COMPARISON OF BASELOAD
LIQUEFACTION PROCESSES
COMPARAISON DES PROCEDES DE LIQUEFACTION
DES USINES DE GRANDE CAPACITE
K.J. Vink
R. Klein Nagelvoort
Shell International Oil Products, B.V.
PO box 451, 2501 CM
The Hague, The Netherlands
ABSTRACT
This paper summarises a recent comparison of five baseload LNG processes. The
range of technology encompasses the Propane/MR process, a Cascade process, a version
of the Dual Mixed Refrigerant process, a simple Single Mixed Refrigerant process and a
pre-cooled Nitrogen expansion process. Each process was assessed on the basis of specific
costs, specific power and fuel efficiency for a nominal 2 train liquefaction plant.
The conclusions are:
• The Propane/MR process appears to be the best choice within the premises of this
comparison study, viz. large capacity LNG trains, employing air cooling in a tropical
climate. Other promising processes are the Dual Mixed Refrigerant process and the
Single Mixed refrigerant process.
• The Cascade process appears to be relatively expensive, partly disadvantaged as it is by
the study premises. Under colder conditions (arctic, water cooling), the costs and
power requirements come closer to the Propane/MR process.
• The pre-cooled Nitrogen Expansion process is not an economic choice for a large
onshore application. It may be an alternative for smaller scale offshore applications.
RESUME
Cet article résume une récente comparaison de cinq procédés à GNL de base. La
gamme de technologies inclut le procédé à propane/Réfrigérant Mélangé (RM), un
procédé à cascade, une version du procédé à réfrigérant double mélangé, un procédé à
réfrigérant unique mélangé et un procédé à expansion d’azote pré-réfroidi. Chaque
procédé a fait l’objet d’une évaluation basée sure les coûts spécifiques, l’énergie spécifique
et la rentabilité de carburant pour une usine de liquéfaction de 2 chaînes nominales.
3.6–1
Les conclusions sont:
• Le procédé à propane/RM s’est révèlé la meilleure alternative dans le cadre de l’étude
comparative, c’est-a-dire des chaînes de GNL de grande capacité, employant le
refroidissement par air sous un climat tropical. Les autres procédés prometteurs sont le
procédé à réfrigérant double mélangé et le procédé à réfrigérant unique mélangé.
• Le procédé à cascade se révèle relativement coûteux, en partie désavantagé comme tel
par le cadre de l’étude. Dans les conditions plus froides (zone arctiques,
refroidissement par eau), les coûts et les besoins en énergie se rapprochent de ceux de
procédé à propane/RM.
• Le procédé à expansion d’azote pré-réfroidi n’est pas une alternative économique pour
une grande installation côtière mais il peut l’être pour des installations offshore de
dimensions plus petites.
3.6–2
COMPARISON OF BASELOAD
LIQUEFACTION PROCESSES
INTRODUCTION
The liquefaction unit accounts for approximately 50 % of the total capital costs of 1-2
billion US$ for a baseload LNG plant. The type of liquefaction process, in combination
with the rotating equipment and ambient cooling system, affects the capacity and
availability of the entire LNG system so it is essential that the best process is selected.
The world of baseload LNG processes is dominated by the propane pre-cooled/mixed
refrigerant (C3/MR) process, introduced by Air Products and Chemicals, Inc. in the early
seventies. Nearly 95 % of the word wide LNG production capacity is based on this
process in various forms, employing steam turbines, and gas turbines as compressor
drivers and water and air cooling for heat rejection. Over the years APCI was able to
enlarge its spoolwound main cryogenic heat exchanger and to adopt the C3/MR process
for a wide range of conditions. Shell has further advanced the C3/MR process by the
introduction of large gas turbines/compressors, liquid expanders, air cooling, new gas
scrubbing and Nitrogen rejection systems, etc., thus improving the competitiveness of the
process. Unit sizes up to 4 MTPA are now possible.
The recent upsurge in the LNG industry stimulates new developments and a revival of
old LNG processes, especially for use in smaller schemes, on floaters, concrete gravity
structures, barges, etc. For instance, with the development of the Atlantic LNG project the
Phillips optimised Cascade process has received renewed interest. Pritchard has also
improved the power efficiency of its Single Mixed Refrigerant (SMR) based PRICO
process and now proposes it for applications onshore and offshore. A new development
originating from LNG peak shaving is the Compact LNG or CLNG process developed by
BHP Petroleum and Linde AG of Germany for an offshore application (Bayu-Undan in the
Timor Sea). The CLNG process is based on a pre-cooled (chilled water) nitrogen
refrigeration system and it utilises a Linde spoolwound heat exchanger.
It is therefore timely that a comparison is conducted to assess the latest process
developments.
BASIS FOR COMPARISON
A “like-for-like” comparison of the different process is difficult to achieve in practice
because the technical content is not available in the public domain. Shell has therefore
conducted a comprehensive study comparing an optimised C3/MR process with the best
of four alternative processes, starting from basics and using the same conditions like
cooling medium, feedgas, standards and cost basis. The processes are:
• The Propane/Mixed Refrigerant process
• The Cascade process
• A version of the Dual Mixed Refrigerant (DMR) process
3.6–3
• A version of the Single Mixed Refrigerant (SMR) process
• The pre-cooled Nitrogen Expansion process.
It is realised that for each of these processes several options exist in the configuration
of the process that will influence the capacity and overall attractiveness. This variation
within a process results from the hardware selected and the particular gas turbine drivers
and cryogenic heat exchangers.
This study focused on large liquefaction units with a capacity in the range of 3-4
MTPA, installed onshore in a tropical location. For each process an optimum capacity has
been assessed within this capacity range. Air cooling is used for heat rejection, minimising
pre-investments for water cooling systems.
The differences in capacity, refrigerant make-up, power consumption, etc. of the
various liquefaction units are also reflected in the fractionation and utility requirements. It
was therefore decided to extend the scope of the study to include all process facilities and
related utilities in a combined liquefaction package. Each package consists of 2 complete
LNG trains with a common fractionation/refrigerant make-up unit and utilities like
electrical power, process heat, etc. Such a liquefaction package can be added to an
existing LNG plant or form the heart of a new LNG plant.
For consistency of comparison the LNG trains have a similar pre-treating line-up, i.e. a
Sulfinol (Shell licence) unit for acid gas removal, molecular sieve dryers, a mercury
removal bed and a simple scrubbing system for removal of heavy hydrocarbons. The
premises of the study are:
1.
2.
3.
4.
Gas feed available at 60 bara and 25 °C; gas composition as shown in Table 1.
Average ambient air temperature 27 °C
LPGs recovered are re-injected into the LNG.
All processes are fully air cooled. The air cooler banks govern to a large extent the
total train plot area. For comparison the process conditions are selected such that
the air cooler banks and train plot areas are approximately equal.
5. LNG storage and loading facilities and general facilities are outside the scope of the
study.
Table 1. Feed gas composition
Component
N2
CO2
C1
C2
C3
C4
C5plus
Total
3.6–4
mol %
1.5
2.2
85.1
6.5
3.0
1.2
0.5
100
THE ALTERNATIVE PROCESSES
The flow schemes of the five liquefaction processes are shown in Figures 1, 2, 3, 4,
and 5 below.
Figure 1. Simplified flowscheme of the propane/MR process
A recently developed C3/MR process is used as the reference case in this comparison
study. The process uses propane as pre-cooling medium and a mixed refrigerant (nitrogen,
methane, ethane and propane) as liquefaction medium. A GE-7EA driven compression
train pumps around MR, which is partly condensed against air and four stages of propane
cooling. The vapour and liquid refrigerant fraction as subsequently auto-cooled and
expanded, such as to achieve matching cooling curves in a spoolwound main cryogenic
heat exchanger of maximum proven size. The natural gas is liquefied in this heat
exchanger. A four-stage propane cycle provides the pre-cooling for the MR and the
natural gas. The propane compressor is also driven by a GE-7EA gas turbine. Moreover,
to further enhance LNG production a Shell patented endflash system has been used.
Within the scope of this study, two of such LNG trains require two electricity
generation gas turbines of 20 MWe to sustain operation. A relatively low electrical power
consumption is achieved because excess power from the C3 compressor gas turbine is
transferred to the MR compression gas turbine through an electric coupling. Hereby the
starter motor of the first gas turbine acts as a generator supplying electricity to the helper
motor of the second gas turbine. This part of the process is covered by a Shell patent.
3.6–5
Figure 2. Simplified flowscheme of the Cascade cycle
The Cascade process is a multiple refrigerant system wherein the lowest boiling
temperature stage of each refrigerant is used in turn to condense the next refrigerant. The
process in this comparison study uses pure refrigerants in the consecutive cooling steps,
viz. propane and ethylene, both in closed, three stage cycles and finally methane in a four
stage open cycle. Core-in-kettle type heat exchangers and plate-fin heat exchangers are
used for cooling of the natural gas and for cold recovery. The use of these exchanger
types allows very low temperature approaches.
To limit the compression ratio of the methane compressor the LNG rundown is above
atmospheric conditions and runs flashing into the LNG tank. Hence boil-off gas
compression forms part of the methane cycle. In order to limit the build-up of nitrogen in
the methane cycle fuel is taken from the methane compressor.
Six GE-5C gas turbines are used as compressor driver, distributed in pairs over the
three refrigerant loops. Each gas turbine drives one compressor. The gas turbines driving
the propane compressors are equipped with 7 MW electrical helper motors to allow
balancing the loads over the ethylene and methane cycles. The methane compressor is
rather complex, requiring three casings.
Two Cascade LNG trains require in total three electricity generation gas turbines of 20
MWe for power supply, and separate ethylene storage/loading facilities. It is assumed that
ethylene needs to be brought in by ship.
3.6–6
Figure 3. Simplified flowscheme of the Dual Mixed Refrigerant process
The Dual mixed Refrigerant process uses a mixture of methane, ethane, propane and
butane as precooling medium. The compressed mixture is fully condensed against air and
subsequently auto-cooled and expanded to provide refrigeration duty. The expansion can
be performed at one, two or three pressure levels. In the comparison study a three stage
pre-cooling cycle was selected. A GE-7EA type driver plus 6 MW helper power and two
maximum size spoolwound heat exchangers are used in the pre-cooling circuit.
Alternatively plate-fin heat exchangers in cold boxes can be used. The liquefaction circuit
and nitrogen rejection system resemble to a large extent the liquefaction circuit of the
C3/MR process.
Two DMR type LNG trains require in total three electricity generation gas turbines of
20 MWe for power supply.
3.6–7
Figure 4. Simplified flowscheme of the Single Mixed Refrigerant Process
In the Single Mixed Refrigerant Process shown here, one refrigerant provides the total
cooling from ambient to LNG temperatures at one pressure level. Two GE-7EA driven
compressor strings, operating in parallel, each with 6 MW helper power are used to pump
around the refrigerant, that contains components ranging from nitrogen to pentane. One
compressor string consists of an axial compressor followed by a centrifugal compressor.
The refrigerant partially condenses in the interstage and discharge air coolers. The
compressed vapour refrigerant fraction is combined with the pumped liquid fractions in the
inlet of plate-fin heat exchangers, where the mixture is auto-cooled and expanded such as
to achieve matching cooling curves. The natural gas is pre-cooled and liquefied in the
same heat exchangers. A large number of parallel exchangers is distributed over a number
of cold boxes. Alternatively a number of spoolwound heat exchangers can be used. The
endflash system employed is identical to the endflash system used for the C3/MR process.
Two of such Single Mixed Refrigerant LNG trains require three electricity generation
gas turbines of 20 MWe for power generation.
3.6–8
Figure 5. Simplified flowscheme of the Nitrogen Expansion process
The Nitrogen Expansion process uses Propane as precooling medium and Nitrogen as
liquefaction refrigerant. A GE-7EA driven axial compressor pumps around Nitrogen that
is pre-cooled against propane. The nitrogen is subsequently auto-cooled and expanded via
three turbo-expander sets, such as to achieve a good match of the cooling curves in the
cryogenic heat exchangers. A plate-fin type was selected here, but a spoolwound heat
exchanger could have been used as well. A four-stage propane cycle provides the
precooling for the nitrogen and the natural gas. A GE-5C gas turbine to drives this
compressor with a helper motor. An advanced nitrogen rejection system, identical to the
one used for C3/MR is used.
Two of such Nitrogen Expansion LNG trains require a total of three electricity
generation gas turbines of 20 MWe for power generation.
HARDWARE CONSIDERATIONS
Rotating Equipment
The gas turbines driving the refrigerant compressors are of the heavy industrial type,
either the GE-5C (dual shaft, variable speed) or the GE-7EA (single shaft, fixed speed),
both from General Electric. These gas turbines have a track record in the LNG industry
and have low specific power costs. Each process is optimised such that it utilises as far as
possible the available powers, supplemented by helper motors where economical. The
multi-stage compressors in pre-cooling and in the Cascade process are of the centrifugal
type, in some cases stretched to the maximum flow and head limits. High efficiency axial
compressors are used for the first stage of mixed refrigerant compression and for nitrogen
3.6–9
compression. The selection is based on in-house experience and feedback from rotating
equipment vendors. Electric motors are used as starter/helper and for driving the endflash
compressor. A summary of the rotating equipment used is given in Table 2.
Table 2.
Summary of main rotating equipment.
Process
C3/MR
Cascade
DMR
Precooling
Gas turbine
Precooling
compressor
Liquefaction
gas turbine
Liquefaction
compressor
GE-7EA (+
generator)
4 stage
centrifugal
GE-7EA (+
helper)
Axial plus 2
stage
centrifugal in
tandem
2 * GE-5C
(+ helper)
2 * 3 stage
centrifugal
4 * GE-5C
GE-7EA (+
helper)
3 stage
centrifugal
GE-7EA (+
helper)
Axial plus 2
stage
centrifugal
in tandem
Ethylene: 2
* 3 stage
centrifugal
Methane: 2
* 4 stage
centrifugal, 3
casings
SMR
N2
expansion
N/A
GE-5C (+
helper)
N/A
4 stage
centrifugal
2 * GE-7EA GE-7EA (+
(+helper)
helper)
2 * Axial
Axial
plus single
Nitrogen
stage
compressor;
centrifugal
Three
in tandem.
expander/
compressor
sets
Heat Exchangers
The cryogenic heat exchangers are of the aluminium spoolwound design, plate-fin or
core-in-kettle type, the selection being based on operating experience, cooling
characteristics and feedback (including sizes and costs) from vendors. A summary of the
cryogenic heat exchanger selection is found in Table 2.
Table 3.
Summary of main cryogenic heat exchange equipment
Process
C3/MR
Cascade
Precooling
exchangers
Liquefaction
exchangers
Kettle
Core-inSpoolwound
kettle
Core-inSpoolwound
kettle/ PlateFin
Spoolwound
DMR
SMR
Plate-Fin
N2
expansion
Kettle
Plate-Fin
Plate-Fin
Table 4 below summarises the minimum number of 20 MW turbo-generators required
for the different process options (excluding sparing)
3.6–10
Table 4. Summary of electricity generation requirements
Process
C3/MR
Number of 20 MW 2
turbo-generators
cascade
DMR
SMR
3
3
3
N2
expansion
3
PROCESS COMPARISON
LNG Production
The process calculations were aimed at maximisation of LNG production within a set
of equipment constraints such as available gas turbine power, approach temperatures on
air coolers, maximum spoolwound heat exchanger size, etc.
Optimisation of the five processes resulted in the LNG production figures shown in
Table 5.
Table 5. Daily LNG production for the five processes
(per LNG train, in t/d, excluding process margin)
Process
C3/MR
Cascade
DMR
SMR
LNG production
11900
10000
13100
11300
N2
expansion
6540
The processes have production rates within the target range with the exception of the
nitrogen expansion process. The low production of this process is due to the low cold
potential of expanded nitrogen (sensitive heat rather than latent heat) and the lower
powers installed.
Specific Power
A crude yardstick for cost and efficiency comparison is the specific power of a
process. This is the ratio of total compressor shaft power absorbed (including all
refrigerant compressors and the endflash compressor, excluding contributions from liquid
or gas expanders) over the LNG rundown in t/d. The result is shown in Table 6.
Table 6.
Process
C3/MR
Specific power 12.2
(kW/(t/d
LNG
rundown))
Specific power of the five processes.
Cascade
DMR
SMR
14.1
12.5
14.5
N2
expansion
15.6
Inspection of table 6 shows that the C3/MR and the DMR process have comparable
low specific powers, indicating a high degree of optimisation for an air cooled process.
3.6–11
The Cascade process appears to suffer disproportionally from the high ambient cooling
temperatures. It proved to be possible to fully load all gas turbines, albeit that the propane
compressor drivers had to be equipped with sizeable helper motors. Due to the nature of
the process no liquid expanders can be used in the LNG and refrigerant streams to boost
production. Furthermore the smaller compressors in the Cascade process tend to have
lower efficiencies than the large centrifugal and axial machines in the C3/MR and DMR
process.
It is also noted that the Cascade process, employing pure components, after
optimisation for one point of operation, contains no longer any degree of freedom, except
the speed of the gas turbines. On the other hand, the C3/MR, DMR, and also the SMR
process do contain process variables that enable optimisation of process performance
outside the exact design point.
The SMR specific power of 14.5 kW/t/d, albeit high as compared to C3/MR, is a good
achievement for such a single pressure process.
Train Efficiency
Each liquefaction train provides for its own fuel, mainly endflash gas, and a proportion
of the fuel required in the utility area.
The Train efficiency of the processes is defined as the total HHV of valuable products
(LNG and condensate) divided by the total HHV of the feed; hence this covers both
process efficiency and the efficiency of the gas turbines. It is hereby assumed that waste
heat of the gas turbines is only used for providing process heat and not for steam/power
generation. Such a kind of combined cycle operation would improve total efficiencies
further, if this can be justified. The train efficiencies of the liquefaction processes are
shown below in Table 7.
Table 7.
Train efficiencies of the LNG processes
Process
C3/MR
Cascade
DMR
SMR
Fuel efficiency (%)
92.9
91.2
92.7
91.6
N2
expansion
90.4
Table 7 clearly shows the high train efficiency of the DMR, C3/MR and the SMR
processes. This is not surprising since these processes use large drivers with a relatively
high fuel efficiency of some 31 %. The Cascade process uses gas turbines with a lower
fuel efficiency (25 %). This process has 6 gas turbines and therefore does not lend itself
very well for full waste heat recovery, should this be required. The Nitrogen expansion
process has the lowest fuel efficiency, mainly due to the low process efficiency.
Plant Capex
Detailed cost estimates were prepared for two complete liquefaction trains and
associated common utilities, using vendor quotes where applicable. The results are
presented in Table 8 below in an indexed format to exclude local cost factors, C3/MR
being indexed to 100 %.
3.6–12
Table 8.
Indexed Capex comparison
Process
C3/MR
Cascade
DMR
SMR
N2
expansion
Indexed Capex
100
119
116
97
95
The Cascade process is more expensive because of the paralleling of the refrigerant
cycles and the large number of equipment items required. Each liquefaction unit has 6
refrigerant compressors with a total of 20 pressure stages, etc. Variable speed drives are
mandatory for the air cooled process to sustain a minimum degree of operability. At
present, variable speed gas turbines of the heavy industrial or aircraft derivative type, are
limited to some 20-25 MW site rating. Larger new machines like the GE LM6000 and the
RR Trent 800 (35-45 MW range) offer prospects for a simpler line-up, albeit that these
gas turbines must yet be proven for mechanical drive applications.
The costs of the DMR process appear high as compared to the C3/MR and the SMR
process. This can be attributed largely to the high costs assumed for the three spoolwound
heat exchangers in the DMR LNG train.
Plant Availability/Annual Capacity
To arrive at annual production volumes the availability of the different processes
should be taken into account. As a first approximation it is assumed that the feed gas is
always available and that there are no restrictions in LNG storage and shipping. Within
this limited scope, the plant availability was broken down in scheduled unavailability
(maintenance, inspections, gas turbine overhauls, etc.), and unscheduled unavailability
(equipment failure, nuisance trips, operational mistakes, etc.). The scheduled unavailability was averaged over a six years maintenance cycle, dictated by gas turbine major
overhauls. Differences are small here because all processes employ industrial type gas
turbines. For the unscheduled unavailability differences in robustness between spoolwound heat exchangers and plate-fin heat exchangers were addressed as well.
The average annual stream day numbers and annual productions are summarised in
Table 9 below.
Table 9.
Process
Plant Availability and annual Capacity - 2 LNG trains
C3/MR
Plant availability 340
(sd/a)
Annual
7.9
production
(MTPA)
Cascade
DMR
SMR
336
340
338
N2
expansion
335
6.6
8.7
7.4
4.3
N.B: The annual production includes a 2.5 % process margin.
3.6–13
With the information in the previous tables it is now possible to compare the processes
on specific costs, US$ Capex/annual tonne of LNG. To exclude local cost factors the
specific costs have been indexed to C3/MR=100 %. The specific costs are summarised in
Table 10 below.
Table 10. Indexed Specific costs (C3/MR=100)
Process
C3/MR
Indexed Specific 100
Costs
Cascade
DMR
SMR
143
105
103
N2
expansion
175
Table 10 clearly differentiates between attractive and less attractive liquefaction
processes for the particular study conditions, viz. large capacity and air cooling in tropical
conditions
The C3/MR, DMR and SMR process come remarkably close. Lower cost spoolwound
heat exchangers or replacement of the pre-cooling spoolwounds by plate-fin heat
exchangers could make the DMR process more attractive. At more or less equal specific
costs the higher capacity of the DMR trains may result in more favourable overall project
economics.
It is furthermore noted that the C3/MR process has the best track record with trains in
operation up to 3 MTPA. So far the DMR process has not been applied in LNG baseload
operation. The SMR “PRICO” process trains operating in Algeria have a capacity of some
1.2 MTPA.
CLOSING REMARKS
Impact of Air Temperature on Production
It is well known from literature that the thermodynamic efficiency of a liquefaction
process decreases by some 0.7 % per degree increase of cooling medium temperature.
Moreover the gas turbine power decreases by some 1 % per degree C air temperature
increase. Based on these two simple yard sticks it can be shown that the LNG production
of an air cooled LNG plant changes by some 30-35 % over an ambient temperature range
of 20 - 40 °C. The C3/MR process as depicted in this article suffers less from this
temperature effect because the GE-7EA gas turbine driving the propane compressor has
excess capacity over most of the temperature range. Shell is able to design the C3/MR
process such that the production swing can be limited to some 10-15 % over a
temperature range of 20 - 40 °C.
Potential for Modularisation
In order to keep a continuous pressure on plant costs, various alternatives are being
explored to reduce construction costs, especially in remote, high labour rate areas
(Australia, Arctic).
3.6–14
Modularised construction of the LNG plant is one of these ideas. The concept is to
divide the process and utility units in large self-supported modules (say 1000 tonnes per
module) and to fabricate these in an area with relatively high productivity and low labour
rates. The modules are subsequently shipped to the plant site and welded together.
Another idea is to build an entire LNG train on a concrete gravity structure (CGS) and
to transport this to an inshore location near the gas supply. Water cooling would then be
the logical choice as ambient cooling medium. BHP goes as far as proposing an entire
LNG plant on an offshore GCS. Since chilled water is used for pre-cooling the Nitrogen
cycle, the hydrocarbon inventory in the entire process is at a minimum. This concept is
outside the scope of this comparison study.
All processes studied can be modularised in this way, albeit that cost savings are more
modest than generally assumed. The extra costs for pre-ordering, early detailed
engineering, extra structural steel, transport, etc., offset to a large extent the savings in
going to lower cost construction yards.
The two processes that seem most suitable for modularised construction are the SMR
process and the Nitrogen expansion process. The SMR process combines a rather high
LNG production with a minimum of equipment. To a lesser extent (more
expander/compressors, etc.) the same applies for the Nitrogen expansion process. The
other processes have either a lot of equipment needing more modules or less equipment
but with considerable height that cannot be transported vertically (spoolwound heat
exchangers). These equipment items can be pre-fabricated, i.e. dressed up with platforms,
etc., to reduce local construction.
CONCLUSIONS
The Propane/MR process appears to be the best choice within the premises of this
comparison study, viz. large capacity LNG trains, employing air cooling in a tropical
climate. Other promising processes are the Dual Mixed Refrigerant process and the Single
Mixed refrigerant process. Shell is further investigating several variations of these three
processes.
The Cascade process appears to be relatively expensive, partly disadvantaged as it is
by the study premises. Under colder conditions (arctic, water cooling) the capacity comes
closer to the C3/MR capacity.
The pre-cooled Nitrogen Expansion process is not an economic choice for a large,
onshore application. It may be an alternative for smaller scale offshore applications
(absence of hydrocarbon refrigerants).
3.6–15