Вы находитесь на странице: 1из 8

Two-phase flashing flow methods and

comparisons

J. C. Leung and F. N. Nazario*


Fauske & Associates, Inc., Burr Ridge, IL. 60521, USA
*Exxon Research Engineering Company, Florham Park, NJ 078OI, USA

Two-phase flashing flow methods used by DIERS, the American Petroleum Institute (API) and
ASME are reviewed and compared. For saturated and low quality inlet conditions, the API
method was found to yield higher theoretical mass flow rates than the DIERS homogeneous
equilibrium flow model (HEM). The degree of disparity can be quantified by two dimensionless
physical property groups. Comparison with flashing flow data for water, refrigerant 12 and a
propane-butane mixture shows good agreement with the HEM prediction, but poor agreement
with the API method. The ASME method only recommends a saturated water flashing flow curve
that is in agreement with the HEM calculation to within 10%.

(Keywords: two-phase flow, flashing flow; modelling)

Two-phase flashing flows are frequently encountered in experimental data. The purpose of this paper is to
both practice and design, ranging from process flow review and evaluate these existing flashing flow proce-
streams through metering devices to pressure relief dures as recommended by the API, ASME4, and
sizing application. The most significant feature of DIERS organizations. This paper, however, will not be
flashing flow is the ease with which choking or critical concerned with the application of various flow/capacity
flow condition is attained. This is a maximum flow coefficients as employed in an actual safety relief valve
situation in which further lowering of the downstream sizing. Rather, the discussion will focus on the theore-
pressure will not lead to further increase in flow. For tical aspect of the flashing flow determination. It is
vapour or gas flow in nozzles or short pipes, the hoped that such a discussion will bring out the
pressure at choking is about half of the upstream differences as well as similarities among these currently
chamber (stagnation) pressure. However, for saturated employed methodologies.
liquid and low quality (vapour mass fraction) inlet
conditions, the pressure at choking is typically only Review of flashing flow methods
l&20% below the upstream pressure. Thus it does not DtEKS recommendation
take much flashing as a result of the depressurization DIERS recommends the HEM as the most appropriate
and fluid acceleration before the flow is ‘choked’, quite flashing flow formulation*. In this model, the flashing
unlike the gas case. two-phase flow mixture is treated much like the
Previously. most of the investigations on flashing classical compressible gas while undergoing an adiaba-
choked flow were related to the so-called ‘loss-of- tic expansion with equal velocities and temperature
coolant’ accident in nuclear reactor safety studies, and (thermodynamic equilibrium) in both phases. Among
therefore water was the fluid being studied most. Much the many other flow models tested in the DIERS
of these earlier works dealt with thermodynamic program, the HEM model yields conservative (low)
non-equilibrium and non-homogeneous (slip) flow estimates of the flow capacity. In flow passages of
behaviour in the attempt to explain the observed greater than 0.1 m in length (frequently encountered in
departure from the reference homogeneous equilib- relief systems), the HEM model in fact provides a
rium flow model (HEM) prediction. In the chemical best-estimate calculation. This observation is consistent
processing and petroleum refining industries, a more with previous investigations5-s.
widely practiced flashing flow method used in relief For flow in frictionless devices, the traditional
valve sizing is based on the API procedure I. The method for evaluating the HEM flow rate would
recently completed DIERS research recommends the require seeking a maximum in the mass velocity
HEM approach?J. which is supported by DIERS expression derived from the first law of
thetmodynamic9, namely,
Received 6 June 1989
Presented at the AlChE Loss Prevention Symposium. 2-6 April
G = Mho - hII”’
1989, Houston, TX, USA V (1)
095c4230/90/020253-08
0 1990 Butterworth &Co. (Publishers1 Ltd

J. Loss Prev. Process Ind., 1990, Vol3, April 253


Two-phase flashing flow methods: J. C. Leung and F. N. Nazario

where ho denotes the stagnation enthalpy, and the greater than 8.0) by
specific enthalpy and specific volume are expressed in
terms of quality and the individual phase properties,
G* - 0.9 (44
i.e. h, = hl, + xohf,,, h = hf + xh,, and u =
6
or simply
UJ + XU,,. In addition, the static quality is to be
(1.5
evaluated along a constant entropy line (isentropic
path)
(4b)
x = (S” - St)/SQ (2) A PI recommendation
The above procedures, while not complicated, are API Recommended Practice 521 provides only a brief
rather tedious and would require extensive thermo- description of how to handle two-phase flashing flow
dynamic property tabulation. The latter poses a special through relief devices. That text is repeated here for
problem in fluids without such extensive property completeness:
tables. The DIERS research had suggested a novel way ‘A pressure relief valve handling a liquid at
of carrying out the calculational procedure while vapour-liquid equilibrium or a mixed-phase fluid
reducing the number of flash computations’ will produce flashing with vapour generation as
substantially2.3. More recently, Leungr” proposed, for the fluid moves through the valve. The vapour
the first time, a generalized HEM correlation applic- generation may reduce the effective mass flow
able to all fluids including multicomponent mixtures capacity of the valve and must be taken into
with ideal solution behaviourlJ. The correlation is account. The quantity of flash vapour is common-
based on a single property parameter w given by ly calculated based on adiabatic flashing from the
relieving condition - either to the critical down-
. stream pressure (the critical pressure drop across
Vtl “0 f higo 1
Here all properties are evaluated at the known up- the valve orifice) or to the back pressure,
stream stagnation conditions (with the vapour prop- whichever is higher - and on an orifice area
erties for the case in which the fluid is a saturated liquid calculated for that vapour flow using the same
being based on the composition of the initial vapour pressure drop. An area is also calculated for the
produced when the liquid flashes). This correlation is remaining liquid quantity (after flash) under
reproduced in Figure 1, in which the normalized critical relieving conditions using the total pressure drop
D u”, and the critical pressure (the relieving pressure minus the back pressure).
mass flux. GT = G,/m
ratio, rlc = P,/P,, are correlated solely in terms of w. The orifice selected should have an area equal to
Curve fit equations were given in the original paperr”. or greater than the sum of the individual areas
For the case of saturated liquid inlet, the first term previously calculated’.
in the w definition (which can be approximated by the We could not locate the original paper that
inlet void fraction (Y(,) drops out and the critical mass discussed this procedure and its supporting data,
flux can be approximated (for fluids with o values although numerous recent papers have made reference
to the API flashing flow method1:-r5. In what follows,
we will attempt to give our interpretations of this
model. Our best understanding is that it is a frozen
separated (stratified) two-phase flow formulation with
an inlet quality evaluated based on an adiabatic flash
from the upstream relief condition to the choked
pressure of the vapour phase. Thus it can be characte-
rized as a two-step model, not necessarily rigorous in
the thermodynamic sense.

Step 1. Adiabatic flash to the vapour phase critical


(choked) pressure given by
/ L1 \k:(k - 11
PC,= p,, H
k + 1 (5)

In our discussion, we will assume that the downstream


back pressure is always less than this choked pressure.
The corresponding quality x,, would be given by
Equation (2). In practice, the use of an isenthalpic
(constant enthalpy) flash suffices here.

Step 2. Frozen flow of two non-interacting (stratified)


Figure 1 Generalized HEM correlation for flashing choked flow streams with an inlet vapour-liquid split as calculated
through a perfect nozzle (adapted from Ref. 10) above.

254 J. Loss Prev. Process Ind., 1990, Vol3, April


Two-phase flashing flow methods: J. C. Leung and F. N. Nazario

According to continuity consideration for sepa-


rated two-phase flow as outlined in Appendix A, we I I I I -50
have the following general result for the overall mass 24-
flux, from Equation (A-7)

f1
xcv 1 -X,, -’
w,IA,+- JJ’,/A
and the total flow area, from Equation (A-9)
A =A,+A, (7)
Although these two equations are quite different
in form, they in fact represent the same continuity
statement. Since Equation (7) is stated explicitly in the
API method. the alternate statement, Equation (6).
would also be implied.
Now the individual mass flux terms, W,/A, and 0 400 800 I200 1600 2WO 2400 2000 3200
W,/A f, in Equation (6) can be derived theoretically as Set pressure(psig~
detailed in Appendix B. This appendix also contains Figure 2 ASME flashing flow correlation for saturated water
the rather lengthy conversion of these formulae to the
form used in APJ. These formulae are demonstrated to
(berg)
be equivalent under the following two conditions: 0 50 100 150 200
I I I I iSO
0 The API method in reality permits two different 24 -
pressures to co-exist at the choked plane, with the
E
vapour stream at its own choked pressure as if -409
flowing alone and the liquid stream at the back E
P
pressure. This is therefore a simplifying assump-
tion. - 36%

l The capacity correction factors, K,, K,, and K,, :


_’
representing the overpressure, back pressure, and I
-20 I
viscosity corrections respectively, are taken to be E
unity. Under most circumstances, K w and K v can P,
be assigned values of 1.0. As for K,, we would -10 E
b
prefer to treat it in combination with the term
I
1.25P, - P such that for a fully-o en relief
zve, K VW. is simply &, where
OF
0

400 800
I I
1200
I I I I ’ lo
1600 2000 2400 2600 3200
AP is thePtotal available pressure drop. Set pressure (psig)

To summarize then, the API method would Figure3 Comparison of HEM prediction (---_) versus ASME
(---) flashing flow curve for saturated water
evaluate the flashing mass flux according to Equation
(6) with x,, calculated as described earlier, and the
liquid and vapour mass fluxes, WI/A, and W,/A,. pressure ratio is calculated in the ASME method. But
estimated from Equations (B-2) and (B-3) in Appendix in the vicinity of the maximum flow evaluated accord-
B. respectively. ing to Equation (l), the value of G decreases only
gradually with either increasing or decreasing pressure
ASME recommendation ratio. This might help to explain the observed discre-
The ASME Code does not address the issue of flashing pancy if indeed the calculated critical pressure ratio in
two-phase flow except for the case of saturated water. the ASME method does not correspond to the max-
It presents a flashing flow curve for saturated water as imum in G.
reproduced in Figure 2. According to the ASME
publication4, this curve is the theoretical, isentropic Saturated water data
flow value arrived by assuming equilibrium flow and Upon examining choked flow data for saturated water
calculated values for the critical pressure ratio. No from various sources, Moody5 found that equilibrium
recommendation is provided for other fluids. however. flow conditions were closely approached in pipes longer
than 0.1 m. On the other end, pipes approaching zero
length (orifice) tend to have non-flashing Bernoulli-
Comparison
type flow rates. For the first time, Moody demon-
ASME curve and HEM strated that these choked flow data with L > 0.1 m
The ASME saturated water flow capacity curve is first could in fact be predicted quite well simply with the
compared with the published HEM tabulationi6. This HEM. Figure 4 reproduces his comparison of the HEM
comparison, as illustrated in Figure 3, shows reason- with extensive literature data17-*l satisfying this equilib-
able agreement to within 10% throughout the entire rium length criteria. Though Moody’s evaluation of the
pressure range. However, it is not clear how the critical HEM was based on Equation (1) and the steam table,

J. Loss Prev. Process Ind., 1990, Vol3, April 255


Two-phase flashing flow methods: J. C. Leung and F. N. Nazario

could be correlated by the incompressible Bernoulli


formula, thus indicative of metastable superheated
liquid flow. Figure 5 reproduces their equilibrium flow
data in pipes of 0.15 m in length. Three HEM
predictiohs are shown here. The top curve is based on
perfect nozzle flow with a discharge coefficient Co of
1.0. The middle curve is based on a friction factor of
0.003 for the 50 L/D flow yielding an equivalent Co of
0.9 according to the generalized two-phase flow
chart23. Finally, the lower curve based on a friction
factor of 0.005 yielding a C, of 0.85. These last two
curves, properly accounting for pipe friction (with f
typically in the range of 0.003 to 0.005 for two-phase
flow), are found to bound the data quite well. Figure 5
also illustrates the two cases of prediction (as in the
saturated water case) based on the API approach.
Again, the flow rates are overpredicted. The lower
curve (case 2) with KD = 0.62 and AP given by
(P, - P,“) for the liquid stream predicts substantially
higher flow this time than the former comparison with
saturated water.

A
0 I I I I I I z”t
0 2 4 6 e IO 12 I
StoQnotion pressure, PO t MPO)
.- L
IR I
Figure4 Comparison of saturated water choked flow data
-in I6
Y -
against HEM and API (case 1) methods (adapted from MoodyS): E
A, Ref. 17, 4.0 mm diameter; o, Ref. 18, 6.35 mm diameter; n , : l4-
Ref. 19, 7.95 mm diameter; 0, Ref. 20; 173.0 mm diameter; 0,
Ref. 21.13.1 mm diameter 5
x I2 -

2 IO-
his prediction can be reproduced closely by the HEM B
generalized correlation of Figure 1. Also shown in this E 8-
figure are the predictions from the API flashing flow E
2
E 6-
method. Two cases are shown:
4-
Case I. This case employs KD = 1.0 for the vapour
stream and KD = 0.62 for the liquid stream. (Note that 2-
API actually recommends using 0.975 for the vapour I
01 I I I I I
K,, but the results are indistinguishable). 0 0.2 0.4 0.6 0.8 1.0
Stagnation pressye,Pc WPo)
Case 2. This is a modified API method. The choked Figure 5 Comparison of saturated R-12 data against HEM and
pressure for the liquid stream also corresponds to the API (case 1) methods: R-12 data of van den Akker et a/.22, 0,
vapour stream, i.e. the available pressure drop for the stainless steel pipe, o, polycarbonate pipe

liquid stream is given by (PO - P,“) rather than


(PO - P,,) as in case 1. Here P,, is given by Equation Multicomponent data
(5). The third and last comparison is based on the multi-
As shown in Figure 4, case 1 (API method) component (binary) propane/butane (C&Z,) data of
overpredicts the flow by nearly 100% at low pressures van den Akkerz4. A 50-50 wt% mixture was dis-
to about 30% at the high pressures. The agreement is charged from the bottom of their container vessel
much improved in the case 2 prediction. through a 2 mm diameter pipe. Their 75L/D data are
shown in Table I, together with the present predic-
Saturated refrigerant I2 data tions. Again the HEM predictions are in good agree-
Another comparison can be made based on the ment with their measurements. The API method (case
refrigerant 12 data of van den Akker et al.22. They 1) overpredicts the data by about 60%. The modified
found that the pipe length required for equilibrium method (case 2) yields overprediction of about 30%, an
flashing flow was somewhat less than 0.1 m (the flow in improvement over case 1. Finally, it is worth noting
a 0.06 m long duct was within 10% of the equilibrium that for this multicomponent system, the HEM predic-
flow value). Again, their near zero pipe length data tion is not too sensitive to the assumed physical

256 J. Loss Prev. Process Ind., 1990, Voi 3, April


Two-phase flashing flow methods: J. C. Leong and F. N. Nazario

Table 1 Comparison with Cp/CI flashing flow data of van den Akker (private communication, 1985). Discharge of 50-5Owt%C&
mixture in a 2 mm dia. x 150 mm length duct

Test no. PJkPa) w API Mass Flux G (kg mm2sm1)


&”
Measured HEM API methodb
correlationa

Case 1 Case 2
D 2.150.1 859 9.1 0.144 4800 5300 8655 7203
+ 10% + 80% + 50%
D 2.150.2 916 8.7 0.148 5700 5520 8951 7448
- 3% f 57% + 31%
D 2.150.3 917 8.7 0.148 5600 5525 8960 7456
- 1% + 60% + 33%
D 2.150.4 911 8.8 0.148 5700 5500 8908 7417
- 4% + 56% + 30%

“A discharge coefficient of 0.9 was used to account for the frictional losses in the 75 L/D duct
bBoth cases employed a K, of 0.62 in Equations B-2 or B-8; case 1 assumed AP = PO - Pb while case 2 assumed AP = PO - P,,. After G
was evaluated according to Equation (61. a factor of 0.9 was multiplied to account for the frictional losses

properties. Table 2 illustrates that the predictions quality x,, in Equation (8) can be related to the ‘vapour
based simply on single-component fluid (i.e. either C3 phase’ critical pressure ratio ncV and w, via a modified
or C,) at the average relief pressure and the corres- form of the energy balance (see Appendix C)
ponding component saturation temperature yield for all
intents and purposes the same result. XC” = (1 - n,.,+ (12)
Combining Equations (8), (11). and (12) yields the
Generalized comparison
desired result
To cast the comparison into a more generalized form,
G HEM
we present a closed-form expression relating the HEM -=
flow rate to the API flow rate for the saturated inlet
case. First the mass flux corresponding to the API
method can be written as

G API =
c;G
XC”
+KoG;\/p,p,
l-x,,

1
-’
(8) This equation can be readily evaluated once the
(13)

where G; and G; are the normalized mass fluxes given parameters w, pr/plO, k ( va p our specific heat ratio) and
P,,/Po are specified. In practice, this ratio G HEM/G Art
is governed primarily by two dimensionless property
groups. w and pr/pgo. Figure 6 illustrates how this
quantity G HEM/G API varies with the density ratio
CT= [2(1 - PJP,,)]‘/Z = [2(1 - r7b)]‘/? (10) pf/+, for the conditions of w = 5 or 10, k = 1.2 and
where 77s is the back pressure to stagnation pressure P,/P,, = 0.1. For the two cases shown, i.e.: API
ratio. As for the HEM mass flux, recall method with K, = 0.62 and AP given by the total
pressure drop for the liquid stream; modified API
G HEM = G&M s (11) method with KD = 0.62 for the liquid stream and AP
where G&rM is given by Equation (4a). The ‘critical’ given by (PO - P,“) only, the ratio GueM/GAri is

Table 2 Sensitivitv of assumed properties in C&, HEM prediction

w G(kg m 2s ‘1
Pll
Test
no. (kPa) Case A Case B Case C Case D Case A Case B Case C Case D Meas.

D2.150.1 859 9.06 9.70 8.12 8.28 5300 5150 5454 5468 4800
+ 10% + 7% + 14% + 14%
D2.150.2 916 8.73 9.39 7.90 7.99 5520 5360 5662 5690 5700
- 3% - 6% - 1% 0%
D2.150.3 917 8.72 9.38 7.89 7.98 5525 5365 5667 5695 5600
- 1% - 4% + 1% + 2%
D2.150.4 911 8.76 9.42 7.91 8.01 5500 5340 5643 5672 5700
- 4% -.6% ~ 1% ~ 1%

Case A, h,,, from rigorous thermodynamic calculation involving C&.,; Case B. h,, from IY,h&, where Y, is vapour mass composition:
Case C. one component assumption based on C,; Case D, one component assumption based on C,

J. Loss Prev. Process Ind., 1990, Vol3, April 257


rwo-phase flashing flow methods: J. C. Leung and F. N. Nazario

entirely on the homogeneous equilibrium flow assump


tion. Here we take the saturated water case at a
stagnation pressure PU of 570 kPa and according to the
HEM evaluation, the flashing flow results are:
P, = 520 kPa and GHEIVI= 4194 kgm-2s-t. Now, the
flashing flow is treated as a two-step process, and the
‘choked quality’ corresponding to the calculated
choked pressure of 520 kPa is evaluated first. This
quality, x,, is calculated to be 0.00725 using the steam
tableZJ properties. The second step is to employ this x,
as the inlet quality and treat the liquid and vapour
streams as entirely in frozen but homogeneous flow.
The calculated mass flux from this two-step scheme is
9542 kgm-‘s-l, which is about a factor of two higher
than the reference HEM case. (Note that the disparity
IO 100 1000 between the calculated mass flux from the API method
Density ratio pf/peO and the HEM method is reduced from about 200% to
Figure6 Mass flux ratio GHEM/GAP, versus density ratio, -
about 160% of the HEM value with hydrocarbons such
w = 10; ---, w = 5. Case 1: K. = 0.62; AP = PO - P,,, Case 2: as C3/C4 due to the smaller density ratio). This
KD = 0.62; AP = P, - P,” calculation clearly illustrates that the flashing flow
process cannot be treated as completely separate from
always less than unity and would decrease with the accelerating process. After all, the key governing
increasing pf/pgo. It is interesting to note that these mechanism for choking in flashing flow is the local rate
curves in Figure 6 would tend to merge towards a of flashing (mass transfer) itself.
G n&G Art ratio of 1 .O as the density ratio goes to 1 .O, For completeness, we also examined the effects of
i.e. at the thermodynamic critical point. For most the above comparisons on relief vent sizing. For a given
conditions of interest with pr/osO ranging from 10 to relief vent rate or capacity requirement, W, the
100, the following results can be obtained from Figure required relief area is inversely proportional to the
6: GHEM/GAPI ratio is 0.5-0.75 for case 1, and calculated mass velocity. Hence for the API and HEM
0.65-0.85 for case 2. methods the area ratio is
One can further extend Equation (13) to cover the A API G HEM
quality inlet conditions (i.e. x, > 0) but the results are (14)
A HEM GAPI
not expected to change greatly until x0 starts to
approach unity, i.e. the case for all-vapour flow, when which,is less than unity as shown in Figure 6. Thus the
potential undersizing of relief valves is 20 to 50%,
GHEMIGAPI would converge to 1.0 for all density
ratios. depending on the fluid type and the temperature and
pressure under consideration, i.e. the density ratio and
the IU parameter. (Hydrocarbons, due to their lower
Discussion
density ratio and as the equations are used in practice,
An attempt has been made here to compare the typically yield the lower values of potential undersiz-
available two-phase flashing flow methods as practiced ing). This does not necessarily imply that the API
in various engineering communities. Our comparison method yields undersized relief devices, since actual
shows that the ASME method is in close agreement relief devices are selected from those offered by the
with the DIERS method, both assuming isentropic manufacturers on the basis that the actual device relief
equilibrium flow in nozzles. The latter is simply the area must be larger than the minimum required area
classical homogenous equilibrium model. The API calculated. In addition, relief vent rates used in sizing
method, on the other hand, gives somewhat higher relief systems are typically very conservative and
theoretical mass flow rates. This result stems mainly assume worst case conditions that are rarely, if ever,
from the apparent neglect of the interfacial momen- obtained in actual accidents.
tum, heat and mass transfer accompanying the flashing Nevertheless. while the API method provided the
and acceleration processes. Although this method practising engineers with a simple to use approximation
considers an initial adiabatic flash to the assumed when no other easy-to-use calculational method was
choking pressure, subsequent evaluation treats the available, the simplicity and ease of computation
two-phase flow streams as entirely ‘frozen’ (i.e. no afforded by the generalized HEM flow correlationsto
mass transfer), in addition to no momentum and heat provide a superior, more realistic and thermodynami-
interchange. It is important to recognise that the cally consistent approach to two-phase relief system
choking phenomenon in flashing flow is principally design.
governed by the rates of these interfacial transport
processes.
Acknowledgement
A two-step approach flashing flow treatment can
also be illustrated via a numerical example based The authors are indebted to J. E. Huff (recently retired

258 J. Loss Prev. Process Ind., 1990, Vol3, April


Two-phase flashing flow methods: J. C. Leung and F. N. Nazario

from Dow Chemical Company) who first presented his CP


liquid specific heat
understanding on this subject at the DIERS User pipe diameter

Group Meeting in April 1988, and to H. G. Fisher


(Union Carbide Corporation) for providing key refer-
E
c*
fanning friction factor
mass flux. mass flow rate per unit area
normalized mass flux
ences. h specificenthalpy
k vapour specific heat ratio
KD flow discharge coefficient in API method
References K,. K,. K, API capacity correction factors due to overpressure,
back pressure and viscosity, respectively
I API RP-521. ‘Guide for Pressure Relief and Depressurizing L pipe length
Systems’, American Petroleum Institute. New York. USA M, molecular weight
September 1982 P pressure
2 Huff, J. E Plant/Operariom Progress, 1985.4 (4). 191 Ph ambient or back pressure
3 Fauske & Associates. Inc.. ‘Emergency Relief Systems for AP pressure drop
Runaway Chemical Reactions and Storage Vessels: A Summary R gas constant
of Multiphase Flow Methods’. DIERS Report FAI/83-27. 19X3 s specific entropy
4 ASME Boiler and Pressure Vessel Code. Appendix 11: Capacity sg specific gravity, I .O for water
Conversions for Safety Valves. American Society of Mechanical T temperature
Engineers. New York. USA, 1986. p. 610 u velocity
5 Moody. F. .I., ‘Maximum Discharge Rate of Liquid Vapour 0 specific volume
Mixtures from Vessels’. Non-Equilibrium Two-Phase Flow, W mass flow rate
ASME Symp. Vol.. Am. Sot. Mech. Eng.. 1975; NEDO-21052. + quality or mass fraction of vapour
General Electric Company. San Jose, CA. USA, 1975 Z compressibility fador
6 Lahey. R. T. and Moody, F J., The Thermal-Hydraulics of a a void fraction
Boiling Water Nuclear Reactor. Am. Nuclear Sot., Hinsdale, 9 pressure ratio
Illinois, USA, 1977 P density
7 van den Akker. H. E. A.. Snoey. H.. and Spoelstra. H., 0 flashing flow parameter, Equation (3)
‘Discharges of Pressurized Liquified Gases Through Apertures
and Pipes‘, 4th Int. Symp. on Loss Prevention and Safety Subscripts
Promotion in the Process Industries. IChemE Symp. Scrics No. b back pressure
80, E23-35, 19x2 c choked or critical
8 Fauske. H. K. Plunt/Oprrution~ Progress 1985.4 (3). 132 C” vapour phase choked condition
9 Starkman. E. S.. Schrock. V. E.. Nuesen. K. F.. and Maneely, f liquid phase
D. J. J. Basic Engineering Trans. ASME Series D. 1964, 86 (2). difference between vapour and liquid phase property
fg
246 vapour phase
g
10 Leung, J. C. AIChEJournal, 1986.32 (IO). 1743 0 upstream stagnation conditions
I1 Leung. J. C. and Fisher. H. G. J. Loss Prev. Process Ind. 1989. s set pressure condition
2.78
I2 Kern. R.. Chemical Engineering 1977, p. 187
13 Richter. S. H., H_~drocnrbon Processing. 1978. p, 145
14 Mukerji. A. Chtvnical Engineering. 1980, p_ 79
15 Drlskell. L. Chemical Enginewing 1983. p. 94 Appendix A: continuity consideration in
16 Hall. D. G. and Czapary,
Equilibrium Flow Parameters
L. S., ‘Tables of Homogeneous
for Water in SI Units’, Report NO.
separated flow
EGG-2056. EG&G Idaho, Inc., Idaho Falls, Idaho. 1980 For separated two-phase flow, we have the following
17 1Jchida. H.. and Nariai, H., Proceedings of the Third Interna-
tional Heat Transfer Conference. 1966.5 identities for each flow stream (Wallis2s, p. 44)
18 Fauske. H. K.. Chem. Eng. Prog. Symp. Ser. 1965.61.210
19 Henry. R. E.. ‘An Experimental Study of Low-Quality, Steam-
Wf=piurAr= W(l-x) (A-l)
Water Critical Flow at Moderate Pressures’, ANL-7740, 1970.
W, = p%u,A, = Wx (-4-2)
Argonne National Laboratory, Argonne. IL, USA
211 Allemann. R. T.. et al., ‘Experimental High Enthalpy Water G‘ = PfU‘(l - a) = G (1 - x)
Blowdown from a Simple Vessel Through a Bottom Outlet’.
(A-3)
BNWL-I41 I. 1970. Battelle Northwest Laboratory, Richland. G, = PBuB~ = Gx (A-4)
WA, USA
21 Sozzi, G. L., and Sutherland, W. A.. ‘Critical Flow of Saturated From Equation (A-3) and Equation (A-4)
and Subcooled Water and High Pressure.‘ NEDO-13418. May
1975, General Electric Company, San Jose. CA. USA I_ ~ = G(1- x)
(A-5)
22 van den Akker. H. E. A. and Bond. W. M.. Protection of
Exothermic Reactors and Pressurized Storage Vessels, Chester, PfUf
Gx
25-27 April 1984. IChemE Symp. Ser. 1984.85.91 (y=- (A-6)
23 Leung, J. C. and Grolmes, M. A. AIChEJ.. 1987.33 (3). 524
24 Keenan. J. H., Keyes. F. G.. Hill. P. G.. and Moore, J. ti.. in PkFg
‘Steam Table (S.I. Units)‘, John Wiley & Sons. New York, 1978 Adding Equations (A-5) and (A-6) yields:
25 Wallis. G. B.. in ‘One-Dimensional Two-Phase Flow’. McGraw-
Hill. New York, 1969 1 x 1-X
--+ (A-7)
26 Shapiro, A., in ‘Compressible Fluid Flow’. Ronald Press, New
E PfUf
- PSU%
York, Vol. I, Ch. 6, 1953
27 API RP-520. ‘Recommended Practice for the Design and Multiplying Equation (A-7) on both sides by W, i.e.,
Installation of Pressure-Relieving Systems m Refineries. Pt. 1 -
Design‘. December, 1976. Am. Petroleum Inst., NY. USA
W
-=- wx + W(1 -xl
(A-8)
G Pk+B PfUf
and making use of Equations (A-l) and (A-2), we
Nomenclature finally obtain
A flow area A=A,+AI (A-9)
C coefficient in API formula Equation (B-4)
CD overall flow discharge coefficient which is the expected result.

J. Loss Prev. Process Ind., 1990, Vol3, April 259


Two-phase flashing flow methods: J. C. Leung and F. N. Nazario

Appendix B: individual mass tlux evaluation where gpm denotes flow rate in gallons per minute.
in API method Again the above equation is in US customary engineer-
ing units (A B is in in2 here). For comparison purposes,
The individual mass flux terms, W,/A, and W,/A I, in
we shall ignore the influence of the discharge coeffi-
Equation (6) can be evaluated separately as follows.
cient K, and the various capacity correction factors,
Under the assumption of frictionless non-interacting
K,, K,, and K,. It is further noted that (1.25 P, - P,,)
flow, the two streams are going to follow isentropic
is simply the available pressure drop A P on the basis of
paths while satisfying the flow area requirement of
25% overpressure for liquid relief sizing. In such cases,
Equation (7). Thus the vapour will expand isentropic-
K, is assigned a value of unity. Now, the Bernoulli
ally in a nozzle according to the well-known equationZh
equation, Equation (B-2), can be rearranged to yield
1 ( Wf/P,)
Af=Kn m
(B-6)
(B-1) and by employing units consistent with the API practice
where a discharge coefficient KD is included to account as shown in the equation below (note that the 6 is
for imperfect nozzles. replaced by 6 here)
Since the liquid is incompressible in the frozen
flow assumption, it will obey Bernoulli’s equation. Its Af(in’)(
li5y;n2)
=
mass flux at the place where the pressure is P will be

1
W, (gpm)(3.785 X 1O-3 m3/gal) kg/m’ ‘I2
- = KD[2pf(Po - P)]‘i” (B-2) sg 2 lOoO-
A, (60 min/sec) ]O
It is widely known that the discharge coefficient K, is
Ko[2AP (psi)(6897~)]“’ “”
about 0.62 for a sharp-edge orifice and this value is
recommended by API as a conservative approach. (B-7)
Theoretically the choked pressure is to be found by
We finally obtain
satisfying the criterion that at the throat dA/dz is zero.
This would yield a choked pressure less than the typical gpmVs&/cc)
A f(in2) = (B-8)
vapour choked pressure given in Step 1. Such a solution 38K&w-.
is, as expected, much more involved. The API method which is identical to Equation (B-5) with K,, K,, and
recommends instead that P in Equation (B-l) for the K, being 1.0, and (1.25P, - Ph) being APon the basis
vapour flow be limited by Equation (5), but P in of 25% overpressure.
Equation (B-2) for the liquid flow be evaluated at the
downstream back pressure P,,. In other words, the total
available pressure drop is allowed in the liquid phase.
Thus for the vapour stream, the classical choked
flow formulaZh is obtained by combining Equations (5)
and (B-l)
Appendix C: approximate solution for flash
quality
(B-3) Based on a constant enthalpy energy balance, the
adiabatic flash quality at a temperature T from an
This is to be compared with Equation (1) in API-520 initially saturated liquid state (x,, = 0) is
[Ref. 27) for gas/vapour relief
C,(To - T)
M X= (C-1)
W, = CKDA~P,, -.L h fE
(B-4)
ZT,,
We invoke the Clapeyron relation
where C is given by
AI-’
-=- h,
,,,[,( k : l )i-y AT (C-2)
or,T
and approximate the properties at the stagnation
and the number 520 contains the proper conversion condition. thus arriving at
when using the US customary engineering units (A g is
x= __-
in in2 here) in Equation (B-4). By noting that pa0 is (C-3)
P,,M,/ZRT,,, Equations (B-3) and (B-4) are there-
and for us,, >> vf
fore equivalent in form.
As for the liquid stream, Equation (B-2) with AP ego
x=xX (C-4)
P = Pb is to be compared with Equation (C-7) in API
RP-520 publication Therefore, at the vapour phase critical pressure
gpm%i
Af= (B-5)
38K,K,K,K,d 1.25P, - P,,

260 J. Loss Prev. Process Ind., 1990, Vol3, April

Вам также может понравиться