The Removal of CO2 and N2 From Natural Gas

32
The removal of CO 2 and N 2 from natural gas: A review of conventional and emerging process technologies T.E. Rufford a , S. Smart b , G.C.Y. Watson a , B.F. Graham a , J. Boxall a , J.C. Diniz da Costa b , E.F. May a,n a The University of Western Australia, Centre for Energy, 35 Stirling Highway, Crawley, WA 6009, Australia b The University of Queensland, FIMLabFilms and Inorganic Membrane Laboratory, School of Chemical Engineering, College Road, St Lucia, Qld. 4072, Australia article info Article history: Received 5 September 2011 Accepted 2 June 2012 Available online 19 June 2012 Keywords: CO 2 capture nitrogen rejection contaminated gas membrane absorption pressure swing adsorption hydrates for gas separation abstract This article provides an overview of conventional and developing gas processing technologies for CO 2 and N 2 removal from natural gas. We consider process technologies based on absorption, distillation, adsorption, membrane separation and hydrates. For each technology, we describe the fundamental separation mechanisms involved and the commonly applied process flow schemes designed to produce pipeline quality gas (typically 2% CO 2 , o3% N 2 ) and gas to feed a cryogenic gas plant (typically 50 ppmv CO 2 , 1% N 2 ). Amine absorption technologies for CO 2 and H 2 S removal (acid gas treating) are well- established in the natural gas industry. The advantages and disadvantages of the conventional amine- and physical-solvent-based processes for acid gas treating are discussed. The use of CO 2 selective membrane technologies for bulk separation of CO 2 is increasing in the natural gas industry. Novel low- temperature CO 2 removal technologies such as ExxonMobil’s Controlled Freeze Zone TM process and rapid cycle pressure swing adsorption processes are also emerging as alternatives to amine scrubbers in certain applications such as for processing high CO 2 concentration gases and for developing remote gas fields. Cryogenic distillation remains the leading N 2 rejection technology for large scale (feed rates greater than 15 MMscfd) natural gas and liquefied natural gas plants. However, technologies based on CH 4 selective absorption and adsorption, as well as N 2 selective pressure swing adsorption technol- ogies, are commercially available for smaller scale gas processing facilities. The review discusses the scope for the development of better performing CO 2 selective membranes, N 2 selective solvents and N 2 selective adsorbents to both improve separation power and the durability of the materials used in novel gas processing technologies. & 2012 Elsevier B.V. All rights reserved. 1. Introduction In 2010 natural gas (NG) supplied 23.81% of the world’s energy demand and the volume of natural gas consumed increased by 7.4% over 2009 levels driven by economic recovery the USA and the continuing economic development of China, India and Russia (BP Statistical Review of World Energy June 2011, 2011). The growth in the global demand for natural gas has led to a re-evaluation of the development potential of unconventional, stranded and contami- nated gas reserves that were previously considered economically unviable. Furthermore, many of the significant natural gas reserves are located far from the large established gas markets in Western Europe, Japan and South Korea. Therefore, significant volumes of natural gas must be transported long distances from exporting countries either by pipeline or by tanker as liquefied natural gas (LNG); the economics of this choice have been discussed by many authors such as Rojey et al. (1997). Table 1 contains a list of the major natural gas importing and exporting countries in 2010 including a breakdown of the amounts transported by pipeline and LNG. The production of LNG is clearly already essential to the international trade of natural gas and its importance is set to increase further over the next two decades, particularly in the Asia-Pacific region. The estimated world volumes of sub-quality natural gas reserves, including sour natural gas reserves, are significant. Sub-quality natural gas reserves are defined as gas fields contain- ing more than 2% CO 2 , 4% N 2 and 4 parts per million (ppm) hydrogen sulphide (H 2 S) (Kidnay and Parrish, 2006). Burgers et al. (2011), for example, estimate that 50% of the volume of known gas resources contain more than 2% CO 2 . The development of sub-quality, unconventional and remote natural gas reserves, including development via LNG production, can present new challenges to gas processing that require more efficient approaches to the conventional absorption and cryogenic con- densation technologies that are most commonly used for the removal of carbon dioxide and nitrogen, respectively, from Contents lists available at SciVerse ScienceDirect journal homepage: www.elsevier.com/locate/petrol Journal of Petroleum Science and Engineering 0920-4105/$ - see front matter & 2012 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.petrol.2012.06.016 n Corresponding author. E-mail address: [email protected] (E.F. May). Journal of Petroleum Science and Engineering 94-95 (2012) 123–154

Transcript of The Removal of CO2 and N2 From Natural Gas

  • lam

    , WA

    choo

    Article history:

    Received 5 September 2011

    Accepted 2 June 2012

    CO2 capture

    nitrogen rejection

    contaminated gas

    membrane

    absorption

    pressure swing adsorption

    hydrates for gas separation

    This article provides an overview of conventional and developing gas processing technologies for CO2and N2 removal from natural gas. We consider process technologies based on absorption, distillation,

    adsorption, membrane separation and hydrates. For each technology, we describe the fundamental

    countries either by pipeline or by tanker as liqueed natural gas

    l gascant.tain-ppm)et al.

    sub-quality, unconventional and remote natural gas reserves,

    Contents lists available at SciVerse ScienceDirect

    .el

    Journal of Petroleum Sci

    Journal of Petroleum Science and Engineering 94-95 (2012) 123154removal of carbon dioxide and nitrogen, respectively, fromE-mail address: [email protected] (E.F. May).(LNG); the economics of this choice have been discussed by many including development via LNG production, can present newchallenges to gas processing that require more efcientapproaches to the conventional absorption and cryogenic con-densation technologies that are most commonly used for the

    0920-4105/$ - see front matter & 2012 Elsevier B.V. All rights reserved.

    http://dx.doi.org/10.1016/j.petrol.2012.06.016

    n Corresponding author.Europe, Japan and South Korea. Therefore, signicant volumes ofnatural gas must be transported long distances from exporting

    (2011), for example, estimate that 50% of the volume of knowngas resources contain more than 2% CO2. The development ofthe global demand for natural gas has led to a re-evaluation of thedevelopment potential of unconventional, stranded and contami-nated gas reserves that were previously considered economicallyunviable. Furthermore, many of the signicant natural gas reservesare located far from the large established gas markets in Western

    The estimated world volumes of sub-quality naturareserves, including sour natural gas reserves, are signiSub-quality natural gas reserves are dened as gas elds coning more than 2% CO2, 4% N2 and 4 parts per million (hydrogen sulphide (H2S) (Kidnay and Parrish, 2006). BurgersIn 2010 natural gas (NG) supplied 23.81% of the worlds energydemand and the volume of natural gas consumed increased by 7.4%over 2009 levels driven by economic recovery the USA and thecontinuing economic development of China, India and Russia (BPStatistical Review of World Energy June 2011, 2011). The growth in

    including a breakdown of the amounts transported by pipelineand LNG. The production of LNG is clearly already essential to theinternational trade of natural gas and its importance is set toincrease further over the next two decades, particularly in theAsia-Pacic region.1. Introductionseparation mechanisms involved and the commonly applied process ow schemes designed to produce

    pipeline quality gas (typically 2% CO2, o3% N2) and gas to feed a cryogenic gas plant (typically 50 ppmvCO2, 1% N2). Amine absorption technologies for CO2 and H2S removal (acid gas treating) are well-

    established in the natural gas industry. The advantages and disadvantages of the conventional amine-

    and physical-solvent-based processes for acid gas treating are discussed. The use of CO2 selective

    membrane technologies for bulk separation of CO2 is increasing in the natural gas industry. Novel low-

    temperature CO2 removal technologies such as ExxonMobils Controlled Freeze ZoneTM process and

    rapid cycle pressure swing adsorption processes are also emerging as alternatives to amine scrubbers in

    certain applications such as for processing high CO2 concentration gases and for developing remote gas

    elds. Cryogenic distillation remains the leading N2 rejection technology for large scale (feed rates

    greater than 15 MMscfd) natural gas and liqueed natural gas plants. However, technologies based on

    CH4 selective absorption and adsorption, as well as N2 selective pressure swing adsorption technol-

    ogies, are commercially available for smaller scale gas processing facilities. The review discusses the

    scope for the development of better performing CO2 selective membranes, N2 selective solvents and N2selective adsorbents to both improve separation power and the durability of the materials used in novel

    gas processing technologies.

    & 2012 Elsevier B.V. All rights reserved.

    authors such as Rojey et al. (1997). Table 1 contains a list of themajor natural gas importing and exporting countries in 2010Available online 19 June 2012

    Keywords:The removal of CO2 and N2 from naturaemerging process technologies

    T.E. Rufford a, S. Smart b, G.C.Y. Watson a, B.F. Graha The University of Western Australia, Centre for Energy, 35 Stirling Highway, Crawleyb The University of Queensland, FIMLabFilms and Inorganic Membrane Laboratory, S

    a r t i c l e i n f o a b s t r a c t

    journal homepage: wwwgas: A review of conventional and

    a, J. Boxall a, J.C. Diniz da Costa b, E.F. May a,n

    6009, Australia

    l of Chemical Engineering, College Road, St Lucia, Qld. 4072, Australia

    sevier.com/locate/petrol

    ence and Engineering

  • T.E. Rufford et al. / Journal of Petroleum Science and Engineering 94-95 (2012) 123154124natural gas. Additional gas processing challenges arise from newenvironmental regulations that may call for capture and seques-tration of CO2 from gas elds and stricter regulation of CH4

    Table 1Volumes of natural gas traded as pipeline and LNG (billions of cubic metres) by top

    natural gas exporting and importing countries in 2010 (BP Statistical Review of

    World Energy June 2011, 2011).

    Top natural gas exporters

    Pipeline LNG Total

    1 Russian Federation 186.5 13.4 199.9

    2 Norway 95.9 4.7 100.6

    3 Qatar 19.2 75.8 94.9

    4 Canada 92.4 0.0 92.4

    5 Algeria 36.5 19.3 55.8

    6 Netherlands 53.3 0.0 53.3

    7 Indonesia 9.9 31.4 41.3

    8 Malaysia 1.5 30.5 32.0

    9 U.S. 30.3 1.6 32.0

    10 Australia 0.00 25.4 25.4

    Top natural gas importers

    Pipeline LNG Total

    1 United States 93.3 12.2 105.5

    2 Japan 93.5 93.5

    3 Germany 92.8 92.8

    4 Italy 66.3 9.1 75.3

    5 United Kingdom 35.0 18.7 53.6

    6 France 35.0 13.9 48.9

    7 South Korea 44.4 44.4

    8 Turkey 28.8 7.9 36.7

    9 Spain 8.9 27.5 36.4

    10 Ukraine 33.0 33.0emissions in N2 vent streams from natural gas productionfacilities.

    Carbon dioxide, as well as H2S and other acid gases, must beremoved from natural gas because in the presence of water theseimpurities can form acids that corrode pipelines and otherequipment. Although this paper is not focussed on the removalof H2S, it should be noted that health and safety is a key driver forremoval of this highly toxic gas from sour natural gas streams.Several of the CO2 capture technologies described are moreselective for H2S than CO2. Furthermore, CO2 provides no heatingvalue and must be removed to meet gas quality specicationsbefore distribution to gas users. The maximum level of CO2permitted in natural gas transmitted to customers by pipeline istypically less than 3% (Hubbard, 2010) although local contractsmay stipulate quality specications different to these values. Thespecications of CO2 removal from natural gas to be processed ina cryogenic plant to produce LNG, recover liqueed petroleum gas(LPG) or natural gas liquids (NGL) are more stringent than thosefor typical gas pipelines. For example, in addition to moreextensive dehydration, the CO2 concentration of the natural gasshould be less than 50 ppmv (Hubbard, 2010) before it enters thecryogenic processes within the plant to avoid the formation ofdry ice.

    A typical gas pipeline specication for N2 is 3% (Kidnay andParrish, 2006); because N2 is inert the main driver for its removalfrom pipeline sales gas is to increase the heating value of the gas.Nitrogen will not freeze or lead to corrosion in a cryogenic gasplant, but a maximum concentration of 1% N2 is often specied forLNG, for example to avoid stratication and rollover of the liquidproduct during shipping. To ensure this specication is met inplants liquefying NG with a high N2 content, cryogenic distillationcolumns known as nitrogen rejection units (NRU) are integratedinto the process; these columns are both expensive and energyintensive. In addition, high N2 concentrations in natural gasprocessed by an LNG plant are energetically parasitic because asignicant amount of energy is wasted cooling the N2 fromambient temperatures to those of LNG (about 161 1C at atmo-spheric pressure). Furthermore, even if the NG does not have ahigh N2 concentration, an NRU may be necessary because theliquefaction process produces two product streams: the LNG andan end-ash gas which is a mixture of N2 and CH4 vapour. Thisend-ash gas can be used as fuel gas for the plant and/or blendedinto sales gas destined from a conveniently located pipeline.However, if production of end-ash gas exceeds the fuel gasrequirements and if no pipeline is available for its disposal anNRU may be necessary to produce a stream pure enough that itcan be vented to atmosphere. Given the increasing regulation of,and costs associated with mitigation of, greenhouse gas emissionsthis relatively expensive NRU option could become increasinglycommon.

    This review provides an overview of conventional, developingand novel gas processing technologies for CO2 and N2 removalfrom natural gas, with particular attention paid to large scale LNGproduction. The introduction sections provide an overview of atypical process ow sheet and the fundamental unit operationsinvolved in natural gas processing. The subsequent sections ofthis review describe the core concepts and industrial applicationof absorption, distillation, adsorption, membrane and hydrate gasseparation technologies.

    Many of the process technologies we describe for CO2 capturefrom CH4 can also be applied to CO2 capture from combustion uegases, for example in coal-red power stations. The recentadvances and future trends in technologies for capturing CO2 atthe power plant are reviewed elsewhere by many other authors,including Figueroa et al. (2008), MacDowell et al. (2010) andEbner and Ritter (2009). However, the process conditions avail-able in the natural gas processing facility can be very different tothose conditions available for post-combustion ue gas (predo-minantly a mixture of CO2, N2 and H2O) treatment. The two mostsignicant differences between these applications are the CO2partial pressure, and the level of CO2 removal required. In the rstcase, the natural gas feed to the CO2 removal unit is typicallyavailable at high pressures (more than 3000 kPa) while ue gasesare typically at close to atmospheric pressure, so the driving forcefor CO2 capture from ue gas (partial pressure CO2 typically lessthan 15 kPa) is much lower than from natural gas. In the secondcase the level of CO2 removal required for natural gas productionis greater, especially for LNG production plants, than bulk separa-tion of CO2 from ue gas. Thus, some promising strategies forCO2 capture in thermal power plants such as oxy-combustion(a modied combustion process that can produce a high CO2concentration ue gas, Plasynski et al., 2009) are not relevant toCO2 removal from natural gas.

    1.1. Overview of conventional gas processing ow schemes

    The major operations that can be used in natural gas proces-sing and LNG production are shown in Fig. 1. Common processoperations include inlet gas compression, acid gas removal,dehydration, LPG/NGL recovery and hydrocarbon dewpoint con-trol, nitrogen rejection, outlet compression and liquefaction.Depending on the available markets, feed gas properties, productspecications and the gas ow rate, the units identied in Fig. 1may not all be required. For example a nitrogen rejection unit(NRU) may not be required for the production of pipeline gas froma feed gas containing only low N2 concentrations, or if a high N2content gas can be blended with richer natural gas streams to

    meet pipeline gas specications.

  • a cr

    of t

    T.E. Rufford et al. / Journal of Petroleum Science and Engineering 94-95 (2012) 123154 125Fig. 1. Conventional gas processing operations in a typical natural gas plant withprocess ow arrangements vary depending on the local feed gas properties, size

    cryogenic heat exchanger).

    Table 2Physical properties of methane, carbon dioxide and nitrogen.

    Property

    Kinetic diameter (A) (Tagliabue et al., 2009)In conventional gas processing H2S and CO2 are removed in anacid gas removal unit (AGRU) using aqueous amine absorptionprocesses. The sweetened gas leaving the amine process is saturatedwith water, so typically the AGRU is located upstream of thedehydration facilities. The nal sections of the gas liquefaction plant(main cryogenic heat exchange (MCHE) in Fig. 1) can operate attemperatures as low as 161 1C, and therefore, it is essential thatcomponents that could freeze, and cause blockages in the cryogenicequipment, at low temperatures are removed from feed to thecryogenic plant. In addition to CO2, components that could freeze inthe cryogenic plant include water (typically removed to less than0.1 ppmv), heavier parafnic hydrocarbons and aromatics such asbenzene. To protect the aluminium plate-n heat exchangers usedin the NRU the feed gas to the cryogenic plant must be free ofmercury (below 0.01 mg/Nm3) (Kidnay and Parrish, 2006). The gasfeed entering the (Pre-Cool section of the) liquefaction plant istypically delivered, or compressed to, a pressure of more than3400 kPa and at a temperature close or slightly above ambient.

    The conventional method of removing N2 from natural gas is bycryogenic distillation; hence NRUs are usually closely integratedwithin the liquefaction process. There are several variations ofcryogenic liquefaction processes including, for example, Air Productsand Chemicals (APCI) C3MR process and the ConocoPhillips Opti-mised Cascade process. Commonly a propane refrigeration loop isused to pre-cool the gas before it enters a main cryogenic heatexchanger (MCHE) which might use a mixed refrigerant or a cascadeof several pure uid refrigeration cycles. If an NRU is incorporatedwith the liquefaction process then it would commonly be locateddownstream of the MCHE and before the nal depressurisationstage of the LNG production process.

    Normal boiling point (NBP) (K) (Lemmon et al., 2010)

    Critical temperature (K) (Lemmon et al., 2010)

    Critical pressure (kPa) (Lemmon et al., 2010)

    DHvap at NBP (kJ/mol) (Linstrom and Mallard, 2011)

    Polarisability (A3) (Tagliabue et al., 2009)

    Quadrapole moment (DA) (Tagliabue et al., 2009)yogenic process for liqueed natural gas production. The required operations and

    he plant, and product specications; and may not all be required (MCHEmain

    CH4 CO2 N2

    3.80 3.30 3.641.2. Overview of gas separation mechanisms

    Separation processes can be designed to exploit differences inthe molecular properties or the thermodynamic and transportproperties of the components in the mixture. Molecular proper-ties that could be exploited to achieve a separation of CO2, N2 andCH4 include the differences in kinetic diameter, polarizability,quadrupole and dipole moments of the molecules (Table 2).Thermodynamic and (interphase) transport properties that couldbe exploited include vapour pressure and boiling points, solubi-lity, adsorption capacity and diffusivity. Based on these propertiesof the components to be separated, the primary operations for theseparation and purication of gases apply one of the followinginherent separation mechanisms: (1) phase creation by heat trans-fer and/or shaft work to or from the mixture, (2) absorption in aliquid or solid sorbent, (3) adsorption on a solid, (4) permeationthrough a membrane and (5) chemical conversion to anothercompound (Kohl and Nielsen, 1997; Seader and Henley, 2006).The rst four of these operations are discussed in this review;direct chemical conversion of CO2 from natural gas to a usefulproduct is beyond the current scope. An example of such directconversion, which is attracting signicant current research inter-est, is the dry reforming process where CO2 reacts with CH4 toproduce syngas (a mixture of H2 and CO2) which can then be usedfor the production of liquid fuels via FischerTropsch reactions.Recent reviews on the conversion of CO2 include articles byHavran et al. (2011), Song (2006) and Zangeneh et al. (2011).Separations involving the creation of a phase include the partialcondensation or partial vaporisation of species with very differentvolatility from a feed mixture; and this category can also include

    111.7 77.3

    3.80 304.1 126.2

    4600 7380 3400

    8.17 26.1 5.58

    2.448 2.507 1.710

    0.02 4.3 1.54

  • desublimation of CO2. If the volatility differences among speciesto be separated are not sufciently large, as is the case forN2CH4, to achieve the desired separation in a single partial vapor-isation or partial condensation contact stage, then a distillationprocess involving multiple vapour-liquid contact stages may berequired.

    One of the most fundamental properties of an inherent separa-tion mechanism is its selectivity with respect to the components iand j being separated. As will be discussed, some separationmechanisms have more than one type of selectivity; however, themost common and applicable type is the equilibrium selectivity, aij,which is dened in terms of phase compositions:

    aij xixj

    yjyi

    1

    Here xi and xj are the mole fractions (or convenient concentra-tion units) of components in one equilibrium phase, and yi and yjare the mole fraction of components in a second equilibrium

    Here C1i and C2i are the concentrations of component i in each of

    the product streams, and C1j , C2j the concentrations of component j

    in each of the product streams. The symbol C for concentration hasbeen used here to distinguish Eq. (2) as the separation power of theprocess from the equilibrium selectivity of the mechanism, althoughin many cases the composition of the product streams used in thecalculation of SPij may be in units of mole fraction. In principle anyselective mechanism with aij41 could be engineered to achieve agiven separation power if no constraints exist on capital, operationaland energy costs. In practice a separation process will be selected if itcan achieve the separation power necessary to meet the desiredproduct specications at a cost that is (i) lower than other processingalternatives, and (ii) economically viable in terms of theproducts value.

    In Table 3, indicative values of aij and SPij are given for the mostcommon industrial processes used to separate CO2 or N2 from naturalgas for the purpose of evaluating the prospects of new and emergingtechnologies. To generalise this measure for any of the four primaryseparation operations the product streams must be dened. For CO2removal from natural gas, a convenient approach may be to compare

    CH

    ion

    2 fr

    10,

    T.E. Rufford et al. / Journal of Petroleum Science and Engineering 94-95 (2012) 123154126phase. This denition of equilibrium selectivity derives from itscentral role in distillation theory, where it is also known as therelative volatility (McCabe et al., 2005). However, the equilibriumselectivity can be applied more broadly than to vapourliquidequilibrium in distillation; it is useful in the analysis of adsorp-tion- and absorption-based processes. The equilibrium selectivitycan also be applied, through the use of appropriate analogies anduse of concentrations of the permeate (x) and feed (y) streams, tothe analysis of membrane processes which are not strictlyequilibrium-based. It is important to note that selectivity is oftenlinked to the characteristic energy of the separation mechanism,and therefore the regeneration energy needed if the mechanism isto be used in a cyclical process.

    The performance of a separation process is governed by twofactors: the inherent selectivity of the separation mechanism beingutilised and the degree to which that mechanism can be exploitedthrough appropriate engineering design. This gives rise to anothermetric, known as the separation power, SPij (Seader and Henley,2006), which can be used to quantify the performance of an entireseparation process in terms of the product stream compositions. Fora single stage equilibrium operation aij SPij; however, for a processutilizing multiple separation stages SPij values much larger than thesingle stage equilibrium selectivity can be achieved.

    SPij C1iC1j

    0@

    1A C2j

    C2i

    !2

    Table 3Inherent equilibrium selectivity, ai,j, for the separation of CO2 from CH4 and N2 fromin example technologies implementing these separation operations. The separat

    containing 5% CO2 to produce pipeline quality gas with 2% CO2, and (b) to reject N

    Process Separating agent

    (a) CO2/CH4 separation

    Amine absorption (MDEA) Liquid absorbent

    Physical solvent (chilled methanol) Liquid absorbent

    AdsorptionCO2 selective Solid adsorbent

    MembraneCO2 selective Membrane

    (b) N2/CH4 separation

    Cryogenic distillation Heat transfer

    AdsorptionN2 selective Solid adsorbent

    AdsorptionCH4 selective Solid adsorbent

    MembraneN2 selective Membrane

    MembraneCH4 selective Membrane

    a Inherent kinetic selectivities of narrow pore adsorbents are reported from 2 toengineered.the aCO2 ,CH4 values of the various separation technologies by deningthe equilibrium phases to be used with Eq. (1) as the CO2 selectivephase: that is the amine solution or physical solvent for absorption,the high-purity liquid CO2 product for distillation, the adsorbed phase(xi or qi) for gassolid adsorption systems, and the permeate streamfrom a CO2 selective membrane stage. Using available data for CO2removal processes, such as the amine absorption data included inKohl and Nielsen (1997), the typical separation power of processes toproduce pipeline quality gas (2% CO2) from a feed mixture containing5% CO2 can be estimated with Eq. (2) as shown in Table 3. Theinherent equilibrium selectivity can be exploited for each separationmechanism to achieve SPijbaij through the arrangement of multipleseparation stages into an engineering process system, for example:trayed absorption and distillation columns, multiple adsorption bedsoperating in adsorptiondesorption cycles, and membrane stageswith interstage recompression and recycle loops. This simpliedanalysis shows clearly that the typical inherent process separationpowers for chemical absorption technologies are much larger thanthe best alternatives currently available to treat large gas ows; henceamine absorption is the most commonly applied type of AGRUdespite the energy required for regeneration and corrosive nature ofthe amine solutions.

    For N2 rejection technologies the conventional cryogenic distilla-tion technology has a separation power more than 8 times that of N2selective adsorption and membrane processes. However, the

    4 by different operations, with the typical process separation power (SPi,j) achieved

    powers are calculated for typical processes to (a) remove CO2 from a feed gas

    om a gas containing 4% N2 to produce a stream with 1% N2 for LNG production.

    Typical inherent

    equilibrium selectivity

    Typical process

    separation power

    aCO2 ,CH4 SPCO2 ,CH4860 3300

    318 1900

    28.5 622

    1520 2040

    aN2 ,CH4 SPN2 ,CH458 320

    1.32a 840a

    0.250.5 1.052

    23 210

    0.250.3 210

    which could allow a N2 selective process with much higher separation power to be

  • separation of N2 from CH4 in adsorption-based processes isenhanced by the differences in rates of diffusion of N2 and CH4 thathave been reported for small pore titanosilicate ETS-4 materials(Marathe et al., 2004a). Guild Associates Molecular GateTM pressureswing adsorption (PSA) process (Guild Associates, 2007) is acommercial example of a N2 rejection process that relies on thekinetic selectivity of N2 over CH4.

    In practice, the achievable separation power may be muchlower than the ideal separation powers estimated in Table 3.Furthermore, there are many other factors beside a separationpower or inherent selectivity that a process engineer designing orselecting a gas separation process will need to consider. Otherfactors that inuence the selection of process technology for CO2/CH4 and N2/CH4 separations include the level of contaminants inthe feed gas, the required level of contaminant removal orproduct purity, the ow rate and condition of the feed gas(temperature, pressure, water content), and for AGRUs the needfor simultaneous or selective removal of H2S. Process selection isalso inuenced by the available disposal routes for the removedcontaminants, which may include reinjection of CO2 for enhancedoil recovery (EOR) or enhanced gas recovery (EGR), and venting ofN2 to atmosphere. The process plant layout and available plant

    new technology and the conventional one with which it willcompete.

    2. Absorption

    This section focuses primarily on CO2 absorption processes,but also introduces technologies for N2 rejection by the selectiveabsorption of CH4 in hydrocarbon solvents (Mehra and Gaskin,1997) and the potential for the development of N2 selectivesolvents. Although this review is concerned primarily with CO2removal, the selection of the AGRU process is more oftendetermined by the H2S removal requirements. Thus, most of thetechnical literature concerning acid gas treating focuses on theabsorption of H2S. Commonly, the capacity of a sorbent isreported as an acid gas loading capacity which includes thecapacity for CO2 and H2S. We have noted in Tables 5 and 10which of the process technologies are suitable for the simulta-neous or the selective absorption of H2S.

    A large number of commercial processes are available for CO2absorption in chemical and physical solvents, including thetechnologies listed in Table 4. Chemical absorption processes

    ing.

    ine

    cele

    red

    carb

    carb

    carb

    lene

    yeth

    2-py

    thanol www.lurgi.com

    thanol Larue and Lebas, (1996)

    DIP

    se

    T.E. Rufford et al. / Journal of Petroleum Science and Engineering 94-95 (2012) 123154 127space for the separation process must be considered. If theseparation process is to be installed on an offshore platform, aoating LNG (FLNG) plant or as a retrot in an existing productionplant, then the process footprint the plan area and/or heightoccupied by the process equipment may inuence the choice ofprocess technology. Each of these process selection criteria mayhave material impacts on the feasibility, energy requirements andcosts of CO2 and N2 removal processes.

    There are many factors that inuence the cost of a separationprocess including the extent to which it has been usedsuccessfully in the past. Consequently, a process engineer design-ing or selecting a gas separation process is likely to be moreinterested in the ratio of its separation power to its cost ratherthan in the inherent selectivity upon which the process is based.However, once a separation process is sufciently mature, itsseparation power to cost ratio will generally only improveasymptotically, unless a signicant improvement in its inherentselectivity can be achieved. Thus, the starting point for scientistsand engineers aiming to develop a new separation technologyshould be an analysis of the inherent selectivities of both the

    Table 4Examples of commercial absorption technologies for CO2 capture and gas sweeten

    Vendor/licensor Sorbent

    Chemical absorptionEconamineSM Fluor Digylcolam

    ADIP-X Shell MDEAacaMDEAs BASF MDEA

    GAS/SPEC Ineos MDEA

    UCARSOL DOW MDEA

    KM CDR Mitsubishi Heavy Industries KS-1 hinde

    Beneld UOP Potassium

    Catacarb Eickmeter and Associates Potassium

    Flexsorb HP Exxon Mobil Potassium

    Physical absorptionFluor SolventSM Fluor Dry propy

    Selexol UOP/DOW Mixed pol

    Purisol Lurgi n-Methyl-

    Rectisol Lurgi Chilled me

    Ifpexol IFP Chilled me

    Mixed-solvent processesSulnol-D Shell SulfolaneAmisol Lurgi MethanolAwater www.shell.com (Rajani, 2004)condary aminewater Kohl and Nielsen, (1997)with amine solutions are the most commonly used acid gasremoval technologies in the natural gas industry (GPSAEngineering Data Book, 2004). The chemical absorption processesrely on reactions of the CO2 with the sorbent to form weaklybonded intermediate compounds, and these reactions can bereversed by the application of heat to release the CO2 andregenerate the sorbent (Olajire, 2010). Physical solvents, such asthe mixture of polyethylene glycoldimethyl ethers used in theSelexols process, selectively absorb CO2 from the natural gas feedaccording to Henrys law so that absorption capacity increases athigh pressure and low temperature.

    The two major cost factors in gasliquid absorption processesare (1) the required sorbent circulation rate, which is determinedby the amount of CO2 that must be removed from the feed gasand the CO2 loading capacity of the sorbent, and (2) the energyrequired to regenerate the sorbent (Kidnay and Parrish, 2006).The acid gas loading capacity of physical solvents at low tomoderate CO2 partial pressures is generally lower than that ofchemical absorbents. However, physical solvent processes havelower energy requirements for regeneration because the heat of

    Reference

    , mono-ethanolamine http://www.uor.com

    rator www.shell.com

    www.basf.de

    www.gasspec.com

    http://www.oilandgas.dow.com

    amine Mimura et al., (1995)

    onate UOP Overview of Gas Processing

    Technologies and Applications, (2010)

    onate (organic additive) www.catacarb.comonate (steric amine) www.exxonmobil.com

    carbonate www.uor.com

    ylene glycol dimethyl ethers www.uop.com

    rrrolidone www.lurgi.com

  • ads

    T.E. Rufford et al. / Journal of Petroleum Science and Engineering 94-95 (2012) 123154128absorption for physical solvents is much lower than the heat ofabsorption for chemical solvents.

    2.1. Chemical absorption processes for acid gas treating

    2.1.1. Aqueous amine processes

    Amines are organic compounds derived from ammonia (NH3)where one or more hydrogen atoms have been substituted withan alkyl or aromatic group. It is the (NH2) functional group of theamine molecule that provides a weak base that can react with theacid gases. The absorption of CO2 occurs via a two-step mechan-ism: (1) the dissolution of the gas in the aqueous solution,followed by (2) the reaction of the weak acid solution with theweakly basic amine. The rst physical absorption step is governedby the partial pressure of the CO2 in the gas feed. The reactionsinvolved in the second step of CO2 absorption in aqueous amineshave been widely studied, with a large number of reference

    Fig. 2. Process ow diagram of a typical amine-solvent (MDEA)-based chemical(Hubbard, 2010; Kohl and Nielsen, 1997).materials on the reaction mechanisms (Bindwal et al., 2011;Kohl and Nielsen, 1997; Penny and Ritter, 1983; Vaidya andKenig, 2007; Versteeg et al., 1996) and guidelines for processoperation (GPSA Engineering Data Book, 2004) available in theliterature. The fundamental reactions involved in CO2 absorptionin amine treating are (Kohl and Nielsen, 1997):

    Dissociation of water:

    H2O"HOH (3)

    Hydrolysis and dissociation of dissolved CO2:

    CO2H2O"HCO3H (4)

    Protonation of the amine:

    RNH2H"RNH3 (5)

    Carbamate formation:

    RNH2CO2"RNHCOOH (6)

    The dissociation reactions are shown here to highlight that thepH of the amine solution is an important process parameterbecause the concentrations of the ionic species H , OH- and HCO3

    -

    in the amine solution affect the other reactions involvingthe amine.Amines can be classied according to the number of hydrogenatoms that have been substituted, as primary (RNH2, where R isa hydrocarbon chain), secondary (RNHR0) or ternary (R0NRR00). For primary and secondary amines, such as monoethanola-mine (MEA) and diethanolamine (DEA), the carbamate formationreaction (Eq. (6)) predominates; this reaction is much faster thanthe CO2 hydrolysis reaction (Eq. (4)). The stoichiometry of thecarbamate reaction suggests that the capacity of primary andsecondary amines is limited to approximately 0.5 mol of CO2 permole of amine. However, DEA-based amine processes can achieveloadings of more than 0.5 mol of CO2 per mole of amine throughthe partial hydrolysis of carbamate (RNHCOO-) to bicarbonate(HCO3

    ), which regenerates some free amine (Kidnay and Parrish,2006).

    Tertiary amines such as MDEA, which do not have a freehydrogen atom around the central nitrogen, do not react directlywith CO2 to form carbamate. Instead, CO2 reactions with tertiary

    orption system for the separation of CO2 and other acid gases from natural gasamines proceed via equivalent reactions to those shown in Eqs.(4) and (5), which are much slower than the reaction in Eq. (6), togive the overall reaction:

    RR0R00NCO2H2O"RR0R00NHHCO3 (7)

    The stoichiometry in Eq. (7) shows that theoretically tertiaryamines can achieve a loading of 1 mol of CO2 per mole of amine,which is double the CO2 loading capacity of primary amines. Alsothe required heat of regeneration is lower for tertiary amines. Adisadvantage of tertiary amines is that the absorption kinetics areslower than for primary and secondary amines. For some naturalgas treating processes the slow kinetics of CO2 absorption intertiary amines can be utilised to achieve selective H2S removalby optimisation of the contact time in the absorber to minimiseCO2 uptake (GPSA Engineering Data Book, 2004). Alternatively, toenhance CO2 the absorption kinetics of tertiary amines anactivator (usually a primary or secondary amine) may be addedto increase the rate of hydrolysis of carbamate and dissolved CO2(GPSA Engineering Data Book, 2004).

    Since the rst application of tertiary amines in the mid-1970s,signicant research has been directed into the further develop-ment of novel amine solvents. To accelerate the reaction rate oftertiary amines with CO2, a primary or secondary amine canbe included as an activator (GPSA Engineering Data Book, 2004).

  • For example, the cyclic diamine piperazine has been studied as apromoter to improve the CO2 mass transfer rates of MDEA andMEA (Bishnoi and Rochelle, 2002); a commercial example of thistechnology is the aMDEA solvent from BASF. The piperazinereacts rapidly with CO2 in the vapour phase, which acceleratesthe dissolution of CO2 into carbonic acid, which can then reactquickly with MDEA. Sterically hindered amines are either primaryor secondary amines with large bulky alkyl or alkanol groupsattached to the nitrogen (Seagraves and Weiland, 2011), whichreduces the carbamate stability. The molecular congurationdictates the amount of CO2 removal: severely hindered amines,such as ExxonMobils Flexsorb SE, have very low rates of CO2absorption and allows selective H2S removal. In contrast, moder-ately hindered amines, such as 2-amino, 2-methyl, 1-propanol(AMP), are characterised by high rates of CO2 absorption and highcapacities for CO2. Weiland et al. (2010) evaluated the use of AMPwith MEA for CO2 capture from ue gases and found it had severaladvantages including at least a 15% reduction in the requiredregeneration energy.

    A typical process ow diagram for the removal of acid gas froma sour gas feed using methyldiethanolamine (MDEA) is shown inFig. 2. Properties and typical operating conditions for commonly

    gas per mole of amine (GPSA Engineering Data Book, 2004). Thepressure of the rich amine stream is reduced to around 6 bar in aash tank, to separate any dissolved hydrocarbons from the richamine, and preheated to 80105 1C before entering the stripping orregenerator column. In the stripping column, heat supplied by asteam reboiler generates vapour, which removes the CO2 from therich amine as the vapour travels up the column. A stream of leanamine is removed from the bottom of the stripper, cooled toapproximately 40 1C and recycled to the amine contactor. Thevapour stream from the top of the stripping column is cooled tocondense and recover water vapour, and the acid gas may be vented,incinerated, sent to a sulphur recovery plant (for H2S rich feed gas)or compressed for reinjection into a suitable reservoir for enhancedoil/gas recovery (Hughes et al., 2012).

    The process disadvantages with conventional amine treatingprocesses include: (1) the large amounts of energy required forregeneration of the amine, (2) the relatively low CO2 loadingcapacity of amines requires high solvent circulation rates andlarge diameter, high-pressure absorber columns, (3) the corrosiveamine solutions induce high equipment corrosion rates, (4) degra-dation of amines to organic acids, and (5) co-absorption ofhydrocarbon compounds such as benzene, toluene, ethylbenzene

    Di

    DE

    ye

    pa

    ye

    SN

    so

    CS

    30

    0.2

    0.2

    0.0

    11

    71

    T.E. Rufford et al. / Journal of Petroleum Science and Engineering 94-95 (2012) 123154 129used aqueous amine solutions are shown in Table 5. The basicprocess ow for other amine absorption systems is similar to thatshown for MDEA, although some commercial process designsoften feature multiple column feeds and contactor sections. Anyliquids or solids in the sour feed gas are removed in an inletseparator before the gas enters at the bottom of amine contactor.Typical operating pressures for amine contactors are in the rangeof 5070 bar (Kidnay and Parrish, 2006). The lean amine solution,typically an aqueous solution containing 1065%wt amine, is fedat the top of the column. As the amine solution falls down thecontactor and mixes with the gas, the acid gases dissolve andreact with the amine to form soluble carbonate salts. Thesweetened natural gas leaves the top of the contactor saturatedwith water and so dehydration is normally required before thegas is sold or fed to a cryogenic gas plant. Process temperaturesinside the contactor rise above ambient temperature due to theexothermic heat of absorption and reaction, with a maximumtemperature observed near the bottom of the column.

    The rich amine leaves the bottom of the contactor at a tempera-ture of approximately 60 1C and containing 0.200.81 mol of acid

    Table 5Properties of common aqueous amine solvents for acid gas treating.

    Solvent Monoethanolamine

    Acronym MEA

    Normally capable of meeting H2S specication

    (Kidnay and Parrish, 2006)

    yes

    Removes COS, CS2, mercaptans (Kidnay and Parrish,

    2006)

    partial

    50 ppm CO2 for cryogenic plant feed (Kidnay and

    Parrish, 2006)

    no, 100 ppm possible

    Solvent degradation concerns (components) (Kidnay

    and Parrish, 2006)

    yes - COS, CO2, CS2, SO2,

    SO3, mercaptans

    Solution concentrations, normal range wt% (GPSA

    Engineering Data Book, 2004)

    15-25

    Acid gas pickup, mole acid gas / mole amine (GPSA

    Engineering Data Book, 2004)

    0.33-0.40

    Rich solution acid gas loading, mol/mol amine

    normal range (GPSA Engineering Data Book, 2004)

    0.45-0.52

    Lean solution acid gas loading, mol/mol normal

    range (GPSA Engineering Data Book, 2004)

    0.12

    Stripper reboiler normal range, 1C (GPSA EngineeringData Book, 2004)

    107-127

    Approximate integral heats of absorption of CO2, kJ/

    mol Kohl and Nielsen, 1997

    84.4and xylene (BTEX) which subsequently are emitted with the acidgas stream (Collie et al., 1998; Morrow and Lunsford, 1997).Operational issues also include solution foaming, emulsions,excessive solution losses, heat stable salts and high-lter changeout frequency (Seagraves and Weiland, 2011). Aqueous ammoniaand hot carbonate systems are among the alternative chemicalabsorption processes to amines. However, many of the disadvan-tages of amine treating are also associated with aqueous ammo-nia and hot carbonate processes.

    Future innovations in conventional absorption column tech-nology (e.g., tray and packing designs) could be expected toachieve only incremental improvements in process efciencies(MacDowell et al., 2010). However, one promising strategy tointensify the CO2 absorption process is the use of a hollow bremembrane as a gasliquid contactor device (Cai et al., 2012;Ebner and Ritter, 2009; Favre and Svendsen, 2012; Zhou et al.,2010). In this concept, the membrane does not show anyselectivity for CO2 over CH4; instead the membrane provides aphysical barrier between the gas and liquid phases, and a largeinterfacial surface area for mass transfer of CO2. The selective

    ethanolamine Digylcolamine Methyldiethanolamine

    A DGA MDEA

    s yes yes

    rtial partial partial

    s, 50 ppmv in

    EA-DEA

    no, 100 ppm

    possible

    no, pipeline

    quality only

    me - COS, CO2,

    2, HCN, mercaptans

    yes - COS, CO2,CS2 no

    -40 50-60 40-50

    0-0.80 0.25-0.38 0.20-0.80

    1-0.81 0.35-0.44 0.20-0.81

    1 0.06 0.005-0.01

    0-127 121-132 110-132

    .6 83.9 58.8

  • T.E. Rufford et al. / Journal of Petroleum Science and Engineering 94-95 (2012) 123154130absorption of CO2 in the liquid solvent occurs at the liquidinterface of the membrane. For the hollow bre membrane tooperate effectively as a gasliquid contact device the pores mustremain gas lled and preventing liquid penetration into themembrane pores is one of the practical challenges hindering thecommercialisation of this technology (Favre and Svendsen, 2012).A gasliquid membrane contactor pilot plant is reported to havebeen tested in Scotland to remove CO2 from a 5000 Nm

    3/h naturalgas feed (Mansourizadeh and Ismail, 2009). Other improvementsin the performance of CO2 absorption processes are likely to comethrough the development of new solvent materials. Some acid gastreating process licensors are also working to develop chemicaladditives to inhibit corrosion and solvent degradation, so thatamine-based solvents can be operated at higher amine concen-trations and regenerated at higher temperatures (Normand et al.,2012). Improved understanding of the thermodynamics andkinetics of CO2-amine systems, including the development ofmass transfer rate-based modelling approaches, is also allowingoptimised design of absorption processes.

    2.1.2. Hot carbonate (alkali salt) systems

    Technologies using hot solutions of potassium carbonate(K2CO3) or sodium carbonate (NaCO3) have been employed sincethe 1950s to remove CO2 from high pressure gas streams (Kohland Nielsen, 1997). The overall reactions for CO2 with potassiumcarbonate can be represented by (GPSA Engineering Data Book,2004):

    CO2(g)K2CO3H2O(l)"2KHCO3(s) (8)

    The basic potassium carbonate process was developed by theUS Bureau of Mines and commercialised as the Beneld Process in1954, and now licensed by UOP with over 700 units constructed(UOP Overview of Gas Processing Technologies and Applications,2010). Other commercial potassium carbonate technologies com-peting with the Beneld Process include the Catacarb Process(Eickmeter and Associates), which is mainly used in the ammoniaindustry, and the Flexsorb HP Process (Exxon Research andEngineering).

    The process ow diagram for a potassium carbonate absorp-tion system shares many features with the general amine processow diagram shown in Fig. 2. In a typical hot carbonate processdesign the absorber and stripping columns operate in a tempera-ture range of 100116 1C (GPSA Engineering Data Book, 2004). IfH2S removal is required or if low CO2 concentrations are requiredin the product gas, then alternative designs with a two-stagecontactor and a lean-solution pumped to the middle of theabsorber may be used. For gas treatment requiring CO2 removalto low levels for cryogenic gas processing, or GTL plant feed, theUOP process design can be modied to a Hi-Pure design thatcombines the potassium carbonate process and an amine process(UOP Overview of Gas Processing Technologies and Applications,2010; Miller et al., 1999). Amines such as DEA and MEA are alsoused as activators to increase the rate of absorption of CO2 in thepotassium carbonate solution (Kohl and Nielsen, 1997). TheCatacarb Process is characterised by the use of a proprietaryorganic additives to improve mass transfer rates (Kohl andNielsen, 1997). The distinguishing feature of the Flexsorb HPProcess is the use of a sterically hindered amine as the activator(Kohl and Nielsen, 1997) which is claimed to improve CO2 loadingcapacity and mass transfer rates.

    2.1.3. Aqueous ammonia solvents

    Similar to the acid gas absorption processes using aqueousamine solutions, processes based on the reaction of CO2 with

    ammonia (NH3) in solution have been developed for the captureof CO2 from natural gas, coal seam gas, and post combustion uegases (Darde et al., 2010; Gonzalez-Garza et al., 2009). There aretwo variants of the aqueous ammonia process (AAP) reported:(i) chilled AAP designs operating with absorber temperatures inthe range 020 1C and (ii) processes operating with absorber atambient temperatures (2540 1C). Both variants are based on thesame reactions of CO2 and ammonia (NH3) described by Bai andYeh (1997), but at low temperatures the chilled AAP allows theprecipitation of ammonium bicarbonate shown in Eq. (9):

    CO2(g)NH3(l)H2O(l)"NH4HCO3(s) (9)

    A further advantage of the chilled AAP design is that absorberoperation at low temperatures reduces ammonia slip into thesweetened gas.

    The process ow scheme of an AAP plant is very similar to theow scheme for amine absorption cycles (Fig. 2) with an absorp-tion column and a solvent regeneration system. The absorptioncolumn in the AAP operates at low pressures, usually close toambient, and the CO2-rich slurry leaving the bottom of theabsorber column must be pumped to a high-pressure, hightemperature regeneration column (Gal, 2006). In the regenerationcolumn the ammonium bicarbonate solid can be decomposed toNH3 and CO2 at temperatures greater than about 50 1C (Dardeet al., 2010; Olajire, 2010), although temperatures of 100150 1Care preferred in some designs (Gal, 2006). Typical AAP solventconcentrations are in the range 1330% wt NH3 (Kim et al., 2008;Olajire, 2010).

    Although the aqueous ammonia process has potentially lowerenergy requirements than amine absorption processes (one studyon AAP for postcombustion CO2 capture suggests 21003100 kJ/kg CO2 compared to 3700 kJ/kg for CO2 capture by MEA, Dardeet al., 2010), the energy savings are not sufciently large to offsetthe additional costs associated with complexity of the ammoniaprocess (Kohl and Nielsen, 1997) and the need to recompress thesweetened gas in AAPs. Also, the removal efciency of chilledAAPs is only 90% (Gal, 2006), hence ammonia processes may notbe capable of achieving very low CO2 concentrations in productgas required for cryogenic gas processing.

    2.2. Physical solvent and hybrid solvent processes

    Physical solvent processes may be competitive with amineabsorption when the feed gas is available at high pressure(generally greater than about 20 bar) or when the acid gas partialpressure is 10 bar or greater (Nichols et al., 2009). For onshorenatural gas processing facilities all the commercial physicalsolvents listed in Table 4 could be used for bulk removal of CO2.Due to their large plant footprints, physical solvent technologiesare generally not suitable for AGRUs on offshore facilities (Nicholset al., 2009). To treat feed gas with very high CO2 concentrations,the leading physical absorption technologies include the Selexols

    and Rectisols processes (Burr and Lyddon, 2008).The regeneration of physical solvents can be achieved by

    reducing the pressure of the rich solvent stream in a series ofmulti-stage ash vessels, as shown in Fig. 3, or by stripping theabsorbed gas species in a regeneration column. Importantly, theheat inputs required for regeneration of a physical solvent aregenerally much lower than the heat required for the regenerationof amine or potassium carbonate sorbents. A potential short-coming of low pressure regeneration cycles is the cost of recom-pressing the acid gas if it is to be further processed for CO2sequestration, EOR or sulphur recovery.

    The Selexols process, based on a mixture of polyethyleneglycol-dimethyl ethers, is able to remove CO2 simultaneously

    with H2S and water (GPSA Engineering Data Book, 2004). In fact,

  • cess

    T.E. Rufford et al. / Journal of Petroleum Science and Engineering 94-95 (2012) 123154 131H2S has a greater solubility in most organic solvents than CO2, aproperty that can be used to design H2S selective processes (Kohland Nielsen, 1997). As many physical solvents also absorb water,in contrast to the aqueous amine, AAP and potassium carbonatetechnologies which saturate the sweet gas with water, therequired capacity of any dehydration units downstream of theAGRU may be smaller if a physical absorption process is used.

    To produce a sweet gas containing less than 50 ppmv CO2 forfeed to a LNG plant, the Rectisols process using a methanolsolvent operating at temperatures as low as 35 to 75 1C hasbeen applied successfully (GPSA Engineering Data Book, 2004). Areported example of a chilled methanol plant is at Riley Ridge,Wyoming, where 200 MMscfd of ultra-rich CO2 gas (70%) istreated (CO2 Extraction & Sequestration Project Riley Ridge, WY,2011).

    The main weakness of physical solvent technologies relative toamine AGRUs remains the issue of relatively low acid gasadsorption capacities of the commercially available physicalsolvents. Consequently physical solvent circulation rates are high;thus large diameter absorption columns and solvent circulationequipment are required in physical solvent processes. For onshore

    Fig. 3. Process ow diagram of a typical physical solvent progas processing facilities, the capital costs associated with the highsolvent circulation rates may be at least partially offset by thelower costs of the carbon steel materials required when usingnon-corrosive physical solvents compared to the more expensivematerials required to handle the highly corrosive aqueous aminesolutions (GPSA Engineering Data Book, 2004).

    Several mixed-solvent (also known as hybrid solvent) gastreating processes combine the effects of physical and chemicalabsorption processes in a single operation. The most well-knownmixed-solvent processes for CO2 absorption are the Sulnol-D

    s

    process licensed by Shell Global Solutions (Rajani, 2004) and theAmisol process licensed by Lurgi (Kohl and Nielsen, 1997). TheSulnol-Ds process based on a mixture of Sulfolane (tetrahy-drothiophene dioxide), DIPA (diisopropanolamine) and water iscapable of deep removal of CO2 to less than 50 ppm from feed gascontaining very high concentrations of CO2 (Kohl and Nielsen,1997). Another variation of the Shell Global Solutions technologyis the Sulnol-Ms process, which is also based on Sulfolane butuses MDEA instead of DIPA, used mainly for selective removal ofH2S. The process ow scheme of the Sulnol-D

    s process isessentially the same as that for an amine absorption process(Fig. 2) with the addition of a ash tank to remove the bulk of theacid gas from the rich solvent upstream of the stripper column.The Amisol process is based on a mixture of methanol, water andeither diethylamine (DETA) or diisopropylamine (DIPAM). Thisprocess has most commonly been used for purication of synth-esis gas derived from coal, peat, or heavy oils (Kohl and Nielsen,1997). The advantages of the hybrid solvent technologies overconventional amine absorption technologies include low energyconsumption for regeneration of the solvent, high acid gas loadingcapacities, low foaming tendency, and reduced corrosion. Hybridsolvent processes are usually only suited for treatment of naturalgas with an acid gas partial pressure of more than 100 kPa. Themain drawback of the mixed-solvent processes is that hydrocar-bon losses (to the solvent) are slightly higher than the typicallosses in conventional amine processes.

    2.3. Ionic liquids and switchable solvents

    Among the materials investigated as new solvents for CO2absorption processes, ionic liquids (ILs) are one of the solventsthat may in the future offer an alternative to amines and the lowcapacity physical solvents. Ionic liquids are commonly dened asorganic salts with melting temperatures of less than 373 K. They

    for absorption of CO2 and other acid gases from natural gas.have a range of properties that may make them useful replace-ments for volatile organic solvents such as extremely low vapourpressure, nonammability, and in many cases low toxicity (Zhao,2006). Furthermore, ILs are often considered to be designersolvents because of the many different possible combinations ofcations and anions which can be used to tune their chemical andphysical properties (Brennecke and Gurkan, 2010). These proper-ties of ILs have generated great scientic interest in their devel-opment and investigation into their use in chemical engineeringapplications, including gas separations. Reviews by Zhao (2006),Plechkova and Seddon (2008) and Werner et al. (2010) describethe present industrial processes which use ionic liquids anddiscuss many other possible industrial applications of ionicliquids. The application of ILs to CO2 capture and natural gassweetening has been discussed by many authors, including Baraet al. (2010b), Karadas et al. (2010), Brennecke and Gurkan (2010)and MacDowell et al. (2010).

    The development of ILs for CO2 capture and natural gas sweet-ening has been driven by the desire to develop a solvent with a CO2capacity comparable to that of amine-based solvents but withgreatly reduced energy requirements for regeneration. Initially,much of the research focussed on the potential of imidazolium-based ILs as alternative physical solvents. Early studies focussed on

  • T.E. Rufford et al. / Journal of Petroleum Science and Engineering 94-95 (2012) 123154132the most convenient ILs to synthesise, such as 1-n-alkyl-3-methy-limidazolium [Cnmim] cations paired with various anions includingbis(triuoromethylsulfonyl)amide [Tf2N] (Bara et al., 2010b; Hugheset al., 2011). However, the alkyl chains (in the imidazolium cation)are not the ideal functional group for separating CO2 from CH4 or N2and, thus, these alkyl chains were substituted with various groupscontaining either ethylene glycol or nitrile units to form[Rmim][Tf2N] (Bara et al., 2010b). Karadas et al. (2010) summarisedthe results of using anions other than [Tf2N] on the solubility of CO2in various ILs, with anions containing uorinated derivatives show-ing a modest increase.

    Despite these efforts to optimise the molecule, the solubilitiesachieved with ILs on a volumetric basis (dissolved moles of CO2 perliquid volume) have remained comparable with most commonorganic solvents (Bara et al., 2010b; Karadas et al., 2010). Muchresearch has thus focussed on the development of task specic ionicliquids (TSIL) which incorporate an amine functional group into theIL, enabling it to serve as a reactive, chemical solvent for CO2. Bateset al. (2002) reported an imidazolium-based TSIL containing an aminefunctional group attached to one of the alkyl chains. The stoichio-metry of the absorption reaction achieved was 0.5 mol of CO2 permole of TSIL; regeneration of the IL solvent was also achieved byheating the carbamate product to 80100 1C under vacuum. Gurkanet al. (2010) and Zhang et al. (2009) used TSILs with amino-acidfunctional groups that improve the stoichiometry to nearly 1 mol ofCO2 per mole of IL, which is important if the IL solvent is to becompetitive with the volumetric capacities of aqueous amine solu-tions (Brennecke and Gurkan, 2010). However, TSILs can be difcultto synthesise and have large viscosities at ambient temperature,which increase further upon complexation with CO2 (Brennecke andGurkan, 2010; Karadas et al., 2010). Brennecke and co-workerssuggest that the viscosity increase of TSILs upon complexation withCO2 can be eliminated through the use of aprotic heterocyclic anionsand have led a provisional patent (Brennecke and Gurkan, 2010).Nevertheless, the synthesis and viscosity challenges associated withTSILs currently limit their commercial viability.

    Currently, the most viable method of applying ILs to CO2 captureor natural gas sweetening is the use of ILamine mixtures, in whichMEA or DEA is dissolved in an [Rmim][Tf2N] solvent; such solutionscan have up to 116 times the CO2 solubility on a volumetric basisthan the IL alone (Bara et al., 2010a). Camper et al. (2008) found thatMEA-IL and DEA-IL solutions could rapidly and reversibly capture1 mol of CO2 per mole of amine and thereby reduce feed gas CO2concentrations to the ppm level, even at CO2 partial pressures below0.133 kPa. Such ILamine solvents offer signicantly reducedenergy requirements relative to conventional aqueous amine sol-vents: Bara et al. (2010a) compared a IL-amine process with ashregeneration (similar to the process ow scheme shown in Fig. 3),against a conventional amine-based gas sweetening plant (similar tothat shown in Fig. 2). They considered the sweetening of 100MMSCFD to a sales gas specication of 2% CO2 for an NG feedcontaining either 15% or 5% CO2 and concluded that over a 20 yrplant life cycle, the amine IL process had combined CAPEX andOPEX savings of about 25%. This was due primarily to the removal ofa regeneration column, the higher amine loadings and lower solventcirculation rates of the amine IL solvent, and the reduced dutiesrequired to cool, heat and regenerate the amine IL solvent. Thedifference in the caloric properties of water and the IL, and inparticular the duty reduction associated with not vaporizing anysolvent during regeneration, is probably the most signicant advan-tage of the amine IL solvent approach to gas sweetening. A pilotscale unit is under construction for planned NG sweetening eldtests in 20112012 (Bara et al., 2010a).

    Brennecke and Gurkan (2010) point out that all ILs arehygroscopic (that is, a material that can adsorb or absorb water

    molecules), even those that are usually designated hydrophobicbecause they are insoluble in water. This fact represents a seriouschallenge for the application of ILs to CCS but it also hasimplications for NG sweetening. If the NG contains some water,then the absorption of that water by the IL could decrease thesolvents capacity, and will also degrade the reduction in regen-eration duty associated with an amine IL solution. Thus incontrast with conventional gas processing practice, a sweeteningprocess utilising an amine IL solvent should probably be situ-ated downstream of the dehydration process.

    Jessop and co-workers reported the development of switch-able solvents (Jessop et al., 2005; Phan et al., 2008) where a basic,non-polar liquid mixture converts into a polar IL upon theaddition of CO2. Several groups are researching the optimisationof these switchable solvents for improved CO2 capture or NGsweetening. For example, Heldebrant et al. (2011) reported thedevelopment of 2nd generation switchable solvents that couldcapture of nearly 1.3 mol of CO2 per mole of solvent. In thesecases, the viscosity of resulting IL is appreciable and its reductionis the focus of ongoing research.

    2.4. Absorption processes for N2 rejection

    Nitrogen rejection using absorption-based technologies is nota common practice in the natural gas industry. There are severalcommercial N2 rejection processes that operate by the physicalabsorption of CH4 in a hydrocarbon oil which have been built toprocess gas feed rates of 230 MMscfd (AET-Technology, 2007;TGPE, 2009). The costs of large solvent rates and CH4 recompres-sion are prohibitively high for the use of CH4 selective absorptionNRUs in large scale LNG production plants; thus there is agrowing need for the development of a N2 selective absorbent.

    An example of a CH4 selective NRU process is that designedand constructed by Advanced Extraction Technologies, Inc. (AET,Houston USA). The ow scheme of the AET process is similar tothe physical solvent process shown in Fig. 3. In a typical CH4selective absorption process the feed gas is cooled before enteringthe absorption column where CH4 is dissolved into a lean oilsolvent (TGPE, 2009). The CH4 is then recovered from the richsolvent through a series of ash vessels to produce sales gascontaining less than 4% N2. The N2 from the feed gas owsthrough the top of the column and is used for precooling the unitfeed, and is then vented. The AET process operates at approxi-mately 32 1C to optimise the CH4 absorption in the solvent.However, the capacity of most solvents for CH4 is relatively lowand large solvent circulation rates are required to achieve eco-nomic recovery of the CH4. Furthermore, because the CH4 isrecovered from the solvent by ashing at low pressure the gasmust be recompressed for pipeline or cryogenic gas plant feed.

    Although most solvents have a greater solubility for CH4 thanN2, some organo-metallic complexes (OMCs) have been reportedto preferentially bind N2 from natural gas mixtures. Reversibletransition metal complexes forming with N2 were rst reported inthe scientic literature by Allen and Senoff (1965) who describedthe reversible reaction of hydrazine (a N2 source) with rutheniumchloride. Several scholarly articles describing the preparation andN2 absorption capacity of these types of transition metals formingcomplexes were also published in the 1970s (Allen et al., 1973;Chatt et al., 1978; Sellmann, 1974). However, there have been fewstudies reported on the application of these N2 binding complexesto natural gas processing. The most extensive applied studies ofN2 selective solvents are reported by Stanford Research InstituteInternational (SRI), on contract for the US Department of Energy(Alvarado et al., 1996), and Bend Research Inc. (Friesen et al.,2000; Friesen et al., 1993). Both the SRI process and the Bendprocess exploit the reversible chemical complexing abilities of

    multi-dentate transition metal complexes as shown in Fig. 4.

  • cryogenic LNG plant the boiling point (BP) of CH4 is 94.7 1C(Lemmon et al., 2010), which provides a sufcient difference inrelative volatility with N2 (NBP 195.8 1C, BP of 148.7 1C at

    ano-

    Fig. 5. Illustration of the binary CH4N2 vapourliquid equilibrium relationshipand the construction of a McCabeThiele diagram to calculate the number of ideal

    equilibrium stages for separation by distillation. The vapourliquid equilibrium

    data shown here represents a relative volatility, aN2CH4 , of 7 (calculated usingREFPROP (Lemmon et al., 2010) for an operating pressure of 2757 kPa. In this

    hypothetical example, the feed contains 50% N2 with product specications of 5%

    N2 in the CH4-rich bottom product and 5% CH4 in the N2-rich overheads product.

    These compositions were selected as an example which could be shown clearly on

    this gure and do not represent a typical set of operating conditions. To meet more

    realistic processing objectives, such as a higher purity CH4-rich bottom product

    and 98% recovery of the methane from the feed gas, a much larger number of

    equilibrium stages would be required.

    T.E. Rufford et al. / Journal of Petroleum Science and Engineering 94-95 (2012) 123154 133The absorption of N2 in OMCs is achieved through using metalions with six coordination sites, the four equatorial sites complexedwith bi, tri, or tetra dentate ligands; with an electron withdrawingion on one axial site leaves the other axial site ready to complexwith a gaseous N2 molecule in an end-on conguration. The strengthand therefore reversibility of the complex formation is dictatedthrough the use of different ligands and transition metals. The mostpromising OMC reported by SRI was a (bis)tricyclohexylphosphinemolybdenum tricarbonyl, in a toluene solution, which formed ayellow precipitate when bound with a N2 molecule (Bombergeret al., 1999). This phosphine complex absorbed up to 0.12 mmol ofN2/mL of solution at 1979 kPa (Alvarado et al., 1996) in equilibriummeasurements. However, during absorptionregeneration cyclesmeasured on a 0.02 MMscfd bench-scale apparatus the SRI research-ers encountered issues with the degradation of the phosphinecomplex and regeneration of the complex to release the N2. Theproblems with the regeneration of the phosphine complex includeddifculties in controlling the size of the N2-bound precipitates,which would bypass the regeneration system if the size was lessthan that of the phosphine solids (412 mm) (Bomberger et al.,1999). Although SRIs economic analysis suggests that this OMCprocess could become cost competitive with cryogenic NRUs, thechallenges to reduce the cost of OMC synthesis, improve the stabilityof the phosphine complex in the presence of water and oxygen, andto overcome the problems encountered with solids handling in theregeneration process are all signicant.

    The patents held by Bend Research Inc. (Friesen et al., 2000,1993) describe OMCs based on several transition metalligandcombinations including complexes with iron, as well as the perfor-mance of these complexes in N2 absorptionregeneration cycles. Thepatents report N2 uptakes of 0.5 mol N2/mol OMC at a N2 partialpressure of approximately 1130 kPa and a selectivity for N2 over CH4close to 6. These complexes are reported to exhibit a high degree ofstability, showing no decrease in capacity through repeated use over100 days, and exposure to an atmosphere of 3% CO2 and 100 ppmO2. Although the publically available information on the Bendprocess indicates the concept of N2 absorption in these OMCs maybe sound, no commercial process has been developed for thistechnology. The main barriers to the use of OMCs in large scale

    Fig. 4. General chemical scheme for the reversible binding of nitrogen with an orggas processing operations remain the high cost of synthesis of theOMCs and improving the chemical resistance of the complexes tocommon gas contaminants such as water and H2S. Furthermore, theBend Research Inc. iron-based OMCs may present safety andmaterials handling issues because these are pyrophoric compounds,which can ignite spontaneously on contact with air.

    3. Condensation, desublimation and distillation

    3.1. N2 rejection by cryogenic distillation

    The normal boiling point (NBP) of CH4 is 161.5 1C and at atypical pressure of 3150 kPa in the intermediate stages of themetallic complex adsorption (Miller et al., 2002). Rgeneric organic functionality.3150 kPa) for separation of CH4N2 mixtures by cryogenicdistillation. Relevant to such separations are vapourliquid equi-librium curves, such as the one shown in Fig. 5 for N2CH4 atconditions accessible in a cryogenic LNG plant. The equilibriumcurve in Fig. 5 represents a typical relative volatility aN2 ,CH4 valueof 58. To illustrate the separation of N2 and CH4 in the cryogenicdistillation column, we have applied the McCabeThiele method(McCabe and Thiele, 1925) to estimate the number of idealequilibrium stages required to produce a liquid CH4 bottomsproduct of 95% from a feed containing 50% N2 and 50% CH4. Thesefeed and production compositions were selected here to allowillustration of the separation process; in a real process to producea CH4 containing less than 1% N2 and with a high rate of CH4recovery, the number of actual equilibrium stages required in the

  • N2-rich vapour from the HP column is condensed to provide reuxfor both the high pressure column and the low pressure column. Thelow pressure column produces a high purity N2 stream (o1% CH4);which is used to cool N2 reux fed to the LP column in the LPcondenser, sub-cool the bottoms of the HP column and to pre-coolthe feed gas. The CH4-rich bottom product of the LP column ispumped through the crude sub-cooler, which provides the LPcolumns reboiler duty by condensing the overheads of the HPcolumn, vaporised and reheated against the feed gas. Alternatively,for LNG production the CH4-rich bottoms product of the LP columnmay remain liquid.

    The three-column NRU process consists of the double-columnprocess described and a prefractionation column. The prefractio-nation column recovers some of the hydrocarbons at a highertemperature than the double-column system and increases the N2concentration of the feed gas. Importantly, the prefractionationcolumn reduces the volume of N2-rich gas that must be processedat low temperatures in the main NRU column(s) and this reduc-tion in gas volume can signicantly reduce the power require-ments of the NRU (MacKenzie et al., 2002; Wilkinson andJohnson, 2010). A further variant of an NRU with a prefractionatordescribed by MacKenzie et al. (2002) is a two column processwith a high pressure prefractionation column, an intermediateliquidvapour separator and a low pressure column.

    3.2. Low-temperature CO2 removal processes

    The separation of CO2 from natural gas by low-temperature

    Fig. 6. A process ow schematic of a typical single-column N2 rejection unit(schematic adapted from GPSA Engineering Data Book, 2004; Agrawal et al., 2003;

    MacKenzie et al., 2002).

    T.E. Rufford et al. / Journal of Petroleum Science and Engineering 94-95 (2012) 123154134distillation column would be higher than the eight ideal traysshown in the McCabeThiele construction on Fig. 5. (The corre-sponding graph would be more difcult to read.) In practice,modern cryogenic NRUs (with SPN2 ,CH4 320) can produce veryhigh purity CH4 at high recovery rates, which also reduces the CH4content in the N2 overheads vapour to less than 3%. This isachieved through the use of columns with large numbers ofstages and the design of systems with multiple columns.

    Currently, cryogenic distillation is the only N2 rejection methodthat has been demonstrated at gas ows above 25 MMscfd toachieve very high methane recovery (typically above 98%) and highpurity N2 (approximately 1% CH4). The selection, and optimiseddesign, of a cryogenic NRU is principally determined by theconcentration of N2 in the feed gas (Wilkinson and Johnson, 2010).The feed gas pressure, feed gas ow rate, concentration of con-taminants and product specications (sales gas or LNG) alsoinuence process selections. The optimum design of cryogenicNRUs, like for most cryogenic processes, is an exercise in balancingthe energy efciency and process ow sheet integration to reducethe power consumption required for compression of the CH4refrigerant loops, which are used to provide the reboiler andcondenser duties (Finn, 2007; Wilkinson and Johnson, 2010). Thecompression requirements of this distillation-based separation arethe largest contributor to capital and operating costs of the NRUprocess. Cryogenic NRU processes have been constructed by most ofthe major process designers such as Linde, Costain, Praxair, Con-ocoPhillips and APCI. The main variants of cryogenic NRU designsare (1) the single-column heat-pumped process, (2) the double-column process (Agrawal et al., 2003) and (3) three or two columndesigns featuring a prefractionation column (MacKenzie et al., 2002;Wilkinson and Johnson, 2010).

    A typical single-column heat-pumped NRU process is illu-strated in Fig. 6. Upstream of the cryogenic NRU, contaminantsthat could freeze at cryogenic temperature such as water, CO2 andheavy hydrocarbons have been removed from the gas. The feedgas to the NRU is cooled, throttled and fed to an intermediatestage of the distillation column operating at pressures from 13002800 kPa (Agrawal et al., 2003). Rejected N2 vapour (typicallyo1% CH4) is drawn from the column overheads and the CH4-richliquid product is drawn from the bottom of the column. Thebottoms product then can be reheated against the NRU feed gas.A closed-loop CH4 heat-pump cycle driven by an external com-pressor provides the reboiler and condenser duties, with theclosed-loop CH4 condensed at a high pressure in the reboilerand revaporised at low pressure in the condenser.

    As the N2 content in the feed gas increases, the CH4 in the upperstages of the column becomes more difcult to condense. Theoperating exibility of a single-column NRU process is limited by(1) the critical pressures of nitrogenmethane mixtures, whichlimits the maximum pressure of the distillation column to approxi-mately 2800 kPa, and (2) the minimum practical temperature of CH4after the throttling valve of the heat-pump cycle (MacKenzie et al.,2002). These limitations mean that the single-column NRU processis, generally, used for feed gases containing less than 20% N2.

    A double-column NRU can provide additional process exibilitycompared to the single-column process to allow the separation ofgases containing higher N2 concentrations or gases in which the feedgas quality varies. In the double-column N2 rejection process theNRU feed gas is cooled, throttled and fed to a high pressure (HP)column operating typically at 10002500 kPa (Agrawal et al., 2003).Having had some of the N2 removed, the crude natural gas liquidstream from the bottoms of the high pressure column is sub-cooled,throttled and fed to the low pressure (LP) column (operating atapproximately 150 kPa). In practice both the HP and LP columns areusually integrated into a single tower to improve process heat

    integration and minimise heat transfer to the atmosphere. The processes (operating at temperatures below 0 1C) can be

  • CO2. The basic thermodynamic path for the CryoCell operationinvolves cooling a dry, feed gas (at 56006600 kPa) to just abovethe CO2 freezing point (for example to 60 1C) to condense someor all of the vapour, followed by an isenthalpic ash to furthercool the mixture to obtain solids, liquid CO2 and a CH4-richvapour. Pilot plant trials of the CryoCells process demonstratedthe production of pipeline quality gas from 2 MMscfd of feedgases containing 3.560% CO2 (Hart and Gnanendran, 2009). Thecryogenic separation of CO2 has potential as a highly selectiveprocess to treat CO2-rich natural gas, although research efforts toovercome the operational issues associated with control of theCO2 freezing and solids handling were underway in 2009.

    The key advantages of phase creation processes for CO2separation from CH4 over amine-based absorption systems forseparation of CO2 from CH4 include the recovery of a high purity,liquid CO2 product at a reasonable pressure, which facilitates thesubsequent transport or injection for use in EOR; the avoidance ofhighly corrosive aqueous amine solvents, and possibly, reducedprocess footprint and reduced hydrocarbon inventories, whichmay be important considerations for offshore or oating produc-tion facilities (Kelley et al., 2011).

    4. Adsorption

    The separation and purication of gas mixtures by the selec-tive adsorption of components from the gas mixture onto poroussolid adsorbents is an established process technology used inthe production of hydrogen, the separation of O2 and N2 from air,and the capture of odorous pollutants from various industrialprocesses. In the natural gas industry adsorption-based separa-

    T.E. Rufford et al. / Journal of Petroleum Science and Engineering 94-95 (2012) 123154 135categorised as: (1) gasliquid phase separations operating attemperatures above the CO2 triple point temperature of56.6 1C (Lemmon et al., 2010) and (2) gassolid phase separa-tions where desublimation of CO2 occurs at temperatures belowthe triple point. Although the term cryogenic is often used bythe vendors and in the literature to describe these types of CO2capture technologies, most of the processes operate at tempera-tures above the scientic denition of cryogenic as 153 1C(Agrawal et al., 2003; Radebaugh, 2007). To overcome theproblems associated with the formation of CO2 solids duringcryogenic distillation two technological approaches have beenpursued: (1) extractive distillation by the addition of a heavierhydrocarbon to alter the solubility of components in the column(Ryan/Holmes process) and (2) controlled freezing and re-meltingof the solids (Controlled Freeze ZoneTM and CryoCells processes).Other low-temperature CO2 removal technologies under devel-opment include systems in which mechanical methods are usedto separate the CO2 rich phase from the natural gas. For example,Willems et al. (2010) report the C3sep (condensed contaminantcentrifugal separation) process in which condensed CO2 dropletsare separated from the natural gas using rotational separators,and Clodic et al. (2005) describe the ALSTOM process whichfeatures a multi-stage thermal swing process that freezes thenmelts CO2 on mechanical ns. We focus our discussions in thisreview on the commercialised Ryan/Homes process and the pilot-plant demonstrated Controlled Freeze ZoneTM (along with thesimilar CryoCells process).

    The extractive distillation approach to solving the problem ofCO2 freezing in CH4CO2 distillation is most well known throughthe Ryan/Holmes process described in the 1982 US Patent4,318,723 (Holmes and Ryan, 1982). The Ryan/Holmes processis representative of several similar technologies patented byvarious other inventors. The addition of a heavier hydrocarbonstream (typically a C2C5 alkane) to the condenser of the distilla-tion column shifts the operation away from conditions that favoursolids formation, because the solubility of CO2 in the liquid phasecan be increased, the overheads temperature can be raised, andthe column can be operated at a higher pressure since themixtures critical pressure increases. A typical four columnRyan/Holmes process conguration incorporates a de-ethanisercolumn, a CO2 recovery column, a demethaniser column, and acolumn for recovery of the hydrocarbon additive. Further detailson Ryan/Holmes congurations and operating issues for separa-tions of methaneCO2, ethaneCO2, and CO2H2S are discussed inthe GPA Engineering Data Book.

    The Controlled Freeze ZoneTM (CFZTM) process was rst patentedby ExxonMobil in 1985 (Valencia and Denton, 1983) and tested in aTexas pilot plant during 19861987 (Nichols et al., 2009). Morerecently a commercial demonstration project designed to treat afeed gas of 14 MMscfd has been constructed in LaBarge, Wyoming(Controlled Freeze ZoneTMincreasing the supply of clean burningnatural gas, 2010). The separation tower of the CFZTM process is splitinto three sections with an upper rectication section and a lowerstripping section (both conventional distillation sections) separatedby the CFZTM section, as shown in Fig. 7 (Fieler et al., 2008). In theCFZTM section, the liquid falling from the rectication section iscontacted with a cold methane stream (90 to 85 1C), whichcauses the CO2 to freeze out of the methane mixture. The CO2 solids(62 to 45 1C) drop to a liquid layer on a melt tray in the lowerstripping section; the solids melt before falling as liquid through thedowncomers of the melt tray. The standard CFZTM process canproduce pipeline quality gas, and when implemented with amodied rectication section is claimed to be capable of producinga sweet gas of less than 50 ppm CO2 (Nichols et al., 2009).

    Cool Energys CryoCells was developed by researchers at

    Curtin University in Western Australia with industrial partnersWoodside Petroleum and Shell Global Solutions (Hart andGnanendran, 2009). Like the CFZTM process, the CryoCells processoperates by the controlled freezing and subsequent remelting of

    s

    Fig. 7. General schematic of the Controlled Freeze ZoneTM process with spraynozzles in the CFZ section (schematic adapted from Fieler et al., 2008).tions are used to remove water, sulphur, mercury and heavy

  • bent with a high surface area (such as microporous carbons)possessing a large number of adsorption sites is likely to be a goodcandidate.

    For gas separations based on differences in sorption rates akinetic selectivity factor bij which incorporates the effects of eachcomponents sorption mass transfer coefcient ki can be denedas follows:

    bij aijkikj

    s-

    yj yibij

    qiqj

    ! kikj

    s11

    As discussed recently by Ruthven (2011) the kinetic selectivitydepends on both the diffusivity ratio (assuming kipDc,i, whereDc,i is the diffusivity coefcient of component i in adsorbentpores) and the equilibrium selectivity, which if being inferredfrom pure uid measurements can be estimated from the secondequality in Eq. (11). Materials that exhibit a kinetic selectivity forCO2 or N2 from CH4 include carbon molecular sieves (Bae and Lee,2005; Cavenati et al., 2005), CuMOF (Bao et al., 2011a) andsmall-pore zeolites such as clinoptilolite (Ackley and Yang, 1991;Hernandez-Huesca et al., 1999).

    4.2. Adsorption-based separation processes

    Similar to the solvent absorption processes, adsorption-based

    T.E. Rufford et al. / Journal of Petroleum Science and Engineering 94-95 (2012) 123154136hydrocarbons (for dew point control) from the natural gas(Tagliabue et al., 2009). Adsorption-based processes for theseparation of mixtures of CH4, N2 and CO2 are also used forpost-combustion CO2 capture (Ebner and Ritter, 2009), the pur-ication of coal mine methane (Richter et al., 1985; Tonkovich,2004; US Environmental Protection Agency, 2008; US EPA, 1997)and coal mine ventilation air (VAM) (Warmuzinski, 2008), and forthe purication of biogas (Alonso-Vicario et al., 2010; Esteveset al., 2008). Central to the development and implementation ofadsorption-based processes are the various selectivity mechan-isms that give rise to the separation of components within the gasmixture.

    4.1. Adsorbent selectivity

    The preferential adsorption of components from a gas mixturecan be achieved by one, or a combination, of the followingmechanisms: (1) differences in the adsorbatesurface interactionsand/or adsorbate packing interactions when the system reachesequilibrium (thermodynamic equilibrium mechanism), (2) differencesin the size and/or shape of gas molecules leading to exclusion ofmolecules with a critical diameter too large to enter the adsorbentpores (steric mechanism) and (3) differences in the diffusion rates ofmolecules through the adsorbent pores (kinetic mechanism) (Li et al.,2009; Ruthven, 2011). The kinetic mechanism can include thequantum sieving effect of different diffusion rates observed forsome light molecules in narrow micropores (Xiao-Zhong et al.,2009). Most industrial adsorption processes such as NG dehydrationusing silica desiccants or molecular sieves (Kohl and Nielsen, 1997)rely on the thermodynamic equilibrium effect. The separation of N2from CH4 in the Molecular Gate

    TM PSA process using ETS-4 and theseparation of O2 and N2 from air using carbon molecular sieves andsmall pore zeolites (Kerry, 2007) are industrial examples based onthe kinetic mechanism. True steric or size exclusion-based processesare unlikely to be viable technologies for CO2/CH4 or N2/CH4separations because the differences in the critical diameters ofCO2, N2 and CH4 are not sufciently large for total exclusion ofone component from the adsorbent pores. Based on these adsorptivegas separation mechanisms there are two types of adsorbateselectivitythe equilibrium selectivity achieved in the limit of longtime periods and the kinetic selectivity (or time dependent selectiv-ity). The equilibrium selectivity aij of the adsorption mechanismdened in Eq. (1) can be written for an equilibrium selectiveadsorption process as:

    aij qi=yiqj=yj

    KiKj

    as yi,yj-0 10

    where qi and qj are the equilibrium adsorption capacities deter-mined from pure gas component isotherms, and yi and yj are themole fractions of the components in the gas mixture. In many cases,including at low partial pressures of the component species, theseparation factor can be estimated as the ratio of the Henrysconstants (Ki/Kj) (Tagliabue et al., 2009). This denition of aij servesas a useful tool to screen potential adsorbents for CO2 and N2removal from natural gas. However, to adequately design anadsorption-based separation process the selectivity of the adsorbentfor components from a real gas mixture must be conrmed and theworking capacity of the adsorbent needs to be evaluated (Ackleyet al., 2003). The working capacity is the difference between theamounts of a component adsorbed and desorbed at the conditions ofthe adsorption and desorption steps, and this capacity is inuencedstrongly by temperature and pressure.

    The equilibrium capacity for a gas species is inuenced by thestrength of the gassolid interaction and the number of availableadsorption sites. The strength of the gassolid interaction is

    determined by the characteristics of the adsorbents surfacechemistry and pore structure; and by the adsorbates propertiesincluding molecule size, polarizability and quadrupole moments.Typical heats of adsorption for CH4 and N2 on commercialadsorbents are in the range of 1522 kJ/mol (Cavenati et al.,2004; Watson et al., 2009; Xu et al., 2008). Carbon dioxideexhibits a large quadrupole moment, thus adsorbents with polarsurfaces that have a high electric-eld gradient, such as zeolites,have a stronger interaction with CO2 than with the non-polar CH4and N2 molecules (Li et al., 2009). For example, Cavenati et al.(2004) report the isosteric heat of adsorption of CO2 on zeolite13X is 37.2 kJ/mol, Xu et al. (2008) report 49.9 kJ/mol on Na b-zeolite, and Watson et al. (2012) report 44.9 kJ/mol on a naturalchabazite. Fig. 8 illustrates the differences in equilibrium adsorp-tion capacity at 298 K for CO2, N2 and CH4 on zeolite 13X. Forselective adsorption of a non-polar molecule like CH4, an adsor-

    Fig. 8. Equilibrium adsorption capacity of CO2, CH4 and N2 at 298 K on zeolite 13X.Figure constructed from data reported in Cavenati et al. (2004).processes for gas separation require both adsorption and

  • ssur

    atso

    T.E. Rufford et al. / Journal of Petroleum Science and Engineering 94-95 (2012) 123154 137regeneration stages. Adsorbent regeneration, or desorption, canbe achieved by utilizing the differences in adsorption capacities atdifferent temperatures (thermal-swing adsorption, TSA) and atdifferent pressures (pressure-swing adsorption, PSA), as illu-strated in Fig. 9. Continuous TSA and PSA processes operate withmultiple beds containing a stationary adsorbent and use a mani-fold of valves to switch gas ow to the beds corresponding toadsorption and desorption cycles. Less commonly used technol-ogies for continuous adsorption processes are uidised andmoving bed operations (Seader and Henley, 2006), and xed-bed electrothermal-swing adsorption (ESA) (An et al., 2011;Grande and Rodrigues, 2008).

    In the TSA method, the adsorbent is regenerated by desorption

    Fig. 9. Schematic representation of a two-bed cyclic adsorption process with preisotherms for CO2 adsorption on a carbon molecular sieve Takeda MSC 3K-171 (Wat a higher temperature than that used during the adsorptionphase of the cycle. In the natural gas industry TSA processes withsilica gel or zeolite molecular sieve lled adsorbent beds havebeen used widely for gas dehydration. The temperature of the bedcan be increased by purging the bed with a hot, inert and non-adsorbing gas, or less commonly by heat transfer from heatingcoils located within the bed. After desorption the bed temperatureis reduced with a cool purge gas, and the adsorption cycle startsagain. In gas dehydration TSA units, heating and cooling the bedcan take several hours (or even days). The adsorbents used fordehydration have a large capacity and high selectivity for watercompared to other natural gas components. However, if theselectivity of the adsorbent for the contaminant is not so strongthe bed will become saturated quickly. To treat a large volumetricgas ow with a long cycle TSA process will then require a bedwith a large total adsorption capacity, and thus a large amount ofadsorbent. In the PSA method, adsorption occurs at elevatedpressure (typically in the range 4002000 kPa) and desorptionoccurs at near-ambient pressure (Seader and Henley, 2006). Usingthe PSA method adsorbent beds can be depressurised and re-pressurised rapidly, allowing cycle times of several minutes oreven several seconds to be utilised. Accordingly, the amount ofadsorbent required for PSA processes can be much smaller thanfor an equivalent TSA processes.

    Examples of commercially available PSA-based systems forCO2 and N2 separation from natural gas are listed in Table 6. Mostadsorption-based CO2 capture technologies are limited toprocessing natural gas feeds containing no more than 2% CO2because the quantity of adsorbent required to capture greatervolumes of CO2 is large. Recent advances in small-footprint PSAsystems for CO2 removal from NG on offshore platforms in