AU2012273829B2

AU2012273829B2 - Method for liquefying natural gas with a triple closed circuit of coolant gas

Info

Publication number: AU2012273829B2
Application number: AU2012273829A
Authority: AU
Inventors: Marc Bonnissel; Bertrand DU PARC; Eric ZIELINSKI
Original assignee: Saipem SA
Current assignee: Saipem SA
Priority date: 2011-06-24
Filing date: 2012-06-22
Publication date: 2016-05-26
Anticipated expiration: 2032-06-22
Also published as: AU2012273829C1; EP2724100A2; WO2012175889A4; HUE053378T2; WO2012175889A2; US20140190205A1; BR112013033341B1; US9557101B2; CY1124080T1; DK2724100T3; HRP20210341T1; ES2854990T3; EP2724100B1; WO2012175889A3; AU2012273829A1; FR2977015B1; FR2977015A1; RS61507B1; BR112013033341A2; PT2724100T

Abstract

The present invention relates to a method for liquefying a natural gas primarily including methane, in which said natural gas to be liquefied is liquefied by performing the following concomitant steps: (a) circulating said natural gas to be liquefied that is circulating in three heat exchangers which are mounted in series and in which the gas is cooled to T3, T3 being no higher than the liquefaction temperature of said natural gas at atmospheric pressure; (b) the closed-circuit circulation of at least: a first stream of coolant gas at a pressure P1 that is lower than P3, which is entering the third exchanger and leaving said first exchanger, said first stream being obtained by expanding, in a first expansion valve, a first portion of said second stream to P3, which is higher than P2, said second stream circulating parallel to said stream of natural gas while entering said first exchanger and leaving said second exchanger; and a third stream at a pressure P2 higher than P1 and lower than P3 circulating parallel to said first stream which is entering said second exchanger and exiting said first exchanger, and which is obtained by expanding, in a second expansion valve, a second portion of said second stream exiting said first exchanger; and (c) said second stream of coolant gas at the pressure P3, which is obtained by compression by means of at least two first and second compressors arranged in series and coupled to said first and second expansion valves.

Description

1 METHOD FOR LIQUEFYING NATURAL GAS WITH A TRIPLE CLOSED CIRCUIT OF COOLANT GAS 1 The present invention relates to a process for liquefying natural gas in order to produce liquefied 5 natural gas (LNG). Still more particularly, the present invention relates to liquefying natural gas comprising a majority of methane, preferably at least 85% methane, with the other main constituents being selected from nitrogen and C-2 to C-4 alkanes, i.e. ethane, propane, 10 and butane. The present invention also relates to a liquefaction installation on board a ship or a floating support at sea, either on the open sea or in a protected zone such as a port, or indeed an installation on land for small or 15 medium natural gas liquefaction units. With an installation on board a ship, the present invention relates more particularly to a process for reliquefying gas on board an LNG transport ship, known as a "methane tanker", said gas for reliquefying being the 20 result of the LNG contained in the tank of said ship heating and evaporating in part, said evaporated gas, generally a majority of methane, being referred to as "boil-off". The methane-based natural gas is either a by-product 25 of oil fields, being produced in small or medium quantities, in general in association with crude oil, or else it is a major product from a gas field, where it is to be found in combination with other gases, mainly C-2 to C-4 alkanes, C0 2 , and nitrogen. 30 When the natural gas is associated in small quantities with crude oil, it is generally treated and separated and then used on site as fuel for turbines or piston engines in order to produce electricity and heat for use in the separation or production processes. 1 Translation of the title as established ex officio.

2 When the quantities of natural gas are large, or indeed substantial, they need to be transported so that they can be used in regions far away, in general on other continents, and in order to do this the preferred method 5 is to transport the gas while it is in the cryogenic liquid state (-165*C) and substantially at ambient atmospheric pressure. Specialized transport ships known as "methane tankers" possess tanks of very large dimensions with extreme insulation in order to limit 10 evaporation while traveling. Gas is generally liquefied for transport purposes in the proximity of the production site, generally on land, and that requires substantial installations in order to achieve capacities of several millions of (metric) tonnes 15 per year, with the largest existing units combining three or four liquefaction units, each having a unit capacity of 3 megatonnes (Mt) to 4 Mt per year. The liquefaction process requires substantial quantities of mechanical energy, with the mechanical 20 energy generally being produced on site by taking a portion of the gas in order to produce the energy needed by the liquefaction process. A portion of the gas is then used as fuel in gas turbines, steam turbines, or piston engines. 25 Numerous thermodynamic cycles have been developed for the purpose of optimizing overall energy efficiency. There are two main types of cycle. A first type is based on compressing and expanding a refrigerant fluid with a change of phase, while a second type is based on 30 compressing and expanding a refrigerant gas without a change of phase. The terms "refrigerant fluid" and "refrigerant gas" are used to designate a gas or gas mixture circulating in a closed circuit and being subjected to stages of compression, possibly of 35 liquefaction, then of heat exchange with the external medium, and subsequently stages of expansion, possibly of evaporation, and finally of heat exchange with the 3 natural gas for liquefying, which gas comprises methane, and cools little by little to reach its liquefaction temperature at atmospheric pressure, i.e. about -165*C for LNG. 5 Said first cycle type with a change of phase is generally used in installations on land and it requires a large amount of equipment and occupies a large footprint. In addition, the refrigerant fluids, generally in the form of mixtures, are constituted by butane, propane, 10 ethane, and methane, which gases are dangerous since in the event of a leak they run the risk of leading to substantial fires or explosions. In contrast, in spite of the complexity of the equipment required, they remain more efficient and they require about 0.3 kilowatt hours 15 (kWh) of energy per kilogram (kg) of LNG that is produced. Numerous variants of that first type of process with a change of phase in the refrigerant fluid have been developed, and suppliers of technology or of equipment 20 have their own formulations of mixtures associated with their specific equipment, both for so-called "cascade" processes and for so-called "mixed cycle" processes. The complexity of those installations comes from the fact that in those stages where the refrigerant fluid is in 25 the liquid state, and more particularly in separators and in connection pipes, it is appropriate to install gravity collectors in order to bring the liquid phase together and direct it to the cores of heat exchangers where it vaporizes on coming into contact with the methane for 30 cooling and liquefying in order to obtain LNG. Those devices are very bulky, but that does not lead to problems for installations on land, since it is generally simple to obtain an area of land that is large enough to house all of those bulky pieces of equipment side by 35 side. Thus, for installations on land, all of the compressor, heat exchanger, and collector pieces of equipment are generally installed side by side on 4 substantial areas, lying in the range 25,000 square meters (M 2 ) to 50,000 M 2 , or even more. The second type of liquefaction process, without any change of phase in the refrigerant gas, is an inverse 5 Brayton cycle or a Claude cycle using a gas such as nitrogen. The efficiency of the second type of process is lower, since it generally requires about 0.5 kWh of energy per kg of LNG produced, i.e.; about 20.84 kilowatt-days per tonne (kWxd/t), but in contrast 10 it presents a substantial advantage in terms of safety since the cycle refrigerant gas, nitrogen, is inert and thus incombustible, which is very advantageous when the installations are concentrated in a small amount of space, e.g. on the deck of a floating support located in 15 the open sea, where said equipment is often installed on a plurality of levels one above another on an area that is reduced to the strict minimum. Thus, in the event of the refrigerant gas leaking, there is no danger of explosion and it then suffices to reinject into the 20 circuit the fraction of the refrigerant gas that has been lost. Furthermore, that process for liquefying natural gas without a change of phase is very advantageous on board floating supports since the equipment is of much simpler 25 design, because there is no liquid phase in the refrigerant gas. In such installations, all of the equipment is moving practically continuously as a result of the movements of the floating support (roll, pitching, yaw, lurch, surge, heave). Managing a process with a 30 phase change involving a liquid phase of the refrigerant fluid would then be extremely difficult, even for small movements of the floating support, and indeed practically impossible for extreme movements, whereas stationary installations on land do not face the problem of 35 movements. In spite of the lower energy efficiency of the liquefaction process without a change of phase of the 5 refrigerant gas, this process remains very advantageous since the equipment used, mainly compressors, expanders, turbines, and heat exchangers is much simpler than the equipment required for a liquefaction process involving 5 cycles with a change of phase in a refrigerant fluid, both in terms of the technology used for said equipment and in terms of maintaining the equipment in an environment that is confined, i.e. on board a floating support that is anchored at sea. Furthermore, the 10 running of such installations in operation remains simpler, since this type of cycle is relatively insensitive to variations in the composition of the gas for liquefying, i.e. a natural gas that is constituted by a mixture in which methane predominates. In the cycle 15 with a change of phase in the refrigerant fluid, in order to ensure that efficiency remains good, the refrigerant fluid needs to be matched to the nature and the composition of the gas that is to be liquefied, and the composition of the refrigerant fluid might possibly need 20 to be modified over time as a function of the composition of the natural gas mixture for liquefying as produced by the oil field. In principle, implementing a cycle of the liquefaction process without a change of phase in the 25 refrigerant gas, such as nitrogen, comprises the four following main elements: - a compressor that increases the pressure of the refrigerant gas and causes it to go from ambient temperature at low pressure to high temperature at high 30 pressure; - a heat exchanger that cools the refrigerant gas from the high temperature at high pressure substantially down to ambient temperature at high pressure; * an expander device, generally a decompression 35 turbine, in which the refrigerant gas expands: its pressure drops and its temperature is then very low; while simultaneously mechanical energy is recovered from 6 the expansion turbine, which mechanical energy is generally reinjected directly to the compressor that is coupled thereto; and * a cryogenic heat exchanger through which there 5 flow both the refrigerant gas at cryogenic temperature and also the gas for liquefying, said refrigerant gas absorbing heat from the gas for liquefying, and thus heating up, while said gas for liquefying gives off heat and cools until it reaches the looked-for liquid state. 10 At the end of the heat exchanger cycle, the refrigerant gas is substantially at ambient temperature and it is then reintroduced into the compressor in order to perform a new closed-circuit cycle. Throughout the duration of the cycle, the 15 refrigerant gas remains in the gaseous state and it circulates in continuous manner, as explained above: it releases its "frigories" little by little, i.e. absorbs calories little by little from the gas that is to be liquefied, i.e. a mixture that is constituted for the 20 most part by methane together with traces of other gases. The gas for liquefying flows as a countercurrent relative to the refrigerant gas, i.e. said natural gas comprising methane enters the heat exchanger substantially at ambient temperature close to the 25 refrigerant gas outlet where the refrigerant gas is substantially at ambient temperature. Thereafter, the natural gas comprising methane advances into the heat exchanger towards colder zones and transfers its heat to the refrigerant fluid: the natural gas comprising methane 30 cools while the refrigerant gas heats up. As the natural gas comprising methane advances into the heat exchanger, its temperature drops, and at the end of its travel it liquefies and its temperature continues to drop until it reaches a temperature T3=-165 0 C for a gas containing 85% 35 methane. Throughout its passage through the heat exchanger(s), the natural gas is liquefied at a pressure 7 PO lying in the range 5 bars to 50 bars, in general in the range 10 bars to 20 bars, in four main stages: - stage 1: cooling the natural gas from ambient temperature TO down to Tl=-50*C approximately (this 5 temperature depends on the composition of the natural gas); - stage 2: liquefaction of the natural gas (passing from the gaseous state to the liquid state). Since the natural gas is a mixture of gases at a pressure PO of a 10 few tens of bars, approximately, this change of state is spread over the temperature range Tl=-50 0 C to T2=-120 0 C, approximately; - stage 3: once the natural gas has liquefied completely (LNG), it is at about T2=-120*C, and still at 15 a pressure PO of several tens of bars approximately. Within the heat exchanger(s), the LNG continues to be cooled until it reaches the temperature T3 of -165"C, which temperature corresponds to LNG being in a liquid phase at atmospheric pressure; and 20 - stage 4: the resulting liquid or LNG is then depressurized down to atmospheric pressure where it remains in the liquid state because its temperature T3 is lower than or equal to -165 0 C, and it can be transferred to an insulated storage tank, or possibly loaded directly 25 on board a transport ship such as a methane tanker. Stage 2 consumes the most energy, since it is necessary to supply the gas with all of the energy that corresponds to its latent heat of vaporization. Stage 1 consumes a little less energy, and stage 3 consumes least 30 energy, but it takes place at the lowest temperatures, i.e. at temperatures around -165*C. The values given above for T1, T2, and T3 are appropriate for a natural gas comprising 85% methane and 15% of said other components comprising nitrogen and C-2 35 to C-4 alkanes, and those temperatures may be significantly different for a gas having a different composition.

8 Figure 1 is a diagram of an installation for performing a standard process for liquefying natural gas using a refrigerant gas constituted by nitrogen without a change of phase in the refrigerant gas, as described 5 above, with the process being described in greater detail below. US 2011/0113825 and WO 2005/071333 describe a process for liquefying natural gas in which said natural gas for liquefying is liquefied by causing the natural 10 gas to flow through three cryogenic heat exchangers, while causing three streams of refrigerant gas that remains in the compressed gaseous state without a change of phase to circulate in three closed circuits. Said natural gas for liquefying is liquefied by performing the 15 following concurrent steps: a) causing said natural gas for liquefying to flow at a pressure PO that is higher than or equal to atmospheric pressure through the three cryogenic heat exchangers connected in series, namely: 20 - a first heat exchanger (101/5) into which said natural gas enters at a temperature TO, is cooled, and leaves at a temperature Tl lower than TO; then - a second heat exchanger (102/6) in which the natural gas is completely liquefied and leaves at a 25 temperature T2 lower than T1 and higher than T3, where T3 is lower than the liquefaction temperature of the LNG; and - a third heat exchanger (103/7) in which the liquefied natural gas is cooled from T2 to T3; and 30 b) causing two streams of the refrigerant gas in the gaseous state at different pressures P1 and P2, referred to respectively as first and third streams, to circulate through two of said heat exchangers in indirect contact with and as a countercurrent relative to the natural gas 35 steam, comprising: - a first refrigerant gas stream at a pressure P1 lower than P3 passing through the three heat 9 exchangers by entering into said third heat exchanger at a temperature T3' lower than T3, then entering said second heat exchanger at a temperature T2' lower than T2, and then entering said first heat exchanger at T1' lower 5 than T1 and leaving said first heat exchanger at a temperature TO' lower than or equal to TO, said first refrigerant gas stream at P1 and T3' being obtained by using a first expander (112/9) to expand a first portion (122/16B) of a second refrigerant gas stream (22/15) 10 compressed to the pressure P3 higher than P2, said first portion of the second stream flowing in indirect contact with and as a countercurrent relative to the natural gas, entering said first heat exchanger at TO and leaving said second heat exchanger substantially at T2; and 15 - a third stream at a pressure P2 higher than P1 and lower than P3 flowing in indirect contact with and as a countercurrent relative to said first stream, passing solely through said second and first heat exchangers, by entering said second heat exchanger 20 substantially at a temperature T2' and leaving said first heat exchanger at TO', said third refrigerant gas stream at P2 and T2 being obtained by using a second expander (111/8) to expand a second portion (121/17) of said second refrigerant gas stream (22/15) leaving said first 25 heat exchanger substantially at T1; and c) said second refrigerant gas stream compressed at the pressure P3 being obtained by compression in three or four compressors and by cooling said first and second refrigerant gas streams leaving said first heat exchanger 30 respectively at Pl and at P2. In US 2011/0113825, first and second compressors 113 and 114 are connected in series to compress the refrigerant gas of the first and second streams to P'3, and two other compressors 115a and 115b connected in 35 parallel compress it from P'3 to P3. In WO 2005/071333, two series-connected compressors 2 and 3 compress said first stream 16d to P'3, and then a 10 third compressor 4 connected in series with the first two compressors compresses said first and third streams to P3. In the report on the "24th International Conference 5 and Exhibition for the LNG" of May 25, 2009, by Olve Skjeggedal et al. published in the GASTECH 2009 journal, a process of the above-described type having three closed-circuit refrigerant gas streams is described in which said first and second streams are compressed to P'3 10 by two compressors connected in series, and two other compressors connected in series compress said first and third streams to P3 in order to deliver said second stream. The process described above is advantageous compared 15 with that of Figure 1 in that, firstly, instead of a portion D2 of the second stream leaving the first heat exchanger by being expanded and recycled in order to join the first stream at the inlet to the second heat exchanger, this portion D2 of the second stream is 20 recycled to the inlet of the second heat exchanger at an intermediate pressure P2 higher than P1 in a third stream S3 independent of and parallel with Si, i.e. as a cocurrent relative to S1. And because the major portion of the energy is consumed by stage 2 of the process 25 within said second heat exchanger, this makes it possible to increase the transfer to heat and the energy efficiency of the process. Nevertheless, in the embodiment of US 2011/0113825, all of the external power delivered to said series 30 connected first and second compressors 113 and 114 relates to the refrigerant gas streams circulating at low and medium pressures P1 and P2, with the energy recovered from the turbines 111 and 112 being reinjected to the two parallel-connected compressors 115a and 115b for 35 compressing the refrigerant gas to high pressure P'3/P3, with no other additional external power being delivered to said parallel-connected compressors 115a and 115b.

11 The two parallel-connected compressors 115a and 115b are powered solely by respective ones of the two energy recovery turbines 111 and 112. The pressure levels P1 and P2 of the gas leaving the 5 turbines 112 and 111 are different and thus the flow rates of the streams passing through the expanders 111 and 112 are different, and in practice they lie in particular in the range 10% to 20% of the total flow rate for the flow rate of the stream coming from the expander 10 112, as compared with 80% to 90% for the flow rate of the stream coming from the expander 111. As a result, the compressor 115b recovers only 10% to 20% of the total recovered power compared with the 80% to 90% of the power that is recovered in the compressor 115a. This mismatch 15 in the powers delivered to the two parallel-connected compressors 115a and 115b leads to a major difficulty in stabilizing the operation of the circuit. Running two compressors in parallel can lead to surge phenomena, i.e. one of the compressors prevails over the others by 20 disturbing their inlet and outlet pressures: there is then a risk of one or more of the smaller-capacity compressors operating in "turbine mode". It is essential to avoid this mode of operation since some or all of the fluid then loops between the compressors, one operating 25 in compressor mode and the other(s) in "turbine mode": the compression process is then greatly disturbed or even interrupted, and the overall efficiency of the installation then collapses. The operation of the circuit can be stabilized in 30 conventional manner by means of regulation valves upstream and/or downstream from said parallel-connected compressors 115a and 115b, and/or upstream and/or downstream from said turbines 111 and 112 in order to control the flow rates and the operation of the 35 compressors. Nevertheless, those regulation valves lead to head losses, and thus to losses of energy, thereby 12 greatly affecting the expected overall efficiency and/or the production capacity of the installation. In WO 2005/071333 and in the report in the above mentioned GASTECH 2009 journal, all of the compressors 5 are mechanically coupled to a common power source, with all of the power being delivered in undifferentiated manner among the various compressors. A preferred embodiment of the present invention aims to provide a natural gas liquefaction process of the type 10 with no phase change in the refrigerant gas that is suitable for being installed on board a ship or a floating support and that presents improved energy efficiency, i.e. that minimizes the total energy consumed by the process in terms of kWh in order to obtain 1 tonne 15 of LNG, and/or that presents increased transfers of heat in the heat exchangers, and/or that makes it possible to implement a liquefaction installation that is more compact and more efficient. Any discussion of documents, acts, materials, 20 devices, articles or the like which has been included in the present specification is not to be taken as an admission that any or all of these matters form part of the prior art base or were common general knowledge in the field relevant to the present disclosure as it 25 existed before the priority date of each claim of this application. Throughout this specification the word "comprise", or variations such as "comprises" or "comprising", will be understood to imply the inclusion of a stated element, 30 integer or step, or group of elements, integers or steps, but not the exclusion of any other element, integer or step, or group of elements, integers or steps. To do this, the present invention provides a process for liquefying natural gas comprising a majority of 35 methane, preferably at least 85% methane, the other components essentially comprising nitrogen and C-2 to C-4 alkanes, wherein said natural gas for liquefying is 13 liquefied by causing said natural gas to flow at a pressure PO higher than or equal to atmospheric pressure (Patm), PO preferably being higher than atmospheric pressure, through that least one cryogenic heat exchanger 5 (EC1, EC2, EC3) by flowing in a closed circuit as a countercurrent in indirect contact with at least one stream of refrigerant gas that remains in the gaseous state and that is compressed to a pressure P1 entering said cryogenic heat exchanger at a temperature T3' lower 10 than T3, T3 being the temperature on leaving said cryogenic heat exchanger, and T3 being lower than or equal to the liquefaction temperature of said liquefied natural gas at atmospheric pressure, wherein said natural gas for liquefying is liquefied by performing the 15 following concurrent steps: a) causing said natural gas for liquefying to flow at a pressure PO higher than or equal to atmospheric pressure, PO preferably being higher than atmospheric pressure, through at least three cryogenic heat 20 exchangers connected in series and comprising: - a first heat exchanger in which said natural gas entering at a temperature TO is cooled and leaves at a temperature Ti lower than TO; then - a second heat exchanger in which the natural 25 gas is fully liquefied and leaves at a temperature T2 lower than Ti and higher than T3; and - a third heat exchanger in which said liquefied natural gas is cooled from T2 to T3; b) causing at least two streams of refrigerant gas 30 in the gaseous state and referred to respectively as the first and third streams to circulate in closed-circuits at different pressures P1 and P2 passing through at least two of said heat exchangers in indirect contact with and as a countercurrent relative to the natural gas stream 35 and comprising: - a first stream of refrigerant gas at a pressure P1 lower than P3 passing through the three heat 14 exchangers entering into said third heat exchanger at a temperature T3' lower than T3, then entering at T2' lower than T2 into said second heat exchanger, then entering at T1' lower than Ti into said first heat exchanger and 5 leaving said first heat exchanger at a temperature TO' lower than or equal to TO, said first stream of refrigerant gas at P1 and T3' being obtained by using a first expander to expand a first portion of a second stream of refrigerant gas compressed to the pressure P3 10 higher than P2, said second stream circulating in indirect contact with and as a cocurrent relative to said stream of natural gas by entering into said first heat exchanger at TO and said first portion of said second stream leaving said second heat exchanger substantially 15 at T2; and - a third stream at a pressure P2 higher than P1 and lower than P3 circulating in indirect contact with and as a cocurrent relative to said first stream, passing solely through said second and first heat exchangers, 20 entering into said second heat exchanger substantially at a temperature T2' and leaving said first heat exchanger substantially at TO', said third stream of refrigerant gas at P2 and T2 being obtained by using, a second expander to expand a second portion of said second stream 25 of refrigerant gas leaving said first heat exchanger substantially at Ti, the flow rate D2 of said second portion of the second stream being higher than the flow rate Dl of the first portion of the second stream; c) said second stream of refrigerant gas compressed 30 to the pressure P3 being obtained by using at least two compressors and by cooling, to compress said first and third streams of refrigerant gas leaving said first heat exchanger at P1 and P2 respectively, a first compressor compressing from P1 to P2 all of said first stream of 35 refrigerant gas leaving said first heat exchanger, and at least one second compressor compressing firstly said third stream of refrigerant gas leaving said first heat 15 exchanger at P2 and secondly said first stream of refrigerant gas compressed to P2 and leaving said first compressor, from P2 to at least P' 3 , where P' 3 is a pressure lower than or equal to P3 and higher than P2, 5 thereby obtaining said second stream of refrigerant gas at P3 and TO after cooling, said second compressor being connected in series with said first compressor; the process being characterized in that: - the series-connected first and second compressors 10 are coupled respectively to said first and second expanders consisting in energy-recovery turbines; and - at least said first compressor is coupled to a first motor; and - at least one gas turbine is coupled: 15 either to said second compressor, which compressor compresses said second stream of refrigerant gas directly to P3, or to a third compressor connected in series after the second compressor, said third compressor compressing 20 said second stream of refrigerant gas from P' 3 to P3, said gas turbine delivering the major portion of the total power delivered to all of said compressors in use, - said first motor delivering 3% to 30% of the total power delivered to all of said compressors in use, said 25 gas turbine supplying 97% to 70% of the total delivered power. In the present description, the terms "compressor coupled to an expander/turbine or motor" or indeed "compressor driven by a motor"! (or vice versa 30 "expander/turbine or motor coupled to the compressor") are used to mean that the outlet shaft from the turbine or the motor, as the case may be, drives the inlet shaft of the compressor, i.e. transfers mechanical energy to the shaft of the compressor. This is thus mechanical 35 coupling of the compressor to the expander/turbine or respectively of the compressor to the motor.

16 More particularly, said motor may either be a fuel burning engine, or else it is preferably an electric motor, or any other installation capable of delivering mechanical energy to the refrigerant gas; the compressors 5 are of the rotary turbine type, also known as centrifugal compressors. Preferably, after step a), the liquefied natural gas leaving said third heat exchanger at T3 is depressurized down from the pressure PO to atmospheric pressure, where 10 appropriate. The process is advantageous compared with the process described in US 2011/0113825 in that all of the compressors are connected in series without requiring flow rates to be controlled by flow rate regulator valves 15 in order to stabilize the operation of the installation. In the process of the invention, there is no separation of streams in the compression line. As a result, energy and/or stream flow rate is/are regulated in the various compressors essentially by 20 regulating the amount of power delivered by said first and second motors and said gas turbine. It is not essential to use regulator valves in association with said compressors and said turbine because said first and second expanders are coupled to said first and second 25 compressors that are connected in series and are therefore not coupled to compressors that are connected in parallel as in US 2011/0113825. Furthermore, the major fraction of the energy delivered to said compressors is injected via the second 30 and/or third compressors compressing the refrigerant gas stream to high pressure P'3/P3, and the energy recovered from the first and second expanders is reinjected via the first and second compressors serving to compress the refrigerant gas flowing at low and medium pressures P1 35 and P2. The fraction of the fluid passing through the compressor Cl is a small fraction of the total flow rate (e.g. 10% to 15%) and the energy needed is of the same 17 order of magnitude as the energy recovered by the turbine El. It is therefore advantageous to couple them together. Furthermore, controlled addition of power at Cl serves to improve the energy efficiency of the system 5 by controlling P1 and P2 independently of each other. Furthermore, the major portion of the power delivered to the compressors is injected into the compressors that supply the greatest pressure (P' 3 , P3), thereby making it possible to increase the production 10 capacity of the process, while improving its energy efficiency. In addition, using said first and second compressors in series and coupled to said first and second expanders in accordance with the present invention, thus makes it 15 possible to improve the compactness of the installation, which is particularly advantageous for performing the process on board a floating support where space is limited. The process as described with reference to Figures 2 20 and 3 is advantageous compared with that of Figure 1 in that firstly instead of expanding a portion D2 of the second stream at the outlet from the first heat exchanger and recycling it in order to join the first stream at the inlet to the second heat exchanger, this portion D2 of 25 the second stream is recycled to the inlet of the second heat exchanger at an intermediate pressure P2 that is higher than P1 in a third stream S3 that is independent of and parallel with Si, i.e. flowing as a cocurrent with S1. And because the major portion of the energy is 30 consumed in stage 2 of the process within said second heat exchanger, this makes it possible to increase the transfer of heat and to improve the energy efficiency of the process. Furthermore, the process is advantageous compared 35 with WO 2005/071333 and with the process described in the above-mentioned journal GASTECH 17A 2009, in that it makes it possible to vary said pressure P2 in controlled manner so that the energy (Ef) consumed for performing the process is minimized. Advantageously, it is possible to modulate and control specifically the 5 value of the pressure P2 by delivering different amounts of power to said first compressor by means of said first motor, thus making it possible to modulate and control the power delivered to the various compressors in different manners, and thus cause the value of P2 to 10 vary. In a preferred advantage, said pressure P2 is caused to vary in controlled manner by delivering power in controlled manner to said first compressor from said first motor, in such a manner that the energy consumed 15 for performing the 18 process (Ef) is minimized, preferably when the composition of the liquid gas for liquefying varies. This process is particularly advantageous since by modulating and controlling specifically the value of the 5 pressure P2 of said third stream, it thus makes it possible to modify and optimize the operating point of the process, i.e. to minimize energy consumption and thus increase efficiency, in particular when the composition of the natural gas for liquefying varies, as happens in 10 operation. More particularly, said first motor delivers at least 3%, and preferably 3% to 30%, of the total power delivered to all of said compressors in use, said gas turbine supplying 97% to 70% of the total delivered 15 power. Still more particularly, it is observed that when the power injected via said first motor is increased, the pressure P1 remains substantially constant, the pressure P2 increases, and efficiency increases, i.e. the energy 20 consumption expressed in kWxd/t decreases down to a minimum, after which any further increase in the power delivered by said motor, in particular to more than 30% of the total power, causes said energy consumption to increase once more. 25 A conventional liquefaction unit is dimensioned relative to the powers delivered by available gas turbines, with high power turbines currently delivering 25 megawatts (MW) or even 30 MW when they are for installation on board a floating support. Stationary gas 30 turbines installed on land may reach maximum powers in the range 90 MW to 100 MW. In general, it is desired to increase the power of the installation and it is then possible to install two identical gas turbines in parallel in order to obtain 35 twice the power, but there are then two rotary machine lines which increases overall bulk, increases the quantity of pipework, and naturally increases costs.

19 By installing a single gas turbine GT that delivers nMW and by adding power of lower than nMW via a said second motor M2, the operation of the process is identical in terms of efficiency to that obtained when 5 using two nMW gas turbines in parallel. Thus, adding power via the second motor M2, preferably using an electric motor, gives greater flexibility in operation and thus enables power to be increased. However overall efficiency remains unchanged. 10 In contrast, if the same power is delivered via a first motor M1, the overall power remaining the same, then the overall efficiency is improved, which represents a saving in energy consumption for the same overall power, compared with injecting power via the second motor 15 M2. Thus, as a function of the nature of the natural gas being produced from the underground reservoirs, both in terms of quantity and in terms of quality, it is advantageous to use a gas turbine GT, e.g. a 25 MW gas 20 turbine, continuously at full power with power being added, and where appropriate modulated, by: - injecting power prior to turbine GT or the second motor M2, without changing overall efficiency; and/or - by injecting power via the first motor M1, thereby 25 having the effect of improving overall efficiency, up to an optimum, i.e. a minimum of energy consumption. In a first variant of the process, two compressors are used that are connected in series, and that comprise: i) at least one first compressor, preferably a said 30 first compressor coupled to said first expander compressing from P1 to P2 all of said first stream of refrigerant gas leaving said first heat exchanger; and ii) at least one second compressor, preferably a said second compressor coupled to said second expander, 35 compressing firstly said third stream of refrigerant gas leaving said first heat exchanger at P2 and secondly said first stream of refrigerant gas compressed to P2 and 20 leaving said first compressor, from P2 to at least P'3, where P'3 is higher than P2 and lower than or equal to P3, in order to obtain said second stream of refrigerant gas at P3 and TO after cooling; and 5 iii) said first compressor being driven by a first motor, said second compressor being driven by at least one said gas turbine. This first variant implementation is advantageous in that it makes it possible to provide an installation that 10 is more compact in terms of the amount of space occupied on board the floating support. In a second variant implementation, use is made of three compressors connected in series, the compressors comprising: 15 i) a first compressor driven by a first motor and coupled to said first expander, compressing from P1 to P2 all of said first stream of refrigerant gas leaving said first heat exchanger; and ii) a second compressor driven by a second motor and 20 coupled to said second expander compressing firstly said third stream of refrigerant gas leaving said first heat exchanger at P2 and secondly said first stream of refrigerant gas compressed to P2 and leaving said first compressor from P2 to P'3, where P'3 is higher than P2 25 and lower than P3; and iii) a third compressor driven by a said gas turbine to supply the major portion of the energy and to compress from P'3 to P3 all of the first and third streams of refrigerant gas compressed by the second compressor in 30 order to obtain said second stream of refrigerant gas at P3 and TO after cooling; and iv) said first motor delivers at least 3%, and more preferably at least 3% to 30%, of the total power delivered to all of said compressors in use, the gas 35 turbine coupled to said third compressor and said second motor coupled to the second compressor together supplying 21 97% to 70% of the total power delivered to all of said compressors in use. This second variant implementation is advantageous in terms of thermodynamic efficiency and in terms of 5 production capacity since it is then advantageously possible to use a gas turbine having the maximum capacity that is available on the market, i.e. lying in the range 25 MW to 30 MW for gas turbines designed to be installed on board a floating support, together with a second 10 electric motor, e.g. having power of 5 MW to 10 MW that is connected to the second compressor, the total power available from the second motor plus the third motor (the gas turbine) then lying in the range 30 MW to 40 MW, and thus being considerably higher than the power available 15 from the largest gas turbine available on the market and suitable for use on board floating supports. Advantageously, the second motor may also be a gas turbine, preferably of power identical to the main gas turbine, thus making it possible to reach an overall 20 power level of 50 MW to 60 MW. By varying the pressure P2 by delivering energy to said first compressor via said first motor, the process of the invention makes it possible to use a minimum amount of total energy Ef consumed in the process that is 25 lower than 21.5 kWxd/t, and more particularly that lies in the range 18.5 kWxd/t to 20.5 kWxd/t of liquefied gas production. In general, a gas turbine GT will be operated at full power, and additional power will be delivered via 30 the first motor M1, said additional power delivery being limited to lower than 30% of the total power so as to optimize efficiency at the minimum value lying in the range 18.5 kWxd/t to 21.5 kWxd/t, and then where necessary, the overall power can be increased by 35 injecting power via the second motor M2, and concurrently the power injected via the first motor M1 should be readjusted so that said power is always substantially 22 equal to less than 30% of the overall power so as to conserve the efficiency of the installation at the optimum power in the range 18.5 kWxd/t to 21.5 kWxd/t. Said optimum efficiency of 19.75 kWxd/t can be 5 obtained for the first motor M1 delivering 24% of the total power when the refrigerant fluid is constituted by 100% nitrogen. When using other gases such as neon or hydrogen or nitrogen-neon or nitrogen-hydrogen mixtures, the power percentage and the optimum efficiency vary in 10 the range 18.5 kWxd/t to 21.5 kWxd/t depending on the gas or the mixture and on the percentages of neon or hydrogen, but the advantages specified above remain valid and can even be cumulative. More particularly, said refrigerant gas comprises 15 nitrogen. In a variant implementation, said refrigerant gas consists in a single gas selected from nitrogen, hydrogen, and neon. Neon is preferred because of the greater risk of 20 explosion with hydrogen and because hydrogen can present a certain propensity for percolating through elastomer gaskets and ,even through thin metal walls. According to other particular characteristics: - the composition of the natural gas for liquefying 25 lies in the following ranges for a total of 100%: - methane 80% to 100%; - nitrogen 0% to 20%; ethane 0% to 20%; propane 0% to 20; and 30 butane 0% to 20%; and using the following temperatures: - TO and TO' lie in the range 10'C to 35 0 C (temperature at AA); and - T3 and T3' lie in the range -160*C to -170*C 35 (temperature at DD); and - T2 and T2' lie in the range -100 0 C to -140*C (temperature at CC); and 23 - Ti and T1' lie in the range -30 0 C to -70 0 C (temperature at CC); with the following pressures: - PO lies in the range 0.5 MPa to 5 MPa (5 bars 5 to 50 bars); and - P1 lies in the range 0.5 MPa to 5 MPa; and - P2 lies in the range 1 MPa to 10 MPa (10 bars to 100 bars); and - P3 lies in the range 5 MPa to 20 MPa (50 bars 10 to 200 bars). The present invention also provides an installation on board a ship or a floating support for implementing a process wherein it comprises: - at least three said cryogenic heat exchangers in 15 series and comprising at least: - a countercurrent flow first duct suitable for causing a first stream of refrigerant gas in the gaseous state compressed to P1 to flow as a countercurrent successively through the third, second, and first heat 20 exchangers; - a cocurrent flow second duct suitable for causing a said second stream of refrigerant gas in the gaseous state compressed to P3 to flow as a cocurrent successively through only the said first and second heat 25 exchangers; - a countercurrent flow third duct for said refrigerant gas suitable for causing a said third stream of refrigerant gas in the gaseous state compressed to P2 to flow as a countercurrent successively through only 30 said second and first heat exchangers; - a fourth duct suitable for causing said natural gas for liquefying to flow successively through the first, second, and third heat exchangers; - a first expander between the outlet from said 35 second duct and the inlet to said first duct; - a second expander between i) a branch connection to said second duct situated between said 24 first and second heat exchangers, and ii) the inlet of said third duct; and - a first compressor at the outlet from said first duct, coupled to a turbine constituting said first 5 expander; - a second compressor at the outlet from said second duct, coupled to a turbine constituting said second expander, and said second compressor being connected in series with said first compressor, in 10 particular at the outlet from said first compressor; and - a duct for passing all of the gas compressed to P2 by the first compressor to the second compressor connected in series in this way with said first compressor; and 15 - at least a first motor coupled to said compressor and suitable for delivering at least 3%, of the total power delivered to all of said compressors in use. Still more particularly, a said installation 20 comprises: only at least two compressors connected in series and comprising: i) at least one said first compressor coupled to said first expander, suitable for compressing from P1 to 25 P2 all of said first stream of refrigerant gas leaving said first heat exchanger; and ii) at least a second compressor coupled to said second expander, suitable for compressing from P2 to P3 firstly said third stream of refrigerant gas leaving said 30 first heat exchanger at P2 and secondly said first stream of refrigerant gas compressed to P2 and leaving said first compressor, in order to obtain said second stream of refrigerant gas at P3 and TO after cooling; and iii) a said first motor coupled to a said first 35 compressor, and a gas turbine coupled to a second compressor, said first motor being suitable for delivering at least 3%, more preferably 3% to 30%, of the 25 total power delivered to all of said compressors in use; and iv) said gas turbine coupled to said second compressor being suitable for supplying 97% to 70% of the 5 total delivered power. Still more particularly, invention preferred embodiment comprises: only three compressors connected in series and comprising: 10 i) a said first compressor coupled to said first expander and to a said first motor; and ii) a said second compressor coupled to said second expander and to a said second motor; and iii) a third compressor coupled to a gas turbine 15 suitable for supplying the major portion of the energy and suitable for compressing to P3 all of the first and third streams of refrigerant gas compressed by said second compressor in order to obtain said third stream of refrigerant gas at P3 and TO after cooling; and 20 iv) said first motor being suitable for delivering at least 3%, more preferably 3% to 30%, of the total power delivered to all of said compressors in use, the gas turbine coupled to said third compressor and said second motor coupled to the second compressor being 25 suitable together for supplying 97% to 70% of the total power delivered to all of said compressors in use. Other characteristics and advantages appear in the light of the following detailed description of embodiments given with reference to the accompanying 30 figures, in which: - Figure 1 is a diagram showing a standard liquefaction process with a double loop using nitrogen as the refrigerant gas; - Figure 2 is a diagram of a liquefaction process of 35 the invention with a triple loop using nitrogen, or a 26 mixture including nitrogen, as the refrigerant gas, in a version that is referred to as "balanced"; - Figure 3 is a diagram of a liquefaction process of the invention with a triple loop using nitrogen, or a 5 mixture including nitrogen, as the refrigerant gas, in a version referred to as "compact"; - Figure 4 is a cooling and liquefaction diagram for natural gas in the context of a liquefaction process of the invention, plotting the enthalpy of the natural gas 10 and of the refrigerant fluid in kilojoules per kilogram (kJ/kg) as a function of temperature from TO to T3; - Figures 5 and 5A are graphs plotting total energy consumption (Ef) in kilowatt days per tonne of LNG produced (kWxd/t) for a liquefaction process of the 15 invention using a mixture of nitrogen and neon as the refrigerant gas, as a function of the pressure P1 and of various percentages of neon in said mixture; - Figures 5 and 5B are graphs plotting total energy consumption (Ef) kWxd/t of LNG produced for a 20 liquefaction process of the invention using a mixture of nitrogen and hydrogen as the refrigerant gas, as a function of the pressure P1 and of various percentages of hydrogen in said mixture; - Figure 6A is a graph plotting the total energy 25 consumed (Ef) in kWxd/t of LNG produced by a liquefaction process of the invention using a mixture of nitrogen and neon as the refrigerant gas, as a function of the pressure P2 and of various percentages of neon in said mixture; 30 - Figure 6B is a graph plotting the total energy consumed (Ef) in kWxd/t of LNG produced by a liquefaction process of the invention using a mixture of nitrogen and hydrogen as the refrigerant gas, as a function of the pressure P2 and of various percentages of hydrogen in 35 said mixture; ' Figure 7 is a graph plotting the total energy consumed (Ef) in kWxd/t of LNG produced in a liquefaction 27 process of the prior art (60) and a liquefaction process of the invention, using nitrogen as the refrigerant gas and depending on the level of the pressure P3; - Figure 7A is a graph plotting the total energy 5 consumed (Ef) in kWxd/t of LNG produced by a liquefaction process of the invention using a mixture of nitrogen and neon as the refrigerant gas, as a function of the pressure P3 and of various percentages of neon in said mixture; and 10 - Figure 7B is a graph plotting the total energy consumed (Ef) in kWxd/t of LNG produced by a liquefaction process of the invention using a mixture of nitrogen and hydrogen as the refrigerant gas, as a function of the pressure P3 and of various percentages of hydrogen in 15 said mixture. Figure 1 is a process flow diagram (PFD) for the standard double loop process without phase change using nitrogen as the refrigerant gas. The process uses compressors C1, C2, and C3, expanders El and E2, 20 intermediate coolers H1 and H2, and cryogenic heat exchangers EC1, EC2, and EC3. In known manner, the heat exchangers are constituted by at least two circuits that are juxtaposed but that do not communicate with each other in terms of said fluids, the fluids flowing in said 25 circuits exchanging heat all along their paths within such a heat exchanger. Numerous types of heat exchanger have been developed in various industries and in the context of cryogenic heat exchangers, two main types predominate: firstly coiled heat exchangers; and secondly 30 brazed aluminum plate heat exchangers, known as "cold box" heat exchangers. Heat exchangers of this type are known to the person skilled in the art and they are sold by the suppliers Linde (France) or Five Cyrog6nie (France). Thus, all of 35 the circuits of a cryogenic heat exchanger are in thermal contact with one another in order to exchange heat, but the fluids that flow through them do not mix. Each of 28 the circuits is dimensioned so as to minimize head losses at the maximum flow rate of the refrigerant fluid and so as to present sufficient strength to withstand the pressure of said refrigerant fluid as it exists in the 5 loop in question. In conventional manner, an expander causes the pressure of a fluid or a gas to drop and is represented by a symmetrical trapezoid, its small base representing its inlet 10a (high pressure) and its large base 10 representing its outlet 10b (low pressure) as shown in Figure 1 with reference to the expander E2, said expander possibly being merely a reduction in the diameter of the pipe, or else being merely an adjustable valve, but in the context of the liquefaction process of the invention 15 the expander is generally a turbine that serves to recover mechanical energy during said expansion, so that that energy is not lost. In the same way, and in conventional manner, a compressor increases the pressure of a gas and is 20 represented by a symmetrical trapezoid, with its large base representing the (low pressure) inlet 11a and its small base representing the (high pressure) outlet lb, as shown in Figure 1 with reference to the compressor C2, said compressor generally being a turbine or a piston 25 compressor, or indeed a spiral compressor. The compressors Cl and C2 (Figures 2 and 3) are preferably mechanically connected to respective motors Ml and M2 that may be electric motors or fuel-burning engines, or any other installation capable of delivering mechanical 30 energy. Natural gas flows in the circuit Sg and enters at AA into the first cryogenic heat exchanger EC1 at a temperature TO, higher than or substantially equal to ambient temperature, and exits at T1=-50'C approximately. 35 In this heat exchanger EC1, the natural gas is cooled but it remains in the gaseous state. At BB it passes into the cryogenic heat exchanger EC2 in which temperature 29 extends over the range Tl=-50*C approximately, to T2=-120 0 C approximately. In this heat exchanger EC2, all of the natural gas liquefies into LNG at a temperature of T2=-120*C, 5 approximately, and then the LNG passes at CC into the cryogenic heat exchanger EC3. In this heat exchanger EC3, the LNG is cooled down to the temperature T3=-165*C, which enables the LNG to be discharged from the bottom portion at DD, and then to be depressurized at EE so that 10 the liquid can finally be stored at ambient atmospheric pressure, i.e. at an absolute pressure of 1 bar (i.e. about 0.1 megapascals (MPa)). All along this path of the natural gas along the circuit Sg in the various heat exchangers, the natural gas cools, transferring its heat 15 to the refrigerant gas, which then heats up and needs to be continuously subjected to a complete thermodynamic cycle for the purpose of continuously extracting the heat in the natural gas entering at AA. Thus, the path of the natural gas is shown on the 20 left of the PFD, and said gas flows downwards along the circuit Sg, its temperature decreasing going downwards from a temperature TO that is substantially ambient at the top at AA, down to a temperature T3 of about -165*C at the bottom at DD. 25 The right-hand portion of the PFD shows the double loop thermodynamic cycle of the refrigerant gas corresponding to circuits S1 and S2. To clarify explanation, the pressure levels in the main circuits are represented by fine lines for low pressure (F1 in the 30 circuit S1), by medium lines for intermediate pressure (P2), and by bold lines for high pressure (P3 in the circuit S2). In a conventional circuit as shown in Figure 1, the stages 1, 2, and 3 are performed by a low pressure loop 35 P1 at very low temperature at the bottom inlet of EC3. The installation is made up of: 30 - a motor, generally a gas turbine GT that drives the compressor C3 and that delivers all of the mechanical power; - three compressors: 5 C3, which compresses all of the refrigerant stream; C2, which is coupled to the turbine E2 and which compresses the fraction D'2 of the total stream D; and 10 Cl, which is coupled to the turbine El and which compresses the complementary fraction D'l of the total stream D; - two turbines: - E2 directly coupled to the compressor C2 and 15 serving to expand the fraction D2 of the total stream D from the high pressure P3 down to the low pressure P1; - El directly coupled to the compressor Cl and serving to expand the fraction Dl of the total stream D, from the high pressure P3 to the low pressure Pl; 20 - a three-portion cryogenic heat exchanger or three heat exchangers in series ECl, EC2, and EC3, corresponding respectively to liquefaction stages 1, 2, and 3, and having three circuits, respectively a natural gas circuit SG and refrigerant gas circuits Sl and S2; 25 and - at least two coolers Hl and H2 situated respectively at the outlet from the main compressor C3 (H1) and on the high pressure loop (H2) before the inlet to the cryogenic heat exchangers. 30 A cooler Hi, H2 may be constituted by a water heat exchanger, e.g. a heat exchanger using sea or river water or using cold air, the heat exchanger being of the fan coil or cooling tower type, such as those used in nuclear power stations. 35 More precisely, Figure 1 shows the circuit for a process and an installation in which said natural gas for 31 liquefying is liquefied by performing the following concurrent steps: a) causing said natural gas for liquefying to flow Sg at a pressure PO higher than or equal to atmospheric 5 pressure (Patm), with PO preferably being higher than atmospheric pressure, the gas flowing through the three cryogenic heat exchangers EC1, EC2, EC3 arranged in series and comprising: - a first heat exchanger EC1 into which said 10 natural gas enters at a temperature TO, is cooled, and leaves at BB at a temperature Ti lower than TO, but at which all of the components of said natural gas are still in the gaseous state; then - a second heat exchanger EC2 in which the 15 natural gas is liquefied in full and leaves at CC at a temperature T2 lower than Ti; and . a third heat exchanger EC3 in which said liquefied natural gas is cooled from T2 to T3, where T3 is lower than T2 and T3 is lower than or equal to the 20 liquefaction temperature of said natural gas at atmospheric pressure; and b) causing a first stream Sl of refrigerant gas in the gaseous state and compressed to a pressure P1 lower than P3 to flow in a closed circuit in indirect contact 25 with and as a countercurrent to the natural gas stream Sg, said first stream Si at a pressure P1 passing through the three heat exchangers EC3, EC2, and EC1, entering at DD into said third heat exchanger EC3 at a temperature T3' lower than T3 and then leaving said third heat 30 exchanger and entering said second heat exchanger EC2 at CC at a temperature T2' lower than T2, and then leaving the second heat exchanger and entering the first heat exchanger EC1 at BB at a temperature Tl' lower than Ti, and leaving said first heat exchanger EC1 at AA at a 35 temperature TO' lower than or equal to TO; - said first stream Si of refrigerant gas at P1 and T3' being obtained by using in a first expander El to 32 expand a first portion Dl of a second stream S2 of refrigerant gas compressed to P3 higher than Pl flowing as a countercurrent to said natural gas entering said first heat exchanger EC1 at AA and at PO, and leaving 5 said second heat exchanger EC2 at CC and substantially at T2; and - a second portion D2 of said second stream S2 of refrigerant gas compressed to P3 flowing as a countercurrent to said natural gas entering said first 10 heat exchanger EC1 at AA and at TO and leaving said first heat exchanger substantially at Tl is expanded in a second expander E2 to said pressure P1 and to a said temperature T2', and is recycled to join said first stream at the inlet at CC of said second heat exchanger; 15 and c) said second stream S2 compressed to P3 is obtained by compression using three compressors Cl, C2, and C3 followed by at least two coolers Hl and H2 acting on said first stream Sl of recycled coolant gas leaving 20 said first heat exchanger ECl at AA via a first compressor C1 coupled to said first expander El; and d) after step a), the liquefied natural gas is depressurized from the pressure PO to atmospheric pressure. 25 More precisely, in Figure 1, three compressors are used including first and second compressors that are connected in parallel, the three compressors comprising: - a third compressor C3 driven by a motor, preferably a gas turbine GT, to compress all of the first 30 stream of refrigerant gas coming from the outlet at AA from said first heat exchanger EC1 from P1 to P'3, where P'3 lies in the range P1 to P3; and - a first compressor C1 coupled to the first expander El, which is constituted by a turbine, to 35 compress from P2 to P'3 a portion Dl' of said first stream of refrigerant gas as compressed by the third compressor C3; and 33 a second compressor C2 coupled to the second expander E2, which is a turbine, to compress from P'3 to P3 a portion D2' of said first stream of refrigerant gas as compressed by the third compressor C3. 5 In Figure 1, C1 and C2 are thus connected in parallel and they operate between the medium pressure P'3 and the high pressure P3 on all of the stream coming from c3. The refrigerant gas leaving the heat exchanger EC1 10 at the high outlet at AA from the circuit Sl has a flow rate D: it is at low pressure P1 and at a temperature T'O that is perceptibly lower than TO and at ambient temperature. It is then compressed in C3 to the pressure P'3, after which it passes through a cooler H1. The 15 fluid at flow rate D is then split into two portions presenting flow rates DlT and D2' that are fed respectively to the compressors C1 (Dl') and C2 (D2') that are operating in parallel. The two streams at the pressure P3 are then reunited and then cooled 20 substantially to ambient temperature TO by passing through the cooler H2. This total flow then enters into the top of the cryogenic heat exchanger EC1 via the circuit S2, and then at the outlet from the first level at BB, a large portion of the stream at flow rate D2 (D2 25 greater than D1) is extracted and directed to the turbine E2 coupled to the compressor C2. The remainder of the flow D1 passes through the second stage of the cryogenic heat exchanger EC2, and then at CC it is directed to the turbine El coupled to the compressor Cl. 30 At the outlet from the turbine E1, the refrigerant gas at a temperature T3' lower than T3=-165 0 C, is then directed downwards from the cryogenic heat exchanger EC3 into the circuit Sl and rises as a countercurrent to the gas for liquefying that is flowing along the circuit Sg, 35 thereby performing the final stage 3 of the liquefaction. The flow D2 of refrigerant gas coming from the turbine E2 is at a pressure Pl and at a temperature T2 of 34 about -120*C and it is recombined within the circuit S1 with the flow Dl coming from the turbine El via the top outlet from the cryogenic heat exchanger EC3 at CC. The separation of the second stream S2 into two 5 portions having different flow rates D1 and D2 at the outlet BB from the first heat exchanger, preferably with D2 greater than D1, is advantageous since most of the energy consumption takes place during stage 2 within the second heat exchanger EC2. Thus, only a minor portion of 10 the flow rate D1 passes through the third heat exchanger EC3 where stage 3 takes place, while the total flow D=Dl+D2 of the circuit Sl passes through the cryogenic heat exchanger EC2 in order to perform liquefaction stage 2 (from temperature Tl=-50*C to T2=-120*C). 15 The same flow D of the circuit S1 finally passes through the cryogenic heat exchanger EC1 in order to perform stage 1 of the liquefaction process (from temperature Tl=-50*C to temperature TO=ambient temperature). At the top outlet from the cryogenic heat 20 exchanger EC1, the flow D of the circuit S1 is at the temperature TO' that is perceptibly lower than ambient temperature. Thereafter, the flow D is once more directed to the compressor C3 in order to perform a new cycle in continuous manner. 25 In this configuration, the compressors Cl and C2 run in parallel and they need to provide the highest pressure level in the cycle. The two compressors Cl and C2 handle different flow rates of refrigerant fluid, respectively D1' and D2', and they are directly coupled to the 30 turbines El and E2, which likewise handle different flow rates, respectively D1 and D2. The following relationship applies: Dl + D2 = D = D'1 + D'2 where Dl is different from D'l and D2 is different from 35 D'2. In practice, and preferably D1/D = 5% to 35%, and preferably 10% to 25%.

35 Thus, in that type of installation, all of the power is injected into the system via the compressor C3 (by the gas turbine GT), with the power transfers via the turbine and compressor pairs E2-C2 and El-Cl varying as a 5 function of the pressures in the various circuits (P1, P2, P3), as a function of the temperature levels at the inlets to the cryogenic heat exchangers, and as a function of the heat transfers within each of said cryogenic heat exchangers. 10 Thus, such an installation presents an operating point that self-stabilizes at a given level of energy consumption Ef which is generally expressed in terms of kWxd/t, i.e. kW-days per tonne of LNG produced, or indeed kWh per kg of LNG produced, said operating point possibly 15 being totally unstable in certain circumstances. It is then very difficult to control the pressures in the high and low pressure loops independently of each other. This may be found to be necessary in the event of variations in the composition of the natural gas for liquefying. It 20 is possible to modify the streams by locally constraining the flows D1, D1l, D2, D'2 in full or in part, e.g. by creating localized head losses, however such arrangements lead to losses of energy and thus to a drop in the overall efficiency of the liquefaction installation. 25 The graph of Figure 4 shows how enthalpy H expressed in kJ/kg of LNG production varies in a natural gas liquefaction process. This graph of Figure 4 is the result of theoretical calculation relating to a natural gas having a majority of methane (85%), with the balance 30 (15%) being made up of nitrogen, ethane (C-2), propane (C-3), and butane (C-4). The graph shows: - stage 1 of cooling the natural gas between points AA and BB and corresponding to the heat exchanger EC1 in 35 the PFD of Figure 1, corresponding to temperatures lying between ambient temperature TO and Tl=-50*C; 36 stage 2 of natural gas liquefaction between points BB and CC, corresponding to the stage EC2 of the PFD of Figure 1, and corresponding to temperatures lying in the range Tl=-50 0 C to T2=-120C; and 5 - stage 3 of cooling the LNG between points CC and DD, corresponding to the heat exchanger EC3 of the PFD of Figure 1, and corresponding to temperatures lying in the range T2=-120*C to T3=-165*C. Curve 50 made up of triangles shows the variations 10 in the enthalpy H of the fluids flowing as cocurrents in the circuits Sg and S2 as a function of the temperature of the gas for liquefying comprising methane and/or LNG for an ideal virtual process. The curve 51 corresponds to the variation in the 15 enthalpy H of the refrigerant gas flowing in the circuit Sl of Figure 1, and thus represents the energy transferred to the circuits Sg and S2 during the liquefaction process. The area 52 lying between the two curves 50 and 51 20 represents the overall loss of the energy Ef consumed in the liquefaction process: this area should therefor be minimized in order to obtain the best efficiency. In land-based processes involving a change of phase in the refrigerant fluid, the curve 51 is not straight, but 25 rather comes close to the theoretical curve 50, thereby implying smaller losses, and thus better efficiency, but the process with a change of phase in the refrigerant fluid is not suitable for use in liquefaction on board a floating support and in an environment that is confined. 30 Figures 2 and 3 are PFD diagrams for the improved process of the invention in which the path followed by the natural gas for liquefying, having a majority of methane and traces of other gases, is identical to that of Figure 1 and takes place in the same manner within the 35 circuit Sg, going from the top (temperature TO substantially ambient temperature) towards the bottom 37 (liquid state at T3=-165 0 C), via three cryogenic heat exchangers EC1, EC2, and EC3. In Figures 2 and 3, instead of expanding a portion D2 of the second stream at the outlet from the first heat 5 exchanger and recycling it so that it rejoins the first stream at the low inlet CC of the second heat exchanger, as shown in Figure 1, this portion D2 of the second stream is recycled to the inlet CC of the second heat exchanger, and this done at an intermediate pressure P2 10 that is higher than P1 in a third circuit S3 that is independent of S1, S2, and Sg, and that is parallel to S1, i.e. in which the flow is a cocurrent relative to S1. Because the major portion of the energy is consumed in stage 2 of the process within said second heat 15 exchanger, this makes it possible to further increase the transfers of heat and the overall energy efficiency of the process. However, and more importantly, this also makes it possible to modulate and control specifically the value of the pressure P2 by connecting the two 20 compressors Cl and C2 in series and by coupling C1 with a motor MI serving to modulate and control the additional power delivered to C1, which is already coupled to the turbine El, thus making it possible to control the pressure value P2 as described below. 25 More precisely, Figures 2 and 3 show the process and the installation in which said natural gas for liquefying is liquefied by performing the following concurrent steps: a) causing said natural gas for liquefying to flow 30 Sg at a pressure PO higher than or equal to atmospheric pressure (Patm), PG being higher than atmospheric pressure, through three cryogenic heat exchangers EC1, EC2, EC3 connected in series and comprising: - a first heat exchanger EC1 in which said 35 natural gas entering at a temperature TO is cooled and leaves at BB at a temperature Ti lower than TO, the temperature Tl being a temperature at which all of the 38 components of the natural gas are still in the gaseous state; then - a second heat exchanger EC2 in which the natural gas is liquefied in full and leaves at CC at a 5 temperature T2 lower than Tl; and - a third heat exchanger EC3 in which said liquefied natural gas is cooled from T2 to T3, T3 being lower than T2, and T3 being lower than the liquefaction temperature of said natural gas at atmospheric pressure; 10 and b) causing refrigerant gas in the gaseous state to flow in a closed circuit in two streams SI and S3, referred to respectively as the first and third streams, having respective different pressures P1 (SI) and P2 15 (S2), the streams passing through two of said heat exchangers in indirect contact with and as a countercurrent to the natural gas stream Sg, the streams comprising: - a first refrigerant gas stream Sl at a 20 pressure P1 lower than P3 passing through all three heat exchangers ECI, EC2, and EC3, by entering said third heat exchanger EC3 at DD at a temperature T3' lower than T3, and then leaving said third heat exchanger and entering said second heat exchanger EC2 at CC at a temperature T2' 25 lower than T2, then leaving the second heat exchanger and entering the first heat exchanger EC1 at BB at a temperature Tl' lower than TI, and leaving said first heat exchanger at AA at a temperature TO' lower than TO, said first refrigerant gas stream at P1 and T3' being 30 obtained by using a first expander El to expand a first portion D1 of a second refrigerant gas stream S2 compressed to the pressure P3 higher than P2, said second stream S2 flowing in indirect contact with and as a cocurrent to said natural gas stream Sg entering said 35 first heat exchanger EC1 at AA and substantially at PO, and leaving said second heat exchanger EC2 at CC and substantially at the temperature T2; and 39 a third stream S3 at a pressure P2 higher than P1 and lower than P3 flowing in indirect contact with and as a cocurrent to said first stream, passing solely through said second and first heat exchangers EC2 5 and ECl, entering said second heat exchanger at CC substantially at a temperature T2' lower than T2 and leaving said first heat exchanger EC1 at AA substantially at a temperature TO', said third stream S3 of refrigerant gas at P2 and T2 being obtained by using a second 10 expander E3 to expand a portion D2 of said second stream D2 of refrigerant gas leaving said first heat exchanger substantially at T1; c) said second stream of refrigerant gas S2 compressed to the pressure P3 being obtained by 15 compressing said first and third refrigerant gas streams leaving said first heat exchanger EC1 at AA and respectively at P1 and P2 by means of first and second compressors respectively C1 and C2 connected in series and coupled respectively to said first and second 20 expanders El and E2, which are constituted by turbines; and d) after step a), the liquefied natural gas leaving said third heat exchanger at DD and at T3 is depressurized from the pressure PO to atmospheric 25 pressure, where appropriate. More precisely, in Figure 2, use is made of: 1) three compressors Cl, C2, and C3 connected in series and comprising: i) a first compressor C1 coupled to said first 30 expander El, and compressing from P1 to P2 all of said first refrigerant gas stream leaving said first heat exchanger EC1 at AA; ii) a second compressor C2 coupled to said second expander E2 and compressing firstly said third 35 refrigerant gas stream S3 leaving said first heat exchanger EC1 at P2 and secondly said first refrigerant gas stream compressed to 22 and leaving said first heat 40 exchanger EC1 from the pressure P2 to P'3, where P'3 is higher than P2 and lower than or equal to P3; and iii) a third compressor C3 driven by a gas turbine GT to deliver the major portion of the energy and to 5 compress from P'3 to P3 all of the first and third refrigerant gas streams compressed by the second compressor C2 so as to obtain said second refrigerant gas stream at P3 and TO after cooling (Hl, H2); and 2) said first compressor C1 is coupled to a first 10 motor M1 serving to vary the pressure P2 in controlled manner by delivering power in controlled manner to said first compressor C1, said first motor M1 delivering at least 3%, and preferably 3% to 30%, of the total power delivered to all of said compressors C1, C2, and C3 that 15 are in use, the gas turbine GT coupled to said third compressor C3 and the second motor M2 coupled to the second compressor C2 together delivering 97% to 70% of the total power delivered to all of said compressors Cl, C2, and C3 that are in use. 20 The installation of Figure 2 is thus made up of: - a plurality of motors, generally a gas turbine GT that drives the compressor C3, and motors Ml and M2, e.g. electric motors or fuel-burning engines, such as gas turbines, that are connected respectively to the 25 compressors Cl and C2; - three compressors: - C3, which compresses all of the refrigerant gas flow D; - C2, which is coupled to the motor M2 and to 30 the turbine E2 and which compresses all of the refrigerant gas flow D; and - C1, which is coupled to the motor Ml and to the turbine El, and which compresses the fraction Dl of the first refrigerant gas stream; 35 two expanders, e.g. turbines: - E2 coupled to the compressor C2 and to the motor M2; and 41 El coupled to the compressor Ci and to the motor M1; a cryogenic heat exchanger made up of three portions or comprising three heat exchangers in series 5 ECl, EC2, and EC3, corresponding respectively to the stages 1, 2, and 3 of the liquefaction and having four circuits, respectively Sg (natural gas) and Si, S2, and S3 (refrigerant gas); - two coolers H1 and H2 situated respectively 10 at the outlet from the main compressor C3 (H1) before the inlet to the circuit S2 of the cryogenic heat exchangers, and on the high pressure loop (H2). The compressors C1 and C2 are connected in series: - C1 operates between the low pressure P1 and the 15 medium pressure P2 on the fraction D1 of the refrigerant gas stream coming from the turbine El and flowing in the circuit Sl upwards through each of the three cryogenic heat exchangers EC3, EC2, ECl; and - C2 operating between the medium pressure P2 and 20 the high intermediate pressure P'3 on all of the flow D, made up of the fraction D1 of the stream coming from the compressor Cl and of the fraction D2 of the refrigerant gas stream coming from the turbine E2 flowing in the circuit S3 upwards through two of the cryogenic heat 25 exchangers EC2 and EC1. The entire refrigerant gas flow D leaving the compressor C2 is cooled in a cooler Hi prior to returning to the pressure P'3 in the compressor C3, which compressor is connected to a motor (GT) generally a gas 30 turbine. Said gas turbine and the motor (M2) together delivering 70% to 97% of the total power Q to the refrigerant gas, with the balance of the power being delivered to the system via the motor Ml, i.e. 30% to 3% of the total power Q. 35 At the outlet from the compressor C3, all of the refrigerant gas flow D is at the high pressure P3. The stream is then cooled in a cooler H2 prior to flowing in 42 the circuit S2 downwards through two of the cryogenic heat exchangers EC1 and EC2. The fraction D2 of the refrigerant gas stream is taken at BB from the outlet from the cryogenic heat 5 exchanger EC1 and is directed to the inlet of the turbine E2, the balance, i.e. the fraction Dl of the refrigerant gas stream being taken at CC from the outlet from the cryogenic heat exchanger EC2 and being directed to the inlet of the turbine El. 10 A cooler H2 operating at a pressure P'3 is installed within the compressor C3 between two compression stages, said cooler H2 handling all of the flow D. In this process of the invention, the following relationships apply: 15 Dl + D2 = D and preferably D1/D2 = 1/3 to 1/20, and more preferably 1/4 to 1/10. The main advantage of the device of the invention as shown in Figure 2 lies in the possibility of optimizing 20 the overall efficiency of installations and of modifying at will the operating points of the various loops corresponding to the circuits Sl, S2, S3, i.e. minimizing energy consumption by increasing or decreasing the power injected into any one of the compressors Cl, C2, C3, or 25 by varying the distribution of the overall power Q injected into the system. These adjustments of the amount of power injected into the various compressors Cl, C2, C3 have the effect of modifying the flow rates in the various loops, and thus of modifying the pressures P1, 30 P2, and P3 and also the mass flow rates D, Dl, and D2 in the various circuits Sl, S2, S3, thereby giving great flexibility in optimizing the operating point of the installation and thus great facility and great speed when readjusting the process as a result of fluctuations in 35 the composition of the natural gas for liquefying coming from underground reservoirs. These fluctuations can be large during the lifetime of the gas production field, 43 which lifetime may be as long as 20 years to 30 years, or even more. Thus, in the graph of Figure 4 relating to a natural gas comprising 85% methane, with the balance being made 5 up of nitrogen, ethane (C-2), propane (C-3), and butane (C-4), the curve 50 made up of triangles shows variations in enthalpy H of the fluids flowing in the circuits Sg and S2 of Figure 2 as a function of the temperature of the natural gas and/or LNG for an ideal virtual process. 10 The curve 53 corresponds to the variation in the enthalpy H of the refrigerant fluid flowing in the circuits S1 and S3 of Figure 2, and it thus represents the energy that is transferred during the liquefaction process to the circuits Sg and S2 of Figure 2. 15 The area 52 lying between the two curves 50 and 53 represents the overall energy loss in the liquefaction process with reference to Figure 2: it is therefore appropriate to seek to minimize this area in order to obtain the best efficiency. 20 During variations over time in the quantity of natural gas delivered by the gas field, and thus of its composition, the low point 54 of the curve 50 corresponding to PO and T2 at the end of liquefying the LNG may vary by a few percent. In the conventional 25 process of Figure 1, the corresponding point 55 of the refrigerant gas circuit remains substantially stationary, so the area 52 and thus the efficiency of the installation cannot be optimized. In contrast, in the device of the invention as shown 30 in Figure 2, by acting on the distribution of mechanical energy, and in particular on the energy that is injected via GT, Ml, and M2, and more particularly M1, it is advantageously possible to vary the position of the point 56, which point is thus caused to move in optimum manner 35 closer to the point 54, thereby enabling the area 52 lying between the curves 50 and 53 to be minimized, and as a result acting in real time to optimize the 44 efficiency of the liquefaction installation, as a function of the composition of the natural gas. Figure 3 is the PFD diagram of a version of the invention that is more compact than the process and the 5 installation of Figure 2, in which the compressor C2 is incorporated on the same shaft line as the compressor C3 and is driven by the gas turbine GT that contributes 85% to 95% of the total energy Q in the form of mechanical energy. In this configuration, the expansion turbine E2 10 is connected firstly to the compressor C2 and secondly to the gas turbine GT. In this version of Figure 3 that is more compact than the version described with reference to Figure 2, there is nevertheless reduced latitude for adjusting the 15 operating points of the various loops, since power adjustments can then only be made via the motor GT connected to C3 and the motor M1 connected to C1. Thus, this compact version is advantageously preferred when the available area to the installation is very limited, and 20 in addition it has only two rotary machine shaft lines and two compressors, whereas in the version described with reference to Figure 2, it is necessary to install three rotary machine shaft lines and three compressors, which represents non-negligible extra cost, even though 25 it provides greater flexibility in fine adjustment of the various pressure loops, and also better final efficiency, and thus better profitability for the installation over the long term, throughout the total lifetime of the installation, which may exceed 20 years to 30 years, or 30 even more. Figures 5 to 9 as described below reproduce the results of tests in which the values of P1, P2, and P3 were modified in order to minimize the total energy consumption Ef expressed in kWxd/t as a function of 35 variations in the composition of the refrigerant gas. Figures 5, 5A, and 5B are energy efficiency diagrams, and more precisely they show Ef expressed in 45 kWxd/t as a function of the pressure P1 and as a function of various variants of the invention. This pressure Pl is constant for a given refrigerant gas composition, which explains that all the points on any one curve lie 5 on a straight line parallel to the ordinate axis. This pressure P1 corresponds to the lowest temperature T3' of the device, i.e. to the temperature at the low inlet to the cryogenic heat exchanger EC3. This pressure P1 corresponds to the dew point of the refrigerant gas at a 10 temperature T3' substantially lower than T3=-165*C, i.e. the temperature at which the LNG remains liquid under a pressure corresponding to atmospheric pressure, i.e. substantially 0.1 MPa absolute, i.e. substantially one atmosphere. 15 In Figures 5, 5A, and 5B, it can be seen that by mixing the nitrogen with neon or with hydrogen, up to a molar proportion of 50%, it is possible to increase the pressure P1, which is accompanied by a decrease in the optimum energy consumption at the stabilized operating 20 point, and thus by improved energy efficiency of the liquefaction process. Furthermore, in Figure 5A relating to a nitrogen neon mixture, the operating point for the conventional process of Figure 1 with pure nitrogen is situated at 60. 25 The curve 70 (straight line portion) represents the variation in the energy efficiency as a function of the power injected into the process via the motor Ml, with reference to Figures 2 and 3. The top point WO=0 of the curve 70 corresponds to an unpowered motor Ml, i.e. a 30 motor delivering no power. The point W1 corresponds to said motor M1 delivering a power Wl>0. Likewise, successive points of the curve correspond to the motor Ml delivering increasing powers to the system, i.e. W4>W3>W2>W1>WQ=0. 35 The points WO to W4 correspond to the following powers being injected via the motor Ml: WO = zero power; 46 W1 = 7% of the total power; W2 = 15% of the total power; W3 = 24% of the total power; and W4 = 33% of the total power. 5 In similar manner, the diagram of Figure 6A shows the energy efficiency as a function of the pressure P2 and as a function of various variants of the invention. The curve 90 represents the process of Figure 2 using a refrigerant gas made up of 100% nitrogen. As in 10 Figure 5A, the top point WO=0 of the curve 90 corresponds to a motor that is unpowered, and thus that delivers no power. The point W1 corresponds to said motor M1 delivering a power W1>0. Likewise, the following points of the curve correspond to the motor Ml delivering 15 increasing powers to the system, such that W4>W3>W2>Wl>WO=0: said powers W1 to W4 being identical in Figures 5A and 6A. Thus, in this same Figure 6A, it can be seen that when the power W injected via the motor M1 increases, the 20 pressure P1 remains constant, but the pressure P2 increases and the efficiency increases, i.e. the energy consumption expressed in kWxd/t decreases, until it reaches a minimum 90a, which in this example coincides substantially with the point W3, after which said energy 25 consumption increases once more towards W4. This minimum 90a corresponds to the low point 70a of the curve 70 in Figure 5A, for minimum energy consumption of about 19.75 kWxd/t, a pressure P1 of about 9 bars, and a pressure P2 of about 28 bars. In comparison, the 30 operating point WO without energy being delivered via the motor M1 corresponds, in a pure nitrogen process, to energy consumption of about 21.25 kWxd/t, to the same pressure P1 of about 9 bars, and to a pressure P2 of about 11 bars: the energy efficiency is thus improved by 35 7.06%. In similar manner in the diagram of Figure 7A, there can be seen the energy efficiency as a function of the 47 pressure P3 and as a function of various variants of the invention, in particular for a mixture of nitrogen and neon. The points WO, W1, W2, W3, W4 correspond to the motor M1 injecting the same levels of power as described 5 above with reference to Figures 5A and 6A. P3 thus represents the maximum pressure of the system in the circuit S3: it increases in proportion to the power injected, and also to the percentage of neon in the refrigerant gas mixture. 10 Thus, an increase in the proportion of the power W injected via the motor Ml in Figures 2 and 3 compared with the total injected power: has no influence on the pressure Pl; increases the pressure P2; 15 increases the maximum pressure P3; and decreases energy consumption Ef down to a minimum value for a given proportion of power W, with the energy consumption then increasing once more beyond said given proportion of power W. 20 In the same way, the use of a nitrogen-neon mixture leads to an improvement in energy performance as shown in Figures 5A to 6A, both in conventional processes as described with reference to Figure 1 and in processes as described with reference to Figures 2 and 3. 25 Thus, giving consideration to a mixture having 20% neon, the pressure P1 is about 12.5 bars and curve 71 in Figure 5A shows the variations in energy consumption for the same increasing powers delivered to the system via the motor M3 (W4>W3>W2>W1>WO=0). 30 For this same neon percentage of 20%, curve 91 of Figure 6A shows the variations in energy consumption for the motor Ml delivering the same increasing powers to the system (W4>W3>W2>Wl>WO=0), as a function of the power P2. It can thus be seen that when the power W injected via M1 35 is increased, efficiency increases, i.e. energy consumption expressing kWxd/t decreases down to a minimum 91a, situated between the points W2 and W3 of said curve 48 91, after which said energy consumption increases once more towards W4. This minimum corresponds to the low point 71a of the curve 71 in Figure 5A for a minimum energy consumption of about 19.4 kWxd/t, a pressure P1 of 5 about 12.5 bars, and a pressure P2 of about 33 bars. In comparison, the operating point WO of the same curve 91 corresponding to a 20% neon mixture without energy being delivered via the motor Ml corresponds to an energy consumption of about 20.45 kWxd/t, to the same pressure 10 P1 of about 12.5 bars, and to a pressure P2 of about 17 bars, thus illustrating the improvement in energy efficiency when combining the increase in the percentage of neon and the increase in the power injected via the motor M1. 15 The same effects are observed using hydrogen, as can be seen in Figures 5B and 6B. In Figures 5 to 7, there can be seen performance diagrams for a conventional process and for a process of the invention for liquefying a natural gas comprising 85% 20 methane and 15% of said other constituents. In the diagram of Figure 7A, the maximum pressure P3 is plotted along the abscissa and the energy per unit mass of gas is plotted up the ordinate. Energy is plotted in units of kWxd/t of natural gas (1 kWxd/t = 25 0.024 kWh/kg). Thus, for a refrigerant gas constituted by 100% nitrogen, the operating point of the conventional process with reference to Figure 1 is situated at 60 in Figure 7A. In contrast, in the process of the invention as described with reference to Figures 2 and 3, for 30 various compositions of the nitrogen-neon mixture, while injecting power via the motor Ml, it is possible to vary the efficiency of the installation in accordance with curve 70 (20% neon) and with other curves (40% or 50% neon). Thus, from an operating point at 45 bars to 35 50 bars using the conventional process, corresponding to energy consumption of about 21.3 kWxd/t, it is possible to increase the thermodynamic efficiency by increasing 49 the maximum pressure. Thus, as shown in this same diagram, for a refrigerant gas constituted by 100% pure nitrogen, while injecting a portion of the power via the motor M1, and while operating at a pressure of about 5 68 bars, the energy consumption drops to about 19.75 kWxd/t, which represents an increase in efficiency of 7.28%. In general, by operating at higher pressure, for a given mass flow rate, the volume flow rates are reduced 10 prorata the increase in said pressure: the pipes are thus of smaller diameter, while their mechanical strength and thus their thickness, their weight, and their cost need to be increased accordingly: in contrast, the footprint is also reduced accordingly, which is most advantageous 15 for installations in a confined environment, such as on a floating support anchored at sea, or indeed on a methane tanker for a unit for reliquefying boil-off. In the same manner, compressors and turbines operating at higher pressures are much more compact. For the cryogenic heat 20 exchangers, an increase in pressure also improves heat transfer, but the heat exchange areas are not reduced by as much as for the pipes and the compressors and the turbines. In contrast, their weight increases significantly because they need to be able to withstand 25 this increase in pressure. Thus, overall, the process of the invention as shown in Figures 2 and 3 leads to installations that are more compact and to a significant improvement in energy efficiency when the refrigerant is pure nitrogen, which 30 energy efficiency is further improved when the refrigerant gas is a mixture of nitrogen and either neon or hydrogen. Figure 7A is a performance diagram for a conventional process as described with reference to 35 Figure 1 and for the process of the invention as described with references to Figures 2 and 3, using a mixture of nitrogen and neon as the refrigerant gas, in 50 which the maximum pressure P3 is plotted along the abscissa and energy per unit mass of gas is plotted up the ordinate. Energy is plotted in units of kWxd/t of natural gas. 5 Thus, for a given gas composition, the operating point of the conventional process described with reference to Figure 1 is situated at 60 in Figure 7A. In the process of the invention, as described with reference to Figures 2 and 3, and using a refrigerant gas made up 10 of 100% nitrogen, while injecting power via the motor Ml, it is possible to vary the efficiency of the installation along curve 61 with an optimum operating point 62 at about 68 bars, corresponding to an energy consumption of about 19.75 kWxd/t, which represents an improvement in 15 efficiency of 7.28% compared with the operating point 60 of the conventional process. By using a mixture of 80% nitrogen and 20% neon as the refrigerant gas, it is possible to increase pressure, as shown by curve 70, without the gas mixture reaching 20 its dew point up to an optimum value 70a of about 88 bars and for an energy consumption of about 19.4 kWxd/t, which represents a thermodynamic efficiency improvement of 1.77% compared with the operating point 62 of the process of the invention with a refrigerant gas made up of 100% 25 nitrogen and a thermodynamic efficiency improvement of 8.92% compared with the operating point 60 of the conventional process. By using a 60% nitrogen and 40% neon mixture as the refrigerant gas, it is possible to increase pressure as 30 shown by curve 71 without the gas mixture reaching its dew point up to an optimum value 71a of about 118 bars, together with minimum energy consumption of about 19.15 kWxd/t, which represents a thermodynamic efficiency improvement of 3.04% compared with the operating point 62 35 of the process of the invention with a refrigerant gas made up of 100% nitrogen, and a thermodynamic efficiency 51 improvement of 10.09% compared with the operating point 60 of the conventional process. By using a mixture of 50% nitrogen and 50% neon as the refrigerant gas, it is possible to increase the 5 pressure, as shown by curve 72, without the gas mixture reaching its dew point, up to an optimum value 72a of about 145 bars in association with minimum energy consumption of about 18.8 kWxd/t, which represents a thermodynamic efficiency improvement of 4.81% compared 10 with the operating point 62 of the process of the invention with a refrigerant gas made up of 100% nitrogen, and a thermodynamic efficiency improvement of 11.74% relating to the operating point 60 of the conventional process. 15 In the same manner, as shown in the diagram of Figure 7B, it is advantageous to use as the refrigerant gas a mixture of nitrogen and hydrogen. Thus, by using a mixture of 80% nitrogen and 20% hydrogen as the refrigerant gas, it is possible to 20 increase the pressure as shown by curve 80 without the gas mixture reaching its dew point up to an optimum value 80a of about 94 bars associated with minimum energy consumption of about 19.2 kWxd/t, which represents a thermodynamic efficiency improvement of 2.78% compared 25 with the operating point 62 of the process of the invention of Figures 2 and 3 using a refrigerant gas made up of 100% nitrogen, and a thermodynamic efficiency improvement of 9.86% relative to the operating point 60 of the conventional process of Figure 1. 30 By using a 60% nitrogen and 40% hydrogen mixture as the refrigerant gas, it is possible to increase the pressure, as shown by curve 81, and without the gas mixture reaching its dew point, up to an optimum value 81a of about 140 bars in association with minimum energy 35 consumption of about 18.8 kWxd/t, which represents a thermodynamic efficiency improvement of 4.81% compared with the operating point 62 of the process of the 52 invention as shown in Figures 2 and 3 when using a refrigerant gas made up of 100% nitrogen, and a thermodynamic efficiency improvement of 11.74% relative to the operating point 60 of the conventional process of 5 Figure 1. As shown by curve 82, by using a mixture of 50% nitrogen and 50% hydrogen as the refrigerant gas, it is possible to increase the pressure without the gas mixture reaching its dew point up to an optimum value 82a of 10 about 186 bars, in association with minimum energy consumption of about 18.7 kWxd/t, which represents a thermodynamic efficiency improvement of 5.32% compared with the operating point 62 of the process of the invention of Figures 2 and 3 using a refrigerant gas made 15 up of 100% nitrogen, and a thermodynamic efficiency improvement of 12.21% relative to the operating point 60 of the conventional process of Figure 1. Thus, for an increasing percentage of the additional gas, whether hydrogen or neon, that is added to nitrogen 20 in order to make up the refrigerant gas, the thermodynamic efficiency of the process is significantly improved, while allowing operation at higher pressure, which implies equipment that is more compact, which in itself is most advantageous when only very small areas 25 are available, as applies to a floating support anchored at sea or on board a methane tanker when applied to reliquefaction units. The process of the invention uses either a mixture of nitrogen and neon or of nitrogen and hydrogen, and in 30 spite of its slightly lower efficiency, it is preferred to use the nitrogen and neon mixture, since neon is an inert gas, whereas hydrogen is combustible and remains dangerous and difficult to use, in particular at high pressure in confined installations on board a floating 35 support. In addition, hydrogen is a gas that percolates very easily through elastomer gaskets and even under certain circumstances through metals, particularly at 53 very high pressure, and as a result the process of the invention based on the use of a nitrogen-hydrogen mixture does not constitute the preferred version of the invention: the preferred version of the invention remains 5 using a mixture of nitrogen and neon as the refrigerant gas in the devices described above with reference to the various figures. In the same manner, the efficiency of conventional processes using nitrogen as the refrigerant gas can be 10 improved by giving consideration to a binary mixture of nitrogen and neon or of nitrogen and hydrogen. Thus, as shown in the diagram of Figure 7A, curve 75 shows variation in the efficiency of a conventional process as described with reference to Figure 1 or one of 15 its variants, as a function of the percentage of neon gas in the refrigerant gas. For neon present at 20%, the operating point is situated at 70b, which corresponds to a maximum pressure P3 of about 63 bars and to an energy consumption of about 20.45 kWxd/t, which represents a 20 thermodynamic efficiency improvement of 3.76% compared with the operating point 60 of the same conventional process using a refrigerant gas made up of 100% nitrogen. With a 40% neon content, the operating point is situated at 71b, which corresponds to a maximum pressure 25 P3 of about 90 bars and to energy consumption of about 19.70 kWxd/t, which represents a thermodynamic efficiency improvement of 7.29% compared with the operating point 60 of the process conventional process with a refrigerant gas made up of 100% nitrogen. 30 For a 50% neon content, the operating point is situated at 72b, which corresponds to a maximum pressure P3 of about 120 bars and to an energy consumption of about 19.35 kWxd/t, which represents a thermodynamic efficiency improvement of 8.94% compared with the 35 operating point 60 of the same conventional process using a refrigerant gas made up of 100% nitrogen.

54 In the same manner, with a nitrogen-hydrogen mixture having 20% hydrogen, as shown in Figure 7B, the operating point is situated at 80b, which corresponds to a maximum pressure P3 of about 68 bars and to an energy consumption 5 of about 20.2 kWxd/t, which represents a thermodynamic efficiency improvement of 4.94% compared with the operating point 60 of the same conventional process with a refrigerant gas made up of 100% nitrogen. For a 40% hydrogen content, the operating point is 10 situated at 81b, which corresponds to a maximum pressure P3 of about 108 bars and to energy consumption of about 19.8 kWxd/t, which represents a thermodynamic efficiency improvement of 6.82% compared with the operating point 60 of the same conventional process with a refrigerant gas 15 made up of 100% nitrogen. With a 50% hydrogen content, the operating point is situated at 82b, which represents a maximum pressure P3 of about 150 bars and an energy consumption of about 19 kWxd/t, which represents a thermodynamic efficiency 20 improvement of 10.59% compared with the operating point 60 of the same conventional process with a refrigerant gas made up of 100% nitrogen. By way of example, a conventional liquefaction unit is dimensioned with reference to the powers of available 25 gas turbines, and high power turbines commonly deliver 25 MW. In general, it is desired to increase the power of the installation so it is possible to install two identical gas turbines (GT1 and GT2) in parallel in order 30 to obtain 30 MW (2x15 MW) or indeed 40 MW (2x20 MW), however there are then two rotary machine lines which increases overall bulk, the amount of pipework, and naturally costs. By using a single 25 MW gas turbine GT at C3 as in 35 Figure 2 and by adding power via the second motor M2, e.g. at 5 MW, in order to obtain a total of 30 MW, or of 15 MW in order to obtain a total of 40 MW, the operation 55 of the process as described with reference to Figure 2 is identical in terms of efficiency to that of the process using two gas turbines (GT1 and GT2) in parallel. Thus, giving consideration to a 25 MW gas turbine 5 GT, and adding 5 MW via the motor (M2), preferably using an electric motor, gives greater flexibility in operation and thus makes it possible to increase power by 20%. In contrast, the overall efficiency remains unchanged being substantially 21.25 kWxd/t of LNG produced, as shown in 10 the diagram of Figure 7 at point 60. In contrast, if the same power of 5 MW is delivered via the first motor M1, the overall power is still 30 MW, but the overall efficiency is then improved, substantially reaching the value of 19.8 kWxd/t of LNG 15 produced, which represents an improvement of 6.59% for the same overall power of 30 MW, compared with injecting a power of 5 MW via the second motor M2, as described above. Said additional 5 MW of power via the first motor M1 then represents 16.6% of the overall power and said 20 efficiency (19.8 kWxd/t) corresponds substantially to the point W2 in the diagram of Figure 7. In the same manner for the embodiment of Figure 3, by using a single 25 MW gas turbine GT at C2 as shown in Figure 3 and by adding power via the turbine GT, e.g. 25 5 MW in order to obtain a total of 30 MW, or 10 MW in order to obtain a total of 40 MW, the operation of the process described with reference to Figure 2 is identical in terms of efficiency to that of the process using two gas turbines (GTI and GT2) in parallel. 30 Thus, in consideration of a 25 MW gas turbine GT, adding 5 MW of power via the turbine GT gives greater flexibility in operation and thus enables power to be increased by 20%. In contrast, the efficiency of the assembly remains unchanged, being substantially 35 21.25 kWxd/t of LNG produced, as shown in the diagram of Figure 7 at point 60.

56 In contrast, if the same power of 5 MW is delivered via the first motor M1, the overall power is still 30 MW, but the overall efficiency is then improved and reaches a value of substantially 19.8 kWxd/t of LNG produced, which 5 represents an improvement of 6.59% for the same overall power of 30 MW, compared with injecting power, of 5 MW via the second motor M2 as described above. Said additional 5 MW of power added via the first motor M1 then represents 16.6% of the overall power and said 10 efficiency (19.8 kWxd/t) corresponds substantially to the point W2 in the diagram of Figure 7. Thus, as a function of the quantities and the qualities of the natural gas produced from underground reservoirs, it is advantageous to use a gas turbine GT, 15 e.g. a 25 MW turbine, that operates continuously at full power: - with additional power being added by being injected via the turbine GT (Figure 2) or via the second motor M2 (Figure 3) without changing the overall 20 efficiency (point WO in Figure 7); and - with constant or variable additional power being injected via the first motor Ml having the effect of improving the overall efficiency as shown by curve 61 in the same Figure 7, up to an optimum, i.e. a minimum 25 energy consumption of 19.75 kWxd/t corresponding substantially to the point W3 of said curve 61: the energy injected via said first motor M1 then represents substantially 24% of the total energy in this situation. In general, a gas turbine GT will be used at full 30 power, and additional power will be delivered via the first motor Ml, said additional power being limited to about 24% of the overall power so as to optimize the efficiency on the minimum value of 19.75 kWxd/t, and then, where necessary, the overall power will be 35 increased by injecting power via the second motor M2 and concurrently readjusting the power injected via the first motor M1 so that the power it injects is still 57 substantially equal to about 24% of the total power so as to conserve the efficiency of the installation on the optimum value of 19.75 kWxd/t. Said optimum efficiency of 19.75 kWxd/t for power 5 from the first motor M1 representing 24% of the total power is valid for a refrigerant fluid constituted by 100% nitrogen. When using a nitrogen-neon or nitrogen hydrogen mixture, the optimum efficiency, and thus also the power percentage vary as a function of the mixtures 10 and of their percentages of neon or hydrogen, but the advantages described in detail above remain valid and are even cumulative.

Claims

1. A process for liquefying natural gas comprising a majority of methane, the other components essentially comprising nitrogen and C-2 to C-4 alkanes, wherein said 5 natural gas for liquefying is liquefied by causing said natural gas to flow at a pressure PO higher than or equal to atmospheric pressure (Patm), through that least one cryogenic heat exchanger (EC1, EC2, EC3) by flowing in a closed circuit as a countercurrent in indirect contact 10 with at least one stream of refrigerant gas that remains in the gaseous state and that is compressed to a pressure P1 entering said cryogenic heat exchanger at a temperature T3' lower than T3, T3 being the liquefaction temperature of said liquefied natural gas on leaving said 15 cryogenic heat exchanger, T3 being lower than or equal to the liquefaction temperature of said liquefied natural gas at atmospheric pressure, wherein said natural gas for liquefying is liquefied by performing the following concurrent steps: 20 a) causing said natural gas for liquefying to flow(Sg) at a pressure PO higher than or equal to atmospheric pressure (Patm), through at least three cryogenic heat exchangers (EC1, EC2, EC3) connected in series and comprising: 25 - a first heat exchanger (ECi) in which said natural gas entering at a temperature TO is cooled and leaves (BB) at a temperature Ti lower than TO; then - a second heat exchanger (EC2) in which the natural gas is fully liquefied and leaves (CC) at a 30 temperature T2 lower than Ti and higher than T3; and - a third heat exchanger (EC3) in which said liquefied natural gas is cooled from T2 to T3; b) causing at least two streams (Si, S3) of refrigerant gas in the gaseous state and referred to 35 respectively as the first and third streams to circulate in closed-circuits at different pressures Pi and P2 passing through at least two of said heat exchangers in 59 indirect contact with and as a countercurrent relative to the natural gas stream (Sg) and comprising: - a first stream of refrigerant gas (Si) at a pressure P1 lower than P3 passing through the three heat 5 exchangers (EC1, EC2, EC3) entering (DD) into said third heat exchanger (EC3) at a temperature T3' lower than T3, then entering (CC) at a temperature T2' lower than T2 into said second heat exchanger (EC2), then entering (BB) at a temperature T1' lower than Ti into said first heat 10 exchanger (ECi) and leaving (AA) said first heat exchanger at a temperature TO' lower than or equal to TO, said first stream of refrigerant gas at Pi and T3' being obtained by using a first expander (Ei) to expand a first portion (Di) of a second stream (S2) of refrigerant gas 15 compressed to the pressure P3 higher than P2, said second stream (S2) circulating in indirect contact with and as a cocurrent relative to said natural gas stream (Sg) by entering (A) into said first heat exchanger (ECi) at TO and said first portion (Di) of the second stream (S2) 20 leaving (CC) said second heat exchanger (EC2) substantially at T2; and - a third stream (S3) at a pressure P2 higher than Pi and lower than P3 circulating in indirect contact with and as a cocurrent relative to said first stream, 25 passing solely through said second and first heat exchangers (EC2, EC1), entering (CC) into said second heat exchanger at a temperature T2' lower than T2 and leaving (A) said first heat exchanger (ECi) at TO' lower than or equal to TO, said third stream (S3) of 30 refrigerant gas at P2 and T2 being obtained by using a second expander (E2) to expand a second portion (D2) of said second stream (S2) of refrigerant gas leaving said first heat exchanger substantially at Ti, the flow rate D2 of said second portion of the second stream being 35 higher than the flow rate Di of the first portion of the second stream; 60 c) said second stream of refrigerant gas (S2) compressed to the pressure P3 being obtained by using at least two compressors (Ci, C2, C3) and by cooling (Hi, H2), to compress said first and third streams (Si, S3) of 5 refrigerant gas leaving (AA) said first heat exchanger (ECi) at Pi and P2 respectively, a first compressor compressing from Pi to P2 all of said first stream of refrigerant gas leaving (Ak) said first heat exchanger (ECi), and at least one second compressor (C2) 10 compressing firstly said third stream (S3) of refrigerant gas leaving said first heat exchanger (ECi) at P2 and secondly said first stream of refrigerant gas compressed to P2 and leaving said first compressor, from P2 to at least P' 3 , where P' 3 is a pressure lower than or equal to 15 P3 and higher than P2, thereby obtaining said second stream of refrigerant gas at P3 and TO after cooling (Hi, H2); in which process: - the series-connected first and second compressors 20 (Ci, C2) are coupled respectively to said first and second expanders (Ei, E2) consisting in energy-recovery turbines; and - at least said first compressor (Ci) is coupled to a first motor (Mi) ; 25 said process being characterized in that: - said first motor enables the amount of power delivered to said first component to be varied relative to the power delivered to the other compressors, and - a gas turbine (GT) is coupled either to said 30 second compressor, which compressor compresses said second stream of refrigerant gas directly to P3, or is coupled to a third compressor (C3) connected in series after the second compressor (C2), said third compressor compressing said second stream of refrigerant gas from 35 P'3 to P3, said gas turbine delivering the major portion of the total power delivered to all of said compressors (C1, C2, C3) in use; 61 said first motor (Ml) delivering 3% to 30% of the total power delivered to all of said compressors(C1, C2) in use, said gas turbine (GT) supplying 97% to 70% of the total delivered power. 5

2. A process according to claim 1, wherein said pressure P2 is caused to vary in controlled manner by delivering power in controlled manner to said first compressor from said first motor, in such a manner that the energy (Ef) 10 consumed for performing the process is minimized.

3. A process according to claim 2, wherein the pressure P2 is increased by increasing the power injected to the first compressor via the first motor, the pressure P1 15 remaining substantially constant.

4. A process according to claim 2 or claim 3, wherein said pressure P2 is caused to vary in controlled manner by delivering power in controlled manner to said first 20 compressor via said first motor when the composition of the liquid gas for liquefying varies.

5. A process according to any one of claims 1 to 4, wherein two compressors (Cl, C2) are used that are 25 connected in series, and that comprise: i) a said first compressor coupled to said first expander (El) compressing from P1 to P2 all of said first stream of refrigerant gas leaving (A) said first heat exchanger (EC1); and 30 ii) a said second compressor (C2) coupled to said second expander (E2), compressing from P2 to P3 firstly said third stream (S3) of refrigerant gas leaving said first heat exchanger (EC1) at P2 and secondly said first stream of refrigerant gas compressed to P2 and leaving 35 said first compressor, in order to obtain said second stream (S2) of refrigerant gas at P3 and TO after cooling (H1, H2); and 62 iii) said first compressor (Cl) being driven by a first motor (Ml), said second compressor (C2) being driven by at least one said gas turbine (GT). 5 6. A process according to any one of claims 1 to 4, wherein three compressors (Cl, C2, C3) are used that are connected in series and that comprise: i) a said first compressor (Cl) driven by a said first motor (Ml) and coupled to said first expander (El), 10 compressing from P1 to P2 all of said first stream of refrigerant gas leaving (A) said first heat exchanger (EC1); and ii) a said second compressor (C2) driven by a said second motor (M2) and coupled to said second expander 15 (E2) compressing firstly said third stream (S3) of refrigerant gas leaving said first heat exchanger (EC1) at P2 and secondly said first stream of refrigerant gas compressed to P2 and leaving said first compressor (Cl), from P2 to P'3, where P'3 is higher than P2 and lower 20 than P3; and iii) a third compressor (C3) driven by a said gas turbine (GT) to supply the major portion of the energy and to compress to P3 all of the first and third streams of refrigerant gas leaving the second compressor (C2) in 25 order to obtain said third stream of refrigerant gas at P3 and TO after cooling (Hi, H2); and iv) said first motor (Ml) delivers at least 3% of the total power delivered to all of said compressors (Cl, C2, C3) in use, said gas turbine (GT) coupled to said 30 third compressor (C3) and said second motor (M2) coupled to the compressor (C2) together supplying 97% to 70% of the total power delivered to all of said compressors (Cl, C2, C3) in use. 35 7. A process according to any one of claims 1 to 6, wherein said refrigerant gas comprises nitrogen. 63

8. A process according to any one of claims 1 to 7, wherein the composition of the gas for liquefying lies within the following ranges to give a total of 100%: - methane 80% to 100%; 5 nitrogen 0% to 20%; - ethane 0% to 20%; - propane 0% to 20; and - butane 0% to 20%. 10 9. A process according to any one of claims 1 to 8, wherein: - TO and TO' lie in the range 10 0 C to 35 0 C; and - T3 and T3' lie in the range -160 0 C to -170 0 C; and - T2 and T2' lie in the range -100 0 C to -140 0 C; and 15 Ti and T1' lie in the range -30 0 C to -70 0 C.

10. A process according to any one of claims 1 to 9, wherein: - PO lies in the range 0.5 MPa to 5 MPa; and 20 P1 lies in the range 0.5 MPa to 5 MPa; and - P2 lies in the range 1 MPa to 10 MPa; and - P3 lies in the range 5 MPa to 20 MPa.

11. A process according to any one of claims 1 to 10, 25 wherein P2 is caused to vary until the minimum total energy (Ef) consumed in the process is lower than

21.5 kWxd/t of liquefied gas produced. 12. A process according to any one of claims 1 to 11, 30 wherein it is performed on board a floating support. 13. A process according to any one of claims 1 to 12, wherein PO is higher than atmospheric pressure. 35 14. An installation on board a floating support for performing a process according to any one of claims 1 to 13, the installation comprising: 64 at least three said cryogenic heat exchangers (EC1, EC2, EC3) in series and comprising at least: - a countercurrent flow first duct suitable for causing a first stream (Si) of refrigerant gas in the 5 gaseous state compressed to P1 to flow as a countercurrent successively through the third, second, and first heat exchangers (EC3, EC2, EC1); - a cocurrent flow second duct suitable for causing a said second stream (S2) of refrigerant gas in 10 the gaseous state compressed to P3 to flow as a cocurrent successively through only the said first and second heat exchangers (EC1, EC2); - a countercurrent flow third duct for said refrigerant gas suitable for causing a said third stream 15 (S3) of refrigerant gas in the gaseous state compressed to P2 to flow as a countercurrent successively through only said second and first heat exchangers (EC2, EC1); - a fourth duct (Sg) suitable for causing said natural gas for liquefying to flow successively through 20 the first, second, and third heat exchangers (EC1, EC2, EC3); - a first expander (El) between the outlet from said second duct and the inlet to said first duct; - a second expander (E2) between i) a branch 25 connection (BB) to said second duct situated between said first and second heat exchangers, and ii) the inlet (CC) of said third duct; and - a first compressor (Cl) at the outlet from said first duct, coupled to a turbine constituting said 30 first expander; - a second compressor at the outlet from said second duct, coupled to a turbine constituting said second expander, and said second compressor being connected in series with said first compressor; and 35 - a duct for passing all of the gas compressed to P2 by the first compressor (Cl) to the second 65 compressor (C2) connected in series in this way with said first compressor; and - at least one compressor (Cl) coupled to a first motor (Ml) suitable for delivering at least 3%, of 5 the total power delivered to all of said compressors (Cl, C2, C3) in use, said first motor enabling the amount of power delivered to said first compressor to be varied relative to the power delivered to the other compressors; and 10 a gas turbine (GT) coupled either to said second compressor compressing said second refrigerant gas stream directly to P3, or to a third compressor (C3) connected in series after the second compressor (C2), said third compressor compressing said second refrigerant 15 gas stream from P'3 to P3, said gas turbine being suitable for delivering the major portion of the total power delivered to all of said compressors (Cl, C2, C3) in use. 20 14. An installation according to claim 13, wherein it has only two compressors (Cl, C2) connected in series and comprising: i) one said first compressor (Cl) coupled to said first expander (El), suitable for compressing from P1 to 25 P2 all of said first stream of refrigerant gas leaving (AA) said first heat exchanger (EC1); and ii) one said second compressor (C2) coupled to said second expander (E2), suitable for compressing firstly said third stream (S3) of refrigerant gas leaving said 30 first heat exchanger (EC1) at P2 and secondly said first stream of refrigerant gas compressed to P2 and leaving said first compressor, from P2 to at least P' 3 , where P' 3 is a pressure higher than P2 and lower than or equal to P3, in order to obtain said second stream of refrigerant 35 gas at P3 and TO after cooling (Hi, H2); and iii) a said first motor (Ml) coupled to a said first compressor (Cl), and at least one gas turbine (GT) 66 coupled to said second compressor (C2), said first motor being suitable for delivering at least 3% of the total power delivered to all of said compressors (Cl, C2) in use; and 5 iv) said gas turbine (GT) coupled to said second compressor being suitable for supplying 97% to 70% of the total delivered power. 16. An installation according to claim 15, wherein it has 10 only three compressors (Cl, C2, C3) connected in series and comprising: i) a said first compressor (Cl) coupled to said first expander (El) and to a first motor (Ml); and ii) a said second compressor (C2) coupled to said 15 second expander (E2) and to a second motor (M2); and iii) a third compressor (C3) coupled to a gas turbine (GT) suitable for supplying the major portion of the energy and suitable for compressing to P3 all of the first and third streams of refrigerant gas compressed by 20 the second compressor (C2) in order to obtain said third stream of refrigerant gas at P3 and TO after cooling (Hi, H2); and iv) said first motor (Ml) being suitable for delivering at least 3% of the total power delivered to 25 all of said compressors (Cl, C2, C3) in use; and v) the gas turbine (GT) coupled to said third compressor (C3) and said second motor (M2) coupled to the second compressor (C2) being suitable together for supplying 97% to 70% of the total power delivered to all 30 of said compressors (Cl, C2, C3) in use. 17. Installation according to claim 15 or 16, wherein said first motor is suitable to deliver 3% to 30% of the total energy delivered to all of said compressors in use. 35