Membrane-Based Oxygen-Enriched Combustion

Membrane-Based Oxygen-Enriched CombustionHaiqing Lin,*,† Meijuan Zhou,† Jennifer Ly,† Jimmy Vu,† Johannes G. Wijmans,† Timothy C. Merkel,†

Jianyong Jin,‡,⊥ Adam Haldeman,‡ Earl H. Wagener,‡ and David Rue§

†Membrane Technology and Research, Inc., 39630 Eureka Drive, Newark, California 94560, United States‡Tetramer Technologies, LLC, 657 S. Mechanic Street, Pendleton, South Carolina 29670, United States§Gas Technology Institute, 1700 S. Mount Prospect Road, Des Plaines, Illinois 60018, United States

ABSTRACT: The use of oxygen-enriched air, instead of ambient air, can significantly improve the energy efficiency ofcombustion processes and reduce the cost of CO2 capture from flue gases throughout manufacturing industries. This studyexamines the overall energy savings and economic benefits that can be obtained using oxygen-enriched combustion based onnovel membranes and processes to produce oxygen-enriched air. Membrane processes using low-pressure air as a countercurrentsweep in the permeate were used to minimize the energy cost of producing oxygen-enriched air. High-performance thin filmcomposite membranes based on a series of perfluoropolymers and bench-scale spiral-wound modules were prepared, and showedoxygen permeance as high as 1200 gpu (1 gpu = 10−6 cm3(STP)/cm2·s·cmHg) combined with O2/N2 selectivity of 3.0. Themembrane-based oxygen-enriched combustion processes show good energy savings (defined as the fuel savings less the energyconsumption of producing oxygen-enriched air) and economic benefits (defined as the value of fuel saved less the operating costof producing oxygen-enriched air), especially at flue gas temperatures higher than 1090 °C (or 2000 °F). For example, at a fluegas temperature of 1649 °C (or 3000 °F), membrane-based oxygen-enriched combustion shows a net energy savings of 35% anda net economic benefit of 29%, compared to the combustion process with air. The effect of oxygen-enriched air on NOxemissions in a natural gas furnace was also experimentally investigated.

1. INTRODUCTION

The benefits of using oxygen-enriched air (air containing morethan 21% oxygen), instead of ambient air, have long beenrecognized in many industrial processes such as catalystregeneration in refinery fluid catalytic cracking (FCC), partialoxidation of sulfur in Claus plants, wastewater treatment, andcombustion applications for glass and foundry operations.1−5

Oxygen-enriched air can lower the capital and operating costs,reduce the CO2 emissions, and increase the process flexibilityand reliability.3 The use of oxygen-enriched air would becomemore attractive for a broader variety and scale of industrialcombustion processes, if oxygen-enriched air can be producedin a lower cost and more energy efficient manner.2

Figure 1 illustrates the significant natural gas savings that canbe obtained when oxygen-enriched air is used instead ofambient air in natural gas-fired furnaces.1 The use of oxygen-enriched air reduces the volume of inert gas (N2); therefore,reducing the heat loss through the furnace exhaust andincreasing the energy efficiency.3,6 As shown in Figure 1, theuse of oxygen-enriched air can lower natural gas fuelrequirements by as much as 40% at a flue gas temperature of2500 °F (or 1371 °C), which is a typical operating temperaturefor the production of glass, cement, and steel.2 The actualenergy savings may be less than 40% due to waste heat recoveryin the furnaces. However, the use of waste heat recovery varieswidely, and it is often not used by small producers to reducedcapital cost. Therefore, to simplify the discussion, the techno-economic analysis in the paper assumes that there is no wasteheat recovery in the furnace.Figure 1 also shows that the increase in natural gas savings

becomes less pronounced at higher oxygen concentrations;

Received: May 8, 2013Revised: July 2, 2013Accepted: July 3, 2013Published: July 3, 2013

Figure 1. Natural gas savings achieved in furnace operations as afunction of the oxygen concentration in combustion air.1 The increasein savings becomes less pronounced at higher oxygen concentrations;consequently, the principal range of interest is 25−35% oxygen.Reprinted from ref 1 with permission. Copyright 1986 Elsevier.

Article

pubs.acs.org/IECR

© 2013 American Chemical Society 10820 dx.doi.org/10.1021/ie401464z | Ind. Eng. Chem. Res. 2013, 52, 10820−10834

pubs.acs.org/IECR

http://pubs.acs.org/action/showImage?doi=10.1021/ie401464z&iName=master.img-000.png&w=187&h=207

consequently, the principal range of interest is 25−35% oxygen.Moreover, if oxygen content is less than 27%, the flametemperature is compatible with most furnace materials.Therefore, existing furnaces can be easily retrofitted.2

Concerns about global climate change provide an additionalimportant incentive for using oxygen-enriched air incombustion.7 Combustion of hydrocarbon fuels generatescarbon dioxide, the leading contributor to climate change. Asshown in Figure 1, the use of oxygen-enriched combustionsaves energy and, therefore, reduces CO2 emissions. Thepercentage of reduction in CO2 emissions is about the same asthe percentage of energy savings.Substantial efforts are underway to develop technologies for

capture, utilization, and sequestration of the CO2 contained incombustion flue gases.7 The cost of capturing CO2 from fluegas decreases significantly if the volume of flue gas to be treateddecreases and the CO2 concentration in the flue gasincreases.8−10 As shown in Figure 2, this is exactly what is

accomplished by using oxygen-enriched air in combustionprocesses. As the oxygen concentration in the combustion airincreases, the amount of nitrogen in the flue gas decreases,resulting in a reduction in flue gas volume and a correspondingincrease in the CO2 concentration. The effect is significant evenat modest oxygen concentrations. For example, when thecombustion air contains 30% oxygen, the flue gas volume isreduced by 57% and the CO2 concentration in the flue gas isincreased by 47%.

2. BACKGROUND2.1. Theory. Permeation of gas A in nonporous polymers is

often characterized by permeability, PA (cm3(STP)·cm/cm2·s·cmHg), which represents the steady state pressure- andthickness-normalized gas flux through a polymer film and isdefined as follows:11

=−

PN l

A p p( )AA

m 2 1 (1)

where NA is the gas flux through the polymer (cm3(STP)/s), lis the film thickness (cm), Am is the active film area (cm2) for

gas permeation, and p2 and p1 are upstream (i.e., high) anddownstream (i.e., low) pressures (cmHg), respectively.Industrial membranes are often comprised of a thin selective

layer governing gas molecular separation. Because the thicknessof the selective layer may not be well-characterized, industrialmembranes are usually characterized by gas permeance, JA(cm3(STP)/cm2·s·cmHg), which is given as follows:12

=JPlAA

(2)

Permeance can be expressed in gpu units, where 1 gpu =10−6 cm3(STP)/cm2·s·cmHg. Industrial membranes with highpermeances are preferred, leading to less membrane area andlower capital cost for the membrane system to treat a given feedstream.13 On the basis of eq 2, membrane permeance can beincreased either by selecting a more permeable selective layermaterial (increasing PA) or by making the selective layer thinner(decreasing l).12

On the basis of the solution-diffusion mechanism governinggas transport in nonporous polymers, permeability of gas A canalso be written as follows:11

=P D SA A A (3)

where DA (cm2/s) is gas diffusivity and SA (cm3(STP)/cm3·cmHg) is gas solubility in the polymer.A measure of a membrane’s ability to separate two

components, A and B, is the membrane selectivity of gas Aover gas B, defined as the ratio of gas permeabilities orpermeances:11

α = =PP

DD

SSA/B

A

B

A

B

A

B (4)

Diffusivity selectivity, DA/DB, depends on the relativemolecular sizes. Oxygen with a kinetic diameter of 3.46 Å issmaller than nitrogen with a kinetic diameter of 3.64 Å, andtherefore, diffusivity selectivity always favors oxygen overnitrogen.14,15 Solubility selectivity, SA/SB, is determined bythe relative condensabilities of the permeants. Oxygen with acritical temperature of 154.6 K is more condensable thannitrogen with a critical temperature of 126.2 K,16 and therefore,solubility selectivity always favors oxygen over nitrogen.Consequently, polymeric membranes are always selective foroxygen over nitrogen to varying degrees.

2.2. Equivalent Pure Oxygen (EPO2). Equivalent pureoxygen (EPO2) is defined as the amount of pure oxygen thatneeds to be mixed with atmospheric air to obtain a specifiedvolume, V, of oxygen-enriched air with a specified oxygenconcentration, [O2] (in vol %):13,17

=−

VEPO[O ] 20.9

79.122

(5)

This metric provides a rational basis for comparingtechnologies that produce oxygen at different levels of purity.Later, we will explain production costs in terms of EPO2.

2.3. Technologies for Oxygen Enrichment from Air. Atpresent, most oxygen is produced by cryogenic separation of airand vacuum swing adsorption.4 However, the high cost ofoxygen from either of the established processes simply prohibitsthe use of oxygen-enriched air for many combustionapplications.2

Membrane processes for oxygen enrichment were developedto the early commercial stage in the 1980s using silicone rubber

Figure 2. Effect of oxygen levels in combustion air on flue gas volumeand CO2 concentration of the flue gas from natural gas-based furnaces,without using excess air. The effect is significant even at modestoxygen concentrations.

Industrial & Engineering Chemistry Research Article

dx.doi.org/10.1021/ie401464z | Ind. Eng. Chem. Res. 2013, 52, 10820−1083410821


and poly(phenylene oxide) membranes.13,17 Other types ofmembranes including polymeric materials,18,19 ceramic mem-branes,20 and facilitated transport membranes21,22 have alsobeen examined in the past two decades. Various process designsare possible; two typical process designs are shown in Figure 3aand b.1,13,17,23 Feed air containing 21% oxygen is passed acrossthe surface of a membrane that preferentially permeates oxygen.In Figure 3a, the pressure differential across the membranerequired to drive the process is provided by compressing thefeed gas. An alternative approach is to draw a vacuum on thepermeate, as shown in Figure 3b.In both cases, the oxygen concentration of the permeate gas

is higher than the desired oxygen concentration (which is 30%in this example), so atmospheric air is mixed with the permeateto achieve the target oxygen concentration. Vacuum operationis shown to be a better choice since it requires much lessenergy.13,17,23 However, to date, these types of membraneprocesses are not economically competitive with cryogenicdistillation and pressure swing adsorption at production rates ofindustrial interest, either because the power needed for the feedcompression was too high (high operating cost) or themembrane area required was too high due to the lack of highflux membranes and the low driving force for oxygenpermeation.12

This study evaluates the overall energy efficiency andeconomic benefits of oxygen-enriched combustion, whereoxygen-enriched air is produced using a new membraneprocess with high energy efficiency and new high-performancemembranes for O2/N2 separation. The thin film compositemembranes were fabricated into spiral-wound modules, whichwere tested at various operation conditions simulatingindustrial conditions. The effect of oxygen-enriched air onNOx emissions was experimentally investigated, and a techno-economic analysis of oxygen-enriched combustion based on thenewly developed membranes and processes is discussed.

3. NOVEL MEMBRANE PROCESS WITHCOUNTERCURRENT/SWEEP OPERATION

Figure 4 shows the new membrane process for oxygenenrichment from air, which has two design features differentfrom the processes shown in Figure 3:

1. a turboexpander that improves energy efficiency and2. a countercurrent/sweep membrane design that improves

separation efficiency.

The feed air to the membrane unit is compressed to increasethe driving force for permeation. A significant amount of thiscompression energy can be recovered by sending thepressurized residue stream to a turboexpander as it leaves themembrane unit. On the permeate side of the membrane,

atmospheric air is used as a sweep stream and mixes with thepermeated gas at the membrane interface, not outside themembrane modules as shown in Figure 3. The sweep designinherently improves energy efficiency because the atmosphericair dilutes the oxygen-enriched permeate flow and increases thedriving force for oxygen permeation without requiring morecompression energy.8,24−26 A simulation of this design wasperformed using a commercial process simulation package,ChemCAD 6.3 (ChemStations, Inc., Houston, TX), enhancedwith proprietary MTR code for membrane unit operations.The membrane process shown in Figure 4 produces 7.1

m3(STP)/s oxygen-enriched air (containing 30% O2) or 100tons of EPO2 per day (see eq 5). Atmospheric air is compressedto 60 psia (4.1 bar), using a compressor with an efficiency of0.85, and passes through a series of countercurrent/sweepmodules containing 4240 m2 of membrane area. About 34% ofthe compressed air permeates the membrane and combineswith the sweep air to produce oxygen-enriched air. Themembrane residue stream passes through a turboexpanderwhere some of the compression energy is recovered. Assumingan 85% conversion efficiency, the turboexpander will generate33% of the power required to compress the feed gas. Theoperating conditions, such as feed pressure and ratio of sweepair to permeate flow rate, are optimized for a membrane with anO2/N2 selectivity of 3.0 and an O2 permeance of 1200 gpu. Asshown later, these membrane properties were achieved in themembrane development activities.Table 1 compares the performance of the countercurrent/

sweep design shown in Figure 4 with the conventional feedcompression of Figure 3a or permeate vacuum design of Figure3b. The process simulation was performed using ChemCAD

Figure 3.Membrane processes to produce 30% oxygen-enriched air with (a) feed compression and (b) permeate vacuum. Both designs operate withthe membrane modules in the cross-flow mode and cannot produce oxygen-enriched air at a cost competitive with conventional technologies.

Figure 4. Energy-efficient membrane process to produce oxygen-enriched air using countercurrent/sweep membrane operation. Theprocess produces 7.1 m3(STP)/s of 30% oxygen-enriched air, whichequals 262 ton O2/day or 100 tons/day EPO2 (defined in eq 5). Themembrane has an oxygen permeance of 1200 gpu and an O2/N2selectivity of 3.0.





6.3, and the flow rates were varied in the three processes tomeet the requirement in oxygen concentration and overall flowrate in the oxygen-enriched air.Countercurrent/sweep operation provides the best combi-

nation of power consumption and membrane area. Comparedto the feed compression process, the countercurrent/sweepprocess consumes 29% less energy and uses 33% lessmembrane area. Compared to the permeate vacuum operation,countercurrent/sweep operation consumes 18% more energybut uses 78% less membrane area. The permeate vacuumprocess uses 44% less energy and requires about 220% moremembrane area than the feed compression process, which isconsistent with results from earlier studies.13,17

4. EXPERIMENTAL SECTION4.1. Membrane Preparation. Figure 5 shows a schematic

drawing of the flat sheet thin film composite membranes

prepared in this work.12 The nonwoven polyester paper layerprovides mechanical strength to the membrane compositestructure. On top of the nonwoven paper, which has a relativelyrough surface, a microporous ultrafiltration support layer isformed by wet-phase separation casting of a poly(etherimide)(Ultem 1000, General Electric, Mount Vernon, IN).27 Themicroporous support has surface porosity of less than 5%, andfine pores with an average pore diameter of less than 100 nm.The proprietary gutter layer and selective layer were formed bya dip-coating process using an industrial-scale coater.12 The

smooth surface of the gutter layer allows a very thin denseselective layer of perfluoropolymers to be deposited. In general,the microporous support and gutter layer contribute essentiallyno resistance to gas flow, and therefore, the resistance to gastransport in the thin film composite membranes lies mainly inthe selective layer. In this manner, the gas selectivity isdetermined primarily by the selective layer. However, for highflux membranes with ultrathin perfluoropolymer selectivelayers, the transport resistance in the gutter layer may not benegligible compared to that in the selective layer, and therefore,the composite membranes could show selectivity slightly lowerthan that of the perfluoropolymer selective layer alone.

4.2. Module Preparation. The flat sheet thin filmcomposite membranes were fabricated into spiral-woundmodules that can operate in countercurrent/sweep mode. Asshown in Figure 6, these modules have four ports: feed, residue,

sweep, and the combined sweep and permeate (sweep +permeate).26,28−31 Feed gas passes down the module parallel tothe permeate pipe in the channel created by the feed spacer. Aportion of the feed gas permeates the membrane and thenenters the permeate channel. The permeate channel can beswept with a sweep gas, allowing the module to operate in apartial countercurrent mode.26,28 Although the module doesnot have a perfect countercurrent flow pattern between the feedand permeate flow, it has been shown that the module can stillhave very high sweep efficiency, despite the presence of anonideal countercurrent flow pattern.26,31,32

Table 1. Results of Simulation Comparing Various Membrane System Designs to Produce Oxygen-Enriched Air Containing30% Oxygena

aThe membranes have oxygen permeance of 1200 gpu and O2/N2 selectivity of 3.0.bF = feed; R = residue; S = sweep; S + P = sweep + permeate. cA

compressor and turboexpander efficiency of 85% and a vacuum pump efficiency of 75% were assumed in the calculations.

Figure 5. Schematic illustration of the industrial thin film compositemembranes prepared in this work.12 The selective layer is aperfluoropolymer with high O2/N2 separation performance.

Figure 6. Exploded view of the membrane envelope for acountercurrent/sweep spiral-wound module, which uses a sweep gason the permeate side and operates in a partial countercurrentpattern.26,28−31 The membrane envelope has a length of 90−100 cm(in the direction parallel to the product pipe) and a width of 30−60cm (in the direction perpendicular to the product pipe).




http://pubs.acs.org/action/showImage?doi=10.1021/ie401464z&iName=master.img-005.jpg&w=239&h=111


4.3. Determination of Pure-Gas Permeances. Pure-gaspermeances in thin film composite membranes weredetermined using a constant-pressure/variable-volume appara-tus.33 Before each measurement, the entire apparatus is purgedusing the feed gas to remove any impurities from the system.The permeate gas flow rate is measured using bubble flowmeters (Alltech Associates, Inc. Deerfield, IL, USA). The steadystate permeance for gas A, can be calculated by33

=−

⎜ ⎟⎛⎝

⎞⎠

Pl A p p T

VT

1( )

273 dd

A

m 2 1 (6)

where the downstream pressure, p1, is atmospheric pressure inthis case, T is the absolute temperature of the gas (K), and dV/dt is the steady state volumetric displacement rate of the soapfilm (cm3/s). Permeance is expressed in gpu units, where 1 gpu= 10−6 cm3(STP)/cm2·s·cmHg.4.4. Determination of Mixed-Gas Permeances. The

composite membranes and membrane modules were testedunder countercurrent/sweep conditions using a system asshown in Figure 7. Compressed air flows on the feed side of the

membrane, and low pressure air flows on the permeate side.The test system is equipped with an Agilent MicroGC 3000portable gas chromatography (GC), to monitor the gascomposition of all streams (including the feed, residue,sweep, and “sweep + permeate” stream) as oxygen andnitrogen permeate across the membrane. The flow rates weremeasured using mass flow meters (Sierra Instruments,Monterey, CA). The experimental data were only consideredvalid when the mass of each component entering and leavingthe permeation cell or module housing was within ±10%.The mixed-gas permeance in countercurrent/sweep oper-

ation can be estimated using the following equation:26,34

=Δ Δ

Δ ΔPl

N p p

A p p

ln( / )

( / )A A A,L A,0

m A,L A,0 (7)

Δ = − Δ = −+p p p p p p, andA,0 A,F A,S P A,L A,R A,S (8)

where pA,F, pA,R, pA,S, and pA,S+P are the partial pressures ofcomponent A in the feed, residue, sweep and sweep + permeatestream, respectively. NA is the flow rate difference ofcomponent A between the sweep and sweep + permeatestream. Equation 7 is similar to eq 1, except that the partialpressure difference is expressed as the logarithmic mean ofΔpA,L and ΔpA,0. The use of logarithmic mean pressure

difference in eq 7 is analogous to that derived for acountercurrent heat exchanger, in which the logarithmicmean temperature difference is used.26,35

The composite membrane sheets were mounted in amodified Osmonics SEPA CF II Med/High Foulant Systemcell (GE Osmonics Labstore, Minnetonka, MN). The test cellwas modified to allow a near perfect countercurrent flow test.36

The countercurrent/sweep spiral-wound modules were in-stalled in a custom-built module housing, with separate ports toaccommodate the four gas streams.Chemical-grade nitrogen, oxygen, and air, each with a purity

of 99%, were used as received from Praxair Inc. (Hayward, CA).An air compressor was also used to provide high-pressure air,the composition of which was determined using the GC.

4.5. Evaluation of Oxygen-Enriched Combustion. Theeffect of oxygen-enriched combustion on NOx emissions wasevaluated experimentally at Gas Technology Institute (GTI,Des Plaines, IL). A Bloom Engineering model 1476 hot airbaffle burner was selected for testing, because it is a commonindustrial burner and shows strong dependence of NOxemissions on oxygen levels in the combustion air. Figure 8

presents the test setup, with the baffle burner mounted on theresearch modular furnace. The modular furnace is equippedwith thermocouples, side view ports, and sampling ports indifferent locations. Exhaust gas is sampled in the test tomonitor the levels of NOx, CO, CO2, and O2 at different testconditions using an HORIBA analyzer (HORIBA InstrumentInc., Illinois, US).

5. RESULTS AND DISCUSSIONS5.1. Pure-Gas Permeance and Stability in Membranes.

We have identified a series of perfluoropolymers (PFPs) withhigh oxygen permeability and good oxygen/nitrogen selectivityfor oxygen enrichment. Thin film composite membranes of thetype shown in Figure 5 were fabricated using theseperfluoropolymers as the selective layer.Figure 9 shows the effect of perfluoropolymer concentration

in the coating solution on pure-gas oxygen permeance and O2/N2 selectivity. As expected, decreasing perfluoropolymerconcentration in the coating solution reduces the selectivelayer thickness, increasing oxygen permeance.37 The oxygenpermeance is inversely proportional to the polymer concen-tration of the coating solutions. On the basis of theperfluoropolymer O2 permeability of 50 Barrers, the selectivelayer prepared from the solution with 0.15 wt % polymer has anstimated thickness of 50 nm, using eq 2.

Figure 7. Schematic of countercurrent/sweep test system for flat sheetcomposite membranes and spiral-wound modules. A rectangularcountercurrent test cell (with a membrane area of 150 cm2) was usedfor membrane testing, and one 4-in.-diameter module housing wasused to test countercurrent/sweep spiral-wound modules (with amembrane area of 0.5−3 m2): (MFM) mass flow meter; (GC) gaschromatography.

Figure 8. Photos of a (a) baffle burner and (b) modular furnace at GasTechnology Institute (GTI) that were used for oxygen-enriched aircombustion tests.





Figure 9 also shows that the pure-gas O2/N2 selectivitydecreases slightly at very high oxygen permeances, presumablycaused by the gas transport resistance of the gutter layer. Forlow flux membranes with O2 permeance less than 600 gpu, thetransport resistance in the selective layer is much higher thanthat in the gutter layer (which has a measured O2 permeance of11 000 gpu). Therefore, resistance to gas transport through themembranes is mainly in the selective layer, and O2/N2selectivity of the composite membrane is similar to that ofthe selective layer.38 However, as the selective layer becomesthinner, the transport resistance of the selective layer decreasesand becomes more comparable with the resistance of the gutterlayer, particularly for the more permeable gas, O2. The resultingselectivity of the composite membranes thus demonstrates avalue between that in the selective layer and gutter layer.Because the O2/N2 selectivity in the gutter layer is lower thanthat in the selective layer (about 3.0), increasing oxygenpermeance in the composite membranes tends to slightly loweroxygen/nitrogen selectivity. A more quantitative analysis is

provided below. The membrane with highest O2 permeance(980 gpu) exhibits O2/N2 selectivity of 2.7, as shown in Figure9. On the basis of the resistance layer model and O2 permeanceof the gutter layer of 11 000 gpu, the selective layer is estimatedto have an O2 permeance of 1080 gpu, which gives an O2/N2selectivity of 3.0 for the selective layer, consistent with theintrinsic property of the perfluoropolymer. This calculationassumes a minimal effect of the gutter layer on the permeanceof the less permeable gas (N2), which is valid here. Based onthis understanding, slight adjustments were made to themembrane configuration to allow better performance, asshown later.Thin films of glassy polymers often show aging behavior; i.e.,

gas permeability declines significantly with time, and thedecrease becomes more substantial as the polymer filmthickness decreases.39−42 For example, oxygen permeability inpoly(2,6-dimethyl-1,4-phenylene oxide) (PPO) with a thick-ness of 400 nm decreased from 20 to 10 Barrers in 40 days.43

The aging behavior in these polymer thin films has beenrationalized by the collapse of free volume with time, leading toa decrease in polymer free volume.39 Higher flux polymers withhigher free volume may show more dramatic aging behavior, asexemplified in poly[1-(trimethylsilyl)-1-propyne] (PTMSP)44

and polymers of intrinsic microporosity (PIM-1).45−47

Significant aging behavior creates a great challenge in systemdesign and increases the capital cost of the membrane system,limiting the commercial usefulness of many polymers thatdemonstrate initially promising (or even spectacular) labo-ratory performance. The perfluoropolymers in this work havehigh free volume, and composite membranes with aperfluoropolymer selective layer as thin as 50 nm have beenprepared. Investigating the aging behavior in these compositemembranes was a necessary next step.Figure 10 shows the stability of various composite

membranes based on perfluoropolymers. In these tests, themembrane sheets were stored in air at 20 °C, and differentmembrane stamps were taken and tested at various times, at afeed pressure of 50 psig.Unlike other high flux glassy polymers (such as a PPO film

with a thickness of 400 nm for which oxygen permeabilitydecreased by 50% in 40 days43), perfluoropolymer-basedcomposite membranes showed stable O2 permeance and O2/

Figure 9. Effect of polymer content in the coating solutions on pure-gas O2 permeance and O2/N2 selectivity in perfluoropolymer-basedcomposite membranes at 4.4 bar and 20 °C. One gpu = 10−6

cm3(STP)/cm2·s·cmHg. The lines are to guide the eye.

Figure 10. Time dependence of (a) pure-gas O2 permeance and (b) pure-gas O2/N2 selectivity in perfluoropolymer-based composite membranes at4.4 bar and 20 °C. The thickness of the selective layer in these composite membranes was estimated using eq 2 with a measured oxygen permeabilityvalue of 50 Barrers. The lines are to guide the eye.





N2 selectivity for 60 days, even for the membranes withselective layers thinner than 100 nm.Optimal membranes for O2/N2 separation should have high

O2 permeance to reduce the required membrane area andcapital cost, and high O2/N2 selectivity to increase the productO2 purity. Figure 11 presents a permeance/selectivity map for

O2/N2 separation in polymeric membranes at 25 °C, where gastransport follows the solution-diffusion mechanism.14,48,49 Thistype of plot was popularized by Robeson.48,49 Each pointrepresents the selected separation properties for one particularpolymer. The upper bound line in the figure gives a roughestimate of the highest selectivity possible for a givenpermeability in polymer-based materials.14 There is a trade-offbetween gas selectivity and permeability; that is, materials withhigher oxygen permeance have lower oxygen/nitrogenselectivity.Figure 11 assumes that all polymers can be made into stable

composite membranes with a selective layer thickness of 1 μm,without further discussions on the feasibility of membranefabrication and physical aging behavior. The perfluoropolymer-based composite membranes prepared in this study show O2/N2 separation performance (high and stable oxygen permeanceof 1200 gpu and O2/N2 selectivity of 3.0) close to the upperbound. Compared with current commercial membranes (madefrom polysulfone and polyimides)12 for nitrogen generationfrom air, these perfluoropolymer-based membranes showorders of magnitude higher oxygen permeance and lower O2/N2 selectivity.Figure 11 also compares the perfluoropolymer-based

membranes with prior membranes developed for oxygenenrichment from air, such as silicone rubber (SR) andpoly(phenylene oxide) (PPO) membranes. The perfluoropol-ymer-based membranes show a good combination of oxygenpermeance and O2/N2 selectivity, compared to these polymers.5.2. Mixed-Gas Permeances in Membranes. Figure 12

shows results for a perfluoropolymer-based membrane stamptested in the countercurrent/sweep mode with air as feed andsweep gas at 20 °C. The feed gas had a pressure of 2.0 bar andflow rate of 50 cm3(STP)/s. The sweep gas was at ambient

pressure. The stage-cuts (defined as the percentage of the feedgas entering the permeate) for all the tests were below 10%.As shown in Figure 12, the use of a sweep gas significantly

increased oxygen flux, especially at low values of sweep/feedflow rate. For example, as sweep/feed flow rate increased from0% to 26%, oxygen flux increased by 53%, from 1.7 to 2.6cm3(STP)/s. The incremental increase in oxygen flux withincreasing sweep flow rate leveled off at higher sweep flow rates.For instance, as the sweep/feed flow rate increased from 26%to 54%, the O2 flux increased by only 7.8%, to 2.8 cm3(STP)/s.Figure 13a and b compares mixed-gas oxygen permeance and

oxygen/nitrogen selectivity at different feed pressures andsweep flow rates. The feed flow rate was 8.7 and 26 cm3(STP)/s for the tests with feed pressure of 2.0 and 4.4 bar, respectively.The sweep gas was at ambient pressure for all the tests. Theexperimental uncertainty is often within 10% in thesemeasurements. As shown in Figure 13, the mixed-gas O2permeance and O2/N2 selectivity were independent of feedpressures and sweep/feed flow ratio.

5.3. Module Fabrication and Characterization. Thenext step in developing a commercially practical membranetechnology is to confirm that a larger-scale module package canbe prepared and that the separation performance of this moduleis nearly comparable to that obtained for the membrane inlaboratory testing. As a preliminary step in this direction, twobench-scale spiral-wound modules with countercurrent/sweepdesign (S and L) were fabricated using membranes describedpreviously, produced on a commercial coater. The smallermodule (module S) has a membrane area of 0.50 m2, and thelarger module (module L) is a full-size semicommercial 4-in.-diameter module containing a membrane area of 1.8 m2. Pure-gas separation properties of these two modules are shown inTable 2. The pure-gas permeances and selectivities of thesemodules were similar to the properties of the membranestamps, indicating that the modules were defect-free.Figure 14 shows the effect of sweep flow rate and permeate

pressure on the module separation performance. All tests wereperformed with air as both feed and sweep at 20 °C. For thetests of module S, the feed flow rate was 2000 cm3(STP)/s atsweep pressure of 1.0 bar and 1000 cm3(STP)/s at sweeppressure of 1.7 bar. For the tests of module L, the feed flow ratewas 1700 cm3(STP)/s at permeate pressure of 1.0 bar, and1400 cm3(STP)/s at permeate pressure of 1.3 bar. The stage-

Figure 11. Comparison of a perfluoropolymer (PFP) based compositemembrane to other membranes in the literature (with assumed 1 μm-thick selective layer) for O2/N2 separation. Current commercialmembranes for O2/N2 separation have oxygen permeance of 10−100gpu and oxygen/nitrogen selectivity of 4−6. PPO: poly(phenyleneoxide); SR: silicone rubber.

Figure 12. Oxygen flux as a function of sweep/feed flow rate for aperfluoropolymer-based membrane stamp (150 cm2) in the counter-current/sweep mode at 20 °C. Air was used as the feed gas and sweepgas. The feed gas pressure was 2.0 bar and flow rate was 50 cm3(STP)/s. The sweep gas was at atmospheric pressure.





cuts for all the tests were below 25%. Since the membraneperformance is independent of feed pressure (as shown inFigure 13), low feed pressures (1.7 and 2.0 bar) were used inthe module testing.As shown in Figure 14, the use of a sweep gas increased

oxygen flux in both modules, especially at high permeatepressure. For example, as sweep/feed flow rate increased from0% to 46% at a sweep pressure of 1.7 bar, oxygen flux inmodule S increased by 54%, from 14 cm3(STP)/s to 22cm3(STP)/s. However, the increase in module L was not as

significant as that in module S. For example, as sweep/feed flowrate increased from 0% to 50% at a permeate pressure of 1.3bar, oxygen flux increased by only 18%, from 62 to 78cm3(STP)/s. More work on module scale-up will be needed, toachieve better and more consistent sweep efficiency in futuremodules.Figure 15a and b compares mixed-gas oxygen permeance and

O2/N2 selectivity at different permeate pressures and sweepflow rates for module L. The oxygen permeance and O2/N2

selectivity were independent of permeate pressures and weresimilar to those of the membrane module (1300 gpu for oxygenpermeance and 3.0 for oxygen/nitrogen selectivity, as shown inTable 2).Figure 16 compares the countercurrent/sweep efficiency

(defined as the increase in oxygen permeate flow rate caused bythe sweep) obtained with a membrane sheet and module S.The test results for the membrane sheet as provided in Figure12 were obtained at near “perfect” countercurrent conditions.In contrast, the module operates in only a partial counter-current mode, as shown in Figure 6. Nevertheless, Figure 16

Figure 13. (a) Mixed-gas O2 permeance and (b) mixed-gas O2/N2 selectivity at 20 °C and different feed pressures and sweep flow rates in amembrane sheet. The feed flow rate was 8.7 and 26 cm3(STP)/s for the tests with feed pressure of 2.0 and 4.4 bar, respectively. The sweep gas was atatmospheric pressure. The lines are to guide the eye.

Table 2. Pure-Gas Separation Performance in Two Bench-Scale Countercurrent/Sweep Perfluoropolymer-BasedMembrane Modules (S and L) at 1.7 bar and 20 °C

permeance(gpu)

bench-scalemodule

membrane area(m2) N2 O2

O2/N2selectivity

S 0.50 480 1200 2.5L 3.0 430 1300 3.0

Figure 14. Relative oxygen flux for (a) module S and (b) module L as a function of sweep/feed flow rate in the countercurrent/sweep mode at 20°C. Air was used as the feed and sweep gas. The feed gas pressure was 2.0 bar for module S and 1.7 bar for module L.





shows that the efficiency of countercurrent/sweep in themodule is very similar to that in the membrane sheet. However,as shown in Figure 14, the larger module (module L) does notshow sweep efficiency as good as that of the smaller module(module S), indicating that further optimization of the moduleconfiguration to improve sweep efficiency will be needed.5.4. Effect of Oxygen-Enriched Air on NOx Emissions.

Combustion with oxygen-enriched air often leads to higherflame temperature, resulting in higher NOx content in the fluegas.2 Higher NOx content in flue gases is undesirable becauseemissions of this pollutant are subject to various environmentalregulations. Therefore, the benefits of oxygen-enriched aircombustion in terms of process efficiency must be weighedagainst potential negative changes to the flue gas emissionsprofile.In this study, the effect of oxygen-enriched air combustion on

NOx emissions was evaluated experimentally using a naturalgas-fired burner at Gas Technology Institute. Firing rate wasfound to affect NOx levels, so all comparative tests wereconducted at a constant natural gas firing rate of 500 scf/h(standard cubic foot per hour). Table 3 shows the effect ofexcess air on NOx level in the burner without modifications.

The NOx level increased with increasing excess air and oxygenenrichment level in the air, as expected.2

NOx levels can be effectively reduced by oxygen lancing,where a portion of the oxidant (air or enriched air) is injectedthrough a lance above the main flame during combustion.50

Figure 17 compares the effect of lance percentage on NOxconcentration in the flue gas for combustion air containing 21%O2, 25% O2, and 29% O2. When firing with 25% enriched air,the NOx decreased from 115 ppm with no lancing to 55 ppm

Figure 15. (a) Mixed-gas oxygen permeance and (b) mixed-gas O2/N2 selectivity at different sweep pressures and flow rates in countercurrent/sweep module L at 20 °C. Air was used as the feed gas and sweep gas. The feed gas pressure was 1.7 bar. The lines are to guide the eye.

Figure 16. Comparison of countercurrent/sweep efficiency (indicatedby the relative oxygen permeate flux) in a membrane sheet and inmodule S at 20 °C. Test condition details are provided in the captionsfor Figures 12 and 14. The lines are to guide the eye.

Table 3. Effect of Oxygen Levels in the Combustion Air onNOx Emissions for an Unmodified Bloom Baffle Burner witha Natural Gas Firing Rate of 500 scf/h

NOx emissions (ppm) at different O2 levels in thecombustion air

excess air (%) 21% 25% 29%

0 64 94 2075 79 142 27410 80 156 304

Figure 17. Effect of lance percentage on NOx concentration in the fluegas using combustion air containing 21% O2, 25% O2, and 29% O2.The furnace was operated at a firing rate of 500 scf/h natural gas with5% excess combustion air.






with 10% lancing of the enriched air above the flame. Atroughly 6% of the enriched air sent to the lance, the burneroperating with 25% oxygen-enriched air generates the sameNOx as the burner operating with normal air and no lancing.These results confirm that lancing is an effective method toreduce NOx emissions and to avoid excess NOx productionwhen oxygen-enriched air is used for combustion. The amountof lancing needed to meet this goal increases as the oxygenenrichment level is increased. It should be noted that the effectof excess air level and lancing on the NOx level also dependssensitively on the type of industrial burners.

6. ENERGY ANALYSIS OF MEMBRANE-BASEDOXYGEN-ENRICHED COMBUSTION6.1. Energy Consumption of Membrane-Based Oxy-

gen Enrichment. The membrane process shown in Figure 4produces 30% oxygen-enriched air at 100 tons of EPO2 per dayusing perfluoropolymer-based composite membranes withoxygen permeance of 1200 gpu and oxygen/nitrogen selectivityof 3.0. One performance metric often used to compare oxygenproduction technologies is the energy required to generate agiven volume of oxygen. Table 4 compares this energy

requirement for the membrane process discussed in this articleand conventional technologiescryogenic distillation, vacuumswing adsorption (VSA), and pressure swing adsorption (PSA).The membrane process in Figure 4 yields oxygen productionenergies considerably lower than those of VSA or PSA, andhigher than those of cryogenic fractionation.6.2. Energy Savings of Membrane-Based Oxygen-

Enriched Combustion. Table 5 shows sample calculation ofthe net energy savings in oxygen-enriched combustion, which isdetermined by calculating the gross savings from reduced fueluse in a furnace and subtracting the cost of generating oxygen-enriched air using membrane systems. Two cases with a firingrate of 1 MMBtu are compared. The base case uses air forcombustion with 1000 scf natural gas (or 1 MMBtu) at anexhaust gas temperature of 1649 °C. The second case usesoxygen-enriched air containing 30% oxygen for combustionand, as shown in Figure 1, saves 45% in fuel to achieve the sameenergy output as the base case (equivalent 1 MMBtu). To keepthis initial energy balance simple, the electricity needed togenerate the oxygen-enriched air is assumed to come from a30%-efficient on-site, gas-fired engine, and not from the grid.

As shown in Table 5, with 30% oxygen in the combustion air,the natural gas burner can save 35% fuel at an exhaust gastemperature of 1649 °C. Such positive savings in energy makemembrane-based oxygen-enriched combustion a very attractiveroute to improving the energy efficiency in combustionprocesses.For membrane-based oxygen-combustion processes to be

economically attractive, the net energy savings need tooutweigh the production cost of oxygen-enriched air. Asshown in Table 5, saving 350 scf of natural gas has a value of$1.4, assuming a value of $4 per 1000 scf natural gas. Therefore,the production of oxygen-enriched air needs to be less than$45/ton EPO2 to demonstrate the economic benefits of usingoxygen-enriched air.Figure 18 shows the effect of oxygen-enriched combustion

on net energy savings at various exhaust temperatures. Twomembranes with oxygen permeance of 1200 gpu are consideredin these process simulations. The first membrane developedhere has an O2/N2 selectivity of 3.0, which results in a powerconsumption of 1.7 MMBtu/ton EPO2. The second membranehas an O2/N2 selectivity of 4.0, resulting in a powerconsumption of 1.4 MMBtu/ton EPO2. The second membraneis hypothetical and is used here to illustrate the benefits ofhigher selectivity membranes on energy savings.Figure 18 also shows that the net energy savings increase

with both increasing oxygen levels in the oxygen-enriched airand exhaust temperatures. The calculated net energy savings at816 °C (or 1500 °F) using both membranes are negative;membrane processes with this level of energy savings are notpractically attractive. The membrane/oxygen-enriched combus-tion process appears much more favorable in applications withexhaust temperatures higher than 1093 °C (or 2000 °F). Forexample, at an exhaust temperature of 1649 °C, the membrane-based enriched oxygen combustion shows a net energy savingsof 35% with membrane oxygen/nitrogen selectivity of 3.0 (asalso shown in Table 5). Membranes with O2/N2 selectivity of4.0 show slightly better net energy savings than those with O2/N2 selectivity of 3.0.

Table 4. Comparison of Energy Required to Produce EPO2Using Various Technologies for Use in Oxygen-EnrichedCombustion

technologyproduction energy

(MMBtu/ton EPO2)production volume(ton EPO2/day)

O2purity

cryogenic51 0.84−1.36 >50 90−99%

VSA51 2.08 20−90 90−93%

PSA51 2.60 <20 90−95%

PFP-membrane

1.4−1.7a up to 100 25−35%

aProduction energy for the membrane process depends primarily onmembrane O2/N2 selectivity, which is 1.7 MMBtu for currentmembranes with a selectivity of 3.0 and 1.4 MMBtu for a hypotheticalmembrane with a selectivity of 4.0. 1 MMBtu = 293 kWh.

Table 5. Calculation of Energy Savings for Membrane-BasedOxygen-Enriched Combustion. With 30% Oxygen in theCombustion Air, the Burner Can Save 35% Energy at a FlueGas Temperature of 1649 °C

two cases with a firing rate of 1 MMBtua values unit

base case: air (20.9% O2)

required natural gas 1000 scfrequired air 10048 scfexhaust gas temperature 3000 °Fcase with membrane-based oxygen-enriched air (30% O2)

required natural gas (45% fuel saving at 1649 °C) 550 scfrequired oxygen-enriched air (30% O2) 3850 scfrequired EPO2 0.018 tonpower consumption for 30% O2 (at 1.7 MMBtu/tonEPO2)

0.030 MMBtu

natural gas required to generate power (30% efficiency) 100 scfoverall natural gas consumption in using 30% O2 forcombustion

650 scf

overall natural gas saving in oxygen-enrichedcombustion

35%

aBoth combustion processes use 5% excess oxygen. These calculationsassume no heat recovery from flue gas to simplify the comparison.



7. ECONOMIC ANALYSIS OF MEMBRANE-BASEDOXYGEN-ENRICHED COMBUSTION7.1. Economics of Membrane-Based Oxygen-Enriched

Air Production. Table 6 shows the cost calculation for

oxygen-enriched air production using the countercurrent/sweep process, compared with feed compression and permeatevacuum processes. The operating conditions (such as feedpressure and flow rate) were shown in Figure 4 and Table 1.These membrane processes produce 30% oxygen-enriched airat 100 tons of EPO2 per day, based on a membrane with anoxygen permeance of 1200 gpu and O2/N2 selectivity of 3.0.The assumptions for the costs of the membrane portion of thedesign and compressors are also shown in Table 6. Althoughthe economic analysis performed here is very primitive, it canshed some light on the effect of membrane separation

properties (oxygen permeance and O2/N2 selectivity) on theproduction cost of oxygen-enriched air.For the feed compression process (Figure 3a), both the

capital cost and operating cost are dominated by thecompressor, which gives a production cost of $50/ton EPO2.The electricity consumption of the feed compression process isthe highest among the three processes. For the permeatevacuum process (Figure 3b), the capital cost is dominated bythe membrane module portion due to the enormous membranearea required, and the operating cost is more evenly distributedbetween the membrane portion and compressor.The countercurrent/sweep process has the lowest capital and

operating costs for oxygen-enriched air production, primarilybecause the membrane area required is the lowest and aturboexpander is used in the residue stream for energyrecovery. Capital costs for this process are highest for thecompressor (3150 kW) and turboexpander (1040 kW).However, the net power requirement is only 2110 kW,resulting in lower power consumption than required for thefeed compression process.Figure 19 shows the required membrane separation proper-

ties (oxygen permeance and O2/N2 selectivity) to produce 30%oxygen-enriched air at $45/ton EPO2, using the three differentprocesses shown in Figures 3 and 4.23 The process simulationswere performed using ChemCAD 6.3, and the costassumptions were presented in Table 6. As shown in Table 5,the production of oxygen-enriched air needs to be less than$45/ton EPO2 to demonstrate the economic benefits of usingoxygen-enriched air for the combustion processes with a fluegas temperature of 3000 °F. In general, increasing O2/N2

selectivity decreases the required oxygen permeance to meetthe production cost; the countercurrent/sweep process showsthe least stringent requirement on the membrane separationproperties among the three processes considered. Interestingly,in the feed compression process, increasing the O2/N2

selectivity above 6 slightly increases the required oxygenpermeance needed to meet the production cost of $45/tonEPO2; presumably, this occurs because the separation is limitedby the process feed-to-permeate pressure ratio (about 4), ratherthan the membrane selectivity (6 or above).12,52

Figure 18. Effect of oxygen-enriched combustion (with 25% and 30% oxygen) on net energy savings at various flue gas temperatures. (a) Membraneprocesses with an O2/N2 selectivity of 3.0 and power consumption of 1.7 MMBtu/ton EPO2. (b) Membrane processes with an O2/N2 selectivity of4.0 and power consumption of 1.4 MMBtu/ton EPO2.

Table 6. Capital and Operating Costs for the MembranePortion of the 100-ton/day EPO2 Plant Illustrated in Figure4

cost

feedcompression[Figure 3a]

permeatevacuum [Figure

3b]countercurrent/sweep [Figure 4]

capital cost ($1,000) ($1,000) ($1,000)membrane moduleskids at $100/m2

600 1,900 424

frame/pipes/valves at$100/m2

600 1,900 424

compressor/vacuumpump at $200/kW

640 360 838

total capital cost 1,840 4,160 1,686operating cost ($1,000) ($1,000) ($1,000)depreciation at 15%of capital

276 624 252

module replacement(every 4 y)

150 475 106

electric power at$0.05/kWh

1,401 788 924

total annual operatingcost

1,827 1,887 1,283

production cost/tonEPO2

$50 $52 $35




Figure 20 shows the required membrane performance toproduce 30% oxygen-enriched air at a cost of $35, $45, and $55

per ton EPO2, respectively, using the countercurrent/sweepprocess. This information is compared with the properties ofcommercial membranes for nitrogen generation from air suchas polysulfone (PSf), polyimides (PI), and poly(phenyleneoxide) (PPO). These commercial membranes have beendeveloped to produce high-purity nitrogen and as such theyhave high O2/N2 selectivity.12 However, they produce 30%oxygen-enriched air at a cost higher than $55/ton EPO2,primarily because of low oxygen permeances, and the resultantlarge membrane area required. The perfluoropolymer-basedcomposite membranes developed in this work would be able toproduce 30% oxygen-enriched air at much lower cost than thecommercial membranes for nitrogen generation, even thoughthe perfluoropolymers have lower O2/N2 selectivity.

7.2. Economics of Membrane-Based Oxygen-EnrichedCombustion. Figure 21 shows the effect of oxygen-enriched

combustion on net economic benefits at various flue gastemperatures, for membrane-based oxygen-enriched combus-tion (30% O2, 5% excess) versus combustion with regular air(5% excess). The net economic benefit is the value of thenatural gas savings obtained using oxygen-enriched combustionless the production cost of oxygen-enriched air. The price ofnatural gas is assumed to be $4/1000 scf. A membrane withoxygen permeance of 1200 gpu and O2/N2 selectivity of 3.0(with $35/ton EPO2) is used for the calculations.As shown in Figure 21, the net economic savings increase

with increasing flue gas temperature. At low exhaust temper-atures (such as below 1000 °C), the net economic savings arenegative, which makes the membrane processes of no practicaluse. The membrane oxygen-enriched combustion processappears much more favorable in applications with flue gastemperatures of 1090 °C or higher. As the example in Table 5showed, if oxygen-enriched air containing 30% oxygen is usedat a flue gas temperature of 1649 °C, the natural gas fuelrequired is lowered by 45% (or $1.8 with an assumption of $4/1000 scf natural gas). With a production cost of $35/ton EPO2using membrane processes and the amount of 0.018 ton EPO2required for the combustion (see Table 5), the net economicsaving becomes 29%.

8. CONCLUSIONS

A techno-economic analysis of membrane-based oxygen-enriched combustion was performed, providing guidelines onmembrane processes and membrane separation propertiesrequired to achieve energy and economic benefits by usingoxygen-enriched combustion, instead of combustion with air. Anew membrane process using perfluoropolymer-based thin filmcomposite membranes was examined for the application,showing that membrane processes can produce oxygen-enriched air in an energy-efficient manner. Integratedmembrane/combustion process trains can significantly improvethe energy efficiency and economics of combustion processes,andalthough not discussed in detail herewill also reducethe cost of CO2 capture from flue gases throughout the

Figure 19. Required membrane separation properties to produce 30%oxygen-enriched air at $45/ton EPO2 in three membrane processes:countercurrent/sweep (sweep, Figure 4), feed compression (Figure3a), and permeate vacuum (vacuum, Figure 3b). The sweep processesuse the countercurrent modules that were fabricated and tested in thisstudy (as shown in section 5.3 Module Fabrication and Character-ization).

Figure 20. Membrane separation properties required to produce 30%oxygen-enriched air at $35, $45, and $55 per ton EPO2 in thecountercurrent/sweep membrane process (Figure 4). The perfluor-opolymer-based membranes studied (PFP) show economic advantagesover commercial O2/N2 separation membranes, such as polysulfone(PSf), polyimides (PI), and poly(phenylene oxide) (PPO).12,19

Figure 21. Effect of membrane-based oxygen-enriched combustion onnet economic savings at various exhaust temperatures. The membraneprocesses use membranes with oxygen permeance of 1200 gpu andO2/N2 selectivity of 3.0. The base case uses 5% excess air forcombustion, as shown in Table 5.






manufacturing industries. More specific conclusions aresummarized below.

1. A new countercurrent/sweep membrane process wasevaluated for producing oxygen-enriched air, whichincorporates a turboexpander into the residue streamto recover energy, and uses low-pressure air as acountercurrent sweep in the permeate to increase theoxygen driving force across the membrane. The counter-current/sweep process shows the best performanceamong the three membrane processes evaluated.

2. Thin-film composite membranes based on perfluoropol-ymers were successfully prepared, with oxygen perme-ance as high as 1200 gpu and O2/N2 selectivity of 3.0.The permeance and selectivity of these membranes arestable over the time period tested (60 days), even with aselective layer as thin as 50 nm. These membranes werefabricated into bench-scale spiral-wound modules withcountercurrent designs for parametric tests, whichshowed oxygen/nitrogen separation properties similarto those of membrane stamps.

3. The effect of membrane separation properties on theenergy consumption and operating cost of producingoxygen-enriched air was studied. The membraneprocesses can produce O2-enriched air at an energycost of 1.4−1.7 MMBtu/ton EPO2, which is competitivecompared with conventional technologies such ascryogenics and vacuum pressure swing adsorptionprocesses, especially for small-scale oxygen production.The perfluoropolymer-based membranes developed inthis work produce 30% oxygen-enriched air at a cost of$35/ton EPO2, which is much lower than that possiblewith the commercial membranes available for nitrogengeneration from air.

4. The membrane-based oxygen-enriched combustionprocesses show good energy savings (defined as thefuel savings less the energy consumption of producingoxygen-enriched air) and economic benefits (defined asthe value of fuel saved less the operating cost ofproducing oxygen-enriched air), especially at flue gastemperatures higher than 1090 °C (or 2000 °F). Forexample, at a flue gas temperature of 1649 °C (or 3000°F), membrane-based oxygen-enriched combustionshows a net energy savings of 35% and a net economicbenefit of 29%, compared to the combustion processeswith air.

5. The effect of oxygen-enriched air on NOx emissions wasexperimentally evaluated using a natural gas furnace,which confirms that oxygen-enriched combustion maynot produce higher levels of NOx than normal air firing,if lancing of combustion air is used and the excess airlevels are controlled.

Looking forward, the main barrier to adoption of membrane-based oxygen-enriched combustion lies in the proven cost-effectiveness of producing oxygen-enriched air. Membranepermeance and selectivity should be further increased to reducethe capital and operating cost of installed systems, improvingtheir competitive position against the firmly entrenched,established technologies. The configuration of countercur-rent/sweep modules can be optimized to further improve thesweep efficiency. Finally, testing of oxygen-enrichmentmembrane systems integrated with specific combustionprocesses will be key to better understanding issues such as

NOx emissions, the actual net energy savings (including wasteheat recovery considerations), the required oxygen concen-tration, and the cost of furnace retrofitting that could impactprocess economics.

■ AUTHOR INFORMATIONCorresponding Author*Tel.: +1-650-543-3359. Fax: +1-650-328-6580. E-mail:[email protected] Address⊥J.J.: School of Chemical Sciences, The University of Auckland,Tamaki Innovation Campus, 261 Morrin Road, St. Johns,Auckland Private Bag 92019, Auckland 1142, New Zealand.NotesThe authors declare no competing financial interest.

■ ACKNOWLEDGMENTSWe gratefully acknowledge the partial financial support of thiswork by the United States Department of Energy, Golden FieldOffice, Advanced Manufacturing Offices, Grand ChallengeProjects (grant number DE-EE0003462).

■ ABBREVIATIONSAm = membrane area (cm2)DA = diffusivity of gas A (cm2/s)DA/DB = diffusivity selectivity of gas A over gas BdV/dt = steady state volumetric displacement rate of thesoap film (cm3/s)EPO2 = equivalent pure oxygen, as defined in eq 5gpu = gas permeation unit [10−6 cm3(STP)/cm2·s·cmHg]JA = permeance of composite membrane to gas A (gpu)l = thickness of the membrane selective layer (cm)MMBtu = million British thermal unitsNA = steady state flux of gas A across the membrane[cm3(STP)/s][O2] = oxygen concentration (vol %)p1 = membrane permeate pressure (cmHg)p2 = membrane feed pressure (cmHg)PA = permeability of membrane selective layer to gas A(Barrer)PA/l = permeance of composite membrane to gas A (gpu)ΔpA,0 = partial pressure difference across the membrane forgas A between feed and sweep + permeate streams (cmHg)ΔpA,L = partial pressure difference across the membrane forgas A between residue and sweep streams (cmHg)SA = solubility of gas A in a polymer [cm3(STP)/(cm3

polymer atm))SA/SB = solubility selectivity of gas A over gas BT = temperature (K)V = volume of oxygen-enriched air [m3(STP)]

Greek LetterαA/B = permeance selectivity of gas A over gas B

Subscript1 = membrane permeate side2 = membrane feed sideA = gas component AB = gas component B

■ REFERENCES(1) Kimura, S. G.; Browall, W. R. Membrane oxygen enrichment I:Demonstration of membrane oxygen enrichment for natural gascombustion. J. Membr. Sci. 1986, 29, 69−77.



mailto:[email protected]

(2) Baukal, C. E. Oxygen-Enhanced Combustion: CRC Press: BocaRaton, FL, 1998.(3) Hendershot, R. J.; Lebrecht, T. D.; Easterbrook, N. C. Useoxygen to improve combustion and oxidation. Chem. Eng. Progress2010, 57−61.(4) Prasad, R.; Shaner, R. L.; Doshi, K. J. Comparison of membraneswith other gas separation technologies. In Polymeric Gas SeparationMembranes; Paul, D. R., Yampol’skii, Y. P., Eds.; CRC Press, Inc.: BocaRaton, FL, 1994; pp 531−614.(5) Prasad, R.; Notaro, F.; Thompson, D. R. Evolution of membranesin commercial air separation. J. Membr. Sci. 1994, 94, 225−248.(6) Bisio, G.; Bosio, A.; Rubatto, G. Thermodynamics applied tooxygen enrichment of combustion air. Eng. Convers. Manage. 2002, 43,2589−2600.(7) Figueroa, J. D.; Fout, T.; Plasynski, S.; McIlvried, H.; Srivastava,R. D. Advances in CO2 capture technology - The U.S. Deparatment ofEnergy’s Carbon Sequestration Program. Int. J. Greenh. Gas Con. 2008,2, 9−20.(8) Merkel, T. C.; Lin, H.; Wei, X.; Baker, R. W. Power plant post-combustion carbon dioxide capture: An opportunity for membranes. J.Membr. Sci. 2010, 359, 126−139.(9) Merkel, T. C.; Wei, X.; He, Z.; White, L. S.; Wijmans, J. G.;Baker, R. W. Selective exhaust gas recycle with membranes for CO2

capture from natural gas combined cycle power plants. Ind. Eng. Chem.Res. 2013, 52, 1150−1159.(10) Belaissaoui, B.; Moullec, Y. L.; Willson, D.; Favre, E. Hybridmembrane cryogenic process for post-combustion CO2 capture. J.Membr. Sci. 2012, 415−416, 424−434.(11) Wijmans, J. G.; Baker, R. W. The solution-diffusion model: Areview. J. Membr. Sci. 1995, 107, 1−21.(12) Baker, R. W. Membrane Technology and Applications, 3rd ed.;John Wiley and Sons, Ltd.: Chichester, UK, 2012.(13) Bhide, B. D.; Stern, S. A. A new evaluation of membranepocessess for the oxygen-enrichement of air. I. Identification ofoptimum operating conditions and process configurations. J. Membr.Sci. 1991, 62, 13−35.(14) Freeman, B. D. Basis of permeability/selectivity tradeoffrelations in polymeric gas separation membranes. Macromolecules1999, 32, 375−380.(15) Lin, H.; Freeman, B. D. Gas solubility, diffusivity andpermeability in poly(ethylene oxide). J. Membr. Sci. 2004, 239, 105−117.(16) Reid, R. C.; Prausnitz, J. M.; Poling, B. E. The Properties of Gasesand Liquids; McGraw-Hill: New York, 1987.(17) Matson, S. L.; Ward, W. J.; Kimura, S. G.; Browall, W. R.Membrane oxygen enrichment II. Economic assessment. J. Membr. Sci.1986, 29, 79−96.(18) Takada, K.; Matsuya, H.; Masuda, T.; Higashimura, T. Gaspermeability of polyacetylenes carrying substituents. J. Appl. Polym. Sci.1985, 30, 1605−1616.(19) Li, X.; Huang, M. Multilayer ultrathin-film compositemembranes for oxygen enrichment. J. Appl. Polym. Sci. 1997, 66,2139−2147.(20) Dyer, P. N.; Richards, R. E.; Russek, S. L.; Taylor, D. M. Iontransport membrane technology for oxygen separation and syngasproduction. Solid State Ionics 2000, 134, 21−33.(21) Shoji, M.; Nishide, H. Fluorophilic cobalt phthalocyanine-containing Nafion membrane: High oxygen permeability and protonconductivity in the membrane. Polym. Adv. Technol. 2010, 646−650.(22) Figoli, A.; Sager, W. F. C.; Mulder, M. H. V. Facilitated oxygentransport in liquid membranes: Review and new concepts. J. Membr.Sci. 2001, 181, 97−110.(23) Bhide, B. D.; Stern, S. A. A new evaluation of membranepocessess for the oxygen-enrichement of air. II. Effects of economicparameters and membrane properties. J. Membr. Sci. 1991, 62, 37−58.(24) Prasad, R. Membrane oxygen process and system. US Patent5,185,014, 1993.(25) Callaghan, K. R.; Nemser, S. M. Method of producing nitrogenenriched air. EP Patent 1,255,605 B1, 2001.

(26) Lin, H.; Thompson, S. M.; Serbanescu-Martin, A.; Wijmans, J.G.; Amo, K. D.; Lokhandwala, K. A.; Low, B.; Merkel, T. C.Dehydration of natural gas using membranes. Part II: Sweep/countercurrent design and field test. J. Membr. Sci. 2013, 432, 106−114.(27) Pinnau, I.; Toy, L. G. Solid polymer electrolyte compositemembranes for olefin/paraffin separation. J. Membr. Sci. 2001, 184,39−48.(28) Reddy, D.; Moon, T. Y.; and Reineke, C. E. Counter current dual-flow spiral wound dual-pipe membrane separation. U.S. Patent 5,034,126,1991.(29) Cath, T. Y.; Childress, A. E.; Elimelech, M. Forward osmosis:Principles, applications, and recent developments. J. Membr. Sci. 2006,261, 70−87.(30) Mehta, G. D. Further results on the performance of present-dayosmotic membranes in various osmotic regions. J. Membr. Sci. 1982,10, 3−19.(31) Bayer, C.; Follmann, M.; Breisig, H.; Wienk, I. M.; Cuperus, F.P.; Wessling, M.; Melin, T. On the design of a 4-end spiral-wound L/Lextraction membrane module. Ind. Eng. Chem. Res. 2013, 52, 1004−1014.(32) Hao, P.; Lipscomb, G. G. The effect of sweep uniformity on gasdehydration modules. In Membrane Gas Separation; Yampolskii, Y.,Freeman, B. D., Eds.; John Wiley & Sons, Ltd: New York, 2010; pp333−352.(33) Lin, H.; Freeman, B. D. Permeation and diffusion. In Springer-Handbook of Materials Measurement Methods; Czichos, H., Smith, L. E.,Saito, T., Eds.; Springer: New York, 2006, pp 371−387.(34) Ku, A. Y.; Kulkarni, P.; Shisler, R.; Wei, W. Membraneperformance requirements for carbon dioxide capture using hydrogen-selective membranes in integrated gasification combined cycle (IGCC)power plants. J. Membr. Sci. 2011, 367, 233−239.(35) McCabe, W. L.; Smith, J. C.; Harriott, P. Unit Operations ofChemical Engineering, 4th ed.; Chemical Engineering Series; McGraw-Hill Book Company: New York, 1985.(36) Lin, H.; Thompson, S. M.; Serbanescu-Martin, A.; Wijmans, H.G.; Amo, K. D.; Lokhandwala, K.; Merkel, T. C. Dehydration ofnatural gas using membranes. Part I: Composite membranes. J. Membr.Sci. 2012, 413−414, 70−81.(37) Zhou, J.; Jin, J.; Haldeman, A. T.; Wagener, E. H.; Husson, S. M.Formation and characterization of perfluorocyclebutyl polymer thinfilms. J. Appl. Polym. Sci. 2013, DOI: 10.1002/app.39048.(38) Henis, J. M.; Tripodi, M. K. Composite hollow fiber membranesfor gas separation: The resistance model approach. J. Membr. Sci. 1981,8, 233−246.(39) Huang, Y.; Wang, X.; Paul, D. R. Physical aging of thin glassypolymer films: Free volume interpretation. J. Membr. Sci. 2006, 277,219−229.(40) Rowe, B. W.; Freeman, B. D.; Paul, D. R. Physical aging ofultrathin glassy polymer films tracked by gas permeability. Polymer2009, 50, 5565−5575.(41) Pfromm, P.; Koros, W. J. Accelerated physical aging of thinglassy polymer films: Evidence from gas transport measurements.Polymer 1995, 36, 2379−2387.(42) Chung, T. S.; Kafchinski, E. R. Aging phenomenon of 6FDA-polyimide/polyacrylonitrile composite hollow fibers. J. Appl. Polym.Sci. 1996, 59, 77−82.(43) Huang, Y.; Paul, D. R. Effect of film thickness on the gas-permeation characteristics of glassy polymer membranes. Ind. Eng.Chem. Res. 2007, 46, 2342−2347.(44) Nagai, K.; Masuda, T.; Nakagawa, T.; Freeman, B. D.; Pinnau, I.Poly[1-(trimethylsilyl)-1-propyne] and related polymers: Synthesis,properties and functions. Prog. Polym. Sci. 2001, 721−798.(45) Carta, M.; Malpass-Evans, R.; Croad, M.; Rogan, Y.; Jansen, J.C.; Bernardo, P.; Bazzarelli, F.; McKeown, N. B. An efficient polymermolecular sieve for membrane gas separations. Science 2013, 339,303−307.(46) Bushell, A. F.; et al. Nanoporous organic polymer/cagecomposite membranes. Angew. Chem., Int. Ed. 2013, 52, 1253−1256.



(47) Budd, P. M.; Elabas, E. S.; Ghanem, B. S.; Makhseed, S.;McKeown, N. B.; Msayib, K.; Tattershall, C. E.; Wong, D. Solution-processed, organophilic membrane derived from a polymer withintrinsic microporosity. Adv. Mater. 2004, 16, 456−459.(48) Robeson, L. M. The upper bound revisited. J. Membr. Sci. 2008,320, 390−400.(49) Robeson, L. M. Correlation of separation factor versuspermeability for polymeric membranes. J. Membr. Sci. 1991, 62,165−185.(50) Ren, J. Y.; Fan, Y.; Egolfopoulos, F. N.; Tsotsis, T. T.Membrane-based reactive separations for power generation applica-tions: oxygen lancing. Chem. Eng. Sci. 2003, 58, 1043−1052.(51) Rue, D. M.; Servaites, J.; Wolf, W. Industrial Glass BandwidthAnalysis. Gas Research Institute Final report to the U.S. DOE, DE-FC36-03GO13092, 2007.(52) Baker, R. W.; Lokhandwala, K. Natural gas processing withmembranes: An overview. Ind. Eng. Chem. Res. 2008, 47, 2109−2121.



Membrane-Based Oxygen-Enriched Combustion

Documents