You are on page 1of 13

COPYING AND DISTRIBUTING ARE PROHIBITED WITHOUT PERMISSION OF THE PUBLISHER

Developments in natural gas liquefaction


04.01.1999 | Finn, A. J., Costain Oil, Gas & Process Ltd., Manchester, United Kingdom; Tomlinson, H. L., Syntroleum
Corp., Tulsa, Oklahoma; Johnson, G. L., Costain Oil, Gas & Process Ltd., Manchester, England

New technologies and design methods will play a large role in the design and construction of LNG projects

Keywords:

Some key ongoing developments can reduce the cost of liquefied natural gas (LNG) projects. Offshore liquefaction will enable cost-
effective use of remote fields, and economical mid-sized single machine LNG trains could improve project viability onshore. Improved
design methods – the increased use of exergy analysis in conceptual design, modular engineering where appropriate and computing
techniques such as dynamic simulation – can be used along with appropriate design margins and engineering standards to reduce
capital costs.

Uncertainty over the viability of large base-load LNG projects has created interest in economical LNG production in mid-sized plants.
Optimizing these plants requires a reappraisal of accepted process technology and traditional design practice. A new mixed-
refrigerant process is suitable for train capacities of up to 1.5 million metric tons per year (MMtpy) using a single machine/ driver
and high-efficiency plate-fin heat exchangers. Power requirement is close to that of a cascade refrigerant cycle, and the relatively
simple machinery configuration compared to a cascade cycle makes it an attractive option for a wide range of plant sizes. Using
multiple trains can give high capacity, with higher reliability than single-train plants.

Currently there is interest in offshore LNG plants for developing remote gas fields. This article shows that proven refrigeration cycles
using expanders are the best choice for offshore developments. Such projects are approaching commercial development, and some of
the key design issues are discussed.

Background. For remote gas fields where pipeline transportation is expensive, natural gas must be either liquefied or converted to
high-value liquid products. LNG has the advantages that it contains about 40% more heating value than liquid fuels derived from
chemical conversion of natural gas and is produced using well-established technology. 1 Although gas-to-liquids conversion
technology is attracting attention, existing plants are small and many processes are still in a developmental stage. Liquid products
from gas conversion are aimed at different markets than LNG. It is unlikely this technology will have a large impact on LNG demand
in the near future.

LNG trade has grown steadily since the mid-1960s. Current annual global demand is about 80-million metric tons. Approximately
two-thirds is imported by Japan. Although the Asian economic downturn has created uncertainty over future demand in Japan and
South Korea, LNG demand is expected to grow in other countries, notably China and India. The emergence of a new market in LNG
for independent power producers (IPPs) has helped to justify new projects in Nigeria, Trinidad, Qatar and Oman. For the IPPs, LNG
can represent a reliable and secure supply of environmentally friendly fuel for combined-cycle gas turbine plants. These plants give
high overall efficiency for relatively low capital cost. 2

LNG projects are inherently capital-intensive, with the liquefier making up around 25% to 50% of the total cost (Fig. 1). The balance
is for storage, send-out terminals, jetties and ships (base load) or vaporizers (peak shave) (Fig. 2). The liquefier is where a process
designer can bring about the largest cost savings, even affecting overall project viability.
Typical breakdown of liquefaction plant capital
Fig. 1
costs.

Fig. 2 Features of the LNG chain.

Since their advent in the 1960s, liquefaction plants have increased in capacity to take advantage of economies of scale.3 The overall
investment in single projects (including the whole production and supply chain) is extremely large; and project development can take
many years. Medium- and small-scale projects are becoming more attractive because they are usually easier to promote and
implement, and can subsequently be expanded in capacity. As a result, technology developments pertinent to smaller-scale projects
are posing challenges to process designers.

Liquefaction plants are generally classed as either peak-shave or base-load plants, depending on their size/ role.

Peak-shave facilities are usually relatively small (typically up to 100,000 tpy) and are used to overcome mismatches between supply
and demand. They liquefy and store excess natural gas during periods of low demand and vaporize it at times of peak demand. Many
peak-shave plants were built in the 1970s and '80s, primarily in Europe and North America.

Base-load plants supply several thousand tpd of LNG, usually for marine transportation. The number of base-load trains operating
or under construction worldwide is now approaching 70, at 15 sites. The maximum achievable liquefaction train size has increased
over the last 30 years, with maximum train capacities now over 3 MMtpy of LNG.
Small- or mid-sized base-load. There are no mid-sized LNG plants in the range of 300,000 tpy to 1.5 MMtpy. Some small plants
(up to 150,000 tpy) exist, in the U.S. for example, delivering locally by road or rail tanker when it is deemed more economical than
pipeline delivery.

Offshore. The potential for offshore liquefaction is in gathering and liquefying gas from several small fields, enabling reasonably
easy movement of the production ship between fields, or liquefying associated gas for transportation to shore. Offshore liquefaction
can give benefits in exploiting remote fields, and give a lower cost LNG project than if based on land.

Liquefaction cycle selection. Natural gas liquefaction requires removal of sensible and latent heat over a wide temperature range
using a refrigerant. The refrigerant may be part of the natural gas feed (open-cycle process) or a separate fluid continuously
recirculated through the liquefier (closed-cycle process).

Three main types of refrigeration cycle are used, namely the cascade, mixed-refrigerant and expander cycle. Each has its own merits,
and more than one of the processes may be economical, depending on capacity of the plant being considered. There are a number of
variants of each cycle, with some common features between them. 4 For example, both the mixed refrigerant cycle and the expander
cycle have variants in which the feed gas is pre-cooled by a conventional propane vapor compression cycle – also a feature of the
cascade cycle.

The proportion of total capital cost due to machinery is always high in liquefaction processes. Reduced power consumption and
smaller machinery generally lead to lower capital cost and operating costs. Therefore, thermodynamic efficiency is greatly
emphasized when designing liquefaction cycles.

Cascade refrigerant cycle. The natural gas is cooled, condensed and sub-cooled in heat exchange with propane, ethylene (or
ethane) and finally methane in three discrete stages (Fig. 3). The three refrigerant circuits generally have multistage refrigerant
expansion and compression, each typically operating at three evaporation temperature levels. After compression, propane is
condensed with cooling water or air, ethylene is condensed with evaporating propane and methane is condensed with evaporating
ethylene.

The cascade cycle requires the least amount of power of all the liquefaction processes, mainly because the flow of refrigerant is lower.
It is also flexible in operation, since each refrigerant circuit can be controlled separately. Composite warming and cooling curves for
the cascade cycle are given in Fig. 4. Mean temperature differences between the curves are wide relative to those of the mixed
refrigerant cycle (Fig. 5). Therefore, the cascade cycle has a comparatively low heat exchanger surface area requirement.

Fig. 3 Cascade cycle (simplified).


Typical composite warming and cooling curves for
Fig. 4
cascade cycle.

Typical composite warming and cooling curves for


Fig. 5
an MRC.

The major disadvantage of the cascade cycle is the relatively high capital cost due to the number of refrigeration compression
circuits, each requiring its own compressor and refrigerant storage. Maintenance and spare equipment costs tend to be comparatively
high due to the large number of machines. The cascade cycle has relatively low power and low surface area, but increased complexity
in machinery configuration.

Economies of scale show that the cascade cycle is most suited to large train sizes, since the low heat exchanger area and low power
requirement offset the cost of having multiple machines. 5 By optimizing machinery selection, the cascade cycle can be competitive
with the precooled mixed-refrigerant cycle, which has been the dominant cycle in base-load mixed-refrigerant plants.

Mixed-refrigerant cycle. The mixed-refrigerant cycle (MRC) uses a single mixed refrigerant instead of the multiple pure
refrigerants in the cascade cycle. The mixture composition is specified so the liquid refrigerant evaporates over a temperature range
similar to that of the natural gas being liquefied. A mixture of nitrogen and hydrocarbons (usually in the C 1 to C 5 range) is normally
used to provide optimal refrigeration characteristics (i.e., close matching of composite warming and cooling curves, as shown in Fig.
5, with small temperature driving forces over the whole temperature range).

Small temperature driving forces give operation nearer to reversibility, leading to greater thermodynamic efficiency, lower power
requirement and smaller machinery. However, a typical MRC has a lower efficiency than the cascade cycle because, although
temperature driving forces are smaller, refrigerant flow is much higher. The MRC does have the advantage of a simpler configuration
and the amount of equipment is minimized since the cycle requires only one compressor and fewer vessels for refrigerant separation.

In a typical MRC, the refrigerant stream, at approximately ambient temperature and low pressure, is compressed and partially
condensed against air or cooling water (Fig. 6). Resulting vapor from the separation is partially condensed in the first stage of the
main heat exchanger. Liquid is subcooled separately in the heat exchanger and then expanded across a valve, reducing its
temperature using the Joule-Thomson effect. This low-pressure stream combines with the equivalent low-pressure stream from the
second (colder) stage and returns counter-currently through the heat exchanger.
Fig. 6 Typical three-stage MRC.

The refrigeration available is sufficient to subcool the incoming refrigerant liquid phase, partially condense the incoming refrigerant
vapor phase and cool/ condense the incoming natural gas. The incoming refrigerant vapor phase stream, now partially condensed,
undergoes a phase separation. The vapor and liquid phases pass to the main heat exchanger's second stage, which operates similarly
to the first. The refrigeration required to subcool liquid refrigerant and condense vapor refrigerant is normally a large proportion of
the total duty.

The optimum number of stages for partial condensation, separation and expansion depends on the relative importance of capital cost,
operating complexity/ flexibility and operating costs. More stages increases energy efficiency, but with greater complexity. For
example, a two-stage process will consume less power than a single-stage process and the reduction in machinery capital cost
normally more than offsets the cost of additional equipment, giving a reduced overall capital cost. As the number of stages is further
increased, an optimum is reached so that further stages have minimal effect on machinery power and cost, and only increase
complexity and overall capital cost.

The results of an evaluation of MRC performance (Table 1) shows the effect of increasing the number of stages against power
consumption. It would be expected that the techno-economic optimum number of stages would be greater for a base-load plant than
a peak-shave plant. In general, a three-stage process gives a balance between energy efficiency and complexity and is close to optimal.

Improved multistage MRC designs have recently been developed 6 that use
compact aluminium plate-fin heat exchangers to give reduced power TABLE 1. Evaluation of MRC performance
consumption. Multistream plate-fin heat exchangers give high thermodynamic
efficiency because small temperature driving forces and a high degree of Power required relative
energy integration are possible, enabling true matching of composite curves. Number of stages to one stage process
Multiple separation stages give a lower refrigerant flow compared with simple
MRC processes and, hence, a lower power requirement (Table 1).
Consequently, power requirements approaching those of a cascade cycle are
attainable, but with a simpler machinery configuration. For a fixed driver, the 1 1
amount of LNG that can be produced can be optimized and, as a result, small-
and mid-sized plants can be built at competitive cost. 2 0.93
Feed gas precooling using a separate propane refrigeration system is usually
economical for larger plants and has been used on virtually all base-load plants 3 0.90
in the last 20 years. It is common to use a 3-stage propane refrigeration
system equivalent to that of the cascade cycle. The propane precooled MRC is
the most common cycle. Dual MRCs have also been proposed, using a separate 4 0.88
mixed refrigerant to precool the natural gas feed, the primary mixed
refrigerant or both. 5 0.87
One operating problem that is often encountered with MRCs is their sensitivity
to changes in feed gas composition. Adjustment of the refrigerant composition is needed to maintain efficient operation and ensure
full capacity. The ability to change refrigerant composition makes the MRC reasonably flexible to variations in feed conditions or
seasonal temperatures, but it can be difficult to evaluate what composition is needed. Modern gas plants use process modeling and
real-time optimization to improve plant performance and stability. Such techniques can also be applied to LNG plants to help
operators predict optimum refrigerant composition for varying feed and ambient conditions and, therefore, run plants more
efficiently.

There is a requirement to make up the losses of each refrigerant component due to seal leakage and to adjust the refrigerant
composition. NGLs are often extracted and fractionated, to give products for sale, so a refrigerant supply is readily available. For
small plants or lean feed gases, NGL recovery may not be economical and refrigerant must be imported and stored. This is one
reason why the expander cycle (described in the next section), which does not require hydrocarbon refrigerant make up, is well
suited to small plants.

Capital costs can be reduced by avoiding excessive design pressures for the refrigerant circuits. In a multicomponent refrigerant
system, the refrigerant's composition and hence the vapor pressure, varies around the circuit as the refrigerant is separated, unlike a
cascade based on pure components. Design pressures for the refrigerant loop are set for the calculated settle-out pressure and often
have a large design margin. Dynamic simulation can accurately calculate design pressure, avoiding unnecessary costs. 7 Dynamic
simulation can also be used to model transient performance of the whole facility to improve control and minimize flare system size.

Various forms of MRC have been used for LNG plant sizes from peak shave to base load. The most important factor in the success of
the cycle is its high efficiency, and consequent low power consumption, without the complexity of the cascade cycle.

Expander cycle. In its simplest form, process refrigeration is provided by compression and work-expansion of a single-component
gas stream. High-pressure cycle gas is cooled in counter-current heat exchange with returning cold cycle gas (Fig. 7). At an
appropriate temperature, the cycle gas is expanded in a near isentropic manner through an expansion turbine, reducing its
temperature to a lower temperature than would be given by expansion through a Joule-Thomson valve. Useful work is generated,
which is normally recovered through a booster compressor brake, which supplements the main-cycle compressor.

Fig. 7 Typical expander cycle (single expander).

The cold, low-pressure gas stream from the expander is returned through various stages of heat exchange, where its refrigeration is
given up to the incoming natural gas and incoming cycle gas. The gas is recompressed by the main-cycle compressor and booster
compressor. The refrigerant cycle gas used can be methane or nitrogen. Using nitrogen allows subcooling to temperatures low enough
to eliminate flashing gas when the LNG is let down.

Expander cycles have several advantages over cascade and MRCs. They enable relatively rapid and simple startups and shutdowns.
This is important when frequent shutdowns are anticipated, such as with peak-shave plants. Because the refrigerant is always
gaseous and the heat exchanger operates with relatively wide temperature differences, it tolerates feed gas composition changes with
minimal requirements for changing the refrigerant circuit. Temperature control is not as crucial as for mixed-refrigerant plants and
the cycle performance is more stable. The problem of distributing vapor and liquid phases into the exchanger is eliminated because
the cycle fluid is maintained in a gaseous phase.

A major disadvantage of the expander cycle is its relatively high power consumption, compared with the cascade and MRCs. The
cycle's simplicity can, however, make up for the high power consumption, especially in small plants where power consumption is less
important.

Changes can be made to the single expander cycle to increase its efficiency. For example, power consumption can be reduced by
approximately 20% using natural gas precooling with a conventional vapor compression cycle, typically using propane. This
introduces extra complexity, but can be cost-effective if the cost of extra equipment can be offset by the reduction in size and cost of
cycle machinery.

A further alternative to propane precooling is to use two expanders, operating over different temperature levels. This has been
conventional practice in the liquefaction of nitrogen and oxygen for more than 20 years. Fig. 8 shows typical composite warming and
cooling curves for a double-expander cycle. Two expanders allow closer matching of the warming and cooling curves than with a
single expander, giving reduced temperature driving forces and higher thermodynamic efficiency. Power consumption similar to a
precooled single-expander cycle can be obtained without using a separate refrigeration system.

Typical composite warming and cooling curves


Fig. 8
for double nitrogen expander cycle.

Methane expansion cycles have been used for peak shaving, usually with a portion of the feed gas being let down into a low-pressure
distribution system. 8 The Isle of Grain plant in the UK takes advantage of pressure letdown to a low-pressure distribution system to
produce LNG with low operating costs. This type of process is well established – similar processes have been used for cryogenic air
separation since the early twentieth century. Such cycles are limited in natural gas liquefaction, since it is unusual for a large volume
low-pressure product gas system to be available.

Expander cycles are usually best suited to smaller plants since they have relatively high power requirements. They have been used
where a short period of annual operation is required and a fast startup and shutdown are important, such as in peak-shave plants.
Offshore, the expander cycle has benefits for any size plant.

Cycle efficiency comparison. The approximate specific power for each cycle relative to a cascade is shown in Table 2. A
comparison of specific power consumption cannot be precise, as parameters such as feed composition and pressure have a different
effect on each cycle. Expander cycle power, for example, is highly dependent on the curvature of the condensing gas's enthalpy-
temperature curve. This is affected by both the operating pressure and C 2 + feed content.

TABLE 2. Comparison of efficiencies for different cycles

Approximate power
consumption relative to
Cycle cascade cycle

Cascade cycle 1.00

Single stage mixed refrigerant cycle 1.25

MRC with propane pre-cooling 1.15

Multi-stage MRC 1.05


Single expander cycle 2.00

Single expander cycle with propane pre-cooling 1.70

Double expander cycle 1.70

Comparing the cascade cycle with the most efficient MRC, the difference in power is small. Taking into account the machinery
complexity of the cascade cycle, it is clear why it has only been used for high capacities.

The actual energy requirement per kg of LNG produced depends on the feed and ambient conditions, and on the compressor
efficiency (which depends on type and size). However, specific powers of around 0.33 kWh/kg of LNG are typical for a cascade cycle.

Exergy analysis in liquefaction cycle design. In energy-intensive liquefaction processes, machinery constitutes a major portion
of total capital cost. Improving thermodynamic efficiency reduces power requirements, machinery size and, hence, capital cost.
Therefore, exergy analysis – the evaluation of process irreversibility and inefficiency – is a fundamental design tool to reduce costs. 9–
11

The minimum work (termed the Carnot work) to produce a given amount of LNG is what would be required if the LNG were
produced by a reversible process and is fixed for a given feed pressure and composition. The Carnot work can be compared with the
actual work required by a given process flowsheet or part of a flowsheet, to identify potential process improvements. The actual work
is always greater than the Carnot work since real processes need finite driving forces that reduce the potential for doing work and
increase process power requirements:

We have used overall exergy balances and typical plant efficiencies (usually around 40%) to identify power consumption for a given
process as a precursor to conceptual design.

The Carnot factor is defined as the ratio of the minimum (reversible) work required, Wmin, to remove a quantity of heat, Q, to a
temperature, T:

Below the temperature of the surroundings, T surroundings the Carnot factor rises steeply as temperature is reduced. Fig. 9 shows that
the minimum work required to produce 1 kJ at 100 K is four times that which would be required to produce 1 kJ at 200 K, based on a
T surroundings , of 300 K. This clearly shows the importance of analyzing work losses and thermodynamic efficiency at low
temperatures to avoid excessive power consumption.

Carnot factor vs. temperature (T surroundings =


Fig. 9
300 K).

Process efficiency can be improved by operating more closely to thermodynamic equilibrium, therefore reducing lost work. During
conceptual design, exergy analysis can be used to fix a nears optimum process flowsheet, and lead to both low operating and capital
costs.

Exergy analysis usually develops a process flowsheet and then assesses the lost work by carrying out an exergy balance around each
unit operation or part of a unit operation. This method is rigorous, but time-consuming, and the analysis results can be difficult to
interpret. As a result, exergy analysis tends to be underused. However, the principles of exergy analysis are valuable in identifying
areas of a flowsheet that lead to energy losses. To enable the principles to be used, design heuristics have been developed that allow a
process flowsheet to be developed and be close to optimal the first time. 9 They help formalize the conceptual design so potential
improvements are not missed.

An example of where exergy-based heuristics can lead to improvements is using hydraulic turbines in place of Joule-Thomson valves
for letdown of LNG or liquid refrigerant. The heuristic is that of using high-pressure streams to generate work. A hydraulic turbine
leads to a reduced overall power consumption. When letting down LNG, not only is shaft power generated, but the quantity of flash
gas is also reduced. Hydraulic turbines have been used on both LNG and mixed refrigerant letdown at Malaysian LNG's plant at
Bintulu. Reportedly, 4% extra LNG production, was achieved with a payback measured in months. 12

In the field of "pinch technology", where formalized design methods have been developed using thermodynamic principles, using
composite enthalpy vs. temperature curves is extremely valuable for process design evaluation and improvement. A development of
this graphical method, for low temperature plants, is to plot Carnot factor vs. composite enthalpy curves. 11 The area between the
warming and cooling streams, as shown in Fig. 10, represents lost work. The process designer can concentrate on manipulating the
process to improve the areas highlighted by the plot as incurring significant losses. This method is rapid and allows the investigation
of more process options and, as with the use of heuristics, enables process configuration to be fixed with minimal effort.

Typical Carnot factor vs. enthalpy curves for a


Fig. 10
cascade refrigeration cycle.

Offshore LNG. Offshore liquefaction of natural gas is an option for cost-effective use of gas from small or remote fields. Many fields
are not economical to develop if the gas must be piped to shore. Offshore liquefaction can also be used to avoid associated gas flaring.
Studies were being made for associated gas liquefaction in the North Sea as early as the 1970s. 13 Offshore liquefaction can give a
more economically attractive LNG project than those on land. This is mainly because the plant does not need to be located near the
gas field(s) – an offshore plant can be quickly and easily relocated. This gives the flexibility to move between fields and produce gas
sooner than would otherwise be possible.

Floating production, storage and off-loading (FPSO) facilities for oil production have become commonplace. A number of concepts
along similar lines for offshore LNG production have been proposed and developed.14–18

The choice of liquefaction cycle has a significant effect on offshore project economics. Plant weight and space requirements have a
much greater influence on overall cost, because they impact the ship's size and cost. A compact plant – with the liquefaction plant,
utilities and storage all in close proximity – is necessary. Cascade cycles and MRCs require separate storage of refrigerant
components. This significantly increases the ship's size and cost. Using expander cycles has been limited to small plants onshore,
since the high power requirement makes them uneconomical for larger plants. The expander cycle is, however, compact and safe,
largely due to the elimination of high-pressure hydrocarbon refrigerant storage. This helps make the expander cycle economical for
larger plants offshore. The reduced ship cost compensates for disadvantages resulting from low energy efficiency. Previous work has
shown the expander cycle to be the best choice for offshore plants of all sizes.

An evaluation nearly 20 years ago concluded that the most suitable process for an offshore LNG plant in the North Sea was a
nitrogen expander cycle with precooling by mechanical refrigeration. 13 During the 1980s, this technology was further developed as
part of an evaluation of suitable technologies for bringing gas to shore from a floating production platform.14 In 1991, a double
expander plant developed for an offshore project came out better than more complicated cycles for virtually all evaluated criteria. A
similar process using a double nitrogen expander cycle has recently been proposed for offshore natural gas liquefaction in the Bayu-
Undan field in the Timor Sea.15

Offshore, fast startup and shutdown are important. Expander plants offer these characteristics and have the advantages of being
simple, reliable and easy to operate.

There are challenges to the process designer other than selecting the process route and optimizing the process design. Designing a
flare system for offshore LNG facilities is a potential obstacle to technical feasibility, and requires careful consideration to limit heat
radiation and noise. Onshore, these objectives can usually be met by adequate spacing between equipment, buildings and flares. With
the space constraints of an offshore facility, the best solution may be to limit the flare rate using high-integrity pressure protection
systems (HIPPS) such as the ones commonly used in onshore and offshore gas processing plants.

One of the biggest challenges in developing floating LNG facilities has been the safe transfer of LNG between two moving ships.
Recent advances in the conceptual design of equipment and off-loading methods have improved the technical viability of offshore
liquefaction.17

Offshore LNG facilities must be designed so as not to be adversely affected by wave motion. Movement must be minimized through
design, and all FPSO equipment must be selected to maintain good operation. Some issues have already been addressed on FPSO
facilities for oil production, and experimental work has been carried out to assess the reduction in equipment performance.18 Cascade
and MRCs require precise distribution of the refrigerant phases for satisfactory heat exchanger operations. A major advantage of the
methane or nitrogen expander cycle is that the refrigerant remains in the gaseous phase. This limits the effect of wave motion, since
refrigerant phase distribution in the heat exchangers is not as important.

Machinery selection. Liquefaction processes, perhaps more than any other, are influenced by machinery selection. Machinery can
constitute around 40% of the liquefaction plant cost.

The most common type of compressor in base-load LNG service is the centrifugal. Developments in centrifugal compressor design,
enabling higher suction volumes, means that they are usually the most economical choice of cycle compressor. Axial compressors are
best suited to higher flowrates at lower overall pressure ratios than base-load MRCs generally require. However, they may have
potential for the first stage of compression, to take advantage of their high efficiency.

Steam turbines (up to 80-MW rating) and gas turbines (up to 78-MW rating) have been used in LNG compressor driver duty. Many
earlier base-load LNG plants used steam turbines drivers as they were already proven for compressor duty, and there are a wide
range of power and speed ranges available. 19 Gas turbines are a less costly alternative, as no steam generation equipment is needed
and the cooling system is smaller. Their use has become widespread of late. Gas turbines have a higher thermal efficiency than steam
turbines, are compact and offer a fast startup. The time taken for manufacture and installation of a gas turbine is also shorter than for
a steam turbine system.

The 26-MW Frame 5, the largest dual shaft industrial gas turbine currently available, 20 is widely used in base-load LNG plants. These
gas turbines have separate shafts for the air compressor and turbine. They are responsive to changes in demand since the air
compressor speed can be controlled independently of the turbine speed. A smaller starter motor can be used because the air
compressor can be started without interference from the turbine.

Recent use of 38-MW Frame 6 and 78-MW Frame 7 turbines, which have little track record in compressor driver duty, shows that
machines previously unproven on LNG plants will be used when the potential rewards are great enough. These single shaft gas
turbines have been used at the MLNG-Dua LNG plant at Bintulu in Malaysia.21 Although there are limitations to their flexibility, they
have a potential for cost savings through economies of scale.

Dual-shaft aero-derivative turbines should be considered since they offer a number of benefits. They generally require less
maintenance downtime than industrial turbines and, being much more compact, are ideal for offshore use. Aero-derivative dual-shaft
gas turbines with a power output and efficiency greater than the Frame 5 are available, but are still unproven in an LNG plant.

For small plants such as peak shavers, a wide range of compressors can be considered. The integration of a machine into the process
is an iterative procedure, with the process eventually being optimized around the particular machine. Driver selection depends on the
availability of a power supply. Electric motors are usually the most economical driver on small plants. Small gas turbines are
relatively expensive, but offer high reliability and can use the gas flashed in the process as fuel. Gas engines are really only suited to
driving reciprocating compressors, and require greater maintenance than gas turbines, but are used in some small plants.

Heat exchanger selection. The liquefaction duty is generally carried out in either aluminium plate-fin or vertical spiral wound
heat exchangers. The propane precooled mixed refrigerant process, which has been the dominant base-load technology in recent
years, uses a combination of plate-fin heat exchangers for precooling and spiral wound exchangers for liquefaction. Processes using
plate-fin heat exchangers for liquefaction do, however, play a role in both small- and large-scale plants. Plate-fin heat exchangers are
being used in the Atlantic LNG base-load project, reportedly offering a lower cost than a spiral wound heat exchanger. 5

The size of an individual plate-fin exchanger core is limited by fabrication and design pressure issues to approximately 1.2-m × 1.3-m
× 8.0-m. Multiple exchanger cores can be grouped in parallel cold boxes as far as is economical to give the required plant capacity.
MRC plants have a relatively large heat exchanger surface area requirement. Thus, there is a limit to the liquefaction capacity (maybe
1.5 MMtpy) for which plate fin heat exchangers are economical due to the cost of multiple cold boxes.

A single spiral-wound exchanger can provide a capacity of up to approximately 4 MMtpy of LNG. The size of these exchangers is
normally limited by transportation issues, typically they are 4.5-m in diameter and over 30-m tall.22

Cost reduction. Because LNG technology is reasonably mature, reducing capital cost is likely to result from a number of
improvements rather than one major technology breakthrough.21

Machinery capital costs increase in a stepwise manner with increasing power, due to the use of standard frame sizes. Reduced power
consumption can substantially reduce compressor cost if a smaller frame size or fewer machines can be used. Reduced power
requirement is achieved by increasing cycle efficiency through reducing process temperature driving forces. A clear balance must be
struck between low-cost standardized machinery and exchanger costs to arrive at an optimum design.

Selecting reliable machinery with the optimum redundancy and sparing philosophy can potentially have a greater impact on costs
than small improvements in cycle efficiency during process design. The availability of equipment databases means it is now relatively
straightforward to perform a reliability, operability and maintainability (ROM) analysis to highlight the areas of low availability in a
proposed design. Analyzing the impact of plant nonavailability against the cost of equipment redundancy can lead to optimum
redundancy and sparing philosophy for given plant availability criteria.

Judicious specification of the design ambient air temperature for gas turbines can lead to significant reductions in capital cost. Gas
turbines have become the driver of choice for most LNG plants, as previously discussed. They exhibit a wide variation in power
output depending on ambient temperature, having a greater shaft power output at low ambient temperature. A gas turbine should be
selected to give low capital cost, while ensuring plant capacity can usually be achieved. An appropriate balance is needed to avoid lack
of capacity or excessive design margin.

Waste heat recovery from gas turbines improves overall thermal efficiency. Gas turbine exhaust streams are typically available
between 400°C and 600°C. The amount of heat available from the exhaust depends on the operating conditions and can be
manipulated to some extent. This heat can be integrated with the process, perhaps by generating steam, or by introducing an
absorption refrigeration system for precooling instead of the more usual propane cycle.

Using industry and even vendor standards for equipment such as plate-fin exchangers can substantially reduce engineering and
capital costs, compared to using client standards that are often based on refinery practice. The Atlantic LNG project used judicious
selection of standards appropriate to a gas plant to drive down capital cost. 5 This exercise requires care and experience to ensure that
safety, reliability and operability are not compromised.

Value engineering procedures can highlight equipment that can be modified, or even deleted, to reduce capital cost. Formalized
techniques used throughout the industry involve splitting the plant into a number of functions that may affect operability, availability
or flexibility. Costs are attributed to the various functions over the plant's life, and cost-saving schemes can be speculated. The design
is revised to include these schemes and the costs re-evaluated.

Understanding equipment design margins is important to optimize plant performance and cost. For example, including an adequate
design margin in plate-fin heat exchangers is necessary to ensure satisfactory performance over all perceived conditions, especially
when temperature driving forces are small or when changes in composition can greatly affect operation. Experience leads to
specification of optimal design margins that ensure satisfactory operation while minimizing cost.

Materials selection philosophy has a significant effect on capital cost, especially if requirements for low-temperature-grade materials
can be reduced. The basis for selecting materials is to account for the worst case coincident temperatures and pressures. The worst-
case conditions can be a matter of conjecture and require experience. Auto-refrigeration during depressurizing can lead to
temperatures that would require low-temperature-grade materials if the system was immediately repressurized. As a policy, using
low-temperature-grade materials can be minimized if it can be ensured that coincident low temperature and high pressure cannot
occur.

Modular construction methods normally result in increased engineering work in the design office and increased materials for module
fabrication compared with onsite fabrication. However, if modular design and fabrication is the standard practice for the designer/
contractor, then any increase in engineering work can be minimized. Within our company, the use of modular techniques is so well
standardized that the increased manhour element is negligible. The additional cost for material, compared to onsite construction,
normally does not exceed 5% of total material cost.

The investment in engineering work with a modular approach normally gives significant savings in both construction labor costs and
indirect costs associated with construction. Savings in construction labor costs come from factory-quality assembly and reduced
rework. Savings in indirect construction costs come from reductions in construction labor force, site infrastructure, time in the field
and rework. Normally, savings of up to 10% can be made on total plant cost due to modular construction techniques and the project
schedule is normally improved, giving faster startup and increased revenue.

Where construction labor costs are low, the arguments in favor of modular construction are less clear. In these cases, look carefully at
labor cost vs. labor efficiency and include costs such as expatriate supervision, project schedule and many other issues in deciding the
correct project/ construction strategy.

Using 3-D modeling systems throughout the design process improves constructability and can potentially shorten construction
schedules. This technology, integrated with automatic generation of material lists, etc., for procurement and intelligent P&IDs can
reduce engineering costs by minimizing rework. The largest cost savings normally come about in the construction phase, where any
necessary rework would be extremely expensive. HP

LITERATURE CITED

1. Liu, Y. N., T. L. Daugherty and J. C. Bronfenbrenner, "LNG Liquefier Cycle Efficiency," Twelfth International
Conference and Exhibition on Liquefied Natural Gas, Perth, Australia, May 1998, p. 3.5.
2. Chiu, C.-H., F. W. Richardson and J. R. Siegel, "Optimisation of the integrated LNG to Independent Power
Producer (IPP) chain," Twelfth International Conference and Exhibition on Liquefied Natural Gas, Perth,
Australia, May 1998, p. 7.2.
3. DiNapoli, R. N., and C. C. Yost, "LNG plant costs: Present and future trends," Twelfth International Conference
and Exhibition on Liquefied Natural Gas, Perth, Australia, May 1998, p. 7.4.
4. Vink, K. J., and R. K. Nagelvoort, "Comparison of baseload liquefaction processes," Twelfth International
Conference and Exhibition on Liquefied Natural Gas, Perth, Australia, May 1998, p. 3.6.
5. Jamieson, D., P. Johnson and P. Redding, "Targeting and achieving lower cost liquefaction plants," Twelfth
International Conference and Exhibition on Liquefied Natural Gas, Perth, Australia, May 1998, p. 7.1.
6. "Refrigeration cycle using a mixed refrigerant," International Patent Application, PCT/6B98/01720.
7. Swinney, J., W. E. Jones and J. A. Wilson, "Dynamic modelling of pool boiling refrigeration systems using
multicomponent refrigerants," GPA European Chapter Meeting, London, November 1997.
8. "Gas Processes 98," Hydrocarbon Processing, April 1998, p.112.
9. Tomlinson, T. R., A. J. Finn and D. I. Limb, "Exergy analysis in process development," The Chemical Engineer,
483 and 484, IChemE, Oct. 11 and 25, 1990, pp. 25–30 and pp. 39–42.
10. Chiu, C.-H., "Evaluate separation for LNG plants," Hydrocarbon Processing, September 1978, pp. 266–272.
11. Linnhoff B., and V. R. Dhole, "Shaftwork targets for low temperature process design," Chem. Eng. Sci., 47, 8,
1992.
12. Verkoelen, J., "Initial experience with LNG/MCR expanders in MLNG-Dua," 17th International LNG/LPG
conference, Gastech '96, Vienna, Vol.2, Session 7, part 1, December 1996.
13. Kennett, A. J., D. I. Limb and B. A. Czarnecki, "Offshore Liquefaction of Associated Gas - A Suitable Process for
the North Sea," Offshore Technology Conference, Houston, May 1981.
14. Terdre, N., "Bringing the gas home without a pipeline," Euroil, November 1990, p. 33.
15. Cottrill, A., "An offshore LNG first for Undan/ Bayu," Offshore Engineer, April 1997, p. 21.
16. Miyake, H., N. Kishimoto and Y. Kakutani, "Small scale LNG FPSO for marginal gas fields," Twelfth
International Conference and Exhibition on Liquefied Natural Gas, Perth, Australia, May 1998, p. 2.2.
17. Naklie, M., "Mobil's floating LNG plant," Twelfth International Conference and Exhibition on Liquefied Natural
Gas, Perth, Australia, May 1998, p. 2.1.
18. Hanawa, K., et al., "An experimental study of float type LNG terminal," Twelfth International Conference and
Exhibition on Liquefied Natural Gas, Perth, Australia, May 1998, p. 5.7.
19. de Kraa, J. A., "Current trends in LNG technology," GPA European Chapter Special One Day Meeting, London,
November 1991.
20. "Performance Specifications 1997–98," Gas Turbine World, Volume 17, December 1997.
21. Coyle, D. A., C. A. Durr and D. K. Hill, "Cost Optimisation, the contractor's approach," Twelfth International
Conference and Exhibition on Liquefied Natural Gas, Perth, Australia, May 1998, p. 7.3.
22. Renaudin, G., J. P. Jouffrey and M. Lazzari, "Spool wound exchangers and/or plate fin exchangers in
liquefaction plant," GPA European Chapter Special One Day Meeting, London, November 1991.

Adrian Finn is technology development manager with Costain Oil, Gas & Process Ltd. He has spent 17 years with
Costain as a process engineer, mainly on gas processing projects including NGL extraction and fractionation,
natural gas liquefaction, nitrogen rejection and acid gas processing. He has authored 16 technical papers and holds
patents in nitrogen rejection, NGL extraction and LNG. He is a Fellow of the Institution of Chemical Engineers and
is a chartered engineer in the UK. He holds a BScTech in chemical engineering and fuel technology from Sheffield
University and an MSc from Leeds University (both UK).

Grant Johnson is a process engineer with Costain Oil, Gas & Process Ltd. Since joining Costain in 1997, he has
worked on a number of conceptual and front-end design projects in natural gas liquefaction and nitrogen rejection.
He has also been involved in the evaluation, development and optimization of in-house gas processing technology,
particularly in natural gas liquefaction and NGL extraction. He holds BA and MEng degrees from the University of
Cambridge, UK.

Terry Tomlinson is business development director for Costain Oil, Gas & Process Ltd., Manchester, England, and
is responsible for all front-end activities at Costain. Since receiving his BSc degree in chemical engineering from
Leeds University, UK, in 1974, he has worked on the process design and development of oil, gas and processing
plants. He is a Fellow of the Institution of Chemical Engineers and is a chartered engineer in the UK. He has been
involved in cryogenic plant process design across a wide variety of applications, including processing natural gas for
LNG, nitrogen rejection, NGL recovery, petrochemical and refinery gas processing and air separation.