Cryogenics 72 (2015) 127–147

Contents lists available at ScienceDirect

Cryogenics
journal homepage: www.elsevier.com/locate/cryogenics

Review

A thermodynamic review of cryogenic refrigeration cycles for

liquefaction of natural gas
Ho-Myung Chang ⇑
Hong Ik University, Seoul 121-791, Republic of Korea

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

Article history: A thermodynamic review is presented on cryogenic refrigeration cycles for the liquefaction process of
Received 20 July 2015 natural gas. The main purpose of this review is to examine the thermodynamic structure of various cycles
Received in revised form 30 September 2015 and provide a theoretical basis for selecting a cycle in accordance with different needs and design criteria.
Accepted 5 October 2015
Based on existing or proposed liquefaction processes, sixteen ideal cycles are selected and the optimal
Available online 14 October 2015
conditions to achieve their best thermodynamic performance are investigated. The selected cycles
include standard and modified versions of Joule–Thomson (JT) cycle, Brayton cycle, and their combined
Keywords:
cycle with pure refrigerants (PR) or mixed refrigerants (MR). Full details of the cycles are presented and
Thermodynamics
Refrigeration cycle
discussed in terms of FOM (figure of merit) and thermodynamic irreversibility. In addition, a new method
LNG of nomenclature is proposed to clearly identify the structure of cycles by abbreviation.
Natural gas Ó 2015 Elsevier Ltd. All rights reserved.
Liquefaction

Contents

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127
2. Definition and performance index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129
3. Classification and proposed nomenclature of refrigeration cycles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 130
3.1. Cycles by expansion process. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 130
3.2. Cycles by type of refrigerant. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 130
3.3. Proposed nomenclature of refrigeration cycles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 130
4. Thermodynamic performance of ideal cycles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131
4.1. JT cycles with pure refrigerants (PR) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 133
4.2. JT cycles with mixed refrigerants (MR) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 133
4.3. JT cycles with pure refrigerants (PR) and mixed refrigerants (MR) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 135
4.4. Brayton cycles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 136
4.5. Combined and Claude cycles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 139
5. Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 142
6. Concluding remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145
Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 146
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 146

1. Introduction of high-density storage and long-distance transport, natural gas

is liquefied to cryogenic liquid called LNG (liquefied natural gas)
Liquefaction of natural gas is one of the most important ther- in a large based-load plant near gas reservoir. The liquefaction of
modynamic processes in cryogenic gas industry. For the purpose natural gas is different in thermodynamic structure from that of
other cryogenic gases, such as air (nitrogen and oxygen), hydrogen,
⇑ Tel.: +82 2 320 1675. or helium [1,2], as illustrated in Fig. 1. In general, a refrigeration
E-mail address: [email protected] system is used to remove a thermal load (Q_ L ) from

http://dx.doi.org/10.1016/j.cryogenics.2015.10.003
0011-2275/Ó 2015 Elsevier Ltd. All rights reserved.
128 H.-M. Chang / Cryogenics 72 (2015) 127–147

Nomenclature

FOM figure of merit Subscripts

h specific enthalpy 0 ambient
m _ mass flow rate 1, 2, 3, . . .location in refrigeration cycle
Q_ heat transfer rate AC after-cooler
s specific entropy C compressor
S_ gen entropy generation rate E expander
T temperature F natural gas feed
U overall heat transfer coefficient HX heat exchanger
W _ power or work rate JT JT valve
L LNG or liquefied natural gas
max maximum
Greek letters min minimum
e heat exchanger effectiveness MIX mixing device

low-temperature source, and it works in a closed cycle, as shown in A variety of liquefaction processes have been developed and
Fig. 1(a). The refrigeration requires a cost of net input power patented over the past half century, and further effort still contin-
(W_ C
W_ E ), and heat (Q_ 0 ) must be rejected to ambient. A liquefac- ues for improving the processes in efficiency, capacity, or reliabil-
tion system is used to receive gas at ambient temperature and deli- ity. Out of the numerous liquefaction processes, however, only a
ver liquid at cryogenic temperature, and it works in an ‘‘open” few are under operation in practice, and one dominant process
cycle (strictly speaking, not a complete cycle), as shown in Fig. 1 (widely called C3-MR or propane-precooled mixed-refrigerant pro-
(b). In most cases, the gas itself is the working fluid that undergoes cess) has been considered the reasonable choice for large-scale liq-
the process of compression and expansion. The liquefaction also uefaction. It was only several years ago that many plant engineers
requires a cost of input power, and heat must be rejected to ambi- began to look for different options in refrigeration cycles, because
ent. On the other hand, a natural gas liquefaction system works the most suitable process may be different for smaller-scale lique-
basically like a closed refrigeration cycle, but the thermal load is faction or under particular conditions, such as peak-shaving plants,
distributed over a temperature range of the natural gas (NG) flow offshore or floating plants, re-liquefaction of boil-off gas, and bio-
from ambient to cryogenic LNG temperature, as shown in Fig. 1 gas/waste application. In accordance with the recent needs, a few
(c). In order to clarify this structural difference, this review is titled monographs and review articles have been published up to date
with the term, ‘‘refrigeration cycles for liquefaction of natural gas.” to compare various LNG processes under operation or for future
Natural gas (NG) is a mixture of different hydrocarbons and development [3–12]. This paper is another review with the same
minor gases. Along the flow of liquefaction, its specific heat varies motivation, but has been prepared with unique features in the fol-
significantly over temperature, depending on the pressure and lowing three aspects.
composition. Fig. 2 shows the temperature–entropy (T–s) diagram First, the main objective of this review is to examine the struc-
and the specific heat (at constant pressure) as a function of tem- ture of cryogenic refrigeration cycles from a thermodynamic point
perature for NG with a composition of 1% nitrogen, 91% methane, of view. As thermodynamics is ‘‘the science of energy and entropy”
5% ethane, 2% propane, 0.6% n-butane, and 0.4% i-butane (on mole aiming at efficient energy conversion [13,14], it is intended to
basis). As the isobar of 5 MPa is indicated by bold curve, for exam- describe how efficiently a variety of refrigeration cycles work with
ple, the liquefaction flow of NG can be divided into three regions by different refrigerants. The refrigeration cycles include possible
the phase: pre-cooling (vapor), condensation (phase change), and modifications and proposals as well as fully developed and proven
sub-cooling (liquid). The specific heat has a sharp peak around ones. Thermodynamic efficiency (to be defined as FOM) is the
200 K, which makes the cooling load uneven along the NG flow. major index to evaluate the cycles, and the loss of thermodynamic
In order to achieve a high thermodynamic efficiency, it is crucial availability (called the irreversibility) is itemized by the
to minimize the entropy generation due to the temperature differ- contribution of each component in the cycle. Other technical
ence between the refrigeration cycle and the NG flow. issues involved in practical liquefaction process (such as the

Fig. 1. Energy and mass flow in cryogenic refrigeration and liquefaction cycles.
 H.-M. Chang / Cryogenics 72 (2015) 127–147 129

Fig. 2. Temperature–entropy (T–s) diagram and specific heat as a function of temperature for NG with a composition of 1% nitrogen, 91% methane, 5% ethane, 2% propane,
0.6% n-butane, and 0.4% i-butane.

pre-treatment of natural gas, the power supply to compressors, or expander work), respectively. The expander work W _ E is applicable
the separation or recovery of gases or liquids) are not directly only to the cycles that have work-producing expanders such as a
considered. turbine. In practice, the expander work may be used for compress-
Second, the ‘‘ideal” refrigeration cycles are the major focus of ing the refrigerant or simply dissipated with a braking device, but
thermodynamic analysis. In theory, there exists an absolute (and ðW_ C
W_ E Þ is considered as the input. A liquefaction system is usu-
unique) minimum in the input power to liquefy a unit mass of nat- ally evaluated in terms of the work per unit mass of liquefied gas,
ural gas. A real liquefaction system needs more power input than _ C
W_ E Þ=m
_ F . For thermodynamic evaluation, however, a dimen-
ðW
the theoretical minimum for many different reasons. These reasons
sionless parameter (valued between 0 and 1) should be adopted,
can be sorted into two groups: the reasons from the refrigeration
and the FOM (figure of merit) is such a performance index defined
cycle itself and the reasons from the imperfect performance of
by the combined first and second laws.
components in the cycle. The former is intrinsic to the selected
From the second law, the entropy balance equation for the sys-
thermodynamic cycle and refrigerants, while the latter is about
tem is written as:
the technical issue associated with economic factors. In engineer-
ing education on thermal design, it is common to introduce an Q_
_ F ðs0
sL Þ þ S_ gen ¼ 0
m ð2Þ
ideal cycle first and then consider the deviation of practical cycles T0
with the effect of components [13,14]. An ideal cycle here means
where s is the specific entropy of NG feed, and T 0 is the ambient
that every component in the cycle is perfect, and the second group
of reasons is excluded. Specifically, it will be assumed that the temperature where Q_ 0 is rejected. The entropy generation rate,
effectiveness of all heat exchangers is 100% (or the minimum tem- S_ gen , is zero in a reversible system, but has a positive value in prac-
perature approach is zero), the pressure drop in all flow compo- tical systems. By eliminating Q_ 0 from Eqs. (1) and (2), the work
nents is zero, and the adiabatic efficiency of all compressors and input is expressed as:
expanders is 100% (or the process is isentropic).
Thirdly, the cryogenic refrigeration cycles are systematically
_ C
W
W _ F ½ðhL
h0 Þ
T 0 ðsL
s0 Þ þ T 0 S_ gen
_ E¼m ð3Þ
classified, and a new method of nomenclature is proposed in accor-
Since S_ gen is non-negative, the minimum work required for the
dance with the thermodynamic classification. Over decades, many
liquefaction is
liquefaction processes have been called by their refrigerant or key
component (e.g. C3-MR or Nitrogen-Expander process), by techni- _ min ¼ m
W _ F ½ðhL
h0 Þ
T 0 ðsL
s0 Þ ð4Þ
cal terms (e.g. Single MR, Dual MR, Optimized Cascade, or MFC), or
which is the absolute minimum given by the thermodynamic limit.
even by proper nouns (e.g. PRICO, AP-X, or LIQUEFIN). The system-
The bracket in Eq. (4) is the flow availability or exergy of LNG. The
atic classification with new nomenclature will be able to clearly
FOM of a liquefaction system is defined as the ratio of the minimum
identify the structure of refrigeration cycles and simply name the
work to the actual work.
cycle with abbreviated characters and symbols.
_ min
W ðhL
h0 Þ
T 0 ðsL
s0 Þ
FOM ¼ ¼ ð5Þ
2. Definition and performance index _
WC
WE_ _ C
W
ðW _ E Þ=m_F
The difference between the actual and minimum works is the
For the entire liquefaction system shown in Fig. 1(c), including
entropy generation rate multiplied by the ambient temperature
the refrigeration cycle and the NG stream from ambient tempera-
ture (subscript 0) to the LNG temperature (subscript L), the first _ C
W
ðW _ min ¼ T 0 S_ gen
_ EÞ
W ð6Þ
law or the energy balance equation is written as:
which is called the irreversibility or the loss of availability (exergy).
_ F ðh0
hL Þ ¼ Q_ 0
ðW
m _ C
W
_ EÞ ð1Þ In Eq. (3), the entropy generation rate can be itemized by the con-
tribution of each component involved in the system.
where m _ F and h are the mass flow rate and specific enthalpy of NG X X X X
S_ gen ¼ ðS_ gen ÞHX þ ðS_ gen ÞJT þ ðS_ gen ÞMIX þ ðS_ gen ÞAC ð7Þ
feed, respectively, and Q_ 0 and ðW
_ C
W_ E Þ are the heat rejected to
HX JT MIX AC
ambient and the work input (the compressor work minus the
130 H.-M. Chang / Cryogenics 72 (2015) 127–147

where HX, JT, MIX, and AC denote heat exchangers, JT valves, mixing cycle”, whose standard structure for liquefaction is shown in
devices, and after-coolers, respectively. It is noted that there is no Fig. 3(b). In order to safely produce work in the expander, it is
entropy generation in ideal compressors (C) and expanders (E). important that the refrigerant should remain in ‘‘gas phase” (or
at least ‘‘vapor-rich” saturated state) throughout the cycle.
3. Classification and proposed nomenclature of refrigeration Both JT and adiabatic expansions can be used in a refrigeration
cycles system, and ‘‘Claude cycle” is one of the examples, whose standard
structure is shown in Fig. 3(c). High-pressure refrigerant is
The refrigeration cycles for NG liquefaction are classified in two branched into two streams, and then one is expanded through an
different ways – by the principle of expansion processes to produce expander in gas phase and the other is further cooled-down to
low temperatures and by the type of working fluids (refrigerants) liquid and finally expanded through a JT valve to the cold end.
used in the cycles. Table 1 lists the refrigeration cycles in a two- Alternatively, one or more JT cycles and Brayton cycles may be
dimensional way in accordance with this classification. combined in many different ways.

3.1. Cycles by expansion process

3.2. Cycles by type of refrigerant
Two distinct expansion processes are used in cryogenic refriger-
Refrigeration cycles can be classified by the type of refrigerants
ation – JT (Joule–Thomson) expansion and adiabatic expansion [1].
used in the cycles – pure refrigerant (PR) and mixed refrigerant
The JT expansion is a throttling process through flow resistance,
(MR). The PR cycles use a single-component fluid, such as propane
such as a valve or a porous plug. Since neither heat nor work is
or nitrogen. The MR cycles use a mixture of multi-component flu-
transferred, the enthalpy of fluid remains the same so that this
ids, for example, a mixture of nitrogen, methane, ethane, and pro-
expansion process is modeled as isenthalpic. The JT expansion is
pane. During the phase change (evaporation or condensation)
an irreversible process, where the cooling effect is strongly depen-
along an isobar, temperature is constant for PR, but varies contin-
dent on the thermodynamic property of refrigerant. A refrigeration
uously for MR.
cycle with JT expansion is called ‘‘JT cycle”, whose standard
In general, the PR cycles are simple and easy to operate, but require
structure for liquefaction is shown in Fig. 3(a). High-pressure
multi-staged refrigeration to achieve a high efficiency by reducing the
refrigerant is cooled in a counter-flow heat exchanger (HX) to
temperature difference between the NG feed and PR in HX’s, as
liquid, and then expanded through a JT valve typically into
explained later. On the contrary, the MR cycles could be thermody-
‘‘two-phase” region for the return flow of cold refrigerant at low
namically more efficient with a small number of components, but
pressure. The cold HX at the bottom may or may not be used,
requires a sophisticated engineering in the design and operation of
depending on the thermodynamic property of refrigerant.
cycles. In many cases, a liquefaction system is composed of two or
On the contrary, the adiabatic expansion is a working-
more separate cycles, which can be a combination of PR and MR cycles
producing process with expanders, such as a turbine or an
to take advantage of the two cycles at the same time.
expansion engine. In an ideal case, the expansion is reversible
and adiabatic so that this expansion is modeled as isentropic. In
general, the adiabatic expansion is thermodynamically more 3.3. Proposed nomenclature of refrigeration cycles
efficient than the JT expansion, but the cryogenic expanders are
costly and difficult to operate. A refrigeration cycle composed of In accordance with the classification, a new method of nomen-
isentropic and isobaric processes is called ‘‘(reversed) Brayton clature is proposed to clearly and simply identify the structure of

Table 1
Classification of sixteen refrigeration cycles with proposed nomenclature.

Pure refrigerants (PR) Mixed refrigerants (MR) Pure and mixed refrigerants (PR
+ MR)
JT (Joule–Thomson) cycles (1) p3J + e0 3J + m3J (2) M1J (7) p4J + M2J
(3) M2J (8) b1J/e1J + M2J
(4) M1J + M1J
(5) M1J + M2J
(6) M2J + M2J
Brayton cycles (9) n1B (13) M1B
(10) n2B
(11) n1B + n1B
(12) m1B + n1B
Combined or Claude cycles (14) p3J + e0 3J + n1B (15) M1C (16) p4J + M1J + n1B

(1) 3-stage propane JT, 3-stage ethylene JT, and 3-stage methane JT cycles p3J + e0 3J + m3J Cascade
(2) 1-stage MR JT cycle M1J SMR
(3) 2-stage MR JT cycle M2J SMR with PS
(4) Dual 1-stage MR JT cycles M1J + M1J PRICO
(5) 1-stage MR JT, 2-stage MR JT cycles M1J + M2J DMR-1
(6) Dual 2-stage MR JT cycles M2J + M2J DMR-2
(7) 4-stage propane JT, and 2-stage MR JT cycles p4J + M2J C3-MR
(8) Parallel 1-stage butane JT, 1-stage ethane JT, and 2-stage MR JT cycles b1J/e1J + M2J
(9) 1-stage nitrogen Brayton cycle n1B N turbine
(10) 2-stage nitrogen Brayton cycle n2B
(11) Dual 1-stage nitrogen Brayton cycles n1B + n1B Dual N turbine
(12) 1-stage methane Brayton, and 1-stage nitrogen Brayton cycles m1B + n1B
(13) 1-stage MR Brayton cycle M1B
(14) 3-stage propane JT, 3-stage ethylene JT, and 1-stage nitrogen Brayton cycles p3J + e0 3J + n1B C3–C2–N2
(15) 1-stage MR Claude cycle M1C NG Claude
(16) 4-stage propane JT, 1-stage MR JT, and 1-stage nitrogen Brayton cycles p4J + M1J + n1B AP-X
 H.-M. Chang / Cryogenics 72 (2015) 127–147 131

Fig. 3. Standard JT, Brayton, and Claude cycles for liquefaction of natural gas (C: compressor, AC: after-cooler, HX: heat exchangers, JT: Joule–Thomson valve, E: expander).

Fig. 4. Examples to illustrate the proposed nomenclature of refrigeration cycles.

cycles. As illustrated in Fig. 4, a refrigeration cycle is denoted by 3-stage ethylene, and 3-stage methane JT cycles is denoted by p3J
three characters listed in order of an alphabet, a number, and + e0 3J + m3J, and more specifically by (p3J + (e0 3J) + m3J), meaning
another alphabet, which denote the refrigerant, the number of that the ethylene of e0 3J is pre-cooled by P3J and the methane of
stages, and the cycle type, respectively. m3J is pre-cooled by e0 3J, which is the structure of conventional
The refrigerant is represented by its first lower-case letter, such cascade liquefaction process described later.
as propane by hpi, and nitrogen by hni. A prime is added over the Sixteen refrigeration cycles were selected and listed in Table 1
alphabet, if a hydrocarbon is unsaturated, such as he0 i for ethylene for detailed examination. As described below, the selected
to distinguish from hei for ethane. The only exceptional use of a cycles represent the simplified version of existing or proposed
capital letter is hMi, which stands for mixed refrigerant to distin- liquefaction processes. Among the numerous proposed cycles, only
guish from hmi for methane. The third capital letter represents a limited number of cycles were selected mainly to examine how
the cycle by the type of expansion, which has three options: hBi the modified structure affects the thermodynamic performance.
for Brayton cycle, hCi for Claude cycle, and hJi for Joule–Thomson The primary modification under consideration here includes the
cycle. The cycle type here includes all standard or modified cycles. multi-staged refrigeration and the combination of multiple cycles.
It is recalled that a JT cycle contains JT expansion only, a Brayton The effect of different refrigerants is also examined, but only a few
cycle contains adiabatic expansion only, and a Claude cycle con- limited choices are included and other feasible ones are briefly
tains both JT and adiabatic expansions. For example, M2J stands mentioned in the text as necessary or appropriate. It is recalled
for 2-stage MR JT cycle, which has the same structure as MR pro- that this review focuses on the thermodynamic nature of cycles,
cess with a phase separator, as described later. instead of technical issues in practical liquefaction processes. In
When two or more separate cycles are involved, their combina- Table 1, the sixteen cycles are classified two-dimensionally by
tions are indicated by either + (series) or/(parallel). The top (i.e. the expansion process (as JT, Brayton, and combined or Claude
warm) cycle comes first, and then the bottom (i.e. colder) cycles cycles) and by the type of refrigerants (as PR, MR, and both).
follow. For example, p4J + M1J + n1B stands for a series combina- The full thermodynamic names for each cycle are footnoted with
tion of 4-stage propane JT cycle, 1-stage MR JT cycle, and 1-stage the commercial or conventional terms of corresponding or closely
nitrogen Brayton cycle, which is known as the AP-X process under related processes.
operation at the largest-capacity plants. In some cases, the refriger-
ant of bottom cycle is pre-cooled by the top cycle, which can be 4. Thermodynamic performance of ideal cycles
optionally indicated by grouping the cycles with (parenthesis).
For example, (p4J + M1J) + n1B means that the MR of M1J cycle is Details of thermodynamic performance are presented for the
pre-cooled by p4J cycle, but n1B cycle works independently from sixteen cycles listed in Table 1. As mentioned above, only ideal
two JT cycles on the top. A series combination of 3-stage propane, cycles are considered in order to focus on the thermodynamic
132 H.-M. Chang / Cryogenics 72 (2015) 127–147

Fig. 5. 3-stage propane JT, 3-stage ethylene JT, and 3-stage methane JT cycles (p3J + e0 3J + m3J).
 H.-M. Chang / Cryogenics 72 (2015) 127–147 133

cycle, but ethylene is superior to ethane because of its lower boil-

ing temperature.
Fig. 5 shows an example of 3-stage propane JT, 3-stage ethylene
JT, and 3-stage methane JT cycles (p3J + e0 3J + m3J). At each cooling
stage, propane is directly fed into compressors, but ethylene and
methane are super-heated before entering the compressors. As
noted in T–s diagram and temperature profiles in HX’s, the cooling
temperature at each stage is determined by the corresponding sat-
uration pressure in each JT cycle. Since the highest and lowest
pressures are fixed, two intermediate pressures are optimized.
The maximum FOM is 69.8% for p3J + e0 3J + m3J with optimized
intermediate pressures. The major portion of irreversibility occurs
in HX’s due to the stepwise temperature profile.
The number of stages in each JT cycle is largely an economic
consideration. Fig. 6 shows how the FOM increases as the total
number of stages increases from 6 (=2 + 2 + 2) to 12 (=4 + 4 + 4),
assuming that the ambient temperature is fixed at 298 K. The three
numbers inside the column (in format of p + e0 + m) indicate the
Fig. 6. FOM of propane–ethylene–methane (p + e0 + m) JT cycles with different number of stages in the top (p), middle (e0 ), and bottom (m) cycles,
number of stages. respectively. As the total number of stages increases, the FOM
increases more or less linearly, but the number of required compo-
nents should increase accordingly. Depending on the operating and
structure and properties of refrigerants. The following assumptions capital costs, the number of stages is determined typically as 3 in
are made for cycle analysis. each JT cycle, but 10 (=4 + 3 + 3) stages are used in tropical climate.

 The ambient temperature is 298 K. 4.2. JT cycles with mixed refrigerants (MR)
 The natural gas (NG) feed has a constant composition (1% nitro-
gen, 91% methane, 5% ethane, 2% propane, 0.6% n-butane, and The structure of JT refrigeration cycle can be significantly sim-
0.4% i-butane on mole basis) at 5 MPa, and the inlet and exit plified by using mixed refrigerants (MR). Fig. 7 shows the simplest
temperatures are 298 K and 113 K, respectively. 1-stage MR JT cycle (M1J), as used in the SMR process [22]. The
 The mass flow rate of NG (that is, the liquefaction rate) is 1 kg/s. performance of M1J is strongly dependent on the composition
 The pressure drop in all heat exchangers (HX), after-coolers and pressure levels of MR. There have been a great number of
(AC), separators, and mixing devices (MIX) is zero. reports to find the optimal conditions that maximize the FOM.
 The effectiveness of all HX’s is 100%, or the minimum tempera- Basically, the optimization needs an empirical and iterative proce-
ture approach between the hot and cold streams is 0 K. dure, but a somewhat robust approach has been developed, such as
 The adiabatic efficiency of all compressors (C) and expanders (E) the sequential quadratic programming (SQP) [3,23] or the nonlin-
is 100%. ear programming (NLP) [24–26]. Recently, a generic algorithm
 The maximum and minimum pressures of refrigerants are lim- (GA) [27–29] and knowledge-based optimization (KBO) [30] are
ited to 10 MPa and 0.1 MPa, respectively. introduced to considerably reduce the number of iterations.
 The number of compression stages is determined such that the The maximum FOM is 57.1% for M1J with the optimized pres-
pressure ratio at each stage is 2–3, and an AC is placed at each sure and composition of MR. The minimum temperature approach
compression stage whose exit temperature is above 298 K. (0 K) is found at an intermediate temperature as well as two ends.
 The exit condition of pure refrigerants (PR) is saturated vapor in The optimized composition (n + m + e + p) of MR is constant over
all evaporating HX’s. the cycle, as listed at the bottom of thermodynamic table. The tem-
perature difference in HX is quite small in the low-temperature
The cycle analysis is performed with a process simulator (Aspen region, but the irreversibility in HX is large due to the temperature
HYSYS Version 8). The thermodynamic properties of refrigerants mismatch near the warm end of HX.
and NG are calculated with the Peng–Robinson equation of state In order to improve the thermodynamic performance, there are
(EOS) linked to the simulator. For each cycle, the results are several schemes to modify the MR JT cycle. An important modifica-
presented as a table of thermodynamic states and graphically in tion is to arrange the cycle for 2-stage refrigeration with a phase
temperature–entropy diagram (excluding the compression and separator (symbolized by a small circle) as shown in Fig. 8
after-cooling processes), temperature profiles in HX’s, and exergy [31,32], which will be called 2-stage MR JT cycle (M2J). From the
expenditure (i.e. the composition of FOM and irreversibility ratio phase separator, liquid with heavy components is expanded at an
by components). The temperature profiles in multi-stream HX’s intermediate temperature, and vapor with lighter components is
are the composite curves of hot and cold streams. expanded at the cold end. It is noted that two cooling streams
are at the same pressure level, but the compositions of MR are dif-
ferent each other, as they are determined by the phase separation
4.1. JT cycles with pure refrigerants (PR) at ambient temperature. The pressure levels and composition of
MR are optimized.
The JT cycles with PR involve a constant-temperature cooling The maximum FOM is 53.3% for optimized M2 J. The minimum
process by evaporating refrigerant. For efficient liquefaction, the temperature approach (0 K) is found at two ends and an interme-
NG feed should be refrigerated by a series of multiple JT cycles, diate temperature. The FOM of M2J may be usually lower than that
and each cycle is composed of multiple stages [15–21], as com- of MIJ, because the mismatch in temperature profiles is even larger
monly called ‘‘cascade”. Typical refrigerants are propane, ethylene near the warm end. In spite of the shortcoming in FOM, this 2-
(or ethane), and methane in the top, middle, and bottom cycles, stage cycle has advantages in practical operation, such as a smaller
respectively. Either ethylene or ethane may be used in the middle flow rate of MR (which may result in a smaller size of heat
134 H.-M. Chang / Cryogenics 72 (2015) 127–147

Fig. 7. 1-stage MR JT cycle (M1J).

Fig. 8. 2-stage MR JT cycle (M2J).

 H.-M. Chang / Cryogenics 72 (2015) 127–147 135

Fig. 9. Dual 1-stage MR JT cycles (M1J + M1J).

exchangers), and less possibility of freezing of heavy components earlier by using two phase separators, as used in KLEEMENKO or
at the cold end [3]. TEALARC processes [38,39].
The FOM of MR JT cycles can be considerably improved, if two More complex structures with three separate MR JT cycles in
separate cycles are employed in series. Out of a variety of combina- series have been developed as well. The MFC (multi-fluid cascade)
tions, three cycles are presented here – dual 1-stage MR JT cycles process [40] is based on 2-stage MR JT (at top) and dual 1-stage MR
(M1J + M1J), 1-stage MR JT and 2-stage MR JT cycles (M1J + M2J), JT cycles (in bottom) in series (M2J + M1J + M1J). The top 2-stage
and dual 2-stage MR JT cycles (M2J + M2J), as shown in Fig. 9, MR-JT cycle works as a pre-cooling cycle for the NG feed as well
Fig. 10, and Fig. 11, respectively. The structure of M1J + M1J is close as the MR streams of two bottom JT cycles. The LIQUEFIN process
to the PRICO process [33], and M1J + M2J and M2J + M2J are [41] is also composed of three MR JT cycles in series, but the two
simplified versions of the DMR (dual mixed-refrigerant) processes top cycles are 1-stage MR JT cycles and the bottom cycle is
[34–37]. 2-stage MR JT cycle (M1J + M1J + M2J). Since the multiple cycles
The pressure levels and composition of MR are optimized in are basically independently each other, much more efforts are
two-dimensional way for each MR JT cycle. The maximum FOM needed to find the optimal conditions, even though the FOM may
is 74.8%, 78.2%, and 78.6% for optimized M1J + M1J, M1J + M2J, reach an even higher value once every cycle is optimized. Another
and M2J + M2J respectively. These values are well above those for advantage of multiple MR cycles is that the performance is less
M1J and M2J, mainly because the temperature profiles get closer sensitive to the conditions of NG feed and ambient climate [3].
each other and the irreversibility in HX’s is reduced. On the other
hand, the irreversibility at JT and MIX is slightly added due to 4.3. JT cycles with pure refrigerants (PR) and mixed refrigerants (MR)
the increased number of stages.
Numerous combinations of MR JT cycles have been designed The robustness of PR JT cycles and the high efficiency of MR JT
and developed, even though the details of cycles are not included cycles can be achieved at the same time, if a PR JT cycle is used at
in this review. An MR process under recent development in Korea, the top and a MR JT cycle is used in the bottom. The most reputed
called the KSMR process [35], is similar to M2J in structure, but the is the combined 4-stage propane JT and 2-stage MR JT cycles (p4J
two cooling stages are configured at different pressure levels and + M2J) shown in Fig. 12, as widely called C3-MR process [42–45].
the overall compression work is minimized with a sophisticated The propane JT cycle has the same structure as the top cycle of
arrangement of multiple compressors and phase separators at p3J + e0 3J + m3J, but the number of stages is 4 in most cases (or
ambient temperature. In some cases of dual M2J cycles (M2J can be 3 in cold climates). Since the pre-cooling temperature is
+ M2J), the evaporating pressure in top M2J cycle can be arranged determined by the low-pressure of top propane cycle, the bottom
as two-stage as well [3]. The multi-pressure evaporation could be MR JT cycle is optimized and operated nearly independently of
effective in reducing the temperature differences near the warm the top cycle.
end, but may also cause more irreversibility due to the mixing of The maximum FOM is 80.5% for optimized p4J + M2J. The opti-
streams [37]. 3-stage MR JT cycles (M3J) have been also designed mization of p4J + M2J includes the three intermediate pressure
136 H.-M. Chang / Cryogenics 72 (2015) 127–147

Fig. 10. 1-stage MR JT and 2-stage MR JT cycles (M1J + M2J).

levels of top p4J, and the pressure levels and composition of M2J. evaporating butane is part of the cold streams. It was reported that
As shown in the diagrams, the temperature profiles of hot and cold the parallel JT cycles can take advantage of high FOM and simple
streams are very close to each other along the bottom M2J, and the structure at the same time. The maximum FOM is 70.2% for
stepwise profile matches reasonably well in the top p4J, too. b1J/e1J + M2J with the optimized pressure levels and flow rate in
Indeed the irreversibility in HX and AC is noticeably reduced and both cycles. This can be considered as a relatively high value, by
the value of FOM is the highest among the cycles presented in this taking into account the simple structure and easy operation with
review. the 1-stage JT cycles with PR.
Apart from the practical applicability to base-load plants, there
are other options in selecting the PR of top JT cycle. Two examples 4.4. Brayton cycles
are ammonia (a) or R-134a (r), since their boiling temperatures are
close to that of propane. The structure and thermodynamic charac- Turbine-based liquefaction processes are operated on standard
teristics of a4J + M2J and r4J + M2J are nearly identical to Fig. 11. In or modified Brayton cycles, as used for peaking shaving plants or
general, the ammonia JT cycle has a merit of small flow rate for recently proposed off-shore application [3,49–57]. While either
because ammonia has the largest value of latent heat of vaporiza- PR or MR may be used in a Brayton cycle, it is essential for the
tion, and the R-134a JT cycle may take advantage of non- refrigerants to remain in vapor phase (or at least vapor-rich condi-
flammable and non-toxic working fluid. The maximum FOM is tion) for the safe operation of turbines. Nitrogen is the most com-
78.2% and 79.4% for optimized a4J + M2J and r4J + M2J, respec- mon refrigerant, and the simplest and standard cycle is shown in
tively. Recently more works were reported on the selection of Fig. 14, as called 1-stage nitrogen Brayton cycle (n1B). In theory,
pre-cooling cycle with different refrigerants (top cycle) for M2J there is no limit in the pressure range, but the optimum conditions
cycle [46,47]. for all Brayton cycles are found between 0.1 and 10 MPa according
A unique combination of two JT cycles with PR on the top of M2J to the assumption.
has been lately proposed and patented [48]. As shown in The maximum FOM is 60.3% for n1B with optimized pressure
Fig. 13 and 1-stage butane JT cycle and 1-stage ethane JT cycle levels. It is recalled that the output power from expander is sub-
are combined in parallel for the pre-cooling of MR and NG feed tracted from the input power to compressors in Eq. (5), and the
(b1J/e1J + M2J). The symbol h/i between b1J and e1J denotes their compression is multi-staged with an inter-cooler between the
‘‘parallel” combination, as listed in the ‘‘combination” box of stages from the assumption. The minimum temperature approach
Fig. 4. As the pressure levels of b1J and e1J cycles are optimally is found at two ends and an intermediate point. The major source
determined, the minimum temperature approach is found not only of irreversibility is the after-coolers (AC) due to the large temper-
at two ends of HX3, but also over a wide range in the middle, ature rise by adiabatic compression and the large flow rate of
where the condensing ethane is part of the hot streams and the nitrogen.
 H.-M. Chang / Cryogenics 72 (2015) 127–147 137

Fig. 11. Dual 2-stage MR JT cycles (M2J + M2J).

The thermodynamic performance of Brayton cycles can also be parallel with a small gap), because the specific heat is nearly con-
improved by modifying the standard cycle or combining two sep- stant over this temperature range for both NG and nitrogen. A
arate cycles in series. A modified Brayton cycle for 2-stage refriger- nitrogen Brayton cycle therefore is an excellent option for the bot-
ation is shown in Fig. 15, as called 2-stage nitrogen Brayton cycle tom of NC liquefaction, as mentioned below.
(n2B). The high-pressure nitrogen is expanded in the first expander Two separate Brayton cycles may be combined in series. Fig. 16
only to an intermediate pressure, where a branched flow is shows the dual nitrogen 1-stage Brayton cycles, where nitrogen is
returned through the warm HX and the other flow is expanded fur- used in both the top and bottom cycles (n1B + n1B). Fig. 17 has the
ther to the lowest pressure and temperature at the cold end. The same structure, but methane is used in the top Brayton cycle, while
pressure levels and flow rates at each stage are optimized. nitrogen is used in the bottom Brayton cycle (m1B + n1B). In either
The maximum FOM is 67.1% for optimized n2B. The minimum case, the pressure levels and flow rate are optimized in the two
temperature approach is found at an intermediate position as well cycles.
as two ends, and the temperature profiles are closer to each other The maximum FOM is 67.1% and 69.7% for optimized n1B + n1B
than those of n1B. It is noticeable in the coldest HX at bottom that and m1B + n1B, respectively. It is interesting to note that the FOM
the temperature profiles appears to be almost overlapped (or of n1B + n1B is nearly the same as that of n2B, which means that
138 H.-M. Chang / Cryogenics 72 (2015) 127–147

Fig. 12. 4-stage propane and 2-stage MR JT cycles (p4J + M2J).

there is no clear merit in using two separate cycles with the same the mass flow rate of methane is larger. In the optimized m1B +
gas. On the other hand, m1B + n1B is somewhat superior to n2B n1B, methane remains in vapor phase throughout the cycle.
and n1B + n1B under this specific condition. When compared with Mixed refrigerants may be used in a Brayton cycle, but the
n1B + n1B, the pressure levels of m1B + n1B are much lower, but selection of MR is subject to the constraint that the refrigerants
 H.-M. Chang / Cryogenics 72 (2015) 127–147 139

Fig. 13. Parallel 1-stage butane/1-stage ethane, and 2-stage MR JT cycles (b1J/e1J + M2J).

should be in vapor phase or at least the liquid content should be specific heat, as mentioned above. Fig. 19 shows a series combina-
within a certain limit (for example, less than 10%). One practical tion of 3-stage propane JT, 3-stage ethylene JT, and 1-stage nitro-
choice to satisfy this constraint is a mixture of nitrogen and gen Brayton cycles (p3J + e0 3J + n1B) [58]. The top and middle
methane [3], which can be denoted by M1B. The pressure levels cycles are basically identical to those of p3J + e0 3J + m3J, but the
and composition of MR are optimized for M1B. As shown in bottom cycle is replaced by an independent n1B. Alternatively,
Fig. 18, the maximum FOM is 61.1% for optimized M1B, where nitrogen of the bottom n1B cycle may be pre-cooled by the top
the MR is a mixture of 74.0% nitrogen and 26.0% methane on mole and/or middle cycles to increase the FOM to an extent. The pres-
basis. When compare with n1B, M1B has minor benefits of a higher sure levels of JT and Brayton cycles are optimized. The maximum
FOM and a smaller flow rate. FOM is 72.6% for optimized p3J + e0 3J + n1B, which is evaluated
as the highest among all PR cycles under consideration.
4.5. Combined and Claude cycles A similar idea was proposed and investigated to combine
1-stage carbon dioxide JT cycle on the top of 1-stage nitrogen
JT and Brayton cycles can be combined in a variety of ways for Brayton cycle (c1J + n1B) [59,60]. Since both refrigerants are
efficient liquefaction. Among the PR cycles, a series combination of inflammable, the combined cycles have an inherent advantage in
either Brayton + JT + Brayton (B + J + B) cycles or JT + JT + Brayton operational safety. It was reported that the cycles require a smaller
(J + J + B) cycles is an intuitive choice, since the bottom Brayton footprint area, and may be suitable for offshore or small-scale
cycle can efficiently cover the liquid region with a nearly constant application.
140 H.-M. Chang / Cryogenics 72 (2015) 127–147

Fig. 14. 1-stage nitrogen Brayton cycle (n1B).

Fig. 15. 2-stage nitrogen Brayton cycle (n2B).

 H.-M. Chang / Cryogenics 72 (2015) 127–147 141

Fig. 16. Dual 1-stage nitrogen Brayton cycles (n1B + n1B).

Fig. 17. 1-stage methane and 1-stage nitrogen Brayton cycles (m1B + n1B).
142 H.-M. Chang / Cryogenics 72 (2015) 127–147

Fig. 18. 1-stage MR Brayton cycle (M1B).

Recently, a closed Claude cycle is proposed to use the natural 5. Discussion

gas (NG) itself as refrigerant [61], as shown in Fig. 20. This cycle
can be termed as 1-stage MR Claude cycle (M1C), since the The results presented above are useful primarily for selecting a
refrigerant is an MR whose composition is the same as the NG feed. suitable cycle under any given environments. For each cycle, the
Even though three HX’s are used in series above the JT expansion, thermodynamic performance was indexed by the presented value
this cycle is called 1-stage, because a Claude cycle always needs of FOM in the range of 53.5–80.6%. In addition, the graphic repre-
three HX’s (in the same context, the Collins cycle used for helium sentation of cycles on T–s diagram, temperature profiles, and
liquefaction or refrigeration [1,2] may be called 2-stage Claude exergy expenditure can provide an intuition on how the structure
cycle, even though M2C is not considered here). An obvious merit of cycles works for the liquefaction of natural gas. The listed data
of using the NG itself as refrigerant is that no refrigerant inventory on thermodynamic properties at each point of the cycle can also
is needed in offshore or floating systems. The pressure levels and serve as a good starting point for the design of practical liquefac-
expander flow ratio (i.e. the ratio of expander flow to total com- tion processes.
pressor flow) are optimized. In examining the results, it is important to recall that the anal-
The maximum FOM is 59.7% for optimized M1C. It is noted in ysis was performed for ideal cycles with perfect components and
the optimized cycle that the exit condition of expander (E) must optimized condition. The deviation of an actual process from the
match exactly with the merging low-pressure stream of MR, as ideal cycle is estimated by modifying the listed assumptions with
called the ‘‘optimally expanded” condition in [61]. In the presented proper performance parameters on imperfect components, includ-
ideal cycle, the vapor fraction of MR is 86.5% at the exit of expan- ing the HX effectiveness, the adiabatic efficiency (for compressors
der, but should be well above this value in practical cycles, as the and expanders), and the pressure drop. Even though the details
adiabatic efficiency of an expander is lower than the unity. A mix- on actual cycles may be beyond the scope of this review, a few
ture of nitrogen and methane has been investigated for similar important comments are made especially about the actual perfor-
Claude cycle [62]. mance of HX’s.
The largest-capacity plants under operation are based on a pro- There are a few different ways how to incorporate the effect of
cess called AP-X [63,64], which is a series combination of 3-stage finite HX size and heat transfer coefficient into the cycle analysis. A
propane JT cycle, 1-stage MR JT cycle, and 1-stage nitrogen Brayton common method is to simply specify the value of minimum tem-
cycle (p4J + M1J + n1B), as shown in Fig. 21. It was reported that perature approach (e.g. DTmin = 3 K) [3]. This is a convenient way
the capacity limit of efficient p4J + M2J has been successfully over- of utilizing the commercial process simulators, but may not be a
come with the addition of nitrogen Brayton cycle at the bottom. fair criterion for comparing the cycles with different configurations
The pressure levels of three cycles and the composition of MR JT of HX’s. A more reasonable method for fair comparison is to specify
cycle are optimized. The maximum FOM is 76.0% for optimized the value of HX effectiveness (e.g. e = 0.98) as a dimensionless
p4J + M1J + n1B. parameter. In order to incorporate the value of effectiveness into
 H.-M. Chang / Cryogenics 72 (2015) 127–147 143

Fig. 19. 3-stage propane JT, 3-stage ethylene JT, and 1-stage nitrogen Brayton cycles (p3J + e0 3J + n1B).
144 H.-M. Chang / Cryogenics 72 (2015) 127–147

Fig. 20. 1-stage MR Claude cycle (M1C).

the cycle, the maximum heat must be calculated first (i.e. with 0 K variational problem of minimizing the entropy generation subject
of minimum temperature approach), and then the actual heat is to a constraint, and successfully solved by the method of Lagrange
determined by its definition. multiplier [58]. The results show that the best thermodynamic per-
formance is achieved when the temperature difference is propor-
Q_ tional to the absolute temperature of the NG feed.
e¼ _ ð8Þ
Q max
 
DT
Another rigorous method is to simultaneously solve the energy  constant ð9Þ
equations for all the streams and determine the temperature distri- T F opt
bution of respective streams. For this calculation, the values of
overall heat transfer coefficient (U) or (dimensionless) the number In other words, it is desired is that the temperature difference
of transfer unit (NTU) should be specified [65]. This analysis should be smaller at colder HX.
requires huge efforts on the numerical calculation, but the detailed Finally, the effect of multi-stream HX’s should be mentioned.
specifications of HX’s (e.g. the dimension of plate-fin HX’s or spiral Most of HX’s used in the refrigeration cycles for liquefaction have
wound HX’s) can be effectively incorporated. three or more streams. In the cycle analysis presented above, it
A thermodynamic optimization theory is cited here with was assumed that all hot streams have the same temperature
regards to the effect of finite HX sizes, especially when the lique- and all cold streams have the same temperature at an axial loca-
faction process is composed of multi-stages of HX’s in series. It is tion, as only two composite curves were plotted. It was reported
always true that as any HX area increases, the temperature differ- that those temperature profiles are difficult to realize in practice,
ence between hot and cold streams in the HX decreases and the and the FOM based on the assumption is always an over-
FOM of liquefaction increases accordingly. It is an important design estimate [65]. Elaboration is needed for proper heat-exchanger
strategy, however, how to allocate the HX area to each stage, if the design in order to achieve the designated thermodynamic
total sum of HX area is fixed. This was formulated with a typical efficiency.
 H.-M. Chang / Cryogenics 72 (2015) 127–147 145

Fig. 21. 3-stage propane JT cycle, 1-stage MR JT cycle, and 1-stage nitrogen Brayton cycle (p4J + M1J + n1B).

6. Concluding remarks requirements, and local climate). This review is prepared with a
goal to provide the primary basis of cryogenic refrigeration for liq-
It is a difficult but important task to determine which uefaction from thermodynamic point of view. Towards the goal, a
refrigeration cycle is the most suitable for a specific liquefaction variety of refrigeration cycles are classified in systematic way by
system of natural gas, taking a number of practical factors into con- the type of expansion processes (JT, Brayton, and combined cycles)
sideration (such as composition/pressure/amount of natural gas, and by the type of refrigerants (pure and mixed refrigerants).
constraints on size/weight, operation simplicity, safety, statutory Sixteen standard and modified refrigeration cycles were selected
146 H.-M. Chang / Cryogenics 72 (2015) 127–147

from the existing and proposed liquefaction processes, and their [28] Shirazi MMH, Mowla D. Energy optimization for liquefaction process of
natural gas in peak shaving plant. Energy 2010;35:2078–85.
ideal versions are fully analyzed with the optimal conditions to
[29] Alabdulkarem A, Mortazavi A, Hwang Y, Radermacher R, Rogers P.
achieve its utmost thermodynamic performance. Detailed refriger- Optimization of propane pre-cooled mixed refrigerant LNG plant. Appl
ation cycles are quantitatively presented and compared in terms of Therm Eng 2011;31:1091–8.
FOM (figure of merit) and thermodynamic irreversibility. Hope- [30] Khan MS, Lee S, Rangaiah GP, Lee M. Knowledge based decision making
method for the selection of mixed refrigerant systems for energy efficient LNG.
fully, this review will be helpful in clearly understanding the struc- Appl Energy 2013;111:1018–31.
ture of refrigeration cycles and designing an efficient liquefaction [31] Coers DH, Sudduth JW. Refrigerant apparatus and process using
process for future development in accordance with various needs. multicomponent refrigerant. US Patent 3,932,154; 1976.
[32] Roberts MJ, Agrawal R, Daugherty TL. Single mixed refrigerant gas liquefaction
process. US Patent 6,347,531; 2002.
Acknowledgments [33] Xu X, Liu J, Cao L. Optimization and analysis of mixed refrigerant composition
for the PRICO natural gas liquefaction process. Cryogenics 2014;59:60–9.
[34] Garier C, Paradowski H. Method and plant for liquefying a gas with low boiling
The author thanks his students at Hong Ik University, especially temperature. US Patent 4,274,849; 1981.
Ms. Hye Su Lim, for the cycle analysis and the preparation of [35] Lee S, Yang YM, Choe KH, Lee CG, Park CI, Choi SH, Lee YB. Natural Gas
Liquefaction Process. Korean Patent 10-1037249; 2011.
figures. This work is supported by a grant from the LNG Plant
[36] Lee S, Long NVD, Lee M. Design and optimization of natural gas liquefaction
R&D Center funded by the Ministry of Land, Infrastructure and and recovery processes for offshore floating liquefied natural gas plants. Ind
Transport (MOLIT) of the Korean Government. Eng Chem Res 2012;51:10021–30.
[37] Newton CL. Dual mixed refrigerant natural gas liquefaction with staged
compression. US Patent 4,525,185; 1985.
References [38] Michelsen FA, Halvorsen IJ, Lund BF, Wahl PE. Modeling and simulation for
control of the TEALARC liquefied natural gas process. Ind Eng Chem Res
[1] Barron RF. Cryogenic systems. 2nd ed. New York: Oxford University Press; 2010;49:7389–97.
1985. [39] Paradowski H, Rojey A. Method and device for Liquefying a natural gas without
[2] Timmerhaus KD, Flynn TM. Cryogenic process engineering. New York: Plenum phase separation of the coolant mixtures. US Patent 6,105,389; 2000.
Press; 1989. [40] Roberts MJ, Agrawal R. Dual mixed refrigerant cycle for gas liquefaction. US
[3] Venkatarathnam G. Cryogenic mixed refrigerant processes. New York: Patent 6,119,479; 2000.
Springer; 2008. [41] Roberts J. Integrated multiple-loop refrigeration process for gas liquefaction.
[4] Mokhatab S, Mak JY, Valappil JV, Wood DA. Handbook of liquefied natural US Patent 7,086,251; 2006.
gas. Gulf Professional Publishing; 2013. [42] Gaumer Jr S, Lee S, Charles NL. Combines cascade and multicomponent
[5] Ikealumba WC, Wu H. Some recent advances in liquefied natural gas (LNG) refrigeration system and method. US Patent 3,763,658; 1973.
production, spill, dispersion, and safety. Energy Fuels 2014;28:3556–86. [43] Pillarella M, Liu YN, Petrowski J, Bower R. The C3MR liquefaction cycle:
[6] Lim WS, Choi KH, Moon I. Current status and perspectives of liquefied natural versatility for a fast growing, ever changing LNG industry. In: Proceedings of
gas (LNG) plant design. Ind Eng Chem Res 2013;52:3065–88. the 15th international conference on LNG (LNG-15), Barcelona, Spain, April
[7] Hwang J, Lee KY. Optimal liquefaction process cycle considering simplicity and 24–27; 2007.
efficiency for LNG FPSO at FEEE stage. Comput Chem Eng 2014;63:1–33. [44] Mortazavi A, Abdullah A, Hwang Y, Fadermacher R. Novel combined cycle
[8] Bosma P. Liquefaction technology; developments through history. In: configurations for propane pre-cooled mixed refrigerant (APCI) natural gas
Proceedings of the 1st annual gas processing symposium, Doha, Qatar, liquefaction cycle. Appl Energy 2014;117:76–86.
January 10–12; 2009. [45] Mortazavi A, Somers C, Hwang Y, Radermacher R, Rodgers P, Al-Hashimi S.
[9] Li Y, Wang X, Ding Y. An optimal design methodology for large-scale gas Performance enhancement of propane pre-cooled mixed refrigerant LNG
liquefaction. Appl Energy 2012;99:484–90. plant. Appl Energy 2012;93:125–31.
[10] Vantani A, Mehdi M, Palizdar A. Energy and exergy analyses of five [46] Castillo L, Dorao C. On the conceptual design of pre-cooling stage of LNG plants
conventional liquefied natural gas processes. Int J Energy Res 2014. using propane or an ethane/propane mixture. Energy Convers Manage
[11] Vink KJ, Nagelvoort RK. Comparison of base load liquefaction processes. In: 2013;65:140–6.
Proceedings of the LNG 12, institute of gas technology, Perth, WA, Australia, [47] Castillo L, Majzoub Dahouk M, Di Scipio S, Dorao CA. Conceptual analysis
May 4–7; 1998. of the precooling stage for LNG processes. Energy Convers Manage 2013;66:
[12] Shukri T. LNG technology selection. Hydrocarbon Eng 2004;9(2):71–6. 41–7.
[13] Borgnakke C, Sonntag RE. Fundamentals of thermodynamics. 8th ed. New [48] Chang HM, Park JH, Woo HS, LEE S, Choe KH. New concept of natural
York: John Wiley & Sons; 2014. gas liquefaction cycle with combined refrigerants. In: Proceedings of the ICEC
[14] Bejan A. Advanced engineering thermodynamics. 3rd ed. New York: John 23 – ICMC 2010; 2011. p. 257–62.
Wiley & Sons; 2006. [49] Finn AJ. Effective LNG production offshore. In: Proceedings of the 81st annual
[15] Andress DL, Watkins RJ. Beauty of simplicity: Phillips optimized cascade LNG GPA convention, Dallas, Texas, March 10–13; 2002.
liquefaction process. ACE 2014;49:91–8. [50] Foglietta JH. Production of LNG using dual independent expander refrigeration
[16] Ransbarger W. A fresh look at LNG process efficiency. LNG Ind 2004;92 cycles. In: Proceedings of the AICHE Spring national meeting, New Orleans, LA,
(10):1719–34. March 10–14; 2002.
[17] Andress DL. The Phillips optimized cascade LNG process. A quarter century of [51] Remeljej CW, Hoadley AFA. An exergy analysis of small scale liquefied natural
improvements. Bartlesville OK: Phillips Petroleum Company; 1996. gas (LNG) liquefaction processes. Energy 2006;31(12):2005–19.
[18] Kanoglu M. Exergy analysis of multistage cascade refrigeration cycle used for [52] Barclay M, Denton N. Selecting offshore LNG processes. LNG J 2005;10
natural gas liquefaction. Int J Energy Res 2002;26(8):763–74. (1):34–6.
[19] Austbo B, Gundersen T. Using thermodynamic insight in the optimization of [53] Bronfenbrenner JC, Pillarella M, Solomon J. Selecting a suitable process. LNG
LNG processes. In: Proceedings of the 24th European symposium on computer Industry; 2009. <http://www.lngindustry.com/>.
aided process engineering, Budapest, Hungary, June 15–18; 2014. [54] Chang HM, Park J, Cha K, Lee S, Choe K. Modified reverse-Brayton cycles for
[20] Zhang J, Xu Q. Cascade refrigeration system synthesis based on exergy efficient liquefaction of natural gas. Cryocoolers 2012;17:435–42.
analysis. Comput Chem Eng 2011;35:1901–14. [55] Li QY, Ju YL. Design and analysis of liquefaction process for offshore associated
[21] Lee I, Tak K, Kwon H, Ko D, Moon I. Design and optimization of a pure gas resources. Appl Therm Eng 2010;30:2518–25.
refrigerant cycle for natural gas liquefaction with sub-cooling. Ind Eng Chem [56] He TB, Ju YL. A novel conceptual design of parallel nitrogen expansion
Res 2014;53:10397–403. liquefaction process for small-scale LNG plant in skid-mount packages. Energy
[22] Swenson LK. Single mixed refrigerant, closed loop process for liquefying 2014;75:349–59.
natural gas. US Patent 4,033,735; 1977. [57] He TB, Ju YL. Performance improvement of nitrogen expansion liquefaction
[23] Wang M, Zhang J, Xu Q, Li K. Thermodynamic-analysis –based energy process for small-scale LNG plant. Cryogenics 2014;61:111–9.
consumption minimization for natural gas liquefaction. Ind Eng Chem Res [58] Chang HM, Chung MJ, Lee S, Choe KH. An efficient multi-stage Brayton-JT cycle
2011;50:12630–40. for liquefaction of natural gas. Cryogenics 2011;51:278–86.
[24] Vaidyaraman S, Maranas CD. Synthesis of mixed refrigerant cascade cycles. [59] Khan MS, Lee S, Hasan M, Lee M. Process knowledge based opportunistic
Chem Eng Commun 2002;189:1057–78. optimization of the N2-CO2 expander cycle for the economic development of
[25] Kim JK, Lee GC, Zhu F, Smith R. Cooling system design. Heat Transfer Eng stranded offshore fields. J Nat Gas Sci Eng 2014;18:263–73.
2002;23(6):49–61. [60] Yuan Z, Cui M, Xie Y, Li C. Design and analysis of a small-scale natural gas
[26] Lee GC, Smith R, Zhu XX. Optimal synthesis of mixed-refrigerant systems for liquefaction process adopting single nitrogen expansion with carbon dioxide
low-temperature processes. Ind Eng Chem Res 2002;41:5016–28. pre-cooling. Appl Therm Eng 2014;64(1):139–46.
[27] Nogal FD, Kim JK, Perry S, Smith R. Optimal design of mixed refrigerant cycles. [61] Chang HM, Lim HS, Choe KH. Thermodynamic design of natural gas
Ind Eng Chem Res 2008;47:8724–40. liquefaction cycles for offshore application. Cryogenics 2014;63:114–21.
 H.-M. Chang / Cryogenics 72 (2015) 127–147 147

[62] Cao WS, Lu XS, Lin WS, Gu AZ. Parameter comparison of two small-scale international conference and exhibition on liquefied natural gas (LNG-14),
natural gas liquefaction processes in skid-mounted packages. Appl Therm Eng Doha, Qatar, March 21–24; 2004.
2006;26:898–904. [65] Chang HM, Lim HS, Choe KH. Effect of multi-stream heat exchanger on
[63] Roberts MJ, Agrawal R. Hybrid cycle for the production of liquefied natural gas. performance of natural gas liquefaction with mixed refrigerant. Cryogenics
US Patent 6,308,531; 2001. 2012;52:642–7.
[64] Roberts MJ, Liu YN, Bronfenbrenner JC, Petrowski JM. Reducing LNG capital
cost in today’s competitive environment. In: Proceedings of the 14th