mersed gears, which experience churning power losses 共Boness
The losses associated with meshing gears are important in the 关11兴 and Luke and Olver 关12兴 provided information on this area兲.
design of many industrial, marine, and gas turbine situations. Meshing losses are a major proportion of gear power losses,
Gearbox efficiency varies from 98% to 99% for the best designed and it is outside the scope of this paper to provide a full review of
high power applications. The highest rated gearboxes now exceed the work in that area. However, some references are provided to
100 MW 关1兴, so a 2% loss of efficiency equates to 2 MW of lost assist the designer. Meshing losses include sliding friction
energy. 关13–16兴, oil rolling or pumping 关14,15,17兴, oil acceleration
There are different mechanisms for losses that have been iden- 关16,18,19兴, and ejection 关13兴.
tified within gearboxes; these are meshing losses, bearing losses, The rest of the paper is divided into experimental and modeling
windage losses, and churning losses. Depending on the applica- studies into WPL for spur, helical, and bevel gears.
tion, the relative importance of these mechanisms varies.
Windage power loss 共WPL兲 is defined as the power loss due to
the fluid drag experienced by the gear when it is running in air or
an air-oil mist. Churning power loss is defined as the power loss
2 Experimental Investigations of Gear Windage
Before reviewing the published data on gear windage, it is use-
Contributed by the Power Transmission and Gearing Committee of ASME for
publication in the JOURNAL OF MECHANICAL DESIGN. Manuscript received May 16,
ful to briefly summarize work that has investigated windage due
2006; final manuscript received June 13, 2007; published online February 14, 2008. to a rotating disk. An enclosed rotating disk can be regarded as a
Review conducted by Teik C. Lim. simplification of an enclosed single spur gear. Since Daily and
Journal of Mechanical Design Copyright © 2008 by ASME MARCH 2008, Vol. 130 / 034001-1
Pw = 0.0311 冋册s
r
−1/4
1/43/411/4r9/2 共1兲
ery, the loss of power due to the pumping of air through the teeth,
and the loss of power associated with the passage of a tooth
through the wake of the preceding tooth. For spur gears, the major
geometric variables in WPL are the radius and module, with the
While this bears some relation to fully enclosed gears 共especially
face width having a less significant impact. Lord found that a
for very low modules兲, it is not fully physically representative.
maximum WPL was experienced for a spur gear of Module 4 and
2.1 Spur Gears. Dawson 关25,26兴, Lord 关27兴, and Diab et al. that the minimum was seen for that of Module 1. However, the
关28兴 investigated windage due to single spur gears in isolation. experimental test only tested spur gears of Modules 1–5. Lord
Dawson 关25兴 investigated how windage losses are affected by also concluded that shrouding a single gear in air reduces windage
speed, size, gear geometry, and shrouding for single spur gears in losses to 25% of their unshrouded values. There were a number of
air. Dawson measured the WPL due to gear disk faces, smooth unresolved questions arising from his work, not least why was a
drum 共spur gear without teeth兲, and a full spur gear for a variety of peak in power loss seen related to a particular module, and was
modules. Thirty-seven spur gears of root diameter 300– 1160 mm this linked to radius and hence tip speed?
with face widths of 32– 187 mm and tooth modules of 2 – 24 mm Some of the testing conducted at Swansea 关27兴 investigated the
were tested. The work undertaken by Dawson was carried out in effect of shrouding single and meshed gears. Axial and peripheral
his own domestic garage using wood and cardboard representative shrouds were tested on a single spur gear. No measurable change
gears. in power loss was noted with changes in axial clearance from
Dawson concluded that for gears with realistic sized teeth and 1 mm to 10 mm. Significant reductions in WPL were measured
face width, the main contributors to windage losses were the teeth with the addition of peripheral shrouding. With a 1 mm peripheral
共e.g., gear of diameter of 760 mm, face width of 187 mm, and clearance, windage loss at high speeds was reduced by 75%. This
module of 8 mm, seeing over 90% of losses due to teeth兲. bears out the conclusion from Dawson’s work 关25,26兴. The rela-
From comparing the power loss at high and low face widths, he tionship between power loss and peripheral clearance was ap-
concluded that the teeth may be acting as a centrifugal fan draw- proximately linear. An equation for predicting WPL of shrouded
ing in air axially at the ends of the teeth and ejecting it radially spur gears 共Module 1兲 was developed 共see Sec. 3.1兲. Lord also
and/or tangentially toward the middle of the face. This was con- noted that the presence of a hole 共e.g., drain hole兲 in the axial
firmed by the visualization. Dawson then blocked axial entry into shroud caused an increase in power loss.
the teeth, and reductions of 22–44% were seen in the WPL. Diab et al. 关28兴 presented an experimental and numerical study
Dawson then investigated the performance of various shrouding of a single gear rotating in air without enclosure. Four different
共enclosure兲 arrangements consisting of axial and peripheral 共ra- spur gears were used with pitch diameters from
dial兲 shrouds with different spacings. All tests were undertaken on 144 mm to 300 mm, face width of 30– 60 mm, and Modules 4
one spur gear, diameter of 760 mm, face width of 187 mm, and and 6. Two different expressions were derived for WPL; one from
module of 8 mm. The percentage reduction in WPL was recorded dimensional analysis and the second based on the addition of
compared to the same gear rotating in free air. The greatest reduc- power loss from the faces of the gear and teeth, which were in-
tion 共66%兲 was seen for a smooth cylindrical shroud totally en- formed correlations from the experimental data. Flanges were
closing the gear and spaced 15 mm from the tooth tips. used to block the axial inlet and outlet to the teeth for a single gear
Dawson concluded that close fitting peripheral shrouds were the of module 6, and a drop of 60% in WPL was seen, which is
most effective, although full axial shrouds were effective, but to a considerably larger than the value seen in Dawson’s work. A sec-
lesser extent. However, as this was only tested for a single gear, ond paper by Diab et al. 关29兴 extends this work to include a large
the conclusions cannot necessarily be read across other situations. diameter double-helical gear and reports on analysis based on a
Dawson ran spur gears in water as an attempt to assess what force balance 共this paper is described in more detail in Secs. 2 and
effect the fluid surrounding the gear had on windage loss. Dawson 3兲.
concluded that the power loss showed a slightly greater depen- All of the experimental studies conducted on single spur gears
dency on rotational speed 共P ⬀ N3.0兲. However, only four points have identified the requirement for shrouding of gears as a method
were plotted and are within the experimental scatter of the previ- of power loss reduction. Visualization studies have shown the im-
ous testing of the gears in air. So, the conclusions would seem portance of the fluid flow around the gears.
difficult to support based on the data.
Dawson 关25兴 also investigated the inclusion of axial holes in a 2.2 Helical Gears. Dawson 关25兴 carried out one test on a
spur gear. He took a 760 mm diameter gear, with a face width of helical gear 共30 deg helix angle兲, which showed 70% of the loss
187 mm and a module of 8 mm, and tested it for a combination of measured for an equivalent spur gear. Flow visualization using
two, four, and eight holes at diameters ranging from smoke showed that air entered the helical gear axially and radially
28 mm to 124 mm 共five tests in total兲. The holes all lie on a at one end and was pumped out at the other end tangentially/
450 mm pitch circle diameter. The small holes showed no detect- axially and tangentially/radially. Windage loss was reduced by
able increase in windage loss over the unmodified gear. However, 42% when this pumping action was interrupted by blocking both
the large holes 共⬎100 mm兲 showed significant increases leading ends of the gear teeth. Blocking only the air inlet gave a 20%
to an 82% increase for eight 124 mm holes. The increase in wind- reduction and blocking only the air exit increased losses by 15%.
age was seen to increase roughly in proportion to the number of In Dawson’s second paper 关26兴, further tests on helical gears
holes. Again, as a single gear was used for these tests, it is difficult were carried out. Dawson studied the effect of varying helix angle
to justify any conclusion other than the claim that including axial from 0 deg to 50 deg for a single helical gear of a diameter of
holes may affect WPL. 0.514 m and a face/diameter ratio of 0.364. A modified tooth
Dawson’s work 关25,26兴, while limited, was the first published number was created to account for the helix angle, n⬘, where n⬘
best performing flange gap, while a 15% reduction was seen in the The power loss 共in horsepower兲 for the edge 共toothed side兲 is
clockwise direction. quoted as
The conclusion reached for helical gears is similar to that for
spur gears: that both shrouding and understanding the flows
around the gears are important in order to produce further effi-
PL =
15
冉 冊 冉 冊 冉 冊冉 冑 冊
N
0.746 1000
3
D
100
4
5L
100
Rf
tan 
共3兲
ciency gains. The term R f is the rough surface adjustment factor, L is the length
of the rotating element, and  is the helix angle. The total power
2.3 Bevel Gears. Winfree 关31兴 investigated the shrouding
loss is then calculated from the sum of Ps and PL. For a double-
共baffling or containment兲 of a single spiral bevel gear. He used a
helical gear, an additional term is employed to assess the flat
15 in. gear constructed from a rigid urethane compound. The test
between the gear teeth. Townsend commented that this method
rig used a vertical rotating shaft with a gear mounted horizontally
“leaves something to be desired” in the assessment of actual
such that the gear inlet 共smaller diameter face兲 was facing toward
losses but is helpful in establishing when windage becomes an
the bottom of the testing chamber with no shaft passing through
issue for a designer.
the inlet gear face.
Dawson 关25兴 presented a preliminary formula for WPL derived
Winfree measured the difference between a single unshrouded
from his test results by treating the sides and periphery of a spur
gear and a gear with the gear teeth completely covered 共shaped
gear separately. Gear shrouding is accounted for by a scaling fac-
disk兲. The WPL dropped 共unshrouded to covered case兲 by 86% tor. The formula is only applicable to spur gears within the vari-
dry and 81% wet 共the wet condition used water to simulate the able range tested. The formula is reproduced below,
amount of lubricant applied to meshing bevel gears兲. Uncovering
the gear but leaving the tooth inlet taped 共to reduce pumping from sides periphery
tooth inlet to outlet兲 reduces the power loss to 77% for both wet P = N2.9共0.16D3.9 + D2.9F0.75M 1.15兲 ⫻ 10−20⌽ 共4兲
and dry cases compared to a completely free gear. Tests were then ⌽ is 1 for an oil free atmosphere, is unity for a gear in free
carried out for fully shrouded gears, with the optimum tested space, 0.6–0.7 for a gear in a large enclosure, and 0.5–0.6 for a
shroud showing a reduction of 79% for a dry condition. “fitting” gear case.
However, it should be noted that Winfree’s test conditions only In his second paper, Dawson 关26兴 rearranged this equation
matches gears where a rotating shaft is not passing through the based on further analysis and testing, and the equation is repro-
gear. In the author’s opinion, where the shaft passes through the duced below,
gear the flow direction at inlet to the teeth is not aligned to the
entry area, and hence the flow is less likely to cause the magnitude P = 1.12 ⫻ 10−8C⬘N2.85d4.70.15 共5兲
of pumping 共and therefore power losses兲 seen by Winfree. where, as before, is the effect of gear enclosure. C⬘ is obtained
Winfree 关31兴 concluded that gears traveling over 51 m / s must from Fig. 3 of Ref. 关25兴 for spur gears, although, as noted in Sec.
be shrouded; this agrees with Ref. 关2兴. He also stated that the 3.1, there is a large degree of experimental scatter within the
spiral bevel gear rotating direction has no effect on baffle design results. For helical gears, an estimate is made based on Figs. 3–5
and that bevel gears pump from the inside to the outside across the in Dawson’s second paper 关26兴.