Supercritical water gasification of glycerol: Intermediates and kinetics

J. of Supercritical Fluids 78 (2013) 95–102

Contents lists available at SciVerse ScienceDirect

The Journal of Supercritical Fluids journal homepage: www.elsevier.com/locate/supflu

Supercritical water gasification of glycerol: Intermediates and kinetics Simao Guo, Liejin Guo ∗ , Jiarong Yin, Hui Jin State Key Laboratory of Multiphase Flow in Power Engineering (SKLMF), Xi’an Jiaotong University, Xi’an 710049, Shaanxi, China

a r t i c l e

i n f o

Article history: Received 6 December 2012 Received in revised form 21 March 2013 Accepted 22 March 2013 Keywords: Supercritical water gasification (SCWG) Kinetics model Glycerol Hydrogen production

a b s t r a c t In this paper, the liquid products from supercritical water gasification (SCWG) of glycerol were analyzed and some intermediates were identified. A simplified reaction pathway for gases production from SCWG of glycerol was proposed. The first quantitative kinetics model for describing the gaseous products (H2 , CO, CH4 and CO2 ) of SCWG of glycerol was developed. The model comprises seven reactions to describe the typical reactions in SCWG, and the reaction rate constant of each reaction was obtained by using the nonlinear least-square fitting method. The reaction rate analysis showed that the main sources of hydrogen yield were glycerol pyrolysis and steam reforming of intermediates, while the hydrogen yield from water–gas shift reaction (WGSR) was very small. The temperature estimated by the kinetics model for completely SCWG of glycerol solution was given. In addition, the sensitivity analysis of rate constant of WGSR was done based on the model. © 2013 Elsevier B.V. All rights reserved.

1. Introduction As the main byproduct of biodiesel industry, glycerol attracts more and more attention. How to realize high-value added utilization of these surplus crude glycerol becomes a hot topic in recent years. Besides producing high-value added chemical reagents [1] from glycerol, gasification of these biodiesel glycerol for renewable hydrogen production may be another promising way [2]. Among various gasification types, supercritical water gasification (SCWG) is considered to be suitable for gasification of wet biomass and organic wastes, due to the unique properties of supercritical water (SCW) [3–6]. In our previous work [7], the effects of various operating parameters (temperature, residence time, concentration and alkali catalysts) on SCWG of glycerol were systematically studied, and the results showed that SCWG of glycerol for hydrogen production had good prospects. The literatures about SCWG of glycerol was increasing in recent years. However, most of these researches focused on the effects of operating parameters on SCWG of glycerol and development of related catalysts [8–12]. Very few study involved the reaction mechanisms and no quantitative kinetics model was reported for individual gaseous products (H2 , CO, CH4 and CO2 ) of SCWG of glycerol. Antal et al. [13] first studied the glycerol decomposition in SCW at 500 ◦ C and 34.5 MPa. They found that the free radical chemistry may play a dominant role in glycerol

∗ Corresponding author. Tel.: +86 29 8266 3895; fax: +86 29 8266 9033. E-mail address: [email protected] (L. Guo). 0896-8446/$ – see front matter © 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.supflu.2013.03.025

decomposition in higher temperatures, and gave the reaction pathway for acetaldehyde formation. Buhler et al. [14] proposed that two competing reaction pathways (ionic reactions and free radical reactions) were existent when glycerol decomposed in nearand supercritical water, and offered a detailed reaction pathways and kinetics. However, their study was in a lower temperature range (349–475 ◦ C) and the gases were not the main products. Byrd et al. [9] reported glycerol gasification in SCW with Ru/Al2 O3 as a catalyst and gave an overall catalytic glycerol conversion kinetics. May et al. [10] used Ru/ZrO2 to catalyze SCWG of glycerol, but the gaseous products were very low. They also analyzed the liquid products and gave a reaction pathway. Chakinala et al. [8] gasified glycerol as a model compound of microalgae in SCW, and gave an overall glycerol carbon gasification kinetics. Our previous work [7] also gave an overall glycerol carbon gasification kinetics. All of the above studies were related to reaction mechanisms of SCWG of glycerol. However, some references only drew reaction pathways without specific reactions and quantitative research. Other references only gave overall reaction kinetics for glycerol conversion or carbon gasification. The quantitative kinetics model which can describe the individual gaseous products of SCWG of glycerol has not been reported yet. In fact, not only for glycerol, but most of SCWG studies generally lack quantitative kinetics model for individual gaseous products, and the main concern on kinetics study of SCWG at present were overall feedstock conversion [8,15], some typical liquid products and carbon distribution [16,17]. The Savage’s team first focused on the kinetics for describing individual gaseous products, making the kinetics study on SCWG had more practical value. They studied extensively on such

96

S. Guo et al. / J. of Supercritical Fluids 78 (2013) 95–102

quantitative kinetics models for SCWG of cellulose, lignin [18] and algae [19]. Goodwin et al. [20] modeled the typical liquid products and individual gaseous products separately for SCWG of xylose. Picou et al. [21] modeled the gas yield of SCWG of jet fuel based on the Sequential Reaction Mechanism. To date, this kind of research for SCWG was extremely limited. From the chemical engineering perspective, the research of quantitative kinetics model for SCWG must be intensified. In this work, intermediates of SCWG of glycerol (487–600 ◦ C, 25 Mpa) were identified and a quantitative kinetics model describing the individual gaseous products for SCWG of glycerol was developed. The model parameters were determined by fitting to the experimental data using nonlinear least square fitting method. By this model, the analysis of rates of reactions was done to find the detailed information for formations of different gases, and the temperature for completely SCWG of glycerol was estimated. In particular, the sensitive analysis of water–gas shift reaction (WGSR) rate constant was also investigated. These studies may be helpful to design a SCWG process for converting biodiesel glycerol to hydrogen-rich gas.

2.4. Data interpretation

where V is the reactor volume, Vo is volumetric flow rate of reaction flow, L is the length of reactor (m), A is the inner cross-section area (m2 ),  is the density of the reaction flow which was assumed to be the pure water at the reaction condition (kg m−3 ), Q is the mass flow rate of reaction flow (kg s−1 ). The carbon gasification efficiency (CE), gasification efficiency (GE), and hydrogen gasification efficiency (HE) were defined as the following equations:

2. Experimental

The residence time was estimated by the equation: =

V LA = Vo Q

(1)

CE =

the total carbon in the product gas the total carbon in the glycerol feed

(2)

GE =

the total mass of the product gas the total mass of the glycerol feed

(3)

HE =

total hydrogen in the product gas the total hydrogen in the glycerol feed

(4)

2.1. Materials 3. Results and discussion The glycerol (>99.0 wt%) used in this study was produced by Tianjin Fu Chen Chemical Reagent Factory. Anhydrous K2 CO3 was provided by Tianjin TianLi Chemical Reagents Ltd. All these reagents were analytical pure. 2.2. Apparatus and procedures The experiments were conducted in a continuous SCWG system developed in our laboratory. The characteristic of this system was that the feedstock solution was first mixed with preheated water before the heated tubular reactor, to realize fast heating and try best to keep the reaction fluid isothermal. The temperature mentioned in this work was the fluid temperature. The details of this system have been described in our previously work [22,23]. It is worth noting that when the reaction temperature was very high (e.g. 600 ◦ C), it is hard for tubular reactor to achieve isothermal state, because the relatively lower temperature range existed at the front of the reactor which was not sufficiently heated. For this reason, the kinetics model developed in this work may underestimate the reaction rate at a specific temperature.

3.1. Reaction intermediates The aqueous effluent was pretreated by SPE technology and identified by GC/MS method. As shown in Table 1, acetaldehyde, propionaldehyde, acrolein, ally alcohol, hydroxyacetone, propanoic acid and unreacted glycerol were found in aqueous effluent. These liquid products were consistent with the previous reports [10,14]. The methanol, formic acid and formaldehyde reported in literatures [10,14] were not found in our detection. This may be because their peaks in GC were very small and immersed in the solvent peak, or these C1 compounds were easily gasified completely in SCW at our reaction condition. Quantitative analysis of these liquid intermediates was beyond the scope of this work due to two reasons. First, some intermediates (e.g. acrolein and ally alcohol) are controlled in China, so the standard substances of some compounds are not commercially available. Second, the purpose of this work focused on modeling the individual gaseous products rather than liquid intermediates. The detailed formation/consumption pathways of these intermediates can be found in other work [10,13,14]. 3.2. Reaction pathway and kinetics model

2.3. Sample analysis The gaseous products were separated and measured by an Agilent 7890A gas chromatograph (GC) equipped with a thermal conductivity detector (TCD). The capillary column C-2000 in GC was purchased from LanZhou Institute of Chemical Physics in China. The carrier gas of GC was high purity helium with a flow rate of 30 ml min−1 . The total carbon amount (TOC) in the liquid effluent was determined by Elemental High TOCII. Solid phase extraction (SPE) technique was used for pretreatment of the liquid effluent before GC/MS analysis. About 10 ml effluent was filtered through preconditioned Agilent SampliQ C8 column. After the column was dry, the compounds were eluted from the column by 2 ml acetonitrile as the eluting agent. Then, the organic extract sample (1 ␮l) was injected into the GC (Agilent 6890) with HP-INNOWAX capillary column. A mass selective detector (MSD5973) was used to identify the organics in the sample.

According to previous works [10,13,14,24] and the intermediates above, a simplified reaction pathway was proposed and shown in Fig. 1. The char and tar formation was not be considered in our reaction pathway due to two reasons. First, no tar or char was observed obviously in our experimental range of SCWG of glycerol and the effluent was colorless, which was consistent with Chakinala’s work [8]. Second, in each experiment for constructing the kinetics model, the sum of carbon amount in liquid (obtained from TOC) and gas accounted for more than 89% of total carbon amount in feedstock, indicating if the tarry material existed, its amount would be very small. Besides the neglect of tar and char formation, a key concept in our kinetics model was using lumped parameter method to handle the various intermediates, and the various intermediates were lumped together and denoted as “Int”. This lumped parameters method is advisable for handling various intermediates [18,19], because our work only focused on modeling the individual gaseous

S. Guo et al. / J. of Supercritical Fluids 78 (2013) 95–102

97

Table 1 Identified liquid intermediate products by GC/MS from supercritical water gasification of glycerol. Retention time (min)

Compound

1.52

Acetaldehyde

1.85

Propionaldehyde

2.24

Acrolein

8.31

Ally alcohol

Structure

Formula

O

C2 H4 O

O C3 H6 O

O

C3 H4 O

OH

C3 H6 O

O 13.75

OH

Hydroxyacetone

C3 H6 O2

O

22.2

Propanoic acid

C3 H6 O2 OH OH

41.0

Glycerol

products rather than individual intermediates. In fact, in a complex reaction process, it was hard to quantitatively describe all of intermediates in detail, and the most common way to get quantitative kinetics describing the relationship between feedstock and products was lumped parameter method. Unfortunately, once the lumped parameter method was adopted, the subjectivity involved in defining lumped intermediates and stoichiometry of reaction equations could not be avoided. For example, in order to balance chemical reaction equation, we assumed the intermediates had an average molecular formula of C2 H4 O, and this assumption was actually made based on two considerations. First, it seemed to be more reasonable that lumping the intermediates as two carbon atoms compound (C2) than that of one (C1) or three (C3) carbon atoms. Because at our experimental temperature range, C1 may be easily gasified, and the C2 were more typical intermediates than C3 due to the broken of glycerol C C bond by free radical pathway [14]. Second, there was evidence that the C2 H4 O (acetaldehyde) was the main liquid product in SCWG of glycerol [13,14,25]. Although the assumption may be insufficient, it was necessary and effective for

HO

C3 H8 O3

OH

establishing the lumped kinetics model, and the same situation and handing also can be found in other work [18–20]. Based on the reaction pathway shown in Fig. 1 and the assumption of lumped intermediate “Int”, we inferred the gaseous products came from three sources: (1) The pyrolysis of glycerol. When the glycerol solution was mixed with preheated water and heated in the reactor, dehydration and pyrolysis reactions occurred. Once the temperature was high enough, the glycerol and other dehydrated C3 compounds (e.g. hydroxyacetone) were pyrolyzed, generating C1 and C2 compounds(e.g. formaldehyde and acetaldehyde [13]). Some CO, CO2 , H2 may be also released due to the broken of glycerol C C bond. Reactions 1 and 2 were used to describe these processes in the kinetics model. Reaction 1. Glycerol pyrolysis I.

K1

C3 H8 O3 −→Int + CO2 + 2H2 ,

R1 = K1 CC3 H8 O3

Fig. 1. Simplified reaction pathway for gases production by supercritical water gasification of glycerol.

(5)

S. Guo et al. / J. of Supercritical Fluids 78 (2013) 95–102

Reaction 2. Glycerol pyrolysis II. K2

C3 H8 O3 −→Int + CO + H2 + H2 O,

R2 = K2 CC3 H8 O3

a

H equilibrium value

(6)

3.0

Int + H2 O−→2CO + 3H2 ,

R3 = K3 CInt CH2 O

-1

(2) The steam reforming of intermediates. Although some references gave reaction equations that the feedstock (e.g. glucose) can be directly reformed by steam reforming reaction [15,26], it seemed to be more reasonable that only small molecular fragments from initial feedstock can be directly steam reformed easily. Thus, the steam reforming of intermediates was considered, and two types of steam reforming reactions were developed and shown by reactions 3 and 4. Reaction 3. Intermediates steam reforming I. K3

3.5

Gas yield / mol mol glycerol

98

(7)

2.5

1.5

0.5 0.0 0

K5

R5 = K5 CInt

(9)

(4) Once the gaseous products were generated, the reactions between different gases may occur. Only WGSR and methanation were taken into account in this work [18,20]. Despite the fact that these two reactions were known as reversible reactions, the reaction rate constant of reverse reactions were very small due to the large equilibrium constants (e.g. 1.2 × 103 for WGSR and 4.9 × 104 for methanation at 567 ◦ C) calculated by using the method from Lu and Yan [27]. So, we neglected the reverse reactions of WGSR and methanation. The result from Resende’s [18] report also showed that this ignorance was reasonable. Reaction 6. Water–gas shift reaction (WGSR) K6

CO + H2 O−→CO2 + H2 ,

R6 = K6 CCO CH2 O

(10)

Reaction 7. Methanation. K7

CO + 3H2 −→CH4 + H2 O, R7 = K7 CCO CH2

(11)

Reactions 1–7 may well incarnate the typical reactions occurred in SCWG. The reaction rates Ri (i = 1–7) for the reactions were assumed to be of first order in concentrations of each species (CC3 H8 O3 , CC2 H4 O , CH2 , CCO , CCO2 , CCH4 , CH2 O ). Then, the variations of concentrations of species with time can be given as follows: dCC3 H8 O3

= −R1 − R2

(12)

dCInt = R1 + R2 − R3 − R4 − R5 dt

(13)

dt

dCH2 dt

= 2R1 + R2 + 3R3 + 5R4 + R6 − 3R7

dCCO = R2 + 2R3 + R5 − R6 − R7 dt dCCO2 dt dCCH4 dt dCH2 O dt

4 6 Residence time / s

8

(8)

(3) The pyrolysis of intermediates. The lumped intermediate “Int” also can be pyrolyzed, and generated gases. The reaction 5 was used to describe this process. Reaction 5. Intermediate products pyrolysis. Int−→CO + CH4 ,

2

(14)

b

4.5 H equilibrium value

4.0 3.5 3.0

-1

R4 = K4 CInt CH2 O

Gas yield / mol mol glycerol

K4

CH equilibrium value

1.0

Reaction 4. Intermediates steam reforming II. Int + 3H2 O−→2CO2 + 5H2 ,

CO equilibrium value

2.0

2.5

CO equilibrium value

2.0 1.5 1.0

CH equilibrium value

0.5 0.0 0

2

4 6 Residence time / s

8

Fig. 2. Model and experimental results for variation of gas yields with residence time from supercritical water gasification of 10 wt% glycerol solution (a) 525 ◦ C (b) 567 ◦ C. ( ) Experimental H2 yield, ( ) experimental CO yield, ( ) experimental CH4 yield, ( ) experimental CO2 yield, ( ) calculated H2 yield, ( ) calculated CO yield, ( ) calculated CH4 yield, ( ) calculated CO2 yield. The solid purple lines were gas yields of thermodynamic equilibrium values. (For interpretation of the references to color in figure legend, the reader is referred to the web version of the article.)

The nonlinear least square fitting method (using functions in MATLAB software) was used to fit the experimental data (SCWG of 10 wt% glycerol at 487, 525, 567 and 600 ◦ C, residence time 3.9–9.0 s, and some data was from our previous work [7]) and minimize the value of the objective function which was the unweighted sum of the squared differences between measured and calculated concentration values of four gases (H2 , CO, CH4 and CO2 ). Table 2 shows the rate constants K1 –K7 at 487, 525, 567 and 600 ◦ C based on data fitting. As expected, the rate constants increased as the temperature increased. The apparent activation energy Ea and apparent pre-exponential factor A for each reaction were obtained according to Eq. (19), and the uncertainty represents the standard deviation.

(15) ln K = ln A −

= R1 + 2R4 + R6

(16)

= R5 + R7

(17)

= R2 − R3 − 3R4 − R6 + R7

(18)

Ea 1 R T

(19)

Fig. 2 shows the experimental results of the fitted cases (525 and 567 ◦ C) along with the calculated results from the model. From Fig. 2, it can be seen that this model captured the gaseous products trends correctly and fitted the experimental data well. The maximum gas yields values of H2 , CO, CH4 , and CO2 when glycerol was completely gasified at 525 and 567 ◦ C were also given in Fig. 2, estimated by thermodynamic equilibrium values obtained from Lu and

S. Guo et al. / J. of Supercritical Fluids 78 (2013) 95–102

99

Table 2 Rate constants and Arrhenius parameters by data fitting.

(s ) (s−1 ) (L mol−1 s−1 ) (L mol−1 s−1 ) (s−1 ) (L mol−1 s−1 ) (L mol−1 s−1 )

487 ◦ C

525 ◦ C

567 ◦ C

600 ◦ C

Ea (kJ mol−1 )

0.0857 0.0450 0.0618 0.0425 0.350 7.17E−04 0.208

0.1219 0.0681 0.1273 0.0886 0.550 1.20E−03 0.347

0.200 0.105 0.372 0.215 0.928 2.20E−03 0.600

0.254 0.158 0.612 0.409 1.40 3.50E−03 0.988

53.3 59.8 114.1 109.6 66.7 76.5 74.3

Yan’s method [27]. As shown in Fig. 2, the equilibrium values of H2 , CH4 , and CO2 were higher than the experimental values, and the CO yield exceeds the equilibrium values (nearly 0 and not shown in Fig. 2), indicating that the experimental result was far away from the thermodynamic equilibrium state. Besides the good fitting to the experimental data, it is expected that the kinetics model can give reasonable reaction constant rates rather than only mathematical fitting results. Unfortunately, all reactions mentioned in the model have not been experimentally studied except WGSR. Fig. 3 shows an Arrhenius plot of our WGSR rate constants K6eff , and compares our results to the previous experimental work [28–31]. Here, for comparison, K6eff = K6 CH2 O to keep consistent with other work. It can be seen from Fig. 3 that our fitting results fell on the region between the results of Holgate [28] and Sato [31], indicating that our fitting results of WGSR was reasonable. For pyrolysis reactions 1 and 2, the active energies were 53.3 and 59.8 kJ mol−1 , respectively. While the active energies of the steam reforming reactions 3 and 4 were both above 100 kJ mol−1 , which were much larger than those of pyrolysis reaction, indicating that steam reforming reactions were more sensitive to the temperature than pyrolysis reactions, and higher temperature may be more beneficial to steam reforming reactions than pyrolysis reactions. Picou [21] recently performed a kinetics study on SCWG of jet fuel, and obtained the active energies of pyrolysis and steam reforming were 60.6 and 197.3 kJ mol−1 , respectively, which were similar to our results.

-3 -4

lnk

-5 -6 -7 -8 -9 -10 1.1

1.2

1.3

1.4

1.5

1.6

-1

1000/T / K

Fig. 3. Arrhenius plot comparing the first-order rate constants k for water–gas shift ) reaction. ( ) This work K6eff . Lines were plotted based on references, ( ) Sato 1999 [30], ( ) Sato 2004 [31], ( ) Rice [29], ( Holgate [28].

ln A

1.7 2.8 4.1 1.8 1.7 1.7 3.3

6.00 6.37 15.28 14.18 9.51 4.86 10.19

± ± ± ± ± ± ±

0.25 0.42 0.61 0.27 0.26 0.26 0.49

4.0 3.5 3.0 2.5 2.0 1.5 1.0 0.5 0.0 0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

4.0

Experimental gas yield / mol mol glycerol Fig. 4. Model predictions for gas yields from supercritical water gasification of different glycerol solutions at 567 ◦ C, 5.8 s residence time. ( H2 yields; (

) represent CO yields; (

) Represent ) represent CH4 yields;

( ) represent CO2 yields. Different colors of symbols represent gas yields from different concentrations of glycerol solution. Red: 15 wt%; green: 20 wt%; blue: 50 wt%. (For interpretation of the references to color in figure legend, the reader is referred to the web version of the article.)

glycerol solution concentrations (15, 20 and 50 wt%). A comparison of gas yield between experiments and predictions was showed in Fig. 4. It can be seen from Fig. 4 that the model can predict the gas yield at various glycerol solution concentration with reasonable proximity.

-1

-2

± ± ± ± ± ± ±

-1

H2 formation/ comsuption rate(mol s )

3.2.1. Model validation A good kinetics model was required to have not only the ability of fitting but also the ability of prediction. The parameters of the model were all obtained from the data fitting based on experimental data of 10 wt% glycerol solution. In order to test the predictive ability of the model, we performed the experiments with various

-1

K1 K2 K3 K4 K5 K6 K7

−1

Calculated gas yield / mol mol glycerol

Rate constant

0.007 0.006 0.005 0.004 0.003 0.002 0.001 0.000 -0.001 -0.002 0

2

4 Residence time (s)

6

8

Fig. 5. Rates of formation/consumption for H2 (567 ◦ C, 10 wt% glycerol solution). ) Glycerol pyrolysis I, ( ) glycerol pyrolysis reaction II, ( ) intermediates steam reforming I, ( ) intermediates steam ( ) water–gas shift reaction, ( ) methanation. reforming II, (

S. Guo et al. / J. of Supercritical Fluids 78 (2013) 95–102

CO2 formation/ comsuption rate(mol s )

0.0040

-1

-1

CO formation/ comsuption rate(mol s )

100

0.0035 0.0030 0.0025 0.0020 0.0015 0.0010 0.0005 0.0000 -0.0005

0

2

4 Residence time (s)

6

0.0035 0.0030 0.0025 0.0020 0.0015 0.0010 0.0005 0.0000

0

8

2

4 Residence time (s)

6

8

Fig. 6. Rates of formation/consumption for CO (567 ◦ C, 10 wt% glycerol solu) Glycerol pyrolysis reaction II, ( ) intermediates steam tion). ( ) intermediates pyrolysis, ( ) water–gas shift reforming I, ( ) methanation. reaction, (

Fig. 8. Rates of formation for CO2 (567 ◦ C, 10 wt% glycerol solution). ( ) intermediates steam reforming II, ( Glycerol pyrolysis I, ( water–gas shift reaction.

3.3. Reaction rate analysis

H2 , and the fast type WGSR mentioned in literature [15,32] did not occur in our experimental condition. The rates of glycerol pyrolysis reaction I and II decreased with increasing residence time due to the decreasing glycerol concentration, while the rates of steam reforming I and II increased firstly and then decreased. This was because there were no intermediates at the beginning of the reaction. As the reactions progress, the intermediates were accumulated and the rates of steam reforming were increased. At last, with intermediates consumption and decreased supplement, the rates of steam reforming reactions decreased. The consumption of H2 was through methanation. Fig. 6 shows the formation/consumption of CO. The largest source of CO was intermediates steam reforming I. Two other sources were glycerol pyrolysis II and intermediates pyrolysis. Two consumption ways of CO were WGSR and methanation, the consumption rate of the latter was larger than that of the former. In general, the methane produced in SCWG is supposed to be not only from methanation but also from the pyrolysis of intermediates [18]. Our model contained these two formation pathways, and the sources of methane were from intermediates pyrolysis reaction and methanation. As shown in Fig. 7, at the beginning of the reaction, the rate of intermediates pyrolysis was larger than that of methanation. As the reaction progress and the accumulations of CO and H2 , the rate of methanation was increased. It is worth noting that the increase of methanation rate could not continue forever and the hydrogen could not be completely consumed because methanation is actually a reversible reaction. It took no account of reversible reaction of methanation in our model may over estimate the methanation reaction rate, especially in long residence time calculation. There are three sources for CO2 formation: glycerol prolysis I, intermediates steam reforming II and WGSR. The detailed variations of rates with time can be seen in Fig. 8.

-1

CH4 formation/ comsuption rate(mol s )

In the previous section, we developed and verified our kinetics model, showing that our model has the ability to describe the non-catalytic SCWG of glycerol under our explored experimental conditions. In this section, this model was applied to analyze the reaction rate of each reaction, and identify which reaction was most responsible for the different gases production. Figs. 5–8 display the variation of formation/consumption rates of H2 , CO, CH4 and CO2 with residence time. It can be seen from Fig. 5 that the main sources of H2 were from glycerol Pyrolysis I, intermediate products steam reforming I and II. The rates of these three reactions were at the same level. A large fraction of H2 was generated by steam reforming confirmed that the H atom in H2 came not only from the glycerol but also from the water. Moreover, as shown in Section 3.2, the steam reforming reactions was more sensitive to the temperature than pyrolysis reactions, which meant more H2 would be produced by steam reforming reactions than pyrolysis reactions at higher temperatures. Although it is believed that WGSR is an important reaction in SCWG [5,31], it can be seen from Fig. 5 that the rates of WGSR was very low, indicating that the WGSR was not the main source of

0.0012 0.0010 0.0008 0.0006

) )

0.0004

3.4. Estimation of temperature for completely gasification

0.0002 0.0000

0

2

4 Residence time (s)

6

Fig. 7. Rates of formation for CH4 (567 ◦ C, 10 wt% glycerol solution). ( ) methanation. Intermediates pyrolysis, (

8 )

It is meaningful to find out a high enough temperature for completely gasification of glycerol solution at a specific residence time. In this work, we investigated the CE, GE and HE of SCWG of 10 wt% glycerol solution at different temperatures (500, 525, 550, 575, 600, 625, 650 ◦ C) with 7 s residence time based on the kinetics model. According to the apparent activation energies Ea and apparent preexponential factors A given in Table 2, different rate constants at

S. Guo et al. / J. of Supercritical Fluids 78 (2013) 95–102

101

Table 3 Sensitivity analysis of water–gas shift reaction rate constant K6 . Gas yield (mol mol−1 glycerol) H2

CO

CH4

CO2

Experimental values [7] (525 ◦ C, 6.45 s, 10 wt% glycerol)

Non-catalytic 0.5 wt% K2 CO3

2.57 3.75

0.86 0.02

0.15 0.21

0.87 1.75

Calculated values by changing water–gas shift reaction rate constant K6 (L mol−1 s−1 )

K6 = 0.0012 K6 = 0.01 K6 = 0.1 K6 = 0.5

2.68 2.81 3.44 3.72

0.93 0.81 0.26 0.04

0.28 0.28 0.26 0.24

0.82 0.95 1.52 1.75

different temperatures for each reaction can be obtained. Then, the kinetics model was used to calculate gases yields, and the carbon gasification efficiency (CE), gasification efficiency (GE) and hydrogen efficiency (HE) can be known (the definitions of CE, GE and HE can be found in Section 2.4). It can be seen in Fig. 9 that when the temperature reached to 600 ◦ C, the CE, GE and HE were up to 94.1%, 107.7% and 111.6%, respectively, indicating that the temperature of 600 ◦ C was enough for almost completely gasification of 10 wt% glycerol solution with 7 s residence time. To test this prediction, we performed an experiment that SCWG of 10 wt% glycerol solution at 600 ◦ C with 6.8 s residence time. The CE, GE and HE obtained by experimental data were 91.3%, 102.2% and 108.0%, respectively, which were consistent with the kinetics model prediction and confirmed our model had some practical value. An interesting result can be found both in this verification experiment and Fig. 9 was that when the temperature was above 600 ◦ C, the GE and HE could exceed 100%. It did not mean that the model went wrong at higher temperatures. In fact, water in SCWG was not only a solvent but also a reactant, and some water was converted to gaseous products by steam reforming and WGSR. So, when rates of intermediates steam reforming and WGSR were high enough at higher temperatures, the total mass of gaseous products which contained extra mass of water may exceed the total mass of glycerol, making the GE exceed 100% according to the definition of GE in Eq. (3). The HE may exceed 100% for the same reason. As mentioned in Section 3.3, the H atom in H2 came not only from the glycerol but also from the water by steam reforming and WGSR, making the HE may exceed 100% in higher temperatures according to the definition of HE in Eq. (4). It also can be found in Fig. 9 that further increase in temperature only resulted in a slight increase in CG, GE and HE.

120

CE, GE and HE / %

110 100

3.5. Sensitive analysis of WGSR rate constant WGSR was considered to be an important reaction in SCWG, but in fact, as shown both in literatures and our results, the rate of the WGSR reaction in lower temperature was very low [28]. From the perspective of using H2 in fuel cells, due to CO poisoning of proton exchange membrane [33], the content of CO in hydrogen-rich gas from SCWG is expected to be as low as possible. Increasing the WGSR rate in SCWG may be a good way to convert CO to H2 , and the most common way to enhance WGSR is the alkali catalysts addition. In order to know how the yields of gaseous spices changed when the WGSR rate increased, a sensitive analysis of WGSR rate constant K6 was done based on the kinetics model. We gradually turned up the rate constant K6 (which was 0.0012 L mol−1 s−1 at 525 ◦ C with no alkali catalysts in Table 2.) to calculate yields of gases, while kept other reaction constants unchanged, meaning no matter what made the rate constant K6 increase, the rates of other reactions (reaction 1–5,7) were not changed. When rate constant K6 increased, the calculated results (shown in Table 3) showed that the yield of H2 and CO2 increased, while the yield of CO was significantly decreased. The yield of gases gradually drew close to the catalytic gasification results as the rate constant K6 increasing. In particular, it was interesting that when the rate constant K6 increased to 0.5, the calculated values of gases yields were almost the same as the catalytic gasification results. Here, it seemed to be that the alkali catalyst K2 CO3 promoted WGSR rate constant K6 from 0.0012 to 0.5. However, it was worth noting that this speculation actually based on the assumption that the alkali catalysts only affected on WGSR but not other reactions. This assumption was supported by reference [12] and our pervious experimental results of SCWG of glycerol [7], which found that alkali catalysts did not increase carbon conversion obviously. But more references showed opposite results. Although the inconsistency in references makes the assumption inadequate, the main purpose of showing this sensitive analysis of the WGSR rate constant is to give an initial concept of the rough order of magnitude changes of the WGSR rate constant with and without alkali catalysts, rather than an exact result.

90

4. Conclusion

80

In this work, the liquid products from SCWG of glycerol were identified and a simplified reaction pathway for gases production was proposed. The first quantitative kinetics model for describing the gaseous products of SCWG of glycerol was developed. This model comprised four typical reaction types in SCWG: pyrolysis, steam reforming, water–gas shift and methanation reaction. The nonlinear least-square fitting method was used to obtain the reaction rate constant of each reaction in the model by fitting to experimental data. The apparent activation energy and preexponential factor for each reaction were also given. By comparing active energies of steam reforming reactions with those of pyrolysis reactions, it seemed that the higher temperatures tend to be more beneficial to steam reforming reactions than pyrolysis reactions.

70 60 50 475

500

525

550 575 600 Temperature / ºC

625

Fig. 9. Model prediction values of carbon gasification efficiency ( GE) and hydrogen gasification efficiency ( cation efficiency ( different temperatures (10 wt% glycerol solution, 7 s residence time).

650

CE), gasifiHE) at

102

S. Guo et al. / J. of Supercritical Fluids 78 (2013) 95–102

The reaction rates analysis based on the model showed that the main sources of H2 production were glycerol pyrolysis and steam reforming of intermediate products. The rate of WGSR in the temperature range 487–600 ◦ C was very low, indicating the WGSR was not the main source of H2 . However, the sensitivity analysis showed that after adding alkali catalyst, the rate constant of WGSR may increase greatly. The temperature estimated by kinetics model for completely SCWG of 10 wt% glycerol solution was 600 ◦ C with 7 s residence time. Acknowledgements This work was financially supported by the National Basic Research Program of China (Contract No. 2009CB220000) and the National Natural Science Foundation of China (Contract No. 50821064). The authors thank Dr. Youjun Lu and Dr. Changqing Cao for their valuable suggestions. References [1] D.T. Johnson, K.A. Taconi, The glycerin glut: options for the value-added conversion of crude glycerol resulting from biodiesel production, Environmental Progress 26 (2007) 338–348. [2] S. Adhikari, S.D. Fernando, A. Haryanto, Hydrogen production from glycerol: an update, Energy Conversion and Management 50 (2009) 2600–2604. [3] P.E. Savage, A perspective on catalysis in sub- and supercritical water, Journal of Supercritical Fluids 47 (2009) 407–414. [4] A.A. Peterson, F. Vogel, R.P. Lachance, M. Froling, M.J. Antal, J.W. Tester, Thermochemical biofuel production in hydrothermal media: a review of sub- and supercritical water technologies, Energy and Environmental Science 1 (2008) 32–65. [5] L.J. Guo, Y.J. Lu, X.M. Zhang, C.M. Ji, Y. Guan, A.X. Pei, Hydrogen production by biomass gasification in supercritical water: a systematic experimental and analytical study, Catalysis Today 129 (2007) 275–286. [6] A. Kruse, Supercritical water gasification, Biofuels Bioproducts and BiorefiningBiofpr 2 (2008) 415–437. [7] S.M. Guo, L.J. Guo, C.Q. Cao, J.R. Yin, Y.J. Lu, X.M. Zhang, Hydrogen production from glycerol by supercritical water gasification in a continuous flow tubular reactor, International Journal of Hydrogen Energy 37 (2012) 5559–5568. [8] A.G. Chakinala, D.W.F. Brilman, W.P.M. van Swaaij, S.R.A. Kersten, Catalytic and non-catalytic supercritical water gasification of microalgae and glycerol, Industrial and Engineering Chemistry Research 49 (2010) 1113–1122. [9] A.J. Byrd, K.K. Pant, R.B. Gupta, Hydrogen production from glycerol by reforming in supercritical water over Ru/Al2 O3 catalyst, Fuel 87 (2008) 2956–2960. [10] A. May, J. Salvado, C. Torras, D. Montane, Catalytic gasification of glycerol in supercritical water, Chemical Engineering Journal 160 (2010) 751–759. [11] J.A. Onwudili, P.T. Williams, Hydrothermal reforming of bio-diesel plant waste: products distribution and characterization, Fuel 89 (2010) 501–509. [12] S.R.A. Kersten, B. Potic, W. Prins, W.P.M. Van Swaaij, Gasification of model compounds and wood in hot compressed water, Industrial and Engineering Chemistry Research 45 (2006) 4169–4177.

[13] M.J. Antal Jr., W.S.L. Mok, J.C. Roy, A.T. Raissi, D.G.M. Anderson, Pyrolytic sources of hydrocarbons from biomass, Journal of Analytical and Applied Pyrolysis 8 (1985) 291–303. [14] W. Buhler, E. Dinjus, H.J. Ederer, A. Kruse, C. Mas, Ionic reactions and pyrolysis of glycerol as competing reaction pathways in near- and supercritical water, Journal of Supercritical Fluids 22 (2002) 37–53. [15] I.G. Lee, M.S. Kim, S.K. Ihm, Gasification of glucose in supercritical water, Industrial and Engineering Chemistry Research 41 (2002) 1182–1188. [16] T.L.-K. Yong, Y. Matsumura, Reaction kinetics of the lignin conversion in supercritical water, Industrial and Engineering Chemistry Research 51 (2012) 11975–11988. [17] A. Chuntanapum, Y. Matsumura, Role of 5-HMF in supercritical water gasification of glucose, Journal of Chemical Engineering of Japan 44 (2011) 91–97. [18] F.L.P. Resende, P.E. Savage, Kinetic model for noncatalytic supercritical water gasification of cellulose and lignin, AIChE Journal 56 (2010) 2412–2420. [19] Q. Guan, C. Wei, P.E. Savage, Kinetic model for supercritical water gasification of algae, Physical Chemistry Chemical Physics 14 (2012) 3140–3147. [20] A.K. Goodwin, G.L. Rorrer, Reaction rates for supercritical water gasification of xylose in a micro-tubular reactor, Chemical Engineering Journal 163 (2010) 10–21. [21] J. Picou, J. Wenzel, A. Niemoeller, S. Lee, H.B. Lanterman, A kinetic model based on the sequential reaction mechanism for the noncatalytic reformation of jet fuel in supercritical water, Energy Source Part A 33 (2011) 785–794. [22] Y.L. Li, L.J. Guo, X.M. Zhang, H. Jin, Y.J. Lu, Hydrogen production from coal gasification in supercritical water with a continuous flowing system, International Journal of Hydrogen Energy 35 (2010) 3036–3045. [23] C.Q. Cao, L.J. Guo, Y.A. Chen, S.M. Guo, Y.J. Lu, Hydrogen production from supercritical water gasification of alkaline wheat straw pulping black liquor in continuous flow system, International Journal of Hydrogen Energy 36 (2011) 13528–13535. [24] A. Corma, G.W. Huber, L. Sauvanauda, P. O‘Connor, Biomass to chemicals: catalytic conversion of glycerol/water mixtures into acrolein, reaction network, Journal of Catalysis 257 (2008) 163–171. [25] J.B. Muller, F. Vogel, Tar and coke formation during hydrothermal processing of glycerol and glucose. Influence of temperature, residence time and feed concentration, Journal of Supercritical Fluids 70 (2012) 126–136. [26] M.J. Antal, S.G. Allen, D. Schulman, X.D. Xu, R.J. Divilio, Biomass gasification in supercritical water, Industrial and Engineering Chemistry Research 39 (2000) 4040–4053. [27] Q. Yan, L. Guo, Y. Lu, Thermodynamic analysis of hydrogen production from biomass gasification in supercritical water, Energy Conversion and Management 47 (2006) 1515–1528. [28] H.R. Holgate, P.A. Webley, J.W. Tester, R.K. Helling, Carbon-monoxide oxidation in supercritical water – the effects of heat-transfer and the water gas shift reaction on observed kinetics, Energy Fuels 6 (1992) 586–597. [29] S.F. Rice, R.R. Steeper, J.D. Aiken, Water density effects on homogeneous water–gas shift reaction kinetics, Journal of Physical Chemistry A 102 (1998) 2673–2678. [30] T. Sato, S. Kurosawa, T. Adschiri, K. Arai, Kinetics of the water–gas shift reaction in supercritical water, Kagaku Kogaku Ronbun 25 (1999) 993–997. [31] T. Sato, S. Kurosawa, R.L. Smith, T. Adschiri, K. Arai, Water gas shift reaction kinetics under noncatalytic conditions in supercritical water, Journal of Supercritical Fluids 29 (2004) 113–119. [32] X.D. Xu, Y. Matsumura, J. Stenberg, M.J. Antal, Carbon-catalyzed gasification of organic feedstocks in supercritical water, Industrial and Engineering Chemistry Research 35 (1996) 2522–2530. [33] J.J. Baschuk, X.G. Li, Carbon monoxide poisoning of proton exchange membrane fuel cells, International Journal of Energy Research 25 (2001) 695–713.