1

European Combustion Meeting (ECM2003) Plenary Lecture Biomass Combustion

J Swithenbank, YB Yang, C Ryu, J Goodfellow, S Shabangu, N V Russell, F M Lewis*, V N Sharifi

Sheffield University Waste Incineration Centre (SUWIC), Department of Chemical and Process Engineering, University of Sheffield, Sheffield S1 3JD, UK

* 535 East Mariposa Ave, El Segundo, CA 90245- 3013, USAABSTRACTSustainable cities require the generation of electrical energy from ‘carbon dioxide neutral’ biomass crops and suitable fractions of wastes that cannot be economically reused or recycled. The energy content of these solid materials can be recovered by burning directly or after processing into refuse-derived fuel (RDF). Alternatively, the combustion process can be staged by the production of intermediate fuels using either pyrolysis or gasification. Co-processing of the biomass with coal generally increases plant utilisation and thus reduces costs.The design, operation and maintenance of solid fuel combustion, pyrolysis or gasification plants requires detailed understanding of the processes occurring within a reacting packed bed of solids, or combustion of the derived liquids or gaseous fuels. Both of the latter can be modelled using well-established computational fluid dynamic codes (CFD). Previously, there has not been available a validated, comprehensive and fundamentally based code for mathematically modelling the combustion/pyrolysis/gasification process within a packed bed of solid particles on a stationary or moving grate. Bearing in mind that pyrolysis and gasification are sub-sets of combustion we have developed a generalised model of bed combustion. This code known as FLIC solves iteratively the flow field within a reacting bed of randomly packed particles, including radiant heat transfer. The equations governing the processes of drying, pyrolysis, de-volatilisation and char burnout within the particles are evaluated. Since the burning of volatiles and CO in the channels is mixing limited, flame reactions also occur in the gas phase above the bed. The conditions evaluated at the surface of the bed are the boundary conditions for conventional CFD modelling of the mixing and reactions in the secondary combustion zone in the freeboard above the bed and in the gas clean-up system. This permits the evaluation and minimization of emissions such CO, VOCs, NOx, heavy metals and dioxins. In fact, dioxins from incinerators now only contribute 3% of the total UK dioxin emissions. Pyrolysis of biomass/waste can generate a storable char fuel. This is achieved by heating a bed of the material slowly in a closed container to about 500°C from which air is excluded. This decomposes the organic material to release liquid products that can be condensed then purified and burned to efficiently generate heat and power. The carbon char remaining is a valuable fuel that can be easily separated whilst it is still hot from any inert material that was originally present. This storable fuel can be transported and used when the ‘renewable’ energy that it contains is required. The char from pyrolysis contains much of the original carbon and has a high-energy content. The validation of our reacting bed modelling code (FLIC) has been achieved by measurements in a pot burner using various biomass materials. Additionally, a small ‘ball instrument’ that has been specially developed to contain instruments has complemented these measurements by withstanding temperatures up to 1000oC for well over an hour. This novel device passes through industrial moving grate furnaces with the ‘fuel’ and records parameters such as oxygen concentration, vibration and several temperatures onto a computer memory chip. The ball is recovered from the ash pit and the information is downloaded onto an Excel spreadsheet for detailed analysis and verification of the FLIC predictions. Our new gasification concept offers the prospect of achieving the goal of high efficiency power generation from char, coal or various biomass sources by utilizing ultra superheated steam (USS). The method uses low-grade steam from sources such as process cooling, waste incineration and local industry, and then enhances it to a temperature above 1600ºC. This is achieved by adding oxygen to the steam to form "artificial air"; gas is then burned in this artificial air to produce an ultra-superheated steam flame. Char, biomass or powdered coal are injected coaxially into the high temperature steam flame where they react in about a second to produce a gas that is free of tar and consists largely of CO, H2 and CH4. Significantly, the high temperature steam provides the enthalpy for the endothermic gasification reactions. This feature ensures that tar formation is avoided and any ash is cooled below its melting point before it arrives at the reactor walls. This process provides a route to the future hydrogen fuel based economy. These innovative technologies form part of an integrated consortium programme towards sustainability.

2

BACKGROUNDWorldwide, there is a progressive change to use a greater proportion of our total energy consumption as electrical energy and citizens in developed countries now use up to about 1 kW of electricity per person. This corresponds to about 2.4 tonnes of coal per person per year. At the same time, citizens bring into being up to one tonne of municipal waste each year having an energy equivalent of 300 kg of coal plus even larger quantities of agricultural and industrial wastes. Bearing in mind that these wastes are largely biomass, the recovery of energy from the waste would not add very much to the net atmospheric CO2 level. Thus recovering energy from waste helps to mitigate the climate change problem. Furthermore, dedicated agricultural production of biomass for energy generation is increasing in popularity. Sustainable cities therefore require the generation of electrical energy from biomass including suitable fractions of waste that cannot be economically reused or recycled. The integration of these fuels with clean coal technology can also result in better utilization of the generation plant and thus reduce the cost of the electricity produced.

INTRODUCTION

The combustion of biomass can be accomplished either on a grate, in a fluidised bed, or as entrained particles. This presentation addresses the phenomena involved in their combustion on a static or moving grate. The four principle stages involved are: drying, pyrolysis/devolatilisation, oxidation of the volatile material and char combustion, finally leaving any residual ash. These processes can take place in separate stages as in pyrolysers or gasifiers however these are both subsets of the overall combustion process; hence they can all be modelled with a comprehensive combustion model. It is also relevant to point out that the total energy available is independent of the process stages used, however the conversion of the energy into electricity may be more efficient for certain processes. Figure 1 The design and operation of plants involving these processes requires the development of a mathematical model that is preferably based on the fundamental principles of physics and chemistry. The reliability of the model must be checked experimentally and the discussion below presents studies that have been carried out to achieve these aims.

Mathematical Description of Biomass Combustion on a Packed Bed A packed bed is an assembly of individual particles and consists of a solid phase (the particles) and a gas phase (gases flowing through the gaps between the particles). Theoretical calculation of the mass and heat transfer inside a packed bed is made complicated by three major factors:

1) The temperature profile inside a single particle is highly 3 dimensional (not 1-D in respect of the radial distance), especially for very thermally thick particles (the Biot number = hs’dp/λp >>1);

2) The number of individual particles in a bed is huge prohibiting calculations based on solving for individual particles; and

3) Lack of models calculating the mixing rate between the under-grate combustion air and the volatile gases released from solid devolatilization.

Bed Combustion ZonesBed Combustion Zones

BiomassBiomass

DryingDryingPyrolysisPyrolysis

Oxidation to CO2 & H2OOxidation to CO2 & H2O

Reducing CO2Reducing CO2

AshAsh

to COto COOxidationOxidation

to CO2to CO2CharChar

Char reaction Char reaction here reduces NOxhere reduces NOx

Bed Combustion ZonesBed Combustion Zones

BiomassBiomass

DryingDryingPyrolysisPyrolysis

Oxidation to CO2 & H2OOxidation to CO2 & H2O

Reducing CO2Reducing CO2

AshAsh

to COto COOxidationOxidation

to CO2to CO2CharChar

Char reaction Char reaction here reduces NOxhere reduces NOx

3

Other uncertainties include irregularity of the particle shapes, the process rate and channelling in the bed, etc. For a moving bed, theoretical calculations are further complicated by the movement and mixing of individual particles which are governed by friction between particles (dependent on particle shape and orientation), gravity, type of the grate and its moving pattern during bed operation. To make the mathematical calculation possible for a packed bed, it is assumed that the major bed properties, i.e., temperatures of gas and solid phases inside the bed, gas compositions (O2,H2, CO, CO2, etc.) and solid compositions (moisture, volatiles, fixed carbon and ash) can be described pseudo one-dimensionally as functions of bed height. It is also assumed that the bed can be treated as a porous medium where mass and heat transfer take place between the solid and gas phases and the shape of the particle is spherical (the surface-volume averaged diameter is used). Under such assumptions, the individual bed processes (moisture evaporation, devolatilisation and char burning) can be viewed taking place layer by layer, from the bed top to the bottom. Employing numerical methodology, the whole bed is divided into many small cells along the bed height and inside each cell the major bed parameters are assumed uniform. One benefit of this approach is that by reducing the cell size (hence increasing the cell number), calculation can be made on a size-scale much smaller than the fuel particles. This means that the non-isothermal behaviour of the single particles can be accounted for to some extent.

Transport equations for gas and solid phases. Peters (1995) has summarised the basic governing equations for both the gas and solid phases in a moving bed. For a stationary bed, the gas-phase equations can be written as follows:

Gas continuity: ( )

tg

∂∂ φρ

+( )

xVgg

∂∂ φρ

= Ssg (1)

where Vg is the gas velocity and x the coordinate along the bed height (x=0 at the bed bottom). The source term Ssg is the conversion rate from solid to gas due to moisture evaporation, devolatilisation and char combustion.

Gaseous species transport:

( )tYigg

∂∂ φρ

+ ( )

xYV iggg

∂∂ φρ

=( )

∂∂

∂∂

xYV

Dx

igggig

φρ+ Syig (2)

Yig represents mass fractions of individual species (e.g. H2, H2O, CO, CO2, CmHnOl, …). The source term Syigaccounts for mass sources of the individual species during evaporation, devolatilization and the combustion of volatile gases and char.

The fluid dispersion coefficient Dig is considered to consist of diffusion and turbulent contributions and is given by the following equation (Wakao & Kaguei 1982)

Dig = E0 + 0.5dpVg (3)

where E0 is the effective diffusion coefficient.

Gas-phase energy conservation:

( )

tHgg

∂∂ φρ

+ ( )

xHV ggg

∂∂ φρ

=∂

∂∂∂

xT

xg

gλ + Sa hs’ (Ts - Tg ) + Qh (4)

where Hg represents gas enthalpy, λg the thermal dispersion coefficient, and Qh the heat gain of the gas phase due to combustion. The thermal dispersion coefficient λg consists of diffusion and turbulent contributions in a similar way as species dispersion, and can be expressed as [20]:

λg = λ0 + 0.5dpVg ρg Cpg (5)

4

Where λ0 is the effective thermal diffusion coefficient.

The equations for the solid phase are:

Solid continuity: ( )( )

t1 p

∂−∂ ρφ

+( )( )

xV1 sp

∂−∂ ρφ

= - Ssg (6)

where ρp is the particle density and Vs solid velocity due to the downward movement of the bed caused by mass loss.

Conservation of solid-phase species:

( )( )t

Y1 isp

∂−∂ ρφ

+( )( )

xYV1 issp

∂−∂ ρφ

= - Syis (7)

where Yis represents mass fractions of particle compositions (moisture, volatile, fixed carbon and ash) and Syis the source term. Syis accounts for the loss of the individual components (moisture, volatile, fixed carbon and ash) during evaporation, devolatilisation and char combustion.

The energy equation for the solid-phase is:

( )( )t

H1 sp

∂−∂ ρφ

+ ( )( )

xHV1 ssp

∂−∂ ρφ

= ∂∂

∂∂

xT

xs

sλ + Sa hs’ (Tg - Ts ) +xqr∂∂ + Qsh (8)

where Hs presents the solid-phase enthalpy, λs is the effective thermal conductivity of the solid bed, and qr denotes the radiative heat flux. The source term Qsh accounts for the heat generation due to heterogeneous combustion.

Radiation Heat Transfer in the Bed. Radiation is the major mechanism of heat transfer between solid particles in a packed bed, and a proper model has to be developed to simulate the process. The already widely used flux model (Smoot & Pratt 1979) for gaseous and entrained-flow combustion is the first choice, although development of a more appropriate model is needed in the future. A two-flux radiation model is presented in the following:

dxdIx

+ = - (ka + ks) Ix

+ +21 ka Eb + 2

1 ks ( Ix+ + Ix

- ) (9a)

dxdIx

−− = - (ka + ks) Ix

- +21 ka Eb + 2

1 ks ( Ix+ + Ix

- ) (9b)

where Ix+, Ix-, represent the two radiation intensities. ka and ks denote the absorption and scattering coefficients respectively. Eb is black-body radiation.

ks is assumed zero as the first approximation, and ka is taken as ( Shin & Choi, 2000)

ka = − )ln(d1

pφ− (10)

More details of the model description can be found in Yang et al’s work (2001, 2002, and 2003).

Mixing Rate of the Under-grate Air with Volatile Combustible Gases Released from SolidsGaseous fuels released from the devolatilization process have first to mix with the surrounding air before their combustion can take place. Obviously the burning of the volatile hydrocarbon gases is limited not only by the reaction kinetics (temperature dependent) but also by the mixing-rate of the gaseous fuel with the under-fire air. The mixing rate inside the

5

bed is assumed to be proportional to energy loss (pressure drop) through the bed and by recalling the Ergun equations can be expressed as:

Rmix = Cmix ρg { 150( )

φ

φ2p

32g

d

1D − + 1.75

( )φ

φ

p

31g

d

1V −} min{

fuel

fuelSC

,2

2

O

O

SC

} (11)

here Cmix is an empirical constant, Dg the molecular diffusivity of the combustion air, Vg the air velocity, dp the particle diameter, φ the local void fraction of the bed, C the mass fractions of the gaseous reactants and S their stoichiometric coefficients in the reaction.

The actual reaction rates of volatile species are taken as the minimum of the temperature-dependent kinetic rates and their mixing-rates with oxygen:

R = Min[Rkinetic, Rmix] (12)

FLIC CODE PREDICTIONS OF BIOMASS COMBUSTION AND COMPARISON TO EXPERIMENTS: Predictions in a large scale moving bed. Based on the mathematical model described above, a special computer code, FLIC (Fluid Dynamics of Incinerator Combustion) was written to solve simultaneously the various parallel equations for solid fuel combustion and gasification in a packed bed. Figure 2 shows some of the results for a large-scale moving bed. Figure 2a) shows the temperature profile for the whole bed and it can be seen that ignition occurs about 2 meters from the fuel entrance. The reaction front is 150 – 300 mm (or 2.5 – 5.0 times the particle diameters) below the bed top. After ignition, intensive burning occurs above the bed and long flame tongues are observed. Further on along the bed length, the flame front travels further into the bed and an increasing proportion of the burning processes occur inside the bed. The flame front reaches the grate about two thirds of the bed length and remaining combustion (mainly char burnout) continues for a further two meters until the whole combustion is completed. Figure 2b) shows the moisture concentration inside the burning bed. It is seen that moisture evaporation occurs in a thin layer. This is because the high radiation flux from the flame front quickly heats up the wet layers inside the particles.

Figure 2a) Figure 2b) Figure 2c)

All the moisture in the bed is evaporated at about half of the whole bed length.

6

Figure 2c) shows the volatile matter profile inside the bed. The devolatilisation process occurs also within a narrow layer and all the volatile material in the bed is also released at about half of the bed length. Figure 2d) shows the individual process rates along the bed length. Moisture evaporation occurs as soon as the fuel is pushed into the burning chamber and exposed to freeboard flame radiation. However, it is not until at 2 meters from the fuel entrance that the top part of the bed has dried out and the temperature is raised above the

threshold for volatile matter release. The subsequent volatile release is intense due to the freshness of the fuel. After that, the devolatilisation rate drops to a stable level. At 6 meters from the fuel entrance, the volatile release rate rises again, due to the dried-out nature of the local bed leading to the raised bed temperature, before finally being reduced to zero. The char starts to burn after the initial intensive release of the volatile matter from the solids and the burning rates undergo a slow increase as the fuel moves along the bed. At 6 m from the fuel entrance, the char burning rate rises sharply due to the total release of the volatile matter and therefore the full access to the under-grate air supply that is no longer consumed by combustion of volatile gases. The whole combustion process is complete at 7.5 meters along the bed length. Figure 2d) also indicates that the whole burning process is divided into three stages: I – the ignition stage; II – the main stage; and III – the final char burnout stage. The “Ball Instrument”. To investigate the local combustion behaviour in a full-scale moving grate incinerator (burning largely biomass) for a range of operating conditions, we have developed a 'ball instrument' that passes through the bed with the waste. This small ‘ball instrument’ contains instruments and withstands temperatures up to 1000oC for well over an hour (Yang et al. 2001). This novel device passes through industrial moving grate furnaces with the ‘fuel’ and records parameters such as oxygen concentration, vibration and several temperatures onto a computer memory chip. The ball is recovered from the ash pit and the information is then downloaded onto an Excel spreadsheet for detailed analysis and verification of the FLIC predictions.FLIC code predictions generally compare very well with the data, but the measured temperature within the bed again shows that the process is dominated by many violent transient fluctuations from 300°C to 1000°C along the bed. It was deduced that these fluctuations were due to the formation and collapse of channels within the bed. A vibration transducer installed within the ball instrument confirmed this concept since the temperature

0

50

100

150

200

250

300

350

400

450

0 2 4 6 8Distance along bed length, m

Pro

cess

rate

s, k

g/m

2.hr

Moisture evaporation

Char burn-

Volatile rele

I II II

Figure 2d). FLIC prediction of individualprocess rates inside a large-scale moving bed.Initial fuel moisture 36%. Particle size: 60mm.Initial bed height 1050 mm.

0

200

400

600

800

1000

1200

1400

1600

0 2 4 6 8 10Distance along bed length, m

Test one Test two:T

Tmax

Tmin

Figure 2e). In-bed measurement of temperature using an electronic device and FLIC calculation of the maximum and minimum local-bed temperatures from the bed top to a distance of 250mm underneath in the full-scale moving bed

The The Ball InstrumentBall Instrumentpasses throughpasses throughthe burning the burning bedbed

7

fluctuations coincided with the mechanical disturbances. The significance of channelling is also apparent in the jets of flame that can be seen above the bed in incinerator plants. Clearly, it is not sufficient to model only the mean values of parameters in the bed, and unsteady aspects of the process have had to be attacked. To investigate this phenomenon, the next stage has been the successful modelling of the formation of gas flow channels by the random packing of particles (discussed below), including the effect on the flow distribution caused by the grate design. Further analysis of the gas combustion within the channels shows that this is limited by gas phase mixing, and it has proved possible to confirm the predicted height of the flames above the bed on full-scale plants. The measured and simulated temperature profiles along the bed length are shown above in Figure 2e). Data from two runs with same operating conditions are presented. During the first run only one thermocouple was used for measurement of the local bed temperature and during the second run, two thermocouples were used. These two thermocouples extended from the pair of sidewalls of instrument opposite each other and were around 200mm apart. The measurement shows a sharp temperature rise at a distance of 2.0m from the fuel entrance, indicating the start of fuel ignition that was then followed by a series of violent fluctuations of the measured temperature. The position of bed ignition was consistent with the visual observation from the viewing ports during the tests in which no flames were seen for the first 1.5 - 2 m of the bed length. Figure 2f) shows the measured local O2 concentration profile inside the bed as the electronic device tumbled along with burning wastes and also the simulated O2concentration profile. Measurement shows that oxygen began to fall at a position of 1.7 m from the fuel entrance. It then fluctuated between 0% and 14% for a significant portion of the bed length (2m – 5m) before settling at a more or less stable level (around 4%) after 5 m from the fuel entrance. Violent fluctuations in both the measured temperature and O2 level are due to three factors:

1) Constant changing of the probe positions as the electronic device tumbled along the bed so that the probe tips could be either, out of or inside, the reaction zone which is about 150 mm to 300 mm deep from the bed top (according to FLIC simulation);

2) Channel formation and destruction in the bed; and3) Diversity of the fuel properties causing uneven local combustion.

The Effect of under-grate air flow rate – combustion or gasification A one-dimensional “pot burner” has been used to provide data for use in FLIC and to confirm the results of FLIC calculations using various biomass materials. The apparatus is illustrated in Figure 3. The FLIC code has been used to calculate the effect of under-grate airflow rate in a wide range and the results are shown in Figure 4. All the calculations were based on fuel analysis (listed in Table 1) used in this stationary bench-top reactor (Yang et al. 2002).

1

13121110

2

9876543

1415

W eighing Scale

Com pressed air

A irPreheater

Rotam eter

Secondary air

Prim aryA ir

Burner

G rate

Data Acquisition

System

To Stack

Sam pling point

Secondary A ir Nozzle

W aste

Figure 3 Schematic D iagram of the Experimental Facility

Batch Incinerator Expt.

0

5

10

15

20

25

0 2 4 6 8Distance along bed length, m

O2

conc

entra

tion,

vol

% (d

ry)

Test two: O2

O2max

O2min

Figure 2f). In-bed measurement of O2 concentration

8

Air to fuel stoichiometric rat io

0.0

0.5

1.0

1.5

2.0

2.5

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7

Primary air mass f low, kg/m2.s

8.3% (M arie2000)30% (Gort1995 )10% thiswork 30% thiswork40% thiswork50% thiswork

10%20%

30%

40%50%

M oisture :

Gasif icat ion

combustion

Figure 5. Combustion stoichiometry vs. airflow rate at different moisture levels. Lines – calculations using FLIC; Symbols – experimental data from this work and references.

Table 1 Fuel Characteristics Moisture Ash Volatile Fixed carbon C H O LCV Size No.

wt% wt% wt% wt% wt% wt% wt% MJ/kg mm 1 10 4.3 74 11.7 39.8 4.7 41.2 13.4 12 2 20 3.9 65.7 10.4 35.3 4.1 36.7 11.6 12 3 30 3.5 57.4 9.1 30.7 3.6 32.2 9.85 12 4 40 3.2 49.1 7.7 26.3 3.0 27.5 9.0 12 5 50 2.7 41.0 6.3 21.8 2.5 23.0 8.1 12

Figure 4 shows the burning rate vs. under-grate airflow rate at different moisture levels, and comparison was made between the FLIC calculations and experimental data. The airflow rate spans a range from 0.03 to 0.6 kg/m2s without preheat (around 15°C) and the moisture level covers from 10% to 50% on a wet basis. At each moisture level, there was a characteristic airflow rate where the burning rate reached a maximum and this characteristic flow rate increases with decreasing moisture level in the fuel.

Figure 5 shows the relationship between combustion stoichiometry and under-grate air flowrate at different moisture levels in the bio-fuel. It is seen that the higher the airflow rate

and wetter the fuel, the richer the combustion will become. For each moisture level, there is a critical flow rate below which the combustion becomes a gasification process (overall air to fuel stoichiometric ratio <1) where a net production of combustible gases (CO, CH4,etc.) would result. This critical flow rate decreases with increase in the fuel moisture level, i.e., wet fuels tend to be combusted and dry fuels gasified if similar conditions are applied.

Burning rate,dry

0

50

100

150

200

250

300

350

0 0.2 0.4 0.6 0.8

Under-grate air mass flow, kg/m2.s

kg/m

2.hr

10%-10mm (WC,Gort1995)

30%-10mm (WC,Gort1995)

40.3% (FW, Thunman2001)

10% this work

30% this work

40% this work

50% this work

fuel moisture = 10%

20%

30%40%

50%

Moisture content:

Figure 4. Burning rate vs. airflow rate at different moisture levels. Lines – calculations using FLIC; Symbols – experimental data from this work and references. WC – wood chips; FW – forest waste.

9

CO emission from the bed top

0

5

10

15

20

25

30

35

40

0 0.1 0.2 0.3 0.4 0.5 0.6Primary air mass flow, kg/m2.s

CO

(vol

%, d

ry)

10%20%30%40%50%

Moisture content:

Figure 7. CO emission at the bed top vs. airflow rate at differentmoisture levels.

Peak temperature in the bed

900

1000

1100

1200

1300

1400

1500

1600

1700

0 0.1 0.2 0.3 0.4 0.5 0.6

Air mass flow, kg/m2.s

K

8.3% (Marie et al. 2000)10% 20%30% 40%50%

Moisture content:

Figure 6. Peak temperature vs. air flow rate at different moisturelevels. Line and symbol – calculations using FLIC; Symbols –Rönnbäck Marie et al. 2000).

Figure 6 shows the peak flame temperature vs. airflow rate at different moisture levels. For very wet fuels (50% moisture, for example), the flame temperature rises quickly with increasing airflow and peaks at a certain critical point. Further increasing the airflow rate would decrease the flame temperature and eventually lead to flame extinction. But for very dry fuels (10% moisture, for example), this critical flow rate is far higher and so is the maximum flame temperature obtainable. It is also seen that at lower airflow rate, wet fuels can produce a higher temperature flame than a dry fuel. This is because the combustion becomes much more fuel rich (gasification) for dry fuels at low air rates.

Figure 7 shows the calculated CO concentration in the flue gases exiting from the bed top. As the under-grate air flow decreases, CO increases in the flue gases as the process becomes air lean and shifts from combustion to gasification. Wet fuels produce less CO at the bed top as they favor fuel-lean combustion in the bed.

PYROLYSIS OF WASTE/BIOMASS AND GENERATION OF STORABLE FUEL Biomass pyrolysis is one of the thermal energy recovery processes, which has the potential to generate oil, char and gas products. It is achieved by heating the material in a closed container in the absence of an oxidizing agent. This decomposes the organic material to release gas and liquid products with solid char that can be used to generate heat or power. The process parameters which have the major influence on the products are the pyrolysis temperature, heating rate, particle size and retort atmosphere (Williams and Besler, 1996; Beis et al, 2002). The process conditions of pyrolysis can be optimized to maximize the production of either: pyrolytic oil, char, or gas, all of which have a potential use as fuels. The carbon char remaining is a valuable fuel that can be easily separated whilst it is still hot from any metal and stones that were originally present in the bio-waste. It is proposed that the char and oil can be produced locally in a small-scale simple unit that uses the gas produced to heat the pyrolyser. The char and oil are than compact fuels that can be transported economically to a central power station where the advantages of scale can be employed for the efficient and economic generation of electricity, possibly using gasification as discussed later.The pyrolysis oils are already being studied elsewhere and the main objective of this study is to address the relevant aspects of production, characterisation and use of a storable char fuel derived from the pyrolysis of biomass materials. This will thus contribute to national

10

sustainability targets. A series of pyrolysis tests were conducted initially using wood cubes in the temperature range of 350°C - 700°C with a heating rate of 10oC/min. The pyrolytic products from the tests were then analysed. Experimental Method The reactor system built was a batch type packed bed pyrolysis unit shown in Figure 8. It consisted of a reactor in a temperature-controlled furnace followed by two liquid condenser/traps. The stainless-steel reactor, 12.5cm diameter × 50 cm high, is placed inside a furnace whose inner temperature (To) was controlled by a temperature controller. Nitrogen gas was supplied from below the reactor to purge the volatile gases released from the sample during pyrolysis. The volatile gases and nitrogen leaving the reactor passed through two water-cooled condensers to separate oil vapour from the gas stream. The oil was collected in a disposable plastic container at the bottom of each trap. The concentrations of CO, CO2 and O2 in the off-gas after the condensers were monitored by gas analysers and recorded by the data logger. The flow rate of the sampling gas was l.0 l/min. Gas samples were also taken into glass bottles for further analysis of their chemical composition using an off-line gas chromatograph.

NITROGEN

BED SECTIONOF SAMPLES

WATER CONDENSER

TEMPERATURECONTROLLER

OIL CONTAINERS

OIL

THERMOCOUPLEST4

T5

EXTRACT

SAMPLING BOTTLE

GAS ANALYSER(CO, CO

2 , O2 )

EXTRACT

SAMPLING BOTTLE

GAS ANALYSER(CO, CO

2 , O2 )

FURNACE

DATA LOGGER

To

T1

T2

T3

REACTOR

Figure 8. The pyrolyser set-up

Sample Pinewood cubes of size 2cm

Proximate analysis

Moisture 8.86%, Volatile matter 78.86%, Fixed carbon 12.08%, Ash 0.20%

Ultimate analysis

C 47.9%, H 6.2%, O 38.3%

Lower heating value 17.8 MJ/kg (dry)

Table 2. Properties of the wood sample

11

At the beginning of a pyrolysis experiment, the reactor was charged with feed material, typically 100 - 400g in weight, and was placed inside the furnace. Then, the furnace was heated up to the final temperature at the given heating rate with a fixed flow rate of nitrogen. Once the furnace attained the set value of To, it was maintained for 2 hours to allow sufficient time to complete pyrolysis. Characterisation of Wood. The waste sample at the reported initial stage of the experiments consisted of cubes of pinewood of size 2cm. Table 2 shows the properties of this material acquired from the proximate, ultimate and calorific value analyses.

Temperature (oC)

0 100 200 300 400 500 600 700

Wei

ght (

%)

0

20

40

60

80

100

-dX/dTWOOD

Figure 9. TGA curve for pine sawdust under nitrogen (heating rate: 10oC/min)

Figure 9 shows the thermo-gravimetric analysis (TGA) for pine sawdust. The pyrolysis commenced around 250°C, and the rate of pyrolysis increased slowly and reached a peak at 386oC. By 400°C, the sample had evolved 67% of its original mass by pyrolysis. The rate of pyrolysis slowed down significantly above 410°C.Experimental Results. Figure 10 shows the solid and liquid products from pyrolysis from a test with a final temperature of 500oC and the heating rate was 10oC/min. The size of the char cubes was around 1.5cm, which was 42% of the original wood cube in volume. The mass yield of the liquid was 45.8% in this case and consisted of a mixture of black greasy oil and translucent solution. The other pyrolytic product was the off gas composed of CO, CO2, H2,CH4 and other hydrocarbons. When the temperature reached 500oC, the gas concentrations were CO 40.4%, CO2 38.9%, CH4 15.8% and H2 4.9%.

Inside the reactor after pyrolysis

Char24.5%wt

Liquidcollected in the traps45.8%wt

Wood cubessize: 2cm

GasCO, CO2, H2, H2O, CH4,Hydrocarbons29.7%wt

Figure 10. The solid and liquid products from pyrolysis

The mass yield, higher heating value (HHV) and corresponding energy yield of char from various final temperatures are plotted in Figure 11. The energy yield was calculated simply

12

by multiplying the calorific value with the mass yield. The mass yield of char was 33% at 350oC, and decreased with increasing the final temperature. The decrease in the mass yield over 500oC slowed down. However, the HHV of char increased with increasing temperature, from 30.9MJ/kg at 350oC to 33.3MJ/kg at 700oC. Since the HHV of char was much higher than that of the original wood, the energy yield ranged from 49% - 33%.

Final Temperature (oC)

300 400 500 600 700 800

Yie

ld, %

0

10

20

30

40

50

60

Calorific Value (M

J/kg)

10

15

20

25

30

35

40

Mass Yield

Calorific ValueEnergy Yield

Figure 11. Mass yield, energy yields and calorific value of char from wood (heating rate=10oC/min)

O/C ratio

0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7

H/C

ratio

0.0

0.5

1.0

1.5

2.0

wood

350oC400oC

500oC

600oC700oC

Figure 12. Van Krevelen diagram: H/C and O/C ratios of char (heating rate=10oC/min)

Figure 12 shows a plot of H/C versus O/C ratios, also known as the Van Krevelen diagram. The char became more carbonaceous at high temperatures. The carbon content in the char at 350oC was 77%, which was 53% of the original carbon in the wood sample. It became 89% at 700oC, which was 40% of the original carbon in the wood sample. At the final temperature of 500oC, the carbon content of char was 83.1% (43% of original carbon), while the hydrogen content was 3.8% (15% of original hydrogen). The decrease in H/C and O/C ratios was close to a linear relationship.

Final Temperature ( oC)

350 400 500 600 700 B A

Wei

ght (

%)

0

20

40

60

80

100

Higher H

eating Value (M

J/kg)

16

20

24

28

32

36

FCV

M

HHV

Final Temperature ( oC)

350 400 500 600 700 B A

Wei

ght (

%)

0

20

40

60

80

100

Higher H

eating Value (M

J/kg)

16

20

24

28

32

36

FCV

M

HHV

Figure 13. Proximate analysis of char (heating rate=10oC/min) (VM: Volatile Matter, FC: Fixed Carbon, B: Bituminous coal, A: Anthracite, Source of the

coal results: Phyllis Database)

Figure 13 shows the volatile matter (VM) and fixed carbon (FC) contents measured from the proximate analysis of char along with its calorific value. The proportion of FC increased with temperature rise, which is the inherent nature of pyrolysis. The chars from 400°C and 600°Cshowed similar FC/VM ratios to bituminous coal and anthracite, respectively. Thus the calorific values of chars were as high as that of bituminous coal and anthracite. The main conclusions drawn from this phase of the work are:

i) The mass yield of char from wood cubes decreased from 33.0% to 21.5% with increasing final temperature from 350oC to 700oC at the heating rate of 10oC/min.

ii) The char became more carbonaceous at higher temperatures, and the decrease of H and O in char showed a linear relationship between H/C and O/C ratios.

13

Ultra Superheated SteamUltra Superheated Steamfor Sustainable Processesfor Sustainable Processes

Steam at 1600°CSteam at 1600°C

ThermocoupleThermocouple

iii) The calorific value of char was over 30 MJ/kg, which was as high as coal. As discussed above, in addition to the well-reported value of the pyrolysis liquids as a fuel oil, the use of the storable char as feedstock for a gasifier required investigation. Its behaviour was therefore studied in our unique gasification system.

ULTRA SUPER-HEATED STEAM GASIFICATION

Introduction. The conversion efficiency of solid fuel via steam raised in a boiler into electricity using steam turbines is only about 20% to 45% depending on the temperature and pressure of the steam. Gasification of solid fuels for IGCC systems produces gas to run gas turbines that are used to produce power together with steam turbines. This enables enhanced efficiency in the conversion of energy from solid fuels and the overall efficiency of a conventional integrated gasification combined cycle is about 51%. In this case, the heat required for the gasification reaction is generated by the reaction of about 15% of the fuel with oxygen. A new process is proposed that uses low-grade steam from an incinerator as the starting point for an ultra superheated steam gasification system for solid fuels. The overall efficiency of the proposed process is about 60%, which is superior to the sum of the separate processes of electricity from a steam boiler together with an independent biomass

gasification-based electricity generator. When used as an embedded CHP system, the energy complex can produce power and heat from biomass (and coal) for a sustainable city at an efficiency of about 85%. If the plant is installed in a city, the biomass can be derived from waste, thus disposing of waste that cannot be effectively reused or recycled. The aim of this aspect of the research project is the development of the novel gasification process that utilises ultra-superheated steam (Figure 14).

Figure 14

Experimental investigation. The new gasification concept offers the prospect of achieving the goal of high efficiency power generation from various biomass (or coal) energy sources by utilizing Ultra Superheated Steam (USS). The concept first reduces the biomass including wastes such as wood waste (and coal) to a fine powder that can be reacted in a small residence time and therefore a small volume. The process uses low-grade steam from sources such as waste incineration, process cooling, and local industry, and then enhances it to a temperature above 1600ºC. This is achieved by adding oxygen to the steam to form "artificial air"; gas is then burned in this artificial air to produce an ultra-superheated steam flame. The biomass (or coal) is injected coaxially into the high temperature steam flame where it reacts in about one second to produce a gas that is free of tar and consists largely of H2, CO and CH4.Significantly, the high temperature steam provides the enthalpy for the endothermic gasification reactions. This feature ensures that any ash is cooled below its melting point before it arrives at the reactor walls. This is important since the high alkali content of biomass ash results in a low melting point and hence a high tendency to form slag.

The USS gasifier can be envisaged as being composed of two distinct main parts, namely: 1. The USS generator (a conventional burner where the gasification also commences). 2. The gasifier (which is the shell where the reactions proceed).

The USS Burner. The North American Manufacturing Company manufacture the burner selected for the generation of USS and subsequent USS gasification (Figure 15). Its design

14

capacities are a maximum airflow of 0.0322 m3/s and a heat output of 120 kW. Its construction features include air and gas inlets into the burner and a standard quarl that is about 230 mm long. This burner was chosen because, being manufactured for dual fuel operation it could be easily adapted for the USS gasification by replacing the liquid fuel pipe with one suitable for the supply of granular material for the gasification of solids, or by just using the burner as supplied for the gasification of liquids such those from pyrolysis, or slurry material made from these oils combined with char. This burner was adapted to feed in particulate materials by removing the liquid fuel inlet pipe, and replacing it with a pipe with a wider bore, connected to the funnel and vibrating feeder system as shown in Figure 15. The USS Gasifier. The rational design of a gasifier requires the best possible understanding of the fundamental chemical and physical processes that occur: Pressure is one of the operating variables of interest in the investigation of USS gasification. In the gasifier design stage, high pressure USS gasification was considered but the extent of the initial pressure was limited by the rapid construction and operation feasibilities. This then limited the pressure operation of the gasifier to near atmospheric for the preliminary tests, where the material to be gasified is fed into the gasifier by gravity. The gasification chamber is a cylindrical mild steel shell lined with a 50mm thick fused alumina based castable refractory with high abrasion resistance. The refractory inside diameter = 285 mm. A catch pot at the bottom collects ash or slag. The USS gasification was carried out using methane as the gaseous fuel for the generation of USS and propane was used to fire the pilot burner. The yield of gas was measured by collecting samples for GC analysis and also evaluating the ash content. A photograph of the equipment is shown in Figure 15. Figure 15 The theoretical calculations were carried out to estimate the gas yield from USS gasification. The Boudouard, water-gas and hydrogasification reactions occur simultaneously. The predicted yield at equilibrium was used for carbon gasification using steam at 1 atm and 2131K (which is the adiabatic flame temperature for a stoichiometric USS flame). The results are:Volume fraction of CO = 54.8% Volume fraction of CO2 = 0.0% Volume fraction of H2 = 45.2% Volume fraction of H2O = 0.001% Volume fraction of CH4 = 0.007% Chemical equilibrium calculations based on minimising Gibbs free energy are shown in the figure 16. These show higher levels of CO2 as observed in the tests.

15

Figure 16

THE INTEGRATION OF FLIC WITH FLUENT In order to design industrial plant based on the various processes discussed above, integrated process modelling techniques are required. These are include advanced procedures such as the recognition and modelling of the channels that form in packed beds of particles such as in

moving grate combustion systems. Such an assortment of particles is illustrated in Figure 17. The flow distribution through such a bed takes place along channels and the results of calculations (that have been verified by measurements) is illustrated in Figures 18 a) and b). These show the calculated flow in a vertical section and that measured across the bed surface respectively. The formation of such burning channels can also be seen clearly by looking at the flames on the surface of a burning bed of coal or other particles. This phenomenon is also included in the FLIC model.

Figure 18 a) Figure 18 b) FLIC thus provides a boundary condition for the FLUENT CFD calculation of the flow in the freeboard above the burning bed. However, FLIC requires information on the radiation heat transfer from the freeboard. Thus the calculation procedure must be performed by the

Measurement of channelling in a Measurement of channelling in a pot bed without combustionpot bed without combustion

V/V0

Simulation of channel formation Simulation of channel formation in a nonin a non--burning bedburning bed

Grate causes initial uniform flow

Figure 17

16

simultaneous (iterative) solution of both FLIC and FLUENT. This procedure for the biomass combustion in a large incinerator is illustrated in Figures 19 a). Figure 19 b) shows the temperature distribution through the plant. Figure 19c) shows that a flat heat transfer distribution used to initiate the iterative procedure leads to convergence after about four cycles. Figure 20 shows the converged results of the combined reacting flow calculation.

F L U E N T /F L I C m o d e l i n t e g r a t i o nF L U E N T /F L I C m o d e l i n t e g r a t i o n

F L U E N T(G a s F lo w M o d e l)

F L IC(M o d e l fo r W a s te B e d )

Q R A D

M G A S , T G A S , V G A S

P R IM A R Y A IR

W A S T E

A S H

G R A T E

S E C O N D A R YA IR

W A T E R W A L L

B O IL E RT U B E S

Figure 19a)

S im u la t io n R e s u lt sS im u la t io n R e s u lt sG A S F L O W F IE L D

F R O M F L U E N T

W A S T E C O M B U S T IO NF R O M F L IC

T e m p e ra tu re

Gas

Pro

perti

es

D is ta n c e a lo n g b e d

V e lo c it y

Rad

iatio

n

D is ta n c e a lo n g b e d

G A S R E L E A S E D A T AF R O M F L IC

IN C ID E N T R A D IA T IO NF R O M F L U E N T

Figure 19b)

C alculation R esultsC alculation R esultsFL U E N T to FL IC : In cid en t rad iation on th e w aste b ed

– E m itted gas affects th e rad iation p rofile sh ap e » (a) B y gaseou s em ission , (b ) V ia th e fu rn ace w all,

(c) B y p articu late em ission . » T h e rad iation p rofile b ecam e stab le after th e 4 th

iterative calcu lation s of FL IC an d FL U E N T

0 2 4 6 8 10

400

600

800

1000

1200

1400

1600

Gas

Tem

pera

ture

(K)

D is tance a long bed length(m )

0 .51 .52 .53 .5 th

0 2 4 6 8 10400

600

800

1000

1200

1400

1600

Effe

ctiv

e R

adia

tion

Tem

pera

ture

(K)

D is ta nce a long bed length(m )

0 (assum ed)1234 th upda te

Figure 19c)

17

FLUENT/FLIC Results: Gas ConcentrationsFLUENT/FLIC Results: Gas Concentrations– Note the active gaseous reactions by fresh

oxygen from the secondary air

CO2 MASS FRACTIONO2 MASS FRACTION

CO MASS FRACTION O2 MASS FRACTION TEMPERATURE

Figure 20

CONCLUSIONThe combustion/pyrolysis/gasification of biomass on both static and moving beds has been investigated. The fundamental processes have been studied and an understanding has been gained of the complexities involved. New mathematical models of these phenomena now give a basis for integrated modeling that can provide design data for industrial exploitation of energy from biomass. Acknowledgments:The authors would like to thank the following organizations for their financial and technical support for the above research programmes: UK Engineering and Physical Science Research Council (EPSRC), Onyx Environmental trust and the UK Incineration Industry. NomenclatureAv pre-exponent factor in devolatilization rate, s-1

C constant; molar fractions of species (fuel, oxygen) Cfuel fuel concentration, kg/m3

Cpg specific heat capacity of the gas mixture, J/(kg K) Cmix mixing-rate constant

Dg molecular diffusion coefficient of volatile hydrocarbons in air, m2/s

Dig dispersion coefficients of the species Yi, m2/s

dp particle diameter, m Dr in-flow dispersion coefficient in bed, m2/sDs particle mixing coefficient due to random movements of particles in the bed, m2/sEv activation energy in devolatilization rate, J/kmol

Hg gas enthalpy, J/kg

Hs solid-phase enthalpy, J/kg hs’ convective heat transfer coefficient between solid and gas, W/m2K

kv rate constant of devolatilization, s-1

18

Qcr heat absorbed by the solids, W/m3

Qh heat loss/gain of the gases, W/m3

Qsh thermal source term for solid phase, W/m3

qr radiative heat flux, W/m2

Rmix mixing-rate of gaseous phase in the bed, kg/m3s

S stoichiometric coefficients in reactions Sa particle surface area, m2

Ssg conversion rate from solid to gases due to evaporation, devolatilisation and char burning, kg/m3s

Syig mass sources due to evaporation, devolatilization and combustion, kg/m3s

Syis source term, kg/m3s

t time instant, s

Tg gas temperature, K

Ts solid temperature, K Vg superficial gas velocity (vector), m/s

Yis mass fractions of particle compositions (moisture, volatile, fixed carbon and ash)

εs system emissivity

σb Boltzmann radiation constant, 5.86 × 10-8 W/m2K4

φ void fraction in the bed ρg gas density, kg/m3

ρsb solid bulk density in the bed, kg/m3

λg thermal dispersion coefficient, W/mK

λs effective thermal conductivity of the solid bed, W/mK References:Beis S.H.; Onay O and Kockar O.M., Fixed-bed pyrolysis of safflower seed: influence of pyrolysis parameters on product yields and compositions, Renewable Energy, 2002, vol. 26, no. 1, pp. 21-32 Gort, R, On the Propagation of a Reaction Front in a Packed Bed: Thermal Conversion of Municipal Waste and Biomass, Academic Dissertation, University of Twente, 1995. Peters B, A detailed Model for Devolatilization and Combustion of Waste Material in Packed Beds, 3rd

European Conference on Industrial Furnaces and Boilers (INFUB), Lisbon, Portugal, 18 – 21 April, 1995.Pyle, D. L. and Zaror, C. A., Heat transfer and kinetics in the low temperature pyrolysis of solids,Chem Eng Sci, Vol.39, No.1, pp.147-158, 1984. Shin, D and Choir, S, The combustion of simulated Waste Particles in a Bed, Combustion and Flame, Vol.121, pp.167-180, 2000. Smoot, L D and Pratt, D T, Pulverized-coal Combustion and Gasification, Plenum Press, 1979. Wakao, N and Kaguei, S, Heat and Mass Transfer in Packed Beds, Gorden & Breach Science Publishers, 1982. Williams P.T. and Besler S., The influence of temperature and heating rate on the slow pyrolysis of biomass, Renewable Energy, Vol.7(3), pp.233-250, 1996 Yang Y B, Goodfellow J, Ward D, Gan S, Swithenbank J and Nasserzadeh V, Cutting Wastes From Municipal Solid Waste Incinerator Plants, Trans IChemE, Vol 81, Part B, pp.143-155, May 2003. Yang Y B, Nasserzadeh V, Goodfellow J, and Swithenbank J, Simulation of Channel Growth in a Burning Bed of Solids, Trans IChemE, Vol 81, Part A, pp.221-232, 2003. Yang Y B, Nasserzadeh V, Goodfellow J, Goh Y R, and Swithenbank J, Parameter Study on the Incineration of Municipal Solid Waste in Packed Beds, Journal of Institute of Energy, September, 2002.Yang Y B, Goh Y.R, Zakaria R, Nasserzadeh V, Swithenbank J, Mathematical modelling of MSW incineration on a travelling bed, Waste Management 22 (2002) 369-380. Yang Y B, Goodfellow J, Goh Y, Nasserzadeh V and Swithenbank J, Investigation of Channel Formation Due To Random Packing In a Burning Waste Bed, Trans IChemE, Vol 79, Part B, 2001.