Kinetic Study on Continuous Sampling of Coal Char from a Micro Fluidized Bed.

We self-design a micro fluidized bed reactor (MFB) with combination of an online char particle sampling system to study the kinetics of coal char combustion and gasification. The system mainly contains two parts: a micro fluidized bed and vacuum online sampling. Vientiane coal was continuously sampled from the MFB. Both combustion and gasification reactivities of the sampled chars were tested in a thermogravimetric analyzer. Kinetic parameters of the sampled char were analyzed. Char reactivity in oxy-fuel combustion in the MFB obeys the rule of decrease-increase-decrease behavior with the sampling time. Pre-exponential factor A and activation energy E of the sampled char increase with the sampling time. The gasification reactivity of the sampled char increases with the sampling time even though there is a minor decrease in an initial gasification stage. The new designed MFB combining with the online sampling system will pave the path for the investigation of gas-solid reaction evolution in the future.

Introduction

Oxy-fuel combustion has many advantages over conventional air combustion.[1,2] In oxy-fuel combustion, O2 is usually separated from air and mixed with the recycled flue gas in a power plant. CO2 can reach as high as 95% in the flue gas, which makes CO2 capture easier. The cyclic flue gas contains high concentration of CO2, which is therefore compressed and stored underground. Oxy-fuel combustion has a good CO2 capture efficiency comparing to other technologies such as absorbing CO2 by solid sorbents.[3−5] As a result, good economic benefits can be realized in oxy-fuel combustion.[6]

Char structure evolution during combustion or gasification has been investigated by many researchers in recent years.[7−15] Char morphology and structure, i.e., size, thickness, porosity, and functional groups, have significant effects on char reactivity,[13−15] as well as other applications.[16−18]

Char is usually generated from a short pyrolysis stage in a boiler followed by combustion or gasification. Char morphology in the pyrolysis stage has been reported in our previous report[12] and hence not discussed in this study. This study investigates char morphology evolution and its impact on reactivity in the combustion and gasification stage.

During combustion, it was reported that char density and size gradually decrease,[19,20] while the char surface area increases due to the formation of a large amount of meso- and micropores.[19] However, in Liu’s calculation,[20] the specific surface area shows different rules during combustion, where it either first increases to a maximum and then decreases or only decreases to the end of the combustion. Apart from physical structure evolution, a preferential loss of functional groups may affect char reactivity.[21] It is reported that char combustion reactivity decreases with the proceeding of the combustion, which is due to the loss of the intrinsic reactivity[19] as well as the functional groups like a carbonaceous matrix.[21]

During gasification, the char surface area decreased rapidly and then was kept constant. The degree of char graphitization gradually increased during gasification by in situ Raman spectroscopy,[22,23] which may result in the decrease in the char reactivity.

However, the char structure and reactivity evolutions during coal combustion or gasification are mostly investigated in a fixed bed or a drop furnace in the previous literature. A fixed bed reactor has disadvantages of oxygen diffusion limitation and temperature deviation for gas–solid reaction measurements. A drop furnace is usually used to measure rapid reactions at high temperature. The disadvantages of the drop furnace are that it is hard to measure a nonrapid gas solid reaction and to realize an isothermal environment due to significant reduction in radiation heat transfer at low temperature. Hence, it is a good choice to measure char structure and reactivity evolutions in an MFB reactor.

The MFB has the advantage of mass and heat transfer for isothermal reactions and has been applied to investigate the kinetic study of gas–solid reactions. We have previously used an MFB to determine coal char combustion[24−26] and gasification[27,28] as well as petroleum coke gasification,[29] which shows good reliability. Zhao et al. used an MFB to determine steam gasification of in situ char.[30,31] Xu et al. used an MFB to measure several kinds of gas–solid reactions.[32,33] All these studies show very good reliability of the MFB for investigating gas–solid reactions.

In this study, we propose to use the MFB reactor combining with a continuous sampling system to investigate coal char reactivity evolution during combustion and gasification. Kinetic evolution is determined by an isothermal method. This study will pave the path for online sampling in the MFB for gas–solid reactions in the future.

Materials and Methods

Sample Preparation

A Vientiane coal sample was selected in this study. Raw coal was ground and sieved to a diameter of 75–100 μm. We have investigated the effect of char diameter on char conversion in such MFBs in our previous study[24] and found that the carbon conversion rate did not increase with decreasing the particle size below 100 μm, indicating that the internal diffusion limitation could be disregarded when the particle size was less than 100 μm.

Later, coal samples were put into the oven and dried at 378 K for 1 hour. The proximate and ultimate analysis are shown in our previous studies.[12,26]

Vacuum Online Sampling

In the MFB reactor, we could measure the char reactivity by detecting the outlet gas by gas chromatography (GC) and/or mass spectrometry (MS). Char reactivity identified from outlet gas measurement is the overall average reactivity, which cannot represent immediate reactivity during the reactions. Then, the question is how we can measure the immediate reactivity. Sampling a char out of the MFB reactor by an online solid particle sampling system and directly measuring its reactivity can be a good choice. To the best of our knowledge, the idea of combining the MFB with an online solid particle sampling system is first reported in this study.

We self-developed an MFB with combination of an online char particle sampling system to study coal char combustion and gasification. The schematic diagram of the experiment apparatus is shown in Figure , which mainly contains two parts: (left) a micro fluidized bed and (right) vacuum online sampling. The reactor in the MFB is in a dimension of 25 mm in inner diameter and 180 mm in height. Two wind boards with a distance of 50 mm in between are placed in the middle of the reactor in which char combustion takes place. The vacuum online sampling system mainly consists of a sampling tube, vacuum box, and vacuum pump. A thermocouple was placed in the sampling tube (not shown in Figure ) to measure the temperature at the sampling point in the MFB. The vacuum box is in a dimension of 50 × 30 × 20 cm. The inner diameter of the sampling tube is 3 mm.

Figure 1

Schematic diagram of integration of the MFB and online sampling system.

Sampling tube is inserted to the fluidized zone in the MFB. All valves were closed, and the sampling box was vacuumed to the pressure as lowest as 0.09 MPa within 10 s before starting online char sampling. Char particles were sampled at different reaction times to different bottles located in the sampling box by opening the corresponding valve. Once sampling was finished, a flow of cold Ar was injected to the sampling tube and sampling bottle to cool the sampled char. The cooled char was then collected for further analysis.

There are some other techniques solving the immediate reactivity during the reaction. For instance, a spatially resolved reactor was designed to solve intermediate reactions, catalyst structure evolutions, and subkinetic parameters from different parts along the axial in the sample bed.[34−36] In this technique, the reactor involves reactants flowing through a solid catalyst bed containing a sampling capillary with a side sampling orifice, where a very small amount of in situ gas is sampled and transferred into an analytical device (GC/MS) for quantitative analysis. The sampling capillary can be moved with μm resolution in or against the flow direction to measure species profiles through the catalyst bed. Some other researches propose to solve immediate species and structures by a so-called spatially resolved infrared radiation (IR) surface analysis by continuously moving a fixed bed.[37] However, all these techniques are measured from the fixed bed and have their own disadvantages.

Experimental Procedure

Raw coal was first pyrolyzed in a fixed reactor at 1173 K in an Ar atmosphere for 4 h to deeply remove the volatile. The pyrolysis experimental setting can be found in our previous study.[12]

For combustion in the MFB, the temperature was set to 823 K. A stream with 125 mL/min O2 and 375 mL/min CO2 was injected to the reactor. The sampling box was vacuumed to 0.09 MPa. The char sample (1 g) was then fed to the MFB by a programmable logical controller (PLC)-controlled Ar pulse after stabilization of temperature and flow stream. Open electromagnetic valve to start sampling at 0.5, 2.5, 5.5 and 10 min, respectively. The sampled char was collected (10–20 mg) and cooled down in the sample bottles, which were placed in the vacuum box. Sampled char surface morphology was measured by field emission scanning electron microscopy (FESEM) with a resolution of 0.8 nm (GeminiSEM 300).

Generally, it is better to use a milligram char sample in kinetic studies in the MFB. However, we have to sample the char particles as enough as possible out of the MFB at different reaction stages to further determine their kinetic parameters in a thermogravimetric analyzer (TGA). We have tried the experiments with the char sample of milligrams, and few char particles were sampled out and not enough to measure the kinetics in the later thermogravimetric (TG) experiment. Comparing with char sample in milligrams, we do not think the char sample mass (1 g) will affect the properties of the char because our char particles are totally fluidized in the MFB.

For gasification in the MFB, the temperature was set to 1123 K. Gasification is an endothermic reaction. It requires heat to start up and a long time to operate the gasification furnace in industrial application. The heat could be from the combustion reaction with supplying of low concentration of O2 (char–O2 reaction). Although in an experimental scale we electrically heat the reactor, we still add a low concentration (5%) of O2 to the reaction atmosphere to simulate the case in industrial application. In this way, we still call it gasification. Experimentally, a stream with 25 mL/min O2, 300 mL/min CO2, and 175 mL/min steam was injected to the reactor. The sampling times selected were 0.5, 1.0, 1.5, and 2.0 min. Other experimental settings are the same with the combustion experiment.

Kinetic Determination

The reactivity of sampled char from combustion in the MFB was determined in a TGA (STA 449 F3) at 773, 823, and 873 K in an O2/Ar atmosphere (21 mL/min O2, 79 mL/min Ar). TGA temperature was first increased to the target temperature at a rate of 10 K/min in an Ar atmosphere. The atmosphere was then shifted to O2/Ar after temperature stabilization. In each measurement, the mass of the sampled char was 5 mg. The mass loss was continuously recorded.

The reactivity of sampled char from gasification in the MFB was determined in the same TGA at 1273 K but with a vapor furnace reactor. Temperature was set from 423 to 1273 K at a heating rate of 20 K/min in an N2 atmosphere. Then, a stream of mixed CO2, steam and N2 with varied concentrations was injected to the reactor.

Kinetic determination of char oxidation (combustion and gasification) can be found in our previous study.[12,26,38] The char conversion X is calculated by the weight loss data as eq :where m0, mt, and m∞ represent the initial mass of char sample, mass of the sample at reaction time t, and mass of the sample when the reaction was finished, respectively.

In this study, the reactivity index Rs (min–1) was defined as eq to evaluate the overall char reactivity.[29,39]t0.5 is the time required to reach the char conversion of 50%.

The reaction rate constant k can be calculated as follows:where f(X) is the reaction model function. The integral format of f(X) can be described as eq :Therefore, the reaction rate constant can be calculated because there is a linear relationship between G(X) and t in the isothermal experiments. The reaction rate constant k is the slope of the fitting straight line.

If the main intention for studying coal reactivity is just to describe the relation between time and conversion, a shrinking core model is preferred.[40] The shrinking core model fits the results well from our previous work[12,38] and other researcher’s publications.[41,42] The shrinking core model (eq ) is selected as the mechanism function model in this study because of its mathematical simplicity, and the conversion time behavior is well described.

As a comparison, the other two models, nucleation and growth model and two-dimensional diffusion model, are calculated from G(X) expressed as eqs and 7, respectively.

After the calculation of reaction rate constant k under different temperatures, the following eq is then straight forward according to the Arrhenius equation:

The plot (ln k, 1/T) fits on a straight line. The activation energy −E/R and pre-exponential factor ln A are the slope and the intercept, respectively.

Results and Discussion

Online Sampling Char Morphology

Surface morphologies of the sampled char at different times are shown in Figure . On the original char surface, there are few fissures and micropores but they are not obvious (Figure a), which is perhaps due to the long-time pyrolysis at high temperature in the fixed bed to deeply remove the volatile content from raw coal. High temperature could result in a slight sintering. In the early combustion stage, there are some micropores and small char particles on the surface of the sampled char at 0.5 min (Figure b). In this stage, char was heated from room temperature to experimental high temperature (823 K) in an O2/CO2 atmosphere. Both heating and oxidant could make char particle active, and active sites on char surface were first consumed in the short reaction time. In the middle combustion stage, macropore occurs, and char pores become much developed (pore enlargement) with char combustion proceeding at 2.5 and 5.5 min (Figure c,d). In this stage, O2/CO2 further diffused into the internal of the char particle, and most of carbon was consumed, which resulted in the pore enlargement. The middle combustion stage takes about 50–70% of the whole combustion process, and the reaction rate is much stable,[12] which is the best stage to determine char kinetic parameters. Details will be discussed in later sections. In the late combustion stage, macropore breaks and collapses on the surface of the sampled char at 10 min, where most of carbon was consumed and coal ash left (Figure e). In this stage, coal ash blocked oxidant diffusion especially when coal ash melt, which resulted in the decrease of combustion rate.

Figure 2

FSEM images of the online sampled char morphology: (a) original char, (b) 0.5, (c) 2.5, (d) 5.5, and (e) 10 min.

Char Combustion Reactivity

The combustion reactivity of the online sampled chars was measured in a TGA. Typical TG curves are shown in Figure from which the carbon conversion in the MFB is calculated. The total carbon in the original char is calculated from Δmoriginal = m0 – m∞, and the resident carbon in the sampled char is Δmt = m0 – m∞. m0 and m∞ represent the initial and final sample mass in TGA, respectively. Therefore, the carbon conversion of the sampled char in the MFB could be calculated as , and the final conversions of the sampled char are shown in Figure . The carbon conversion is from 1.4 to 65.3% for the sampled char from sampling time of 0.5 to 10 min.

Figure 3

Carbon conversion in the MFB at different sampling times.

Char conversion (X) and reaction rate (dX/dt) versus oxidation time are shown in Figure . Char conversion rate is enhanced with increasing temperature. For example, it takes about 12 min to finish combustion at 773 K, while only 5 min is required at 873 K. Further, we can see that the conversion rate of the sampled char shows nonmonotonic behaviors with the sampling time. This is the main finding in our study, which will be discussed in more detail in later sections.

Figure 4

Combustion conversion of the sampled char at different temperatures: (a) 773, (b) 823, and (c) 873 K.

Figure shows the reaction rate of different sampled chars. Obviously, char has different reaction rates in different combustion stages (reaction time) with an increase–stable–decrease feature. In early combustion stage (i.e., carbon conversion rate <10%), the reaction rate decreases from the original char (square) to 0.5 min sampling char (circle). In the middle combustion stage (i.e., carbon conversion rate 10–60%), it increases from 0.5 min sampling char to 5.5 min sampling char (down triangle). In the late combustion stage (i.e., carbon conversion rate >60%), it decreases from 5.5 min sampling char to 10 min sampling char (diamond). Sampled char reactivity is sensitive to measuring temperature in TG, where sampled char at 5.5 min shows at the highest reaction rate at 823 and 873 K.

Figure 5

Combustion reaction rate of the sampled char at different temperatures: (a) 773, (b) 823, and (c) 873 K.

Reaction rate curves in Figure show the information for the whole conversion range from 0 to 100%, while reaction index Rs shows the overall average char reactivity as shown in Figure . Similarly, it also shows that char reactivity obeys the rule of decrease–increase–decrease behavior along with the sampling time.

Figure 6

Reactivity index Rs of the sampled char at different temperatures: (a) 773, (b) 823, and (c) 873 K.

.

For the original char, the carbon structure is much disordered from a Raman spectrometer, which will be detailedly presented in our further report. Although there are some small pore openings in the 0.5 min sampling char (Figure b), char quickly tends to ordered from original char to 0.5 min sampling char in the early combustion stage. Therefore, the reactivity of 0.5 min sampling char is lower than original char (Figure ). However, the disordered char structure will not dominate in later combustion. The pore structure and carbon matrix will dominate in the middle combustion stage, while the ash component will dominate in the late combustion stage. With pores developing from 0.5 to 5.5 min (Figure b–d), both reaction contact area and carbon active sites increase, leading to the increase in char reactivity (Figure ). However, at the late combustion stage, pore structure collapses, and the ash component blocks the pore path (Figure e), which reduces the char reactivity (Figure ).

Char Combustion Kinetic

G(X) fittings versus oxidation time calculated from three different models are shown in Figure at the range of conversion rate from 0.1 to 0.9. Overall, the shrinking core model fits the experiment well at the measured temperatures for different sampled chars (Figure a–c). However, the nucleation and growth model (Figure d–f) and two-dimensional diffusion model (Figure g–i) fail to fit as a straight line particularly at a low conversion rate. That is to say that the shrinking core model is better than the nucleation and growth model and two-dimensional diffusion model to fit char combustion in the TG reactor. Therefore in later sections, we used the shrinking core model to calculate kinetic parameters. The reaction rate constant k is the slope of the fitting straight line from the shrinking core model (Figure a–c).

Figure 7

Linear fitting of G(X) vs reaction time using three different models: (a–c) shrinking core model, 773, 823, and 873 K; (d–f) nucleation and growth model, 773, 823, and 873 K; and (g–i) two-dimensional diffusion model, 773, 823, and 873 K.

Table shows the value of the reaction rate constant measured at various temperatures for the sampled chars. R2 from the fittings are over 0.99, which indicates the reliability of the fittings. For a specific char, either original char or the sampled char, reaction rate constant k increases with the increase in temperature. Take the sampled char at 10 min for example, the reaction rate constant increases from 0.00284 to 0.00525 s–1 with increasing temperature.

Table 1

Reaction Rate Constants and Kinetic Parameters of the Sampled Char Calculated from the Shrinking Core Model

sampling time	T (K)	R2	k (s–1)	ln k	A (s–1)	E (kJ/mol)
original	773	0.9999	0.00322	–5.738	0.153	25.05
	823	0.9979	0.00357	–5.635
	873	0.9948	0.00506	–5.286
0.5 min	773	0.9938	0.00293	–5.833	0.204	27.40
	823	0.9972	0.00352	–5.649
	873	0.9966	0.00479	–5.341
2.5 min	773	0.9999	0.00298	–5.816	0.336	29.79
	823	0.9950	0.00521	–5.257
	873	0.9989	0.00501	–5.296
5.5 min	773	0.9999	0.00319	–5.748	0.726	34.38
	823	0.9957	0.00558	–5.189
	873	0.9938	0.00583	–5.145
10 min	773	0.9980	0.00284	–5.864	0.684	34.89
	823	0.9967	0.00473	–5.354
	873	0.9945	0.00525	–5.250

When comparing the reaction rate constant between original and different sampled chars, it obeys the rule of decrease–increase–decrease behavior with the sampling time. For example, at 773 K, the reaction rate constants are 0.00322 s–1 (original char), 0.00293 s–1 (0.5 min sampling char), 0.00298 s–1 (2.5 min sampling char), 0.00319 s–1 (5.5 min sampling char), and 0.00284 s–1 (10 min sampling char). Reaction rate constant decreases first from original char to 0.5 min sampling char, followed by increasing from 0.5 min sampling char to 5.5 min sampling char. Finally it decreases again to 10 min sampling char. The similar results could be also found in other temperatures, which agree with the results in Figure .

When the values of k at different temperatures were determined, and the kinetic parameters could be obtained according to eq . The values of ln k at different cases are listed in Table , as well as the calculated values of pre-exponential factor A and activation energy E. Both pre-exponential factor A, i.e., from 0.153 to 0.684 s–1, and activation energy E, i.e., from 25.05 to 34.89 kJ/mol, of the sampled char increase with the sampling time, which is much different from the rule of char reactivity (Figure ). This indicates that the sampled char reactivity is determined by not only kinetic parameters but also the char structure (i.e., pore structure and carbon matrix). A high pre-exponential factor means high molecule collision. High activation energy results in the reaction taking place difficulty. With the combustion proceeding in the MFB, coal char becomes more and more reactive because of the increase of pre-exponential factor A. However, because the char structure becomes ordered and activation energy increases, char reactivity slightly decreases at the early combustion stage (≤0.5 min in this study). In late combustion stage, the activation energy has a little change from 5.5 to 10 min sampling char. However, due to the pore collapse and ash blocking, char reactivity decreases at the end (Figure ).

Char Gasification Reactivity

Coal char was continuously sampled in CO2/H2O gasification atmospheres. Char gasification conversion (X) and reaction rate (dX/dt) are shown in Figure . Compared to Figure a, increasing either CO2 (Figure b) or H2O (Figure c) partial pressure can enhance the gasification conversion. Moreover, in Figures and 8, there is not much difference between the combustion and gasification rate. This is mainly because our combustion experiments are performed at 773, 823, and 873 K, while gasification experiments are performed at 1273 K. In Figure , it takes about 12 min to finish combustion at 773 K, while only 5 min are required at 873 K. If the temperature increases to 1273 K, the time for finishing combustion would be much lower than 5 min and would be much faster than the gasification rate in Figure .

Figure 8

Gasification conversion of the sampled char at different gasification atmospheres: (a) 20% CO2 + 20% H2O, (b) 33.3% CO2 + 20% H2O, and (c) 20% CO2 + 33.3% H2O.

Figure shows the gasification rate of different sampled chars. The difference in gasification rates originates from different gasification atmospheres. There is not much difference for the gasification rate of the sampled char at an atmosphere of 20% CO2 + 20% H2O (Figure a). A minor increase was observed for 2.0 min sampling char. The gasification rate obviously increases along with the sampling time at both 33.3% CO2 + 20% H2O (Figure b) and 20% CO2 + 33.3% H2O (Figure c).

Figure 9

Gasification rate of the sampled char at different gasification atmospheres: (a) 20% CO2 + 20% H2O, (b) 33.3% CO2 + 20% H2O, and (c) 20% CO2 + 33.3% H2O.

As for the reaction index Rs from the overall average, char reactivity is shown in Figure . In Figure a, a minor decrease from 0.5 to 1.0 min sampling char is followed by the increase from 1.0 to 2.0 min sampling char. Overall, the difference of Rs in 20% CO2 + 20% H2O is not much. In Figure b, a minor decrease from 0.5 to 1.0 min sampling char is followed by a large increase from 1.0 to 2.0 min sampling char. In Figure c, Rs increases with the sampling time. Overall, in different gasification atmospheres, the reactivity of the sampled (partial gasified) char increases with the sampling time even though there is a minor decrease in the initial gasification stage. The same phenomenon can be found in the gasification reaction rate constant, as shown in Table . In Section , a reactivity decrease was observed in the late combustion stage in our combustion test, while it was not found in the gasification test probably due to the short sampling time in the MFB.

Figure 10

Reactivity index Rs of the sampled char at different gasification atmospheres: (a) 20% CO2 + 20% H2O, (b) 33.3% CO2 + 20% H2O, and (c) 20% CO2 + 33.3% H2O.

Table 2

Reaction Rate Constants of the Sampled Char (Calculated from the Shrinking Core Model)

	k (s–1)
sampling time	20% CO2 + 20% H2O	33.3% CO2 + 20% H2O	20% CO2 + 33.3% H2O
0.5 min	0.00225	0.00403	0.00343
1.0 min	0.00220	0.00363	0.00400
1.5 min	0.00221	0.00442	0.00431
2.0 min	0.00260	0.00556	0.00530

In gasification, a small amount (5%) of O2 was injected together with CO2/H2O. In a low oxygen O2/CO2/H2O atmosphere, CO2 concentration is high, which accelerates the char polycondensation reaction. Hence, gasification reactivity slightly decreases in the initial stage. Later, char pores become developed after initial gasification. The developed char structure results in the increase in the later gasification rate. Increasing CO2 and H2O (Figures b,c) concentrations in the gasification test further enhances the gasification reaction as well as reactivity. CO2-enhanced gasification reactivity dominates for the highly gasified char, i.e. 1.5 and 2.0 min sampling char (Figure b). H2O-enhanced gasification reactivity dominates for the low gasified char, i.e. 0.5 and 1.0 min sampling chars (Figure c).

Conclusions

In this study, we self-design an MFB with combination of an online char particle sampling system to study the kinetics of coal char combustion and gasification. For Vientiane coal, the online sampled char reactivity in the MFB obeys the rule of decrease–increase–decrease behavior with the sampling time. However, both pre-exponential factor A and activation energy E increase with the sampling time from the original char to sampled char. For the effects on the sampled char reactivity, the disordered char structure dominates in the early combustion stage (i.e., carbon conversion rate <10%); the pore structure dominates in the middle combustion stage (i.e., carbon conversion rate 10–60%); and the ash component dominates in the late combustion stage (i.e., carbon conversion rate >60%). In gasification, the reactivity of the sampled (partial gasified) char increases with the sampling time even though there is a minor decrease in the initial gasification stage. CO2-enhanced gasification reactivity dominates for highly gasified char, while H2O-enhanced gasification reactivity dominates for lowly gasified char. This study shows the reliability of the MFB combining online particle sampling to investigate char combustion and gasification, but the tests for different coal ranks as well as the application to other gas–solid reactions are certainly required in the future.