Post-combustion CO2 capture using super-hydrophobic, polyether ether ketone, hollow fiber membrane contactors

Journal of Membrane Science 430 (2013) 79–86

Contents lists available at SciVerse ScienceDirect

Journal of Membrane Science journal homepage: www.elsevier.com/locate/memsci

Post-combustion CO2 capture using super-hydrophobic, polyether ether ketone, hollow fiber membrane contactors Shiguang Li a,n, Dennis J. Rocha a, S. James Zhou a, Howard S. Meyer a, Benjamin Bikson b, Yong Ding b a b

Gas Technology Institute, 1700 S. Mount Prospect Road, Des Plaines, IL 60018, United States PoroGen Corporation, 6C Gill St., Woburn, MA 01801, United States

a r t i c l e i n f o

abstract

Article history: Received 14 October 2012 Received in revised form 30 November 2012 Accepted 2 December 2012 Available online 7 December 2012

The feasibility of utilizing a super-hydrophobic polyether ether ketone (PEEK) hollow fiber membrane contactor in combination with chemical solvents to separate and capture CO2 from simulated flue gases was investigated. Greater than 90% CO2 capture with greater than 95% CO2 purity has been achieved in one stage with both activated methyldiethanolamine (aMDEA) and activated K2CO3 solvents. The measured volumetric mass transfer coefficient was as high as 1.7 s  1, which is more than 20 times greater than the mass transfer coefficient of a packed column. Preliminary tests indicated that the CO2 capture performance was not affected by flue gas contaminants, including O2, NO2, and SO2, with aMDEA solvent. The PEEK membrane module showed good mechanical properties and stable permeation properties at process design conditions for duration of the test (120 h). The process economics were evaluated assuming direct substitution of the conventional absorber by the membrane contactor, while using a conventional packed column in the regeneration step. Relative to the cost estimation with no CO2 capture, the evaluation indicates a 56% increase in the levelized cost of electricity (LCOE) for the PEEK membrane contactor technology, which is 29% lower than DOE’s benchmark amine absorption technology (85% increase in LCOE). & 2012 Elsevier B.V. All rights reserved.

Keywords: Gas separation Carbon capture Membrane contactor CO2 capture Polymer membrane Hollow fiber membrane Polyether ether ketone membrane

1. Introduction The greatest concern to climate change is the emission of greenhouse gases, especially CO2 from a range of sources. According to the Environmental Protection Agency, the US emitted 6.1 billion metric tons of CO2 to the atmosphere in 2007 [1]. About 40% of this CO2 was produced by electric generating power plants, of which  50% are fueled by coal. Existing coal-fired power plants emit  85% of the total CO2 of all power plants’ generating CO2. Therefore, to address concerns about global climate change and to reduce US greenhouse gas emissions of 17% by 2020 and 83% by 2050 from a 2005 baseline [2], Federal legislation targeting coal-fired power plants is likely. The US Department of Energy (DOE) has a 2020 Carbon Capture Program post-combustion capture goal of achieving 90% capture in existing plants with less than a 35% increase in the levelized cost of energy (LCOE). To achieve this goal by any technological means is very difficult, because flue gas is hot, dilute in CO2 content, near atmospheric pressure, high in volume, and often contaminated with other impurities (O2, SOx, NOx, and ash). Amine absorption is the current industry and DOE benchmark technology for capture of CO2 from power plant flue gas. Residual oxygen in the flue gas is

n

Corresponding author. Tel.: þ1 847 544 3478; fax: þ 1 847 544 3470. E-mail address: [email protected] (S. Li).

0376-7388/$ - see front matter & 2012 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.memsci.2012.12.001

especially troublesome for conventional amine plants because of oxidative degradation of the amine. These factors result in enormous amine circulation rates, large equipment, and large energy requirements. In addition, the heat duty of the stripper places a substantial burden on the steam supply. It is estimated that for every pound of CO2 captured 2 to 3 pounds of steam is required [3,4]. DOE/NETL systems analysis studies estimated that using aqueous monoethanolamine absorption process to capture 90% of the CO2 from flue gas in existing plants will result in an increase in LCOE services by 75–85% [5]. These values are well above the 2020 DOE NETL Carbon Capture Program post-combustion capture cost goal (35% increase in LCOE). Therefore, it is important to develop new advanced CO2 capture technologies in order to maintain the cost-effectiveness of U.S. coal-fired power generation. Recently published system analysis and feasibility studies demonstrate that gas separation membranes are a technically and economically viable option for CO2 capture from the flue gas exhaust in pulverized coal-fired (PC) power plants [6]. Furthermore, membranes are compact and can be retrofitted onto the tail end of a power-plant flue gas stream without complicated integration schemes. Conventional gas separation membrane process operates by a solution/diffusion mechanism, and the separation driving force is provided by the partial pressure difference of each component across the membrane. This process requires either flue gas compression, permeate sweep, application of permeate side vacuum, or a combination of these steps to provide the separation driving

80

S. Li et al. / Journal of Membrane Science 430 (2013) 79–86

force required. Elaborate process design and optimization become prerequisite for conventional membrane processes in CO2 capture from flue gases [7]. The main limitation of conventional membrane processes is the process pressure ratio (feed gas pressure/permeate gas pressure) limitation [8]. A flow of component m across the membrane can only occur if the partial pressure of m on the feed side is greater than that on the permeate side. That is: ym,f pf Zym,p pp

ð1Þ

the operating conditions encountered in typical flue gas applications. It can withstand contact with all common treating solvents or advanced absorption solvents under development. Specifically, a ‘‘real-world’’ assessment of the necessary and desirable features including the following elements has been undertaken:

 Membrane module: membrane modules with their perfor

Thus, the maximum separation achieved by the membrane can be expressed as: pf ym,p Z pp ym,f

 ð2Þ

where, pf and pp are the feed and permeate side pressures, ym,f and ym,p are mole fractions of component m in the feed and permeate sides, respectively. Eq. (2) simply illustrates that the enrichment achievable in the permeate relative to the feed (ym,p/ym,f) is always less than the feed-to-permeate pressure ratio (pf/pp), no matter how selective the membrane. In practical gas separation applications, the pressure ratio across the membrane is usually between 5 and 15 [9]. When the membrane separation process is pressure ratio limited, the product CO2 concentration will be limited even when the membrane selectivity is much larger than the pressure ratio. Thus, multiple membrane stages are required to generate greater than 95% pure CO2 product (DOE’s target) from flue gases using the conventional membrane process. Membrane contactor process (also known as hybrid membrane/ absorption process) combines advantageous features of both absorption and membrane processes to provide a cost-effective solution for CO2 capture from flue gases. Hollow fiber membrane contactors for CO2 capture, especially the absorption process, have been reported [10–14]. In this process, CO2-containing gas passes through smalldiameter membrane tubes (hollow fibers with porous walls) while a CO2 selective solvent (typically an amine solution) flows on the shell side of the membrane. CO2 permeates through the membrane and is absorbed in the solvent. The CO2 rich solvent can be regenerated in a second membrane module operated in a reverse process. The specific surface area per volume for hollow fiber membrane contactors can be as high as  1000–9000 m2/m3, which is up to two orders of magnitude greater than conventional contactors (free dispersion columns:  3.0–35 m2/m3, packed and tray columns: 30–300 m2/m3, mechanically agitated columns: 160–500 m2/m3) [15,16]. Thus, the use of a membrane contactor instead of a conventional amine scrubber leads to a much smaller space requirement. This technology is well suited for new and existing PC power plants due to the reduced footprint requirement and a much lower visual impact as well as providing more options for placement in the confines of existing plants. In the hybrid membrane/absorption process, the permeate side partial pressure of CO2 can be considered close to zero due to the chemical reaction of CO2 with the absorption solvent, and thus overcomes the pressure ratio problem encountered by the conventional gas membrane process. Feed compression or permeate vacuum is not required to create the separation driving force for gas molecules to be transported through the membrane. The process selectivity for the hybrid membrane/absorption process is determined by the chemical affinity of the absorption solvent to CO2. Therefore, high purity CO2 product can be realized in a single stage hybrid membrane/absorption process. The objective of the current study was to develop and demonstrate a practical and cost effective hybrid membrane/absorption process for CO2 separation and capture. The membrane contactor is based on chemically and thermally resistant commercially engineered plastic PEEK [17], which is virtually non-destructible under

 

mance essentially capable of being linearly scaled to commercial size were used. Solvent: solvents used in our membrane contactor technology are commercially available. Four types of solvents were investigated. Feed: simulated flue gas compositions were used for existing pulverized coal power plants. The effects of flue gas contaminants, such as O2, NOx, and SOx on membrane contactor performance were investigated. Stability: membrane contactor stability at process design conditions was investigated by continuously performing CO2 capture for 120 h. Techno-economic analysis: experimental results were combined with process design to determine the technical performance for a variety of process scenarios and economic costs associated with each scenario.

2. Experimental methods 2.1. Hollow fiber membrane fabrication and surface modification The hollow fiber membranes are manufactured from the best in class commercial engineered plastic, PEEK. Porous PEEK hollow fibers used in preparation of super-hydrophobic membranes are manufactured by a high temperature melt extrusion process. In this process, a mixture containing PEEK and PEI (polyether imide), is melt extruded to form a hollow fiber, which is then cooled in air and sent to reagent bath, where PEI is removed. The resulting product is porous PEEK fiber. The details of the process were described previously [18]. The super-hydrophobicity of the porous PEEK membrane was generated by surface modification with a functional perfluoro oligomer. Prior to grafting with the perfluoro oligomer the surface of the porous PEEK was functionalized with –OH groups by reacting ketone groups in the PEEK polymer backbone with monoethanolamine. The functionalized porous PEEK was prepared in a single step Reactive Porogen Removal process during porous PEEK fiber preparation according to US Patent 7,176,273 [19]. The process is illustrated schematically in Fig. 1. 2.2. Membrane module preparation Module design and construction have significant impact on the overall gas mass transfer coefficient by minimizing liquid side resistance and maximizing the driving force. The hollow fiber membrane modules were of the four-port, counter-current flow design, which took into account gas side and liquid side pressure drops. The hollow fiber cartridges were formed by computercontrolled helical winding hollow fibers around a mandrel. The cartridges (Fig. 2) were nominally 2-inch- (5-cm-) diameter and 12-inch- (30.5-cm-) long. The effective membrane surface areas were about 0.12–1 m2. The cartridge was sealed with o-rings and housed in a stainless steel pressure vessel. 2.3. Membrane module characterization The non-wetting characteristics of membrane modules were determined by pressurizing a feed liquid in the shell side of the module and observing collection of liquid in the tube side if any.

S. Li et al. / Journal of Membrane Science 430 (2013) 79–86

Any liquid collected on the tube side would indicate wet out of the membrane. The liquid used in these quality control tests was MDEA/water (50/50 volume) solution. The wettability tests were extended to include longer term exposure to the aqueous MDEA solution ( Z100 h) and exposure at higher temperatures (50–60 1C). The latter was comparable to flue gas conditions. The membrane intrinsic gas permeation properties for CO2 and N2 were measured in a flow system shown in Fig. 3. The gases were fed to the tube side of the module. Fluxes were measured using a soap film bubble flow meter. The pressure normalized flux, permeance, is: P¼

J

ð3Þ

Dp

O

+ H 2N

N

N OH

OH

N OH

+

O

Rf

O Rf OH

Hydrophobic

Hydrophilic

Fig. 1. Membrane surface modification: (a) functionalization of porous PEEK with –OH groups, and (b) reaction to form hydrophobic surface (Rf is a perfluorinated group with average molecular weight about 5000 Da).

81

Dp ¼ P tube Pshell

ð4Þ

where J is the total steady state flux through the membrane; Dp is pressure differential between the tube and shell sides. Note that pressure drop was observed between the inlet and outlet of the tube side, and thus an average tube-side pressure Ptube was used for calculation in Eq. (4). 2.4. Membrane absorption The membrane modules were mounted in a membrane contactor skid for CO2 capture testing. The skid was designed for 25 kWe equivalent CO2 capture (50.6 lb/h of CO2). The membrane absorption process test apparatus process flow diagram is shown in Fig. 4. Mass flow controllers were used to control feed gas composition from pure CO2 and N2 gases, which were in the molar ratio of 13:87 if not indicated otherwise. The mixed gas stream was heated by a flowthrough heater and then sent to a bubbler filled with water to humidify the gas. Humidity measurement indicated that the stream after the knock-out vessel was saturated with water at given temperature. The stream was then directed to the tube side of the module. Next, the lean solvent was directed to the shell side of the module, as shown in Fig. 4. During membrane absorption, gas-side CO2 permeated through the membrane and was absorbed in the solvent. The CO2 concentrations of the simulated flue gas feed (gas inlet) and CO2-depleted gas residue (gas outlet) were measured by CO2 analyzers 1 (the QUBIT SYSTEMS S158-15 IRGA Carbon Dioxide (CO2) Analyzer, 0–15% range) and 2 (the QUBIT SYSTEMS S158 IRGA Carbon Dioxide (CO2) Analyzer, 0–10% range), respectively. Four types of commercial solvents were tested in the current study as follows: (1) 30 wt% diethanolamine (DEA)/H2O, (2) 40 wt% methyldiethanolamine (MDEA)/H2O, (3) 20 wt% activated K2CO3/water, and (4) activated methyldiethanolamine (aMDEA, 40 wt% amine solution). Other operating conditions are listed in Table 1. For comparison to conventional contactors, volumetric mass transfer coefficient (KGAv, (s  1) ) was used in the current study and calculated as follows, K G Av ¼ K G  Av

ð5Þ

where KG is the mass transfer coefficient (m/s), and Av is the specific surface area per volume of the membrane module. 2.5. Effects of flue gas contaminants The effects of O2, NOx and SOx flue gas contaminants on CO2 capture a PEEK- membrane contactor were studied using aMDEA solvent and the following pre-mixed feed gases (Matheson Tri-Gas) at levels consistent with Illinois (a state of the US) basin coal feeds: a) 145 ppmv SO2, 3.06% O2, 14.96% CO2, and balance N2, b) 66 ppmv NO2, 3.27% O2, 12.98% CO2, and balance N2.

P

Fig. 2. Helically wound structured hollow fiber cartridge.

Membrane module in oven

Flow Check meter valve

Pressure gauge Backpressure regulator Gas to vent

Valve Filter P CO2orN2

P

Pressure gauge

Pressure gauge

Flow meter Gas to vent

Fig. 3. Process diagram for membrane intrinsic permeance testing.

82

S. Li et al. / Journal of Membrane Science 430 (2013) 79–86

Treated gas to vent CO2 analyzer 2

Membrane module Flow through Heater

N2 Knock-out vessel

Filter

CO2 Flow through Heater Water column

CO2 analyzer 1 Rich solvent

Lean solvent

Fig. 4. Process diagram for bench-scale membrane contactor CO2 absorption testing.

Table 1 Laboratory test parameters.

Table 2 Stability operating conditions.

Parameter

Condition

Parameter

Lean solvent CO2 loading Gas inlet temperature Gas feed

o 0.02 mol CO2/mol amine or K2CO3 37–57 1C Simulated flue gas with CO2: N2 molar ratio of 13:87 if not indicated otherwise 0.12–1 m2 o 6 psig Saturation

Absorption (membrane) Gas inlet temperature Simulated flue gas CO2 inlet concentration Membrane contactor surface area Gas flow rate Inlet gas pressure Solvent Liquid flow rate Inlet liquid pressure Desorption (tower) Liquid temperature of desorption tower Liquid flow rate N2 stripping flow rate

Membrane contactor surface area Inlet gas pressure Moisture

2.6. Stability Membrane contactor stability at process design conditions was investigated by continuously performing CO2 capture with aMDEA solvent for  120 h using a membrane contactor module that has been in contact with the aMDEA solvent for more than 9 months. In this test, the CO2-loaded rich solvent was sent to a regeneration tower in which N2 gas was used as a stripping gas. The membrane absorber and regeneration tower were integrated so that the CO2 loading of lean solvent remained low and constant during the test. The operating conditions are listed in Table 2.

3. Results and discussion 3.1. Membrane and module properties Post-combustion CO2 capture process conditions require development of hollow fibers with large diameter bore dimensions to minimize the feed side pressure drop; even a small increase in the gas side pressure drop can introduce a significant energy penalty on the entire power plant. The fiber dimensions manufactured in the current study were  0.43 mm outside diameter and 0.28 mm inside diameter. The constructed membrane modules had specific surface areas per volume of  2400 m2/m3. The membrane was not wetted by ethanol (liquid drop on membrane surface is shown in Fig. 5) indicating a superhydrophobic surface. This super-hydrophobicity is generated via

Condition

 41 1C 17.2 mol% (balance N2) 0.13 m2 1 L(STP)/min 3.6–3.8 psig aMDEA 0.40 L/min 4.4–6.6 psig 85–94 1C 0.40 L/min 1 L (STP)/min

the so called ‘‘Lotus effect’’ [20], due to a combination of nanometer size surface pores, exceptionally uniform pore size distribution, and the perfluoro-hydrocarbon surface chemistry. The wettability tests indicated that membrane modules were impermeable to a 50/50 (volume) MDEA/water solution at 50–60 1C and trans-membrane pressure differential of 5–60 psi between the gas and the liquid sides for greater than 100 h. High membrane intrinsic CO2 permeance is important in attaining high CO2 capture rates in membrane contactor mode. By modifying and optimizing membrane preparation procedures and module construction conditions, a membrane intrinsic CO2 permeance as high as 1000 GPU (1 GPU ¼10  6 cm3 (STP)/ (cm2  s cmHg)) was achieved (modules 2PG285 and 2PG286 in Table 3). Table 3 also lists CO2 permeances for other experimental modules used in this study that varied in membrane morphology and design of module structured packing. 3.2. CO2 capture performance in membrane contactor Table 4 shows CO2 capture performance for module 2PG285 with MDEA, DEA, and activated K2CO3 aqueous solvents. The equivalent CO2 permeances measured from contactor testing with MDEA and DEA solvents were only 230 and 330 GPU,

S. Li et al. / Journal of Membrane Science 430 (2013) 79–86

83

Table 4 CO2 capture performance for Module 2PG285 with various solvents. Solvent

Total gas flow rate, L(STP)/min

Alcohol droplet

40 wt% MDEA/H2O 1.0 1.4 30 wt% DEA/H2O 3.5 activated 20% K2CO3/H2O a

CO2 CO2 removal, capture % rate, kg/m2/h

Volumetric CO2 permeance, mass GPUa transfer coefficient, (s  1)

74.8 93.1 87.4

0.40 0.58 1.4

0.14 0.17 0.40

230 330 760

1 GPU ¼ 10  6 cm3 (STP)/(cm2 s cmHg).

Fig. 5. Alcohol drop on the porous PEEK membrane surface.

CO2 permeance (GPUa)

2PG283 2PG285 2PG286 2PG368

590 1000 1000 810

a

1 GPU ¼ 10  6 cm3 (STP)/(cm2 s cmHg).

CO2 removal (%)

Membrane module

80

0.6

60 0.4 40 0.2

20 0

respectively (Table 4), which were much lower than the intrinsic CO2 permeance (1000 GPU). The overall resistance for CO2 to transport through the membrane contactors includes three parts: (1) the gas phase, (2) the membrane, and (3) the liquid phase. The resistance in the gas phase is typically very small as compared to other two [21]. As demonstrated by the results from module 2PG285, the equivalent CO2 permeance is much lower than the membrane intrinsic permeance when MDEA and DEA solvents were used, indicating a high liquid phase transport resistance. This is mainly due to liquid side concentration polarization created by the low reaction rate between CO2 and the solvents. Two factors affect this concentration polarization: (1) CO2 diffusion rate in the solvent, and (2) the reaction rate of CO2 with the solvent. The activated K2CO3/H2O solvent showed the best performance in terms of CO2 capture rate and volumetric mass transfer coefficient (Table 4). The equivalent CO2 permeance measured from contactor CO2 capture testing (760 GPU) was closer to the membrane intrinsic CO2 permeance (1000 GPU) than those of other solvents (230–330 GPU), indicating a lower liquid-side transport resistance when using activated K2CO3/H2O solvent. The lower liquid-side transport resistance was mainly due to the higher reaction rate between CO2 and the activated K2CO3/H2O solvent. Same strategy was thus attempted to the MDEA aqueous solvent. To improve reaction kinetics, a piperazine promoter was added to the MDEA solvent (i.e., aMDEA). The measured equivalent CO2 permeances in subsequent contactor testing were closer to the membrane intrinsic CO2 permeances. For example, for module 2PG286 with intrinsic CO2 permeance of 1000 GPU, the measured CO2 permeance in membrane contactor with aMDEA solvent was 970 GPU at 90% CO2 removal. This is a good indication that there is little concentration polarization when using aMDEA solvent. Fig. 6 shows the CO2 flux through the membrane increases as the total gas flow rate increases. The percentage of CO2 captured, however, decreases with increasing gas flow rate. These

CO2 flux (kg/m2/h)

0.8

100 Table 3 Single-gas permeation properties of the modules.

0 3

4 5 6 Total gas flow rate (L/min)

7

Fig. 6. CO2 removal percent and CO2 flux as a function of total gas flow rate for membrane module 2PG286.

measurements were made for a 13/87 CO2/N2 feed at 49 1C through membrane module 2PG286. Note that 90% CO2 removal can be achieved in one membrane contactor stage. The CO2 flux was 0.41 kg/m2/h at 90% CO2 removal. The volumetric mass transfer coefficient was as high as 1.7 s , which is more than 20 times greater than the highest mass transfer coefficient of a packed column reported (0.075 s ) [15]. 3.3. Effects of flue gas contaminants Carbon dioxide removal from simulated flue gas mixture containing 145 ppmv SO2, 3.06% O2, 14.96% CO2, and balance N2 was tested utilizing membrane module 2PG286 with aMDEA solvent. At steady state operation, the outlet was 1.46% CO2, 22 ppmv SO2, 3.5% O2, and balance N2 (Fig. 7). As shown in Table 5, 91% CO2 removal was achieved and the volumetric mass transfer coefficient was 1.6 s . The slightly lower volumetric mass transfer coefficient (relative to the feed without SO2 and O2) may be due to the higher percentage of CO2 removal (91%) as seen by the trade-off in Fig. 6. Considering this factor, the CO2 capture performance of the membrane had hardly been affected by SO2. Similar behavior was observed when testing a feed containing 66 ppmv NO2, 3.27% O2, 12.98% CO2, and balance N2. 3.4. Stability Membrane contactor stability at process design conditions was investigated by continuously performing CO2 capture through module 2PG283 with aMDEA solvent for 120 h. As shown in Fig. 8, the CO2 removal rate had been greater than 90% throughout the

84

S. Li et al. / Journal of Membrane Science 430 (2013) 79–86

1.46% CO2 22 ppmv SO2 3.5% O2 Balance N2

CO2 removal (%)

90

Lean aMDEA

90% CO2 removal line

60

30

0 0

15.0% (mol) CO2 145 ppmv SO2 3.1% (mol) O2 Balance N2

20

40

60 80 Time (h)

Table 6 Comparison: membrane contactor vs. conventional gas membrane process.

Table 5 Module 2PG286 performance for a feed containing SO2. CO2 removal

91%

Mass transfer coefficient, (s  1) Gas side DP, (psi) CO2 capture rate, (kg/m2/h)

1.6 1.6 0.5

Membrane technology

Need to create driving force?

CO2/N2 selectivity (a)

Can achieve Z 90% CO2 removal and high CO2 purity in one stage?

Conventional gas membrane process

Yes. Feed compression, permeate sweep or permeate vacuum are required No. Liquid side partial pressure of CO2 close to zero

Determined by the dense ‘‘skin layer’’, typically a ¼20–50

No. Limited by pressure ratio, multi-step process is required Yes

Membrane contactor

test. The gas side pressure drop reached steady state in 45 h and remained at  0.58 psi for the rest of the testing period. After this stability test, the membrane intrinsic permeances for CO2 and N2 was re-measured and were found to be identical to the initial values, indicating good mechanical stability and stable permeation properties of the module. 3.5. Techno-economic analysis Table 6 compares process characteristics of the membrane contactor with the conventional gas separation membrane process as reported by Merkel et al. [7]. Based on results shown above, the hybrid membrane/absorption process can operate at close to atmospheric pressure. This is because the permeate (solvent) side partial pressure of CO2 can be considered to be close to zero due to the fast chemical reaction of CO2 with the absorption solvent. The inlet flue gas pressure must be slightly higher than the ambient pressure in order to ensure uniform flue gas flow through the hollow fibers. Yoon et al. [22] described a theoretical calculation and an experimental method to verify pressure drop through fiber membranes where there is a mass transfer between the tube and shell sides. In our post-combustion CO2 capture process, majority of feed gas ( 88%) remains in the tube side. Thus, the pressure drop for flow through the hollow fiber membrane can be simply estimated by the Hagen–Poiseuille equation: 8Q ZL p  r4

120

Fig. 8. CO2 removal rate as a function of operating time for module 2PG283.

Rich aMDEA

Fig. 7. Separation performance for membrane module 2PG286 with a feed containing SOx and O2.

DP ¼

100

ð6Þ

where Q is the volumetric flow rate, Z is the absolute viscosity of the fluid, L is the length of the hollow fiber, and r is the hydraulic

Determined by the solvent, a 41000

radius of the hollow fiber. Eq. (6) suggests the gas-side pressure drop increases linearly with fiber length for a fixed fiber structure and gas flow rate. Extrapolating the pressure drop obtained experimentally through module 2PG283 (0.59 psi), the gas side pressure drop would be approximately 2.3 psi when using the same type of fibers for the construction of 8-inch-diameter, 60-inch-long commercial size modules for practical application. With 12% of the feed stream permeating to the shell side, the gas side pressure drop is expected to be even lower than 2.3 psi. For such a low pressure drop, commercially available blowers used for flue-gas desulfurization (FGD) can be adopted to the CO2 capture membrane contactor system. Note that membrane selectivity is not required in the hybrid membrane/absorption process. The overall CO2/N2 selectivity is determined by the chemical affinity of the absorption solvent to CO2. That is why over 90% CO2 removal and high purity of CO2 product were achieved in a single stage hybrid membrane/ absorption process (Figs. 6 and 8). To meet the same target by using conventional gas membrane process, Merkel et al. [7] have reported a two-stage and two-step membrane systems with boiler combustion air as a sweep gas. In terms of energy consumption, the hybrid membrane/absorption process still requires energy to regenerate the CO2-loaded solvents. However, the process results in significant energy savings by removing the intense compression required to create the driving force in the conventional gas separation membrane process. Process economics modeling was carried out assuming the substitution of a membrane contactor for CO2 absorption, while still using a packed column for solvent regeneration. We have used cost estimates for the DOE Case 9 (cost estimation with no CO2 capture) and Case 10 (cost estimation with CO2 capture using MEA plant) as the Base Case that represents the current benchmark technology (monoethanolamine

S. Li et al. / Journal of Membrane Science 430 (2013) 79–86

Estimated reboiler heat duties, Btu/lb CO2

(MEA) plant) for electric power generation with CO2 removal (including transport, storage and monitoring) from flue gas generated in a nominal 550 MWe pulverized coal boiler [23]. These scoping economic numbers for the membrane contactor technology, based on activated K2CO3 and aMDEA solvents, were developed to estimate economic advantages of a hybrid membrane absorption/conventional regeneration process over the DOE Case 10. For this comparative study, the total coal feed rate for all the design cases is the same (at 646,589 lb/h). The experimental CO2 flux data at 90% CO2 removal obtained for these solvents (activated K2CO3: 0.39 kg/m2/h, and aMDEA solvent: 0.41 kg/m2/h) were used in our cost estimates. The total CO2 removal rate for these designs cases is about 626.2 Mt/h, corresponding to 90% CO2 capture from a nominal 550 MWe, subcritical pulverized coal power plant. Differences in the reboiler heat-duty requirements for the regeneration of CO2-rich solvent would lead to changes in (1) net electric power generation, and (2) capital costs for the reboiler as well as for the low-pressure (LP) steam turbine units. The estimated reboiler heat duties for the three design cases are shown in Fig. 9. For the membrane contactor application using activated K2CO3, there would be significant reduction in the usage of LP steam for the solvent regeneration unit. For the DOE Case 10, total LP steam flow (at 168 psia and 395 1C) to the amine unit is about 1.995 million lb/h. For the membrane contactor case, the total LP steam required would be about 685,900 lb/h, which would result in an excess of LP steam of about 1.309 million. This extra steam can be used in LP steam turbine to generate about 117,400 kW of additional electric power. In this context, we would need to correspondingly increase the capital expenditure (CAPEX) (estimated for the DOE Case 10) of the turbine system. This reduction in steam usage would also reduce the CAPEX of the reboiler unit of the stripper system. For the membrane contactor design case using the aMDEA solvent, the total LP steam required for solvent regeneration would be about 1.557 million lb/h. Based on a study by Nexant/Bechtel [24], a typical capital investment (Table 7) for the absorber unit is approximately 27% of the total cost of the Amine-based CO2 removal process (estimated at $436 MM, 2006$, for the DOE Case 10). This absorber will be replaced by a membrane contactor unit. According to this Nexant study, the typical investment for the reboiler unit is

1500

Table 7 Key capital cost distribution factors for a typical amine plant for CO2 removal [25]. Absorber Rich/lean exchanger Reboiler and other heat exchangers Stripper Feed cooler Flue gas blower Pumps Others

500

0

MEA

Activated K2CO3

aMDEA

Fig. 9. The estimated heat duties for the three design cases: (1) MEA: the one used in DOE Case 10, estimated from the total LP steam need in the Regenerator Unit; (2) Activated K2CO3: For the K2CO3-based Enhanced LoHeat Benfield process [24], this value has been reported as 18,000–25,000 Btu/lbmol CO2; we have assumed a value of 23,000 Btu/lbmol CO2; and (3) aMDEA: estimated as the sum of the heat of desorption (14.0 kcal/gmol), heat of vaporization of water (10.3 kcal/gmol), and sensible heat required to bring the rich solution to the temperature of the stripper (4.7 kcal/gmol).

27% 19% 15% 10% 9% 9% 8% 3%

Table 8 Key changes in CAPEX (Yr 2006$). Item

DOE case 10 Membrane (amine plant) contactor K2CO3 aMDEA solvent solvent

Absorber unit of the amine plant, $MM (@27% of total amine plant) Reboiler unit of the amine plant (@ 15%), $MM Stripper, $MM Membrane unit, $MM Other equipment, $MM Total CAPEX for the CO2 capture unit, $MM

118

–

–

65

22

39

44 – 209 436

10 104 209 345

12 117 209 377

Table 9 Comparative data on LCOE. Parameter

As-received coal feed rate, (metric tons/day) Capital cost, (mills/kW h)a Fixed operating costs, (mills/kW h) Variable operating costs, (mills/kW h) Coal, (mills/kW h) CO2 transport, storage and monitoring, (mills/kW h) Total LCOE, (mills/kW h) Increase over no capture LCOE, (%) a

1000

85

DOE case 9

4765

DOE case 10

7039

Membrane contactor K2CO3 solvent

aMDEA solvent

5937

6584

34.14 3.99 5.80

68.05 5.81 10.82

58.5 4.82 8.65

61.10 5.43 9.73

20.14 –

29.78 3.91

24.68 3.48

27.83 3.48

64.00 –

118.36 85

100.11 56

107.57 68

Mills/kW h¼ tenths of a cent per kilowatt hour.

approximately 15% of the total cost of the amine process. The reboilers for the membrane plants are prorated on steam requirements. The changes in total CAPEX for the two design cases relative to the DOE Case 10 are summarized in Table 8. The key data on various LCOEs for the design cases are summarized in Table 9. The LCOEs for our hybrid membrane/absorption process using either activated K2CO3 (56%) or aMDEA (68%) solvents are better than DOE’s benchmark amine absorption technology (85% increase in LCOE). The overall goal of DOE/NETL’s carbon capture R&D is to develop advanced technologies that achieve 90% CO2 capture at less than a 35% increase in LCOE for post-combustion capture for new and existing coal-fired power plants. This target can be met by increasing membrane intrinsic CO2 permeance, decreasing membrane module cost, utilizing membrane contactor in regeneration process, and by utilizing new, advanced solvents.

86

S. Li et al. / Journal of Membrane Science 430 (2013) 79–86

4. Summary

 The feasibility of utilizing PEEK membrane based hollow fiber

 

  

contactor in combination with chemical solvents to separate and capture CO2 from simulated flue gases with high efficiency (at least 90% removal) has been successfully established. The specific membrane area per volume was as high as 2400 m2/m3 in scalable membrane modules. The measured volumetric mass transfer coefficient in a PEEK membrane contactor was as high as 1.7 s  1 , which is more than 20 times greater than the highest mass transfer coefficient of a packed column reported. Preliminary tests indicated that the CO2 capture performance was not affected by flue gas contaminants O2, NO2, and SO2 when testing with an aMDEA solvent. The membrane module showed good mechanical properties and stable permeation properties during a stability test at process design conditions for 120 h. Economic evaluation based on membrane contactor testing data indicates a 56% increase in the LCOE, which is much lower than DOE’s benchmark amine absorption technology (85% increase in LCOE).

Acknowledgments We gratefully acknowledge support by the US Department of Energy (Contract No. DE-FE-0004787) and Illinois Clean Coal Institute (ICCI) and the Department of Commerce and Economic Opportunity’s Office of Coal Development (DCEO/OCD). We thank DOE/NETL Project Manager, Jose´ Figueroa, for his assistance and many valuable discussions. We also thank Dr. Debalina Dasgupta of ICCI for her assistance and insights. We are grateful to other GTI’s Gas Processing Group members Eduardo Tolentino, Osman Akpolat, and Timothy Tamale for their help in constructing and operating the testing unit.

References [1] US Environmental Protection Agency, Climate Change—Greenhouse Gas Emissions, US Greenhouse Gas Inventory Report, Inventory of US Greenhouse Gas Emissions and Sinks: 1990–2007, April, 2009. [2] /http://www.whitehouse.gov/the-press-office/president-attend-copenhagenclimate-talksS. [3] T. Guenther, Power Plant CO2 Capture Overview, LRGCC, March 2008.

[4] M. Ramezan, et al., Carbon Dioxide Capture from Existing Coal-Fired Power Plants, Final Report, DOE/NETL-401/110907, Nov. 2007. [5] NETL, Existing Plants—Emissions and Capture Program Goals, 2009, US Department of Energy. [6] E. Favre, Carbon dioxide recovery from post-combustion processes: can gas permeation membranes compete with absorption? J. Membr. Sci. 294 (2007) 50–59. [7] T.C. Merkel, H.Q. Lin, X.T. Wei, R. Baker, Power plant post-combustion carbon dioxide capture: an opportunity for membranes, J. Membr. Sci. 359 (2010) 126–139. [8] Y. Yampolskii, et al., Materials Science of Membranes for Gas and Vapor Separation, John Wiley and Sons, 2006. [9] R.W. Baker, Membrane Technology and Applications, second ed., John Wiley & Sons Ltd., Chichester, England, 2004, p. 544. [10] S. Khaisri, D. deMontigny, P. Tontiwachwuthikul, R. Jiraratananon, Comparing membrane resistance and absorption performance of three different membranes in a gas absorption membrane contactor, Sep. Purif. Technol. 65 (2009) 290–297. [11] D. deMontigny, P. Tontiwachwuthikul, A. Chakma, Using polypropylene and polytetrafluoroethylene membranes in a membrane contactor for CO2 absorption, J. Membr. Sci. 277 (2006) 99–107. [12] N. Nishikawa, M. Ishibashi, H. Ohta, N. Akutsu, H. Matsumoto, T. Kamata, H. Kitamura, CO2 removal by hollow fiber gas–liquid contactor, Energy Convers. Manage. 36 (1995) 415–418. [13] D. deMontigny, P. Tontiwachwuthikul, A. Chakma, Comparing the absorption performance of packed columns and membrane contactors, Ind. Eng. Chem. Res. 44 (2005) 5726–5732. [14] J.L. Li, B.H. Chen, Review of CO2 absorption using chemical solvents in hollow fiber membrane contactors, Sep. Purif. Technol. 41 (2005) 109–122. [15] B.W. Reed, et al., Membrane Contactors, in: R.D. Noble, S.A. Stern (Eds.), Membrane Separations Technology: Principles and Applications, Elsevier Science B.V., 1995, p. 478, Chapter 10. [16] R.P. Lively, M. Dose, L. Xu, J.R. Johnson, K. Zhang, J. Vaughn, M.E. Lydon, J.A. Thompson, J.S. Lee, L. Liu, Z. Hu, M. Realff, W.J. Koros, A. High-Flux Polyimide, As a Contender for CO2 Recovery From Flue Gas, 2012 North American Membrane Society Meeting, New Orleans, LA, USA, 2012, June 9–13. [17] T.E. Attwood, P.C. Dawson, J.L. Freeman, L.R.J. Hoy, J.B. Rose, P.A. Staniland, Polymer 22 (1981) 1096–1103. [18] Y. Ding, B. Bikson, Preparation and characterization of semi-crystalline poly(ether ether ketone) hollow fiber membranes, J. Membr. Sci. 357 (2010) 192–198. [19] Y. Yuan, Y. Ding, Functionalized Porous Poly(aryl ether ketone) Materials and Their Use, US Patent 7,176,273, Feb 13, 2007. [20] L. Feng, S. Li, Y. Li, H. Li, L. Zhang, J. Zhai, Y. Song, B. Liu, L. Jiang, D. Zhu, Superhydrophobic surfaces: from natural to artificial, Adv. Mater. 14 (2002) 1857. [21] P.T. Nguyena, E. Lasseuguette, Y. Medina-Gonzalez, J.C. Remigy, D. Roizarda, E. Favrea, A dense membrane contactor for intensified CO2 gas/liquid absorption in post-combustion capture, J. Membr. Sci. 377 (2011) 261–272. [22] S. Yoon, S. Lee, I. Yeom, Experimental verification of pressure drop models in hollow fiber membrane, J. Membr. Sci. 310 (2008) 7–12. [23] DOE/NETL-2010/1397Cost and Performance Baseline for Fossil Energy Plants, Volume 1: Bituminous Coal and Natural Gas to Electricity Final Report (Revision 2, November 2010), p. 375. [24] A.L. Kohl, R.B., Nielsen, Gas Purification, fifth ed., Gulf Professional Publishing, p. 340. [25] G.N. Choi, cost efficient amine plant design for post combustion CO2 capture from power plant flue gas, DOE/NETL In: 3rd Annual Conference on Carbon Capture & Sequestration, Alexandra, Virginia, May 3–6, 2004.