HYDROCARBON ENGINEERING MAY 2004 81

In recent years, process selection studies for baseloadLNG production have grown in scope and have increas-ingly considered multiple cycle options. The reasons

are varied and include interest in larger train capacities, adesire to minimise CO2 emissions, different compressordriver arrangements, plus specialised applications suchas floating production storage and offloading (FPSO)plants, LNG peak shaving and associated gas liquefac-tion. This article describes the work process used in aid-ing the LNG plant owner to select the optimal processtechnology, along with some specific examples of appli-cations.

Work process Typically, a process selection study will start with a designbasis prepared by the plant owner or his representativethat defines many of the design parameters to be used inthe study, such as ambient conditions; feed compositionand conditions; process equipment design assumptionssuch as pressure drop and heat exchanger approachtemperatures; fuel balance requirements; possible driverconfigurations and desired production of LNG and LPGs.The owner will also give some information on the criteriato be optimised in the design, for example, maximise pro-duction, minimise power, or minimise CO2 emissions.Preferably, the owner will provide the basis for a costfunction that can incorporate several parameters. Thisallows the process licensor to optimally trade off capitalinvestment for efficiency. By receiving as much informa-tion as possible from the plant owner on the criteria that itwill be used for the evaluation, the process licensor canprovide an optimised design that will best meet theowner’s objectives.

The first step after the design basis is established is toperform process simulations. Even when the licensor isprovided with a highly detailed design basis, there areseveral unspecified equipment performance parametersthat need to be defined for the process simulation. In par-ticular, parameters that define the performance andcapacity of the compressors and the main cryogenic heatexchanger (MCHE) are crucial to assessing the perfor-mance of a cycle. Ideally, such parameters are calculatedwith a model that is embedded in the process simulationrather than being calculated in another software applica-tion. This approach can save time and result in a moreoptimal design, particularly if numerical optimisation tech-niques are used.

For several years Air Products has been using propri-etary sequential quadratic programming based tech-niques to optimise process designs. Several (typically 20 - 50) flowsheet parameters are varied to optimise per-formance relative to an objective function, while incorpo-rating multiple constraints. The objective function used inthe optimisation can be simply to maximise production orminimise power at fixed production; or a more sophisti-cated cost based objective function can be used that con-siders both capital and operating costs. The method issuperior to older design optimisation methods such asparametric studies. This is particularly true for optimisingflowsheets that have a large number of variables that arenot linearly independent, like for LNG plants. Air Productshas found that the use of these numerical techniques notonly allows the process engineer to produce significantlybetter designs, but perhaps just as important, they alsoserve as a check against any unconscious biases theprocess designer may have.

To assess compressor performance, Air Productsutilises an in-house compressor efficiency correlation

Mark J. Roberts, Dr Yu-Nan Liu, James C.

Bronfenbrenner and James Solomon,Air

Products and Chemicals, Inc., USA,

describe the work process used in

aiding the LNG plant owner to select

the optimal process technology.

Technology selection

81-84 12/15/04 2:01pm Page 81

(ICEC™). ICEC™ is a program that accurately predictsinline compressor efficiency and impeller size. The pro-gram calculates performance as a function of impellersize, stage shape characteristics, and Mach numberrelated effects. ICEC™ predicts compressor efficiencywithin 2 percentage points accuracy, and can be adjustedto reflect known differences between manufacturers.

Air Products has integrated ICEC™ into the processsimulation models so that constraints on parameters suchas inlet flow coefficient, peripheral Mach number andimpeller diameters can be met in the converged and opti-mised simulation. The optimisation routine ‘considers’ theeffect of the flowsheet variables (pressures, compositionsetc) and the compressor speed on the compressionequipment efficiency and design. Compressor designconstraints are periodically updated to reflect the latestproven technology available from the compressor suppli-ers. The use of ICEC™ eliminates time consuming iterationswith the suppliers, and assures a feasible and near-optimaldesign on the first pass. The information from ICEC™ canthen be translated into known manufacturers frame sizes.After the final process selection is made, compressor per-formance parameter specifications are sent to the vendorsto verify both the aerodynamic and mechanical considera-tions, and to validate the calculated design.

Applications ofprocess technology

Baseload LNG plantsThe propane pre-cooled, mixed refrig-erant LNG process (C3MR) has beenthe workhorse of the LNG industry formore than 30 years. It has been appliedin LNG plants producing from 1 - 5 mil-lion tpy of LNG per train, using steamand gas turbine drivers, sea water andair cooling, rich and lean feeds contain-ing varying levels of nitrogen, with andwithout LPG extraction. The processhas proven to be efficient, flexible, reli-able and cost competitive.

For many cases, the C3MR processremains the optimal solution. In somecases however, depending on thedesign basis and owner preference,other solutions should be considered.For example, with very cold ambientconditions a dual MR process can offeradvantages over C3MR.

In recent years, interest has grownin ever larger train capacities so thateconomies of scale can significantlyreduce the unit cost for LNG. To meetthis requirement, Air Products hasdeveloped and patented the AP-X™

LNG process1. The AP-X™ LNG process will

increase single train capacities to 7 - 10 million tpy. By producing warmerLNG at pressure, today’s C3MR or dualMR process capacity can produce up to80% more LNG with the same provenequipment in use today. In the AP-X™

process, the warmer LNG is then sub-cooled to storage temperature by agaseous nitrogen refrigerant in the

same compression/expansion cycle that has been used foryears to liquefy oxygen, nitrogen, and for peak-shaver nat-ural gas liquefiers. The efficiency of this process is as high asany in the industry, while the capital cost of the liquefier perunit of LNG is reduced significantly.

The C3MR version of the AP-X™ process cycle isdepicted in Figure 1. Similar to the traditional C3MR process,propane is used to provide cooling to a temperature ofapproximately -30 ˚C. The feed is then cooled and liquefiedby mixed refrigerant, exiting the MCHE at a temperature ofapproximately -120 ˚C. Final subcooling of the LNG is per-formed using cold gaseous nitrogen from the nitrogenexpander. Figure 2 shows the equipment layout for the liq-uefaction and subcooling sections of an AP-X™ LNG train.Coil wound heat exchangers are used to liquefy and subcoolthe LNG, while the nitrogen economiser uses brazed alu-minum plate fin heat exchangers.

The power split between C3, MR, and N2 is flexible, andcan be manipulated by changing the temperature range ofthe three refrigerant loops. This feature, and the use of theSplit-MR™ machinery configuration2, allows considerableflexibility in matching compressor driver sets, whether Frame7s, Frame 9s, electric motors or combinations of the above.

The use of electric motors as compressor drivers is gain-ing increasing interest in the LNG industry. Electricity can be

82 HYDROCARBON ENGINEERING MAY 2004

Figure 1. Air Products AP-X™ process.

Figure 2. AP-X™ LNG process layout.

81-84 12/15/04 2:01pm Page 82

generated in a combined cycle power plant togreatly reduce the CO2 emitted per unit LNG pro-duced. Electric motors are also a very flexible drivesystem for the baseload LNG production. Singletrain liquefaction capacities from 7 - 10 million tpycan be easily achieved using electric motors withthe AP-X™ process, with the specific arrangementdepending upon the desired capacity and the max-imum motor size considered proven by the LNGowner. Systems may also be considered, whichuse combinations of electric motor and gas turbinedrive for the refrigeration compressors. With theuse of variable frequency devices, the speed ofeach compressor can be optimised not only toachieve maximum compressor aerodynamic effi-ciency at a design point, but also across the rangeof ambient temperature extremes to maximiseannualised production.

FPSO PlantsThere is continued interest in a Floating ProductionStorage and Offloading (FPSO) facility for base-load LNG service. The design of such a facility pre-sents significant technical challenges, not the leastof which is the selection of the liquefaction cycle.Developing and selecting a liquefaction cycle suit-able for a FPSO facility presents unique chal-lenges to the process licensor. Safety and com-pactness are particularly important criteria forFPSO service. Due to the very limited space avail-able aboard a FPSO vessel, there is increasedemphasis on compactness. The heightened con-cern with safety is related to the geographical iso-lation and compact nature of a FPSO plant.Inevitably, operating personnel will be locatedcloser to the plant equipment than on a land basedplant, and evacuation on a floating plant facility isproblematic.

In addition to the special emphasis on com-pactness and safety, the typical issues guiding theselection of a liquefaction cycle for a land basedplant such as flexibility, efficiency and operabilityapply to FPSO facilities as well. The cycleselected should have the operating flexibility tohandle varying feed compositions and ambienttemperatures, high efficiency in order to maximiseproduction for a given driver configuration, andsimplicity and operability to enable operating per-sonnel to maximise production and avoid down-time.

To some extent the above considerations canbe competing objectives. For example, more effi-cient cycles often require more pieces of equip-ment, which can lead to a less compact design. Tofurther complicate issues, FPSO facilities havebeen proposed to handle a wide variety of feedcompositions and capacities. Air Products hasreceived inquiries for facilities to handle feed com-positions from rich to lean and at design capacitiesfrom 1 - 5 million tpy LNG production. For thesereasons, it is not possible to select a single cycleas the preferred one for all FPSO opportunities.The cycle selected for a particular opportunity willdepend on the train capacity, driver selection (gasturbine or electric drive), project economics (capi-tal versus power) and owner preference.

The Air Products DMR2™ cycle2 is a recently developeddual mixed refrigerant cycle. It represents a good compro-

mise between the competing objectives at a size range ofapproximately 2 - 4.5 million tpy. A simplified schematic of

HYDROCARBON ENGINEERING MAY 2004 83

Figure 3. Air Products DMR2™ cycle.

Figure 4. Air Products DMR2™ process plant layout.

Figure 5. Peak shaver LNG cycle.

81-84 12/15/04 2:01pm Page 83

this cycle is shown in Figure 3. The condensation of liquidsbetween compression stages of the warm (pre-cooling)mixed refrigerant allows this cycle to achieve very close align-ment of cooling curves in the warm heat exchanger (HX1),resulting in a high thermodynamic efficiency.

This cycle offers safety advantages over the propane pre-cooled mixed refrigerant cycle (C3MR) or cascade cycles.The primary reason is the use of a mixed refrigerant ratherthan propane for pre-cooling. In the C3MR and cascadecycles, there is a significant inventory of propane refrigerantin multiple vessels and interconnecting piping runs, with thepotential (however remote) for leakage. The Piper Alpha off-shore gas processing platform disaster of 1988 providesmuch of the basis for this concern. In that incident, an explo-sion occurred after dense propane gas accumulated at a lowpoint on the platform4.

The use of a single vapourising pressure for the pre-cool-ing mixed refrigerant leads to a more compact design com-pared with other dual mixed refrigerant cycles that use mul-tiple vapourising pressures in the pre-cooling circuit. Thisallows the use of a single wound coil heat exchanger for pre-cooling, saving space compared with multiple pressuredesigns with multiple pre-cooling exchangers and intercon-necting piping. Figure 4 shows the equipment layout for theliquefaction and subcooling sections of a DMR2™ train.

Peak shaving LNG plantsSmall LNG plants producing 5 - 20 MMSCFD of LNG arecommon in industrialised countries. Typically, these plantsare used for ‘peak shaving’ in that they produce LNG frompipeline gas during warm months to be re-vapourised andconsumed during peak demand winter months.

Figure 5 shows a process flowsheet for a typical peakshaving plant. A closed-loop nitrogen expander cycle pro-vides refrigeration for making LNG. Advances in compressor

loaded expander, or compander technology over the last 15- 20 years have made this type of cycle particularly attractivefor small plants. State-of-the-art high efficiency companderdesigns allow this type of cycle to be competitive in efficiencyfor small plants. The use of pure nitrogen as a refrigerantalso saves significant cost over a mixed refrigerant cycle,since there is no need to procure or store hydrocarbon refrig-erant components. The use of nitrogen also saves cost in therefrigerant recycle compressor. Efficient, relatively inexpen-sive integral gear compressors that are typically used in theair separation industry can be used for peak shaver LNGplants using nitrogen recycle technology.

ConclusionThe selection of the optimal LNG liquefaction technology inthe early phases of a project can have a large impact on theoverall profitability. To provide the best possible solution, it isimportant that the process licensor has not only a completeand detailed design basis, but also as much information aspossible on the objective criteria used by the LNG owner toevaluate proposed solutions. There are several differentcycle and equipment options that can be considereddepending on the project specifics. Air Products’ objective asa process licensor is to provide proven process and equip-ment technology that maximises return on investment to theLNG owner.

References1. ROBERTS, M.J., and AGRAWAL, R., ‘Hybrid Cycle for the Production

of Liquefied Natural Gas’, US Patent 6,308,521.2. LIU, Y.N., LUCAS, C.E., BOWER, R.W., and BRONFENBRENNER,

J.C., ‘Reducing LNG Costs by better Capital Utilization’, LNG 13, May2001, Seoul, Korea.

3. ROBERTS, M.J., and AGRAWAL, R., ‘Dual Mixed Refrigerant Cycle forGas Liquefaction’ US Patent 6,119,479.

4. LEES, F.P., ‘Loss Prevention in the Process Industries’, Volume 3, PiperAlpha, Appendix 19/11, 1996.

84 HYDROCARBON ENGINEERING MAY 2004

81-84 12/15/04 2:01pm Page 84