Beruflich Dokumente
Kultur Dokumente
1, 2007 25
Markus A. Reuter
Faculty of Engineering, Department of Chemical and Biomolecular Engineering,
University of Melbourne, Melbourne, Australia
E-mail: mreuter@unimelb.edu.au
Joep Verwoerd
AVR, P.O. Box 59144, 3008 PC Rotterdam, The Netherlands
E-mail: Joep.Verwoerd@avr.nl
Reference to this paper should be made as follows: Yang, Y., Pijnenborg, M.J.A., Reuter, M.A.
and Verwoerd, J. (2007) Analysis of transport phenomena in a rotary-kiln hazardous waste
incinerator, Progress in Computational Fluid Dynamics, Vol. 7, No. 1, pp.2539.
Biographical notes: Yongxiang Yang received his BSc and MSc Degrees at Northeastern
University in Extractive Metallurgy (China, 1982 and 1988), and the degrees of Licentiate and
Dr. of Tech. at Helsinki University of Technology in Materials Processing (Finland, 1992 and
1996). He has worked as design engineer, researcher and lecturer in metallurgical engineering.
Since 1998 he has been working at Delft University Technology (The Netherlands) as an
Assistant Professor in Process Metallurgy. His main interests include metallurgical and materials
processing, transport phenomena, and applications of CFD in process simulation. He is a
co-author of the book Metrics of Material and Metal Ecology.
Marc J.A. Pijnenborg got his MSc Degree in Resource Engineering in 2003, Delft University of
Technology (TU Delft). He has made his MSc Thesis in modelling combustion and flow
behaviour in hazardous waste incineration process. He is now working outside TU Delft as
Process Engineer.
Markus A. Reuter: Chief Executive Technologist Ausmelt Ltd. Australia. Degrees: B.Eng.
(Chemical Engineering 1981), M.Eng. (1985), PhD (1991), D.Eng. (2006) all University of
Stellenbosch (South Africa) and Dr.Habil. (1995) Aachen University of Technology (Germany).
Industry: Manager in the Measurement and Control Division, Mintek, Metallurgist at Anglo
American Corporation (both South Africa) and registered Professional Engineer (South Africa).
Academic: 19962005 Professor at the Delft University of Technology (Netherlands) and since
July 2005 Professor at University Melbourne (Australia). He has ca. 300 publications (journal
and conference proceedings) and has written a book Metrics of Material and Metal Ecology
Elsevier (Amsterdam).
Joep Verwoerd works as a Process Engineer at AVR Rozenburg, the largest MSW and hazardous
waste incineration company in the Netherlands. Before January 2005 he has worked for 32 years
in the plant of hazardous waste incineration at AVR-Chemie, part of the AVR Rozenburg.
After the kilns have stopped operation he is in charge of waste treatment related research. He has
written a book on hazardous waste processing, together with other authors, called gevaarlijk afval
(hazardous waste).
both thermal and chemical aspects of the waste combustion Figure 2 Illustration of waste combustion system: the rotary kiln
process are emphasised. and the Secondary Combustion Chamber (SCC)
waste. Therefore, the transport phenomena within the whole PHOENICS 3.5 (Cham, 2004). The governing partial
combustion system play a very critical role in the waste differential equations for conservation of mass, momentum,
incineration process. To catch the information about and energy in a turbulent flow system are expressed by
distribution of turbulent mixing, temperature and Residence time-averaged 2nd order partial differential equations.
Time Distribution (RTD) within the whole combustion The general form is expressed in Equation (1), and it is
space, CFD simulation can provide a good solution. solved with a common numerical algorithm in the code. Due
Hazardous waste in various forms are fed into the rotary to the turbulent nature of the flow, different turbulence
kiln through a number of burners and injection lances/pipes, models such as the standard k- and its modifications were
or even the loading door so called load chute, as is described tested, and the standard k- model was finally chosen in the
previously. Combustion air is supplied partially together study. Due to the assumed adiabatic wall boundary
with the waste as carrier gas and primary air, and partially conditions to be explained later, the simulation with
supplied through load chute at the kiln front as well as radiation model did not bring any difference for heat
separate injection lances in SCC. To promote earlier transfer with PHOENICS and has been thus neglected in the
combustion, a supporting burner (also called main burner at combustion heat transfer model. This is justified by the fact
AVR) is installed to burn high calorific waste or fuels to that a relatively low heat is lost in the furnace operation.
ensure a high temperature zone near the feeding end of the
kiln. The cold waste in liquid or solid form is firstly heated
( ) + div( u ) = div( ,eff grad ) + S (1)
up by the radiation flame and walls, it gets dried and solid t
vapourisation/gasification takes place before combustion where
reactions. The main combustion will take place in the
gas phase between the fuel vapour and combustion air. : a general flow variable (velocity component,
The mixing between the fuels vapour and the air is not temperature, turbulence kinetic energy or its
perfect, which will form partially diffusion and partially dissipation rate, radiosity or radiation temperature,
pre-mixed flames. Due to the insufficient length of etc.)
the kiln and thus the short residence time for the mixture of u: the velocity vector
the fuel vapour and air, the combustion will continue in the : density of the fluid phase
SCC. To promote complete combustion reactions, additional ,eff: the effective exchange coefficient, which is the sum
air lances are installed. After certain travelling time, the of laminar and turbulent exchange coefficients
remaining fuel components including CO gas generated S: general source term for flow variable .
as intermediate combustion product at relatively high
For the flow simulation, the continuity equation and the
temperatures (10001100oC) are expected to react with
equation of momentum conservation (the Navier-Stokes
oxygen before they get into the WHB. It is prescribed by the
equations) are solved together with a turbulence model.
regulation that a minimum of 2 seconds is required for the
For heat transfer simulation, the energy conservation
gas mixture at minimum of 850C for non-halogen waste or
equation ( becomes enthalpy) is solved, where the
1100C for halogen-containing waste after the last air
source term S includes the gas phase combustion energy
injection point in the furnace (Directive 2000/76/EC, 2000).
estimated from a combustion model (global 3-gas or 7-gas
It is very important that the remaining CO (very critical for
combustion model in this study). The combustion model
the environmental quality of the off-gas) is fully combusted,
will be discussed in more detail.
since it cannot be combusted any further downstream in the
Apart from waste combustion modelling in the gas
APC devices downstream. In practice, minor amount of CO
phase, separate simulation on the flow of shredded solid
in the off-gas is often the main cause for the operation
waste was conducted by using the Inter-Phase Slip
failure.
Algorithm IPSA, the Eulerian Eulerian two phase
model of PHOENICS (Cham, 2004). In the current study,
only the isothermal particle flows were simulated. In order
4 CFD simulation to include the effect of gas temperature variations
The details of the coupled transport phenomena in the (gas cooling, and non-isothermal gas mixing) on the gas
hazardous waste incinerator are very difficult to obtain flow field, the convective heat transfer within the gas phase
through online measurement and high temperature (non-isothermal gas flow) was included in the simulations.
experiments. CFD simulation offers a lot of benefits for Combustion models were not included in the gas phase flow
investigating coupled transport phenomena of fluid flow, and heat transfer simulation, while the combustion effect
heat transfer and chemical reactions in the incineration was represented with hot gas jets where the incoming
system. However, care needs to be taken of the validity of temperatures were estimated from the 7-gas combustion
physical models and validation of the results, particularly model. The effect of gravity on the particle momentum
due to the empiricism introduced by the models of conservation is included. The gas phase turbulence was
turbulence, multi-phase flow and fuel combustion. simulated with the standard k- model, however, the effect
In the current study, the simulation of gas flow and of the particles presence on the gas phase turbulence was
temperature distribution as well as waste combustion neglected. In the current simulation, the inter-phase mass
in the incinerator has been conducted with CFD code and heat transfer were not taken into account.
Analysis of transport phenomena in a rotary-kiln hazardous waste incinerator 29
Table 1 summarises the assumptions, the physical grid of 359,100 cells was used to study effect of extra
models, boundary conditions, and other model parameters, cooling air from the ventilation system to the rotary
for the simulations presented in this paper. kiln cooling ring.
dissipation rate. In the current study it is assumed that the lies in the areas near the burner zones where gas velocity is
reaction rate for the combustion of the different waste types very high.
is turbulent mixing rate limited and the EBU is chosen to Both steady and transient situations of solid waste flow
determine the reaction rate. On the basis of extensive were simulated. For transient studies, the time interval of
parametric testing a value at 2550 for CEBU was found to 0.11 second was used with 28 time steps to catch the
be reasonable. The stoichiometric ratio s, was taken with smooth change of the particle flow, especially in the first
value of 6.52, based on the information in the literature second of the transient flow. In total 15 seconds was
(Wardenier and van den Bulck, 1997) and estimation from simulated in most of the transient models, which could
the average heating value of the waste at AVR. approach roughly (if not completely) the steady flow of the
solid. In these studies, the steady gas phase flow was firstly
4.4 7-gas combustion model solved for the single gas-phase flow. Afterwards, the
particles were added into the system from the solid waste
More recently, an improved combustion model was studied burner. The variation of the particle volume fraction was
in order to understand and predict some chemical aspects of followed to see the development of the solid flowing
the waste combustion process. Not only the overall thermal process and the statistical particle trajectories, especially the
behaviour could be estimated, but the species distribution dispersion extent of the solid waste in the system.
could be predicted as well, such as fuel (hydrocarbons),
oxidant (O2), intermediate and final combustion products
(CO, H2, CO2, H2O), and the inert component of nitrogen 5 The Virtual Fuel and waste combustion
(N2). An extended SCRS model from Phoenics is explored models
for this purpose. In order to model the species distribution
together with combustion, the first step is to convert the 5.1 Virtual fuel determination
different waste streams into a sort of virtual fuel or fuel Based on the off-gas composition and mass and energy
mixture. Then the virtual fuel is used for combustion balance of each waste streams at AVR-Chemie, a virtual
reaction modelling with both equivalent thermal and fuel was determined. The organic compound was estimated
chemical contribution as the complex wastes. A similar as C102H131O110, however, a hydrocarbon has to be found
concept has been proposed and used for Municipal Solid due to the limitation of the CFD code that it cannot handle
Waste (MSW) incineration (Themelis and Kim, 2003), and oxygen at present in the organic compounds. To get the
this proved to be a very successful approach in handling generalised fuel, two constraints were used:
the complicated waste materials. The details of the 7-gas
heat of combustion = 42,000 kJ/kg for the dry mass
combustion model are described later in this paper.
H/C molar ratio is in a range of 1.31.7.
4.5 Solid flow model for the shredded solid waste For all closely related real existing hydrocarbons
In the present study, the standard drag coefficient for calculations were carried out to determine the generalised
spherical solid particles moving in a gas phase suggested by virtual fuel composition. Figure 4 illustrate the concept,
Clift et al. (1978) was chosen. The drag coefficient Cd is where one can see how a poorly defined waste is converted
determined by the particle Reynolds number Re: to the better defined fuel. To get the balance of oxygen in
the waste, oxygen is introduced to the combustion streams
24.0 0.42 through other oxygen containing compounds such as CO,
Cd = (1.0 + 0.15 Re0.687 ) + . (4)
Re 1.0 + 4.25 104 Re 1.16 O2, or H2O. Combinations with C2H3 or C3H4 with other
components and for all incoming streams are put in the
Equation (4) is valid for the Reynolds number of solver of Microsoft Excel, both C2H3 and C3H4, as
Re 3.38 105. The particle Reynolds number is based on the main virtual fuel component, are found to fit to the
the slip velocity Vslip (the velocity difference between the constraints. In this paper further calculations are performed
particle and the carrier gas), the particle diameter Dp and the with C3H4. C2H3 and C3H4 have very similar heat of
kinematic viscosity of the gas phase g: combustion to methane (CH4), and are in gaseous form at
room temperature. Here it may be necessary to emphasise
Vslip D p that the hydrocarbon C2H3 or C3H4 is only the main
Re = . (5)
g component of the fuel streams, and the used fuel streams in
the model are defined as a mixture of the main component.
In the particle gas incineration system, the maximum Other allowable fuel component (by the CFD code) of CO
particle Reynolds number is estimated to be around and non-combustibles such as CO2, H2O, O2 and N2, as will
105, within the boundary value in the criterion be explained in the following section (waste stream
(Re 3.38 105). The highest particle Reynolds number optimisation).
Analysis of transport phenomena in a rotary-kiln hazardous waste incinerator 31
Figure 4 Concept of artificial fuel definition for the hazardous Step 2: Secondary reactions
waste incineration plant
2CO + O 2 = 2CO 2 (7)
2H 2 + O 2 = 2H 2 O. (8)
Table 3 Stream distribution, the mass and thermal input for all inlet streams
Compositions (wt. %)
Input data Streams Load chute Main burner Sludge burner Solid burner1 SCC burner Air lances Ash sump Total
Stream distribution (wt. %) Fuel rich 0.00 34.45 40.00 37.77 13.10 0.0 0.0
Fuel lean 100.0 65.55 60.00 62.23 86.90 100.0 100.0
2
Mass flow rate (kg/h) 27,000 11,250 3,900 8,193 6,620 9,030 4,000 69,993
Total energy input (MW) Practice 0.00 11.67 4.33 11.67 2.84 0.00 0.00 30.51
Model 1.00 11.39 4.56 10.67 2.70 0.33 0.15 30.80
1
In definition of solid waste burner, 15.04% was assumed as inert materials reported to slag. The total mass flow rate and the
stream distribution are based on the mass flow rate to the gases phase only.
2
The mass flow rate reported here includes the wastes, oxidation air and water content. The waste processing rate is estimated at
7,200 kg/h, based on the operational data in practice.
32 Y. Yang, M.J.A. Pijnenborg, M.A. Reuter and J. Verwoerd
Table 3 shows the mass and energy input for all the process. The residence time of the gas particles is crucial for
incoming waste streams as well as additional air supply. completing the combustion. From the CFD combustion
For heat of combustion and energy input from each models, the flow pattern and gas mixing are predicted, and
incoming stream, the difference of the model data from the Figure 5 illustrates the predicted velocity distribution
real operation conditions in practice is illustrated, and the across a few regions of the incinerator, and a clear
deviation is relatively small. view of the complexity of the flow pattern and mixing
behaviour could be seen. The 3-dimensional nature and
complex mixing pattern are clearly demonstrated.
6 Simulation results Although the flow pattern has not been validated
by flow measurements due to extremely hostile
6.1 Main flow characteristics environment within the incinerator, the prediction could
It is known that the flow and mixing of the combustion air well indicate the general characteristics of the gas flow and
and fuel vapour determine essentially the waste combustion mixing pattern.
Figure 5 Gas flow pattern inside the rotary kiln and SCC during combustion process
The RTD of different gas streams and the in-flight time The RTD curve has been obtained by adding inert
distribution of each incoming streams have been tracers from the individual inlets, which follow the same
studied (Yang et al., 2002). Together with temperature flow path as the flowing gases. Through transient simulation
distribution maps, this could well indicate combustion by measuring the outgoing tracer amount along time from
status. With PHOENICS, it is possible to calculate the outlet, the RTD curves were constructed with which the
the time elapsed for a fluid since its entry into the flow characterisation was performed for the incinerator
flow domain. This time variable must only be (mean residence time, dead volume etc.). A typical RTD
convected by the fluid. The accumulation of time is curve obtained for the incinerator is illustrated in Figure 7
accounted for by special sources, which are related to for the main burner combustion stream (E1()) and the load
the fluid residence time in a cell. The calculation of the chute air stream (E2()). Based on this analysis, a dead
in-flight time could be easily performed in a steady volume of 16% and 20% were identified for the above two
state model. Figure 6 illustrates the predicted time streams. The incinerator is essentially a back-mixed flow
distribution of the gas streams since its entry point reactor with a back-mixing portion of over 75% for the
into the incineration system. studied streams.
Analysis of transport phenomena in a rotary-kiln hazardous waste incinerator 33
Figure 6 Gas in-flight time distribution since its entry into the the solid bed for part of the solid waste could be clearly
incinerator system (based on the 3-gas combustion seen. However, in the 3-gas model, all the solid waste was
model)
fed through the load chute, and the combustion was
assumed to take place completely on a solid bed at the
bottom of the rotary kiln.
It is important to notice the distribution of the main
combustion air, coming from the load-chute. Due to the
physical arrangement, the air enters the kiln front from the
load-chute at an angle of 50 downward. This causes a poor
mixing of the air with waste vapours. From the temperature
distribution maps, one can see the colder zone at the lower
part of the kiln from both Figures 8 and 9. In addition, rather
high temperature gradients of the combustion system can be
observed. The hotter zones in the kiln lie in the upper-half,
and very long flames were formed with a maximum
temperature above 2000C. Combustion from the solid-bed
waste takes place somewhat downstream away from the
bed, due to the cooling effect of the load-chute air.
However, the maximum temperature from the flames is
Figure 7 RTD curve, obtained from the transient simulation highly dependent on the rate constant (CEBU) of the EBU
model (numerical tracer experiment) model. In the model shown here, the CEBU constant from
both the primary and secondary reactions was set at 2.0.
The constant at various levels have been tested, the
fine-tuning of the constant needs to be conducted carefully
in the next stage. The cooling effect by the cold air from
injection lances seems to be significant, and this effect could
extend up to the outlet region.
Figure 9 Temperature distribution (C) across various burners, the load-chute and near the outlet (7-gas model)
6.3 Distribution of chemical species supply from load-chute and air lances makes the combustion
also partially diffusion-controlled. Therefore, air supply and
The 7-gas combustion model provides the mass-fraction
mixing with the fuel streams become very important.
distribution of all defined seven chemical species: main fuel
Figure 10 illustrates the species distribution along the
component C3H4, two intermediate products (also as the
main-burner axis plane and the line profiles along the burner
fuels for the secondary combustion reactions) CO and H2,
axis. Quick consumption of the fuel components can be
final combustion products CO2 and H2O, and oxidant O2
and inert species N2 brought by air. C3H4 is present in all clearly seen. The CO and H2 concentrations depend very
fuel-rich streams, but also in minor amounts in the fuel lean much on the relative generation and destruction rate.
streams. CO originates both from the fuel-rich and fuel-lean At present, all the rate constants were set equal, and
streams, and from the primary combustion reaction of C3H4. thus C3H4, CO and H2 follow similar trend. However,
From the air lances, minor amount of CO could be if the reaction rate of primary combustion is higher
observed. It should be noted that H2O and CO2 originate than the secondary combustion reactions, CO and H2 will
from both fuel-rich streams and the secondary combustion exist more appreciably for a long time (more downstream).
reactions. This will be tested further in the following stage of research.
According to the detailed analysis of the simulation Also obvious is the peak values of both CO2 and H2O as
results, all the primary and secondary fuel components are final combustion products and CO as an intermediate
reacted quickly upon entry to the system. Since all the product lie in the location about 3 m from the burner exit
burner streams contain certain amount of oxygen (O2), down stream. The minimum value of CO2 at about 6 m
partial combustion takes place at early stage upon entry as downstream from burner axis is due to the air mixing
partially premixed flames. However, the requirement of (bending up from the kiln floor originated from the
more air from the surrounding air inlets and majority O2 load-chute).
Analysis of transport phenomena in a rotary-kiln hazardous waste incinerator 35
Figure 10 Distribution of CO mass fraction across the main In addition, other chemical species are calculated from the
burner plane 7-gas model, and typical distributions of the main fuel and
products of the combustion are illustrated in Figure 12. H2 is
not present in the main fuel, and it is only generated as the
secondary fuel component and is quickly combusted. It has
almost the same distribution pattern as CO.
Air preheating: Air preheating was expected to For non-isothermal and non-uniform gas flow system,
have a big influence on thermal homogenisation. Air simple averaging by cross sectional area could not give
preheating at different levels were tested from 50C, to correct mean temperature data. Therefore, a mass flow
100C, 200C, 300C. It shows that the air preheating weighted approach is developed to obtain the average
up to 200C results in a big improvement in reducing temperature and other flow variables (e.g., chemical
the temperature stratification. This will help to form species). A typical average temperature profile along the
more uniform combustion reactions, and reduce CO main flow direction of the rotary kiln and the SCC can be
concentration possibly originated from low temperature illustrated in Figure 13. This can be very useful for design
zones. and control purposes of industrial furnaces. The data points
in the figure also include the maximum and minimum
Introducing extra cooling air to the kiln: In order to
temperatures at each cross-section, and a temperature profile
safely utilise the ventilation air from the shredding
simply averaged with cross-sectional area. The difference
machine, which contains small amount of combustibles,
between the two different averaging methods could be
it is introduced into the rotary kiln through a cooling
clearly seen.
ring. The influence of this air stream on the combustion
and the temperature distribution was studied at different Figure 13 Comparison of temperature profiles along the main
flow rate of 20006500 Nm3/hr. The results show that flow direction of the incinerator
introducing certain amount of extra air through the
cooling ring reduces the temperature gradient near the
kiln front, and the cooling effect can also protect the
refractory lining from overheating. However, too much
extra air (e.g., 6500 Nm3/hr) will increase the
temperature gradient again and cool the zone near the
kiln front.
Air incoming angle from the load chute: The primary
combustion air from the load chute is introduced to the
rotary kiln at an angle of 50 downward from the
horizontal (50). The simulation results indicate a
large temperature gradient and poor mixing of the air
with the waste streams near the kiln front. Thus
variation of the air incoming angle was studied
from 50, 10, 0, +10 and +50 from the horizontal.
The study indicates that horizontal inflow of the air
results in the least radial temperature gradients, but
also in less mixing with the waste streams. The angle
change has to comply with the overall feasibility in
practice and a more systematic study should be
conducted in accordance with the incoming direction of
all the waste streams.
6.6 Flow of the shredded solid waste by introducing more carrier gas is an effective method for
solid dispersion.
From the Eulerian-Eulerian 2-phase flow model, the volume
To see the particle size effect on the solid particle
fractions of both gas and solid phases can be computed. For
dispersion, three different particle sizes at low injection
the hazardous waste incineration kiln, distribution of the
velocity of 5.5 m/s were studied: 5 mm, 10 mm, and 20 mm.
volume fraction of the shredded solid waste gives a good
However, no obvious difference is found in this range.
indication about where the solid waste could be dispersed
In principle, larger particles tend to fall down to the bottom
and likely be combusted if they are mixed with the
walls due to the greater volume and the resultant gravity
oxidation air.
force. However, the effect may be dumped when the particle
Figure 14 shows the solid volume fraction of 20 mm
momentum is already very low at the injection point. In the
particles across the solid burner plane from transient
2-phase flow models after 15 seconds of solid injection, for future, high gas injection velocity will be simulated to see
both low and high injection velocities: (a) with a low the influence of the particle size.
injection velocity of 5.5 m/s, (b) with a high injection Compared to particle size, particle density has more
velocity of 55 m/s with the same total solid mass flow rate. significant effect on the particle flow and accumulation.
It can be seen that the lower injection velocity leads to In the current study, variation of particle density from 1200,
much earlier drop of the solid waste particles to the bottom 1400, and 4000 kg/m3 has been investigated under steady
of the kiln, and a clear solid bed could be formed. When the state conditions. Results indicate that increase in density
injection velocity is 10 times higher, a much faster solid from 1200 kg/m3 to 1400 kg/m3, a visible difference could
stream could be seen and the solid waste will fly to and already be seen. Compared with a higher density of
settle on the back wall of the SCC and further down to the 4000 kg/m3 a more significant difference can be found.
ash sump. However, a thinner solid bed is still formed. Solid particles with higher density have more momentum
Similar distribution patterns are also observed in the steady (inertia) if they are injected at the same horizontal velocity
flow models. To confirm the inject velocity effect, two (which is the case here), and the particles can travel longer
injection velocity for the lower feeding rate of 2.4 t/h with a distance before they are dragged down to the bottom by the
solid density of 1400 kg/m3 was simulated. Results indicate gravitational force. In practice, heavier particles may not
that the five times higher injection velocity confirmed the gain the same injection velocity as for the lighter particles
expected solid flow pattern. Higher solid injection velocity due to the inter-phase slip effect.
Figure 14 Illustration of the accumulation of the shredded solid waste (dp = 20 mm): solid volume fraction across the solid burner plane:
transient model after 15 seconds solid injection (a) low injection velocity (5.5 m/s) and (b) high injection velocity (55 m/s)
7 Model validation waste incinerator through the viewports. Both 2-meter and
5-metre long thermocouples were used, and the 5-meter
Model validation plays a critical role in all CFD studies.
long thermocouple was designed to use flowing water as
In the current model system, it is important to obtain
cooling medium. Along with the measurement campaigns,
measurement data both for gas flow and temperature as well
temperature recordings from permanent thermocouples at
as chemical species distribution. Because of the hostile
SCC were analysed to see the process dynamics, and
environment inside the incinerator and technical restrictions,
general trend of process stability was observed.
only the exit plane of the kiln and the SCC furnace interior
In order to validate the combustion model, validation
could be accessed through viewports. At present, only
cases from both 3-gas and 7-gas combustion models were
temperature was measured with radiation pyrometer and
built. The predicted temperature profiles were compared
thermal couples. In the current project, two temperature
with the measurement data, as is illustrated in Figure 15.
measurement campaigns were carried out at AVR-Chemie
38 Y. Yang, M.J.A. Pijnenborg, M.A. Reuter and J. Verwoerd
During the measurement campaign, the burner for shredded waste combustion process. The incinerator is a highly
solid waste has not been installed. The total energy input heterogeneous reactor, with high gradients in combustion
was about 37.06 MW, and the waste processing rate was species and temperature, especially along the rotary kiln, the
6.67 ton per hour. primary combustion space. Simulation results indicate that
Since the temperature measurement lasted about the gas flow is clearly 3-dimensional and complicated
2 hours, the process was not really running steadily as was because of the multi-gas streams and strong heterogeneous
monitored from the permanent thermocouples, in spite of reactive nature of the system, and there is a large room to
the great effort from the operators. If taking into account the improve the mixing of the air and the waste streams.
process fluctuations, it can be seen that the model The 3-gas combustion model, especially the 7-gas
predictions are in a reasonable agreement with the measured combustion model could provide a lot of information about
data. The general trend is well predicted, and the larger the incineration process, provided that the waste streams
discrepancy in all 3-cases near the SCC burner side is with very complex nature are properly defined both in
caused by the remaining air from the load chute. The 7-gas chemical and thermal aspects. The present concept of
combustion model predicted somewhat higher temperature the virtual fuel definition is a novel approach and proved to
in general than the measured data, if heat loss is not be a useful exercise and a testing method for the fuel
subtracted in the model. Negligence of the heat loss from optimisation, and any of the similar steps is a precondition
the furnace walls and by the slag formation as well as dust for more detailed combustion modelling of the complex
carry-over is estimated approximately at 10%, which could waste materials. Temperature measurements provided very
drag the temperature down roughly by 120C. This means useful information for calibrating the CFD model, and
also that the 3-gas combustion model may have predicted reasonable agreement has been reached between the
lower temperatures than the measured data. The 7-gas measured and predicted temperature data. It is obvious
model prediction with 10% heat loss shows a better that the detailed combustion modelling of the multi-phase
agreement with the measurements. reacting system still requires more in-depth research, and
further work with multi-phase flow and combustion of the
Figure 15 Temperature profile crossing view-ports of the rotary shredded solid waste is highly needed.
kiln incinerator at AVR-Chemie Through the present study it is found that one of the
most effective ways to increase the solid particle flow
velocity and hence increase the particle dispersion, is to
increase the particle injection velocity. This is already
demonstrated by an extreme case at very high injection
velocity and other models from both steady and transient
simulations. However, it should be noted that too high
injection velocity will result in a shorter residence time of
the particle and less effective suspension combustion.
An optimal injection velocity has to be found to gain a
higher dispersion and a longer mean residence time. Further
research is required for thorough parametric studies and
various design alterations.
Directive 2000/76/EC (2000) European Parliament and the Veranth, J.M., Dachun, G. and Silcox, G.D. (1996) Field
Council, of 4 December 2000 on the incineration of waste, investigation of the temperature distribution in a commercial
Official Journal of the European Communities, Vol. 288, hazardous waste slagging rotary kiln, Environmental Science
No. 12, pp.L332/91111. and Engineering, Vol. 30, No. 10, pp.30533060.
Ficarella, A. and Laforgia, D. (2000) Numerical simulation of Veranth, J.M., Silcox, G.D. and Pershing, D.W. (1997) Numerical
flow-field and dioxins chemistry for incineration plants modeling of the temperature distribution in a commercial
and experimental investigation, Waste Management, Vol. 20, hazardous waste slagging rotary kiln, Environmental Science
No. 1, pp.2749. and Technology, Vol. 31, No. 9, pp.25342539.
Jakway, A.L., Sterling, A.M., Cundy, V.A. and Cook, C.A. (1996) Wardenier, K. and van den Bulck, E. (1997) Steady-state waste
Three-dimensional numerical modeling of a field-scale combustion and air flow optimization in a field scale rotary
rotary kiln incinerator, Environmental Science and kiln, Environmental Engineering Science, Vol. 14, No. 1,
Technology, Vol. 30, No. 5, pp.16991712. pp.4354.
Jenkins, B.G. and Molar, F.D. (1980) Modelling of heat transfer Williams, J.D., Becker, A.R. and Gerovich, M.J. (1988) 3-D flow
from a large enclosed flame in a rotary kiln, Chemical modeling of a hazardous waste incinerator, JAPCA, Vol. 38,
Engineering Research and Design, Vol. 59, No. 1, pp.1725. No. 8, pp.10501054.
Khan, J.A., Pal, D. and Morse, J.S. (1993) Numerical modeling of Wolbach, C.D. and Garman, A.R. (1984) Modeling of destruction
a rotary kiln incinerator, Hazardous Waste and Hazardous efficiency in a pilot-scale combustor, American Flame
Materials, Vol. 10, No. 1, pp.8195. Research Committee International Symposium on Alternative
Leger, C.B., Cundy, V.A. and Sterling, A.M. (1993) A three Fuels and Hazardous Waste, Tulsa, Oklahoma.
dimensional detailed numerical model of a field-scale rotary Yang, Y., Reuter, M.A. and Hartman, D.T.M. (2001) Combustion
kiln incinerator, Environmental Science and Technology, modeling of a rotary kiln waste incinerator with phoenics,
Vol. 27, No. 4, pp.677690. Phoenics Journal of Computational Fluid Dynamics and its
Niessen, W.R. (1995) Combustion and Incineration Processes, Applications, Vol. 4, No. 1, pp.5381.
Marcel Dekker Inc., New York. Yang, Y., Reuter, M.A. and Hartman, D.T.M. (2003) CFD
Spalding, D.B. (1979) Combustion and Mass Transfer, Pergamon modelling for control of a chemical waste rotary kiln
Press, Oxford. incinerator, Control Engineering Practice, Vol. 11, No. 1,
pp.93101.
Themelis, N.J. and Kim, Y.H. (2003) Modeling of transport and
chemical rate phenomena in flash combustion of municipal Yang, Y., Reuter, M.A., Voncken, J.H.L. and Verwoerd, J. (2002)
solid wastes, in Sohn, H.Y., Itagaki, K., Yamauchi, C. and Understanding of hazardous waste incineration through
Kongoli, F. (Eds.): Proceedings of Yazawa International computational fluid-dynamics simulation, Journal of
Symposium: Metallurgical and Materials Processing Environmental Science and Health, Part A: Toxic/Hazardous
Principles and Technologies Vol. 1: Materials Processing Substance and Environmental Engineering, Vol. A37, No. 4,
Fundamentals and New Technologies, Warrendale, PA, TMS, pp.693705.
pp.10311051.