0% found this document useful (0 votes)
435 views14 pages2 Phase Flow Orifice
orifice
Uploaded by
Pinjala AnoopCopyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF or read online on Scribd
0% found this document useful (0 votes)
435 views14 pages2 Phase Flow Orifice
orifice
Uploaded by
Pinjala AnoopCopyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF or read online on Scribd
/ 14
Manmatha K. Roul
Dipset Mecha Ere
hata tuto Engneing ad Teena)
tack ca 7565918
emai: hou} tgp atin
Sukanta K. Dash
Departmen Mecha rlesig
a stu at Tzanalgy,
rap, in 727300
Single-Phase and Two-Phase
Flow Through Thin and Thick
Orifices in Horizontal Pipes
Tworphave flow pressure drops through thin and thick orifices have been numerically
investigated sith air-water flows in horizontal piper Twosphase computational fluid
‘dynamics (CFD} calculations, using the Eulerian-Eulerian model have been employed to
calculate the pressure drop through orifices. The operating conditions cover the gas and
liquid superfsal velocity ranges Vyq—0.3-4mis and Vu 05-2 mis, respectively, The
local pressure drops have been obtained by means of extrapolation from the computed
upstream and downstream linearized pressure profiles tothe ovfce section, Simulations
forthe single-phare flow of water have been carried out for local liguid Reynold number
{Re based on orifice diameter) ranging from 3 x 10" to 2% 10° to obtain the discharge
coefficient and the two-phase local multiplier, which when multiplied with the pressure
‘drop of water (for same mass flow of water and two phase mixture) will reproduce the
pressure drop for two phase flow through the orifice. The effect of orifice geometry on
twosphase pressure losses has Been considered by selecting two pipes of 6 mm and
40mm inner diameter and eight diferent orifice plates (for each pipe) with two area
ratios (00.73 and 6~0.4) and four diferent thicknesses (sid 0.025-0.59). The
results obtained from numerical simulations are validated against experimental data
fiom the literature and are found to be in good agreement. (DOL 10.1115/1.477267]
1 Introduction
‘The calculation of pressure drop due to gas-liquid two-phase
flow through an orifice is a problem yet to be solved in engineer
ing design Knowledge of pressure dkop for two-phase flows
through valves, orifices, and other pipe fitings are important for
the control and operation of industrial devices, such as chemical
reactors, power eneration tints, refrigeration apparatuses, oil,
‘wells, and pipelines. The ofiice is one of the most commonly
used elements in flow tate measurement and regulation, Because
fof ils simple structure and reliable performance, the ortice is
increasingly adopted in gas-liquid two-phase flow measurements.
Single orifices or arrays of them constituting perforated plates, are
‘often used to enhance flow uniformity and mass distribution
downstream of manifolds and distributors. They are also used (0
tenance the heal-mass transfer in thermal and chemical processes
(eg, distillation trays). Single-phase flows across orifices have
‘been extensively studied, as has been shown by Tdelehik eta. [1
in ther handbook, The available correlations do not always (ake
into account Reynolds number effeet and a complete set af ge0-
smeisical parameters. Some investigations have been made on the
theory and experiment of resistance characteristics of orifices
[17} and some useful correlations have been proposed. However,
Some of them cover only limited range of operating conditions
and the errors of some are far beyond the limit of tolerance. So
they are not widely used in engineering design. Major uncertain
ties exist with reference to two-phase flows through orifices. Few
experimental studies reported in the literature often refer 10 a
limited set of operating conditions. With particular reference (0
office plates, some of the correlations and models [24] are dis-
cussed by Friedel (5]. Other references are the early study by
Janssen [6] on two-phase pressure loss across abrupt area contac
tions and expansions taking steam water flow, the work by Li [7]
‘on two-phase flow measurement with sharp edge orifices: the
recent experimental investigation by Saadawi et al [8], which
Journal of Fluids Engineering
Copyright © 2012 by ASME
refers to two-phase flows across orifices in large diameter pipes
and the work by Kojasoy et al. [9] on multiple thick and thin
torfice plates, Rertola [10) studied void fraction distribution for
ai-water low in a horizontal est section with sudden area con:
traction and found thatthe sudden contraction considerably affects
the gas distbution in both the upstream and the downstream
pipe. Fossa and Guglielmini [11], Fossa et al. [12], and Jones and
Zuber [13] experimentally investigated two-phase flow pressure
{drop through thin and thick orifices and observed that the void
faction generally increases across the singularity and altains a
‘maxinvum value just downstream of resection, Shedd and Newell
[14] found a unique set of Liquid film thickness and pressure drop
data for horizontal, annular flow of air and water though round,
square and tnangular tube using @ noninvasive, optical uid film
thickness measurement system,
‘The information regarding the elfects of orifice thickness on
two-phase pressure losses is not available inthe literature. Most
models require the knowledge of local void fraction, which i usu
ally calculated by means of corelations for straight pipes without
tlices and hence the actual void fraction distribution du to oi
fice interactions is not considered. In the present study the effect
of orifice geometry on twe-phase pressure losses has been consi
cred by selecting two pipet of 60mm and 40mm inner diameter
and eight different onifce plates (for each pipe) with two area
ratios (60.73 and = 0.54) and four different thicknesses three
thin and one thick orifice) (s/d = 0.025, 0.05, 0.2, 0.59). When the
value of sid is below 0.5 itis called a thin orifice otherwise tis a
thick orice [3]. The results presented in this study provide useful
Jnformation on the reliability of available models and correlations
‘when applied to intermittent lowe through onfices having high
values ofthe cooteaction atea ratio,
2. Theoretical Background
24
igle-Phase Flow. For the flow through a thin orifice
(Fig. 1(@), the flow contracts with negligible losses of mechanical
energy, © a vena contracta of area A, that forms outside the
restriction, Downstream of the vena contracta the flow expands in
an irreversible process to the pipe wall of flow area A. If the
SEPTEMBER 2012, Vol. 134 / 091301-1 Fig.1. Single-phase flow across (a) thin and (b) thick orifices
cotfce is thick (Fig. 1(6)), downstream ofthe vena contracts, the
flow reataches tothe wail within the length of the geomettical
contraction and can even develop a boundary layer flow til it
finally expands back into the pipe wal, Acconding to Chisholm
[3] the thick orice behavior sakes place when the dimensionless
nice thickness to diameter ratio, ade greater than 0.5. Assume
ing that each expansion occurs ireversibly and the fii come
preeible, the single-phase presse drop AP, in atin once can
bye expressed at function of the flow area fatio 7 ~ (4/D) and
the contraction cocticent «2 — A./Ae as
y ©
wey
@ [lew
‘where p is the fluid density and V its mean velocity
If the orifice is thick, the loss of mechanical energy is due to
the dotie expansion a described! above. For these conton the
sinleshase veal prestte drop can be exeiied
) @
[eae
“The local pressure drop can also be expressed as a function of ori
lice discharge coefficient C, (Lin {7}, Grace and Lapple (13),
Py
Py
@
From Bqs. (1) and (3), 6 fo thin oiices can be writen as
1
ae ®
ot vio /Cy “
Similarly from Eqs. (2) and (3), 0; fr thick orifices can be
1
o
1+ /[a-A/G] 1
“The well-known Chisholm expression for contraction coefficient
fn terms of the area flow ratio only (Chisholm (3). Bullen eta
[16], and Benedict [17] is given as
091301-2 / Vol, 134, SEPTEMBER 2012
—_
coe
]
o
Inthe present study the pressure drops across different orifices for
single phase flow of water are obtained numerically, from which
the discharge coeticient is calculated using Eq. (3). The coated
tuon coefficient is calculated using Egs. (4) and (S) for thin and
thick orifices, respectively. This contraction coefficient is com.
pared with the Chisholm correlation az given by Eq, (6).
22. TwosPhase Flom. Accoxing to Chi [3] nd Mons
{eh the dpa ined a he ao of ga pase vel fo
th gud pase eos any pon in the Now path Cole and
Thome 18) and ae function of te quality yan thet of
Aides When the ny of xe, ey Tw (ch
ts thre conideved in the pest nvesiguon, 20008) fe
tip, Sean beexprewedsr
‘)} o
5 | ot
fe
The quality, is dened as th rato of mass fax of gas the tal
tsa of mote at ny ros secon [18] The mass fay of
fuss gus mss fow rv ide by ol crosses
Stier wkeren the teal mason the tt mass Row re of
tty divided by iol cevesectonal ea athe pipe Mahe
tatty the qty team exeetsed
BBV Aco
PieVal Aan + PrRVaT As
os
Vn Fa
o
‘The Armand and Threschev correlation [19] for fully developed
flow. near atmospheric pressure, can be expressed in terms of the
gas volume fraction x, = Vig (Vig + Vt).
0.833,
Som —«)
o
Kojazoy etal. [9] adopted the Chisholm expression for sip rato S
but they suggest a correction to account for the effect of flow
restriction on sli ratio,
a0)
The exponent mis zero at the vena contracta and downsteam of i
ies the slip ratio is expected to be one) while n is equal to 0.4
and’ 0.15 in the upstream region for thin and thick orifices,
respectively
‘Simpson etal (2] adopted different correlation for sip eval
tion that does not account forthe quality ofthe mixture,
ay
The prediction of the two-phase multiplier 2, can be calculated
using diferent madels as deserted hele. Te two-phase rl
plier is defined asthe ratio ofthe two-phase presture drop through
the once to the single-phase pressre drop obtained at Liquid
‘ass foe alto the overall ro-phace mast Sx
the nntie i considered homogenous (S
ing expression can be obtained
2), the follow.
x) a2)
‘Transactions of the ASME Chisholm 3] developed te felowing expression:
ey -1+ (2-1) eae a
Pe
where he prance B can be assumed to be for thin ies
Tats fortek one,
“ti Moms tlaeshp [i] sles to thin ences and gat
vale nd has Ge follwing expression
ay,
where, the slip ratio Sis given by Ba (7)
Simpson etal. [2] proposed the following relationship based on
slip predictions given by Eq, (11)
= 1 +x(S — 1) [0 +468 ~ 1) as)
“The Simpson model is based on data collected with arg diameter
pipes (up to 127mm) at mixture qualities generally higher than
those obtained inthis work (2 < 0.005).
‘Finally the comlation of Saadawi ct al. [8], based on experi-
ments caried out at nea atmospheric pressure with a very lage
ameter pipe 203 mm). is given by
Oh = 14 1-795
a6)
In the present study the pressine drops acrts different orifices for
two-phase flow of air-water mixtures are obtained numerically
‘The single-phase pressure drops at liquid mass fux equal to the
overall two-phase mass flux are obtained by interpolating the
single-phase pressure drop results. The two-phase mulplier is
‘obtained by taking the ratio of the two-phase pressure drop to that,
fof the single phase pressure drop. The two-phase multiplier thus
‘obtained is compared with the theoretical predictions from the
above equations (Eqs. (12)-(16)).
3 Numerical Modeling
‘The flow Geld is modeled using the averaged Reynolds equa-
tions with realizable per-phase k-s turbulence model, with the two
layer near-wall eatment. The governing equations are briely
described below
34 Governing Equations. Here we considered the 1wo-fuid|
‘method or Eulerian-Hlerian model, which considers both the
phases as interpenetrating continuum, with each computational cell,
ff the domain containing respective fractions ofthe continuo and
dlspetsed phases. We have adopted the following assumptions in our
study which are very realistic forthe present situation.
Assumptions
1. The ffuids in both phases are Newtonian, viscous and
incompressible
2, The physical properties remain constant
5, Nomass tancier between the two phases
44, The pressure is assumed to be comtmon to both the phases
5, The realizable kee turbulent model is applied to describe the
behavior ofeach phase
6, The surface tension forces are neglected; therefore, the pres
sures of both phases are equal at any cross section,
7. The flow is assumed to be isothermal, so the energy equa-
tions are not needed
With all the above assumptions the governing equations for
phase q can be writen as (Drew (20). Drew and Lahey [21],
Crowe etal, (22):
Journal of Fluids Engineering
Continsity equation:
ay, (eap.t) =
Flore) + ¥ (279i) =0 an
The volume fractions are assumed to be continuous fonctions of
space and time and their sum is equal to one
uy bya) a3)
Momentum equation:
8 con) a¥- lupin @)
EZ toaegie) +9 (ayo) = ah + V+ () + pa +My
a)
2 the ath phase stress tensor
> tani (TH, VE) (20)
He = B= tq en
where My is the interfacial momentum transfer term, which is
given by
My = th + Mi 2
where the individual terms on the righthand side of By, 22) are
the drag force, vitual mass force, and hit force, respectively. The
drag force is expressed as
3
i= seco
23)
The drag coetfcient Cp depends on t
as given below (Wallis [23).,
particle Reynolds number
Co~ 24(1 +015 Re /Re,) Rey < 1000
= oat Reo 100
Panicle Reynolds number for primary phase q and seconds
phase pis given by
“ald
%
Re,
? @5)
Equation (23) shows thatthe dag force exerted by the secondary
phase (bubbles) on the primary phase is a vector directed along,
the relative velocity of the secondary phase. We have varied the
diameter of the particle trom 10 to 100 micron and have not seen,
any change in the pressure profile within this range of ditmeter
change
‘The second term in Eq, (22) represents the virtual mass force
‘which comes into play when one phase is accelerating relative to
the other one. In cate of bubble accelerating ina continuous phase,
this force can be described by the following expression (20)
Mi = MM = Cinapy @ 6)
where Cyiy is the virtual mass coefficient, which for a spherical
particle is equal to 0.5 [20]
‘The third term in Eq. (22) i the lift force, which arises from a
velocity gradient of the continuous phase in the lateral tection,
and is given by Drew and Lahey [21],
tp = My
H)x(VXR) 2D
Cir 04liy
SEPTEMBER 2012, Vol. 134 / 091301-3 ‘where Cis the lit force coefficient, which for shear flow around
spherical droplet is equal to 0.5.
“Turbulence modeling:
Here we considered the realizable perphase k-e turbulence
‘model (Launder and Spalding [24], Shi etal. (25], Troshko and
Hassan (26),
‘Transpo equations fork
3 loeeahe) + ¥-(sep,eke)
= feet
“Transport eq
a, /
2 (ae) +¥ (see)
+ C3 [a Ce hi) —B(Wp~0,)- PEW
tt, -0) | »
where Us the phase-weighted velocity Here,
C= max|043,
pea
“The terms Cyy and Cy can be approximated as
ont nif) w
ty tt ep
oo
oo
where #4 is characteristic time ofthe energetic turbulent eddies
and is defined as
eelot @
sai
cya 18 138059 os
where 0 is the angle between the mean particle velocity and the
‘mean relative velocity, The characteristic particle relaxation time
091301-4 / Vol, 134, SEPTEMBER 2012
connected wth nial lfts acting on a seed pase pis
‘etre
sanekes (P+
oak (2
Ky is defined as the inter phase momentum exchange coef
cient In flows where there are unequal amounts of two fiids, the
predominant Muid is modeled as the primary Avid, since’ the
sparser fuid is more likely to form droplets or bubbles. The
exchange coefiicient for these types of bubbly, liquid-liquid or
‘zar-liguid mixtures can be writen inthe following general form:
ron 66
yh
were ihe ag fnction is etd ferent er he dren
exchange-coelfcient models and ty. the “particulate relaxation
time," is defined as,
out
Th
were dis the diameter of the bubbles or droplets of phase p
“The eddy viscosity model is used to calculate averaged facto
ing quantities. The Reynolds sress tensor for continuous phase q
ts piven as
+ var)
fm) 3D
The turbulent viscosity i, ie written in terms ofthe turbulent ki
netic energy of phase
&
Pha = see 8)
The production of turbulent kinetic energy, Gz is computed from
Guy = hg C5498) o
The production em in the eguaton (he second temo he
‘chon sie of Ey) doesnt cont te se Cea
tte incu, knots Seale ae nc ene
Aguas Bet 9) doesnt ve ey sag Sh
‘nmittorover ute evn atl was onfeane set
arta enue h coated wih sete he el
hk hvet singly dc to tin donate Caeser
TETERG Cause Eo ot 2 staat ee compe
com
G 40)
where
VSS) +0405 ay
and
By = Oy drow
24 = Th ~ son
‘Transactions of the ASME ‘where, Ty ie the mean rate of rotation tensor viewed ina rotating
reference frame with the angular velocity Ty, The constants Ap
and A, are given by
Ao= 404, A, = VBcosd
where 9
“(e-8)
Jeos!(VEW), w= 2, §
Cur 14s, G19, = 10, 9.12
Interphase turbulent momentum transfer:
‘The turbulent drag term Kpy (Fp ¥,)i8 modeled as follows
ale ~F) = Kra(Hp Us) Kina)
Here U, and Oy ase phaseaweighed velocities, and Fay isthe
Re}, Re
rmodele are employed (Launder and Spalding [24). In the viscoe-
ity affected near wall region (Re, < Res), the one-equation model
of Wolfstein [27] is employed. In one equation model, the
‘momentum equations and k equation are retained. However, the
turbulent viscosity, xis computed from
see = PCL E (46)
‘while the rate of dissipation of turbulent kinetic energy « is com:
puted from
e
7
ap
The length scales fy and 1, in Bes. (46) and (47) are computed
fom Chen and Pate! [28],
a(t
be pa(t ets)
as)
49)
Here, the twe-layer definition is smoothly blended withthe igh
Reynolds number 1, definition (as described in the k-s models)
from the outer region,
Baga = Bite + ~ Ae aine 60)
‘A blending function J, is deined in such way that iis equal to
‘unity far Esom the walls and is zero very near to the wall,
som)
“The constant A determines the width of the Blending function. By
defining a width such that the value of 2, will be within 156 ofits
far field value given by a variation of AR,
on
are, |
aan 95) a
The constants are cy = AC", = 70,4, = Ber
34 Numerical Schemes. The governing equations of mass,
‘momentum, and turbulent quantities have been integrated over a
control volume and the subsequent equations have een disere
tized over the control volume using the finite volume technique
(Patankar [29)) to yield a set of algebraic equations. Boundary
conditions were implemented tothe nite volume equations which
could be solved hy the algebraic multigrid scheme of Fluent 6.3,
The flow field was assumed to be axisymmetric and solved in two
dimensions. The algebraic equations were solved using double
precision selver with an implicit scheme for all variables with a
vasiable time step stating at 0.00001 and finally going up to
0.0015 for quick convergence. The diseretization scheme for
momentum, ‘volume fraction, urbulent kinetic energy and
SEPTEMBER 2012, Vol, 134 / 091301-5 turbulent dissipation rale were taken tobe fist order up winding
intially for beter convergence Slowly as time progressed the dite
cretization forms were switched over to second one sp winding
and then slowly towards the QUICK scheme for beter accuracy
Iie tobe noted here that inthe Bslerian scheme of solution, wa
continuity equationt and two momentum equations were solved
for two phases. The phase-
