A computational model of hydrogen production by steam reforming of dimethyl ether in a large scale CFB reactor. Part II: Parametric analysis

A computational model of hydrogen production by steam reforming of dimethyl ether in a large scale CFB reactor. Part II: Parametric analysis

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y x x x ( 2 0 1 6 ) 1 e1 0 Available online at www.sciencedirect.com ScienceDi...

2MB Sizes 0 Downloads 33 Views

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y x x x ( 2 0 1 6 ) 1 e1 0

Available online at www.sciencedirect.com

ScienceDirect journal homepage: www.elsevier.com/locate/he

A computational model of hydrogen production by steam reforming of dimethyl ether in a large scale CFB reactor. Part II: Parametric analysis Francis A. Elewuwa a, Yassir T. Makkawi b,* a

European Bioenergy Research Institute (EBRI), School of Engineering and Applied Science, Aston University, Birmingham B4 7ET, UK b Chemical Engineering Department, American University of Sharjah, P.O. Box 26666, Sharjah, United Arab Emirates

article info

abstract

Article history:

This study presents a computational parametric analysis of DME steam reforming in a

Received 19 May 2016

large scale Circulating Fluidized Bed (CFB) reactor. The Computational Fluid Dynamic (CFD)

Received in revised form

model used, which is based on EulerianeEulerian dispersed flow, has been developed and

9 August 2016

validated in Part I of this study [1]. The effect of the reactor inlet configuration, gas resi-

Accepted 11 August 2016

dence time, inlet temperature and steam to DME ratio on the overall reactor performance

Available online xxx

and products have all been investigated. The results have shown that the use of double sided solid feeding system remarkable improvement in the flow uniformity, but with

Keywords:

limited effect on the reactions and products. The temperature has been found to play a

Dimethyl ether

dominant role in increasing the DME conversion and the hydrogen yield. According to the

Hydrogen

parametric analysis, it is recommended to run the CFB reactor at around 300  C inlet

CFD modelling

1 temperature, 5.5 steam to DME molar ratio, 4 s gas residence time and 37,104 ml g1 cat h

Steam reforming

space velocity. At these conditions, the DME conversion and hydrogen molar concentration

Parametric analysis

in the product gas were both found to be around 80%.

Fluidized bed

© 2016 Hydrogen Energy Publications LLC. Published by Elsevier Ltd. All rights reserved.

Introduction Hydrogen production through Dimethyl Ether (DME) steam reforming in a fluidized bed is a relatively new process, therefore, there is a need of reliable design and optimization tool in order to take this process into industrial scale application. Improved hydrogen production, enhanced operability of the reactor and increased economic viability of the process can be achieved by optimizing the reactor design and using the appropriate range of operating conditions. In Part I of this

study [1], a valid computational model was developed and the prediction of the reactor hydrodynamics (in terms of phases distribution, velocities and residence time) and thermochemical performance (in terms of temperature, species distribution and gas composition) were studied at one selected operating condition of temperature ¼ 300  C, steam to DME 1 molar ratio ¼ 7.6 and a space velocity ¼ 37,104 ml g1 cat h . The reported studies on methanol (MeOH) and DME steam catalytic reforming suggest that the ratio of steam to methanol or DME and the reactor temperature are the two most important parameters that influence the degree of chemical conversion

* Corresponding author. Fax: þ971 65152979. E-mail address: [email protected] (Y.T. Makkawi). http://dx.doi.org/10.1016/j.ijhydene.2016.08.072 0360-3199/© 2016 Hydrogen Energy Publications LLC. Published by Elsevier Ltd. All rights reserved. Please cite this article in press as: Elewuwa FA, Makkawi YT, A computational model of hydrogen production by steam reforming of dimethyl ether in a large scale CFB reactor. Part II: Parametric analysis, International Journal of Hydrogen Energy (2016), http:// dx.doi.org/10.1016/j.ijhydene.2016.08.072

2

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y x x x ( 2 0 1 6 ) 1 e1 0

and hydrogen production in fluidized bed reactors (e.g. Refs. [2,3]). There are several reported experimental and theoretical studies on parametric analysis of catalytic steam reforming of DME or methanol (e.g. Refs. [3,4]). None of these have looked into the performance of a circulating fluidised bed reactor at a range of operating conditions. This is despite of the great potential of a circulating fluidised bed for industrial scale application compared to fixed or bubbling fluidized bed reactors. It is therefore of interest to shed light on the performance of this type of reactor at a range of operating conditions relevant to industrial scale processing. The parameters that are commonly investigated are the temperature, steam to DME ratio, space velocity and the catalyst type. In a theoretical and experimental study of DME steam reforming (DME-SR) in a micro-reactor [5] it was reported that higher DME conversion and carbon monoxide concentration in the product gas occurs at increasing the reactor wall temperature. A wall temperature of 270  C was recommended as optimum for the hydrogen production. Another experimental study on DME-SR over a metallic catalyst (CuO/ZnO/Al2O3/ZrO2 þ ZSM5) in a fixed bed reactor [6] has shown increased DME conversion and hydrogen yield when increasing the temperature and steam to DME ratio. On the contrary, the increase in the space velocity was reported to decrease both the DME conversion and hydrogen yield but increased the carbon dioxide selectivity [4e6]. A similar experimental and numerical study on a fixed bed reactor has shown the hydrogen production to decrease with increasing the temperature beyond 300  C [4]. The experimental results at various operating conditions (space velocity of 2420e4615 h1 and temperature of 270e310  C) have been found to reasonably match the numerical results obtained using a simple one-dimensional heterogeneous model. Hydrogen production by steam reforming of DME is particularly an attractive option because the process can be carried out at a relatively low temperature compared to the conventional method of natural gas reforming, which takes place around 600  C. This is in addition to other advantages related to DME properties such as high hydrogen to carbon ratio, non-corrosive, non-carcinogenic, non-toxic nature and increased worldwide mass production from various resources [7]. Hydrogen produced by methanol steam reforming is another alternative option for low temperature processing (200e300  C) [8], however methanol is more expensive to produce and entails the risk of biological poisoning [4,9]. Recent theoretical studies on DME steam reforming have looked into the potential of making the process thermal efficiency more attractive by utilizing exhaust gas as a source of heat to derive the endothermic DME reforming reaction [10,11]. These studies included the solution of CFD models (mass, energy, momentum and reaction equations) [10] and Aspen plus software [11] to predict the hydrogen yield and process thermal efficiency with respect to varying the DME to steam ratio and heat supply. Both parameters were found to play an important role in the overall reactor performance. Studies on the types of catalyst (e.g. Refs. [12e14]) are generally in agreement that the DME steam reforming process occurs in two-steps involving the hydrolysis of DME and the

steam methanol reformation. These two steps led to the requirement of a bifunctional catalyst to facilitate both reactions [15]. However depending on the catalyst system utilised and reaction parameters some side reactions may occur, which could include water-gas shift reaction and DME decomposition reaction. In this second part of the study, the computational fluid dynamic (CFD) model developed and validated in part I [1] is used to study the reactor design improvement by altering the feeding points, carry out parametric analysis and identify the optimum operating conditions for increased hydrogen production in a circulating fluidized bed reactors. A simple energy balance is first described to allow estimation of the thermal input required to derive the reaction. This is followed by studying the impact of changing the solid catalyst feeding points on the overall flow hydrodynamics and product quality. In the last section, parametric analysis focused on the effect of steam to DME ratio, reactor temperature and the gas residence time is presented. The study is then concluded with a summary of the optimum conditions recommended for increasing the hydrogen yield.

Computational model and chemical reactions The DME steam reforming reactions, rate laws and the computational model used to simulate the overall reactor performance and product gas composition have been given in details in Part I of this study [1]. Here, only the main model equations (hydrodynamics, heat transfer and reactions) in addition to the computation method are briefly described. Conservation of mass: v ðai ri Þ þ V$ðai ri vi Þ ¼ Si vt

i ¼ gas or solid

(1)

Gas phase momentum:    v ag rg vg þ V$ ag rg vg ¼ ag VPg þ V$tg þ ag rg g vt 2   X  j ¼ solid 1 or 2 bsj vg  vsj j¼1

(2) Solid phase momentum:    v as r vs þ V$ asi rsi vsi ¼ asi VPsi þ V$tsi þ asi rsi g vt i si i 2     X þ bsi vg  vsi þ Ksi;j vsi  vsj j¼1

js i (3) Granular kinetic energy:        3 v rSi asi qsi þ V$ rsi asi qsi vsi ¼  Psi I þ tsi : Vvsi 2 vt   þ V$ ksi qsi VQsi  gsi þ ∅gsi

(4)

j ¼ solid 1 or 2

Please cite this article in press as: Elewuwa FA, Makkawi YT, A computational model of hydrogen production by steam reforming of dimethyl ether in a large scale CFB reactor. Part II: Parametric analysis, International Journal of Hydrogen Energy (2016), http:// dx.doi.org/10.1016/j.ijhydene.2016.08.072

3

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y x x x ( 2 0 1 6 ) 1 e1 0

Heat transfer: 2 X v vPi ðai ri hi Þ þ V$ðai ri vi hi Þ ¼ ai þ ti : Vvi  V$qi þ Ri þ Qs j g vt vt j¼1

i ¼ solid or gas;

Chemical reactions: CH3 OCH3 þ H2 O42CH3 OH DHor ¼ þ24:5 kJ=mol

(6)

CH3 OH þ H2 O/CO2 þ 3H2

(7)

DHor ¼ þ49:1 kJ=mol

DHor ¼ þ90:0 kJ=mol

CO þ H2 O4CO2 þ H2

Parameter

This study

j ¼ solid 1 or 2 (5)

CH3 OH4CO þ 2H2

Table 1 e Range of operating conditions considered in the simulations.

DHor ¼ 41:17 kJ=mol

(8) (9)

The reaction rates and kinetics of the above reactions ((6)e(9)) have been derived from experimental work on catalytic DME steam reforming using the bifunctional catalyst CuO/ZnO/Al2O3 þ ZSM-5 (see Table 2 in part I of this study [1]). The metallic part of the catalyst (CuO/ZnO/Al2O3) is widely reported as being effective for methanol steam reforming, while the ZSM-5 zeolite is added to enhance the DME hydrolysis reaction (due to its high acidity) [3,16e18]. In practise, such a bifunctional catalyst is usually prepared by mill crushing of H-ZSM-5 and Cu/ZnO/Al2O3 powders in equal weights to produce a mixture of particles with the required size distribution [4]. The reactions have been implemented in a EluerianeEulerian model and solved for the prediction of the multiphase flow hydrodynamics, heat transfer and reactions using ANSYS Fluent CFD commercial code. The flow mixture is assumed to consist of two solid phases of different physical properties, namely, CuO/ZnO/Al2O3 and ZSM-5, and a gas phase consisting of several species, namely, hydrogen, carbon monoxide, carbon dioxide, methanol, DME and steam. The chemical reactions rates and kinetics have been incorporated in the model using in-house developed user defined function. For further details on the full computational model and solution procedure the reader is referred to Part I of this study [1].

Steam Inlet temperature ( C) Flow rate (kg/s) DME Inlet temperature ( C) Flow rate (kg/s) Steam to DME molar ratio () Reactor dimension Diameter (m) Height (m) Reactor solid feeding configuration 1 Space velocity (ml g1 cat h )

Base case study [1]

200e350 4.7e9.0

300 9.0

200e350 3.0 4e7.6

300 3.0 7.6

3.0 10e20 Double sided inlet 37,104

3.0 15 Single sided inlets 37,104

investigation by considering a new reactor configuration with two gas-assisted catalyst feeding points from opposite sides (symmetric) as demonstrated in Fig. 1. In order to assess the effect of the reactor temperature on the product gas quality, the temperature of the feed (DME, steam and catalyst) was varied within the range of 200e350  C. The effect of the steam to DME molar ratio was studied by varying the ratio within the range of 3.0e7.6 while maintaining the inlet temperature at 300  C. The effect of the gas residence time, which is directly related to the degree of gas solid contact and reaction equilibrium, has been investigated by varying the reactor height within the range of 10e20 m, while maintaining all other operating conditions same as in the base case. It is recognized that the reactor operating conditions are interrelated, for instance, increasing the temperature will result in reducing the residence time, while the same effect could be obtained by

Parametric analysis procedure The reactor hydrodynamics and thermochemical performance have been investigated through parametric analysis by modifying the reactor design (solid feeding point and total reactor height) and varying the operating parameters (temperature and steam to DME ratio). In each case, a comparison has been made against the reactor performance at the base case conditions given in Part I of this study [1]. Table 1 gives a summary of the range of all parameters used in comparison to the base case. The choice of the range of operating condition has been based on reported studies on DME SR. In the base case study, the solid catalyst was introduced to the reactor through one feeding point located at the lower part of the reactor. As a consequence, flow non-uniformity was observed all along the reactor height (asymmetric flow structure). It was not clear how such a flow structure affects the overall reactor performance. This prompted further

Fig. 1 e The computational domain and meshing of the CFB riser reactor (a) base case study reactor design with one single sided catalyst feeding [1] (b) modified reactor design with two sided catalyst feeding (symmetric feeding).

Please cite this article in press as: Elewuwa FA, Makkawi YT, A computational model of hydrogen production by steam reforming of dimethyl ether in a large scale CFB reactor. Part II: Parametric analysis, International Journal of Hydrogen Energy (2016), http:// dx.doi.org/10.1016/j.ijhydene.2016.08.072

4

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y x x x ( 2 0 1 6 ) 1 e1 0

reducing the reactor height. Therefore, care is exercised here to take into consideration such a complex relation when discussing the impact of the inlet temperature and steam to DME ratio. The chemical performance of the reactor has been assessed in terms of the product gas composition, DME conversion (CDME), methanol conversion (XMeOH), hydrogen yield (YH2 ) and carbon dioxide selectivity (SCO2 ). The mathematical formulas defining these parameters are given in Table 2. Note that, in calculating the carbon dioxide selectivity (Eq. (13)), some of the reported literature assumes the DME to decompose producing a small fraction of methane and hence added this to the total gas products containing carbon (the denominator). However, no quantified quantity of methane has been experimentally proven for the range of conditions considered here, hence this is assumed negligible.

Results and discussion Simple energy balance In the DME steam reforming process, the DME hydrolysis reaction and the subsequent methanol production and side reactions are all reversible and strongly dependent on the operating temperature. Here, it is of interest to show how a simple energy balance calculation can be used for first estimate of the required heat supply to the reactor, assuming that the reactions are predominantly forward at the range of temperature considered here. The general energy balance for a steady state flow reactor is given by Ref. [19]: Qr ¼ FAo

n X

Qi Cpi ½T  TR  þ DHRX ðTÞFAo X

(14)

i¼1

where Qr is the heat supplied by the circulating solid catalyst. The first term in the right represents the rate of energy introduced to the reactor and the second term represents the rate of energy generation or consumption. If, for simplicity, it is assumed that the temperature is uniform within the reactor and that the heat capacities remain constant, then the heat of reaction, DHRX, is given by: DHRX ðTÞ ¼ DHoRX ðTR Þ þ DCp ðT  Tio Þ þ Dl

(15)

where the first term in the right represent the heat of reaction at the reference temperature of 25  C, the second term represent the overall change in the heat capacity per mole of the limiting reactant and the last term represent the overall change in latent heat in case of phase change within the range of reactor temperature and reference temperature. Now, summing-up all the reactions given in Section “Computational model and chemical reactions” (Eqs (6)e(9)) according to Hess's law and assuming that little or no reversibility takes place, as stated above, then the following single step reaction is obtained: CH3 OCH3 þ 3H2 O/2CO2 þ 6H2

DHor ¼ þ122:4 kJ=mol

(16)

Further, if it is assumed that the reactor temperature remains high, with only 15  C drop between the inlet and exit, as shown in Part I of this study [1], and that complete DME conversion takes place, then the total thermal input by the solid feed is Qr ¼ 9762.4 kW. For a solid catalysts supplied to the reactor at 300  C, this thermal input corresponds to a solid flow rate of 591.6 kg/s. Accordingly, a total solid feed rate of 600 kg/s (300 kg/s each catalyst) at 300  C is assumed sufficient to derive the reaction and maintain the reactor close to the desired temperature.

Effect of the reactor inlet design In Part I of this study [1], it was observed that the flow structure exhibits uneven distribution of the solid catalyst (asymmetric) and this was mainly attributed to the entrance effect (solid feeding from one side of the reactor). In this study, symmetrical feeding was adopted to improve the solid distribution and contact between the gas and catalyst, which is expected to enhance the overall DME conversion and hydrogen yield. Fig. 2 shows the result of solid distribution in the reactor with the modified inlet in comparison with the original design with one side catalyst feeding. Clearly, the radial solid distribution in the modified reactor is symmetrical and the flow maintain better uniformity in the upper level compared to the original design. It is also interesting to note that the middle of the modified reactor exhibits higher solid volume fraction. While this is a clear deviation from the classic feature of dilute core and dense annular flow commonly observed in fluidized bed reactors, it is to a great extent resembling the flow pattern in the upper part of a spouted bed with aeration [20].

Table 2 e Definition of the parameters used in the analysis. Parameter DME conversion (%)

MeOH conversion (%)

Hydrogen yield (%)

Carbon dioxide selectivity (%)

Formula nDME;in nDME;out nDME;in

CDME ¼  100 (10) where nDME,in and nDME,out are the molar flowrate of dimethyl ether at the inlet and outlet of the reactor, respectively n nDME;out  100 (11) CMeOH ¼ MeOH;pro nMeOH;pro where nMeOH,pro and nDME,out are the molar flow of methanol produced and that is leaving the reactor, respectively nH2  100 (12) YH2 ¼ u1 nDME;in where u ¼ 6 represents the stoichiometric coefficient of the hydrogen component produced in the reactions ni  100 (13) SCO2 ¼ nCO þn CO2 where ni represents the molar flow rate of CO or CO2.

Please cite this article in press as: Elewuwa FA, Makkawi YT, A computational model of hydrogen production by steam reforming of dimethyl ether in a large scale CFB reactor. Part II: Parametric analysis, International Journal of Hydrogen Energy (2016), http:// dx.doi.org/10.1016/j.ijhydene.2016.08.072

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y x x x ( 2 0 1 6 ) 1 e1 0

5

quantitatively negligible. The slight increase in the hydrogen yield and DME conversion in the modified reactor is believed to be due to the improved solid distribution, i.e. better catalyst-gas contact.

Effect of the gas residence time The gas residence time is an important operational parameter that is of direct impact on the reactants conversion. It is also commonly used as a scaling parameter in reactor design. In Part I of this study [1], it was shown that the gas residence time distribution at a given operating conditions can be obtained by a numerical procedure monitoring the exit concentration of a tracer as function of time. The mean gas residence time, t, is then given as function of the residence time distribution function, E(t), as follows: Z



tEðtÞdt Fig. 2 e Contours of solid concentration and distribution in the CFB reactor (a) original design with one side solid feeding point (b) modified design with two sides solid feeding. Operating at inlet temperature ¼ 300  C and steam to DME molar ratio ¼ 7.6. Further details on the effect of modifying the reactor inlet is shown in Fig. 3 in terms of the contours and vectors of the solid (catalyst) velocity. The introduction of the solid from two radially opposite directions appear to create two vortices that disrupt the flow pattern in the lower region. The disruption in the entrance region led to the transformation of the classic core-annular flow into a flow pattern similar to that observed in the upper part of a spouted bed, as noted earlier. Some of the solid appears to deflect towards the wall to fall under intense upward gas drag force, hence, attaining high particle velocity in this region. In a previous study on the effect of inlet boundary [21], it was shown through CFD modelling and experiment that a dense core and high solid velocity at the wall of a CFB riser, as a predicted here, is a phenomenon related to the entrance effect. This was specifically attributed to the significant loss of particle kinetic energy as they radially collide at the entrance. This in turn, result in altering the radial particle concentration and velocity profiles up to the top section of the riser. It is worth noting that, there is little knowledge on the effect of dual solid feeding on CFB riser hydrodynamics, and perhaps, the CFD approach is the best option for such investigation, especially for a large scale system as considered here. The impact of re-arranging the solid feeding and the associated hydrodynamic changes on the reactions and products, the conversion, yield, selectivity and gas composition has been compared with the base case design in Fig. 4. Clearly, there is little effect on the conversion, yield and selectivity despite the difference in the flow hydrodynamics as demonstrated earlier in Figs. 2 and 3. Fig. 4b shows that the hydrogen and carbon dioxide concentrations remain almost the same, while there is around 50 mol% difference in the concentration of the DME, carbon monoxide and methanol. These last three components together represent less than 8 mol% of the overall product gas composition, hence, are

t ¼ Z0 ∞

(17) EðtÞdt

0

The reactor temperature, molar feed flow of the reactants, space velocity and reactor height all contribute to the variation of the gas residence time. The choice has been made here to vary the residence time by changing the reactor height while maintaining all other parameters at the base case condition. Fig. 5 shows the gas residence time distribution as function of the reactor height. The residence time distribution curves shift to the right with increasing the reactor height, hence, indicating an increase in the gas mean residence time. The calculated mean residence times were 2.5 s, 3.9 s and 5.1 for the three heights of H ¼ 10 m, 15 m and 20 m, respectively. At the height of 15 m, it is suggested that the flow closely resemble plug flow behaviour due to the narrow residence time distribution. This permits better operation and control but with the plenty of reduced axial mixing. Interestingly, the data suggest linear relationship between the mean residence time and height given by: t ¼ 0:256H. The product gas composition corresponding to the various reactor heights discussed above is shown in Fig. 6. It is clear that the change in the gas composition is negligible, particularly beyond 15 m height. This suggest that, for the operating conditions considered here, a residence time of around 4 s is sufficient to reach chemical equilibrium.

Effect of temperature Reported experimental studies on DME steam reforming have shown that the temperature plays a dominant role in the reactor performance. Fig. 7a shows the effect of temperature on the DME conversion, hydrogen yield, carbon dioxide conversion. The result suggests that the DME hydrolysis reaction, which is slightly endothermic, shifts towards the right as the temperature increases. Consequently, the overall hydrogen yield increases as demonstrated in Fig. 7b. On the contrary, the carbon dioxide selectivity slightly decreases beyond 300  C because of the effect of equilibrium in the water gas shift reaction at higher temperature, which favours the formation of more carbon monoxide. These trend agree well with the

Please cite this article in press as: Elewuwa FA, Makkawi YT, A computational model of hydrogen production by steam reforming of dimethyl ether in a large scale CFB reactor. Part II: Parametric analysis, International Journal of Hydrogen Energy (2016), http:// dx.doi.org/10.1016/j.ijhydene.2016.08.072

6

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y x x x ( 2 0 1 6 ) 1 e1 0

Fig. 3 e Time averaged contours (a1 and b1) and vectors (a2 and b2) of solid velocity in the CFB reactor (a) original design with one side feeding point (b) modified design with two sides feeding. Operating at inlet temperature ¼ 300  C and steam to DME molar ratio ¼ 7.6.

reported literatures for similar operating conditions. However, some studies have shown complete DME conversion at a much lower temperature and space velocity [6]. It is also clear in Fig. 7b that significant increase in the hydrogen concentration occurs within the range of 200e250  C. At a high temperature of 350  C, the DME and methanol are almost fully converted, hence, the hydrogen yield is maximum. The set of reactions given in Eqs. (6)e(9) suggest that the carbon dioxide and monoxide concentrations in the product gas are interrelated through the water gas shift reaction (Eq. (9)). At the temperature beyond 300  C, the reverse reaction results in reducing the carbon monoxide as clearly demonstrated here. It should be noted that, some of the studies reported deactivation of the metallic part of the bifunctional catalyst (CuO/ZnO/Al2O3) by coke deposition or

Cu sintering at the high temperature around or beyond 300  C [3,6]. Such a phenomena has not been considered in the model used in this study, therefore, the results at the temperature of 350  C should be treated with caution.

Effect of steam to DME ratio The effect of steam to DME ratio was studied while maintaining the inlet temperature constant at 300  C. Fig. 8a shows that the hydrogen yield and the DME conversion both consistently increase with increasing steam to DME ratio. Within the range considered here, the percentage increase in the hydrogen yield is around 30%, while for the DME conversion it is just around 10%. This trend is to some extent similar to the one observed when increasing the temperature,

Please cite this article in press as: Elewuwa FA, Makkawi YT, A computational model of hydrogen production by steam reforming of dimethyl ether in a large scale CFB reactor. Part II: Parametric analysis, International Journal of Hydrogen Energy (2016), http:// dx.doi.org/10.1016/j.ijhydene.2016.08.072

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y x x x ( 2 0 1 6 ) 1 e1 0

7

Fig. 6 e Variation of the gas composition with reactor height. Operating at temperature ¼ 300  C and steam to DME molar ratio ¼ 7.6.

Fig. 4 e Comparison between the original and modified reactors design (a) DME conversion, hydrogen yield and CO2 selectivity (b) product gas composition. Operating at temperature ¼ 300  C and steam to DME molar ratio ¼ 7.6.

Fig. 5 e Variation of the gas residence time with the reactor height. Operating at temperature ¼ 300  C and steam to DME molar ratio ¼ 7.6.

Fig. 7 e Effect of the operating temperature on (a) DME conversion (CDME), hydrogen yield (YH2) and carbon dioxide selectivity ðSCO2 Þ (b) product gas composition. Operating at steam to DME molar ratio ¼ 7.6.

Please cite this article in press as: Elewuwa FA, Makkawi YT, A computational model of hydrogen production by steam reforming of dimethyl ether in a large scale CFB reactor. Part II: Parametric analysis, International Journal of Hydrogen Energy (2016), http:// dx.doi.org/10.1016/j.ijhydene.2016.08.072

8

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y x x x ( 2 0 1 6 ) 1 e1 0

Fig. 8 e Effect of steam to DME molar ratio on the (a) DME conversion (CDME), hydrogen yield ðYH2 Þ and carbon dioxide selectivity ðSCO2 Þ (b) product gas composition. Operating at inlet temperature ¼ 300  C.

however, the carbon dioxide selectivity appear to be almost independent of the steam to DME ratio. This is due to the excess steam allowing full conversion of the carbon monoxide in the forward WGSR (Eq. (9)). The slight decrease in the carbon dioxide selectivity at the ratio beyond 5 is mostly due to the WGSR reaching equilibrium, hence more carbon monoxide ending in the product gas. Similar trend has been previously demonstrated experimentally [6], therefore, operating at high steam to DME ratio is favoured for reduced carbon monoxide concentration. Fig. 8b shows the effect of the steam to DME ratio on the product gas composition. It is clear that the changes are insignificant despite of the considerable increase in the steam to DME ratio. The DME, Methanol and carbon monoxide are relatively more sensitive to this parameter, but they exist in the final product in very small quantities. It is therefore reasonable to conclude that changing the steam to DME mass ratio beyond 5 is unjustifiable in terms of increasing the hydrogen product, bearing in mind that increasing the steam implies higher operation cost.

Fig. 9 e Comparison between this study and the literature data on the effect of temperature on (a) the DME conversion and (b) the hydrogen yield. This study conditions: space ¡1 and steam to DME molar velocity ¼ 37,104 ml g¡1 cat h ratio ¼ 5. Yan et al. (2014) [5] conditions: space ¡1 and steam to DME molar velocity ¼ 3600 ml g¡1 cat h ratio ¼ 5. Feng et al. (2009) [6] condition: space velocity ¡1 and steam to DME molar ratio ¼ 3.5. 2461 ml g¡1 cat h

Comparison with the literature data Studies on catalytic DME steam reforming, either experimental or theoretical, have so far mainly been focused on fixed and bubbling bed type of reactors. In part 1 of this study [1], the numerical model predictions were compared with experimental data obtained in a bubbling fluidized bed reactor at one selected operating condition. Here, comparison is made against wider literature data produced in different types of reactors (micro-reactor and fixed bed) with the same catalyst considered here. This comparison is only meant to confirm the model capabilities for such analysis and to compare the parametric trends analysis discussed above rather than to validate the numerical data itself. Fig. 9 shows the reported literature data in comparison with this study predictions of the DME conversion and hydrogen yield as function of temperature. The trend is generally in good agreement showing consistent increase in the DME conversion and hydrogen yield for the temperature within the range of 200e250  C. The literature data show the

Please cite this article in press as: Elewuwa FA, Makkawi YT, A computational model of hydrogen production by steam reforming of dimethyl ether in a large scale CFB reactor. Part II: Parametric analysis, International Journal of Hydrogen Energy (2016), http:// dx.doi.org/10.1016/j.ijhydene.2016.08.072

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y x x x ( 2 0 1 6 ) 1 e1 0

curve for both parameters to level off or reach a peak within the temperature range of 250e275  C, thus suggesting reaching overall chemical equilibrium within this range. Another factor, that has been noted earlier, is the possible catalyst decay at high temperature, leading to slower conversion in practical experimental test. In this study, the DME conversion and hydrogen yield appear to approach 100% at a higher temperature around 350  C. It should be noted that both, the micro reactor and fixed bed, have been reported to be operating at a much lower space velocity, hence having longer gas residence time, and consequently better conversion and yield at lower temperature compared to the circulating fluidized bed reactor considered in this study. Fig. 10 shows the predicted variation of DME conversion and hydrogen yield with respect to steam to DME ratio in comparison with the literature data. Again, it is confirmed that the trends are generally in good agreement, despite of the recognized differences in the type of reactors and condition used. Clearly, all reactors show increased conversion and hydrogen

Fig. 10 e Comparison between this study and the literature data on the effect of steam to DME ratio on (a) the DME conversion and (b) the hydrogen yield. This study ¡1 and conditions: space velocity ¼ 37,104 ml g¡1 cat h  temperature ¼ 300 C. Yan et al. (2014) [5] operating ¡1 and conditions: space velocity ¼ 3600 ml g¡1 cat h  temperature ¼ 240 C. Feng et al. (2009) [6] operating ¡1 and condition: space velocity ¼ 2461 ml g¡1 cat h  temperature ¼ 240 C.

9

yield with increasing the steam, but with little evidence of reaching chemical equilibrium at these conditions, even at the high range of steam to DME ratio beyond 6.5. This confirm that the temperature is the dominant factor in shifting the reactions towards equilibrium. A reactor operated at a temperature and a steam to DME ratio around 300  C and 5.5, respectively, appear to be a common ground for achieving high DME conversion and hydrogen yield.

Conclusions This study presented a theoretical parametric analysis of hydrogen production by DME steam reforming in the riser section of a circulating fluidized bed reactor. The theoretical model used has been developed and validated in Part I of this study. The problem of flow non uniformity, associated with the entry effect (single sided solid feeding) has been found to be diminished by changing the inlet to double sided feeding. This resulted in more uniform and symmetric flow structure across the whole reactor. However, such an improved hydrodynamics has been found to have negligible effect on the reactant conversion and product gas composition. The gas residence time, investigated through changing the reactor height, suggests that a gas residence time of around 4 s is sufficient to reach chemical equilibrium. The analysis on the temperature effect clearly indicates that this parameter plays the dominant role in enhancing the hydrogen production. The hydrogen yield and conversion have been found to linearly increase with increasing the inlet temperature within the range of 200e350  C, reaching 82 mol% at the temperature of 350  C. However, the results at such a high temperature should be treated with caution due to the possible catalyst deactivation, which is ignored in the model used in this study. Comparison with the reported literature data on various types of reactors suggest that circulating fluidized bed reactor approaches chemical equilibrium at a higher temperature and steam to DME ratio, and this is mainly attributed to the high space velocity used in the circulating fluidized bed reactor. However, there is agreement in terms of the general trends with respect to variation of steam to DME ratio or temperature. Finally, based on the parametric analysis conducted here and taking into consideration the interrelated effects of the various parameters it is recommended to run the circulating fluidized bed reactor around the temperature of 300  C and steam to DME ratio around 5.5. These conditions achieved high hydrogen yield and concentration in the product gas.

Acknowledgement This research has been financially supported by the EPSRC (Ref EP/J501797/1).

Nomenclature Cp FAo

specific heat capacity molar flow rate of DME (mol s1)

Please cite this article in press as: Elewuwa FA, Makkawi YT, A computational model of hydrogen production by steam reforming of dimethyl ether in a large scale CFB reactor. Part II: Parametric analysis, International Journal of Hydrogen Energy (2016), http:// dx.doi.org/10.1016/j.ijhydene.2016.08.072

10

i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y x x x ( 2 0 1 6 ) 1 e1 0

g gravity (m s2) h enthalpy (kg m2 s2) DHoRX ; DHRX heat of formation and reaction respectively (kg m2 s2 mol1) heat supplied to the reactor (kg m2 s3) Qr Q rate of heat exchange between solid and gas phase (kg s3 m) q heat flux (kg s3) diffusion coefficient of granular energy (kg m1 s1) ks P pressure (Pa) R rate of heat transfer due to chemical reaction (kg s3 m) S volumetric mass transfer rate (kg m3 s1) reactor temperature and reference temperature T, TR respectively ( C) v velocity (m s1) X conversion () Greek letters a volume fraction () b solidegas momentum exchange (drag) coefficient (kg m3 s1) g collisional energy dissipation (kg s3 m) mols of component i per mole of DME () Qi granular temperature of solid phase i (m2 s2) qsi r densities respectively (kg m3) t shear stress tensor (kg m1 s2) ∅ energy exchange between the gas and solid phase (kg s3 m) l latent heat (kg m2 s2 mol1) Subscript g gas phase s solid phase

references

[1] Elewuwa F, Makkawi Y. Hydrogen production by steam reforming of DME in a large scale CFB reactor: part I: computational model and predictions. Int J Hydrogen Energy 2015;40:15865e76. http://dx.doi.org/10.1016/ j.ijhydene.2015.10.050. [2] Shi Y, Du X, Yang L, Sun Y, Yang Y. Experiments on hydrogen production from methanol steam reforming in fluidized bed reactor. Int J Hydrogen Energy 2013;38:1397e481. http:// dx.doi.org/10.1016/j.ijhydene.2013.08.073. [3] Vicente J, Eren J, Oar-arteta L, Olazar M, Bilbao J, Gayubo AG. Effect of operating conditions on dimethyl ether steam reforming in a fluidized bed reactor with a CuO- ZnO- Al2O3 and desilicated ZSM-5 zeolite bifunctional catalyst. Ind Eng Chem Res 2014;53:3462e71. [4] Amirshaghaghi P, Eliassi A, Taghizadeh M. Hydrogen production by steam reforming of dimethyl ether over Cu/ ZnO/Al2O3 and H-ZSM-5 catalysts: an experimental and modeling study. Iran J Hydrogen Fuel Cell 2014;3:153e61. [5] Yan F, Ye W, Guo Q, Huang L, Li B, Luo M. Numerical simulation and experimental study of hydrogen production from dimethyl ether steam reforming in a micro-reactor. Int J Hydrogen Energy 2014;39:18642e9. http://dx.doi.org/10.1016/ j.ijhydene. 2014.02.133.

[6] Feng D, Zuo Y, Wang D, Wang J. Steam reforming of dimethyl ether over coupled ZSM-5 and Cu-Zn-based catalysts. Chin J Catal 2009;30:223e9. http://dx.doi.org/10.1016/S1872-2067(08) 60098e4. [7] Kawabata T, Matsuoka H, Shishido T, Li D, Tian Y, Sano T, et al. Steam reforming of dimethyl ether over ZSM-5 coupled with Cu/ZnO/Al2O3 catalyst prepared by homogeneous precipitation. Appl Catal A Gen 2006;308:82. [8] Iulianellia A, Ribeirinhab P, Mendesb A, Basilea A. Methanol steam reforming for hydrogen generation via conventional and membrane reactors: a review. Renew Sustain Energy Rev 2014;29:355e68. [9] Dybkjir I, Hansen JB. Large scale production of alternative synthetic fuels from natural gas. Book series: studies in surface science and catalysis, vol. 107; 1997. p. 99. [10] Li C, Gao Y, Wu C. Modeling and simulation of hydrogen production from dimethyl ether steam reforming using exhaust gas. Int J Energy Res 2015;39:1272e9. http:// dx.doi.org/10.1002/er.3330. [11] Luo Z, Li C. Simulation and analysis of hydrogen production by dimethyl ether steam reforming for PEMFC. J Power Energy Eng 2016;4:25e30. http://dx.doi.org/10.4236/ jpee.2016.45004. [12] Takeishi K. Steam reforming of dimethyl ether. Appl Catal A General 2004;260:111e7. http://dx.doi.org/10.1016/ j.apcata.2003.10.006. [13] Faungnawakij K, Tanaka Y, Shimoda N, Fukunaga T, Kikuchi R, Eguchi K. Hydrogen production from dimethyl ether steam reforming over composite catalysts of copper ferrite spinel and alumina. Appl Catal B Environ 2007;74:144e51. [14] Yang M, Men Y, Li S, Chen G. Hydrogen production by steam reforming of dimethyl ether over ZnOeAl2O3 bi-functional catalyst. Int J Hydrogen Energy 2012;37:8360e9. http:// dx.doi.org/10.1016/j.ijhydene.2012.02.070. [15] Li C, Wang Y, Fan P. Numerical analysis and experimental study of hydrogen production from dimethyl ether steam reforming. Sci China Chem 2012;55:1982e7. http://dx.doi.org/ 10.1007/s11426-012-4603-0. [16] Gayubo AG, Vicente J, Eren J, Oar-Arteta L, Azkoiti MJ, Olazar M. Causes of deactivation of bifunctional catalysts made up of CuO-ZnO-Al2O3 and desilicated HZSM-5 zeolite in DME steam reforming. Appl Catal A General 2014;483:76e84. [17] Suh JS, Lee MT, Greif R, Grigoropoulos CP. A study of steam methanol reforming in a microreactor. J Power Sources 2007;173:458e66. http://dx.doi.org/10.1016/j.jpowsour. 2007.04.038. [18] Nehe P, Kumar S. Methanol reformation for hydrogen production from a single channel with cavities. Int J Hydrogen Energy 2013;38:13216e29. http://dx.doi.org/ 10.1016/j.ijhydene.2013.07.119. [19] Fogler H, Scott. Elements of chemical reaction engineering. 4th ed. New Jersey: Pearson Education, Inc; 2006. [20] Vinayak S. Sutkar, Tom J. K. van Hunsel, Niels G. Deen, Vitalij Salikov, Sergiy Antonyuk, Stefan Heinrich, and J.A.M. Kuipers, Experimental investigations of hydrodynamics of a spout fluidized bed with draft plates. in “The 14th international conference on fluidization e from fundamentals to products”, Eds, ECI symposium series, Volume (2013). [21] Shah Milinkumar T, Utikar Ranjeet P, Evans Geoffrey M, Tade Moses O, Pareek Vishnu K. Effect of inlet boundary conditions on computational fluid dynamics (CFD) simulations of gas-solid flows in risers. Industrial Eng Chem Res 2012;51:1721e8. http://dx.doi.org/10.1021/ie200940f.

Please cite this article in press as: Elewuwa FA, Makkawi YT, A computational model of hydrogen production by steam reforming of dimethyl ether in a large scale CFB reactor. Part II: Parametric analysis, International Journal of Hydrogen Energy (2016), http:// dx.doi.org/10.1016/j.ijhydene.2016.08.072