You are on page 1of 15

ARTICLE IN PRESS

Energy 31 (2006) 20052019


www.elsevier.com/locate/energy

An exergy analysis of small-scale liqueed natural gas (LNG)


liquefaction processes
C.W. Remeljeja, A.F.A. Hoadleyb,
a
BHP Billiton, 180 Lonsdale Street, Melbourne, Vic. 3000, Australia
b
Department of Chemical Engineering, Building 36, Clayton Campus, Monash University, Vic. 3800, Australia
Received 20 October 2004

Abstract

Four processes for small-scale liqueed natural gas (LNG) production are evaluated. These include a single-stage mixed
refrigerant (SMR), a two-stage expander nitrogen refrigerant and two open-loop expander processes. Steady-state
simulations were undertaken to ensure that each process was compared on an identical basis, was fully optimised and was
in agreement with published results. Composite curves for the feed and recycle streams and the refrigerant or cold recycle
stream showed the degree of optimisation available within each process. The full exergy analysis showed the relative
contributions to the total shaft work requirements, with the lowest being the SMR process. The lower efciency of the
expander-driven compressors is the main difference between processes. A more general comparison suggested that the
nitrogen refrigerant process and the New LNG open-loop process are the leading candidates for offshore compact LNG
production.
r 2005 Elsevier Ltd. All rights reserved.

Keywords: Refrigeration; Cryogenic processes; Energy efciency; Shaftwork; Mixed refrigerants; Nitrogen

1. Introduction

A small-scale offshore-liqueed natural gas (LNG) process employed on a oating production facility
would be capable of converting solution gas and/or gas-cap gas from remote marginal oil or condensate elds
into a saleable product. Energy efciency is important to LNG production as feed gas is consumed in order to
carryout the liquefaction process. However, the energy efciency is not the only factor of importance, as any
project must be cost effective, reliable, and tolerant of reasonably foreseeable feed changes. It must also be
safely operated and maintained and for offshore, it should also be relatively compact.
Four different processes have recently been reported in the literature as suitable for small-scale offshore
liquefaction producing less than 1 million tonnes per annum of LNG. The four processes can be grouped into
three generic process types based on the different refrigerant cycles used. These are a single mixed refrigerant
cycle (SMR process [13]), a nitrogen cycle (cLNG process [4]) and two different open-loop cycles (New LNG
[5] and GCL [6]).

Corresponding author. Tel.: +61 3 99053421; fax: +61 3 99055686.


E-mail address: andrew.hoadley@eng.monash.edu.au (A.F.A. Hoadley).

0360-5442/$ - see front matter r 2005 Elsevier Ltd. All rights reserved.
doi:10.1016/j.energy.2005.09.005
ARTICLE IN PRESS
2006 C.W. Remeljej, A.F.A. Hoadley / Energy 31 (2006) 20052019

World-scale LNG plants typically have a capacity greater than 2.5 million tonnes per annum through each
train, and often 23 trains are installed to provide the required economies of scale. These processes have
reported efciencies of 5.56 kW h/kmol LNG [4]. A consistent trend was noticed amongst the published
comparisons of different LNG processes, which suggest that the most efcient LNG processes overall, are the
propane pre-cooled mixed refrigerant cycles (C3/MR) and the dual mixed refrigerant cycle (DMR). Liu et al.
[7] and Foerg et al. [8] report that the C3/MR cycle is the most efcient, whereas Vink and Nagelvoort [9]
reported that DMR is more efcient. All three published comparisons are consistent with the efciency
ranking for the remaining generic processes. The next most efcient process is reported to be the cascade cycle,
followed by the SMR and then the nitrogen cycle. No literature was found to compare the above processes
with the open-loop processes (where feed gas is also used as refrigerant and non-condensed feed/refrigerant is
recycled). Foerg et al. [8] considered the capital cost aspects of the processes with the relative capital cost being
essentially the reverse of the efciency seriatim listed above. In all cases of process comparison, the underlying
efciency of the key pieces of machinery that consume the majority of the work is not stated, nor is the feed
composition or ambient conditions stated. These omissions from the papers make proper quantitative
comparison difcult, if not impossible [9].
The four processes evaluated in this paper all differ signicantly with respect to process design by using
different refrigerants and rotating machinery congurations. Yet all claim high efciencies, low capital cost
and offshore suitability. The objective of this paper is therefore to compare the processes on an identical basis
and evaluate both their energy efciency, potential for improvement and suitability for an offshore
application.

2. Small-scale LNG processes

A refrigerant self-cooling/condensing loop is required to some degree in all LNG processes because of the
low temperatures required for LNG production. To achieve such low temperatures the refrigerant system
must include nitrogen (cLNG) or methane (open-loop) or a mixture of the two (SMR), which cannot be
condensed at ambient temperature. This self-cooling/condensing requirement results in the need for high
refrigerant circulation rates compared to the feed gas ow rate and as a result the characteristics of the
refrigerant often dominate the temperature proles within the heat exchanger system.

2.1. SMR process

The PRICO SMR process is the simplest of the four processes studied. The process was rst used in 1981 at
the Skikda LNG plant in Algeria. Three liquefaction trains using the process have been built and operated
over the last 23 years. Fig. 1 shows the original ow sheet, which consists of a single LNG heat exchanger, a
separate feed/product and refrigerant system, a compressor with an associated after-cooler, suction scrubbers,

LNG to
Natural Gas
storage

Cold refrigerant at 40
LNG Heat Exchanger (cold box) bar and LNG exit
Warm refrigerant at 40
Temperature
bar and 30 oC

MR
Compressor

subcooled refrigerant
at 3.7 bar

Fig. 1. Single-stage mixed refrigerant (SMR) ow sheet.


ARTICLE IN PRESS
C.W. Remeljej, A.F.A. Hoadley / Energy 31 (2006) 20052019 2007

a separator and pump. Price [2,3] claims that this signicant operating experience has led to improvements in
the process efciency of 38% compared to the original design. For example, compression is now usually in two
stages and with higher compression ratios. The SMR process has the lowest equipment count compared to the
other processes.
In the SMR process the feed gas enters the LNG exchanger at feed conditions and is cooled against the cold
refrigerant stream to the required LNG storage conditions of less than 155 1C. The cold low-pressure
refrigerant stream also acts to condense the high-pressure refrigerant stream prior to the pressure letdown
stage that provides the necessary heat exchanger cold side temperature differential.

2.2. New LNG scheme

The New LNG Scheme has not yet been trialed. It is based on a proven LNG peak-shaving operation.
The process is shown in Fig. 2. It is considerably more complex than the SMR process and differs in
three major areas. It is an open-loop scheme where the feed gas is also the refrigerant. As a result
the low-pressure warmed refrigerant gas is recompressed and recycled back to the raw feed stream. The
entire feed stream is pre-cooled before being split into two streams, one for further cooling to become
the LNG product and the other for the refrigerant recycle loop. Cooling and condensation of the refri-
gerant is attained using external chilling, self-cooling and isentropic expansion via a turbo expander.
The nal temperature of the LNG product stream is attained by ashing a portion of the chilled feed stream
from feed pressure to near atmospheric pressure and the resulting cold ash gas is recycled as a refrigerant
stream. In this scheme, the work recovered from the turbo expander is used for refrigerant recycle
recompression.
Foglietta [5] claims that the New LNG Scheme is more cost effective than licensed technologies with
comparable processing efciencies. In addition, due to the use of a simple propane cycle and the feed gas as the
refrigerant he claimed reduced refrigerant costs compared to other technologies. As a consequence,
installation is simplied, and the plot plan is reduced. Thus, the technology is suitable for small-scale
applications, especially offshore.

FUEL GAS
3 RD STAGE 2 ND STAGE 1ST STAGE
Low pressure
flash gas recycle
FEED GAS RECYCLE COMPRESSOR

FUEL

LNG HEAT EXCHANGER PROPANE


REFRIGERANT
PACKAGE
Recompression and
recycle of main
refrigerant stream

COMPRESSOR/ LNG Flash


EXPANDER Separator

LNG
PRODUCT

Cooled feed gas side stream taken for


isentropic expansion to provide main
refrigeration duty

Fig. 2. New LNG ow sheet.


ARTICLE IN PRESS
2008 C.W. Remeljej, A.F.A. Hoadley / Energy 31 (2006) 20052019

Chiller

Feed

LNG
Possible recycle to
minimise chiller or open
a pinch depending on
Fuel gas
TE outlet liquid fraction.

Fig. 3. The GCL concept ow sheet.

2.3. GCL concept

As with the New LNG Process Scheme, the GCL LNG process is a design concept with no
operating history. The ow sheet for the GCL LNG concept was constructed from the patent application
[6] and is shown in Fig. 3. The GCL concept differs from the New LNG Scheme in one major area.
This process requires the stream exiting the expander to be greater than 30% liquid by mass as this
stream is separated with the liquid being further cooled and ashed to become the LNG product stream.
The vapour phase is recycled as refrigerant. (By comparison, in the New LNG scheme the entire expander
two-phase outlet stream is recycled as refrigerant because the feed stream was split prior to entering the
expander.)

2.4. cLNG

The cLNG scheme is shown in Fig. 4. This process is most similar to the SMR process, as the LNG stream
is separate from the refrigerant system. The cLNG refrigerant is a single component, nitrogen, which is always
in the gaseous phase. The cLNG refrigerant is a two-stage system. In order to cool the nitrogen refrigerant to
below the product temperature the cLNG process uses both self-cooling (within the LNG heat exchanger) and
turbo expanders. The power extracted from the turbo expanders is recovered and used to supplement the
refrigerant recompression duty.
cLNG has a similar development history as the New LNG Process scheme, having been derived
from existing LNG peak-shaving technology, in this instance using a nitrogen cycle [4]. An operating
cLNG plant has not yet been built, but several LNG nitrogen peak-shaving plants exist. As a result
the technology, like the two other open-loop processes, claims its validity from the use of proven
technology elements. However, unlike the other two processes, the technologys inventor, built and
successfully operated a nitrogen rejection unit that processed about 1 million Sm3/day of feed gas between
1994 and 2002 [10].
The cLNG technology also claims two other advantages over the other compact LNG processes intended
for offshore use. First, the heat transfer process is easier to model and the equipment is easier to design,
because the nitrogen refrigerant is always in a single phase. Second, the cLNG process is safer for offshore use
because of the reduced hydrocarbon inventory compared to the other processes, which in turn is due to the use
of nitrogen as the refrigerant.
ARTICLE IN PRESS
C.W. Remeljej, A.F.A. Hoadley / Energy 31 (2006) 20052019 2009

Cooling
M
M M
M

Treated Natural Gas Gas Turbine


Cycle
Compressor

M
MM M

Chilling Cooling

LNG Heat Warm


exchangers Expander/Booster

Cold Expander/Booster

LNG Storage
Tank
LNG Transfer Pump

Fig. 4. The cLNG process ow sheet.

3. Theory

3.1. Exergy analysis

Exergy is a measure of the maximum amount of useful energy that can be extracted from a process stream
when it is brought to equilibrium with its surroundings in a hypothetical reversible process [11]. This is a
thermodynamic measure dened only in terms of stream enthalpy, H, and entropy, S, for the given stream
conditions relative to the surroundings. For ow sheet unit operations at steady-state conditions, the kinetic
and potential energy effects are ignored. The exergy, Ex, or useful available work, of a stream is therefore
expressed as
Ex H  T 0 ST;P  H  T 0 ST 0 ;P0 , (1)
where T0, P0 are the equilibrium temperature and pressure, normally considered as the ambient temperature
and pressure. When matter is taken from one state to another via a hypothetical reversible process, the
reference terms cancel out and the change in exergy is given by
DEx H  T 0 Sstate 2  H  T 0 Sstate 1 . (2)
This change in exergy represents the minimum amount of work to be added or removed to change from
state 1 to state 2 when there is an increase or decrease in internal energy or enthalpy resulting from the change.
Irreversibilities exist in real systems. As a result the actual work required to be input to a process to change
state is more than that which would be required in the ideal case. The actual work input or output can be
calculated from an energy balance, as given by the rst law of thermodynamics, over the actual system. Lost
work can thus be dened as the difference between the work involved with the ideal reversible process and the
real process chosen. Lost work, Wlost, can be expressed as
W lost W actual  DEx. (3)
ARTICLE IN PRESS
2010 C.W. Remeljej, A.F.A. Hoadley / Energy 31 (2006) 20052019

For a chosen feed condition and LNG product specication, the minimum possible amount of work
required to produce the LNG product is determined by the difference in the exergy of the LNG and the feed.
This can be expressed as

W rev SH  T o SLNG  SH  T o Sfeed . (4)

For a feed pressure of 55 bar absolute, feed temperature of 25 1C and ambient temperature of 20 1C, the
minimum reversible specic power requirement is 2.13 kW h/kmol LNG. The actual amount of work required
to produce LNG is greater than the minimum reversible work in all processes studied because of the
irreversibilities within the processes. The major irreversibilities in the LNG processes are due to losses within
the compression (and associated after-cooling) system, driving forces across the LNG heat exchanger and
other exchangers, and losses due to refrigerant letdown. World-scale optimised LNG plants require more than
2.5 times the minimum theoretical power requirements.
Liu et al. [7] discuss the use of exergy as a tool for analysis. They concluded that the individual
processes have signicantly different patterns of lost work and that all liquefaction cycles can approach
the same efciency by modifying the cycle conguration and equipment choice. However, due to the
different patterns of lost work the cost of achieving comparable efciencies for different processes is
considerably different. As a result the economically optimum congurations are dependent on the specic
project factors at hand that affect these costs. An attempt was made by these authors to show using exergy
analysis that power requirements can be reduced through both compressor and refrigerant staging for the
SMR process. However, the nitrogen cycle and open-loop turbo expander processes were not covered in their
paper.

3.2. Shaftwork targeting

For heat engine and heat pump processes driven by temperature change, the reversible shaftwork can be
calculated from the heat duty DH and the Carnot factor according to

Ex W rev DHO; (5)

where the Carnot factor O 1  T 0 =TU Linnhoff and Dhole [12] showed that this could be incorporated
into composite curve analysis by plotting the composite curves on a Carnot factor rather than temperature
basis against heat ow. Therefore, for heat ow processes it is possible to estimate the minimum shaftwork by
comparing the composite hot streams and cold streams, since the area between these two curves equates to the
minimum driving force for heat exchange and also the lost work in the heat exchange network. Knowing this
target value signicantly aids the process optimisation problem by providing an ultimate goal with which
possible process improvements can be compared.
Lee [1] conducted a detailed investigation and optimisation of the original SMR process. He concluded that
a graphical targeting approach could not be used directly to optimise the SMR process, because the
optimisation required small adjustments to the refrigerant composition, which in turn required changes to the
refrigerant conditions. Lee did use the Carnot factor plots to visualise the results of the optimisation that was
achieved.
Another problem with the shaftwork targeting method is in handling pressure losses. These may be
due to friction losses across equipment, such as exchangers or valves, or work extracted from the
process by liquid expanders and turbo expanders, which operate as irreversible machines. All the processes
involve these operations to some extent and the turbo-expander processes are dominated by these equipment
losses.
Taking a similar approach to Lee [1], the shaftwork targeting method is employed in this study to evaluate
the temperature driving force losses associated with the LNG heat exchange and determine the potential for
further improvement. An exergy analysis of the whole ow sheet is used to determine the exergy losses
occurring across individual components of the different ow sheets.
ARTICLE IN PRESS
C.W. Remeljej, A.F.A. Hoadley / Energy 31 (2006) 20052019 2011

Table 1
Feed and ambient conditions and other assumptions

Property Condition Basis Comments

Feed temperature 25 1C Case Study 1 [1]


Feed pressure 55 bar Case Study 1 [1]
Nominal pressure drop 5 bar Case Study 1 [1]
Feed (mol fraction) CH4 0.96929 Case Study 1 [1] Feed data from [13]
C2H6 0.02938
C3H8 0.00059
nC4H10 0.0001
N2 0.00064
Ambient temperature 20 1C Compressed refrigerant is A 10 1C heat exchanger
cooled to 30 1C approach was assumed for
after-coolers except in open
the loop systems where the
nal recycle cooler has a
5 1C approach.
Compressor polytropic 83.5% Case Study 1 [1] Centrifugal compressor
efciency polytropic efciencies range
typically between 75 and
85%.
Turbo expander efciencies Expander adiabatic Manufacturers
efciency 80% recommended net
efciencies for expanders in
LPG & LNG applications.
Compressor polytropic
efciency 70%
Liquid expander adiabatic 70% [15]
efciency
Gas turbine fuel gas 4.3 MW of fuel per MW 23% thermal efciency Fuel gas allowances for the
requirements shaft power GCL process to minimise
the effect of nitrogen build
up.

4. Results

4.1. Processing conditions and constraints

The processing conditions are summarised in Table 1. Because of the level of detail provided by Lee [1] and
in follow-up communications [13], the conditions of his Case Study No. 1 were selected for the comparison of
the four liquefaction processes. Identical compressor and expander efciencies, and fuel gas basis were used
for all process models. The LNG production rate was specied as 22.6 kg/s (0.7 million tonnes per annum) of
LNG to storage at atmospheric pressure.
The nal temperature of the product was not the same in all the processes studied. It was therefore assumed
that any ash gas was recompressed to the feed pressure. Both SMR and cLNG produces a product stream at
around 157 1C and the amount of ash gas is low. Similar temperatures were not attainable in the other
processes without signicantly changing the reported process conditions and hence the performance of the
design. However, it should be noted that it can be advantageous to take the fuel gas from this recycled ash
gas at low or intermediate pressures, saving a small amount of compression. It also offers a way of removing
nitrogen from the open-loop schemes, which is important for the GCL concept and would be important for
ARTICLE IN PRESS
2012 C.W. Remeljej, A.F.A. Hoadley / Energy 31 (2006) 20052019

the other processes if the feed gas has higher nitrogen content. For much higher nitrogen content, a nitrogen
rejection step would need to be included in all the processes.
The compression systems have not been optimised to match a specic frame sizes and/or machine types.
However, some practical rules have been applied to ensure a valid comparison of the processes. The criterion
used to minimise the compressor power in each model was to keep all compression ratios less than a maximum
of 51. Where multiple stages were required, the compression ratios were balanced and the gas was cooled
between stages to 30 1C.
The feed gas preparation steps (water, CO2 and heavy ends removal) were omitted from the comparison.
These steps generally operate above ambient temperature, thus requiring heat input that can be supplied by
waste heat from the compressor gas turbine drivers. All four processes studied would be affected in a similar
manner if feed gas preparation processes were included [8].

4.2. Process simulation and validation

All four processes were simulated using commercial mass and energy balance software with the
PengRobinson equation of state [14]. The feed gas temperatureenthalpy relationship was checked
extensively against Lees data [1] and the PengRobinson model was found to give good agreement. Further
details can be obtained from [10].

4.2.1. SMR process simulation


Four different cases for the SMR model were reported by Lee and each of these was resimulated by
Remeljej [10]. The shaftwork power in each case was within 3% of the value reported by Lee. However, the
pinch points in the LNG exchanger (minimum temperature differences) in all of Lees cases were in the order
of 0.5 1C, which are considered too small for a practical design. In order to be able to compare processes in this
study, it was decided that the minimum temperature difference in the LNG exchanger should be 2 1C.
Therefore some reoptimisation was required. Increasing the LNG exchanger temperature difference can easily
be achieved by increasing the refrigerant ow rate. As the refrigerant ow was increased the three pinch points
opened up until a single pinch at the cold end existed. The single pinch equalled the difference between the
LNG product temperature and the sub-cooled refrigerant. For this single pinch situation to occur the
refrigerant ow rate had to increase by 30% with a corresponding 20% increase in power. This ow increase
resulted in the actual power to theoretical minimum power ratio being the same as that reported for the
commercial PRICO process.
Using a liquid expander rather than a valve to let down and cool the refrigerant is another way to open up
the pinch point at the cold end while also reducing the actual compressor power required. The liquid expander
reduces the refrigerant temperature for the same pressure levels compared to the valve and can potentially
recover useable work. The effect of adding the expander and increasing refrigerant ow from 3.4 to 3.6 kmol/s
gave pinch points at the cold end of greater than 2.0 1C. These changes only increased the shaftwork by 4%
compared with Lees design. The composite curves for the expander case are shown in Fig. 5.

30

-20
Temperature (oC)

-70

-120

-170
0 20 40 60 80 100
Heat Load (MW)

Fig. 5. Hot and cold composite curves for the SMR process with the liquid expander.
ARTICLE IN PRESS
C.W. Remeljej, A.F.A. Hoadley / Energy 31 (2006) 20052019 2013

30

Temperature (o C)
-20

-70

-120

-170
0 10 20 30 40
Heat Load (MW)

Fig. 6. Hot and cold composite curves for the New LNG process.

30
Temperature (o C)

-20

-70
Chiller duty required
to open pinch
-120

-170
0 10 20 30 40 50
Heat Load (MW)

Fig. 7. Hot and cold composite curves for the GCL concept process.

4.2.2. New LNG process simulation


Signicant changes were required to the New LNG process from the published process in order to
accommodate different feed conditions. Foglietta [5] gave the inlet pressure as 67 bar (compared with 55 bar),
the inlet temperature as 37.8 1C (compared with 25 1C) and the feed gas composition was not reported.
With the feed conditions chosen for this study, it was found that the propane refrigeration levels in [5] were
no longer optimal. By inspecting the composite curves in this paper, it was concluded that the basis of
optimisation of the propane levels in the New LNG Scheme was two stages with roughly equal duties (colder
duty 57% and warmer 43%). Applying this same principle gave a new intermediate temperature of 5 1C,
which also gave the lowest shaftwork requirement.
Apart from the re-specication of the intermediate propane refrigeration level, the model of the New LNG
Process with Lees feed was constructed by matching the temperatures for the exit of the propane pre-cooling
section, the entry to the expander, and the inlet to the ash for the nal LNG production [5]. The expander
pressure drop and recycle ow were manipulated to ensure a feasible solution, as shown in Fig. 6. The
different feed conditions resulted in a minor reduction in the power consumption of less than 1% to
6.28 kW h/kmol compared with [5].

4.2.3. GCL concept


The GCL LNG ow sheet was constructed from limited information in [6]. The LNG product stream is
taken directly from the expander outlet with a liquid fraction of around 30%. A high expander liquid outlet
fraction is required to minimise the recycle ow and hence the total power input. To achieve this liquid
fraction, it is necessary to operate the expander inlet at, or just to the left, of the critical point of the gas. These
conditions are the current upper operational limits of isentropic efciency, size and liquid exit fraction for
expander technology.
A critical design parameter is the owrate of the purge stream to reduce the nitrogen content in the
expander feed. Without any purge the nitrogen accumulation is severe as the liquid drawn off from the
ARTICLE IN PRESS
2014 C.W. Remeljej, A.F.A. Hoadley / Energy 31 (2006) 20052019

30

-20

Temperature (o C)
-70

-120

-170
0 10 20 30 40
Heat Load (MW)

Fig. 8. Hot and cold composite curves for the cLNG process.

Table 2
Power requirements for all processes

SMR New LNG cLNG GCL concept

Specic power at 157 1C (kW h/kmol) 5.67 6.28 6.79 6.64


Ratio of power to minimum theoretical power 2.81 3.47 3.52 3.56

separator on the exit of the expander is essentially nitrogen-free, causing all the nitrogen to be recycled. This
accumulation of nitrogen drives the process power requirements to impractical and uncompetitive levels. In
order to provide a practical model for this study, the purge rate was set equal to the typical fuel requirements
for the gas turbine drivers required to provide the entire schemes compression power requirements.
The composite curves for the process are shown in Fig. 7. This shows that a chiller is required to open up the
pinch. The pinch temperature decreased as more liquid was recovered from the TE outlet and the slope of the
cold composite line increased. The chiller duty would need to increase, on decreasing the recycle compressor
duty.

4.2.4. cLNG process


A model of the cLNG scheme was matched with the temperature data presented in [4]. It was found that the
power requirements were a little lower than those originally reported. This may be due to using a higher
compressor and/or expander efciencies in this study. However, the composite curve for the model using this
feed gas and that in [4] indicate from the similarity between the curves that the model is a suitable
representation of the cLNG process. The composite curve for the cLNG scheme with no fuel gas take-off is
shown in Fig. 8.

4.3. Exergy analysis

The exergy analysis is conducted from a converged simulation of each process. The output le from the
simulation is used to calculate the lost work for each unit operation using Eqs. (1)(3). The reference
conditions are 20 1C and 1 atm pressure. The PengRobinson thermodynamics package is again important in
generating the enthalpy and entropy data at the stream conditions. Table 2 summarises the lost work, power
requirements, and specic power for each process model.
ARTICLE IN PRESS
C.W. Remeljej, A.F.A. Hoadley / Energy 31 (2006) 20052019 2015

0
SMR
New LNG
cLNG
-0.5 GCL
Feed

Carnot Factor -1

-1.5

-2
0 20 40 60 80 100
Heat Load (MW)

Fig. 9. Hot Composite Curves for all processes and the feed stream plotted as Carnot factor versus heat ow.

-0.5
Carnot Factor

-1 SMR
New LNG
cLNG
-1.5
GCL
Feed
-2
0 20 40 60 80 100
Heat Load (MW)

Fig. 10. Cold Composite Curves for all processes and the feed stream plotted as Carnot factor versus heat ow.

5. Discussion

The models developed for each process allow the composite curves and exergy loss across the main unit
operation of each process to be directly compared. Comparison of these quantities provides insight into the
performance differences between processes and in some cases indicates potential for improvement.

5.1. Composite curves

Fig. 9 shows the Hot Composite curves for each process plotted as Carnot factor versus heat load. Also
included is the feed gas-cooling curve for the feed gas stream at 50 bar. The area between the horizontal axis
and the feed curve represents the absolute minimum reversible work required to cool the feed stream to the
product LNG conditions of 165 1C. The different slopes of the remaining curves, relative to the feed curve,
represent the different heat capacity ow rates associated with the self-cooling duty of the recycle and
refrigerant streams within the process that is additional to the feed gas-cooling duty.
ARTICLE IN PRESS
2016 C.W. Remeljej, A.F.A. Hoadley / Energy 31 (2006) 20052019

The steepest curve of the four LNG processes is that of the cLNG process. This shows that the cLNG
refrigerant ow rate is relatively small. The kink in the curve represents the intermediate pressure level between
the two stages of the nitrogen cycle where only part of the refrigerant ow is required in the coldest stage.
The hot composite curves for both open-loop systems terminate well above the bubble point of LNG. This
reduces substantially the amount of theoretical reversible shaft work required by the process. However, in
both cases there are exergy losses across the letdown valve, the expander, and the ash/recycle gas (refrigerant)
must be recompressed.
The cold composite curve reects the irreversibilities existing within the LNG heat exchanger system. Fig. 10
shows the Cold Composite curves for each process plotted as Carnot factor versus heat load. Again for
comparison, the feed curve is shown with no minimum temperature difference between the hot and cold
composite curves. In each case, the area between the axis and the cold composite curve represents the
reversible work required by each process.
The heat load for the SMR curve is substantially greater than the other three processes. In the SMR process,
the cold refrigerant stream must remove all of the heat in the LNG and refrigerant recycle, through the heat
exchange network. For the other three processes, the expanders remove a substantial amount of energy below
ambient temperature. The substantial exergy losses associated with the expanders are not reected in Fig. 10.
The cLNG scheme and the GCL process have the lowest heat demand. The cLNG scheme must remove
heat continuously down to 160 1C. By contrast, the GCL process has only a small contribution below
120 1C due to the ashed vapour.
The ability to match the hot and cold composite curves presented in Figs. 58 demonstrates the degree of
optimisation available within each process. The area between the hot and cold composite curves for each
process is directly proportional to the exergy loss over the LNG heat exchanger system. In the case of the
SMR process (Fig. 5), the use of a multi-component system gives the design a large number of degrees of
freedom, including the precise refrigerant composition, ow rate and pressure, which in turn allows for a close
match between the composite curves. In the case of the New LNG scheme Fig. 6, the degrees of freedom
include the expander ow rates and pressures. The propane pre-cooling produces at spots, which are not
desirable for a close approach between the composite curves, in the cold composite curve.
In the case of the cLNG process (Fig. 8), the points of optimisation are in the choice of both the warm and
the cold expander ow rates and pressures. The two sections of the cold curve are linear and are able to match
well the hot composite curve because the nitrogen does not change phase.
The process with the least degrees of optimisation is the GCL process. The end point of the hot composite
curve is xed by the very specic conditions required on the inlet of the turbo expander (Fig. 7). At the same
time, the ash gas temperature exiting the rst separation vessel xes the cold end of the cold composite. The
slope of the cold composite curve is also xed by the ash gas ow rate. The large gap between hot and cold
composite curve indicates that there may be room for further improvement to this process.

5.2. Exergy analysis

In the previous section the processes were compared using the LNG exchanger composite curves only. The
heat load gives the amount of heat to be removed, while the Carnot factor gives an indication of the degree of
difculty of this heat removal. The lower the Carnot factor the more difcult it is to remove the heat.
However, the Carnot factor is based on a reversible process. In reality the refrigeration processes which
include the compressors, expanders and let down valves all involve irreversible operations. In this section, all
of the exergy losses across the different processes will be compared by unit operation.
Fig. 11 shows graphically the specic power requirement and also the patterns of lost work. The combined
losses for the compression system, including after-coolers, are the most signicant contributor to lost work.
The lost work in this area is a direct result of the efciency of the compression system and must be evaluated in
conjunction with the compressor.
In the case of the SMR process, two-thirds of the total lost work is attributable to the compression and
after-cooling of the refrigerant stream while about a one-third is due to the driving force across the LNG
exchanger. Clearly, methods of reducing the losses in the compression and after-cooling section would
produce the greatest effect in this scheme. For example, the current practice of providing two-stage
ARTICLE IN PRESS
C.W. Remeljej, A.F.A. Hoadley / Energy 31 (2006) 20052019 2017

Min Reversible specific power Specific Power


Specific Power kWh/kmol

7
6
5
4
3 Large scale LNG
2
1
0

3% 11% 11% 8% 10%


22%
30%
17% 22% 19%
SMR New cLNG
46% 27% GCL 45%
LNG
19%

21% 33% 28% 25%

LNG HX Compression x% Aftercooling

Compander Mixers Valves / liq. expanders

Fig. 11. Specic power requirements and patterns of lost work for each process.

compression with inter-cooling reduces the lost work by 17%, with a corresponding shaftwork savings of
11%. Three equal stages gives a lost work reduction of 27% compared to the single stage and a shaftwork
reduction of 16%.
For all the schemes adopting a multistage compression system makes it practicable to add additional stages
to the refrigeration system and further reduce power requirements. Splitting the compression system into
multiple stages with inter-cooling does not affect the characteristics of the refrigerant, nor the LNG heat
exchanger system, because it does not change the composite or grand composite curves in any way. It may be
possible in some specic instances to increase the compressor polytropic efciency (e.g. use an axial
compressor for the rst stage) or take advantage of heat integration with the surrounding plant, should
opportunities exist to cool the inter-stage exhaust to lower temperatures than can be achieved with utilities. As
a result small efciency gains made in this area on these processes will have a measurable effect on the overall
process efciency.
In the case of the three turbo-expander processes, Fig. 11 shows the turbo-expander losses grouped as a net
loss over the machine so that the value represents the net inefciency of the compressor side relative to the
work extracted. Any improvements sought should focus on the efciency on the compressor side as its
isentropic efciency is low compared to separately driven compressors (refer to Table 1). Any such
improvements would be analogous to improvements discussed above for the compressors and associated after-
cooling systems. Improvements in this area would not affect the LNG heat exchanger system. Also, any slight
increase in the isentropic efciency of the expander side would reduce the required refrigerant ow rate,
resulting in a power saving. An increase in efciency may be approached with small plants if the expander is
used to drive a generator and the electricity is recovered for use in an electrically driven compressor with a
higher efciency.
Fig. 11 also shows that the LNG exchanger contributes 30% of the exergy loss in the SMR process, 20% for
GCL and only 10% in the New LNG and cLNG schemes. For the SMR process, Fig. 5 shows the gap between
the hot and cold composite is predominately at the hot end, indicating that this is the area with most
opportunity for lost work reduction. However, given that the process has already been mathematically
optimised for the single-stage refrigerant system, the opportunities to be explored to remove the hot end bulge
no longer include both refrigerant composition and pressure as long as the system is maintained as a single
ARTICLE IN PRESS
2018 C.W. Remeljej, A.F.A. Hoadley / Energy 31 (2006) 20052019

stage refrigeration system. This leaves two options, viz., further refrigerant system pressure staging and/or
additional external chilling (for example propane pre-cooling). Both options will allow the mixed refrigerant
composition and operating pressure stages to be optimised further.
As already mentioned, most of the 20% lost work across the GCL exchangers occurs at the cold end of this
process, between the cold recycle gas and the feed. Improvements to this process could involve the use of a
cold nitrogen cycle operating between 160 and 120 1C to condense more of the feed gas and close this gap.
Similarly for cLNG, a small improvement can be gained through the use of an expander on the feed stream to
extract a small amount of work and close the composite curves.
There are also measurable losses in the order of 10% of the total loss over valves within the GCL and New
LNG Process Scheme that may be reduced using additional expanders with work recovery.
Table 2 provides the comparison of the specic power consumption for each process and the ratio of
required specic power to minimum reversible specic power. Due to the different product temperatures for
each of the processes studied, the specic power for each process has also been calculated for a standardised
product temperature of 157 1C. In terms of overall process efciency, the SMR process has the lowest
specic power requirement of 5.67 kW h/kmol with single-stage compression, and 5.10 kW h/kmol with a two-
stage system. The New LNG scheme has a 10% higher energy cost compared with the single-stage SMR
process and the other two processes are 67% higher again. However, the possible changes to both GCL and
cLNG indicate that a hybrid of these processes might give further efciency improvements and close the 10%
gap with new LNG.

5.3. Other factors for a development project

In addition to the energy efciency, the choice of the process for a small-scale LNG development onshore or
offshore would be based on the two main parameters of safety and project lifecycle cost. Capital cost in all
cases would signicantly inuence the lifecycle costs.
Life cycle cost is primarily a function of capital cost and operating cost both of which increase as the
number and size of equipment items increase. For the LNG processes studied here, the relative equipment size
is expected to increase as power requirements increase due to higher refrigerant circulation rates. These costs
are also affected by the complexity of the process arrangement, and its sensitivity to start up, feed gas
composition changes and potential errors in the thermodynamic modeling requiring higher design margins.
Table 3 summarises the basic equipment requirements for all four processes. It should be noted that not all
the equipment required to make the schemes practicable, for example compressor suction scrubbers, are
shown in the ow sheets diagrams (Figs. 14). The simplicity of both the SMR and cLNG processes is
demonstrated by their low equipment counts.

Table 3
Process scheme equipment count

Equipment No. of equipment items

SMR New LNG GCL cLNG

Hydrocarbon N2

Compressors (stages) 2 6 5 1 4
Expanders 0 1 1 0 2
Suction scrubbers 2 6 4 1 0
After coolers 2 5 5 1 4
Separators 1 2 2 0 0
Pumps 1 2 2 1 0
LNG heat exchanger 1 1 1 0.5 0.5
Equipment count 9 23 20 4.5 10.5
Total overall 9 23 20 15
ARTICLE IN PRESS
C.W. Remeljej, A.F.A. Hoadley / Energy 31 (2006) 20052019 2019

6. Conclusions

Four different LNG processes have been investigated in this study. It was found that a single general analysis
method, such as proposed by Linnhoff and Dhole [12] for cold processes, was not able to provide a direct compa-
rison because of three main differences between the processes. These were the type of refrigerant used, whether the
scheme is open-loop or closed-loop, and the use and placement of the expanders. As a result, optimisation
methods have to be developed for each type of process, as demonstrated by Lee [1] for the SMR process.
However in the absence of a single optimisation method, an exergy analysis was able to provide valuable
insight into areas that could be targeted for improvement within a particular process. The Carnot factor versus
heat ow diagrams for the heat exchanger systems were used to identify where in the temperature range there
might be signicant inefciencies. The lost work for the heat exchanger was directly proportional to the area
between the hot and cold composite curves. However, the trade-off for a decrease in lost work over the LNG
exchanger system was increased area and hence cost.
The performance of the two open-loop schemes was found to be more sensitive to feed gas composition and
expander performance compared to the non-open-loop processes. As a result the open-loop schemes may not
be as practicable in as many cases as closed-loop schemes, especially for offshore use. Analysis of the GCL
concept showed that expanders placed in the feed stream to produce the LNG directly provided no overall
signicant advantage compared to the other processes studied. This was the case even if the expander was able
to provide the current maximum permissible liquid recovery fractions of 40% by weight.
The exergy analysis showed that the SMR process had the lowest specic power requirement of 5.67 kW h/kmol
(single-stage compression) and 5.10 kW h/kmol (two-stage compression), followed by the New LNG process and
then the other two processes. The exergy analysis also showed areas for improvement in efciency still exist within
each of the schemes. Examples include adding a third compression stage to the SMR process or separating the
expander from the compressor using an electrical generator. A hybrid of the GCL and cLNG schemes may also
lead to an improvement in efciency to both processes. However, any further optimisation of the LNG processes
would be a trade-off between cost (capital and operating) and energy efciency, each of which may be affected by
factors peculiar to a particular project. Other factors of importance are operability, reliability, and safety.

References

[1] Lee G-C. Optimal design and analysis of refrigeration systems for low temperature processes. PhD, School of Chemical Engineering
and Analytical Science, The University of Manchester, UMIST, Manchester, UK, 2001.
[2] Price BC. PRICOa simple, exible proven approach to natutral gas liquefaction. In: Proceedings of the 17th international LNG/
LPG conference, Gastech 96, Vienna, 1996.
[3] Price BC. Small-scale LNG facility development. Hydrocarbon Process 2003;82(1):379.
[4] Dubar C, Forcey T, Humphreys V, Schmidt H. A competive offshore LNG scheme utilising a gravity base structure and improved
nitrogen cycles. In: Proceedings of the LNG 12, Institute of Gas Technology, Perth, WA, Australia, May 47, 1998.
[5] Foglietta JH. New LNG process scheme. In: Proceedings of the 78th GPA Annual Convention, GPA, 1999. p. 2816.
[6] Maunder AD, Skinner GF; Gas Consult Ltd., assignee. Method for liquefying methane rich gas. WO patent 03/019095 A1, 20 August
2002.
[7] Liu LY, Daugherty TL, Brofenbrenner JC. LNG liqueer efciency. In: Proceedings of the LNG 12, Institute of Gas Technology,
Perth, 1998.
[8] Foerg W, Bach W, Stockmann R, Heiersted RS, Fredheim AO. A new LNG base load process and the manufacturing of the main
heat exchangers. In: Proceedings of the LNG 12, Institute of Gas Technology, Perth, WA, Australia, May 47, 1998.
[9] Vink KJ, Nagelvoort RK. Comparison of base load liquefaction processes. In: Proceedings of the LNG 12, Institute of Gas
Technology, Perth, WA, Australia, May 47, 1998.
[10] Remeljej CW. An exergy analysis of small scale LNG processes. Masters minor thesis, Department of Chemical Engineering, Monash
University, Melbourne, Australia, 2004.
[11] Seider WD, Seader JD, Lewin DR. Process design principles. New York: Wiley; 1998. p. 20742.
[12] Linnhoff B, Dhole VR. Shaft work targets for low-temperature process design. Chem Eng Sci 1992;47(8):208191.
[13] Smith R. Personal communication between Professor Robin Smith and C.J. Remeljej, 2002.
[14] Twu CH, Coon JE, Cunningham JR. A new generalized alpha function for a cubic equation of state. Part 1. PengRobinson
equation. Fluid Phase Equilibria 1995;105:4959.
[15] Rawlins H, Bond T. LNG and Industrial refrigeration performance improvements with multiphase turbines. In: Moneitization de
reserves a traves de LNG-GTL, Buenos Airies, Argentina, July 31August 1, 2001.

You might also like