International Journal of Greenhouse Gas Control 39 (2015) 194–204

Contents lists available at ScienceDirect

International Journal of Greenhouse Gas Control journal homepage: www.elsevier.com/locate/ijggc

Investigating the influence of the pressure distribution in a membrane module on the cascaded membrane system for post-combustion capture Torsten Brinkmann a , Jan Pohlmann a , Martin Bram b , Li Zhao b,∗ , Akos Tota c , Natividad Jordan Escalona d , Marijke de Graaff e , Detlef Stolten b,f a

Institute of Polymer Research, Helmholtz-Zentrum Geesthacht, Max-Planck- Str. 1, D-21502 Geesthacht, Germany Institute of Energy and Climate Research, Forschungszentrum Jülich, D-52425 Jülich, Germany c Linde Engineering, D-82049 Pullach, Germany d RWE Power Aktiengesellschaft, D-45128 Essen, Germany e EnBW Energie Baden-Württemberg AG, Durlacher Allee 93, D-76131 Karlsruhe, Germany f Chair for Fuel Cells, RWTH Aachen University, D-52056 Aachen, Germany b

a r t i c l e

i n f o

Article history: Received 24 July 2014 Accepted 12 March 2015 Keywords: Carbon capture Gas separation Membrane module Pressure drop Efficiency loss Post-combustion

a b s t r a c t Polyactive® membranes show promising properties for CO2 separation from flue gas. An investigation of different module types using Polyactive® membranes was carried out for this paper. A test rig was built to explore, amongst other process parameters, the pressure drop in envelope-type membrane modules. The experimental data and simulation results were compared with quite good consistency. This validation enabled further simulations for different modules in a virtual pilot plant configuration. Applying the data from the pilot plant simulation to a reference power plant, the scaled-up cascaded membrane system was analyzed using different membrane modules. Considering the required membrane area, energy consumption and pressure drop in different modules, a counter-current membrane module configuration exhibited the best performance and had a marginal advantage in comparison with the chemical absorption process. © 2015 Elsevier Ltd. All rights reserved.

1. Introduction Energy-related CO2 emissions reached a record 31.2 gigatonnes in 2011, representing by far the largest source (around 60%) of global greenhouse-gas emissions measured on a CO2 -equivalent basis (World Energy Outlook, 2012). An update released by the World Bank warns about the potentially disastrous consequences that an increase of four degrees Celsius in the global temperature could have by 2100 (World Could Be 4 Degrees Hotter By End of This Century, 2013). Forecasts by the IEA and others show that “decarbonizing” electricity and enhancing end-use efficiency could make major contributions to the fight against climate change (Climate Electricity Annual, 2011). In spite of increased energy efficiency, the electricity demand is projected to increase substantially – by up to 50% between today and 2050. Renewable energy systems (RES) will generate at least 40% of the electricity required to meet this demand, and the rest will be generated by nuclear sources and

∗ Corresponding author. Tel.: +49 2461 614064; fax: +49 2461 616695. E-mail address: [email protected] (L. Zhao). http://dx.doi.org/10.1016/j.ijggc.2015.03.010 1750-5836/© 2015 Elsevier Ltd. All rights reserved.

fossils with carbon capture and storage (CCS) (Decarbonizing the European Electric Power Sector by 2050). CCS is a series of technologies and applications which capture CO2 from large point sources, transport it via pipelines and ships and safely store it in geological formations, such as saline aquifers and depleted oil and gas fields (Metz et al., 2005a). The principle of CCS is clear: continue using fossil fuels and capture and store the released CO2 underground. However, the technology is currently still being developed and has yet to be demonstrated as feasible on a large scale at acceptable cost. Apart from high investment costs, a high CO2 price or regulations will be required to encourage actual use of CCS, as a significant share of the power generated by a CCS plant is needed to drive its gas separation units, thereby lowering the plant’s net efficiency and flexibility (Decarbonizing the European Electric Power Sector by 2050). In different regions in the world, post-combustion, pre-combustion and oxy-fuel combustion processes are considered options for CO2 capture in the large-scale demonstration of CCS in the power generation sector. To date, no individual capture route or technology can claim a general competitive advantage over other processes (Climate Electricity Annual, 2011).

T. Brinkmann et al. / International Journal of Greenhouse Gas Control 39 (2015) 194–204

195

Fig. 1. Types of investigated membrane modules.

The competing technologies for post-combustion carbon capture are absorption, adsorption and membrane methods (Metz et al., 2005a; Post-Combustion CO2 Control U.S. Department of Energy, 2014). As the first-generation technology for CO2 capture, amine absorption is a mature and proven purification technique that is widely employed in the industrial treatment of acid gases (Kohl and Nielsen, 1997). Nevertheless, the high energy consumption of the absorbent (monoethanolamine, MEA) regeneration step with efficiency losses of 10–14% points and corrosion problems associated with solvent degradation increase the operation and maintenance costs of this technology (Wang et al., 2011; Luis et al., 2012; Mangalapally et al., 2012; Svendsen et al., 2011; Blomen et al., 2009; Galindo-Cifre et al., 2009). Gas separation membrane technologies, a potential second-generation technology for postcombustion capture, are gaining more and more attention. The advantages of these technologies are their potentially lower environmental impact and the fact that membrane modules can be used as add-on equipment requiring with fewer modifications to power plants. The other potential advantage is that for low degrees of CO2 separation, a membrane array demands a lower specific energy than that required for MEA absorption. Furthermore, membrane systems are easier to scale-up and more suitable for intermittent, dynamic operation. Membrane science and technology can be divided into two classes, namely materials research and process engineering. Many groups and researchers worldwide have been involved in materials and process development (Zhao et al., 2010; Bounaceur et al., 2006; Favre, 2007; Car et al., 2008a; Follmann et al., 2011; Ho et al., 2008; Deng et al., 2009; Hussain and Hägg, 2010; Merkel et al., 2010; Brinkmann et al., 2011; Kai et al., 2008; Powell and Qiao, 2006; Lin and Freeman, 2005; Reijerkerk et al., 2010; Brunetti et al., 2010; Bram et al., 2011), with some institutions covering the entire research and development chain from material synthesis to process engineering (Hussain and

Hägg, 2010; Merkel et al., 2010; Abetz et al., 2006; Sijbesma et al., 2008). Important progress has been achieved and relevant experience obtained in the past by testing membrane modules in real flue gas environments. In Europe, under the framework of the projects MemBrain (MEM-BRAIN Alliance, 2011), METPORE (METPORE, 2014), Nanoglowa (CO2 Capture Using Membrane Technology, 2015) and iCap (The iCap project), different polymer and ceramic membranes are being investigated to meet the harsh requirements in coal-fired power plants (Car et al., 2008a; Hussain and Hägg, 2010; Brinkmann et al., 2011; Reijerkerk et al., 2010; Bram et al., 2011). Membrane modules equipped with Polyactive® thin-film composite membranes (Car et al., 2008c; Brinkmann et al., 2013; Brinkmann et al., 2012; Brinkmann et al., 2010) in a parallel configuration (12.5 m2 and 1 m2 ) are currently being tested in the EnBW power plant Rheinhafen-Dampfkraftwerk (Bram et al., 2011; METPORE, 2014). A 1 MWel pilot-scale Polaris® membrane separation system at the Department of Energy’s National Carbon Capture Center in Wilsonville, USA, was announced by the National Energy Technology Laboratory (NETL) in November 2012. It will test a postcombustion membrane capture technology on the largest scale in the world to date (Merkel et al., 2010; NETL Greenlights 1 MW Field Test For Membrane Capture Tech, 2012). In order to realize the potential of gas permeation for industrial applications, advanced membrane module concepts are desirable. Membranes can be inserted into three major types of modules for gas separation applications: envelope-type, spiral wound, and hollow fiber modules (Favre, 2010; Melin and Rautenbach, 2003; Baker, 2012; Ohlrogge and Wind Brinkmann, 2010). In general, the membrane module must be used in commercial processes as a package with as much surface area per unit volume as possible, good flow distribution and efficient contact of the feed gas within the membrane. The main flow configurations in membrane

196

T. Brinkmann et al. / International Journal of Greenhouse Gas Control 39 (2015) 194–204

Table 1 Packing density range of different membrane module forms (Favre E., 2010). Module form

PD [m2 /m3 ]

Hollow fiber Spiral wound Envelope module

2000–5000 700–1000 500–900

gas permeation processes are cross-plug flow, co-current flow and counter-current flow (Melin and Rautenbach, 2003). More specifically, hollow fiber modules may be designed with reasonable precision to approximate an idealized counter-current when the permeate pressure is significant, or to approximate a cross-flow pattern when the permeate pressure is low enough (Favre, 2010). The most commonly used flat sheet membrane module for gas separation is the spiral wound module, which is dominated by cross flow (Merkel et al., 2010). Within an envelope-type module, the first half of a membrane envelope can be assumed to be in co-current flow configuration, whilst the second half is in counter-current flow configuration (Brinkmann et al., 2013). The packing density (PD) of each type of membrane, which is the surface area per volume [m2 /m3 ], is shown in Table 1. Furthermore, the flow patterns do not only influence the concentration distribution on the feed and permeate sides of the module but also give rise to different behavior in terms of pressure drops (Melin and Rautenbach, 2003; Brinkmann, 2006) and hence the utilization of the available driving force. Although hollow fiber modules appear to be superior in terms of membrane area utilization, they have substantial drawbacks when high-flux membrane materials are employed in a thin-film composite configuration. In order to transfer the attractive CO2 /N2 selectivities at 20 ◦ C of approx. 60 of poly(ethylene oxide) block copolymers as Polyactive® or Pebax® 1657 (Car et al., 2008c,b) into a membrane module, whilst also achieving high CO2 permeances in the order of 3 Nm3 / (m2 h bar) at 20 ◦ C, ultrathin layers with a thickness in the order of 70 nm are required. This ambitious goal can be met by manufacturing flat sheet membranes on a 100 m2 scale (Brinkmann et al., 2012). To the authors’ knowledge, no comparable coating technology is available for hollow fiber membranes. In the present work, different flow patterns will be investigated in flat sheet membrane modules within the framework of the METPORE II project (METPORE, 2014; Brinkmann et al., 2013). The envelope-type modules used in our experiments were developed by Helmholtz-Zentrum Geesthacht (HZG) (Brinkmann et al., 2013; Baker, 2012; Ohlrogge et al., 2006). They were tested both on a pilot plant scale and in the flue gas of the power plant Rheinhafen-Dampfkraftwerk (EnBW). The pressure drop in the modules was investigated. The relevant results were integrated in a total system analysis using flue gas data from the Reference Power Plant North Rhine-Westphalia (RKW-NRW) (Konzeptstudie: Referenzkraftwerk Nordrhein-Westfalen (RKW NRW), 2004). The equation-oriented process simulator Aspen Custom Modeler®1 was used for membrane module simulation. The cascaded membrane process was investigated by Aspen Process Modeling V8.0. 2. Investigation of pressure distribution in a flat sheet membrane module (Brinkmann et al., 2013) 2.1. Introduction of membrane material and module The membrane considered in this study is a multilayer, thin-film composite membrane specifically developed for the separation of CO2 from N2 . Details on this membrane can be found in (Car et al.,

1

http://www.aspentech.com/products/aspen-custom-modeler.aspx, last access on 27th June 2014.

Table 2 Free-Volume parameters for Polyactive® composite membrane (Brinkmann et al., 2013).

CO2 N2 O2

L0 ∞ [Nm3 /(m2 h bar)]

E [kJ/kmol]

m0 [1/bar]

mT [1/K]

 [Å]

1084.66 68646.60 39776.50

14279.81 34251.30 30463.00

0.1568 0.0000 0.0000

-0.0054 0.0000 0.0000

3.941 3.798 3.467

2008c; Brinkmann et al., 2013, 2012, 2010; Metz et al., 2005b). The 70 nm thin separation layer consists of the polyethylene /polybutylteraphtalate block copolymer Polyactive® . The active layer is sandwiched between two highly permeable polydimethylsiloxane layers cast in turn onto a porous support structure consisting of polyacrylonitrile and polyester. The CO2 permeance and the CO2 /N2 selectivity of this membrane are similar to the published values for the MTR Polaris® membrane (Lin et al., 2014). The Polyactive® membrane used here has been extensively tested for applications as diverse as biogas processing (Brinkmann et al., 2012, 2010), separation of CO2 from reaction products (Stünkel et al., 2011; Song et al., 2013) and flue gas treatment (Brinkmann et al., 2013, 2012). The relevant permeation properties are given in Table 2, in which L0 ∞ is the permeance for T → ∞ and p → 0, E is the activation energy, m0 and mT are free-volume model parameters, and  is the Lennard-Jones molecule diameter. Only CO2 , N2 and O2 are listed. The permeance of Ar was assumed to be identical to that of O2 whilst that of H2 O was assumed to be ten times the value of that for CO2 . The parameters are given for the free-volume model (Brinkmann et al., 2013; Fang et al., 1975; Alpers, 1997) which has been employed to express multicomponent permeation behavior in modeling. The parameters were determined using single-gas measurements employing the pressure increase method. The membrane was produced on a 100 m2 scale and was installed in different module types, mostly into envelope-type modules. These modules had a maximum capacity of 75 m2 with a length of 1250 mm and a diameter of 310 mm, depending on envelope thickness. The membrane material was manufactured into membrane envelopes consisting of two sheets of membrane material with their separation layer on the outside and a permeate spacer between the sheets. They were thermally welded at the outer circumference. The permeate was withdrawn radially toward a central hole. The membrane envelopes were stacked on a perforated permeate tube and subdivided into compartments, thus allowing the cross-sectional area to be adjusted depending on the amount of permeate withdrawn. Hence, a nearly constant flow velocity on the feed side could be realized. More details on this module type can be found in (Brinkmann et al., 2013; Ohlrogge and Wind Brinkmann, 2010; Ohlrogge et al., 2006). The spiral wound membrane module is widely used in membrane technology, and its design features are outlined in (Baker, 2012). A novel concept for flat sheet membranes is closely related to the envelope type (Brinkmann et al., 2011; Brinkmann et al., 2013, 2012) but the envelopes are rectangular and are housed in a module with a rectangular cross section. The design allows for different points of permeate withdrawal so that different flow patterns can be realized, i.e. co- and counter-current design. Furthermore, it is possible to divide the envelope into segments, thus realizing different permeate withdrawal points along the flow path from the feed to the retentate, hence minimizing the pressure drop on the permeate side. Fig. 1 illustrates the investigated module types. 2.2. Test equipment The envelope-type module has been extensively studied in pilot plant investigations for the separation of CO2 from various gas streams using Polyactive® membranes (Brinkmann et al., 2011,

T. Brinkmann et al. / International Journal of Greenhouse Gas Control 39 (2015) 194–204

197

Fig. 2. Gas permeation pilot plant for power plant flue gas installed at EnBW Rheinhafen-Dampfkraftwerk, Karlsruhe, Germany.

Table 3 Experimental conditions of the pilot plant experiments using synthetic gas mixtures and experimentally determined as well as simulated retentate pressures at a temperature of 20 ◦ C for an envelope type membrane module containing 9.52 m2 of Polyactive® membrane (Brinkmann et al., 2013). Experiment no.

1 2 3 4 5 6 7 8

Feed volumetric flowrate [Nm3 /h]

Feed pressure [bar]

Permeate pressure [bar]

Feed CO2 mole fraction [–]

Retentate Pressure [bar] Experiment

Simulation

43.64 44.41 36.00 26.31 34.24 33.64 20.07 42.61

4.41 4.48 4.38 4.32 2.55 2.49 2.52 4.44

0.20087 0.111 0.099 0.095 0.201 0.100 0.101 0.200

0.181 0.177 0.180 0.182 0.174 0.176 0.182 0.167

4.34 4.41 4.33 4.29 2.45 2.40 2.48 4.36

4.33 4.40 4.32 4.28 2.45 2.39 2.48 4.36

2013, 2012, 2010). It is employed in numerous industrial applications in the chemical and petrochemical industries (Ohlrogge and Wind Brinkmann, 2010; Ohlrogge et al., 2006, 2005). The analysis of different module types presented in (Brinkmann et al., 2013) was constrained to two applications. In addition to synthetic gas mixtures, a bypass stream from the Rheinhafen-Dampfkraftwerk was supplied to a dedicated pilot plant within the scope of the project METPORE II, which was funded by the German Ministry of Economics and Energy. A simplified flow sheet is shown in Fig. 2. A separate publication describing its design and the results achieved in detail is in preparation (Pohlmann and Brinkmann, 2014). 2.3. Modeling validation The permeation behavior of multicomponent mixtures through Polyactive® multilayer composite membranes can be described by the free-volume model using the parameters given in Table 2. In (Brinkmann et al., 2013; Brinkmann, 2006), it was shown that the combination of this model with an appropriate model describing the membrane module flow patterns can very accurately predict the separation performance of envelope-type modules. It was assumed that this is also true for the other membrane module types considered here. However, it should be emphasized that this has yet to be experimentally verified. The details of the models employed here can also be found in (Brinkmann et al., 2013). They were implemented in the equation-oriented process simulator Aspen Custom Modeler (ACM).1 The operating conditions investigated in the pilot plant as well as the experimental and simulation results for the retentate pressure are summarized in Table 3. It is apparent that the

simulation results closely reflect the experimental values. The prediction of the separation performance is equally good (Brinkmann et al., 2013). The relevant results for pressure drops employing real flue gas at the EnBW Rheinhafen-Dampfkraftwerk are shown in Table 4. The accuracy of the model predictions allows the developed models to be applied for the estimation of the required compression energies. 2.4. Simulation of different module types in a virtual pilot plant configuration Fig. 3 illustrates a possible pilot plant system. A feed flow of 1000 Nm3 /h was assumed to enable the investigation of membrane modules commonly employed in today’s gas permeation installations. The blower C1 was used to compensate the pressure drops of the membrane modules installed in the first stage of the process, i.e., the retentate pressure was set to atmospheric pressure and the outlet pressure of C1 was adjusted according to the pressure drop. The simulation assumed adiabatic compression with an efficiency of 70%. The cooler H1 was used to cool the feed gas to 25 ◦ C and to separate condensed water whilst the heater H2 increased the feed temperature above the dew point in order to prevent condensation in the downstream system. The membrane modules in MemStage1 separated CO2 from the feed gas according to the set recovery target, i.e., the CO2 separation degree, which was assumed to be 50% for this investigation. The vacuum pump C2 compressed the permeate of the first stage. Its suction pressure was set to 100 mbar. Since water preferentially permeates the membrane, i.e., even more strongly than CO2 , the separator S1 was employed to draw off the

198

T. Brinkmann et al. / International Journal of Greenhouse Gas Control 39 (2015) 194–204

Table 4 Experimental conditions of the pilot plant experiments using flue gas and experimentally determined as well as simulated retentate pressures at temperatures of 24–28 ◦ C for an envelope type membrane module containing 12.5 m2 of Polyactive® membrane. Experiment no.

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22

Feed volumetric flowrate [Nm3 /h]

Feed pressure [bar]

Permeate pressure [bar]

Feed CO2 mole fraction [–]

Retentate Pressure [bar] Experiment

Simulation

50.00 50.30 60.20 69.90 80.00 50.30 59.70 65.00 69.60 70.40 79.60 80.10 80.10 80.30 80.60 60.30 79.70 80.00 80.30 60.00 70.00 80.00

1.13 1.14 1.18 1.24 1.31 1.14 1.19 1.22 1.23 1.25 1.29 1.31 1.31 1.30 1.31 1.20 1.32 1.31 1.32 1.20 1.26 1.32

0.060 0.062 0.063 0.057 0.062 0.075 0.066 0.075 0.065 0.069 0.067 0.069 0.069 0.067 0.069 0.111 0.116 0.100 0.100 0.150 0.151 0.150

0.135 0.130 0.135 0.135 0.129 0.129 0.131 0.128 0.136 0.129 0.138 0.130 0.135 0.136 0.136 0.128 0.129 0.137 0.132 0.128 0.131 0.128

1.10 1.11 1.14 1.19 1.25 1.11 1.15 1.18 1.18 1.20 1.22 1.25 1.25 1.24 1.25 1.16 1.25 1.25 1.25 1.16 1.20 1.26

1.09 1.11 1.13 1.19 1.25 1.11 1.15 1.17 1.18 1.19 1.23 1.25 1.25 1.24 1.25 1.15 1.26 1.25 1.26 1.16 1.21 1.26

Fig. 3. Pilot plant flow sheet.

liquid water. The vacuum pump was assumed to operate isothermally with an efficiency of 40%, as can be realized by liquid ring vacuum pumps. In operation, this lower efficiency was offset by a simple, robust design, the partial condensation of permeated water in the service liquid circuit and a lower demand of cooling utilities since the heat of compression was transferred to the service liquid which was cooled down in a straightforward fashion. The adiabatic compressor C3 increased the pressure to 6 bar, i.e., the operating pressure of the second stage, at an efficiency of 70%. The heat exchanger H3 cooled down the gas and removed the liquefied water. The membrane stage MemStage2 was employed to further increase the CO2 concentration of the permeate stream of MemStage1 to the required purity, i.e., 95 vol%, in the permeate of the second membrane stage at a pressure of 1.0130 bar. In order to achieve the required recovery, the retentate was recycled upstream of H1. Part of the compression energy was recovered by the turbo expander T1 (efficiency 70%). For comparison, the required membrane areas were calculated for the described separation task (i.e., a CO2 purity of 95 vol% at a CO2 separation degree of 50 %). It was assumed that the membrane stages could be described as cross-flow stages with unhindered permeate withdrawal and no pressure drops on the feed or permeate side as well as no concentration polarization. The dependency of the

permeances on temperature, pressure and composition was considered as well as real gas behavior for the calculation of the driving forces and non-isothermal operation due to the Joule–Thomson effect. The latter two phenomena were of minor effect, especially in the first stage, due to the low pressures and the small stage cut. The simulation rendered area requirements of 300 m2 for stage 1 and 8.2 m2 for stage 2. Using these area requirements, flat sheet membrane modules were dimensioned for both stages. The following module types were considered (see also Fig. 1):

• Envelope-type membrane module. • Spiral wound membrane module. • Counter-current membrane module concept with one segment. The permeate was led counter-currently to the feed flow and withdrawn at the feed side of the module. • Co-current membrane module concept with one segment where the permeate was withdrawn co-currently with the feed at the retentate location. • Counter-current membrane module concept with four segments. The module was divided into four sections of equal length, and at the start of each segment, permeate was withdrawn at the location of the feed.

T. Brinkmann et al. / International Journal of Greenhouse Gas Control 39 (2015) 194–204

199

Table 5 Membrane module details stage 1.

Membrane area per module A [m2 ] Number of modules in parallel Module diameter D [m] Number of envelopes Compartments Envelope breadth b [m] Envelope length l [m] Feed channel height hR [mm] Permeate channel height hP [mm]

Envelope

Spiral wound

Counter-current

Co-current

Counter-current 4 segments

60.21 5 0.310 506 9 – – 1.500 1.500

20.11 8 and 7 0.203 11 – 0.914 1.000 1.500 1.500

300.72 1 – 127 – 0.600 1.973 1.500 1.500

300.72 1 – 127 – 0.600 1.973 1.500 1.500

300.72 1 – 127 – 0.600 4 × 0.493 1.500 1.500

Envelope

Spiral wound

Counter-current

Co-current

Counter-current 4 segments

8.33 1 0.310 70 22 – – 1.500 1.500

8.33 1 0.159 9 – 0.693 0.668 1.500 1.500

8.21 1 – 5 – 0.300 2.736 1.500 1.500

8.21 1 – 5 – 0.300 2.736 1.500 1.500

8.21 1 – 5 – 0.300 4 × 0.684 1.500 1.500

Table 6 Membrane module details stage 2.

Membrane area A [m2 ] Number of modules in parallel Module diameter D [m] Number of envelopes Compartments Envelope breadth b [m] Envelope length l [m] Feed channel height hR [mm] Permeate channel height hP [mm]

In addition to the previously described phenomena, pressure drops were considered on the feed and permeate sides, as well as concentration polarization on the feed side. Details on the simulation models employed, their implementation in Aspen Custom Modeler®1 and the employed geometrical values can be found elsewhere (Brinkmann et al., 2013). Tables 5 and 6 show the module details required for the simulation of stages 1 and 2, respectively. All membrane modules are assumed to be equipped with a Polyactive® membrane (Car et al., 2008c; Brinkmann et al., 2013, 2012). For the co- and counter-current membrane module concepts, one membrane module can be used for each of the stages, since this concept is quite versatile in terms of adjusting it for different flow rates. However, an important point to mention is that the simulation predictions for this module type have not yet been validated experimentally. For the envelope-type and spiral wound modules, it was assumed that five or eight modules were mounted in parallel in the first stage, respectively. In the case of the spiral wound modules, a second sequential set of seven parallel modules was required to realize the required membrane area. For the second stage, one membrane module was sufficient for both envelope and spiral wound module types. The results of the simulations are summarized in Table 7 and Figs. 4–6. Since stage 1 governs the process, the discussion will focus on this stage. The differences in pressure drops for the investigated modules are rather small, as shown in Fig. 4. It is apparent that both the non-segmented counter-current and the co-current modules have the highest pressure drops on the permeate side and low pressure drops on the feed side. The latter is also true for the four-segment counter-current concept. However, due to the segmentation, low pressure drops can also occur on the permeate side. This causes a superior pressure ratio (Fig. 5) and a good CO2 driving force (Fig. 6). The pressure on the permeate side of the spiral wound modules is quite high. This is due to the long permeate pathways. High pressure drops also cause the high pressures on the feed side. In combination, these factors lead to a good pressure ratio and a good driving force (Figs. 5 and 6). The module with the best fluid dynamics is the envelope-type module. It has very short permeate pathways resulting in small pressure drops on the permeate side (Fig. 4). Using the calculated performance data for the different modules to compare CO2 purities and recoveries, as well as the required power for the blower C1, the vacuum pump C2 and the

Fig. 4. Pressures on the retentate/feed side and permeate side in the different module types.

Fig. 5. Pressure ratios of the different module types.

200

T. Brinkmann et al. / International Journal of Greenhouse Gas Control 39 (2015) 194–204

Table 7 Simulation results.

CO2 -purity [kmol/kmol] CO2 -recovery [–] Membrane area stage 1 [m2 ] Membrane area stage 2 [m2 ] Total membrane area [m2 ] Power blower C1 [kW] Power vacuum pump C2 [kW] Power compressor C3 [kW] Power turbine T1 [kW] Total power [kW] Total cooling duty [kW] Pressure drop stage 1 retentate [bar] Max. pressure drop stage 1 permeate [bar] Pressure drop stage 2 retentate [bar] Max. pressure drop stage 2 permeate [bar] Permeate flowrate stage 1 [Nm3 /h]

Envelope Type

Spiral Wound

Counter-Current 1 Segment

Co-Current

Counter-Current 4 Segments

0.944 0.451 301.070 8.330 309.400 7.991 17.674 9.016 −0.933 33.748 −137.607 0.092 0.0002 0.340 0.0015 97.901

0.943 0.451 301.620 8.330 309.950 9.260 17.875 9.177 −0.985 35.326 −139.318 0.120 0.0088 0.005 0.0318 98.990

0.946 0.431 300.716 8.207 308.923 7.260 17.782 8.769 −0.952 32.859 −136.351 0.076 0.0118 0.088 0.1638 98.500

0.932 0.401 300.716 8.207 308.923 7.296 15.902 8.320 −0.928 30.592 −135.971 0.077 0.0152 0.084 0.2545 88.040

0.945 0.455 300.716 8.207 308.923 7.262 17.751 8.940 −0.927 33.026 −136.623 0.076 0.0008 0.089 0.0152 98.300

lar behavior. It was employed to recompress the permeate from stage 1 before feeding it to stage 2. Since envelope-type and spiral wound modules had the highest pressure drops on the feed side¸ the blower powers were also higher for these modules with consumption being the highest for the spiral wound module. This was due to the high velocity on the feed side, i.e., 2.5 m/s, compared to ca. 2 m/s for the other modules. This high velocity resulted from distributing modules of a given area and geometry to achieve the required membrane area. It can be concluded that the segmented counter-current membrane module appears to be the best choice in terms of CO2 purity and recovery, as well as energy consumption of the rotating equipment.

3. Power-plant data used for simulation

Fig. 6. CO2 driving forces of the different module types.

compressor C3, it is apparent that the segmented counter-current module renders the highest CO2 purity with the highest recovery. The envelope-type and spiral wound modules are close contenders. Considering the one-segment co- and counter-current modules, the situation regarding CO2 recovery is more pronounced: recovery is strongly decreased, especially for the co-current module. This is due to the inferior pressure ratio and driving force distribution in these module types (Figs. 5 and 6). In terms of compression power consumption, the highest demand is caused by the vacuum pump C2 followed by the compressor C3 and the blower C1. The lowest power demand for the vacuum pump was observed for the co-current module, which had the lowest permeate flow rate due to its inferior driving force utilization. In other words, the vacuum pump power follows the permeate flow rate to be withdrawn from stage 1. The power demand of the compressor C3 showed simi-

A scaled-up cascaded membrane system, shown in Fig. 7, was used for the process analysis. The first membrane (Mem 1) was driven by a vacuum pump and the second membrane (Mem 2) by a compressor. The retentate of the second membrane was recycled as the feed of the first membrane. Part of the compression energy was recovered by a turbo expander. The membrane cascade was located downstream of the SCR-DeNOx, dust removal (E-filter) and desulphurization (FGD) processes and prior to the cooling tower. Here, the flue gas had a pressure of approximately 1 atm and a temperature of 50–70 ◦ C. The basic data of RKW-NRW and the flue gas data are listed in Table 8. The flue gas data were obtained by simulation using Klein Kopje hard coal. The residue of pollutants in the flue gas consisted of approximately 50 vppm SO2 and approximately 200 vppm NO2 . The temperature, H2 O permeance and relative humidity had an evident influence on the CO2 selectivity of polymer membranes (Low et al., 2013). In order to develop a feasible membrane system, these influence factors must be considered.

Fig. 7. Schematic illustration of the scaled-up cascaded membrane system.

T. Brinkmann et al. / International Journal of Greenhouse Gas Control 39 (2015) 194–204

201

A wet scrubber was adopted upstream of the membrane system, referring to a post-combustion process presented by RWE (Peters and Wallus, 2012). This scrubber had two functions: the first was to further reduce the pollutants (SOx, NOx and dust) in the flue gas, e.g., SOx can be decreased to 10 vppm; the second was to cool the flue gas down further to 25–30 ◦ C. The latter function is a crucial factor for the membrane process, because the working temperature of the Polyactive® membrane influences its performance strongly (Zhao et al., 2013). Furthermore, applying a dewatering process prior to CO2 separation leads to a lower energy consumption of the whole system (Zhao et al., 2013). Here, 2/3 water content of the flue gas was removed.

4. Integrating the module data with the scaled-up cascaded membrane system 4.1. Simulation of single-stage system In order to investigate the influence of flow patterns on membrane separation performance, a case study was carried out. Two different cases were explored with a feed flow rate of 100 kmol/h, composition of 14.91 mol% CO2 , 0.59 mol% H2 O and 84.50 mol% N2 , membrane CO2 permeance of 3 Nm3 /(m2 h bar), CO2 /N2 selectivity of 50, H2 O permeance of 30 Nm3 /(m2 h bar). Case 1 had a feed pressure of 4 bar and a permeate pressure of 0.1 bar; case 2 had a feed pressure of 1 bar and a permeate pressure of 0.1 bar. The results are shown in Fig. 8(a) and (b), respectively. The working temperature of the membranes was 25 ◦ C. Comparing these two diagrams, it is apparent that using both compressor and vacuum pump for a membrane leads to a smaller membrane area. In case 1, co-current flow had a slightly better CO2 purity than cross- and counter-flow at a low CO2 separation degree (