Microsoft Word - LNG Large Scale Background Paper Oct2006 PDF

SantosSantosSantosSantos

C:\TEMP\LNG BACKGROUND PAPERS\LNG LARGE SCALE BACKGROUND PAPER OCT2006.DOC PAGE 1

LARGE SCALE LNG

Revision 0

LARGE SCALE LNG

BACKGROUND PAPER

PR E P AR E D BY

T I T L E :

P R IN C IP A L E N G IN E E R ST R A T E G I C TE C H N O L O G I E S

NA M E :

G .N.HU N T E R

RE V I S IO N H IS T O R Y

RE V IS IO N DE T A IL S : DA T E :

0 IS S U E D 25 T H O C T O B E R , 2006



LARGE SCALE LNG

Revision 0

CONTENTS

1. INTRODUCTION 5

2. LNG SUPPLY CHAIN 5

2.1 Gas Pre-treatment 8

2.2 LNG Liquefaction 8

2.2.1 Gas Liquefaction Basics 8

2.2.2 Liquefaction Processes 10

2.2.3 Economy of Scale 10

2.2.4 Operating Performance/Availability 13

2.2.5 Process Selection 13

3. TECHNOLOGY VENDORS 14

3.1 APCI 15

3.1.1 Propane Pre-Cooled Mixed Refrigerant (PMR or C3-MR) Process 15

3.1.2 AP-X Process 16

3.2 Phillips 17

3.3 Shell 21

3.3.1 SMR process. 21

3.3.2 DMR process. 21

3.3.3 PMR process 23

3.4 Linde 25

3.5 Axens 27

4. ASSOCIATED FACILITIES 28

4.1 LNG Storage Tanks 30

4.1.1 Total Storage Capacity 30



LARGE SCALE LNG

Revision 0

4.1.2 Number of Tanks 30

4.1.3 Type of Containment 31

4.1.4 Pump Column for In-tank Pumps 35

4.1.5 Tank Pressure Control 36

4.1.6 Purging and Cooldown 36

4.1.7 Insulation 37

4.2 Jetty and Marine Facilities 37

4.2.1 Ship Size 37

4.2.2 Berth Occupancy 38

4.2.3 LNG Tanker Berth and Loading Dock 38

4.2.4 Safety in Port and Jetty Design 40

4.3 Shore-To-Ship Interface And Transfer Piping 41

4.3.1 LNG Loading Arms 41

4.3.2 Loading Line 43

4.4 Vapour Handling (Boil-Off Gas) 45

5. LIQUEFACTION EQUIPMENT SELECTION. 46

5.1 Main Cryogenic Heat Exchangers (MCHE) 47 5.1.1 Spiral Wound (Coil Wound) Heat Exchangers 47

5.1.2 Plate Fin (Brazed Aluminium) Heat Exchangers 49

5.1.3 Core-in-kettle 51

5.1.4 Cold Boxes 51

5.2 Compressors and Drivers 54

5.2.1 Combustion Gas Turbines 54

5.2.2 Electric Motors 55

6. ATTACHMENTS 55

6.1 LNG Trains Operating or being executed 55



LARGE SCALE LNG

Revision 0

6.2 LNG Train Current Maximum Capacity 55

6.3 LNG EPC Contractors Experience 55



LARGE SCALE LNG

Revision 0

1. INTRODUCTION This Background Paper presents an overview of Large Scale Baseload LNG export plants, provides an introduction to the processes available for the liquefaction section of a Baseload LNG plant, and discusses some of the technology selection issues that affect LNG plant configuration.

Since the first LNG liquefaction train came into operation in 1964, some 110 trains have been brought on line (or are currently in design or construction phases). Tables listing trains, sorted by process licensor and EPC contractor, are included as Attachments 6.1 and 6.3.

Starting in the 1960s, with single train capacity less than 1Mtpa (million tonne per annum), train sizes have increased eight-fold, with 7.8Mtpa trains under construction in Qatar. Refer to Attachment 6.2 for a table of current maximum train capacities for each technology.

For the purposes of this paper I have used the term Mini for plants less than 300,000tpa, Mid-scale for plants between 0.3 and 2Mtpa and large for plants over 2.0Mtpa. In this document I have concentrated on providing background regarding Large Scale LNG plants. Refer to the separate October 2nd, 2006 Background Paper which covered Mini & Mid-scale LNG.

No attempt has been made herein to cover floating LNG.

2. LNG Supply Chain The typical LNG supply chain is comprised of facilities similar to those below, of which only those within the Base Load Liquefaction Plant box will be discussed in this memorandum.



LARGE SCALE LNG

Revision 0

The production of LNG from natural gas involves two distinct process steps, being:

Gas Pre-treatment - the removal of impurities and contaminants such as acid gas, mercaptans, water and mercury; and

LNG Liquefaction - Pre-cooling and the removal of heavy hydrocarbons to prevent hydrocarbon

freeze-up and plugging of the cryogenic equipment and to ensure that the calorific value of the LNG meets specification; and liquefaction and sub-cooling.

The typical process scheme for an LNG plant (using the Phillips Process):



LARGE SCALE LNG

Revision 0

The Darwin LNG plant (which uses the Phillips LNG Process):



LARGE SCALE LNG

Revision 0

2.1 Gas Pre-treatment

In a typical LNG plant the feed gas will be delivered at high pressure (e.g. up to 90 bar, 1300psi) from upstream gas fields. The gas is metered and its pressure controlled to the design pressure of the plant. The gas is first pre-treated to remove any impurities that interfere with processing or are not desired in the final products.

Pre-treatment upstream of a liquefaction unit traditionally consists of an acid gas removal step, in which CO2 and sulphur compounds (H2S, COS and mercaptans) are removed, a dehydration step and a mercury removal step. Treating unit requirements are determined by the liquefaction unit requirements (water, CO2), specifications of the LNG product (H2S, COS, organic sulphur compounds), material protection (mercury) and environmental restrictions (SO2 and hydrocarbon emissions). In addition waste streams have also to fulfil minimum specifications.

Where there are high levels of H2S and limitations on the SO2 emissions, the removed sulphur components are recovered as elemental sulphur. Environmental limitations to hydrocarbon emissions can require incineration of CO2 acid gas even in the absence of sulphur compounds. The mercury removal step can be positioned upstream of the acid gas removal or downstream of the dehydration step.

Most of the operational base load LNG plants process feed gases with only low concentrations of CO2, mercury and water. This type of gas requires the minimum of treating, often comprising a CO2 removal unit, molecular sieves for drying and a carbon bed for mercury removal. The relative capital investment for acid gas removal in a LNG plant increases significantly with increasing CO2 content. At 2 mol% CO2 the acid gas unit represents 6% of the processing equipment cost but at 14 mol% CO2 it represents 15% of the processing equipment cost. New developments such as membrane technologies are starting to be considered as an option for bulk removal of CO2 but solvent absorption remains the most cost effective treatment process for meeting LNG specifications.

The LNG product specification (e.g. heating value/Wobbe number etc) for the end market for the LNG will also determine the pre-treatment (and liquefaction) processing requirements. Most LNG contracts specify a range of acceptable heating values for the LNG sold into a particular market. In most cases, this requires that a certain fraction of the heavier hydrocarbon components found in natural gas be removed prior to liquefaction, so that the LNG does not exceed the upper heating value limit. Some natural gases also require removal of the heavy ends to prevent operating problems in the liquefaction cycle, such as freezing of aromatic hydrocarbons at low temperatures.

The remaining gas is made up mainly of methane and typically contains less than 0.1 mol% of pentane and heavier hydrocarbons.

2.2 LNG Liquefaction

Studies of the different liquefaction processes by independent consultants suggest there is not one of them, on its own, that is substantially more efficient than the others in all situations. Rather, each technology can be competitive within a certain range of train sizes. The ultimate choice of which process to select will remain dependent on project-specific variables and the potential development state of novel processes.

2.2.1 Gas Liquefaction Basics

The liquefaction section is the key LNG plant element. Liquefaction processes mainly use mechanical refrigeration, in which heat is transferred from the natural gas, through exchanger surfaces, to a separate closed loop refrigerant fluid. The refrigerant loop uses the cooling effect of fluid expansion, requiring work input via a compressor.



LARGE SCALE LNG

Revision 0

LNG plants consist of parallel units, called trains, which treat and liquefy natural gas and send the LNG to storage tanks. Liquefaction train capacity is primarily determined by the liquefaction process, refrigerant used, and largest available size of the compressor/ driver combination that drives the cycle and the heat exchangers that cool the natural gas.

Basic principles for cooling and liquefying gas using refrigerants involve matching, as closely as possible, the cooling/ heating curves of the process gas and the refrigerant. This results in a more efficient thermodynamic process requiring less power per unit of LNG produced, and it applies to all liquefaction processes. Typical natural gas/refrigerant cooling curves are shown in this figure:

Observing the cooling curve of a typical gas liquefaction process, three zones can be noted in the process of the gas being liquefied. These include a pre-cooling zone, followed by a liquefaction zone, and completed by a sub-cooling zone. All of these zones are characterized by having different curve slopes, or specific heats, along the process. All of the LNG processes are designed to closely approach the cooling curve of the gas being liquefied. This is done by using specially mixed multi-component refrigerants that will match the cooling curve at the different zones/ stages of the liquefaction process to achieve high refrigeration efficiency and reduce energy consumption.

The natural gas, being a mixture of compounds, liquefies over a wide temperature range. Matching of heat curves by minimising the temperature difference between the cooling process gas and refrigerant streams can be achieved by using more than one refrigerant to cover the temperature range and using the refrigerant at different pressure levels to further split the temperature ranges to closely matching ones. The process gas side is normally operated at high pressure (e.g. 40 50 bara) to reduce equipment size and provide more efficient refrigeration.

The liquefaction cooling curve performance is a benchmark that is reviewed in LNG technology comparisons and is often misunderstood or incorrectly applied when considering energy performance relative to lifecycle cost. Caution should be used with this type of comparison. Detailed knowledge of each liquefaction process design, the options they can achieve at different performance levels along this curve, and these options' cost impact, is required for a valid comparison.

Because LNG liquefaction requires a significant amount of refrigeration energy, the refrigeration system represents a large portion of an LNG facility. A number of liquefaction processes have been developed, with the differences mainly confined to the type of refrigeration cycles employed. The most commonly utilized LNG technologies are described below. These processes are used in current plants or are applied in projects in progress. However, there are other processes developed for baseload LNG applications, which are being considered for future projects (and of these only the Axens process is discussed here).

The liquefaction section typically accounts for 30% to 40% of the capital cost of the overall liquefaction plant, which in turns accounts for 25% to 35% of total LNG export plant costs. Key equipment items include compressors used to circulate the refrigerants, compressor drivers, and heat exchangers used to cool and liquefy the gas and exchange heat between refrigerants. For recent baseload LNG plants, this equipment is among the largest of its type and at the leading edge of technology.



LARGE SCALE LNG

Revision 0

The composition of the refrigerant gives an added control parameter as it can be made either from pure or mixed components. With a mixed refrigerant the composition can be adjusted to suit the process conditions. The aluminium heat exchangers used [e.g. the spiral wound heat exchangers (SWHE) or the plate fin heat exchangers (PFHE)] have very large surface areas and a large number of passes, to enable close temperature approaches.

2.2.2 Liquefaction Processes

Variations of three liquefaction processes have used for large scale base-load plants. These are:

The classical cascade in which a three-stage pre-cooling cycle is followed by a three-stage ethylene liquefaction cycle and a three-stage methane sub-cooling cycle;

The single flow mixed refrigerant process in which a mixture of nitrogen, methane, ethane, propane and normal butane is used as the refrigerant.

The propane pre-cooled mixed refrigerant (PMR) process in which pre-cooling is undertaken in a three-stage propane cycle compressor and pre-cooling heat exchangers. Liquefaction and sub-cooling are undertaken using a two-stage mixed refrigerant compressor, separator, liquefier and sub-cooler.

As a general statement, LNG technology licensors have focused on three aspects of LNG production; these are:

The compression required in the refrigeration cycles;

The power to drive the refrigeration cycles with the exception of the Kenai plant, all of the earliest plants used steam turbine driven compressors. Now, combustion gas turbines are the norm. Certain licensors are looking to utilize the largest frame turbines, whilst others are considering aero-derivative GTs; and

The Main Cryogenic Heat Exchanger (MCHE) that is used to chill the incoming gas. Until recently, APCI with its spiral-wound heat exchanger (SWHE) dominated the design and manufacture of this component. Linde also manufactures such a unit. Brazed aluminium plate-fin heat exchangers (PFHE) are now challenging the dominance of the SWHE. These high efficiency units used for MCHE service in both the Phillips and Axens processes, and are also used for lesser services in the APCI AP-X and the Statoil/Linde MFCP processes.

2.2.3 Economy of Scale

The trend towards larger train size seems inevitable as the still relatively young LNG industry seeks to lower unit capital costs through economies of scale. Moreover, there is a growing consensus that the hyper trains will prove technically feasible by introducing new line-ups and/or paralleling known equipment. But as train size increases, an array of factors including overall capital requirements, added complexity during construction, specific venture conditions such as resource availability and market considerations make it unlikely that hyper trains will be a "one size fits all" solution for the industry.

To meet the demand for larger LNG trains, the providers of LNG technologies (including APCI, Phillips, Linde, Shell and Axens) have been engineering ever larger throughput trains, with nominal capacities of around 5million tonne per annum (Mtpa) now in operation and up to 7.8Mtpa being constructed (refer to Attachment 6.2).



LARGE SCALE LNG

Revision 0

The technical limitations involved in scaling up train size vary between the different liquefaction processes. The size of the combustion gas turbines used to drive the compressors is a limitation in most of the processes and in the case of the APCI PMR process, the size of the spiral-wound exchangers is also a limitation. Other challenges include the size of the large diameter cryogenic piping headers and control valves. Economies of scale as a result of bigger individual trains have significantly reduced liquefaction plant capital costs and enhanced the competitiveness of LNG in international energy markets. In the 1960s 1Mtpa was regarded as a large train and train capacity increased slowly throughout the 1970s and 1980s such that a decade ago, the largest LNG production capacity per train was around 2.5Mtpa. By the late 1990s, this had risen to 3Mtpa and the most recent trains have been constructed with capacities of 4 to 5Mtpa. Now the industry is setting its sights on super sized trains that are capable of producing 7.8Mtpa. The following chart compares the growth in train size over time (cerise curve) with the associated increase in LNG Carrier (LNGC) capacity (blue diamonds and green curve):

One limitation to train size has been the feasibility of manufacturing larger heat exchangers and transporting them from the manufacturing site to the field. For spiral wound heat exchangers (SWHE) employed by APCI, winding the tubing and manufacturing and transporting the heat exchanger has been strongly related to the shell size, which is limited to 18-20 feet in diameter. This manufacture and transport limit for SWHEs capped the APCI C3-MR and Shell DMR processes to around 5Mtpa. The technology developers have overcome this (in APCIs case, refer to their AP-X process, and in Shells case refer to their PMR process) so that the same size of spiral wound heat exchanger being manufactured today can be used to produce much higher output. As a result, heat exchangers compatible with a 7.8Mtpa train are within the industry's capability and the ultimate true single LNG Main Cryogenic Heat Exchanger (MCHE) could perhaps support 8-10Mtpa trains from a technical feasibility standpoint. As train sizes have increased, operators have employed ever-larger gas turbine drivers to power refrigerant compressors. Earlier gas turbine driven LNG plants used smaller, dual shaft gas turbines as compressor drivers. The 28 MW ISO rating of the GE Frame 5 gas turbine limited maximum possible train capacity to about 2.7Mtpa without resorting to multiple compressor-drivers in parallel (such as used in the Phillips process). LNG train production of 3.3Mtpa or more for the C3-MCR liquefaction process was made possible with the use of GE Frame 6 gas turbine (ISO rating of 38.5 MW) as driver for the propane refrigeration cycle compressor and a GE



LARGE SCALE LNG

Revision 0

Frame 7 gas turbine (ISO rating of 80 MW) for the driver for the mixed refrigerant compressor. The use of the GE Frame 7 gas turbine drivers for both C3 and mixed refrigerant cycles compressors enabled individual train capacity up to 4 to 5Mtpa. Now APCI is employing GE Frame 9 gas turbine drivers in their AP-X process. Such increases come at a price, including more expensive drivers, more difficult start up and greatly reduced flexibility of operation. Further optimization and technical innovation is required for gas turbine drivers and compressors to meet these high planned production rates and to cope with the increasing requirements for availability and operability.

Another potential bottleneck in the ultimate realizable train size is the ability to extract impurities from the feed gas. If the feed gas contains a concentration of CO2 and/or H2S of 20-50% or greater, treating plant could easily limit train size. Removal of these two most common impurities in natural gas is a costly and complex step in LNG manufacture. Such treating often requires high pressure, thick walled, large diameter vessels. While these vessels are limited in size to around 18 to 20 feet, this bottleneck can be removed if parallel design is employed. While technical feasibility appears within the industry's grasp, LNG export ventures must consider an array of other factors when considering whether a super sized train fits their needs, including:

The overall capital cost of the project, rather than unit costs;

Whether the plant is to be onshore, on a gravity-based structure or floating;

Adequacy of gas resource base;

Complexity of scheduling, and additional manpower and other resources during construction phase;

The robustness of process and design (requires 20-40 years of operation);

Safety over the life of the project;

Compatibility of operation and maintenance with existing trains for an expansion project, as well as similarity of equipment for spare parts;

General ease of maintenance and accessibility of major components, including supplemental refrigerant;

Ease of start-up (and restart);

Minimum emissions; and

Experience of the technology partner with the process.

Therefore, choosing a liquefaction process is not a matter of simply comparing one technology against another. Rather, it is part of a whole range of factors that must be considered in order to achieve maximum value in project implementation. Furthermore, as engineering and construction firms come up with innovative combinations of equipment, the concept of an LNG "train" could blur. Finding sales outlets for the large incremental LNG production from a hyper train remains a significant commercial challenge. There are only a few buyers willing to make large purchase commitments from single trains. In general, customers are seeking small tranches of LNG to match the requirements of newly liberalizing and competitive markets. Moreover, many buyers are insisting on long ramp-up periods in their purchase contracts. Delays in production build-up could quickly erode any specific cost benefits from the new hyper trains. At the same time, potential economies of scale do little to help the producer maintain a steady cash flow stream, particularly if it is a relatively new seller with most of his "eggs" in a single large train "basket."



LARGE SCALE LNG

Revision 0

2.2.4 Operating Performance/Availability

The ability of a facility to achieve the required annual production and export volumes is dependent on the overall system availability which is in turn a function of equipment reliability, sparing philosophy, planned shutdowns and unscheduled shutdowns.

Normal strategy for a baseload facility is that a single component failure or malfunction must not cause a shutdown of the plant through a cascade of events and may only affect the operation of a specific facility or unit. Some projects adopt a policy whereby in general, sparing of equipment is minimized and only selected equipment has installed spares. In establishing the economics for a project, it is important to establish the overall system availability as soon as practicable so that technical limitations and commercial expectations are properly aligned. Optimized sparing is determined by reliability modelling and cost-benefit analysis.

The design availability for LNG plants ranges from 340 days per year to 350 days per year corresponding to 93 to 95+ percent availability. Downtime for schedule maintenance of refrigeration gas turbines is generally set at an average of 7 days per year. In practice the compressors become the critical path for maintenance and downtime. Best in class performers achieve around 10 days per year for unplanned downtime.

2.2.5 Process Selection

Each process has its merits and, depending on plant capacity, more than one process may be economical. The

choice of optimal process can vary based on site location, feed gas price and ambient conditions, and

evaluation of a number of processes may be necessary to determine the best economically over a

developments full life cycle.

Choosing the optimum process is crucial to reducing plant capital cost as reduction in liquefier costs also reduces utilities and offsites costs. The choice of liquefaction cycle depends on many factors of which the major ones are:

Machinery configuration and available drivers;

Specific power requirement (affecting machinery capital cost and operating cost);

NGL recovery or nitrogen rejection requirement;

Heat exchanger type and surface area;

Required flexibility; and

Ease of operation/start-up/shutdown. All of these issues should be considered in process technology selection. Contacts will be made with the LNG liquefaction licensors, LNG experienced EPC contractors and main equipment vendors to obtain data and develop designs to enable valid comparisons and optimum selections to be made. Technology selection of process and equipment will be based on technical and economic considerations:

Depending on the stage of project development, sufficient process details must be developed to define main equipment and operating parameters to evaluate options using relevant criteria;

Technical considerations include process and equipment experience, reliability, process efficiency, site

conditions and environmental impact among others;



LARGE SCALE LNG

Revision 0

Economic issues include capital cost, operating cost and life cycle costing. All of these aspects will need to be evaluated to arrive at the optimum solution;

Technical risks associated with a process relate to the track record of the process in operation and any

developments required for the project e.g. capacity increase;

Process efficiency, for example energy required to produce LNG, is not solely related to the thermodynamic efficiency of the liquefaction process but also to the efficiency of the main equipment, such as the main refrigerant compressors and drivers;

Site conditions may favour one type of process over other. For example, with very cold ambient

temperatures multi-mixed refrigerant processes may offer the optimum solution;

Process requirements and configuration will have an influence on selection. A requirement for greater LPG recovery may suit processes with lower pre-cooling temperatures;

Wider feed gas range will require better process adaptability and may favour mixed refrigerant

processes with the added flexibility of changing refrigerant composition; and

Refrigerants made up from components that can be produced in the process (in the fractionation unit) will obviate the need for external supply to make up refrigerant losses.

3. Technology Vendors

Some 80% of total liquefaction capacity uses the Air Products and Chemicals, Inc. (APCI) propane/mixed-refrigerant system. APCIs strategy was so effective that they were the only successful liquefaction process supplier for about 25 years. Their pre-cooled propane mixed refrigerant (C3-MR) system became the standard in which project investors could rest assured it would operate as advertised.

Early patents for APCI's precooled propane mixed refrigerant (C3/MR) process have expired. So, not only are new competitors vying for market share, but also former customers (such as Shell) are devising systems in-house.

APCI dominance has recently been successfully challenged by Phillips, Shell, Linde-Statoil and IFP-Axens (although the latter has not yet found a commercial application).

The following technologies are potential candidates for a Large Scale LNG Project (listed in order of experience refer to Attachment 6.1):

Air Products and Chemicals, Inc. (APCI) Propane Pre-Cooled Mixed Refrigerant (PMR or C3-MR) Process and the AP-X Process Technologies;

ConocoPhillips Optimized Cascade (OCP) Process Technology;

Shell Dual Mixed Refrigerant (DMR) Technology;

Statoil-Linde Mixed Fluid Cascade (MFC) Process Technology;

IFP-Axens Liquefin Technology.

Selection of an EPC Contractor will be run in parallel with LNG liquefaction technology selection. Attachment 6.3 provides background regarding contractors experience with designing and constructing LNG trains since the first in 1964.



LARGE SCALE LNG

Revision 0

3.1 APCI

APCI is based in Allentown, Pennsylvania.

APCI markets both their Propane Pre-Cooled Mixed Refrigerant (PMR or C3-MR) and the AP-X processes:

3.1.1 Propane Pre-Cooled Mixed Refrigerant (PMR or C3-MR) Process

The Propane Pre-cooled Mixed Refrigerant (C3-MR) process, developed by APCI, began to dominate the industry from the late 1970s on. This process accounts for a very significant proportion of the worlds baseload LNG production capacity and train capacities of up to 5.0Mtpa have been built.

There are two main refrigerant cycles. The pre-cooling cycle uses a pure component, propane. The liquefaction and sub-cooling cycle uses a mixed refrigerant (MR) made up of nitrogen, methane, ethane and propane. The pre-cooling cycle uses propane at three or four pressure levels and can cool the process gas down to 40 C. It is also used to cool and partially liquefy the MR. The pre-cooling is achieved in kettle-type exchangers with propane refrigerant boiling and evaporating in a pool on the shell side, and with the process streams flowing in immersed tube passes. A centrifugal compressor with side streams recovers the evaporated C3 streams and compresses the vapour to 15 25 bara to be condensed against water or air and recycled to the propane kettles. In the MR cycle the partially liquefied refrigerant is separated into vapour and liquid streams, which are used to liquefy and sub-cool the process stream by cooling from typically -35C to between -150C to -160C. This is carried out in a proprietary spiral wound heat exchanger (SWHE), the main cryogenic heat exchanger (MCHE). The MCHE consists of two or three tube bundles arranged in a vertical shell, with the process gas and refrigerants entering the tubes at the bottom and flowing upward under pressure.. The process gas passes through all the bundles to emerge liquefied at the top. The liquid MR stream is extracted after the warm or middle bundle and is flashed across a Joule Thomson valve or hydraulic expander



LARGE SCALE LNG

Revision 0

on to the shell side. It flows downwards and evaporates to provide the bulk of cooling for the lower bundles. The vapour MR stream passes to the top (cold bundle) and is liquefied and sub-cooled, and is flashed across a JT valve or expander into the shell side over the top of the cold bundle. It flows downwards to provide the cooling duty for the top bundle and, after mixing with liquid MR, part of the duty for the lower bundles. The overall vaporised MR stream from the bottom of the MCHE is recovered and compressed by the MR compressor to 45 48 bara. It is cooled and partially liquefied first by water or air and then by the propane refrigerant, and recycled to the MCHE. In earlier plants all stages of the MR compression were normally been centrifugal, however, in some recent plants axial compressors have been used for the LP stage and centrifugal for the HP stage. The SEGAS Damietta LNG plant in Egypt (at 5Mtpa, the largest APCI C3-MR plant):

3.1.2 AP-X Process

A limitation of the C3-MR process, which caps the capacity of this process to around 5Mtpa for a single train, is the physical size of the SWHE required, and the ability to manufacture and transport the largest examples of these. While the SWHE could be paralleled to increase single train capacity (refer to the Shell PMR process below), APCI introduced a new technology called AP-X Liquefaction Process Technology, which allows LNG production trains to produce approximately 8Mtpa, over a 50 percent increase from today's 5Mtpa standard. This new design marries APCIs standard LNG technology with the company's air separation technology.

A modification of the APCI process, the APX process is a hybrid C3-MR cycle adding a third refrigerant cycle (nitrogen expander) to conduct LNG sub-cooling duties outside the MCHE. The new process employs three refrigeration stages, propane (C3), mixed refrigerant (MR) followed by a nitrogen (N2) expander cycle, and there are several options of a suite of refrigerants. It is expected to be similar to C3-MCR in that it is highly efficient. The addition of the nitrogen expander cycle to sub-cool LNG can reduce the refrigerant flow requirements of C3 and MR per unit of LNG production. Thus, by using proven compression equipment without duplicating or paralleling requirement, the individual train size can be greatly increased.

With the new AP-X Hybrid LNG Process, train capacities up to eight million metric tons per year are feasible in tropical climates, in existing compressor frame sizes without duplicate-parallel compression equipment, and



LARGE SCALE LNG

Revision 0

using a spool-wound main cryogenic heat exchanger (MCHE). The AP-X process is being utilised for the 6 trains, each of 7.8Mtpa capacity, that are currently under development for RasGas and Qatargas in Qatar.

The process cycle is an improvement over the propane-precooled, mixed refrigerant (C3-MR) process in that the LNG is sub-cooled using a simple, efficient nitrogen expander loop instead of mixed refrigerant. In addition to improving the efficiency, the use of the nitrogen expander loop makes greatly increased capacity feasible. It does this by reducing the flow of both propane and mixed refrigerant. The nitrogen-expander loop is a simplified version of the cycle employed in hundreds of air separation plans and nitrogen liquefiers worldwide.

Volumetric flow of mixed refrigerant at the low-pressure, compressor suction is about 60 percent of that required by the C3-MR process while mass flow of propane is about 80 percent of that needed by the latter process. Propane is used to provide cooling to a temperature of about -30C. The feed is then cooled and liquefied by mixed refrigerant, exiting the MCHE at about -120C. Final sub-cooling of the LNG is done using cold, gaseous nitrogen from the nitrogen expander.

The AP-X train can be operated at a reduced production rate of about 65 percent (5.2Mtpa) without the nitrogen expander loop by adjusting the composition of the mixed-refrigerant inventory. The producer can expand capacity later by adding the nitrogen-expander cycles.

The power split between the propane, mixed-refrigerant and nitrogen loops is flexible and can be manipulated by changing the temperature range of the three refrigerant loops.

3.2 Phillips

ConocoPhillips is based in Houston, Texas.

Phillips Petroleum Company developed the original Optimized Cascade Process (OCP) in the 1960s. The objective was to develop a liquefaction technology that permitted easy start-up and smooth operation for a wide range of feed gas conditions. This process was first used in 1969 at their Kenai, Alaska LNG facility. That facility was constructed by Bechtel and was the first plant to ship LNG to Japan, and it has achieved more than 37 years of uninterrupted supply to its Japanese customers.

The first new use of this process was for Train 1 of the Atlantic LNG plant in Trinidad which came on line in 1999. The process has emerged as a serious competitor to APCI's dominant liquefaction technology and other



LARGE SCALE LNG

Revision 0

LNG trains using this process have since been installed in Trinidad (4), Darwin (1), Idku, Egypt (2) and Equatorial Guinea (1). Train capacities of up to 5.2Mtpa are now operating. Bechtel, which is the engineering and construction firm that has exclusive rights to the Phillips process, is working with ConocoPhillips to design even larger trains, with an 8Mtpa train having been evaluated. Refrigeration and liquefaction of the process gas is achieved in a cascade process using three pure component refrigerants; propane, ethylene and methane, each at two or three pressure levels. This is carried out in a series of brazed aluminium plate fin heat exchangers (PFHE) arranged in cold boxes. Pre-cooling can be carried out in a corein-kettle type exchanger. The refrigerants are circulated using centrifugal compressors.

This process uses two pure refrigerants - propane and ethylene circuits, and a methane flash circuit - cascaded to provide maximum LNG production by utilizing the horsepower available from 6 gas turbines. Each refrigerant circuit has parallel compression trains using two 50% compressors with common process equipment. Frame 5 gas turbine drivers have been used for most plants, although Darwin LNG uses GE LM25000+ aero-derivative gas turbines. Brazed aluminium heat exchangers and core-in-kettle exchangers are used for the feed gas, propane, ethylene and methane circuits. All of these heat exchangers, with the exception of the propane chillers, are housed in two "cold boxes." All compressor inter-cooling, after-cooling and propane refrigerant condensing is provided by fin-fan heat exchangers.

The Phillips' "two-in-one" design provides an added advantage as train size increases, especially when production availability is important. This design incorporates two drivers per refrigeration service. As a consequence, if one driver goes down, the entire train's production capacity is not lost. According to Bechtel, production capacity using the Cascade process has an inherent availability advantage of about 4 to 5% over a typical single driver-compressor train arrangement. Despite negative perceptions that the 2-in-1 compressor concept means more drivers and compressors to maintain, those OCP plants operating achieve greater than 95% availability. For example, Train 1 at Atlantic LNG has operated with an availability of over 96% and the original Kenai plant has achieved some 97% availability over its 37 year life. The stated benefits of the 2-in-1 OCP concept include:

The ability to operate with one compressor down and still produce at approx. 60% throughput. The ability to turndown throughput even lower with only 3 of the 6 compressors operating (I recall

from published papers that this can be as low as some 40%).



LARGE SCALE LNG

Revision 0

High availability as maintenance can be undertaken on individual compressors (shut down) while the

remainder of the plant continues to produce (albeit at reduced capacity as noted above). In fact Atlantic LNG debottlenecked their Train 1 by changing out all of their compressor drivers (from GE Frame 5Cs to 5Ds) one by one while still producing from the plant - no shutdowns involved! Because of this, availability of these OCP plants is high. Bechtel believe that because APCI plants do not have this flexibility, that the best that an APCI plant can achieve is circa 93%.

They can take advantage of the smaller proven gas turbines (such as GE Frame 5 and LM2500+ ) that

are available, rather than having to take the risk with larger KW drivers (note that Darwin LNG is the first use of the more efficient aeroderivative gas turbines in an LNG liquefaction plant).

Unlike the APCI plants which have little turn down capability, the OCPs 2 x 50% compressor design means that the following operating ranges that can be utilised (in addition to running in full recycle mode):

Operating Range % of design capacity

Full plant 80 to 105 One compressor offline 60 to 80 Three compressors offline 30 to 60

Another feature of the 2-in-1 OCP process is that should a project be initially constrained to gas feedstock volumes or reserves that can only support an initial 1.5Mtpa plant, then there is a way that a Phillips OCP process can manage this as a phased 1.5+1.5 plant. In discussion with Bechtel they recounted work that they did on a study some 8 years ago for a client who wanted to start at 1.5Mtpa and grow to 3.0Mpta. This project did not proceed, however the idea is one worthy of consideration. Due to the 'standard' 2-in-1 design's 2 x 50% compression trains, the compression can be installed in 2 stages with 3 compressors initially (for the 1.5Mtpa capacity) and with the other 3 compressors (and drivers) being installed when the plant has gas reserves and/or market to support a 3.0Mtpa plant. For this, the gas pre-treatment and utilities would be designed and installed in 2 x trains (and if their were two distinct gas feedstocks, such as lean CSM gas and richer/hotter pipeline sales gas, there is the possibility of dedicating each of the pre-treatment trains to a different gas feedstock). Of course this 1.5+1.5 facilities arrangement and phased execution costs more than going ahead with a Greenfield 3.0Mtpa plant in a single go. Even if this resulted in the same total costs of two separate 1.5 trains, from efficiency/operability/maintainability points of view the integrated 1.5+1.5 would be preferable.

The OCP can provide a facility with high thermal efficiency. The process utilizes proven technology and equipment, and has a wide range of operational flexibility. Turndown rates to 10% are achievable for long-term operation. Due to the pure component systems, the plant has easy start-up and operation. The plant boasts low utility and reduced flaring requirements, because refrigerants are not flared on typical upset conditions. This leads to reduced requirements for maintenance and operational staffing.



LARGE SCALE LNG

Revision 0

The four train Atlantic LNG plant, with the 5.2Mtpa train (currently the worlds largest) in the foreground:

A schematic of the two train BG Idku plant in Egypt:



LARGE SCALE LNG

Revision 0

3.3 Shell

Shell is based in the Hague.

The APCI C3/MR designs were traditionally the industry and Shells standard. Shell has developed its own processes based on its experience with operating APCI process LNG trains in Brunei, Malaysia, Australia, Oman and Nigeria.

Shell has designed Single-Mixed-Refrigerant (SMR), Double-Mixed-Refrigerant (DMR) and Parallel-Mixed-Refrigerant (PMR) processes. With three different mixed-refrigerant cycles, Shell can design LNG trains with capacities ranging from 0.5 to 7.0Mtpa. The process options in the Shell portfolio are all based on two refrigeration cycles in series. Depending on the required capacity and local circumstances, different choices are made with respect to the line-up of rotating equipment, type of refrigerant, ambient cooling medium and types of cryogenic heat exchangers.

The Shell Double-Mixed-Refrigerant (DMR) process overcomes the inherent limitations of using a single component refrigerant in pre-cooling in the C3/MR design; the additional degree of freedom resulting from the use of two MR cycles allows full utilisation of power in a design with two mechanically driven compressors. Furthermore, it allows keeping the compressors at their best efficiency points over a very wide range (up to 50C) of ambient temperature variations and changes in feed gas composition.

3.3.1 SMR process.

The traditional C3/MR design loses its advantage in economy of scale for small capacity liquefaction trains. For this purpose, the more cost-effective Shell single-mixed-refrigerant (SMR) process has been developed. This type of design is most suitable for the lower capacity ranges of 0.5 to 1.5Mtpa of LNG.

The process uses one refrigerant loop, which is used both for pre-cooling the natural gas circuit as well as for pre-cooling the refrigerant. A dedicated, spool-wound heat exchanger provides for pre-cooling the mixed refrigerant. A bundle break in the cryogenic heat exchanger allows overhead cooling of the scrub column for removal of heavy hydrocarbons.

The unit operates at two mixed-refrigerant pressure levels. Three-stage compression drives the mixed-refrigerant loop. The driver could be electric drive or mechanical-drive gas turbines.

3.3.2 DMR process.

The Shell DMR process has two separate mixed-refrigerant loops, hence the name. The DMR process is designed for the mid and high capacity ranges of 1.5 to 4.5Mtpa of LNG. Shell is also developing a promising air-cooled DMR version with capacities from 2.0 to 4.0Mtpa for tropical conditions.

The pre-cool mixed refrigerant circuit is used just like the propane pre-cool loop in the C3/MR process. This circuit pre-cools both the natural gas circuit and the main mixed-refrigerant loop. The main difference between this process and the C3/MR design is in the use of two spool-wound heat exchangers, rather than a multiple of kettle exchangers for extracting heat from the circuits. Also, a less complicated two-stage centrifugal compressor is included in this design.

The choice of an end flash system integrated with the liquefaction design is dependent on the nitrogen content in the feed gas and the requirement of increased design capacities. For the large-scale Shell DMR process, the capacity of the system can be boosted by some 5.0 percent to 10 percent by applying an endflash system.

The DMR design was selected for the Sakhalin LNG Project currently under construction, with an annual design capacity of 9.6Mtpa (two trains @4.8Mtpa each). It is also being used for the Woodside NWS Train 5 and Pluto Projects.



LARGE SCALE LNG

Revision 0

Process configuration is similar to the APCI propane pre-cooled mixed refrigerant process, with the pre-cooling conducted by a mixed refrigerant (made up mainly of ethane and propane) rather than pure propane. Another main difference is that the pre-cooling is carried out in SWHE rather than kettles. The pre-cooling and liquefaction SWHEs for Sakhalin were supplied by Linde.

The DMR process has two separate mixed refrigerant cooling cycles. One is for pre-cooling gas to about - 50C (PMR cycle), and the other is for final cooling and liquefaction (MR cycle). This concept allows the designer to choose the load on each cycle. It also uses proven equipment, e.g. spiral-wound heat exchangers (SWHEs), throughout the process.

PMR vapour from the pre-cool exchangers is routed via knock-out vessels to a two-stage centrifugal PMR compressor. De-superheating, condensation and sub-cooling of the PMR is achieved by using induced-draft air coolers.

The PMR compressor is driven by a single gas turbine, equipped with an electric starter motor/ generator. The refrigerant compressors are driven by two Frame 7 gas turbines, equipped with a separate variable speed starter/ helper motor. An axial compressor is also used as part of the cold refrigerant compression stages.

The cooling for liquefaction of the natural gas is provided by a second mixed refrigerant cooling cycle (MR cycle). This cycle's refrigerant consists of a mixture of nitrogen, methane, ethane and propane. Mixed refrigerant vapour from the shell side of the main cryogenic heat exchanger is compressed in an axial compressor, followed by a two-stage centrifugal compressor. Inter-cooling and initial de-superheating is achieved by air cooling. Further de-superheating and partial condensation is achieved by the PMR pre-cooling cycle. The mixed refrigerant vapour and liquid are separated, and further cooled in the main cryogenic heat exchanger, except for a small slipstream of vapour MR, which is routed to the end flash exchanger.

Shell has also developed technology to further push the propane cycle capacity, by employing double casing instead of single casing equipment. This reliable method brings the propane-MR process closer to a capacity of 5Mtpa. Another possibility for the propane-MR process is to transfer power from the propane cycle to the mixed refrigerant cycle. The closer coupling between the two cycles by mechanical interlinking of compressors is an operational challenge.



LARGE SCALE LNG

Revision 0

A further development of the DMR process is the electrically-driven DMR (LNG GameChanger) design. The Gamechanger concept is based on parallel line-up of electrically driven refrigerant compressors around a common set of cryogenic spool wound exchangers. Electric motors of 65MW have already been constructed for LNG service. Motors up to 80MW are considered feasible. The current electrically driven DMR design is particularly attractive in the 58Mtpa capacity range. Electrically driven LNG trains can compete with mechanically driven trains because the increase in cost is compensated for by the increased availability. Other benefits of the electric option are the variable size and speed of the driver, the increased vendor base and the potential to make a step change reduction in overall plant carbon dioxide (CO2) emissions, by using combined cycle electric power generation.

3.3.3 PMR process

Shell's portfolio also includes designs for ultra large trains based on the company's own liquefaction technology.

The company's Parallel-Mixed-Refrigerant (PMR) process employs a common pre-cool cycle serving two parallel liquefaction cycles using three large industrial-type gas turbines, which enable a production of 6 to 7Mtpa. Much of the equipment has been used in previous designs. The train can run at 65% capacity when one of the liquefaction cycles is down.

Shell has developed the PMR process to meet the current challenge of the industry for larger train sizes in tropical conditions. With a single pre-cooling cycle and two PMR cycles, the capacity can be boosted up to 8Mtpa with three GE Frame 7 compressors in a tropical climate. The process can either use C3 or MR in pre-cooling. Proven refrigerant cycles can be used and the design can currently be applied, without step changes in technology. The capacity can be increased further with different (larger) drivers.

Gas receipt and natural gas treating is followed by a single propane pre-cooling cycle. After pre-cooling, the flow is distributed over two parallel strings, each having a scrub column for NGL extraction and an MR cycle for liquefaction and sub-cooling of the natural gas. The scrub column overheads are cooled by the MR to create reflux and ensure the required extraction level. Each liquefaction cycle has its own MR circuit, driven by a gas turbine. The sub-cooled liquid from the two liquefaction cycles is combined in an end-flash system, where fuel gas is flashed off and LNG is sent to storage at atmospheric pressure. The split-propane technology is applied to limit the propane suction flows to acceptable levels. In this arrangement, the four-stage propane compressor is split over two casings the first machine compresses the low-pressure (LP) and high pressure (HP) propane to discharge pressure, whereas the second machine handles the medium pressure (MP) and high-high-pressure (HHP) flows. This split arrangement results in a lower volumetric flow per stage, for the same compression duty. In order to achieve the targeted 8Mtpa production capacity with the driver configuration chosen, an extended end-flash system is required for the conditions prevalent in this study. The application of such a system allows an increase in the run-down temperature from the main cryogenic heat exchanger (MCHE) in the liquefaction cycle. The additional flash gas that is generated as a result of this, which cannot be accommodated in the fuel gas system, is compressed, condensed and recycled back into the end-flash gas system. In this way, power in the MR circuit is freed up and replaced by power in the end-flash gas compressor. This relieves the power constraint in the MR circuit and enables higher LNG production capacity. The extended line-up as chosen here results in some 4% to 5% additional LNG capacity. As an alternative for the extended end-flash system, larger compressor drivers like GE-Frame 9 or Siemens V84.2 can be used to achieve a production capacity of 8Mtpa or higher. Two of the key parameters in this process design are the scrub column overheads temperature and the cut-point temperature between the C3 and MR cooling cycles.



LARGE SCALE LNG

Revision 0

The scrub column overheads temperature sets the level of propane recovery. In order to achieve a specification of an LNG product quality of say 1,110 British thermal units (BTU) per standard cubic foot (scf), a propane recovery level of approximately 40% to 45% is required. At the pressures prevalent in the scrub column, this is accomplished by using MR cooling to approximately -45C. The lowest propane cooling level is the cut-point temperature between pre-cooling and liquefaction. This temperature is optimised so that the utilised power is balanced to a 1/2 ratio between the pre-cooling and the liquefaction cycles. The ability to tune the power balance exactly to the installed mechanical refrigeration capacity in the PMR process is an advantage over the conventional C3/MR process where, due to other constraints, the pre-cooling cycle cannot be fully loaded.

The Shell PMR technology for large LNG trains has a number of advantages. The processes are robust through the application of well-proven equipment (e.g. spool wound heat exchangers, proven rotating equipment) without requiring further scale-up. In fact, the main equipment in the PMR process is already in operation in plants such as Nigeria LNG, North West Shelf LNG and Malaysia LNG. The parallel line-up of the liquefaction cycles improves the reliability of the train since the LNG production can be designed to continue at 60% of the train capacity when one of the liquefaction cycles trips. Moreover, it allows high production capacity with only two refrigeration cycles in series, compared with three in most other large train concepts. Due to the PMR line-up and the application of split-propane technology, the installed power of three GE-Frame 7 drivers can be fully utilised. Since these machines form a large part of the liquefaction unit cost, this makes PMR designs very cost effective. The Shell PMR process has a high efficiency through the use of two very efficient refrigeration cycles. The parallel line-up reduces the pressure drop in the system, which also helps to improve efficiency. Another advantage is the absence of a third cycle with associated efficiency losses due to temperature approaches in the cooling of the third cycle refrigerant. Comparison by Shell of different large-train processes for similar conditions and design premises has shown that the Shell PMR process has an efficiency up to 10% better than alternative processes.

In a tropical climate, where air-cooling and GE-Frame 7 drivers are used, an LNG production capacity of 8Mtpa can be achieved with the Shell PMR process. This production capacity can be achieved without stopouts in main equipment or process technology. The Shell PMR process can be designed to produce different grades of LNG. The installed driver power can be utilised fully by changing the cut-point temperature between the pre-cooling and liquefaction refrigerant cycles.



LARGE SCALE LNG

Revision 0

3.4 Linde

Linde Engineering is based in Munich.

Their business model is to market their patented technologies from early concept through to implementation. Therefore the Statoil/Linde MFCP is exclusively available through Linde.

Linde is highly experienced with cryogenic processing technologies (to be expected for a company whose founder Karl Linde in 1873 invented the first mechanical refrigeration), with many decades of ethylene, gas processing, air separation and LNG experience.

Linde is well established in providing small scale LNG peak shaving plants (with 5 plants, starting with their first LNG peak shaver in 1972), base load plants (with 4 very small plants plus Xinjiang and Snohvit) and satellite plants (with 7 small plants).

Their flagship LNG project is the Statoil-Linde process 4.3Mtpa Snohvit LNG project due for commissioning mid-2007. Their next biggest plant constructed is the 0.43Mtpa Xinjiang Phase 1 plant in China, and like Black & Veatch they have submitted a LSTK bid for the ~0.8Mtpa Phase II expansion of this.

They are co-developer with Statoil of the LNG liquefaction technology being utilised for the Snohvit LNG development in Norway. The Statoil/ Linde LNG Technology Alliance was established to develop alternative LNG baseload plants for the North Sea and this work resulted in a new LNG baseload process, the Mixed Fluid Cascade Process (MFCP).

Additionally, Linde manufacture cryogenic exchangers (both plate fin heat exchangers and spiral wound heat exchangers) and have done so for decades. Linde pioneered spiral wound heat exchangers in early 1900s. As well as being used within their own process, Linde spiral wound heat exchangers (SWHE) are being installed within APCI and Shell liquefaction processes, in new projects and as replacement for old APCI cryogenic exchangers, on many world-scale LNG trains. The MFCP uses three mixed refrigerants to provide the cooling and liquefaction duty. Pre-cooling is carried out in a plate fin heat exchanger (PFHE) by the first mixed refrigerant, and the liquefaction and sub-cooling are carried out in spiral wound heat exchanger (SWHE) by the other two refrigerants. The SWHE may also be used for the pre-cooling stage. The refrigerants are made up of components selected from methane, ethane, propane and nitrogen. The 3 refrigerant compression systems can have separate drivers or integrated to have 2 strings of compression. Frame 6 and Frame 7 gas turbine drivers have been proposed for large LNG trains (> 4Mtpa). A novel feature of the Snhvit project is that all electric motor drivers will be used for the main refrigerant compressors, with sizes up to 60 MW.

Within this proprietary process, purified natural gas is pre-cooled, liquefied and sub-cooled by three separate mixed refrigerant cycles. The pre-cooling cycle's cold is transferred to natural gas via two PFHEs, whereas the cold of the liquefaction and sub-cooling cycle is transferred via two SWHEs by the other two refrigerants. The three refrigerant compression systems can have separate drivers or be integrated to have two strings of compression.

The MFCP is a classic cascade process with one important difference - mixed component refrigerant cycles replace single component refrigerant cycles, and thereby improve thermodynamic efficiency and operational flexibility. The MFCP concept is built up by well-known elements. The size and complexity of the separate SWHE applied in the MFCP are considerably less when compared with today's single unit used in dual-flow LNG plants. Last, but not least, MFCP allows larger, single compressors to handle refrigerant over a larger temperature scale.



LARGE SCALE LNG

Revision 0

For Snohvit the process facilities are barge-mounted for ease of fabrication, to save costs since the site is remote. The barge forms the permanent foundation for the process equipment. The barge, with process plant facilities installed, was fabricated in Spain and dry towed to Hammerfest on a heavy lift vessel.

The Snohvit process barge being floated into its prepared space at site:



LARGE SCALE LNG

Revision 0

The completed Snohvit LNG plant on Melkoya Island, Hammerfest:

3.5 Axens

Axens is based in Paris.

The Institute Francais de Petrole (IFP) developed the Liquefin process that is licensed through Axens. While this process has not yet been accepted for a commercial application, BP invested in its evaluation for commercial application and it is being considered for several projects in Iran (where companies such as APCI and ConocoPhillips are excluded due to USAs trade embargo).

Claims made by Axens for the Liquefin process are that it produces LNG cheaper than with any other process and that very high capacities can be reached with a simple scheme and standard compressors. No plants have been built so these claims are yet not proven.

The Liquefin process is a two-mixed refrigerant process designed for base load projects of train sizes up to 6Mtpa. All cooling and liquefaction is conducted in plate-fin heat exchangers (PFHE), arranged in cold boxes. The PFHE arrangement is at the heart of the liquefaction technology. The PFHEs are non-proprietary and can be supplied by a number of independent vendors.

The refrigerants are made up of components from methane, ethane, propane, butane and nitrogen. The first mixed refrigerant is used at three different pressure levels to pre-cool the process gas and pre-cool and liquefy the second mixed refrigerant. The second mixed refrigerant is used to liquefy and sub-cool the process gas.

Using a mixed refrigerant for the pre-cooling stage, the temperature is decreased down to a range of - 50C to - 80C depending on refrigerant composition. At these temperatures, the cryogenic mixed refrigerant can be completely condensed, no phase separation is necessary and, moreover, the quantity of cryogenic refrigerant is substantially reduced.



LARGE SCALE LNG

Revision 0

Overall necessary power is decreased, as the quantity of cryogenic mixed refrigerant is lower, and a good part of the energy necessary to condense it is shifted from the cryogenic cycle to the pre-refrigeration cycle. Moreover, this shifting of energy allows a better repartition of exchange loads. The same number of cores in parallel can be used all along between the ambient and the cryogenic temperature, allowing a very compact design for the heat exchange line. A very significant advantage of this new scheme is the possibility to adjust the power balance between the two cycles, making it possible to use the full power provided by two identical gas drivers.

The Liquefin process is very flexible. It offers more than one possibility to reach large, highly competitive capacities, either by using very large gas turbines (combined cycle) to produce electricity, and large electrical motors (up to 70 MW) in parallel on each cycle, or by using larger gas turbines. Frame 7 gas turbines are proposed for large LNG trains. The Frame 9 has very recently been qualified for mechanical drive. With Liquefin, this would allow capacities of 7 to 8 Mtpa with only two main drivers.

The process has been reported by Axens as representing a real breakthrough with a total cost reduction per ton of LNG of 20% when compared to the APCI C3-MR process. The cost reductions arrive from: 1) increasing the plant capacity; 2) reducing the heat exchanger costs; 3) all-over plate-fin heat exchangers; 4) a compact plot area; and 5) multi-sourcing of all equipment, including heat exchangers.

It is particularly well-adapted to the range of 4 to 8Mtpa per train.

4. Associated Facilities As well as the LNG liquefaction process plan described above, other specialised facilities are required for an LNG development including those systems for; storage, loading, LNG pumpout and boil-off handling.

Storage

The system consists of one or more specially designed tanks. Ships to transport the LNG arrive at the terminal at specific intervals. The minimum required storage capacity is the volume of LNG discharged to the largest ship expected at the terminal. In practice, the installed storage is larger than this minimum. The extra storage provides a cushion to account for scheduled and unscheduled delays in ship arrival. The storage tanks represent a substantial capital cost. The volume of LNG stored in these tanks is large and a failure of one or more tanks could have disastrous consequences. Because of the exacting design and operational techniques used, the modern LNG industry has had an excellent safety record.



LARGE SCALE LNG

Revision 0

Loading

The LNG loading system consists of all the facilities, infrastructure and equipment required to safely dock the LNG ship, to establish the necessary ship-to-shore interfaces, and for transferring the cargo from the onshore tanks to the ship's tanks. The system also includes facilities for disconnecting the ship-to-shore interface at the end of the loading operation, and for undocking the ship prior to its return voyage.

Specifically, the loading system consists of:

Breasting and mooring dolphins for securing the LNG ships to the loading berth;

The loading platform which supports the loading arms and the control building;

The control and emergency safety systems housed within the control building;

The liquid loading arms for transferring LNG from the shore piping to the ship;

The vapour return arm for returning vapour from the ship to shore (created from the ship's cargo tanks because of the LNG being pumped in);

Connections for transfer of utilities (e.g. nitrogen) from the shore to the ship;

The piping, valves and vessels required for transferring the LNG;

The return vapour and the utilities between the loading berth and the main terminal facilities; and

If the loading berth or jetty is some distance from shore, a pier connecting the jetty to the shore, will be required to provide both access to the jetty and support the ship-to-shore piping.

LNG Pumpout

The LNG tanks operate at very low pressure, just slightly above atmospheric pressure. Pumping of cryogenic liquids, especially at the high rates required in LNG facilities, is specialized technology. In modem terminals the first stage pumps are almost always installed inside the storage tanks, and referred to as in-tank pumps. The second stage pumps, when required (if for instance there is a long loading jetty), are located outside the tanks. These second-stage pumps discharge at a pressure sufficiently high to satisfy the battery limit pressure at the terminal fence.

Boil-Off Handling

LNG is a cryogenic liquid having a temperature, at atmospheric pressure, of about -1620C. Heat entering the LNG (often referred to as "heat in-leak") causes the LNG to warm up. However, in the storage tanks the LNG needs to be maintained at a sufficiently low temperature, consistent with the low operating pressure. Hence, heat absorbed by the LNG has to be released by "flashing" (or boiling-off) some of the liquid to gas.

Handling of boil-off gas requires compression equipment that is costly to install and operate. Every effort is made to reduce the amount of boil-off gas produced. Three main factors cause LNG boil-off;

The LNG loaded into the ship may be slightly warmer than the temperature in the storage tanks;

The energy used by the loading pumps is ultimately transferred to the LNG as heat; and

Ambient heat transferred into the LNG through the cryogenic insulation in pipes, equipment and storage tanks.



LARGE SCALE LNG

Revision 0

Boil-off gas is essentially gasified LNG at atmospheric pressure. It has substantial fuel value and, except in emergencies, should not be vented or flared. Design and operation of the boil-off gas handling system involves setting priorities for its efficient disposition, including its utilization as fuel gas and reliquefaction. Boil-off gas generated in the loading mode can be many times greater than the gas generated in the period between loadings (the period between loadings is referred to as the "holding mode"). Hence, larger compression equipment may be needed for the loading mode.

4.1 LNG Storage Tanks

LNG storage tanks account for a large portion of the cost of an LNG plant. LNG is stored in double-walled tanks at atmospheric pressure. The storage tank is a tank within a tank, with insulation between the walls of the tanks.

Important factors to consider while specifying the LNG storage system include:

Total storage capacity required;

Number of tanks required;

Type of containment preferred;

Applicable codes to be used; and

Other considerations like tank internals, commissioning and insulation.

4.1.1 Total Storage Capacity

Determination of total storage capacity is seldom a simple and straightforward exercise. Clearly, the minimum required capacity would be the volume of the largest LNG tanker expected at the terminal, plus a small margin above this (buffer). Another way to look at the storage requirement is in terms of number of days of LNG production.

The number of days of buffer (spare volume) storage capacity varies. For instance, the Darwin LNG storage tank at 188,000m3 capacity has only some 2.5days of buffer when loading 145,000m3 ships. 5 to 10 days is more typical however.

Computer simulations are helpful in fine-tuning decisions regarding the LNG storage capacity, and also the number of LNG ships, their size and speed, and their utilization among different facilities. It is important to note that the primary determinant of storage capacity is the philosophy adopted by the owners. Computer simulations can be used as a tool for fine tuning the capacity, after the basic philosophy has been established.

The theoretical volume of storage required, assuming there are no delays in LNG ship arrivals and no variations in LNG production rate, is easy to calculate. In practice there will be events-both scheduled and unexpected-that will cause deviations from this theoretically ideal situation. These could include, for example, predictable events like maintenance turnaround at the liquefaction plant, scheduled maintenance for the LNG ships, seasonal variations in LNG delivery, maintenance at downstream power plants, or seasonal variations in sendout requirements. Other disruptions that are anticipated, but whose timing cannot be predicted, might include unscheduled downtime at the liquefaction plant, weather-related ship delays, or unexpected downtime at the power plant.

4.1.2 Number of Tanks

Once the total storage capacity is established the number of tanks should be decided.



LARGE SCALE LNG

Revision 0

The minimum number of tanks can be determined based on the total storage volume required and the maximum capacity of a single tank. The latter number however is not fixed and will depend on the type of containment, type of construction and applicable codes.

For Greenfield LNG export plants the norm has been to install two 50% tanks as part of the initial development. This was to provide assurance that at least one tank would be available should one tank require scheduled or unscheduled maintenance. For Darwin LNG however, the Operator (Phillips) weighed up the risks and made the decision to install a single tank (at 188,000m3, the largest above-ground tank installed at that time). Despite the excellent record of reliability with LNG tanks many owners prefer two smaller sized tanks instead of a single large tank. The baseload nature of the facility and the implications of a long-term take or pay contract often favour multiple tanks.

Most of the LNG tanks in service have capacities of 100,000 cubic meters or less. Aboveground tanks with an inner metal wall have now been built for capacity as high as 200,000 cubic meters. However, the maximum capacity is limited by the availability of the required wall thickness 9% nickel plate. Below-ground tanks using the membrane type design with reinforcing concrete have been built for capacities as high as 200,000 cubic metres, and above-ground tanks with concrete inner and outer walls have been proposed for 250,000 cubic metres capacity.

In specifying storage tank capacity it is important to remember that the "usable" volume in the tank is less than the total tank volume. The minimum level to which the LNG in the tank can be lowered, will be limited by the LNG pumps' ability. Similarly, to avoid tank overfill it will be necessary to limit the maximum fill level to less than the full height of liquid container. The ratio of usable volume to built-up volume will depend on the tank height, the pumpout arrangement, the LNG pump characteristics, and the instrumentation/control philosophy Typically only about 95% of the volume is usable (for example the Darwin LNG of nominal 188,00m3 capacity has actually some 200,000m3 of built capacity, or 94% usable). The LNG at the bottom of the tank that isnt used is called the heel.

4.1.3 Type of Containment

A variety of storage tank designs have been employed in LNG service, all of which are characterized by their heavy insulation and special material requirements.

LNG storage tanks are classified in terms of containment type (single, double or full), and erection method (in-ground, semi-buried and above-ground). All three containment types are designed to store LNG safely and contain any spills in the unlikely event of a leak in the primary liquid container, and are designed with an inner and outer wall separated by insulation materials. The inner wall must be designed for LNGs low cryogenic temperature and the material used most extensively is 9% nickel steel, as this remains ductile at cryogenic temperatures.

Safe use of LNG, or any cryogenic substance, requires an understanding of how materials behave at cryogenic temperatures. At extremely low temperatures, carbon steel loses its ductility and becomes brittle. Therefore, the material selected for tanks, piping, and other equipment that comes in contact with LNG is critical. The use of high nickel content steels, aluminium, and stainless steels is costly but necessary to prevent embrittlement and material failures. High alloy steels composed of 9% nickel and stainless steel will be used for the inner tank of LNG storage tanks and for other LNG applications. In much of the discussion above the primary liquid containment is assumed to be constructed of 9% Ni steel. In addition to 9% Ni, other materials that are suitable for cryogenic service include aluminium and stainless steel. Aluminium is no longer considered economical for large LNG tanks. However, stainless steel is a viable material and is routinely used in the membrane-type design. The membrane technology for LNG tanks relies on a post-tensioned concrete outer tank for structural strength and a steel-corrugated membrane for liquid and gas tightness. Membrane type tanks have been used extensively in Japan where in-ground tanks have been built with capacity as high as 200,000 cubic meters. Membrane technology has also been used successfully for above ground LNG tanks.



LARGE SCALE LNG

Revision 0

The main features of the typical single, double and full containment tanks are summarised below:

1. Single Containment Tank

Double-walled with an interior tank is made of 9% nickel, while the outer tank is made of carbon steel;

Only the inner primary container is required to meet the low temperature ductility requirement for

storage of the product;

The outer container serves primarily to contain insulation and vapour and to provide a weather shield. In the event of leakage from the primary container the outer tank is not designed to contain the refrigerated liquid;

The tank is surrounded by a bund wall or dike to contain any leakage;

Single containment tanks are less expensive and rely on a separate impoundment to contain the design

spill;

The required distance between the earthen type bund wall and the tank adds significantly to the total land area. This type of impoundment system has a large footprint, resulting in a large heat flux exclusion zone;

The cost of a single containment tank is about 65% that of a corresponding full-containment tank. If land is scarce this cost advantage might be reduced; and

The construction time for a single containment tank will be about four months less compared to a full or double containment tank.

2. Double Containment Tank

Both the inner self-supporting primary container and the secondary container are capable of

independently containing the refrigerated liquid;

The secondary tank, typically a concrete wall, is located outside the primary tank. In the event of a leak, the secondary tank contains the cryogenic liquid and limits the surface area and vaporization of an LNG liquid pool. However, it is not intended to contain any vapour resulting from such a leakage;



LARGE SCALE LNG

Revision 0

The primary liquid container and the shell to contain the insulation are similar to a single-containment tank. However, in addition to it there is a surrounding concrete wall that is capable of containing the cryogenic liquid in the event of a leakage from the primary container. Unlike the bund wall surrounding a single-containment tank, this wall is located close to the primary container. This ensures that the liquid pool, in the event of LNG leakage, has a smaller surface area compared to the single-containment system.

The 2 x 140,000m3 LNG double containment storage tanks at the BG Idku, Egypt LNG facility:

3. Full Containment

A full containment tank typically consists of a 9% Ni inner tank with a prestressed concrete outer tank. The reinforced concrete roof is lined with carbon steel, with the liner also functioning as formwork for the concrete.

Both the self-supporting primary container and the secondary container are capable of independently containing the refrigerated liquid;



LARGE SCALE LNG

Revision 0

The inner tank contains the LNG under standard operating conditions. The outer shell, bottom and roof are made out of pre-stressed concrete;

The outer tank supports the outer roof and is also intended to contain the LNG;

Full containment tanks offer the highest level of safety;

The outer tank or wall composed of approximately 1 metre of concrete is one to two meters away from

the inner tank;

The outer tank is capable both of containing the refrigerated liquid and of controlled venting of the vapour resulting from product leakage after a credible event;

The outer tank, which includes a reinforced concrete roof lined with carbon steel, can be designed to

withstand realistic impacts from missiles or flying objects;

Concrete provides good resistance to heat radiation from nearby LNG fires. There will be a significant time delay before structural weakening of the reinforcement occurs;

Concrete also provides good protection against possible LNG spills on the tank roof. The effects of cold-shock, if any, will most likely be restricted to a small area, and generally should not affect the vapour-tight integrity of the tank;

The cost significantly more and require about six months longer to construct than the equivalent single containment tanks; and

Typically, full containment type tanks are used in sites where the public is nearby or where security

issues exist. Outside the U.S., virtually all new above-ground LNG storage tanks have been either full-containment or double-containment designs.



LARGE SCALE LNG

Revision 0

4.1.4 Pump Column for In-tank Pumps

Modern storage tanks have no side or bottom penetrations. All penetrations, including those for LNG sendout, are through the roof of the tank to avoid siphoning of the full content of the tank in case of piping failures. This design substantially reduces the amount of LNG spilled in the unlikely event of a rupture or leakage in the sendout piping.

In-tank pumps are provided to transfer the LNG out of the tanks and into the sendout system. In older facilities the LNG pumps were usually located external to the tank, and a cryogenic line from the bottom of the tank conveyed the LNG to the pump suction. In modern facilities, for safety reasons, LNG tanks are designed with no bottom or side penetrations. Instead, in-tank pumps, located at the bottom of the tank and inside a pump column, are used. The fabrication and installation of the pump columns requires coordination with the pump supplier.

The LNG transfer pumps comprise in-tank pumps submerged within pipe deepwells extending from the top of the domed roof to the bottom of the inner tank. Typically, four pump deepwells are installed, but only three wells will contain pumps, leaving one well as a spare. Crane/winch facilities above the roof are used to extract and re-install pumps undergoing maintenance.



LARGE SCALE LNG

Revision 0

The 188,000m3 Darwin LNG storage tank showing the external riser and pump access platform:

4.1.5 Tank Pressure Control

The tank pressure must be controlled within a narrow range. During normal operation, the vapour handling system will increase or decrease the boil-off' gas removal rate to maintain the required pressure. However, properly designed over-pressure protection and vacuum protection systems must be installed to handle upset conditions and unusual circumstances. Typically, metal roof tanks are restricted to a design pressure of less than 150 millibar gauge. Concrete roof tanks can be designed to withstand a much higher internal pressure, perhaps as high as 300 millibar gauge. A higher design pressure allows a greater range of operating pressures, and may also permit direct return of vapour to the ship, without the need for compression. LNG tanks are usually designed for vacuum conditions between 0 and -10 millibar gauge. Under normal operation a vacuum condition is not expected, but a vacuum protection system is required to safeguard against upset conditions.

4.1.6 Purging and Cooldown

When an LNG tank is put into service, such as during initial commissioning, the atmosphere in the tank has to be changed from air to natural gas. Natural gas vapour is primarily methane which, in certain concentration ranges, can form a flammable mixture with oxygen. To avoid the possibility of forming a flammable mixture the oxygen content in the tank must be reduced to less than 12%. In practice a margin of safety is included and the oxygen content should be reduced to around 8% or 9%. This is accomplished by purging the tank with nitrogen, which is an inert gas. The annular space between the inner and outer tanks contains the insulation, usually loose perlite. Effective purging of the perlite is also a requirement and means to accomplish this must be provided.

Cooldown of the tank is a sensitive operation, and must be completed prior to filling it with LNG. Cooldown is accomplished in a slow and gradual manner with cooldown rates (degrees per hour) limited by the tank vendor specifications. Cooldown must be not only gradual but also uniform, so that temperature gradients within the tank are within the limits specified by the tank vendor. Cooldown is accomplished by spraying liquid nitrogen or LNG into the tank. A spray ring, located below the suspended deck of the tank, ensures uniform spraying and cooldown. Sufficient number of thermocouples, located at suitable intervals, are provided to monitor the



LARGE SCALE LNG

Revision 0

cooldown progress. If a source of liquid nitrogen is conveniently available it would be preferable to have the first tank cooled and ready to receive LNG when the first ship arrives. Subsequent tanks can then be cooled using LNG from the first tank.

4.1.7 Insulation

Insulation is necessary to limit heat leak into the LNG tanks. Heat leak typically averages around 0.05% to 0.06% of full tank contents per day. Different types of insulation are used in different parts of the tank. Typically, the annular space between the inner and outer tanks is filled with loose perlite (expanded mica). In addition, a resilient blanket, such as fibreglass material, is installed on the outside of the inner tank. This blanket provides resiliency for the perlite as the tank contracts due to temperature changes, and prevents settling of the perlite. The blanket also facilitates flow of the purge gas during the tank inerting process. In membrane type tanks an internal insulation such as rigid PVC foam is used to transmit liquid pressure from the membrane to the concrete tank.

Heat leak from the roof of the LNG tank is limited by installing insulation on the suspended deck (which is suspended from tile roof). There is no insulation immediately beneath the roof, and the vapour space between the suspended

Documents

Microsoft Word - LNG Large Scale Background Paper Oct2006 PDF