Three-pipe thermal network Field of Invention The present invention generally relates to thermal networks, and more precisely to thermal networks that use an energy transfer medium to supply heat and refrigeration to end-user locations. State of the art A thermal network is a system that provides thermal energy services to several end-user locations within a same building or within separate buildings. A thermal network is generally connecting one or multiple central plants (heating/cooling injection points) to said end-used locations (heating/cooling utilization points). When heating is needed the thermal network vehiculates energy from one or several plants to end-user locations via decentralized heat exchangers or heat-pumps through pipelines that carry water, steam or CO ₂ . If cooling is needed the thermal energy vehiculated in the thermal network follows an opposite path, i.e. heat from heat exchangers or refrigeration systems is extracted through the said pipelines. A thermal network is made of at least a central plant, a pipe system that contains an energy transfer medium such as water or CO ₂ , and at least an end-user location. European patent EP 2 122257 B1 discloses a district energy system comprising two main pipes, the first pipe containing CO ₂ in liquid state and the second pipe containing CO ₂ in gaseous state. Both pipes are adapted in a way as to each act as a supply or return pipe, depending on the heating or cooling requirements of the end-user locations. In the system disclosed in this prior art, rotating machines (pumps or compressors) are in most cases needed at the end-user locations as one of the pipes must be maintained at a higher pressure than the other all along the network between end-user locations and between end- user locations and the central plant. This results in an increased complexity of the equipment installed at end-user locations which negatively impacts the footprint of the system within the end-user locations (larger size of the substations) as well as the operational reliability (more maintenance required, more numerous sources of breakdowns, rotating machines are more sensitive to inadequate operating conditions than the other pieces of equipment involved, potential difficulty regarding access right to the location). Swiss patent CH 712294 A2 discloses a thermal energy superstructure with multiple temperature levels and exchange points. The possibility of using CO ₂ as a heat transfer fluid in such superstructure is presented. The CO ₂ transfer between different temperature stages can take place at temperature below 30°C and pressures below 70 bar, with total or partial evaporation or condensation of the CO ₂ . For the superstructure disclosed in CH 712294 A2 there is no characterization on how the circulation of the fluid is realized within a temperature stage as well as between different temperature stages. General description of the invention The present invention provides alternatives and improvements with respect to existing thermal networks. More precisely, the present invention concerns a thermal network comprising at least one plant, at least one end-user location, a pipe system and an energy transfer medium, said end- user location(s) being connected to the plant through the pipe system. The invention is characterized in that it comprises three main pipes that are each connected to the plant and wherein the energy transfer medium is in a liquid state in the first and the third main pipe, and in a gaseous state in the second main pipe. The present invention is the synthesis resulting from the following considerations that are mostly linked to energy efficiency: - The thermal network aims principally at supplying heat or cooling services to the end- user locations. It should aim at being as simple as possible in the way it supplies these services while not excluding other services less related to thermal energy. - In case where there is a simultaneous demand of one or several end-user locations that demand heat and one or several end-user locations that demand cooling. the network has to be able to recover the heat injected in it by the cooling user and transfer it to the end user location that extract heat from the network, exploiting as much as possible the waste heat recovery potential and as a result improving the energy efficiency of the system. - The temperature of the heating service required can vary in a fairly wide range from one end-user location to another and the network has to be able to deal with it in the most energy efficient way. - The temperature of the cooling service required can vary in a fairly wide range from one end-user location to another and the network has to be able to deal with it in the most energy efficient way. The above considerations can be met by a network that uses heat pumps installed at the end- user locations to supply heat always at the temperature required by each end-user. It guarantees a lower electricity consumption, as compared to the more traditional approach of thermal network where the heat pumping is done at the plant with a supply temperature imposed by the end-user locations requiring the highest temperature. The use of direct heat exchange for supplying most of the cooling services is also a way to meet said considerations. Direct heat exchange consumes far less electricity as compared to systems with compression chillers. It results from the proposed arrangement of decentralized heat pumps and direct heat exchange for cooling that the temperature of the network needs to simultaneously be high enough to supply heat to the cold source of heat pumps installed at the end-user locations while being cold enough to be able to supply the cooling to most of the end-user locations through direct heat exchange. The following constraints have also been taken into account in the proposed invention: - The combination of the use of heat pumps and direct heat exchange for cooling implies that the temperature difference between the hottest part and the coldest part of the pipe system is limited to a few Kelvin (in general less than 10 K). - The pipe system should also be very compact to be able to facilitate its installation in the congested underground that is often found in urban areas. - Practical considerations regarding maintenance and reliability call also for minimizing the complexity of the elements of the network at the end-user locations, particularly for the ones that deliver the heating and cooling services. The present invention complies with these constraints by exploiting the latent heat of vaporization of the medium instead of its sensible heat, that way it renders possible to deliver the same amount of heating, respectively cooling service with many times less mass flowrate than if the sensible heat was used. The temperature difference constraint is easily met because the evaporation and condensation of a pure compound is isothermal and even in the case of a mixture of several chemical species, the evaporation is in general within the acceptable range of temperature difference. The small mass flowrate translates directly into a higher compacity of the pipe system because of the smaller diameter of the pipes needed. The use of three pipes, two filled with liquid and one with gas is a way to reduce the complexity at the end-user locations by, in most cases, removing the need for rotating machines (pumps or compressor) dedicated to the handling of the medium within said locations. Of course, there still might be heat pumps or refrigeration units within those locations but in most of the cases they are not part of the network in the sense that the medium of said network does not flow in the rotating machines (compressor or pumps) of said heat pumps and refrigeration units. The result is that the only rotating machines necessary to the operation of the network itself are located at the plant, which is very favourable for their monitoring, maintenance and the overall operational reliability of the entire network. Moreover, the configuration of two liquid pipes and one gas pipe leads to a more compact pipe system than an alternative that would use two gas pipes and one liquid pipe. This result comes from the fact the pressure drops tend to be higher in the gas than in the liquid which lead to larger pipe diameters. In other words, given a predefined amount of thermal energy to be supplied by the network to end-user locations (heating or cooling) and given a maximum amount of pumping energy to circulate the medium, a configuration with two gas pipes and one liquid pipe would require larger pipe diameters than a configuration with two liquid pipes and 1 gas pipe. In the present document, the expressions “liquid” has to be understood as either fully constituted of liquid or mostly of liquid (i.e. more than 50 % by mass) with a smaller fraction of gas. The expressions “gas” has to be understood as either fully constituted of gas or mostly of gas (i.e. more than 50 % by mass) with a smaller fraction of liquid. In the following text, the term “medium” has to be understood as “energy transfer medium”. The end-user locations may directly or indirectly be connected to the plant through the pipe system. The energy transfer between the plant and the pipe system, the pipe-system and the end-user location(s), or between the end-user locations is realized by a thermodynamic transformation (e.g cooling, evaporation or condensation) of the medium in energy exchange devices (e.g. assemblies of heat exchangers, valves, sensors, etc.) at said plant and end-user locations. To be able to guarantee that the right state of the medium is always selected before being sent to the pipe system by the end-user locations and the plant the following methods are used: Whenever liquid has to be used it is done: o Either by extracting it from a part of a receiver, preferably a lower part of a receiver o Or by controlling the flow of medium in a condenser, preferably using a valve, based on the level of subcooling at the outlet. Indeed, a value of subcooling high enough is an indication that the flow leaving the condenser is fully in liquid state. Whenever gas has to be used it is done: o Either by extracting from another part of a receiver, preferably an upper part of a receiver o Or by controlling the flow of medium in an evaporator, preferably using a valve, based on the level of superheat at the outlet. Indeed, a value of superheat high enough is an indication that the flow leaving the evaporator is fully in gaseous state. Note that the “selection” of the right phase occurs not necessarily right before the injection point in the pipe-system but also upstream of machines such as pumps or compressors for guaranteeing their reliable operation and lifetime. Preferably, the first main pipe is the supply pipe. It is a pressurized line inside which the medium always circulates from the plant(s) towards the end-user location(s). The first main pipe acts as a supply pipe, the third main pipe acts as a return pipe. It is a pressurized line inside which the medium always circulates always from the end-user location(s) towards the plant(s): - Ideally, during operation, the pressure in all the main pipes should be equal. However, because of the various pressure drops (friction, gravity, momentum…) it is preferable to guarantee a pressure difference between the pipes. Said pressure difference should be kept as low as possible for the sake of energy efficiency. As a result, the pressure in the pipe system is controlled by pumps (or compressors) in such a way that, at every location along the pipe system and by extension at the boundaries between the end- user locations and the pipe respectively between the plant and the pipe system: The pressure of the medium in the first main pipe (supply pipe) is higher than the pressure in the second main pipe. - The pressure of the medium in the second main pipe is higher than the pressure of the medium in the third main pipe (return pipe) To ensure the compactness of the pipe system, of the components located at end-user locations and of those located at the plant, the fluid in the liquid pipes is kept as close as possible to the saturated liquid or even in a slightly subcooled liquid state, while the fluid in the gas pipe is kept as close as possible to the saturated gas or even in a slightly superheated gaseous state. Advantageously, the present invention provides a pipe configuration that allows to guarantee the energy circulation of the fluid in the system between the central plant and any end-user locations, or between at least two end-user locations. This is guaranteed by imposing the fluid direction within the supply and return liquid pipes. The flow direction in the supply pipe is imposed by allowing connections only from said pipe to the inlet port at the end-user locations. The flow direction in the return pipe is imposed by allowing connections only from the outlet port at the end-user locations, to said pipe. The present invention also encompasses the use of a series of circuit elements to improve the reliability and energy efficiency of the circulation of the medium that comprise: - One or multiple bypass valves connected between the supply pipe and the return pipe - Depending on the circuit topology, gas traps may be located in the supply pipe and the liquid return pipe, to collect eventual gas bubbles - Depending on the circuit topology, liquid traps may be located in the second gas pipe, to collect possible drops of liquid The invention also relates to the use of the thermal network as defined above, wherein when the heating requirements are higher than the cooling requirements (e.g. in winter), the second main pipe (gas pipe) is used as a supply pipe, and vice-versa, when the cooling requirements are higher than the heating requirements (e.g. in summer), the second main pipe is used as a return pipe. More specific aspects of the invention are defined in the dependent claims. With respect to EP 2122257 B1, the use of the three pipes according to the invention provides several advantages, in particular: ^ a substantial reduction in the complexity of the end-user locations, as for the most common cases, it suppresses the need for local rotating machines that are in direct contact with the medium. ^ a substantial reduction of the footprint of the equipment installed at the end-user locations, which greatly improves the deployment capability of such systems. ^ a substantial reduction of the maintenance work required, as the equipment requiring heavy and/or regular maintenance work are reduced in number and concentrated in one single location (the plant). ^ A substantially improved reliability and serviceability of the system in general as the sensitive equipment is located preferentially at the plant and not in premisses at the end- user locations that will unavoidably by more problematic to access to. ^ monodirectional flow in liquid lines allows to ease the fluid control, avoiding consequently the risk of simultaneous pump operation (at end-user and plant sides) with opposite flow directions, responsible for potential flow instability and pumps malfunctioning/failure. Detailed description of the invention The invention will be better understood in this chapter, with the support of non-limiting examples illustrated by the following figures: Figure 1: 1 ^st configuration of a thermal network according to the invention Figure 2: 2 ^nd configuration of a thermal network according to the invention Figure 3: 3 ^nd configuration of a thermal network according to the invention Figure 4: 4 ^th configuration of a thermal network according to the invention Numerical references used in the figures: 1. Intermediate pressure receiver 2. Medium pressure liquid pump 3. Low pressure receiver 4. Low pressure compressor 5. Low pressure liquid pump 6. Sub-cooling apparatus 7. Anti-flash condenser 8. Heat pump apparatus 9a. Intermediate pressure evaporator 9b Source preheater 10. Intermediate pressure condenser 11. Liquid supply pipe 12. Gas pipe 13. Liquid return pipe 14. Cooling user 15. Heating user 16. Intermediate pressure compressor 17. Refrigeration apparatus 18. Liquid pipe bypass 19. Gas pipe trap 20. Liquid return pipe gas trap 21. Liquid supply pipe gas trap 22. Receiver isolation valve 23. Receiver flash purge valve In all examples, the thermal network comprises a plant, end-user locations 14,15 three main pipes, two of which, namely the first 11 and the third 13, respectively form a liquid supply pipe and a liquid return pipe. The medium is preferably CO ₂ : The second main pipe 12 contains the same medium but in a gaseous state. The energy transfer is realized by evaporation/condensation of the said medium. Both first and third main pipes 11, 13 are arranged and connected in a fashion such that the medium flows in a single direction between the end user locations 14, 15 and between the plant and the end user locations 14,15. The second main pipe 12 is arranged and connected in a fashion such that the medium can flow indifferently in both directions in every segment connecting the end-user locations 14,15 between them, as well as those connecting the plant and end-user locations 14,15. Preferably, the end-user locations 14,15 are provided with any suitable technology that can: - extract the network medium from one or several lines and/or - inject the network medium into one or several lines in a thermodynamic state close to the one desired in said lines. For instance, when considering the examples illustrated in the following figures, at end-user locations 14 the medium is extracted from the first main pipe 11 in a thermodynamic state corresponding to that prevailing in said pipe at this location and time, ideally saturated or slightly subcooled liquid but in some cases a liquid/gas mixture at the saturation with a gas content as low as possible. Within the equipment at the end-user location 14, the medium can undergo all sorts of thermodynamic processes, the simplest of which being an evaporation in a heat exchanger device in which the flow of the medium is regulated, preferably using a valve upstream of the heat exchanger inlet, so as to guarantee that the thermodynamic state of the medium, at the outlet of the end user location corresponds to that desired in the gas pipe 12 to which said outlet is connected. In that case the desired state is saturated gas or slightly superheated gas with no liquid content. In the other end-user location 15, the medium is extracted from the second main pipe 12 in a thermodynamic state corresponding to that prevailing in said pipe at this location and time, ideally saturated or slightly superheated gas but in some cases a liquid/gas mixture at the saturation with a gas content as high as possible. Within the equipment at the end-user location 15, the medium can undergo all sorts of thermodynamic processes, the simplest of which being the condensation in a heat exchanger device in which the flow of the medium is regulated, preferably using a valve downstream of the heat exchanger device’s outlet, so as to guarantee that the thermodynamic state of the medium, at the outlet of the end user location 15, corresponds to that desired in the third main pipe 13, to which said outlet is connected. In that case the desired state is saturated or slightly subcooled liquid with no gas content. Advantageously, the plant includes any suitable element that allows to: - maintain pressure at suitable set points by exchanging energy with an energy source/sink, denominated “The Source”. - ensure the circulation of the medium in the different pipes 11-13, in accordance to the flow requirements generated at end-user locations 14,15. - ensure that the medium supplied by the plant to the first (supply) main pipe 11 is in liquid state and, when applicable, the medium supplied to the second main pipe 12 is in gaseous state. - Ensure that the plant can continue its operation even in the event that the medium collected from the third (return) main pipe 13 departs from the desired liquid state (within reasonable bounds). - Ensure that the plant can continue its operation even in the event that the medium collected from the second main pipe 12 departs from the desired gas state (within reasonable bounds). In those examples, the network is arranged to ensure that the medium in the first main pipe 11 is always and everywhere at a higher pressure than that in the second main pipe 12 and the pressure in this later one always and everywhere higher than the pressure in the third pipe 13. At end user locations 14,15, the above-mentioned pipe arrangement and their respective pressure allows for the medium to flow either from the first main pipe 11 to the second main pipe 12 or from the second main pipe 12 to the third main pipe 13, without the need of any active machine such as a pump or a compressor. This provides an important benefit for the end-user locations dedicated to cooling 14, heating 15 or both. Among other advantages, it reduces the footprint of the equipment installed at said locations 14,15 as well as the operational risk link to active machines that tend to be more sensitive than passive equipment (p.ex. valve or heat exchangers). Also, active machines generally require more maintenance than passive equipment and, if located within the premisses of end-users, such machines will likely be more difficult to service than if located in the plant, for basic access right reasons. For a system where the end-user locations 14,15 exploit the network only for cooling purposes 14 or heating purposes 15 or both, there is no medium that leaves or enters the system (the system is closed, it does not exchange mass with its environment, only energy). During steady state operation the mass flow of the gas leaving, respectively entering the plant, exactly compensates the difference between the mass flow of liquid entering the plant from the third main pipe 13 and the mass flow of liquid leaving the plant through the first main pipe 11. During transient, the mass flows will not compensate each other until a new steady state is reached. As an example if the end-user locations cumulated demand for cooling generates a mass flow in the liquid supply line of “X” kg/s and the cumulated demand for heating generates a mass flow in the liquid return line of “Y” kg/s, then a mass flow of “Z = X-Y” kg/s entering the plant from the gas line will be observed (provided that the system is in steady state). In reverse, if the mass flow from the liquid return line is “Y” kg/s and that to the liquid supply line is “X” kg/s, a mass flow of “Z = Y – X” kg/s of gas leaving the plant will be observed (provided that the system is in steady state.) The flow of liquid entering the plant from the third main pipe 13 is first directed into a receiver 3, the function of which is to separate the liquid from the gas. Indeed, because of the pressure drops along the third main pipe 13 but also through some of the valves at the end-user locations 14,15 and also, in some conditions, because of heat input coming from the environment through the pipe wall, a certain amount of “flash gas” will enter the plant. Therefore, a separation receiver must be used to allow the pump 5 being fully fed with liquid. However, over time, gas will build up in the receiver and a suitable mean must be used to either send it back to the gas line, for instance by condensing it, with an “anti-flash” condenser 7 (see figure 1)or using a dedicated compressor 4 (see figure 2). The anti-flash condenser 7 can be cooled by any suitable mean. For instance, if a source cold enough is available, it can be used directly to cool down said condenser. In the absence of source at a suitable temperature for direct cooling, a heat pump apparatus 8 can be used for providing the necessary cooling to the anti-flash condenser via its cold source. A combination of both direct cooling and cooling via the heat pump apparatus can be used, which would be particularly suited for cases where the temperature of the source available varies significantly over the time. To improve the operational reliability and energy efficiency of the system several devices can be incorporated in the system. One or several liquid pipe bypass 18 can be installed, preferably with at least one bypass at the furthest point of the pipe system from the plant. Said bypass consist in a valve, the opening of which can be fixed, manually set or automatically actuated, that connects the first main pipe 11 to the third main pipe 13. It ensures that even in absence of medium being extracted from, respectively injected into said pipes at the end-user locations 14, 15 a minimum flow of the medium is guaranteed within the network. The purpose of having a minimum flow of medium guaranteed always and everywhere in the pipe system is two-fold: to ensure that the medium supplied by the first main pipe 11 at the inlet of the enduser locations 14 is always in an acceptable thermodynamic state, ideally saturated or slightly subcooled liquid, so as to allow for a start-up with no or minimal time delay of the device at the end-user location. Without this minimal flow, thermal input from the environment could evaporate the liquid in the pipe when the medium is at rest and the start-up of the equipment at end-user locations 14 would be delayed of the time needed to bring back liquid to said end-user location from the plant, via the first main pipe 11. to ensure that the medium coming back to the plant via the third main pipe 13 is always in the adequate thermodynamic state, ideally saturated or slightly subcooled liquid but it is also conceivable to have a liquid/gas mixture at saturation with a gas content as low as possible. This is particularly of importance when there is no or very little medium injected in the third main pipe 13 by the device at the end-user location 15 since in such cases thermal energy from the environment (e.g heating of liquid pipe or equipment) could evaporate a too large fraction of liquid in the pipe for the low pressure liquid pump 5 to be able to operate, even with the help of the other devices that are incorporated in the system to mitigate that problem, namely the low pressure receiver 3, the anti-flash condenser 7 or the low pressure compressor 4 or the low pressure receiver isolation and flash purge valves 22, 23.

One or several gas traps 19 can be installed to collect and drain the gas bubbles that may be present in the medium flowing in the first main pipe 11. These traps could for instance be advantageously placed at locations where there is a local maximum in elevation in the network. Such a trap comprises a receiver connected via a pipe to the Liquid Supply Pipe 11, and connected via an automatic valve to the gas pipe 12 (second main pipe). The arrangement is made in such a way as to ensure that the automatic valve will drain gas from the receiver and not liquid. During normal operation the valve is closed and gradually the receiver of the trap will fill up with the gas collected. Once the liquid level in the receiver reaches a value low enough, the automatic valve opens and drains the gas towards the gas pipe 12. When the liquid level in the receiver of the trap reaches a value high enough the valve closes back. It is also possible to use a modulating valve commanded via a suitable control loop, to stabilize the level of liquid in the receiver at a desired value. With the latter option the drainage of the gas would be a continuous process instead of a batch one.

One or several gas traps 20 can be installed to collect and drain the gas bubbles that may be present in the medium flowing in the third main pipe 13. These traps could for instance be advantageously placed at locations where there is a local maximum in elevation in the network. Such a trap comprises a receiver connected via automatic valves to all three main pipes of the pipe system 11, 12 and 13. The arrangement is made in such a way that the valve that connects the receiver to the second main pipe 12 will drain gas and not liquid. During normal operation the valve connecting the receiver of the trap to the pipes 11 and 12 are closed and the one connecting said receiver to the third main pipe 13 is open. In this way the receiver will gradually fill-up with gas collected from the third main pipe 13. Once the liquid level in the receiver of the trap reaches a value low enough, the valve that connects the receiver to the third main pipe 13 is closed, the valves that connects the receiver to the main pipes 12 and 11 are opened. Because of the higher pressure in the pipe 11 than in the pipe 12, some liquid mixture is admitted from the pipe 11 into the receiver, acting as a liquid piston that pushes out the gas from the receiver through the valve and into the second main pipe 12. Once the level of liquid in the receiver reaches a value high enough, the valves that connect it to the main pipe 11 and 12 closes back, the one that connects it to pipe 13 opens again and normal operation resumes.

One or several liquid traps 21 can be installed to collect and drain liquid that may be present in the medium flowing in the second main pipe 12. These traps could for instance be advantageously placed at locations where there is a local minimum in elevation in the network. Such a trap comprises a receiver connected via a pipe to the second main pipe 12 and connected via an automatic valve to the third main pipe 13. The arrangement is made in such a way as to ensure that the automatic valve will drain liquid from the receiver and not gas. During normal operation the valve is closed and gradually the receiver of the trap will fill up with the collected liquid. Once the liquid level in the receiver reaches a value high enough, the automatic valve opens and drains the liquid towards the liquid return pipe 13. When the liquid level in the receiver of the trap reaches a value low enough the valve closes back. It is also possible to use a modulating valve commanded via a suitable control loop, to stabilize the level of liquid in the receiver at a desired value. With the latter option the drainage of the liquid would be a continuous process instead of a batch one.

In order to further increase the reliability of the system one can help the pump 5 be fed fully with liquid at its inlet (which equivalent to say that the net positive suction head available must be higher than the net positive suction head required by the pump) using a subcooling apparatus 6 located either between the low-pressure receiver 3 and the pump 5 or alternatively directly in the receiver. Alternatively it is also possible to impose some subcooling at the anti-flash condenser outlet 7 either by the mean of a controlled valve that would be operated in such a way as to impose actively said subcooling, or by the mean of a syphon at the anti-flash condenser outlet that would guarantee that at any time the bottom of the anti- flash condenser is filled with an appropriate amount of liquid that will be subcooled. Another advantage of providing subcooled liquid at the pump inlet is to reduce the necessary height difference between the bottom of the receiver and the inlet of the pump.

Similarly to the anti-flash condenser 7, the cooling circuit of the subcooling apparatus 6 can be fed directly if a source cold enough is available. Alternatively, it can also be fed by the cold source of heat pump apparatus 8, It may be advantageous in term of space to use the same heat pump apparatus for both the subcooling apparatus and the anti-flash condenser. The heat exchangers can be connected in series or in parallel to the heat pump apparatus and/or the cooling source.

In the case where a heat pump apparatus 8 is used, during its operation, the waste heat discharged at its hot sink can advantageously be used to preheat the source before it enters the intermediate pressure evaporator 9a. Alternatively it is also possible to install a dedicated flooded evaporator on the intermediate pressure receiver 1. It is particularly well suited since the maximum load on the anti-flash condenser 7 and subcooling apparatus 6 will occur simultaneously with the maximum demand for gas to be provided by the plant, through the second main pipe 12, to the concerned end-user locations, 15. Meaning that said discharged heat can always be fully valorised within the system.

The liquid from the low-pressure receiver 3 is pumped back into the intermediate pressure receiver 1 using the low-pressure liquid pump 5. As an alternative to the anti-flash condenser 7 it is possible to use a compressor 4 that extracts the flash gas from the low-pressure receiver compress it and send it into the intermediate pressure receiver 1 or directly into the second pipe 12. In any case the liquid pump 5 is still required and the beneficial effects of having a subcooling apparatus 6 remains even in the absence of the anti-flash condenser 7. At the intermediate pressure, the gaseous phase from the gas line and/or in the intermediate pressure receiver can be condensed 10 (the case when gas comes back to the plant from the pipe 12) or the liquid be evaporated 9a in the intermediate pressure receiver (the case when gas is sent from the plant into the pipe 12). Depending on the temperature of the source available for the evaporator’s heating circuit 9a, respectively the condenser’s cooling circuit 10, said source can either feed directly those heat exchangers or a heat pump apparatus, respectively a refrigeration apparatus 17 (see figure 3), can be used to supply said heat exchangers, thus decoupling the saturation temperature of the intermediate pressure gas in the network to that of the source. The source feeding the evaporator’s heating circuit 9a can also be pre-heated in heat exchanger 9b using the heat available from the heat pump apparatus 8. In cases where the temperature of the source varies significantly over time, a combination of direct heat exchange and use of a heat pumping apparatus or a refrigeration apparatus can be realised advantageously. As an alternative to a separate refrigeration apparatus 17 that would interface the intermediate pressure receiver to a source with a temperature too hot for direct condensation, it is possible to use a compressor 16 that extract gas from the intermediate pressure receiver, compress it and sends it to the condenser 10 where it will condense and be sent back through an expansion valve to the intermediate pressure receiver 1. In term of energy efficiency, this solution may be especially of interest when the temperature difference required between the source available and the desired saturation temperature in the intermediate pressure receiver is relatively small. In a rather similar way, it is possible to decouple the temperature in the intermediate pressure receiver 1 from that in the evaporator 9 by extracting liquid from said receiver, expand it in a valve, evaporate the liquid in said evaporator and recompress it and send it back to the intermediate receiver. Here also, this solution might be of a particular interest if the temperature difference required is relatively small A pump 2 is also used to extract from the intermediate pressure receiver 1, pressurise and send the liquid demanded by the end-user locations via the first main pipe 11. A subcooling apparatus may also be installed for improving the performance and reliability of said pump as well as reducing the static head required. The subcooling can be imposed and the cooling be provided by means analogous to those described for the subcooling apparatus at the low pressure 6. As an alternative to both methods of elimination of the flash gas in the low-pressure receiver 3, namely via the use of the anti-flash condenser 7 or the low-pressure compressor 4, it is possible to provide the same functionality using a suitable arrangement of the receiver 1 and 3 namely: - By installing the intermediate pressure receiver 1 at a slightly higher elevation than that of the low-pressure receiver 3, in order to allow for a gravity driven gas purge of the latter - By installing two automated Receiver Isolation Valves of the low-pressure receiver 22 one located on the third main pipe 13 and one on the pipe upstream of the low- pressure pump 5 - By installing two automated receiver flash gas purge valves 23, one located on a pipe that connects a liquid filled part of the intermediate pressure receiver 1 to the low- pressure receiver 3 and the other located on a pipe that connects a gas filled part of the intermediate pressure receiver 1 to a part of the low-pressure receiver 3 that also contains gas. During normal operation, the receiver isolation valves 22 are open and the receiver flash gas purge valves 23 are closed. Ideally the third main pipe 13 returns saturated liquid or even slightly subcooled, however because of pressure drops, thermal energy input from the environment and/or possible injection of medium in an inadequate thermodynamic state from end-user locations 15, it can be expected that some gas is also returned to the receiver 3. Gradually the gas will build up in said receiver until the liquid level reaches a value low enough to trigger a purge cycle by closing the isolation valves 22 and opening the purge valves 23. As a result, the pressure of the low-pressure receiver 3 will rise up to that of the intermediate pressure receiver 1 and thanks to the lower density of the gas with respect to that of the liquid phase, the volume of gas in the receiver 3 will migrate through the purge valve 23 into receiver 1 and be replaced by liquid flowing down from the intermediate pressure receiver 1 into the low-pressure receiver 3. Once the liquid level in the low-pressure receiver 3 reaches a value high enough the purge valves 23 close back and the isolation valve 22 reopen and normal operation can resume. During the time gas is being purged from the low-pressure receiver 3, the flow from the third main pipe 13 and the flow through the low-pressure pump 5 are both interrupted. This may be detrimental to the stability of operation of the whole system but can be overcome by installing in parallel two or more aggregates that comprise the receiver low-pressure 3, the isolation valves 22 and the purge valves 23 (and possibly the subcooling apparatus 6). That way it is possible to continue operating the system while one of the aggregates proceeds to a purge cycle. Alternatively (see figure 4), it is also possible to change the geometry of the low - pressure receiver 3 by having one part of the volume of said receiver, preferably smaller, installed above the main part of the low-pressure receiver 3, both parts being linked by a pipe on which is installed the receiver isolation valve 22. When the said valve is open both volumes constitute the low-pressure receiver 3 and during normal operation the gas coming from the third main pipe 13 will accumulate within the top part of said receiver by buoyancy. Once the liquid level in the top part of low-pressure receiver 3 reaches a value low enough because of gas build-up in said part, a purge cycle analogue to the one described previously is carried out by closing the receiver isolation valve 22 and opening the purge valves 23. Once the liquid level in the top part of the low-pressure receiver 3 reaches a value high enough, valves 23 close back and the valve 22 opens up again to resume normal operation. The advantage of this version of the receiver gas purge is to avoid the interruption of the flow coming from the third main pipe 13 and going to the pump 5 while also avoiding the necessity of putting aggregates in parallel as described earlier. Moreover, the number of isolation valves 22 is reduced to only one valve. Note that the purge valves 23 can also be installed on pipes that link the low-pressure receiver 3 to respectively the first main pipe 11 and second main pipe 12 instead of the intermediate pressure receiver 1. This latter solution could be advantageous if it is not possible to install the intermediate pressure receiver 1 slightly above low-pressure receiver 3. The invention is of course not limited to those four illustrated examples but to any alternative covered by the claims.