Techno-economic assessment of membrane assisted ﬂuidized bed reactors for pure H 2 production with CO 2 capture

This paper addresses the techno-economic assessment of two membrane-based technologies for H 2 pro- duction from natural gas, fully integrated with CO 2 capture. In the ﬁrst conﬁguration, a ﬂuidized bed membrane reactor (FBMR) is integrated in the H 2 plant: the natural gas reacts with steam in the catalytic bed and H 2 is simultaneously separated using Pd-based membranes, and the heat of reaction is provided to the system by feeding air as reactive sweep gas in part of the membranes and by burning part of the permeated H 2 (in order to avoid CO 2 emissions for heat supply). In the second system, named membrane assisted chemical looping reforming (MA-CLR), natural gas is converted in the fuel rector by reaction with steam and an oxygen carrier (chemical looping reforming), and the produced H 2 permeates through the membranes. The oxygen carrier is re-oxidized in a separate air reactor with air, which also provides the heat required for the endothermic reactions in the fuel reactor. The plants are optimized by varying the operating conditions of the reactors such as temperature, pressures (both at feed and permeate side), steam-to-carbon ratio and the heat recovery conﬁguration. The plant design is carried out using Aspen Simulation, while the novel reactor concepts have been designed and their performance have been studied with a dedicated phenomenological model in Matlab. Both conﬁgurations have been designed and compared with reference technologies for H 2 production based on conventional ﬁred tubular reforming (FTR) with and without CO 2 capture. The results of the analysis show that both new concepts can achieve higher H 2 yields than conventional plants (12–20% higher). The high electricity consumptions of membrane-based plants are associated with the required low pressure at the retentate side. However, the low energy cost for the CO 2 separation and compression makes the overall reforming efﬁciency from 4% to 20% higher than conventional FTR with CO 2 scrubbing. FBMR and MA-CLR show better performance than FTR with CO 2 capture technology in terms of costs mainly because of lower associated CAPEX. The cost of H 2 production reduces from 0.28 € /Nm 3H2 to 0.22 € /Nm 3H2 (FBMR) and 0.19 € /Nm 3H2 (MA-CLR). (cid:1) 2016 The Authors. Published by Elsevier Ltd. ThisisanopenaccessarticleundertheCCBYlicense(http://


Introduction
With more than 600 billion Nm 3 /year capacity installed worldwide, H 2 represents an important product for the chemical industry [1], while the H 2 demand is constantly increasing due to its potential use as automotive fuel [2]. Natural gas steam reforming is the established process for H 2 production [3]. Generally, H 2 production is carried out in a multi-tubular fixed bed reactor (fired tubular reformer-FTR) using an external furnace to provide the heat of reaction or in an auto-thermal reforming (ATR) system in which an oxidant (air or pure oxygen) is fed to the system [3][4][5][6]. The following equilibrium limited reactions of steam methane reforming (SMR, Eq. (1)), irreversible methane partial oxidation (POX, Eq. (2)) and water gas shift (WGS, Eq. (3)) play a role: Conventional steam reforming processes are based on several conversion and separation steps which include the feedstock pretreatment and sulfur compounds abatement, high temperature reforming, water gas shift reactor(s) and final H 2 separation in a pressure swing adsorption (PSA) unit in order to reach a high H 2 purity of 99.999%. The conventional plants lead to large CO 2 emissions (about 380-420 kg CO2 /Nm 3 H2 ) to the atmosphere, associated to the combustion of the PSA-offgas and part of the fuel feedstock to provide the heat for the endothermic reactions.
When including CO 2 capture, different separation systems may be adopted for different emission sources. The first emission source is CO 2 in the syngas, which is typically separated by MDEA scrubbing, prior purification in the PSA-unit. Capturing this CO 2 leads to an overall CO 2 capture ratio of 60% as discussed in [7][8][9][10]. The second emission source is the stack of the FTR furnace. In order to capture the CO 2 from this source and achieve overall capture efficiencies higher than 85-90%, it is possible to either use part of the decarbonized hydrogen as fuel in the furnace [10] or include a post-combustion capture unit at the gas stack of the reformer by MEA absorption [11].
The increased number of units also increases the CAPEX of the H 2 production with CO 2 capture. In order to reduce the costs of the plant, membrane reactors have recently been proposed as a valid alternative for both small and large-scale applications [12][13][14][15][16][17]. The advantage of membrane reactors is the possibility to shift the equilibrium toward the products by continuously extracting the H 2 through perm-selective membranes. Natural gas to H 2 conversion and H 2 separation can thus be carried out in one single unit which can be operated at a lower temperature (typically between 450°C and 600°C). At the retentate side it is possible to obtain a CO 2 -rich gas stream at pressurized conditions which can be easily separated and sent to long-term storage. Pdbased membranes have been amply studied in the last decades as the best candidate because of their very high H 2 permeability and extremely high perm-selectivity at intermediate temperatures (400-700°C) [13,[18][19][20][21]. Based on this technology, De Falco et al. [22] studied the use of multi-stage reforming and membrane separation to enhance the methane conversion at 650°C. In this configuration, after the CO 2 separation from the retentate, part of the remaining gas (mostly H 2 and unconverted CO and CH 4 ) is used as fuel in a gas turbine to produce additional electricity for the plant. The exhaust gases leaving the expander are used to provide the heat of reaction for the reforming unit leading to an overall decrease in the energy consumption by 10%. An economic assessment and comparison of this plant has been discussed in Iaquaniello et al. [23], and about 30% of cost reduction compared to the conventional steam methane reforming plant was predicted. Manzolini et al. [24] presented a techno-economic analysis of a steam reforming membrane reactor operated at 600°C and 10 bar, fully integrated in an advanced large scale combined cycle with CO 2 capture. 100% of carbon capture rate (CCR) was achieved by using an Air Separation Unit to produce pure O 2 for the complete retentate combustion and the resulting N 2 is used as sweep gas at the permeate side to enhance the hydrogen recovery factor (HRF). Depending on the thermodynamic and economic assumptions, the achieved CO 2 avoidance ranges from 38.9 €/ton CO2 to 44.5 €/ton CO2 (to be compared with 48.5 €/ton CO2 of the reference NGCC with post-combustion capture). Jordal et al. [25,26] have proposed a different plant configuration using two different membrane reactors based on Pd and microporous membranes integrated in a NGCC with CO 2 capture, reaching almost 48% of electric efficiency. Atsonios et al. integrate H 2 membrane reactor for enhanced WGS for power generation reaching 50.6% of net electric efficiency [27]. Finally, Chiesa et al. [28] presented different configurations with integration of O 2 and H 2 perovskite membranes for power production yielding an electric efficiency ranging from 45% to almost 50% and more than 87% of CO 2 avoided.
Among the several solutions already proposed for CO 2 capture, the chemical looping technology represents one of the most promising and efficient alternatives since CO 2 separation is inherently integrated in the fuel conversion step [29]. In CLC systems, a metal oxide (named oxygen carrier) is oxidized by reacting it with air and reduced by converting a fuel into CO 2 /H 2 O (chemical looping combustion, named CLC) or a syngas (chemical looping reforming, named CLR). CLR consists of two reactors operated at high pressure in which the oxygen carrier and catalyst are circulated to transfer the oxygen and heat for the reforming reaction from the air reactor to the fuel reactor. CLR has been proposed by Ortiz et al. [30,31] using Ni-based oxygen carriers because this oxygen carrier can also act as catalyst for the methane reforming reactions. Based on a simplified thermodynamic analysis, it has been found that a H 2 yield of 2.74 mol H2 /mol CH4 can be achieved with a NiO/CH 4 molar ratio equal to 1.18 to sustain the endothermic reactions at the fuel reactor [30]. Up to now no detailed thermodynamic analysis of the CLR plant has been presented. Due to differences in catalytic activity and oxygen transfer capacity, different oxygen carrier materials have been considered for the CLR process [32]. Finally, a different approach for steam methane reforming has been proposed by Ryden et al. [33], where the chem- The main objective of this paper is the thermodynamic analysis and optimization of two different membrane assisted reforming processes and the comparison from a techno-economic point of view with the benchmark technologies based on FTR plant with and without CO 2 capture. The two new processes are fully integrated with the other parts of the plant. Specifically, both H 2 plant configurations are optimized by varying the operating temperatures and pressures, steam-to-carbon ratio and other operating conditions. The design of the novel membrane reactor is carried out supported by a dedicated phenomenological model. The model is used for the reactor design and evaluation of its performance. For the selected configurations, an economic assessment is carried out and a sensitivity analysis on natural gas cost, reactor, oxygen carrier durability and membrane area costs is presented to provide an overview of the two technologies and their profitability in comparison with current technologies with and without CO 2 capture.

Reference plants
A state-of-the-art fired tubular reforming plant (FTR) has been considered as benchmark technology reproducing the plant described in Martinez et al. [10]. In this plant (Fig. 1), natural gas is first pre-heated up to 365°C to convert sulfur compounds to H 2 S and saturate any olefins in a catalytic hydrogenation reactor over a Co-Mo based catalyst operating in the range of 290-370°C. H 2 S is then adsorbed over a bed of ZnO. NG is then mixed with H 2 O to achieve a S/C ratio of 3.4 in the reformer. An adiabatic pre-reforming is carried out at 490°C over a Ni-based catalyst (generally the pre-reforming is carried out in the temperature range of 300-525°C), which removes the higher hydrocarbons that are present in the natural gas to prevent excessive coke depositions. The pre-reformed natural gas is pre-heated up to 620°C and introduced in an externally heated fired tubular reformer where the gas mixture is converted into synthesis gas (syngas) at 890°C in tubes filled with catalyst (generally Ni-based) where 80% of the CH 4 is converted into H 2 /CO. The selection of the S/C for the reformer is based on industrial practices for the production of H2 via SMR [34][35][36] to avoid Ni-based catalyst de-activation [37]. For the present benchmark reforming process, a reforming pressure of 32.7 bar has been selected based on industrial applications. After reforming, the syngas is cooled down to 338°C by producing HP-steam at 100 bar and 485°C. Syngas cooling is generally carried out in two different heat exchangers. The first one is a shell and tube gas boiler in which steam is produced. In this heat exchanger, the temperature of the syngas is reduced avoiding excessive tube overheating, which may otherwise be damaged by crevice corrosion [38]. Saturated steam leaving the steam drum is superheated downstream the gas boiler in a shell and tube which is typically U-shape and baffle plates type heat exchangers [38].
The syngas is later fed to an adiabatic WGS reactor operated between 330 and 430°C, where more than 70% of the CO in the syngas reacts with H 2 O and is converted into CO 2 and H 2 . HT-WGS is carried out in multiple reactors over a Fe-Cr based catalyst. For the reference plant without CO 2 capture a single-stage WGS suffices, since the unconverted CO is later used as fuel in the furnace. After cooling the shifted syngas to ambient temperature, the H 2 -rich syngas is fed to a PSA (Pressure Swing Adsorption) unit, which consists of multiple adsorption beds, filled with molecular sieves and activated carbon. The H 2 is produced with a purity higher than 99.999% and the PSA-off gas is combusted in the reformer burner to supply the heat for the endothermic reactions. The heat for the steam reforming reaction is supplied burning the PSA-offgas and part of the NG with air in the furnace which is surrounding the catalytic bed. Waste heat recovery downstream of the furnace is carried out in convective banks in which the feed reformers pre-heating as well as steam production is carried out. These heat exchangers are typically hairpin type HE and the air combustion pre-heating takes place in tube bundle-type equipment [38]. The steam produced is then sent to a steam turbine to produce electricity. Part of the intermediate pressure (IP) steam is sent to process to meet the S/C ratio, while the remaining stream is expanded to 6 bar and delivered to the steam line of the refinery for thermal use. The thermal integration of the plant and the heat recovery from syngas cooler (orange 1 ) and exhaust gases (green) are depicted in the cumulative grant curves in Fig. 1.
When integrating CO 2 capture technologies in the FTR plant new components as well as different operating conditions are required. The considered benchmark plant with CO 2 capture plant (and the related thermal integration) is depicted in Fig. 2. The reformate syngas is shifted in two-stages WGS to enhance both CO 2 and H 2 production: after the LT-WGS which is carried out at 213°C using a Cu-Zn catalyst supported on Al 2 O 3 , the CO content is below 0.2% with an overall CO conversion in WGS section higher than 98%. The H 2 -CO 2 rich mixture is cooled down to ambient temperature and H 2 O is condensed, the CO 2 is separated from the H 2 by means of methyl-di-ethanol amine (MDEA) absorption. Here, the CO 2 scrubbing unit is based on a chemical absorption process, which involves reversible reactions between the CO 2 and the solvent to form weakly bonded compounds. The pure CO 2 is then released, compressed and sent to the pipeline for long term storage, while the regenerated solvent is fed back to the absorber. Solvent regeneration is undertaken in a desorber column operated below 180°C, in which low pressure (LP) steam is used in the reboiler. MDEA has been selected as solvent considering the intermediate CO 2 partial pressure and the limited heat requirement for solvent regeneration.
Part of the impure H 2 from MDEA plant is used as fuel in the furnace, where it is burned with pre-heated air producing low CO 2 exhausts gases. The remaining part of the H 2 is sent to the PSA for final purification. Compared to the FTR without CO 2 capture, in this process a S/C ratio of 4 is used to enhance the NG-to-H 2 conversion and reduce the amount of unconverted species which are later burnt with the H 2 stream in the furnace.

Fluidized bed membrane reactor (FBMR)
The fluidized bed membrane reactor considered in this work has been proposed by Gallucci et al. [15] and several experimental and numerical works have been published on this concept [16,39,40]. The system (shown in Fig. 3) is based on a membrane reactor using Pd-based membranes, where the H 2 produced (#15) via the SMR/WGS reactions is separated at high purity and compressed after cooling down to ambient temperature (#18). Part of the H 2 permeates through other U-shaped membranes and reacts with air (#9) at the permeate side in order to supply the heat for the reforming reaction. The gases leaving this membrane (#10) are rich in O 2 and therefore they are used to burn some CO, H 2 and CH 4 from the cryogenic CO 2 purification unit (#7) and subsequently used to pre-heat the feed gases and produce part of the steam required for the process. At the retentate side the exiting gas (#4) mostly contains CO 2 , H 2 O and traces of H 2 and CO. Due to the high concentration and low volatility of the CO 2 compared to other components, a simplified CO 2 processing step based on low temperature phase separation can be used instead of a high energy demanding process based on physical and chemical absorption [41,42]. A two-flash auto-refrigerated cryogenic system has been designed according to Chiesa et al. [43]: after water condensation, the CO 2 rich stream is cooled down to À30°C and part of the liquefied CO 2 is separated from the stream (#6) while the remaining gas is further cooled down in a second recuperative heat exchange below À53°C to reach a deeper CO 2 recovery. The CO 2 is then throttled in order to supply the required cooling duty to the system  and after the recuperative heat exchangers is compressed to the final storage pressure. The vented gases are sent to the combustor after the membrane reactor. Due to the required S/C ratio, the total amount of steam produced from the gas cooling heat exchanger network has to be used for the process and therefore there is no steam cycle in this configuration however some steam is produced for the export depending on the operating conditions. This plant configuration can be also scaled-down to a smaller unit in which   the H 2 production plant can be integrated with other processes including also chemical processes using CO 2 as feedstock. Different process parameters have been varied to carry out the optimization of the process as discussed in later sections.

Membrane assisted chemical looping reforming (MA-CLR)
The membrane-assisted chemical looping reforming has been recently presented by Medrano et al. [44]. In this configuration, shown in Fig. 4, two reactors (named air reactor, AR and fuel reactor, FR) are used. In the FR, the fuel (stream #4) reacts with a metal oxide (NiO supported on CaAl 2 O 4 [45]) to form H 2 O and CO 2. Here, the Ni also acts as catalyst for the SMR and WGS reactions. The produced H 2 permeates through the membranes and therefore the CH 4 conversion is enhanced. Fresh NiO (#a) is fed from the top of the FR and therefore the unconverted fuel reacts with the oxygen carrier particles in the upper part of the reactor with two significant advantages: (i) at the reactor outlet it is possible to have a mixture of CO 2 and H 2 O (#5) which does not require any downstream separation treatment and (ii) part of the NiO is reduced to Ni enhancing the catalytic properties of the solid material. The oxygen carrier also acts as heat carriers because it is introduced at higher temperature than the FR temperature (200°C higher than in the fuel reactor). The oxygen carriers can also be de-activated in presence of low steam content. However, in case of chemical looping the solid circulation is also used to control the C formation since a high amount of O 2 is also available in the system as discussed by Cho et al. [46]. The H 2 from the permeate side (#13) is cooled down to ambient temperature and fed to the compressors before being delivered for final use (#17), while the retentate stream pre-heats the fuel and, after H 2 O condensation, is compressed and sent for long term storage (#7). The CLR reactors operate high pressure (up to) 50 bar. The air reactor is fed with compressed air (#9) and the exiting O 2 depleted air is used first to produce the required steam for the process and then expanded before being released to the stack. An inter-cooled air compression is adopted to avoid too high temperatures at the compressor outlet which would require very expensive materials to withstand the severe conditions resulting from a pressure ratio of up to 50. Although from a thermodynamic point of view it may be more efficient to first expand the gas in the gas turbine and then recover the heat from the expanded air at the turbine outlet, in this work a different layout has been used because of the a lack of high temperature heat and an excess of low temperature heat available from other gases that makes this solution preferable from the thermodynamic and economic points of view.

Assumptions and methodology
In this section, the main assumptions related to both the thermodynamic assessment and the estimation of the costs of the plant are discussed. The main assumptions adopted are listed in Table 1.
Most of the assumptions related to the heat exchangers, reactor conditions, CO 2 process unit for compression and purification as well as steam cycle and turbomachines efficiencies are taken from EBTF [47]. The FBMR has been modeled at chemical equilibrium for a fixed minimum H 2 partial pressure difference between the retentate and the permeate side of 0.2 bar. In the FBMR, the amount of H 2 burnt in the U-shaped membranes is calculated to achieve auto-thermal operation and keep all the streams at the same outlet temperature. In the MA-CLR, the retentate gas at the end of the membrane is fully oxidized, as predicted by the reactor model discussed further on, resulting from a counter-current flow of the OC and the retentate gas within the reactor. The HRF is calculated in order to achieve auto-thermal operation and full conversion of the reactants at the retentate side. Thus, the minimum H 2 partial pressure difference across the membranes depends on the operating conditions. Complete reduction of the oxygen carrier is achieved in the fuel reactor, while the Ni to NiO conversion in the AR is not complete, but dictated by the amount of O 2 (air), which is the amount required to close the energy balance of the system. Therefore no O 2 is contained in the AR off-gases. The amount of air is varied to reach the desired operating temperature in the FR and the solids circulation rate is used to control the temperature difference between the AR and FR. In order to define the component costs, several literature sources have been used as references (see Table 2). The cost of the syngas coolers have been calculated with Aspen Exchanger Design and Rating and compared with some data form manufacturers, the fluidized bed reactors cost has been calculated according to [48]. The gas turbine cost has been derived from [49]. Exponential scaling law has been used to calculate the equipment cost as function of scaling parameters, as indicated in Table 2. Each cost has later been adapted to the current equipment size and the cost actualized according to the chemical engineering cost index. The membrane specific cost is based on a Pall report [50] equal to 1000 $/ft 2 (corresponding to 7900 Є/m 2 ). The membrane lifetime is assumed to be 2 years [50].

Thermodynamic analysis
A schematic representation of the mass and energy balance of the four processes is depicted in Fig. 5. Each plant will convert the chemical energy from the natural gas into H 2 , electricity and heat (as steam export) and will release to the environment part of the CO 2 while the remaining part will be captured and stored. Following the same approach from Martinez et al. [10] different indices have been chosen to determine the plant performance. In particular, in addition to standard hydrogen production efficiency and specific emissions (Eqs. (6) and (8)), equivalent efficiency and emissions are calculated (Eqs. (7) and (9)) based on equivalent natural gas input (Eq. (4)) and equivalent emissions (Eq. (8)) associated to electricity and steam import/export. In these way the calculation of performance takes into account the produced H 2 , the heat and the electricity production/demand as the same amount would be provided using state-of-the-art technologies as NGCC and industrial boiler represented by the g ref . The same approach has been used for the calculation of the CO 2 emissions by using specific CO 2 emissions associated to the electricity and heat production from an external plant. The SPECCA eq (Eq. (10)) quantifies the energy cost in terms of primary energy to capture 1 kg of CO 2 and therefore takes into account both the energy penalization (to separate CO 2 ) and the reduced emissions with respect to a reference plant (in this case FTR without CO 2 capture).Finally the HR (Eq. (11)) is the typical efficiency (and units) used to quantifies the efficiency of an H 2 plant in the industry and therefore it has been also added in the comparison.

Equivalent natural gas flow rate
Steam export equivalent H 2 production efficiency CO 2 specific emissions (E CO2 ) equivalent CO 2 specific emissions (E CO2, eq )

Economic analysis
The economic assessment has been carried out in order to compare the cost of H 2 following the methodology adopted from the global CCS institute [56]. The cost of H 2 is calculated as follows: Cost of Hydrogen In which the TOC is the total overnight cost defined according to NETL [57] which includes any ''overnight" capital expenses incurred during the capital expenditure period, except for the escalation and interest during construction ( Table 3): The TOC computes the COH using the capital charge rate factor (CCF) which defines a characteristic unit cost of the plant over the life of the plant accounting for all the expenditures that occur in different periods on a common value basis. Since the ''first-year cost" methodology is applied, the rate for general inflation and real cost escalation are equal to zero. In order to determine the CCF different financial parameters have been used according to [11,47]. The resulting CCF for the entire plant equals 0.153. Operating and maintenance costs (C O&M ) are divided in two parts: the fixed costs which account for the insurance, maintenance, labor wages, chemicals and membranes replacement, and the variable costs which are related to the cost of the fuel and water, as well as the revenue or cost associated with electricity production/ consumption and steam/heat export ( Table 4). The plant availability is assumed equal to 90%.

Results
The results of the techno-economic analyses have been divided into three sections: in the first part, the optimization of the FBMR and MA-CLR plants is discussed. For the selected operating conditions, the reactor design is presented. Finally, the techno-economic assessment is discussed in the third part.

Thermodynamic optimization
The thermodynamic optimization focused on the most important operating conditions of the integrated membrane reactor. The membrane reactor operating temperature has been changed between 600 and 700°C (assuming that in the near future stable high-flux membranes operating at 700°C become available, while 600°C should be considered as the maximum temperature for today's state-of-the-art high-flux membranes), the feed pressure has been varied from 32.5 bar to 50 bar and also the effects of the permeate pressure and steam-to-carbon ratio have been investigated.

FBMR
The results of the sensitivity analysis are summarized in Figs. 6 and 7. Referring to the plant configuration in Fig. 3, the following trends are highlighted: -The g H2 , as well as the g H2,eq , decreases by increasing the steamto-carbon ratio because the amount of heat required for the steam generation, reduces the possibility to achieve a higher reactants pre-heating upstream of the membrane reactor and therefore more H 2 is required for the combustion membranes thus decreasing the net overall pure H 2 produced.
-Increasing the feed pressure increases the g H2,eq : although the H 2 yield increases slightly (thereby changing the g H2 ) a larger amount of steam for export is produced when the reactor works at higher pressure, because the steam is produced at higher pressure (and therefore the evaporator is working at higher temperature according to the saturation point). As consequence, more heat is available for the IT/LT heat recovery where the LP steam is produced.  -A lower reforming temperature slightly reduces the efficiency of the plant due to a small decrease in the CH 4 conversion. -Increasing the permeate pressure (from 1 to 4 bar), at 700°C leads to an increase in g H2,eq , despite the decrease in g H2 due to the reduction of the H 2 yield by about 1%. The lower permeated H 2 flow causes an increase of the H 2 partial pressure at the retentate side and therefore a larger amount of H 2 /CO is available at the cryogenic vent gas to pre-heat the reactants thus decreasing the amount of H 2 to be used for the combustion. On the other hand, the electricity consumption decreases by 23% due to the lower H 2 compression consumption and this effect dominates over the decrease in g H2 . A different trend is observed when the system is operated at 600°C especially at lower S/C ratios, because the lower H 2 permeation and lower temperature cause a decrease in CH 4 conversion (0.2% vs 4% CH 4 vol. fraction at the retentate side by increasing the permeate pressure). Therefore, at this temperature the effect of a lower g H2 prevails over the decrease in electricity consumption.
In terms of CO 2 emissions, the CCR varies from 70% up to 95%, depending on the operating conditions. In general, the CO 2 emissions originate from the cryogenic vent gas which is burnt for heat recovery and released as stack gas: (i) at lower temperature, the amount of carbon containing compounds at the retentate side decreases because the lower CH 4 conversion is compensated for by a higher CO conversion into CO 2 which is subsequently separated, and therefore, at the cryogenic unit the overall CO 2 content in the exhaust gases decreases; (ii) the increase in the feed pressure and/or the decrease in the permeate pressure enhances the conversion to CO 2 at the retentate side due to the higher H 2 permeation and therefore the CO 2 emissions decrease. The E CO2,eq are higher than E CO2 because of the contribution associated with the electricity consumption. Overall, the SPECCA eq is slightly lower when working at 700°C due to the g H2,eq , in particular when the permeate side is at pressurized conditions (4 bar).
In Table 5, the thermodynamic conditions of the streams depicted in Fig. 4 are reported for the case at 700°C, S/C of 2.7 and 50/4 bar at feed/permeate pressure.

MA-CLR
The fuel reactor is fed with pre-reformed syngas which reacts with the oxygen carrier to form reformate. The temperature difference between AR and FR is controlled by the solids circulation rate: the higher the solids circulation rate, the lower the temperature difference. As for the conventional ATR where the air/O 2 is varied in order to keep the system auto-thermal, in the CLR system the oxygen content (in terms of NiO) to the FR increases by increasing the air flow rate. For a high CH 4 -to-H 2 conversion, a large amount of heat is required in the FR to sustain the reactions thermally. As    Fig. 7. Environmental performance of the FBMR in terms of CO 2 specific emissions (gray bars), equivalent CO 2 specific emissions (black bars) and SPECCA (points) as defined in the methodology section. Labels indicate temperature, feed pressure and permeate pressure.

Table 5
Thermodynamic conditions of the selected streams represented in Fig. 3.  Fig. 8 and can be summarized with the following points: -When increasing the S/C ratio the equivalent efficiency decreases because less hydrogen is produced: due to the higher amount of steam to the process, the syngas is fed to the FR at a lower temperature and therefore the amount of heat required at the FR increases (the amount of NiO converted in the fuel reactor slightly increases from 0.14 to 0.15 kmol/s), while the electric consumption and the heat generated as steam export remain substantially the same. -At a higher reforming pressure (from 32 to 50 bar) the reforming efficiencies increases due to a reduced electric consumption (CO 2 is delivered at higher pressure). -At a lower temperature (650°C in this case), the overall efficiency increases: this is due to the reduced amount of heat from the reaction with oxygen carriers. Compared to the FBMR, the low S/C required enhances the performance at lower temperature because the heat available for gas cooling is sufficient to produce the steam for the process without compromising the reactants pre-heating.
-Increasing the pressure at the permeate side, the g H2 remains constant (the H 2 yield is the same), the steam to export does not change and the electricity consumption decreases due to the reduced H 2 compression consumption (e.g. at 50 bar and 700°C about 16% less of the total power is required when increasing the p per from 1 to 5 bar) resulting in an increase of g H2,eq .
-The E CO2 are zero because the retentate gas is fully converted before leaving the reactor with the fresh oxygen carrier and therefore no carbon species are released to the atmosphere; therefore the equivalent CO 2 emissions are mainly associated to the electricity consumption.
In Table 6, the thermodynamic conditions of the streams depicted in Fig. 4 are reported for the case at 700°C, S/C of 1.75 and 50/4 bar at feed/permeate pressure. The O 2 -depleted air is cooled to 630°C and after the expansion is released to the stack at 80°C, while the heat from H 2 cooling is used exclusively for steam production (including LP steam). The release of gas to the ambient from the expander at temperature as low as 80°C confirms the good process integration strategy deriving from high temperature heat recovery by steam generation, followed by gas expansion. Such high heat recovery efficiency would not be possible if heat recovery were performed after expansion of hot nitrogen stream exiting the air reactor, due to the lack of low temperature heat sink.
Based on the operating conditions of the membrane reactor, a phenomenological model [58] has been used in order to design the reactor and calculate the number of membranes and amount of oxygen carrier/catalyst required for the plant considered as detailed in the next section.

Reactor design
The membrane reactor design has been carried out using a onedimensional two-phase phenomenological fluidized bed reactor model for the freely bubbling fluidization regime at steady state conditions. The model considers the gas moving upward through an emulsion and a bubble phase. The solids in the bubble wake move upward with the bubble and this flow is balanced by the downward motion of the solids in the emulsion phase. The system is assumed to be operated in the freely bubbling fluidization regime with a u/u mf ratio in the range of 5-12 with the exception for some zones of the bed where the u/u mf drops to 1.5 due to H 2 extraction. The membranes inside the reactor act as internals for the system and therefore the bubble diameter is limited enhancing in this way the mass transfer between bubble and emulsion phase. All the constitutive equations related to the hydrodynamics and mass transfer have been taken from Kunii-Levenspiel [59][60][61] and adapted as proposed by Iliuta [62]. Note that the correlations for the mass transfer have been derived for low pressure operation, which may considered as a worst case, since the mass transfer is expected to be improved when operating at higher pressures. The kinetics of the gas-solid and heterogeneous reactions have been selected according to some experimental data on the oxygen carrier characterization undertaken at Eindhoven University of Technology (TU/e) [45]. The permeation law has been derived from permeation tests that have been carried out on cylindrical highflux Pd-based membranes using a metallic support [63,64]. The resulting H 2 permeation expression is given by:  Fig. 8. Results of the MA-CLR sensitivity analysis at different conditions. Table 6 Thermodynamic conditions of the selected streams represented in Fig. 3. For the FBMR the reactor design is carried out by changing the reactor geometry (diameter D r and length L r ) and the membrane areaas a combination of membrane length (L m ), membrane diameter (d m ) and number of membranes (N m ) in order to achieve the same H 2 yield and retentate composition as from the Aspen simulation. For the MA-CLR, the AR and the FR are operated simultaneously. The AR is designed by changing the reactor length and diameter to achieve the required solids conversion, while the FR design is carried out by varying different parameters: the number, length of the membranes have been varied to obtain the same H 2 flow permeated as in the thermodynamic analysis above discussed, while the reactor geometry is changed to obtain complete NiO reduction, and complete combustion of the fuel species at the retentate side. The energy balance of the system is not implemented in the reactor model, therefore, the presence of a temperature profile along the reactor could change the results. However, due to the high internal solids circulation, the temperature profile is expected to be virtually homogeneous for this system.
Based on the reactor design, the cost has been estimated (respectively for the reactor, membranes and particles) and a cost sensitivity analysis has been carried out in the next part of the paper, accounting also for the uncertainties associated to the cost estimation of the membrane reactor. The preliminary results of the simulations are shown in Fig. 9. It is possible to notice that: (i) for both cases, the CH 4 is converted along the entire reactor until it reaches the exit of the reactor; (ii) some influence of bubble-toemulsion phase mass transfer limitation is evident from the differences of the concentration profiles for the emulsion and bubble phases; (iii) the presence of the membranes reduced the bubble size acting as internals in the reactor, with the advantage that the composition at the emulsion and bubble phases is not very different; (iv) for the MA-CLR the H 2 is produced at the beginning and the H 2 mol fraction decreases continuously until 10.2 m and after that the H 2 , the CO and CH 4 are completely consumed because they react with NiO that is present at high concentration in the upper part of the reactor; (v) in the MA-CLR, the permeated H 2 increases from the beginning of the reactor and remains constant after 10.2 m which corresponds to the end of the membrane which leads to a strong gradient in the gas composition as shown in Fig. 9b) while in the FBMR the permeated H 2 increases until the end of the reactors; (vi) about 21% of the permeated H 2 in the FBMR reacts with air in the U-shaped membranes and therefore the number of membranes used for H 2 separation and combustion are respectively 109 and 8 (taking into account the different H 2 partial pressures at the retentate side). Based on the geometry calculated from the model, the cost of the reactor has been calculated as described in the next part.
The reactor costs have been defined by considering them as pressure vessels and the reference value used for the calculation is the one of a fluidized catalytic cracker (FCC) defined by NREL report [55]. The scaling factor assumed is the weight of the vessel. The starting point to calculate the size of the fuel reactors, are the reactor diameter and length obtained from the model simulations: following the procedure described in [65] it is possible to define the volume and consequently the total weight. The air reactor, which is usually considered as a riser is divided in two parts: a mixing zone that behaves as a bubbling fluidized bed, and a riser where the solids are transported upwards in order to be recirculated. Considering the superficial velocity of 1.2 m/s and the volumetric flow rate resulting from the mass balance, the sections of the reactor can be determined, and thus the diameter. For the mixing zone a ratio of 1.5 between length and diameter has been used, whereas for the riser the length has been assumed equal to 10 m. After having defined these values, it is possible to follow the same procedure described in [65] to calculate the weight of the air reactor.

Techno-economic evaluation
The energy balance and performance indices are shown for the selected cases in Table 7. Compared to the SMR plant without CO 2 capture, the conventional reformers with CO 2 capture with chemical absorption show a decay of more than 14 percentage point in g H2,eq , because the heat required in the reformer is higher (due to the higher S/C), the electricity consumption is higher due to CO 2 compression and the steam export is significantly lower due to the steam required for the MDEA reboiler. The FBMR shows a higher g H2 , mostly because the temperatures of the gases are kept at relatively low temperature (around 700°C) and therefore more input heat (as NG LHV ) is converted into H 2 heating value instead of sensible heat of the outlet gas streams. Due to the high reforming pressure, the CO 2 is available at high pressure and therefore the CO 2 compressor energy demand drops in comparison to the conventional system. On the other hand, H 2 is separated at a relatively low pressure (4 bar for the selected case) and therefore the electricity requited for the H 2 compression increases significantly. As explained before, the CO 2 emissions (both E CO2 and E CO2,eq ) are higher than for SMR with CO 2 capture because of the lower CO 2 capture efficiency of the cryogenic unit. In order to increase the CO 2 capture ratio, a WGS reactor can be added downstream of the FBMR to further enrich the retentate of CO 2 . This would allow increasing the CO 2 recovery rate of the CO 2 purification unit, thereby reducing the overall E CO2 to 26 kg CO2 /MWh H2,LHV and increasing the CO 2 avoidance from 72% to 80%, without any relevant change in the thermal balance of the plant.
For the MA-CLR the g H2 is above 90% because the amount of air used in the air reactor is the amount required to provide the stoichiometric O 2 (no excess is required in this case) and therefore less sensible heat is released through the gases. Moreover, a complete fuel oxidation is achieved at the retentate side, which increases the thermal input to the reforming reaction increasing the amount of H 2 produced. In terms of electricity demand, the air compressor consumption is only partly compensated by the gas expander production. However, the H 2 compressors are the most significant consumers of electricity. Similarly to the FBMR plant, the heat for the production of steam-to-export is significantly lower when compared to the conventional plant without capture, because waste heat is mostly available at low temperature and the steam produced is mostly consumed by the process itself. The direct CO 2 emissions from the plant (E CO2 ) are zero because all the CO 2 is condensed and sent to the final storage. However, the CO 2 avoidance is between 87 and 91% due to the emissions associated to the electricity demand. The electricity consumption associated with H 2 compression significantly affects the membrane-based plants both from a technical and environmental point of view. The use of sweep gas (in this case only H 2 O can be used) would be beneficial to increase the pressure at the permeate side. However, the low amount of heat available for steam generation makes the sweep gas solution less practical as a stand-alone unit, unless steam can be imported from other unit or externally produced with a conventional boiler.

Cost assessment
The results of the economic analysis are presented in Table 8. The BEC has been listed for the main group of components. For the conventional SMR the largest cost is associated to the reactor (in particular the reformer unit) and the heat exchanger that is used to cool down the exhaust gas from the furnace. The cost of the PSA, which is operated with a high volumetric flow rate, is also above 20% of the BEC. In case CO 2 capture is carried out with a MDEA unit the investment cost increases by about 1/3 mostly due to the cost of CO 2 , separation (absorption columns and CO 2 compressors). The cost of the reactors in the FBMR is comparable with the cost required for the reactors in conventional H 2 plants. The cost of H 2 Pd-based membranes accounts for only 5% of the total BEC. The most expensive unit of the plant is represented by the convective heat exchangers network (HEN) for the H 2 and air cooling: due to the low heat transfer coefficient (assumed 35 W m À2 K À1 because of the low pressure) and the high volumetric flow rate (especially when H 2 is cooled), the resulting heat transfer area is very high with associated high cost (43% of the BEC). In case of MA-CLR, the highest cost is associated to the CLR reactor units, especially the fuel reactor (ranging from 51% to 57% of the total reactors costs). The cost of the membranes is higher than in the case of FBMR with a slightly higher impact on the BEC. Although the amount of H 2 separated in the MA-CLR membranes is lower compared to the FBMR, the required membrane area is slightly higher. Two main reasons explain this difference: (i) the conversion of CH 4 in the FBMR is higher in the bottom part of the reactor and therefore the p H2,ret is always higher along the reactor than in the case of MA-CLR; (ii) in case of FBMR, part of the H 2 required to supply the heat of reaction is separated with air used as sweep gas which consumes instantaneously the permeated H 2 and therefore the p H2,per is virtually zero. The differences in the reactor geometry for the MA-CLR are dictated by the operating pressure at the retentate/permeate side of the membrane which influence the hydrodynamics of the reactor and the H 2 permeation.
According to the thermodynamic and the economic assessment, the variable costs are higher than the fixed costs. This is particularly pronounced for the case of SMR + CA-MDEA unit due to the low efficiency that leads to high fuel cost. As already anticipated in the previous part of the paper, the energy and economic cost for the electricity-to-import in the membrane reactor plants   significantly contribute to the COH, which indicates that one of the solutions to improve the plant performance is to separate the H 2 at higher pressure, possibly increasing the reforming pressure. In terms of overall economic performance, compared to the conventional plant without CO 2 capture, the SMR with chemical absorption has a cost of CO 2 avoided of 83 €/t CO2 , while for FBMR the cost of CO 2 avoided is around 6 €/t CO2 and it becomes even negative for the MA-CLR due to the reduced operating costs associated with the high efficiencies of the plant. For the MA-CLR, the results show that an investment in R&D for the scale-up and commercialization of the technology would already be beneficial at this stage due to the lower COH compared to the other technologies including the conventional H 2 production plant.

Sensitivity analysis
A sensitivity analysis has been carried out on the COH by varying the main parameters affecting the COH and the results are presented in Fig. 10. This analysis is required, mainly due to the uncertainties associated on the cost of membranes, of reactors vessels, of the oxygen carrier and on the cost of NG. Since the cost for the membranes accounts for 5-8% (as showed in Table 8), an increase in the cost of Pd membranes up to 4 times the one considered in this work increases the COH up to approximately 0.24 and 0.21 €/m 2 for the FBMR and MA-CLR (@50 bar/700°C) respectively, but H 2 production with MA-CLR still compares favorably with the state-of-the-art technology without CO 2 capture. In case the cost of the reactor vessels increases by more than twice, the membranebased plant would cost more than the conventional steam reforming plant. The effect of the cost of the oxygen carrier on the COH is negligible. Finally, in case the cost of NG increases from 0.5 to 2 times the cost used as reference scenario, the COH increases more for the conventional and CA-MDEA plants due to the lower efficiencies of H 2 production. It must be noted that in this analysis it is not considered how the cost of NG would influence the cost of electricity: as the membrane plants require a high electricity import due to the H 2 compressors, an increase in the electricity cost would also increase the COH. For all the sensitivity analyses presented here, the membrane based plants are always more convenient than the plant with conventional CO 2 capture technology by chemical absorption.

Conclusions
Two membrane based plants, FBMR and MA-CLR, for the production of H 2 with CO 2 capture using natural gas as feedstock have been studied and compared with reference technologies from a techno-economic point of view. In order to improve the efficiency of the system a lower S/C ratio and higher operating pressures are beneficial for both systems. The reactors temperature affects the heat recovery significantly and the optimal value depends on the specific plant considered. For the membrane based plants the cost of H 2 compression is the most relevant energy cost due to the relatively low H 2 pressure at the permeate side. The membrane reactor can improve the gas conversion and reduce the number of components required in the plant. In case of FBMR, the cost of the HEN is drastically limiting the economic performance of the plant, while for the case of MA-CLR the design of an interconnected fluidized bed reactor operated at high pressure needs to be experimentally demonstrated. The COH associated to the FBMR is only slightly higher than the conventional system without CO 2 capture and the cost of CO 2 avoided is about 6 €/t CO2 , while for the MA-CLR the economic advantages are evident in terms of reforming efficiencies, CCR and COH.