Improvements in Nitrogen Rejection Unit Performance with Changing Gas Compositions.

Improvements in Nitrogen Rejection Unit Performance with Changing Gas Compositions. Gas Processors Association Annual Conference. Warsaw 21 st 23 rd September 2005 Katarzyna Cholast, Andrzej Kociemba KRIO, Odolanów Harry Isalski Technical Director, Tractebel Gas Engineering John Heath Special Projects Manager, Ebara International Corporation, Cryodynamic Division ABSTRACT The KRIO Nitrogen Rejection Plant in Odolanów, Poland was constructed in the 1970s to upgrade a lean natural gas to pipeline specification and to produce liquid helium for sale. This plant has seen changes to the gas feed composition which necessitated modifications to it in order to continue economic performance and to reduce methane emissions. Several studies were undertaken to examine options for modifications that would satisfy environmental needs and changing market requirements. Whilst the feeds gas became richer in methane, the needs to retain plant flexibility became more important. The paper describes the history of the plant, the changes that have occurred over its years of operation, the decisions taken to implement the modifications and the final configuration of the cryogenic part of the process. The paper also covers the initial start-up trials, the results and the final acceptance test data. The paper discusses the operational aspects of the facility and the benefits in having the main modifications in the selected locations. The paper describes the use of Variable Frequency Drive (VFD) of liquid methane pumps to improve cold production in the process as the first modification. Also covered is the subsequent installation of LNG expanders, now an important part of every new LNG liquefaction plant, where expansion of methane-rich, sub-cooled liquid enters the two-phase region. The paper reports the successful operational experience of two such turbine Expanders implemented in the nitrogen rejection unit revamps in order to improve their performance. The implementation of these 2-phase expanders heralds a new chapter in the use of expanders in the LNG and general cryogenic industry. This paper also serves to update the audience with results of a further years operation of these turbines. 1

Improvements in Nitrogen Rejection Unit Performance with Changing Gas Compositions. INTRODUCTION Natural gases which contain a significant amount of nitrogen are often not saleable since they do not meet minimum heating value requirements of the grid into which they are compressed. As a result the feed gas will generally undergo processing, wherein heavier components such as heavy hydrocarbons, water and carbon dioxide are initially removed and the remaining stream containing nitrogen and methane, and also possibly containing lower boiling or more volatile components such as helium, is usually separated cryogenically when passing through a nitrogen rejection unit. The nitrogen rejection unit (NRU) comprises compact heat exchangers, pumps, cryogenic distillation columns and Joule-Thomson (J-T) valves linked together in a highly integrated process to carry out efficient separation into the required products. J-T valves are applied to reduce the pressure of streams entering the distillation columns in order to decrease the stream temperature and assist in the provision of refrigeration to the process. Two such plants were constructed in the 1970s in Poland to achieve this and subsequently, these were modified by the application of novel turbine expanders with Variable Speed Constant Frequency Control (VSCF) to accommodate a feed gas having a different composition to the original design range and the application of Variable Frequency Drive ( VFD) of the methane liquid pumps. HISTORY OF THE NITROGEN REJECTION FACILITIES IN POLAND The nitrogen rejection facility is located in south-western Poland in Odolanów. It comprises two trains each processing up to 136000 NCMH of raw feed gas. Figure 1 shows a block diagram of one of the processing trains and shows how it is linked to the supply and the sales gas compression. Figure 1 Block Diagram of one train of the Odolanów plant 2

Raw gas is admitted from the plant supply header into each train. The first processing unit in each train is an amine unit which removes the carbon dioxide from the gas down to about 50ppm by volume. The next process block is a thermally regenerated adsorption system which consists of two adsorbers and a regeneration system. Each of the adsorbers has both molecular sieve to remove water to less than 0,1ppm and an activated carbon guard bed to remove benzene and heavy hydrocarbons. The adsorbents are regenerated using hot, low pressure gas to drive off the water and heavy hydrocarbons. The adsorption and regeneration cycle is controlled by a pre-determined sequence using changeover valves. The dry gas is then admitted to the cryogenic unit which separates the nitrogen from the methane, producing a sales gas of appropriate quality for compression into a high pressure gas. In addition to nitrogen/methane separation, crude helium is also produced in the process which is subsequently passed to a helium purification unit and then liquefaction. Helium has always been a valuable product from this facility and the initial concentration was as high as 0.4% in the raw gas. The original incoming gas had a composition approximately as indicated in Table - 1 below which compares the composition today. Table 1. Feed Gas Composition Changes. Original Current Helium 0.4 vol% 0.2 vol% Nitrogen 42.7 vol% 33.0 vol% Methane 56.0 vol% 65.7 vol% Ethane 0.5 vol% 1.0 vol% Propane+ 0.1 vol% max. 0.1 vol% CO 2 0.3 vol% Water Saturated at inlet conditions. Pressure 5.6MPa Temperature 10 15 o C Flow 136000 NCMH The sales gas required from the plant had to contain less than 4% nitrogen whilst the waste nitrogen vented to atmosphere was required to have less that 1% methane. In addition, the helium recovery was required to be better than 85% having a final purity of 99.999% as liquid at -269 o C. The original design required that the sales gas emanated from the cryogenic unit at about 1.8MPa. Over the years of operation, the raw feed gas has gradually changed and today, has about 32 35 % nitrogen in the feed. In addition, the helium content has decreased to a little over 0.2%. Therefore, the refrigeration potential of the feed gas has been reduced because there is now proportionally less nitrogen to expand to near atmospheric and by difference, more methane that is produced at 1.8MPa. This has gradually caused more severe operational problems for the two plants. The plant was originally capable of producing small amounts of liquid methane or nitrogen for sale to third parties. All of this was severely impaired, or even curtailed, with the lowering of the nitrogen content in the feed gas. In addition, there was a further negative environmental effect in larger than normal methane losses in the waste gas. 3

In summary, there are several main factors associated with the change of initial composition of the feed gas : - The environmental impact of increased methane content in the waste gas - The economic loss as the consequence of methane emission - Less stable operation of the cryogenic unit. - Inability to produce liquid products. - As the feed gas pressure decreases, the plant may require expensive investment in a precompression step. It would also be desirable to improve the efficiency of the NRU operation so that even with the drop of the feed gas pressure there will be still enough cold produced in the process to be able to operate in a stable manner. The current feed gas pressure is about 5.3 MPa and is being maintained by adding new gas sources. As a result, various studies were carried out in the late 1990s and 2000 to address the problem and find some solutions. In addition, the helium, a high value product, has decreased in concentration in the feed gas. This made it imperative that any changes to the process would not have any negative effect on the helium recovery. The study work examined many conventional, radical and novel solutions. In order to understand these better the process is described below. BRIEF DESCRIPTION OF THE CRYOGENIC PROCESS In the original process, all energy needed for natural gas separation in low temperature units was provided by pressure reduction of the natural gas across Joule-Thomson valves. The process is shown in Figure 2. Figure 2. Process Flowscheme for the NRU Cryogenic Section. The dry feed gas is admitted into E1 of the cryogenic unit where it is cooled at 5.3MPa, condensed and sub-cooled before it is passed through JT1 for expansion to the lower column K1. The pressure in the lower column is about 2.1MPa and thus there is some flash into the lower column. The rich liquid from the base of K1 is then sub-cooled in E2 against 4

returning colder methane liquid and gaseous nitrogen streams to about -150 o C before passing through JT3 where the sub-cooled rich liquid expands to 0.2MPa, thereby creating some flash before entry into the upper column. The upper column operates at about 0.15MPa and carries out the main distillation between nitrogen and methane. Reboil to this column is provided by condensing nitrogen at 2.1MPa in the top part of the lower column. This condensed nitrogen is very pure having less than 10ppm methane, and is drawn from a downcomer seal as a sidestream from the lower column, sub-cooled in exchanger E3 before it is expanded across JT2 and admitted to the top of the upper column, K2. This nitrogen stream provides the necessary reflux to purify the nitrogen waste gas which returns through E3, E2 and finally, E1, before it is vented to atmosphere. Therefore, the expansion valves JT1,2 & 3 provide all of the refrigeration to the process. The methane product pump, P1, raises the pressure of the methane product from 0.2MPa to about 1.8MPa before the methane is evaporated and reheated in E2 and E1 respectively. Over the years of operating the plant, KRIO has seen the nitrogen reduce from 42% to about 32% now. As the nitrogen in the feed decreased, the amount of this nitrogen reflux also decreased in proportion to the methane that entered the upper column. Consequently, the distillation process did not perform as well as before and methane content in the waste gas increased. CONCEPT DEVELOPMENT Initial study work was carried out to select the most effective way of providing refrigeration to the upper column. The analysis included the options listed in Table 2 below. Improving the performance of the upper column was the key area which had to be addressed to reduce methane losses. If LNG or liquid nitrogen was also required, then more cold had to be supplied to the process. Dropping the column pressure with blowers, adding external cycles or using liquid nitrogen (Options 1, 4 & 5) were eliminated mainly for cost reasons, though Option 1 also had limited benefit. Simply adding distillation trays or high efficiency packing (Option 2) would marginally improve the methane losses in the overheads, but would have no effect on the cold production. Adding other separation equipment to condition the feed to the lower column (Option 3) was also felt to be prohibitively expensive and time consuming to implement. These options would be very expensive to implement and were discarded quickly. The use of gas expanders (Option 7) was seriously considered, but the cold production was not in the correct location, i.e. too warm, and hence this option was not utilised. Adding gas from high nitrogen fields was a complex issue and also a costly one. This could still be an option in the future but will require high investment costs and a long project schedule. Therefore, two options remained: VFD control of the methane pumps and installation of flashing liquid expanders and these were implemented in sequence. The interesting point about this approach was that the VFD control scheme implemented and tested on the methane level control in the upper column applied to the methane pumps was already fully proven in the VSCF control of the liquid expanders selected. 5

Table 2 Options Studied to improve cold production to upper column. Option Description Number 1 Lowering Pressure in Upper Column with Blower on the waste gas at the outlet of the cold box. 2 Addition of extra trays to the upper column or replacing its internals with high efficiency packing. 3 Addition of pre-separation column or separator(s) to increase nitrogen content into double column. 4 Adding a separate cycle to provide refrigeration. This could be a nitrogen cycle taking the waste gas as the working fluid. 5 Importing liquid Nitrogen to assist in cold production. 6 Modification of methane pump control scheme to VFD controller. 7 Application of gas expanders in various locations. 8 Application of liquid expanders in various locations: (a) feed to lower column, (b) rich liquid & (c) poor liquid 9 Bring in other gases with higher nitrogen content. Likely Cost/Benefit This would increase the relative volatilities hence needing less reflux to the column. High Capital and Operating Cost. Some benefit but reduces overall plant reliability. Limited benefit. Long duration shutdown to implement and very high cost to implement. This is a complete revamp with high cost, long shut-down and thus considered impractical. This is potentially a good solution since it gives the ability to produce large amounts of LNG or liquid nitrogen. Very High Capital and Operating Cost. Complex solution which requires strong market drive to achieve the benefit of the high CAPEX. Long schedule. High operating cost. Simple implementation and with good benefit. Low cost, with good benefits, with limited scope for LNG production. Now operational and described in this paper. Difficult to provide cold at correct temperature level. Quite expensive. The (b) & (c) options provide cold at the appropriate temperature level. Limited sources of machinery. Testing was required. Option (b) now operational. These were not easy to implement at the time of the study work. This option also requires a high capital investment for pipelines & controls. Revamping the Methane Pump Control Scheme. The original plant used a pump spillback line to control level in the upper column. The pump basically operated at its full speed/load and the surplus flow was returned back to the upper column at 0.15MPa having been pumped to 2.1MPa. The original design of the pumps were such that the power absorbed was about 130kW. With a significant portion of the pumped liquid being returned via the level controller, about 30 40 kw of power was wasted across the level control valve and heat input to the process. The level control valve expanded this liquid back into the upper column from 2.1MPa to 0.15MPa. By adding the VFD control to adjust the speed of the motor, the correct amount of liquid methane could be withdrawn from the column without wasting power. The pumps themselves were capable of delivering a 6

higher than required head and it was possible to remove one of the stage impellers and still meet the demand. These modifications were successfully implemented in 2002. The benefits immediately showed with either lower methane losses in the waste gas or LNG production in small quantities. Other benefits observed were greatly improved plant stability and an immediate power saving of 30kW on each of the two plants. This system has been operational since early 2003. This gave the plant operations staff the confidence that this kind of control would operate successfully before utilising it on a more radical revamp i.e. the liquid turbine expanders. Original Scheme New Scheme L L M M VFD Control L Level Controller; M Electric Motor; VFD Variable Frequency Drive FIGURE (3) Pump Control scheme modification, before and after. Rich Liquid Expanders. There were three possibilities for adding liquid expanders in parallel with JT1, JT2 or JT3. Adding an expander at JT1 location would only be of partial benefit since the cold was not produced for the upper column (i.e. it was at too warm a temperature to be fully useful in reducing the methane level in the waste gas or producing cold for LNG production). The addition of an expander in parallel with JT2 would be of benefit, but the flow was not as great as the rich liquid flow so the location in parallel to JT3 was considered to have the best potential. Therefore a liquid expander which produced two-phases at its outlet was located in parallel with Joule-Thomson valve, JT3. Expansion turbines are inevitably more efficient since they carry out isentropic depressurisation which generates work (which can be extracted as drive power or electricity) instead of isenthalpic depressurisation across a Joule-Thomson valve, which generates no work.. The choice of this location for an expander meant that the liquid entering the expander was well sub-cooled ensuring a good quality single phase at the turbine inlet since it exits exchanger, E3, at below -150 o C at that point, and is at a pressure of 2.1 2.4 MPa. The outlet produces about 80 95% by volume of vapour. This issue represented a significant challenge from two aspects. The first is the two-phase flow stability for a fluid rising to the upper column entry point for all the cases. The second, and by far the most challenging, was the 7

design of an efficient turbine which converted not only the hydraulic energy into useful work, but also the gas expansion energy. The first issue was successfully addressed by correct selection of the fluid regime for all the practically expected cases, use of specially design piping and entry systems to prevent surges and vortices causing problems with pressure stability at the turbine outlet and flow stability at the upper column inlet. The second issue of two phases within the turbine itself had already been proven on several test programs with various companies dating as far back as the 1980s. However, no sustained commercial application of a similar nature was found. There was confidence in the success of the flashing flow from the turbine based on some tests done over the past 20 years in several cases with success over short periods of time but with less vapour generated at the turbine outlet. It was decided that the Ebara turbine lent itself well to producing power for export, since the generator is located inside the vessel that houses the turbine. The control of the unit was facilitated by Variable Speed, Constant Frequency (VSCF) control of the turbine speed making it act like a control valve. The level signal from the lower column, K1, was successfully used for the control of the turbine speed. With the rich liquid cooled down by up to 6 deg C more than from the control valve, JT3, more cold is routed to the upper column and the separation of methane from the nitrogen waste gas stream is considerably more effective. FIGURE 4) Picture of the VSCF control cabinet. 8

TWO-PHASE EXPANDER DESIGN CONCEPT Two-phase expander design concepts essentially follow existing single-phase turbine and expander technology. The hydraulic energy of the pressurized fluid is converted by first transforming it into kinetic energy, then into mechanical shaft power and finally to electrical energy through the use of an electrical power generator. The generator is submerged in the cryogenic liquid and mounted integrally with the expander on a common shaft. The cryogenic induction generator uses insulation systems specifically developed for cryogenic service giving submerged windings significantly superior dielectric and life properties. Thrust balancing and lubrication for the bearings is provided by an internally designed system which takes a small portion of the liquid passing into the turbine and routing it to the bearings. This produces some inefficiency, but the design of this system is far simpler than a typical external oil lubrication system. This was another attractive feature of the canned turbine design since the revamp required a smaller plot area, less connections to the main process (no needs for seal gas etc). Figure 5 Ebara Two-Phase Expander Cross Section. Figure 5 shows the cross section of a typical Ebara International Corporation cryogenic twophase submerged expander. The expander consists of a nozzle ring generating the rotational fluid flow, a radial inflow reaction turbine runner and a two-phase jet exducer. Symmetrical flow is achieved in the two-phase expander by utilising a vertical rotational axis to stabilize the flow and to minimize flow induced vibrations, with the direction of flow being upward to take advantage of the buoyant forces of the vapour bubbles. (Expanders with horizontal rotational axis generate asymmetric flow conditions which can result in higher vibration levels.) The hydraulic assembly is designed for continuously decreasing pressure to avoid any cavitation along the two-phase flow passage. The low vibration levels observed on site with the turbines operating at design flows support the design thinking very well. 9

FIELD EXPERIENCE USING TWO-PHASE EXPANDERS To upgrade low-methane natural gas by extracting undesired nitrogen, two Ebara two-phase expanders (each located in parallel to JT3, see Figure 2) were installed. During the early running, optimisation of the turbine runners and exducers was carried out in order to improve their power output and hence cold production giving successive improvements. The operations and maintenance staff gained excellent confidence in the operational and maintenance monitoring issues of the turbine as well as a good impression of the cryogenic unit behaviour and hence were able to contribute considerable guidance in process and turbine improvements. This gave an optimum process operating regime, a sound mechanical design and very high plant overall availability. Briefly, based upon the operational experience the following statements can be made: - The expanders required limited modification of the existing equipment and consequently their installation is normally very easy and retrofitting can be done very quickly. - The two-phase expanders have been in stable operation for more than 18500 hours now. Throughout that period regular inspections have shown no incipient failures in bearings or materials, vibration levels have been less than 20% of API 610 allowable limits. - The expanders operate surprisingly quietly; not being audible alongside neighbouring equipment of average noise level below 80dB. - The employed expanders have made the process really flexible in terms of its adjustment to changing mass flows, varying even by 100%. Even with such considerable changes they assure easy and precise regulation of level in the lower columns of both trains, which is of fundamental value for stable running of the process. - Due to the greater temperature difference achieved by the expander, the heat exchangers (particularly E3) operate in a more efficient and flexible way. - The use of this two-phase expander in place of JT3 allows the sales gas product from the NRU to be at higher outlet pressure, increasing by approximately 0.2MPa debottlenecking the compression. - The NRU has continued to produce either liquid nitrogen or LNG for sale to third parties. The Holy Grail in such duties is to extract more than just the hydraulic power from the expansion. For the duty in question, the hydraulic power that could be extracted by the turbine was estimated to be about 65kW. The actual power extracted from the turbine generator into the grid as export was between 80 and 85kW clearly demonstrating that gas expansion energy was being utilised effectively. When calculating the overall expansion, thermodynamic efficiency, the value was found to be relatively low (below 60%) compared with large gas turbines of today which regularly achieve over 80%. However, this is a relatively small machine, with an onerous duty which is also using the process fluid as bearing and generator coolant. BENEFITS TO THE PROCESS MODIFICATION During the modifications to the expander runners, the efficiency of the turbine rose to a high enough level that gas expansion energy was being utilised, and subsequently a test run was carried out on both the trains during the 4 th quarter of 2004. Table 3 below also shows recent plant data which demonstrates that there is no deterioration in turbine and hence plant performance after a further year or operation. 10

Table 3 Examples of plant data taken during the test runs and current Operation. Train Case LNG T/D Liq. N2 T/D Measured kw Waste CH4 % Date 1 1 21.6 0 82.2 0.93 25 Oct 04 1 2 0 45 80.9 >1 26 Oct 04 1 3 32.9 0 78.7 0.3 25 Oct 04 2 1 15.7 0 81.2 0.58 25 Oct 04 2 2 0 52 84.7 >1 26 Oct 04 2 3 22.6 0 80.5 1.0 25 Oct 04 1 1 - - - - Aug 2005 1 2 0 44 84 >1 Aug 2005 1 3 31.6 0 82 <1 Aug 2005 2 1 - - - - Aug 2005 2 2 0 48 81.6 >1 Aug 2005 2 3 34.8 0 83.5 <1 Aug 2005 Case 1 Minimum/no liquid production Case 2 Maximum liquid nitrogen production Case 3 Maximum LNG production It should be noted that during extraction of liquid nitrogen, reflux to the upper column is reduced, hence the rise in methane loss from the waste gas. Also, during 2005, market reasons dictated that the plant was not run without liquid production. Note that the original design was for 42% nitrogen in the feed. With the composition as tested, the waste gas methane content rose to above 3% with very poor plant stability and frequent operator intervention. In addition, the liquid production was severely impaired, and in some cases not possible when using just the J-T valves as cold producers. However, with the turbine and methane pump control scheme in operation, the following benefits were observed: - Because of the higher efficiency of the described process employing Ebara liquid twophase expansion turbines the upper column feed is colder which helps to further cool the upper column top section leading to a lower methane content in the waste gas, when processing the lower nitrogen content feed gas. - With the turbine in operation the NRU can run even with short-term carbon-dioxide increases without shutdown. The plant can then accept short term higher carbon dioxide concentration with no danger of plugging or consequent shut-down of the whole NRU. This added tolerance is only temporary, since the plant would have to be thawed out later on in any case. Whilst one would not design a new plant in this way, it gives operations staff more flexibility and a greater on-stream time for the facility. - The higher process efficiency allows easier operation at a lower feed gas pressure, though this is not a big issue at the moment. In case of reducing pressure of the feed gas from depleting sources one can postpone the decision to install an expensive pre-compression step. - Due to the high process efficiency, there is a possibility of withdrawing considerable amounts of low-pressure LNG or a liquid nitrogen stream, running the nitrogen methane separation in a stable manner at the same time. The possibility of producing LNG may be useful for Peak Shaving opportunities. If liquid nitrogen production is considered, one should be aware of the increased methane content in waste gas. - Employing liquid two-phase expansion turbines in the separation of nitrogen and methane will allow generation of energy that can be used for export or as a drive power for another duty. 11

- One of the most significant benefits of the turbine operation has been the enormous flexibility that the plants now have. The operators can easily switch from one mode of operation to another within a matter of a few hours and are therefore able to rapidly respond to market needs to achieve: - - Maximum sales gas output. - Maximum helium recovery. - Below 1% methane in the waste gas. - Up to 50t/day liquid nitrogen. - Up to 35 t/day LNG per train.. CONCLUSIONS The installation of the VFD control of the liquid methane pumps and subsequently the Ebara expanders fulfilled and in some instances, exceeded the initial design expectations and offered an excellent solution to ensure efficient operation when the plant is given feed gas with much lower than design nitrogen content. Apart from reducing emissions of methane from the plant the turbine has: Improved operational stability & availability with low nitrogen in the feed gas. Allowed greater LNG or liquid nitrogen production. The VFD control of the methane pumps has now been operating for a total of more than 38000 hours and the turbine expanders have now been in operation for over 18500 hours which demonstrate that the two modifications are an excellent way of adding refrigeration and plant stability to cryogenic processes. For more information please email 2PhaseExpander@btconnect.com 12

BIBLIOGRAPHY and REFERENCES - Ross, Greg; Davies, Simon; Vislie, Geirmund; Hays, Lance; "Reductions of Greenhouse Gas Emissions in Oil and Gas Production and Processing by Application of Biphase Turbines", 1996, www.mpptech.com/techpp/tech_home.htm - Hays, Lance, "History and Overview of Two-Phase Turbines", International Conference on Compressors and Their Systems", Institution of Mechanical Engineers, London, 1999. - Bond, Ted, "Replacement of Joule-Thomson Valves by Two-Phase Flow Turbines in Industrial Refrigeration Application", 2000, www.mpptech.com/techpp/tech_home.htm - Shively, R.A. and Miller, H., Development of a Submerged Winding Induction Generator for Cryogenic Applications, in Proceedings of the IEEE Electrical Insulation Conference, Anaheim, California, 2000. - Gebhart, Benjamin et al.; "Buoyancy-Induced Flows and Transport" Hemisphere Publishing Corporation, New York, 1988, ISBN 0-89116-728-5 - Boom, R.W. et al.; "Experimental Investigation of the Helium Two Phase Flow Pressure Drop Characteristics in Vertical Tubes", Proc. ICEC 7, pg 468-473, 1978 - Elliott, D.G.; Weinberg, E; "Acceleration of Liquids in Two-Phase Nozzles", Technical Report no.32-987, Jet Propulsion Laboratory, California Institute of Technology, Pasadena, California, USA, 1968 - Filina, N.N.; Weisend II, J.G.; "Cryogenic Two-Phase Flow: Applications to largescale systems", Cambridge University Press, 1996, ISBN 0-521-48192-9 - Vislie, Geirmund; Davies, Simon; Hays, Lance; "Further Developments of Biphase Rotary Separator Turbine", Paper presented at IBC Separation Systems Conference, May 1997, Oslo, Norway. - Jones, J.K. (1973), Upgrade Low BTU Gas, Hydrocarbon Processing, Sept., 193. - Isalski, W.H. (1989), Separation of Gases, Monographs on Cryogenics 5, Oxford Science Publications. - Cholast, K, et al, Two-Phase LNG Expanders, GPA Conference, Amsterdam, 24 Feb, 2005 13