Journal of Natural Gas Science and Engineering 36 (2016) 496e509

Contents lists available at ScienceDirect

Journal of Natural Gas Science and Engineering

journal homepage: www.elsevier.com/locate/jngse

Performance analysis of liquefied natural gas storage tanks in

refueling stations

rida*
Amir Sharafian, Omar E. Herrera, Walter Me
Clean Energy Research Centre, The University of British Columbia, 2360 East Mall, Vancouver, BC, V6T 1Z3, Canada

a r t i c l e i n f o a b s t r a c t

Article history: Liquefied natural gas (LNG) could replace diesel in the transportation sector. However, fugitive emissions
Received 5 August 2016 including boil-off gas (BOG) across the LNG supply chain have revealed uncertainties on the overall
Received in revised form environmental benefits of such replacement. In this study, time-dependent thermodynamic models were
28 September 2016
developed to study the LNG holding time of storage tanks in refueling stations before BOG releases to the
Accepted 26 October 2016
atmosphere. Previously overlooked factors, such as the thermal mass of storage tanks and the actual
Available online 28 October 2016
operating conditions at refueling stations, were included explicitly in the models. The effect of the
thermal mass of storage tanks on holding time is illustrated by an analysis of 57.20 m3 storage tanks filled
Keywords:
Liquefied natural gas
with LNG at
150  C and
126.5  C. The tank with the lower temperature fills shows 3.7-times longer
Boil-off gas holding time. Further investigations highlight the importance of the ratio of heat transfer surface area to
Storage tank the LNG volume as a key factor in proper sizing of storage tanks to maximize the holding time. Finally,
Holding time the modeling of a 57.20 m3 storage tank with a heat transfer coefficient of 0.022 W/m2K shows that fuel
Refueling station delivery rates as low as 1.89 m3/day are sufficient to maintain the tank pressure within allowable limits.
© 2016 Elsevier B.V. All rights reserved.

1. Introduction
Climate change is one of the main concerns of today's world
(Richter, 2014), but greenhouse gas (GHG) emissions from indus-
trial and transportation processes have steadily increased (Global
greenhouse, 2016; Sources of greenhouse gas emissions, 2016;
Liimatainen et al., 2014). For instance, the GHG emissions from
the U.S. medium- and heavy-duty trucks increased by 76% between
1990 and 2014 and reached 407.4 Mt CO2,eq (Inventory of U.S.,
2016). According to the announcements at the 21st Conference of
Parties in Paris, mitigation of climate change and reaching the 2  C
scenario targets would require immediate and significant changes
over the next three decades (as opposed to changes occurring over
centuries) (21st session of the Conference of the Parties et al., 2015).
It has been claimed that replacing conventional petroleum fuels,
e.g., diesel and gasoline, with low-carbon content fuels reduces
GHG emissions and climate change (van Der Hoeven, 2015). Natural
gas (NG) is considered a low-carbon content fuel (Van Den Broek
et al., 2015) and several studies reported the benefits of NG on
economic and market growth (Van Den Broek et al., 2015; Imran
* Corresponding author.

rida).
E-mail address: walter.merida@ubc.ca (W. Me
et al., 2016; Hao et al., 2016; Wang and Li, 2016; Furuoka, 2016;
Balitskiy et al., 2016; Wang et al., 2016; Şevik, 2015; Wang and
Lin, 2014; Kakaee et al., 2014; Wang et al., 2014; Khan et al.,
2015). However, and despite this significant body of work, recent
studies revealed uncertainty in the overall benefits associated with
NG use (Alvarez et al., 2012; Howarth et al., 2011; Venkatesh et al.,
2011; Davis and Shearer, 2014; McJeon et al., 2014; Delgado and
Muncrief, 2015). According to the Global Warming Potential
(GWP), methane emissions contribute up to 72 times more to
climate change than CO2 in a 20-year horizon (Solomon, 2007).
Therefore, the reduction in CO2 emissions from NG use must be
compared to the impact of the corresponding methane emissions.
Without reliable data on the actual deployment technologies, most
of the models and analyses comparing widespread NG use to the
existing energy options will remain incomplete.
Liquefied natural gas (LNG) is the condensed form of natural gas
with 60% volumetric energy density of diesel (Study on natural gas,
2014). The combustion of LNG in comparison with ultra-low sulfur
diesel can reduce CO2, NOx, and particulate matter emissions by up
to 20%, 90%, and 100%, respectively (International Gas Union, 2015).
These features make LNG a candidate fuel in the transportation
sector to reduce GHG emissions, especially for large, mobile ap-
plications, such as heavy-duty trucks (Bassi, 2011; Nicotra, 2013),
trains (Dunn and LeBlanc, 2014; Energy Carriers, 2014; Al Ali, 2015),

http://dx.doi.org/10.1016/j.jngse.2016.10.062
1875-5100/© 2016 Elsevier B.V. All rights reserved.
 A. Sharafian et al. / Journal of Natural Gas Science and Engineering 36 (2016) 496e509
and ships (Æsøy et al., 2011; Seo et al., 2016). Nevertheless, fugitive
emissions from LNG are a major concern.
LNG is a cryogenic liquid stored at temperatures as low
as
162  C. Due to its large temperature gradient with the envi-
ronment, LNG is gradually heated and evaporates (boil-off gas
(BOG)). Methane (the major component in LNG mixtures) is a
potent GHG and has more impact on climate change than CO2 due
to methane higher radiative forcing (Solomon, 2007). The BOG
leads to undesirable pressurization across the LNG distribution
chain (Chen et al., 2004). In large LNG carriers, the BOG is re-
liquefied or used as fuel to keep the LNG at atmospheric pressure
and low temperature (Miana et al., 2015). In large LNG regasifica-
tion plants and storage facilities, the BOG can be used to generate
electricity (Querol et al., 2010). However, in small-scale applica-
tions, such as LNG refueling stations, the BOG management is more
challenging. BOG generation and methane emissions from LNG
facilities are originated from (Lowell et al., 2013): 1) heat transfer to
LNG and pressurization of storage tanks, 2) ventilation of displaced
BOG when filling a tank, 3) heat transfer to hoses, lines, and pumps,
4) precooling of equipment prior to LNG transfer, and 5) LNG
transfer from a high pressure tank to a low pressure tank.
Burnham et al. (2015). classified the natural gas (NG) well-to-
tank into four sectors to quantify fugitive emissions rate: 1) gas
field, 2) processing, 3) transmission and storage, and 4) distribu-
tion. Their analysis showed that on average 1.12%e1.15% of NG was
emitted to the atmosphere across the value chain. Their in-
vestigations also indicated that transportation and storage sector
accounted for 35% of methane emissions followed by the distri-
bution sector, which includes refueling stations and fueling process,
with 28%. Therefore, the transportation, storage, and distribution
sectors have significant opportunities to improve by preventing the
BOG release and fixing existing leaks. The main focus of this study is
on the distribution sector.
According to Burnham et al. (2015), the average methane
venting from LNG refueling stations was about 0.32% per delivery of
LNG. Hailer's measurements from two LNG refueling stations in the
U.S. indicated that one of the operating stations had methane
emissions of 0.1%e1.5% of fuel dispensed to vehicles (Hailer, 2015).
However, the second station had methane emissions of 0.9%e5.3%.
Field surveys from 2400 LNG refueling stations in China demon-
strated that more than 1600 stations had daily methane emissions
of greater than 5% and in some cases 10% due to improper insu-
lation (G boil off gas (G) em, 2015). China has currently about 3200
LNG refueling stations (development status, 2015) which account
for 94% of the total refueling stations around the world (Sharafian
et al., 2016). It is expected that the number of LNG refueling sta-
tions in the U.S. and China will reach 223 and 5000 by 2020,
respectively.
Developing appropriate thermodynamic models to determine
the weaknesses in the design of LNG infrastructures and quantify
their BOG generation rates is the first step toward the design of zero
or near zero fugitive emissions LNG facilities. A summary of rele-
vant thermodynamic studies on LNG infrastructures are listed in
Table 1.
The LNG facilities investigated in Table 1 illustrate the recent
work on LNG storage tanks (Chen et al., 2004; Barclay et al., 1998;
Adom et al., 2010; Pellegrini et al., 2014; Liu et al., 2015; Migliore
et al., 2015), transportation (Hasan et al., 2009; Miana et al., 2010;
Pellegrini et al., 2014), unloading (tank-to-tank transfer) (Yan and
Gu, 2010), regasification terminals (Querol et al., 2010; Park et al.,
2012; Fahmy et al., 2015; Kurle et al., 2015), and LNG weathering
(Miana et al., 2010; Pellegrini et al., 2014). There are limited studies
available in the literature on LNG storage tanks with less than
113.5 m3 capacity for LNG refueling station applications, such as
Refs. (Chen et al., 2004; Barclay et al., 1998; Pellegrini et al., 2014).
497
These tanks can hold LNG at pressures about 1300 kPa which are
higher than those of large storage tanks (100,000e300,000 m3)
used in tanker ships with operating pressures close to atmospheric
pressure. The difference between these operating pressures affects
the thermo-physical properties of LNG, LNG holding time, and the
BOG generation rate. LNG holding time refers to the time a storage
tank can hold the LNG without venting (Delgado and Muncrief,
2015).
In this study, we analyze LNG storage tanks of refueling stations
by using practical parameters such as tank storage size, LNG initial
temperature, tank thermal insulation, and fuel delivery rate. The
main differences between this study and previous models in Refs
(Chen et al., 2004; Barclay et al., 1998; Adom et al., 2010; Pellegrini
et al., 2014). are to consider the thermal mass of storage tank and
use real operating temperatures and pressures for the LNG at
refueling stations. Previous studies used LNG at
162  C (the dew
point of NG at atmospheric pressure) for the modeling which is not
the case in a LNG refueling station. This study highlights the
importance of these parameters on the model predictions, such as
heat transfer rate to tank, LNG holding time, and the BOG genera-
tion rate.
2. LNG refueling stations and vehicles' fuel supply system
LNG is dispensed to vehicles in two conditions: 1) Unsaturated
(cold) LNG at a less than
143  C and 340 kPa, and 2) saturated
(warm) LNG at
125 to
131  C, and 690e930 kPa (Roche, 2009).
This is due to variations in vehicles' fuel supply systems (Sharafian
et al., 2016; Powars, 2010; Wiens et al., 2001). In a simple vehicle's
fuel supply system, the tank pressure transfers the LNG to the
vaporizer and engine at appropriate temperature, pressure, and
flow rate (Powars, 2010; Wiens et al., 2001). It is customary to fill
these LNG tanks with saturated LNG, which has been heated prior
to filling to increase its saturation pressure, to supply the required
fuel flow rate. Other, more sophisticated fuel supply systems have
auxiliary equipment, such as a pump or compressor, for trans-
ferring the LNG from the onboard tank to the vaporizer and engine
(Sharafian et al., 2016; Powars, 2010). These fuel supply systems are
capable of accepting unsaturated LNG.
The process of heating the LNG to increase its saturation tem-
perature and pressure is known as “LNG conditioning”. There are
two methods of LNG conditioning at refueling stations: bulk and
on-the-fly conditioning. Fig. 1a shows a simplified schematic of a
LNG refueling station with bulk conditioning. After filling the
storage tank with unsaturated LNG, the pump pushes the LNG to
the heater to rise its temperature and pressure. This process con-
tinues until the LNG pressure stored in the storage tank reaches the
set point.
In contrast, a LNG refueling station with on-the-fly conditioning
increases the pressure and temperature of unsaturated LNG
simultaneously with the fueling process, as shown in Fig. 1b. This
method helps the storage tank to store more LNG with higher
density for a longer time. However, the heater needs to be precisely
designed to heat the LNG on-the-fly within a short time without
overheating it.
3. LNG storage tank modeling and assumptions
LNG storage tanks in refueling stations with 22.7e113.5 m3 ca-
pacity have a maximum allowable working pressure (MAWP) of
about 1300 kPa (Powars, 2010). Fig. 2 shows a schematic of a
double-wall LNG storage tank with a net capacity of 57.2 m3.
A combination of Aspen Plus and Aspen Plus Dynamics software
(Aspen Engineering, 2016) was used to model a LNG storage tank.
The LNG storage tank was initially designed in Aspen Plus and was
498 A. Sharafian et al. / Journal of Natural Gas Science and Engineering 36 (2016) 496e509

Table 1
Thermodynamic modeling of different LNG facilities across the value chain to determine their BOG generation rates.

Ref. Scope of modeling Approach* Results

Barclay Optimization of LNG storage tank capacity  Conducted a thermo-economic analysis to size LNG  Capacity and capital cost of LNG storage tanks showed
et al., for a fleet-size refueling station storage tanks with respect to the fleet size a linear relationship
1998  Investigated the effects of LNG storage capacity,  Use of onsite liquefier assisted to use less efficient and
insulation material, and tank pressure on the capital cost less expensive storage tanks
of LNG refueling stations  A single-wall storage tank with polyurethane insu-
lation equipped with a liquefier had lower life-cycle cost
than a double-wall LNG storage tank
Chen Thermodynamic modeling of LNG storage  Developed steady-state and dynamic models  Thermal conductance of LNG storage tanks could be
et al., tank and refueling station  Analyzed the effects of number of vehicles fueled every calculated indirectly by using dynamic thermodynamic
2004 day on methane emissions modeling and measuring temperature and pressure over
 Modeled 22.7e113.6 m3 LNG storage tanks time when no fuel is delivered
 Delivering more than 2.8 m3 LNG per day eliminated
methane emissions from a 49.0 m3 storage tank with
thermal conductance of 2 W/K
 An electric generator was recommended to be used for
the BOG management and power generation
 A liquefier was recommended to be used for the BOG
management
Hasan BOG management in LNG transportation  Used Aspen HYSYS for thermodynamic modeling  Important factors on the BOG generation during LNG
et al., from liquefaction plant to regasification  Used Soave-Redlich-Kwong (SRK) equation of state transportation were LNG sloshing, LNG composition,
2009 terminal ambient temperature, tank thermal conductance, and
tank pressure
 The BOG generation was increased by increase in
voyage distance
Yan and Thermodynamic modeling to determine  Used Aspen Plus software for thermodynamic  The BOG generation rate was minimized at a specific
Gu, important parameters on BOG generation modeling LNG flow rate and pump head
2010 during LNG unloading in LNG terminals  Used Peng-Robinson (PR) equation of state  Increasing the elevation difference of pipelines
 Analyzed the effects of LNG flow rate, and pipe between two tanks increased the BOG generation rate
elevation change, roughness, and diameter on the BOG  Larger pipe diameter reduced the BOG generation rate
generation rate and pump head
 Pipe roughness had negligible effect on the BOG
generation rate
Adom Thermodynamic modeling to calculate the  Used Lee-Kesler-Plocker (LKP) and the Starling modi-  Structure of the tank affected the BOG generation rate
et al., BOG generation rate of LNG storage tanks fied Benedict-Webb-Rubin (BWRS) empirical models to  Smaller tanks had larger BOG generation rate because
2010 simulate compressibility factor and equation of state of large surface area to volume ratio
 Assumed LNG evaporation only at the liquid-vapor  The BOG generation rate decreased as the tank
interface pressure increased
 Assumed constant LNG temperature and density  Constant temperature and density during the whole
during the whole process process were reduced the accuracy of modeling
 The BOG generation rate did not necessarily mean
methane emissions to the atmosphere
Querol Thermodynamic modeling to determine  Used Aspen Plus software for thermodynamic  Recondenser was the best option to manage the BOG
et al., the daily BOG generation rate in LNG modeling due to its low power consumption and maintenance
2010 terminals  Analyzed the usage of recondenser, compressor, and  Integration of cogeneration plants with LNG terminals
reliquefaction plant on the BOG management in were proposed to manage the BOG and generate the
receiving terminals electricity required for operating terminals
 The BOG was a mixture of methane and nitrogen not
heavy hydrocarbons
Miana Thermodynamic and intelligent neural  Used MOLAS software to predict LNG composition and  The intelligent neural network model had better
et al., network modeling to determine LNG thermo-physical properties prediction in LNG composition at the destination than
2010 weathering in tanker ships  Used a lumped body model which is independent from thermodynamic modeling because of using historical
the shape of LNG storage tanks data
 For new tank designs, thermodynamic modeling
provided better results
Park et al., Thermodynamic modeling to calculate the  Used Aspen Plus combined with Sequential Quadratic  Proper design of receiving terminals could reduce the
2012 BOG generation rate in receiving terminals Programming (SQP) optimization solver in Matlab operating cost of compressors for handling the BOG
software for thermodynamic modeling  Optimal design of receiving terminals provided 23%
 Used Peng-Robinson (PR) equation of state more energy savings and a payback period of less than
0.2 year
 Direct compression of the BOG had higher costs than
recondensation
Pellegrini Thermodynamic modeling to study LNG  Developed an in-house code for thermodynamic  LNG composition had to be considered for an accurate
et al., weathering in LNG storage tanks modeling of LNG storage tanks thermo-economic analysis
2014  Used Soave-Redlich-Kwong (SRK) equation of state  Using average BOG value reported in previous
 Assumed constant heat flux on walls of a 190 L storage literature was misleading in prediction of LNG
tank composition and density change
Fahmy Thermodynamic modeling and cost  Used Aspen HYSYS software to determine the  Removing heavier hydrocarbons from LNG resulted in
et al., optimization of regasification plants minimum lower heating value range of natural gas by higher net gain
2015 recovering heavier hydrocarbons and allowing  A regasification plant with appropriate heavy
maximum net gain hydrocarbon removal provided $14M/year net gain
 Used Penge-Robinsone-Stryjeke-Vera (PRSV) equation
of state
Thermodynamic modeling to minimize the  Used Aspen Plus software for thermodynamic  Usage of the BOG as fuel gas was the most effective way
BOG generation rate in exporting terminals modeling to manage the BOG
 A. Sharafian et al. / Journal of Natural Gas Science and Engineering 36 (2016) 496e509 499

Table 1 (continued )

Ref. Scope of modeling Approach* Results

Kurle  Used Soave-Redlich-Kwong (SRK) equation of state
et al.,  Used C3-MR (mixed refrigerant) process in the LNG
2015 liquefier
Liu et al., Thermodynamic modeling to determine  Not clarified modeling software and assumptions
2015 the energy required to liquefy the BOG of  Analyzed four re-liquefiers powered by a coal driven
100,000 to 300,000 m3 LNG storage tanks Claude cycle, a Brayton cycle run on the BOG, a Claude
cycle run by a BOG-driven spark-ignition engine, and a
liquid nitrogen condenser
Migliore Thermodynamic modeling to investigate  Developed an in-house code for thermodynamic
et al., LNG weathering in storage tanks of modeling of LNG storage tanks
2015 regasification terminals  Used the revised Kloseke-McKinley method (an
empirical correlation) for calculating the LNG density
 Assumed variable heat flux to tank due to the
temperature difference between the ambient and LNG
 Less than 20% of the liquefied BOG energy was
consumed for the BOG recovery
 The liquefier powered by the Claude cycle driven by a
spark-ignition engine had the highest BOG energy
recovery
 The liquefied nitrogen condenser provided the lowest
BOG energy recovery
 Initial composition and thermal mass of LNG affected
the BOG generation rate
 Presence of 0.5% nitrogen in LNG composition resulted
in a 7% decrease in the BOG generation rate
 Increasing the ambient temperature as low as 1  C
increased the BOG generation rate by 0.2% independent
from initial composition of LNG and tank size

* Common assumption in all studies: 1) Liquid-vapor phases are in equilibrium, and 2) steady-state assumption unless otherwise noted.

Pressure relief Pressure relief

valve valve

Dispenser Dispenser
Vapor return Vapor return
line line

Heater
Storage Storage
tank tank

Heater
Sump Sump
tank tank

(a) (b)
Fig.1. LNG conditioning at a refueling station by: (a) bulk conditioning method (Wiens et al., 2001) and (b) on-the-fly conditioning method.

solved for the steady-state condition. Then, the model was expor-
ted to Aspen Plus Dynamics where the inlet and outlet valves were
Din= 2.4 m closed to model a stationary tank with no fuel delivery. The ther-
modynamic model included time-dependent conservation of mass
and energy, heat and mass transfer equations and equation of state.
Outer tank In this study, pure methane was considered as the LNG because
more than 90% of NG is composed of methane. A constant ambient
temperature boundary condition was applied on the walls of the
LTank = 12.64 m storage tank. The liquid-vapor phases, i.e., LNG and BOG, were
Inner tank
assumed to be in equilibrium. The Peng-Robinson (PR) equation of
state was used to determine the density of vapor phase for a given
temperature and pressure.
According to technical recommendations (Powars, 2010), LNG
Feed storage tanks at refueling stations should be filled to a maximum of
80% of their volume with unsaturated LNG. In the case of saturated
LNG, storage tanks should not be filled more than 90% of their
To volume. In this study, the rationale behind this practical suggestion
dispenser is discussed. To have a fair comparison, the initial volumes of un-
Fig. 2. Schematic of a double-wall LNG storage tank with the net capacity of 57.2 m3.
saturated and saturated LNG in Aspen Plus Dynamics were set at 80%
The dimensions on the figure belong to the inner tank (LNG storage vessels, 2016). of the tank net volume. The pressure and temperature of LNG in the
tank increased until the tank pressure reached its MAWP due to the
500 A. Sharafian et al. / Journal of Natural Gas Science and Engineering 36 (2016) 496e509

heat transfer from the environment. This time is defined as the
“LNG holding time” of the tank with no methane emissions. During
the modeling, there is no LNG flow in and out of the storage tank
unless otherwise specified. Further details about the specifications
of LNG storage tank, ambient temperature, initial LNG level in the
tank, and temperatures and pressures of saturated and unsaturated
LNG are listed in Table 2.
The overall heat transfer coefficient, Uinsulation, of the storage
tank was calculated based on LNG at
162  C and 101.325 kPa, and
the storage tank average daily BOG rate of 0.3% (Powars, 2010):
respectively.
For solving the time-dependent governing equations, the mixed
Newton method was selected in Aspen Plus Dynamics. The relative
error differences for all variables, namely, density, temperature,
pressure, and liquid and vapor mass fractions, were set at 10
4 at
each time step. The mixed Newton method uses the Newton
method for initialization and the fast Newton method for dynamic
iterations. Therefore, it provides a fast iteration speed and high
convergence rate for most of time-dependent simulations (Aspen
Plus Dynamics user, 2016).

Q_ ¼ BOG%  rLNG  Vtank  hfg ¼ 4. Results

  

0:3% kg 3 kJ
 423:0  57:20 m  512:59 ¼ 430 W 4.1. Model validation
24  3600 m3 kg
(1) Experimental data (Advanced LNG onboard storage system,
2003) of a horizontal onboard tank with net capacity of 0.257 m3
Q_ 430ðWÞ were used to validate the model. The holding time of the tank was
ðUAÞtank ¼ ¼ ¼ 2:3W=K (2) measured according to the SAE J2343 Recommended Practice
Tambient
TLNG 25
ð
162Þ
standard (Society of Automotive, 2008). The tank was filled to 75%
of its net capacity and changes in its pressure was recorded over
ðUAÞtank 2:3ðW=KÞ .
Uinsulation ¼ ¼   ¼ 0:022W m2 K (3) time until the tank pressure reached its MAWP. Further details
Atank 104:35 m2 about the tank and initial temperature and pressure are shown in
Table 3.
where, in Eq. (1), Q_ is the heat transfer rate to the tank, rLNG and hfg Fig. 3 shows the comparison of experimental data and the re-
are the LNG density and heat of evaporation at the given temper- sults of the model. The LNG pressure stored in the tank increased
ature and pressure, and Vtank is the storage tank net volume. In Eqs. gradually due to the heat leak rate to the tank and after about 5.7
(2) and (3), (UA)tank is the thermal conductance of the storage tank days (137 h), it reached the MAWP. As shown in Fig. 3, the model
and Atank is the total heat transfer surface area of the inner tank, as predicted the experimental data with good accuracy. The
shown in Fig. 2. maximum and average relative differences between the experi-
The average daily BOG generation rate in the model is calculated mental and numerical modeling were 6.2% and 3.1%, respectively.
by Eq. (5):
4.2. Baseline model analysis
Q_ i Dti
BOGi % ¼  100 (4)
rLNG  VLNG  hfg Variations in LNG pressure, temperature and level in the tank,
and heat leak rate to the tank over time are discussed in this section
Pi¼tholding time for the given values in Table 2. Fig. 4a and Fig. 4b show the pressure
i¼0
BOGi % and temperature variations of unsaturated and saturated LNG in a
BOGð%=dayÞ ¼  24  3600 (5)
tholding time 57.2 m3 tank over time. Unsaturated LNG with initial pressure of
230 kPa is gradually heated and after about 87 days, the tank
where in Eq. (4), BOGi , Q_ i , and Dti are the BOG generation, the heat pressure reaches its maximum allowable working pressure
leak rate to the tank, and the time interval at time step i, (MAWP). Whereas, the pressure of saturated LNG initially at

Table 2
Specifications of LNG storage tank, and the base-case initial and boundary conditions.

Parameter Value Description

Tank net capacity, Vtank 57.20 m3 A vertical double-wall storage tank with a vacuum insulation between two walls (LNG storage
vessels, 2016)
Length of inner tank, Ltank 12.64 m Calculated from the length of outer tank
Outer diameter, Do 2.896 m Ref. (LNG storage vessels, 2016)
Inner diameter, Di 2.4 m Calculated from the net capacity of the tank
Insulation thickness, tinsulation 0.248 m Calculated from the tank inner and outer diameters
Mass of storage tank (empty) 21,455 kg Ref. (LNG storage vessels, 2016)
Specific heat capacity of storage tank, cp,tank 477 J/kgK Assumed from stainless steel 304
Surface area of inner tank, Atank 104.35 m2 Calculated from inner tank geometry
Overall heat transfer coefficient of tank, Uinsulation 0.022 W/ Calculated from BOG generation rate of 0.3%/day for methane at
162  C and 101.325 kPa
m2K (Powars, 2010)
Thermal conductance of tank, (UA)tank 2.30 W/K
Ambient temperature, Tambient 25  C
Maximum allowable working pressure of storage tank 1300 kPa (Powars, 2010)
(MAWP)
Initial conditions
Initial LNG level in tank 10.11 m Calculated based on 80% of the length of inner tank
Unsaturated LNG temperature
150  C
Unsaturated LNG pressure 230.65 kPa
Saturated LNG temperature
126.5  C
Saturated LNG pressure 900 kpa
 A. Sharafian et al. / Journal of Natural Gas Science and Engineering 36 (2016) 496e509 501

Table 3
Specifications of LNG storage tank and initial operating conditions.

Parameter Value Description

Tank net capacity, Vtank 0.257 m3 A horizontal double-wall tank with a vacuum insulation between two walls (Advanced LNG onboard
storage system, 2003)
Length of inner tank, Ltank 0.909 m Calculated from the net capacity of the tank
Outer diameter, Do 0.66 m Ref. (Advanced LNG onboard storage system, 2003)
Inner diameter, Di 0.6 m Ref. (Advanced LNG onboard storage system, 2003)
Insulation thickness, tinsulation 0.03 m Calculated from the tank inner and outer diameters
Mass of storage tank (empty) 175 kg Ref. (Advanced LNG onboard storage system, 2003)
Specific heat capacity of storage tank, cp,tank 477 J/kgK Assumed from stainless steel 304
Surface area of inner tank, Atank 2.279 m2 Calculated from inner tank geometry
Overall heat transfer coefficient of tank, Uinsulation
0.075 W/ Calculated from average heat leak rate of 20.9 W given in Ref. (Advanced LNG onboard storage
m2K system, 2003)
Thermal conductance of tank, (UA)tank 0.17 W/K
Ambient temperature, Tambient 25  C Assumed
Maximum allowable working pressure of storage tank 1585 kPa Ref. (Advanced LNG onboard storage system, 2003)
(MAWP)
Initial conditions
Initial LNG level in tank 0.4212 m Calculated based on 75% of the volume of inner tank in the horizontal position.
LNG temperature
147.4  C
LNG pressure 283 kpa

1800 This will be discussed further in Fig. 6.

MAWP The BOG generation of the storage tank filled with unsaturated
1500
LNG pressure [kPa]

and saturated LNG are depicted in Fig. 4e and f, respectively. The

1200
900
600
300 Experiement
Model
0 of evaporation of LNG steadily decreases. For example, the enthalpy
0 1 2 3 4 5
Time [day]
Fig. 3. Comparison of changes in LNG pressure between experimental data (Advanced
LNG onboard storage system, 2003) and the model developed in this study.
900 kPa exceeds the MAWP of the tank after 23 days. As shown in
Fig. 4b, temperatures of unsaturated and saturated LNG increase
linearly from
150  C and
126.5  C, respectively, to
118.3  C.
Fig. 4c demonstrates the LNG level variations in the tank due to
the heat transfer from the environment and changes in the LNG
density. Unsaturated and saturated LNG levels are initially at
10.11 m (corresponding to 80% of the length of the inner tank, Ltank,
as shown in Fig. 2). The unsaturated LNG level increases by
increasing the LNG temperature and reaches 11.77 m when the tank
pressure reaches its MAWP. This shows a 1.66 m change between
the initial and the end levels of LNG in the tank due to decrease in
the LNG density by 16.5% from 405.7 to 348.2 kg/m3. Saturated LNG
level also increases by 0.46 m and reaches 10.57 m. The level of
saturated LNG increases less because the density of saturated LNG
only decreases by 4.5% from its initial density. These data support
the practical recommendation on filling storage tanks up to 80%
and 90% of their net volume with unsaturated and saturated LNG,
respectively.
The amount of heat leak rate to the LNG storage tank is shown in
Fig. 4d. The heat leak rate to unsaturated and saturated LNG de-
creases over time as the LNG temperature gradually increases. The
heat leak rate to the tank filled with saturated LNG is less than that
filled with unsaturated LNG. As shown in Fig. 4d, the heat leak rate
to the tank shows an exponential decay in the first few days and
then linearly decreases. This is because of the thermal mass of the
storage tank that stores a portion of heat transferred to the LNG.
BOG generation of the tank filled with unsaturated LNG is greater
than that of the tank filled with saturated LNG. This is due to a
larger temperature gradient between the unsaturated LNG and the
environment. As shown in Fig. 4e and f, the BOG generation de-
creases over time to a minimum and then increases. Increase in the
pressure of the LNG storage tank minimizes the BOG generation.
However, as the tank temperature and pressure increase, the heat
6 of evaporation of LNG decreases by 24% from 488.9 to 394.8 kJ/kg
by increasing the LNG temperature and pressure from
150  C and
230.65 kPa to
118  C and 1300 kPa. Therefore, a competing trend
between increase in the pressure and decrease in the enthalpy of
evaporation controls the BOG generation. These results refute the
assumption of constant BOG generation and BOG generation
reduction by increase in the tank pressure, Refs (Hasan et al., 2009;
Adom et al., 2010). for instance.
Fig. 5 shows the LNG holding time and average daily BOG gen-
eration rate of the 57.2 m3 storage tank at the baseline operating
conditions. It can be seen from Fig. 5 that storage of unsaturated
LNG in comparison with saturated LNG increases the LNG holding
time by almost 3.74 times from 23.4 days to 87.4 days. As a result,
the average daily BOG generation rate of the tank filled with un-
saturated LNG is 12.5% less than that of the tank filled with satu-
rated LNG.
Effects of the thermal mass of storage tank on the modeling
results are shown in Fig. 6. The thermal mass of the tank stores a
portion of the heat transferred from the environment to the LNG as
shown in Fig. 6a and b, as an example. The tank with zero thermal
mass filled with saturated LNG has faster pressure and temperature
rise than the tank with thermal mass. Also, the LNG level increases
in a shorter time for the tank with zero thermal mass compared to
that with thermal mass as shown in Fig. 6c. These results indicate
that thermodynamic models with zero thermal mass for storage
tanks underestimate the LNG holding time and overestimate the
average daily BOG generation rate.
Fig. 6d shows that the heat leak rate to LNG in the storage tank
with zero thermal mass decreases linearly with time. Whereas, the
heat leak rate to the LNG in the storage tank with thermal mass
initially reduces because a portion of the heat is stored in the walls
of the storage tank. The remainder of the heat is transferred to the
502 A. Sharafian et al. / Journal of Natural Gas Science and Engineering 36 (2016) 496e509

-115
1400 MAWP -120

LNG temperature [°C]

LNG pressure [kPa]
1200 -125
1000 -130
800 -135
-140
600
-145
400 -150
Unsaturated LNG Unsaturated LNG
200 -155
Saturated LNG Saturated LNG
0 -160
0 10 20 30 40 50 60 70 80 90 0 10 20 30 40 50 60 70 80 90
Time [day] Time [day]

(a) (b)
12.0 370

Heat leak rate to tank [W]

LNG level in tank [m]

11.6 350
330
11.2
310
10.8
290
10.4 Unsaturated LNG 270 Unsaturated LNG
Saturated LNG Saturated LNG
10.0 250
0 10 20 30 40 50 60 70 80 90 0 10 20 30 40 50 60 70 80 90
Time [day] Time [day]

(c) (d)
0.006 0.0017

0.0058 0.00165
BOG generation%

BOG generation%

0.0016
0.0056
0.00155
0.0054
0.0015
0.0052 0.00145
Unsaturated LNG Saturated LNG
0.005 0.0014
0 10 20 30 40 50 60 70 80 90 0 4 8 12 16 20 24
Time [day] Time [day]

(e) (f)
Fig. 4. Variations in unsaturated and saturated LNG (a) pressure, (b) temperature, and (c) level in tank, (d) heat leak rate to tank, (e) BOG generation% of unsaturated LNG, and (f)
BOG generation% of saturated LNG vs. time.

0.5 LNG. When the thermal mass of storage tank cannot store more
LNG holding time BOG rate
Average BOG rate [%/day]

heat, the heat leak to the tank is transferred mostly to the LNG. At
100
LNG holding time [day]

87.4 0.4 this point, the slope of the heat leak rate curve is parallel to that of
0.36
80 0.32 the storage tank with zero thermal mass, as shown in Fig. 6d. This
0.3 analysis highlights the importance of thermal mass of storage tanks
60 that has been overlooked in the literature (Chen et al., 2004; Adom
0.2 et al., 2010; Pellegrini et al., 2014; Migliore et al., 2015).
40
23.4
20 0.1 4.3. Performance analysis of LNG storage tanks without fuel
delivery
0 0.0
Unsaturated LNG Saturated LNG
Proper usage of insulation materials in LNG infrastructures can
Fig. 5. Unsaturated and saturated LNG holding time and average daily BOG rate of a minimize the heat transfer to LNG. In a storage tank, the thermal
57.2 m3 storage tank at the base operating conditions given in Table 2. resistance of metallic walls can be neglected in comparison with
those of insulation materials. The heat transfer resistances due to
insulation materials on the side walls, roof, and ceiling of the
 A. Sharafian et al. / Journal of Natural Gas Science and Engineering 36 (2016) 496e509 503

1400 -117

LNG tank temperature [°C]

Saturated LNG Saturated LNG
LNG tank pressure [kPa]
MAWP
1300 -119
1200
-121
1100
-123
1000
900 Tank with zero thermal mass -125 Tank with zero thermal mass
Tank with thermal mass Tank with thermal mass
800 -127
0 5 10 15 20 25 0 5 10 15 20 25
Time [day] Time [day]

(a) (b)
10.6 350
Saturated LNG Saturated LNG
10.5 330

Heat leak rate [W]

LNG level [m]

10.4
310
10.3
290
10.2
10.1 Tank with zero thermal mass 270 Tank with zero thermal mass
Tank with thermal mass Tank with thermal mass
10 250
0 5 10 15 20 25 0 5 10 15 20 25
Time [day] Time [day]

(c) (d)
Fig. 6. Variations in saturated LNG (a) pressure, (b) temperature, and (c) level in a 57.2 m3 storage tank with and without thermal mass (Other parameters are as given in Table 2).

storage tank are assumed to be in parallel. Equation (6) gives the
thermal conductance of the storage tank, (UA)tank:
1 1
ðUAÞtank ¼
 þ tinsulation
þ tinsulation
kinsulation Aroof
ln Do=
Di
2pkinsulation Ltank
The parameters in Eq. (6) are given in Tables 2 and 4. By
replacing (UA)tank in Eq. (3), the overall heat transfer coefficient,
Uinsulation, of the storage tank can be calculated. The thermal con-
ductivity, kinsulation, thermal conductance, (UA)tank, and the overall
heat transfer coefficient of the tank, Uinsulation, for different insu-
lation materials are summarized in Table 4.
Effects of Uinsulation ranging from 0.01 to 0.25 W/m2K on the
unsaturated and saturated LNG pressures and heat leak rates to the
tank are shown in Fig. 7. For Uinsulation of 0.01 W/m2K, the unsatu-
rated and saturated LNG pressures take more than 80 and 20 days,
respectively, to reach the MAWP of the storage tank as shown in
Fig. 7a and b. While, the storage tank with Uinsulation of 0.25 W/m2K
can only hold unsaturated and saturated LNG for about 10 and 3
days, respectively, before releasing the BOG to the atmosphere. As
shown in Fig. 7c and d, the heat leak rate to the tank varies from 165
to 2870 W for unsaturated LNG and 154 to 2500 W for saturated
1 LNG under Uinsulation of 0.01e0.25 W/m2K. The amount of heat leak
(6)
kinsulation Aceiling rates shown in Fig. 7c and d can be used as benchmark values in the
design or selection of proper liquefier in LNG refueling stations.
Fig. 8 shows the LNG holding time and average daily BOG gen-
eration rate of the storage tank filled with unsaturated and satu-
rated LNG under different Uinsulation. It can be seen in Fig. 8a that the
tank with Uinsulation of 0.01 W/m2K (similar to that of LCI insulation
material under vacuum pressure of 133.3 Pa listed in Table 4) can
hold unsaturated and saturated LNG for 188 and 50 days, respec-
tively. Using conventional insulation materials, such as poly-
urethane with Uinsulation of 0.10 W/m2K, and no vacuum results in
unsaturated and saturated LNG holding times of 22 and 5.7 days,
respectively; more than 8 times reduction in the LNG holding time
compared to that of Uinsulation of 0.01 W/m2K.
Fig. 8b shows that the average daily BOG generation rate in-
creases from 0.15% to 2.59% for unsaturated LNG and 0.16e2.93% for
saturated LNG when Uinsulation increases from 0.01 to 0.25 W/m2K.

Table 4
Thermal properties of different insulation material with/without vacuum pressure.

Insulation material (Fesmire and Augustynowicz, 2005) Vacuum pressure kinsulation (UA)tank Uinsulation
[Pa]a [W/m.K] [W/K] [W/m2K]

Layered composite insulation (LCI) 133.3 0.0016 0.735 0.007

Aerogel blanket 133.3 0.0034 1.561 0.015
Multi-layer insulation (MLI) 133.3 0.010 4.592 0.044
Fiberglass 133.3 0.014 6.429 0.062
Perlite powder 133.3 0.016 7.348 0.070
Polyurethane No vacuum 0.021 9.644 0.092
Cellular glass foam No vacuum 0.033 15.155 0.145
a
133.3 Pa ¼ 1 torr.
504 A. Sharafian et al. / Journal of Natural Gas Science and Engineering 36 (2016) 496e509

1350
1400 MAWP MAWP
0.25
LNG pressure [kPa]

LNG pressure [kPa]

1200 0.25 1250
1000
1150 0.1
800 0.1 0.05
0.05 0.02
600 1050 0.02
400
UinsulaƟon= 0.01 W/m2K 950 UinsulaƟon= 0.01 W/m2K
200
Unsaturated LNG Saturated LNG
0 850
0 10 20 30 40 50 60 70 80 0 4 8 12 16 20
Time [day] Time [day]

(a) (b)
3500 2800

Heat leak rate to tank [W]

Heat leak rate to tank [W]

Unsaturated LNG Saturated LNG

3000 2400
0.25
2500 2000
0.25
2000 1600 0.15
1500 0.15 1200 UinsulaƟon= 0.01 W/m2K
UinsulaƟon= 0.01 W/m2K 0.1
1000 0.1 800 0.05
0.05
500 0.02 400 0.02
0 0
0 10 20 30 40 50 60 70 80 0 4 8 12 16 20
Time [day] Time [day]

(c) (d)
Fig. 7. Effects of Uinsulation on unsaturated and saturated LNG (a)e(b) pressure, and (c)e(d) heat leak rate to tank (Other parameters are as given in Table 2).

240 140
Unsaturated LNG Unsaturated LNG
112
188

LCI
LNG holding time [day]
LNG holding time [day]

210
104

Saturated LNG 120 Saturated LNG

Aerogel
97
180 Polyurethane 100
87
blanket

82

78
150
MLI Cellular glass foam 80
120
87

60
90
30.7
50.4

28.2

26.0

23.4

21.9

20.5
60 40
23.4

40
10.6

22

16

20
13
5.7

30
4.1

11
3.3

2.8

0 0
0.01 0.02 0.05 0.10 0.15 0.20 0.25 -10 0 10 25 35 45
Uinsulation [W/m2K] Tambient [°C]
(a) (a)

3.5 Unsaturated LNG Unsaturated LNG

2.93

Average BOG rate [%/day]

Average BOG rate [%/day]

0.5
0.40
2.59

Saturated LNG
0.38

Saturated LNG
2.51

3.0
0.36
0.36

0.34
2.22

0.32

0.32

0.4
2.03

0.29

0.29

2.5
0.27

0.27
1.79

0.25

0.3
1.45

2.0
1.28

1.5 0.2
0.78
0.69

1.0
0.36
0.32
0.16
0.15

0.1
0.5
0.0 0.0
0.01 0.02 0.05 0.10 0.15 0.20 0.25 -10 0 10 25 35 45
Uinsulation [W/m2K] Tambient [°C]
(b) (b)
Fig. 8. Effects of Uinsulation on (a) LNG holding time and (b) average daily BOG gener- Fig. 9. Effects of Tambient on (a) LNG holding time and (b) average daily BOG generation
ation rate of a 57.2 m3 storage tank filled with unsaturated and saturated LNG (Other rate of a 57.2 m3 storage tank filled with unsaturated and saturated LNG (Other pa-
parameters are as given in Table 2). rameters are as given in Table 2).
 A. Sharafian et al. / Journal of Natural Gas Science and Engineering 36 (2016) 496e509 505

120 fuel throughput LNG refueling stations, one idea is to only refill half
Unsaturated LNG of the storage tanks to increase their LNG holding time. Fig. 10
LNG holding time [day]

95
100 Saturated LNG shows the effects of LNG filling level in a 57.2 m3 storage tank on

87
80
LNG holding time and average daily BOG generation rate. Initial

72
80 63
LNG level in the tank represents the volume percentage of the tank
60 occupied with LNG. As shown in Fig. 10a, increasing the initial LNG
level in the tank from 50% to 90% increases the holding times of
40

25
23
unsaturated and saturated LNG from 63 and 17 days to 95 and 25

21
19
17

20 days, respectively. Lower thermal mass of LNG, e.g., 50%, in the tank
causes higher BOG generation rate as shown in Fig. 10b. These data
0 disprove the idea of filling the tank partially in order to increase the
50% 60% 70% 80% 90% LNG holding time in refueling stations.
Initial LNG level in tank/LTank Smaller LNG storage tanks can be used in low throughput LNG
(a) refueling stations. Table 5 shows the dimensions of LNG storage
tanks with 21.84e107.32 m3 net capacity. The surface area to tank
0.7 net capacity ratio, Atank/Vtank, affects the BOG generation rate of
Average BOG rate [%/day]

Unsaturated LNG
0.53

0.6 LNG storage tanks (Adom et al., 2010); Larger tanks have smaller
0.48

Saturated LNG
0.46

Atank/Vtank as indicated in Table 5. This means that small Atank and

0.41

0.40

0.5
0.36

0.36

large thermal mass of LNG stored in storage tanks are desirable

0.32
0.32

0.4
0.3
0.2
0.1
0.0
50% 60% 70% 80%
Initial LNG level in tank/LTank
(b)
Fig. 10. Effects of initial LNG level in tank on (a) LNG holding time and (b) average daily
BOG generation rate of a 57.2 m3 storage tank filled with unsaturated and saturated
LNG (Other parameters are as given in Table 2).
In LNG refueling stations with high fuel deliver rate, Barclay et al.
(1998). suggested to use a single-wall tank insulated with tradi-
tional insulation materials and install an on-site liquefier to liquefy
the BOG. This idea is investigated in Section 4.4.
LNG refueling stations are exposed to different ambient tem-
peratures, Tambient, over the course of a year. Also, they are installed
in different geographic locations where Tambient can deviate from
the design point of LNG storage tanks. Fig. 9 shows the effects of
Tambient on the LNG holding time and average daily BOG generation
rate of the tank filled with unsaturated and saturated LNG.
Increase in Tambient from
10  C to 45  C decreases the unsatu-
rated LNG holding time from 112 to 78 days. For the saturated LNG,
the tank holding time decreases by 50% when Tambient increases
from
10  C to 45  C. It can be seen in Fig. 9b that Tambient of 45  C
causes the average daily BOG generation rates of 0.36% and 0.40% in
the tank filled with unsaturated and saturated LNG, respectively.
The thermal mass of LNG stored in the tank also affects the LNG
holding time and BOG generation rate (Migliore et al., 2015). In low
0.29
features to reduce the BOG generation rate and increase the LNG
holding time.
Fig. 11 demonstrates the holding time and average daily BOG
generation rate of LNG storage tanks ranging from 21.84 to
107.33 m3. It can be seen in Fig. 11a that a 21.84 m3 storage tank
filled with unsaturated LNG has a holing time 56% shorter than a
90% 107.33 m3 storage tank. Accordingly, the average daily BOG gener-
ation rates of the 21.84 m3 storage tank filled with unsaturated and
saturated LNG are equal to 0.43% and 0.48% which are higher than
those of the 107.33 m3 storage tank. This analysis shows that
storing LNG in a storage tank with the right size provides longer
holding time than storing the same amount of LNG in two or three
lower-capacity storage tanks.
The 57.2 m3 storage tank filled to 60% of its net capacity, as
shown in Fig. 10, holds the same amount of LNG as the 41.49 m3
storage tank filled to 80% of its net capacity, as shown in Fig. 11.
However, the 41.49 m3 storage tank holds the LNG for a longer time
with lower average daily BOG generation rate than the 57.2 m3
storage tank. This is because of the difference between the surface
area to LNG volume ratios, Atank/VLNG, of these tanks. The Atank/VLNG
of the 41.49 m3 storage tank is equal to 2.36 (¼ 78.19 m2/
(80%  41.49 m3)) whereas that of the 57.2 m3 storage tank is equal
to 3.04 (¼ 104.35 m2/(60%  57.20 m3)). It can be concluded that
when comparing storage tanks of different sizes, smaller Atank/VLNG
is preferred in order to achieving longer LNG holding time.
4.4. Performance analysis of LNG storage tanks with fuel delivery
In this section, the performance of a 57.20 m3 LNG storage tank
with fuel delivery to vehicles is investigated similar to the study of
Powars (2010). Three fleet sizes of 5, 10, and 20 vehicles with fuel
requirements of 1.89, 3.79, and 7.56 m3/day, respectively, are

Table 5
Specifications of different LNG storage tanks.

Parameter Storage tank sizes and specifications

Tank net capacity, Vtank (LNG storage vessels, 2016) 21.84 m3 41.49 m3 57.20 m3 107.32 m3
Length of inner tank, Ltank 9.29 m 9.17 m 12.64 m 19.0 m
Outer diameter, Do 1.978 m 2.896 m 2.896 m 2.93 m
Inner diameter, Di 1.73 m 2.4 m 2.4 m 2.682 m
Insulation thickness, tinsulation 0.248 m 0.248 m 0.248 m 0.248 m
Mass of storage tank (empty) 9525 kg 15,694 kg 21,455 kg 38.555 kg
Surface area of inner tank, Atank 55.19 m2 78.19 m2 104.35 m2 171.37 m2
Initial LNG level in storage tank 5.94 m 7.34 m 10.11 m 15.2 m
Surface area of inner tank to net capacity ratio, Atank/Vtank 2.53 1/m 1.88 1/m 1.82 1/m 1.60 1/m
506 A. Sharafian et al. / Journal of Natural Gas Science and Engineering 36 (2016) 496e509

modeled. In this model, it is assumed that each vehicle is fueled

120 Unsaturated LNG

100
with a 0.38 m3 of LNG. The fueling process is started every day at 6
LNG holding time [day]

Saturated LNG
100 p.m. and takes 5 min per vehicle with a fuel delivery rate of

87
84
0.076 m3/min. There is a 5 min time interval between the fueling of
80
64
two consecutive vehicles. For example, fueling process for a fleet
60 size of 5 vehicles takes 45 min (¼ 25 min fueling þ 20 min in-

26.6
23.4
22.6
40 tervals). Fig. 12 shows the unsaturated and saturated LNG pressures
17.0

and levels in the tank for daily fuel delivery rates of 1.89, 3.79, and
20
7.56 m3/day.
0 Fig. 12a and b show that unsaturated and saturated LNG pres-
21.84 41.49 57.20 107.33 sures do not reach the MAWP of the tank and decreases gradually,
(5,770) (10,960) (15,110) (28,350) especially for fuel delivery rates of 3.79 and 7.57 m3/day. LNG
Tank size [m3] or [gal] removal from the tank causes a portion of LNG evaporate to
displace the liquid volume. LNG gains heat from the LNG thermal
(a) mass to evaporate and the LNG temperature reduces. Accordingly,
the LNG pressure reduces. Fig. 12c and d show that the unsaturated
Average BOG rate [%/day]

0.6 Unsaturated LNG

0.48

and saturated LNG levels in the tank steadily decrease and the tank
Saturated LNG
0.43

0.5 is fully depleted without releasing the BOG to the atmosphere. It

0.37

0.36
0.33

should be noted that these data is true for the case of regular fuel
0.32

0.31
0.4
0.28

delivery with appropriate rate. In the case of low fuel delivery rates,
0.3 the LNG tank pressure can reach the MAWP of the tank, such as
0.2 ones reported by Chen et al. (Chen et al., 2004).
Fig. 13 shows the idea of using a single-wall storage tank insu-
0.1
lated with conventional insulation materials and no vacuum con-
0.0 dition, such as polyurethane, and daily fuel delivery rates of 1.89,
21.84 41.49 57.20 107.33 3.79, and 7.56 m3/day to fleet sizes of 5, 10, and 20 vehicles,
(5,770) (10,960) (15,110) (28,350) respectively.
Tank size [m3] or [gal] Uinsulation in this modeling is equal to 0.1 W/m2K corresponds to
that of polyurethane listed in Table 4. For the storage tank filled
(b) with unsaturated LNG, Fig. 13a indicates that fuel delivery rate of
1.89 m3/day fails to maintain the storage tank pressure below its
Fig. 11. Effects of storage tank size on (a) LNG holding time and (b) average daily BOG
generation rate for unsaturated and saturated LNG (Other parameters are as given in
MAWP. At this condition, 25% of unsaturated LNG is still available in
Table 2). the tank as shown in Fig. 13c. However, higher fuel delivery rates of

Fig. 12. Effects of fueling fleet sizes of 5, 10, and 20 vehicles on unsaturated and saturated LNG (a)e(b) pressure, and (c)e(d) level in tank (Uinsulation is equal to 0.022 W/m2K. Other
parameters are as given in Table 2).
 A. Sharafian et al. / Journal of Natural Gas Science and Engineering 36 (2016) 496e509 507

Fig. 13. Effects of fueling fleet sizes of 5, 10, and 20 vehicles on unsaturated and saturated LNG (a)e(b) pressure, and (c)e(d) level in tank (Uinsulation is equal to 0.1 W/m2K. Other
parameters are as given in Table 2).

3.79 and 7.56 m3/day would be sufficient to contain the BOG in the
tank with no release to the atmosphere.
Fig. 13b shows that the single-wall tank with Uinsulation of 0.1 W/
m2K and fuel delivery rates of 1.89 and 3.79 m3/day cannot hold
saturated LNG for more than 6.5 and 8.6 days, respectively, below
the MAWP of the tank. At these times, 62% and 27% of saturated
LNG is still available in the tank, as shown in Fig. 13d. These results
highlight that a single-wall storage tank cannot be used for LNG
applications if the on-site liquefier does not operate efficiently and
the refueling station has not a high fuel throughput greater than
7.56 m3/day. In this model, the insulation thickness corresponding
to Uinsulation of 0.1 W/m2K was equal to 0.248 m. In order to achieve
Uinsulation of 0.02 W/m2K (similar to that of the baseline model), the
insulation thickness made out of polyurethane should be 1.5 m
thick which is not practical. As a result, using a single-wall storage
tank with conventional insulation materials and on-site liquefier
cannot be a practical solution to reduce the BOG emissions and
upfront capital cost of equipment in LNG refueling stations.
 High vacuum insulation materials, such as LCI at vacuum pres-
sure of 133.3 Pa, assisted to increase the LNG holding time up to
188 days.
 Changes in the ambient temperatures (
10  Ce45  C) and
geographic location of refueling stations reduced LNG holding
time up to 50%.
 The ratio of heat transfer surface area to LNG volume, Atank/VLNG,
was introduced as a crucial factor in comparing the holding
times of storage tanks with different sizes.
 Storage tanks with proper vacuum insulation (Uinsulation of
0.022 W/m2K) could function with no BOG emissions under fuel
delivery rates as low as 1.89 m3/day.
 The single-wall tank insulated with polyurethane (Uinsulation of
0.1 W/m2K) could not hold BOG for fuel delivery rates less than
3.79 m3/day and was not a reliable solution for LNG refueling
stations.
Acknowledgment

The authors gratefully acknowledge the financial support of the

5. Conclusions Natural Sciences and Engineering Research Council of Canada
(NSERC), Mitacs Elevate, and Westport Power Inc.
In this study, the performance of LNG storage tanks in refueling
stations with and without fuel delivery were studied. Our main Nomenclature
findings are summarized in the following list:
A area (m2)
 Storage tanks filled with unsaturated LNG provided more than 3 BOG boil-off gas
times longer LNG holding time than those filled with saturated D diameter (m)
LNG. Dt time interval (s)
 The thermal mass of storage tanks had direct impact on LNG GHG greenhouse gas
holding time and the BOG generation rate. hfg heat of evaporation (J/kg)
508 A. Sharafian et al. / Journal of Natural Gas Science and Engineering 36 (2016) 496e509

k thermal conductivity (W/mK) Furuoka, F., 2016. Natural gas consumption and economic development in China
and Japan: an empirical examination of the Asian context. Renew. Sustain.
L length (m)
Energy Rev. 56, 100e115. http://dx.doi.org/10.1016/j.rser.2015.11.038.
LCI Layered composite insulation LNG Boil off Gas (BOG) Emissions Cannot Be Ignored, 2015. http://paper.people.com.
LNG liquefied natural gas cn/zgnyb/html/2015-12/21/content_1642385.htm (accessed May 6, 2016).
MAWP maximum allowable working pressure Global greenhouse gas emissions data, 2016. US Environ Prot Agency. https://www.
epa.gov/ghgemissions/global-greenhouse-gas-emissions-data (accessed April
MLI Multi-layer insulation 15, 2016).
NG natural gas Hailer, J.T., 2015. LNG Station Analysis for the Prediction of Pressure Rise and Vented
Q_
Emissions. West Virginia University. http://ezproxy.library.ubc.ca/login?
heat transfer rate (W) url¼http://search.proquest.com/docview/1752230020?accountid¼14656.
r density (kg/m3) Hao, H., Liu, Z., Zhao, F., Li, W., 2016. Natural gas as vehicle fuel in China: a review.
t time (s) Renew. Sustain. Energy Rev. 62, 521e533. http://dx.doi.org/10.1016/
j.rser.2016.05.015.
T temperature ( C)
Hasan, M.M.F., Zheng, A.M., Karimi, I.a., 2009. Minimizing boil-off losses in liquefied
U heat transfer coefficient (W/m2K) natural gas transportation. Ind. Eng. Chem. Res. 48, 9571e9580. http://
UA thermal conductance (W/K) dx.doi.org/10.1021/ie801975q.
Howarth, R.W., Santoro, R., Ingraffea, A., 2011. Methane and the greenhouse-gas
V volume (m3)
footprint of natural gas from shale formations. Clim. Change 106, 679e690.
http://dx.doi.org/10.1007/s10584-011-0061-5.
Imran, M., Yasmeen, T., Ijaz, M., Farooq, M., Wakeel, M., 2016. Research progress in
Subscripts
the development of natural gas as fuel for road vehicles: a bibliographic review
ambient ambient (1991e2016 ). Renew. Sustain. Energy Rev. 66, 702e741. http://dx.doi.org/
ceiling tank ceiling 10.1016/j.rser.2016.08.041.
holding time holding time International Gas Union, 2015. LNG as Fuel. 26th World Gas Conf, Paris, France,
pp. 1e120.
i ith time step, inner Washington, DC, U. S. Inventory of U.S. greenhouse gas emissions and sinks:
insulation insulation material 1990e2014, 2016. EPA 430-R-16-002.
o outer Kakaee, A.-H., Paykani, A., Ghajar, M., 2014. The influence of fuel composition on the
combustion and emission characteristics of natural gas fueled engines. Renew.
roof tank roof Sustain. Energy Rev. 38, 64e78. http://dx.doi.org/10.1016/j.rser.2014.05.080.
tank storage tank Khan, M.I., Yasmin, T., Shakoor, A., 2015. Technical overview of compressed natural
gas (CNG) as a transportation fuel. Renew. Sustain. Energy Rev. 51, 785e797.
http://dx.doi.org/10.1016/j.rser.2015.06.053.
References Kurle, Y.M., Wang, S., Xu, Q., 2015. Simulation study on boil-off gas minimization
and recovery strategies at LNG exporting terminals. Appl. Energy 156, 628e641.
2016-development status and investment strategy report of China LNG filling sta- http://dx.doi.org/10.1016/j.apenergy.2015.07.055.
tions consulting industry in 2022, 2015. Chi Institute Consulting Group (in Liimatainen, H., Kallionpa €€ € ll€
a, E., Po anen, M., Stenholm, P., Tapio, P., McKinnon, A.,
Chinese language). 2014. Decarbonizing road freight in the future - detailed scenarios of the carbon
21st session of the Conference of the Parties, 2015. The United Nations Framework emissions of Finnish road freight transport in 2030 using a Delphi method
Convention on Climate Change. http://dx.doi.org/FCCC/CP/2015/L.9. approach. Technol. Forecast. Soc. Change 81, 177e191. http://dx.doi.org/10.1016/
Adom, E., Islam, S.Z., Ji, X., 2010. Modelling of Boil-off Gas in LNG tanks: a case study. j.techfore.2013.03.001.
Int. J. Eng. Technol. 2, 292e296. Liu, S., Li, X., Huo, Y., Li, H., 2015. An analysis of the primary energy consumed by the
Advanced LNG onboard storage system, 2003. http://www.osti.gov/scitech/servlets/ Re-Liquefaction of boil-off gas of LNG storage tank. Energy Procedia 75,
purl/835120 (accessed September 19, 2016). 3315e3321. http://dx.doi.org/10.1016/j.egypro.2015.07.716.
Al Ali, M., 2015. Development of Novel Energy Systems for LNG Locomotives. Uni- Chart Industries LNG storage vessels, 2016. http://files.chartindustries.com/
versity of Ontario Institute of Technology. 20810915_VerticalStorage.pdf.
Alvarez, R.A., Pacala, S.W., Winebrake, J.J., Chameides, W.L., Hamburg, S.P., 2012. Lowell, D., Bradley, M.J., Wang, H., Lutsey, N., 2013. Assessment of the Fuel Cycle
Greater focus needed on methane leakage from natural gas infrastructure. Proc. Impact of Liquefied Natural Gas as Used in International Shipping. http://www.
Natl. Acad. Sci. U. S. A. 109, 6435e6440. http://dx.doi.org/10.1073/ theicct.org/sites/default/files/publications/ICCTwhitepaper_MarineLNG_
pnas.1202407109. 130513.pdf.
Æsøy, V., Einang, P.M., Stenersen, D., Hennie, E., Valberg, I., 2011. LNG-fuelled en- McJeon, H., Edmonds, J., Bauer, N., Clarke, L., Fisher, B., Flannery, B.P., et al., 2014.
gines and fuel systems for medium-speed engines in maritime applications. Limited impact on decadal-scale climate change from increased use of natural
SAE Tech. Pap. http://dx.doi.org/10.4271/2011-01-1998. gas. Nature 514, 482e485. http://dx.doi.org/10.1038/nature13837.
Aspen Engineering (Version 8.8), 2016. http://www.aspentech.com/. Miana, M., Del Hoyo, R., Rodriga
lvarez, V., Valde
s, J.R., Llorens, R., 2010. Calculation
Aspen Plus Dynamics User Guide (Version 8.8), 2016. http://www.aspentech.com/. models for prediction of Liquefied Natural Gas (LNG) ageing during ship

Balitskiy, S., Bilan, Y., Strielkowski, W., Streimikien _ D., 2016. Energy efficiency and
e, transportation. Appl. Energy 87, 1687e1700. http://dx.doi.org/10.1016/
natural gas consumption in the context of economic development in the Eu- j.apenergy.2009.10.023.
ropean Union. Renew. Sustain. Energy Rev. 55, 156e168. http://dx.doi.org/ Miana, M., Legorburo, R., Díez, D., Hwang, Y.H., 2015. Calculation of boil-off rate of
10.1016/j.rser.2015.10.053. liquefied natural gas in Mark III tanks of ship carriers by numerical analysis.
Barclay, M.A., Corless, A.J., Nelson, E.H., 1998. Optimized lng storage tanks for fleet- Appl. Therm. Eng. 93, 279e296. http://dx.doi.org/10.1016/
size refueling stations with local lng liquefiers. In: Adv. Cryog. Eng. Springer j.applthermaleng.2015.09.112.
Science & Business Media, pp. 1199e1206. Migliore, C., Tubilleja, C., Vesovic, V., 2015. Weathering prediction model for stored
Bassi, A., 2011. Liquefied natural gas ( LNG ) as fuel for road heavy duty vehicles liquefied natural gas (LNG). J. Nat. Gas. Sci. Eng. 26, 570e580. http://dx.doi.org/
technologies and standardization. SAE Tech. Pap. http://dx.doi.org/10.4271/ 10.1016/j.jngse.2015.06.056.
2011-24-0122. Nicotra, A., 2013. LNG, a Sustainable Fuel for All Transport Modes.
Burnham, A., Han, J., Elgowainy, A., Wang, M., 2015. Updated Fugitive Greenhouse Park, C., Song, K., Lee, S., Lim, Y., Han, C., 2012. Retrofit design of a boil-off gas
Gas Emissions for Natural Gas Pathways in the GREET1_2015 Model. handling process in liquefied natural gas receiving terminals. Energy 44, 69e78.
Chen, Q.S., Wegrzyn, J., Prasad, V., 2004. Analysis of temperature and pressure http://dx.doi.org/10.1016/j.energy.2012.02.053.
changes in liquefied natural gas (LNG) cryogenic tanks. Cryog. Guildf. 44, Pellegrini, L.A., Moioli, S., Brignoli, F., Bellini, C., 2014. LNG technology: the weath-
701e709. http://dx.doi.org/10.1016/j.cryogenics.2004.03.020. ering in above-ground storage tanks. Ind. Eng. Chem. Res. 53, 3931e3937.
Davis, S.J., Shearer, C., 2014. Climate change: a crack in the natural-gas bridge. http://dx.doi.org/10.1021/ie404128d.
Nature 514, 436e437. http://dx.doi.org/10.1038/nature13927. Powars, C.A., 2010. Best Practices to Avoid LNG Fueling Station Venting Losses. St.
Delgado, O., Muncrief, R., 2015. Assessment of Heavy-Duty Natural Gas Vehicle Croix Research report prepared for Brookhaven National Laboratory.
Emissions: Implications and Policy Recommendations. Querol, E., Gonzalez-Regueral, B., García-Torrent, J., García-Martínez, M.J., 2010. Boil
Dunn, M.E., LeBlanc, V.N., 2014. Two Engine System with a Gaseous Fuel Stored in off gas (BOG) management in Spanish liquid natural gas (LNG) terminals. Appl.
Liquefied Form, US Pat. 8,763,565. Energy 87, 3384e3392. http://dx.doi.org/10.1016/j.apenergy.2010.04.021.
Brussels, Belgium. Energy Carriers for Powertrainsefor a clean and efficient Richter, B., 2014. Beyond smoke and mirrors. Climate Change and Energy in the 21st
mobility, 2014. www.ertrac.org (accessed March 8, 2016). Century. Cambridge University Press.
Fahmy, M.F.M., Nabih, H.I., El-Rasoul, T.A., 2015. Optimization and comparative Roche, G., 2009. Fueling infrastructure for natural gas trucks. In: Faster Freighte-
analysis of LNG regasification processes. Energy 91, 371e385. http://dx.doi.org/ Clean. Air Calif. 2009 Conf., Long Beach, CA.
10.1016/j.energy.2015.08.035. Seo, S., Han, S., Lee, S., Chang, D., 2016. A pump-free boosting system and its
Fesmire, J.E., Augustynowicz, S.D., 2005. Cryogenic Thermal Insulation Systems, in: application to liquefied natural gas supply for large ships. Energy 105, 70e79.
16th Therm. Fluids Anal. Work., Cryogenics Test Laboratory. NASA Kennedy http://dx.doi.org/10.1016/j.energy.2015.09.052.
Space Center, Orlando, Florida, U.S. Şevik, S., 2015. An analysis of the current and future use of natural gas-fired power
 A. Sharafian et al. / Journal of Natural Gas Science and Engineering 36 (2016) 496e509
plants in meeting electricity energy needs: the case of Turkey. Renew. Sustain.
Energy Rev. 52, 572e586. http://dx.doi.org/10.1016/j.rser.2015.07.102.

rida, W., 2016. A review of
Sharafian, A., Talebian, H., Blomerus, P., Herrera, O., Me
liquefied natural gas refueling station designs. Renew. Sustain. Energy Rev.
(under review).
Society of Automotive Engineers (SAE) Standard J2343: Recommended practice for
LNG medium and heavy-duty powered vehicles, 2008. http://standards.sae.org/
j2343_200807/.
Sources of greenhouse gas emissions, 2016. US Environ. Prot. Agency. https://www.
epa.gov/ghgemissions/sources-greenhouse-gas-emissions (accessed September
12, 2016).
Solomon, S. (Ed.), 2007. Climate change 2007-the physical science basis: Working
group I contribution to the fourth assessment report of the IPCC, vol. 4. Cam-
bridge University Press.
Study on natural gas research and development priority setting for transportation
in Canada, 2014. Canadian Natural Gas Vehicle Alliance.
Van Den Broek, M., Berghout, N., Rubin, E.S., 2015. The potential of renewables
versus natural gas with CO2 capture and storage for power generation under
CO2 constraints. Renew. Sustain. Energy Rev. 49, 1296e1322. http://dx.doi.org/
10.1016/j.rser.2015.04.089.
van Der Hoeven, M., 2015. CO2 Emissions from Fuel Combustion IEA Statistics.
http://dx.doi.org/10.1787/co2-table-2011-1-en.
509
Venkatesh, A., Jaramillo, P., Griffin, W.M., Matthews, H.S., 2011. Uncertainty in life
cycle greenhouse gas emissions from United States natural gas end-uses and its
effects on policy. Environ. Sci. Technol. 45, 8182e8189. http://dx.doi.org/
10.1021/es200930h.
Wang, Q., Li, R., 2016. Natural gas from shale formation: a research profile. Renew.
Sustain. Energy Rev. 57, 1e6. http://dx.doi.org/10.1016/j.rser.2015.12.093.
Wang, T., Lin, B., 2014. Impacts of unconventional gas development on China's
natural gas production and import. Renew. Sustain. Energy Rev. 39, 546e554.
http://dx.doi.org/10.1016/j.rser.2014.07.103.
Wang, Q., Chen, X., Jha, A.N., Rogers, H., 2014. Natural gas from shale formation - the
evolution, evidences and challenges of shale gas revolution in United States.
Renew. Sustain. Energy Rev. 30, 1e28. http://dx.doi.org/10.1016/
j.rser.2013.08.065.
Wang, J., Jiang, H., Zhou, Q., Wu, J., Qin, S., 2016. China's natural gas production and
consumption analysis based on the multicycle Hubbert model and rolling Grey
model. Renew. Sustain. Energy Rev. 53, 1149e1167. http://dx.doi.org/10.1016/
j.rser.2015.09.067.
Wiens, J., Powars, C., Pope, G., 2001. LNG vehicle fuel pressure strategy alternatives.
SAE Tech. Pap. http://dx.doi.org/10.4271/2001-01-1919.
Yan, G., Gu, Y., 2010. Effect of parameters on performance of LNG-FPSO offloading
system in offshore associated gas fields. Appl. Energy 87, 3393e3400. http://
dx.doi.org/10.1016/j.apenergy.2010.04.032.