-

CyclonicDevices

8.I INTRODUCTION

Cyclone separatorshave been used in the United Statesfor about 100 yr, and are still one
of the most widely used of all industrial gas-cleaning devices. The main reasons for the wide-
spreaduse of cyclones are that they are inexpensive to purchase,have no moving parts, and can
be constructed to withstand harsh operating conditions. Cyclonic devices by themselvesare gen-
erally not adequateto meet stringent particulate emission standards,but they serve an important
purpose. Their low capital cost and their maintenance-freeoperation make them ideal for use as
precleanersfor more expensivecontrol devices such as fabric filters or electrostaticprecipitators.
In the past, cyclones were regarded as low-efficiency collectors. However, efficiency
varies greatly with particle size and with cyclone design. Recently, advanced design work has
greatly improved cyclone performance. They can now achieve efficiencies greater than 98Vo for
particles larger than 5 trrm.In general, as efficiencies increase,operating costs increase,primarily
becauseof the resulting higher pressuredrops.
Cyclones use the centrifugal force createdby a spinning gas stream to separateparticles
from a gas. Figure 8.1 shows a tangential inlet, reverse flow cyclone separator.The particulate-
laden gas enters tangentially near the top of the device. The cyclone's shape and the tangential
entry force the gas flow into a downward spiral. Centrifugal force and inertia causethe particles to
move outward, collide with the outer wall, and then slide downward to the bottom of the device.
Near the bottom of the cyclone, the gas reverses its downward spiral and moves upward in a
smaller inner spiral. The "clean" gas exits from the top through a vortex-finder tube, and the pani-
cles exit from the bottom of the cyclone through a pipe sealedby a spring-loadedflapper valve or
a rotary valve.
The centrifugal force is proportional to the squareofthe tangential velocity and inversely
proportional to the radius of curvature of the gas trajectory. Therefore, the efficiency of a cyclone
increasesas the diameter of the device is reduced. To achieve higher efficiencies dictates the use
of smaller cyclones. However, the pressuredrop through the cyclone increasesrapidly as the tan-
gential velocity increases.A way to maintain high efficiencies with a moderatepressuredrop is to
use a large number of small cyclones placed in parallel.

333
 Chap.b Devices
Cyclonic Sec.8.2
334

Co
at 13,as sho
Figure 8.1 Schematicdiagram of a tan-
gential inlet reverseflow cYclone ward across
a radial cor
component
stant. For ar
m may ran
ideal, lamin
ized casest
been widell
Dusty gasin

8.2.1 ld€

Th
streamline
This chaPter covers the sizing and of areaW (
costingof singlecyclonesbasedon theoretical
and empirical considerations,and the optimal
designof systemsof multiplecyclonesto mini'
mizeTAC. Becausecyclonicdevicesare usual-
ly usedas precleanersfor more sophisticated
particulatecontrol equipment,such as ESP or
aerosol
fabric filters, it is important to be able to characterizethe particle size distribution of the
This chapter presents computational tech-
that penetratesthe cyclonesand entersthe nextdevice.
niquesfor thatpurpose.

FLOW
8.2CYCLONIC

Yourobjectivesin studyingthis sectionareto

1.Developa fractionalcollectionefficiencyequationfor particulatematter
in ideal,laminarcyclonicflow.
2. Developa fractionalcollectionefficiencyequationfor particulatematter
in ideal,turbulentcyclonicflow.
3. Usea semiempirical practicaldesignequationto predictcyclone
performance under real conditions.
 Sec.8.2 CyclonicFlow 335

Consider a particle entering tangentially onto a horizontal plane of a spinning gas stream
at 13,as shown in Figure 8.2. Becauseof the centrifugal force, the particle will follow a path out-
ward acrossthe the flow streamlines.Its velocity vector will have a tangential component (ll9) and
a radial component (u). The velocity of the spinning gas is assumedto have only a tangential
component, v0, with vr = 0. Tangential gas flows of this type usually are of the form v6/" = con-
stant. For an ideal, inviscid fluid in such a vortex flow m = 1, although in real flows the value of
m may range downward to 0.5. The analysis of cyclone performance that follows begins with
ideal, laminar flow. Then, it considersideal, turbulent flow. Becauseboth of theserepresentideal-
ized casesthat are not attained in real cyclones, it turns finally to a semiempirical theory that has
been widely used in practical cyclone design.

8.2.1 ldeol, lqminor Cyclonic Flow

The so-called ideal laminar cyclonic flow refers to a frictionless flow in which the
streamlines follow the contours of the cyclone. When the flow enters through a rectangular slot
ofarea W (rz- 11), the gasvelocity componentsare (Crawford i976):

Vr=0' v g = v= O
-rM (8.1)

Figure 8.2 Trajectoryof a

particle in laminar cyclonic
flow

'*llllr[t'
 336 Chap,d CyclonicDevices Sec.8.2

To determine the collection efficiency consider a particle entering the cyclone at a radial
position 13that strikes the wall at an angular position 01.The particle's velocity componentsat any
point on its trajectory are u, and u6. The radial velocity component is the terminal velocity of the
particle when the centrifugal force, F, = tnpuo2lr,acts on it. When the drag force can be given by
Stokes'slaw, this velocity is An
hits the outer

,r=L (8.2)
3npDo

Since the O-componentof the particle's velocity is that of the fluid, ae = ve, &nd

Toc
Q2 ing particle I
r,=f;otDi (8.3) Figure8.3).I
w2 f (tnnzlrr)z
over0<0<
that hit the ot
CombineEqs.(8.2)and(8.3)to obtain:

ppe2 D;
ur= (8.4)
t8p 13w'(n r/r1)2
SubstitutingI
The next stepis to obtainan equationfor the particle'strajectoryin the cyclone.Thedrs-
tancetraveledin the 0-directionin a time intervaldt is us dt = rd9. Also, the distancethe particle
movesin the r-directionin timedt is dr = udt.Then,

rd0 -ue (8.s)

dr ur

Eqs.(8.1)and(8.4)in (8.5)gives
Substituting

d g = 18pWh(r2l4)r (8.6)
dr Poad
=
a differential equationdescribingthe particle's trajectory.If the particle entersthe ryclone at r
rr andhits theouterwall at 0 = 0n thenintegratingEq. (8'6) gives

I
I
-
I

n
 Sec.8.2 CyclonicFlow 337

0; -,:)
er-evwn(r'd
(8.7)
PPQD;

An alternative is to solve Eq. (S.7) for 13to find the entranceposition

of a particle that
hits the outer wall at 0 = 0r,

l, -
fz=lfo-
poeolerf'''
l' w*;Gil) (8.8)

To obtain an expressionfor the collection efficiency of a cyclone, assume

that the enter-
ing particle concentration and gas velocity are uniform across thl
enrance cross section (see
Figure 8'3). If the cyclone has an angle 01,all particles that enter the
device at r ) rzhit the wall
over 0 S 0 ( 0r. The collection efficiency is just that fraction of the particles
in the entering flow
that hit the outer wall before 0 = 0p.Therefore,

w( "- ")
n (D p) = (8.e)
wl'z-')
SubstitutingEq. (8.8)inro (8.9)andrearranging

+-* Figure 8.3 Schematicdiagram of a cyclone

12- 13 entrancechannel
€
12- rl
338 Chap.8. CyclonicDevices S e c . 8 . 2C

l t 1l/2
I
1 -J Ll -
o-ODi$r
I P-/ r
II
r r 0.8
/ \ L su
' r wr ; n|zn) ) (8.10)
n(D:ll=
l-r11ry
' .: | :. :

Thevalueof Oyatwhichn(Dp)= 1 is givenby

'l-l
0.2
f t

,,=]
" :o-!??i,l' V;- ,?)
tnlrztd)
lgvw
(8.11)
'0
0

Example 8.1 ldeal, Lanrinar Cyclonic Flow only at theen

trifugalforce
At at298K and 1 atmflows at therateof 5.0 m3lsand,carries',with,it particulatema{er lentductflw
' ' Con
with a densityof 1,500kg/m3.The streamentersa cyclonicregionwith 11= 0.2 m ?nd12=
Figure8.5.O
0.4 m in ideal, laminar flow. Throughwhat anglemustthe flow turn in the cycloneif the
theouterwd
efficiencyis to be unity for 30-pmparticles?Theheightof thechannel,lV, is I m. Plot the
sublayerove
efficiencyasa functionofparticle sizefor this angle.
thelaminars
Solution : i

^ ,: : '
^ g (r.a +"1 ^-5 1 /,\,-
u,.=-# - 0.22|=2.041rad= lti"
l:n.+' whererraist
" t,soo(5.0)(ex
ro-to) ftactionof &
i

Equation(8.10) gives the efficiency as a function of particle size.Substitutingthe value of

',,,',,
qfcalculatedearlier, , ,..

\ ( n ) = z (-r a
whereDnis in pm. Figure8.4 is a plot of this equation.

8.2.2 ldeol, TurbulentCyclonic Flow ,. l

Figure 8.5 showsthe model of the turbulentflow cycloneseparator.Assqme.thatttre

effect of the turbulenteddiesis to distributethe particlesuniformly over the crosssection,not
Sec.8.2 Cyclonic
Flow 339

0.8

0.6
n(Dp)
0.4

0.2

Figure 8.4 Efficiencycurve for

10 t5 E
Example8.1
Do$rm)

only at the enfrance,but at any given angle0. This is a conservativeassumptionbecausethe cen-

trifugal force effects may serveto dampout the turbulenteddieswhich natumlly occur in turbu-
lent duct flow. Relatively little is known aboutthis area.
Considerthe effectof a laminarlayernext to the oubr edgeof the cyclone,as shownin
Figurq8.5.Oncea particle,vigorouslymixedin thecoreof theflow, entersthis layer,it travelsto
the outer wall and is removed.The distancethe particle travels in the 0 direction in the laminar
sublayerovera time intervaldt is us2dt = rzd0, whereils2is evaluatedat r = rz.The thicknessof
the laminar sublayeris

dr=ur2dt=
urr# (8.r2)

whereu,2is also evaluatedztr = rz. The fractionaldiminutionof particlesover the angled0 is the
fraction of the particleswhich lies in the boundarylayer,m

Figure 8.5 Turbulent cyclonic flow

'.'.
340 Devices
Chap.S Cyclonic S e c . 8 . 2C

rz (b) Dete
_dn = dt =4rZ 4g
fl u o z1 2 - 1 1 (8.13)
r2-\
Solution

lntegrating this equation between the entrance(0 = 0) and any angular position, (a) Thmr

n=no"*lffih!\ (8.14)
(b) Solv

where n6 is the initial numberof particlesof diameterDrper unit volume of gas' The collection
efficiencyof a cyclonethathasan angle01is

q(D,)=1-#=1
,plffiA (8.1s) Substin
1 clesis Q

For lack of a betterapproximation,usethe inviscid gas velocity componentsgiven by

the fluid velocityfield in the turbulentflow cyclone.Therefore,
Eq. (8.1)to represent
Com

Uao = ---;
Q
,
A cory
Wr,- :rrilLl (8.16) cally the
\rrl flow ini
collectir
flow cor
poe, Do,
llr2 =
(8.17)
t8p 123w2 0n f2lr)2
8.2.3 Prqc

The j
andthe collectionefficiency of
previously re1
eitherof these
n(np (8.18) L,eitl
Alexander(1!
yetn = COnSta

Example8.2Ideal, Tlrrbulent CyclonicFlow

(a) Consider the data of Example 8.1. Estimate the collection efficiency for 30-pm particles
assumingthat theflow is'turbulentandthecycloneangleis 2.041rad'.
 Sec.LE,@onic Flow
841

(b) Determinethe angleof turlr to obtwiggvocollection

efficiency.foi tt .,
"'ro-}hipurti"t
Solution
--::---
. . .:

'{a)::?h,oughdirectsubstitution,;.rn.r,.'*l,.:..':.'

t1(30-pm)= 0.528(52.8Va) ..
.. ,. .
.. ,
(b) SolveEq..(8.18)for the angleof turn; 0r,for
a givenefficiency. o-
r v1r vruvrvrrvJ ! ,....
'
l' ' :
.
; .'.... ,,'

_ ,fu!rtn)t"(,_rl)
o"__l8ttrzw(,rz
v/---

Substituting
lyneri:ar ya]ues,the anglefor 99%coilectionefficiency-ofthe vv
clesis 07= 12.53
30-um parti-
r'r vqu-
rad= 2 frrll tums.
:.::" . 1 r...
.....,_t.

ir. -:::r:: .::.i.- - : ,:i'. - :,: , , . . . " : . 1 ., ,, .,

-
Comments . .t.:,,,..,.....-.,,,.

].:.'.:]:]-,':.11:..,.,.;....:::.-].l1.

A comp'artson'ofthgrlesults.ofExauplesS,l.and f,.2,sho\^{s
thaf turbuleneeiEduces drarnati-
e!figrencyof auy"ton", ni"cise criter,ionfor transition
lallf $e { from,laminar to turbulent
notexisr(Fragan
andseinreld lettii;;;;;""tauy determined
:^"xlnr#:1"-f*
L;uilecuonemcrencycurvesappearto conforrnmore
,'.'flowconditions; closely to turbulentthanto laminar.

8.23 PrccticslCpe" DesignEgudion

flow *tte'fl::,?^rP5__tvroy comptexoneandtherwo modelspresenred
,The
previously represent extremesin performance.
Because op"rutinf"yclonesdo not conformto
eitherof thes; d*,en
Lerrn Yr,Tf^::lltl?1'l:*"TnT.d "qlutioo,
anc Lrcht (Licht 1980) developed a theory
f*ii", rheirperforrnance.
useful in practicai cyclone design.
riuna.experimentailv
thattheexponent ;;;;il,*genriar.verociry
|t:*^d:l lil]fl is,given profie,
veln =€onstq!i.t:, by:

|4)Lt)'=
. = r -o.67D,o .,.(g:19)
,-(1
342 Chap.B CyclonicDevices S e c . 8 ' 3S

Exte
where D, is the cyclone body diameter in meters and Z is the gas temperaturein K. The collection
efficiency, according to the model by Leith and Licht, is given by cyclones aftt
sufhcientlr t
sions are reli
q(Do)=t -.^p(- ,vD{) (8.20) along rtith r
pressure drc
dimension:
where
M=Il(m+l)

Toble8'
^lxQp,c,(^+ tl)Mtz (8.21)
V=/,---- , I Term
| 18PDi l
Ko=a/D,
K'"=brD
where K is a dimensionlessgeometric configuration parameter and C" is the Cunningham correc-
sfl),
tion factor. Be consistentwith the units in Eqs. (8.20) and (8.21).
DtD'
h"D,
Example8.3 Leith-Licht Model for Efficiency of Cyclone HD,
BD
Consider the gas streamof Examples 8.1 and 8.2. It flows through a cyclone with a body
diameter of 2.0 m and a value of K, the geometric configuration parameter,of 55 1.3.
Estimate the removal efficiency of the cyclone for 10-pm particles' f,
r5t^rca
Solution
=
For a cyclone body diameter of 2.0 m and a gas temperatureof 298 K, Eq. (8'19) yields ru
= particles,
Eq. (8.20) yields q =
0:734,M = 0.57'7. From Eq. (8.21),Y 1041.For 10-pm
tq'e::r
0.742 04.2Vo).
iafilfi

rn fiet'
iror\
Lnl.r
CYCTONE
8.3 STANDARD RATIONS
CONFIGU

Your objectivesin studying this sectionare to 1r:f

1. Understandthe concept of a cyclone of standardproportions.

2. Calculate the efficiency ofvarious standardcyclone configurations.
3. Estimate the pressuredrop through cyclones of standardproportions.

'.,..#
 Sec.8.3 StandardCycloneConfigurations
343

Extensive work has been done to determine in

what manner the relative dimensions of
cyc^lonesaffect their performance. A number
of configurations have been proposed and studied
sufficiently to be regarded as "standards." In
thesestandard cyclone conJigurations all dimen-
sions are related to the cyclone body diameter.
Table g.1 pr"r"nt, the dimension ratios of these,
along with values of the geometric configuration
parameter, K, anda constant, 1/6, relating the
pressure drop through the cyclone to the
inlet velocity head. Figure g.6 illustrates the various
d i m e n s i o nisn T a b l e8 . 1

Toble8.I Stqnd@loqe Configurotions

Term Description Stairmanda Swiftb Lapplec
Ko= a/D, inlet height 0.5 0.44 0.5
Kt= b/D" inlet width 0.2 0.21 0.25
s/D" outlet length 0.5 0.5 0.625
D/D" outlet diameter 0.5 0.4 0.5
h/D" cylinder height 1.5 r .4 2.0
H/D, overall height 4.0 3.9 4.0
B/D, dust outlet 0.375 0.4 0.25
Ns Eq.(8.22) 6.4 o 1,1
8.0
Eq.(8.21) 551.3 699.2 402.9
aStairmand(1951) bSwift(1969)
andLappte(19t9).
"Shepherd

The other major.consideration in cyclone specification,

besides collection efficiency, is
pressuredrop' while forcing the gas through
the cyclone at higher velocities results in improved
removal efficiencies, todo so increasesthe pressure
drop andlhe operating costs. There is ulti-
mately an economic trade-off between efficiency
and operating cost. Several methods have been
proposed to estimate the total drop in the fiow or"gu, through a standard cyclone.
.pressure
unfortunately, there is no definitive study to determine which is the best to use. Most
agreeto expressthe pressuredrop in terms of a methods
multiple of the inret velocitv head. or

M = Nspyvfl2 (8.22)
rr here
'\g is a constantwhich dependson the cyclone
configuration (seeTable g.1)
p-'is the gasdensity
r'6is the -easr-elocityin the cycloneinlet duct =
O/ab.
 Chap.a Devices
CYclonic
w
Exau

(
I

ll'
L-J
(
t
De SoM
-\t!\
_-

\
.l
H -t

Figure 8.6 Dimensions of a standard

cyclone

andthegasvolumefric
Equation(8.22)canberewrittenin termsof thecyclonebodydiameter
flow rate.
E,{ 5
rc=-W (8.23)
zxlxtP,4

Ka= alDc
Kt= blD"
sensitivet9 -th: cyclone
It is evidenrftom Eq. (8.23) that the pressure'eropis 9xtrglely
bodydiameter, *!!t 9*l& becomes
increasinJraiiOily Jmaler'Noticein Table8'1that'for a
configuration
Swi{tstrandard is moreeffi-
ti"Oitl*t andbodYdiar,neter,
givensetof operating
pressuredrop (higher value of Nr) ' I-apple con-
cient (higher value or x), uut resultsin a higher
nearlyasefficient asthe other two'
figuratio-n,with a relatively high pressuredrop,is not

,. fft- {l '*'r=-;
Sec.8.3 StandardCycloneConfigurations' 345

Example 8.4 PressureDrop through Cyclone of Standard Configuration

(a) Estimatethepressuredropfor the cdnditionsof Example8.3.Accordingto the valueof

K in that example,the cycloneproportionscorrespondto Stairmandconfiguration.
(b) If the cyclonebody diameterdecreasesto 1.0m, estimatethe removalefficiency for 30-
pm particles,andthe correspondingpressuredrop.

Solution
l
a) FromTable8.1,,forStairmandconfiguration,Nrr 6.4,Ko= 0.5,Kr = 0.2.The idealgas
=
law givesthefluid density,Pf= 1.186kg/m:. Thecyclonebodydiameter,D,,is 2.0 m, the
gasvolumetricflow rateis 5 m3/s.FromEq. (8.23),AP = 593 Pa.
(b) For a bodydiameterof 1.0m, andStairmandstandardconfiguration,m=0.665, M =
0.6,Y =2,49I, n(30 ttm) =0.992. Becausethebody diameteris twice assmallasthat of
the cycloneof parta, thepressuredropis 16timeshigher.Therefore,AP = 9,500Pa.

Commenls
Usually, the pressuredrop is the limiting factor in the designof cyclones.To maintainthe
pressuredrop within acceptablelevels,the removalefficiency for:small particlesmust
remainrelativelv low.

8.4 SIZEDISTRIBUTION
OF FENETRATING
PARTICLES

Your objectivesin studying this section are to

1. Developan equationrelating the cumulativemassdistibution function
ofthe particlespenetratinga cycloneto the inlet conditionsandthe
devicegradeefficiency function.
2. Estimatethe resultingintegralswith Gauss-Legendre quadrature
fcmulas.
3. Esimate MMD ando, for the aerosolthat penetratesthe cyclone.

ffi*n- q-;.
 346 Device$
Chap;,8.,,Gyelonic Seeil

. ,,GycJQncs.are mgre.sophisticatedparticula.le,eontrol
f,requentlyuse.das..B-rgeleaners,,.for
devices.The size distribution characteristicsof the aerosolpopulation entering the cyclone are
known,but theymustbe calculatedfor thoseparticlespenetratingthedevice. cjdr
The size distributionfunction of the particlespenefratingthe cyclone,by cumulative
massfractionlessthan size,Go(Dp),maybe determinedby a materialbalancetakenoverall parti-
cles finer than a given sun Do as shownin Figur-e8.7. Thq maqs,,flowqatg,oJpartiqleqfiner than
Drenteringthecycloneis givenby QciG;(Dp), where
O = volumetric flow rateof the gas
ci = massof,total particulatematterper unit volume of enteringgas
Gi (Dp) = cumulativemassfraction of particlesfiner thanD, at theentrance Sinfl
Thq;massfloy rate of particlesfiaer thanpopengtrating the cycloneis,Qc;P1ffin(D)nwherg
Ptu =overallpenetration =1 -rlu
, " Ga(&) = cumgJativemassfraction o{ particlesfiner,thar,rDr.atlhg outlet . .,

of paricles finer t{an D" hthe collectedrlu t is giyeg by.,

The masq,{lgw-.1-1t9 : , ,1

smd
a,,f n(q)*^/(o)a@o'\
1y[g16 '.,".,.'.
n^;(Do)d.(Dp') = the mass fraction of particles with diametersaround D o' at the inlet of
the cYclone
For a log-normal size distribution at the cyclone inlet,

{lno;1+,f"l4nOi)' Itr
(8.?A)

Gas outlet

-e:'-
gcici(np) gCirtnG|{oi\

Figure 8.7- Material

q
balance on particles
fuo finer thaff',D;
Q",l r{oo')r^{or'laoo'
lo

rt
Ir
I
Sec.8.4 SizeDistribution
of Penetrating
Particles 347

where MMD; and og; are characteristic of the inlet aerosol population.
Assuming that the size of the individual particles do not change as they flow through the
cyclone, a material balance yields

lry
gcirtyGo(or) = pcic;(no)
* ert I n(o;)"^,(oo')doo' (8.25)
to

Simplifying andrearrangingEq. (8.25)becomes

c,(o)- n,*,a(oo']
c"(oo)= f (8.26)
Ptu

Substitutingin Eq. (8.26) the definition of G;(D,),

(r-n)n^,ano'
c,(oo)= f
(8.27)
Ptv

Definethe gradepenetation,Pt(Do')= 1 - n(Dp'). Ttien,Eq. (8.27)becomes

tDp
rdoo')n-;(oo')dnr'
c"(oo)=J"
(8.28)
Ptu

For a log-normal distribution, the integral in Eq. (8.28) can not be evaluated in closed
form. Define a new variable, x, such that

2D-',
' -7
x- (8.2e)
Dp
 Devices
Gyclonic Sec.8
348 Chap.A

Soluft
Then,

r =+L', TI
ftr
rt(oo')n*i(oo')aoo' Pt(x)n^;(x)ilx (8.30)

numerically with a R
The integral on the right hand side of Eq. (8.30) can be approximated d
Gauss-Legendrequadratureformula: ll
T
h
N

Pt(x)n*{x)dx = w;Pt(x;)n*i{xi) (8.31)

i=l

degree Legendre polynomial' Equation

where w; andxi arethe weight factors and roots of the Nth
(8.28) becomes
Do$
.1J w;Pt(x)na;@)
;L
i =1
(8.32)
c"(o) =
Ptu

quadrature
The overall penetrationin Eq. (8.32) can be estimatedwith a Gauss-Hermite
gradeefficiency equationfor the
formula (as illustrated in Cttup*t r) given MMD;,'os; and the
chmacterize the particle
cyclone.The following exampleiliustrates'the-computational'schemerto
sizedistributionfunction at the cyclone gas outlet'

a cyclone
Example 8.5 SizeDistribution Function of Particles Penetrating

. consider Example8.3. The aerosolpo_pulation enteringthe cycloneis log-normallydistrib-

removalefficielcy forthe
utedwith MMD = 8.0 pm ando* = 2.5. Erti*ut" thg over,.all
fractionof particlespenetratingthecyclone.for vari-
;"t;. A*late the cumulative"mass
follow a log-norrnaldlstnDuuon'
ousparticlesizes.Determineif thesecumulativeresults
*0, if so, estimatethe correspondingMMD ando, values'
i
Se.e.,i84Size,Distribution of Penetrating Particles 349

Solutlon,.

The gradeefficiency firnction for the cycloneat the operatingeonditionsof E*ample 8.3 is,
: ' ir
f r o m E q . ( 8 . 2 0 ) , ' : :: :

\ (Dp')= 1 * exp( - 1,041Do'o'stt

1 (8.20a)

sChemeoutlinedinrExample'7,4to estimatGr,the
:-Foli6w,tlle colnp{r,tatiioeal ovgrall removal
i*;ef.fici6ticfthrorgh.applioatioii.ofGdussjHermitequadrature-formulb:rThe result;basedon a
''
16-pointformula,isqy=0.685'Therefore,theoverallpenri:trationis'Ptna=9;314''
The procedureta charaeteize'theparticle sizpdissi-butionponetr'aling the cycloneli'sasfol-
lows: : :
. ChooseN, the numberof quadraturepoints for Gzuss:Legertdxe formula.
. Obtainthe valuesof the root$ &, Andweight factors,ltr, of the.corresponding
Nth degreeLegendrepolynornial,either from a mathematicaltabfeor from the
computersubprogram GAIJLEGalreadyprovided(Sectio44.3.I).
.ChooseavalueofDp.,. "', - I ', , 'r 'l ,, ' r ': : :
. r'For eachof the roots calculatethe conOsrrontlingDri frorlr-'Eq.(8.29).
.rr.omEq.(8.20a)calculatePt(Do)=1..r1.(D7,).:.'....
, , rr":F.ronlt&i,.{S:24)calculateno1;{Dp'\. ,." 1,,, :
, . FromF,q.,(8,32) calculateGo(Dp).
. Chooseotftet outo"sof Do andrepeatthe calculations.
. Plot G, versusD, on log-pntbability paper.
. If a reasonabt"rnuigntUnc resblts,estimateMMD andos at the cycloneoutlet.
The-following table sunrna*izesthe resultsof the outle.tcumulativemass&actioncalcu-
lations with a 16-pointGauss-I,qgendre quadratrreformula.

1.0 4
,,.2O',.,,.'. 0.13_ss.
..
' ' 0:3958r ,:
4.0
'
5.0 [363t':r"i'
',t7'0 ,," 0.66s1
.,,,lQJ . t 0.809,
11.0 0.8404
12,0;, .,:0.8659,,
13.0 ,0.8868
15.0 , 0.9184
350 Chap.8 Devices
Cyclonic Sec.8

Figure 8.8 is a log-probabilityplot of ttretabulatedresults.It showsthat the cumulative

massfraction distributionof theparticlespenehatingthe cycloneis log-normal,with MMD tional r.-
=4.96pm and6e=2.22. of the gi
is reduct
sure dro
Comments maintar
cyclone
The net effectsof the cycloneareto remove68.6Voof all theparticulatemass,to displace tem. the
the size distributiontowardfiner particles,and to reducethe spreadin particle sizesexhibit- cy of tlx
ed by the original distribution.
20
{
10 il
^
F4uan.'

.$
a

Figure 8.8 Size distri-

bution of penetrating
particles in Example c1-clon
- 3 R3 g R8 g3 8.5 high re
e 3A
O\ O\
similar
o\
Go(7o)

Eran
8.5 MULTIPLE
CYCLONES Svste

Your objectivesin studyingthis sectionare to

1. Understandthe advantageof multiple small-sizecyclonesoperatingin
parallel-as comparedto a singlecycJone-for high particulateremoval
efficiency.
2. Calculatethe overall removalefficiency andgaspressuredrop for a
multiplecyclonesystem.
Sec.8,5 MultipleCyclones 351

The centrifugalforce-the ddving force for particulate'rernovalin'cyclslss is propor-

tional to the squareof the tangentialvelocity and inverselyproportionalto the radiusof curvature
of the gastrajectory.'Therefore,the efficiencyof a cycloneincreases asthe diameterof the device
is reduced.To achievehigher,efficienciesdictatesthe useof smallercyclones.HoweVer,the pres-
suredrop,throughthe cycloneincreasesrapidly as the tangentialvelocity increases.A way to
maintain high efficiencieswith a moderatepressuredrop is to use a large number of small
cyclonesplacedin parallel. If N, is thq numberof cyclonesin parallel in a multiple cyclone sys-
tem, the gasvolumetricflow ratethrougheach oneis Q/N,. Equation(8.20) for the gradeefficien-
cy of the systemcanberewrittenas(seeProblem8.4):

tt(Do)= 'I-expl'rr;t\ (8.33)

Equation(8.23)for the pressuredrop throughthe multiple cyclonesystembecomes:

tp =-*!"!E-
n1t4 (8.34)
2Ka'K6" Ni Dg

Equation(8.34) suggeststhat the pressuredrop'decreases rapidly as the numberof

cyclonesincreases. The following example shows that a multiple cyclone systemcan achievea
high removalefficiency for small particlesat a muCh lower pressure drop than d singlecycloneof
similarefficiencv.

- - ' - - - - - _ _ Efliciency and PressureDrop of lVlultiple Cyclone

Example 8.6 Collection
System .:
A papermill in Lorrisianaoperatesa powerplant that producepsteamfor the papermaking
process,andenoughelectriciryto run a city of 200,000.Theplanthasfour boilers-two
recoveryboilersthatburn "black liquor" from thepulp mill, andtwo powerboilersthat
burnbark,coal,oil, andgas.Thepowerboilershavea flue gascleaningsystemthatconsists
of a multiple cycloneprecleaner,followed by an elecfostatic precipitator.The flue gasflow
rafeis 165m3/s,measuredat the actualconditionsof 450 K and 1 atm.The averageparticle
:, densityis 1,600kg/m3.The precleanerconsistsof 2 units in parallel,eachwith 450
, Stairmandcycloneswith body diameterof 0.25 m. Assumingthat the flue gasesbehaveas
afu,

,Flult :&
352 Devices
Chap.8 Cyclonic Sec.8

(a) Estimate the removal efficiency of the multiple cyclone system for 10-pm particles and 8.6.rC
the resulting pressuredrop.
(b) If a single Stairmandcyclone removes 10-pm particles with the sameefficiency as the
multiple cyclone system of part a, estimate the correspondingpressuredrop. (updated
support !
Solution

(a)Thetotalnumberof cyclones, N., is 900;D"=0.25 m. FromEq. (8.19),m=0.485,M =

0.674.Calculatethe characteristic or relaxationtime for 10-pmparticles.Theviscosityof where
air at450K is 2.48x 10-5kg/m-s,the Cunninghamcorrectionfactoris approximately1.0.
Therefore,t = 3.58x 10r s.FromEq. (8.33),tt (10 ttm) =0.952(95.2Vo). Thepressure
dropis from Eq. (8.34): =
A,P 2.16h'Pa
(b) Calculate,by trial-and-error,
thebodydiameterof a singleStairmandcyclonefor 95.2Vo
removalof 10-pmparticles.The answeris D, = 2.37m. The pressuredropis from Eq.
itsef. Tl
(8.23):AP = 216.8kPa.

Commenls Examg

This example illustrates the advantagesof multiple cyclone systemsover single cyclones as C(
high efficiency preclealers. Many combinations of number of cyclones and body diameter an
all
result in the desiredefficiency. The optimum combination is obtained from a cost analysis.
sr

Solutir
8.6 COSTANATYSIS
FORCYCLONES
Tt
an

Your objectivesin studying this sectionare to b

1. Estimate the initial and annual cost for single and multiple cyclones. TI
2. Design a multiple cyclone systemto minimize its total annual cost. gr
cl
ft
fa
Cyclones are very inexpensive, having capital costs at least an order of magnifude less s
than final control devices such as baghousesand electrostaticprecipitators. Becauseoftheir sim-
plicity, the only significant operating expenseis the cost of electricity to overcome the pressure
drop through the device.
 Sec.8.6 CostAnalysisfor Cyclones 353

8.6.1Costsof SingleCyclones
Vatavuk (1990) gives a correlation to estimate the equipment cost of single cyclones
(updated to June 1990) for a range of inlet duct areas.The cost includes a carbon steel cyclone,
support stand,a fan and motor, and a hopper for collecting captureddust. The correlation is:

EC = 57,800fablo.soz (8.3s)

where
EC = equipment cost, in $ of June 1990
a and b are the inlet height and width, respectively
Equation(8.35) is valid in the range 0.020 < ab,m210.4

Installation costs and ductwork connections are often more expensive than the cyclone
itself. The total capital investment is about twice the purchasedequipment cost.

Example8.7 TCI and TAC for Cyclone

Consider the cyclone of Examples 8.3 and 8.4a. Estimate the total capital investment and
annual cost if the device operates8,000 h/yr. Assume a useful life of 10 yr and a minimum
attractive rate of return on investment of 157o.The mechanical efficiency of the motor-fan
system is 65Vo,and the cost of electricity is $0.08/kW-h.

Solution

The cyclone is of Stairmand standardconfiguration with a body diameter of 2m. The inlet
areaisab = KoKaD.2 = (0.5)(0.2)(2.0)2 = 0.4 m2.The equipmentcost is from Eq. (8.35),EC
= $25,300. AddSVoto EC to account for freight and taxes, and multiply by 2.0 to estimate
the total capital investment: TCI = $54,650.

The only two significant expensesin the total annual cost calculation are the capital recov-
ery cost and the cost of electricity. For i = 0.15 and n = 10 yr, the capital recovery factor is
CRF = 0.201yr.Therefore, the capital recovery cost is CRC = $10,930/yr. For a volumetric
flow rate of the gasesof 5 m3/s and a.pressuredrop of 0.593 kPa, the power to operatethe
fan is (5X0.593)/(0.65) = 4.56 kW. The annual cost of electricity is (4.56X8,000)(0.08)=
$2,9201yr.TAC = 10,930+2,920 = $13,850/yr.

,5'
854 Chap.8 Cyclonic
Devices Sec.I

8.6.2Goslsof MultipleCyclones , .
where
The following correlationfor ttreequipmentcost of multiple cyclonesystemsis basedon
suggestionsby Crawford (1976)and cost datapresentedby Coopu and Alley it986;, and n"ters
andTimmerhaus (1991):
..i. -.'r. !1..i-.
$

! EC =7,C0f,N..ab+72i,1, (8.36)
;_. , s

where
EC= equipment cost,in $ of June1990 : :
Nc= numberof cyclones. Equatio
(8.36)is validin therange1.0SN"ab,mz
Equation <6.0

Example8.8 TCI,and',TAG,of,a
Multiple CycloneSystem ,,,, :
-
Estimatethe total capitalilveitrnent andannualcostfor the muitiple cyclonesystemof where
Example8.6a.Assumethattheconditionsaresimilarto thoseof Example:8.7., :, i ?
I
I wrirtEr
Solution i : : .. .r ::: ; ... I
t

{
;
Fora totialof 900Stairmandcycloneswithbodydiameter of 0.25m,tlretotalinletareais t
I
J
=
N"ab N"KJ(p"2 = (900X0.5)(0.2X0.25P = 5.625m2.Theequipment costis fromEq. 1

(8.36): ' 1
EC=$104,200.TCI=(2Xl.0SXlM,200)-$225,000.'
,,
Ii
t
Fora capitalrecoveryfrcnr of O.2lyr,thecapitalrccoverycostis g45,000/yrFora pressure $
+ tU.IIof
...
dropof 2.16kPaanda gasflowra'teof 165m3/s,thepowerto pushthegasthroughtte sys- i; stantri
temis (2.16)(165)/(0.65)
= 548kW.Theannual is (548)(8,000X0.08)
cosrof elecrricity = condili
" $351,000/yr.TACr='r$396,000/yr.' =N.D.3

8.6;3'Optimizolion'ofMultipteCyclone SystemDesign ::
:..1 . , .: : ,'

' It is possibleto estimatethe mosteconomicaldesignfor a multiplecyclonesystem,

given the-required'collection
efficiency.The total annualcostfor the systemis , , l lh

TAC= KlNgab+X2Wt+K3N,
(8.37)

1'
l!

'\.
 I
Sec.8,6 CostAnalysisfor Cyclones 355

where
Kr = the capitalrecoveryfactor times the installedcost of a cycloneof unit inlet
area,$/yr-m2
Kz. = the cosi of electricenergy,in $/kW-h

W = electricpower,in kW
t = the numberofhours theunit operatesper year
K: = the capitalrecovbryfactor timesthe portion of the installedcost of the
systemthatis proportionalto the numberof cyclones,in $/yr.
Equation(8.37)canbe rewrittenas

KzlNaPtQ-'
TAC= KlNsKaK6D] + +K3N" (8.38)
znxlrlu?o"4

.l
where E = mechanicalefficiency of the motor-blowersystem.
? Equation(8.33),which givesthe gradeefficiencyof the multiplecyclonesystem,maybe
l
I
written as
I
I
a' KQT
NrD: =
*l-t^(,-n)f''*
I
1 (8.3e)
j
I
t
a
I The only term on the right hand side of Eq. (8.39) that dependson the systemdesignis
t

t M. Howevbr,it is only a weak function of the cyclone body diameterand may be assumedcon-
stantwithout introducingsignificant errorsin the following analysis.For a given set of operating
$ conditions,then,the combinationN"D,3 is fixed. Equation(3.38) can be written in termsof ND3
{
in
rl: = NrDr3.
T rrr"rr(No:) KztNnpfQ3D? . rr(Nn:)
f
i
TAC --
Dc
'
znrlx?(Noz)2 D:
+-
(8.40)

I
i
t Define
t
I K1' - KlKo!fi(ND3),

,, t K2tNpp1Q3
-
t\? =-
2EK:Kf(Noa)'

-{'rt'
 356 G h a p . 8 .eyclonicDevices Sec-|

1
K:' = KI(ND3) I
Then, c
d
TAC=U+ xz'o?+\ (8.41)

.. :., : Exan
To minimize TAC, takethe derivativeof'Eq. (8al) wittr respectltoD" a4d set it equalto
zeto.
I

iaracl Kl',nurn
= -::r- + ZK2,D" _ 3":.i
Kt
=O (8.42)
LT4l D"' D;

ReanangingEq. 6.a2L
'' Solu

Dj - --6Lp"2-3Kl' = 0
"' 2Kr, ZKc, (8.43)
: -
, . ::'.,',

Theoptimumcyclonebodydiameteris therealrootof Eq. (8.a3).

Exampte.,S;9Optimal Designof Multiple CycloneSystem

. .t : :' :

Determinethe numberandbody diameterof Stairmandcyclonesthat minimize the TAC in

Example8.8.Your designshouldachievethesameefficiencyfor l0-pm particles,namely,
95.ZVo.
'r''"- '" :' ':'
Soiution " "'r' ',

Assume M - 0.7.FromEq.(8.39),NP3= 14,!m:,lhe qApital

recovery
factoris 0.20lyr.
BasedonEq.(8.36),Kr = (7,000X1.08X2X0.2)
=3,024$/n-m2.Accordingto theirdefini-
=4265$-mlyr:,Kz'=iO.OSXg,0o0X6.4X0.7g5X165)3/
tions,K12=(3,024X0.i)(0.2X14.1)
(2X0.65X0.5)2(0.2)2(14.1)2 = 5.62x106$/yr-m2,
(1,000)l h' = QZ)(r.0SX2X0.2X14.1) =
438.5$.m3/5n. (S.+l)blomes
Equation
.1.
, D:'-a.w3lsD? -0!:000116z
= 0.,
. , .. .r, :i:.

The real root of this polynomial is D" = 0.166 m. The number of cyclones is N" =

ll
ti
 957
Sec.8.6 CostAnalysisfor Cyclones

(8'19)' m = 0'45' M = O'69'

14.Il(0.166)3= 3,080.Check the value of M assumed'From Eq'
the optimal design is 3,080
which is sufficiently close to the assumedvalue. Therefore,
(8.41) gives the total annual cost for this
cyclones with body diameter of 0.166 m. Equation
the TAC for Example 8'8!
derign, TAC = $Zi6,OOOlyr.That is $120,000/yr lower than

Example8.10 optimal DesignBasedon overall Efficiency

Example 8'9 is log-normally
The aerosolpopulation entering the multiple cyclone system of
design that achieves
distributed with MMD = 4.0 pm and o, = 2.5. Determine the optimum
aerosol population pene-
an overall particulate removal efficiency of 70Vo' Chancterize the
trating the sYstem.

Solution

efficiency' The relation

The efficiency specified is not a grade efficiency, but an overall
than Eq. (8.39). The solution involves trial and error'
between ND3 and q,y is more complex
The computational schemeis as follows:

. Assume a value of ND3.

. CalculateKt' , Kz', and K:''
' Solve Eq. (8.43) for D"; calculateM andN, = ND3/D'3'
. Apply Gauss-Hermite quadrafure formula to estimate the resulting overall
efficiencY.
. If the calculated efficiency differs from the specified efficiency by less than a
peneffatirrgaerosol
specified tolerance,the design is optirnal; characterizethe
according to the schemepresentedin Example 8'5'
. otherwise, assumea new value of ND3 and repeatthe procedure.

program calls the subpro-

The computer program MLTCYC implements this scheme.This
and HERMIT (see
grams nis,Lre (ie Example 2.g), GAULEG (seeSection 4.3.1),
Example 7.4)The input data are defined as follows:

PENR = specified overall penetration,fraction

MMD = massmedian diameter of initial aerosol,mm
SIGG = op dimensionless
Q = volumetric flow rate, m3/s
LAMBDA = mean free Path, Pm
TEMP = temperature, K
 358 Chap.8' Cyclonic
Devices Sec.8,6C

DENP = particle density,kg/m:

VIS = gasviscosity,kg/m-s
KIP = K1K.K6
K2P = K2tNsppz | (2EKozK 62)
K3P = Ks

Theoptimaldesignfor an overallefficiencyof 7TToisiy',= 1,310cyclones,D" = 0.30m.

Thetotal annualcostis TAC = $155,700.Thepenetratingaerosolis log-normallydistrib-
utedwith MMD = 2.39 trtmand o, =2.16.

BLOCKDATA
REAL PENR,MMD, SIGG,Q, LAMBDA, TEM?. DENP,VIS. KIP, K2P,K3P,K
coMMoN/BLOCKI/PENR, MMD, SIGG,Q,LAMBDA, TEMP,DENP,vIS,
. * KlP, K2P,K3P,KIPP,K2PP,K3PP,K
DATA PENR/0.30/,MMD/4.00/, SIGG/2.50/,Q/165./,LAMBDA/0. 1500/,
* TEMP/450.0/,DENP/1600.0/, vrst2.48B-st,Ktpt302.40t,
* KzpI I.t tBgI, K3p/3l. l0/, w 55l.3t
END

PROGRAMMLTCYC
REALPENR, MMD, SIGG, K, Q, LAMBDA, TEMP, DENP
REAL VIS, KIP, K2P,K3P,X1, X2, XACC, KIPP, K2PP,K3PP ' '
REAL CA, CB, DIAM, ND3, DPI, RTSAFE,RTBIS,SUMA
REALW Y, G, D, TAU, ENE,PEN,QO .. ...
INTEGER N,, NC
LOGICAL SUCCES .o

EXTERNAL FUNCD, FI.JNCP,FI.]N

COMMONiBLOCKI/ PENR,IVIVID,SIGG.Q, LAMBDA, TEMP,DENP,VIS,
. '* KlP, '' :.
K2P,K3P,K1PP,K2PP,K3PP,K
COMMONDIAM ..
PARAMETER( N: 10,FACTOR=1.6,NTRY = 50) i : : :
DMENSTONY(N),W(N)
TAU l=(MMD* 1.0E-6)**2*DENP/(l8.*VIS)
X I =l .43x149' *Q*IV(-0.5*LOG(PENR))**2.86
X2=Xl/SICG
F1 = f'gY6P111,
F2 = FI.JNCP(X2)
SUCCES=.TRLIE.
D O 1 J = 1 ,N T R Y
IF(FI*F2.LT.0) cO TO 2
IF(ABS(FI) .LT. ABS(F2))TIIEN
X1 =X1+FACTOR*(X1-X2) , ..
F1=FLINCP(X1)

''
-+*r:U6f,!tll li
!'
 See;,8i6;eostAnalysis.,for.eyclones 359

j' r r':.: : j
ELSE
' 'r'"r':r:'.!'' ' ': : "' :
X2=X2+FACTOR*112-X1)
F2=FLINCP(X2)
ENDIF
CONTINI.IE.
SUCCES=.FALSE.
.";'rr' I' '
IF(SUCCES)THEN"' "::::':r:'
XACC=1.0F-3
ND3=RTBIS(FUNCP,X l,X2,XACC)
',
PRINT *, 'ND3 = ND3
t :lrl

I NC=n$T(ND31(DIAM**3))
: :r U,qM=(ND3/NC)**(1./3.) ' j':.,
.. K3PP/DIAM**J :'::.! - .' :'
TAC;.KIPSDfAIVI' +I(2PP*DIAM*t2.'
:'i:
PRINT*,'NUMBEROFCYCLONES=',NC'i
,DIAMETER,OF ,; ', i , : .:
PRINT *, EACH CYCLONE= DIAM
rpffi1'*':;TOTAIANNUAL|95I=i.;TAC . ., :r':
-i . ': :
. PAUSE'PRESSTIIEENTERKEYTOC'OI{TINUE'
PRINT *, 'ENTER PARTICLE STZE,ENTER O.OTOITERMD{ATE
READ *. DPI " "' . I
IF (DPI .NE. O.O)TIIEN
Y14.0
CALL GAIJLEG(Y I,DPI,Y.W,N)
*0.3
*0.14)*(TEMP/283.)*
--ENE=1.-(1.-0.67*DIAII*
SIIMA=0.0
DO l0I=l,N
':D=Y{D
,,,trI(0'55+DILAMBDA.GT.80')TIIEN
: I'CUN=.1..+,?.*tAMBDA/D
ELSE
.,' CUN=I.+(2.tLAMBDA/D)*(1.25?'$.4*EXP(-0-55*D/
-' ''' .::;
''LAI4BDAII . :: ':
ENDIF
TAU=CIIN*(1.08-6*DJ**2*DENP(18.*WS)': "
ARGUM=2.*(0<re*TAU*@NE;t))ArID31**11,7. : r;''
: 'r ' " ': I rr:; '' r
, : (2.*ENE+2)) r' "'r "l
rF(ARGUM.GT. 80.0)THEN :
: Pf,lf = Q.Q
', r ' .i i i r '
ELSE : . ' : i . rr' , : . ,
.]i:..I:] :::.:
PEN=E)(P(-ARGUM)
i rr":-l:1
ENDIF
-
Q0=(1.(D*SQRT(2.*3. 14I 6)*LOG(SIGG))*EXP(-(
ICIG@/I\{MDy(SQRfi2:)*DoG(StCC1;**2;''
SLJMA=SUMA+W(I)*PEN*Q0
10 . coI{nNUE
.G=-SUMA/PENR

I
I
360 Chap.A CyclonicDevices Sec.'8.

PRINT *, 'MASS CLJMIJLATIVE FRACTION=', G

PRINI *, 'FOR PARTICLE SITF = DPI
GOTO5 -
" :
ENDIF
GOTO 15
ENDIF
PRINT *, 'FAILURE IN BRACKETING OPTIMI,JMDIAMETER'
GO TO 15
15 STOP
END

FIJNCTIONFUNCP(ND3) :
REAL KIP,K2P,K3P,KIPP,K2PP,K3PBND3,K,MMD,
LA\8DA; NMD
LOGICALSUCCES
EXTERNAL FUNCD, F{]N
coMMoN /3LOCKl/ PENR,MMD, SrGG,Q, LAMBDA, TEMP, DENP, VIS,
+ KIP, K2P,K3P,KlPP,K2PP,K3PP,K
PARAMETER(N=10)
COMMONDIAM
DTMENSTONX(N),A(N) :
KIPP=KlP*ND3
K2PP=K2P.lD3**2
K3PP=K3P*ND3
CA=KIPP(2.*K2PP)
CB=(3.*K3PP/(2.*K2PP)
Xl=0.0
X2=1.0 '
,, ,.
XACC=1.0E-3
CALL ZBRACI(FUN,XT,X2,SUCCES,CACB)
rF(succEs)THEN
DIAM=RTSAIIE(FUNCD,XI,X2,XACC,CA,CB)
EPS=1.0E-6 ,, :
CALL IIERMIT(N,X,A,EPS)
ENE=1.-(1.-0.67*DIAM'<*0.
I 4){'(TEMP/283.)**0.3
NMD = E)P(LOG(MMD)-3*(LOG(SICG))**2)
SUMA=O.0
DO l0I=1,N
D=E)(P(SQRT(2.)*LOG(SIGG)*X(I)+LOG(NMD))
IF (0.55*D/LAMBDA .GT. 80.0)THEN : i ,
CUN=1.+2.*LAMBDA/D
ET,sE
CUN=1.+(2.*LAMBDA/D)I(1.257+0.4*EXI,(-0.55'&D/
LAMBDA)
ENDIF
TAU=CLJN*(1.0f-6*D)*'i2l'DE|11p(I 8.*VIS)
ARGUM=(2.*((K{,Q*TAU*(ENE+ I ))ND3)* *( l./
 for Oyclones
Sec.8.6 CostAnalysis 361

(2.*ENE+2.)))
rF (ARGUM .GT. 80.0) THEN
PEN = 0.0
ELSE
PEN=H(P(-ARGIJM)
ENDIF
SUMA=SUMA+A(I)*PEN*HG(3. *SQRT(2.)*LOG(SIGG)*X(I))
IO CONTINTJE
PENC=SUMA{SQRT(3. *2))
l4 I 6)*E)(P(4.5*(LOG(SIGG))*
ABC=PENC-PENR
FUNCP=ABC
ENDIF
END

STJBROUTINE FI,]NCD(X,FN,DF,CA,CB)
FN=X**5-CA*X**2-CS
DF=5.*X4'*4-2.*CA*X : i :
RETURN
END
,.
FTJNCTTONFLTN(X,CA,CB) :
FUN=X**5-CA*X{<t2-CB
RETIJRN
END

FUNCTTONRTBIS(FLTNCPXI,X2XACC)
@ 1986by NumericalRecipesSoftware.Repdntedwith permissionfrom Numerical
Recipes:The Art.of Scientific Conputing Canbridge University Press,New Yodr (1986).
PARAMETER (JMAX=40)
FMID=FUNCP(X2)
F=FLINCPCXI) . tl, '
'Root mustb€ bracket€dfor hisoction.'
trr(F*FMID.GE.0.) PAUSE
IF(F.LT.O.)THEN
RTBIS=XI .:i, '. ,'...
r :
DX=X2-Xl
ELSE
p1g1i=X2 ':
ot=it*z
ENDIF
' r
DO llI=l.JMAX
DX=DX*.5
XMID=RTBIS+DX
FMD=FUNCP()CvIID)
IF(FMID.LE.0.)RTBIS=XVIID
rF(ABS(DX).LT.XACC .OR. FMID.EQ.O.)RETURN
362 :' .Cyelonic
Ohap::8:, Dbviees R€#ft

11 CONTINUE :
PAUSE 'too manybisections' . .l
END
, :,
SUBROUTINEZBRACI(FI,JN,X1,X2,SUCCES,CA,CB) .
@ 1986by NumericalRecipesSoftware.Reprintedwith permissionfrom Numerical
Recipes:The Art of ScientifieConputing, CambtidgeUniversityPress,New York (1986)'
EXTERNALFI,]N
PARAMETER(FACTOR=I.6"T'{TRY=50)
LOGICALSUCCES : . '.]'
IF(X1.EQ.X2)PAUSE 'You haveto guessan initial range' :
Fl=FUNCXI,CA,CB)
F2=FIINCX2,CA,CB)i \ . :
SUCCES=.TRIIE.
DO 11J=1,NTRY
IF(F1*F2,LT.O.)RETURN
rF(ABS(Fl ).LT.ABS(F2))THEN
X1=X1+FACTOP*1f,1-X2)
F1=FI.]N(XI,CA,CB)
ELSE
X2=X2+FACTOR*(X2-X1)
F2=FUlrl(x2,CA,CB)
ENDIF
II CONTINIJE
SUC'CES=.FALSE.
RETURN
END r:.

8.7 CONCLUSION
Cyclones are very simple devices for particulate removal. They arel' effective to remove
relatively big particles from waste gas streams. The pressure drop as the gas flows through the
cyclone is the main factor limiting the removal efficiency achieved. Multiple cyclone systems
exhibit high removal efficiencies at moderate pressure drops. Careful selection of the number and
size of cyclones operating in parallel in a multiple cyclone system can result in significant annual
savilgs. When these devices are used as pre-cleanersfor more sophisticatedparticulate conftol
equipment, such as electrostatic precipitators or fabric filters, the size distributiorr of the aerosol
penetrating the cyclonic device must be characterized. The next chapter Shows hsw the design of
an electrostaticprecipitator dependson the entering aerosolpopulation characteristics.
Fleferences 363

REFERENCES

Alexander,R. M. Proc. Austral. Inst. Min. and Met. (N.5.), 152:202(1949).

Cooper,C, D., andAlley, F. C. Air PollutianControl;A DesignApproach,PWS

Engineering,.Boston,MA(1986)., . , . ,'

Crawford,M. Air PollutionControlTheory,McGraw-HillNew York (1976).

..
of Air PollutionEngfnbering,Prentice
Flagan,R. C., andSeinfeld,J'.H. Fundamentals
Hall, EnglewoodCliffs, NJ (1988).

Licht, W. Air Pollution Control Engineering,Basic Calculntionsfor Particulate

Collection,MarcelDekker,New York (1980).

Peters,M. S.,andTimmerhaus,K.D. Plnnt Designand EconomicsforChemical

Engineers,4th ed.,McGraw-HilI, NewYork ( I 99I ).

G. 8., andLapple,C.E. btd. Eng. Chem.,3l:972(1939).

Shepherd,

C. J. Trans.Inst.Chem.Engrs.,29:356(1951).
Stairmand,

Swift, P: SteamHeatingEng'g.,38:453(1969).

of AirPollutionControl,Lewis,Chelsea,MI (1990).
Vatavuk,W. M. Estimating Costs
364 Chap.I CyclonicDevices Proble*

.PROBLEMS The gasv

while flor
': :

The,problems,atthe end of eaeh.chapterhave beengroupedinto

four classes(designatedby a superscript after the problem num-
and I
ber)
,'{a)r
_ - Illusftatesdirect numericalapplicationof the formulasin the
- _ _ _ ' a:
Class
text.
'b)r
Classb: Requireselementaryanalysisof physicalsituations,basedon
the',srrbject:material in thechapter. I,'.
Classg;'Requifessomewhatmorematureana1ysis:.:.-::i:::: c) IJ
Class,d: Requirescomputels-olution.

8.1a.Ideal,laminarcyclonicflow
8.4b. AX
A streamof 15m3/sof air at 298,K apd I atmflows in.laminarcyclonic flow Sho
througha dUctwheretheradii are0.5 m and 1.0m, andtheheightis 2.0 m. What angle
efficiency.?The,pafticledensi-
of turn is necessaryto collect 20-pm particlps,withperf.eCt
tv is 2.000ks/m3. t. '

Answer:14.35rad wherc
lV. =
l

8.2a. Ideal, turbulent cyclonic flow a-

RepeatProblem8.1 assunringturbulentflow- What angleo-fturq is necessaryto

collect 20-pm particleswirh.99Voefficiency?
Answer: 88 rad

&5b. DcC
8.3u. Collectioir efficiency for a cyclone of standard proportions .Dqd
,,
kg/mrfiu
A cyclonewith a body diameterof 1.0m andwith Stairmandstandardproportions themqir
processesair at298K and I atm,which carriesparticleswith a densityof 1,000kg/m3. chosen

** -*
?.|*
Probleins 365

The gasvelocityin the entranceductis 20:0m1s.The numbetof,turnsthatthegasmakes

while flowing througha cycloneof standardproportions,N,, is givenby

y ,' 2
-Ha+ h

and0J= ZnNlrad,.Estimate
the collectionefficiencyfor 10-pmparticlesassuming
, :. l,'
-, .i .a'

(a) Ideal"laminar cyclonic flow l

Answer: 56Vo
b) Ideal,turlulent cycloniiflow
Answer:38.2Vo
c) Leith-Licht semiempiricalmodel applies
:
Answer:81.3Vo

8.4b. Alternative form of Leith-Licht model

Showthat,for a multiplecyclonesystem,Eq. (8.20)canbe rewritten'as

L .M\
I - expl-rl-!e -P\
n(po)=
\ tuN"o;1 l
where
& = the numberof cyclories
t = relaxationtime for a particle of diameterDo, which is given by
:'.
,. _r. , D - zYnP,"cc _ ,, . :.
':'r, t-"P
i : :.' ,':,:
lgp '-
,,. r' :, i . -' 1,.
. . ,l , - ,

8.5b.Designof a cycloneof standard proportions

Designa cycloneto remove80Voof pafiiclesof 20 pm diameteranddensity1,500
kg/mr from a streamof 20 m3/sof air at zg}Kand I atm.Determinesuitablevaluesfor
thgmajordimensionsof thecyclone,assumingSwift standardconfigurationhasbeen
chosen.
Answer:H=18.25m
366 Chip.'8 GyclonicDevices Problefi

' ' :i ::'. I

8.61.Design,of,'acyeloneof standardproportions ' ' E.10b.sd
RepeatProblern8.5for Lapple standardconfiguration' Efl
meimshil
100%'in
v.eloc*yr
8.7d.Overall efficiency of a cyclone of standard proportions is due.bl
Z,enz(El
Estimatethe overall efficiency for the conditionsof Problem8.5 if the particlesare
posdtb
log-norrnallydistributedwith MMD = 15 pm andor = 2'0' Approximatethefuitegral
the,rya4
with a 16-pointGauss-Hermitequadfatureformula.
mqotrtd
i :, ,..',; AnsWr:73.8Vo
.......-.

' I
' ;l::rl:_ : I

8.84.Pressuredrop through a cyclone Cil

,__-
Estimatethe pressuredrop andthe powerto overcomeit if the mechanicalefficien- tion$,fi
cy of the motor-blowersystemis 60Vo
(a) For Problem8.5'
- : :.
Answer:17-8kW
8.flq01
(b) For Problem8.6 Ttx
Answer: 18.3kW norma[51
enlrryr
8.9d.Pressuredrop and overall efficiency be remor

A streamof ur at298K and 1 atm flows at the raie of 10 m3/sandcanies with it

particleswith a densityof 2000kg/m3. The particlesarelog-normally disnihred with 8.1.2c.8
iuffrrfO= 10 pm ando, = 25, You must designa cycloneof Swift standardconfiguration
,Tq
to serveasa precleanerfor this stream.A blower is availablewith a motor rasedat'20 plecyck
kW and a mechanicalefficiency of 6|Vo:Incorporatingthis blower in your desrgp specilI ISVo)is
'
fy the cyclonebody diameterandthe overall efficiencyrachievable. a siiafiI
An*ver:76.l%o the sd
': .a
cyclone
52m3lm
a densig
Problems 367

8.10b.Saltation effect
Equations(8.20) and (8.21) suggestthat as the sizeof a cyclonedecreases, which
meanshigher inlet velocities, the grade efficiency will continue to increaseand approach
I00% inthe limit. However, it is known that this is not so. There is a limit to the inlet
velocity above which further increaseresults in a decreasein collection efficiency. This
is due to the reentrainmentof particles by the saltation effect describedby Kalen and
Zenz(Kalen,B. andZenz,F.A.A.I.C4.E.Sympos.Ser, 70:388,1974).Licht (1980)pro-
posedthe following empirical correlation to estimatethe inlet velocity, vy,thatresults in
the maximum cyclone collection efficiency, just before the saltation effect becomes
important:
2
Ku1
vm= 3,025W ,ro'zot
-
072| t<'

Consider the multiple cyclone systemof Example 8.6. Estimate vx4for those condi-
tions, and compareit to the actual inlet velocity.
Answer:vtv = 26.7m./s

8.LLd.Overall efficiency of a multiple cyclone system.

The aerosolpopulation entering the multiple cyclone systemof Example 8.6 is log-
normally distributed with MMD = 4.0 pm and o, = 2.5.The particulate loading at the
entranceof the systemis 0.028 kg/m3. Estimate the rate at which the collected dust must
be removed from the system,in kg/d. Neglect the Cunningham correction factor.
Answer: 308,600kg/d

8.12". Effects of hopper evacuation on the performance of multiple cyclone systems

Tucker et al. (JAPCA39:1614,1989)showedthat the collectionefficiency of multi-
ple cyclone systemsincreaseswhen a small amount of the gas flow (typically less than
l5fto) is withdrawn from the dust hopper.This hopper evacuationflow then goesthrough
a small baghousebefore joining the remainder of the systemexhaustflow and going out
the stack. They demonstratedthe concept experimentally in a systemcontaining 10
cyclones of 0.25-m body diameter and 3.61-m overall body height. The air flow rate was
52 m3lmin at298 K and 101.3kPa. The particulatematterwas sphericalglassbeadswith
a density of 2,200 kg/m3. The size distribution function was approximately log-normal
 Chap."8 CyclonicDevices Proble
368

effi- numbera
with MMD = 12 pmand o, = 2.0. With no hopper evacuation,the measuredoverall
to 95Vo . the aeros
ciency was89.5Vo. With I4Eo hopper evacuation,the overall efficiency increased
The authors suggestthat the improved performancewith hopper evacuationis due to a
higher gas velocity and a more uniform gas distribution through the individual cyclones.
(a) Estimate the value of the geometric configuration parametet,K, for the cyclones 8.17dM
in the experimental work describedand compareit to the Swift and Stairmand standard The
configurations.
MMD =
(b) If the improved performancewas due solely to the effect of higher gas veloci-
gasfrom
ties, estimatethe percentincreasein gas velocity to explain the observedefficiency of
95Vo.
ertiesare
To satis
Answer:l90%o
removed
sideredf
8.L3a.Cost of a single cyclone De
Consider the cyclone of Problem 8.9. Calculate the total capital investment and Specify
total annual cost. The useful life of the cyclone is 5 yr and the minimum attractive return cyclone
on investment is 207olyt The cyclone operates8,000 h/yr; the cost of electricity is calculat
s0.06lkw-h.
TAC= $37,800
Answer:

8.14b.Optimization of multiple cyclonesystemdesign

RepeatExample8.9,but for Swift standardconfiguration.Calculatethe numberof
cyclonesandbodydiameterthat minimizethetotal annualcost.
Answer: TAC= $312,000/yr

8.15bEffect of the costof electricity on multiple cyclonesystemdesign

RepeatExample8.9,but for a costof electricityof $0.041kW-h.Calculatethe num-
ber of cyclonesandbody diameterthatminimizetheTAC'
Answer: Dc = 0.192 m

8.16d.Optimization of multiple cyclonesystembasedon overall efficiency

RepeatExample8.10for cyclonesof Swift standardconfiguration.Calculatethe

Problems

numberandsizeof cyclonesthatminimizeTAC. EstimateTAC, andMMD ando* for

the aerosolpenetratingthe system.
Answer:TAC= $175,800

8.1.7dMultiple cyclonesystemfor a Portland cement kiln

The cementdustfrom a Portlandcementkiln is log-normallydistributedwith
MMD = 12pm ando* = 3.08,havinga densityof 1,500kg/m3(Licht 1980).The waste
gasfrom thekiln flows at therateof 377m3lmin,at394.4K and 101.3kPa,andits prop-
ertiesaresimilarto thoseof air.Theparticulateloadingof the wastegasis 0.023kg/m3.
To satisfythe correspondingNSPS, 99.8Voof theparticulatematterin the gasmustbe
removed.A multiplecyclonesystem,followedby a fabricfilter or an ESPis beingcon-
sideredfor this purpose.
Designthemultiplecyclonesystemfor overallefficienciesof 80Vo,85Vo, and907o.
Specify,for eachcase,which standardconfigurationto use,the sizeandnumberof
cyclones,TCI, andTAC. Characterize the sizedistributionof thepenetratingaerosol,and
calculatethe particulateloadingat the outlet.The following dataapply:
. Usefullife, 10yr; nosalvage
value
. Minimum attractivereturn on investment.207o
. 8,400h/yr ofoperation
. Costof electricity,$0.08&W-h