Attrition of spherical electrode carbon particles during combustion in a turbulent fluidized bed

Chmaicd Engineering Science. Vol. 47, No. 3, pp. 527-532.1992. Printed in Great Britain.

ooos2509192 SS.00 + 0.00 Q 1991 Pcrgamon Rear plc

ATTRITION OF SPHERICAL ELECTRODE CARBON PARTICLES DURING COMBUSTION IN A TURBULENT FLUIDIZED BED P. K. HALDERf Department

of Mechanical

(First

received

and P. BASU*

Engineering, Technical University of Nova Scotia, P.O. Box 1ooO. Halifax, NS, Canada B3J 2X4

8 February

1989; accepted for publication

in revised form

26 June

1991)

Abstract-Experiments were conducted to study combustion-assisted attrition in a 41 mm diameter and 1.2 m tall, open-top, turbulent fluid&d-bed apparatus. The difference between the observed weight loss rate and the burning rate calculated from the gas analysis is taken as the attrition rate. Both mechanical and combustion-assisted attrition were measured. Results are compared with available data on bubbling fluidized beds. The effect of oxygen concentration on the combustion-assisted attrition was also studied.

INTRODUCTION While

modelling

the

combustion

phenomena

in

beds, earlier researchers (Basu, 1977; Chakraborty and Howard, 1978; La Nauze and Jung, 1982a, b) neglected the attrition of carbon particles. Later it was shown by Donsi et al. (1981) and Arena et al. (1983) that attrition contributed to the elutriation of carbon fines from a continuously bubbling fluidized bed. Though the loss of carbon due to attrition may not constitute a major fraction of the total weight loss due to combustion, it makes a major contribution to the loss of carbon from the bed through elutriation. This results in lower combustion efficiencies of these combustors. It was observed that combustion and attrition are simultaneous phenomena (Chirone et al., 1985) and that combustion enhances attrition. Halder et al. (1988) showed that an enhancement of attrition occurs due to combustion even for particles that are spherical, homogeneous, and smooth. However, these studies were carried out in bubbling fluidized beds. In a circulating fluidized bed, the carbon fines generated due to attrition may be lost through the cyclone. It is essential to consider this in any performance models of high velocity turbulent or fast fluidized bed. Basu (1984) and Basu and Subbarao (1986) measured mechanical attrition in a turbulent fluidized bed and used the results to calculate burning rates of carbon particles from observed weight loss rates. Their work did not take account of the perceived enhanced attrition rate due to combustion as observed by Arena et al. (1990) using an indirect measurement technique in a fast-bed mmbustor. This paper describes a more direct method to study the effect of combustion on the attrition of spherical electrode carbon particles in a turbulent fluidized bed. The gas-solid mixing in a turbulent bed is similar to that of a fast bed. where experimental difficulties

fluidized

inhibit direct measurement of attrition. So the results of attrition in a turbulent bed may be the nearest approximation for the fast bed. Results were obtained for mechanical as well as for combustion-assisted attrition. Measured data for combustion-assisted attrition are compared with those for mechanical attrition. Data obtained from the turbulent bed are compared with that from bubbling fluidized beds. The effect of oxygen concentration on attrition of burning particles in a turbulent fluidized bed is also discussed. EXPERIMENTAL of a 41 mm i.d. and 1.2 m high combustor made of stainless steel with open-top arrangement was used for these experiments. The schematic diagram of the experimental set-up is shown in Fig. 1. Air entering the bed was pre-heated using both an electrical heater located inside a quartz tube and an electrical furnace outside the air tube. The combustor was placed inside a cylindrically shaped electrical furnace which provides the heat required to maintain the bed temperature. The bed is operated in the turbulent regime at velocities in excess of terminal velocities of individual bed particles. The top of the bed was expanded to about 300 mm, so that the gas velocity is reduced at the bed exit. Lancia ct al. (1988) found that the angle of the expansion cone must exceed a critical angle which was maintained in the present bed. The absence of a secondary bed on the top was confirmed through visual inspection. Solids are entrained in the expanded section of the furnace, but return to the bed as their velocity drops below the terminal velocity of individual particles. This arrangement helped operate the bed in the turbulent regime while avoiding the problem of excessive elutriation of bed particles. Sieved sand in the size range 212-355 p was used as the bed material. About 600 g of sand was placed in the bed, the temperature of which was maintained at 1073 K using electrical controls. An apparatus

‘Present address: Department of Power Plant Engineering, Jadavpur University, Calcutta 7ooo91, India. ‘Author to whom correspondence should be addressed. 527

consisting

P.K.

528

For each test, 20-30 particles

were prepared

spherical electrode by machining carbon

HALDER carbon rods of

dry-cell weight carbon Each

batteries. Each particle had nearly the same and diameter. The weight of each batch of was between 9 and 11 g. batch of carbon was heated for about 2 min inside the bed by maintaining the bed near the minimum fluidized condition using nitrogen. Then the nitrogen was substituted by a mixture of oxygen and nitrogen. The particles were allowed to burn in this for a specified period of time. Concentrations of oxygen, carbon dioxide, and carbon monoxide in the outlet gas were continuously monitored using a computer. The oxygen concentration at the bed exit dropped by about lO-25% of its value at the inlet. For low inlet oxygen concentrations (5.2% and 10.5%) particles were allowed to burn for 10 min. For 15.75% and 21% oxygen concentrations, they burned for 8.5 and 5 min, respectively. This was done to avoid a large change in diameter during combustion. At the end of the pre-determined time period, the fluidizing gas was switched back to nitrogen and the bed was maintained at the minimum fluidizing condition. Carbon particles were sieved out using a basket and were then immersed quickly in a stream of cold nitrogen. The weight of the batch of carbon and the diameter of individual particles were measured before and after the test. The weight of the carbon burnt during the experiment could be calculated from the exit concentrations of carbon dioxide and carbon monoxide. The measured difference of the weight of the batch of carbon before and after the test will constitute carbon burnt and attrited during the experiment. The difference between the calculated weight loss rate (from the gas analysis) and observed weight loss rate is the attrition contribution of the weight loss rate.

and P. BASU The experimental procedure neglects any post-attrition combustion of fines within the bed. This assumption may not be strictly true (Chirone et al., 1985). Turnbull and Davidson (1984) showed that small particles ( > 0.5 mm) burn under kinetic control. They burn slowly as their temperature is close to the bed temperature. Tumbull and Davidson (1984) further pointed out that fine particles are expected to be elutriated from a bubbling bed before they are burnt; this is one reason for poor combustion efficiency in bubbling fluidized beds (Beer et al., 1980; Chirone et al., 1982; Arena et al., 1983). Massimilla et al. (1985) have shown that attrited fines are very small (75% below 50 p on numerical basis). They are therefore expected to be elutriated much faster than coarser particles because the gas velocity is many times greater than the terminal velocity of attrited- fines. Nevertheless, there is a definite amount of post-combustion of attrited fines; results are therefore presented within those error limits. Similar assumptions of post-attrition combustion was used by Arena et al. (1990) in their study of attrition in fast beds. To verify elutriation of fines from the bed, a handful of carbon below 50 ,Y in diameter was mixed with average-bed particles present in the bed at 1073 K and at the operating velocity of attrition tests. Rapid elutriation of finer particles was observed. A separate test was also conducted at 1073 K using nitrogen. This was done to determine the rate of mechanical attrition at the operating temperature and velocity for combustion-assisted attrition tests. A batch of ten spherical electrode carbon particles of known weight and diameter were dropped inside the bed, which was maintained at 1073 K. Particles were attrited for 30 min. Then they were taken out using the basket and quenched in cold nitrogen. Attrition rate was determined from the observed weight loss rate. A series of tests was also carried out at the room temperature to identify the parameters which influence attrition rate. In bubbling beds, it is generally assumed that the attrition rate is proportional to the exposed surface area of the carbon particle as well as to the excess gas velocity over minimum fluidization velocity. To check the validity of the above dependence in turbulent fluidization, tests were carried out in a turbulent bed at room temperature. A batch of five spherical electrode carbon particles of different diameters and weights was attrited inside the cold bed for 1 h. After the test, the particles were measured for diameter and weight. The bed was maintained at five different velocities in the turbulent-bed regime. Experimental conditions are given in Table 1. Results are discussed in the next section.

RESULTSANDDISCUSSION

Fig. 1. Experimental set-up showing air pre-heater, cylindrical furnace, and open-top turbulent fluid&d bed.

All tests were carried out in the turbulent regime. To verify the regime of fluidization during hightemperature studies, static pressure signals were taken from the top and from the bottom of the bed. The gas

Attrition of spherical electrode carbon particles

529

Table 1. Experimental conditions for measuring attrition in a turbulent fluid&d Combustionassisted attrition tests Gas velocity (m/s) Bed temperature (K) Inlet O2 concentration

Mechanical attrition tests at high temperatures

1.7 1073 5.2, 10.5, 15.7 and 21

(%)

Sand size (JJ)based on aperture opening of the sieve Bed voidage

Cold-bed attrition tests

1.7 1073 0

212-355 0.9687

bed

1.8-3.77 300 21

212-355 0.9687

212-355 0.9684.97

20

c

18

70. 1

BD-

-

14

-

gg 12. 0 =

so-

2

10 6-

40.

6-

so-

4

.

al., 0

a2

. . . . a4

. .

0.B

0.6

.

1.0

1.2

supERFlcuLoAs-

*.

1.4

.

‘.

*.a

1 1.a

was changed

slowly

and the amplitude

of the

pressure fluctuation was noted. The amplitude of pressure fluctuation is plotted against the velocity in Fig. 2. It peaks at a certain velocity and then begins to level off at higher velocities. The velocity at which the amplitude of the pressure fluctuation peaks is known as ZJ,and the velocity at which the amplitude of the pressure fluctuation levels off is known as 1.4~~ Turbulent fluidization is believed (Avidan, 1980) to extend from ulr to the transport velocity, y,. The experimental velocity exceeds uk, confirming the existence of the turbulent fluidization. THEORY

At present, only limited information on attrition in a turbulent fluidized bed is available_ The relationship between attrition rates and parameters like fluidizing velocity, particle size and weight is not very clear. Donsi et al. (198 1) derived the following equation for carbon attrition in a bubbling fluidized-bed combustor: E, = K,(U,

0.08

0.12

O-18

0.2

0.24

w, ug j&v

(ksIlS>

0.26

0.32

I 0.36

Fig. 3. A plot of carbon attrition rate, E,, vs [ W,U,/D,,] as measured at 300 K in a bed of 212-355 pm sand. The fluidizing velocity was 2.8-3.77 m/s.

wm

Fig. 2. A plot of amplitude of pressure fluctuation v-ssuperficial gas velocity. The test was conducted with air as the lluidizing medium. The sand size was between 212 and 355 g. The temperature of the bed was 1073 K.

velocity

-

21 0.04

20.

10

/

16

-

u,,gk d”

(1)

The assumption in the derivation of the equation is that the instantaneous rate of detachment of carbon fines from the surface of a carbon particle is proportional to the exposed surface area of the particle as well as to the excess of gas velocity over the minimum fluidizing velocity. In a bubbling fluidized bed, the attrition may be attributed to the relative velocity between the carbon particles and neighbouring particles. In an incipiently fluidized bed, the relative velocity between particles is negligible. The bubble motion, which is proportional to (U, - U,), stirs up the emulsion phase causing abrasion between particles which leads to the attrition. However, bubbles and emulsion phases are absent in a turbulent fluidized bed. A turbulent bed is characterized by vigorous motion of agglomerates and interconnected gas pockets. The exact mechanism controlling the relative velocity between the particle agglomerates and the coarse-attrited particle is not fully understood. However, it would be fair to assume that the abrasion between the particle agglomerates and the attrited particle would increase with the gas velocity, Ug. As a first approximation, we take the attrition to depend on the first power of U, so that eq. (1) can be written for turbulent iluidization as

In view of the above assumptions, a special experiment was conducted to examine the validity of relationship (2). The measured attrition rate, EC, is plotted

Petroleum coke

South African coal

Snibstone

Electrode carbon Electrode carbon

Electrode carbon

Bubbling

Bubbling

Bubbling

Turbulent Turbulent

Bubbling

Massimiha et al. (1985)

Massimilla et al. (1985)

Canunarota et al. (1985)

Basu and Subbarao (1986)

Present experiment

Halder et al. (1988)

Fuel

Type of bed

Researcher

1.0 1.0 1.0

1.7 1.7 1.7 1.7 1.7 2.8-3.8

1123 1123 1123

1073 1073 1073 1073 1073 300

309

1123 1123 1123 1123

0.8 0.8 0.8 0.8 0.58-2.15

1123

1123 1123 1123

1123 1123 1123 1123

Temperature (K)

0.8

0.8 0.8 0.8

0.8 0.8 0.8 0.8

Velocity (m/s)

0

0 5.2 10s 15.7 21 21

21

10 4.5 1.4 0

21

350 350 350

184 284 284 284 284 284

240

350 350 350 350

350

350 350 350

350 350 350 350

0 4.5 10 0 2.9 4.5 21

Sand size (P)

0, W)

Experimental conditions

Table 2. Attrition rate constants in bubbling and turbulent beds

0.05 5.59 3.37

0.21 4.8 3.0 2.92 2.57 0.05

0.1

0.95 0.83 0.79 0.31

0.5

2.69 2.3 1.9

0.05 3.28 5.73 0.13

k,,x 10’

0

2

Attrition of sphericalelectrode carbon particles against U, WC/D,, in Fig. 3. The linearity of this variation supports the validity of the above relationship. The presence of combustion complicates the attrition process. Combustion at the outer surface of the carbon does not occur uniformly. As a result, some parts are burnt faster than others leaving asperities on the carbon surface. These fine ridges are separated from the surface either through combustion with other particles or through burning away of the roots of the ridges. This enhances the rate of attrition and it is known as combustion-assisted attrition. The other complication lies in the measurement of attrition rate because the attrited carbon fines burn when passing through the hot bed in the presence of oxygen. The weight loss of a carbon particle burning in a bed may be written as the sum of weight loss due to combustion of the parent particle and that lost through attrition: kUW

dW, dt

=

*+%-

0”

where I& is the burning rate of the parent carbon particle having diameter D,,, which is known from the flue gas analysis. Before leaving the bed, a part of the attrited carbon fines burns. This amount is not of immediate interest to the designers whose primary concern is how much of the attrited carbon fines is lost through entrainment. Furthermore, it was shown earlier that the extent of post-attrition combustion is expected to be less significant in the present case. Burning rates l@ were determined from observed gas analysis in turbulent fluidized beds. Though there may be some error as the particle diameter varied during bum-out, clearly the results suggest higher burning rates than those observed in bubbling beds. Basu (1984) and Basu and Subbarao (19f36) also noted higher burning rates in a turbulent bed. Burning rates were compared with Basu (1977) and Halder et al. (1988). Attrition rate constants evaluated from eq. (3) are shown in Table 2 where experimental results are compared with those of other fuels. Experimental results agree well with those obtained in bubbling beds for identical fuels. Lower values of attrition rate constants when compared with bubbling beds for 5% and 10% inlet oxygen concentrations may be attributed to the increase in post-combustion and/or the lower average size of sand particles used for experiments in the turbulent bed. It is apparent that attrition rate constants for burning particles are many times higher than those observed for mechanical attrition. This supports earlier observations in a bubbling fluidized bed (Halder et al., 1988).

RESULTS

Table 2 presents a list of attrition constants calculated from the attrition rates of electrode carbon measured over a range of operating conditions in the turbulent fluidized bed. These data are contrasted with the attrition rate constants determined by CES

47:3-B

531

Halder et al. (1988) from experimental data on the same carbon but in a bubbling Auidized bed. We also present data for bubbling beds for a few other fuels (Massimilla et al., 1985) for comparison. To set the stage for a discussion on the present results, we start with a few comments on their results of bubbling beds. It is generally believed that the rate of attrition increases as the burning rate increases with oxygen concentration. Most fuels exhibit this behaviour; but some do not. This can be seen from Table 2. Attrition rate constant has a peak somewhere between 0 and 21% O2 for both Snibstone coal and South African coal. Electrode carbon also exhibits the same behaviour. This can be explained from the R, vs X, relationship of the char (Massimilla et al., 1985) where X, is the degree of carbon conversion at the particle surface and R, is the rate of removal of weakened solid layers from the particle surface. Carbon attrition rate is proportional to the amount R,(l - X,) of the unconverted material being detached, per unit time, as the external surface moves inward on the particle. Considering that R, increases while (1 - X,) decreases when oxygen concentration becomes higher, a maximum in the product R,(l - X,) may eventually be expected at some intermediate oxygen concentration in the experimental range as a consequence of opposite effects (Salatino, 1987). In the case of bubbling beds, the post-combustion of attrited fines is minimum because the freeboard of the bed is chilled. The data on the mechanical attrition in a turbulent bed show that its rate constant increased from 0.05 x lo-’ to 0.21 x iOm7 when the bed temperature was increased from 300 to 1073 K. The hot bed was fluidized by nitrogen and hence there was no combustion of any kind. Thus, this effect of temperature is exclusively on mechanical attrition. The oxygen concentration was varied from 5 to 21% by mixing oxygen with nitrogen. In the presence of oxygen at combustion temperature (1073 K), the attrition rate constant jumped to a range between 2.57 x lo-’ and 4.8 x lo-‘. These values are an order of magnitude higher than that found in the absence of oxygen, but at the same high temperatures. This rise is clearly due to the combustion-assisted attrition; nonuniform combustion on the carbon surface creates large asperities which then break loose. As explained earlier, these constants include the post-attrition combustion of fines. As a result, the values shown here are lower than actual attrition rates. The difference between the apparent and actual attrition rate constants is the extent of combustion of fines. The effect of oxygen concentration as discussed earlier may be to some extent shadowed by the varying degree of postattrition combustion of fines. An interesting observation made from the present results is that the attrition rate constant for mechanical attrition in bubbling beds (i.e. in the absence of combustion) for all fuels and all temperatures varies between 0.05 x lo- ’ and 0.31 x fOe7. For a turbulent fluidized bed under similar operating conditions, it is 0.21 x lo- ‘. This value is not considerably different

532

P. K. HALDER

from those in bubbling beds especially when we are relying on only one set of data points. Now if we look at the attrition under combustion conditions, we find that the attrition rate constant is 10-100 times larger than that for mechanical attrition. The attrition rate constant f6r electrode carbon is in the range 3.3 x lo-’ -5.7 x lo-’ in bubbling beds, while in the turbulent bed the constant is in the range 2.5 x 40 ‘-4.8 x LO-‘. So, we find a considerable overlap of attrition constant between the two regimes. Thus, the above similarity in the values of rate constant suggests an insensitivity of the attrition rate constants to the fluidization regime. This is a significant observation. If more experiments over a wider range of operating conditions can substantiate this conclusion, the attrition rates for fuels can be measured under the relatively simple experimental condition of bubbling beds and used in turbulent and fast fluid&d-bed conditions with the use of an appropriate equation like eq. (2).

CONCLUSIONS

(1) The attrition in a turbulent fluid*

bed can be expressed as E, = K, U, W,lD,, , where U, is the superficial gas velocity. (2) Combustion greatly enhances the attrition in a turbulent bed. The attrition rate constant, K,, under combustion conditions is (2.574.8) x 10-7, while that in the absence of combustion is only (0.03-0.05) x lo- 7. (3) A comparison of attrition rate constant in bubbling and turbulent beds determined using eqs (1) and (Z), respectively, suggests that these constants are independent of the regime of fluidization.

NOTATION

Drz” &I k, KZ U, u, uk

average carbon particle diameter, m elutriation rate of carbon, kg/s attrition rate constant, dimensionless rate of regression of particle external surface, m/s velocity at which transition turbulent fluidization starts, m/s gas velocity, m/s velocity at which transition to turbulent fluidization becomes complete, m/s minimum fluidization velocity, m/s transport velocity, m/s weight of the carbon particle in the bed, kg degree of carbon conversion, dimensionless

and P. BASU REFERENCES

Arena, U., Cammarota, A., Sicilliane, L., Massimilla, L. and Basu, P., 1990, Attrition of char in CFB combustors. Combust. Sci. Technol. 73, 383-394.

Arena, U., D’Amore, M. and Massimilla, L., 1983, Carbon attrition during the fluidized combustion of a coal. A.I.G.E.

J. 29, 40.

Avidan, A. A., 1980, Bed expansion and solid mixing in high

velocity fluidized beds. Ph.D. dissertation, The City College, New York. Basy P., 1977, Burning rate of carbon in fluidiied beds. Fuel 56, 390. FXasu, P., 1984, Measurements of burning rates of carbon particles in a turbulent fluidized bed, in Proceedings of 3rd International Ffuidized Bed Combustion Conference. Dis. cussion, The Institute of Energy, London. Basu, P. and Subbarao. D., 1986, An experimenral investigation of burning rate and mass transfer in a turbulent bed. Combustion and Flame 66, 261-269. Beer, J. M., Massimilla, L. and Sarofim, A. F., 1980, Fluidized coal combustion: the effect of coal type of carbon load and carbon elutriation. Inst. Fuel Svma. Ser. 4. IV 5.1. Cammarota, A., Chirone, R., D’Amo&, &I. and Massimilla, L., 1985, Carbon attrition during the fluidized combustion of coals, paper presented at 8th International Conference

on FBC, Houston. TX. Cammarota, A., D’Amore, M., Donsi, G. and Massimilla, L., 1981, Attrition of carbon particlesin fluid&d combustion. Fluidized Combustion Conference, Energy Research Institute, Cape Town, South Africa. Chakraborty, R. K. and Howard, J. R., 1978, Burning rates and temperatures of carbon particles in a shallow fluidized bed combustor. J. Inst. Fuel S&220-224.

Chirone, R., Cammarota, A., D’Amore, M. and Massimilla, L., 1982, Fragmentation and attrition in the fluidiid bed combustion, p. 1023. U.S. Department of Energy, Washington, DC. Chirone, R., D’Amore, M., Massimilla, L. and Mazza, A., 1985, Char, attrition during the batch fluid&d bed combustion of a coal. A.I.Ch.E. J. 31. 812. Donsi, G., Massimilla, L. and Miccio, M., 1981, Carbon fines Droduction and elutriation from the bed of a fluid&d coal combustor. Combustion and Flame 41, 57-64. Halder, P. K., Salatino, P. and Arena. U.. 1988. Attrition of spherical elkrode c&bon particles-during b&h fluidii combustion. Can. J. CJlem. Engng 66, 163. La Nauze, R. D. and Jung, K., 1982a, Combustion kinetics in fluidized beds, in Proceedings of 7th International Conjkrence of Fluidized Bed Combustion, Philadelphia, p_ 1040. La Nauze, R. D. and Jung, K., 1982b, The kinetics of

combustion of petroleumcoke particlesin a fluid&d bed

combustor, in i9th Symposium-International on Combustion. The Combustion Institution. D. 1087. Lancia, A., Nigro, R., Volpicelli, G: and Santoro, L., 1988, Transition from slugging to turbulent flow regimes in fluidized beds detected bv means of caoacitance nrobes. . _~ ~~ Powder Technol. 56, 45&S&. Massimilla, L., Chirone, R., D’Amore, M. and Salatino, P., 1985, Carbon attrition during the ffuidized combustion

and gasification of coal, Final Report-DOE Grant No. DE-FG22-81PC40796. Dipartimento di Ingegneria

Chin&a-Universita di Napoli. Turnbull, E. and Davidson, J. F., 1984, Fluidbustion of char and volatiles from coal. A.I.Ch.E. 881-889.

comJ. 30,