A NEW LNG BASELOAD PROCESS AND THE MANUFACTURING OF THE MAIN HEAT EXCHANGERS

Transcription

1 A NEW LNG BASELOAD PROCESS AND THE MANUFACTURING OF THE MAIN HEAT EXCHANGERS UN NOUVEAU PROCEDE DE LIQUEFACTION DU GAZ NATUREL ET LA FABRICATION DES PRINCIPAUX ECHANGEURS THERMIQUES Wolfgang Foerg Wilfried Bach Rudolf Stockmann Linde AG, Process Engineering and Contracting Division (Linde) Roy Scott Heiersted Pentti Paurola Dr. Arne Olav Fredheim Den norske stats oljeselskap a.s. (Statoil) ABSTRACT The LNG industry needs innovative ideas to enhance reduction in cost and schedule. In this context, Statoil and Linde have entered into a long-term LNG Technology Alliance, aiming at more cost effective technology, project execution and operation of LNG Baseload Plants. As a prerequisite, innovative LNG process design and careful machinery selection have to be complemented by the capability of producing large heat exchangers for the cryogenic process section. Accordingly, the Alliance has developed a new LNG process resulting in a plant concept which allows significant reduction of plant investment cost and increased thermal efficiency. The Baseload Plant will be equipped with LNG heat exchangers from Linde's own workshops. The design and manufacturing of these heat exchangers are based on reputable scientific methods and the experience of a large number of gas separation and liquefaction plants where spiral wound as well as plate-fin heat exchangers have been tailor-made for each specific application. The paper will present important results of the Alliance's work. Also the technical and commercial features of the LNG process and plant design will be evaluated and compared with the present state-of-the-art technology

2 RESUME L'industrie du GNL a besoin d'idées innovatives pour aider à la réduction des coûts et des délais. C'est dans ce cadre que Statoil et Linde ont conclu une alliance à long terme pour le développement d'une technologie de liquéfaction du gaz naturel, et pour une optimisation des coûts du procédé, de l'exécution des projets, et de l'exploitation des usines de GNL. Comme au préalable, la conception d'un procédé innovatif de liquéfaction du gaz naturel et le choix ciblé des équipements, devront être associés à la capacité de produire des échangeurs thermiques de grande capacité pour la section cryogénique du procédé. Ainsi, l'alliance Statoil-Linde a mis au point un nouveau procédé de GNL permettant à la fois une réduction remarquable des coûts d'investissement de l'installation et une augmentation du rendement thermique. L'usine de liquéfaction de gaz naturel sera équipée d'échangeurs thermiques fabriqués dans les propres ateliers de Linde. La conception et la fabrication de ces échangeurs thermiques sont basées sur des méthodes de calcul éprouvées et l'expérience de nombreux projets de séparation ou liquéfaction de gaz où des échangeurs bobinés ou à plaques sont faits sur mesure pour chaque cas spécifique. Cet article va présenter des résultats importants des travaux de l'alliance Statoil-Linde. Il évalue et compare les aspects techniques et commerciaux du procédé mis au point ainsi que la conception de l'usine avec les règles de l'art actuelles

3 A NEW LNG BASELOAD PROCESS AND THE MANUFACTURING OF THE MAIN HEAT EXCHANGERS INTRODUCTION The formation of the Statoil - Linde LNG Technology Alliance reflects the LNG industry's desire for reduced costs and schedules for Baseload Plants. The evaluation of existing liquefaction processes led to the development of a new LNG Baseload Process. The capability of manufacturing large cryogenic heat exchangers is necessary for building LNG plants, since such heat exchangers cannot be purchased on a competitive basis. The installation of the LNG plant on a purpose built barge will shorten the construction time under specific conditions set by remote locations. LNG PROCESS DESIGN Historically, we can distinguish between three different liquefaction processes which have been built and operated until today: the Classical Cascade the Single Flow Mixed Refrigerant and the Propane Precooled Mixed Refrigerant Process. The Classical Cascade has been established at three locations: Process Year of Start-up Arzew Algeria TEAL 1965 [1] Kenai Alaska Phillips Petroleum 1969 [2] Point Fortin Trinidad & Tobago Phillips Petroleum 1999 [3] The Single Flow Mixed Refrigerant Process has also been used three times: Process Year of Start-up Marsa el Brega Libya APCI 1970 [4] [5] Skikda Algeria TEALARC 1972 [6] Skikda Algeria PRICO 1981 [7] These three latter processes may have remarkable differences within their designs, but they have in common that one mixed refrigerant stream is compressed by one single compressor. The most successful process so far, however, has been the Propane Precooled Mixed Refrigerant Process of APCI, which has been used with start-up dates from 1972 to 1999 in eight different countries [8]

4 The liquefaction plant represents approx. 50 % of the investment cost of the entire LNG value chain. The Alliance therefore scrutinized and compared the existing processes in order to find a benchmark for further improvements. The conditions for the comparison were set as follows: Natural gas composition (in mol %): N 2 2.0; C ; C 2 5.5; C 3 2.5; C Natural gas inlet temperature 13 C, inlet pressure 60 bar, gross flow rate 4 million tons per annum of LNG; cooling water temperature 6 C. Heavy hydrocarbon removal and N 2 -rejection were not considered in this evaluation. It was assumed that their influence on each process would be similar, which means they would not affect the process comparison. NG E1A E1B 1 E1C C1 E2A E2B 2A 2B E2C E2F E2E E2D C2 E3D E3A X1 3A/3B C3 E3B E3C LNG Fig.1: Classical Cascade Process 2.6 4

5 Fig. 1 shows the Classical Cascade Process (CCP). A three-stage propane precooling cycle is followed by a three-stage ethylene liquefaction cycle and a three-stage methane subcooling cycle. While propane is compressed leaving the different suction drums at its dewpoint, ethylene and methane are vaporized and superheated before being compressed. Fig. 2 shows the Single Flow Mixed Refrigerant Process (SFMRP) which was the result of extensive optimization work. It was not the intention to compare the three NG E1 E2 E3 E4 LNG D4 X1 1 C1 D1 D2 D3 2A P1 C2 2B Fig.2: Single Flow Mixed Refrigerant Process different processes mentioned before, but we have chosen our own design as representative for this type of process. The mixed refrigerant is composed of N 2, CH 4, C 2 H 6, C 3 H 8 and n-c 4 H 10. The condensate from separator D1 downstream the first compressor stage C1 and water cooler 1 is pumped to the discharge of the second compressor stage. The condensate from separator D2 of this stage is used as refrigerant in the first cryogenic heat exchanger E1. The condensate from the third compressor stage is collected in separator D3 and used as refrigerant in the second cryogenic heat exchanger E2. The gas from the third compressor stage is partially condensed in the first cryogenic heat exchanger E1. The condensate from separator D4 is subcooled, expanded in liquid expansion turbine X1 and used as refrigerant in the third cryogenic heat exchanger E3. The gas from separator D4 is condensed in the third cryogenic heat exchanger E3, subcooled in the fourth cryogenic heat exchanger E4 and then used for final cooling in the fourth cryogenic heat exchanger E4 after throttle expansion. All refrigerants are subsequently vaporized and warmed up in heat exchangers E4, E3, E2 and E1 before being compressed in compressor C1/C2. E1A E1B E1C NG C1 1 Fig. 3 depicts the well-known Propane Precooled Mixed Refrigerant Process (C3MRC). Precooling is achieved by a three-stage propane cycle compressor C1 and precooling heat exchangers E1A, E1B and E1C. Liquefaction and subcooling is accomplished by two-stage mixed refrigerant compressor C2/C3, separator D1, liquefier E2A and subcooler E2B. For large plant capacities a liquid turbine X1 is advantageous. D1 2 3A 3B E2A X1 C2 C3 E2B LNG Fig.3: Propane Precooled Mixed Refrigerant Process 2.6 5

6 NG LNG X1 E2 E3 E1A E1B C2 C3 C1 1 2A/2B 3A/3B Fig.4: Mixed Fluid Cascade Process Fig. 4 represents a sketch of the Alliance's LNG Baseload Process consisting of three mixed refrigerant cycles, called Mixed Fluid Cascade Process (MFCP). The precooling cycle consisting of a mixture of C 2 H 6 and C 3 H 8 is compressed in compressor C1, liquefied in sea water cooler 1 and subcooled in cryogenic heat exchanger E1A. One part is throttled to an intermediate pressure and used as refrigerant in E1A. The other part is further subcooled in heat exchanger E1B, throttled to the suction pressure of compressor C1 and used as refrigerant in heat exchanger E1B. The liquefaction cycle is compressed in compressor C2, cooled in sea water coolers 2A and 2B, further cooled in heat exchangers E1A, E1B and E2. It is throttled and used as a refrigerant in liquefier E2. The subcooling cycle is compressed in compressor C3, cooled in sea water coolers 3A and 3B, further cooled in heat exchangers E1A, E1B, E2 and E3, expanded in liquid turbine X1 and used as refrigerant in subcooler E3. All compressor suction fluids are slightly superheated above their dewpoints. A comparison of the main data of the above processes is given in the following table: compressor shaft power at 100 % adiabatic efficiency MFCP C3MR SFMRP CCP C % Heating surface cooling water section % Heating surface refrigerant section % Heating surface total % number of compressor casings number of suction lines maximum flow in suction line approx. m³/h eff. 120, , , ,000 Special effort was made to investigate the different processes on a comparable basis. Linde's proprietary design optimization program Optisim was used for this purpose [9]. It was attempted to have similar heat transfer surfaces for all processes within the cooling 2.6 6

7 water and the cryogenic section. This was not always possible as the limitation of a minimum temperature pinch was given priority. Under these assumptions the compressor shaft power at 100 % adiabatic efficiency for the refrigeration cycles turned out to be 70.4 MW for the Mixed Fluid Cascade Process. If one compares the compressor shaft power with real adiabatic efficiencies the advantage of the MFCP versus the C3MRC may disappear, because no axial machine can be applied at the MFCP, while compressor C2 within the C3MRC can be built as an axial machine, with higher adiabatic efficiency. In spite of this fact we see an advantage for the MFCP, since heat exchangers E2 and E3 of this process are of similar size and well within the limits of manufacturability of spiral wound heat exchangers. With other words, those heat exchangers are not the limiting factor for the size of one liquefaction train. As far as the SFMRP is concerned we believe that the limits of manufacturability are exceeded in several areas, e.g., suction line, separators D1 to D4 and compressor C1. Therefore, the SFMRP is suitable for smaller train capacities than 4 million tons per annum only. MANUFACTURING OF CRYOGENIC HEAT EXCHANGERS For the cryogenic process section Plate-Fin Heat Exchangers as well as Spiral Wound Heat Exchangers can be applied. Aluminum Plate-Fin Heat Exchangers (PFHE) have been installed in air separation -, in gas separation - and in the cryogenic section of ethylene plants as well as natural gas separation - and LNG peakshaving plants for several decades. For approx. 20 years Linde has been manufacturing such heat exchangers [10] using the vacuum-brazing process as shown in Fig. 5. Raw materials for separator plates, fins and side-bars are measured, cut and stamped. All parts are washed to remove oil, dried and stacked. They are brazed in a vacuum furnace under precise temperature control at approx. 600 C. Thereafter, headers and nozzles are welded to the block to complete the heat exchanger. Before shipment the final product is subjected to X-ray-, helium leakage-, flow- and pressure testing. More than 4000 blocks have been built until now. PFHEs are well-suited for applications in LNG Baseload Plants. They are, however, susceptible to large and rapid temperature changes and are therefore regarded as less robust in comparison to Spiral Wound Heat Exchangers. Also the two-phase flow distribution to a battery of blocks is considered to be more difficult when compared to a single Spiral Wound Heat Exchanger

8 Fig.5: Manufacturing of Vacuum Brazed Aluminium Plate-Fin Heat Exchangers In the precooling section of an LNG Baseload Plant, however, vacuum-brazed aluminum PFHEs are a good choice. Fig. 6 shows four assemblies of four blocks each, finally forming one heat exchanger unit

9 Fig.6: 4 x 4 Blocks to be Assembled to Form one Heat Exchanger Spiral Wound Heat Exchangers (SWHE) or Coiled Tubular Heat Exchangers [11] as they are called sometimes have been used in the cryogenic industry since the early days. The heat exchanger used by Carl von Linde when he liquefied air on an industrial scale for the first time in Munich in May 1895 consisted of two concentric tubes, which were wound to form a coil [12]. Until today, more than thousand SWHEs have been manufactured in Linde's workshops. Copper has been the traditional material for many decades. It has been almost completely replaced by stainless and low temperature alloy steels and by aluminum. For weight reasons and since the service can be kept clean and free from corrosive substances aluminum is the material of choice for the liquefier and subcooler of an LNG Baseload Plant. Many of the construction details and fabrication procedures such as tube support system and tube spacing arrangements are proven design gained on many heat exchangers in low temperature service for more than 25 years. For the thermodynamic, hydraulic and geometrical design Linde's proprietary computer program Genius is used [13]. Process data as flows, pressures, temperatures and temperature differences are provided by the process calculation using Optisim. Pressure drops are set by an iterative optimization. Genius determines the temperature and pressure profiles of the individual streams by calculating the heat transfer coefficients, pressure drops and temperature differentials as driving forces for discrete elements. The number of elements used in the calculation is determined dynamically and depends on the accuracy required and by the non-linearity of the enthalpy-temperature curves of the individual streams. The dew and bubble points as well as the composition of each stream are taken 2.6 9

10 into account by the simulation. All methods used for heat transfer and pressure drop implemented into the program have been carefully tested. Our own research is complemented by evaluation of literature (e.g. [14], [15], [16]), and cooperation with major research and development institutions like Heat Transfer Research, Inc. (HTRI) and Heat Transfer and Fluid Flow Service (HTFS). Especially for falling film evaporation at the shell-side of the SWHE Linde and Statoil [17] have performed their own measurements and evaluated available literature (e.g. [18]) and produced corresponding calculation methods. As final product, Genius calculates the number and length of the tubes for the individual streams, the number of layers, the dimensions of the spacer bars and the distribution of the tubes to the different layers, resulting in the geometry of the bundle. 1. Mandrel 1.1 Race Ring Race Ring 3. Shell Fabrication 3.1 Lower Part 1.2 Tubesheets Support Arms Tubesheets 3.2 Upper Part Liquid Distribution Tray 2. Tube Bundle Winding 2.1 First Tube Layer 3.3 Bundle Assembly 2.2 First Tube Layer 3.4 Completing Second Tube Layer 4. Final Position 2.3 Tubing Completed 2.4 Shroud to bundle 2.5 Bundle Completing Bonnets Liquid Distribution Ring Fig.7: Manufacturing Procedure for Spiral Wound Heat Exchangers

11 The manufacturing procedure of the SWHE is as follows (Fig.7): It starts with the core cylinder or mandrel (1.1), which can be rotated on a winding bench by means of two race rings in horizontal position. Support arms and liquid distribution trays are connected to what will be the upper side of the heat exchanger (1.2) when it is erected to its final vertical position. Tube sheets are brought into their position at both ends of the heat exchanger. Spacer bars will keep the designed distance between the mandrel and the first layer of tubes. Now the winding of the inner layer of tubes can begin. Therefore, each tube of this layer is inserted into its tube sheet at one side, wound onto the mandrel and then inserted into the corresponding tube sheet at the other side (2.1). When the winding of the first layer is performed from left to right, the next layer is performed form right to left (2.2), and so on. Proper distance between the individual layers is kept by spacer bars (2.3). To avoid a by-pass at the shell side the bundle is wrapped into a shroud (2.4). After the tubes have been welded to the tube sheets, bonnets are welded to the tube sheets (2.5). Parallel to the manufacturing of the bundle the lower (3.1) and the upper part (3.2) of the shell are being fabricated. During assembly the completed bundle is inserted into the lower part of the shell and the shroud is connected to the shell (3.3). The upper part is added, the closing seams and the nozzles are welded to the shell (3.4). Fig. 7.4 shows the SWHE in its final upright position. Fig.8: Three LNG SWHE s during Manufacturing on Winding Benches

12 Fig. 8 shows on the left-hand side a mandrel equipped with tube sheets, support arms and liquid distribution tray. In the foreground and in the background two bundles during different phases of the winding process can be seen. PROCESS PLANT ON PURPOSE BUILT BARGE During the work of the Statoil - Linde LNG Technology Alliance, it was assumed that future LNG Baseload Plants are to be installed in remote locations without the availability of infrastructure and subjected to extreme weather conditions. It was envisaged at an early stage that for this purpose a conventional construction approach would not be cost and time effective. Therefore, one of the most fundamental project decisions was to install the major part of the LNG Process Plant in a most compact format on a purpose built barge (Fig. 9), thereby minimizing construction work on site. The installation takes place within a strategically well located shipyard making use of local skilled labor force and weather protected dock facilities. This in turn meant that basic engineering, design methods and project execution strategies had to be tailored to meet the specific constraints of this barge-mounted concept. Fig.9: LNG Plant on Purpose Built Barge The barge serves as a permanent base and foundation for the LNG Process Plant, both during construction and plant operation. Likewise the barge is also used as a transportation vessel, when the fully tested, mechanically completed and precommissioned plant is towed from the construction/assembly yard to the production site. Here the barge is brought into a prepared dock and settled down on ground. The dock will then be closed to seaside, filled with gravel and hooked up with the other site systems. It was also realized during the work that an alliance type of project execution and contract strategy is a necessity to accomplish construction and scheduling flexibility, which is required to achieve the reduced cost and schedule of the plant

13 CONCLUSION The work done by the Statoil - Linde LNG Technology Alliance with respect to process design, selection of main compressors and drivers, manufacturing of cryogenic heat exchangers along with the installation of the process plant on a purpose built barge has led to significant savings in investment cost and a considerable shortening of the project execution time. REFERENCES CITED [1] Operating Experience of the Arzew Plant Pierot, M. LNG 1, Chicago, Illinois, USA, April 7-12, 1968 Session No. 2, Paper 10b [2] Phillips Optimized Cascade LNG Process Houser, C. G.; Krusen, L. C. (Phillips Petroleum Co.) Gastech 96, 17 th Int. LNG/LPG Conf., Vienna, Dec. 3-6, 1996 Conference Papers [3] The LNG Industry 1996; GIIGNL; page 9 [4] Esso Libya Venture Latimer, D. M. LNG 1, Chicago, Illinois, USA, April 7-12, 1968 Session No. 3, Paper 15 [5] The Design, Fabrication and Operation of Large Cryogenic Heat Exchangers Gaumer, L. S.; Geist, J. M.; Harnett, G. J.; Pfannenstiel, L. L. LNG 3, Washington D. C., Sept , 1972 Session II, Paper 15 [6] Experience of Arzew and its Effect on the Design of the Skikda Natural Gas Liquefaction Plant Bourget, J. M. LNG 3, Washington D. C., Sept , 1972 Session V, Paper 6 [7] PRICO - A Simple, Flexible Proven Approach to Natural Gas Liquefaction Price, B. C.; Mortko, R. A. (Pritchard Corp.) Gastech '96, 17 th Int. LNG/LPG Conf., Vienna, Dec. 3-6, 1996 Conference Papers [8] The Air Products Propane Precooled/Mixed Refrigerant LNG Process Bronfenbrenner, James C. (Air Products and Chemicals, Inc.) LNG Journal, Nov./Dec. 1996, page [9] The Design of Optimal Air Separation and Liquefaction Processes with the OPTISIM equation-oriented Simulator and its Application to on-line and offline Plant Optimization

14 Burr, Peter S. AIChE Spring National Meeting, Houston, Texas, April 7-11, 1991 Paper 50a [10] The Manufacture of Plate-Fin Heat Exchangers Diery, W. Linde Reports on Science and Technology 37/1984 [11] Coiled Tubular Heat Exchangers Scholz, W. H. Linde Reports on Science and Technology 18/1973 [12] The History of Air Separation Foerg, W. MUST 1996, Refrigeration Science and Technology Proceedings Munich (Germany) Oct , 1996 [13] Optimised Calculation of Helical-Coiled Heat Exchangers in LNG Plants Steinbauer, M.; Hecht T. Eurogas 96 Conference, Trondheim, Norway, June 3-5, 1996 [14] Coiled Tubular Heat Exchangers Abadzic, E. E.; Scholz, H. W. Advances in Cryogenic Engineering, Vol. 18, Plenum Press, 1973 [15] Waermeuebertragung im Gegenstrom, Gleichstrom und Kreuzstrom Hausen, H. Springer Verlag, 1950 [16] Condensation of Hydrocarbon Mixtures in Coil-Wound LNG Heat Exchangers Neeraas, B. O. Ph. D. Thesis, University of Trondheim, Norwegian Institute of Technology, 1993 [17] Thermal Design of Coil-Wound LNG Heat Exchangers Shell-Side Heat Transfer and Pressure Drop Fredheim, A. O. Ph. D. Thesis, University of Trondheim, Norwegian Institute of Technology, 1994 [18] Echange de Chaleur et Pertes de Charges en Ecoulement Diphasique dans la Calandre des Echangeurs Bobines Barbe et al. Proceedings of the XIII International Congress on Refrigeration, Vol