Journal Pre-proof

A review on modeling and simulation of blowdown from pressurized

vessels and pipelines

Umar Shafiq, Azmi M. Shariff, Muhammad Babar, Babar Azeem,

Abulhassan Ali, Mohamad Azmi Bustam

PII: S0957-5820(19)30262-9
DOI: https://doi.org/10.1016/j.psep.2019.10.035
Reference: PSEP 1972

To appear in: Process Safety and Environmental Protection

Received Date: 14 February 2019

Revised Date: 24 September 2019
Accepted Date: 31 October 2019

Please cite this article as: Shafiq U, Shariff AM, Babar M, Azeem B, Ali A, Bustam MA, A
review on modeling and simulation of blowdown from pressurized vessels and pipelines,
Process Safety and Environmental Protection (2019),
doi: https://doi.org/10.1016/j.psep.2019.10.035

This is a PDF file of an article that has undergone enhancements after acceptance, such as
the addition of a cover page and metadata, and formatting for readability, but it is not yet the
definitive version of record. This version will undergo additional copyediting, typesetting and
review before it is published in its final form, but we are providing this version to give early
visibility of the article. Please note that, during the production process, errors may be
discovered which could affect the content, and all legal disclaimers that apply to the journal
pertain.

© 2019 Published by Elsevier.

A review on modeling and simulation of blowdown

from pressurized vessels and pipelines

Umar Shafiq1, Azmi M Shariff*1 azmish@utp.edu.my, Muhammad Babar1, Babar Azeem1,

Abulhassan Ali2, and Mohamad Azmi Bustam1

of
1
CO₂ Research Centre (CO2RES), Universiti Teknologi PETRONAS, 32610 Bandar Seri

ro
Iskandar, Perak, Malaysia

2 -p
Department of Chemical and Materials Engineering, University of Jeddah, Jeddah, Kingdom

of Saudi Arabia
re
lP

*
Corresponding Author. Tel.: +60 5 3687530, (Azmi Shariff)
na

Graphical abstract
ur
Jo

Page | 1
 of
ro
-p
re
Abstract
lP

In process industry, failure or rupture of pressurized vessel is very dangerous especially

when there is an escape of flammable gaseous mixture that can cause potential fire or
na

explosion. One of the scenarios that causes such accidents is the blowdown process.

Therefore, it becomes crucial to control blowdown process to prevent such accidents. It is

ur

important to design optimally to make sure that blowdown valve is according to the

requirements. For the safe use of a pressure relief system, some of the parameters are critical,
Jo

for example, selection of construction material, sizing of relief valves, temperature, and

pressure, etc. There is no literature currently available that discusses all the mathematical

models or simulation tools for optimum design of the blowdown process. This subject matters

because the available models or tools cover different aspects of blowdown process. A

meticulous review is required to present the applications of these models and tools based on

Page | 2
the accidental scenarios. Therefore, this paper critically reviews the models and tools that are

developed purposely to calculate optimum blowdown parameters based on fluid and vessel

conditions. Recommendations are given for the development of new simulation tool to

simulate phase change conditions especially when solid formation is involved.

Keywords: Blowdown, high-pressure vessels, hazards, depressurization, simulation tools,

modeling

Nomenclature

of
BP British Petroleum MSM: Marginal Stability Two-Fluid Model

ro
CCS: Carbon Capture and Sequestration PR: Peng & Robinson

CFD:

EoS:
Computational Fluid Dynamics

Equation of State
PRV:

QRA:
-p Pressure Relief Valve

Quantitative Risk Assessment

re
FBR: Full Bore Rupture SCWR: Supercritical Water-Cooled Reactor
lP

FDM: Finite Difference Method SDM: Simplified Dynamic Model

FEM: Finite Element Method SRK: Soave-Redlich-Kwong

na

HC: Hydrocarbon TEM: Thermodynamic Equilibrium Model

ur

HEM: Homogenous Equilibrium Model VBsim: Vessel Blowdown Simulator

HRM: Homogenous Relaxation Model V-L Vapor-Liquid

Jo

ID: Inner Diameter VPM Vent Pipe Model

MAD Maximum Absolute Deviation V-S Vapor-Solid

META: Multi-Component Equilibrium Two-Phase Analyser VLE: Vapor Liquid

Equilibrium

Page | 3
 MOC: Method of Characteristics

1 Introduction

Air-tight pressurized vessels and pipelines are widely used to transport or hold gases and

process liquids or fluids in petrochemical plants, refineries, and process industries. These

vessels are generally subjected to high-pressure loading and unidentical external or internal

pressures in comparison to the ambient pressure [1]. Consequently, it becomes potentially

hazardous and dangerous because of characteristic operational pressure of the vessel [2]. The

of
rupture of pipeline and subsequent eruption of fire broke down the platform into seabed

resulting in a loss of 167 lives during the Piper Alpha incident [3]. Many other tragedies have

ro
been witnessed that caused a significant number of fatalities due to pipeline and vessel

-p
rupture [4, 5]. After the Piper Alpha tragedy, the offshore industry carried out a great deal of

research work to understand the characteristics of hydrocarbon fires and explosions [6, 7].
re
The results indicate that a pressure above 4 bar developed in typical topsides of the vessels is

enough to inflict the structural damage such as weld failure, local instability, or extensive
lP

inelasticity.
na

The National Board of Boiler and Pressure Vessel Inspectors recorded a 24 % increase in

the number of accidents involving pressure vessels during 1999-2000. Steam heating, power

boilers, and unfired pressure vessels were the major sources of such accidents [8]. From 1992
ur

to 2001, a total of 23,338 accidents related to pressure vessels were recorded [8-10].
Jo

Similarly, according to Swedish Nuclear Power Inspectorate [11], more than 66 % (2476) of

the pipeline failure incidents recorded during 1994-1999 were triggered due to leakage or

puncture. Common reasons for pressure vessel failures are illustrated in Figure 1.

Page | 4
 4% 8%
Design 2%
Misapplication 23%
Installation
Maintenance
2%
Repair
Operation
61%
External

Figure 1: Common reasons for pressure vessel failures

of
In oil and gas platforms or offshore operations, overpressure of pressurized vessels also

ro
arises during emergency situations due to fire or valves' malfunction. These industries usually

work with CO₂, natural gas, and fossil fuels [12, 13]. Most of the accidents at offshore oil and
-p
gas industries, nuclear power plants, and chemical plants cause the outflow of radioactive,
re
toxic, explosive, or flammable materials. Particularly in the oil and gas industry, failure or

rupture of pressurized vessels poses higher risk due to the possible escape of flammable
lP

gaseous mixture in a fire event. Therefore, process vessels are depressurized for the

transportation of flammable or environmentally toxic gaseous mixture to a safer place [14-

na

17]. Blowdown is a typical way of minimizing the failure hazard of vessels when an

emergency arises in a process industry. However, for the safe use of a pressure relief system,
ur

some of the parameters are crucial; for example, temperature, pressure, selection of

construction material, and sizing of relief valves. Prediction of precise minimum vessel wall
Jo

temperature during depressurization process can affect the selection of construction material,

reduce over-design, and subsequently lower the project cost. Similarly, over-design can also

be reduced by precise prediction of maximum flow rate during blowdown. Critical

parameters for a safe blowdown process are portrayed in Figure 2.

Page | 5
 of
Figure 2: Important parameters for safe blowdown process

1.1 Concerns and Considerations

ro
In recent years, safety and process engineers have encountered several safety-related

issues while dealing with high-pressure blowdown. The key principle of sustainable
-p
depressurization process design is to decrease the pressure and discharge the inventory safely
re
in the minimum possible time [18]. The discharge rate of compressible fluids generally

depends upon pressure, pressure drop, temperature, friction factor, Reynolds number,
lP

pipeline diameter, roughness, length, and gas properties [19]. However, the rapid expansion

of gas and generation of vapors can expose the pressure vessel to pressurized thermal shock
na

during an emergency blowdown process. This shock initiates due to integrated stress from

rapid transitions in pressure and temperature that result in the non-uniform distribution of
ur

temperature in vessel walls and, therefore, lead to the differential contraction and expansion.

Pressurized thermal shocks can also cause embrittlement of vessel walls and subsequent
Jo

“failure by fatigue” due to very low temperature generated inside the vessel because of Joule-

Thomson effect. Therefore, to predict the effects of vessel/pipeline rupture or leak, or to

design a sustainable blowdown process, there is dire need of a technique to predict the mass

efflux. Moreover, to calculate optimum blowdown time, a precise balance between maximum

acceptable depressurization time [20] and minimum fluid and wall temperature that may be

Page | 6
safely considered, is required. Some of the important factors to be considered before

designing the blowdown system are illustrated in Figure 3.

of
ro
Figure 3: Factors to be considered before designing a sustainable pressure relief system
-p
Conventionally, several numerical models and simulation tools are reported in literature
re
for the prediction of maximum possible blowdown time and fluid and vessel conditions. The

numerical models have been developed by commonly adopting the Homogenous Equilibrium
lP

Model (HEM) assumption such that each phase is considered to be at the phase equilibrium

and thermal equilibrium. Some of the researchers also coded simulation tools based on their
na

numerical models. However, the available mathematical models or simulation tools cover

different aspects of blowdown or depressurization process. Therefore, a literature review of

ur

the existing correlations, numerical models, and computational tools that have capability to
Jo

calculate optimum blowdown parameters for pressure vessels and pipelines, is presented in

this paper to summarize existing work and set out most overlooked aspects.

2 Design of Blowdown System for Pressure Vessels

In case of blowdown of vessels, a hazard may emerge due to very low temperature gained

by the material inside the process vessel. This can potentially lead to rupture even with a

Page | 7
small amount of force if temperature falls below the ductile-brittle transition temperature of

vessel's material of construction [21]. If free water is present inside vessel, it can result in

growth of hydrates or formation of liquid condensate that can get carried-over into the flare

or vent system. If large droplets of liquid condensate or continuous liquid stream gets into the

flare, they will start falling from the top covering entire flare into flames. This condition is

extremely dangerous as this falling and burning liquid may cause a major fire incident in

vicinity of the plant [22]. To date, several numerical models and simulation tools have been

developed that can simulate different blowdown scenarios of vessels based on vessel sizing,

of
blowdown conditions, and compositions etc. A timeline of these simulation tools or

ro
numerical models is illustrated in Figure 4 followed by a brief description (Some of the

authors didn't name their simulation tools; therefore, we have mentioned the name of the

authors instead of the simulation tool).

-p
DIERS FRICRUP PHAST Fredenhagen SDM VBsim
re
1990 1999 2001 2002 2014 2018
lP

1976 1993 1999 2001 2011 2015

na

BLOWDOWN BLOWSIM Cumber Mahgerefteh Zhang Park

Figure 4: Timeline for blowdown simulation tools and numerical models for vessels
ur

2.1 DIERS
A consortium of several companies founded DIERS program in 1976 to design pressure
Jo

relief system for runaway reactions and develop the additionally required technology. DIERS

research program performed several experimental studies, coded a comprehensive simulation

tool, and designed a bench-scale prototype apparatus. The aim was to predict two-phase flow

venting and analyze the possible applicability of various two-phase (vapor-liquid) flashing

flow methods for sizing of relief systems. In this model, it is considered that boiling takes

Page | 8
place through the entire volume of liquid rather than solely at the surface. Each bubble

occupies volume and displaces the liquid surface upward. Individual bubbles are able to rise

(slip) through the liquid with a velocity that depends on buoyancy and surface tension,

whereas they are retarded by viscosity and foamy character of the fluid [23]. DIERS program

evolved over time; the initial phase included study of vapor-liquid phase dynamics, second

phase involved the experimental investigations for small and large scale integral blowdown

and vented runaway reaction, and the final phase included the development of a coded

computational package along with a bench scale prototype experimental apparatus [24].

of
2.2 BLOWDOWN

ro
A brief description of prediction and experimentation of fluid and vessel behavior during

blowdown process was presented by Haque et al. in 1990 [25]. Experiments were performed
-p
on blowdown of pressurized vessels containing pure N₂, 70 % N₂–30 % CO₂, and natural

gas/hydrocarbon mixture. Based on the experiments performed and assessment of available

re
data, a computational tool called BLOWDOWN was coded to simulate rapid depressurization
lP

of vessels. The developed BLOWDOWN package has the ability to measure the temperature,

pressure, and composition as a function of time.

na

For the experimental investigation, a 6.325 mm orifice was used to blowdown the mixture

from a 0.086 m³ vessel with 25 mm of wall thickness. Pure N₂ took almost 100 s to reach the
ur

atmospheric pressure from 15 MPa. Whereas, the vessel containing 70 % N₂–30 % CO₂

binary mixture reached the atmospheric pressure in 60 s. The dynamic change in gas and wall
Jo

temperature is depicted in Figure 5. To analyze the predicted results, the coefficient of

determination, R2, and geometric mean bias of the model data are calculated. For this model,

the R2 for temperature prediction is ≈ 0.977. A comparative illustration of R2 values and

geometric mean bias of different models is presented in Figure 33 and Figure 34 respectively.

Page | 9
 295
(a)
275
Temperature/K

255

235

215

195

175
0 10 20 30 40 50 60 70 80 90 100
Time/s

0 0
(b)

of
-20 -5

Wall temperature/K
Gas temperature/K

-40 -10

ro
-60

-80

-100

-120
-p -30
re
0 20 40 60 80 100
Time/s
lP

Figure 5: Dynamic temperature changes during blowdown for (a) 70 % N₂–30 % CO₂
mixture and (b) pure N₂ [25]

Subsequently, an extension in the simulation tool was made that can also be used to
na

simulate the blowdown of pipeline, particularly a long subsea pipeline. Extension of this

work by Haque et al. in 1992 covers simulation study of free water (produced during
ur

depressurization) for the arbitrary combinations of vessels and pipelines [26]. This extended

version of BLOWDOWN tool was validated again in 1992 by Haque et al. while supported
Jo

by several case studies [27, 28]. In the first case study; 53804 m long, 16.5″ subsea gas line

containing 51 MMSCF methane at 283 K was blown down from a pressure of 12 MPa. In the

second case study, full-bore depressurization of 2000 m long pipeline containing 1975 tonnes

of methane condensate was carried out. The initial pressure inside the pipeline was 20 MPa

and initial temperature was around 313 K.

Page | 10
 Another comparison of BLOWDOWN predictions and experimental data was made

during LPG test performed by British Petroleum (BP) and Shell on the Isle of Grain in 1985

[29-31]. Further experimental investigations on the prediction of blowdown parameters using

BLOWDOWN package are also available in literature [32]. In all the experimental studies,

BLOWDOWN package provides a reasonable understanding of the physical processes

occurring and an ability to make predictions with adequate certainty. However, use of

extended corresponding states principle introduces uncertainties that make the simulation

computationally difficult. Moreover, the limitations of a cubic equation of state (EoS) near or

of
at the critical region are well documented but their effect on the accuracy of blowdown

ro
simulation is still unknown. Further improvement and simulation of practical problems imply

the assumptions of homogeneity when the flow is two-phase and the quasi-steadiness could

be neglected.
-p
re
2.3 FRICRUP
A simple mechanistic model named FRICRUP was coded by Norris et al. in FORTRAN
lP

program to predict single and multi-phase flow behavior during blowdown of vessels or

pipelines [33]. HEM and thermodynamic equilibrium model (TEM) assumptions along with
na

no relative velocities between liquid and vapor phases were taken into account. Moreover,

FRICRUP tool was validated with the available data for gaseous mixture of CO₂, carbonated
ur

water, and air. The comparison shows that a reasonable mechanistic model has been

developed that can predict single and multi-phase blowdown from the vessel as well as the
Jo

pipe. Regardless of its complexities, the mechanistic model indicates disagreement with

multi-phase flow possibly due to TEM assumptions. A year later, Norris et al. conducted

another experimental study using HC gases [34]. The results were similar to the ones

achieved while using the non-HC gases because the basic mechanical model assumptions

remained the same.

Page | 11
2.4 BLOWSIM
BLOWSIM model [14, 35], developed by Mahgerefteh and Wong is based on three

cubic-EoSs including Peng and Robinson (PR) [36], Soave-Redlich-Kwong (SRK) [37], and

newly developed TCC Cubic-EoS [38]. BLOWSIM (BLOWdown-SIMulation) requires a

minimum number of input parameters. It is computationally efficient and can predict

pressure, mass flow rate, and fluid and wall temperature, all as a function of time.

BLOWSIM model considers the followings:

1. Heat transfer between fluid phases and their corresponding sections of the vessel wall.

of
2. Non-equilibrium effect between phases.

3. The effect of sonic flow at the orifice.

ro
4. Interphase fluxes because of condensation and evaporation.

-p
The performance of the featured model was assessed by the correlation of predicted data

generated from BLOWDOWN and experimental results obtained from blowdown of

re
condensable gas. Typical output results include dynamic temperature and pressure profiles

for liquid and vapor phases as well as unwetted and wetted walls. The achieved results and
lP

their comparison with different EoSs as well as the experimental data are illustrated in

Figure 6 and 7. BLOWSIM's predictions are quite insensitive to the Cubic-EoS used. In
na

general, it can be inferred that choice of the Cubic-EoS has a negligible effect on predicted

results. Moreover, both numerical models are capable of reasonably predicting the dynamic
ur

pressure variations. The R-square noted for TCC (TWU) model for temperature predictions is

≈ 0.992.
Jo

Page | 12
 12
Exp.
BLOWDOWN
10
SRK
PR
8
Pressure/MPa

TCC

0
0 200 400 600 800 1000 1200 1400

of
Time/s
Figure 6: Correlation between predicted and experimental pressure-time profiles [14]

ro
295
Exp.
290 BLOWDOWN
Bulk vapor temperature/K

285
280
-p SRK
PR
TWU
275
re
270
265
lP

260
255
250
245
na

240
0 300 600 900 1200 1500
Time/s
ur

Figure 7: Correlation between predicted and experimental vapor temperature profile [14]

2.5 PHAST
Jo

The Unified Dispersion Model (UDM) developed by Woodward and Cook in the early

90's was later implemented into DNV software package, termed as PHAST in version 6.0

[39-41]. PHAST is a helpful tool to study the initial release of mixture as well as the far-field

dispersion. Previously, PHAST was limited only to vapor and liquid phase release of the

chemicals. However, the upgraded version allows to study the occurrence of fluid to solid

transition during release of CO₂. PHAST was validated against several experimental studies

Page | 13
for unpressurized release of CO₂ [39, 42]. Later, Witlox et al. performed some more

experimental studies for the validation against pressurized release of CO₂ [43]. An overview

of the validation of different models including flammable effect, dispersion and pool

spreading/evaporation, and discharge models was provided along with extensive

experimental database for validation of the above models and scenarios [44].

2.6 P. S. Cumber
A numerical model was coded by Cumber for the prediction of vent sizing of pressurized

vessel during blowdown batch process [45]. The pressure vessel containing gas mixture is

of
considered as a single unit with thermodynamic equilibrium assumptions. This model has a

ro
vast range of venting models implemented for a two-phase system. However, selection of the

most suitable model for a specific case study necessitates the user to have a comprehensive
-p
information on the scope of application of the respective model. For this particular case,

predicted blowdown data was plotted against the experimental data as shown in Figure 8 and
re
9. The trend indicates that predicted gas temperature in the initial phases of depressurization
lP

is in agreement with the experimental data. However, ultimately it poorly predicts

temperature with an R-square ≈ 0.445 because of the adiabatic vessel wall assumption as
na

shown in Figure 9. The significant difference is with respect to the phase temperature

measurement that shows that thermodynamic equilibrium assumption is not effective for this
ur

case. However, the measured pressure of vessel was precisely predicted. For the vessels

depressurizing over long time-scales, the temperature of the vessel was under-predicted after
Jo

first few minutes as a consequence of assuming no heat transfer through the vessel walls.

This did not have a substantial effect on the simulated mass flow rate though. Another

discrepancy with the application of this model is the robustness that needs to be addressed

[46].

Page | 14
 14
Pre.
12 Exp.

10
Pressure/MPa

0
0 200 400 600 800 1000 1200 1400
Time/s

of
Figure 8: Dynamic pressure profile during blowdown of LPG [45]

ro
310
Pre.
300
Exp. LIQUID
290
280
-p Exp. GAS
Temperature/K

270
re
260
250
lP

240
230
220
0 200 400 600 800 1000 1200 1400
na

Time/s

Figure 9: Dynamic temperature profile during blowdown of LPG [45]

ur

2.7 A. Fredenhagen
Previously accomplished research work on depressurization of pure CO₂ leads towards
Jo

successful modeling of the pressure transitions and axial void profile during the expansion

process [47-50]. However, an experimental investigation on blowdown of the top vented

pressurized vessel containing CO₂ with impurities (N₂) was performed by Fredenhagen and

Eggers in 2001 to study the effect of impurities [51]. In this study, the axial void profiles,

axial temperature profiles, and pressure and mass were measured. Top vented cylindrical

Page | 15
vessel with 0.05 m³ volume, 0.242 m ID, and a 4:5 height to diameter ratio was used. The

venting line was linked to a quick opening ball valve with a cross-sectional area of 50 mm².

PR EoS was used for the calculation of phase equilibrium. For the prediction of this process,

a simulation tool was also coded based on the energy and mass balance and a drift flux

model. The dynamic pressure and temperature profiles predicted by featured simulation tool

and the experimental results are illustrated in Figure 10 and 11. The computational model

predicted reasonable results of pressure transition throughout the process. The overall R-

square for temperature predictions is ≈ 0.955, however, the temperature predictions at low

of
pressure were not much precise. Consequently, an improvement needs to be made for more

ro
precise modeling of thermodynamic behavior of the mixture.

14
Pre.
12

10
-p Exp.
re
8
Pressure/MPa

6
lP

2
na

0
0 10 20 30 40 50 60 70 80
Time/s
ur

Figure 10: Calculated and measured pressure transients for 17 mm² orifice [51]
Jo

Page | 16
 300
Pre.
290 Exp.

280
Temperature/K

270

260

250

240

230
0 20 40 60 80 100 120 140

of
Time/s

Figure 11: Calculated and measured temperature transients [51]

ro
2.8 H. Mahgerefteh
Most comprehensive blowdown model reported in literature was BLOWDOWN, later on
-p
extended as BLOWSIM as discussed earlier. Although validated extensively with the
re
experimental results, the prediction is still limited to depressurization under ambient

temperature. This is a serious drawback since the blowdown situations may involve a fire
lP

case most of the time. To overcome this issue in 2002, Mahgerefteh et al. described the

establishment of a numerical model for quantitative risk assessment (QRA) of rupture

na

following the depressurization of vessel under fire [52]. The model is able to predict the tri-

axial transient pressure and thermal stress produced in both unwetted and wetted wall
ur

sections of the vessel. The fluid/wall heat transfer was obtained using the following equation.

𝑄 = ℎ[𝑇𝑤′ (𝑎, 𝑡)]𝑛 [𝑇𝑤′ (𝑎, 𝑡) − 𝑇𝑍 ]𝐴Δ𝑡 (1)

Jo

Where ℎ is the heat transfer coefficient, 𝑇𝑤′ (𝑎, 𝑡) refers to the unknown wall temperature

at time 𝑡, 𝑇𝑍 is the liquid or vapor temperature, 𝐴 is the heat transfer area and 𝑛 stands for

power index. A hypothetical case study for the depressurization of a vessel containing higher-

HC was performed. The vessel was supposed to be at the initial pressure and temperature of

11.6 MPa and 293 K respectively. Two types of blowdown scenarios were simulated: (1) the

Page | 17
vessel under ambient conditions and (2) the vessel under fire radiating a heat flux of 90

kW/m². The predicted pressure and temperature profiles for a vessel under ambient and fire

conditions are illustrated in Figure 12. The predicted results show that thermal radiation

compensates the temperature drop to a large extent. A comprehensive analysis of pressure

stress and triaxial thermal profiles indicates that rupture occurs in dry section of vessel

because of the combined effect of resident total stress and thermally induced mechanical

weakening of vessel walls. The featured numerical model serves as a powerful tool for the

calculation of minimum required blowdown time to eradicate the risk of vessel puncture

of
under fire.

16 650

ro
Pressure (FC) FC = Fire Condition (Temperature)
14 Pressure (AC) AC = Ambient Condition (Temperature) 600
Vapor temperature (FC)
12 550
Vapor temperature (AC)
-p

Temperature/K
Pressure/MPa

Liquid temperature (FC) 500

10 Liquid temperature (AC)
450
8
re
400
6
350
4 300
lP

2 250
0 200
0 200 400 600 800 1000 1200 1400
na

Time/s
Figure 12: Temperature and pressure histories under ambient and fire conditions [52]
ur

2.9 SDM
Over the years, the researchers' primary focus has been on the performance of pressure
Jo

relief valve (PRV) using dynamic modeling approach. Unfortunately, most of the processes

were simulated for dynamic properties of PRV at the fixed inlet scenarios and not during the

reclosing process [53]. To predict the dynamic properties of PRV during reclosing process, a

simplified dynamic model (SDM) was presented by Song et al. [54]. When the disc moves,

PRV behaves as one degree-of-freedom system and motion of the moving part (stem and

Page | 18
disc) was resolved based on force balance when open. The motion of value was simulated

based on second-order differential equation deduced from Newton's second law (Eq. 2).

𝑚ẍ + 𝑐ẋ + 𝐹𝑠 = 𝐹𝑓 − 𝐹𝑔 + 𝑓𝑐 (2)

Where 𝑚, ẍ, and ẋ represent the mass of the moving parts, acceleration, and velocity of

the disc part in the moving direction respectively. 𝑐 represents the damping coefficient and

𝐹𝑠 stands for the spring force acting on the disc that was calculated using 𝑘(𝑥0 + 𝑥(𝑡)). 𝐹𝑓 ,

𝐹𝑔 , and 𝑓𝑐 are the flow force acting on the disc part, the gravity of disc, and the coulomb

friction parameters respectively. One degree-of-freedom system based on the equation of

of
motion was calculated using Eq. 3.

ro
𝑚ẍ/𝜓 + 𝑐ẋ/𝜓 + 𝑘(𝑥0 + 𝑥)/𝜓 = (𝜋/4). 𝐶𝑓 . 𝜙 (3)

Where 𝑘 is the spring stiffness. 𝜓 and 𝜙 are the reference force and pressure difference
-p
ratio. The dynamic numerical model was based on the major principles of the steady
re
computational fluid dynamics (CFD) analysis and rigid-body motion of the PRV. Afterwards,

the results from a case study revealed that SDM predicted the blowdown of conventional
lP

PRV reliably as shown in Figure 13. In addition, the effect of reclosing/opening process and

spring stiffness can also be easily studied using this dynamic numerical model.
na

5.25
4.75
4.25
ur

Test Signal Mean Value

3.75
3.25 Reclose
Lift/mm

2.75
Jo

2.25
Popping
1.75
1.25
0.75
0.25
-0.25
17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34
Time/s
Figure 13: Comparison of experimental results and SDM [54]

Page | 19
2.10 J. Zhang
To Simulate the blowdown process for the supercritical water-cooled reactor (SCWR), a

transient analysis code using a comprehensive numerical model was developed by Zhang et

al. in 2013 [55]. Three different phase separation models were implemented for the

calculation of stagnation enthalpy of two-phase fluid. Firstly, a homogenous mixture model

was used to calculate subcooled, supercritical, and overheated fluid by neglecting phase

separation effect. Secondly, assuming that bubble rising and blowdown periods are

comparable and ignoring neither discharge rate nor the bubble rising velocity, bubble rising

of
model describes this process properly. Thirdly, a complete separation model is considered

suitable for conditions when bubble rising velocity is faster as compared to the discharge

ro
speed and when the droplets falling velocity is higher than discharge speed for condensation

-p
because there is enough time for vapor dome development in both the cases.

Supercritical CO₂ and H₂O were blown down initially from 313 K temperature and 25
re
MPa pressure for experimental investigation and verification of the developed code. The
lP

experiments were performed on a cylindrical vessel with a 4 m height, 2 m diameter and a 17

mm² orifice for depressurization. The experimental study and predicted pressure drop results

with different models as shown in Figure 14. The R-square for pressure predictions is ≈ 0.377
na

by complete separation model. The consequence of the initial condition on pressure drop was

considered for different regions divided by the relationship between corresponding pseudo-
ur

critical and initial temperature. Furthermore, it was concluded that both void fraction and
Jo

blowdown speed increase with the decrease in initial pressure and an increase of initial

temperature, yet vice versa for the fluid inventory.

Page | 20
 20
Initial Conditions: Exp.
18
P₀ = 20 MPa Homogeneous model
16 Bubble rising model
T₀ = 313 K
14 A₀ = 17 mm² Complete separation model
Pressure/MPa

12
10
8
6
4
2
0
0 20 40 60 80 100 120 140

of
Time/s
Figure 14: Dynamic pressure comparison of different models [55]

ro
2.11 VBsim
To predict the depressurization behavior of vessel containing two-phase (V-L) HC
-p
mixture with non-ideal behavior, Alessandro et al. described the modeling and experimental

validation of the new unsteady model, VBsim (Vessel Blowdown Simulator) [56]. The
re
featured model accounts for non-equilibrium effects between constituent fluid phases such as
lP

temperature difference between phases. The numerical model evaluates dynamic fluid

temperature, pressure, and composition using EoS for non-ideal gas. SRK and PR EoS with
na

Van der Waals mixing rules have been implemented. Vapor phase discharge (𝜓𝐺,𝑜𝑢𝑡 ) was

estimated using equations of motion for compressible fluid through the orifice (Eq. 4 and 5).
ur

𝜓𝐺,𝑜𝑢𝑡 = 𝐶𝐷𝐺 𝜓𝑖𝑠 /𝑀𝑊

𝐺
(4)

𝑆𝐺,𝑜𝑟 √𝛾𝑝𝜌𝐺 (2/(𝛾 + 1))(𝛾+1)/2(𝛾−1) 𝑝/𝑝𝑒𝑥𝑡 ≥ (2/(𝛾 + 1))𝛾/(𝛾−1)

Jo

𝜓𝑖𝑠 = { 𝛾/(𝛾−1)
(5)
𝑆𝐺,𝑜𝑟 ɼ1/𝛾 √2𝛾/(𝛾 − 1)𝑝𝜌𝐺 (1 − ɼ(𝛾−1)/𝛾 ) 𝑝/𝑝𝑒𝑥𝑡 < (2/(𝛾 + 1))

𝐺
Where 𝑀𝑊 is the vapor phase molar weight and 𝐶𝐷𝐺 is fixed to a default value of 0.84. 𝜌𝐺

is the vapor density, 𝑆𝐺,𝑜𝑟 is the orifice area, 𝑝 is the vessel pressure, 𝑝𝑒𝑥𝑡 is the atmospheric

pressure, ɼ is the ratio of 𝑝 & 𝑝𝑒𝑥𝑡 , and 𝛾 is the isentropic coefficient of the real gas. The

Page | 21
equation for an incompressible fluid through an orifice was used for the calculation of liquid

phase discharge rate (Eqs. 6 and 7).

𝜓𝐿,𝑜𝑢𝑡 = 𝐶𝐷𝐿 𝜓 𝐿 /𝑀𝑊

𝐿
(6)

𝜓 𝐿 = 𝑆𝐿,𝑜𝑟 𝜌𝐿 √2((𝑝 − 𝑝𝑒𝑥𝑡 )/𝜌𝐿 + 𝘨𝐻𝐿 ) (7)

Where 𝑆𝐿,𝑜𝑟 , 𝜌𝐿 and 𝐻𝐿 are the orifice area, liquid density, and liquid static head

𝐿
respectively. The 𝑀𝑊 stands for liquid phase molar weight. The liquid discharge coefficient,

𝐶𝐷𝐿 , is fixed to a default value of 0.61. The liquid and the vapor daughter phase compositions

were calculated by 𝑥𝑖 and 𝑦𝑖 respectively.

of
𝑥𝑖 = 𝑧𝑖 /(1 − 𝛽 + 𝛽𝐾𝑖 ’) (8)

ro
𝑦𝑖 = 𝐾𝑖 𝑧𝑖 /(1 − 𝛽 + 𝛽𝐾𝑖 ) (9)

Where 𝐾𝑖 is the initial equilibrium coefficient and the overall vapor fraction, 𝛽, is
-p
calculated by solving the Rachford–Rice equation. For experimental validation of the
re
developed model, the predicted results were plotted against the experimental data from

Haque et al. (PR and SRK) and HSE (SRK) [26, 28, 57]. A comparison with the previously
lP

defined blowdown model (BLOWDOWN) prediction was also made as shown in Figure 15.

Considering the overall performance, VBsim produced reasonable results with R-square ≈
na

0.682 for temperature predictions and less CPU time requirements. However, the impact of

the model could be significantly improved if compared with further experimental data.
ur
Jo

Page | 22
 320
Vbsim
310
BLOWDOWN
300 Exp. UP
290 Exp. DOWN
Temperature/K

280
270
260
250
240
230
220
0 200 400 600 800 1000 1200
Time/s

of
Figure 15: VBsim and BLOWDOWN predictions vs experimental data (S9) [56]

ro
2.12 A. Park
Previously, Haque et al. and Mahgerefteh et al. developed models to study heat transfer

-p
between the mixture and the vessel walls. However, the correlation used and the value of the

heat transfer coefficient were unclear. Therefore, Park et al. (2018) presented an elaborated
re
heat transfer model including combined convection (forced and natural), multi-layer transient

conduction, and the nucleate boiling using PR EoS [58]. Mass and energy conservations for
lP

the fluid were calculated using equations presented by Speranza and Terenzi [59]. The model

consists of sub-algorithms of flash calculation, discharge rate calculation, and heat transfer
na

rate calculation in order to estimate the realistic expansion path. The predicted results were

compared with the measured and simulated data from literature including renowned
ur

simulation tools such as BLOWDOWN, BLOWSIM, VBsim, BLOW, and commercial tools
Jo

such as VessFire 1.2 and Aspen HYSYS v9. The model reasonably predicted the vapor

temperature within the experimental range while BLOWDOWN, VBsim, and HYSYS

predicted 2–10 K higher temperature. Similarly, the model also precisely predicted the wall

temperature within the experimental range while VessFire v1.2 and BLOW predicted almost

2–8 K higher wall temperature in contact with vapor, and VBsim predicted 10 K higher wall

temperature in contact with liquid as shown in Figure 16.

Page | 23
 300
Vapor
Inner Wall Temperature/K

290

280

270

Exp. Liquid
260
Park et al
BLOWDOWN
250 HYSYS v9®
BLOW
240
0 200 400 600 800 1000 1200

of
Time/s
Figure 16: Vessel wall temperature histories and comparison with other simulation tools and
numerical models [58]

ro
3 Design of Blowdown System for Pressure Pipelines
-p
The blowdown of a pipeline and a vessel differs from each other due to a considerable

pressure difference within the vessel but not within the pipeline. For blowdown of a vessel,
re
pressure drop is across the orifice, whereas for a pipeline blowdown, pressure drop is along
lP

the line and across the orifice if it is small enough. In case of blowdown of pipelines, the

hazard emerges not only following the low temperature generated in the pipeline walls but
na

also due to the high efflux rates and large total efflux that arise when a massive inventory in a

typical line is depressurized [22, 60]. To avoid accidents caused by these issues, several

numerical models have been developed. Later, some of them have also been coded as
ur

computerised program usually known as simulation tools. A timeline for the available
Jo

simulation tools and numerical models capable to simulate the depressurization process is

illustrated in Figure 18 followed by a brief description (Some of the authors didn't name their

simulation tools; therefore, we have mentioned the names of those authors instead of the

simulation tool).

Page | 24
 1995 1999 2011 2013 2014

Fairuzov Oke OLGA Martynov

META CNGS-MOC VPM Brown Drescher

1998 2003 2012 2013

Figure 17: Timeline for blowdown numerical models

of
3.1 META
Marginal stability two-fluid model (MSM) based on extended Geurst's variational

ro
principle to generalize multi-component two-phase dispersion was presented by Chen et al. in
-p
1995 [61-64]. Thermodynamic equilibrium assumptions were made to develop the equations

of motion and energy conversion for compressible single or multi-component vapor-liquid

re
(V-L) mixture. To make the mathematical simulation of physical process possible and

profound, the flow was considered to be marginally stable to achieve hyperbolicity of the
lP

equations.

In the later volume of publication, the simplified numerical method developed by Chen et
na

al. in 1993 [65] was modified and applied to the MSM for multi-component mixtures [66].

Subsequently, MSM and HEM were incorporated on the basis of simplified numerical
ur

method, and a computational package termed as META (Multi-component Equilibrium Two-

Jo

phase Analyser) was coded. A comparison of different case studies was made between HEM

and the MSM with experimental data for single or binary component two-phase blowdown. A

100 m long horizontal commercial steel pipeline with 0.15 m inner diameter (ID) containing

LPG (95 mole % propane and 5 mole % butane) was blown down. The testing temperature

and pressure were kept between 288–293 K (ambient temperature) and 800–2100 kPa

respectively. The bore size was varied from 0.05 m to full-bore in diameter and controlled

Page | 25
with the circular orifice plate. Figure 18 and 19 show the history of the pressure and

temperature variations with respect to time at the closed ends. An R-square ≈ 0.957 was

found in case of META-MSM CS for the temperature predictions.

1200
Exp.
1000 META-HEM
META-MSM CS
800
Pressure/kPa

600

400

of
200

ro
0
0 5 10 15 20 25
Time/s -p
Figure 18: Dynamic pressure changes' histories at the closed end [65]
re
300
Exp.
290 META-HEM
lP

280 META-MSM CS
Temperature/K

270
260
na

250
240
ur

230
220
0 5 10 15 20 25
Jo

Time/s
Figure 19: Dynamic temperature changes' histories at the closed end [65]

The comparison of HEM predictions and the measured data has been exceptionally

reasonable. All the predicted temperature, pressure, inventory, and void fraction transitions

correspond with the measured data. The only peculiarity was in case of the void fraction that

Page | 26
was over-predicted after 7 s at the closed end. The most significant findings are presented as

follows.

1. The simplified numerical method predicted reasonable results for blowdown through

experimental validations as shown in Figure 18 and 19.

2. In case of a full-bore short pipeline blowdown (< 10 m), both mechanical and thermal

non-equilibriums are significant. However, for a long pipeline blowdown, thermal

non-equilibrium is found to be insignificant.

3. Precise predictions of the phase equilibrium and thermodynamic behavior are

of
important for the transient flow calculations.

ro
3.2 Fairuzov
Previously, most of the work encompassed the modeling of depressurization from vessels
-p
or relatively short pipelines [67-69]. Therefore, a numerical simulation model for the
re
depressurization of relatively large pipeline transporting multi-component flashing mixtures

was coded by Fairuzov in 1998 [70]. The effect of heat transfer between fluid flow and
lP

pipeline wall during blowdown phenomenon was also described. Heat transfer model, the

hydrodynamic model, and break-flow model were the building blocks of the featured model.
na

No experimental study was conducted, however, the simulation tool was validated against the

already available experimental data from BP and Shell on the Isle of Grain [30]. The
ur

calculations were carried out on an HP Apollo Series 700 working station using a double-

precision FORTRAN 77 code.

Jo

Fairuzov proposed a higher amount of heat transfer amongst fluid and pipe walls during

the depressurization process, hence, the adiabatic assumption for simulation became invalid.

Therefore, effect of thermal capacitance was integrated into the model using a different

method in the development of energy conversion equation. The study unveiled considerable

effect of thermal capacitance of pipeline on the behavior of two-phase fluid flow.

Page | 27
Consequently, thermal capacitance should not be ignored especially in case of the flashing

fluids. The dynamic pressure and temperature profiles are reported in Figure 20 and 21

respectively. Figure 20 illustrates that the model slightly over-predicted depressurization rate

at the closed end of pipe. However, when the fluid flow was considered to be adiabatic, the

temperature at the open end was significantly predicted with R-square ≈ 0.914.

1200
Closed end (Exp.)
Closed end (Pre.)
1000
Open end (Exp.)
Open end (Pre.)
800

of
Pressure/kPa

600

ro
400

200 -p
0
0 5 10 15 20 25
re
Time/s
Figure 20: Dynamic pressure transitions at the open and closed ends of the line [70]
lP

300
290
na

280
270
Temperature/K

260
ur

250
Closed end (Exp.)
240
Closed end (Pre.)
Jo

230 Open end (Exp.)

Open end (Pre.)
220
0 5 10 15 20 25
Time/s
Figure 21: Dynamic temperature transitions at the open and closed ends of the line [70]

Page | 28
3.3 CNGS-MOC
An efficient mathematical model (CNGS-MOC) for the simulation of full-bore rupture

(FBR), based on the method of characteristics (MOC) was developed by Mahgerefteh et al. in

1999 [71]. The MOC was implemented to simulate FBR or depressurization of a long pipe

containing two-phase or condensable HC mixtures. This method was used to develop a more

suitable method than the Finite Difference Method (FDM) and Finite Element Method (FEM)

because both had troubles in handling the choking scenarios at the ruptured end. Moreover,

the MOC is more precise than FDM and FEM because it can resolve the blocked flow

of
intrinsically using Mach line characteristics. The MOC significantly reduces number of

iterations involved while solving the system of simultaneous equations. The use of curved

ro
characteristics in conjunction with fast-mathematical algorithm and compound nested grid

-p
system leads to a dramatic reduction in CPU time in addition to improved accuracy. The

model was extensively validated using field data obtained during Isle of Grain [29] as well as
re
Piper Alpha tragedy [72]. The simulation studies were performed on the basis of HEM

assumptions. In Figure 22, the predicted data with CNGS-MOC and field data are correlated
lP

with those based on other simulation tools including BLOWDOWN [25], META-HEM [64,

66, 72], MSM-CS [64, 66, 72] as well as PLAC [73]. Both META-HEM and CNGS-MOC
na

produced reasonably precise predictions while other models' predictions were relatively poor.

The CNGS-MOC predicted the physical process with R-square ≈ 0.850 for pressure
ur

predictions.
Jo

Page | 29
 1200
Exp.
CNGS-MOC
1000
META-HEM
MSM-CS
800
Pressure/kPa

BLOWDOWN
PLAC
600

400

200

0
0 5 10 15 20 25

of
Time/s
Figure 22: Dynamic pressure histories at the open end for the P42 test LPG [71]

ro
3.4 A. Oke
Regardless of the fact that most common pipeline tragedies include rupture, majority of
-p
the reported models so far are based on 1-D axial flow. Others just treat the pipeline as a

vessel discharging through an orifice. Therefore, Oke et al. developed a model that can
re
predict the transient release rate through the punctured plane based on MOC [74]. The model
lP

accounts for axial and radial flow and real fluid behavior as well as the rupture locality with

respect to pipeline length. The PR EoS was employed to get suitable phase and
na

thermodynamic equilibrium data that is particularly appropriate for high-pressure HC

mixtures [75, 76].

ur

In case of 1-D unsteady flow, assuming phase and thermodynamic equilibrium among

phases, the energy, momentum, and continuity conservation equations for a fluid in a rigidly
Jo

clamped pipeline were taken from literature [77]. The continuity equation was reformulated

as follows.

[𝜌𝑇 + 𝜑]d𝑃/d𝑡 − 𝜌𝜑 ∗ dℎ/d𝑡 + 𝜌2 𝑎2 𝑇 ∗ 𝜕𝑢/𝜕𝑥 = 0 (10)

Where ℎ and 𝜑 represent enthalpy and thermodynamic function respectively. 𝜌 and 𝑢 are

density and velocity as a function of time (𝑡) and distance (𝑥) respectively. Using MOC, the

Page | 30
conservation of continuity, energy, and momentum equations can be substituted with three

compatibility equations along with their corresponding characteristic lines. The isolated flow

was modeled based on termination of the inlet feed upon rupture. In case of non-isolated

flow, inlet feed is supposed to be stopped after a prescribed period of time. Figure 23 is

schematic illustration of the fluid flow analysis following rupture and emergency isolation

with various boundary conditions from B₁-B₆.

Valve End

of
ro
Pump End
Control
Volume

-p
Figure 23: Rupture plane and fluid flow study following line puncture [74]

In Figure 23, C stands for match line, C0 stands for path line, and J represents solution
re
joints. The simulation of pressure and velocity profiles specifies three separate fluid flow

regimes predominant within the pipe that manage the dynamic transitions of the release rate.
lP

For validation of the numerical model, the simulated outcomes were plotted against the data

from experiments performed by BP and Shell on the Isle of Grain [30]. The results ended up
na

with a reasonable prediction of the physical process with R-square ≈ 0.942 for temperature at

the closed end as shown in Figure 25.

ur
Jo

Page | 31
 1200
Exp. (closed end)
1000 Pre. (closed end)
Exp. (opened end)
800
Pressure/kPa

Pre. (opened end)

600

400

200

0
0 5 10 15 20 25

of
Time/s
Figure 24: Pressure transitions at closed and opened end of P40 (LPG) test [74]

ro
300
290
280
-p
Temperature/K

270
re
260
250 Exp. (closed end)
lP

240 Pre. (closed end)

Exp. (opened end)
230
Pre. (opened end)
na

220
0 5 10 15 20 25
Time/s
Figure 25: Temperature transitions at closed and opened end of P40 (LPG) test [74]
ur

3.5 VPM
Until 2011, no numerical model was developed to analyze the blowdown conditions for a
Jo

vent pipeline. Therefore, Rajiwate developed a vent pipe model (VPM) to investigate the

behavior of compressible gas in a vent pipeline using thermodynamic and fluid dynamic

approach [78]. Integrated with the GERG EoS, the use of REFPROP makes a highly adequate

prediction with the VPM [79, 80].

Page | 32
 Temperature, pressure, and density relations in terms of stagnation properties were

defined previously [81-83]. Mass efflux prediction for pipe is the central step for the

development of blowdown model for vent system in terms of static pressure and static

temperature that is expressed as:

𝐺 = 𝑃𝑀√𝛾/𝑍𝑅𝑇 (11)

The above equation for mass flux in terms of stagnation properties was represented as:

𝐺 = 𝑃𝑜 /√𝑇𝑜 ∗ √𝛾/𝑍𝑅𝑀/(1 + (𝛾 − 1)/2 ∗ 𝑀2 )(𝛾+1)/2(𝛾−1) (12)

Where 𝑇𝑜 , 𝑃𝑜 are the stagnation temperature and pressure respectively, 𝛾 is the specific

of
heat ratio, and 𝑍, 𝑅 and 𝑀 are the compressibility factor, universal gas constant, and match

ro
number respectively. For a given downstream and stagnation pressure, geometry, and

assumed friction factor, Eq. 11 and 12 define the flow [84].

-p
The predicted results were then validated with the experimental study performed on a 12
re
m long stainless-steel pipeline. A total of 9 experimental studies were conducted with an

initial pressure and temperature of 200-400 kPa and ≈ 19 °C respectively. The experimental
lP

data were plotted only for a steady-state period that showed a good agreement with the

predicted values.
na

3.6 OLGA
Most of the reported data on experimental study of CO₂ blowdown [49, 85-88] describes
ur

findings from small-scale laboratory setups of flow loops and vessels. Thus, it is of great

interest for carbon capture and sequestration (CCS) community to collect field data from
Jo

existing large-scale CO₂ facilities. Therefore, a large-scale CO₂ pipeline blowdown was

carried out by Clausen et al. in 2012 while pointing the possible limitations of existing

dynamic multi-phase flow simulator (OLGA) [89]. An onshore buried 50 km long 24″ ID

pipeline containing 93000 tonnes of CO₂ (>99 % Pure) was initially depressurized from

supercritical conditions. The initially reported temperature and pressure were 304 K and 8.1

Page | 33
MPa respectively. The blowdown process was modeled using OLGA V5.3.2, a multi-phase

thermo-hydraulic simulation tool developed by SPT Group [90]. The new coded module of

OLGA for CO₂ single component was used that utilizes Span-Wagner EoS for pure CO₂ [91,

92].

The total blowdown time was noted to be nearly 10.50 h while OLGA predicted duration

of 10 h and 25 min. Both simulation and experimental data showed that the blowdown stayed

well above the CO₂ triple point, and no traces of solidification were found as shown in Figure

26. However, the minimum simulated temperature was almost 17 K lower than the minimum

of
measured temperature. The deviation between simulated and measured blowdown path

ro
reveals inability of the simulation tool to handle CO₂ along with impurities. Even a very

small amount of impurities will disturb the blowdown path.

9
8
CO₂ VLE
Exp.
-p
re
7 Pre.
Phase Envelop
6
lP
Pressure/MPa

5
4
3
na

2
1
0
250 260 270 280 290 300 310 320
ur

Temperature/K
Figure 26: Measured upstream pressure and temperature [89]
Jo

To further study the effect of impurities on CO₂ blowdown and reliability of the OLGA

tool, Huh et al. conducted experimentation in 2014 with 2–8 % N₂ addition by mass fraction

[93]. The experimental apparatus comprises of the pressurization part, liquefying part, mixing

zone, and the test section. Total length of the testing apparatus was 51.96 m with 3.86 mm

ID. The temperature and pressure measurements were observed at four different positions as

Page | 34
illustrated in Figure 27 and 28. However, the simulation tool did not predict temperature drop

precisely. The results show that the numerical simulator predicted better pressure drop at

higher concentration of impurities. Furthermore, the R-square value at initial portion (7.93 m)

of pipeline is ≈ 0.697. While, at last portion (51.85 m), the temperature predictions are much

precise with R-square ≈ 0.859. It can be inferred that the numerical simulator requires further

improvements in phase change model.

10
7.93 m
9
23.13 m
8

of
38.32 m
7 51.85 m
Pressure/MPa

6 7.93 m

ro
5 23.13 m
4 38.32 m
51.85 m
3
2
1
-p
re
0
-5 0 5 10 15 20 25
Time/s
lP

Figure 27: Pressure transition of simulated and measured data for 92 % CO₂–8 % N₂
blowdown [93]

296
na

294
292
ur

290 7.93 m
Temperature/K

23.13 m
288 38.32 m
Jo

286 51.85 m
7.93 m
284 23.13 m
282 38.32 m
51.85 m
280
-10 -5 0 5 10 15 20 25 30 35 40
Time/s

Figure 28: Temperature changes of simulated and measured data for 92 % CO₂–8 % N₂
blowdown [93]

Page | 35
3.7 S. Brown
The commonly reported vessel/pipeline outflow numerical models were based on HEM

assumptions where the fluid phase was considered to be at the mechanical and thermal

equilibrium. Consequently, there was no-equilibrium V-L transition and phase-slip

phenomena, that is, delayed bubble formation was ignored. In 2013, Brown et al. extended

the prior work by establishing and validating homogeneous relaxation model (HRM) for

depressurization of dense phase CO₂ pipeline [94]. Initially, the phases were assumed to be at

mechanical and thermal equilibrium, hence, the mass, momentum, and energy conservation

of
were presented by Eq. 13, 14, and 15 respectively.

𝜕𝜌/𝜕𝑡 + 𝜕𝜌𝑢/𝜕𝑧 = 0 (Conservation of Mass) (13)

ro
𝜕𝜌𝑢/𝜕𝑡 + (𝜕𝜌𝑢2 + 𝑃)/𝜕𝑧 = −𝑓𝑤 𝜌𝑢2 /𝐷𝑝 (Momentum Balance) (14)

𝜕𝐸/𝜕𝑡 + 𝜕𝑢(𝐸 + 𝑃)/𝜕𝑧 = −𝑢(𝑓𝑤 𝜌𝑢2 /𝐷𝑝 ) -p (Energy Balance) (15)

Where 𝜌, 𝐷𝑝 , 𝑃 , 𝑢, and 𝑓𝑤 are the mixture density, pipeline diameter, pressure, velocity,
re
and fanning friction factor respectively. All of these factors were evaluated using Chen's
lP

correlation (1979) [95] as function of space (𝑧) and time (𝑡). 𝐸 stands for the energy of the

mixture. For the non-equilibrium V-L transition, HRM was used with the mechanical

equilibrium assumption such that no phase slip was maintained. The non-equilibrium V-L
na

transition equation with thermodynamic equilibrium for the vapor mass fraction is presented

in Eq. 16.
ur

𝜕𝑥/𝜕𝑡 + 𝑢(𝜕𝑥/𝜕𝑧) = (𝑥𝑒𝑞 − 𝑥)/𝜃 (16)

Jo

Where 𝑥 represents the dynamic vapor quality and 𝜃 stands for relaxation time accounting

for the phase change transition delay. Angielczyk et al. proposed an empirically determined

correlation for the relaxation time as given in Eq. 17 [96].

𝜃 = 2.15 × 10−7 (𝑥(𝜌/𝜌𝑠𝑣 )−0.54 ((𝑃𝑠 (𝑇𝑖𝑛 ) − 𝑃)/(𝑃𝑐 − 𝑃𝑠 (𝑇𝑖𝑛 )))−1.76 (17)

Page | 36
 Where 𝑃𝑐 , 𝜌𝑠𝑣 and 𝑃𝑠 are critical pressure, saturated vapour density at given pressure, and

saturation pressure at given temperature respectively. To validate numerical flow model's

reliability, the hypothetical shock tube test was conducted initially for both single (Liquid)

and two-phase CO₂. Table I presents the initial conditions and pipeline characteristics for the

COOLTRANS FBR dense-phase CO₂ shock tube test. Moreover, CO₂PipeHaz CO₂ pipeline

rupture experiment conditions were used as an experimental study [97].

Table I: Predominant conditions and pipeline characteristics for the COOLTRANS shock
tube test

of
Pipeline characteristics Initial conditions
Pipeline length 144 m Fluid temperature 278.35 K

ro
Internal diameter 150 mm Fluid pressure 15.335 MPa

Pipeline wall roughness 0.05 mm Ambient pressure 0.101 MPa

-p
A 37 m long with 40 mm ID instrumented pipeline carrying pure CO₂ was blown out
re
using an explosive charge. Pipeline's initial conditions were 7 MPa pressure and 298.35 K

temperature. In general, HRM produced plausibly good estimations of the experimental data
lP

especially during the initial period of blowdown as shown in Figure 29. R-square for pressure

predictions is ≈ 0.924 that indicates reliability of the model. It was observed that FBR clearly
na

indicates fully dispersed flow and, hence, the phase-slip can't be ignored.
ur
Jo

Page | 37
 8
Exp.
7
HRM
6
Pressure/MPa

5
4
3
2
1
0
0 0.5 1 1.5 2 2.5 3

of
Time/s
Figure 29: Pressure-time profile at 0.1 m from release end of the CO₂PipeHaz

ro
FBR test [94]

3.8 S. Martynov
-p
Since most of the prior reported blowdown simulation models have been developed for

V-L phase depressurization, they were not appropriate for simulating solid CO₂ release.
re
Martynov et al. in 2013 extended HEM for the development of outflow model accounting for
lP

CO₂ solidification during blowdown [98]. A set of equations describing the HEM flow

including mass conversion, and momentum and energy balance equations were used. For

mass conservation, Eq. 13 was used. Eq. 14 and 15 were modified to Eq. 18 and 19 to
na

calculate the momentum and energy balance respectively.

ur

𝜕𝜌𝑢/𝜕𝑡 + (𝜕𝜌𝑢2 + 𝑃)/𝜕𝑧 = −2𝑓𝑤 𝜌𝑢2 /𝐷 (Momentum Balance) (18)

𝜕𝐸/𝜕𝑡 + 𝜕𝑢(𝐸 + 𝑝)/𝜕𝑧 = 4𝑞𝑤 /𝐷 − 2𝑓𝑤 𝜌𝑢3 /𝐷 (Energy Balance) (19)

Jo

To calculate the properties of vapor-solid (V-S) and vapor-liquid (V-L) equilibrium

mixtures formed during the depressurization of vessel/pipeline containing CO₂, previously

developed PR EoS by Martynov et al. [99] was extended and applied.

𝑝 = 𝑅𝑇/(𝑣 − 𝑏) − 𝑎/(𝑣(𝑣 + 𝑏) + (𝑏(𝑣 − 𝑏)) (20)

Page | 38
 Where 𝑅, 𝑇, 𝑝, and 𝑣 are the universal gas constant, temperature, pressure, and specific

volume respectively. 𝑎 and 𝑏 are the empirical parameters accounting for the intermolecular

attraction forces and molecular volume respectively. The instantaneous heat flux at the pipe

wall was calculated based on pipeline wall temperature transitions with time using Eq. 21.

𝑎𝑣𝑒𝑟𝑎𝑔𝑒 𝑎𝑣𝑒𝑟𝑎𝑔𝑒
𝑞𝑤 = 𝜌𝑤 𝐶𝑝,𝑤 𝛿𝑤 ∗ ((𝑑𝑇𝑤 )/𝑑𝑡) (21)
𝑎𝑣𝑒𝑟𝑎𝑔𝑒
Where 𝑇𝑤 represents the average pipeline wall temperature. 𝜌𝑤 , 𝐶𝑝,𝑤 , and 𝛿𝑤 stand

for density, pipeline heat capacity, and wall thickness of pipeline respectively.

of
The featured numerical model was validated against experiments conducted by Dalian

University of Technology (DUT) under CO₂PipeHaz project. The model precisely predicts

ro
the pressure transitions in the pipeline for initially supercritical fluid (Test 1 and 2). The

-p
initial experimental conditions for all the case studies are presented in Table II.

Table II: Initial experimental conditions of the release

re
Initial fluid condition Test 1 Test 2 Test 3
Pressure/MPa 8.0 8.6 5.3
lP

Temperature/K 307 312 290

Inventory/tonne 6.5 4 6.5

na

In case of an initially two-phase fluid (Test 3), the given HEM model is not much reliable

because of R-square ≈ 0.661 for pressure predictions as shown in Figure 30. Moreover, the
ur

release model also predicted the dry ice formation and explained the transition phenomena

from V-L to V-S equilibrium at a triple point of CO₂.

Jo

Page | 39
 6
Pre. (heating)

5 Pre. (adiabatic)
Exp.
4
Pressure/MPa

0
0 50 100 150 200 250 300 350

of
Time/s
Figure 30: Time variation of the pressure measured in test 3 [98]

ro
3.9 M. Drescher
HEM was validated mostly against the experimental data from CO₂ pressure release.
-p
However, Dresher et al. (2014) executed experiments with different CO₂–N₂ compositions to

study the potency of the modified HEM for a multi-component system [100]. For simulation
re
study, PR EoS was used and experiments were performed on a 140 m long horizontal tube
lP

with 10 mm ID. Initially, the binary mixture in the supercritical region was at 12 MPa

pressure and a temperature of approximately 293 K. The numerical results for all three fluid
na

compositions regarding the relative and absolute value match the experimental results very

well as illustrated in Figure 31 and 32. With the increase in composition, the model

predictions become more precise as maximum R-square ≈ 0.919 for temperature prediction is
ur

noted in case of 90 % composition. Despite few complexities in the HEM employed, it still
Jo

contains several simplifications in view of the comparatively reasonable results. Further work

could involve the advanced heat-transfer models considering the effect of flow regimes or

boiling.

Page | 40
 300
70 % HEM
295 70 % Exp.
290 80 % HEM
80 % Exp.
285 90 % HEM
Temperature/K

280 90 % Exp.

275
270
265
260
255
0 5 10 15 20 25 30

of
Time/s
Figure 31: Comparison of numerical and experimental (PT-40, TT-40) dynamic

ro
temperature changes for different compositions at 50 m [100]

12

10
-p 70 % HEM
70 % Exp.
80 % HEM
re
8 80 % Exp.
Pressure/MPa

90 % HEM
6 90 % Exp.
lP

2
na

0
0 5 10 15 20 25 30
Time/s
ur

Figure 32: Comparison of numerical and experimental (PT-40, TT-40) dynamic pressure
changes for different compositions at 50 m [100]
Jo

4 Major Findings

Table III presents a matrix of modeling outlooks presented in literature that could aid in a

holistic simulation of blowdown use in the industrial sector. Based on the literature reviewed,

Table III illustrates the followings:

Page | 41
 Commonly, the models focused on pressure and mixture temperature transitions with

time to understand the blowdown process.

Limited studies on vessel wall temperature and void fraction are also available,

however, further research is required when applied to multi-phase processes.

HEM is extensively used to solve the blowdown process with thermodynamic

assumptions.

Mostly, the models were validated against previously published experimental data,

especially for pure or impure CO₂.

of
Table III: Summary of modeling approaches

ro
Title Pressure Temperature Flowrate Void
Mixture Wall Fraction
BLOWDOWN ✓ ✓ ✓ ✓ ✓
BLOWSIM
P. S. Cumber
✓
✓
✓
✓
-p
✓
✗
✓
✓
✗
✓
A. Fredenhagen ✓ ✓ ✗ ✓ ✓
re
SDM ✓ ✗ ✗ ✗ ✗
J. Zhang ✓ ✓ ✗ ✓ ✓
VBsim ✓ ✓ ✓ ✓ ✗
lP

A. Park ✓ ✓ ✓ ✓ ✗
META ✓ ✓ ✗ ✓ ✓
Fairuzov ✓ ✓ ✗ ✓ ✓
CNGS-MOC ✓ ✓ ✗ ✓ ✗
na

A. Oke ✓ ✓ ✓ ✓ ✗
OLGA ✓ ✓ ✗ ✓ ✗
S. Brown ✓ ✓ ✓ ✓ ✗
ur

S. Martynov ✓ ✓ ✓ ✓ ✗
M. Drescher ✓ ✓ ✗ ✓ ✗
Jo

Some of the numerical models or simulation tools produced incredibly precise predictions

of the physical phenomenon occurring during the blowdown process. The BLOWSIM

produced most precise results with R-square ≈ 0.9925 followed by BLOWDOWN with R-

square ≈ 0.977. BLOWSIM, META, OLGA etc. also produced the accurate results before the

process, however, they lack the ability to simulate some of the significant parameters. On the

Page | 42
other hand, BLOWDOWN has the ability to predict most of the important parameters

accurately. Cumber and J. Zhang produced poor results with R-square lower than 0.4. A bar

chart (Figure 33) illustrates R-square value for most of the reviewed simulation tools or

numerical models. For the quantitative comparison of reviewed models, the geometric mean

bias is taken into account (Figure 34). A perfect model is said to have a geometric mean bias

equal to 1.0. A geometric mean bias less than 1.0 refers to underprediction and above 1.0

refers to overprediction [101].

of
0.8

ro
0.6
R-square

0.4
-p
A. Fredenhagen
BLOWDOWN

P. S. Cumber

CNGS-MOC

S. Martynov

M. Drescher
BLOWSIM

re
S. Brown
Fairuzov
J. Zhang

0.2
A. Park

A. Oke
VBsim

META

OLGA
lP

Figure 33: Comparative illustration of R2 values for different models

na
ur
Jo

Page | 43
 1.8
1.6
Brown
Zhang
1.4
BLOW CNGS
1.2 DOWN META
MOC OLGA
Cumber
1
BLOW Freden Park Fairuzov A. oke
0.8 SIM hagen
0.6
0.4 Martynov

0.2 VBsim Drescher

of
Figure 34: MG bias values for different models

ro
5 Summary

Different simulation tools and numerical models for the calculation of optimum
-p
parameters during blowdown process are reviewed. Table IV summarizes the operating
re
conditions, vessel dimensions, characteristics, and key challenges. Several novel emerging

blowdown models have been reviewed and a comparative study has been presented.
lP
na
ur
Jo

Page | 44
 Table IV: Comparative analysis of different simulation tools and numerical models.

f
oo
Name/Title Year Composition Condition Vessel/Pipeline Orifice Characteristics Key Challenges Ref
Volume Size
(mm)
Models + Tools
N₂, 70 % N₂ – - -

pr
BLOWDOWN 1990 150 bar & 1.5 m long vessel 6.325 Good understanding of the physical No work in cryogenic conditions [25,
(1990) 30 % CO₂, 20 °C with 10.6″ processes occurring or emergency 26, 28]
Model + Tool NG/C₃ & HC diameter - Ability to make predictions with an - Use of the extended principle of
acceptable uncertainty corresponding states for

e-
- Based on a three fluids model (vapor, generating the pertinent fluid
liquid and free water) thermophysical properties
- A simulation study is restricted
to depressurization under

Pr
ambient conditions only
- Not fully documented
1991 Case 1: C₁–C₆ 117 bar & 40 km long 0.4191 - Good understanding of the physical - Homogeneity assumption when [27]
+ 4.03 % N₂ 10 °C pipeline processes occurring the flow is two-phase
- Ability to make predictions with an - Quasi-steadiness assumption
acceptable uncertainty - No work on cryogenic conditions
Case 2: C₁– l
200 bar & 20000 m long 0.4064 - Suitable for planned blowdown
na
C₅, C₁₀, C₂₀ + 40 °C pipeline
3 % H₂O
1996 LPG (95 % C₃ 11.3-22.5 100 m long 35-154 - Predictions have been shown to be in - No work on Natural gas [29]
– 5 % C₄) bar & pipeline with 2-6″ at least adequate, and often good, - No work in cryogenic conditions
ur

13.3-20 °C diameter agreement with the Isle of Grain or emergency situation

measurements
META 1995 LPG (95 % 8 to 21 bar 100 m in length, 0.05 m - Well prediction of experimental data - More efficient thermodynamic [64,
(1994) propane & 5 & 15 to 20 0.15 m measurement methods needed 66]
Jo

Model + Tool % butane °C

FRICRUP 1993 Air, CO₂, 69-138 bar 25.5 km long 1.5 to - Predict P, T, time and mass flowrate. - Did not predict very well for [33,
(1993-94) 1994 carbonated & 37.8 °C pipeline with 12.7 for a - For single and multi-phase multi-phase flow 34]
Model + Tool water, CH₄ (Vessel) 7.992″ diameter bottle - For multi-component - Can predict Vessel/Pipeline
and HHC and 40 bar & 84 for Conditions
& 15.5 °C Pipeline - No work on Natural gas
(pipeline).

Page | 45
Fairuzov 1998 LPG (95 % 11.25 bar 100 m long Full-bore - Multi-component, two-phase pipe - Thermal equilibrium between the [70]

f
(Model + Tool) propane & 5 & 19.9 °C pipeline with 150 rupture flow model fluid and pipe wall was not

oo
% butane mm internal - Effect of heat transfer is discussed reached
diameter - For non-adiabatic fluid flow - Rigorous conjugated heat
- Well prediction of experimental data transfer model should be used for
transient heat transfer description
BLOWSIM 1999 64 % C₁, 6 % 116 atm & N/A 10 mm - Multi-component, the multi-phase - Limited data on experimental [14]

pr
(Model + Tool) C₂, 28 % n-C₃ 293 K pipe flow model validation
& 2 % n-C₄ - BLOWSIM performs better than - No work on Natural gas
BLOWDOWN - Thermal and mechanical
- The cubic equation of state has equilibrium assumption [94]

e-
positive effects on data - No work in cryogenic conditions
or emergency situation
CNGS-MOC 1999 LPG (95 % 11.25 bar 100 m long FBR - Reasonably accurate predictions - No Experimental Work [71]
(Model + Tool) propane & 5 & 19.9 °C pipeline with 150 - Less CPU time - Less data on experimental

Pr
% butane mm internal - improving accuracy validation
diameter
P. S. Cumber 2001 LPG (C₁ 66.5 120 bar & 3.24 m height 10 mm - Precise prediction of pressure and - No experimental study conducted [45]
(Model + Tool) %, C₂ 3.5 % 25 °C with 0.54 m ID mass flow rate - Imprecise temperature prediction
& C₃ 30 % - Less computational time - No heat transfer through vessel
wall assumption
A. 2001 N₂ – CO₂ l
25-15 bar 0.05 m³ vessel 29-17 - A good prediction of the process - Tool commercially not available [51]
na
Fredenhagen & 298 K with 0.242 m mm² - Applicable for Multi-component - Unable to predict equipment
(Model + Tool) diameter process conditions
- Limited data on experimental
validation
- No data on multi-phase process
ur

H. 2002 64 % C₁, 6 % 116 bar & 3.24 m vessel 10 mm - Prediction of thermodynamic - No experimentation validation [52]
Mahgerefteh C₂, 28 % n–C₃ 293 K with 1.13 m properties under fire and ambient - No simulation tools developed
(Model) & 2 % n–C₄ diameter and 59 conditions
mm wall - For multi-component and multi-phase
Jo

thickness process
- Vessel conditions calculations
A. Oke 2004 LPG 21.6 bar & 100 m long with 150 - The model accounts for axial and - No experimental study
(Model) 20 °C 0.154 m ID radial flow, real fluid behavior as well performed
as the locality of puncture with respect - No validation for natural gas
to the length of pipe mixture and cryogenic conditions

Page | 46
VPM 2011 78.12 % N₂, 200-400 120 m long N/A - For multi-component process - It can only predict steady state [78]

f
(Model) 20.96 % O₂ & kPa and ≈ pipeline - Ability to calculate pipe wall conditions

oo
0.92 % Ar 19 °C temperature - It is not suitable for the multi-
phase process
OLGA 2012 CO₂ 81 bar & 50 km long, 24″ 8″ - Experimental data from large-scale - Non-reliable predications of [89]
(1991) 31 °C pipeline CO₂ unit experimental data
Tool - Initially supercritical conditions - No work on multi-component &

pr
- Multi-phase process multi-phase process
2014 CO₂, 2–8 % 85b bar & 51.96 m tube with N/A - Studied Effect of impurities on CO₂ - No study on the effect of initial [93]
N₂ 20 °C 3.86 mm ID blowdown conditions
- Good predictions of initial pressure

e-
drop
J. Zhang 2013 CO₂ & Water 25 MPa & 4 m height & 2 m 0.008 m² - A good prediction of the process - Limited data on experimental [55]
380 °C diameter - Supercritical conditions validation
- Multi-component and multi-phase - No work on natural gas

Pr
process - Only work on supercritical
conditions
S. Brown 2013 CO₂ 15.335 144 m long FBR - Model predictions produced a - Lack of sufficient and reliable [94]
(Model) MPa & pipeline with 150 reasonable agreement experimental data
278.35 K mm diameter - Multi-phase process - No work on the multi-component
- Delayed phase transition has an process
l
7 MPa & 37 m long insignificant effect on the - No work on natural gas and
na
298.35 K pipeline with 40 depressurization rate of pipe cryogenic conditions or
mm ID emergency situation
S. Martynov 2013 CO₂ 53-86 bar 256 m long 50 mm - Good prediction for initially - Unable to predict initially two- [98]
(Model) & 34-39 pipeline with 233 supercritical fluid phase fluid
°C mm diameter - Prediction of dry ice formation - No work on cryogenic or
ur

emergency conditions
- Experimental study performed
only for single component
M. Drescher 2014 CO₂ + 10–30 120 bar & 140 m long 9.5 mm - Good predication of experimental - Require advancement of heat- [100]
Jo

% N₂ 20 pipeline with 10 results transfer models

mm ID - Experimental validation of HEM with - Temperature prediction was less
binary components accurate
VBsim 2015 N₂ CO₂–HC 290-323 K 8.429 m length 0.635-1.4 - Good predication of experimental - No experimental study [56]
(up to C₁₀) & & 118.5- with 1.97 m cm results performed.
HC (up to C₃) 22 bar diameter - For multi-component and multi-phase - No study in the cryogenic region.
process

Page | 47
 - Consider non-equilibrium effects

f
between constituent fluid phases

oo
A. Park 2017 HC Mixture 293-303 K 3.24m length with 10 mm - Two-phase multi-component system - No experimental work performed [58]
& 117.5- 1.13m ID prediction. No study in the cryogenic region.
120 bar - Can predict wall temperature
contracting liquid and vapor.
- More precise predictions than

pr
previously available models

e-
l Pr
na
ur
Jo

Page | 48
6 Conclusion

In this paper, the available models for the optimum blowdown parameters' calculation are

discussed. The challenges for depressurization in process industry through a review of

existing literature are also highlighted. The available models are capable of solving the

blowdown scenarios, however, there are still some limitations listed as follows.

Only a few lab-scale experimental studies are reported for the validation of

models. The reliability of models will be improved if also validated against the

of
large scale experimental facilities.

More extensive experimental studies on natural gas and other gaseous mixtures

ro
could help to improve the validity of models.

-p
Limited studies are reported for the blowdown of pressure vessels under fire

conditions.
re
Experimental studies for blowdown from the cryogenic conditions could also be

considered. It would help to study the blowdown process under worst case
lP

scenarios especially when a phase change is involved that can lead to the

formation of solids.
na

A new numerical model can also be developed to overcome the problems related

to dry ice formation during the blowdown process.

ur

Studies on blowdown of vessels/pipelines through multiple chokes needs to be

Jo

considered.

To avoid solidification during blowdown, the effect of additives can be studied.

In addressing these research areas, industry should be empowered to make effective and

sustainable retrofit decisions while simultaneously reducing hazards and improving the

safety. This paper has focused on solutions for industrial hazards related to depressurization.

Page | 49
However, the outcome of the identified research areas has wide-ranging applications with

techniques that could be applied to solve many other system simulation challenges.

Acknowledgement

of
The authors would like to extend their most profound gratitude to the CO₂ Research Centre

(CO2RES), Universiti Teknologi PETRONAS (UTP), Malaysia, for the accomplishment of

ro
this Review article.

-p
re
lP
na
ur
Jo

Page | 50
 References

[1] Barma M, Saidur R, Rahman S, Allouhi A, Akash B, Sait SM. A review on boilers energy

use, energy savings, and emissions reductions. Renew Sustain Energy Rev. 2017;79:970-83.

[2] Ladokun T, Nabhani F, Zarei S. Accidents in pressure vessels: Hazard awareness.

International Association of Engineers; 2010.

[3] Cullen W. The Public Inquiry into the Piper Alpha Disaster.(Department of Energy,

HMSO, London, UK). 1990.

of
[4] Bond J. Institute of Chemical Engineers accidents database. Rugby. UK: Institute of

ro
Chemical Engineers. 2002.

[5] Fletcher S. US Senate ready to act on pipeline safety as public attention sharpens. Oil Gas
-p
J. 2001;99:58-60.
re
[6] Selby C, Burgan B. Blast and fire engineering for topside structures-phase 2: final

summary report. SCi publication. 1998.

lP

[7] Tolloczko J. Interim guidance notes for the design and protection of topside structures

against explosion and fire: Steel Construction Institute; 1992.

na

[8] Transporter B. Boiler, Pressure Vessel Accidents Increase URL:

ur

http://bulktransporter.com. 2001.

[9] The National Board of Boiler and Pressure Vessel Inspectors, National Board Bulletin
Jo

URL: http://www.nationalboard.org. 2002.

[10] AirConditioning H, Refridgeration-The News. Boiler Accident Report : To Err Is

Human. 2002 November 8th.

Page | 51
[11] Lydell BO. Pipe failure probability—the Thomas paper revisited. Reliab Eng Syst Safe.

2000;68:207-17.

[12] Talbi B. CO₂ emissions reduction in road transport sector in Tunisia. Renew Sustain

Energy Rev. 2017;69:232-8.

[13] Milano J, Ong HC, Masjuki H, Chong W, Lam MK, Loh PK, et al. Microalgae biofuels

as an alternative to fossil fuel for power generation. Renew Sustain Energy Rev.

2016;58:180-97.

of
[14] Mahgerefteh H, Wong SM. A numerical blowdown simulation incorporating cubic

equations of state. Comput Chem Eng. 1999;23:1309-17.

ro
[15] Moss DR. Pressure vessel design manual: Elsevier; 2004.
-p
[16] Cui H, Wang W, Li A, Li M, Xu S, Liu H. Failure analysis of the brittle fracture of a

thick-walled 20 steel pipe in an ammonia synthesis unit. Eng Fail Anal. 2010;17:1359-76.
re
[17] Onyebuchi VE, Kolios A, Hanak DP, Biliyok C, Manovic V. A systematic review of key
lP

challenges of CO₂ transport via pipelines. Renew Sustain Energy Rev. 2017.

[18] Gradle R. Design of Gas Pipeline Blowdowns. Enrgy Proced Cdn. 1984:15-20.
na

[19] Ouyang L-b, Aziz K. Steady-state gas flow in pipes. J Petrol Sci Eng. 1996;14:137-58.

[20] Api R. Recommended practice 521. 1990.

ur

[21] Byrnes W, Reid R, Ruccia F. Rapid depressurization of gas storage cylinder. Ind Eng
Jo

Chem Process Des Dev. 1964;3:206-9.

[22] Richardson S, Saville G. Blowdown of vessels and pipelines. Institution of chemical

engineers symposium series: hemsphere publishing corporation; 1993. p. 195-.

[23] Simon LL, Introvigne M, Fischer U, Hungerbühler K. Batch reactor optimization under

liquid swelling safety constraint. Chem Eng Sci. 2008;63:770-81.

Page | 52
[24] Fisher H, Forrest H, Grossel SS, Huff J, Muller A, Noronha J, et al. Emergency relief

system design using DIERS technology: The Design Institute for Emergency Relief Systems

(DIERS) project manual: John Wiley & Sons; 2010.

[25] Haque A, Richardson S, Saville G, Chamberlain G. Rapid depressurization of pressure

vessels. J Loss Prevent Proc. 1990;3:4-7.

[26] Haque M, Richardson S, Saville G. Blowdown of pressure vessels. I. Computer model.

Trans IChemE, Part B, Proc Safe Env Prot. 1992;70:3-9.

of
[27] Richardson S, Saville G. Blowdown of pipelines. Offshore Europe: Society of

Petroleum Engineers; 1991.

ro
[28] Haque M, Richardson S, Saville G, Chamberlain G, Shirvill L. Blowdown of pressure
-p
vessels. II. Experimental validation of computer model and case studies. Trans IChemE, Part

B, Proc Safe Env Prot. 1992;70:10-7.

re
[29] Richardson S, Saville G. Blowdown of LPG pipelines. Process Saf Environ.
lP

1996;74:235-44.

[30] Cowley L, Tam V. Consequences of Pressurized LPG Releases: The Isle of Grain Full
na

Scale Experiments, submitted 13th Int. Conf LNG/LPG Conference and Exhibition, Kuala

Lumpur, Malaysia1988.
ur

[31] Tam V, Higgins R. Simple transient release rate models for releases of pressurised liquid

petroleum gas from pipelines. J Hazard Mater. 1990;25:193-203.

Jo

[32] Saville G, Richardson S, Barker P. Leakage in ethylene pipelines. Process Saf Environ.

2004;82:61-8.

[33] Norris III H, Puls R. Single-phase or multiphase blowdown of vessels or pipelines. SPE

Annual Technical Conference and Exhibition: Society of Petroleum Engineers; 1993.

Page | 53
[34] Norris III H. Hydrocarbon blowdown from vessels and pipelines. SPE Annual

Technical Conference and Exhibition: Society of Petroleum Engineers; 1994.

[35] Wong MA. Development of a mathematical model for blowdown of vessels containing

multi-component hydrocarbon mixtures: University of London; 1998.

[36] Peng D-Y, Robinson DB. A new two-constant equation of state. Ind Eng Chem Fund.

1976;15:59-64.

[37] Soave G. Equilibrium constants from a modified Redlich-Kwong equation of state.

of
Chem Eng Sci. 1972;27:1197-203.

[38] Twu CH, Coon JE, Cunningham JR. A new cubic equation of state. Fluid Phase

ro
Equilibra. 1992;75:65-79.
-p
[39] Witlox H, Holt A. A unified model for jet, heavy and passive dispersion including

droplet rainout and re-evaporation. CCPS International Conference & Workshop on

re
Modeling the Consequences of Accidental Releases of Hazardous Materials Sept1999.
lP

[40] Woodward JL, Cook J, Papadourakis A. Modeling and validation of a dispersing aerosol

jet. J Hazard Mater. 1995;44:185-207.

na

[41] Woodward JL, Papadourakis A. Reassessment and reevaluation of rainout and drop size

correlation for an aerosol jet. J Hazard Mater. 1995;44:209-30.

ur

[42] Witlox H, Holt A. “Validation of the unified dispersion model against Kit Fox field

data”, Contract 44003900 for Exxon Mobil. 2001.

Jo

[43] Witlox HW, Harper M, Oke A, Stene J. Phast validation of discharge and atmospheric

dispersion for pressurised carbon dioxide releases. J Loss Prevent Proc. 2014;30:243-55.

Page | 54
[44] Witlox HW, Fernandez M, Harper M, Oke A, Stene J, Xu Y. Verification and validation

of Phast consequence models for accidental releases of toxic or flammable chemicals to the

atmosphere. J Loss Prevent Proc. 2018;55:457-70.

[45] Cumber PS. Predicting outflow from high pressure vessels. Process Saf Environ.

2001;79:13-22.

[46] Skouloudis AN. Benchmark exercises on the emergency venting of vessels. J Loss

Prevent Proc. 1992;5:89-103.

of
[47] Eggers R, Green V. Pressure discharge from a pressure vessel filled with CO₂. J Loss

Prevent Proc. 1990;3:59-63.

ro
[48] Gebbeken B, Eggers R. Thermohydraulic phenomena during vessel release of initially
-p
supercritical carbon dioxide. International symposium two-phase ow modelling and

experimentation1995.
re
[49] Gebbeken B, Eggers R. Blowdown of carbon dioxide from initially supercritical
lP

conditions. J Loss Prevent Proc. 1996;9:285-93.

[50] Gebbeken B, Eggers R. Investigations on Vessel Blowdown and Flashing of Pressurized

na

Carbon Dioxide. Proc of the 4 World Conference on Experimental Heat Transfer, Fluid

Mechanics and Thermodynamics (ExHFT4)1997. p. 1709-16.

ur

[51] Fredenhagen A, Eggers R. High pressure release of binary mixtures of CO₂ and N₂.

Experimental investigation and simulation. Chem Eng Sci. 2001;56:4879-85.

Jo

[52] Mahgerefteh H, Falope GB, Oke AO. Modeling blowdown of cylindrical vessels under

fire attack. AIChE journal. 2002;48:401-10.

[53] API Recommended Practice 520 Part 1—Sizing and Selction, sixth ed. 1993.

Page | 55
[54] Song X-G, Park Y-C, Park J-H. Blowdown prediction of a conventional pressure relief

valve with a simplified dynamic model. Mathematical and Computer Modelling.

2013;57:279-88.

[55] Zhang J, Zhu D, Tian W, Qiu S, Su G, Zhang D. Depressurization study of supercritical

fluid blowdown from simple vessel. Annals of Nuclear Energy. 2014;66:94-103.

[56] D’Alessandro V, Giacchetta G, Leporini M, Marchetti B, Terenzi A. Modelling

blowdown of pressure vessels containing two-phase hydrocarbons mixtures with the partial

phase equilibrium approach. Chem Eng Sci. 2015;126:719-29.

of
[57] Roberts T, Medonos S, Shirvill L. Review of the response of pressurised process vessels

ro
and equipment to fire attack. 2000.

-p
[58] Park A, Ko Y, Ryu S, Lim Y. Numerical modeling of rapid depressurization of a

pressure vessel containing two-phase hydrocarbon mixture. Process Saf Environ.

re
2018;113:343-56.
lP

[59] Speranza A, Terenzi A. Blowdown of hydrocarbons pressure vessel with partial phase

separation. Applied and Industrial Mathematics in Italy: World Scientific; 2005. p. 508-19.
na

[60] Jarrah AA. Modeling and simulation of rapid depressurizations of hydrocarbon gas

mixture flowing at unsteady high velocity in piping transport system. International Journal of
ur

Industrial Engineering (SSRG-IJIE). 2016;3.

[61] Geurst J. Virtual mass in two-phase bubbly flow. Physica A. 1985;129:233-61.

Jo

[62] Geurst J. Two-fluid hydrodynamics of bubbly liquid/vapour mixture including phase

change. Philips J Res. 1985;40:352-74.

[63] Geurst J. Variational principles and two-fluid hydrodynamics of bubbly liquid/gas

mixtures. Physica A. 1986;135:455-86.

Page | 56
[64] Chen J, Richardson S, Saville G. Modelling of two-phase blowdown from pipelines—i.

A hyperbolic model based on variational principles. Chem Eng Sci. 1995;50:695-713.

[65] Chen J, Richardson S, Saville G. A simplified numerical method for transient two-phase

pipe flow. Trans IChemE. 1993;71:304-6.

[66] Chen J, Richardson S, Saville G. Modelling of two-phase blowdown from pipelines—II.

A simplified numerical method for multi-component mixtures. Chem Eng Sci. 1995;50:2173-

87.

of
[67] Lahey RT, Moody F. The Thermal Hydraulics of a Boiling Water Nuclear Reactor

(Monograph series on nuclear science and technology). 1977.

ro
[68] Reibold W, Reocreux M, Jones O. Blowdown phase. Nuclear Reactor Safety Heat

Transfer. 1981:325-78. -p
[69] Jackson J, Liles D, Ransom V, Ybarrondo L. LWR system safety analysis. Nuclear
re
Reactor Safety Heat Transfer. 1981;12:415.
lP

[70] Fairuzov YV. Blowdown of pipelines carrying flashing liquids. AIChE journal.

1998;44:245-54.
na

[71] Mahgerefteh H, Saha P, Economou IG. Fast numerical simulation for full bore rupture of

pressurized pipelines. AIChE Journal. 1999;45:1191-201.

ur

[72] Chen J-R. MODELLING OF TRANSIENT FLOW IN PIPELINE BLOVVDOWN

PROBLEMS: University of London; 1993.

Jo

[73] Hall A, Butcher G, Teh C. Transient simulation of two-phase hydrocarbon flows in

pipelines. European Two-Phase Flow Group Meeting (Hannover), paper1993. p. I4.

[74] Oke A, Mahgerefteh H, Economou I, Rykov Y. A transient outflow model for pipeline

puncture. Chemical engineering science. 2003;58:4591-604.

Page | 57
[75] DeReuck K, Craven R, Elhassan A. Transport Properties of Fluids: Their Correlation,

Prediction and Estimation, J. Millat, JH Dymond, and CA Nieto de Castro, eds., IUPAC.

Cambridge Univ. Press, Cambridge; 1996.

[76] Assael MJ, Trusler JM, Tsolakis TF. Thermophysical properties of fluids: an

introduction to their prediction: World Scientific; 1996.

[77] Veersteg K, Malalasekera W. An introduction to computational fluid dynamics: the

finite volume method. Pearson Schweiz Ag. 1995;20:400.

of
[78] Rajiwate FLH. Investigation of compressible fluid behaviour in a vent pipe during

blowdown: Curtin University; 2011.

ro
[79] Lemmon DEW, D. M. L. Huber, and D. M. O. McLinden. REFPROP Manual. Boulder:

U S Department of Commerce. 2009. -p

[80] Kunz O, Klimeck R, Wagner W, Jaeschke M. The GERG-2004 wide-range equation of
re
state for natural gases and other mixtures, Fortschr. Ber VDI, VDI-Verlag, Düsseldorf. 2007.
lP

[81] Bansal RK. Fluid mechanics and hydraulic machines. 9 ed ed. New Delhi: Laxmi

Publications (P) Ltd; 2005.

na

[82] Saad MA. Compressible Fluid Flow. 2 ed. ed. New Jersey: Prentice Hall; 1993.

[83] Shapiro AH. The dynamics and thermodynamics of compressible fluid flow (vol2).
ur

1954.
Jo

[84] Parker G. ‘Pop’safety valves: a compressible flow analysis. International journal of heat

and fluid flow. 1985;6:279-83.

[85] de Koeijer G, Borch JH, Jakobsenb J, Drescher M. Experiments and modeling of two-

phase transient flow during CO₂ pipeline depressurization. Enrgy Proced. 2009;1:1683-9.

Page | 58
[86] Clausen S, Munkejord ST. Depressurization of CO₂–a numerical benchmark study.

Energy Procedia. 2012;23:266-73.

[87] de Koeijer G, Borch JH, Drescher M, Li H, Wilhelmsen Ø, Jakobsen J. CO₂ transport–

Depressurization, heat transfer and impurities. Energy Procedia. 2011;4:3008-15.

[88] Cho M-I, Huh C, Jung J-Y, Kang S-G. Experimental study of N₂ impurity effect on the

steady and unsteady CO₂ pipeline flow. Enrgy Proced. 2013;37:3039-46.

[89] Clausen S, Oosterkamp A, Strøm KL. Depressurization of a 50 km long 24 inches CO₂

of
pipeline. Enrgy Proced. 2012;23:256-65.

[90] Bendiksen KH, Maines D, Moe R, Nuland S. The dynamic two-fluid model OLGA:

ro
Theory and application. SPE Prod Eng. 1991;6:171-80.
-p
[91] Span R, Wagner W. A new equation of state for carbon dioxide covering the fluid region

from the triple‐point temperature to 1100 K at pressures up to 800 MPa. J Phys Chem Ref
re
Data. 1996;25:1509-96.
lP

[92] Aursand P, Hammer M, Munkejord ST, Wilhelmsen Ø. Pipeline transport of CO₂

mixtures: Models for transient simulation. Int J Greenh Gas Con. 2013;15:174-85.
na

[93] Huh C, Cho M-I, Hong S, Kang S-G. Effect of impurities on depressurization of CO₂

pipeline transport. Enrgy Proced. 2014;63:2583-8.

ur

[94] Brown S, Martynov S, Mahgerefteh H, Proust C. A homogeneous relaxation flow model

Jo

for the full bore rupture of dense phase CO₂ pipelines. Int J Greenh Gas Con. 2013;17:349-

56.

[95] Chen NH. An explicit equation for friction factor in pipe. Ind Eng Chem Fund.

1979;18:296-7.

Page | 59
[96] Angielczyk W, Bartosiewicz Y, Butrymowicz D, Seynhaeve J-M. 1-D modeling of

supersonic carbon dioxide two-phase flow through ejector motive nozzle. 2010.

[97] Mahgerefteh H, Fairweather M, Falle S, Melheim J, Ichard M, Storvik I, et al.

CO₂pipehaz: quantitative hazard assessment for next generation CO₂ pipelines. Institute of

Chemical Engineers, Rugby. 2011:606-10.

[98] Martynov S, Brown S, Mahgerefteh H, Sundara V, Chen S, Zhang Y. Modelling three-

phase releases of carbon dioxide from high-pressure pipelines. Process Saf Environ.

of
2014;92:36-46.

[99] Martynov S, Brown S, Mahgerefteh H. An extended Peng‐Robinson equation of state for

ro
carbon dioxide solid‐vapor equilibrium. Greenh : Sci Tech. 2013;3:136-47.

-p
[100] Drescher M, Varholm K, Munkejord ST, Hammer M, Held R, de Koeijer G.

Experiments and modelling of two-phase transient flow during pipeline depressurization of

re
CO₂ with various N₂ compositions. Enrgy Proced. 2014;63:2448-57.
lP

[101] Hanna S, Chang J, Strimaitis D. Hazardous gas model evaluation with field

observations. Atmospheric Environment Part A General Topics. 1993;27:2265-85.

na
ur
Jo

Page | 60