trinidad triples its lng capacity.the engineering challenges
Transcription
trinidad triples its lng capacity.the engineering challenges
TRINIDAD TRIPLES ITS LNG CAPACITY—THE ENGINEERING CHALLENGES AND LESSONS LEARNED TRIPLEMENT DE LA CAPACITE DE TRINIDAD—LES DEFIS DE L’INGENIERIE ET LES LEÇONS TIREES Ali Riche Project Engineering Manager Trinidad LNG Project Bechtel Corporation Tony Diocee Engineering Manager Trinidad LNG Project Atlantic LNG Company of Trinidad & Tobago William Woodard Principal Process Engineer Phillips Petroleum Company ABSTRACT Work is well underway to triple the capacity of Trinidad's LNG industry. Key decisions to proceed with the plant expansion were made in 1999, the same year as first LNG production. The owners, contractors, and government entities faced the challenge of making this expansion move quickly and equitably on all sides. There is certainly a race-to-market competition for LNG in the Atlantic basin. The speed of this major investment decision surprised almost all interested parties. The market, once identified, required an additional 3+ million tonnes per year in 2002 and another 3+ million tonnes soon thereafter. The engineering challenges in maintaining a low unit cost for LNG and still moving quickly to more than triple the size of this facility are discussed . Lessons learned from Train 1 are incorporated. Little pre-investment was done with the construction of Train 1. Trains 2 and 3 were originally expected to be duplicates of Train 1. However, the development of the upgraded Frame 5D gas turbine provided an opportunity for increased capacity with less incremental cost . Train 1 was "designed to capacity" and therefore the fifteen percent additional power provided by the 5D design created numerous production bottlenecks that had to be overcome quickly and for little added cost. Geotechnical conditions for the expansion presented a new challenge. For Train 3 and the new LNG Tank locations, land reclamation and soil improvement were required to establish ground and a good working platform. With the fast track schedule dictated for the expansion, the project team worked diligently to come up with a land reclamation plan and ground improvement technology that will support the aggressive construction schedule. Trinidad Train1 LNG project demonstrated the economic viability of a single train LNG export plant - the approval of the new Trinidad LNG expansion project will achieve a new standard of lower cost per ton of LNG. PO-35.1 RESUME Le projet de tripler la capacité de production de GNL à Trinidad est à un stade bien avançé. Des décisions importantes ont été prises pour l’expansion de ce projet dès 1999, la même année que la première production de GNL. Les propriétaires, les entrepreneurs et les différentes entités du gouvernement ont dû relevé le défi de continuer l’expansion de ce projet de façon rapide et équitable pour tous les partis concernés. Il y a évidemment une compétition assez serrée sur le marché du GNL dans le bassin atlantique. La vitesse, avec laquelle cet investissement majeur a été décidé, a donc étonné presque tous les partis impliqués. Le marché, une fois identifié, a requis une production supplémentaire de plus de trois millions de tonnes par an, en an 2002, suivi rapidement d’une deuxième production de plus de trois millions de tonnes par an. Les défis d'ingénierie, qui ont permis de maintenir un coût de revient unitaire relativement bas pour le GNL et, en même temps, progresser rapidement afin de tripler la taille de ce site, sont discutés avec les leçons apprises du premier train. Très peu de préinvestissements ont été nécessaires pour la construction du premier train. Les deuxième et troisième trains devaient initialement être des répliques du premier. Cependant, le développement de la nouvelle turbine à gaz modèle 5D représentait une occasion tellement inespérée d’accroissement de capacité qu’elle fût retenue. Le premier train avait été "conçu à capacité" et, par conséquent, la puissance supplémentaire de quinze pour cent, créée par la conception de ce modèle 5D, a révélé de nouveaux problèmes limitants la production qui ont dû être résolus rapidement et pour un faible coût ajouté. Les conditions géo-techniques pour cette expansion ont présenté un nouveau défi, plus spécifiquement en ce qui concerne l’emplacement du troisième train et du nouveau réservoir GNL; la réutilisation des terres et l’amélioration du sol ont été nécessaires pour établir une bonne plate-forme de travail au sol. Afin de maintenir un programme très serré pour mener à bien cette expansion, l’équipe de projet a dû travailler de façon diligente afin de proposer un plan de réutilisation des terres et une technologie d'amélioration des sols qui supporteront le programme agressif de construction. Le premier train GNL à Trinidad a démontré la viabilité économique d’un site GNL avec un seul train; l’accord pour continuer l’expansion du site GNL à Trinidad va donc réaliser un nouveau record avec un coût de revient encore plus faible de production de GNL par tonne. PO-35.2 Process Development The liquefaction technology used for the Train 1 facility is the Phillips Optimized Cascade LNG Process. The same technology is applied to the Expansion project. The following schematic illustrates the major process units. Figure 1: Process Flow Schematic ATLANTIC LNG SIMPLIFIED PROCESS FLOW FUEL GAS PROPANE SYSTEM INLET FEED INLET SEPARATION AND NGL RECOVERY GAS TREATMENT ETHYLENE SYSTEM METHANE SYSTEM LIQUEFACTION VAPORS FROM STORAGE AND LOADING LNG STORAGE NGL STORAGE TO NGL PIPELINE TO LNG TANKER The decision to proceed with two additional trains originally called for duplication of the existing Train 1. Strong arguments could be made for duplication given the success of the Train I start-up and the urgency to provide additional product to a growing market. However, the decision was made to upgrade the refrigeration turbines from Frame 5C to Frame 5D. The Frame 5D gas turbine offered an excellent opportunity for increasing the capacity of the Train I design with little impact on the project schedule. It also provided the possibility of a further reduction in the unit cost for the produced LNG. The inclusion of an additional 10% throughput to a facility that had originally been “designed to capacity” presented quite an engineering challenge. To facilitate the process development a very detailed rating model of the Train I facility was developed. This incorporated the actual compressor curves from the Train I compressors and equipment parameters and hydraulic data from the Train I “as built”. This simulation package quickly and accurately identified the process bottlenecks. The process simulation was verified by performing a capacity test and the model was then used to predict potential bottlenecks at higher capacities. Capacity Test A capacity test was conducted on Oct 22 to 24, 1999 to determine the ultimate capacity of the facility and to help identify bottlenecks. The results of this work were used to optimize the Train 2/3 designs and to lay the foundation for starting on a de-bottlenecking exercise for Train 1. The same data points and simulations used for the performance test were used to evaluate the overall performance against the original design. Additional PO-35.3 information was gathered to evaluate hydraulic and equipment capacity limitations. Data collected around the air coolers and propane condenser indicated that the fan blade pitches could be adjusted slightly to improve the performance, this was carried out immediately after the capacity test with very good results. Liquid lines around the ethylene heat exchangers were identified as bottlenecks limiting some of the flexibility of the design. The data was then compared to simulations of the plant using Hysys which revealed possible leaking of recycle valves and liquid carryover from the propane chillers to the propane suction drums. Once the simulation was adjusted to account for these items the actual measured data matched very closely to the simulation results. The results of the capacity test indicate that in a short period after plant takeover Atlantic LNG operations staff learned to maximize LNG production. The following table compares the original design, performance test results and the capacity test results: Table 1: Production Comparisons Original Simulation New Simulation Performance Test Capacity Test Net Guarantee LNG Production (MMBtu/day) Fuel Gas Consumed (MMBtu/day) 418000 56425 416700 425239 436810 56300 55419 56312 Lessons Learned Utilization of the “lessons learned” practice was a key work process for developing Trains 2 and 3 design scope and in shaping the engineering, execution and construction of the expansion project. Meetings were held with Atlantic LNG operations staff and project E&C team early on during the FEED phase of the project to collect lessons learned. This process was ongoing during the early parts of the EPC phase of the project. Some key items were identified during this period, which allowed last minute design improvements to be incorporated. Additionally Bechtel and Phillips teams met to brainstorm and document the lessons learned during the engineering, procurement, construction and commissioning/startup phases Train 1 project. The integral team reviewed recommendations and adopted solutions as the basis for train 2 and 3 design. With the 15% additional horsepower provided by the Frame 5D’s, one of the first significant bottlenecks to be defined were the ethylene compressors. Working closely with the compressor vendor, a new rotor was designed which required no change to the compressor casing. This allowed the use of the additional available horsepower but minimized the effect on the rest of the ethylene refrigeration system. The propane condensers were also found to be a significant bottleneck due to the significant impact on propane compressor horsepower consumption. It was possible to increase the capacity of these air fin coolers by increasing the fan motor horsepower and PO-35.4 by making modifications to fans without affecting tube bundle size. Additional bays were added to the propane condensers to maintain condensing pressure during the hot part of the day. This required some piping modifications Overall plant pressure drop is very important to the efficiency of the LNG process. The Train I design was optimized hydraulically. Trains 2 and 3 design required a few selective changes to line sizes and, in some cases, piping layouts to efficiently allow for the increased capacity. The project team minimized these types of changes to maintain cost and schedule goals. The major pressure vessels, such as the CO2 absorber, dehydrators and mercury removal beds were determined to be adequate for the new design. However, some internal modifications were made. Some pieces of equipment were deleted and units combined for all three trains. These included one single defrost system for all three trains with a cleaner fuel gas supply, and a single fuel gas system for all three trains. These modifications will result in a more efficient plant with easier and faster start-up and shutdown times Ground Improvements Ground improvement for the expansion project presented a different type of challenge. The proposed Train 3 and the new LNG tank were located on reclaimed land which required extensive ground improvement technology be employed. The project team optimized the ground improvement technique that would support the very aggressive construction schedule. Cement Deep Soil Mixing (CDSM) technology was adopted for the ground improvement method. With this foundation system, cement is mixed into loose weaker soil to improve its properties and make it a much better foundation material versus the original soils. The CDSM technique was primarily developed in Japan for coastal structures sited on loose materials and prone to earthquake excitation and potential liquefaction. A CDSM foundation is constructed by making a series of interlocking rows and columns of improved soil. A grid pattern is then obtained which prohibits the unimproved from moving during an earthquake event. The improved cemented soil basically surrounds the weaker unimproved soil. The soil conditions at Trains 2 and 3 were highly variable. However, the most susceptible soil was the fill material in the upper 4 to 7 meters. The material below the fill was marine clay deposits which are not susceptible to liquefaction because of their cohesive behavior. The design for Trains 2 and 3 therefore required that the CDSM columns penetrate into these cohesive marine layers. The total length of the soil columns was about 8 to 9m. CDSM has resulted in Train 2 and 3 ground improvement being completed in a record four months. This has allowed the concrete placement subcontractor to mobilize and start earlier than planned and has improved the train 2 schedule by four weeks. PO-35.5 Figure 2: CDSM Train 2 Environmental Improvements Due to the close location of the plant site relative to the town of Pt Fortin the project was faced with utilizing additional measures to control noise from the facility. Bechtel specialists predicted noise levels at the plant fence and in the community using models to generate noise isopleths. The model was calibrated to actual operating data by conducting a noise survey. The model was then used to predict reductions in noise levels by applying noise mitigation measures at the sources. The three areas the project identified as targets were: 1. Add additional exhaust silencers on the refrigeration compressor drivers. 2. Utilize low noise fans on air coolers. 3. Insulate some compressor suction and discharge piping. Conclusion Staying with the same liquefaction technology and duplicating the train 1 design with some minor modifications has allowed Atlantic LNG, its shareholders and the government of Trinidad and Tobago to move quickly to make the decision to approve the Expansion project. Train 2 is expected to load the first ship by Dec 2002 and Train 3 by Aug 2003 as shown in Figure 3. The LNG production of Trains 2 and 3 is roughly 10% above train 1 capacity. This increase was achieved by utilizing additional refrigeration compressor horsepower and applying the lessons learned during the performance test and during the extensive capacity test exercise. PO-35.6 Table 2: Capacity Comparisons Original Simulation Train 2 and 3 Net Guarantee LNG Production (MMBtu/day) 418000 Fuel Gas Consumed (MMBtu/day) 56425 460320 57900 The unit cost based on design rates and EPC contract cost of all three trains is less than $200/mtpa. This does not include the feed gas pipelines as these are in the gas supplier’s scope of work. The design for the second and third trains of the Atlantic facility is believed to be an excellent compromise providing for the additional capacity while minimizing the amount of new engineering required and achieving a new standard of lower cost per ton of LNG Figure 3: Project Timeline ALNG Facilities Expansion Timeline July Apr il May Jun e Dec Jan 200 Feb 3 Mar July Aug Sep t Oct Nov Apr il May Jun e Jan 200 2 Feb Mar Jan 200 1 Ma r Apr il Nov Dec Jan 200 0 Feb Aug Sep t Oct Mar Apr il May Jun e July LNG Sales Window FEED Train 2 EPC 33 Months Early Commitments First Ship Feed Gas 15 September 2002 1st LNG Ship 15 November 2002 Take Over 15 December 2002 Full Commercial Operations 15 February 2003 Feed Gas Plant Takeover Full Commercial Operations Dec Jan 200 4 Sep t Oct Nov July Aug Jan 200 Feb 3 Ma r Apr il Ma y Jun e Nov Dec Sep t Oct July Aug Apr il May Jun e Feb Mar Jan 200 2 LNG Sales Window Train 3 EPC Tank 42 Months Feed Gas 1 June 2003 1st LNG Ship 1 August 2003 Take Over 1 September 2003 Full Commercial Operations 1 November 2003 PO-35.7 Feed Gas First Ship Plant Handover. Tank 3 Complete Full Commercial Operations Site Photos Train 2 Compressor Area 29-Sep-00 PO-35.8 R S C L C o m p ressor A rea CDSM Grid PO-35.9 Overall Site Photo PO-35.10