Development and characterization of a lab-scale entrained flow reactor for testing biomass fuels

Development and characterization of a lab-scale entrained flow

reactor for testing biomass fuels

E. Biaginia, M. Cionib, L. Tognottia,*

aDipartimento di Ingegneria Chimica, Chimica Industriale e Scienza dei Materiali, Universita di Pisa, via Diotisalvi, 2-Pisa 56100, ItalybEnel Produzione Ricerca, Pisa, Italy

Received 22 October 2004; received in revised form 2 February 2005; accepted 3 February 2005

Available online 17 March 2005

Abstract

Alternative fuels exhibit different features respect to traditional fuels and require an experimental characterization in conditions similar to

those of practical applications (high temperature, high heating rate, low residence time).

In this work, a lab-scale drop tube reactor is characterized and an experimental procedure is developed to test a bituminous coal and a

biomass fuel at high heating rate in oxidative conditions. Thermogravimetric, size and SEM analyses are used to determine the conversion

degree, the reactivity and the morphological variations (swelling, fragmentation, agglomeration) of solid residues in different operating

conditions.

Furthermore, a model is developed in order to simulate the fluidynamics, the energy balance and the mass transfer during the partial

oxidation of fuel particles. The application of this model allows the residence time and the thermal history of the particle inside the drop tube

to be estimated. The experimental and model results are in agreement, considering both configurations, namely, constant diameter and

density models.

q 2005 Elsevier Ltd. All rights reserved.

Keywords: Solid; Fuel; Combustion

1. Introduction

Biomass fuels cause problems from both economical and

technological points of view during the preparation

(seasonal availability, transport, drying, milling) or during

the combustion process (ignition, slagging, emissions)

because of specific characteristics: low calorific value,

high moisture content, presence of alkali compounds, high

ash content. Despite these drawbacks, the advantages are of

current interest: biomasses represent a renewable energy

source, are CO2 neutral fuels and give lower emissions of

SO2, NOx and heavy metals respect to coals, residues can be

used as an energy source (instead of a more and more

problematic disposal), big quantities of biomasses on a

geographic local scale can be exploited. As a matter of fact,

they do not satisfy the requirements for a safe and optimum

0016-2361/$ - see front matter q 2005 Elsevier Ltd. All rights reserved.

doi:10.1016/j.fuel.2005.02.001

* Corresponding author. Tel.: C39 050511240; fax: C39 050511266.

E-mail address: [email protected] (L. Tognotti).

performance in existing boilers. So, the use of these fuels

requires a high flexibility of the boiler. Otherwise,

alternative processes (co-combustion, pyrolysis, gasifica-

tion) can be taken into account [1–4].

Peculiar features of these fuels during the thermal

treatments should be investigated in detail in order to obtain

an exhaustive characterization and provide important

information on their behaviour during the different steps

of the process. The reactivity of these fuels during the

devolatilization is generally higher respect to traditional

coals, with a release of a large amount of volatile

compounds at relatively low temperature. Also, volatiles

released by biomasses contain more oxygen respect to coals

and this can be of interest for pyrolysis processes with the

aim of chemical recovery [5,6]. In combustion units,

ignition is favoured by the release of volatiles at low

temperature, making biomass fuels attractive for cofiring

with low volatile coals. However, the flame stability can be

compromised [7]. Chars obtained after the devolatilization

of biomass fuels are more reactive than those obtained from

coals [8]. Even, the direct oxidation of the parent material

can be a competitive mechanism respect to the classical

Fuel 84 (2005) 1524–1534

www.fuelfirst.com

Fig. 1. A schematization of the drop tube reactor: 1 feed inlet, 2 secondary

air inlet, 3 gas exit, 4 heating element, 5 water cooled head, 6 flow

straightener, 7 injector tip, 8 quartz tube.

E. Biagini et al. / Fuel 84 (2005) 1524–1534 1525

devolatilization/char oxidation path [9]. Size and morpho-

logical variations are of crucial importance for each step of

the entire combustion process [10,11].

It is important to develop methods for the experimental

characterization of these alternative fuels which exhibit so

different features respect to traditional fuels. In fact, the

usual thermogravimetric (TG) analyses should be used only

for a preliminary characterization. Experimental facilities

and techniques performing operating conditions more

similar to those of practical applications (high temperature,

high heating rate, low residence time) are suitable to provide

fundamental data for the optimization of the parameters and

the fuel properties for large scale systems [12].

In this work, a lab-scale entrained flow reactor is

developed and characterized in order to study the behaviour

of different alternative fuels (biomass, agricultural, munici-

pal and refuse derived fuels) during the first steps of

combustion. This experimental solution combines high

performances (moderate to high heating rates, low residence

times, continuous feed of the sample) typical of large scale

reactors with a relative simplicity in both operation handling

and interpretation of results [13–16]. Some preliminary

results are discussed in the following as for the conversion

and the properties of solid residues of a biomass fuel. Also, a

traditional coal is studied for comparison. Furthermore, a

model is developed to simulate the behaviour of a fuel

particle during the experimental run and thus to assess the

effective residence time, temperature and heating rate, mass

and size variations.

The present study is part of the European Community

project BioFlam (project no.: NNE5-1999-00449) on the

combustion of ‘clean’ fuels in power generation.

Fig. 2. A schematization of the feeding device: 1 sample glass tube,

2 pulverized sample, 3 motorized raising support, 4 primary air inlet,

5 gas/solid exit to the drop tube, 6 filter, 7 safety valve.

2. Experimental section

2.1. Experimental apparatus

The Drop Tube Reactor (DTR) is available at ENEL

Produzione Ricerca—Pisa—Italy. It is constructed using a

quartz tube (0.6 m long, 30 mm internal diameter) inserted

into a vertical oven and fixed to the head as shown in Fig. 1.

The oven is electrically heated and insulated for 0.4 m in

length. The maximum temperature is 1473 K. The head is

cooled with water and has two inlets of primary and

secondary air, respectively. At the top of the quartz

tube is centrally positioned a stainless steel tube (1 mm

i.d.) through which the solid fuel is pneumatically

transported with the primary air flow from the glass tube

(illustrated in Fig. 2). The primary air is maintained at 1 Nl/

min (1.7!10K5 m3/s n.c.), which guarantees a constant

flowrate of 0.15 g/min of pulverized material in the DTR.

The secondary air flow can be fed to the DTR and this is

used to increase the air flow inside the reactor and allows also

for variations of the residence time. The secondary air can be

varied between 0 and 2 Nl/min (0–3.3!10K5 m3/s n.c.).

A small glass cyclone is installed at the exit of the DTR

in order for a preliminary separation of the biggest particles

of solid residue. Finally, a fine glass filter is placed to stop

the smallest particles.

The materials successfully fed to the DTR are in the

range 150–300 mm, i.e. a size of interest for practical

applications (pulverized furnaces or fluidized beds).

Fig. 3. Thermal profile along the axis of the drop tube at different nominal

temperatures (973–1273 K). Air flowrate: 1 Nl/min.

Table 1

Ultimate and proximate analyses of materials

Fuel Ultimate analysis (% d.a.f.) Proximate analysis (%)

C H N S Mois. VM FC ash

Hazelnut

shell

51.0 5.4 1.3 – 7.0 73.0 18.8 1.2

Coal

Kema

71.4 4.5 1.1 0.8 5.7 28.7 52.6 13

E. Biagini et al. / Fuel 84 (2005) 1524–15341526

This depends on the present configuration of the feeding

device: a smaller dimension of the particles creates a

problematic pneumatic transport because packing is more

likely to occur, whereas a larger dimension results in a

discontinuous feeding of the solid particles.

2.2. Characterization of the experimental apparatus

The temperature profile inside the DTR is measured with a

thermocouple and related to the nominal value indicated by

the control system of the oven. A shielded thermocouple is

used in order to avoid the interference of radiation from the

resistances of the oven and this allows for a more reliable

measurement of the gas temperature inside the DTR. The

same air flow is maintained during all measurements varying

the nominal temperature between 973 and 1273 K. The

results are reported in Fig. 3. Also, the effect of the air flow is

evaluated measuring the thermal profile varying the

secondary air flowrate at constant nominal temperature.

In all cases, the thermal profile has a very narrow interval

of isothermal conditions. In fact, only a length of 0.05–

0.10 m at the centre of the DTR can be considered practically

isothermal and 5–10 K lower than the nominal value of the

oven. This behaviour is caused by the thermal dispersions

due to the effects of the extremities, that is the fin at the

bottom of the DTR and the forced cooling of the head.

Also, the air flowrate affects significantly the thermal

profile: the higher the air flowrate, the more remarked the

effect on the dispersions at the extremities: differences even

of 50 K are observed doubling the reference value of the air

flowrate, from 1 to 2 Nl/min. Vice versa, negligible effects

are practically noticed at the centre of the DTR, in all

conditions of air flowrate.

The residence time cannot be varied significantly in the

actual configuration of the DTR, because no collecting

probe is available nor partition in the heating elements of the

oven is possible. It is measured only with a rough precision

and at room temperature: it results in the range 0.5–1 s for

particles of 150 mm. However, a modeling approach is

required in order to get more accurate information on

the thermal history of the particles inside the DTR in

different operating conditions.

2.3. Materials

Two materials are successfully used in DTR runs,

namely, a bituminous coal (coal Kema) and a biomass

fuel (hazelnut shells), owing to their characteristics and the

requirements of the feeding apparatus. The ultimate and

proximate analyses are listed in Table 1. The fraction used is

150–300 mm.

Also runs with paper sludge samples have been

attempted but the non-uniformity of the particle dimensions

and the tendency to packing of this material hindered a

continuous feed to the DTR. Improvements to the feeding

device are actually required in order to extend the study to

problematic materials. So, only the results for two classes of

fuels are discussed in the following, that is a traditional

fossil fuel and a lignin-cellulosic material.

The oxidation of fuels and solid residues collected after

DTR runs is carried out using a TG balance (Mettler TA

3000) with samples of 5 mg, dried at 378 K and performing

a constant HR program at 20 8C/min (0.33 K/s) till the final

temperature (typically 1073–1173 K). Air is fed to the

apparatus with a flowrate of 300 Nl/min (5!10K3 m3/s in

normal conditions), thus, assuring a high concentration of

oxygen on the sample and removing promptly the combus-

tion products.

3. Experimental results

The programmed experimental runs differ for the

nominal temperature of the DTR, which represents

the most important operating parameter to evaluate the

conversion and the variations occurring during the

experiments. The characterization of the DTR revealed

that these tests are far from isothermal conditions.

Nevertheless, the simulation of DTR runs will allow to

evaluate the effective thermal history of each particles in

Section 4. Just as an indication, the estimated heating rate

(HR) on particles with size 150 mm at 1073 K is 6000 K/s,

that is a quite high value which allows to provide

important information on the behaviour of the fuel in the

initial steps of the combustion. The residence time is

approximately 0.4 s.

Fig. 5. Comparison of dtg oxidation curves of coal Kema and its solid

residue after partial oxidation in DTR at 1273 K.

E. Biagini et al. / Fuel 84 (2005) 1524–1534 1527

The solid collected after each run is weighted and

analyzed. Proximate, size and Scanning Electron Microscopy

(SEM) analyses of solid residues are carried out. The

conversion of the material in the different conditions is

evaluated considering the unburned matter which is measured

from TG analysis. As a matter offact, the conversion obtained

comparing the weight of the sample and that of the solid

collected leads to results with a poor accuracy. In fact, a

rigorous recovery of all the solid residue is hampered by the

adhesion of some material to the walls of the reactor and the

cyclone, especially for the biomass fuel. However, consider-

ing the TG analysis of the parent material and comparing the

results with the TG analysis of the solid residue, the content of

unburned matter can be evaluated. The final conversion is the

result of both a partial release of volatiles from the material

and a partial oxidation of the solid material, because an

exclusive devolatilization cannot be actually performed. The

conversion is calculated comparing the content of unburned

matter in the dry material and in the solid residue.

The conversion of hazelnut shells and coal Kema in the

reactor temperature range of 973–1273 K is reported in

Fig. 4. The conversion of both classes of fuels is comparable,

conversion of hazelnut shells being higher than for coal only

at the highest temperatures (1173–1273 K). However, at

973 K the conversion of the biomass is extremely low. This

can be in contradiction with the reactivity of the materials,

because hazelnut shells are expected to be more reactive than

coal. At least two considerations are useful to explain this

result (which is reproducible as a result of 3–4 runs) and are

derived from experimental confirmations which will be

reported in the following.

First of all, the effective dimension of the biomass

particles is quite larger than for coal, even if the nominal

fraction is the same (150–300 mm). This will be confirmed

by the dimensional analysis on the parent materials reported

in the following. Obviously, a larger particle experiences

a minor residence time in the DTR and, also, has a higher

thermal inertia respect to a smaller particle. As a

consequence, the larger particles will reach a lower

Fig. 4. Conversion of hazelnut shells and coal Kema (particle size 150–

300 mm) in drop tube runs.

temperature during the experimental run and this gives a

lower final conversion.

Furthermore, SEM analysis will reveal that biomass

residues obtained at low temperature are covered by a

stratum of condensed matter. In this case, the temperature

achieved by the particles can be supposed high enough for

the devolatilization of the large volatile content of the

biomass but not for the subsequent oxidation. Even, the heat

required for the devolatilization can reduce the temperature

of the particle during the process. Once released, the volatile

compounds reach the outer surface when the temperature of

the particle is decreasing. Consequently, they condense

giving a limited weight loss and, thus, a low conversion.

The derivative weight loss curves for solid residues

obtained after DTR runs from coal Kema are compared with

the parent material in Fig. 5. A higher value of the final

weight loss is noticed increasing the temperature of the drop

tube run. This is due to the enrichment in ash of the solid

residues as the material is consumed for both devolatilization

and oxidation mechanisms during the run. An evident delay

in the weight loss can be observed from the dtg curves of the

solid residue (at the reactor temperature of 1273 K), the

temperature for the maximum rate of weight loss passing

from 804 to 840 K. A similar behaviour can be found also

comparing the TG oxidative analysis of the solid residue at

973 K and the parent material for hazelnut shells (Fig. 6).

Fig. 6. Comparison of dtg oxidation curves of hazelnut shells and its solid

residue after partial oxidation in DTR at 973 K.

Fig. 7. SEM analysis of coal Kema and solid residues after DTR runs at

different reactor temperatures.

E. Biagini et al. / Fuel 84 (2005) 1524–15341528

In this case, the behaviour is that typical of biomass fuels [9,

17–19] with two different macro-steps, the first at

lower temperature imputable to the devolatilization and the

second at higher temperature imputable to the oxidation of

the char. From the comparison of the dtg curves of the DTR

solid residue and the parent material, a delay in the

characteristic temperatures can be noticed, that is from 583

to 604 K for the devolatilization peak and from 679 to 718 K

for the char oxidation peak. Furthermore, comparing the

weight loss of the DTR solid residue and the parent material,

the weight loss due to the devolatilization is more remarked

respect to the char oxidation, so that the correspondent DTR

run leads to a prevalent devolatilization. However, in

this case the volatile content of the solid residue is large

and this confirms the previous hypothesis on the limited

conversion of the biomass at low temperature DTR runs. The

macro-peak imputed to the devolatilization decreases

considering the oxidation of solid residues at higher

temperature.

These results can be explained considering the trans-

formations occurring during the thermal treatment. The

decrease in the dtg peak is obviously imputable to the lower

volatile matter (VM) content of solid residues respect to the

parent materials caused by the removal of volatiles during

the DTR run, and this is due to both devolatilization and

oxidation mechanisms. The delay of the dtg peak can be

ascribed to chemical transformations of the solid (recombi-

nation of the carbonaceous matrix) or, even, to morpho-

logical variations of the structure (for instance, porosity

evolution, softening of the material, swelling of the

particle). A less reactive solid residue is the result in both

hypotheses.

SEM images of DTR solid residues from coal Kema are

reported in Fig. 7 and are compared with those of the parent

material. A dramatic scenario can be observed even at the

mildest thermal conditions, that is, at the reactor tempera-

ture of 973 K. The particles assumed a spherical shape,

several cracks emerge at the surface. The internal structure

is visible from some fragmented particles. The particle is

mostly empty, the outer shell being still quite thick. As the

reactor temperature increases, this shell becomes more and

more riddled and thin. At the highest reactor temperature

(1273 K), large holes emerge at the particle surface. The

outer shell is perforated in several points due to the

consumption of the solid matrix.

These observations are deepened by the size analysis

carried out on the solid residue collected. This is done using

a Fritsch particle size Analysette 22 available at ENEL

produzione ricerca in Pisa. The size cumulative curve for

the solid residues collected at different nominal tempera-

tures is compared with the size cumulative curve of the

parent material in Fig. 8. The particles become larger at

the lower temperatures. Therefore, the prevalent factor is

the swelling of the particles due to the rapid release of

volatiles (from both devolatilization and solid oxidation),

which occurs simultaneously with the softening of

the material. Also, Hurt and Davis [20] noticed an increase

in the swelling factor for low rank coals. The particle size

increases for solid residues obtained till 1173 K, whereas at

higher temperatures, the particle size becomes smaller and

this can be the result of two mechanisms: the consumption

of the outer shell due to the oxidation and the fragmentation

of the particles owing to their thinner surface (as observed

from SEM analysis).

The dimensional analysis results of hazelnut shells and

their residue after partial oxidation in DTR at 973 K are

compared in Fig. 9. Results for the solid residues at higher

Fig. 8. Dimensional analysis of the solid residues after partial oxidation in

drop tube at 973–1273 K for coal Kema.

E. Biagini et al. / Fuel 84 (2005) 1524–1534 1529

temperatures are very close to that reported in the graph. A

decrease of the smallest particles can be observed in this

case from the cumulative curve of the solid residue.

However, the particle size decreases of 40% comparing

the average size of the solid residue with the parent material.

From SEM analysis (Fig. 10), residue particles differ

significantly from particles of the parent material. Several

melted surfaces can be observed in the solid residue, large

holes appear at the surface and a more spherical shape of

particles of the biomass residue can be observed as in the

case of coal Kema residues. Even, agglomeration products

are recognized, and this can explain the low number of small

particles from size analysis.

In the case of hazelnut shells, at the lowest reactor

temperatures, a sort of coat seems to cover the particles: this

can be due to the condensation of part of volatiles which had

not enough time to escape or, even, the temperature was not

sufficiently high to burn the volatile matter released. This

can be imputed to the higher VM content of the biomass

respect to the coal. In fact, a large and rapid release of

volatiles can generate a cloud which prevents oxygen to

reach the particle surface. Vice versa, at the highest reactor

temperature, more spherical particles with perforated shells

can be observed as in the case of coal Kema.

Fig. 9. Comparison of the dimensional analysis of the solid residues after

partial oxidation in drop tube at 973 K and the parent material for hazelnut

shells.

Fig. 10. SEM analysis of hazelnut shells and solid residues after DTR runs

at different reactor temperatures.

4. Modeling section

4.1. Development of the DTR model

The conversion of the fuel in the DTR should be related

to the residence time and the effective thermal history of the

particles. The simulation of main phenomena is suitable in

order to get a reliable interpretation of experimental results

and provide significant parameters for practical appli-

cations. Also, variables which cannot be easily measured

(the temperature history of the particle, its residence time

during the run and its mass) can be assessed this way.

Fig. 11. Schematization of the modeling approach to the simulation of the partial oxidation of a fuel particle in the drop tube reactor.

E. Biagini et al. / Fuel 84 (2005) 1524–15341530

For these reasons, a model is developed to simulate the

behaviour of a single particle in the DTR assuming the basic

relations for the dynamics, the energy balance and the mass

balance. A schematization of the algorithm comprehending

the set of hypotheses, the main equations and the objective

is reported in Fig. 11. The temperature profile obtained

experimentally is assumed as the gas temperature inside the

reactor. The jet theory is adopted to model the flow of the air

at the inlet of the DTR. The chemical and physical

properties of the material are known from direct analysis

and literature data.

The model assumes a single particle of the fuel falling

along the axis of the DTR. No interactions are supposed to

exist among the particles during the run. The particle is

considered a sphere fed to the reactor through a small orifice

with a velocity u0 (i.e. the same of the carrier gas). The

particle is subjected to a thermal transfer by both radiation

from the electrical resistances of the oven and convection

from the environment gas. The temperature of the particle is

assumed uniform, conditions used producing negligible

thermal gradient within the particle (Biot number

hd/k/1). The temperature achieved can cause the devola-

tilization and/or the combustion of the particle: in both cases

its mass and dimension decrease. The mass variation is

assumed to follow a global first-order kinetic, with

parameters evaluated from experimental investigation on

the direct oxidation of the material. In this case, kinetics

should be abstracted from experimental results obtained in

Fig. 12. TG isothermal runs for the oxidation of hazelnut shells.

E. Biagini et al. / Fuel 84 (2005) 1524–1534 1531

conditions similar to those encountered in the DTR [12]. In

particular, the heating rate is determinant and the values

calculated from constant HR runs in TG cannot be applied,

because they range between 0.1 and 2 K/s against values of

1000–10,000 K/s estimated for DTR runs. This is the reason

why isothermal TG runs are carried out in oxidative

conditions, because the higher HR achieved is more similar

to that expected in the DTR.

The TG balance described above is used for this purpose.

After the furnace reached the programmed temperature, a

5 mg of dried material is lowered into the furnace and the

sample weight variations are recorded as function of time.

The weight loss curves for hazelnut shells are reported

in Fig. 12 as functions of time. The temperature range is

523–673 K. The global kinetic parameters are abstracted

considering a first-order reaction model. The values are

49.2 kJ/mol for the activation energy E and 2500 sK1 for the

pre-exponential factor A.

Likewise, for coal Kema, isothermal TG runs are carried

out in the range 623–773 K, obtaining an activation energy

of 55.0 kJ/mol and a pre-exponential factor of 1530 sK1.

The model can be based on three sub-models which are

developed separately:

†

†

†

Fig. 13. Effect of the initial diameter Dp0 of the particle on the residence

time (tr), particle temperature (Tp) and diameter of the particle (Dp) along

the axial coordinate of the reactor. Hazelnut shells (initial diameter Dp0

150 mm G10%). Nominal reactor temperature: 1073 K. Constant density

configuration.

energy balance

These sub-models are correlated with the z coordinate

(along the axis of the DTR), the time and the variables

referring to the particle (temperature, mass, size, etc.).

Finally, from the model output, variables such as the

temperature of the particle in the DTR and the residence

time are obtained in different operating conditions and, also,

the variation of size and mass of the particle, so that the

conversion can be calculated and compared with

the experimental results. This allows for a validation of

the model confirming the hypotheses and the parameters

adopted and the thermal history of the particle can be related

to the conditions used.

Two different configurations are developed for the mass

sub-model: a constant density model and a constant

diameter model. From a phenomenological point of view,

a single particle is subjected to a fast heating in presence of

air during the DTR run. A constant density model could be

applied in these conditions, where the oxygen diffusion is

limited and the consumption of material reduces continu-

ously the external surface. However, biomass materials

contain a very high amount of volatile species which are

released at relatively low temperature. This mechanism acts

independently with the diffusion of oxygen and can even

remove the oxygen from the particle surface. A constant

diameter model can be applied in these conditions, where

the release of volatiles affects the entire particle, thus

emptying the internal structure which results in a decrease

of the solid density. Practically, both the direct oxidation

and the devolatilization mechanism act in competition and

influence the fluidynamics of the particle and the heat

exchange. The comparison of the results from these

configurations can simulate the extreme cases of what

realistically happens to the fuel particle in the DTR.

4.2. Modeling results

The model described above is applied to simulate the DTR

run for both fuels studied in this work. The aim is to verify the

conversion of a material considering its main physical and

chemical properties; also, to evaluate the residence time and

the temperature of the particle in the DTR. Furthermore, the

effect of the particle size and the transformations during the

process are evaluated. The reactor temperature is ranged

between 973 and 1273 K, so that 4 simulations spaced of

100 K are carried out for each fuel.

A typical output of the model is reported in Fig. 13 as for

the residence time, the temperature and the diameter of a

150 mm particle of hazelnut shell fed to the reactor at

1073 K. In particular, the effect of the initial size of

E. Biagini et al. / Fuel 84 (2005) 1524–15341532

the particles is evaluated for the constant density configur-

ation, considering variations of G10% respect to the

reference value. The results are reported as function of the

axial coordinate along the reactor.

To quantify, the residence time is 0.39 s for the reference

initial diameter, whereas it decreases to 0.36 s for particles

of 165 mm because of the larger initial mass and increases to

0.42 s for particles of 135 mm. These are actually global

values. In fact, the particle spends a longer time in the more

extreme zone of the DTR, because the velocity of the

particle decreases along the axis of the reactor due to

the expansion of the gas inside the DTR and the attrition of

the particle with the gas. This influences the temperature

trend of the particle, which is clearly delayed respect to the

temperature profile of the reactor shown above. Obviously,

also the thermal inertia of the particle determines this

behaviour. The different initial size influences also the

maximum temperature reached by the particle, that is

1027 K for the reference diameter, 1059 K for the smallest

particle, 995 K for the largest particle. This is due to the

different thermal inertia.

The particle remains unaltered (as for mass and size) till

the temperature is high enough to activate the chemical

transformations. After this point, the size decreases and

approaches an asymptotic value at the exit of the DTR, that

is from 150 to 96 mm in the reference case of Fig. 15, with a

less remarked decrease for the largest particle.

The same qualitative trend can be noticed for the

constant diameter model with density variations of G10%

respect to the reference value of 500 kg/m3. In general, an

increase of 10% on the initial values of the size or the

density decreases the residence time of 7–8%, the maximum

temperature achieved of approximately 3% and the conver-

sion of 3–4%.

Fig. 14. Effect of the model configuration on the residence time (tr), particle

temperature (Tp) and weight loss of the particle (W/Wo) along the axial

coordinate of the reactor. Hazelnut shells (initial diameter Dp0 150 mm,

initial density 500 kg/m3). Nominal reactor temperature: 1073 K.

The comparison of the model output for both constant

density and constant diameter configurations is shown in

Fig. 14 for a reactor temperature of 1073 K. Similar values

of the residence time (approximately 0.4 s) and maximum

temperature are obtained, even if in the case of constant size

a more rapid cooling is noticed. This leads to a lower final

conversion and can be observed from the weight loss

reported in Fig. 14.

The effect of the size and the model configuration on the

conversion reduces as the temperature increases and

practically are negligible at the highest reactor temperature

of 1273 K. The results of the simulations in the entire

temperature range allow to compare the conversions with

the experimental results, as reported in Fig. 15 considering

both configurations, i.e. constant density and constant

diameter model. The conversions calculated considering

particles maintaining a constant diameter are generally

lower than those calculated considering particles with

density constant, with maximum variations of 6% at

973–1073 K. At higher temperatures, both model configur-

ations give similar results, thus with no significant

differences. In general, a quite good agreement can be

observed comparing the experimental and the model results

for hazelnut shells. The experimental results lie between

both model configuration results at the lowest temperatures.

However, from these results the more appropriate

Fig. 15. Comparison of experimental conversions and the results of model

simulations at different reactor temperatures for hazelnut shells and coal

Kema.

E. Biagini et al. / Fuel 84 (2005) 1524–1534 1533

configuration cannot be recognized, also considering the

experimental reproducibility which is approximately 2–4%.

At higher temperatures, the discrepancies between the

experimental and the model results decreases, the effect of

the size being less remarkable. In all cases, the conversion

trend qualitatively agrees. Also from a quantitative point of

view, the results of the model are satisfying considering the

relatively simplex modeling approach.

Qualitatively, the sensitivity analysis on the effect of the

initial size and the model configuration for coal Kema leads

to similar results as in the previous case of hazelnut shells.

However, the effect of the temperature is more remarked as

observable in Fig. 15, where the experimental conversions

are compared with the model results in both configurations.

The variation of results considering both the model

configurations is 10% at the lowest temperature and reduces

as the temperature increases.

5. Conclusions

Alternative fuels exhibit different features respect to

traditional fuels and require an experimental characteriz-

ation in conditions similar to those of practical applications

(high temperature, high heating rate, low residence time). A

preliminary work on a lab-scale entrained flow reactor has

been carried out in order to develop an experimental

procedure for studying the behaviour of different classes of

solid fuels during the combustion. This experimental

solution combines high performances (moderate to high

heating rates, low residence times, continuous feed of the

sample) typical of large scale reactors with a relative

simplicity in both operation handling and interpretation of

results.

A lab-scale drop tube reactor has been characterized to

test a bituminous coal and a biomass fuel at high HR in

oxidative conditions. The most important handling and

operating conditions have been defined in both cases (for

instance, primary air flowrate and optimal particle size

range for the pneumatic feeding, degree of uniformity and

properties of the material to avoid differential fluidynamic

inside the reactor). Analyses on solid residues collected

after DTR runs (proximate, size and SEM analysis)

determined the conversion degree and the morphological

variation in different operating conditions. The reactor

temperature has been varied between 973 and 1273 K. The

global reactivity of solid residues is evaluated and compared

with that of the parent material. The SEM and size analyses

revealed important morphological variations (swelling,

fragmentation, agglomeration).

A model has been developed in order to simulate the

main phenomena occurring during the partial oxidation of

fuel particles: fluidynamics (adopting the round jet

theory), the energy balance and the mass transfer, which

determine the particle size (according to a constant

diameter or a constant density configuration) and, thus,

its characteristics during the entire run. The application of

this model allowed the residence time and the thermal

history of the particle inside the drop tube to be estimated.

The experimental and model results are in agreement,

experimental results lying between the model results in

both configurations, with maximum differences of 6% at

the lowest temperature.

The experience on running the drop tube reactor with

alternative fuels allowed to provide suggestions for further

activities, for modeling improvements and for future

design aiming to solve technical problems. Particular

attention will be focused on the feeding device (in order

to extend the versatility of the apparatus in order to study

different materials) and the collection probe (in order

to vary the residence time). Improvements to the

drop tube model could be made as for the mass balance,

for instance considering the competition between the

direct oxidation and the devolatilization/char oxidation

paths in preference to a global kinetics approach.

In particular, the different heat exchange involved

during both mechanisms (endothermic for devolatilization

and exothermic for oxidation) could be more realistic and

give a more accurate description respect to a global

balance.

Acknowledgements

The authors wish to thank ENEL Produzione Ricerca,

Pisa—Italy for the collaboration, the equipment and the

access to the laboratory. Special Thanks to Dr Linda Tomei

and Angelo Bianchi for the analyses performed.

References

[1] Werther J, Saenger M, Hartge EU, Ogada T, Siagi Z. Combustion of

agricultural residues. Prog Energy Combust Sci 2000;26:1–27.

[2] Sami M, Annamalai K, Wooldridge M. Co-firing of coal and biomass

fuel blends. Prog Energy Combust Sci 2001;27:171–214.

[3] Bridgwater AV. Renewable fuels and chemicals by thermal proces-

sing of biomass. Chem Eng J 2003;91:87–102.

[4] Biagini E. PhD thesis in Chemical Engineering—University of Pisa,

Italy; 2003.

[5] Horne PA, Williams PT. Influence of temperature on the products

from the flash pyrolysis of biomass. Fuel 1996;75:1051–9.

[6] Koufopanos CA, Maschio G, Lucchesi A. Kinetic modelling of the

pyrolysis of biomass and biomass components. Can J Chem Eng 1989;

67:75–84.

[7] Gurgel Veras CA, Saastamoinen J, Carvalho JA, Aho M. Combust

Flame 1999;116:567.

[8] Wornat MJ, Hurt RH, Yang NYC, Headly TJ. Structural and

compositional transformations of biomass chars during combustion.

Combust Flame 1995;100:131.

[9] Senneca O, Chirone R, Salatino P. Energy Fuels 2002;16:661–8.

[10] Arena U, Chirone R, Salatino P. Twenty-sixth symposium (inter-

national) on combustion 1996 p. 3243.

[11] Mitchell RE, Akanetuk AEJ. Twenty-sixth symposium (international)

on combustion 1996 p. 3137–44.

E. Biagini et al. / Fuel 84 (2005) 1524–15341534

[12] Biagini E, Fantozzi C, Tognotti L. Combust Sci Technol 2004;176:

685–703.

[13] Chenevert BC, Kramlich JC, Nichols KN. Twenty-sixth symposium

(international) on combustion, 1998. p. 1719–25.

[14] Storm C, Rudiger H, Spliethoff H, Hein KRG. J Eng Gas Turbines

Power 1999;121:55.

[15] Dacombe P, Pourkashanian M, Williams A, Yap L. Fuel 1999;78:1847.

[16] Zanzi R, Sjostrom K, Bjornbom E. Rapid pyrolysis of agricultural

residues at high temperature. Biomass Bioenergy 2002;23:357–66.

[17] Orfao JJM, Antunes FJA, Figueiredo JL. Pyrolysis kinetics of

lignocellulosic materials—three independent reactions model. Fuel

1999;78:349–58.

[18] Grotkjaer T, Dam-Johansen K, Jensen AD, Glarborg P. An

experimental study of biomass ignition. Fuel 2003;82:825–33.

[19] Calvo LF, Otero M, Jenkins BM, Moran A, Garcia AI. Heating

process characteristics of rice straw in different atmospheres. Fuel

Process Technol 2004;85:279–91.

[20] Hurt R, Davis K. Combust Flame 1999;116:662.