Improved Performance of a PBM Reactor for Simultaneous CO2

Nov 24, 2014 - Department of Chemical and Biomolecular Engineering, University of Cantabria, Av. de los Castros s/n, 39005 Santander, Spain. ‡...
7 downloads 0 Views 1MB Size
Article pubs.acs.org/IECR

Improved Performance of a PBM Reactor for Simultaneous CO2 Capture and DME Synthesis Nazely Diban,*,† Ane M. Urtiaga,† Inmaculada Ortiz,† Javier Ereña,‡ Javier Bilbao,‡ and Andrés T. Aguayo‡ †

Department of Chemical and Biomolecular Engineering, University of Cantabria, Av. de los Castros s/n, 39005 Santander, Spain Department of Chemical Engineering, University of Basque Country, Apdo. 644, E-48080 Bilbao, Spain



ABSTRACT: The use of “nonideal” zeolite membranes for the in situ H2O removal in a packed-bed membrane reactor (PBMR) during the synthesis of dimethyl ether (DME) allows the recovery of CO2 but unexpectedly reduces DME yield by 50% in comparison to a packed-bed reactor (PBR) as previously reported [Diban et al. Chem. Eng. J. 2013, 234, 140]. Due to the advantageous performance of PBMR, the present work aims to the theoretical analysis and optimization of the working conditions and system configuration that enhance both DME yield and CO2 recovery. Here, the previously developed mathematical model able to predict the mass transport rate of all the components present in the reactive system through zeolite membranes has been modified and accounts for the sweep gas recirculation. The influence of the sweep gas flow-rate in the range 0.06−1.80 molCOx·h−1 (laboratory scale) and sweep gas recirculation factor (0 < α < 1) has been analyzed. Sweep gas flow-rates >0.18 molCOx·h−1 favored CO2 conversion but only partial recirculation of the sweep gas promoted DME yields beyond those obtained in a PBR due to the synergism between effective H2O removal and MeOH retention in the feed side. Although energetically challenging, these results show promising prospects to apply the existing zeolite membranes for the chemical transformation of CO2 into DME on a large scale.

1. INTRODUCTION The concerns on the global warming are stressing the interest of the scientific community for CO2 capture and sequestration1−3 and the chemical valorization of CO2 into the synthesis of valuable products, e.g., formaldehyde, acetic acid, propylene, methanol (MeOH), or hydrocarbons, have been explored in the literature.4−7 The synthesis of dimethyl ether (DME) is a catalytic process with promising prospects for CO2 valorization on a large scale.6 In addition to being a propellant and coolant, DME has a broad range of applications as an alternative “clean” fuel for diesel engines, a source of H2 for fuel cells and a key intermediate for producing automobile fuels and raw materials, e.g., olefins and BTX aromatics.8−10 The catalytic synthesis of DME is industrially performed in two steps; in a first packed-bed reactor (PBR), syngas is transformed into MeOH using a metallic catalyst and in a subsequent PBR, MeOH is converted into DME on an acidic catalyst. The set of chemical reactions that take place is described below. In the first PBR: hydrogenation of CO

CO + 2H 2 ⇔ CH3OH

water gas shift (WGS) reaction

hydrogenation of CO2

H 2O + CO ⇔ H 2 + CO2

CO2 + 3H 2 ⇔ CH3OH + H 2O

The use of a bifunctional catalyst with metallic and acidic functions allows the synthesis of DME in a single step PBR with the benefit of switching the equilibrium of the synthesis of MeOH (eq 1) toward the dehydration of MeOH into DME (eq 5) and consequently increasing the conversion of CO and CO2 even working at higher temperatures and lower pressures.11 This bifunctional catalyst promotes cofeeding of CO2 with the syngas. The conversion of the reactions of CO2 hydrogenation (eq 3) and MeOH dehydration (eq 5) are limited by the presence of H2O in the reaction site, as previously seen during the synthesis of DME using feed mixtures of CO2+H2 and CO +CO2+H2.12−14 Hence, the in situ H2O removal from the reaction site with a H2O permeable membrane would enhance CO2 conversion and DME yield. This strategy has been experimentally tested for the Fischer−Tropsch (FT) reaction with zeolite membranes (ZSM5, Mordenite (MOR) and Silicalite-1 (SIL)), providing successful results in terms of product yields (Espinoza et al., 1999 and 2000; Rohde et al., 2005, 2006 and 2008; Schaub et al., 2008).15−20 Iliuta et al.21 have theoretically explored the efficiency of a packed-bed membrane reactor (PBMR) configuration in the synthesis of DME considering an ideal H2O selective membrane under different permeance properties. Among the different membrane materials tested in the literature,22 only zeolite membranes would withstand the high demanding operating conditions (225−325 °C, 10−40 bar) required during DME synthesis.

(1) (2) (3)

formulation of C1−C10 paraffins (HC) (nondesired byproducts) nCO + (2n + 1)H 2 → CnH 2n + 2 + nH 2O, (1 ≤ nc ≤ 10)

(4) Received: Revised: Accepted: Published:

In the second PBR: dehydration of MeOH to DME

2CH3OH ⇔ CH3OCH3 + H 2O

(5) © 2014 American Chemical Society

19479

September 16, 2014 November 12, 2014 November 24, 2014 November 24, 2014 dx.doi.org/10.1021/ie503663h | Ind. Eng. Chem. Res. 2014, 53, 19479−19487

Industrial & Engineering Chemistry Research

Article

Table 1. Characteristic Dimensions of the Membrane Reactor, Catalyst Properties and Operational Conditions in the Feed and Sweep Gas Streams Employed in the Simulations and Mass Transport Properties of Zeolite (ZSM5, MOR and SIL) Membranes (250−300 °C)a parameter

value 6.13 × 10−3 10 × 10−3 13 × 10−3 100 0.015 0.5 2000 2.4 × 10−3 1/7 3/1 0.06

2

effective membrane area, Am (m ) internal membrane support diameter (m) external membrane support diameter (m) selective layer thickness (μm) effective length of the fixed-bed (m) voidage of the fixed-bed (−) catalyst density (kg/m3) catalyst mass (kg) catalyst mass dilution (−) b reactants ratio in the feed and sweep gas streams (H2/COx)c b F COx molar flow rate in the feed stream, FCO (molCOx·h−1) x,0 d

COx molar flow rate in the sweep gas stream, FCOx,FSG (molCOx·h−1)

0.06−1.8

feed and sweep gas stream pressure, PF and PS (bar) reactor temperature, T (°C) operation time (h) H2O permeance, 7H2O(mol·s−1·m−2·Pa−1)

40 275 30 6.8 × 10−8

H2O/comp. i selectivity

SH2O/H2

SH2O/COe

SH2O/CO2

SH2O/MeOH

SH2O/DMEf

SH2O/HC

real zeolite membrane ideal zeolite membrane

49 0.5

19.6 ∞

17.7 ∞

2.8 ∞

43.2 ∞

57.3 ∞

a Determination of H2, CO, CO2, H2O, DME and HC properties by Espinoza et al.15,16 in multicomponent mixtures and MeOH properties by Piera et al.25 in binary H2O/MeOH mixtures. bValues at the entrance of the PBMR. These variables change with the position. cCOx refers to the total amount of CO and CO2 being the CO/CO2 molar ratio of 50%. dInitial values of molar flow rate in the sweep stream at the entrance of the PBMR S FCOx,FSG = FCO . eValues only available at 350 °C. fValue of permselectivity of C8 hydrocarbons. It was considered to be similar to DME x,τ=Tao,t=0 properties.

smallest molecules (H2 and H2O) over the rest of molecules present in the system. This type of zeolite membrane was already reported as hydroxy sodalite (H-SOD)19,24 but it presented low hydrothermal stability and thus it was inapplicable in the present system. Meanwhile this ideal and stable zeolite membrane is developed, further assessment of the system aroused the following question: Could the DME yield be enhanced by using an operational approach, that is, the operational variables and the system configuration, with the hydrothermally stable zeolite membranes currently available? The present work aims at going deeper into the evaluation of the present PBMR process and the knowledge of the influence of the flow rate and recirculation of the sweep gas stream on the process performance. The hypotheses to be demonstrated is how these working conditions would minimize the driving force of the mass transport of the reactants and intermediate products from the feed toward the sweep gas side. Additionally, this work intends to evaluate future experimental research priorities based on the application of theoretical modeling tools for the development of an efficient PBMR aimed at the chemical valorization of CO2 into DME.

Particularly, the zeolite membranes applied by Espinoza et al.15,16 under FT conditions for in situ H2O removal were ZSM5, MOR and SIL and showed improved reaction performance. Therefore, we theoretically evaluated in a previous study23 the potential application of these zeolite membranes in a catalytic PBMR for DME synthesis with CO2 cofeeding using a bifunctional catalyst CuO-ZnO-Al2O3/γAl2O3. Interestingly, we observed that the DME yield was importantly reduced when introducing these zeolite membranes to remove H2O from the feed stream. This low DME yield was caused by the low selectivity of the zeolite membranes that allowed not only the permeation of H2O but also of reactants and intermediate products, mainly the intermediate MeOH that could not be further converted into DME. According to these results, it seems that the selectivity of the membrane had a strong influence on the viability of using a PBMR to capture CO2 and to efficiently valorise it into DME under the simulated conditions. These results led to the logical conclusion that research efforts should be directed toward the improvement of the membrane H2O selectivity and to reduce the mass transport of MeOH through the membrane. This would be a challenging defy for zeolite membrane researchers. Attending to the selectivity of the zeolite membranes evaluated by Diban et al.,23 it is clear that their hydrophilic characteristics were determinant in the transport mechanism of the components (MeOH selectivity was similar to that of H2O and higher than the selectivity of smaller molecules such as H2, CO and CO2, see also selectivity values in Table 1). The synthesis of a zeolite membrane with ideal H2O selectivity over MeOH selectivity would require that the mass transport of the molecules through the membrane proceeds according to a sieving mechanism, thus favoring the permeation of the

2. MATHEMATICAL MODEL MODIFICATIONS 2.1. PBMR Working at Variable Gas Velocities. A mathematical model describing the mass transport through a hydrophilic membrane of the components involved in the catalytic synthesis of DME from a feed mixture of CO/CO2/H2 in a PBMR with in situ H2O removal was developed in our previous work.23 Briefly, the mass transport conservation equations for component i along the PBMR, with the axial 19480

dx.doi.org/10.1021/ie503663h | Ind. Eng. Chem. Res. 2014, 53, 19479−19487

Industrial & Engineering Chemistry Research

Article

position expressed in terms of space time, τ, are (I) in the feed side F

FCOx ,0 dFiF(τ ) F + ji (τ ) ·A m · + ri(τ ) ·FCO =0 x ,0 dτ Wcat

∀ i; ∀ τ (6)

(II) in the permeate side F

FCOx ,0 dFiS(τ ) − ji (τ ) ·A m · =0 dτ Wcat

∀ i; ∀ τ

(7)

And the mass transport flux of the component i through the membrane can be expressed as ji (τ ) = 7i(piF (τ ) − piS (τ ))

∀ i; ∀ τ

(8)

There, the simulations were conducted considering the dimensions of a laboratory scale PBR (Table 1) that had been previously employed to determine experimentally the equilibrium thermodynamics and kinetics of the DME synthesis with a bifunctional CuO-ZnO-Al2O3/γ-Al2O3 catalyst.26,27 It must be noted that the CO2 hydrogenation in eq 3 (section 1) is linearly dependent on eqs 1 and 2 and thus, it has not been included in the mathematical modeling of the process thermodynamics and kinetics. The former reaction conditions together with the boundary conditions have been used in the present work (for further details, please see Diban et al.23). The mathematical model in Diban et al.23 considered that the feed and sweep gas streams, circulating counter-currently through the PBMR, flowed with constant feed and sweep gas flow velocities and under plug flow conditions, the stream pressures were fixed at the entrance of the PBMR and suffered a pressure drop with the axial position according to the consumption of reactants and formation of products and mass transport through the membrane. However, industrially the pressure of the gas streams is usually fixed at the PBR exit while the velocity of the gas stream is widely affected within the PBR axial position, in particular when there is an important change in the total number of moles during the reaction progress. It must be noted that considering the particle size (>150 μm) and the applied gas flow rate, the pressure drop along the reactor is negligible and thus, constant pressure through the PBMR length can be assumed for both the feed and sweep gas streams. This was confirmed by simulations using COMSOL software. Therefore, in the present work, the simulation conditions were adapted to describe a PBR with variable fed gas flow rate by changing the complementary equations in the mathematical model developed by Diban et al.23 as follows Q F(τ ) =

∑ FiF(τ) ∀i

Q S(τ ) =

∑ FiS(τ) ∀i

R·T PF

∀τ

R·T PS

∀τ

Figure 1. Diagram of a PBMR system with partial sweep stream recirculation.

partial condensation of H2O (separation unit), partial vent and refreshment of the sweep stream (mixer unit) and the elements required to restore the sweep gas temperature and pressure conditions necessary in the catalytic PBMR (compressor and heater). It must be noted that the unit for H2O and MeOH removal has been simplified to allow only H2O condensation in order to focus on the effect of the working configuration on the PBMR performance. For the simulations, the mathematical model reported in section 2.1 was used. Additionally, the mathematical description of condenser and mixer units has been incorporated. Regarding the condenser, 95% of H2O condensation was considered (none of the other components were condensed), and thus, only 5% of H2O remains in the dry sweep stream, FSi,dry, to be recirculated, which is mathematically described as follows for H 2O: FHS 2O,dry = 0.05 × FHS 2O, τ= 0

∀t

(11)

for the rest of the components: FiS,dry = FiS, τ= 0 ∀ t , i ≠ H2O

FHS 2O,dry,

(12)

FSi,dry

with and corresponding respectively to the H2O and component i molar flow rates in the recycled sweep stream coming from the partially dehydrated sweep gas stream. The subscript τ = 0 indicates the axial position at the entrance of the feed stream to the PBMR that is the position where the sweep stream exits the PMBR. The mass balance in the mixer unit (Figure 1) is

(9)

FiS,recirculated = α × FiS,dry + (1 − α) × Fi ,FSG

(10)

∀ i, t

(13)

where α is the recirculation factor of the sweep gas stream defined as the fraction of the molar flow rate of the sweep gas stream that is recycled to the PBMR to the total molar flow rate of the sweep gas stream. α took values ranging from 0 for no recirculation to 1 for total recirculation. Fi,FSG is the molar flow rate of each component i in the fresh sweep gas (FSG) stream. The FSG stream was formed only by H2/CO/CO2 and at t = 0 FSi,recircualted = Fi,FSG. Attending to Figure 1, it can be seen that S FSi,recircualted = Fi,τ=tao , which is the boundary condition for the

2.2. PBMR with Sweep Gas Recirculation. Figure 1 shows a flow diagram of the catalytic PBMR with in situ H2O removal for DME synthesis with CO2 cofeeding where the sweep gas could be recirculated to the PBMR at will. The feed and sweep streams circulated counter-currently and the feed stream flowed in single-pass mode, similarly as in section 2.1. For the recirculation of the sweep gas, additional elements were included in the process diagram (Figure 1) to prevent H2O accumulation in the sweep gas. Those elements aimed at the 19481

dx.doi.org/10.1021/ie503663h | Ind. Eng. Chem. Res. 2014, 53, 19479−19487

Industrial & Engineering Chemistry Research

Article

real zeolite membrane previously considered (see characteristics in Table1) was also conducted. 2.4. Definition of the Process Performance Parameters. To evaluate the effectiveness of the recirculation of the sweep gas stream in the catalytic PBMR on the process performance, the conversion of CO2 (XCO2) and the yields of the main product DME (YDME), the intermediate MeOH and the byproduct HC (YMeOH and YHC, respectively) were defined similarly to Rhode et al.19 in eqs 15 and 16. This definition accounted both for component losses to the permeate side and/or cofeeding to the feed side.

sweep gas stream at the entrance of the catalytic PBMR and at any time t (being τ=Tao also the exit position of the feed stream). The boundary condition at the entrance of the catalytic PBMR for the feed stream is FiF, τ= 0 = FiF,0

∀ i, t

(14)

The set of equations in this section (eqs 11−14) were added to the mathematical model equations described in section 2.1 and in Diban et al.23 and were implemented in the simulation software Aspen Custom Modeler v2004.1 (Aspen Technology, Inc., Cambridge, Massachusetts U.S.A.). The characteristic dimensions of the PBMR and operational conditions employed in the simulations are summarized in Table 1. The total pressure of the sweep gas, PS, was the same as in the feed side, PF, to keep partial pressures of reactants (H2/CO/CO2) similar in the feed and sweep gas sides, thus minimizing the driving force of the mass transport of the reactants and the consequent reactant losses to the sweep gas side. The operational conditions of temperature, pressure and reactants molar ratio in the feed stream were selected in order to give the highest DME yields in a PBR with the bifunctional catalyst CuO-ZnO-Al2O3/γ-Al2O3 according to previous results.26,27 The operational variables studied were (i) the molar flow rate of the FSG stream in terms of COx (CO + CO2) composition,FCOx,FSG, and (ii) the recirculation factor, α, of the sweep gas stream leaving the PBMR. 2.3. Membrane Characteristics. As previously indicated, in the present application zeolite membranes were selected because they could withstand with the demanding experimental conditions used in the catalytic PBMR in the present application (temperature of 275 °C and pressure 40 bar). Particularly, the zeolite materials ZSM5, MOR and SIL were tested by Espinoza et al.15,16 for in situ H2O removal in a PBMR configuration for FT reactions that employed operational conditions very similar to those used in a catalytic PBR for DME synthesis. The influence of the mass transport properties of these zeolite membranes on the process performance of the present system was evaluated theoretically in Diban et al.23 Due to the poor fabrication reproducibility of the zeolite membranes,28 it was reported that these membranes presented a wide range of values of H2O permeance, 7H2O, and H2O/component i selectivity, SH2O/i. Diban et al.23observed that the use of real zeolite membranes with low 7H2O and high SH2O/i led to important CO2 conversions in systems operating in once-through mode and constant feed and sweep gas velocities despite the yield of DME was reduced almost 50% in comparison to that attained in a PBR. Therefore, the zeolite membranes with low 7H2O and high SH2O/i characteristics (see values in Table 1) were selected in the present work to address the influence of the recirculation of the sweep gas stream on the process performance. It must be noted that the permeance value used in Table 1 for DME was actually for C 8 hydrocarbons,15,16 and thus, the results herein presented are considered a first approximation. Due to the higher DME polarity, water solubility, etc. in comparison to those parameters for octane (C8), it is expected that DME permeance would be higher than octane permeance. Finally, the simulation of the PBMR performance with an ideal zeolite membrane allowing only H2 and H2O permeance (size exclusion mechanism hypothesis) with the same H2O permeance as the

Χi =

Yi =

FiF|τ = 0 − FiF|τ = Tao − ji ·A m FiF|τ = 0 − Ftmb, i

(15)

nci ·FiF|τ = Tao + nci ·ji ·A m F | − Ftmb,COx FCO x τ=0

(16)

⎧ ⎫ if Ftmb, i ≥ 0 (reactant loss) ⎪0 ⎪ ⎬ being Ftmb, i = ⎨ ⎪ ⎪ ⎩(ji ·A m ) if Ftmb, i < 0 (reactant cofeeded)⎭ and Ftmb,COx ⎧0 ⎫ if Ftmb,COx ≥ 0 (reactant loss) ⎪ ⎪ ⎬ =⎨ ⎪((jCO + jCO )·A m ) if Ftmb,COx < 0 (reactant cofeeded)⎪ ⎩ ⎭ x

It must be remarked that the process suffers a slow catalyst deactivation and thus, the conversion of reactants and the product yields change with the time on stream. Therefore, the reactant conversions and the product yield were average values after 30 h of time on stream. The selectivity to the production of DME regarding the production of the organic products has been defined as follows ςDME =

YDME ∑ Yi

∀ i = DME, MeOH and HC

(17)

3. SIMULATION RESULTS AND DISCUSSION 3.1. Influence of the Sweep Gas Streamflow Rate. In Figure 2, the influence of the sweep gas flow rate, FCOx,FSG, in the range 0.06−1.8 molCOx·h−1 under once-through mode operation, on the process performance (XCO2, YDME, YMeOH and YHC) is depicted. It can also be seen that FCOx,FSG had a high influence on the process performance in the range 0.06−0.60 molCOx h−1, whereas above 0.60 molCOx·h−1, an asymptotic trend is observed. This was in agreement with the observations of Rhode et al.,17,18 which reported higher XCO2 when increasing the sweep gas molar flow rate in the PBMR configuration with in situ H2O removal for FT synthesis. At FCOx,FSG = 0.06 molCOx·h−1, the values of XCO2 were negative, similarly to those of a PBR. This result is againts the hypothesis that the in situ removal of H2O from the feed side in a PBMR would favor the chemical capture of CO2. To understand better why a PBMR with FCOx,FSG of 0.06 molCOx·h−1 resulted in CO2 formation, a detailed analysis of the effect of the sweep gas flow rates on the concentration profiles of H2O and CO2 in the feed and sweep streams flowing through the PBMR is presented in Figure 3. 19482

dx.doi.org/10.1021/ie503663h | Ind. Eng. Chem. Res. 2014, 53, 19479−19487

Industrial & Engineering Chemistry Research

Article

Figure 2. Influence of the sweep gas flow rate in the PBMR on the process performance (CO2 conversion and products yields (DME, MeOH and HC)). Comparison between the PBMR (solid color lines) and a PBR (black dotted lines).

Figure 3. Influence of the sweep gas flow rate, FCOx,FSG (0.06 and 0.60 mol·h−1), on the partial pressure of the H2O in the feed (solid color lines) and sweep gas (dotted color lines) sides and of the CO2 in the feed side with the position, τ, in the PBR (black line) and PBMR (color lines) at the time on stream of 30 h.

Figure 3a depicts the change of H2O partial pressure in the feed and sweep gas sides within PBMR position, τ, at sweep gas flow rates of 0.06 and 0.60 molCOx·h−1. This plot shows that at FCOx,FSG = 0.06 molCOx·h−1 the sweep gas stream is rapidly saturated with H2O at the entrance of the PBMR (τ = 40 gcat·h· molCOx−1) and the partial pressure of H2O in the feed side,PHF 2O, remains very similar to that in a PBR, indicating that H2O is not effectively removed from the feed side. According to eq 2, the high H2O pressure displaced the WGS reaction toward the formation of CO2 and caused that the partial pressure of CO2 in the feed side, pFCO2, was always above 5 bar (the initial value in the entrance of the PBMR) in Figure 3b. When FCOx,FSG increased, the H2O partial pressure difference between the feed and sweep sides, that is the driving force of the H2O mass transport through the membrane, also increased. Therefore, FCOx,FSG values higher than 0.06 molCOx·h−1 reduced significantly the presence of H2O in the feed stream. To keep PFCO2 below 5 bar in Figure 3b and thus to achieve positive XCO2, it was found in Figure 3a that the maximum PHF 2O allowable in the feed side is 0.58 bar. Regarding the product yield, it is observed in Figure 2a,b that YDME decreased when FCOx,FSG increased in the range 0.06−1.80 molCOx·h−1, contrary to YMeOH and YHC trends. Particularly, at low FCOx,FSG, the partial pressure of MeOH in the feed side,

pFMeOH, rose up to about 1.3 bar and decreased as FCOx,FSG increased, as illustrated in Figure 4. The higher the values of pFMeOH, the higher is the dehydration of MeOH into DME in eq 5. As it can be seen in Figure 4, pFMeOHreaches the highest value

Figure 4. Influence of the sweep gas flow rate, FCOx,FSG (0.06 and 0.60 mol·h−1), on the partial pressure of the MeOH in the feed (solid lines) and sweep gas (dotted lines) sides with the position in the PBMR, τ, at the time on stream of 30 h. 19483

dx.doi.org/10.1021/ie503663h | Ind. Eng. Chem. Res. 2014, 53, 19479−19487

Industrial & Engineering Chemistry Research

Article

Figure 5. Influence of the recirculation factor, α, and sweep gas molar flow rate in terms of COx composition, FCOx,FSG (0.18 and 0.60 mol·h−1), on (a) the CO2 conversion (XCO2), (b) DME yield (YDME), (c) MeOH yield (YMeOH) and (d) HC yield (YHC). The feed gas molar flow rate FFCOx,0 was 0.06 mol·h−1.

at FCOx,FSG = 0.06 molCOx·h−1 because the difference between the feed and sweep MeOH pressures is the lowest at this sweep gas flow rate and increases when increasing FCOx,FSG, similar to what happened to the H2O partial pressures in Figure 3a. The low selectivity of the zeolite membrane to MeOH with SH2O/MeOH values around 2 (see Table 1) caused a high permeation of MeOH from the feed toward the sweep side. To favor higher YMeOH values, MeOH should be retained in the feed side as much as possible and consequently, low values of FCOx,FSG would be recommended. However, at low FCOx,FSG, XCO2 are very low or even negative. The solution to this low XCO2 values comes from using FCOx,FSG > 0.06 molCOx·h−1. To enhance DME yields by minimizing MeOH losses from the feed side when using FCOx,FSG > 0.06 molCOx·h−1, the recirculation of the sweep stream according to the system explained in section 2.2 is evaluated in section 3.2. 3.2. Influence of the Sweep Gas Stream Recirculation. Figure 5 shows the effect of the molar flow rate of the FSG stream in the mixer unit, FCOx,FSG, between 0.18 and 0.60 molCOx·h−1, and the recirculation factor, α, between 0 and 1, on the average XCO2, YDME, YMeOH and YHC values after 30 h of time on stream. The increase of the sweep gas molar flow rate

allowed higher XCO2 values, as it was already observed in oncethrough mode operations, thus favoring the CO2 capture in the process. Albeit XCO2 remains almost independent of α at FCOx,FSG = 0.18 molCOx·h−1, at FCOx,FSG, =0.60 molCOx·h−1, XCO2 decreased 27% when increasing α from 0 to almost 1. The efficiency of the H2O removal at high FCOx,FSG is much more pronounced than at low FCOx,FSG; therefore, the effect of the recirculation of sweep gas may affect more significantly the XCO2 results at high FCOx,FSG, values. YDME values in Figure 5b increased always at increasing α while YMeOH decreased (Figure 5c). By increasing α, the MeOH concentration in the sweep gas stream increased reducing the driving force for the MeOH mass transfer across the membrane as expected. MeOH was then retained in the feed stream and could be transformed into DME. This effect is better illustrated in Figure 6. In this figure, the values of pFMeOH and pSMeOH along the position of the PBMR, τ, are depicted for different recirculation factors, α, for FCOx,FSG of 0.18 molCOx·h−1. It is clearly observed that the increase in the values of α from 0 up to 0.975, leads to a noticeable increase in the values of pFMeOH and pSMeOH in the PBMR. Furthermore, the driving force for the mass transfer of MeOH, that is the difference between the F S values of pMeOH and pMeOH , is significantly reduced by 19484

dx.doi.org/10.1021/ie503663h | Ind. Eng. Chem. Res. 2014, 53, 19479−19487

Industrial & Engineering Chemistry Research

Article

Figure 7. Results of the simulation of CO2 conversion DME yield and DME selectivity under PBR and PBMR with a real zeolite membrane (i) without recirculation (α = 0), (ii) with high recirculation factor (α = 0.9) and (iii) PBMR with an ideal zeolite membrane without recirculation (α = 0), at a sweep gas molar flow rate FCOx,FSG = 0.18 mol·h−1. The feed gas molar flow rate FFCOx,0 was 0.06 mol·h−1.

Figure 6. Influence of the recirculation factor, α, on the partial pressures of MeOH in the feed, pFMeOH (solid line), and sweep streams, PSMeOH (dotted line), with the position in the PBMR, τ, at the time on stream of 30 h. The FSG molar flow rate in terms of COx composition, FCOx,FSG was 0.18 mol·h−1 and the feed gas molar flow rate FFCOx,0 was 0.06 mol·h−1. The entrance of the sweep gas stream was on τ = 40 gcat· h·molCOx−1. The effect of α on PFMeOH was evaluated in comparison to the PBR.

3.9% in a PBR to 8.2% in a PBMR working at 0.18 molCOx·h−1. This was attributed to the reduction of the H2O content in the feed side. The H2O removal improved simultaneously the CO2 conversion and the HC formation. This was in agreement with the findings by Sierra et al.10 that observed that cofeeding H2O in the feed stream at 0.2 H2O/syngas molar ratio led to lower deactivation by coke deposition due to a lower HC formation in comparison with systems without H2O cofeeding. Despite the high values of YHC, the ςDME in a PBMR with sweep gas recirculation are still comparable to those in a PBR. However, as it was previously explained, a high formation of paraffin byproducts has to be avoided because not only it deactivates the catalyst but also adds difficulty to the efficient separation of the desired DME from the gas streams exiting the PBMR. Therefore, paraffin formation must be reduced. Finally, the performance of a PBMR working with an ideal zeolite membrane at a FCOx,FSG = 0.18 molCOx·h−1 and in oncethrough mode operation, is presented in Figure 7. It can be seen that the PBMR at high recirculation factors simulate the conditions to approach the performance results to that of a PBMR that uses an ideal zeolite membrane. The quest for the ideal zeolite membrane for selective H2O removal seems very unrealistic as the hydrophilic character of the zeolite membranes applicable in the present system also favor the MeOH permeance in addition to that of H2O. At the sight of these results, it seems that PBMRs working with low sweep gas molar flow rates (FCOx,FSG, of 0.18 molCOx·h−1) and reasonable recirculation factors (α = 0.9) resulted in a compromise among the results of XCO2, YDME and ςDME. Furthermore, these conditions reduce the consumption of reactants in the sweep gas stream, soften the recirculation settings and therefore reduce the costs of this stage. However, a detailed evaluation of the costs and technical feasibility of this configuration must be further addressed. In different simulations, independently of the operation conditions (pressure, flow rate, recirculation and ideal/real membrane selectivity) the values of YDME and YMeOH reached asymptotic values of 30 and 10%, respectively (data not shown). However, XCO2 could increase up to values >80% at the cost of the raise of YHC. Further enhancements of the process performance (high XCO2 and YDME values and low YMeOH and

increasing α. The lower driving force reduced considerably the permeation of MeOH from the feed phase toward the sweep side at higher values of α and it is negligible in the case of α = 0.975. It is worthy to note that at α = 0.975, pFMeOH in the PBMR reaches values similar to those found in a PBR. The formation of the undesired paraffin byproducts has to be taken into account seriously during the analysis of the process performance. Figure 5d shows the high influence that the sweep gas flow rate exerted on YHC. The change in FCOx,FSG, from 0.18 to 0.60 molCOx·h−1 at α = 0.975, almost doubled the value of YHC (from approximately 8 to 15%) but the YDME change was less noticeable (from 27 to 31%). The formation of HCs provokes deactivation of the catalyst by coke deposition on the active sites of the metallic function,29 limits the formation of MeOH and hinders the selective recovery of DME from the gas stream. Therefore, the formation of HCs should be minimized. In Figure 7, a comparison of the values of XCO2, YDME and ςDME between a catalytic PBR and a PBMR configuration under different working conditions is presented. As it has been previously indicated in section 3.1, a PBR under the operational conditions of reactants composition, pressure and temperature employed in the present work does not allow the conversion of CO2 into products; instead, CO2 was formed (see the negative values of XCO2). The use of a PBMR configuration at FCOx,FSG = 0.18 molCOx·h−1 gave positive XCO2 values around 25%. At this sweep flow rate, the YDME in a PBMR increased from almost 18% by working without recirculation of the sweep stream (α = 0) to 26% using a recirculation factor α = 0.9 and were always higher than in a PBR (∼16%). The improvement in the YDME was caused, as it was previously explained, by the increase of the MeOH concentration in the recycled sweep stream that reduced the driving force of MeOH transfer from the feed side to the sweep side of the PBMR. This benefitted the conversion of MeOH into DME and thus YMeOH dropped from approximately 29 to 11% and the DME selectivity ςDME increased from 33 to 55%, similar to the SDME in a PBR. It is important to remark the high increase of YHC, which rose from 19485

dx.doi.org/10.1021/ie503663h | Ind. Eng. Chem. Res. 2014, 53, 19479−19487

Industrial & Engineering Chemistry Research

Article

Notes

YHC) will inescapably pass through the improvement of the characteristics of the bifunctional CuO-ZnO-Al2O3/γ-Al2O3 catalyst employed in the present system.

The authors declare no competing financial interest.



ACKNOWLEDGMENTS Financial support by postdoctoral grant JCI-2011-10994 from the Spanish Ministry of Science and Innovation and by projects CTQ2008-00690, CTQ2010-19188 and ENE2010-15585 is kindly acknowledged.

4. CONCLUSIONS The present work aims at demonstrating the CO2 capture and recovery in the DME synthesis performed in a packed bed catalytic membrane reactor (PBMR) that incorporates a zeolite membrane for in situ H2O removal. While in the traditional packed bed reactor (PBR) CO2 was formed, the zeolite membrane of the PBMR allowed in situ H2O removal from the feed side and therefore the water gas shift (WGS) reaction was displaced toward CO2 conversion. However, the low selectivity of the zeolite membranes toward H2O permeation could cause an important reduction of the DME yield in comparison to that obtained in a PBR depending on the working conditions. In this work, it is concluded that a wise conception of the operational conditions such as the control of the sweep gas flow rates and the recirculation of the sweep gas stream could overcome the practical limitations on building an efficient PBMR for CO2 transformation into DME, given the difficulty to produce a hydrophilic zeolite membrane with selective characteristics similar to those of an ideal membrane and hydrothermally stable. The PBMR configuration proposed in the present work gives the following outstanding advantages that make the PBMR configuration to surpass the PBR performance: (1) Increases the conversion of CO2 up to 85% when high sweep gas molar flow rates (FCOx,FSG > 0.60 molCOx· h−1) are employed. (2) The recirculation of the sweep gas stream reduces the loss of MeOH across the membrane toward the sweep gas side due to a reduction of the MeOH driving force, and thus the yield of DME obtained in the PBMR can reach values around 30% at high recirculation factors. The values of YDME in the PBMR were always higher than those achieved in a PBR (16.3%) and similar ςDME (55%). (3) The recirculation of the sweep gas stream allows for savings in the consumption of reactants (H2/CO/CO2). In summary, the strategic design of the operational mode in a PBMR would favor the presence of the same components in the feed and sweep gas side, leading to a minimization of the driving force of the mass transport through the membrane and thus, would allow a higher conversion of reactants (mainly H2 and CO2) and intermediate products (MeOH) toward the aimed DME in the feed side of the PBMR. A careful evaluation of the technical difficulty of the recirculation and the high costs associated with the energy of cooling, heating and compressing the sweep gas stream must be done. Anyway, although challenging, the quest for a stable and H2O selective membrane still remains the preferable option. Additionally, a yield of the byproduct paraffins (HCs) higher than 4% is not acceptable in this system. The formation of HCs promotes the catalyst deactivation by coke deposition and hinders the MeOH formation. Attending to the results of the simulations in the present work, further improvements of the catalyst selectivity to reduce the formation of HCs becomes a research priority to improve the process efficiency and therefore, this study is currently in progress.





NOMENCLATURE Am = effective membrane area (m2) F = molar flow rate (mol h−1) Ftmb = transmembrane flux in eqs 12 and 13 (mol m−2 h−1) j = membrane partial flux dependent on time and space time (module position) (mol m−2 h−1) nc = number of carbon atoms in each component p = partial pressure (bar) P = total pressure (bar) Q = volumetric total flow rate (m3 h−1) r = reaction rate (mol.h−1.(g of catalyst)−1) R = ideal gas constant (bar m3 mol−1 K−1) S = membrane permselectivity (dimensionless) t = time (h) T = temperature (K)

Greek Letters

α = recirculation factor ςDME = selectivity of the process to the formation of DME τ = space time ((g of catalyst)·h·molCOx−1) Y = products yield (%) X = reactants conversion (%)

Other Symbols

7 = membrane permeance (units in the model, mol·bar−1· m−2·h−1; units in the text, mol·Pa−1·m−2·s−1)

Subscripts

CO = carbon monoxide CO2 = carbon dioxide COx = sum of CO and CO2 reactants DME = dimethyl ether dry = fraction of the sweep gas after partial H2O condensation in Figure 1 FSG = fresh sweep gas (new sweep gas introduced in the mixer unit in Figure 1) HC = hydrocarbons (paraffins) H2 = hydrogen H2O = water i = component (reactant or product) MeOH = methanol recirculated = stream leaving the mixer unit in Figure 1 that enters the PBMR Tao = feed stream exit position 0 = initial Superscripts



F = feed side S = sweep gas or permeate side

REFERENCES

(1) Favre, E. Membrane processes and postcombustion carbon dioxide capture: Challenges and prospects. Chem. Eng. J. 2011, 171, 782. (2) Florin, N. H.; Harris, A. T. Enhanced hydrogen production from biomass with in situ carbon dioxide capture using calcium oxide sorbents. Chem. Eng. Sci. 2008, 63, 287.

AUTHOR INFORMATION

Corresponding Author

*N. Diban. E-mail: [email protected]. 19486

dx.doi.org/10.1021/ie503663h | Ind. Eng. Chem. Res. 2014, 53, 19479−19487

Industrial & Engineering Chemistry Research

Article

selection; Basile, A., Gallucci, F., Eds.; John Wiley & Sons, Singapore, 2011; pp 243−273. (25) Piera, E.; Salomón, M. A.; Coronas, J.; Menéndez, M.; Santamaría, J. Synthesis, characterization and separation properties of a composite mordenite/ZSM-5/chabazite hydrophilic membrane. J. Membr. Sci. 1998, 149, 99. (26) Aguayo, A. T.; Ereña, J.; Mier, D.; Arandes, J. M.; Olazar, M.; Bilbao, J. Kinetic modeling of dimethyl ether synthesis in a single step on a CuO-ZnO-Al2O3/γ-Al2O3 catalyst. Ind. Eng. Chem. Res. 2007, 46, 5522. (27) Ereña, J.; Sierra, I.; Aguayo, A. T.; Ateka, A.; Olazar, M.; Bilbao, ́ J. Kinetic modelling of dimethyl ether sintesis from (H2+CO2) by considering catalyst deactivation. Chem. Eng. J. 2011, 174, 660. (28) Gascon, J.; Kapteijn, F.; Zornoza, B.; Sebastián, V.; Casado, C.; Coronas, J. Practical approach to zeolitic membranes and coatings: State of the art, opportunities, barriers, and future perspectives. Chem. Mater. 2012, 24, 2829. (29) Ereña, J.; Sierra, I.; Olazar, M.; Gayubo, A. G.; Aguayo, A. T. Deactivation of a CuO-ZnO-Al2O3/γ-Al2O3 catalyst in the synthesis of dimethyl ether. Ind. Eng. Chem. Res. 2008, 47, 2238.

(3) Li, B.; Duan, Y.; Luebke, D.; Morreale, B. Advances in CO2 capture technology: A patent review. Appl. Energy 2013, 102, 1439. (4) Alvarez-Guerra, M.; Quintanilla, S.; Irabien, A. Conversion of carbon dioxide into formate using a continuous electrochemical reduction process in a lead cathode. Chem. Eng. J. 2012, 207, 278. (5) Centi, G.; Perathoner, S. Opportunities and prospects in the chemical recycling of carbon dioxide to fuels. Catal. Today 2009, 148, 191. (6) Olah, G. A.; Goeppert, A.; Prakash, G. K. S. Chemical recycling of carbon dioxide to methanol and dimethyl ether: From greenhouse gas to renewable, environmentally carbon neutral fuels and synthetic hydrocarbons. J. Org. Chem. 2009, 74, 487. (7) Ribeiro, A. M.; Santos, J. C.; Rodrigues, A. E. PSA design for stoichiometric adjustment of bio-syngas for methanol production and co-capture of carbon dioxide. Chem. Eng. J. 2010, 163, 355. (8) Arcoumanis, C.; Bae, C.; Crookes, R.; Kinoshita, E. The potential of di-methyl ether (DME) as an alternative fuel for compressionignition engines: A review. Fuel 2008, 87, 1014. (9) Semelsberger, T. A.; Borup, R. L.; Greene, H. L. Dimethyl ether (DME) as an alternative fuel. J. Power Sources 2006, 156, 497. (10) Sierra, I.; Ereña, J.; Aguayo, A. T.; Arandes, J. M.; Olazar, M.; Bilbao, J. Co-feeding water to attenuate deactivation of the catalyst metallic function (CuO-ZnO-Al2O3/γ-Al2O3) by coke in the direct synthesis of dimethyl ether. Appl. Catal., B 2011, 106, 167. (11) Chen, H. J.; Fan, Ch.W.; Yu, Ch.S. Analysis, synthesis, and design of a one-step dimethyl ether production via a thermodynamic approach. Appl. Energy 2013, 101, 449. (12) Aguayo, A. T.; Ereña, J.; Sierra, I.; Olazar, M.; Bilbao, J. Deactivation and regeneration of hybrid catalysts in the single-step synthesis of dimethyl ether from syngas and CO. Catal. Today 2005, 106, 265. (13) Bonura, G.; Cordaro, M.; Spadaro, L.; Cannilla, C.; Arena, F.; Frusteri, F. Hybrid Cu-ZnO-ZrO2/H-ZSM5 system for the direct synthesis of DME by CO2 hydrogenation. Appl. Catal., B 2013, 140− 141, 16. (14) Ereña, J.; Garoña, R.; Arandes, J. M.; Aguayo, A. T.; Bilbao, J. Direct synthesis of dimethyl ether from (H2+CO) and (H 2+CO2) feeds. Effect of feed composition. Int. J. Chem. React. Eng. 2005, 3, A44. (15) Espinoza, R.; du Toit, E.; Santamaría, J.; Menendez, M.; Coronas, J.; Irusta, S. Production of hydrocarbons, World Patent WO 1999064380 A1, December 16, 1999. (16) Espinoza, R.; du Toit, E.; Santamaría, J.; Menendez, M.; Coronas, J.; Irusta, S. Stud. Surf. Sci. Catal. 2000, 130 A, 389. (17) Rohde, M. P.; Unruh, D.; Schaub, G. Membrane application in Fischer−Tropsch synthesis to enhance CO2 hydrogenation. Ind. Eng. Chem. Res. 2005, 44, 9653. (18) Rohde, M. P.; Schaub, G.; Vente, J. F.; van Veen, H. M. Fischer−Tropsch synthesis with in-situ H2O removal by a new hydrophilic membrane  An experimental and modelling study. DGMK Tagungsber. 2006, 4, 215. (19) Rohde, M. P.; Schaub, G.; Khajavi, S.; Jansen, J. C.; Kapteijn, F. Fischer−Tropsch synthesis with in situ H2O removal  Directions of membrane development. Microporous Mesoporous Mater. 2008, 115, 123. (20) Schaub, G.; Unruh, D.; Pabst, K. Fischer-Trropsch synfuels from biomass  Hydrocarbon yield and carbon efficiency. DGMK Tagungsber. 2008, 2, 93. (21) Iliuta, I.; Larachi, F.; Fongarland, P. Dimethyl ether synthesis with in situ H2O removal in fixed-bed membrane reactor: Model and simulations. Ind. Eng. Chem. Res. 2010, 49, 6870. (22) Diban, N.; Aguayo, A. T.; Bilbao, J.; Urtiaga, A.; Ortiz, I. Membrane reactors for in situ water removal: A review of applications. Ind. Eng. Chem. Res. 2013a, 52, 10342. (23) Diban, N.; Urtiaga, A.; Ortiz, I.; Ereña, J.; Bilbao, J.; Aguayo, A. T. Influence of the membrane properties on the catalytic production of dimethyl ether with in situ water removal for the successful capture of CO2. Chem. Eng. J. 2013b, 234, 140. (24) Tellez, C.; Menéndez, M. Zeolite membrane reactors. In Membranes for membrane reactors: Preparation, optimization and 19487

dx.doi.org/10.1021/ie503663h | Ind. Eng. Chem. Res. 2014, 53, 19479−19487