LARGE CAPACITY LNG TRAINS TRAINS GNL DE GRANDE CAPACITE Marc Le Métais Jean-Claude Garcel Christian Bladier Exploration & Production TOTAL Paris, France ABSTRACT The capacity of natural gas liquefaction trains has increased from about one million metric tonnes per annum in the 70's to about five million tonnes today. This five fold increase is the result of the efforts made by the LNG industry to reduce the liquefaction costs through economies of scale. Today this quest seems to have reached its limits as the size of several key equipment cannot readily be increased. Despite these limitations, means have been identified to achieve even larger LNG trains capacity, in the range of eight to nine million metric tonnes per annum, with minimum concession to technical risk. These solutions are being engineered to provide a technically robust way to further develop the Qatar North Field. All components of the LNG train have been reviewed : acid gas removal, sulphur recovery, dehydration, liquefaction process, refrigeration and liquefaction equipment (compressors, turbines and heat exchangers), NGL extraction, nitrogen removal. LNG shipping aspects have also been studied but are not presented in this paper. Indications on the economies of scale that can still be achieved will be provided. RESUME La capacité des trains de liquéfaction de gaz naturel est passée d'environ un million de tonnes par an dans les années 70 à cinq millions de tonnes aujourd'hui. Ce quintuple accroissement découle des efforts entrepris par l'industrie du GNL pour réduire les coûts de liquéfaction par des économies d'échelle. La taille de plusieurs équipements clés ne pouvant être augmentée, cette voie semble avoir à ce jour atteint ses limites. Malgré ces contraintes, pour accroître encore la capacité des trains de GNL d'autres moyens ont été identifiés permettant d'atteindre huit à neuf millions de tonnes par an sans concession majeure vis à vis des risques techniques. Afin de poursuivre le développement du gisement gazier du North Field au Qatar des solutions techniques solides sont actuellement en cours de conception. Tous les composants du train de GNL sont examinés : l'extraction des gaz acides et la récupération PS5-7.1
de soufre, la déshydratation, le procédé de liquéfaction, les équipements de réfrigération et de liquéfaction (compresseurs, turbines et échangeurs), l'extraction de LGN, la déazotation ainsi que le transport maritime. Ce dernier aspect ne sera toutefois pas présenté dans cet article. Des indications sont données sur les économies d'échelle ainsi obtenues. INTRODUCTION Since the first baseload LNG plant built in Arzew, Algeria, the capacity of subsequent natural gas liquefaction trains has continuously increased from 0.4 up to 5 Mt/a at present time. TOTAL a key player in the LNG industry, present in several major LNG plants, is striving to reduce the LNG chain cost to develop its expanding gas resources. Economy of scale is one axis to achieve this goal. The purpose of this paper is to present a technically robust way to design a large capacity train to develop the production from the Qatar North Field. Today, LNG trains capacities up to 5 Mt/a are achievable with slight improvements on present designs. An ambitious production step increase of 50% has been set (7.5 Mt/a LNG production). The principal objective of the study was to conceive a technically feasible train achieving this production target while still benefiting from economies of scale. Several schemes involving different process schemes, driver selections, compression arrangements, availability have been identified. The technical and economical assessment of these schemes concludes to the benefit of very large (higher than the target) capacity trains. MAIN FACTORS LIMITING TRAIN CAPACITY As a basis of comparison, a reference project has been set with the following features : conventional APCI propane pre-cooled, mixed refrigerant (MR) process, LNG capacity of 5.0 Mt/a. one APCI main cryogenic exchanger, two FRAME 7 turbines drive the refrigerant compressors, limited additional power is provided by the electric motor helpers, one propane compressor, the power imbalance between the two cycles is compensated by the "split MR" arrangement (one turbine drives the propane compressor and part of the MR compression), no LPG extraction, indirect sea water cooling, liquid expander turbines, conventional end flash unit, pre-treatment : amine based acid gas removal unit for H 2 S and CO 2. PS5-7.2
Each piece of main equipment is then reviewed to identify which are the limiting ones. In terms of size the critical equipment on the reference train are the following : Gas turbines These equipment being of standard design, have therefore a given power. A larger capacity train requires more power, this can be achieved by several means : use of larger electrical motors (helpers) to provide the additional power (this solution requires that additional gas turbines for power generation are provided). use of larger gas turbines, for example the FRAME 9. This is the most promising solution in terms of cost as it would result in direct economy of scale (lower $/MW), however the use of the FRAME 9 as mechanical drive is still under development. use of a third FRAME 7. Propane compressor This equipment is known to be the most critical. Already deficient performances have been experienced with compressors recently constructed. Due to the very large propane volumetric flows required, the design of these compressors is very close to technological limits. To achieve a larger train capacity several solutions can be envisaged : use mixed refrigerant instead of propane for the pre-cooling duty such as done with the AXENS LIQUEFIN process or the APCI dual mixed refrigerant (DMR) process. modify the liquefaction process in order to reduce the propane flow per unit of LNG produced. This is achieved with the AIR PRODUCT AP-X TM or the TECHNIP MLP processes. reduce the compressor rotating speed which allows to increase the propane flow limit. This can be achieved by the use of the FRAME 9 gas turbine (3,000 rpm) instead of the FRAME 7 (3,600 rpm). modify the compression stages arrangement : in the standard way (figure 1 left) each stage adds the side stream to the flow from the previous stage. This results in high volumetric flow in particular in the third stage. Another arrangement (figure 1 right) can be envisaged whereby the second compression stage discharges at the condensing pressure instead of being added to the third stage. LLP LP MP HP LLP LP MP HP Figure 1 : Possible propane compressor arrangements PS5-7.3
operate two 50% compressors in parallel. Though this solution is feasible provided special care to the design is made to avoid the imbalance of load sharing, it is not recommended for its flow instability risk. have two propane cycles : one for the pre-cooling of the natural gas and the fractionation needs, and another one for the MR pre-cooling. The separation of these two functions while avoiding the parallel operation reduces significantly the load of the main propane compressor and would allow with a feasible propane compressor to produce 30% more LNG. in the above cases where two compressor casings are required, they can be driven either by the same turbine or by two different turbines. Mixed refrigerant compressor Though not as critical as the propane compressor, this equipment is also close to technical limits due to the very high MR volume flow at the low pressure suction. To achieve a larger train capacity several solutions are available : use axial compressors as already the case in some LNG plants. as indicated above, modify the liquefaction process in order to reduce the MR flow per unit of LNG produced. This can be achieved by different means such as the TECHNIP MLP or the AP-X processes. Main cryogenic heat exchanger The largest exchanger that APCI can manufacture in their present workshop is suitable for a 5.0 Mt/a LNG capacity. A larger exchanger could be manufactured in a new workshop to be installed close to the embarkation port. This would allow about 20% increase in LNG production (~6.0 Mt/a). Therefore, for higher rates, either 50% exchangers are used or the process needs to be modified. Three possibilities have been identified : AP-X process : it has been developed by APCI to solve this issue, as it allows to reach about 7.5 to 8.0 Mt/a (about 50% LNG production increase) without exceeding APCI manufacturing limit. By adding a third refrigerant cycle (nitrogen) the duty of the exchanger is reduced. With three refrigerant cycles, the AP-X process actually splits in series the liquefaction duty. APCI process with dual liquefaction strings, two 50% MR cycles. The above AP-X process split needs to be compared with the solution consisting in splitting by half the liquefaction duty with two 50% MR cycles. Like the AP-X, this process also has three refrigerant cycles. MLP (Maximum LNG Production) process patented by TECHNIP : That process (figure 2) has been developed to provide debottlenecking solution to existing trains with no modification of the liquefaction process. This solution adds also a third refrigerant cycle, though contrarily to the AP-X, the third cycle is applied only to a slip stream of feed gas and the resulting capacity increase is limited to about 10 to 15% only. Since the objective is to identify solutions leading to a 50% capacity increase step, that process has not been further considered. PS5-7.4
NG Main Exchanger Fuel Gas LNG Figure 2 : TECHNIP MLP process Piping and valves Large diameter stainless steel lines and valves are required between the main exchanger and the suction of the MR compressor (66" in present trains) as well as at the suction of the propane compressor. At that size, valves are tailor made. Any solution which reduces the refrigerants flows per unit of LNG is beneficial, but this issue is not critical as reducer can be used though at the expense of higher pressure drops resulting in decreased liquefaction efficiency. Pressure vessels and heat exchangers The critical equipment (CO 2 absorber, scrub column, propane evaporators, compressors suction drums) have been evaluated in this study. All have been found feasible. VERY LARGE CAPACITY TRAINS DESCRIPTION The above consideration have led to studying for comparison purpose different train arrangements with a particular emphasis on the most promising potential developments : the use of the FRAME 9 turbine (whatever the process is), the AP-X process and the AXENS LIQUEFIN process to evaluate their value in terms of economy of scale. With the objective of designing a 7.5 Mt/a LNG train, three cases have been considered : AP-X process (figure 3) : This process presented in GasTech 2002 has been developed by APCI "in response to continuing customer demand for larger train capacity and lower unit cost". The use of a nitrogen expander loop, allows to reduce the flow of both propane and mixed refrigerant cycles. It consists of : PS5-7.5
one FRAME 9 turbine driving the propane compressor. one FRAME 9 turbine driving the mixed refrigerant compressor. one FRAME 7 turbine driving the nitrogen compressor. Since the site rated power is 102 MW for a FRAME 9 and 71 MW for a FRAME 7, different arrangements are possible. For the cost estimate of this case, the following has been assumed : a FRAME 9 driving the propane compressor and also drives an electrical generator, a FRAME 9 driving the mixed refrigerant compressor needs an helper motor, and a FRAME 7 to drive the nitrogen compressor though for standardization purposes probably a FRAME 9 would be used. This solution would allow to produce some more electrical power within the train. one main spool wound heat exchanger (MCHE) for natural gas condensing. one nitrogen spool wound heat exchanger for LNG subcooling. nitrogen cold box. two turbo-expander / compressor ("compander") units. electrical power required for helper motors and end flash gas compressor is produced in the generator(s) installed within the train. AP-X TM Hybrid LNG Process LNG Subcooler FEED C3 Pre-Cooling MCHE Nitrogen Expander Nitrogen Coldbox Mixed Refrigerant Figure 3 : Typical scheme for AP-X TM Hybrid LNG Process The above schematic represents the AP-X process in a simplified way. The limitation imposed by the main cryogenic exchanger is overcome by splitting its duty in two. Liquefaction and sub-cooling of the natural gas is achieved with two independent refrigerant cycles, instead of having one exchanger, two exchangers in series are needed. In addition the second exchanger does not come alone, turbo-expanders, cold boxes and nitrogen compressors (the whole right part of the schematic) are also required. The two refrigerant cycles in series found in the conventional APCI process are replaced by three cycles in series. PS5-7.6
According to APCI the train can be built in two phases, the nitrogen refrigerant cycle being added later. In the first phase the train capacity is 65% of the final capacity. However in case of nitrogen cycle shut-down, the whole production needs to be stopped. Resumption of production at 65% capacity cannot be achieved before several hours until the mixed refrigerant inventory has been changed to match the new operating conditions. Though the resulting scheme fits better with the image one has of a single train concept as opposed to splitting into two smaller trains, the capacity increase has nevertheless been obtained at the cost of splitting the process in two and adding a third cycle. The same can be achieved by splitting the main exchanger duty in two parallel rather than series duties. For this reason, a conventional APCI process with dual liquefaction strings, has been studied for comparison with the AP-X process. Conventional APCI process with dual liquefaction strings (figure 4) : Another way of splitting the liquefaction process to overcome the limitation imposed by the main cryogenic exchanger is to install two main cryogenic heat exchangers. Each exchanger being associated with one mixed refrigerant cycle. The pre-cooling of the natural gas and the mixed refrigerant can be achieved with either common or separate loops driven by the same gas turbine. Mixed refrigerant 1 FEED Pre-treatment Propane pre-cooling MCHE 1 MCHE 2 Flash Mixed refrigerant 2 Figure 4 : APCI Process with dual liquefaction strings The above simplified schematic shows a possible arrangement for a 7.5 Mt/a LNG train : one propane cycle. two identical liquefaction strings, each string composed of one main cryogenic exchanger and one mixed refrigerant cycle (corresponding to 3.75 Mt/a each). three FRAME 7 turbines with larger helper motors. the total electrical power demand within the train (helper motors and flash gas compressor) needs to be supplied from the plant power generator. the overall propane compressor duty cannot be achieved in a single compressor therefore the solutions previously described would have to be implemented. in this scheme the main equipment (turbines, compressors and main exchanger) are all within present proven design. PS5-7.7
in this scheme also the train can be built in two phases, the first phase being 50% of the final capacity. In addition in the event of one mixed refrigerant cycle being shutdown, 50% of the production is maintained without delay. Axens Liquefin process This process is based on a mixed refrigerant rather than propane for the pre-cooling which allows compressors design to be within proven domain. This 7.5 Mt/a scheme is considered for comparison purpose to assess whether the LIQUEFIN process is more adapted to providing economies of scale when compared to a 5 Mt/a case. The liquefaction process arrangement for a 7.5 Mt/a LNG production is basically the same as the APCI process with dual liquefaction strings (similar to figure 4 where all exchangers would be grouped in cold boxes). one pre-cooling cycle driven by one FRAME 7 with a helper motor. one liquefaction cycle having two sets of compressors in parallel, each set driven by a FRAME 7 with a helper motor. in the LIQUEFIN process the pre-cooling exchangers and the main cryogenic heat exchanger are made of plate-fin exchangers arranged in cold boxes. For 5.0 Mt/a, four cold boxes are required, six are necessary for 7.5 Mt/a. turbines and compressors are of proven design, however the LIQUEFIN process has no industrial reference for the time being. the electrical power required within the train for the helpers and for the flash gas compressor needs to be supplied from a power generation plant. in this scheme also the train can be built in two phases and 50% production maintained when one liquefaction cycle is shutdown. EVALUATION OF APCI's AP-X VERSUS DUAL STRING PROCESSES FRAME 9 as mechanical drive One important issue when comparing these two arrangements is the assessment of the risk associated with the FRAME 9 turbine as mechanical drive. Some technical developments are ongoing to validate this turbine as mechanical drive for the refrigerant compressors. Today it is expected, that all above developments will be successfully achieved. Therefore, it should be the best way to achieve economy of scale since the price of the FRAME 9 is about 120% of the FRAME 7 price for 44% more power. It should be noted that the use of a FRAME 9 turbine is not specific to the AP-X process. PS5-7.8
AP-X process The process is more complex than the conventional process, it requires several additional pieces of equipment : the added nitrogen cycle includes compressors, turboexpanders, plate fin exchangers, spool wound exchanger, etc We believe that all these additional pieces of equipment result in some complication of the process for the sole sake of avoiding the duplication of the main heat exchanger. Scheduled maintenance Over a six year period, the duration required for the maintenance of a FRAME 9 is ten days longer than what is needed for a FRAME 7. On a yearly basis, this corresponds to an additional production losses of 1.7 days. Ideal maintenance based on condition monitoring may allow to reduce the gap to eight days over a six year period i.e. 1.3 days per year. Reliability There is no experience yet concerning the reliability of FRAME 9s as mechanical drives. However, an estimate can be made based on the following considerations : experience from FRAME 5s, FRAME 6s and FRAME 7s as mechanical drives show that as power increases, reliability decreases. when used as electrical power generators, the FRAME 9s have one percentage point less reliability than the FRAME 7s. only half a percentage point difference of reliability has been used to compare the FRAME 7 and FRAME 9 when used as mechanical drive. Availability The availability of the schemes have been compared on the basis of ideal operation and maintenance and assuming that the FRAME 9 is only half a percentage point less reliable than the FRAME 7. The following additional lost days are obtained when compared to the reference two FRAME 7 train : three FRAME 7 dual string scheme : +1 lost day four FRAME 7 dual string scheme : same as reference three FRAME 9 dual string scheme : +5 lost days (+9)* three FRAME 9 AP-X scheme : +11 lost days (+15)* * : corresponding figures with one percentage point difference between FRAME 9 and FRAME 7. PS5-7.9
Comparative table summary AP-X SCHEME DUAL STRING SCHEME (WITH 3 FRAME 7) New and complex process. Well known proven process. 3 different cycles : C3, MR, N2 3 cycles : C3, MR1, MR2 (MR1 & MR2 are identical. 5 or 6 compressors (all different). 6 compressors (3 or 4 different). Several turbo-expanders with recompressors. None. Cold boxes If any of the gas turbines trips the whole LNG production is stopped. Reduced availability : 10 more days lost. Operating complexity : dual function on some turbines (mechanical drive and power generation). Risk inherent to unproven equipment. None. C3 turbine trip : no production, MR1 turbine trip : 50% LNG production, MR2 turbine trip : 50% LNG production. Better availability : additional LNG production of 220,000 t/a Direct full use of gas turbines capacities. Well proven compressors, turbines and exchangers. There is no significant cost benefit for the EPC (Engineering, Procurement & Construction) cost between the two cases. Consequently, the cost is not the leading parameter for the process selection. CONCLUSIONS This study has established the feasibility to build a very large capacity train of about 7.5 Mt/a, a significant step increase compared to present trains capacity (near to 5 Mt/a). However two factors are reducing the economies of scale that were expected : the limitation imposed by the maximum size achievable today for the main cryogenic heat exchanger which requires to split the liquefaction in two parts (either in series in the AP-X scheme, or in parallel in the dual string case), the use of the FRAME 9 gas turbine instead of the FRAME 7 is by itself inducing economies of scale, however since 7.5 Mt/a requires three turbines, and 8.5 Mt/a requires four FRAME 7s or three FRAME 9s, while 5 Mt/a require two, overall economy of scale on the number of turbines is not obtained. It is generally admitted that economy of scale is expressed by a power factor of 0.65 to 0.70 on the capacity. This means that a 50% increase in capacity should result in a 30 to 33% increase in cost. PS5-7.10
Analysis of the different cost estimates in this study indicate that the step change from 5 Mt/a to 7.5 Mt/a, a 50% capacity increase, induces a 35% cost increase for the process train only (equivalent to a power factor of 0.74) which indicates that there is still an economy of scale. When one compares the cost of the whole plant excluding storage and loading : process, utilities and only the off-site facilities which are directly associated with the process, the cost increase is by 30% only. The specific Qatar context where feed gas pre-treatment and sulphur processing are significant explains the better than expected economy of scale achieved. In the AXENS LIQUEFIN case the cost increase is slightly higher, which indicates that this process does not generate more economies of scale than the APCI process. This result reflects the fact that in this process also, to increase the capacity from 5 to 7.5 Mt/a, three turbines are required instead of two and six cold boxes instead of four are needed. Way forward Despite having reached equipment limits in the 5 Mt/a train capacity range, a larger capacity train (7.5 Mt/a) can be made with limited or none technical risk and still achieve a significant economy of scale : 50% more production for about 30% more cost. The AP-X process, though more conform to the concept of a one-line train is more complex and less reliable than the dual string arrangement without significant cost benefit. The development of the FRAME 9 gas turbine is well under way. To use them in a liquefaction train will be an acceptable risk and will provide some cost reduction. Based on the above considerations, TOTAL considers that very high capacity trains of around 9 Mt/a of "lean" LNG (i.e. with partial extraction of ethane and LPG) are technically feasible with either of the following schemes : the three FRAME 9 dual string arrangement as shown in figure 4 : one FRAME 9 driving a two casing propane compressors. two FRAME 9 driving each the mixed refrigerant compressors. the four FRAME 7 and one FRAME 5 dual string arrangement as shown in figure 5 (with added ethane plus extraction) : two FRAME 7 split MR arrangement on each string, propane cycles duty being dedicated to the mixed refrigerant pre-cooling. The propane compressor being thus well within proven design range. separate propane cycle (FRAME 5 driven) dedicated to the feed gas precooling and fractionation duties. PS5-7.11
Mixed refrigerant 1 NATURAL GAS Propane pre-cooling MCHE 1 LNG Pre-treatment C2+ extraction MCHE 2 Flash Mixed refrigerant 2 Figure 5 : Independent natural gas pre-cooling 9 Mt/a train arrangement PS5-7.12