Computers and Chemical Engineering 23 (1999) 1309 1317

www.elsevier.com/locate/compchemeng

A numerical blowdown simulation incorporating cubic equations

of state
Haroun Mahgerefteh *, Shan M.A. Wong
Department of Chemical Engineering, Uni6ersity College London, Torrington Place, London WC1E 7JE, UK
Received 22 January 1999; received in revised form 15 September 1999; accepted 15 September 1999

Abstract
The development of a numerical simulation, based on cubic equations of state for blowdown of vessels containing high pressure
hydrocarbons is described. The models performance is evaluated by comparison with experimental data relating to the blowdown
of a condensable hydrocarbon mixture in a full sized vessel at a starting pressure of 118 bar. Typical output data including
timetemperature profiles for the vapour and liquid phases as well as the wetted and unwetted walls are reported. These are
shown to be within 95 K of experimental data. In general, it is found that the choice of the cubic equation of state has little
effect on the data. 1999 Elsevier Science Ltd. All rights reserved.
Keywords: Blowdown; Depressurisation; Numerical simulation; Computer modelling

1. Introduction
The rapid depressurisation or blowdown of process
vessels or sections containing high pressure hydrocarbons is contemplated when emergencies arise, particularly in offshore operations where there is a fire or
escape of flammable gases. In recent years however,
such operations have presented process and safety engineers with a dilemma.
The primary purpose for blowdown is to reduce
pressure and remove inventory in the least amount of
time possible. However, rapid depressurisation results
in a dramatic drop in the fluid and hence the vessel wall
temperatures. If the wall temperature falls below the
ductilebrittle transition temperature of the vessel material, rupture is likely to occur. Low fluid temperatures
can also lead to the formation of solid hydrates in cases
where free water is present. The presence of solid
hydrate can cause great difficulties in operations (Katz
& Lee, 1990).
Clearly the optimum blowdown time requires a delicate balance between the maximum permissible depressurisation time (see for example, API Recommended
Practice 521, 1990) and the minimum wall and fluid
* Corresponding author. Tel.: +44-171-4193835.

temperatures that may be safely contemplated. The

accurate predictions of these parameters are central to
addressing this issue. The traditional approach has been
to overdesign. However, this option is increasingly
becoming less attractive given the rapidly decreasing
profit margins that the oil industry is faced with.
Consequently, in recent years there have been a
number of theoretical and experimental studies relating
to blowdown simulation with varying degrees of sophistication. A comprehensive review is given by Wong
(1998). The simplest and by far the most commonly
used methods simulate the expansion of the fluid inside
the vessel as either isentropic or isenthalpic (see for
example API Recommended Practice 521, 1990). Such
an approach generally results in predicting unrealistically low fluid and wall temperatures, thus leading to
significant capital equipment expenditure required in
order to meet the prescribed safety margins. Although
the rapid expansion of a gas in a vessel may initially
follow an isentropic path, heat transfer from the vessel
walls ensures that the gas temperature will never reach
the isentropic value.
A further problem is the assumption of equilibrium
between the different fluid phases within the vessel (see
for example, API Recommended Practice 521, 1990).
Experiments relating to (see for example Haque,

0098-1354/99/$ - see front matter 1999 Elsevier Science Ltd. All rights reserved.
PII: S 0 0 9 8 - 1 3 5 4 ( 9 9 ) 0 0 2 9 6 - 3

1310

H. Mahgerefteh, S.M.A. Wong / Computers and Chemical Engineering 23 (1999) 13091317

Richardson & Saville, 1992b; Overa, Stange & Salater,

1994) depressurisation of a number of different hydrocarbon mixtures have revealed significant temperature
differences between the fluid phases and their associated section of vessel wall. This is mainly because of the
entirely different fluid wall heat transfer mechanisms
in the case of a gas as compared to a liquid.
Recently, two numerical models, BLOWDOWN
(Haque, Richardson & Saville, 1992a) and SPLIT
FLUID MODEL (Overa et al., 1994) have been developed to account for the non-isentropic nature of fluid
expansion as well as the effect of non-equilibrium between the fluid phases during depressurisation.
Despite its sophistication, the main drawback associated with BLOWDOWN is the use of the extended
principle of corresponding states (Rowlinson & Watson, 1969) for generating the pertinent fluid thermophysical properties which as well as introducing
uncertainties associated with its accuracy, makes the
simulation computationally demanding. This also presents serious consistency problems as in practice, cubic
equations of state (CEOS) are almost universally used
in process simulations. Although the shortcomings of
CEOS at or near the critical region are well documented, (Millat, Dymond & Nieto de Castro, 1996)
their effect on blowdown simulation accuracy is not
known.
The less sophisticated SPLIT FLUID MODEL, on
the other hand assumes single phase discharge through
the orifice and a constant liquid wall heat transfer
coefficient. Also, in contrast to the BLOWDOWN
model, the effects of nucleation and gravitational settlement of liquid droplets inside the vessel on blowdown
simulation are not considered.
In this paper we briefly present the development of a
BLOWDOWN model based on three cubic equations
of state including SRK (Soave, 1972), PR (Peng &
Robinson, 1976) and the newly developed TCC CEOS
(Twu, Coon & Cunningham, 1992) for simulating vapour space blowdown of vessels containing multicomponent hydrocarbon mixtures. The model, here termed
as BLOWSIM is computationally efficient and requires
the minimum number of input parameters. Its predictions are compared against those obtained from
BLOWDOWN as well as published field data for high
pressure blowdown of a full-size vessel containing a
condensable hydrocarbon mixture (Szczepanski, 1994).
We have particularly selected the TCC CEOS in view of
its superior performance in predicting liquid densities
(Twu et al., 1992; Mahgerefteh & Wong, 1999).

2. Development of the model

We account for the most important processes taking
place during blowdown of an isolated vessel containing

either gas or two-phase mixture through a single orifice

from the top of the vessel. These include non-equilibrium effects between phases, heat transfer between
each fluid phase and their corresponding sections of
vessel wall, inter-phase fluxes due to evaporation and
condensation, and the effects of sonic flow at the
orifice. Typical outputs include the variations of discharge rate, pressure as well as fluid and wall temperatures with time.
In common with the BLOWDOWN model, the depressurisation process is approximated by a series of
variable pressure increments. The advantages of employing pressure increments are its thermodynamic convenience and computational efficiency as opposed to
choosing time intervals (as employed by the model of
Overa et al., 1994) which are thermodynamically
irrelevant.

2.1. Fluid phase material balances

The fluid in the vessel during blowdown is divided
into two zones. Zone 1 comprises sub-cooled vapour
and condensed vapour, whereas zone 2 represents the
liquid or condensed vapour from zone 1. If the mixture
is single phase (vapour only), zone 1 contains pure
vapour whilst zone 2 is eliminated. The appropriate
assumptions and the material balances for both zones
are described separately in the following.

2.1.1. Zone 1: condensation in sub-cooled 6apour

In the absence of an appropriate predictive theoreticalempirical model, we assume that the delay in the
formation of liquid droplets is negligible when compared with depressurisation duration for a given pressure increment. Additionally, the condensed vapour is
assumed to be at equilibrium with the vapour phase
and for a given pressure interval, it settles immediately
either at the bottom of the vessel (vapour filled vessel)
or above the existing liquid phase (two phase filled
vessel).
2.1.2. Zone 2: e6aporation in boiling liquid
In contrast to the BLOWDOWN model, we simplify
the boiling process in zone 2 by assuming that bubbles
rise rapidly and separate freely from the boiling liquid
which is at equilibrium with the evaporated vapour.
The latter is assumed to be well mixed with zone 1
fluid.
2.2. Thermodynamic trajectories for fluid phases
The effect of non-isentropic expansion of each fluid
phase inside the vessel during blowdown is accounted
for by assuming polytropic expansion (Bett, Rowlinson
& Saville, 1975) in which both heat and work are
transferred.

H. Mahgerefteh, S.M.A. Wong / Computers and Chemical Engineering 23 (1999) 13091317

On the basis of the second law of thermodynamics

and assuming that the amount of heat transfer to each
fluid phase during a given small time interval is infinitesimal so that the fluid temperature remains unchanged, the specific entropies in zone 1, Si,Z1, for a
vessel containing either vapour or two-phase mixture
are given by
Vapour only:
Si,Z1 =Si 1,Z1 +

Qi,Z1
Ti,Z1Ni 1,Z1

(1)

Two-phase mixture:
Si,Z1 =

Ti,Z1[(Si 1,V)Ni 1,V +(Si,EL)Ni,EL]+ Qi,Z1

Ti,Z1(Ni 1,V +Ni,EL)

(2)

Where Ni 1,Z1, total number of moles of zone 1 fluid in

the vessel at stage i 1; Qi,Z1, total heat transfer to
zone 1 fluid at stage I, Si,EL, specific entropy of evaporated liquid from zone 2 at stage i; Si 1,V, specific
entropy of vapour at stage i 1; Si 1,Z1, specific entropy of fluid in zone 1 at stage i 1; Ti,Z1, fluid
temperature in zone 1 at stage i.
The mole fraction of condensed vapour and fluid
temperature are determined by pressure entropy flash
calculation at vessel pressure, Pi, and entropy, Si,Z1.
Similarly, the specific entropy in zone 2, Si,Z2, for a
two-phase mixture is given by
Si,Z2 =

Ti,Z2[(Si 1,L)Ni 1,L +(Si 1,CV)Ni 1,CV]+ Qi,Z2

Ti,Z2(Ni 1,L +Ni 1,CV)
(3)

Where Si 1,L = specific entropy of liquid at stage i1;

Si 1,CV = specific entropy of condensed vapour from
zone 1 at stage I, Ti,Z2 =fluid temperature in zone 2 at
stage i.
The mole fraction of evaporated liquid and its temperature are determined by performing pressureentropy flash calculations at vessel pressure, Pi.

2.3. Heat transfer between 6essel wall and fluid phases

The energy balance between each fluid phase and the
associated section of vessel wall is derived on the basis
of the following assumptions:
Blowdown is performed under no fire mild weather
conditions where heat transfer between the vessel
and surrounding is negligible when compared with
other modes of heat transfer within the vessel.
Vessel wall temperature in contact with each fluid
zone is assumed to be uniform along the vessel
surface and any temperature gradient across the
thickness of the wall is negligible.
Hence, the heat balance between each zone and
corresponding section of vessel wall is given by

1311

Qi,Zx = (T iW,Zx TZx )UZxAZx Dt

1
T iW,Zx)
= MW,ZxCp,W(T iW,Zx

(4)

Where x=zone number, either 1 or 2; AZx = internal

area of vessel wall in contact with zone x; CpW =specific heat capacity of vessel wall material at constant
pressure; MW,Zx, mass of section of vessel wall in con1
tact with zone x; T iW,Zx
, T iW,Zx, average wall temperatures adjacent to zone x at stages i 1 and i,
respectively; UZx = overall heat transfer coefficient between zone x and corresponding section of vessel wall;
CpW, taken for mild steel is assumed to be a negligible
function of temperature below 400 K (Incropera & De
Witt, 1996c).
The assumptions and correlations for predicting the
heat transfer coefficient and heat flux between vessel
wall and each fluid zone are given in the following.

2.4. Heat transfer between 6essel wall and zone 1

Based on published experimental observations
(Haque et al., 1992a; Overa et al., 1994) we assume that
during blowdown, natural convection dominates forced
convection induced by the discharging material in the
vapour space.
At a given pressure increment, we further assume
that the amount of condensed vapour in zone 1 is small
enough so that its effect on heat transfer with associated section of vessel wall may be ignored. The mixture
in zone 1 is therefore treated as vapour only with the
appropriate thermophysical properties being calculated
at the film temperature, (T iW,Z1 + T iZ1)/2. For a given
vapour temperature, TZ1 the wall temperate, TW,Z1 is in
turn determined from Eq. (4) using Brents iteration
method (Press, Teukolsky, Vetterling & Flannery,
1992).
A correlation for predicting the natural convection
heat transfer coefficient for a fluid in an isothermal
container is not available. In this work, a vertical vessel
is approximated by ignoring the effects of curvature on
rate of heat transfer and treating the cylindrical section
as a vertical plane with the same inner surface area. The
two ends on the other hand are approximated by two
horizontal planes.
Churchill and Chus correlation (Incropera & De
Witt, 1996a) is used to calculate the heat transfer
coefficient for cylindrical section of the vessel wall in
contact with vapour.
The advantage of the above heat transfer correlation
over those used by Haque et al. (1992a) and Overa et
al. (1994) is its applicability to the entire range of
Rayleigh number (Incropera & De Witt, 1996a) thereby
covering both laminar and turbulent flows.
The heat transfer coefficients between vapour and
both ends of the vessel are calculated from McAdamss
correlation (Incropera & De Witt, 1996a).

1312

H. Mahgerefteh, S.M.A. Wong / Computers and Chemical Engineering 23 (1999) 13091317

2.4.1. Heat transfer between 6essel wall and zone 2

In zone 2, nucleate boiling is most likely to be
dominant during blowdown as the differences in wall
and liquid temperature are expected to be small due to
the very high heat transfer rates. Accordingly, the corresponding heat flux is estimated using the widely recommended (Burmeister, 1993; Bejan, 1995; Incropera &
De Witt, 1996b) Rohsensows correlation (Incropera &
De Witt, 1996b). The total heat transfer to the liquid
from the wall is in turn obtained by substitution into
Eq. (4).
It is important to note that Rohsenows correlation
applies only to clean surfaces where according to Incropera and De Witt (1996b) any variations can give rise
to as much as 9 100% error in the calculated heat flux.
In their blowdown simulation, Overa et al. (1994)
simply use experimentally estimated liquid wall heat
transfer coefficients in the range 1000 3000 W/m2K. In
this study, for the sake of comparison, the total heat
transfer is also calculated in the same manner and the
effect on blowdown predictions is evaluated by comparison with those based on Rohsensows correlation
(Incropera & De Witt, 1996b).

Vapour thermal conductivity and viscosity required

for the heat transfer calculations are obtained using Ely
and Hanleys method (Ely & Hanley, 1981, 1983). The
main advantage of this method is that only the critical
constants and the acentric factor are required for the
mixture transport property calculation.
Liquid mixture viscosity and thermal conductivity as
required in Rohsenows correlation are determined by a
semi-empirical scheme proposed by Dymond and Assael (Assael, Trusler & Tsolakis, 1996). The range of
application is generally from about 280 to 400 K with
pressures from saturation up to 986.9 atm. The uncertainty of the correlations is claimed not to be greater
than 5% (Assael et al., 1996). Two thousand experimental values of viscosity and thermal conductivity were
employed to optimise the coefficients used in the
scheme.
Surface tension is predicted by MacleodSugden correlation (Reid, Prausnitz & Sherwood, 1986) which is
applicable to polar and non-polar systems. A typical
accuracy is claimed to be: 95 10% (Reid et al.,
1986).

2.5. Discharge calculation

3. Validation of BLOWSIM
The fluid approaching a relief valve during depressurisation can either be single or two-phase. If the
back-pressure is sufficiently low, the fluid accelerates
through the orifice at its maximum velocity and expands rapidly from the upstream to the orifice pressure.
Consequently, condensation may occur which results in
two-phase flow at the orifice. In practice, such phase
changes have profound implications on downstream
process design and hence a prior prediction is
important.
The determination of the discharge rate and the state
of fluid phase at the orifice is described elsewhere
(Haque et al., 1992a). In essence, it is based on carrying
out an energy balance calculation across the orifice
assuming insentropic expansion coupled with thermodynamic and phase equilibrium.

2.6. Thermophysical properties

Thermodynamic and phase properties are predicted
by Soave Redlich Kwong (Soave, 1972), Peng
Robinson (Peng & Robinson, 1976) and the newly
developed TwuCoon Cunningham (Twu et al., 1992)
CEOS. The latter has been specifically developed to
address the drawbacks of SRK and PR CEOS in
predicting liquid densities and vapour pressures at low
temperatures as well as those for hydrocarbons with
acentric factors above 0.5. The number and the appropriate fluid phase present are obtained using the stability test based on the tangent plane criterion of Gibbs
developed by Michelsen (1982).

3.1. Selection of experimental data

Due to its hazardous nature and the associated high
capital equipment expenditure required, the amount of
experimental data relating to blowdown of vessels containing hydrocarbon mixtures is very scarce. The limited data that is reported is mainly confined to small
vessels (:0.050.85 m3, see for examples Norris, 1994
and Overa et al., 1994) without sufficient instrumentation required for detailed measurements. Haque et al.
(1992b) recently reported the results of an extensive
series of blowdown experiments on a full-size vessel
(2.78 m3) at the British Gas test site in Spadeadam.
However, the absence of detailed information about the
vessels dimensions and precise composition of the hydrocarbon mixtures tested render the study of limited
value for validation purposes.
Approximately 2 years later, the missing information
relating to Haque et al.s blowdown experiments was
produced in a publication by Szczepanski (1994). The
vessel is a full-size, vertically oriented suction scrubber
for a gas compressor with length and inside diameter of
3.240 m (2.75 m tan-to-tan) and 1.130 m, respectively.
The wall thickness is 59 mm. The vessels material of
construction has not been given. In this study, it is
assumed to be carbon steel. Table 1 shows the equivalent choke diameter, feed conditions and composition
prior to blowdown of a condensable hydrocarbon
mixture.

H. Mahgerefteh, S.M.A. Wong / Computers and Chemical Engineering 23 (1999) 13091317

1313

The above data are used in this study to evaluate the

performance of BLOWSIM. It should be noted that the
measured and predicted wall temperatures obtained
from BLOWDOWN refer to the inner wall temperatures whereas BLOWSIMs predictions represent the
average values as the temperature gradient across the
thickness of the wall is assumed to be negligible. The
claimed uncertainties associated with the measured values of vessel temperature and pressure were reported to
be 9 0.5 K and 90.2 atm, respectively.

3.2. Validation
Fig. 1 presents predicted vessel pressure time profiles
for the condensable gas mixture based on BLOWSIM
and BLOWDOWN (Haque et al., 1992b) as compared
with experimental results. The data indicate that
Table 1
Experimental conditions for the condensable gas hydrocarbon mixture prior to blowdown (Szczepanski, 1994)a
Composition (% mole)

P0 (atm)

T0 (K)

dC (mm)

64C1, 6C2, 28n-C3, 2n-C4

116

293

10.00

P0, initial pressure; T0, temperature; dC, equivalent choke diame-

ter.

Fig. 1. Comparison between measured (Exp) and predicted pressure

time profiles from BLOWDOWN and BLOWSIM based on various
CEOS (SRK, PR and TCC CEOS) for condensable gas mixture (64
mol% C1, 6 mol% C2, 28 mol% n-C3 and 2 mol% n-C4).

Fig. 2. Comparison between measured (shaded area) and predicted

bulk vapour temperature time profiles from BLOWDOWN and
BLOWSIM based on various CEOS (SRK, PR and TCC CEOS) for
condensable gas mixture (64 mol% C1, 6 mol% C2, 28 mol% n-C3 and
2 mole% n-C4).

BLOWSIMs performance is relatively insensitive to the

type CEOS used. Also, both mathematical models are
capable of predicting pressuretime history relatively
accurately.
Fig. 2 shows the corresponding bulk vapour temperaturetime profiles. The shaded area showing measured
data, indicates the presence of temperature gradients
within the gas. According to the data, the depressurisation process results in approximately 50 K drop in the
gas temperature. The corresponding temperature drop
based on isentropic expansion where the effect of heat
transfer from the vessel walls to the gas is ignored is
132 K
All three CEOS used in conjunction with BLOWSIM
give rise to similar and relatively accurate predictions of
field data throughout the blowdown process. Assuming
a minimum measured average bulk gas temperature of
247 K, BLOWSIM predicts this temperature to :2 K.
Fig. 3 shows predicted temperaturetime profiles for
the unwetted wall based on BLOWSIM and BLOWDOWN. The corresponding experimental data are represented by the shaded area. Once again, BLOWSIM
data are relatively insensitive to the choice of CEOS.
The model over estimates the unwetted wall temperatures by : 4 K. On the other hand, relatively accurate
predictions are obtained from BLOWDOWN.

1314

H. Mahgerefteh, S.M.A. Wong / Computers and Chemical Engineering 23 (1999) 13091317

Fig. 4 shows the corresponding predicted and measured bulk liquid temperatures vs time profiles. As
before, the data indicate that BLOWSIMs performance is insensitive to the choice of the CEOS used.
Also during the first 600s of blowdown, where the rate
of formation of condensed liquid is most rapid, more
accurate liquid temperatures are predicted compared to
those during the subsequent period. This is likely due to
the actual proportion of the heavier hydrocarbons in
the vessel during the latter stages of depressurisation
being greater than the predicted values. Consequently,
the estimated boiling liquid temperature will be lower
than the measured value.
In general, BLOWSIM performs better than BLOWDOWN. It is interesting to note that the experimental
data indicate that the formation of the liquid phase is
nearly instantaneous while BLOWSIM and BLOWDOWN predict liquid formation :30 and 100 s following blowdown, respectively.
In Fig. 5, the measured temperature time profiles for
the wetted wall are compared with the predicted profiles based on BLOWSIM and BLOWDOWN. Interestingly, although liquid formation is instantaneous (see
Fig. 4) no wetted wall temperature is recorded until
:100 s after commencement of depressurisation.

Fig. 4. Comparison between measured (shaded area) and predicted

bulk liquid temperature time profiles from BLOWDOWN and
BLOWSIM based on various CEOS (SRK, PR and TCC CEOS) for
condensable gas mixture (64 mol% C1, 6 mol% C2, 28 mol% n-C3 and
2 mol% n-C4).

On the basis of the data in Figs. 4 and 5, it is clear

that BLOWSIM predicts earlier formation of the liquid
phase compared to BLOWDOWN. This is likely attributed to the following reasons:
1. BLOWSIMs assumption of instantaneous formation and immediate settlement of liquid droplets
from vapour phase as soon as the two-phase
boundary is crossed.
2. The application of CEOS may result in higher saturated pressure and temperature (hence a later crossing of the phase boundary) when compared to those
predicted on the basis of the extended principle of
corresponding states (Rowlinson & Watson, 1969)
used in BLOWDOWN.
In order to investigate the relative importance of
point 1 above, the nucleation time and settling velocity
relative to the vapour in the vessel are estimated based
on the same correlations as those used in BLOWDOWN (Haque et al., 1992a). These are given by:
1
t=
(5)
nliq
(6)
6settle = 0.03+ 3nliq
Fig. 3. Comparison between measured (shaded area) and predicted
unwetted wall temperaturetime profiles from BLOWDOWN and
BLOWSIM based on various CEOS (SRK, PR and TCC CEOS) for
condensable gas mixture (64 mol% C1, 6 mol% C2, 28 mol% n-C3 and
2 mol% n-C4).

where t (s) is the nucleation time, 6settle (ms 1) is the

settling velocity of liquid droplets relative to the gas
and nliq is the equilibrium liquid mole fraction condensed from the gas phase inside the vessel.

H. Mahgerefteh, S.M.A. Wong / Computers and Chemical Engineering 23 (1999) 13091317

At the onset of condensation, nliq, predicted from

TCC CEOS, is 0.7. Hence, based on the above two
equations, the values of t and 6settle are 1.4 s and 2.2
m/s, respectively. The average velocity of evaporating
gas in the vessel is estimated as 0.0097 m/s. Hence, the
absolute settling velocity of liquid droplets is 2.19 m/s.
The time required for the liquid droplets to settle
gravitationally is determined by dividing the total
height of the vessel by, 6settle. The corresponding value
is 1.4 s. The total time delay for formation of the liquid
phase at the bottom of the vessel is therefore 2.8 s
( = 1.4+1.4) which is comparatively small. In the absence of any other arguments, it is therefore likely that
the earlier formation of liquid phase predicted from
BLOWSIM as compared to BLOWDOWN is attributed to the use of different types of CEOS.
Returning to Fig. 5 although both simulations perform reasonably well, BLOWSIM predicts a lower
wetted wall temperature (: 5 K) compared to the
measured value. This is most probably a consequence
of higher heat transfer coefficient as predicted from
Rohsensows correlation (Incropera & De Witt, 1996b)
than the actual value. Fig. 6 shows the effect of incorporating constant heat transfer coefficients in the range
1000, 2000, and 3000 W/m2K on BLOWSIMs predictions of the wetted wall temperature. The heat transfer

1315

Fig. 6. Comparison between measured (shaded area) and predicted

wetted wall temperature time profiles from BLOWSIM based on
Rohsenows correlation (correlation) for predicting boiling heat flux
of liquid phase and various constant heat transfer coefficients between liquid phase and vessel wall (1000, 200 and 3000 W/m2K) for
condensable gas mixture (64 mol% C1, 6 mol% C2, 28 mol% n-C3
and 2 mol% n-C4).

coefficients are in the same range as those recommended by Overa et al. (1994) in carrying out their
blowdown experiments. In order to avoid congestion,
only the data generated using the TCC CEOS are
shown here. Clearly as the heat transfer coefficient is
decreased a better agreement between simulation results
and experimental data is obtainable.

4. Conclusions

Fig. 5. Comparison between measured (shaded area) and predicted

wetted wall temperaturetime profiles from BLOWDOWN and
BLOWSIM based on various CEOS (SRK, PR and TCC CEOS) for
condensable gas mixture (64 mol% C1, 6 mol% C2, 28 mol% n-C3
and 2 mol% n-C4).

The development of a robust numerical model,

BLOWSIM based on three CEOS for simulating blowdown of vessels containing multi-component hydrocarbon mixtures has been described. The performance of
the model is evaluated by comparing predictions with
those generated from BLOWDOWN and experimental
data for depressurisation of a condensable gas from
high pressure. The application of all three CEOS leads
to similar accuracies in BLOWSIMs predictions. Also,
both simulations produce accurate predictions of vessel
pressure as a function of time.
The minimum average bulk gas temperature is predicted to within 2 K while the unwetted wall temperature is overestimated by : 4 K when compared with
measured data. Based on the conventional method

1316

H. Mahgerefteh, S.M.A. Wong / Computers and Chemical Engineering 23 (1999) 13091317

where the vessels inventory is assumed to expand adiabatically, the minimum bulk gas temperature is underestimated by 80 K. This result clearly demonstrates the
importance of taking into account the effects of heat
transfer from the vessel wall on the fluid temperature.
The observed slightly lower (:5 K) predicted wetted
wall temperature as compared to the experimental
value is most probably a consequence of a too high
boiling heat transfer coefficient obtained from Rohsensows correlation. By assuming constant liquidwall
heat transfer coefficients in the ranges recommended by
Overa et al. (1994), we obtain more accurate wall
temperatures as the heat transfer coefficient is
decreased.
Based on the assumption of negligible temperature
gradients across thickness of the vessel wall, we are able
to obtain reasonably good predictions for both liquid
and average wetted wall temperatures. However, boiling heat transfer between the liquid phase and vessel
wall is expected to result in a rather significant temperature difference across the vessel wall. For example, by
assuming a boiling heat transfer coefficient of 3000
W/m2K and steady-state conduction, the estimated
temperature difference across the wetted vessel wall is
:10 K. This value can be far higher in a fire situation
where the resulting hoop stress may easily lead to
catastrophic rupture of the vessel during blowdown.
Hence a useful extension of this work would be simulation of the temperature gradients across the vessel wall
under such adverse conditions.
It is interesting to note that the measured liquid
temperature time profile indicates that the formation of
the liquid phase is nearly instantaneous while BLOWSIM and BLOWDOWN predict liquid formation sometime later. The earlier formation of liquid phase
predicted from BLOWSIM as compared to BLOWDOWN is attributed to the use of different types of
EOS rather than the assumptions of instantaneous formation and immediate settlement of liquid droplets
condensed from the vapour phase as employed in the
former model.
In conclusion, we believe that the value of this study
can be significantly improved by comparison of our
models predictions with a wider range of hydrocarbon
systems and conditions. Such data would be important
for validation as well as being directly useful for modelling purposes. The latter is because blowdown involves a large number of relatively complex interactive
processes that may be extremely difficult to model on a
purely theoretical basis. It is therefore inevitable that a
robust blowdown model will rely, to some extent on
empirically obtained correlations.
Unfortunately experimental data of this nature are
often difficult to obtain due to the high capital equipment costs involved as well as the associated safety
implications. Industry seems to be the only source with

sufficient resources to address this important issue.

However, the limited data that are available are often
proprietary. Our experience has been that even in the
case of published data relating to the more sophisticated blowdown models, important information relating to the precise experimental conditions and
mathematical formulations are missing. We believe that
the pace of progress for such studies much depends on
a change of culture in the process engineering industry.
References
API Recommended Practice 521. (1990). Guide for pressure-relie6ing
and depressuring systems (3rd ed.) API.
Assael, M. J., Trusler, J. P. M., & Tsolakis, T. F. (1996). Thermophysical properties of fluids, an introduction to their prediction.
London: Imperial College Press.
Bejan, A. (1995). Con6ection heat transfer (2nd ed.). John Wiley and
Sons (Chapter 10).
Bett, K. E., Rowlinson., J. S., & Saville, G. (1975). Thermodynamics
for chemical engineers (p. 196). London: Athlone.
Burmeister, L. C. (1993). Con6ecti6e heat transfer (1st ed.). John
Wiley and Sons Chapter 11.
Ely, J. F., & Hanley, H. J. M. (1981). Prediction of transport
properties. 1. Viscosity of fluids and mixtures. Industrial Engineering Chemistry Fundamentals, 20, 323 332.
Ely, J. F., & Hanley, H. J. M. (1983). Prediction of transport
properties. 2. Thermal conductivity of pure fluids and mixtures.
Industrial Engineering Chemistry Fundamentals, 22, 90 97.
Haque, M. A., Richardson, S. M., & Saville, G. (1992a). Blowdown
of pressure vessels. I. Computer model. Transactions of the Institute of Chemical Engineers Part B: Process Safety En6ironmental
Protection, 70(BI), 1 9.
Haque, M. A., Richardson, S. M., Saville, G., Chamberlain, G., &
Shirvill, L. (1992b). Blowdown of pressure vessels. II. Experimental validation of computer model and case studies. Transactions of
the Institute of Chemical Engineers Part B: Process Safety En6ironmental Protection, 70(BI), 10 17.
Incropera, F. P., & De Witt, D. P. (1996a). Fundamentals of heat and
mass transfer (4th ed.). John Wiley and Sons (Chapter 9).
Incropera, F. P., & De Witt, D. P. (1996b). Fundamentals of heat and
mass transfer (4th ed.). John Wiley and Sons (Chapter 10).
Incropera, F. P., & De Witt, D. P. (1996c). Fundamentals of heat and
mass transfer (4th ed.). John Wiley and Sons (Appendix A).
Katz, D. L., & Lee, R. L. (1990). Natural gas engineering
production and storage (1st Edition). McGraw-Hill (Chapter 5).
Mahgerefteh, H., & Wong, S. M. A. (1999). Evaluation of the
performance of Twu Cunningham cubic equation of state in
predicting vapour and liquid speed of sound for hydrocarbon
systems. High Temperatures High Presures, 31 (5), 126 136.
Michelsen, M. L. (1982). The isothermal flash problem. Part I,
stability. Fluid Phase Equilibria, 9, 1 19.
Millat, J., Dymond, J. H., & Nieto de Castro, C. A. (1996). Transport
properties of fluids their correlation prediction and estimation
(1st Edition, pp. 165 186). IUPAC Cambridge University Press.
Norris III, H. L. (1994). Hydrocarbon blowdown from vessels and
pipelines. In SPE 69th Annual Technical Conference and Exhibition (pp. 593 602). New Orleans, LA.
Overa, S. J., Stange, E., & Salater, P. (1994). Determination of
temperatures and flare rates during depressurization and fire. In
Proceedings of 72nd GPA Annual Con6ention (pp. 235 247).
Peng, D. Y., & Robinson, D. B. (1976). A new two-constant equation
of state. Industrial Engineering Chemistry Fundamentals, 15, 59
64.

H. Mahgerefteh, S.M.A. Wong / Computers and Chemical Engineering 23 (1999) 13091317

Press, W. H., Teukolsky, S. A., Vetterling, W. T., & Flannery,
B. P. (1992). Numerical recipes in FORTRAN 77: the art of
scientific computing (2nd ed.). Cambridge University Press (Chapter 9).
Reid, R. C., Prausinitz, J. M., & Sherwood, T. K. (1986). The
properties of gases and liquids (4th ed., p. 642). McGraw-Hill
Book Company.
Rowlinson, J. S., & Watson, I. D. (1969). The prediction of the
thermodynamic properties of fluids and fluid mixtures, i. the
principle of corresponding states and its extensions. Chemical
Engineering Science, 24, 15651574.

1317

Soave, G. (1972). Equilibrium constants from a modified Redlich

Kwong equation of state. Chemical Engineering Science, 27,
1197 1203.
Szczepanski, R. (1994) Simulation programs for blowdown of pressure vessels. IChemE SONG Meeting.
Twu, C. H., Coon, J. E., & Cunningham, J. R. (1992). A new cubic
equation of state. Fluid Phase Equilibria, 75, 65 79.
Wong, S. M. A. (1998) Development of a mathematical model for
blowdown of vessels containing multi-component hydrocarbon
mixtures, PhD Thesis, Department of Chemical Engineering, University College London, UK.