https://ntrs.nasa.gov/search.jsp?

R=19730013703 2018-11-29T08:12:51+00:00Z

N73 22430

NASA TECHNICAL NASA TM X-68214

MEMORANDUM

CM
00
vO
I
X

<
CO
<
Copy

SELF-ACTING AND HYDRODYNAMIC SHAFT SEALS

by Lawrence P. Ludwig
Lewis Research Center
Cleveland, Ohio 44135

TECHNICAL PAPER proposed for presentation at

Seal Education Course of the Annual Meeting of the
American Society of Lubrication Engineers
Chicago, Illinois, April 30-May 3, 1973
 SELF-ACTING AND HYDRODYNAMIC SHAFT SEALS

by Lawrence P. Ludwig

National Aeronautics and Space Administration

Lewis Research Center
Cleveland, Ohio UUl35 ..

.'. . ABSTRACT

Self-acting and hydrodynamic seals are described. The analytical

procedures are outlined for obtaining a seal force balance and the operat-
ing film thickness. Particular attention is given to primary ring re-
sponse (seal vibration) to, rotating seat face: runout. This response
analysis revealed three different vibration modes. Proposed applications
of self-acting seals in gas turbine engines and in rocket vehicle turbo-
pumps are described.. Also experimental data on self-acting face seals
operating under simulated ,gas turbine conditions are given; these data
show the feasibility of operating the seal at.conditions of 3^5 newtons
per square centimeter (500 psi) and 152 meters per second (500 ft/sec)
sliding speed.

INTRODUCTION

Considerable effort has been put into development of self-acting

and hydrodynamic seals because conventional contact seals cannot meet
the ever increasing pressure, temperature and speed requirements of some
modern and advanced rotating equipment. As an example, in advanced air-
craft gas turbine operation, seal speeds are expected to reach 183 meters
per second (600 ft/sec), and pressures may reach 3^5 newtons per square
centimeter (500 psi) (ref. l). .

The present solution to severe operating conditions is to use

clearance seals such as labyrinths; these seals are reliable but carry
performance penalties because of high leakage.

In advanced industrial compressors, pressures to 3^50 newtons per

square centimeter (5000 psi) are under consideration. The current sealing
system for these high pressures usually is a liquid buffer bushing that
requires a high pressure pump and electric drive motor (also for fail-
safe reasons a backup pump and motor is required). These compressor
seal systems with their allied plumbing and controls take up considerable
space and can cost as much as the compressor.

Thus, there is a need; for reliable, low leakage shaft seals that
require no external controls or equipment and that can operate at the
pressures, temperatures and speeds of advanced machines. Recent studies
(refs. 2 and 3) demonstrated that the self-acting seals can operate at
some of the advanced aircraft engine conditions and that they have much
higher capability than conventional contact seals and much lower leakage
than labyrinth seals. These types of seals may also have applications
in industrial compressors.
The objectives of this work are to review the development of self-
acting and hydrodynamic seals and to describe aerospace seal problem
areas for which these seals have application. In particular the develop-
ment of an aircraft gas turbine engine seal will be described in detail.

TERMINOLOGY

First a word about terminology. The terms "hydrodynamic" and . •

"hydrostatic" are often applied to describe seals for compressible fluids
as well as for incompressible fluids. But following the bearing terminology,
the terms , "selfr-acting" and "pneumostatic" will be used to describe seals .
for compressible fluids; and "hydrodynamic" and "hydrostatic" will be
reserved for seals for incompressible fluids.
It should be noted that conventional contact seals often operate
with separation of the sealing surfaces because of forces produced by
self-acting, pneumostatic, hydrodynamic or hydrostatic effects. (These
types of seals are not the subject of this paper). For example, many
conventional radial face seals for liquids (such as water pump seals)
run with hydrodynamic separation of the sealing surfaces; the hydro-
dynamic forces.being produced by surface waviness caused by local thermal
expansions and.wear. Thus in a conventional radial face seal, the seal-
ing and hydrodynamic bearing functions are performed by the same surfaces.
This dual performance is speed limited^because the heat generated in
shearing of the lubricating fluid film,causes fluid vaporization which
degrades the hydrodynamic bearing action.

Data from reference k, shown in figure 1, shows that oil boiling

temperatures (539 K (510° Fj^etre readily achieved even under low speed
shearing of a thin lubricating film of oil. In the case depicted (fig. l),
the sliding speed was 11.3 meters per second (37 ft/sec). (This result
reveals why seal cooling and material conductivity are critical design
parameters in high speed seals).

References 5 and 6 describe a radial face seal that operates hydro-

dynamically and, as a consequence, has much higher pressure and speed
capability than conventional face seals. The construction of this hydro-
dynamic seal is similar to that of a conventional face seal except that
the primary ring face contains circular grooves (see fig. 2). Seat
rotation causes the sealed liquid to circulate through these grooves and
thus the grooved region is cooler than the other regions. Therefore
thermal deformations produce wedge geometries that give rise to the hydro-
dynamic forces. Figure 2 also illustrates the hydrodynamic force (pressure
profile) that is built up in the region of the circular grooves. This
type - of seal apparently maintains a more efficient hydrodynamic action
(than a conventional face seal) as pressure and speeds increase, and
thus it has superior performance. Successful .operation at pressures to
2^13 newtons per square centimeter (3500 psi) and 101 meters per second
(333 ft/sec) for thousands of hours have been reported (refs. 5 and 6).

It should be noted that, as with conventional seals, the sealing

and thrust bearing functions, of the seals described by references 5 and 6,
are performed by the same surfaces. Since the surfaces do not contain
any "machined in" bearing geometry this type of seal is not covered in
this paper; although', dit > as with some'-conventional: seals, operates hydro-
dynamically.

SELF-ACTING AND HYDRODYNAMIC SEALS

The subject of this paper is the type of self-acting and hydrodynamic

seals in which the bearing and sealing functions are separated, the self-
acting or hydrodynamic action being produced by a "machined in" bearing
geometry and not by thermal distortion or wear. Figure 3 shows this type
of self-acting seal. (Since a hydrodynamic seal assembly would be the
same in principle, the discussion can be limited to the self-acting seal.)

Self-Acting Face Seal Nomenclature

A self-acting face seal is similar to a conventional face seal except

for the added feature of a self-acting geometry (gas lubricated thrust
bearing) as shown in figure 3 (ref. 7). As with a conventional face seal,
it consists of a rotating seat that is attached to the shaft and a non-
rotating primary ring assembly that moves in an axial direction; and,
thus, the seal can accommodate axial motion such as engine thermal expansion.
The secondary seal (piston ring) is subjected only to the axial motion
(no rotation) of the primary ring assembly. Several springs provide
mechanical force to maintain contact at start and stop. In operation,
the sealing faces are separated a slight amount (in the range of 0.00025
to 0.00127 centimeters, (0.0001 to 0.0005 i n . ) ) by action of the self-
acting lift geometry. This positive separation results from the balance
of seal forces and the gas film stiffness of the self-acting geometry.
The self-acting geometry can be any of the various types used in gas
thrust bearings; the Rayleigh step type is illustrated here.

Within the seal industry there is a wide variety of terms used to

describe similar seal parts. The ASLE seal glossary (ref. 8) has provided
much needed guidance in seal nomenclature, and the self-acting nomen-
clature that follows is mainly an extension of this ASLE work. The
 u
nomenclature applying to an'assembly of parts (fig. 3 ( a ) ) is:

1. Primary seal - Seal formed by the sealing faces of the seat and
primary ring. Relative rotation occurs between these sealing faces.

2. Secondary seal - Seal formed by the sealing surfaces of the :sec^- „

ondary ring^ • In-^the'''case(-:6foa-bell^vs -seal:-the "secbndiry^seal is^ifie-bel-
lows'itself.
3. Static seal - Seal formed by the mating surfaces of the primary
ring and its carrier (in some designs the static seal is an interference
fit).

H. Self-acting geometry (the term hydrodynamic geometry is used

for incompressible fluids) - Lift-pad geometry (Rayleight step bearing)
and mating face which together produce the thrust bearing action to
separate the sealing surfaces.

5. Film thickness (h) - Distance between primary sealing faces or

between surfaces forming the self-acting geometry. For parallel surfaces,
the film thickness at the primary seal is the same as at the self-acting
geometry. (Note that h may vary with radial and circumferential
position and with time.)

6. Seal head - Assembly that is axially moveable and consisting of

primary ring, its retainer (if any), and its carrier. (The retainer
and the carrier are combined into one part in some designs.)

The nomenclature applying to single parts (fig. 3 ( a ) ) is:

1. Seat - Part having a primary sealing face and mechanically con-

strained with respect to axial motion..

2. Primary ring - Part having a primary sealing face and not con-
strained with respect to axial motion.

3. Retainer, primary ring - Part that retains primary ring.

4. • Secondary ring - Part having secondary sealing surfaces which

mate to the secondary sealing surfaces of the carriers.

Force Balance, General Description

To determine film thicknesses and leakage in a self-acting seal,

the axial forces acting on the seal head (assembly of the primary ring
and its carrier) must be determined for each operating condition. These
forces are comprised of the self-acting lift force, the spring force,
and the pneumatic force due to the sealed pressure. Essentially, :
the analysis requires finding the film thickness for which the opening
forces balance the closing forces. When this equilibrium film thickness
is known, the leakage rate can be calculated, (it should be noted that
this force balance procedure is for the case in which the seat face has
zero runout with respect to the shaft centerline. Seat face runout
introduces dynamic film thickness changes, as discussed in a later section.)

To illustrate force balance calculation, an aircraft gas turbine

seal (ref. 9) was chosen as an example that is given in the following
sections. The force balance analysis considers four design points
(table l) namely: idle, takeoff, climb and cruise. The seal is a
nominal 16.76 centimeter (6.6 in.) diameter size suitable for large gas
turbine engines. Pertinent dimensions are given in figure 3.

ANALYSIS OF.SELF-ACTING SEAL •

Primary Seal Pressure Gradient (Opening Force)

In calculating the axial force balance of the primary ring, the

pressure gradient in the primary seal must be considered (see fig. 3
for primary seal location). The quasi-one-dimensional flow model der-
scribed in reference 10 was used for these calculations. From a gas
leakage flow standpoint, the primary seal is a long passage. For
example, a typical operating film thickness of a self-acting seal is in
the range of 0.00102 centimeter (0.000** in.), and a typical radial length
of the primary seal is 0.127 centimeter (0.050 in.). Thus, the length-
height ratio of the flow channel is 125/1. Previous work (refs. 11 and
12) has shown that this leakage passage has the following .qualitative
features: . .' .,
1. Laminar leakage flow prevails over much of the range of interest
in seals for gas turbines (pressure range of 3^5 N/cm abs (500 psia)).

2. Sonic velocity (choking) exists at the passage exit when pressure

ratios are approximately U/l or greater at a mean film thickness of
0.00102 centimeter (0.000*4- in.).

3. Pressure pr6files across the primary seal for choked and non-
choked flow can be very different.

k. Since the primary seal radial width is small compared with its
diameters, the area expansion effect on flow can be ignored.

5. The leakage flow and pressure profile are significantly different

if the surfaces of the primary seal are not parallel. (See ref. 13 for
a discussion of the effects of converging and diverging sealing surfaces.)

The primary seal and mathematical model used in the quasi-one-

dimensional analysis of reference 10 is shown in figures U and 5. As
mentioned., the area expansion effects are ignored and the model is a
passage of height h and length 1.

From stagnation source conditions of PQ and Vo, see figure 5, an

isentropic expansion is considered to occur ahead of the entrance to the
primary seal gap. Thus, the entrance pressure, P^> is less than the stag-
nation pressure PQ, and the entrance velocity, V-j_, is a finite value.
To account for entrance loss and viscous friction, it was found necessary
to use an entrance loss coefficient of 0.6 (ref. 10). Thus, the entrance
velocity, V-i, is less than that calculated by isentropic expansion.

Flow in the sealing gap is assumed one-dimensional and a friction

factor is introduced to account for viscous losses. At the exit, three
conditions are considered in the analysis. First, exit velocity ¥2 is
subsonic and exit pressure, ?2, is equal to reservoir pressure, Po. Seconds
exit velocity, VVij is sonic and exit pressure, Pg, is equal to reservoir
pressure Pos and third, exit velocity, ¥2, is sonic, the flow is choked,
and exit pressure, ?£, is greater than the reservoir pressure Po. In
addition, the following restrictions apply to the mathematical model
(see ref. 10 for details):

1. Effect of seat rotation on flow is neglected.

1/2
2. For subsonic flow at a Mach number $ 1/(specific-heat ratio) , the
flow is isothermal and the model yields the classical cubic dependence
of mass flow on film thickness.
1/2
3. For.flow at a Mach number > !/•(specific-heat ratio) ,/theoviscous
effects are approximated by a mean friction factor of 2U/Re (Re = Reynolds
2n
number).

Typical pressure gradients across the primary seal for two design
points (idle and take off) are shown in figure 6 (ref. 9). The important
point -is that choked and nonchoked flows can have pressure gradients with
very different shapes thus affecting the opening force (that is, the
integrated force under the pressure-gradient curves). ' .. ''.

Figure 7 shows this primary seal opening force as a function of film

thickness for the four design points. Note that for all four design points,
the'opening force is nearly independent of film thickness. (True only for
parallel faces.)

Self-Acting Geometry (Opening Force)

The self-acting geometry (lift pads) consist of a series of shallow

recesses 0.0025 centimeter (0.001 in.) deep arranged circumferentially
around the seal under the primary seal face as shown in figures 8 and 9«
An important point is that the lift pads are bounded at the inside diameter
and the outside diameter by the sealed pressure P-j_. (This is accomplished
by feed slots connecting the annular groove directly under the 'primary
seal face.) Therefore, a pressure gradient due to gas leakage occurs
only across the primary seal.

The complete primary ring and its 20 lift pads are shown in-figure 9,
and as indicated, motion of the seat over these shallow recesses drags
air from .the feed slots.into the shallow pad recesses. Since the: air is ;
restricted from leaving the recesses "by .the side and back lands, a lift
force, or thrust bearing action, is produced. ,

This self-acting lift pad is approximated by the mathematical model

shown in figure 10. Note that the .curvature effects have been-neglected
in the model. Therefore, the model corresponds to a Cartesian coordinate
system. (The mathematical model is described in detail in reference- 1^.)
In this analysis the following restrictions apply:

1. The fluid is Newtonian and viscous.

2. A laminar flow regime is assumed.

3. Body forces are negligible. . .

The analysis of reference lU admits nonparallel surfaces; thus, the

effects of surface deformation on lift force can be evaluated.

Figure. 11 shows the calculated lift force (ref. .lU) produced.by the
self-acting geometry for the four operating points of table I. Inspection
of figure 11 reveals at film thicknesses of 0.00127 centimeter (0.0005 iri.J-
and greater the lift force is small. However, at film thicknesses less
than 0.00127 centimeters/iO'.:OOQ5:.;iri'iv)-'jthe;'lift,;'force-;-rapidly increases as
the film ;thiqkness:-.decreases'. cThus.,' thelselfrracting geometry •.-has: a very ,
high film stiffnessj.thathenablfes .the sealrheadsto track..the::mot.ions..of • the
rotatingoseat face*.

As mentioned previously, the self-acting lift force tends to open

the seal, and is added to the primary seal opening force to. obtain the
total opening force. The resulting combined opening force given in figure 12.
Again, this is for parallel surfaces only.

. Closing Forces

The closing forces, (see fig. 13) acting on the primary ring are a ..
spring force and a pneumatic force. Since the full sealed pressure acts .
to the inside diameter of the primary seal, the net pneumatic closing
force acts only on the annular area between the primary-seal inside
diameter and the secondary-seal outside diameter. For the seal design
under consideration this annular area is (see fig. 13):
 A = - D = U.66 cm2 (0.122 in.2)

and the resulting closing forces due to the sealed pressure are listed in
table II . It should be noted that these closing forces are for average
dimensions at room temperature. At operating temperature, a thermal
growth difference may cause a change in the relation between the secondary-
seal outside diameter and the inside diameter of the primary seal. Thus,
the closing force could be a function of temperature. However, in this
report all force balance calculations are based on room-temperature di-
mensions. In critical.applications the force balance at operating tempera-
ture may have to be evaluated.

Net Force Balance

In.a conventional seal, the net closing force is resisted by solid-

surface rubbing contact; thus, a total force balance is achieved. But
in self-acting seals the force balance is achieved without rubbing contact.
Therefore, for a given design point, the seal will operate at a film thick-
ness such that the total opening force .exactly balances the total closing
force. This•operating film thickness is obtained by plotting total open-
ing forces (fig. 12) and total closing forces (table II) as a function of
film thickness. The intersection (see fig. lM of these curves is the
equilibrium operating film thickness. This film thickness determination
does not take into account dynamic running factors such as se.at face run-
out and piston ring damping.

Once the equilibrium film thickness is found, the calculated leakage

can be determined by using the quasi-one-dimensional method outlined in
reference 10. For the four design points, this leakage rate is given in
figure 15 for various film thicknesses, and table III shows the calculated
ieakagie-tor:;ths--eiiuill-bri^=filffl^thickn"esses-''e;f figAirer'l4:-( for ^parallel
faces1)'.-' -; •
In designing a seal, each operating point should be .checked for
equilibrium film thickness. If these film thicknesses are not satis-
factory, the closing force should be altered to bring all operating points
within a satisfactory film thickness regime. Experience has shown that
the satisfactory film thickness regime is about 0.00025 centimeter
(0.0001 in.) on the low end (some tolerance to thermal deformation must
be maintained) and 0.0013 centimeter (0.0005 in.) on the high end. These
limits are only approximate and depend to a large extent on the dynamic
and thermal condition to which the seal is subjected. The high limit of
practical film thickness is established by seal dynamics and leakage
considerations. In particular, the primary.ring response to the seat
face runout becomes excessive as the mean film thickness increases (ref. 15).
This is because the stiffness of the gas film decreases with increasing
film thickness. These dynamic considerations are covered in the section
on face runout effects.
 Effect of Nonparallel Seat Face

Figure 16 shows, in an exaggerated manner, the coning displacement

of the seal seat. (The primary ring could also be coned.) This type
of coning displacement, which can "be caused by thermal gradients, results
in nonparallel faces within the primary seal and the self-acting geometry.
These nonparallel faces have a significant effect on load capacity of the
self-acting geometry;"also the primary seal opening force is affected.
Thus, in design, the equilibrium operating film thickness should also be
calculated for anticipated coning displacements.

As an example of the effect of this coning, design point 2 was

checked (using the methods of reference lU) for equilibrium film thick-
ness for a distortion of 0.0013 centimeter (0.0005 : 'in. : )'across'the self-
acting pad. This is a distortion of 2 milliradians and is judged typical
of some seal operation (ref. 15). Figure IT shows this assumed distortion
in an exaggerated manner; note that the self-acting geometry and primary
seal dimensions are given.

Figure 18 shows the self-acting lift force for the 2-milliradian

distortion of the seat f a c e , . Note that-the force is plotted as a func-
tion of the mean film thickness of the self-acting pad. Also plotted,
for comparison, is force generated for a parallel film. Note the signi-
ficant reduction in lift force due to distortion, especially at the lower
film thicknesses.

As noted previously, the primary seal opening force is also affected

by nonparallel faces; and this was calculated by using an analysis similar
to reference 12 for the 2-milliradian distortion. The results are given
in figure 19. For the divergent deformation shown in figure IT, there
is a marked reduction in opening force as the film .thickness decreases.
For convergent deformation, the opening force would increase as film
thickness decreases; and this is desirable. Unfortunately, in aircraft
mainshaft seals, the divergent deformation is a natural tendency due to
thermal gradients.

Finally, in figure 20, the equilibrium film thickness for a 2

milliradian distortion is found by finding the intersection between the
total closing force and total opening force. The mean film thickness is
about 0.00169 centimeters (0.00066 in.). • Thus the minimum film thickness
is 0.0010U centimeters (O.OOOUl).

With the equilibrium film thickness values, the gas leakage was
calculated by using the method previously outlined. The results revealed
that the leakage rate for the 2-milliradian deformation was nearly twice
that of the parallel-face case.
 10

Effect of Seat Face Runout

The preceding analysis was for operating film thicknesses that did
not vary vith time and this would be the situation if the rotating seat
face had zero runout. However, the seat face will always have some run-
out with respect to the axis of rotation; and in particular, the maximum
runout used in practice is of interest since it will induce the maximum
time-dependent film thickness changes.
Of interest, then, is how the primary ring responds to the runout
motions of the seat face. This response determines the film thicknesses
at any instant. Experimental data reported in reference 15 reveal.that
the primary ring can follow (dynamically track) the seat face motion over
a considerable range of face runouts. This data was obtained by mounting
two proximity probes (90° apart) on the ring retainer and recording the
change in film thickness as a function of time. A schematic showing the
probe location is given in figure 21. Some results from reference 15
are given in figure 22 which shows that for a seat face runout of 0.0020
centimeter (0.00085 in.) full-indicator reading (F.I.R) the ring response
is nearly in.phase and the total change in film thickness is 0.0017 centi-
meters (0.00067 in.).
Also, the film thickness varies circumferentially; that is, the film
thickness is not axisymmetric. This response of the primary ring can be
explained by observing the time sequence at one point fixed with respect
to the laboratory observer. This is shown in figure 23. The upper part
of figure 23 illustrates seat and primary ring with maximum film thick-
ness (exaggerated) at a time arbitrarily designated as zero. At some
later time, when the seat has rotated thru 180°, the film thickness at
this point is a minimum. Figure 2B could also represent the simultaneous
film thicknesses l80° apart.

An analytical.program has been developed for the purpose of pre-

dicting this ring response to seat face runout (ref. l6). Analysis of
the seal depicted in figure 3 revealed that the primary ring response
is markedly affected by secondary seal friction and by inertia of the
primary ring assembly. The friction effect is illustrated in figure 2k;
as runout increases, there is a friction level that, if exceeded, will
retard the primary ring motion to such an extent that rubbing contact
will occur (line (l)); also for the higher face runouts there is a
friction level below which the inertia forces are so high that the primary
ring cannot follow the runout (line (2)). Therefore, some friction is
probably desirable for most applications because of the practical limits
on control of face runouts. Further, the data suggest that the primary
ring assembly inertia should be kept as small as practical in order to
maintain good response (avoid unstable operation). Figure 2h in general
shows only how the nosepiece response is affected by secondary seal
friction and seat face runout; in a detailed analysis by reference 17,
 11
three different types of nosepiece responses were revealed by a parametric
study using different magnitudes of seat face runout and secondary seal
friction. These three cases are:

Case 1 - primary ring motion duplicates seat face runout motion

and can be described by rotation (rocking) about tvo orthogonal axes.
However, because of primary ring inertia and/or friction, the face of
the ring has an angular misalinement with respect to the face of the
seat. Therefore, the film thickness between the faces is not uniform.

Case 2 - Same as case one plus an additional axial vibration

component.
Case 3 - Seal failure (film thickness reaches zero). This case can
occur when the frictional forces are either too low (when inertia forces
are high) or too high for the available load capacity of the self-acting
pads.

Results from reference 17 predict case 2 type operation for the seal
shown in figure 3. These results are shown in figures 25; to 27, - The.'
primary ring assembly weight is 0.80 kilograms (1.75 Ibv), and other
operating conditions (which simulate operation in a gas turbine) are:

Sealed pressure, newtons per square centimeter. . ... 207 (300 psi)
Sealed gas temperature, K 8ll (1000° F)
Sliding speed, meters per second 152 (500 ft/sec)
Initial film thickness, centimeters 0.00081 (0.00032 in.)

Figure 25 shows the rotation (rocking) about the X-axis (motion about
the Y-axis is the same except for a 90° phase shift). The trace indi-
cates that rocking is not continuous with time, the ring "sticks" twice
per revolution. This "sticking" also occurs in the motion.about the
Y-axis; therefore, a discontinuity in motion occurs four times per
revolution. The effect of this discontinuity (four per revolution) is
seen in the opening force of the self-acting geometry (figure 26). Each
"sticking" leads to a decrease in film thickness and corresponding build-
up of axial fluid film force. Thus the primary ring has a translation
(axial) motion and this is shown in figure 27.

POTENTIAL APPLICATIONS

Gas Turbine Mainshaft Seals

Mainshaft seals are used in gas turbine engines to restrict gas

leakage into the bearing compartments (sumps). The seal at the front
of the compressor (fig. 28) usually presents no problem since the sump
can be surrounded by compressor bleed air of moderate pressure and
 12

temperature. Bleed air leaks through the seal and, thus, pressurizes
the sump. For moderate bleed air pressures and temperatures, almost
any type.of seal is adequate (ring, circumferential, or face type). As -
an example, ring seals have been used at 37.2 newtons per square centi-
meter (5^ psi) and U38 K (330° F). Circumferential seals have been used
to 58.6 newtons per square centimeter (85 psi) and 6UU K (TOO0 F) with a
sliding speed of 73.2 meters per second (2^0 feet per second). Face
seals are also adequate, as some are operating at 86 newtons per square
centimeter (125 psi), TOOK (800° F), and a sliding speed of 10T meters
per second (350 feet per second). (The final selection of the seal type
for the front compressor location depends on the many design and system
considerations, including designer preference.

The seal operational requirements become more severe at the middle

of the engine and at the turbine bearing sump. The thrust bearing must
be large enough to carry the net thrust, load of the rotating parts. In
addition, the sump is usually surrounded by higher pressure gas, and
corresponding higher temperature, than the front of the engine.

Some shaft seal arrangements for the turbine bearing sump location
are shown in figures 29 to 31. Here the basic problem is protection of
the bearing sump from the turbine cooling gas. In early engines, since
the cooling gas pressure .and temperature were relatively low, a single
labyrinth seal, which restricted turbine cooling gas leakage into the
sump, was adequate (fig. 29). (Some engines, use a once-through lubrica-
tion system in which the sump is pressurized by relatively cool bleed
air, and this air and the oil lubricant are allowed to vent overboard.)
At these pressures, the efficiency loss (and oil loss) due to seal leak-
age was not significant.

A disadvantage of the labyrinth seal, as compared to the closer-

clearance seals (ring, circumferential, and face), is high leakage with
corresponding larger ducting requirements and easier passage of airborne
water and dirt into the sump. In addition, for labyrinth seals, reverse
pressure drops must be avoided to preclude high oil loss. Furthermore,
ram air temperature (minimum-temperature air available for sump-pressuriz-
ing) increases with increasing flight speed. Therefore, a flight speed
limit exists beyond which the low-leakage close-clearance seals are
desirable in order to preclude sump fires.

As turbine cooling gas pressure requirements increased with develop-

ment of engines with higher turbine inlet pressures, the single labyrinth
seal (fig. 29) was no longer suitable. A conventional face seal (fig. 30)
has satisfactory performance to pressures of 86 newtons per square centi-
meter (125 psi), temperatures of TOO K (800° F), and sliding speeds of
10T meters per second (350 feet per second). Recent seal studies (ref. 2)
showed that at higher pressures, temperatures, or sliding speeds perfor-
mance was unsatisfactory from a wear and leakage standpoint; subsequent
analysis revealed that the seal limitations were primarily due to thermal
 13

deformations that induced seal imbalance, which, in turn, caused high

wear and leakage.

When the pressure, speed, and temperatures exceed the capability of

conventional contact seals (such as that of fig. 30), a pressure staging
seal system of multi-labyrinths is used (such as that in figure 31). In
this seal system, a labyrinth seal restricts the leakage of relatively
high pressure (P-|_), high-temperature turbine cooling gas to an over-
board vent or to a low-pressure region ( P ^ ) . Low-pressure compressor
bleed (P-p surrounds the sump and, hence, provides thermal protection;
and leakage (through the labyrinth) into the sump provides required sump
pressurization. It should be noted that, as the turbine cooling gas
pressure (P^,) increases, the labyrinth seal leakage increases and the
detrimental effect on efficiency increases. For example in some modern
engines the loss in thrust specific fuel consumption amounts to 0.9
percent at one seal location.

It is anticipated that advanced engines will have compressor dis-

charge pressures in the range of 276 to 3^5 newtons per square centi-
meter (1*00 to 500 psi) and seal temperatures of approximately 1033 K
(lUOQ 0 F). The seals for these engines will have sliding speeds of 152
to 183 meters per second (500 to 600 feet per second). Engines for
supersonic aircraft may generally have lower compressor discharge
pressures, but the turbine cooling gas temperatures will be near 922 K
(1200° F)/ The high sliding speeds in these advanced engines dictate
that the sealing surfaces should not be in rubbing contact. Thus, the
self-acting seals, since they operate without rubbing contact, are
potentially useful in the advanced engines.

Simulated Gas Turbine Engine Operation

The self-acting face seal (fig. 3) was evaluated under conditions

that simulated the environment of a gas turbine engine sump over a
wide range of combinations of oil'temperature, seal sliding speed,
sealed pressure differential, and sealed air temperature (ref. 3). The
rig used is shown in a cross-sectional view in figure 32. The rig shaft
was mounted on a 130-millimeter-bore roller bearing and on a 110-
millimeter-bore duplex ball bearing that carried the thrust, from the
sealed pressure; both bearings are typical of those used in large air-
craft engineso The air to be sealed (maximum of 922 K (1200° F) and
3^5 newtons per square centimeter (500 p s i ) ) is introduced through tubes
as shown in figure 32. Thus, the sealed pressure is at the inside diameter
of the primary ring, and air leakage is across the primary sealing faces.
(dam) into the bearing sump (also see fig. 3). Air also leaks across the
secondary sealing surfaces into the sump.

The test rig is driven by an industrial internal combustion engine

driving through a four-speed truck transmission and 12:1 step-up gear
box (fig. 33). The operating speed range was TOGO to 21 000 rpm. Oil
was supplied to the rig bearing at temperatures typical of engine sumps
(39^ K.(250° F) to 1*50 K (350° F)). The sealed air is heated to operating
temperature by a resistance heater. In the later runs, a 25-micrometer
filter was introduced (shown dashed in fig. 33) for the purpose of reduc-
ing the amount of hard particles carried by the air leakage flow. (The
hard particles were found to be oxides generated in the resistance heater.)

A general increase in air leakage was noted as speed increased. This

means that the average gas film thickness between the sealing surfaces
increased as the speed increased. This result was attributed to primary
ring response as discussed in the section on face runout and secondary
seal friction effects.
To check wear and leakage characteristics the seal was subjected to
long term runs (10 to 300 hours). A typical example of seal leakage is
shown in figure 3^ for a 100-hour run at a sliding speed of 122 meters
per second (1*00 ft/sec), a sealed pressure differential of 138 newtons
per square centimeter (200 psi), and a gas temperature of 8ll K (1000° F).
From an initial seal leakage of 1.13 standard cubic meters per minute
(ho scfm), the seal leakage decreased to 0.79 standard cubic meters per
minute (28 scfm) at the end of the 100 hours. This gradual leakage
decrease was attributed to wear-in of the secondary ring causing better
sealing, (in some runs fretting wear of the secondary seal caused a
gradual increase in seal leakage rate.)

Inspection after the 100 hour run (150 hours total time on the seal)
revealed that the chromiumr-.carbide-coated molybdenum seat was in excel-
lent condition with no sign of severe contact between the seat and the
seat ring nosepiece. The self-acting pad land surfaces were polished,
and the primary ring sealing face (dam) had circumferential scratches
indicating abrasion due to air entrained debris.. (Probably hard particles
from the air heater operating at I2.Uk K (l600° F).)

The abrasive effect of hard particles was checked in a separate 10

hour test run using crushed-quartz fed into the seal air at a rate of
3.6 grams per hour (0.125 oz/hr). The wearJ.of.rtlie;:^self-acting:'.padsowas
ins ignificant ah'dethe; .sealing>-damlsliowed.iSlv.7,5.-micrometers.:•. (-0•;•.O.Q125,nin.)
wear:; at It he inneraedge of .fhe. is easing •.: dam < . I:'..' ;....] •••.,.• •-.'•. : . ' • L .•

In addition to the endurance test running, the seal was subjected

to a wide range of operating conditions. No rubbing contact occurred
under conditions as low as 69 newtons per square centimeter (100 psi)
and 6l meters per second (200 ft/sec), to as high as 3^5 newtons per
square centimeter (500 psi) and 152 meters per second (500 ft/sec). In
addition, after 38 hours at pressures that ranged between 21*1 to 3^5
newtons per square centimeter (350 to 500 psi) and at sliding speeds
between 122 to 183 meters per second (kOO to 600 ft/sec) the primary ring
 15

(carbon) wear was insignTficahtr{>:see'Jfig'. 35)'.;:L::Typieal leakage- - •'

values are shown in table IV.
In order to check the'effect of starting and stopping, the seal was
subjected to Uo starts and stops .with room-rtemperature air at 69 newtons
per square centimeter (100 pst). The seal-speed was increased from 0 to
76 meters per second (0 to 250 ft/sec) in about 15. seconds. The seal
was then allowed to coast to a stop in about 25 seconds. A comparison of
the surface profile traces before and after the Uo starts and stops re-
vealed carbon wear to be less than 1.27 micrometers (0.00005 in.).

Rocket Engine Turbopump Applications

In.rocket engine turbopumps, shaft seals are used to seal propellants
such as liquid oxygen and hot hydrogen gas. Experience has shown that
the shaft seal system is a very critical item and seal failure can result
in total destruction of the vehicle. Figure 36 shows a typical rocket
engine seal system used in current practice. The purpose of the system
is to separate liquid oxygen at pressures as high as 310 newtons per
square centimeter (^50 psi) from ,the hot turbine gas at 689 newtons
per square centimeter (1000 psi). The seal system consists of three
sealing areas with a central helium purge and a dual vent.

The liquid oxygen seal is a conventional rubbing contact seal with

a bellows secondary seal. Oxygen is prevented from leaking past the
helium seal by the purge flow of helium, and both the oxygen and helium
pass out the vent indicated in figure 36. A similar action occurs on
the hot hydrogen gas side. However, the hydrogen gas seals are conven-
tional ring (clearance) seals. The helium purge seals are :also conven-
tional ring (clearance); seals.

In the seal system such as depicted in figure 36 there are a number

of problem areas, the most critical of which is compatibility of the seal
material with liquid oxygen. Therefore, as indicated, welded bellows
designs are very often used. Another problem area has been wear life.
The typical wear life is now in the range of 2 to 3 hours, which is a
satisfactory margin for the current rocket vehicles. However, for the
shuttle vehicle a total life of 10 hours is needed with a start-stop
(mission) capability of 100. Clearly, the current seal technology is
not adequate for the shuttle life requirements; and the self-acting
seal may provide a solution.

An advanced seal system is under development (fig. 37) with a

hydrodynamic seal for liquid oxygen and a self-acting circumferential
seal for the helium. The seal for liquid oxygen is a hydrodynamic type
with Regleigh step type thrust bearing geometry. Instead of a bellows
secondary, a carbon piston ring is used. Liquid oxygen pressure is near
 16
newtons per square centimeter (500 psi) and seal sliding speed about
1ST meters per second (U50 ft/sec)/ The predicted operating film thick-
ness is 0.00076 centimeters (0.0003 i n . ) . .

The helium purge seal is a circumferential seal, modified to have •

self-acting geometry similar to that shown in figure 38. This self-
acting geometry, as in the case of the self-acting face seal, acts to
maintain positive separation between the rotating shaft and the non-
rotating carbon rings.. The helium gas pressure is 35 to 69 newtons per
square centimeter (50 to 100 psi). In comparison, the clearance ring
seal shown in the conventional configuration (fig. 36) has a leakage
rate 50 times greater. Thus, by use of the self-acting helium seals,
the on-board helium inventory can "be reduced and significant weight
saving will 'acrue because of the smaller tanks and lines.

SUMMARY

Self-acting and hydrodynamic seals and their potential applications

in gas turbine engines and in rocket engine turbopumps are described.
In particular the analytical procedure is given for predicting the
operating film thickness of a 16.76 centimeter (6.6 i n . ) diameter seal
designed for gas turbine applications. The analysis and subsequent
operation of this seal under simulated gas turbine conditions revealed
the following:

1. Analysis

a. Noncontact operation with acceptable leakage is predicted at

the selected four.design conditions of idle, takeoff, climb, and cruise.

b. The operating film thickness ranged between 0.000lv6 and 0.00119

centimeter (0.00018 and 0.000^7 i n . ) for the four design conditions.

c. The calculated seal leakage rates ranged between 0.01 and O . U o

scmm ( O . U and 14.0 scfm) for the four design conditions.

d. For a typical operating condition, noncontact operation was

predicted under the assumption of a 2-milliradian face deformation. Gas
leakage was about twice that for parallel-face operation.

2. Experiment-Simulated Engine Operation

a. The self-acting seal ran without rubbing contact during all of

the simulated engine operating conditions. Of particular interest was
that noncontact operation was maintained at the advanced engine conditions
of a 152-meter-per-second (500 ft/sec) sliding speed, a 3^5 newtons per
square centimeter (500 psi) sealed pressure differential, and a 8ll K
(1000° F) -sealed air temperature.
 IT
b. The leakage rates during endurance tests ranged from 0.79 to
1.63 standard cubic meters per minute (28 to 57.5 scfm).
c. Self-acting geometry vear due to rubbing contact proved to be
insignificant.
d. Two seal wear problems were noted. These are (l) fretting wear
of the secondary ring (piston ring) and (2) erosion of the seal dam
(primary seal) by hard particles.
e. Wear due to starts and stops proved to be insignificant. For
Uo starts and stops wear was less than 1.27 micrometers (0.00005 in.).

REFERENCES

1. Anon., "Aircraft Propulsion," NASA SP-259 (1971).

2. Parks, A. J., McKibbin, R.- H., Ng, C. C. W., and Slayton, R. M.,
"Development of Main-shaft Seals for Advanced Air Breathing Propulsion
Systems," Pratt & Whitney Aircraft Rep. PWA-3l6l, NASA CR-72338 (1967).

3. Povinelli, V. P., and McKibbin, A. H., "Development of Mainshaft Seals

for Advanced Air Breathing Propulsion Systems - Phase II," Pratt &
Whitney Aircraft Rep. PWA-3933, NASA CR-72737 (1970).
k. Orcutt, F. K., "An Investigation of the Operation and Failure of Mechan-
ical Face Seals," Presented at the Uth International Conference on
Fluid Sealing, British Hydromechanics Research Association, Cranfield,
Bedford, England (1969).

5. Mayer, E., "High Duty Mechanical Seals for Nuclear Power Stations,"
Paper A5, Presented at the 5th International Conference on Fluid Seal-
ing, British Hydromechanics Research Association, Cranfield, Bedford,
England (1971).
6. Mayer, E., "Thermodynamics in Mechanical Seals," Presented at the kfh.
International Conference on Fluid Sealing, British Hydromechanics
Research Association, Cranfield, Bedford, England (1969).

7. Ludwig, L. P., and Johnson, R. L., "Design Study of Shaft Face Seal
with Self-Acting Lift Augmentation. Ill - Mechanical Components,"
NASA TN D-6161* (1971).
8. Brovn, P. F., "A Glossary of Seal Terms," Special Publication SP-1,
American Society of Lubrication Engineers.

9. Ludwig, L. P., Zuk, J., and Johnson, R. L., "Design Study of Shaft Face
Seal with Self-Acting Lift Augmentation. IV - Force Balance," NASA
TN D-6568 (1972).
 18
10. Zuk, J., Ludwig, P., and Johnson, R. L., "Quasi-One-Dimensional Com-
pressible Flow Across Face Seals and Narrow Slots. I - Analysis,"
NASA TN D-6668 (1972).

11. Zuk, J., and Ludwig, L. P., "Investigation of Isothermal, Compressible

Flow Across a Rotating Sealing Dam. I - Analysis," NASA TN D-53^
(1969).
12. Zuk, J., and Smith, P. J., "Computer Program for Viscous Isothermal
Compressible Flow Across a Sealing Dam with Small Tilt Angle,"
NASA TN D-5373 (1969).
13. Johnson, R. L., and Ludwig, L. P., "Shaft Face Seal with Self-Acting
Lift Augmentation for Advanced Gas Turbine Engines," NASA TN D-5170
(1969).
lU. Zuk, J., Ludwig, L. P., and Johnson, R. L., "Design Study of Shaft
Face Seal with Self-Acting Lift Augmentation. I - Self-Acting Pad
Geometry," NASA TN D-571^ (1970)..
15. Hady, W. F., and Ludwig, L. P., "Experimental Investigation of Self-
Acting-Lift-Pad Characteristics for Main-Shaft Seal Applications,"
NASA TN D-638U (1971).

16. Colsher,..R-., .and Shapiro:, .W.,.,"Predicting Performance of Gas-

Lubricated Seals Using Advanced Numerical Techniques," Franklin
Inst. FIRL-I-C2429-1*, AD-7^5333 (1972).
17. Colsher, R., and Shapiro, W., "Steady State and Dynamic Performance
of Gas-Lubricated Seals," NASA CR-121093 in process.
 TABLE I. - DESIGN POINTS

Design point Sealed pressure, Sealed gas Seal sliding

P temperature speed
l
2
N/cm abs psia K °F m/sec ft/sec

1- Idle 45 65 311 100 61 200

2- Cruise 148 215 700 800 153 500
3- Takeoff 217 315 977 1300 137 450
4 - Climb 148 215 811 1000 122 400

TABLE II. - CLOSING FORCE

Design point Pneumatic Spring force, Total closing

F
s force,
Sealed pressure, Pressure Sealed-pressure
F
P change, closing force, t = Fp + Fs
l
AP F
P
2 2
N/cm abs psia N/cm psi N Ibf N Ibf N Ibf

1- Idle 45 65 34.5 50 160.6 36.1 71 .2 16 231.7 52.1

2 - Cruise 148 215 138 200 641.0 144.1 711.1 160.1
3- Takeoff 217 315 207 300 963.4 216.6 1034.6 232.6
4 - Climb 148 215 138 200 641.0 144.1 i t
711.1 160.1
 TABLE IE. - EQUILIBRIUM FILM

THICKNESS AND GAS LEAKAGE

THROUGH PRIMARY SEAL

Design point Equilibrium Gas leakage

film
thickness

cm in. scmm scfm

1- Idle 0.0004 0.00018 0.008 0.3

2- Cruise .0011 . 00044 .27 9.5
3- Takeoff .0012 . 00047 .39 14.0
4- Climb .0010 . 00040 .18 6.4

TABLE IV. - TEST CONDITIONS AND TYPICAL LEAKAGE RATES

Time Sliding Speed Seal Press Air Temp Air Leakage
Total
(Mrs.) Hrs. (N/cm 2 ) (psi) ( K) (°F) (scms x 103) (scfm)

.5 .5 122 400 207 300 542 517 16.0 34.0

.5 1.0 122 400 275 400 616 650 19.8 42.0
.5 1.5 137 450 275 400 683 770 20.5 43.5
.5 2.0 153 500 275 400 709 816 20.5 43.5
1.0 3.0 168 550 275 400 742 876 20.3 43.0
175 4.5 153 500 275 400 714 825 20.1 42.5
.5 5.0 168 550 275 400 666 740 21.7 46.0
.25 5.25 168 550 310 450 639 690 24.8 52.5
.25 5.50 183 600 310 450 633 680 27.1 57.5
 PRIMARY RING
(NONROTATING)? OUTSIDE
RfVTATIMft
i\\J in 1 1 t*O
i DIAM.

SEAT

2-PHASE
REGION
1

-r
r\i\ r 1 1 nfl
/ UIL rlUVl
\THICKNESS

\i
/ SINGLE
PHASE INSIDE
/ REGION DIAM.
478 505 533 K
'400 450 500 Op
O (a) SCHEMATIC OF OIL FILM (b) PRIMARY SEAL RADIAL TEMPERATURE
^
r- IN PRIMARY SEAL PROFILE (DATA FROM REF. 4).

W Figure 1. - Conventional oil lubricated radial face seal; seal siding speed,
11.3 meters per second (37 ft/sec).

PRESSURE PROFILE IN
PRESSURE PRIMARY SEAL DUE TO
THERMAL DISTORTION

PROJECTION A-A

Figure 2. - Circular grooves in face of primary ring

(from ref. 6).
 ,-PRIMARY SEAL
/^SELF-ACTING GEOMETRY
-' -' -STATIC SEAL
STATIC SEALING SURFACES
-CARRIER,
PRIMARY RING:
SECONDARY
SEALING
DIAMETERS
-CARRIER.
SECONDARY ^_ "^-PRIMARY
RING P SEAL
1
SECONDARY SURFACES
SEALING
FACES (a) PRIMARY SEAL LOCATION. Pj> Pg.
SECONDARY
RING
(a) NOMENCLATURE

PRIMARY SEA
PRIMARY SEAL FACE r 20 SELF-ACTING LI FT PADS SURFACES -^
' EQUALLY SPACED

DIMENSIONS (b) EXAGGERATEn VIEW OF PRIMARY SEAL

IN CM (IN.) Figure 4 - Primary seal.
(b) SELF-ACTING GEOMETRY

Figures. -Self-acting face seal.

 PRIMARY/
RING I fv 2 <(M=l), P 2 -
-1 }—< V2 - (M -1), P2 •
J Lv 2 .(M-l). P2> >'l

-pi<po
v
_l" U.ISENTROPIC
ISENTROPIC V
1 = 0-6UISENTROPIC
EXPANSION

Vn-0

Figure 5. - Mathematical model of primary ring.

r-
i P 0 = 10.3N/CM2 ABS (15 PSIA)

CHOKED FLOW
TAKEOFF
DESIGN POINTS
P! - 217 N/CM2
ABS (315 PSIA)

DESIGNPOINT1:
P,= 45 N/CM2 ABS
(65 PS I A); NON- 16. 84 CM
CHOKED FLOW (6.63 IN.)
DIAM

200 100
PRESSURE, N/CM2 ABS
16.59 CM
(6.53 IN.)
300 200 100 DIAM
"PRESSURE, PSIA~

Figure 6. - Pressure gradient in primary seal, illustrating

choked and nonchoked flow. Parallel face; mean film
thickness hm, 0.0010 centimeter (0.0004in.). (From
ref. 9.)
 SEALED TEMPERATURE, DESIGN
PRESSURE, K (°F) POINT

N/CM2 (PS I A)

977
(215) 700 (800)
(215) 811 (1000)
(65) 311 (100)

16.84CM(6.63IN.)DIAM
1000 16.59CM(6.53IN.)DIAM
3 (TAKEOFF)
200
800

150
2 AND 4(CRUISE +• CLIMB)
600

o 100 s 400-
e

50 200 1(1 OLE)

1 I
2 4 6 8 1 0 1 2 14x10"
MEAN FILM THICKNESS, nm, CM

I 1
1 2 3 4 5x10'"
MEAN FILM THICKNESS, hm, IN.

Figure 7. -Opening force acting on primary sealing face,

fluid, air; sump pressure Pn, 10.3N/cm^abs (ISpsia);
parallel faces.

PRIMARY

i /
GROOVE-J / '-FEED SLOT^
r\E *r '^-RESILIENT RING
^-SHALLOW RECESS PAD
(SELF-ACTING GEOMETRY)
VIEW A-A

Figure 8. - Self-acting seal with pads on primary ring.

 RETAINER RING-* - CARBON RING

Figure 9. - Primary ring assembly.

NO
o

i
W

i i
L
—* SELF-ACTINGPAD
-w-h
(a) SELF-ACTING PAD LOCATION.

- LAND RECESS

>FEED GROOVE
MOTION

0.0025 CM
(0.001 I N . )
VIEWB-B

(b) MATHEMATICAL MODEL OF SELF-ACTING PAD

WITH CURVATURE EFFECTS NEGLECTED

Figure 10. - Self-acting pad.

 900
SLIDING SPEED, TEMPERATURE, DESIGN POINT
M/ SEC (FT/ SEC) K <°F)

— 137 (450) 977 (1300) 3 (TAKEOFF)

180 i— 800
— 153 (500) 700 (800) 2 (CRUISE)
•— 122 (400) 811 (1000) 4 (CLIMB)
— 61 (200) 311 (100) 1(1 OLE)
160 SEALED SLIDING TEMPERATURE, DESIGN POINT
700 PRESSURE, SPEED, K <°F>
M/SEC (FT/SEC)
N/CM2 (PSIA)
140 217 (315) 137 (450) 977 (1300) 3 (TAKEOFF)
600 • 148 (215) 153 (500) 700 (800) 2ICRUISD
148 (215) 122 (400) 811 (1000) 4 (CLIMB)
45 (65) 61 (200) 311 (100) 1(1 OLE)
120
z 2000 F
, 500
LU*
O
Q£
400,
100 S 1600
o t
t£
o = 400
300'
1200
LJ_
CO
I
o
C£
300 B 200 O
800
60
e
100
200 400
40

_L
100 4 6 8 10 12 14x10"
20 MEAN FILM THICKNESS, nm, CM

I I
1 2 3 4 5x10"'
46 8 10 12 14x10 r4 MEAN FILM THICKNESS, hm, IN.
MEAN FILM THICKNESS, hm, CM
Figure 12. - Total opening force - self-acting pad lift
force plus primary seal pneumatic force, (From
I I I j ref. 9.)
1 2 3 4 5x10'"
MEAN FILM THICKNESS, hm, IN.

Figure 1L - Lift force of self-acting geometry. Number

of pads, 20; recess depth, 0.0025 centimeter (0.001
in.); fluid, air; parallel faces. (From ref. 9.)
 PO o
t\R
\ \ \ \ \ \ \ \ "*~:>H>KING KJKCt
SEAT V) y o vi
PRIMARY ^CLOSING FORCE AREA
]RING

Pi-'
3 *->-•
\ X ra 1
1 \SECONDARY-SEALDIAMETER
\ [J D2- 16. 76 CM (6. 60 1 N. )

\ 1 ' lf\ SIDE DIAMETER D, OF

Pi \ PR IMARYSEAL, 16. 59 CM
(6.53 IN.)
'LPNEUMATIC CLOSING FORCE

Figure 13. - Closing forces - spring force and net closing force due
to sealed pressure.

sD
o
OPENING FORCE
CLOSINGFORCE

SLIDING SPEED, TEMPERATURE, DESIGN PO I NT

M/SEC (FT/SEC) K <°F)

137 (450) 977 (1300) 3 (TAKEOFF)

153 (500) 700 (800) 2 (CRUISE)

150- 122 (400) 811 (1000) 4 (CLIMB)

-EQUILIBRIUMOPER-^
100-
BATING FILM THICKNESS"
\
50-
61 (200) 311 (100) 1(1 OLE)

0- 0.
' 2 4 6 8 1 0 1214xlO-«
MEAN FILM THICKNESS, hm, CM

1 2 3 4 5xW4
MEAN FILM THICKNESS,

Figure 14 - Equilibrium gas film thickness as determined by total seal

opening and closing forces. Parallel faces.
 SEALED PRESSURE. TEMPERATURE, DESIGN POINT
PI>
N/CM2
, (PS I A)
K (°F)

217 (315) 977 (1300) 3 (TAKEOFF)

148 (215) 700 (800) 2 (CRUISE)
148 (215) 811 (1000) 4 (CLIMB)
45 (65) 311 (100) 1IIDLE)

.48

.44
15
.40

.36

I <_> ~j
£ 10 — in . a
o
p
.»

.12

.08

.04

0
46 8 10 12 14xlO'4
MEAN FILM THICKNESS, hm, CM

L I I J
1 2 3 4 5x10""
MEAN FILM THICKNESS, hm, IN.
Figure 15. - Leakage rate as function of film
thickness. Fluid, air; sump pressure Pg,
10.3 N/cm2 abs (15 psia); parallel faces.

PRIMARY RING ASSEMBLY^

rf nPf

Figure 16. - Coning displacement of seat, causing non-

parallel faces in primary seal and in self-acting
geometry.
 1 r2
n
min-
Lin PAD \

SPRIMARY SEALING
\_

FACE
PRIMARY RING

0. 63 CM 127 CM
(0.25 IN.) i (0.050 IN.)
"-0.089 CM
(0.035 IN.)

Figure 17. - Two-mi III radian deformation of seal seat

causing nonparallel faces in primary seal.

NONPARALLEL FACES, 2-MILLIRADIAN

vO
o1 FACE DEFORMATION
-3 PARALLEL FACES
t-
i
W
200

160

QL 1 MD
2 t
t 20
-1 80

40

I I I I J_ J
8 10 12 14 16 18 20x10,-4
MEAN FILM THICKNESS OF SELF-ACTING PAD. hm, CM
I
3 4 5 6 7x10,-4
MEAN FILM THICKNESS OF SELF-
ACTING PAD, hm, IN.
Figure 18. - Lift force of self-acting geometry.
Number of pads, 20; pad depth, 0.0025
centimeter (0.001 in.); fluid, air. Design
point 2; sealed pressure, 148N/cm 2 abs
(215psia); sliding speed, 153 meters per
second (500 ft/sec); fluid temperature,
700 K (800° F).
 -- CLOSING FORCE
— OPENING FORCE FOR PARALLEL FACES
— OPENING FORCE FOR NONPARALLEL FACES
2-MILLI RADIAN DEFORMATION

800r

600
QO
1
- . 100 o
o a: 400
1 50
e ^
200
^
n 0 i i i i i l i i
- 4 6 8 10 12 14 16 18xlO"4
MEAN FILM THICKNESS OF PRIMARY SEAL, hm. CM

1 2 3 4 5 6 7x10 -4
MEAN FlLM THICKNESS OF PRIMARY
SEAL, hm, IN.

Figure 19. - Sealed pressure forces acting on

primary ring assembly. Sealed fluid, air.
Design point 2: sealed pressure, 148N/cm2
(215 psia); fluid temperature, 700 K (800°F).
(From ref. 9.)

1000

200 900

_ _,800 ^CLOSING FORCE

u_ 175 /EQUILIBRIUM OPERATING FILM THICKNESS
QO

i150
o*700 """• —~_^___^ f TOTAL OPENING FORCE
- OC.

2
125
500
100
400 i l i i i l
5 16 17 18 19 20 21xlO'a
MEAN FILM THICKNESS OF PRIMARY SEAL, hmm, CM
I | | i i
6.0 6.5 7.0 7.5 aOxlO"4
MEAN FILM THICKNESS OF PRIMARY SEAL, hm, IN.

Figure 20. - Equilibrium gas film thickness as determined

by total opening and closing forces for 2-milliradian face
deformation. Design point 2: sliding speed, 153 meters
per second (500 ft/sec) ; sealed pressure, 148 N/cm 2 abs
(215 psia); sealed gas temperature, 700 K (1800° F).
 Figure 2L - Schematic showing priximity probe location.

sO
o

i
W
T
PRO BE ON SEAL RING
0.0017 CM (0.00067 IN.)

SEAT FACE MOUNT F. I. R.

0.0020CM (0.00085 IN.)

Figure 22. - Oscillograph traces showing response of ring to seat

face runout Recess-pad length to land-length ratio, 2:1;
recess-pad depth, 0.0013 centimeter (0.0005 in.); sliding ve-
locity, 61 meters per second (200 ft/sec); ambient pressure,
10 newtons per square centimeter (14 7 Ib/in. \ room tem-
perature, 300 K (80° F); spring load, 1.13 kilograms (2.50 Ib).
(From ref. 15.)
 t—MAXIMUM
GAP

1
2xRPM

— MINIMUM
GAP

y COORDINATE
FIXED IN SPACE

Figure 23. - Response of nose to seat face runout.

UNSTABLE

SEAT FACE RUNOUT

Figure 24. - Typical stability map

of primary ring response to
seat face runout.
 <
o 50
ce
g 30
o io
;
o
5 -10
§ -30
X
_J
55 -50
.50 LOO L50 2.00 2.50 3.00 3.50 400 450
SHAFT REVOLUTIONS

Figure 25. - Calculated rotation response of the primary

ring; primary ring assembly weight, 0.78 kilogram
(1.75 Ibw); sealed gas pressure, 207 newtons per square
centimeter (300 psi); sealed gas temperature, 811 K
(1000° F); sliding speed, 152 meters per second (500 ft/
sec); initial film thickness, 0.00081 centimeter
(0.00032 in.); seat face runout 0.00635 centimeter
(0.0025 in.) full indicator reading.
NO
O
-^
r-
i
w
300r
64
u_
00
56
"I
uj~
o
g
U_
48

I 40
U_
o
3 32
LJ_

24
"0 .50 1.00 1.50 ZOO Z50 3.00 3.50 400 450
SHAFT REVOLUTIONS

Figure 26. - Calculated fluid film force of the self-

acting geometry; primary ring assembly weight,
0.78 kilogram (1.75 Ibw); sealed gas pressure,
207 newtons per square centimeter (300 psi);
sealed gas temperature, 811 K (1000° F); seal
sliding speed, 152 meters per second (500 ft/sec);
initial film thickness, 0.00081 centimeter (0.00032
in.); seat face runout 0.00635 centimeter (0.0025
in.) full indicator reading.
 .9x10
15x10

.50 1.00 1.50 2.00 Z 50 3.00 3.50 4 00 4 50

SHAFT REVOLUTIONS

Figure 27. - Calculated translation motion of the

primary ring assembly; primary ring assembly
weight, 0.78 kilograms (1. 75 lbw); seal gas
pressure, 207 newtons per square centimeter
(300 psi); sealed gas temperature, 811 K (1000° F);
seal sliding speed, 152 meters per second
(500 ft/sec); initial film thickness, 0.00081
centimeter (0.00032 in.); seat face runout,
0.00635 centimeter (0.0025 in.) full indicator
reading.

-COMPRESSOR
END SEAL

COMPRESSOR
DISCHARGE
PRESSURE

V
RING SEAL ^CIRCUMFERENTIAL SEAL \ ^-FACE SEAL
CFRONT BEARING >- REAR BEARING

Figure 28. - Schematic of engine showing typical seal locations.

 Figure 29. - Single labyrinth seal - early engine or low pressure designs.

xD
O

r-
i

SUMP COMPRESSOR
PRESSURE DISCHARGE
PRESSURE

Figure 30. - Schematic of a conventional radial face main shaft seal.

 SUMP VENT COMPRESSOR
PRESSURE PRESSURE

Figure 31. - Labyrinth seal system schematic.

-COOLING
DUPLEX OIL
THRUST INLET
BEARING 7 /"HOUSING A SEAL ASSEMBLY
a

HOT AIR
BEARING COM- ,' ROLLER BEARING
PARTMENT (SUMP)-'

Figure 32. - Rig used for experimental evaluation of seals.

 AIR INLET

GEAR AIR FILTER -AIR

INDUSTRIAL TRANS-
' HEATER
INTERNAL MISSION ^BREATHER
COMBUSTION (FOUR -BREATHER^
ENGINE-i SPEED)7
'• _L1 1 •rORIFICES
l_r
- 3EA Rd TEST Rl BLEED
BOX
0 IL
JJUILL- . i iT"^
I t 1 t i
1

Q
OIL HEATER
OIL INLET

Figure 33. -Test facilities.

I I I I I I I
8 16 24 32 40 48 56 64 72 80 88 % 104
TIME, HR

Figure 34 - Air leakage self-acting seal during 100-hour

endurance run. Seal sliding speed, 122 meters per
second (400 ft/sec); sealed pressure differential,
138 N/cm 2 (200 psi); sealed gas temperature, 811 K
(1000° F).
Figure 35. - Surface profile traces across self-acting pad and sealing dam before and after
38 hours of operation. Seal sliding speed, 122 to 183 meters per second (400 to 600 ft/sec);
sealed pressure differential. 241 to 345 newtons per square centimeter (350 to 500 psi)-.
sealed gas temperature, 811 K (1000° 0.

VENT He GAS VENT VENT

LI QUID OXYGEN
HYDROGEN
310N/CM 2 ABS
GAS
(450 PS I A)
90 K (-298° F)

FACE CONTACT SEAL RING SEALS RING SEALS

Figure 36. - Schematic of lox turbopump seal system, current practice.

 VENT He GAS VENT VENT

LI QUID OXYGEN
344N/CM 2 ABS HYDROGEN
(500 PS I A) GAS
90 K (-298° F)

HYDRODYNAMIC SELF-ACTING RING SEALS

SEAL
J
Figure 37. - Schematic of lox turbopump seal system under development

i
Is)

Lift pads in
carbon ring

Figure 38. - Circumferential seal self-acting type.

NASA-Lewis