Chemical Engineering and Processing 39 (2000) 275 – 281

www.elsevier.com/locate/cep

Vapour adiabatic exponent for flashing flow in nozzles

S.D. Morris *
Joint Research Centre (Ispra), T.P. 421, European Commission, 21020 Ispra (VA), Italy

Received 10 May 1999; received in revised form 4 October 1999; accepted 4 October 1999

Abstract

An expression is derived for the adiabatic exponent of the vapour phase during isentropic flashing flow in nozzles. One area of
practical application is in design methods for determining required orifice sizes of safety valves. © 2000 Elsevier Science S.A. All
rights reserved.

Keywords: Two-phase flow; Emergency pressure relief; Loss prevention; Nozzles; Safety valves

1. Introduction p6 GG = constant (1)

Compressible two-phase flow is a subject of continu- Since an expression for G, applicable specifically to
ing interest in the design of emergency pressure relief flashing flow, has not until now been derived, current
systems in chemical and process plants and in the design practice is to use either G= 1 or the expression
estimation of flow rates from vessel ruptures and of Tangren et al. [13], described by:
pipeline leaks. In recent years there has been a particu-
x0cpG0 + (1− x0)cL0
lar interest in the development of improved methods for G= (2)
determining the orifice sizes of safety valves for two- c
x0 pG0 +(1− x0)cL0
phase service. Research projects in this area are cur- k
rently underway in Europe and the USA. Relevant
research work can be found in Refs. [1 – 12]; in particu- This expression was derived for the case of isentropic
gas–liquid flow (non-flashing) but nevertheless is some-
lar, Ref. [7] provides a comprehensive source of litera-
times applied to flashing flows on the reasoning that it
ture on the subject of two-phase flow through nozzles
provides a better description of the thermodynamic
and safety valves.
behaviour of the vapour than either G= 1 (isenthalpic
Safety valves are often the flow-controlling compo-
path for vapour) or G=k (isentropic path for vapour).
nents in pressure relief systems (both emergency and In a two-phase flow that is described by an isentropic
routine) and it is important for the integrity of the (or isenthalpic) flowpath, it is the mixture that is flow-
protected vessel that the safety valve has sufficient flow ing isentropically (or isenthalpically), not the individual
area to handle the required relief loading (mass dis- phases, so neither G= 1 nor G= k can be strictly
charge rate). Compressible two-phase flow models for correct. Nevertheless, in the sizing of safety valves for
safety valves are based on a nozzle geometry with gas or vapour flow, an isenthalpic flowpath is fre-
critical (choked) flow occurring at the nozzle throat. quently chosen as it gives a conservative result with
All models employ an adiabatic exponent (here de- respect to the required orifice size (flow area of valve
noted by G) that describes the vapour expansion along nozzle). In other words, the orifice size necessary to
the chosen thermodynamic flowpath (isentropic or isen- permit a required mass discharge rate is slightly larger
thalpic), usually by: for G= 1 than for G= k.
The choice of thermodynamic flowpath for two-
* Tel.: +39-0332-789754; fax: +39-0332-785584. phase flow through a nozzle also deserves some com-
E-mail address: [email protected] (S.D. Morris) ment. The energy equation is:

0255-2701/00/$ - see front matter © 2000 Elsevier Science S.A. All rights reserved.
PII: S 0 2 5 5 - 2 7 0 1 ( 9 9 ) 0 0 0 9 6 - 3
  
276 S.D. Morris / Chemical Engineering and Processing 39 (2000) 275–281

1 2
dhm + d u =0 (3) sG = sG(p, T) (9)
2 m
for which the total differential is:
and this is entirely consistent with the assumption of
isentropic flow:
dsG =
  (sG
dp +
  (sG
dT (10)
(p (T
T dsm =dhm − 6m dp =0 (4) T p

By definition:
and the momentum equation, i.e.:

6m dp = − d
 
1 2
 
(sG
=
cpG
(11)
u (5) (T T
2 m p

and from the Maxwell relations [14]:

The integration of Eq. (5) is the basic step then
followed in the development of an expression relating  
(sG
=−
  (6G
(12)
mass flow rate to pressure drop but this development is (p T (T p
not a concern of the present work.
Now, a measure of ‘non-ideal’ gas behaviour may be
In the above equations:
introduced through use of a constant compressibility
hm =xhG +(1−x)hL (6) factor Z, i.e.:
sm = xsG +(1−x)sL (7) 6G = ZRT/p (13)
and, for homogeneous flow: The utility of this ‘non-ideal’ equation-of-state
should not be over-estimated however, since the intro-
6m = x6G + (1− x)6L (8)
duction of constant Z simply changes one ideal gas to
The choice of isenthalpic two-phase flow (dhm = 0) another of different gas constant (R is simply replaced
for a nozzle is an approximation that implies, from Eq. by constant ZR and all ideal gas flow theory remains
(3), that changes in kinetic energy are negligibly small valid).
Eq. (12) then gives:
   
compared with changes in enthalpy (per unit mass of
flowing mixture). This is an acceptable approximation (sG ZR c k− 1
in many cases of interest and, conveniently, G =1 sim- =− = − pG (14)
(p T p p k
plifies the derivation of flow models from Eq. (5).
However, care should be exercised in cases of initially since:
low-quality and saturated liquid flows where kinetic
ZR= cpG − c6G (15)
energy and enthalpy changes can be of the same order.
In such cases, a flow model based on isentropic flow and k= cpG/c6G is the ratio of specific heats. Eq. (10)
should be more reliable. Since isentropic flow models then becomes:
are believed to be generally more sound for the nozzle
flows of interest here, there is an incentive to derive an dsG = cpG
dT
−cpG
k− 1 dp  (16)
adiabatic exponent for the vapour phase of an isen- T k p
tropic flashing flow. A comparison with the result of A corresponding expression for the enthalpy change
Tangren et al. [13] for the isentropic non-flashing case can be obtained from the familiar thermodynamic
should then quantify the effect of flashing. The present relationship:
work is not concerned further with the details of two-
phase flow models per se but rather with the discussion dhG = T dsG + 6G dp (17)
and derivation of adiabatic exponents; this requires which reduces to:
considerations only of basic thermodynamics. It is use-
ful first of all to recount the case of gas flow in a nozzle dhG = cpG dT (18)
for the cases of isentropic and isenthalpic flowpaths. on substitution of Eqs. (13), (15) and (16). It remains to
express the quantity dT in terms of dp. Presuming the
gas flow behaviour described by Eq. (1) and combining
2. Gas flow this with Eq. (13) gives:

The objective here is to express differential entropy T Gp 1 − G = constant (19)

and enthalpy changes in terms of differential pressure Differentiating this with respect to p yields the
change. Familiar results emerge when the differential relationship:
entropy and enthalpy changes are set to zero.
The gas entropy sG is a function of the process dT (G−1) dp
= (20)
variables p and T, i.e.: T G p
 S.D. Morris / Chemical Engineering and Processing 39 (2000) 275–281 277

 
which, when substituted in Eq. (16) and Eq. (18), gives since:
the required expressions for differential entropy and (6G ZR 6G
enthalpy changes: = = (30)

dsG =
 
1 1 cpG
− dp (21)
(T p p T
Substitution of Eqs. (28) and (29) in Eq. (26) and
k G p

dhG =
 
G−1 TcpG
dp (22)
re-arranging gives:
dx
xZR− psLG
G p p dT dp
= (31)
Hence, in Eq. (21), isentropic flow (dsG =0) requires T dp xcpG + (1− x)cL
that: It is now required to eliminate the temperature T
G= k (23) from the left hand side of this equation, in favour of p
and 6G. The starting point is to write the temperature T
while, in equation Eq. (22), isenthalpic flow (dhG = 0) as a function of p and 6G as follows:
requires that:
T= T(p, 6G) (32)
G= 1 (24)

   
for which the total differential is:
A further observation is that isenthalpic flow is also
(T (T
isothermal (T=constant from Eq. (19)). dT= dp + d6G (33)
In regard to nozzle flow, saturated vapours are often (p 6G (6G p

treated as gases. In fact, saturated vapours do behave

 
Now, from Eq. (13):
differently. Normally, for one-component two-phase
(T 6G
systems, a saturated vapour that flows isentropically = (34)
will experience net condensation while an isenthalpic (p 6G ZR
flowpath will result in superheating of the vapour. Such
 
and:
flows will be discussed later, as they are more conve-
(T p
niently treated once the case of two-phase flow has been = (35)
developed. (6G p ZR
It then follows from Eq. (33) that:
p dT p d6G
3. Two-phase flow = 1+ (36)
T dp 6G dp

: ;
The overall objective here is to derive and solve a This right hand sides of Eqs. (31) and (36) may now be
differential equation that relates d6G to dp for an equated to give the following interim result:
isentropic process defined by: c dx
x pG +(1− x)cL + psLG
sm0 =xsG + (1− x)sL =constant (25) d6G k dp dp
=− (37)
6G xcpG + (1− x)cL p
where sm0 is the specific entropy of the mixture at the
nozzle inlet. Differentiating this with respect to pressure This equation offers the possibility of analytical inte-
p gives: gration providing that the quantities x, dx/dp and sLG
can be expressed in terms of the pressure p.
dsG ds dx
x +(1−x) L + sLG = 0 (26) It is of interest to note that, following through the
dp dp dp

: ;
above procedure but now for an isenthalpic flowpath
The thermodynamic relation which applies to both gives the corresponding equation:
phases [14]:
 
dx
psLG
ds cp dT (6 d6G dp dp
= − (27) = − 1+ (38)
dp T dp (T p
6G xcpG + (1− x)cL p

gives, for the incompressible liquid phase: but where x and dx/dp for this isenthalpic flowpath will
differ from those in Eq. (37).
dsL cL dT
= (28)
dp T dp
and for the vapour phase: 4. Non-flashing flow

dsG cpG dT 6G Defined by dx/dp = 0 and x= x0 throughout the flow

= − (29)
dp T dp T path, this case is easily analysed. Noticing that the
278 S.D. Morris / Chemical Engineering and Processing 39 (2000) 275–281

dx dx
quantity in large brackets is now independent of pres- psLG :hsLG0[1+ (1− h)c] (49)
dp dh
sure, Eq. (37) integrates readily to give the form of Eq.
(1) where now the adiabatic exponent for isentropic where:
non-flashing two-phase flow is:
dx x0c− f
x c +(1−x0)cL0 : (50)
G = 0 pG0 (39) dh [1+ (1− h)c]2
cpG0
x0 +(1− x0)cL0
k Eqs. (45)–(50) may now be substituted into Eq. (37)
to give, after some algebra:

This is identical to the expression derived previously
by Tangren et al. [13]. By comparison, for isenthalpic
flow, G= 1 from Eq. (38). A
d6G dh
= − + AB
−C
dh
n (51)
6G h (D− Ch)
where:
5. Flashing flow
! ZRT0
x0cpG0 + (1−x0)+ c
"
! "
To facilitate the integration of Eq. (37) for the case hLG0 L0
A= (52)
of flashing flow, the quantities x, dx/dp and sLG must cpG0 ZRT0
{x0 − (k− 1)f} + (1−x0)+ cL0
be expressed in terms of pressure. Eq. (25) defines the k hLG0
mass flow quality x at any pressure p on the saturation (f− x0c)[ZRcL0 − sLG0 D]
line, i.e.: B= (53)
CD
sm0 − sL(p)
x= (40) C= fcpG0 + (c− f)cL0 (54)
sLG(p)
where: D= x0cpG0 + (1− x0)cL0 + fcpG0 + (c− f)cL0 (55)

sm0 =x0sG0 +(1−x0)sL0 (41) Integration of Eq. (51) is straightforward and yields:
The quantities sL(p) and sLG(p) can be expanded as the
first two terms of a Taylor series as follows: ln (h6 A
G)= constant+ ln
! C
1− h
"
AB
(56)

 n
D
dsL
sL(p) :sL0 − (p0 −p) (42) An examination of the magnitudes of the quantities
dp p=p 0 on the right-hand-side of this equation shows that
and: C/D 1 and the product AB 1. The logarithmic term

sLG(p):sLG0 − (p0 −p)


dsG dsL
−
n (43)
is then little different from ln (1)=0 and may be ig-
nored. Eq. (56) then reduces to the form of Eq. (1),
dp dp p=p 0 where now:
Now, defining the pressure ratio:
G= A= 1/(1 − U) (57)
h= p/p0 (44)
and where:
and making use of the previous Eqs. (28) and (29) for


dsL/dp and dsG/dp, the above equations yield the fol- ZR(x0 + f)
lowing expression for x in terms of the pressure ratio h:
U=
x0cpG0 + (1−x0)+
ZRT0
c
n (58)

x0 +(1−h)f hLG0 L0
x: (45)
1+(1−h)c Eqs. (1) and (57) and Eq. (58) define the main result
where the constants f and c are given by: of this paper. Interestingly, if the above procedures for
flashing flow are followed through for an isenthalpic
f = p0T0cL06LG0/(hLG0)2 (46) flowpath the result G=1 is obtained, which is identical

c = 1−
 
cpG0
f+
ZRT0
(47)
to the results for both isenthalpic gas flow and isen-
thalpic non-flashing two-phase flow.
cL0 hLG0
Fig. 1 shows a comparison of the present flashing
and also: flow theory with the non-flashing flow result of Tangren
et al. [13] for steam–water mixtures over the whole
sLG : [1+(1 −h)c] (48)
quality range, and for pressures of 1, 10, 20 and 50 bar
Note that the parameters f and c are generally small abs. It is evident that values of the exponent for flash-
compared with unity (i.e. f 1 and c  1). ing flow are significantly lower than in the case of the
The flashing term in Eq. (37) then becomes: non-flashing flow assumption.
 S.D. Morris / Chemical Engineering and Processing 39 (2000) 275–281 279

Fig. 1. A comparison of the present flashing flow theory with the non-flashing flow result of Tangren et al. [13] for steam – water mixtures over
the whole quality range, and for pressures of 1, 10, 20 and 50 bar abs.

6. Vapour flow over the pressure range 1–100 bar abs. The curves are
both quantitatively and qualitatively different, values of
Saturated vapours behave differently from non-con- G decreasing with increasing pressure while those of k
densable gases. Eq. (26) gives, for any isentropic two- increase.
phase flow:
! "
However, it is known from experiments that satu-
dx dsG ds rated (and slightly wet) steam expanding isentropically
=− x +(1− x) L /sLG (59) through a nozzle can become supersaturated, in which
dp dp dp
case the condensation demanded by equilibrium ther-
Therefore, dx/dp =0 for x=x  where:
,
modynamics does not appear until some distance
dsL dsLG downstream of the nozzle throat. In such cases, the
x= (60) ratio of specific heats, k, is normally recommended as
dp dp
the more appropriate vapour expansion exponent. In
This ‘critical’ value of quality, shown in Fig. 2 for this work, however, where a main concern is the sizing
steam–water mixtures over a range of pressure, defines of safety valves, Eqs. (57) and (61) are recommended
a boundary between vaporising flows (dx/dp B0) and since this value of G will introduce a small amount of
condensing flows (dx/dp \0). conservatism into the calculation of the required flow
An initially saturated vapour (x0 =1) flowing isen- area. This means, then, that the general result given by
tropically through a nozzle under equilibrium thermo-
dynamic conditions will therefore experience
condensation and the exit flow will be two-phase. Thus,
such flows of initially saturated vapours are more cor-
rectly treated a priori as two-phase flows. The adiabatic
exponent in this case is given by Eqs. (56) and (57) with
x0 = 1, i.e. with:

 n
ZR(1+f)
U= (61)
ZRT0
cpG0 + c
hLG0 L0
Fig. 2. A ‘critical’ value of quality for steam – water mixtures over a
A comparison of this result with the ratio of specific range of pressure, defines a boundary between vaporising flows
heats k is shown below in Fig. 3 for saturated steam (dx/dpB0) and condensing flows (dx/dp\ 0).
280 S.D. Morris / Chemical Engineering and Processing 39 (2000) 275–281

Appendix A. Nomenclature

A, B, C, D expressions defined by Eqs. (52)–(55)

cL specific heat of liquid at constant
pressure (J/kg per K)
cL0 value of cL at inlet conditions
cpG specific heat of vapour at constant
pressure (J/kg per K)
cpG0 value of cpG at inlet conditions
hG specific enthalpy of vapour (J/kg)
hL specific enthalpy of liquid (J/kg)
Fig. 3. A comparison of the result derived from Eq. (61)with the ratio
hLG specific enthalpy of vaporisation (J/
of specific heats k for saturated steam over the pressure range 1 – 100
bar abs. The curves are both quantitatively and qualitatively differ- kg)
ent, values of G decreasing with increasing pressure while those of k hLG0 value of hLG at inlet conditions
increase. hm specific enthalpy of the mixture (J/kg
per K)
hm0 value of hm at inlet conditions
Eqs. (57) and (58) for isentropic flashing flow is recom-
K ratio of specific heats (cpG/c6G)
mended for use throughout the quality range 05 x0 5 1
P pressure (Pa)
(initially saturated liquid, two-phase and saturated va-
p0 inlet pressure (Pa)
pour conditions).
R gas constant (J/kg per K)
The situation for an isenthalpic flow of initially satu-
sG specific entropy of vapour (J/kg per
rated vapour is different since there is no possibility of
K)
condensation. In fact, the above derivation from Eqs.
sL specific entropy of liquid (J/kg per K)
(25) –(57) may be repeated but now for an isenthalpic
sLG specific entropy of vaporisation (J/kg
flowpath (dhm =0). The details are not reproduced here
per K)
but the simple result G = 1 is obtained. This means that
sLG0 value of sLG at inlet conditions
T = T0 =constant and, since the flow cannot therefore
sm specific entropy of the mixture (J/kg
follow the saturation line, the vapour becomes super-
per K)
heated as the pressure is decreased. Referring again to
sm0 value of sm at nozzle inlet
Fig. 3 above, it can be seen that, with increasing
T0 inlet temperature (K)
pressure, the exponent G decreases towards the isen-
T temperature (K)
thalpic flow solution (G =1).
um mixture velocity (m/s)
6G vapour specific volume (m3/kg)
6L vapour specific volume (m3/kg)
6LG specific volume difference (6G−6L)
7. Conclusions 6m mixture specific volume (m3/kg)
X mass flow quality
An adiabatic exponent for the vapour phase has been x0 mass flow quality at nozzle inlet
derived for isentropic flashing flow in nozzles. The final Z constant compressibility factor (p6G/
result is applicable to non-ideal vapours in as much as RT)
the compressibility factor Z can be treated as constant
throughout the flow path. In view of current interests to Greek symbols
improve the design basis for emergency pressure relief h pressure ratio (p/p0)
systems, the exponent derived here is recommended for U parameter related to G by Eq. (57)
use in two-phase flow models for safety valve sizing. f parameter defined by Eq. (46)
c parameter defined by Eq. (47)

Acknowledgements

Funding for this research was provided by the Eu-

References
ropean Commission (DGXII -Science, Research and
Development) under the Industrial Safety Area of the [1] T. Lenzing, L. Friedel, J. Cremers, M. Alhusein, Prediction of
Environment and Climate Programme 1994-1998 (IN- the maximum full lift safety valve two-phase flow capacity, J.
OVVATOR Project: Contract No. ENV4-CT97-0609). Loss Prev. Process Ind. 11 (1998) 307 – 321.
 S.D. Morris / Chemical Engineering and Processing 39 (2000) 275–281 281

[2] R.E. Henry, H.K. Fauske, Two-phase critical flow at low quali- [9] R. Darby. Viscous two-phase flow in safety relief valves: phase II
ties, Nucl. Sci. Eng. 41 (1970) 79–91. — experimental program design, Report prepared for Design
[3] D.W. Sallet, Critical mass flow rates through pressure relief Institute for Emergency Relief Systems/American Institute of
valves, Warme-Stoffubertragung 26 (1991) 315–321. Chemical Engineers, September 1998.
[4] S.D. Morris, Choke pressure in pipeline restrictions, J. Hazard. [10] D.H. Sherman. The effects of the flow rate and operational
Mater. 50 (1996) 65 –69. characteristics of spring-loaded pressure relief valves on flashing
[5] S.D. Morris, Liquid flow through safety valves: diameter ratio sub-cooled and saturated water service, presented at ISO/TC185/
effects on discharge coefficients, sizing and stability, J. Loss WG1 (Doc. N52), 15 – 16 September 1993, Ljubljana, Slovenia.
Prev. Process Ind. 9 (3) (1996) 217–224.
[11] H.G. Fisher. Emergency relief system design: the DIERS Users
[6] S.D. Morris, Choked gas flow through pipeline restrictions: an
Group safety relief valve example problem, Int. Symp. on Run-
explicit formula for the inlet Mach number, J. Hazard. Mater. 50
away Reactions and Pressure Relief Design, 2 – 4 August 1995,
(1996) 71 – 77.
Boston, USA, pp. 418 – 449.
[7] R. Darby. Viscous two-phase flow in safety relief valves: phase I
— present status and recommended design procedures, Report [12] S. Selmer-Olsen. Offshore PSV design problem – model evalua-
prepared for Design Institute for Emergency Relief Systems/ tion, Technical Report No. 96-2000, March 1996, Det Norske
American Institute of Chemical Engineers, November 1997. Veritas Research AS, Hovik, Norway.
[8] R. Darby, Perspective on relief valve sizing for two-phase flow, [13] R.F. Tangren, C.H. Dodge, H.S. Seifert, Compressibility effects
in: G.A. Melhem, H.G. Fisher (Eds.), International Symposium in two-phase flow, J. Appl. Phys. 20 (7) (1949) 637 – 645.
on Runaway Reactions, Pressure Relief and Effluent Handling, [14] M.M. Abbot, H.C. Van Ness, Thermodynamics. Schaum’s Out-
AIChE/DIERS, New York, 1998. line Series, McGraw-Hill, New York, 1972.