DEPARTMENT OF MECHANICAL

AND AEROSPACE ENGINEERING

OBSERVATIONS ON A VISCOSEAL IN A TRANSPARENT HOUSING
THE PREVENTION OF LEAKAGE AND BREAKDOWN
bY
R. J. G. Boon
A Transl at i on by
J. P. B. Vreeburg
March 1969
Knoxvi l l e, Tennessee 37916
n i s unl i mi t ed.
ME 69-T57-4
OBSERVATIONS ON A VISC3SEAL IN A TRANSPARENT HOUSING-
THE PREVENTION OF LEAKAGE AND BREAKDOWN
R. J . G. BOaN
A n/Ia;ter':7 Thesis, J une 1968
Chemische VVerktuigen
TECd I.!ISC l i E HOGESC iiOOL,
Delft, The Netherlands
Translated by
J. P. B. Vreeburg
Fluid Sealing Research Laboratory
Mechanical and Aero space Engineering Department
University of Tennessee
March 1969
Knoxville, Tennessee 379 16
Table of Contents
Abstract
Conc l u s ion s
L i st of references
L i st of symbols
1.
2.1
2. 2
3.
4.
5.
6.
6.1
6.2
6.3
6.4
6.5
6.6
7.
7. 1
7.2
7.3
I ntroduction
Literature Survey
Synopsis of the observations of the investigators
mentioned in the literature
Purpose of thi s investigation
Method of investigation
Description of the apparatus
General remarks with regard to the research
Definition of the Reynolds number
Reproducibility of vi sual observations
Non-exact indication of numbers of revolution a2d
Reynolds numbers
The influence of the vi scosi ty of the seal i ng fluid on gas
inclu s ion
L eak due to loss J f sealing fluid
The defining of the conceptions:
inclusion and l eak
Observati oos, measurements and concl usi ons with regard to
gas inclusion and l eak Qf a si ngl e V. S. S., seal i ng agai nst
a chamber of constant volume
The influence of the volume of the liquid on the observed
phenomena
Observations with seal i ng fluid no. 1; Tellus 13
Gas inclusion observations with seal i ng fluid no. 3
gas ingestion; gas
Page
v
V
vi
vi i
1
3
6
6
7
7
8
8
8
10
10
12
12
15
15
15
16
ii
7.4
7.5
8.
9.
9.1
9.2
9.3
9.4
9.5
9.6
10.
10.1
10.2
11.
11.1
11.2
12.
12.1
Determination of the rate of the gas l eak
Concl usi ons from the observati ons on vi scoshaft seal no. 1
Observati ons and concl usi ons with regard to gas inclusion
and l eak of a si ngl e V. S. S. as a seal agai nst a chamber
of constant pressure
Prevention of l eak of a simple V. S. S.
Prevention of l eakage by installing a gas barrier
Prevention of l eakage by separating and returning the
encl osed gas (Gas recircuiation)
Synopsis of the resul ts of 9.1 and 9. 2
Prevention of l eakage by a part of the seal over which no
axial transport is possi bl e
Prevention af gas inclusion and l eakage by changing the
surface energy of the shaft
Prevention of gas inclusion and l eakage by making the
grooves of the seal deeper
I ntroduction to the investigation for the prevantion of
l eakage and break down of double visco shaft seals
The significance of the observed measured gas l eak
The appearance of l eakage at the double V. S. S. in the
Laboratorium for Chcmical Equipment
Prevention of l eakage of a double V. S. S. by recirculating
the ingested gas and/or the i nstal l ati on of a gas barrier
The manipulation of the l eak using the center bushing
The prevention of l eakage by separating and recirculating
the encl osed gas
The influence of the geometry of the seal on the phenomena
of gas ingestion and l eak
Observations on the double V. S. S. no. 3 (single groove)
20
23
26
27
28
32
35
38
39
40
41
44
45
45
47
47
50
50
iii
12.2
12.3
12.4
12.5
12.6
12. 7
12.8
12.9
13.
13.1
13.2
13.3
13.4
13. 5
14.
15.
16.
Observati ons on the double V. S. S.
Observati ons on the double V. S. S. no. 4 (three grooves)
Comparison of the double V.S.S.'s nos. 3, 4, and 5 that
differ in the number of grooves and the landwidth
Comparison of the double V.S.S.'s nos. 4, 8, and 10
that differ in the depth of the groove
Observati ons on and comparison of the double V. S. S. s nos.
no. 5 (four grooves) 53
56
57
58
6 and 7
Observations on and comparison of the double V. S. S. ' ::
nos. 11, 13 ana 15
Observati ons on and comparison of the shafts nos. 12,
14, and 16
Compariscn of the geometries of shaft seal s with in-
creasi ng helix angl es
The influence of foaming on the seal i ng parameter
Theoretical val ues for the seal i ng parameter
The measurement of <he sealitlg parameter
I naccuraci es due to temperature influences in the
determination of the seal i ng parameter
Resul ts ef measurements
Explanation of break down
Thz influence of the recirculation tystem on the sealing
parameter
Comparison of theoreti cal and practi cal val ues of the
(foam free) seal i ng parameter
Check of the resul ts of measurements on the double
V. S. S. apparatus in the Lab. for Cl-axical Inztrument::
62
66
70
70
72
72
73
75
75
76
77
78
78
82
89
Tables
Figures
iv
Abstract
A solution i s sought for the l eakage and break down phenomena as
encountered by many researchers by means of visual observations on a
vi scoshaft seal in a transparent housing.
Our observations are compared with the phenomena mentioned in
the literature.
An effort is made, notwithstanding gas inclusion, to prevent l eakage
in vi scoshaft seals.
We tried to reduce gas inclusion by changing the surface energy of
the seal.
The influence of the geometry of the seal on gas inclusion and leakage
has been studied.
The influence of gas inclusion on the val ue of the sealing parameter
has been looked into.
The measurements of the double V. S. S. in the Lab. of Chem. Equip.
have been checked because it could be expected that gas inclusion had
affected the measured val ues.
Conclusions
Our observations with regard to the gas inclusion mechanism check with
those of Stai.r, Hal e and Ludwig.
Leakage may be prevented by the installation on the shaft of bushings
that have a smal l radial cl earance and by the installation of a recirculation
system for enclosed gas.
A teflon coating on the V. S. S. does not influence the severity of gas
inclusion.
Viscoshaft seals with a large helix angl e, multiple groove, land widths
that are not too smal l and large groove depths, do not occasi on gas inclusion.
Due to gas inclusion, the val ue of the sealing parameter decreases. Thks
drop i ncreases with the number of revolutions. We found a maximum drop of
34 percent.
The dupl i cate measurements on the V. S. S. test apparatus in the Lab. for
Chem. Equip. were in very cl ose agreement with earl i er measurements.
V
References
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
E. F. Boon and S. E. Tal, "Hydrodynamische Dichtung fur rotierende
Wellen" Chem. Ing. Technik, vol. 31 no. 3, J anuary 31, 1959.
J. G, Pape and W. J. Vrakking, "Visco Seal-Pressure Generation and
Friction Loss Under Turbulent Conditions": Trans. WSLE, 11 (1968),
310 - 320.
W. J. Vrakking, "Visco Afdichting, theorie en experiment bi j turbulente
stromingscondities, I ' Rapport Chemische Werktuigen, J uly, 1967.
W. Frossel , I ' Untersuchung von Gewindewellendichtungen, I' Konstruktion,
18 Heft 4, 152 (1966).
Husenbl as and K. H. Switala, "Versuche an Oldichtgewinden bei hoheren
Wellenumfangsgeschwindigkeiten, 'I Konstruktion 14 Heft 5, 197 (1962) e
J . M. McGrew and J . D. McHugh, "Analysis and Test of the Screw Seal
in Laminar and Turbulent Operation, I' J ournal of Basic Engineering,
March 153 (1965).
A. E. King, "Screw Type Shaft Seals for Potassium Lubricated Generators,"
Supplement to I. E. E, E. Transactions on Aerospace, J une (1965).
W. K. Stair and R. H. Hale, "The Turbulent Visco-Seal-Theory and
Experiment I I' Third I nternational Conference on Fluid Sealing
England, April (1967) paper H 2.
Cambridge
J. M. McGrew and A. J . Orsino, "Non Contacting Dynamic Seal s for Space
Power Alternators Bearing and Seal Desi gn in Nuclear Power Machinery, ' I
ASME, Edited by R. A. Burton, New York (1967).
L. P. Ludwig, .T. N. Strom and G. P. Allen, "Gas I ngestion and Sealing
Capaci ty of Hel i cal Groove Fluid Fi l mseal (visco seal) using Sodium
and Water as Sealed Fl ui ds," NASA Report TN-3348, March 1966,
Washington, D. C.
D. van der Meulen, "Onderzoek aan een visco-afdichting, 'I 5e jaars rapport
nr. 2-67, Lab. for Chem. Equip., Delft University of
vi
L i st of symbols
C
cE
d
h
1
M
N
P
Re
Re 1
t
U
a
b
cl earance between land and housing
effective friction coefficient
shaft diameter (m)
groove depth =height of land (m)
active length of the vi sco shaft (m)
momen t (Nm)
number of revolutions per minute
pressure (N/m )
Reynolds number connected with the groove =
Reynolds number over the land =
tangent of the helix angl e
circumferential vel oci ty od/2 (m/sec)
groove width (m)
land width (m)
2
@UC/V
Uc/V
helix angl e (degrees)
(h +c)/c
Y a/(a +b)
2
dynamic vi scosi ty (Nsec/m )
v kinematic vi scosi ty (m / s ec)
@ seal i ng parameter
2
3.
P densi ty (kg/m )
W angular velocity, 23t.N/60 =(l/sec)
vi i
1
1. I ntroduction
The most important requirements for a shaft seal are: little wear and
a l eakage that is as smal l as possi bl e.
O-rings and labyrinth seals do all exhibit some leakage. During the last
years, the visco shaft seal, in the following abbreviated as V. S. S., has
become more prominent.
smal l and a long period of operation assured, moreover theoretically the
l eakage i s vzry smal l .
Stuffing boxes, mechanical seals,
This seal is contactl ess so that the wear is
The V. S. S. resembl es a screwpump with zero capaci ty. The screw
profile may bz placed in either the shaft or the housing.
Only in the last few years have cal cul ati ons of the operation o
This seal devel ops, in the space between V. S. S o' s been performed.
shaft and housing, an axial pressure gradient which can be expressed
in a seal i ng parameter @ .
I n 1959, Boon and Tal (I ) published their derivation for thi s parameter
under laminar flow conditions for which
geometry of the seal under consideration;
is dependent only on the
1
For turbulent flow conditions $ al so appears to be dependent on the
Reynolds number.
for thi s case.
Pape and Vrakking (2) computed the sealing parameter
I n the last few years, many investigators, also in the Lab. for Chem.
Equip., compared the theory with experiments.
has been found (e.g. see 3).
With the gi ven formulae, the machine desi gner can determine a desi gn.
In general a good agreement
In general, the shaft diameter, angular velocity and pressure seal ed agai nst
will be the given quanti ti es.
The minimal radial cl earance between shaft and housing, c, is restri cted
by manufacturing feasabi l i ty. The sealinq pressure has to be real i zed with
2
the geometry of the screw profile, the length of the shaft and the vi scosi ty
of the seal i ng liquid.
The V. S. S. can, for example, be used i n the following cases:
---
in center of V. S. S.
of environment
A B
A.
This one may be used as a seal between a gas and a liquid-filled space.
The outwardly escapi ng liquid is pumped back by the screw.
B.
This one may be used as a seal between two gas-filled spaces. Using the
seal i ng liquid, the seal generates a pressure at the center of the shaft. A s
long as the difference i n pressure between the gases is smaller than the
maxi mal pressure generated by the shaft, no leakage wi l l occur.
The single V. S. S;
The double V. S. S.
3
Especially thi s last case offers interesting possi bi l i ti es for appl i cati ons.
During pumping or compression of p3isonous, flammable or expl osi ve gases,
hence i n general with substances that are not to be in contact with the out-
si de ai r, the neglectable l eak is a great advantage. This l eak is possi bl e
by diffusion of the gases that are to be seal ed through the seal i ng liquid.
I ndeed, at. low circumferential val oci ti es, visco shaft seals give
excel l ent performance. However, at higher vel oci ti es, several investi-
gators report a strong l eakage of the V. S. S. and a decrease or ceasi ng of
the pressure generation, indicated by "break down. I' Due to these phenomena
the V. S. S. cannot longer be used as a seal.
To get an impression of the leak aqd break dow!i phenomena, a literature
survey has been carried out.
failure of the V. S. S. are quoted.
Only those data that are connected to the
2. 1 Literature Survey
Frossel (4) i nvesti gated double V. S. S. ' s . At high circumferential
vel oci ti es there was a decrease in sealing pressure for all seals con-
si dered.
accompanied by foam generation in the sealing liquid.
thi s phenomenon increased at still higher vel oci ti es and Sinally resui ted
in a complete ceasi ng of the pressure generation,
f oam di mi ni shes the pressure generation in the seal due to i ts very low
vi scosi ty.
Through a si ght gl ass in the housing, he observed that thi s was
The i ntenstty of
He postul ates that the
Hausenbl as (5) investigated double V. S. S . ' s with a grooved housing.
He reports: at high circumferential vel oci ti es, the seal i ng gressure de-
creases and finally ceases to exist.
been formed around the smooth shaft due to the centrifugal force.
the oil in the grooves was not set in motion hence no pressure could be
generated. The oi l , supplied in the center of the seal, l eaks towards the
low pressure ends i f there is a smal l pressure difference.
oi l contai ns gas bubbles. .
He supposed that an air cushi on had
Hereby,
The departing
4
McGrew and McHugh (61, usi ng a simple V. S. S. i n vertical position,
found that a seal ed pressurized vessel filled with liquid showed a de-
crease in pressure at higher seal velocities. Furthermore, they observed
that thi s phenomenon was accompanied by oi l l eakage from the lower end
of the shaft (which was exposed to ambient atmosphere). They postulated
that the Re number al one could not gi ve a sati sfactory explanation because
l eakage appeared in the laminar, turbulent and transi ti on regimes.
supposed that the Weber number =U cp/a
They
2
influenced the l eakage of
liquid especi al l y because i t contai ns the surface tensi on.
do riot excl ude other influences.
However they
King (7) used a verti cal double V. S. S. in which seal i ng liquid was
recirculated at the upper and of the system. He fcund a so cal l ed
''secondary l eak" by which liquid l eaks down out of the oil/air interface.
This l eak depends on the number of revolutions and the cl earance, c, and
does also appear when there is no pressure gradient.
characteri sti c numbers of Weber, Froude and Reynolds are influential.
During experiments with liquid Potassium as the liquid and Argon ac the
gas, he observed that gas was pumped into the liquid through the seal.
He supposes that the
Stair and Hale (8) tested a simple V.S. S. that rotated in a ha-izontal
position. The strong decrease of the seal i ng parameter at high Re number
for a shallow grooved seal was attributed to encl osed gas. This partially
fills the grooves resulting in a decrease of their acti ve depths. Extra
barriers with smal l radi al cl earances at the shaft ends gi ve a slightly
better performance.
--The Reynolds number for which gas inclusion appears is dependent on
The following concl usi ons are made:
the geometry of the V. S. S.
--Also in the turbuler,t regime a stabl e $J may appear.
--At high Re number, the pressure generation al ong the shaft fl uctuates.
In thi s case they di sti ngui sh three situations:
1. Originally there are smal l pressure vdriations followed by a strong
decrease in pressure. Hereafter the seal may generate a constant
seal i ng pressure also at high Re numbers.
5
2. There are sl ow pressure fluctuations accompanied by a greater
wetted length. At higher velocities, amplitude and frequency
increase. A pulsating l eakage of liquid appears when the necessary
wetted iength i s greater than the real length.
3 . Fast pressure fl uctuati ons with increasing amplitude and pulsating
l eakage of liquid.
smal l vessel filled with liquid.
This appears especi al l y at the seal i ng of a
Finally is summarized that gas inclusion i s Dpposed by large values of
B and Y and smal l val ues of a .
The investigations of McGrew and Orsino (9) indicated that the stabi l i ty
of the interface was not improved by a smal l helix angl e.
the double V. S. S. wds grooved. Ddring testing under atmospheric pres-
sure no observabie oil l eakage appeared; after imp3sing a smal l pressure
difference the low pressure end l eaked.
as a possi bl e cause.
The shaft of
Gas inclusion was suggested
Ludwig (10) investigated a V. S. S. in a transparent housing. He ob-
servsd that l eakage of liquid and breakdown as described by McGrew
and Mc:Hugh can be caused by gas inclusion in a cl osed vessel .
a seal with a groovzd shaft, he found that over a Re
al ways appeared.
Using
of 2300 gas inclusion
g
When the housing was grooved, he found gas inclusion in a low R e
region. Over Re =8400 there was no more gas inclusion, however, he
found that gas was sti l l ingested a: the immediate oil-gas interface.
g
Furthermore he observed that due to gas inclusion the seal i ng parameter
(J
decreases: l eakage of oil did not take pl ace. Also, with hi s vi sual
observati ons he is abl e to gi ve a gaod explanation for the pulsating l eak
and seal break down as obaervzd by Stair and Hal e.
van der Meulen (11) made observati ons on the double V. S. S. of the
Lab. for Chem. Equip. SeaJ ing two gas filled spaces between which there
was a pressure difference, he found ths onset of l eakage at low pressures.
This phenomenon is transpDsed in a strongly fluctuating l eak at a constant
6
2.2
seal i ng pressure. At a pressure difference of 30% of the seal i ng pressure,
the pressure build-up ceased after a few minutes. After that there was a
strongly changing gas transport through the seal from the high pressure end
to the low pressure end. This l eak i ncreases with the pressure difference.
These phenomena are attributed to gas inclusion from the oil-air interface.
He supposes that the important factors are: the geometry of the V. S. S., the
circumferential velocity, the surface tensi on, and the vi scosi ty of the
sealing liquid.
Synopsis of the observati ons of the investigators mentioned in the literature
For a B. S. S. with a groaved shaft, the seal i ng pressure decreases for
high circumferential vel oci ti es.
of oi l .
of new seal i ng fluid in the center of the double V. S. S.
out thi s supply an open connection betwzen the gas filled spaces wi l l
develop.
In thi s situation there often is a l eakage
The pressure generation has to be maintained by a continuous supply
I t i s cl ear that with-
Because the leaked oi l contai ns gas bubbles or i s in the form of
f oam, it may be assumed that by means of thi s foam a connection between
the tw3 gas-filled spaces is establ i shed.
inclusion may be the cause of the leak and that the seal i ng parameter de-
creases due to the formation of f oam in the seal i ng liquid.
drop s the pressure generation may be caused by thi s.
I t has been observed that gas
Even a complete
One investigator found that gas inclusion did not appear anymore for a
seal with a grooved housing at high Re numbers.
shaft, nor a grooved housing did l eakage of seal i ng fluid from the seal take
pl ace.
Neither with a grooved
3. Purpose of thi s investiqation
The purpose of thi s investigation is the prevention of leak and break down
of vi scoshaft seals (V. S. S. ).
7
4. Method of investigation
I n order to attai n the aforementioned purpose we decided upon a
vi sual investigation of the phenomena in a V. S. S.
survey it became cl ear that gas ingestion causes l eaks and break down.
It is expected that there wi l l al ways be gas inclusion at very high cir-
cumferential vel oci ti es (high numbers of revolutions).
From the literature
Therefore we
i nvesti gate if, notwithstanding gas inclusion and maintaining the shaft
geometry, l eak and break down may be prevented by changes in construction.
Also we i nvesti gate i f by changing the geometry of the V. S. S. the
gas inclusion di mi ni shes whereby the goal may be attained al so.
Besides these vi sual observati ons we checked the influence on the
seal i ng parameter by foam formation in the seal i ng fluid.
down may be caused al so by a decrease of the seal i ng parameter.
I ndeed, break
5. Description of the apparatus. See Fig. 1
The essenti al part of the construction is a gl ass tube that has an
inner diameter which, wi thi qvery close tol erances, is the same over
its enti re length. In thi s tube the vi sco shafts may turn verti cal l y in
bearings. The principle of a vi sco shaft is indicated in Fig. 2 and a
summary of the geometries of the investigated shafts is l i sted in Table 1.
A motor powers the vi sco shaft by means of a V-belt. With the use of a
number of pul l eys el even different numbers of revolutions can be obtained.
Table 3 i ndi cates the possi bi l i ti es and al so the Re
obtained with the seal i ng fluids used.
properties are summarized in Table 2.
l ands of the shafts and the gl ass tube
fluid may be supplied from a reservoir at the si de or through the opening
at the upper end of the shaft formed by the upper bearing. With the si ngl e
V. S. S., the turn direction of the shaft is such that there is a pressure
generation towards the lower end.
manometer.
numbers that can be
1
The fl ui ds and their physi cal
The radi al cl earance between the
is 0.10 mm for all cases. Sealing
The pressure can be read from the
The double V. S. S. generates the hi ghest pressure in the
8
middle of the seal. With the use of the mounting in Fig. 3, a pressure
difference can be imposed over the seal. To thi s end the mountings de-
picted in Fig. l and Fig. 3 are connected at the poi ntsb. For a certai n
imposed pressure difference we can measure an eventual gas l eak with
the U-tube. Then the val ve2 will be cl osed when the system has been
pressurized. A displacement of the liquid l evel in the U-tube corresponds
then to a certai n leakage.
revolving vi sco shaft permits vi sual observations of the motionless picture.
Using a stroboscopic illumination of the
6. General remarks with regard to the research
Already during the first experiments we confirmed suppositions of
other investigators, that the cause of leakage and break down is gas
inclusion in the sealing fl ui d. See pictures 37-21 and 37-25. Before
examining the resul ts of thi s experiment cl oser, the following remarks
are made.
6.1 Dafinition of the Reynolds number
For the computation of the sealing parameter one has to use a laminar
or turbulent flow theory. I t appears that the border between these two
regions can be indicated cl earl y with the Reynolds number Re .
Thi s Re number is thus related to the cl earance between the l ands and the
housing (c in Fig. 2). I t appears from thi s investigation that thi s number
Re
depth i s influential on the gas inclusion, it is better to rel ate the Reynolds
=Uc/v
1
i s not significant with regard to gas inclusion. Because the groove
1
number to the groove depth. Re =R el . p ,
was not sui tabl e as a criteria for gas inclusion ocurred, we used the
definition of Re
resul ts from the literature.
. Because thi s definition
g
In thi s way it i s easi er to compare the resul ts with
1'
6 e 2 Reproducibility of vi sual observations
I t appeared several times that, under the same conditions, different
experiments yielded different pictures. This could be observed especi al l y
9
37-21 37-25
N =1802 rpm
Rel =151
The cause of leak and breakdown appeared to be gas
inclusion in the sealing fluid
(Shaft no. 2 single grooved)
N =2590 rpm
Re =362
1
10
with the double V. S. S. It even happened that the picture changed
during one experiment. The photographs 36-9 and 36- 11 exemplify
thi s. Without regularity these pi ctures changed into each other via
many transitions. With the double V. S. S. ' s it appeared that generally
the gas inclusion at the lower end of the shaft was less prominent than
at the upper end. However, thi s is no rule, pictures 33-12 and 37-17,
show that under the same conditions sometimes gas inclusion does
occur and sometimes not. The pi ctures are not reproducible satisfactorily.
6 3 Non-exact indication of numbers of revolution and Reynolds numbers
In connection with the earl i er mentioned nonreproducibility of pictures,
it i s cl ear that no number of revolution of Reynolds number can be indicated
for which a certai n phenomenon occurs or not. However we often can sti l l
gi ve some indication. "At low Reynold sl' implies that a certai n phenomenon
has been observed mainly in that region. The chance that during many
experiments, it will not be observed in that region is smal l j ust l i ke the
chance that it will be observed for high Reynolds numbers.
exact indication is possi bl e and thi s will be indicated cl earl y.
Sometimes an
6.4 The influence of the vi scosi ty of the seal i ng fluid on gas inclusion
In the beginning of thi s investigation it became apparent that the seal i ng
fluids no.'s 1 and 2 (Table 2) did not cause gas ingestion or gas inclusion.
A seal i ng fluid with a lower vi scosi ty than those of the no. ' s 1 and 2 did
exhibit gas inclusion in the same seal. Also based on the resul ts of other
investigators (lit. survey) it is expected that, for sufficiently high cir-
cumferential vel oci ti es, gas inclusion will al so occur with highly vi scous
sealing fluids. This is why the influence of the vi scosi ty has not been
investigated and we did only use fluid no. 3 to study the gas inclusion
phenomenon.
the seal.
I t is cl earl y indicated when other fluids have been used in
36- 9 36- 11
Different pictures of gas inclusion for equal Re-numbers.
pictures can not be reproduced sati sfactori l y.
test run these pictures may change into each other.
The
During the same
3 3- 12 37-17
Different picture at the lower end of the shaft. Often there i s no
gas inclusion on the lower end of the shaft but sometimes thsre is
under the same conditions. I rreproducible situation.
12
6.5 Leak due to loss of seal i nq fluid
I t appeared from the literature survey that by mentioning l eak one means
a loss of fluid possi bl y mixed with gas bubbles. The cause of thi s fluid l eak
is attributed to gas inclusion which gi ves ri se to an unstabl e interface.
I n our investigation such oil l eakage was observed at some double
V. S. S. I s . Due to thi s downward leaking, the seal i ng fluid di sappears
from the seal; if no new supply occurs thi s wi l l resul t in a decrease and
finally a complete drop in the pressure generation. This then resul ts in
a gas l eak through the seal.
The observed oi l l eakage (picture 36-6), which was al ways accompanied
by gas ingestion, is primarily seen as a l oss of seal i ng fluid. A verti cal l y
positioned seal appears to be necessary for thi s loss to occur. Leak due
to gas inc1u:;ion and l eak due to loss of seal i ng fluid were observed as
two particular phenomena.
6.6 The defining of the cEcepti ons: gas ingestion; gaancl usi on and leak
1)
In the course of thi s investigation it appeared to be useful to depart
from general nomenclature and to use more refined definitions.
A. Gas ingestion
Gas ingestion is defined as an instability at the interface between seal i ng
fluid and gas by which a zone occurs containing foam and/or bubbles.
(See picture 38-40).
B. Gas inclusion
This phenomenon.means that a quantity of gas in the form of an emulsiop,
form or bubbles fills the acti ve length of the seal for some di stance from the
high pressure end. Almost al ways gas inclusion occurs by transport of the
gas trapped by gas ingestion. (Pictures 44-39 and 44-40). The phenomenon
may appear al so by inclusion of a quantity of gas without the occurrence
of gas ingestion (picture 36-13).
C. L eak
This means for the si ngl e V. S. S. that axial transport of gas occurs from
1) Translator's note: In current literature, gas ingestion i s usual l y used for
both gas ingestion and gas inclusion as defined by thi s author.
13
36-6
Observable l eak of sealing fluid from the seal
37-25
Double vi sco
shaft gas in-
33-16 gesti on, gas
inclusion and
(visible l eak)
Single vi sco shaft gas ingestion
gas inclusion and l eak
14
38-40
Gas inclusion
36-13
Gas inclusion without gas ingestion
44-39 44-40
Very little gas ingestion in 44-39 leads to gas inclusion and gas
ingestion in 44-40 after 1 112 min.
'
15
7.
7. 1
the oi l /gas i nterface to the space filled with fluid that has to be
seal ed (picture 37-25).
For the double V. S. S. axial transport of gas occurs through the seal
from interface to interface. Due to the l eak the seal i ng acti on of
V. S. S. vani shes.
gas inclusion according to the aforementioned definitions.
For the double V.S. S. the l eak al ways is vi si bl e because of the presence
of gas in the central part of the seal (picture 33-16).
may al so be an exchange.
di fference over the seal will not decrease; there is no measurable l eak.
Prior to l eak w2 will have had gas ingestion and
This vi si bl e l eak
In the i atter case an imposed pressure
Observations- measurements and collclusions with regard to gas inclusion-
- and l eak of a si ngl e V A S . , seal i ng agai nst a chamber of constant volume
The shaft used in the seal is shaft no. 1 (Fig. 4) of which the round-
machined shaft ends are due to changes of the original desi gn.
-- The influence of the vol umuf the liquid on_ the observed phemmena
The seal ed liqliid-filled chamber is bordered on the lower si de of the
mounting by the system between the val ves, See Fig. 1. Furthermore, the
total volume is determined by the height of the liquid in the seal.
height can be determined accuratel y during the state of rest, however, the
total oil vol ume does not appear to be fixed.
of gas in the system that i s compressed during operation; hereby the wetted
length decreases and therefore al so the pressure generation.
This
This is caused by the quantity
The oil volume cannot be corrected during the operation, fi rstl y because
then the picture becomes distorted by gas inclusion and secondly because
it is not feasi bl e practically. No degassi ng channel has been i nstal l ed be-
cause the variable quantity of gas primarily influences the observed picture.
The l atter turned out to be hard to reproduce during the enti re investigation.
7.2 Observations with seal i nq fluid no. 1; Tellus 13
Although it is already mentioned :hat no gas inclusion occurs with thi s
seal i ng liquid, some observations have still been made. This to & tai n
16
7.3
insight in the real conditions at a supposedly ideal seal.
For these observations, the speci al form of the oil/air interface was
noticed (Fig. 7). The indicated interfaces are not characteri sti c for one
particular number of revolutions. The si tuati ons overlap, they are not
reproducable sati sfactori l y and during the observation they sometimes
become merged.
I n all cases there is a sharply defined interface of while the l i nes
exhibit sharp corners.
1st. The dry part of the seal.
2nd. A pdrt where a thin oiifilm can be observed. I t is supposed that thi s
oi l is hurled agai nst the glass housing and does not have contact
anymore with the rotating shaft.
The seal coxpl etel y filled with oi l .
Three zones become apparenb
3rd.
For numbers of revolutions larger than 1900 rpm bubble traces originate
from the sharp interface.
not give ri se to gas inclusion.
However these bubbles di ssol ve rapidly and do
Gas inclusion observations with sealing fluid no. 3
Except for the l owest possi bl e number of revolutions, N =1082 rpm,
In all cases the level in the seal gas ingestion occurs at the interface.
goes down strongly immediately after the start.
caused by the ai r present in the cl osed chamber.
The compressibility i s
For N =1082 rpm the stationary state is reached immediately after
starting, pressure and level are constant (picture 8-7) For the highest
number of revolutions possi bl e, N =4120 rpm, the stationary state is only
reached after about 20 minutes (picture 8-33). The picture is reproducable
in both cases j ust as the maximum pressure built up by the seal.
Between these mentioned extremes many observations are made. Exact
reproducibility was not possible; see 7.1.. It is observed that for a smaller
volume oi liquid to be seal ed, the ti me necessary to attai n a stationary
condition i ncreases. (Table 4). For three numbers of revolutions, N =1522,
1910 and 3480 rpm, the progress af the gas incluzion i; reported:
17
8-7
N =1082 rpm
Re =151
Vi&o shaft seal
without gas ingestion
stationary immediately.
8-33
N =4210 rpm
R e =575
V. 8. S. with gas ingestion gas
inclusion and l eak stationary
after 20 min.
18
Gas inclusion and l eak for N =1522 rpm; Rel =212
The apparent volume of oil i n the system in rest was 196 m.m. on the
scale. Obviously there was still much air present in the system because
the compressibility was high.
back to 125 m.m. (picture 8-13).
I mmediately after the start the level dropped
The dependence of the measured seal i ng pressure as a function of
ti me has been recorded graphically in Fig. 8.
of fluid i s indicated as a difference in wetted shaft length (196 m. m. =
82%).
The difference in quanti ti es
Air bubbles are ingested at the interface and transported to the space
below the shaft. Hence l eak occurs due to thi s axial transport of gas.
Apparently the enclosed ai r supersedes the oi l because the liquid level
ri ses (picture 8-16). Concurrent with thi s rising, an i ncrease in pressure
generation takes place that corresponds with the theoretical value
2
P =0.026 kg/cm per cm.
After some ti me no di scernabl e ri se of the level takes place: the fluid then
wets the full acti ve length of the seal. In thi s stationary state the i ntensi ty
of the gas ingestion i s decreased; the pressure build-up and apparently al so
the amount of enclosed gas are constant (picture 8-17).
It i s possi bl e that the round-finished shaft ends are the cause of the
dim in i s hed ga s ingestion.
I t is cl ear that bubbles are transported especi al l y on the low pressure
si des of the l ands .
At an other experiment there was more oi l in the system; the rest level
was at 220 m.m. whereby the shaftlength i s wetted for 100% (see Fig. 8).
After the start the level goes down to 165 m. m.thereafter the level ri ses agai n
until close to the shaft end. The phenomenon is stationary earlier than with
a smal l oi l volume.
Gas inclusion and l eak for N =1910 rpm, Re. =267
1
The pressure change for thi s case is recorded graphically in Fig. 9.
19
8-13 8-16 8-17
Gas inclusion and l eak for N =1522 rpm; R e =212; Fig. 8
4 sec. after starting
7 1/2 min. after starting
gas ingestion and level ri ses because gas ingestion decreases
gas inclusion
when the level coi nci des
with the machined shaft
end. Stationary situation
1
35 sec . after starti ng,
liquid below the V. S. S.
is replaced by gas
20
Analogous to the former case, the level goes down immediately after the
start whereby gas ingestion begi ns (picture 8- 23) . After about 4 minutes
the encl osed ai r had superseded so much oil that the seal i ng fluid level
rose over the activa seal . After that the pressure remained constant
(picture 8-27).
----- I t is cl ear that for thi s case oil l eakage would have started i f the
shaft were rotating in a horizontal position.
by the system in the given verti cal position. ---- -
Now thi s l eakage is prevented
When the full active length of the seal is wetted by the fluid the
The transport of gas to i ntensi ty of the ingestion diminishes strongly.
the space below the shaft decreases.
After about 8 minutes a large quantity escaped from the lower part
This so cal l ed gas ( = high pressure end) in one bubble (picture 8-28).
front, did fill the whole seal with foam ana gave ri se to a sharp pressure
drop (Fig. 9).
escaped. Apparently at the maximum pressure, there has to be a certai n
amount of ai r in the seal ed space to cause "back l eakage" of gas through
the seal towards the low pressure end. .
Hereafter the pressure increased until agai n a gas front
Gas inclusion and leak for N =3480 rpm; R e =486
1--
The dependence of the pressure on time i s recorded graphically in
Fig. 10.
with foams (picture 8-31).
shaft the l evel ri ses very strongly and after about 2 minutes reaches the
upper end of the visco shaft ;(picture 8-33). This is accompanied with a
smal l decrease in pressure after which smal l pressure fluctuations occur
without disturbing the picture.
I mmediately after the start the seal is al ready completely filled
Due to l eakage of gas to the space below the
This is the stationary state.
7.4 Determinationof the rate of the gas l eak
A s became apparent from the previous observati ons the encl osed gas
The volume of the seal that may be filled supersedes a quantity of liquid.
21
8-23
Gas inclusion
Gas ingestion
8- 27 8-28
1
and l eak for N =1910 rpm; Re =267; Fig. 9
Enclosed ai r did supersede Stationary case. Gas
so much oi l that the l evel escapes upward in
ri ses over the seal. Gas bubbles. Due to thi s
ingestion diminishes but strong pres sure drop,
the transport does not. later pre s sure increase s
again. After some time
a new gas front appears.
22
8-31 8-33
Gas inclusion and l eak for N =3480 rpm; Re =486; Fig. 10.
Already after 2 sec. the seal
is completely filled with f oam
1
After 4 min. the level has ri sen
due to gas inclusion and l eak
Stationary case with smal l
pressure variations
23
with liquid is simple to cal cul ate.
rnl/mm. According to formula (1) on page 1, 1 is proportional to P/@xN.
From thi s it follows that dl/dt =dP/dt. C / JI N i n which C
The volume of liquid in the seal V, is proportional to the wetted length.
x C2 in which C Hence we can write 1% V and dl /dt =d\l/dt
Combining both derivations we find:
Per unit length we fi nd V =0.09
is a constant.
1 1
is a constant.
2
Table 6.
Resuit s:
ingested
general,
The relation between the volume flow of the encl osed gas and the thereby
di spl aced liquid is difficult to cal cul ate.
and not reproducible, moreover the pressure of the encl osed gas changes, as
more gas causes the wetted length to i ncrease.
The ga; ingestion i s not constaqt
In the following manner an approximate val ue or the ingested gas flow
has been abtai ned.
gas in the chamber below the shaf t.
pressure bu,ld up immediately after the start (Take a value within 10 sec.) .
For the ca.'.culation af the gradient dP/dt we use val ues within 30 sec. after
start.
From graphs 8, 9 and 10 we find val ues for P and dP/dt.
B . We dssume the mean temperature at 20C and by setti ng P.V. =const.
we reduce to gasflow equal to fi
The volume flow of the liquld fi , when there is not yet
V
The pressure of the ingested gas i s the
Formula (2) gi ves us
V
at. pressure P to a gasflow fi at 1 atm. See
V g
In Table 6 the resui ts are summarized: the maximum quantity af
gas turns out to be
the leak does i ncrease with (he number of revolutions.
=8 inl/min (at P : = 1 atm; T =20" C) . In
g
7.5 Concl usi ons from the observations on vi scoshaft seal no.-l
1. The phenomena observed with regard to the gas inclusion is in very accurate.
agreement with the experiments of Stair and Hale (8) and with the observati ons
of Ludwig (10). Compare the summary of the literature survey in 2. 1
24
2. For the l owest number of revolutions, N =1082 rpm, there is a wel l
defined oi l /ai r interface whereby no gas ingestion occurs. For higher
vel oci ti es there is no longer a wel l defined interface: gas i ngesti on
does occur.
3. Gas ingestion may be prevented at high vel oci ti es by putting so much
sealing fluid in the system that the l evel is far above the grooved part
of the seal.
construction of the apparatus. At a high l evel , the fluid may contact
the upper bearing.
through the seal (see also Fig 1.).
4. If the liquid l evel reaches the machined end of the shaft, decreased in-
gesti on resul ts. See pi ctures 8-20 and 8-21 and Fig. 4. This effect
cannot be observed at high vel oci ti es where strong ingestion occurs.
5. For N =1082 rpm whereby no gas ingestion occurs, i t has been observed
that a quantity of gas that was originally present in the space below the
shaft does di sappear slowly.
6. For the lower numbers of revolutions gas transport primarily takes pl ace
along the low pressure si des of the l ands.
7. For numbers of revolutions higher than 1910 rpm pressure vari ati ons occur
in the stationary state caused by the escape of encl osed quanti ti es of gas.
Hereby so cal l ed gas fronts are vi si bl e.
maximal for N =1910 rpm and decrease for increasing number of revolutions.
The height of the l evel is restricted in our case by the
Thereby foam will be formed that i s transported axi al l y
The pressure fluctuations are
8. For numbers of revolutions smaller than 1910 rpm, it takes longer for a
smaller volume of liquid to reach the stationary state than for a larger
volume of liquid.
9. Due to the verti cal mounting of our seal, fluid cannot possi bl y leak out of
it. I t has been observed that seal i ng fluid was transported beyond the acti ve
length of the seal at the escape of gas fronts and for high numbers of
revolutions.
l eakage of oi l . For the intermittantly escapi ng encl osed quanti tes
See Table 4.
For a shaft rotating in a horizontal position thi s could l ead to
25
8-20 8-2 1
For low numbers of revolutions the gas ingestion decreases when the
liquid l evel begi ns to coi nsi de with the machined shaft end.
26
of gas (gas fronts) a pulsating l eak occurs.
Ludwig (10). If the flu hamber to be seal
wi l l not accumulate at the high pressure end
horizontally.
fi l l ed space.
This has been observed by
The resul t is that gas is pumped conti nual l y in the liquid
This i s accompanied by a continuous l eak of fluid.
10. An approximate cal cul ati on of the quantity of gas i ngested i s possi bl e
with the use of the sl ope dP/dt (Fig. 8, 9 and 10) a An i ncrease with the
number of revolutions can be observed; the maximum quantity i ngested
was fi =ml/min. (for P =1 atm; T =2OOC ) (Table 6)
g
8. Observati ons and concl usi ons with reqard to gas inclusion and l eak of
a si ngl e V. S. S. as a seal aqai nst a chamber of constant pressure.
For the observati ons descri bed so far under 7, the space to be seal ed
was filled with a constant volume of liquid. After the start the pressure
in the liquid i ncreased until the pressure generated by the V. S. S. was
reached.
I n practi ce one will often be concerned with a seal agai nst a chamber
of constant pressure, for example pumps and compressors. The shaft
seal must then at least be abl e to generate that pressure in order to prevent
l eakage.
With shaft no. 2 (see Fig. 5) a seri es of experiments is conducted; the
sealing ri ngs on the shaft had been removed.
Observations
The apparatus depi cted in Fig. 3 was connected with the apparatus of
Fig. 1 at the points &. After the start a pressure was introduced in the
space below the shaft whic'h was about 3/4 of the theoreti cal maximum sealing
pres sure.
For all numbers of revolutions larger than 1082 rpm gas inclusion and
l eak appeared. Thereby the fluid i s superseded and the pressure is inclined
to ri se. A continuous correction with a regulating valve was necessary.
Except for the hi ghest number of revolutions a steady state was attai ned
within 5 minutes. Starting at N =1382 rpm gas fronts are observed such as
27
those observed under 7.3 where they start to appear
Concurrently with the gas fronts strong
again: the maximumpessure i s the imp0
table 5, shaft no. 2. For an increasing number of r
pressure fluctuations decrease, and at N =2590 rpm the pressure has
become constant. I n thi s situation large gas bubbles still escape
upwards from the space below the shaft (gas fronts). For N =4120 rpm
the steady state is reached only after 15 minutes.
C o nclu s
There is a sharp di vi si on between the number of revolutions N =1082
rpm at which no ingestion appears aqd the seal does not l eak and the
number of revolutions N =1382 rpm at which strong l eakage occurs. The
wetted length of the seal vari es strongly due to the escape of gas fronts.
J ust as under 7.5 conclusion 9, gas escapes continually here also3 from the
high pressure space below the seal throug5 the l atter to the low pressure
end. If the chamber to be seal ed is a flowing system, then the gas which
leaks through the seal towards the high pressure space will be carried off.
Gas fronts will not occur and the transport of gas through the seal takes pl ace
only in the direction of the pressure build-up.
liquid will be mixed with a stream of gas which, basi des undesirable
contamination, may have a detrimental effect on the pump capaci ty.
I n thi s way the stream of
9. Prevention of-leak of a si mpl e V. S. S.
Since an i nsi ght in the l eak phemmena i s obtained from the observations
of the si ngl e shaft no. 1 we can try to prevent l eakage. One can:
a) Alow gas inclusion but prevent an axial transport of gas over the full
length of the seal.
b) Try to prevent gas ingestion and gas inclusion.
Ad.a) A solution along these l i nes does not prevent gas inclusion. The
uestion remains i f the cal cul ati on method is still satisfactory as
it does not take gas inclusion into account. An effort will be made
to prevent l eakage without changing the geometry.
28
Ad. b) After a solution has been foun
velocity, the chance alway
l eak do appear
fore method a) is a more
l eak. Method b) will be investigated by changing the surface
energy of the shaft and by deepening the groove in the shaft.
In the l atter case the geometry does change.
9.1 Prevention of l eakaqe by installing a qas barrier
I t was supposed that l eakage might be prevented by installing a
barrier agai nst the gas included.
end of the shaft (low end) the gas cannot penetrate the space containing
the fluid to be seal ed.
If it i s placed at the high pressure
To real i ze thi s i t is necessary to i nstal l a part that must be fitted
in the housing with very cl ose tolerances. A s the theory i ndi cates, it
follows from the absence of net axial transport that the gas will not cross
thi s smooth part of the shaft.
Because i t was not feasi bl e to construct a fitting bushing on the lower
end of the exi sti ng shaft no. 1 a new shaft was machined with the same
geometry as the fi rst one having a closing ring at its ends (Fig. 5). The
diameter of thi s closing ring is equal to the diameter taken over the lands:
due to the tolerance on the radial cl earance a larger diameter is not
possi bl e.
With a cl osi ng ring at its lower end thi s shaft is cal l ed no. 2-A.
A. Observations
For the seal i ng with shafts nos. 1 and 2-A that have the same geometry,
the initial gas ingestion is almost equal (pictures 8-8 and 9-28).
steady states differ (pictures 8-12 and 9-31).
The
as to the space below the shaft took
o or very little
29
8-12
8- 8
9-31
Comparison of the stationary states of shafts 1 and 2A;
N =1522 rpm.
2A gi ves gas inclusion but no l eak.
Shaft 1 gi ves ri se to gas inclusion,
shaft
9- 28
Comparison of shafts nos. 1 and 2% with and without cl osi ng
ring. Pictures are made 3 sec. after starting; the initial gas
ingestion is equal for both shafts.
30
l eakage of gas to
thi s case the i nges
to i ntense format
from the upper part of the clo
in the pressure. The steady state i s attai ned faster for sha
for shaft no. 1 possi bl y because
for the attainment of a steady state (see Table 4 at N =15
ring but thi s doe
I t can be observed in picture 9-31 that gas i s present over a large
part of the cl osi ng ring, although on the average only at about 5 mm
from the upper end. The penetration over a larger part coi nci des with
the escape of a gas front. Apparently, the accumulated gas l ooks for
an exi t and in thi s case still finds it upward.
----- Starting with N =1720 rpm l eakage of gas takes pl ace across the
cl osi ng ring.
formation of gas fronts from the space below the cl osi ng ring. Other-
wi se the picture is analogous to picture 9-31. When the fluid l evel
rises so strongly due to the gas l eak that i t becomes si tuated above
the acti ve part of the seal, then no more gas ingestion occurs. A
steady state ari ses for which there is a certai n amount of gas below the
cl osi ng ring.
----- For N =2175 rpm sometimes a large quantity of gas accumul ates
under the cl osi ng ring although gas fronts do occur (picture 10-23).
----_ For N =4120 rpm a very dense foam devel ops immediately as for
shaft no. 1.
This process is very sl ow but resul ts finally in the
When a large volume of liquid was seal ed
There was gas at 3/4 of the cl osi ng ring, but no l eak was
el y after the start (picture 10-35).
process
sure build-up i 34 min.
the state was
31
10-35
Shaft 2A; N =4120 rpm. Gas
appears on a large part of the
closing ring but no l eak occurs.
10-11
10-23
Shaft 2A; N =2175 rpm. Some-
ti mes much gas is accumulated in
the space below the shaft, gas
fronts are possi bl e .
10-13
Transport across the cl osi ng ring for all numbers of revolutions
occurs only in the pressure of a dense layer of foam above the
cl osi ng ring.
32
B. .Conclusions
1. Gas ingestion th shaft 2-A as w
N larger than
2. If gas ingestion occurs is more and dens
than with shaft no. 1. This is especi al l y apparent for low numbers of
revolutions.
3 . Leak across the closing ring and escape of gas in fronts towards the
low pressure end occurs first at N =1720 rpm for shaft 2-A.
for shaft no. 1 l eak occurred before that (at N =1382 rpm) whi l e gas
fronts were observed firstly at N =1910 rpm.
If leak occurs, then it takes considerably longer to reach the steady
state for shaft 2-A than for shaft no. 1 (See Table 4).
However
4.
5. Gas dses only exist on the closing ring or is transported across it i f
a dense layer of foam having a width of at least 10 mm is present
j ust above the cl osi ng ring (pictures 10-11 and 10-13).
The shaft 2-A exhibits edge effects. When the liquid l evel coi nci des
with the upper edge of the seal increased gas ingestion occurs. This
is easy to observe for N =1082 rpm where no gas ingestion occurs
when the liquid l evel is above or below the acti ve length of the seal
(pictures 38-11, 12 and 13).
7. I t was observed at N =1082 rpm that a smal l quantity of ai r in the space
below the shaft di sappeared in 15 minutes.
The situation is not reproducable for the l owest number of revolutions.
Sometimes little gas ingestion occurs.
6.
8.
9.2
rec ircula tion)
method was the foll
ingested gas b
ing fluid the gas bubbl es must be re-
33
38-11 38-12 38-13
Edge effect of Shaft 2-A. When the fluid level coi nci des
with the shaft end, increased gas ingestion occurs.
12-3 12-43
Shaft 1-D. If the upper recirculation channel ends in the liquid
(picture 12-3) then less gas ingestion occurs as for a lower
l evel whereby the channel ends in the gas.
34
due to i ts lower densi ty and may di sappear i n it i f an outlet is avai l abl e.
In order to accomplish thi s a number of changes were made i n shaft no. 1
(Fig. 4); these will be indicated by adding a l etter to the shaft number.
At fi rst we wi l l try to eliminate the observed large gasflow at the upper
sides of the l ands.
Observations
----- See al so Fig. 11.
the axi s. The upper end i s cl osed again. TWO radi al channel s, one in
the center and one at the upper end, connect the shaft surface with the
central cavi ty.
For thi s shaft 1-A an oi l gasilow from the upper charinel was observable
during operation.
----- In the center of the shaft, j ust above a l and, another radial channel
was drilled in the same plane as the one drilled previously. For thi s shaft
1-B an observable decrease in the transport of gas occurred j ust beyond
the point where the center channel s were.
the gas was vi si bl e during an observation at N =1382 rpm.
case gas fronts escaped from the space below the shaft. Arriving at the
center channel s the gas disappeared in them while at the same ti me the
Ioarn intensity increased at the upper channel.
----- Around the shaft a groove was machined (depth 1 mm; width 2 mm)
in the plane oi the center channel s. It was supposed that the gas pressed
agai nst the shaft could separate itself more easi l y from the oi l in thi s
groove to be subsequentiy returned. Al so due to the cutting through the
land and the absence of net axial transport no gas may cross the groove.
I ndeed, for thi s shaft 1-C a strong improvement is observed; however not
yet sufficient.
----- Therefore the grQove is deepened and brought to 4 mm.
radial channel s are drilled to fl =1.5 in order to be able to carry off the
prcbably large quantity of gas. This shaft 1-D shows a strong decrease
In vi sco shaft no. 1 a hole has been drilled along
Both channel s begin directly below a land (,d =1 mm).
No +crease of the l eakage flow vas observed.
The good recirculation of
In thi s
Both the
35
of gas ingestion beyond the groove.
complete wetting of the seal and time necessary to reach a stationary state
i ncrease.
i ncreases with a factor from 5 to 10.
Hereby both the time necessary for the
See also tabl e 4 and compare with shaft 1. The necessary ti me
I t appears as i f the foam densi ty above the groove i s i ncreased with
respect to shaft 1 (pictures 12-43 and 8-19). When the seal is almost
completely filled with 1iqu;d and i f the upper channel ends in the liquid,
the gas ingestion decreases agai n (pictures 12-3 and 12-43).
Gas fronts occu; in l arga- intervals with shaft l-D than with shaft 1. The
ingestion at N=4120 rpm i s analogous to shaft 1 (compare picture 12-3 with
8-33 at 7.3).
about 15 minutes the fi rst gas front escapes from the space below the shaft,
it does not move farther than the groove and di sappears (picture 12-5).
However beyon3 the groove there i s little gas transport. After
The castrahsport beyond the groove may be caused by a smal l decrease in
pressure behind the inclining land.
a part of the land was invastigated.
That i s why the effect of machining away
This yielded shaft l - E Fig. 11.
This shaft did not qi vs an 3bservable l eak under any ci rcusstances if
the liquid level was above the groove. The increasing improvement i s
recorded in the pictures 8-19, 12-41 and 13-22 at N =1720 rpm.
The shaft obtained a more or l ess "self stabi l i zi ns" mode of operation.
If the l evel is too low originally i ntense gas inclusion and l eak takes pl ace,
completely anal ogms to shaft 1 (picture 8-33 at 9. 3). Due to the accom-
panying ri se of the level, gas inclusion stops at a certai n moment (pictures
13-24 and 26).
gas a3bserved as a bubble around the shaft.
I n the stationary state there is then a certai n amount of
9.3 Synopsis of the resul ts of 9.1 and 9-2
-_--_ The solution of shaft l - E is a complete solution.
no l eak through the seal.
of the shaft the acti ve length of the seal i s cut i n half.
There occurs absolutely
However by putting the groove around the center
36
12-8
Shaft 1-D
12-4
N =4120 rpm
12-5
Re =575
1
J ust as with shaft 1 strong ingestion above the groove occurs
immediately after starting (picture 12-8) The little ingestion
below the groove l eads after 15 min. to the escape of a gas front
(12-4). Immediately thereafter the foam on the lower halve of
the shaft disappears; below the shaft there is a cl ear "air bell"
(picture 12-5) Gas ingestion and l eak cause a new gas front to
escape after about 1 112 min. whereby picture 12-4 appears again.
37
8-19 12-41 13-22
Shaft 1 Shaft l-D Shaft l - E
Re =240 internal recirculation system
1
N =1720 rpm
I ncreasing improvement of the prevention of l eak. 8- 19 original shaft:
12-41 shaft with groove and recirculation system; 13-22, the land is
machined away on both si des of the groove, no more l eak occurs.
13- 24 13 -26
"Self stabilizing" operation of shaft l -E. After initial gas inclusion (13-26)
a stationary state without new inclusion occurs due to the rise of the level.
----- Shaft type 2-A only gave a del ay of the l eak, see Table 4:
combined with the changes in construction of shaft 1 a solution ou
to be gotten al so.
----- For thi s reason shaft 2-C has been made, see Fig. 12. Here the
groove is machined out of the upper part of the cl osi ng ring. I nternal
recirculation is possi bl e. The shafts 2-C and 2-D that only differ in
the position of the upper radi al channel , al so give a seal without any
leakage
9.4 Prevention of leakage by a part of the seal over which no axi al transport
is possi bl e
For the aforementior,ed succesi ve changes of the shafts 1 and 2 it
has not become certai n i f a groove without a possi bi l i ty of internal
recirculation can prevent l eakage.
Theoretically no net axial transport occurs for the seal so that over
smooth parts of the shaft or across the grooves no l eak should be abl e
to occur. In order to i nvesti gate thi s we cl osed the radi al channel s i n
the shafts 1 and 2; the shaft are indicate with l-G and 2-E.
Observations
----- It was thought initially that these shafts al so gave a complete
solution for the prevention of leakage. Shaft l -G sati sfi ed best. Very
sporadically gas bubbles were transported from the groove towards the
lower end of the shaft. However they di ssol ve along the way. If we
allow a little oi l to flow away suddenly a net transport resul ts and gas
is cl earl y transported along the full length of the seal. This is contrary
to the shaft with a possi bi l i ty for recirculation because there no transport
was observed after a sudden pressure drop. However the l eak of shaft l-G
is of rel ati ve val ue because we found at all numbers of revolutions that the
ai r present bel the shaft di sappears after some ti me. A slow diminishing
of the oi l volume does not yield an observable gas transport.
39
---a- More l eak appears for shaft 2-E. The phenomena are not
reproducable. Here al so it is observed that a smal l quantity of gas
i n the space under the shaft di sappears for all numbers of revolutions.
Especially when oi l is drafted from the system gas is transported
across the cl osi ng ring.
while for the same number of revolutions and high liquid levels in the
seal l eak does take pl ace.
C o nc 1 us i on:
Sometimes no l eak occurs at low oi l levels
Fcr systems in which pressurs variations occur we may observe gas
inclusion and l eak in the absence of an internal recirculation system.
Shaft l-G i s best suited to remove the enclosed gas.
Gbserved in what way the gas di ssol ves or how it is transported.
It cannot be
For shaft 2-E more ingestion occurrs and the di sappearance of the
encl osed gas progresses slower.
cases the gas inclusion and l eak are only delaye6.
Therefore we can say only that for both
9.5 Prevention of qas inclusion and leakaqe bv chanqinq the surface energy
of the shaft
Till so far only constructive solutions have been di scussed; thi s is
a physi cal solution of the leak problem.
by a liquid depends on the surfacn energks. For a solid thi s can be
varied by applying or having absorb foreign matter. The surface energy
of a liquid can be changed by the addition of certai n elements. One then
may investigate i f the wettability of the shaft surface influences gas
inclusion and l eak.
The degree of wetting of a solid
In general one may assume that a liquid with high surface energy does
not spread over a solid with low energy.
the surface energy of the fluid is l over than a certai n cri ti cal surface
For wetting it is necessary that
tensi on of the solid.
The seal i ng fluid used, no. 3, does cpread wel l on the shafts (material
-3
messing and steel). The surface energy of the liquid i s about 30 x 10 N/m.
40
To induce a possi bl e lesser spreading shaft no. 2 has been coated with
a thin l ayer of Teflon. This material has a wel l known low cri ti cal
- 3
surface tensi on of 18 x 10 N/m.
McGrew and Orsino (9) measured that for a V. S. S. coated with Teflon
less l eakage occurs. We cannot measure a l eakage of liquid at a single
V. S. S., however we can say on the basi s of vi sual observations and
the comparison of those if and how gas inclusion is affected.
----- During our investigation a Teflon coating did not appear to affect
the i ntensi ty of gas inclusion. I t i s possi bl e to obtain a sl i ght decrease
in the maximum attai nabl e pressure generation: data concerning the pres-
sure build-up of shaft no. 2 without coating were not available.-----
9.6 Prevention of gas inclusion and leakage by making the grooves of the seal
deeper
The groove of shaft no. 2 has been deepened and brought to 0.90 mm.
Therefore the geometry is changed; the shaft will from now on be indicated
as shaft no. 9.
The groove may not be deepened too much because it can be derived
theoretically that due to thi s process a strong decrease of the seal i ng
parameter will occur after an initial increase (12). I t is true that thi s
investigation is primarily concerned with the prevention of l eak and one
will therefore be sati sfi ed with a lower sealing parameter; however i t is
cl ear that a l i mi t has to be imposed.
supplied the chosen groove depth.
0 b serva t io n s
Experience with double V. S. S. ' s
The resul ts are shown by the pictures. I nternal recirculation is not
possi bl e. Compared with shaft no. 2 cl earl y less gas ingestion and gas
inclusion occurs especi al l y for the low numbers of revolutions. Only at
the hi ghest velocity (picture 41-10) a small l eak occurred every once i n
a while. The vi si bl e quanti ti es of gas in the space below the seal (pictures
ore the start.
for the lower numbers of revolutions.
2. Leak does not appear except sporadically at the hi ghest numb
revolutions. For shaft no. 2 l eak appeared for N =1720 rpm.
3. For a strong decrease in pressure (due to letting off oil) l eak does
take place across the cl osi ng ring; thi s is analogous to shaft no. 2.
4. If gas has accumulated below the shaft, for example due to a pressure
drop, then thi s gas di ssol ves and di sappears. This phenomenon was
al so observed with the shafts 1 and 2 but the di ssol uti on of the gas
was not as fast.
5 min, for shafts 1 and 2 from 15 to 30 min. The ti me is dependent
on the quantity of gas.
5. For two numbers of revolutions that is N =1720 rpm and 1910 rpm,
only gas i ngesti on occurs. Gas is hardly transported to the lower
end of the seal. See picture 41-12.
6, Contrary to shafts 1 and 2 gas ingestion anci gas inclusion does occur
for the l owest number of revolutions.
with the shaft no. 8 to be di scussed l ater, for which the geometry is
equal to shaft no. 9 except for the number of grooves.
The el apsed ti me for thi s case is of the order of
Hence there is no gas inclusion.
This has been observed al so
The effect of a further deepening of the groove wi l l be investigated
for the double V. S. S. I s .
10.
In the apparatus depi cted in Fig. 1 the si ngl e vi sco shaft has been
ouble grooved shaft. I t is drawn schematically in Fig. 13.
st of a steel core on which messi ng vi sco bushings can be
42
41-13
N =1382 rpm
41-12
N= 1720 rpm
41-10
N =4120 rpm
Effect of the deepening of the groove on the gas inclusion and l eak. Above
shaft no. 9, below shaft no. 2. Due to deepening the gas ingestion decreases
and no leak occurs. For (41-12) only ingestion occurs and no l eak or
inclu s ion.
9-32 9-35 19-8
43
mounted. One shaft seal should consi st of at least a bushing with left-
handed thread and a bushing with right-handed thread. The bushings
must be mounted on the shaft such that during operation pressure will
be generated towards the center of the shaft. The geometry that is
obtained using the bushings, of the obtained seal is defined in tabl e 1.
Both bushings can be mounted separatel y. To thi s end a number of central
bushings are available. Figure 14.
The double V. S. S. is suited to seal two gas filled spaces from each
other. In our case these spaces are the space below the bottom bushing
and the space above the top bushing. The l atter has an open connection
with the environmental ai r. Using the mounting in Fig. 3, the lower space
can be pressurized as need may be.
The sealing fluid is supplied through the opening in the seal formed by
the upper bearing when the shaft is not in operation.
For the investigation the following observations are of interest:
The degree of foaming in the seal. I t appears that the sealing
parameter is influenced by thi s and break down may occur. In
another part of thi s investigation the influence of foaming on the
sealing parameter will be investigated, now only the intensity of
foaming will be establ i shed.
The observing and measuring of gas leakage and
The observation of an oi l leak indicated by loss of sealing liquid.
The non-reproducity of the foaming intensity and gas inclusion
phenomenon has already been di scussed in 6.2. Because the
l eak i s dependent on the foaming intensity these phenomena are
emphasized agai n here.
This phenomenon including its consequences has been di scussed
al ready in 6.5.
a)
b)
c)
Ad a)
Ad c)
44
10.1 The siqnificance of the observed measured gas l eak
The gas l eak phenomenon has already been descri bed in 6.6. For
thi s phenomenon at the double V. S. S. a further explanation follows.
Gas l eak, being an axial transport of encl osed gas through the seal,
ought to te observable vi sual l y, if necessary with instruments.
possi bl e that l eakage is observed although over the seal as a whole no
net axial gas transport occurs. We then experience d l eakage of gas
in the form of an exchanqe procLs2. Al so i f a pressure difference i s
imposed bztween the two spaces there may be l eak only due to exchange.
I n that case no change in pressure occurs.
I t is
If during the above mentioned l eakage process, the transport of gas
---- in one direction -- is dominant then a net gas transport wi l l occur through
the seal .
of both gas-filled spaces with fi:reign gas will take pl ace.
can be measured.
of the seal and there is mainly gas transport to the top of the seal, then
the pressure in the space below the V. S. S. drops.
in Fig. 3, is now connected with the apparatus in Fig. 1, the l evzl in
the left leg of the U-tube ri ses if val ves is cl osed. This i s cal l ed a
posi ti ve leak.
Dependent on the rate of the ensuei ng, exchange contamination
This gas l eak
If there is an encl osed quantity of gas at the bottom
Because the rig
If the direction of the net gas transport is opposite to the one in the
former case, then the quantity of gas in the space below the shaft in-
creases. The pressure ri ses and the l evel i n the left l eg of the U-tube
drops. NCIW negative i eak occurs.
The leak observations have ai ways been made at imposed high pr-es-
Then, for a positivz leak, a pressure sure in the space below the shaft.
drop will follow.
Ti-igether with a negative leak often a great l oss of seal i ng fluid is
observed.
volume decreases.
crease of the volume below the shaft will resul t in an i ncrease of pressure
The fluid appears in the space below the shaft hence the gas
If no gas transport occurs through the seal the de-
45
there. This is also noted by the appearance of a negative l eak.
Conclusion
An observed negative 1 ak does not need
ul l eak in the
seal. It may be caused by a l oss of seal hg fluid.
10.2
Chemical Equipment
Using vi sco bushings no. 3 a shaft was constructed similar to the one
in the Lab. for Chem, Equip. The grooved parts of the shaft are in di rect
contact for thi s case (picture 38-37).
For gas inclusion l eak will al ways appear. The gasses will contact
in the middle of the seal (picture 38-38) whereby exchange occurs. Dependent
on the ci rcumstances we may al so encounter positive or negative l eak.
11. The Prevention of l eakase of a double V. S. S. by recirculating the ingested
qas and/or the installation or' a qas barrier
We may expect that the observed l eakage phenomenon for the single
This is confirmed by V. S. S. will now occur from both ends of the seal.
the observations of 10.2.
We investigated i f l eak can be prevented in the same way as for the
si ngl e V. S. S,
----- A gas barrier can be installed between the two hal ves of the seal
using center bushings (Fig. 13 and 14).
----- In Fig. 6 i s indicated how the recirculation system for the enclosed
gas is i nstal l ed in thi s double V. S. S.
The gas ingested at both hal ves of the seal must be recirculated to
the space from which it was ingested. The presence of two separate
recirculation channel s is therefore necessary. I n the grooves of center
bushing B the encl osed gas is caught whereafter it can flow back through the
internal channel s (Fig. 6). To prevent l eak through the cl earance between
bushing and shaft. I n thi s way the recirculation channel is defined accurately:
46
38- 38 38- 37
I mitation of the V. S. S. in the Lab. for Chem. I nstruments.
The grooved parts of the shaft are next to each other. If
gas inclusion occurs, at least exchange of gas will take
pl ace, thi s means l eak.
47
the low pressure end is formed by the radi al channel in the vi sco bushing.
11.1
11* 2
The manipulation of the l eak usinq the center bushinq
The center bushing A consti tutes a gas barrier for both hal ves of the
seal as investigated for the si ngl e V. S. S. at 9.1 and 9.4.
l eak only occurs in the presence of a certai n i ntensi ty of foaming (pictures
33-16 and 33-26). I n thi s case center bushing A appears to be less abl e
to prevent the l eak than bushings D or C. For the l ast one there are two
grooves next to the l ands adjoining the vi sco bushings.
pears to be abl e to prevent l eakage in many cases; the larger volume of
liquid that is then present in the center of the seal may be better abl e to
absorb smal l pressure vari ati ons so that no axial transport of gas is
possi bl e.
investigation the influence of the center bushing wi l l be investigated for
each case. Also the resi stance of the seal aqai nst sudden chanqes of the
pressure difference between the gas-filled space: will be investigated. I t
turns out that the least leakage occurs for a shaft with a center bushing C.
Werealso
This bushing ap-
(Compare 32-32 with 36-10). In the remaining part of the
The prevention of leakage by separating and recirculating the encl osed gas
From the earl i er di scussi on it appeared that center bushing C was
the best solution for the prevention of l eak. Therefore a second center
bushing was made with the same geometry but with t wo radial channel s (see
Fig. 6 and Fi g? 14). This is center bushing B. Via a groove in the shaft
and radi al channel s i n the vi sco bushings two recirculation systems are
establ i shes.
Under no ci rcumstances does l eak appear now. Gas transport to the
grooves in the bushing is clearly vi si bl e (picture 36-4).
If the capacity of the recirculation channel s is too great then, besi des
gas, fluid wi l l be pumped around al so. This is vi si bl e on pi ctures 36-41
and 36-0. Although neither vi si bl e nor measurable l eak was found, such a
system does not appear to be reliable. The large quanti ti es of gas that
48
33-16 33-26
Double V. S. S., in all cases l eak occurs only i f there is1
a certain amount of foam i n the center of the seal.
36-10 36-4
Possibilities for preventing leak using center bushing.
center bushing A; l eak due to exchange of gas. Picture 36-4, center
bushing B; no leak but internal recirculatioq transport of gas occurs
vi a the grooves.
Picture 32-32,
49
36-41 36-0
Due to the too great capaci ty of the recirculation channel s,
fluid is pumped around too.
fluid to fi l l the whole ci rcui t there i s intermittently a l arge
quantity of gas in the center of the seal.
Because there is not enough
50
intermittently appear i n the center of the seal need only to cover a short
di stance to cause leak.
I n the ideal situation only gas is recirculated. An investigation as
to which is the necessary diameter of the restriction channel s will be
conducted in the chapter about the manipulation of the seal i ns parameter
by recirculation.
12. The influence of the geometry of the seal on the phenomena of gas inqestion
and l eak
Because now, as evidenced in the preceding chapter, one has a way to
prevent the l eak caused by gas inclusion, one may i nvesti gate how the
geometry of the seal controls the l eak. One will strive for a seal without
any gas inclusion and gas ingestion because the cannot cause l eak. How-
ever, by changing the geometry, the pressure generation through the seal
and the di ssi pated energy change. A geometry optimized with re2pect to
these two quanti ti es will then have to be revi sed because of the prevention
of l eak.
The i ntensi ty of th2 gas ingestion and the gas inclusion is only observable
vi sual l y in our experiment. A measure for the quantity of gas encl osed is
the vi si bl e exchange of gas across the center bushing of the seal. Due to
the man-I; irreproducable pi ctures concerning the l eak and foam formation
we concentrated our observations on the intensity of foaming for a certai n
seal. Observations of l eak and of l oss of sealing fluid (6.5) will be
ccnsidered independent of the intensity of the foaming.
Due to its character thi s must be a comparative investigation. On
the basi s of vi sual observations changes are made in the vi scoseal geometry
that decrease gas inclusion.
bi l i ti es. In all cases we used seal i ng fluid no. 3.
----- Table 7 shows the comparison possi -
12.1 Observations on the double V. S. S. no. 3 (single groove)
The geometry of thi s seal is the same a? those of the si ngl e seals nos.
1 and 2 (Table 1).
5 1
a) The i ntensi ty of foaminq
The picture is strongly dependent on the number of re
pared with the anal ogous si tuati on for the simple V. S, S.
and 37-14) less formation of foam does occur. Situations with more foam
than shown in picture 37-14 do occur. For all low numbers of revolutions
the picture of gas inclusion on the lower bushing is not the same as the
one on the upper bushing. The gas inclusion on the upper bushing is
divided over a large number of smal l bubbles; on the l over bushing the
gas is concentrated i n a few l arge air pockets (picture 37-14). A s became
also evident for the single V. S. S., gas inclusion can be prevented by a
liquid level that is above the acti ve part of the seal. For the double V. S. S,
thi s condition could be obtained by pressurizing the space below the seal .
In thi s way only the l evel at the upper si de of the seal could be brought
above the grooved part of the shaft.
si bl e to reduce the gas inclusion by a high level in any case. Because
the space above the shaft is equal for al l double seals one does have a
means to compare.
complete wetting of the active length, is a measure for the sensi ti vi ty to
gas inclusion in general.
b) Observations of l eak
However it did not appear to be pos-
I t is supposed that the ingestion of gas, even for
For N =1082 rpm, there hardly is ingestion or gas inclusion. Sometimes
a little foam cl i ngs agai nst the center bushing (picture 37-15). For the
low numbers of revolutions no vi si bl e l eak is observable. During a sudden
pressure drop axial transport of gas occurs with all center bushings except
bushing .B.
We can observe l eakage across center bushing A up from N =1720 rpm.
A varying l eak is measured usi ng the U-tube rig. A posi ti ve l eak of 20 mbmin;
a smal l negative l eak of 1 ml/min and no l eak at all. This all for different
experiments at N =1720 rpm. Using center bushings B and C no l eak is
observed up from the above mentioned number of revolutions.
However a
52
37-15 37- 14
N =1082 rpm N =1720 rpm
9-36
N =1720 rpm
Stable interface with- Different inclusion picture Single V. S. S. In thi s case
out gas ingestion. on upper and lower bush- foaming al ways occurs
I nitially present gas ing. The lowei- bushing accompanied by gas fronts
is accumulated along contai ns less foam, or going up and eventual l eak
downwards .
1
the center bushing and bubbles at all low Re
does di sappear very numbers. The gas is
slowly or not at all. concentrated in large
air-pockets.
53
negative l eak is measured. This l eak i ncreases for higher numbers of
revolutions and is imputed to the following phenomenon:
c) Loss of sealing fluid
For thi s seal the rapid l oss of sealing fluid was striking. Due to thi s
all fluid di sappears from the seal in a short ti me because no fresh supply
was possible.
l eak.
N =4120 rpm and a pressure difference AP =0.15 kg/cm the following
l eak was observed:
This phenomenon is accompanied by a strongly negative
The phenomenon could not be observed below N =2270 rpm; for
2
I n 7.4 we computed that the volume of the liquid in thi s seal was:
V =0.09 ml /mm.
at most. For an initially partially wetted seal, as was the case in the
l eak measurement mentioned above, a leak of 4.52 ml . can be explained
as being due to l oss of sealing fluid,
Therefore thi s double V. S. S. can contain about 16 ml
12.2 Observati ons on the double V. S. S. no. 5 (four grooves)
a) the intensity of foaming
For low numbers of revolutions there often is a dense and homogeneous
foam along the center bushing. For N =1082 rpm the gas ingestion is
neglectable and an initially present dense foam di sappears within 15 min,
For other numbers of revolutions someti mes large quanti ti es of gas escape
from the center of the seal through the foam: these gas fronts we also
encountered at the single V.S.S. (pictures 33-13 and 35-35). For high
velocities the homogeneous dense foam is replaced by a less homogeneous
54
gas inclusion picture. (Pictures 35-31 and 35-32). For a sufficiently
high level of seal i ng fluid above the upper bushing gas ingestion can
be prevented completely and the encl osed gas di sappears from the bushing.
For low vel oci ti es, the foam is not al ways present initially: there is a
little ingestion observable and foam is formed slowly from the center bushing
towards the ends.
b) Observations of l eak
Here the vi si bl e leak is strongly dependent on the type of center bushing
and appears only if foam is preseiit in the center of the seal as has already
been noted in 11.1.
The difference between center bushings A and B is shown cl earl y by
the pi ctures 33-13 and 35-35, and 33-9 and 35-21 respectively.
----- For center bushing A a cl earl y vi si bl e l eak al ways occurs from N =
1382 rpm.
with foam over a large part but the two gas covered parts are not connected
anywhere (picture analogous to 32-32).
--_-- Less l eakage occurs with bushings B, C and D, From N =1910 rpm a
vi si bl e l eak is observable for bushing D: bushings B and C never show
leaks. Fer bushing C both grooves do sometimes f i l l with gas but the
center part sti l l remains completely wetted.
----- If a sudden drap of the pressure difference between the two gas-filled
spacas is imposed, bushings A, C and D show l eakage. For bushing B, with
internal recirculation; vi sual l y no l eak occurs but photographs do show
Below thi s number of revolutions the center bushing is covered
leakage.
bushing to the lower groove.
----- Leak was not al ways measured for thi s seal: only negative l eak ac-
Picture 35-36 shows that transport of gas took place from the upper
companied by a strong loss of seal i ng fluid has been observed.
c) Loss of seal i ng fluid
Up to N =1720 rpm no l eak was measurable for any center bushing at
thi s seal; a smal l loss of seal i ng fluid i s vi si bl e. For higher vel oci ti es
55
For N =1382 rpm foam i s present on the center
bushing but no leak occurs.
32--33
2
33-13 32-35 35-36
I nternal recirculation P =0.130 kg/cm
R e =240; bushing no. 5; Observations of le&. In the presence of foam
center bushing A al ways gi ves l eak. Center bushing B or C, with or without
recirculation are abl e to prevent i t.
appears al so i f transport to the lower groove took place which would mean
l eak.
1
For a sudden pressure variation; it
33-9 35-31 35-32
=575; bushing no. 5. With center bushing II there still is accumu,ation of
number, as in thi s case,
Re 1
foam in the grooves and therefore leak. For high Re
no accumulation of foam takes pl ace at center bushings B, C, and D and no
l eak occurs.
1
They a;e2wdl abl e to bar the gas al so for a pressure drop of
P =0.40 kg/cm (35-321.
56
at N =4120 rpm a negative l eak of max. 0.27 ml gas/min has been
measured. A particularity was the very strong negati ve l eak that oc-
curred using bushing C. It was 1.35 ml/min, although no vi si bl e
l eak could be observed across the center of the seal.
12.3 Observations on the double V. S. S. no. 4 (three grooves)
a) The intenSity of foaminq
No generation of foam has been observed for these bushings at all
low numbers of revolutions. I nstead l arge air pockets can be observed
now j ust as for shaft no. 3 (see 12.1).
For higher numbers of revolutions the ai r pockets di sappear and gas
inclusion occurs especi al l y vi a the low pressure si des of the l ands.
For a high fluid level gas ingestion can be prevented for every number
of revolutions; then the already encl osed gas di sappears.
b) Q&servations of l eak
For every number of revolutions the seals with bushings B and C are
l eak proof; also after imposing a sudden drop of the pressure difference.
Over N =1720 rpm the use of center bushings A and D gi vas sporadically
vi si bl e l eak.
intermittently posi ti ve and negative from 0 to 0.0 ml gas.
been conducted also for which only negative leak was measured.
c) Loss of seal i nq fluid
For these cases a slightly fluctuating l eak is measured,
Experiments have
For .this seal thi s phenomenon has been investigated more dccurately.
With seal i ng fluid no. 1 whereby.absolutely no qas ingestLon occurs
negative l eak has been measured too, accompanied by a loss of seal i ng
fluid downward. With the usual seal i ng fluid no. 3 al ready at the l owest
number of revolutions a strong loss does occur together with a negative
l eak. I n all cases the l eak decreases i f the lower bushing is not wetted
anymore by the seal i ng fluid.
generation still decreases some more and a complete reversal of l eak
from negative to positive occurs. However it often happens that the
Then by a smal l loss of fluid, the pressure
57
system is abl e to be in the unstabl e situation for a long ti me; the im-
posed pressure can be supported.
The observations are summarized in Table 8 . We can check if the
volume of the leaked fluid is the same as the volume of the measured gas
l eak.
This turns out to be the case for the fi rst two experiments whereby no
vi si bl e l eak was observed. However in the last experiment vi si bl e l eak
occurs and the measured gas l eak is smaller than was expected on the
basi s of l oss of fluid. This i ndi cates that gas l eak towards the low pres-
sure end of the seal took place.
12.4 Comparison of the dsubl e V.S.S.'s nos. 3, 4, and 5 that differ in the
number of grooves and the landwidth
For the geometries see Tables 1 and 7.
a) Comparison of the bushings 3 and 5, that differ only in the number of
grooves.
the geometry is equal but the groove and landwidth differ in an absol ute
sense. As we observed during the course of thi s investigation l eak may
easi l y appear i f a dense l ayer of foam occurs.
l eak (picture 33-15) although no l eak can be observed for bushing 3 in the
absence of foam.
use of a different center bushing, for example bushing B or C.
Bushing 3 has one groove bushing 5 has four grooves. Therefore
Bushing no. 5 causes
The l eak at bushing 5 could be prevented easily by the
For both bushings 3 and 5 mainly negative l eak was observed; for high
vel oci ti es loss of seal i ng fluid occurred.
----- The gas ingestion seems to be less for the bushings with four grooves
than for those with one; for the former foam occurs mainly due to accumulation.
For the lower numbers of revolutions thi s can be observed wel l because the
build-up of the foam l ayer requires some ti me,
----- The si ngl e grooved bushings' are suscepti bl e to gas i ngesti on because
it was not possi bl e to prevent gas inclusion by a high fluid level.
58
----- A di sadvantage of the single grooved bushings seems to be the
presence of on the shaft seal. Due to thi s large parts wi l l
not contribute to the pressure generation.
C o nc l u s ion
The effect of a diminished gas ingestion and a lower sensi ti vi ty to
thi s phenomenon is l ost for more grooves due to more intense foaming.
Hereby l eak occurs easi l y but is prevented by a right choi ce of the center
bushing.
single grooved shaft: i ts effect on the stabi l i ty is not known.
b)
and unequal landwidth.
detrimental foaming effects of bushings 5 that have a narrow land.
For lower numbers of revolutions ai r pockets develop for the
Comparison of the bushings 4 and 5 with a different number of grooves
Bushings 4 with a wide land do not exhibit the
Compare
pictures 32-40 and 33- 15.
much smaller.
for a high l evel of the sealing fluid.
Due to thi s vi si bl e l eak for bushing 4 is
J ust as bushing 5, bushing 4 is not sensi ti ve to gas ingestion
For a high number of revolutions there is gas inclusion for bushing no. 4
but the i ntensi ty is still less than f cr buAhingr; noe. 3 and 5; compare
pictures 35-1, 33-28 and 37-20. For bushing no. 4 the grooves are not
completely filled with dense foam.
C onc lu s ion
3y increasing the land width the foam formation as observed for bushing
5 di sappears.
less than for the single grooved. At a high number of revolutions the
For the two bushings with more grooves the gas ingestion is
i ntensi ty of foaming is less for bushings no. 4 than for bushings nos. 3 and
5; on thi s basi s we must choose: A multiple grooved bushing of which the
land width may not be too smal l , decreases the i ntensi ty of the gas inclusion
(bushing 4).
12.5 ComDarison of the double V. S. S. ' s nos. 4, 8, and 10 that differ i n the depth
of the qroove
The bushing that was found most ideal i n 12.4, the bushing no. 4 with
59
32-40
Bushing 4
33-22
Bushing 3
33-15 Re =193
1
Bushing 5
Comparison of bushings with one and four grooves with the same geometry
(bushings 3 and 5) and of bushings with three and four grooves (nos. 4 and
5) that only differ in land width. The single grooved bushinq no. 3 does not
often exhibit foaming but ai r pockets occur on the lower bushing. Bushing no.
- 5 with 4 grooves al ways gi ves dense foaming for low R e numbers. This occurs
due to accumulation because the actual ingestion i s very little.
causes l eak across the center bushing.
almost never causes foaming. Air pockets do occur, there i s no measurable l eak
and hardly any vi si bl e l eak.
The foam
Bushing no. 4 with 3 grooves almost
35- 1 33-28 37-28
For Re =575 there i s a more homogenous gas-liquid distribution on the shaft.
The pictures can be reproduced wel l .
This foam hardly penetrates on the center bushing or does not get there at all.
There i s no observable l eak.
1
Contrary to the foam at low Re number,
60
three grooves and wide l ands, may be investigated further now
to changes in the geometry. To thi s end only a epening of th
is considered and has been executed in two ste
a)
Comparison of the bushings 4 and 8; groove depths 0.60 and 0.97 mm.
On the basi s of vi sual observati ons we cannot immediately deci de if
there is improvement or not.
deepening the sensi ti vi ty to gas ingestion i ncreased strongly. For high
fluid levels gas ingestion al ways occurs. Also if the level is on the grooved
part of the shaft gas ingestion occurs for N =1082 rpm however not on the
lower bushing .
See pi ctures 35-43 and 41-1. Due to the
The nature of the gas inclusion is changed for low numbers of revolutions
(pictures 35-43 and 41-1).
completely are replaced by smaller bubbles along the upper edges of the
l ands.
of the shaft surface that resul ts.
The air pockets that filled the shallow grooves
This must be regarded as an improvement due to the better wetting
A s far as l eak is concerned, bushing 8 seems to have yielded an improve-
ment. For N >
observable for a sudden pressure drop.
respect to bushing 4 where l eak did occur al ready for center bushings A and
D. The l oss of seal i ng fluid is decreased for the deep-groove bushing and
is only l arge for N =4120 rpm.
----- I n order for axi al transport of gas to occur gas bubbles fi rst must l eave
the groove agai nst the centrifugal force. Only then can l eak occur across
the center bushing so that a deep groove appears to be favorable. 'Therefore
we decided to deepen the groove even further.
b)
1522 rpm and use of center bushing A there even is no l eak
Thi s is a cl ear improvement with
I
Comparison of bushing 10 with bushings 4 and 8.
Groove depth 1.5 mm.
----- For no number of revolutions does gas inclusion and l eak appear.
pictures 44-35 and 44-33. Gas ingestioh can still be observed at the
interface; the center bushing does not contribute anymore to the prevention
See
61
1
Re
357
Re =
1
57 5
35-43 and 35-2 41-1 and 41-2
Bushing no. 4 Bushing no. 8
44-35 and 44-33
Bushing no. 10
Bushings with three grooves. Deepening the grooves of bushing no. 4
finally resul ts in bushing no. 10 which does not give any gas inclusion
in the region considered. Even using tetra as sealing fluid (increasing
Re ) does not give enough gas inclusion to cause leak across the center
bu s hing .
1
62
of l eak. The width of the interface over
i ncreases with the number of revolutions.
h gas i ngest
For N =4120 rpm al so, there is still no g
abl e. Although it became apparent during the course of thi s
that the Re number is no cri teri on for the occurance of ga
l eak. bushing 10 is i nvesti gated al so using a different seal i ng fluid so
that a higher Re number could be attai ned.
1
1
We used seal i ng fluid no, 4, (tetra chlorine carbon). The region of
gas ingestion appears to be i ncreased but there is still no transport of
gas across the center bushing (picture 46-37).
For no number of revolutions has l oss of seal i ng fluid or l eak been
observed.
I mmediately after the start there often is gas present in the seal i ng
fluid on the bushing. Sometimes a very smal l transport of bubbles is
vi si bl e at a hi gh number of revolutions (picture 44-33), however, thi s
does not l ead to the formation of foam or the occurance of l eak. Within
2 mi nutes most of thi s gas has di sappeared.
C o nc lu s ion
Usi ng a deeper groove makes it possi bl e to prevent gas inclusion and
leak. Also no l oss of sealing fluid occurs anymore, hence break down
due to too little prassure generation in the seal does not occur.
12.6 Observations on and comparison of the double V. S,S.'s nos. 6 and 7
With these bushings a start'was made with the investigation of the
influence of the hel i x angl e on the gas inclusion.
using the shafts described so far because the groove depth (for bushing 6)
has been decreased initially by one half,
Observations on shaft no. 6
No comparison is possi bl e
a)
For N =1082 rpm no gas ingestion occurs although initially present ai r
may cause foaming. For higher vel oci ti es the ingestion is still very smal l
6 3
1
Re
46-37
Bushing no. 10
=1805 (with tetra)
64
(pictures 36-39 and 36-5). After a short ti me however
changed because gas accumul ates in the seal and caus
(picture 36-14).
are used, but axial transport does not take place. The formation of foa
is not a certainty; pi ctures 36-39 and 36-5 have been observed also with
foam in the seal at the same number of revolutions.
This foams extends to center bushings A
Up from N =2175 rpm, th am on the bushings is replaced by gas
ingestion or inclusion of bubbles (picture 36-5).
no gas reaches the center bushing anymore which would mean that only
gas inge stion occurs .
Here it even appears that
For N =4120 rpm gas inclusion and ingestion occurs (picture 36-7).
This seems to be less than for bushing 4, compare with 35- 2.
b) Observations of l eak
For low numbers of revolutions vi si bl e l eak occurs in the pressure of
a dense layer of foam.
fluid in the seal.
The leakage decreases if there is very little seal i ng
For number of revolutions N =2 174 rpm no l eak can be ob served anymore;
thi s apparently is due to the absence of foam.
In no case was l eak observed using center bushing B and either a
negative l eak or no l eak was measured.
c) Loss of sealinq fluid
These bushings are characterized by a very strong l oss of seal i ng fluid.
When both bushings are wetted for 67% all seal i ng fluid has disappeared in
50 sec. for N =3480 rpm and no pressure difference.
d) Observations while usinq center bushing E (Fig. 14)
This center bushing was made to obtain an accurate imitation of the
center part of the shaft of si ngl e V. S. S. no. 1, see Fig. 11. Contrary to
the si ngl e V. S. S, thi s solution was not sati sfactory here; for every number
of revolutions l eak was cl earl y present.
or a low number of revolutions these hi ngs cause an i ntense formation
of f oam due to which leak appears. For high numbers of revolutions the gas
65
36-39
Re =193
Bushing no. 6;
1
36-14 36-5 36-7
1
Re =575
1
At low Re often foaming occurs the
gas inclusion i ncreases for high Re
Re =362 Re =240
1
6 grooves B =3.94
t = 0.177
Deepening of the groove of bushing no. 6 yi el ds a new bushing no. 7 that does
not exhibit gas inclusion anywhere in the range considered up to Re =575.
1
38-0
Re =193
1
38-41
1
Re ~' 362
38-40
Re =575
1
Bushing n3. 7; 6 grooves; B =8.25
t =0.177
66
ingestion and gas inclusion are less than for shi ngs nos. 3, 4,
and 5.
From'intercomparison of the bushings with three grooves, it could
have been concluded that a shallow groove promotes foaming and gas
inclusion. The good resul t for the described bushings No. 6, high
vel oci ti es should then be explained by the larger hel i x angle. If the
different effects are cumulative, then it may be expected that by deepening
the groove even less gas inclusion occurs.
Observations on shaft no. 7
Shaft no., 7 is obtained from shaft no. 6 by machining the grooves
from a depth of 0.3 to a depth of 0.74 mm.
----- Shaft no. 7 in no case exhibits gas inclusion or leak: a little gas
ingestion occurs.
bushing no. 7 with the pictures of the bushings 8 and 10, then we see
that bushing no. 7 shows the least ingestion.
Conclusion
If we compare pictures 38-0, 38-41 and 38-40 of
For a larger helix angl e a less deep groove is necessary to prevent
gas inclusion and leak than for a geometry with a smal l helix angl e.
12.7 Observations on and comparison of the double V. S. S. ' s nos. 11, 13 and 15
On the basi s of the resul ts obtained so far it was expected that a
seal with a still larger hel i x angl e would not exhibit gas inclusion even
for shallow grooves
For a certai n chosen large pitch (Table 1) and a land width that is not
too smal l , the number of gr.ooves was chosen to be 10. We started with
the very shallow groove of 0.2 mm.
expectati ons expressed above e
0 b servat ion s
a) The intensity of foaminq usinq shaft no. 11
This seal did not live up to the
I n order to get an impression of the gas ingestion caused by thi s seal
pictures 44-24, 44-27, 44-30, and 44-32 may be studied. True, they
67
44-24
Re =193
1
44- 27 44-30 $4-32
Ra =267
1
Re =575
1
Re =362
1
Gas inclusion decreases for i ncreasi ng number of revolutions for bushings
nos. 11 and 12. After deepening the grooves, so that B =3.94, the same
phenomenon occurs: The Re
anymore is reduced then from Re =418 to R e
=304 (bushing 13) and to Re =
240 for bushing 14.
number whereby no gas reaches the center bushing
1 1 1
1
68
depi ct a different shaft (no. 12) but a completely anal ogous picture was
obtained for shaft no. 11.
For N =1082 rpm absol utel y no ingestion occurs. At the next number
of revolutions, N =1382 rpm, we could observe that the gas ingestion
was very smal l but due to accumulation of gas there resul ted a shaft fully
covered with foam within 2 min. (picture 44-24).
For higher numbers of revolutions the intensity of foaming decreases
and at N =3000 rpm the gas is not ingested anymore up to the center-bushi ng
(pictures 44-30 and 44-32).
A similar phenomenon was observed for shaft no. 6. For high numbers
of revolutions a gas inclusion phenomenon remained present there whereas
in thi s case it di sappears completely. In order to i nvesti gate i f the
fcrmation of foam is continued in a lower region of Re
1
observation with seal i ng fluid no. 2.
numbers, we did some
Up to N =1910 rpm no gas ingestion can be observed using thi s seal i ng
fluid. After that the zone at the interface where ingestion takes pl ace
i ncreases. For N =2270 rpm there is ingestion across an interface down to
20 mm and thi s is about 30 mm for N =4120 rpm.
Conclusion
During seal i ng with bushing no. 11 gas ingestion appears at Re =64.
1
I t i ncreases and resul ts in gas inclusion for Re =193.
gas inclusion diminishes and for Re
1
decreases further towards Re =575.
----- B-, de2pening the groove a different geometry appears, it is called
bushing no. 13.
high Re
1
----- By a still further deepening of the groove, shaft no. 15 appears. This
seal is investigated using seal i ng fluids nos. 3 and 4.
investigated region from Re =151 to Rel =1805 did gas inclusion occur.
b) Observations of l eak
Hereafter thi s
1
=418 only ingestion remains which
1
Now the region of gas inclusion decreases down from the
number. Already for N =2185 rpm no more gas inclusion occurs.
Nowhere in the
1
Bushings nos. 11, 13, and 15 have not been made double. Only bushings
69
with left handed thread are avai l abl e. The other hal ve of the double V. S. S.
in thi s investigation is formed by bushing no. 7 with right handed thread.
This one did not show gas inclusion at any number of revolutions.
We now di spose of a system whereby gas i ncl usi on occurs only from
one of the oil/air interfaces. This fnterface is the upper interface be-
cause bushing no. l l is there. Hence l eak is only possi bl e from the
space abl ve the seal to the space below the seal. Furthermore it is ceftai n
that an eventual measured l eak is not due to loss of seal i ng fluid.
I n the region where gas inclusion is observed we al so al ways measured
a negative l eak. Hence from thi s l eak val ues can be obtained for the
transport of gas due to gas inclusion and l eak. See the tabl e below.
Measured negative l eak in ml/min for
2
A p =0.1 kg/cm .
N rpm bushing 11 bushing 13 bushing 12 bushing 14
1082 0.0 0.0 0.0 0.0
1382 0.0 0.0 3.6 0.0
1522 5.3 8.5 - 0.0
0.1 no inclusion
1720 7.6 -
1910 15.1 0.1 0.0 I t
2175 0.0 no inclusion 0.0
3000 no inclusion no inclusion I I
I 1
11
Conclusion
The expectation that by increasing the helix angl e a very shallow groove
can oppose l eakage has been born out only partially. Over N =3000 rpm no
gas inclusion appears for a shaft with a large pitch and a groove depth from
0.2 to 0.7 mm.
region where l eak occurs due to gas inclusion.
Below thi s number of revolutions there is a comparatively smal l
This region becomes smaller
by deepening the groove. For a depth of 0.72 mm the region ceases to exist.
The order of magnitude of the l eak due to gas inclusion agrees with the
measured l eak for a single V. S. S. in 7. 4
70
12.8 Observations on and comparison of the shafts nos. 12, 14, and 16
The only change of thi s bushing geometry with respect to the bushings
in 12.7 is the larger helix angle.
a) The intensity of foaming
For bushing no. 12 there is a shaft fully covered with foam from the
l owest number of revolutions N =1082 rpm. The i ntensi ty of the inclusion
decreases with the number of revolutions and for N =3000 rpm there only
is gas ingestion at the interface.
In a lower region of re
face for Re =64. An ingestion of very smal l gas bubbles is vi si bl e, they
are transported over a long di stance in the seal.
numbers, f oam starts to appear at the inter-
1
1
Before they reach the
center bushing, they are di ssol ved. The same phenomepon occurs for
Re
----- By deepening the groove the geometry of shaft no. 14 appears.
region of gas inclusion has become smaller and the i ntensi ty has decreased.
For N >1720 rpm only ingestion still occurs and thi s decreases with an
i ncrease of the number of revolutions.
----- If the groove is deepened further, shaft no. 16 appears. This one
=11 5 where it is more cl earl y observable.
1
The
does not show gas inclusion anywhere in the
Re =151 toRe =1805.
b) Observations of leak
1 1
These are given in the preceding tabl e.
cf the shaft was bushing no. 7.. Observable
l eak.
investigated region from
Here al so the lower bushing
l eak accompanies mea surable
12.9 Comparison of the qeometries of shaft seals with increasing helix angl es
1) The bushinqs nos. 11 and 12.
From the preceding tabl e it becomes apparent that the region in which
gas inclusion occurs becomes smaller i f.the helix angl e is increased. The
l eak a1 so decrease s .
2) The bushings nos. 6, 13, and 14.
For shaft no. 6 gas inclusion occurs for all numbers of revolutions;
71
46-29 46-29
Bushing 7 Bushing 15
t =0.177 t =0.268
R s =575
1
46-28
Bushing 16
t =0451
B = 8.05
The i ncrease of the helix angl e form t =0.177 to 0.268 t o 0.431 has no
influence anymore for thi s groove depth.
from Re
interface ingestion occurs over at most 5 mm.
Nowhere in the explored region
=151 to Rel =575 daes gas inclusion occur. A t the oil-air
1
46-36 46-36 46-27
Bushing 7 Bushing 15 Bushing 16
Re =1805 Sealing fluid n3. 4
B = 8.05
Al so with thi s seal i ng fluid there is no gas i ngesti on anywhere in the exp1.ored
region from Re =475 to Re =1805
1
The ingestion at 1 the air- sea,ing 4 fluid interface increased very slightly.
72
only for low velocities thi s causes le
numbers of revolutions di sappears for
pitch is increased. The region of ga
revolutions in which gas is present as foam, decreases w
helix angle. The intensity of foaming diminishes too; for shaft no. 14
there is no vi si bl e l eak anymore.
3) The bushings nos. 7, 15, and 16.
The photographs on the other page show these shafts with sealing
fl ui ds no. 3 and 4. Shaft no. 7 with the zmallest helix angl e still gives
the most gas ingestion. There is no more gas inclusion and l eak. These
three bushings are completely l eak proof i n the investigated region. The
depth over which gas ingestion occurs still decreases with lower numbers
of revolutions and at N =1082 rpm a completely .=table interface is present.
13. The influence of foaming on the seal i ng parameter
We saw that, due to gas inclusion, a connection may open between the
two gas-filled spaces to be seal ed with a double V.S. S. However so far
it has not been apparent that the vi sco-shaft seal usi ng foam as the
seal i ng medium would not be abl e to generate pressure.
The break down as noted by several investigators (lit. 5, 6, 8, 11)
means that, for an imposed pressure difference between the gas-filled spaces,
the fluid is blown out of the cl earance between shaft and housing.
l eakage flow is then the l eak of gas through a narrow slit.
The
13.1 Theoretical val ues for the seal i ng parameter
I n the laminar region formula (1) holds true. For turbulent flow conditions
the seal i ng parameter ((I is dependent al so on the Reynolds number besi des
on the geometry of the seal. Pape and Vrakking (lit. 2) found:
73
For certai n cases thi s formula can be simplified to:
where:
tS 0.268 and 0. 2<Y <0.8
B > 3.6 G =f (Re)
For the computation of these parameters one assumes a homogeneous
fluid in the seal. Hence one does not take gas inclusion into account.
13 e 2 The measurement of the seal i ng parameter
2
For thi s one uses the fopmula = pc / q 0 dl
for a certai n seal , c, d and 1 are given quanti ti es. If the pressure
build-up is measured for a certai n number of revolutions one can compute
the seal i ng parameter.
I n our case the pressure build-up can be determined accuratel y for
a completely wetted shaft length 1. I t appars that the pressure build-
up depends on the i ntensi ty of foaming in the seal.
must be determined; the parameter without f oam and the parameter with foam
i n the seal.
cl osi ng pre s sure:
A) The cl osi nq pressure without f oam in the seal
Therefore two parameters
This is done as follows vi a the determination of the maximum
For single shafts thi s can be done by usi ng a large quantity of seal i ng
fluid so that the level is high above the grooved part. The pressure measured
is the maximum pressure without gas ingestion.
For double shaft a pressure difference must be imposed over the seal by
using air under pressure. For a l arge quantity of seal i ng fluid and increasing
pressure there wi l l be no more liquid in the lower part of the V. S. S. after a
certai n moment.
upper busing no gas inclusion wi l l occur.
If, at the same ti me, the level is far enough above the
If gas was present it wi l l di sappear
32-2
74
32-3
Determination of the sealing parameter without encl osed gas. Due
to a pressure difference the wetted length of the lower bushing de-
creases and it i ncreases on the upper bushing. For a sufficiently
high l evel above the upper bushing no new i ngesthn i s vi si bl e and
the encl osed gas di sappears. When the lower bushing i s completely
dry, the maximum cl osi ng pressure has been reached.
13.3
13.4
from the upper part of the seal within a short ti me.
The pressure measured in thi s way is the maximum cl osi ng pressure.
B) The cl osi nq pressure with encl osed gas in the seal
(pictures 32-2 and 32-3).
If gas inclusion occurs in a V. S. S. when the oil/air interface is on
the grooved part of the shaft, then thi s wi l l happen al so i f the level is only
a short di stance above the grooved end of the shaft. Compared with case A)
the procedure is repeated only with less seal i ng fluid. With the so obtained
maximum pressures the seal i ng parameters can be computed with foam as
the seal i ng medium.
I naccuraci es due to temperature influences in the determination of the
seal i ng parameter
It is not possi bl e to measure or keep constant temperatures of the
seal i ng system due to the simple construction of the apparatus. For high
numbers of revolutions a cl earl y observable heat generation occurs.
if we neglect the influence on the densi ty then the influence of the tempera-
ture on the vi scosi ty and vi a that one on the seal i ng parameter is an uncer-
tai n factor. We estimate the maximum temperature differences to be 20C
and from that it easi l y follows that the seal i ng parameter may change 25%.
The error due to temperature sensi ti vi ty can be reduced by conducting
Even
the experiments swiftly and by usi ng fresh seal i ng fluid from a storage
container each ti me. On the basi s of thi s procedure a temperature of 20C
has been adopted for the determination of the seal i ng parameter.
Resul ts of measurements
For the seals with the shafts nos. 1,2,4,5, and 6 the seal i ng parameters
have been determined with and without foaming.
@ with foam where lower than those without foam in the seal.
16.
and that is not reproducable. Consequently, the data points are spread out
in a region below the curve that is measured without gas ingestion.
In all cases the val ues of
See. Fig. 15 and
The val ues of the parameter with foam depend on the i ntensi ty of foaming
76
The drop of the seal i ng parameter due to foaming increases with
the number of revolutions and a maximum drop of 34% of the val ue
without gas ingestion has been observed.
13.5 Explanation of break down
I n the accompanying figure the behavior of the pressure along the
length of the seal has been
sketched. The pressure
agai nst which must be seal ed
is P . For a certai n number
of revolutions curve 2 represents
the pressure build-up by the
shaft.
of revolutions no gas ingestion
occurs then the maximum pres-
sure build-up will continue to
increase: the seal remains
i ntact (curve h) .
gas ingestion occurs then leak
X
If for a higher number
However if
may occur due to the drop of the maximum closing pressure i f thi s pressure
i s lower then P
----- In order to prevent break down the shaft therefore al ways has to be
cal cul ated for the "f.oam parameter." We may only ask oursel ves how the
sealing parameter is influenced by foaming at higher vel oci ti es than
measured by us.
region, it does not seem to be unlikely that the sealing parameter approaches
zero for higher vel oci ti es.
opposing of l eak will then be futile. One will have to find a solution that
opposes the gas inclusion.
(curve c>.
X
In view of the strong decrease of 34% in a smal l velocity
Hereby break clown would occur again. The
14,
For the V. S. S.
cal cul ati on. From thi s it follows that
occurs (see figure). The
center bushings A, C, and
D did not appear to have any
influence on the foam-free
parameter. For bushing B
there is dependence on the
diameter of the recirculation
channel . The channel s i n
bushing B (see Fig. 14) have
their diameter not changed
contrary to the diameters of the channel s in the vi sco shafts and the visco
bushings.
The diameter for several bushings has been varied from 6 =0 . 3 to
B =1.5 mm.
The tabl e below shows the resul ts.
~~
Shaft 6 recirculation foamfree, reduction with respect
No channel in mm to $4 without recirculation
2 1.5 37.6%
2 0.8 19.5%
3
.4
0.6
0.6
0 %
0 %
5 0.3 0 %
6 1.0 22 %
7 0.6 0 %
The foamfree parameters with and without recirculation have been
compared.
encl osed gas; in no situation did we observe that they were too smal l .
The channel s appeared l arge enough to di spose of the
Therefore one has to take the channel diameter smaller than 0.6 mm
in order to prevent unnecessary l oss of pressure. The decrease of
due to a too large recirculation channel is given in Fig. 15 and 16.
15. Comparison of theoreti cal and practi cal val ues of the (foamfree) seal i nq
parameter
Although it does not belong to the investigation, the resul ts of the
measurements are given in Fig. 15, 16, 17, and 18. Except for seal
1
2 a smal l decrease of the measured parameter occurs for all low Re
numbers.
Re
cal cul ati on are given in 13.1.
Thereafter the measured val ues i ncrease faster for increasing
number than is predicted theoretically. The formulae used for the
1
An explanation of the di screpancy between theory and practi ce possi bl y
may be found in the fact that our measurements have been made j ust in
the transi ti on region between laminar and turbulent flow.
16. Check of the resul ts of measurements on the double V. S. S. apparatus
in the Lab. for Chemical Equipment
Earlier investigators concerned themsel ves mainly with the comparison of
theory and practi ce for the parameters of the V. S. S. Vrakking (lit. 3) sum-
marizes a l arge amount of data. . He arri ves at the conclusion that the theory .
checks very well with the experiments. At the end of thi s investigation con-
cerning gas inclusion some doubt appeared as to the correctness of thi s
conclusion. The following points played a role in thi s.
1) The geometry of the shaft investigated by Vrakking is accuratel y equal
to the shafts nos. 1, 2, and 3 employed in thi s investigation.
2) The seal i ng fluid used, Tel l us 11/Petrol 1: 5, is al so used by Vrakking-
3) Our possi bl e numbers of revolutions are all in the range investigated by
Vrakking .
79
4) Under the ci rcumstances mentioned under 1) , 21, and 3) we al ways
observed gas inclusion and a lowering of th
respect to its val ue without gas inclusion.
al i ng parameter with
On the basi s of our observati ons we argue that Vrakking very probably
al so must have encountered gas inclusion. I n view of the influence of
thi s phenomenon on the seal i ng parameter it does not appear very l i kel y
to us that hi s val ues can agree with theoreti cal values.,
Verification of the val ues measured by Vrakkinq
Picture 38- 38 cl earl y shows that gas inclusion occurs for the shaft
which is equal to the one investigated by Vrakking i n the Lab. for Chem.
Equip. sxcept for its length. I n order to measure the seal i ng parameter
without foam an al terati on has been made in the shaft in the Laboratory
{picture 38- 39) . Two bushings of which the diameters are equal to the
one taken over the l ands have been installed at the ends. We assumed
that these would form a sufficient barrier agai nst gas to prevent inclusion.
I n the transparent mounting thi s indeed appeared to be the case {see
picture 38- 39) . A necessary condition is that the seal between the l ands
and the shaft ends al ways has to be filled with seal i ng fluid.
a smal l continuous flow of oi l was supplied to the seal.
Therefore
We verified that thi s flow of oi l did not exert a measurable influence
on the seal i ng parameters.
Results of measurements
The whole range of Reynolds' numbers from Re =125 to.Re =4630 has
1 1
beenremeasured. The maximum cl osi ng pressure in the center of the seal
and the moment exerted on the vi sco shaft could be measured. For an
accurate descri pti on of the apparatus we refer to, for example, lit. 3 .
The val ues of the seal i ng parameter were computed usi ng the formulae (1) e
The val ue of CE, the effective friction coefficient, has been computed
80
If
38-38 38-39
i V =4120 rpm shaft no. 3
the l evel s are on the grooved part of the shaft gas
inclusion certai nl y will occur (picture 38- 38) .
be prevented by installing to end bushing that have a
small radial cl earance in the housing. There al so must
be so much sealing fluid that complete wetting of the
shaft occurs. (38- 39).
This may
usi ng the formula:
measured
64 M
-
41
E -
H p U d
The resul ts are plotted in Fig. 19 together with the data of Vrakking taken
from lit. 3.
The val ues of the seal i ng parameter agree very cl osel y with the val ues
measured by Vrakking.
a very sl i ght i ncrease at low Re
measured by Vrakking.
of the i nstal l ed end bushings.
The val ue of the effective friction coefficient shows
number as compared to the val ues
1
This can be explained as being due to the friction
Conclusion
Apparently no gas inclusionoccurred for the earlier obtained val ues of
the V. S. S.
enough sealing fluid under all ci rcumstances to maintain a liquid level
above the grooved end of the seal.
This can be explained by assuming that the system contained
Because the shaft is hanging i n a cylinder with a seal ed bottom, fluid
l oss is not possi bl e downward.
with enough fluid does not cause inclusion on the shaft.
I t appears reasonabl e that a cylinder filled
82
W h W " " N
" N d N " "
a = = m W c r h a 3
w
0 M O m 0 m m v ) m v ) m v )
cn c n m m C n c n c n W T r T r T r T r T r
0 0 0 0 0 0 0 0 0 0 0 0 0
m h a - m h v ) w a w W w a
. . . . . . . . . . . . .
h - -
. - -
Qa
I
m
0 -
4 E
X "
a
N m uJ C O N C 0 0 h O R O h
m 0 0 o m o h m h m h m
m = = m = m = m N m N c r N T r 6 i - q
. . . . . . . . .
" 4
I
m
0 -
" E
X -
a
h N N L n N h N 0 0 0 0 0 0
N - -
m N O " N m N 4 N " N 4 h l
" C O C D - 4 N " v ) c r v ) T r v ) T r
. . . . . . . . . . . . . . - -
Tr
0 -
" E
c X'
E
;r
I
0
4
X
N
0
4
I I
u
.-
E
v)
0
0
1
a
..
m
c,
u-4
a
5
83
Table 2. Sealin!
Tellus 11/Petrol. 1: 5
Fl ui ds Used
2 3
1 N. s/m =10 centi poi se
Table 3. Possi bl e numbers of revolutions and Reynolds numbers
Gear
Combinat ion
60/180
80/180
80/160
100/180
100/160
100/140
120/160
120/140
120/120
140/120
140/100
Revolu-
ti ons per
minute
1082
1382
1522
1720
1910
2175
2270
2590
3000
3480
4120
Re, (20" C) for Sealing Fluid
No. 1
13
17
19
21
23
27
28
32
37
43
51
No. 2
30
38
43
48
54
61
64
73
84
97
115
No. 3
151
193
212
240
267
3 04
317
362
418
486
575
No. 4
475
605
667
755
837
9 50
996
1139
1315
1528
1805
84
I
Gi
I
d I W
V
c9
n
I
4
d Lo
N d
I I
l a -F I C Z ) v ) N
I d I f
I l W
1 I
I I
I
I
I I
I I
7
d
E I I
07 v;
E I I
E I I
m l v) Tr
. E
d l CD
N d
I
i-4
224
a,
m
r:
0
m
m
0
u
0
c
0
m
7
u
c
m
m
tr,
m
4
4
-4
..
-4
4
.-I
85
c,
m
c
0
u
N P *
m
i
l-4
-
?
l-4
-
N
E
E
? %
Nl-4
0
0
0
l-4
0
co
l-4
0
v)
co h 4 7 r
0 0 l-4 4
0 0 0 0
v)
N h
l-4 0
0 0
o m
Nl - 4
l-4
0
N
4 0 - 4 c o c o
NNl - 4l - 4 0 0
. . . .
0 0 0 0
. .
0 0
0 N
CD d
N * * O h v m
N 4 d N 0 0 0
0 0 0 0 0 0 0
. . . . . . .
E
I E
I
~ ala
86
I n m c R o 00
4 0 0 w I n 4
1 o c u m m m
. . . .
4 0 h O O c n v ) o w
N O - ' m O m m w w
1 o T r m - P T r m T r m w
. . . . . . . . .
T r c o c o C n O d 4 C r l m o
d 4 4 d w N m C V m m
0 0 0 0 0 0 0 0 0 0
. . . . . . . . I .
c R c o c o 0 3 0 ~ w o
W T r C O h W 4 - 3 4
1 1 0 T r m m m v ) c n c o
. . . . . . . .
T r w a w T r v ) T r c \ 1 0 0 3
n I m m m m P 3 T r - P T r T r
0 0 0 0 0 0 0 0 0 0
. . . . . . . . . .
cu 0 0
N 0 cu
In 0 4
4 m Tr
87
TABLE 7
COMPARISON POSSIBILITIES OF GEOMETRIES OF INVESTIGATE1
LAMD WIDTH
I
I ,
COMPARISON
BETW E& N
SHAFT NOS.
( 6) & ( 7 )
NUMBER
OF
GROOVES
1; 4
6; 10
10
1
3
6
10
10
3;4
Y=0.79 8=6.89
t= 0.099
Y = 0.79 t 0.099
Y =0.593 t =0.099
YzO.65 t=O.117
Y = 0.643 t =0. 268
Y = 0.645 t=0.431
Px 6 . 8 9 t =0.099
Yz 0. 593 y -0.79
88
1 2
% wetted shaft
Table 8. Loss of Sealinq fluid for Shaft No. 4
Loss of
Fluid
Volume of the sh
Upper
67.9
52.2
60.1
59.0
56.8
56.8
100
96.6
92.2
89.0
Lower ml
17.8
11.1
6.7
3.9
0
0 -1.81
11
6.7
3.3
0 -1.38
0
6
10
Et: about 0.07 ml/mm
i
0 0
-0.17 -0.29
-0.49 -0.83
15
26
50
0
6
9
18
0
4
6
12
-0.63 -1.07
-0.45 -0.77
-0.80 -1.36
P = 1. 2
kg/cm2
-0.45 -0.54
-0.63 -0.76
-0.89 -1.07
0 0
P = 1.4
kg/cm2
0 0
-0.36 -0.50
-0.36 -0.50
-0.63 -0.88
100
75.2
66.7
66.7
Tellus 13
N=1522 rpm
no visible
l eak.
27.8
5.6
0
0 -3.78
Sealing fluid No. 3
N=1082 rpm
no vi si bl e leak.
Sealing fluid No. 3
N=4 120 rpm vi si bl e
l eak.
89
I
Sealing Fluid
Motor
-Glass
1
b
FIG. 1
90
91
92
Re = 13
1
= 27
Re 1
Re = 19
1
INTERFACES ON THE VISCO-SHAFT
n GAS
LIQUID FILM
KT7 LIQUID
Re = 51
1
Rel = 23
Re = 193
1
SCALE 1:l
FIG. 7
93
0
94
0
0
II
z
e
c
P
t
<
I
v)
v)
z
I-
<
0, <
&I
<
I &=
I-
W
0
0
I
I
1
0
0 1
0
0
0
p!
II
f
p O j a=
I
I,
t
c
N
c
11
00
7
(Y
0
6,
95
w
x
F-l
96
CHANNEL CLOSED
RADIAL CHANNEL
UNDER LAND =1 MM
I N THI S GROOVE TWO
RADIAL CHANNELS
SHAFT TYPE IE
FIG.11
97
RADI AL CHANNEL 01 CHANNEL CLOSED
&
Cr )
X
Y
I N THIS GROOVE, A
RADIAL CHANNEL +15
C
SHAFT TYPE II
FIG. 12
98
4 1.0
2
A B
E
CENTER BUSHI NGS
OF VISCOSHAFT
C D
UPPER BUSHI NG
CENTER BUSHI NG -
/
LOWER BUSHI NG
SCALE 1:l
SCHEME OF DOUBLE
VI SC 0 SHAFT
FIG. 14 FIG. 13
99
0 THEORETICAZ. ACCORDI NG TO
(14)
A MEASURED WI THOUT GAS
-
- SHAFT
- NO. 2
I NCLUSI ON
SI ON
0 MEASURED WI TH GAS INCLU-
0 MEASURED WI THOUT GAS
I NCLUSI ON WI TH RECI RCU-
LATION Q, =1.5
1 1 1 I I I I l l
2 4
6 8 lo3
2
- 0 THEORETICAL ACCORDI NG TO (14)
- NO. 6
SI ON
-
0 MEASURED WI TH GAS I NCLUSI ON
-
SION WI TH RECIRCULATION
SHAFT D MEASURED WI THOUT GAS I NCLU-
0 MEASURED WI THOUT GAS I NCLU-
=1.5
I ,, I I I I 1 I l l
a
fi' IO2 2 4 6
FIG. 15
FIGS6
100
FIG. 17
101
\
0
I i I I 1 I I i
0 u 7 N c
-
' S
3 - FIG. 18
102
10
8
6 o MEASURED BY US
X MEASUREMENTS OF
VRAKKING
2
1
8
6 THEORETICAL CURVE
ACCORDING TO EQUATION
(26) REF, 2
4
2
-1
10 2 4 6 ~ 8 104 2 4 6 8 2
- ~~~
Rei-
lo2
4 - - THEORETICAL CURVE
ACCORDING TO EQUATION
(21) REF. 2
d
2 -
I I I I I I l l I I I I I I I I I I I I I I l l
2 4 6 8 103 2
4 6 I 2 4 @ I
16
lo2
FIG. 19
Defense Document at i on' Center
Cameron Stati on
Al exandri a, V i rgi ni a 22314
M r , P. H. Broussard
Marshal l Space Fl i ght Center
Huntsvi l l e, Alabama 35812
M r . J ohn J . Gurtowski
Naval A i r Systems Command
Code A I R 52032C - Dept. of the Navy
Washi ngton, D. C. 20360
Mr, Robert L . J ohnson
Chi ef , L ubri cati on Research Branch
NASA, Lewis Research Center
Cl evel and, Ohi o 44135
Mr. W. C. Karl
Of f i ce of Naval Research, Code 473
Department of the Navy
Washi ngton, D. C. 20360
M r . Lawrence P. Ludwig
Head, Seal s Secti on
NASA, Lewis Research Center
Cl evel and, Ohi o 44135
M r . J oseph Mal tz
Materi al s Research Program
Nati onal Aeronauti cs & Space Adm.
Washi ngton, D. C. 20546
Col . W. Metscher
Of f i ce of Di rector of Defense Research
and Engi neeri ng
Room 3C128, Pentagon
Washi ngton, D. C. 20301
M r . E. R. Tayl or
React or Engi neeri ng Di vi s i on
Oak Ridge Nati onal L aboratory
Oak Ri dge, Tennessee 37830
Y r . Marshal l J . Armstrong, J r.
U. S. Army Mobi l i ty Equipment Research
and Development Center
Fort Bel voi r, V i rgi ni a 22060
Y r . Ri chard N . Bel t
U. S. Army Mobi l i ty Equipment Research
and Development Center
Fort Bel voi r, V i rgi ni a 22060
NASA, R-ASTR-GC
Mr. Fl oyd Lux
U . S. Tank Automoti ve Center
Propul si on Systems L abpratory, SMOTA-RCP.4
Warren, Mi chi gan 48090
M r . J . R, Crowder
Fl i ght Vechani cs & Fl ui d Systems Secti on
(AIK 5303)
Naval A i r Systems Command - Dept. of the Navy
Washi ngton, D. C. 20360
Mr. Roy R. Peterson
Naval Shi p Systems Command (Code 03413)
Department of the Navy
Washi ngton, D. C. 20360
DISTRIBUTION LIST
- A-
Dr . Earl Quandt
Naval Shi p Research & Devel opment Center
Annapol i s Di vi si o
Annapol i s, Mary1
Mr. Al an Schrader
Naval Shi p Research & Devel opment Center
Ann apo 1 i s D i vi s i on
Annapol i s, Maryl and 21402
M r . Lyman Carl yl e Fi sher
N aval O r dn ance Lab or at o ry
Department F (Code S10) - Whi te Oak
Si l ver Spri ng, Maryl and 20910
M r . Stanl ey Dorof f
Of f i ce of Naval Research (Code 438)
Department of the Navy
Washi ngton, D. C. 20360
Mr. J ohn W. Zmurk
A i r Force Aero Propul si on L aboratory (API P- 1)
Wri ght- Patterson A i r Force Base, Ohi o 45433
Capt. Kenneth R. Hooks
A i r Force Weapons L aboratory WLDC
K i rtl and A i r Force Base, New Mexico 87117
D r . J oseph F. Masi
A i r Force Of f i ce of Sci enti f i c Research, SREP
1400 Wi l son Boul evard
Arl i ngton, V i rgi ni a 22209
Mr. Frank J . Mol l ura
Rome A i r Development Center, EMEAM
Gri f f i ss A i r Force Base, New York 13440
Mr. Geral d S. L ei ghton
SEPO - Di vi si on of Space Nucl eqr Systems
U . S . Atomic Energy Commission
Washi ngton, D. C. 20545
M r . C. E; . Mi l l er, Jr.
Di vi si on Reactor Development 6 Technol ogy
U. S. Atomic Energy Commission
Washi ngton, D. C. 20545
Bureau of Naval Personnel
Department of the Navy
Washi ngton, D. C. 20370
Attn: Techni cal L i brary
Naval A i r Systems Command'
Department of the Navy
Washi ngton, D. C. 20360
Attn: Techni cal L i brary Di vi si on, AI R-604
Naval Appl i ed Sci ence L aboratory
Fl ushi ng and Washi ngton Avenues
Brookl yn, New York 11251
Attn: Techni cal L i brary, Code 222
U, S , Naval Oceanographi c Of f i ce
Sui tl and, Maryl and 20390
Attn: L i brary, Code 1640
U . S. Naval Postgraduate School
Monterey, Cal i f orni a 93940
Attn: L i brary, Code 0212
DISTRIBUTION LIST (Cont'd)
Naval Shi p Engi neeri ng Center
Phi l ade l ph i a D i vi s i on
Phi l adel phi a, Pennsyl vani a 19 112
Attn: Techni cal L i brary
Naval Shi p Research 4 Development Center
Ann apo 1 i s D i vi s i on
Annapol i s, Maryl and 21402
Attn: L i brary, Code A214
Naval Shi p Systems Command
Room 1532 Main Navy
Washi ngton, D. C. 20360
Attn: Techni cal L i brary
U. S. Atomic Energy Commission
Argonne Nati onal L aboratory
9700 South Cass Avenue
Argonne, I l l i noi s 60440
Attn: L i brary
Nati onal Bureau of Standards
Washi ngton, D. C. 20025
Attn: L i brary
Battel l e Memorial I nsti tute
505 King Avenue
Columbus, Ohio 43201
Attn: L i brary
Y assachusetts I nsti tute of Technology
Cambri dge, Massachusetts 02139
Attn: L i brary
Naval Undersea lVarfare Center
3202 East Foothi l l Boul evard
Pasadena, Cal i f orni a 91 10 7
Attn: Techni cal L i brary
Navy Underwater Sound L aboratory
Fort Trumbul l
New London, Connecti cut 06320
Attn: Techni cal L i brary
U. S. Naval Weapons L aboratory
Dahl gren, Vi rgi ni a 22448
Attn: Techni cal L i brary
Power I nf ormati on Center
Uni versi ty Ci ty Sci ence I nsti tute
3401 garket Str eet, Room 2107
Phi l ade l phi a, Pennsyl vani a 19 10 4
Ae ro j et - Gene r a1 Co rp orat i on
Von Karman ,Center
Azusa, Cal i f orni a 91702
Attn: L i brary
Aero j e t - General Nucl eoni cs
San Ramon, Cal i f orni a 94583
Attn : L i brary
Uni versi ty of Ari zona
Department of Aerospace and
Y echani cal Engi neeri ng
Tucson, Ari zona 85721
Attn: Prof essor Kececi ogl u
A i Res e arch Manufacturi ng Company
9851 Sepul veda Boul evard
Los Angel es, Cal i f orni a 90045
Attn: L i brary
Ai Research Manufacturi ng Company
402 South 36th Street
Phoeni x, Ari zona 85034
Attn: L i brary
General El ectri c Company
Fl i ght Propul si on Di vi si on
Ci nci nnati , Ohi o 45215
Attn: L i brary
General El ectri c Company
?.I e c h an i c a 1 T e c hn o 1 o g y L ah o r at o r y
Research and Development Center
Schenectady, New York 12301
Attn: L i brary
-B-
UNCUS SIFIED
R. J. G. Boon Translation by J. P. B. Vreeburg
ME 69-T57-4
96. O T H E R R E P O R T Not51 (Any other numbers that may be as s f gne d
t hi s report)
6. P R O J E C T N O .
This document has been approved for public release and sale; its distribution
is unlim ited .
1 1 . S U P P L E M E N T A R Y N O T E S 12 S P ON S OR I N G M I L I T A R Y A C T I V I T Y
A solution is sought for the l eakage and break down phenomena as encountered by
many researchers by means of vi sual observations on a vi scoshaft seal in a transparent
hou sing.
Our observations are compared with the phenomena mentioned in the literature.
An effort is made, notwithstanding gas inclusion, to prevent l eakage in vi scoshaft
seals .
We tried to reduce gas inclusion by changing the surface energy of the seal.
The influence of the geometry of the seal on gas inclusion and l eakage has been
studied .
The influence of gas inclusion on the val ue of the seal i ng parameter has been
looked into.
The measurements of the double V.S.S. in the Lab.of Chem. I nstr. have been
I checked because it could be expected that gas inclusion had affected the measured
1473 (PAGE 1 )
DD 1 N O V 6 5 UNCLASSIFIED
S/N 0 10 1 - 807- 680 1 Securi ty Cl assi f i cat i on
-
R OL E
-
W T
-
111
L I NK C
__.
R OL E
-
-
W T
_s