Solid Fuel Combustion Method And Apparatus

This invention relates to a method and an apparatus for solid fuel combustion, and in particular to the in situ gasification of a solid fuel, such as coal, and chemical looping.

Introduction

Electricity generation accounts for ~38% of global anthropogenic carbon emissions to the atmosphere or ~2,400 Mt/y (carbon basis), projected to exceed 4,000 Mt/y by 2020 ¹ . To control its environmental impact, there is an urgent requirement to sequester CO ₂ from the combustion of coal, or fuels derived from it, in the earth ² . The cost of sequestration is small (e.g. $4-8/t C) compared to the costs of separating CO ₂ from typical flue gases ($100-200/t C) ² , so that disposal approaches viability only if pure CO ₂ is available, largely free of nitrogen and other inert gases. One means of obtaining pure CO ₂ from a power plant burning a gaseous fuel, e.g. natural gas, is to use chemical looping combustion ² (CLC). The fuel, in gaseous form, is oxidised with a metal oxide, generalised as MeO, in:

(2/7 + m) MeO + C _n H _2n , -» (2A7 + m) Me + /77H ₂ O + πCO ₂ , (1) to produce mainly CO ₂ and steam. Consequently, almost pure CO ₂ is left when the steam is condensed from the off-gases. The reduced form of the metal oxide, Me, is then transferred to a different reactor, where it is re-oxidised by contact with air in:

Me + 1/2O ₂ → MeO. (2) The gas from this reactor is N ₂ containing unused O ₂ . Algebraically adding (1) and (2) shows that the net effect is that the fuel has been burned, but the resulting CO ₂ has been separated from the nitrogen in the air. Of course, the total heat evolved is the same as for combustion of the fuel in air. This technique for the combustion of natural gas is an area of active research ² ' ³ . Solid fuels can be gasified to syngas in a gasification reactor before being used in a CLC cycle ⁴ .

Summarv of Invention

The invention may advantageously provide a method and an apparatus for combusting a solid fuel, for example for generating power, as defined in the appended independent claims. Preferred or advantageous features of the invention are set out in dependent subclaims.

In a preferred embodiment of the invention, chemical looping may thus be applied to solid fuels. This may be carried out, for example, in a semi-batch mode, involving say a fluidised bed reactor containing a chemical looping agent (CLA) such as a metal oxide (MeO). In the embodiment, the reactor would be operated in a steady cycle of three consecutive periods, t _h t ₂ and t ₃ . During t _h the bed would be fluidised by a gasification agent, such as steam or CO ₂ , and a solid fuel, such as coal-char, would be fed, preferably in a steady supply, to the bed (advantageously at ~ 900 ⁰ C) so that:

• the char undergoes gasification by the gasification agent (steam or CO ₂ ) to yield a synthesis gas (syngas) containing mainly CO and H ₂ :

C(S) + H ₂ O(g) → CO ₍ g ₎ + H _2(Q) (3)

C( _S ) + CO _2(g) → 2CO _(g) (4)

• the syngas reacts with the surrounding CLA (MeO particles) to give CO ₂ and steam by a version of reaction (1):

MeO _(S ) + H ₂₍ B ₎ → Me _(s) + H ₂ O _(g) (5)

MeO _(s) + CO _(g) → C0 _2(g) + Me _(S) (6)

This system can only function down to a certain degree of reduction of the metal oxide. Thus, after time t _1t the feed of coal-char ceases and the remaining inventory of bed carbon is allowed to gasify for a further period of time, t ₂ , until the inventory is sufficiently small. At the end of t _∑ , the bed is fluidised by air instead of steam or CO ₂ for a period of time, t ₃ , during which reaction (2) regenerates the bed of metal oxide. During t ₃ some residual carbon might be burned off, leading to a small release of CO ₂ with the regenerating air, but very much less than that from the direct combustion of coal in air. .

In more general terms, the invention may thus advantageously allow the use of CLC with solid fuels directly, in situ inside a chemical looping reactor. This has not previously been possible for reasons including the problem of separating the solid fuel and the metal oxide for the oxidation (regeneration) phase.

This may therefore advantageously provide a technique for isolating the CO ₂ from burning a solid fuel.

A preferred embodiment may use chemical looping combustion with a solid fuel, such as coal-char, with a gasification agent like steam (or CO ₂ ) introduced into the reactor. The gasification agent is believed to transfer solid carbon to gaseous CO, which like H ₂ can be reacted with a solid, e.g. Fe ₂ O ₃ , carrying oxygen, to yield CO ₂ and H ₂ O. On the basis of a limited series of tests of specific embodiments, the reaction of sintered compacts of Fe ₂ O ₃ appears to be sufficient to make a semi-batch process feasible.

The inventors have demonstrated the feasibility of this technique when used with CO ₂ or steam as the gasifying agent.

Description of Specific Embodiments of the Invention

Specific embodiments of the invention will now be described by way of example, with reference to the accompanying drawings, in which:

Figure 1 is a plot of the rate of production of CO from a bed of sand (i.e. the product of the total molar flow rate and mole fraction of CO) in which a single batch of char (0.0904g) was gasified in 27.5 mol% CO ₂ at 900 ⁰ C;

Figure 2 illustrates the results of a series of experiments in which successive batches of char were gasified in the active bed of Fe ₂ O ₃ and silica sand. The bed was regenerated after the fourth batch using 5% O ₂ in N ₂ . In each case the mass of char in the batch is shown, together with the recovery;

Figure 3 illustrates the results of a series of experiments in which batches of coal (+1.4mm, -1.7mm) were gasified in the active bed of Fe ₂ O ₃ and silica

sand. The bed was regenerated after the fourth batch using 5% O ₂ in N ₂ . The mass of coal used in each experiment is shown;

Figure 4 is a plot of the mole fractions of CO, CO ₂ and H ₂ (dry basis) for an experiment in which 0.0946g of char was gasified in steam and N ₂ , in a bed of silica sand at 900 ⁰ C. The mole fraction of H ₂ , in this case, has been calculated from the reaction's stoichiometry;

Figure 5 is a plot of the rate of production of CO and CO ₂ for experiments in the active bed containing Fe ₂ O ₃ . The fluidising gas was either 27.6 mol % steam in N ₂ or 5.2 mol% O ₂ in N ₂ . The triangles in the plot show the ratio of the yields of CO and CO ₂ produced during each experiment. The mass of char (in grams) added for each experiment is indicated. The temperature of the bed was 900 ⁰ C; and

Figure 6 is a plot of measured mole fractions of H ₂ , CO, CO ₂ in the off-gases for the last two experiments in Figure 5. The concentrations shown are on a dry basis.

Experimental The first embodiments involve preliminary experiments to test the feasibility of the above technique using iron oxide (Fe ₂ O ₃ ) as the CLA, and a lignite fuel gasified by CO ₂ .

Materials. The chemical looping agent was produced from Fe ₂ O ₃ powder (Aldrich > 99% purity), which was mixed with a small amount of distilled water, in a food mixer. The resulting particles were sieved to +300, -710μm (i.e. to a range between approximately 300 and 710μm particle size), and any larger lumps broken up. The procedure was repeated until a sufficient quantity of particles was in this size range. The agglomerated particles of Fe ₂ O ₃ were then placed in a furnace, heated to 900 ⁰ C and maintained at this temperature for 5h. The resulting particles were then sieved into two size ranges, 300 to 425μm and 425 to 71 Oμm.

The fuels used were a lignite (Hambach) and its char. The char was manufactured in a bed of silica sand (sieved to 355-425μm) contained in a quartz reactor, initially at 900 ⁰ C, fluidised by N ₂ . The lignite (1.4 to 1.7mm) was added slowly; ~15g was added over a time period of ~30min. The bed was then allowed to regain a temperature of 900 ⁰ C, and then cooled (whilst fluidised by N ₂ ) until the char could be recovered. The reactivity of a char is known to be affected by its heating history; hence, this method of producing the char would not be ideal if detailed measurements of the char's reactivity were the object of this work. However, for this study the method is sufficient. Once recovered,

10 the char was sieved to 1.18-1.4mm. Analyses of the char and the parent coal are given in Table 1.

Table 1. Analysis of Hambach lignite and its char by microanalysis (as received basis)

I ₅

Chemical looping experiments. Experiments were performed in a fluidised bed of either (i) 20ml of silica sand (+355,-425μm) or (ii)10.308g of Fe ₂ O ₃ (+300, -425μm) made up to 20ml with silica sand (+355μm,-425μm), so the total mass of particles in the bed was 30.971 g. These particles were contained in a quartz

20 tube (internal diameter 30mm) with a sintered disk for a distributor. It was fluidised by N ₂ (85ml s ^"1 at STP (standard temperature and pressure)) mixed with either pure CO ₂ (32ml s ^"1 at STP) for gasification, or air (28ml s ^"1 at STP) to regenerate the Fe ₂ O ₃ . U/U _mf was between 9 and 12. The reactor was placed in a tubular furnace and the bed was heated to 900 ⁰ C, as monitored by a

25 K-type thermocouple within the bed.

The off-gases were sampled continuously into two NDIR (non-dispersive infrared) analysers, one measuring [CO ₂ ] and [CO] (with ranges of 20mol% and

1 mol%, respectively) and the other [CH ₄ ] and [CO] (with ranges of 6mol% and 11mol%, respectively), via a trap at 0 ⁰ C (to remove tars etc) with a glass wool filter and a Permapure™ membrane drier (MD 070 44P), which was purged at 5L min ^'1 with dry nitrogen. Each analyser received 1 L min ^"1 of sample gas.

The first experiments used the bed of silica sand alone and served as a control. With the hot bed fluidised by the mixture of N ₂ and CO ₂ , a batch (~ 0.1 g) of the char was added to the bed. The char was allowed to gasify until completion in every experiment. The experiment was repeated several times. The second set of experiments used the bed containing the particles of Fe ₂ O ₃ mixed with silica sand. As in the first experiments, a batch of char (-0.1 g) was added to the bed and gasified to completion. Further batches of char were added, until it was clear that the chemical looping agent (Fe ₂ O ₃ ) had been used up. At this point, the bed was re-oxidised with the mixture of air and N ₂ ; then the gasification experiment was repeated. Finally, following a further regeneration of the Fe ₂ O ₃ , the gasification experiments were repeated with the parent coal (sieved to 1.4 to 1.7mm) added, instead of its char.

Gasification of char in an inert, bed of silica sand. A typical result from such an experiment is shown in Figure 1 , indicating that gasification of the char was complete within ~400s. The peak rate of gasification in Figure 1 corresponded to a concentration of 1.8mol% of CO in the off-gas from the bed. The conversion of the carbon in the char to CO, as measured from the area under the curve in Figure 1 , was between 90.0% and 91.7% in four replicated experiments. Since a small amount of fine carbon was collected by the trap in the sampling line, the discrepancy of -10% in the mass balance can be partly attributed to the elutriation of fine particles of char, once most of a batch had been gasified.

Gasification of char in an active bed OfFe ₂ O ₃ and silica sand. Figure 2 shows the results of experiments in which a batch of char was gasified in CO ₂ , in a bed initially of sand and Fe ₂ O ₃ particles. In this case reaction (4) again produces CO, which is subsequently oxidised in reaction (6) by the solid particles of Fe ₂ O ₃ . The recovery is here defined as the number of moles of CO actually produced in the off-gases to the amount which would have been

produced had all the carbon in the char been gasified to CO. The bed was not regenerated between the first four experiments; consequently, the ability of the Fe ₂ O ₃ particles to react with CO was gradually reduced. The largest peak concentration in Figure 2 corresponds to 1.9mol%; it occurs when the Fe ₂ O ₃

5 was fully reduced and exhausted. This is evident from the fact that the maximum [CO] agrees well with the 1.8mol% observed with a bed of sand, in Figure 1. The bed was regenerated after the fourth experiment using 5% O ₂ in N ₂ . The very last (fifth) plot for CO in Figure 2 is identical to the first, showing that all the Fe ₂ O ₃ had been regenerated after being exposed to 5mol% O ₂ for o ~5min.

Gasification of raw lignite in an active bed OfFe ₂ O ₃ and silica sand. When a batch of parent coal was added to the bed of Fe ₂ O ₃ and sand, there was an initial, very rapid production of volatile matter, corresponding to the initial sharp s spikes on the broader peaks shown in Figure 3. During devolatilisation, a small amount of methane was also detected, together with the CO. The extent to which the volatiles reacted with the Fe ₂ O ₃ in the bed is not totally clear at this stage. However, the fact that the first CO spike in Figure 3, i.e. associated with the volatiles, was much smaller than say the third peak, indicates that the 0 volatiles do react with Fe ₂ O ₃ . A similar difference was also seen for the methane produced, although this is not shown. It is also noteworthy that the sampling line's filter (located in an ice bath) did not contain any tarry material. This again suggests, qualitatively, that the above procedure results in the improved conversion of volatile matter, including its tarry components. ₅ Of course, it is possible that the volatile matter, before it is oxidised by the Fe ₂ O ₃ , is also cracked by it.

Discussion of reaction of CLA

If it is assumed that the Fe ₂ O ₃ in the bed reacts to form Fe ₃ O ₄ in reactions (5) 0 and (6), so that the overall reaction in the system is:

C + CO ₂ + 6 Fe ₂ O ₃ → 2 CO ₂ + 4 Fe ₃ O ₄ , (7)

the theoretical capacity of a bed (containing 10.308g of Fe ₂ O ₃ as above) is 0.0106mol of carbon. The assumption that the Fe ₂ O ₃ reacts to form only Fe ₃ O ₄ s (rather than Fe ₀ ^ ₇ O or metallic Fe) can be justified by the following

thermodynamic argument. For the reaction

0.788 CO + 0.947 Fe ₃ O ₄ → 0.788 CO ₂ + 3 Fe ₀ . ₉₄₇ O, (8)

the equilibrium constant K _p = (Pcc^/Pco) ^0'788 = 1 -749 (using NASA-Glenn thermodynamic coefficients ⁸ ) at 900 ⁰ C. Thus, at 900 ⁰ C, for the Fe ₃ O ₄ to be reduced to Fe ₀ ^ ₄₇ O would require pcoφco2 > 0.49, which it was not. A rough estimate of the actual capacity of the bed can be calculated by noting that in Figure 2, the Fe ₂ O ₃ had been depleted after the third batch had been added. Using the difference in yields of CO between the inert bed of sand and the active bed containing Fe ₂ O ₃ , the amount of carbon consumed by reaction (7) was 0.0117mol, in close agreement with the above theoretical value of 0.0106mol. Although this calculation is relatively coarse, it does indicate that the Fe ₂ O ₃ goes completely to Fe ₃ O ₄ .

Further Embodiments

The second embodiments involve preliminary experiments to test the feasibility of the technique using iron oxide (Fe ₂ O ₃ ) as the CLA, and a lignite fuel gasified by steam.

Materials. Details of the materials used are as above. The fuel used was char made from a lignite (Hambach) as above.

Chemical looping experiments. Experiments were performed in a fluidised bed at atmospheric pressure of either (i) 20ml of silica sand (+355,-425μm) or (ii) 9.9676 g of Fe ₂ O ₃ (+300,-425μm) made up to 20ml with silica sand

(+355,-425μm). In (ii) the mass of iron oxide was twice that of the sand. The bed was contained in a quartz tube (i.d. 30mm) with a sintered disk for a distributor. The bed was fluidised by N ₂ (84.6ml s ^"1 at NTP) mixed with either steam for gasification, or air (28ml s ^"1 at NTP) to regenerate the Fe ₂ O ₃ . The steam was generated by vaporising a carefully-controlled flowrate of 86ml h ^"1 of water, which was injected into a heated section of the gas supply line to the bed. The reactor was placed in a tube furnace and the bed was heated to 900 ⁰ C; the temperature of the bed was monitored by a K-type thermocouple within the bed.

The off-gases were continuously sampled into two NDIR analysers, one measuring [CO] and [CO ₂ ] (with ranges of 20mol% and 1mol%, respectively), the other [CH ₄ ] and [CO] (with ranges of 6mol% and 11mol%, respectively). Before entering these analysers, the samples were passed through two traps maintained at 0 ⁰ C (to condense the water), a glass wool filter and a Permapure™ membrane drier (MD 070 44P), purged at 5L min ^"1 with dry nitrogen. Each analyser received 1 L min ^"1 of sample gas. For some experiments, a mass spectrometer (Hiden HPR-20) was used to detect H ₂ in the off-gases.

The first set of experiments used the bed of silica sand without Fe ₂ O ₃ and served as a control. With the bed fluidised by the mixture of N ₂ and H ₂ O a batch (0.1g) of the char was added to the bed. The char was allowed to gasify to completion in all experiments. The experiment was repeated several times. The second set of experiments used the bed containing the particles of Fe ₂ O ₃ . As in the first experiments, a batch of char (0.1g) was added to the bed and allowed to gasify to completion. Subsequent batches of char were added, until it was clear that the chemical looping agent had been depleted. At this point, the bed was re-oxidised with the mixture of air and N ₂ , after which, the gasification experiment was repeated.

Results

Experiments in an inert bed of silica sand only.

Figure 4 shows the mole fractions of H ₂ , CO and CO ₂ in the off-gas (dry basis) for an experiment in which a batch of char was added to the (inert) bed of silica sand, when fluidised by steam and N ₂ . The average yield of carbon recovered in the off-gas as CO and CO ₂ was between 103 to 108% of the amount of carbon added in the char, for five repetitions of this experiment.

For the case where the chemical looping agent had been depleted, or is not present, H ₂ can only be produced either by gasifying the char with steam in reaction (3), or by the water-gas shift equilibrium:

CO _(B) + H ₂ O _(g) = CO _2(g) + H ₂ (B). (9)

When this reaction occurs, extra H ₂ is produced. Additionally, the CO ₂ formed might react with the char in reaction (4). The stoichiometry of these reactions is such that the rate of production of H ₂ is equal to the rate of production of CO, plus twice the rate of production of CO ₂ . Accordingly, the value of [H ₂ ] in Figure 4 has been calculated in this way. This argument neglects the fact that the char contained a small amount of hydrogen; in fact the char used here had a molecular formula of CH ₀ .iN ₀ .o7- The fact that large amounts of CO ₂ are produced indicates that the water-gas shift reaction is important in this system.

Experiments in an active bed of silica sand and particles OfFe ₂ O ₃ , The results from a series of experiments in which batches of char were added to the bed containing Fe ₂ O ₃ , fluidised by steam and nitrogen, are shown in Figure 5. Initially, when there was unreacted Fe ₂ O ₃ in the bed, large amounts CO ₂ were produced and the ratio of the yield of CO to that of CO ₂ was small (~0.02 as shown in Figure 2). In subsequent experiments, the ratio of C0/C0 ₂ increased (to ~0.2) due to the depletion of Fe ₂ O ₃ , until, after the second batch, the Fe ₂ O ₃ was spent. After the fourth experiment shown in Figure 5, the bed was regenerated by fluidising with 5.2 % O ₂ in N ₂ to reoxidise the iron oxide back to

Fe ₂ O ₃ . In fact when the chemical looping agent is present CO ₂ can be produced (in addition to that produced by reaction (9)) and H ₂ destroyed, by reaction with the metal oxide in:

3 Fe ₂ O _3(s) + H _2{B ) → 2 Fe ₃ O ₄ (S) + H ₂ 0 _(g) (10)

3 Fe ₂ O ₃ (S) + CO(g) → 2 Fe ₃ O ₄ (S) + C0 _2(g) . (11 )

Therefore, there is now no method of calculating the amount of H ₂ in the off- gases from the measured concentrations of CO and CO ₂ . Thus, for the last two experiments in Figure 5, the H ₂ content of the off-gases was measured by the mass spectrometer, with Figure 6 showing the measured concentrations of CO, CO ₂ and H ₂ . For Figure 6, the calibration of the mass spectrometer was obtained by assuming that for the first experiment in Figure 6, the chemical looping agent was exhausted, so the concentration of H ₂ could be calculated from the measured values of [CO] and [CO ₂ ]. It is interesting to note that, in the presence of the iron oxide, even when it is spent as an oxygen carrier, the initial ratio of CO to CO ₂ is smaller than when iron is not present, as seen by comparing Figures 4 and 5, respectively. This indicates that the iron oxide is

promoting the shift reaction (9) and that reaction (9) was certainly not in equilibrium in the experiments with sand alone; the measured and inferred values [H ₂ ] were also lower than would be expected if reaction (9) had reached equilibrium.

5

Discussion

The above preliminary experiments demonstrate that it is possible to use a solid fuel, such as coal or coal-char, directly within a chemical looping combustion cycle involving gasification of the char as well as combustion of o both the volatiles and the products of gasification by reaction with Fe ₂ O ₃ . Despite the fact that the chemical looping agent used (pure Fe ₂ O ₃ ) was not optimised, it was still able to oxidise most of the CO and H ₂ produced by gasifying the carbon. There is a substantial literature on the production of particles suitable for chemical looping, using e.g. iron ⁵ ' ⁶ and other metallic 5 oxides ⁷ , so that there is considerable scope to improve the design of particles suitable for chemical looping, using iron or other metals. In each of the described embodiments, the same bed was used for each of the series of experiments; thus the iron oxide particles were reoxidised three times in Figures 2 and 3, and twice in Figure 5. The particles did not appear to degrade 0 during these reduction-oxidation cycles and the bed evidently recovered its initial reactivity when reoxidised.

If it is assumed that the Fe ₂ O ₃ in the bed in the experiments of Figures 4, 5 and 6 is reduced to Fe ₃ O ₄ in reactions (5) and (6), the overall reaction in the 5 system is

C + 6 Fe ₂ O ₃ → CO ₂ + 4 Fe ₃ O ₄ , (12)

and the theoretical capacity of a bed containing 9.976g of Fe ₂ O ₃ , as above, is 0.0104mol of carbon. Figure 5 shows that the Fe ₂ O ₃ was approximately o depleted after two batches of carbon had been used (~0.17 g ≡ 0.014mol C); thus the closure on the mass balance is satisfactory. Although this calculation is relatively coarse, it does indicate that the Fe ₂ O ₃ goes completely to Fe ₃ O ₄ . Such an assumption that the Fe ₂ O ₃ forms only Fe ₃ O ₄ , rather than FeO or Fe can be justified by the following thermodynamic argument. For the reaction 5 0.788 CO + 0.947 Fe ₃ O ₄ → 0.788 CO ₂ + 3 Fe _0-947 O, (13)

the equilibrium constant K _p = (pcW Pco) ^α788 = 1 -749 (using NASA-Glenn thermodynamic coefficients ⁸ ) at 900 ⁰ C. Thus, at 900 ⁰ C, for the Fe ₃ O ₄ to be reduced to Feo.9 ₄₇ 0 would require pcoφco2 > 0.49, which it was not. Similarly, for:

0.788 H ₂ + 0.947 Fe ₃ O ₄ → 0.788 H ₂ O + 3 Fe _0-947 O, (14)

at 900 ⁰ C, Kp = (P _H2 0/P _H2 ) ⁰⁷⁸⁸ = 2.116, and for the Fe ₃ O ₄ to be reduced to Feo.9 ₄ 70 would require P _H2 φ _H20 > 0-39, which was not the case.

References

1. IPCC. Climate Change 2001. Third Assessment Report, Intergovernmental

Panel on Climate Change (IPCC). U.N. (available on http://www.grida.no/climate/ipcc tar/index.htm). 2. Lyngfelt A, Leckner B, Mattisson T. Chem. Eng. Sci. 2001 ; 56; 3101 -3113. 3. Lyngfelt A, Leckner B. Technologies for CO ₂ separation. Minisymposium on Carbon Dioxide Capture and Storage, Chalmers University,

22 ^nd Oct 1999.(available on http://www.entek.chalmers.se/~anlv/symp/sympco2.html). 4. Jin H, lshida M. A new type of coal gas fueled chemical-looping combustion. Fuel. 2003; 83; 2411 2417. 5. Mattisson T, Lyngfelt A, Cho P. The use of iron oxide as an oxygen carrier in chemical looping combustion of methane with inherent separation of

CO ₂ . Fuel. 2001 ; 80; 1953-1962. 6. lshida M, Takeshita K, Suzuki K, Ohba T. Application of Fe ₂ O ₃ -AI ₂ O ₃ composite particles as solid looping material of the chemical-loop combustor. Energy & Fuels. 2005; 19; 2514 2518.

7. Mattisson T, Jardnas A., Lyngfelt A. Reactivity of some metal oxides supported on alumina with alternating methane and oxygen-Application for chemical-looping combustion. Energy & Fuels. 2003; 17; 643 651.

8. McBride BJ, Zehe MJ Gordon S. NASA Glenn Coefficients for Calculating Thermodynamic Properties of Individual Species. NASA. 2002 report TP-2002-21155.