Note: Descriptions are shown in the official language in which they were submitted.
~ 3 ~ 0
The presen-t i.nvention relates to separating
dev:ices and methods, and more particularly to separa-ting
devices and me-thods for gas turbines in environments having
rnoisture par-ticles in the air such as gas turbines for
marine applications. For e~ample, the method of the present
inven-tion is particularly useful for removing moisture
particles entrained in air entering the air in-take of
a ~as turbine for ships.
Moisture separators are provided for gas turbines
for marine applications as the moisture particles in the
air generally contain sal-t which, if it should be in-
troduced into the turbine, would deleteriously affect the
component parts of the turbine, as for example, by chemical
corrosion. Further, the dry particles entrained in the air,
for example sand and/or salt crystals, can cause "pitting"
of the -turbine components if they are not removed. However,
by far the greatest concern is the moisture particles
containing salt.
One early prio~ ar-t device for removing moi~ture
and other particles from the air introduced into -the tur-
bine comprised a single stage wire mesh pad placed in the
air intake duct. The mesh pad was generally comprised of a
plurality of layers forming an approximately two inch
thick pad wi-th each oE the layers in turn comprised of
a plurality of .006 inch diame-ter wires knitted into a
grid or screen having approximately five to six stitches
per inch, i.e., there were approximately five wires per
inch of length stitchecl or joined to~ether. The mois-ture
particles were removed as a result of the par-ticles impact-
ing acJainst and being cap-tured by the wires of the pad
1 3 6 ~
as -the air passed -through the device. Howe~er, such a
wire mesh pad did not adequately and efEiciently remove
all sizes of moisture particles entrained in -the air and
thus, was not acceptable.
Another prior art device for removing moisture
particles which was found to be more acceptable, comprised
a three stage separa-tor in which the first and -third stages
were iner-tia separating vanes and the second stage was
a coalescer. The first and third stage vanes had generally
similar performance characteristics, one o~ which was
tha-t they exhibited very poor efficiency in removing
smaller sized droplets, namely, eigh-t microns and below.
The reason for this is simply that inertia separa-tlng
devices work on the principle that as -the air flow turns to
bypass the vanes or other impacting medium the moisture
- particles, being of larger size arld more mass t cannot make
the turns and instead impinge or impact upon the impacting
medium, thereby being removed from the air stream. However~
these conventiorlal inertia vanes do not prove efficient to
remo~e lower size droplet particles since such particles,
being of lighter mass, can make the turns to avoid impact-
ing on the vanes. ~s a consequence, the prior art employed
a second stage filter comprised generally of a plurality of
polyester fibers of .001 inch diameter or smaller. The duty
of the second stage was to cap-ture -the fine droplets by
inertial impaction that had passed through the first stage.
Owing to their smaller diameter, the capture efficiency of
the fibers was high so that some of the f:ibers would tend
to collec-t several droplets which then coalesced or grouped
together untll a drople-t was formed which was large enough
6 0
--3---
-to be re-en-tralned by the aerodynami.c drag forces of the
air passing therethrough. These re-entrained droplets were
then captured by the third stage inertial device (either
vanes or cyclones).
However, the a~dition of the third stage added
grea-tly to -the overall size and weight of the moisture
- separator, which already was quite large in order to
aci~ieve the high mass flow rates needed by the gas turbines
for ships (on the order of 2000 cubic feet per second).
Further still, the use of a polyester fiber pad having a
plurality of closely spaced and dense fibers, each of a
small diameter (on the order of .001 of an inch or
smaller), tended to increase the flow resistance, and thus
the pressure drop across the separator, for a given
velocity of air flow. Thus, to achieve the desired flow
rate, -the size of the polyester pad and -thus the size of
the air duct in which the moisture separator was supported
also had to be increased, thereby further adding to the -
weight, bulkiness, cost, etc. of the separating device.
It should be noted in this connection that
various types of other filtering or separating arrangements
and, in particular filtering pad arrangements, not speci-
fically directed for use in marine applications are known
: in the prior art. For example, in U.S. Patent No. 4,086,070
to Argo et al, -there is di.sclosed a fiber bed separator and
method for separation of aerosols from gases without re-
entrainment. In the improved fiber bed separator of Argo
et al., ~here is provided a pair of fiber beds each com~
prised of a plural.ity of fiber elements or wires. In this
reference, the distinction is made between the amount
3 ~ 0
of wa-ter which is held by the bed after draining by gravity
and that held against the drag of flowing gas or air there-
through. More part:icularly, the fiber diameter and bed
~oidage for -the first or front bed in Argo et al. is
selected such that at the design bed velocity and aerosol
loading, the first bed will not be flooded and the ~sidual
saturation thereof against gas drag on the liquid collected
will be less than the residual saturation of the bed
against gravity drainage of the liquid. In other words, the
front or first bed of Eibers in Argo et al. is such that
the collected liquid in the bed tha-t does not drain by
gravity will be blown through to -the second ~iber bed. In
the second fiber bed on the other hand, the fiber diameter
and bed voidage are chosen so that the residual saturation
thereof against gas drag will be greater than the residual
saturation of the bed against gravity drainage so that
the liquid collected by the second bed will drain of~
by gravity as oppo~ed to being re-entrained.
Thus, in accordance with -the ~rgo et al teachi~,
a more dense fiber pad is used as a first stage i~ a filter-
ing device in which -the residual saturation against gravity
drainage is greater than the residual saturation against
gas phase drag so tha-t liquid which does not drain from
the first stage will be blown through to a second pad
where the residual saturation against gas phase drag is
greater than the residual saturation against gravit~
drainage so that liquid wi]l drain off rather than be
re-entrained in the air. Accordingly, the Argo et al solu-
tion to the problem of re-entrainment experienced in the
prior art is to provide a second pad ha~ing specific charac-
teristics chosen so that the pad has a greater tendency
J 1 ~3~(~
-to dra:in Gaptured or collected l:Lquid by means o:f: gravity
than to permit re-entrainment.
However, upon a deta:il.ed examination of the
Argo et al system, it is seen that the flow veloci-ties
through the fiber pads w:ith which the Argo et al ar-
rangement is concerned are relatively low in comparison to
the flow velocities through filtering arrangement used in
connection with marine applications; i.e., the Argo et al
arrangement is specifically designed to be utilized in
connec-tion wi-th flow veloci-ties which are generally less
than 10 feet per second, whereas in marine applications
flow velocities of greater than 20 feet per second are expe-
rienced. Furthermore, it is to be noted that ~rom an
: examination of the Argo et al reference, in particular
Figures 1 and 7, Argo et al suggests that the residual
saturation of the second fiber bed against gas phase drag
in virtually all instances decreases as the velocity
through the bed increases, and that the upper limit of
velocity through the bed is usually set hy re-entrainment
problems. This may be due to the fact that the Argo et al
reference teaches that the particles captured by the fiber
beds coalesce and drain by gravity.
In this regard, it is to be noted that re-entrain-
ment of any collected or coalesced liquid droplets would
require an additional device downs-tream of the filter
pad for the purpose of recapturing the re-entrained
particles in order to prevent the introduction of such
moisture particles which are re-entrained into -the turbine.
An~ther type of arrangement of the prior art~
again not specifically directed for use as a moisture
separator in marine applica-tions, is disclosed :in U.S.
J ~ ~3~0
-6-
Paten-t No. 4,158,~-~49 to Sun et al. This patent discloses
an in]et air cleaner assembly for -turbine engines which
are mainly used in connection with agricultural aircraft
for the application of chemicals over large and not readily
accessible land areas. In the arrangement of -this
reference, there is provided a first stage vortex air
cleaner and a second stage mist eliminat,or assemb-y. The
first stage vortex air cleaner operates generally on the
principles of inertia separating devices for removing
heavier and larger sized contaminate particles in the air
to provide partially processed air. The second stage of the
filtering arrangement comprises a plurality of superimposed
knitted wire mesh sheets which define passages therethroug~
for removing light-weight or well disperse~ solid contami-
nate particles and liquid drople-ts. The wire mesh sheets
are comprised of a plurality of monofiliments which may
have a diame~er ranging from .0005 inches to 0.035 inches
and which are compressed substantially throughout the
surface area. The monofiliments preferably are coated with
oil or another nonvolatile liquid so that liquid which is
removed by impingement on the oil coated wires of the
mesh will tend to flow by gravity downwardly and collect
in the bottom of the mist eliminator stage for subsequent
removal.
Thus, it is apparent that even in this reference
which is not mainly concerned or directed to a separator
assembly for marine applications, there is the sugges-tion
that liquid which is removed by the mist eliminator as-
sembly coalesces and drains by gravity to a lower portîon
of the assembly. As such the use of this asse~bly in a
~ ~ 6~3~0
marlne appllca-t:ion where rnois-ture particles are to be
removed, would sugges-t -that a further separating device is
required do~nstream thereo~ for capturing any coalesced
particles which become re-entrained. That is, because of
- the suggestion of coalescence occurring~ it would be
expected, especially at the particularly hi.gh ~low
velocities with which separating devices for marine applica-
tions are concerned that -there is the possibili-ty of the
coalesced particles becoming re-entrained in the air
stream. Thus, it would be expected that such re-entrainment
o* coalesced particles would create a problem, such as
experienced in connection with some of the other prior art
arrangements discussed hereinabove if the Sun et al arrange-
ment were used, and thus necessitate the use of additional
separator devices or means downstream of the filtering pad,.
In this regard, as noted above, coalescense would create a
problem of re-entrainment at the relati~ely high flow
veloci-ties such as experienced in connection with marine
applications since the aerodynamic drag forces produced
thereby have a greater tendency to cause re-entrainment of
the enlarged coalesced droplets.
According to the presen-t invention, there is
pro~ided a method of removing particles entrained in the
air, the air including particles of moisture, the method
comprising the steps of: passing the air at a velocity
greater than 20 feet per second through an inertia separat-
ing means for inertially removing a-t least a portion o~
the larger sized particles from the air to provide par
tially processed air; and passi~g the part~ally processed
air at a velocity which is greater than a predetermined
velocity -through an impact fil-tering pad for removing
o
particles entrained in said partially processed air, said
impact filtering pad compri~ing at least one layer of a
plurali-ty of fibers, each of said fibers having a diameter
greater -than .001 :inches and less -than .006 inches, and the
ratio of total surface area of said fibers in said pad to
the volume of said pad being greater than ~5 ft. l and less
than 1400 ft. l, and said predetermine~ ~elo~ity being
greater than 20 feet per second and chosen accordins~ to
the diameter of said fibers of said impact filtering pad
so that coalescence of moisture particles captured by
said impact filtering pad is minimized.
In order that the invention may be fully under-
stood, it will now be described with reference to the
accompanying drawings in which:
Figure 1 is a perspective view, partially broken
away, of a separa-tor device employed in accordance with
the presen-t invention as a moisture separa-tor for the
air intake to a gas turbine on a ship.
Figure 2 is a partial sectional view of the
moisture separator illus-trated in Figure 1, taken along
lines 2-2 of Figure 1.
Figure 3a is a plan view of a portion of one
layer of the impact filtering pad of the present invention.
Figure 3b is a side view taken along lines 3b-3b
of Figure 3a.
Figure 4 is a graphic representation illustrating
the coalescense of moisture particles for -the impact filter-
ing pad as a function of velocity through the pad and
the fiber diameter of fibers comprising the pad,
Turning now to the drawings in which like
reference characters represent like elemen-ts, there is
l :~ B~360
-9
shown -in Figure 1 a mois-ture separating device 10 in accor-
dance w:ith the presen-t invention in which the moi.s-ture
separator 10 i5 p:Laced in the air inta~e duc-t 12 o~ a
gas -turbine (not shown). As the separa-ting device 10 in
accordance with the present invention is particularly
use~ul ~or removing moisture par-ticles entrained in the
air in which the moisture particles contain salt, the
presen-t invention is particularly useful for use with
gas turbines on ships or other amphibious vehicles. It
should also be understood that -the separating device of
the present inven-tion is also useful in removing dry
particles entrained in the air, such as for example, sand,
salt crystals, etc. which also have a tendency to damage
turbine components. However, as -the separating device
is primarily intended to remove moisture par-ticles e~-
trained in air, the present invention will be describad
with re~erence to such application.
It should be noted here that it is not the mois-
ture per se which is undesirable but rather con-taminate
particles such as salt and the like which are dissol~ed
in or carried by the moisture particles which are un-
desirable. However, as can be appreciated, the elimina-
tlon.of the moisture particles from the air will also
serve to eliminate these undesirable con-tamina-ted particles.
In gas turbines or marine applications, the
air ducts are placed high up in the ship or the vehicle
so that as lit-tle moisture as possible might be entrained
in the air being in-troduced into the gas turbine, and
in such instance, the moisture separators would be mounted
at the entrance o~ the air duct. The re~ui.red air flow
to be introduced to the -turbine is on th~ order of
.
J :~ ~43~0
- --10-
~agnitude of 2000 cubic feet per second and greater. The
mass fLow ra-te in turn is dependen-t upon the cross-
sec-tional area of a duct through which the air is in-
troduced and -the veloci-ty of the flow -therethrough For
example, if the flow velocity through the duct can be
increased, the cross-sectional.area of the duct (and thus
the size of the duct) can be reduced subs-tantially. On the
other hand, -the pressure drop across the moisture separator
placed in the duct must be maintained at an acceptable
level. As the pressure drop across the moisture separator
is dependent on both the flow resistance pre~en~ed by the
moisture separator and also the velocity ~-E the flow there-
through, the pressure drop thus serves as a limi-t on the
increase in flow velocity which can be obtained -to satisfy
. the requirements for a given mass flow rate. Therefore, as
- it is desirable to decrease the size of the air cl~ct ~to
save on costs, weight, etc.), it is desirable to design
the moisture separator so as to have as low a flow resis-
tance as possible. At the same time, the reduction in
flow resistance presented by the moisture separator must
not be such as to impair the effici.ency to remove the
molsture particles from the ai.r passing through -the
separator.
With these principles in mind, the moisture
separator in accordance with the present invention will
now be discussed. As shown in Figure 1, the moisture
separator 10 comprises a two stage separating device.
The first stage 14 of the separating device 10 comprises
a conventional inertia device through which the air flow
to be introduced to the turbine first passes. This inertia
~ :~ 6 ~
-Ll-
device, in the preferred ernbodiment, co~prises, as is
conventional, a plurali-ty of chevron or V-shaped vanes
16 vertically oriented and closely spaced to one another.
However, it should, of course, be understood that other
types of inertia devices could be used, as for example,
cyclone separators. The plurality of vanes 16 are supported
in a support frame 17 which holds the uppe~ and lower
ends of the vanes 16 in fixed position. ~ur-ther, a drain
for mois-ture removed from the air is generally provided
in the bottom of -the first stage 14, althou~h it has not
been shown.
As best seen in Figure 2, the ~anes 16 of the
first stage inertial device 14 each have three peaked
sections 18. The peaked sections 18 of adjacent vanes
16 thus define a tortuous path for the air ~lowing there-
-through. In o-ther words, as the air flows t~rough the
vanes 16, it must turn or bend several ~imes to follow
the path be-tween the peaked sections 1~ of adjacent vanes
16. As the entrained particles, such as f~r examp7e, mois-
ture particles containing salt, sand or other par~icularmatter, are generally of a larger mass than the air
particles, the en-trained particles are thrown outwardly
during the turns against the surfaces of the vanes 16
due to the centrifical force exerted thereon. That is 7
the lighter air particles are capable o~` making the turns
through the series of peaked sections 1~, whereas the
heavier mass particles are not, thereby resul-ting in the
larger particles impacting on the surface of the vanes
16. Each of the peaks 20 of the vanes 16 in~clude st~ps
Z2 for preventing the impac-ted par-ticles from sliding
along -the surface of the vanes 16 and becoming re-entrained
`J 1 ~
-12
by virtLle of the aerodynamic drag force exerted by the
air flow.
As can be apprecia-ted, the heavier the particles
entrai.ned in the air, -the more likely such particles will
impac-t on the vanes 16 and thus be removed from the air
10w. On the other hand, lighter particles may have a
tendency to successfully follow the flow path and remain
entrained in the air. For example, inertia separa-tor
devices such as the stage 14 have been found to be effi-
cient in removing moisture particles of a size over 8microns (25.4 microns approximately equals .OOl inches~ in
.the range of the high flow velocities with which moisture
separators for gas turbines are concerned, i.e., greater
than 20 -feet per second. However, the plurality of vanes 16
have not exhibited a high efficiency for removing moisture
particles in the lower size droplet range, nar~ely 8 microns
and below. Thus, it is necessary to utilize further stages
or devices for removing such lower sized particles.
Ater being partially processed by the first
stage 14 of the moisture separator 10, the air is in-
troduced through the second stage 24 which serves to
further process the air to remove particles still en-trained
therein. As best seen in Figures 2, 3a and 3b, the second
stage of the separati.ng device comprises an impact separat-
ing pad 26 comprised o at leas-t one layer 28 of a
plurality of fibers 30. In the embodiment shown in Figures
2, 3a and 3b, the pad 26 is actually comprised of a
plurality of layers 28 of kni-tted wettable fibers 30. The
layers 28 of pad 26 are supported in a frame 32 which is
afEixed to the downstream side of the firs-t sta~e 14.
~ :~ fi~6~)
In the preEerrecl embodiment, the fibers 30 forrn-
ing the l,ayers 28 of the pad 26 are comprised of monel
wire. Monel is an alloy oon-tainirlg primarily nickel and
copper, and other elements, and has been found to be
extremely useful in marine type applications since it is
especially resistant -to corrosion by salt entrained in the
mois-ture of sea water.
As best seen in Figure 3a which is a plan view
of a portion of one layer of the pad 26, the wire fibers
30 have been knitted -together to form a mesh or grid havin~
the wires 30 joined or tied together. Preferably, the
diameter of the fibers or wires 30 is on the order of
.002 inches, but the diameter can range be-tween .001 inches
to .006 inches. With .002 inch diame-ter wire, it has been
found with conventional knitting apparatus that there
are approximately 10 to 12 stitches or joints per inch
of mesh. That is, in a one inch length of the wire mesh
28, there are appro~imately lO -to 12 wires 30.
Also in the preferred embodiment, during the
knitting operation, the wires 30 are crimped or ~ent
slightly so that each of the layers 28 does not lie com-
pletely flat. This is shown in Figure 3b. Thus, when the
layers 28 are combined -to form the pad 26, the layers
28 will not lie flat against one another bu-t ins-tead will
provide a cushioning or "lofty" medium. In the preferred
embodiment, the crimp is sufficient to allow no more than
20 or 24 layers per inch of depth of -the pad 26. In this
regard, it is to be noted that without -the crimping of
the wires 30, it has been found tha-t there would be ap-
proxima-tely 70 to 75 layers 28 of wires 30 per inch of
pad depth. The reason that crimpin~ of the wires 30 is
3 ~ ()
desirable i.s that for a given number of layers 2~, the
res:i.stance -to flow (which is propor-tional to the pressure
drop across -the pad) wi.11 be lower if the distance between
adjacent layers 28 is increased.
It should be notecl tha-t wire fibers 30 have been
~ound to be preferable to ~orm the layers 28 since the
wires 30 will retain the crimp which is provided during the
kni-tting or stitching opera-tion to form the layers 28.
Further, wires 30 provide a wettable surface which is most
useful in capturing -the particles of mois-ture, which impact
on the wires 30 as the air flows thereacross.
The mechanism by which the wire fiber pad 26
serves to remove particles from the air is inertial impac-
tion. As the air "weaves" its way through the -thickness
of pad 26, the particles entrained in the air (for example
particles of moisture.containing salt, particles of sand,
etc.) are not able to avoid the wires 30 and thus impact
on the wires 30. With moisture particles, the surface
tension or adhesion between the metallic wires 30 and
the moisture particles which impact thereon is grea-t and
thus, -the moisture particles are captured and removed
from the stream of air. This would also be -true with
respect to o-ther types of particulate matter en-trained in
the air, such as for example, salt crystals ~produced when
water of sea water.evaporates) or sand, but it is to be
noted that the adhesion or surface tension between such
par-ticles an~ the surface of the wires 30 is much less
than with moisture particles. However, -this is acceptable
since the ~reatest concern is the sea water containin~
~0 salt.
- It should be noted that the smaller -the slze
of the fibers 30, the more effi.cient the fibers 30 are
3 ~ !)
--15-
in capturing f:ine mois-tl.lre particles en-trairled in -the
air. That is, small cl:iameter fibers have a relatively
high captu~e effici.ency by impaction. However, with too
small of a ~iber size, the chances are significantly in-
creased that several drople-ts of moisture will be captured
by the same fiber. When this occurs, the small droplets
tend to coalesce together to form large droplets, such
as for example, over 8 microns in size. Since the aero-
dynamic drag force wi].l be greater on these coalesced
droplets, there is a good chance that they will be re-
entrained in the air.
This is the reason why, wi-th prior art separa-ting
devices which used a fiberous pad having fiber ~iameters
of .001 inch or smaller, it was necessary to provide a
third stage for capturing the large coalesced droplets.
While it has been found -that the degree of coalescence
of smaller size par-ticles, say in the range of 2 to ~
microns growing to 10 microns in size and above, is not
great ~on the order of 3 to 5%~, it is significant enough
that a third inertial impac-ting stage, such as for example,
vanes or cyclones, was necessary to remove such coalesced
particles from the air. However, in the present invention,
this problem of coalescing is avoided b-y using a slightly
larger size fiber 30, preferably having a diameter of
.002 inches.
With respect to the various parameters for design-
ing mois-ture separating devices for gas turbines, it is
to be noted that the efficiency of a pad 26 for removing
a particular size par-ticle from the air is expressed by
the well-known rela-tionship for capture in a lattice ar-
rangement:
~ 3 ~i~3~0
r~p = 1 _ e--2l2~f~t
where:
np is -the eff-iciency for removing a particular
size particle;
nf is the efficiency o-f a sing~Le ~fiber :For remov-
ing such particular size particle;
~ is the ratio of the total sur*ace area o-
the fibers o.E.the pad to the volume of the pad; a~d
- t is the thickness of the pad.
Thus, as it is desirable to maintain -the efficiency as
close to one as possible, it is desirable to increase
-the product ~t. On the o-ther hand, the pressure drop
across the pad is rela-ted to the quan-tity ~2-t. Thus, in-
creasing ~ to too great a quantity, will cause the pressure
to drop across the pad to increase, which as noted above,
is undesirable. On the other hand, it should be n~ted that
the -thickness of the pad 26 is one dimension which also
goes into determining the volume of the pad. Thus~ the
efec-t on the pressure drop by simply increasing the pad
20 thickness (but maintaining the number and ~ize of the wires
30 the same) actually serves to decrease the pressure drop.
In accordance with the.present invention, the
values for ~ should range be-tween 45 ft. 1 to approximately
1400 ft~ 1. In this way, the efficiency can be maximized
while at the ~ame tlme the pressure drop maintained at an
acceptable level. More preferably, the value for ~ should
range between 75 and 200 ft. 1, Also, it has been found
desirable to maintain the thickness o:E the pad 26 between
1/2 inch and five inches, although of cour~e, a lesser
thickness, e~en down to a single layer 28 of wire mesh,
3 ~ ~
-17-
could be utiLi~ed depending on the desired efficiency
characteris-tics ancl pressure ~Irop characteristics for a
par-ticular application. In the preferred embodiment, the
thickness of the pad 26 is appro~imately three inches
and comprises on the order of 70 layers 2~ of the .002
inch diame-ter wire 30 ~laving lO to 12 s-ti-tches per inch.
For such a pad 26, the value for ~ is approximately 84
ft. . If the pad thickness is compressed to I l/2 inches,
the value for ~ would double to approxima-tely 168 -ft.
It is to be noted that wi-th respect to the
polyes-ter pads utilized in the prior art, the value for
~ was on the order of 1800 ft. . Accordingly, for a given
thickness of pad, the pressure drop across a polyester
fiber pad of the prior art is significantly higher than
is across the pad 26 according to the presen-t invention.
Therefore, -the velocity of the air flow through -the pad
26 of the present inven-tion can be increased over that
which was possible with the polyester pad of the prior
art. Thus, the siæe and weight of the moisture separator
lO can be substantially reduced, while at the sam~ time
maintaining an accep-table pressure drop level. Furthermore,
since higher velocities for the air flow through the
separator 20 can be attained with the pad 26 o-f the present
invention, the particle separa-tion efficiency is also
increased. This is the result of the fact tha-t droplet
retention by impaction improves with an increase in
velocity.
Moreover, owing to the larger diameter of the
fibers over that of -the polyes-ter pad, the tendency of
the drops to coalesce is elimina-ted. The reason ~or this
is twofold. First, as the air velocity is increased, the
~ 1 6~ 3~ 0
~18-
ability of drople-ts to coalesce is impaired. Secondly,
the aerodynamic drag forces a-t higher velocities is larger,
and thus, the drople-ts are re-entrained in -the a;r at
lower sizes thereby pre-venting the grow-th of droplets
by coalescing to larger sizes, say in the 1~ to 13 micron
range and abo~e. The re-entrained droplets can then be
captured by another fiber. Since the moisture droplets
are not coalesced, -there is no fur-ther need for a third
stage inertia device in the moisture separator 10, which
in the prior ar-t was only useful for recap-turing the
coalesced rnoisture droplets. The elimination of the need
for a -third stage filter results in substantial savings,
not only in cost, but also in the weight and size of the
moisture separator 10.
An important point to note in this regard is
-the fact that for any given wire or fiber size in the
impact filtering pad 26, there is a minimum flow velocity
above which no coalescence will occur. I-t is thi~ realiza-
tion which makes such an impact filtering pad 26 as con-
templated by the present invent1on particular~y usefulfor the flow veloci-ty ranges incurred in marine applica-
tions. That is, for wire sizes which are efficient in
removing particulate matter from air, the flow velocity for
marine applications can satisfactorily be chosen so that no
coalescence will occur in the filtering pad 26, thus avoid-
ing the necessity of any additional filtering means or
- stage downstream of -the pad 26.
More particularly, Figure 4 shows the relation-
ship be-tween the occurrence of coalescence and fiber
diameter as a function of flow veloci-ty through -the impact
3 6 0
-19---
f:i:L-tering pa~l 26. The abscissa represents a ~le~lsure of
coalescence and khe ordina-te ir, ~i.gure ~ represen-ts f:lc~w
velocity. Three curves are shown in r':igure ~ or three
di*ferent fiber d:iaMeters. 'rhe lowermosk curve represents
the relationship between coalescence and ~~I.ow ve~ocity for
a filtering pad comprised of 0.001 inch d-lame~er polyester
fibers; the ~iddle c~r~e is for a pad c~mprise~ of 0 002
inch diameter monel wires; and ~he uppermost cur~e is
for a pad comprised oE 0.006 inch diameter monel wires.
In this regard, these curves were generated using me~s~red
test data for lower flow velocities and impaction theory
for higher flow velocikies.
It is to be noted that impaction theory su~ges-ts
tha~ the capture efficiency for a given wire size increases
with the square of the par-ticle diameter. There~ore, in the
absence of coalescence, the efficiency of cap-tured drop~ets
in the range of lO to 13 microns would be grea-ter than -tha~
in the 2 to 4 micron range. However, when coalescing
occurs, some of the droplets in the 2 -to ¢ micron range are
coalesced together and promoted -to the lO to 13 micron
regime. This has the effect of improving the efficiency for
the 2 to 4 micron droplets and lowering i~ for the 10 to ~3
micron droplets. Therefore, the d;.fference be-tween the
efficiencies for removing 10 to 13 micron particles and for
removing 2 to 4 micron parkicles represents a measure of
coalescence, i.e.:
: n1o-13~ - n2_~ n
where n10_l3~ is the ef~iciency for removin~ particles
i.n the 10-13 micron range; n2_~ is the ~ffi~ency for
removing particles in the 2-4 micron ran~e; and ~n is -the
~ ~ 64~
-20-
di*fererlce in such efficiencies and ~epre~ents a measure of
coalascence. This difference in efficiences accordingly
constitutes a measure of coalescence for the impact filter-
ing pad. ~hus, for an greater than 0, no coalescence is
occuring; however, if An is less than 0, then coalescence
does occur.
Figure 4 illustrates the flow velocities which
must be attained to eliminate coalescence with pads com-
prised of 0.001 and 0.002 inch diameter fibers. For the
0.001 inch fiber pad, a flow velocity of 60 feet per second
is needed, whereas for the 0.002 inch wire diame-ter pad
only a flow velocity of 30 feet per second is required
to eliminate coalescence. The 0.006 inch wire diameter
pad never exhibits coalescence in the 2 to 13 micron range.
- Generally speaking, a flow veloci-ty o* 60 feet
per second, required by the 0.001 inch fiber diameter
pad to elimina-te coalescence, will lead to an unacceptably
high pressure drop for marine gas turbine inlet separator
applica-tions. This wire size for the impact filtering
pad thus represents a lower limit on fiber diameter size
for the impact filtering pad 26. On the other hand, -the
0.006 inch wire diameter pad, while it never e~hibits
coalescence in the 2 to 13 micron range, has a poor capture
efficiency for smaller droplets and a higher pressure
drop. Therefore, this wire size represents an upper limit
on wire or fiber diameter size for moisture separator
applications. In terms of the preferred embodiment in
which the irnpact fil-tering pad 26 is comprised of 0.002
inch diameter wires, this presents an optimum choice since
only a flow velocity on the order to 30 fee-t per second
~ ~ ~4.~6~
is required to e].:iminate coalescence. Owing to thi..s rela-
tively small wire s:ize, the capture eff:iciency for small
droplets is rela-tively high and the pressure drop across
the filtering pad is reLatively low.
Thus, it will be appreciated tha-t for a given
fiber size there is a minimum flow velocity above which
no coalescence will occur. This realization is most impor-
tant since the absence of coalescence allows the elimina-
tion of a third or additional stage downstream of the
impact filtering pad-26. Fur-ther, it is the absence of
this realization by any of the prior art which negates
- any suggestion in such prior art to use an impact filtering
pad in accordance with the principles of the present inven-
tion in combination with an inertia separating device
14 for a moisture separator 10 for a gas turbine. In this
regard, once the particular wire size is chosen so as
to be in the desired range for providing a highly efficient
pad for remo~ing par-ti.cles of all sizes, the size or cross
sectional area of the pad 26 need only be chosen so as
to provide a flow ~eloci-ty above the predetermine~ limit
for -that particular chosen size of fibers so that no coales-
cence will occur, taking into consideration of course
- the mass flow velocity which is re~uired for turbine opera-
tion.
In this regard, as has been noted hereinabove,
the wire size for such an acceptable efficiency for par-
ticles of all ranges is preferably be-tween 0.001 inches
and 0.006 inches, and most preferabl~ is 0.002 inches.
The pad 26 is then constructed to provide a flo~ velocity
greater than 20 feet per second and, accordin~ -to the
7 ~ & 0
fiber diameter size whi.ch has been chosen, -to insllre the
absence of coalescence dur-iny operation. For instance,
if 0.002 inch diameter wires or fihers 30 are u-tilized
for the impact fil-terlng pad, the flow veloci-ty should
be greater than 30 fee-t per second. In terms o~ gas tur-
bines for marine applications, such a flow velo~ity is
acceptable to provide a relatively low pressure drop across
the moisture separa-tor 10. Consequently, for such an ar-
rangement, a -two s-tage moisture separa-tor 10 may be uti-
lized for efficiently removing particles from the air flowtherethrough, and in particular mois-ture particles there-
from, without the need for additional separating stages
downstream thereof.
. I-t is to be noted that the closer tha-t the diam-
eter of each individual wire 30 is to .001 inch, the higher
the required ve]ocity to eliminate coalescence as the
wire has a grea-ter tendency to coalesce which/ as noted,
is undesirable if the third s-tage of the prior art is
to be eliminated. On the other hand, increasing ~he wire
diameter -to almost .006 inches tends to reduce the effi-
ciency of the pad 26 to remove the particles from the
air and further serves to increase the resistance to flow
and.thus, pressure drop across the pad. Accordingly, be-
cause of these competing considerations 7 it has been found
preferable that the diameter of the ~ibers 30 be on the
order of .002 inches. This will lead to acceptable ranges
of flow veloei-ty to eliminate or minimize coalescence
while also providing an acceptable pressure drop.
Accord:ingly, it is seen that the -two stage separ-
ating device 10 in accordance with -the present invention
o
-23--
results in signifi.can-t aclvan-tages over -the prior ar-t.
~he effi.ciency for removing par-ticles entraine~l in the
air is maintained (if no-t increasecl) wh:ile reducing the
number of s-tages that is necessary. This in turn results
i.n a substantial savings in weight, cost, and size for
the device 10. ~urthermore, the air flow v~locities through
the air duct ean be increased wi-thou-t increasing the pres-
sure drop across the moisture separator 10, thus allowing
for a still further reduction in size of the overall as-
sembly and a reduction in the chances o~ eoalescenee.
While the present invention has been describedmainly in the context of removing moisture particles en-
- trained in air for the gas turbine for ships, it will
of course be understood by persons skilled in the art~
that the filter pad 26 of the present invention can also
be used for removing other types of particles. For example,
the pad 26 is also effieient for removing sancl or salt
erys-tals whieh may be entrained in the air.
Aceordingly, there is diselosed herein an im-
proved method for removing partieles entrained in air,
and in partieular for removing moisture particles entrained
herein. In accordance with the improved method, the air
is initially passed through an inertla separating means
14 at a flow velocity greater than 20 feet per second.
This flow velocity is chosen as such relativ~ly high flow
velocities provide for more efficiency of the inertia
separating device 1~ to separate partieles from the air.
The par-tially proeessed air from the inertia separating
~eans 14 is -then passed at a flow ve].oeity greater than
30 a predetermined flow veloeity to an impact fil-tering pad
26 for removal of par-ti.e:les entrained in the par-tially
3 fi O
---24-
processed air. The impact filteriny pad 26 comprises at
leas-t one layer o:f a plurality of f:ibers 30, each of the
f-ibers 30 having a diameter greater than .001 inches and
less than .0~6 inches, and the ratio of tota] surface
area of the :~ibers in the pad 26 to the volume of the
pad 26 being greater than ~5ft. 1 and less than 1400 ft. 1,
The predeterminèd flow veloci-ty is greater than 20 feet
per second and is chosen according to the diameter of
the fibers 30 in the impact fil-tering pad 26 so that
coalescence of moisture particles captured by the impact
filtering pad 26 is minimized. In this manner, further
filtering stages or devices will not be required downstream
of the impact filtering pad 26 for recapturing re-entrained
particles. This is a result of the minimization of
coalescence which might otherwise cause captured particles -
- to be promoted to a larger size which have a greater pos-
sibility of becoming re-entrained by the air flow passing
through the impact filtering pad 26.
While the preferred embodiment of the present
invention has been shown and described, it will be under-
stood -that such is merely illustrative and -that changes
may be made without departing from the scope of the inven- -
tion as claimed~
