Mathematical and Physical Modeling of Mixing and Flow Phenomena of

Municipal Solid Waste Particles on a Reverse Acting Grate

Masato Nakamura
















Submitted in partial fulfillment of the
requirements for the degree
of Doctor of Engineering Science
in the Fu Foundation School of
Engineering and Applied Science

COLUMBIA UNIVERSITY

2008








































Ⓒ2008
Masato Nakamura
All Rights Reserved


Mathematical and Physical Modeling of Mixing and Flow Phenomena of
Municipal Solid Waste Particles on a Reverse Acting Grate

Masato Nakamura


ABSTRACT

The dominant process for the recovery of energy from municipal solid wastes
(MSW) is mass burning on a moving grate. Also, the leading grate technology is the
Martin reverse acting grate that was modeled in this study. The drying, volatilization,
and combustion phenomena during the travel of the solid waste fuel over the grate
depend on the physical and chemical properties of the diverse particles in MSW, the
design of the grate and combustion chamber, and the distribution of air flowing
through the bed of solids. On the average, it takes one hour for particles to traverse a
10-meter long grate. Therefore, the bed behaves essentially as a fixed bed moving
downwards, by gravity because of the inclination of the grate. In the case of the
Martin grate, the reverse action motion of the grate results in some upward flow and
consequent mixing of the materials within the bed.
This study of the flow and mixing phenomena on a reverse acting grate
included both theoretical and experimental research. In order to characterize the
heterogeneous particle behavior, a two-dimensional stochastic model of particle
mixing within the bed on the reciprocating grate was developed. The model was
calibrated and validated by means of a full-scale physical model of the reverse acting
grate, using tracer particles ranging from 6 – 22 cm in diameter. This size range was
based on a quantitative analysis of the size and shape factor distributions of MSW
samples collected from different locations in New York City. It was found that
different particle sizes result in different residence times because of the Brazil Nut
Effect (BNE). In the BNE, larger particles rise to the surface while smaller particles
migrate to lower depths of the bed where the reciprocating bars push them backward
against the main direction of the MSW flow. The motion of the reverse acting grate
(whose speed ranged from 15 to 90 reciprocations/hour) increases the mean residence
time of small and medium particles by 69 % and 8%, respectively, while decreasing
the mean residence time of large particles by 19%. Also, within this speed range, the
mixing diffusion coefficient of each particle size was determined.
The ratio of particle diameter to the height of moving bar, d/h, was found to be
a major parameter for affecting the mixing diffusion coefficient, D
e
, and the residence
time, t, at reciprocation speeds above 30/hr. When the ratio, d/h, increases from 0.46
to 1.69, the mixing diffusion coefficient, D
e
at 60/hr., decreases from 96 to 38.4. This
indicates that the height of the moving bars should be grater than the diameter of
targeted particles, usually the mean diameter of MSW particles. Accordingly, the
traveling grate operation and moving bar height can be optimized in a more efficient
way, based on these quantitative results and the local MSW particle size distribution.
In the future research, when the stochastic model is combined with particle
size change due to combustion, the stochastic model simulation will become much
more complicated, because the Markov property will be applicable. In this case, the
calculation should be carried out for each single time step (reciprocation), since the
initial condition, instead of only the previous step.
i
Table of Contents


List of Tables ………………………………………………………………...v

List of Figures ……………………………………………………………......vi

Nomenclature …………………………………………………….…............xii

Acknowledgment ..……………………………………………………...xiv

Chapter 1: Introduction …………………………………………………1

1-1. Background …………………………………………………………1
1-1-1. Sustainable waste management and current WTE
technologies …………………..….………………….1
1-1-2. Outline of the dissertation …………………………3

1-2. Overview of waste-to-energy (WTE) technologies ……...……….....4
1-2-1. The mass-burn system ………..………………………..4
1-2-2. The RDF combustion system ...……………………….5
1-2-3. Other combustion systems ..………………………..5

1-3. Traveling grate systems in WTE chamber ………….…..………..8
1-3-1. Reverse acting grate (Martin type) ………..………..8
1-3-2. Forward acting grate (Von Roll type) ………………..10
1-3-3. Roller grate (Riley type) ………………………..10

1-4. Physical and chemical properties of Municipal Solid Waste (MSW)
and combustion residues (ash) ………………………………..12
1-4-1. Physical properties .……………………………….12
1-4-2. Proximate and ultimate analysis …….………………….14
1-4-3. Mass reduction ………………………………………..17
1-4-4. Calorific values of different materials ………………..18

1-5. Physical and chemical transformations of MSW in the bed ………..21
1-5-1. Moisture evaporation in the drying process ………..21
1-5-2. Volatilization in the gasification process ………………..22
1-5-3. Combustion of volatile matter ………………………..24
1-5-4. Combustion of fixed carbon within the bed ………..25

1-6. Mathematical modeling for development of WTE technologies .25

1-7. Shortcomings of existing bed models ...……………………...27
ii

1-8. Objectives of this study ..………………………………………31


Chapter 2: Mathematical Models .……..………………...………………38

2-1. Prior models of grate transport and combustion phenomena ..38
2-1-1. Bed modeling (solid waste combustion) .……………….38
2-1-2. Chamber modeling (volatile combustion) ...……………...45

2-2. Solids mixing theories ………………………………………..46
2-2-1. Diffusion model ………………………………………..47
2-2-2. Diffusion-convection model ………………………..49
2-2-3. Convection model ………………………………………..50
2-2-4. Stochastic model ………………………………………..52

2-3. Mass and volume reduction model ………………………………..53
2-3-1. Progressive-Conversion Model (PCM) ………………..53
2-3-2. Shrinking-Core Model (SCM) ………………………..54

2-4. Heat transfer through the bed ………………………………..59
2-4-1. Radiation among walls and particles ………………..60
2-4-2. Radiation between a gas and a particle ………………..61
2-4-3. Convection between a particle and the primary air though
the bed ………………………………………………..62
2-4-4. Conduction with two particles …………………..……63
2-4-5. Equation describing mass and heat balance ………..64

Chapter 3: Experimental Work ………………………………………68

3-1. Introduction ..……………………………………………....………68
3-1-1.Study of particle size and shape …………..……………68
a) Sieve analysis ………………………………………..68
b) Image analysis ………………………………………..69
3-1-2. Study of mixing and flow …………………..……..……..69

3-2. Measurement of particle size and Shape Factor (SF)
distributions ………………………………………..71
3-2-1. Methodology ……………………...…………...……71
a) Sample collection by field trip in New York City
and by WTE facility visit ………………………..71
b) Sample measurement by image analysis ………..73
3-2-2. Definitions of particle size and shape ………...……...75

iii
3-3. Physical modeling ………………………………………......……78
3-3-1. Methodology ………………..…………...…….……78
a) Geometry of physical model ………………………..78
b) Making tracers in different size ………………………..80
c) Loading NYC-MSW particles ………………………..83
d) Operation conditions ..………………………………83

Chapter 4: Mathematical Work ………………………………………..84

4-1. Mass and volumetric flow of solids and air on the grate ...……...84
4-1-1. MSW flow in the chamber ..………………………………84
4-1-2. Monte Carlo modeling for MSW feeding and flow ..85

4-2. Stochastic model for MSW particles mixing on the traveling
grate ………………………………………. 90
4-2-1. One-dimensional stochastic model
for the reverse acting grate ………………..91
4-2-2. Expansion to a two-dimensional stochastic mode .……….98

4-3. Size (mass and volume) reduction model ………………………100
4-3-1. Introduction ..……………………………...……...100
4-3-2. Modeling ..……………………………...……………...103

4-4. Stochastic model for MSW mixing combined with a flow ….…...105

Chapter 5: Results and Discussion ...…………………………………….112

5-1. Results from experimental work ……………...………...……..112
5-1-1. Particle size and shape distributions ………………112
5-1-2. Probabilities from full-scale physical model ………115

5-2. Results from mathematical work ………………...…………….130
5-2-1. Mass and volumetric flow of solid
(Monte Carlo simulation results) ..…………...………...132
5-2-2. Simulation results from the stochastic model for MSW
mixing ………………………………………………135
a) One-dimensional simulation ………………135
b) Two-dimensional simulation ………………138
5-2-3. Results from size reduction model …...………….144
5-2-4. Integrating model for mixing and flow with data
from the full-scale physical model ………..……..146

Chapter 6: Conclusions ………………………………………………………159

iv

References ………………………………………………………………164

Appendices ………………………………………………………………173

Appendix A: Bed calculation by FLIC ………………………………173

Appendix B: Chamber calculation by CFD ………………………174

Appendix C: Comparison of combustion models ………………178

Appendix D: Movie files on CD-ROM ………………179

v
List of Tables

Table 1-1: Reported WTE Capacity in Europe (IWSA 2003)
Table 1-2: Comparison of reverse acting grate, forward acting grate, and roller grate
systems
Table 1-3: Typical proximate analysis data of MSW and Coal (Tchobanoglous 1993,
Smoot and Smith 1985)
Table 1-4: Typical ultimate analysis data of MSW and Coal (Tchobanoglous 1993,
Smoot and Smith 1985)
Table 1-5: Volatile products of MSW
Table 1-6: Composition of volatile matters
Table 3-1: Various definitions of particle diameters (Allen 1997)
Table 4-1 Ultimate analysis of dry stream municipal solid wastes in New York City
(Themelis et.al 2002, Tchobanoglous et al. 1993)
Table 4-2: Assumed values of probabilities satisfying Eqs (4-10), (4-11) and (4-12)
for the transition matrix used in this calculation
Table 5-1: Particle size parameters obtained from residential MSW and combined
ashes in NYC
Table 5-2: Mixing Diffusion coefficient with different densities [full-scale test] (Yang
et al 2005)
Table C-1: Comparison of combustion models
vi
List of Figures

Figure 1-1: Mass-burn grate combustion system (Von Roll, Brochure of Combustion
System)
Figure 1-2: Combustion Chamber in a RDF system (Themelis 2003)
Figure 1-3: Schematic diagram of fluidized bed system (ref. Takuma)
Figure 1-4: Schematic diagram of rotary kiln system (ref. MHI)
Figure 1-5: Schematic Diagram of Reverse Acting Grate (Martin Grate)
Figure 1-6: Fixed and moving bars of Reverse acting grate (Martin): The
reciprocating bars move against the direction of the feed by gravity: a)
geometry of fixed and reciprocating bars, b) primary air (arrows) injected
through the grate openings.
Figure 1-7: Schematic Diagram of Forward Acting Grate (Von Roll type)
Figure 1-8: Schematic Diagram of Roller Grate (Duesseldorf/Babcock Grate)
Figure 1-9: Particle size distribution in FL (J. Ruf 1974)
Figure 1-10: Waste-To-Energy Combustion System
Figure 1-11: Effects of sample size and heating rate
Figure 1-12: Sample Size effect of pelletized Baker Cellulose, 5.8mg and 180 mg (S.
Gaur and T. B. Reed 1998)
Figure 1-13: Comparison of experimental heating values of various waste materials
(Themelis et. al. 2002)
Figure 1-14: Combustion and transport phenomena of one particle in a MSW bed on
the traveling grate
Figure 1-15: Trend of current bed models
Figure 1-16: Schematic Diagram of Integrated Modeling
Figure 1-17: MSW flow and mixing in the reverse acting grate: The MSW particles
are subjected to different temperatures and their size and composition
change with time and location (e.g., whether a particle is located at the
surface or bottom of the bed)
Figure 1-18: Mass balance of a MSW bed in the combustion chamber
vii
Figure 2-1: Discretization of the fuel bed (Ahmed et al. 1989)
Figure 2-2: Combustion Steps of a Stoker Firing Process and Model Idea (Beckman
1995)
Figure 2-3: Schematic view of the Combustion in a Solid Fuel Bed (Shin et el., 2000)
Figure 2-4: The volume distribution of the various defined components in the bed
(Yang et al. 1999, 2002)
Figure 2-5: The particle-based model using Discrete Element Method (Peters 1994,
Peters et al. 2005)
Figure 2-10: A grid structure of CFD modeling (Nakamura et al.2002)
Figure 2-11: Formation of a striated mixture (Lai et. al. 1978)
Figure 2-12: Progressive-Conversion Model (Levenspiel 1999)
Figure 2-13: Shrinking-Core Model (Levenspiel 1999)
Figure 2-14: Shrinking-Core Model without ash formation
Figure 2-15: Combustion and transport phenomena of one particle in a MSW bed on
the traveling grate
Figure 2-16: Radiation from particle i to all neighboring particles j
Figure 2-17: Conduction between two neighboring particles (Peters, B. 2005)
Figure 3-1: Particle diameter and sieve size
Figure 3-2: Physical Scale Modeling (Lim et al. 2001)
Figure 3-3: New York City map with per capita income (Bowen 2004). X marks are
the locations that black bags were collected
Figure 3-4: The image analyzer system, outlining the shapes of particles regarding
length, width, perimeter, and area
Figure 3-5: Digital camera images of MSW samples (left) and ash samples (right)
Figure 3-6: Various types of diameters (Allen 1997)
Figure 3-7: Full-scale physical section model of a combustion chamber: positions of
CCD cameras (top) and geometries of reverse acting grate (bottom)
Figure 3-8: Spherical tracers of different sizes (small 6cm, medium 14cm, large 22cm)
(left) and NYC-MSW particle size distributions
viii
Figure 3-9: Side views and top views of NYC-MSW beds with 20, 40, 60, and 80 cm
heights and angles of chamber bed decline: 15 degree
Figure 3-10: Cell patterns of 20, 40, 60, 80 cm beds
Figure 4-1: Two-dimensional square lattice site percolation model used in the solid
waste on the combustion grate
Figure 4-2: Channeling and break-up of MSW bed in the percolation model on the
grate
Figure 4-3: Percolation Model in the combustion chamber
Figure 4-4: Illustration of the spatial disposition of cells on the reverse acting grate:
Each cell represents either the reciprocating bar or the fixed bar.
Figure 4-5: Interchange of particles between successive cells (transition graph): The
final cell of this model is cell 17(ash bin); all particles in the MSW feed
after combustion eventually arrive at cell 17, although the paths of
individual particles differ considerably.
Figure 4-6: Illustration of the spatial disposition of cells on the reverse acting grate
Figure 4-7: Particle movement by reciprocating and fixed bars
Figure 4-8: Combustion and transport phenomena of one particle in a MSW bed on
the traveling grate
Figure 4-9: Particle size density function (Gamma distribution) for different values of
α (=1, 2, 3, 4, 5) and constant β (= 5)
Figure 4-10: MSW bed on the reverse acting grate of a WTE plant showing divided
cells (mesh) for the stochastic simulation
Figure 4-11: The elements of the flow matrix F
Figure 4-12: Diagram of probabilities and directions of the flow for each cell of the
MSW bed on the grate
Figure 4-13: The elements of the transition matrix P
Figure 5-1: Distributions by particle numbers of residential MSW and combined
ashes in NYC a) Particle size distributions (PSDs), b) Sphericity
ix
distributions, c) Aspect Raito distributions (ARDs), d) Roundness
distributions
Figure 5-2: Observed particle movements and corresponding state diagram
Figure 5-3: Observed particle movements and corresponding state diagram
Figure 5-4: Observed particle movements and corresponding state diagram
Figure 5-5: Observed particle movements and corresponding state diagram
Figure 5-6: Observed particle movements and corresponding state diagram
Figure 5-7: Observed particle movements and corresponding state diagram
Figure 5-8: Observed particle movements and corresponding state diagram
Figure 5-9: Observed particle movements and corresponding state diagram
Figure 5-10: Observed particle movements and corresponding state diagram
Figure 5-11: Observed particle movements and corresponding state diagram
Figure 5-12: Observed particle movements and corresponding state diagram
Figure 5-13: Observed particle movements and corresponding state diagram
Figure 5-14: Measured probabilities that particles stay in the same position (cell) after
one reciprocation of the moving bars
Figure 5-15: Time series of continuous variation of MSW components by Monte
Carlo simulation
Figure 5-16: Calculation results of the MSW combustion on the Martin grate
Figure 5-17: Combustion probability in each zone
Figure 5-18: Transient phenomena during the combustion process
Figure 5-19: Simulation results of the movement of a solid waste particle on the
Reverse Acting Grate: Top graph: tracing a behavior of one particle on
the bed. In cells #2 and 5 the particle seems to stay put for a while. In
cells #11 and 12, the particle goes back and forth because of the
movement of the reciprocating bar. Bottom graph: Visualization of the
particle travel based on the results of the calculations shown on top
graph.
x
Figure 5-20. Change in tracer particles distribution profiles over grate with number of
movements of reciprocating bars (n=5, 10, 40, 80). The regular
perturbations in the n=40 and n=80 profiles are due to the effect of
alternatively passing over stationary and reciprocating bars.
Figure 5-21: Change in particle distribution profiles over grate with number of
movements of reciprocating bars (a) n=5, (b) n = 10, (c) = 20,(d) n =
40,(e )n= 80
Figure 5-22: Movements (paths) of MSW particles on the Reverse Acting Grate: (a)
n=67, (b) n=106, (c) n=213 and (d) n=401
Figure 5-23: Particle size distributions (PSD) by particle numbers of residential MSW
and combined ashes in NYC: (lines: estimated gamma distributions, dots:
experimental data)
Figure 5-24: Calculated particle path over the grate as a function of size: small
particles (S), medium particles (M), and large (L) particles
Figure 5-25: Brazil Nut Effect (BNE) in a packed MSW bed
Figure 5-26: C (top) and F diagrams (bottom) for small, medium, and large particles
with a reciprocation speed of 90 recip./hr.
Figure 5-27: C and F diagrams for small particles: Dimensionless exit concentration
C (top) and Dimensionless cumulative concentration F (bottom) versus
residence time (min) with different reciprocation speeds ranging from15
to 90 recip./hr
Figure 5-28: C and F diagrams for medium particles: Dimensionless exit
concentration C (top) and Dimensionless cumulative concentration F
(bottom) versus residence time (min) with different reciprocation speeds
ranging from15 to 90 recip./hr
Figure. 5-29: C and F diagrams for large particles: Dimensionless exit concentration
C (top) and Dimensionless cumulative concentration F (bottom) versus
residence time (min) with different reciprocation speeds ranging from15
to 90 recip./hr
xi
Figure 5-30: Mixing Diffusion Coefficients versus grate reciprocation speed for
different particle sizes
Figure 5-31: Wood particle conversion time vs particle diameter (Petek J, 1998)
Figure 6-1: Stochastic models, calibrated by full scale grate tests, can simulate
concept design for novel traveling grate systems
Figure A-1: Calculation results of Bed modeling (Nakamura et al.2002)
Figure B-1: A grid structure of CFD modeling (Nakamura et al.2002)
Figure B-2: Calculated contours of Velocity (m/s) (top) and Total temperature (K)
(bottom)
Figure B-3: Calculated contours of Turbulence intensity (top) and mass fraction of
hydrocarbon concentration (bottom) ((bottom)
Figure B-4: Calculated mass fraction of oxygen concentration (top) and of CO2
concentration (bottom)






xii
Nomenclature

S(0) initial state vector (initial distribution of MSW)
n number of transition (number of the stroke of reciprocating bars)
S(n) state vector after n transitions (distribution of MSW after nth stroke of
reciprocating bars)
P transition matrix
F flow matrix
u
f
feed rate of MSW in the inlet of the mass-burn WTE chamber (cm/min)
d diameter of MSW particle (cm)
D
e
mixing diffusion coefficient (cm
2
/min)
R
r
reciprocation speed of moving bars (recip./hour)
n number of transition (number of the reciprocation of moving bars)
k ratio between feed rate of MSW and frequency of reciprocating bars (cells /
reciprocation) = u
f
/R
r
(how many cells MSW travels during one reciprocation)
t residence time (min)
ˮ

mean residence time (min)
h height of moving bars (cm)
p
i
probability of remaining in state1 after one transition near the inlet
p
r
probability of remaining in a state after one transition on the reciprocating bar
r
r
probability of transiting to the adjacent state on the reciprocating bar
p
f
probability of remaining in a state after one transition on the fixed bar
r
f
probability of transiting to the adjacent state on the fixed bar
p
o
probability of remaining in state16 after one transition near the outlet
f
f
function of R
f
, S, ρ, Rr , L, α, H and θ to determine p
f

g
f
function of R
f
, S, ρ, Rr , L, α, H and θ to determine r
f

f
r
function of R
f
, S, ρ, Rr , L, α, H and θ to determine p
r

g
r
function of R
f
, S, ρ, Rr , L, α, H and θ to determine r
r

R
f
feed rate of MSW in the inlet of the mass-burn WTE chamber
xiii
S particle size of MSW components
ρ particle density of MSW components
R
r
frequency of reciprocating bars
L the length of the reciprocating bars travel
H height of the reciprocating bars
θ angle of the reciprocating bars
α angle of chamber bed decline





xiv
Acknowledgements

First and foremost, I would like to express my deepest gratitude to Professors
Nickolas J. Themelis and Marco J. Castaldi for their support and advice throughout
this research and guiding me to the right direction.
Also, I would like to acknowledge Professors Upmanu Lall, Hoe I. Ling, and
Vijay Modi for helping me as committee members of my dissertation defense.
The financial support of several organizations made this research possible:
Columbia University’s Department of Earth and Environmental Engineering (Henry
Krumb School of Mines), the Earth Engineering Center (EEC), the Waste-to-Energy
Research and Technology Council (WTERT), the Earth Institute (EI), and the Solid
Waste Processing Division (SWPD) of the American Society of Mechanical
Engineers (ASME). I am deeply indebted to these organizations.
Discussion with the engineers and researchers of Covanta Energy Corporation,
especially Dr. Hanwei Zhang, Mark White, and Greg Epelbaum greatly helped me
regarding experimental and modeling work. Dr. Ralf Koralewska from Martin GmbH
introduced me useful theoretical research carried out by European WTE industry and
academics.
I would also like to recognize Liliana Themelis, Peter J. Rennee, Gary S. Hill ,
as well as my research colleagues: Dr. Heidi C. Butterman, Dr. Karsten Millrath,
Dionel Albina, Federico Barrai, Scott Kaufman, Eilhann Kwon, Shang-Hsiu Lee,
Werner Sunk, and Yue Zhou. I have special appreciation for Anuta Belova, Dr.
Fabiola Brusciotti, Dr. Joseph Di Dio III, and Dr. Frank S. Zeman, who organized a
doctoral study group and encouraged me during my dissertation writing process.
Finally, thanks to my wife, Yoshie Tomozumi-Nakamura, my father Kazuo
Ryuwa Nakamura, and my mother, Toyoko Nakamura for helping me along the
whole way through.
1
Chapter 1: Introduction


1-1. Background
1-1-1. Sustainable waste management and current WTE technologies

The United States generates approximately 1.31 short ton per capita of
Municipal Solid Waste (MSW) annually (Kaufman et al. 2002) and the daily
generation of MSW is estimated as 3.6 kg per capita. Sustainable Waste Management
attempts to optimize the energy and resource recovery from collection, transport, and
processing, while minimizing the environmental impact due to final deposition of
residual MSW such as landfilling. Effective collection and transportation of MSW
from residential areas to waste management facilities or landfills are required for
reducing the transportation cost. Materials separation and processing technologies are
used in the recovery of materials, by means of recycling of paper, plastics, metals and
glass. Recovery of energy, known as waste-to-energy (WTE), is the conversion of
post-recycling MSW into energy though the combustion or gasification processes.
Landfilling is the disposal of MSW or the residuals from recycling and combustion
processes. These technologies are utilized in Sustainable Waste Management.
Current WTE technologies are mainly used for the transformation of waste
materials though combustion which reduces the volume and weight of residues for
landfilling and recovers energy and metals. These advantages of WTE have increased
WTE capacities throughout the world. For example, WTE capacities in Europe are
2
shown in Table 1-1. Total capacity in 1999 was more than 40 million tons annually
and the generation of electrical and thermal energy were 41 million GJ and 110 GJ,
respectively.
Country
Tonnes/year
(in 1999)
Kilograms
/capita
Thermal energy
(gigajoules)
Electric energy
(gigajoules)
Austria 450,000 56 3,053,000 131,000
Denmark 2,562,000 477 10,543,000 3,472,000
France 10,984,000 180 32,303,000 2,164,000
Germany 12,853,000 157 27,190,000 12,042,000
Hungary 352,000 6 2,000 399,000
Italy 2,169,000 137 3,354,000 2,338,000
Netherlands 4,818,000 482 9,130,000
Norway 220,000 49 1,409,000 27,000
Portugal 322,000 32 1,000 558,000
Spain 1,039,000 26 1,934,000
Sweden 2,005,000 225 22,996,000 4,360,000
Switzerland 1,636,000 164 8,698,000 2,311,000
UK 1,074,000 18 1,000 1,895,000
Total reported
40,484,000 154.5 109,550,000 40,761,000
(average)

Table 1-1 Reported WTE Capacity in Europe (IWSA 2003)

Combustion of MSW is carried out using WTE units that are broadly divided
into two classes: Those that combust bulk MSW as received (“mass burn”) and those
that pre-process MSW, by magnetic separation, shredding, etc., to what is called
refuse-derived fuel (RDF) units. See Section 1-2-2 for detail.
3
In recent years, WTE technologies have advanced rapidly. For example, the
oxygen-enriched combustion system, which injects industrial oxygen in the primary
air, has been developed by Martin GmbH and Mitsubishi Heavy Industry in order to
generate a better grade of ash and reduce NO
x
and dioxin concentrations (Kira et al.
2001). This technology was tested at a demonstration plant in Germany and is
combined with a flue gas recirculation system in the Martin Syncom process (Gohlke
and Busch 2001). High temperature furnaces, like the thermal plasma torch ash
melting system, have been developed to vitrify fly ashes and eliminate the chance of
future contamination. A combustion improvement system with air injection tubes,
called “Ecotube,” has been developed by ECOMB company in order to reduce NO
x
,
CO, and Volatile Organic Compounds (VOCs). The “prism” system developed by
IBB Engineering GmbH provides for additional secondary injection ports in the
chamber with the objective of stabilizing stable temperature and gas velocity
distributions in WTE units (Klasen et al. 2002). These advanced technologies have
started to be applied to practical use in current WTE facilities.
1-1-2. Outline of the dissertation
This dissertation consists of six chapters. This chapter introduces a
background of the WTE technologies and MSW properties, shortcomings of existing
bed models, and objectives of this research. Chapter 2 gives mathematical models of
WTE combustion including literature review of other related theoretical work.
Chapter 3 is about experimental work of this study, using image analysis for a
4
measurement of MSW properties and a full-scale physical section model of the
reverse acting grate system. Chapter 4 describes stochastic modeling of the flow and
mixing of particles on the traveling grate. Chapter 5 is results and discussion of both
experimental and mathematical work. The final chapter concludes this research with
its major finding and limitation, including suggestions for future work. Abstract of
this study is shown at the beginning of this dissertation.
1-2. Overview of WTE combustion systems (WTE units)
1-2-1. The mass-burn system
The mass-burn system is used by most existing WTE facilities in the world. A
typical mass-burn WTE system is shown in Figure 1-1. It consists of the following
Figure 1-1: Mass-burn grate combustion system (Von Roll, Brochure of Combustion System)
5
sections: 1) combustion chamber, 2)
energy recovery system (boiler), 3)
air pollution control systems (APC).
MSW is brought in by collection
vehicles and is “tipped” on the
unloading bay in an enclosed
building. From there, it is loaded into
the charging hopper by a crane, and
is pushed by feeders into the
combustion chamber. Each mass-
burn combustion chamber is
equipped with a traveling grate the motion of which slowly propels the MSW feed
gradually down the grate. There are several types of traveling grates and will be
described in Section 1-3.
1-2-2. The RDF combustion system
Refuse-Derived Fuel (RDF) is produced from MSW by means of several
processes, such as shredding, separation, and compaction before it is fed into the
combustion chamber of a WTE facility. The RDF system has a combustion chamber
that is physically smaller than a mass-burn chamber, because the heating value of
RDF is higher than that of MSW as-collected. A typical combustion chamber of a
Figure 1-2: Combustion Chamber in a RDF system
(Themelis 2003)

6
RDF system is shown in Figure 1-2. RDF is fed by a fuel distributor into the
combustion chamber, which in this case is equipped with a horizontal traveling grate.
1-2-3. Other combustion systems
There are other combustion systems that are less common in WTE: Fluidized
beds and rotary kiln. A fluidized bed system consists of a vertical steel cylinder and a
sand bed shown in Figure 1-3. This system is operated on a variety of fuels such as
MSW, coal, sludge, and chemical wastes. Rotary kiln systems are also versatile and
can treat a range of materials, ranging from solid to liquid wastes. The rotary kiln,
shown in Figure 1-4, is often used to process medical wastes as well as various
industrial wastes.










7



Figure1- 3: Schematic diagram of fluidized bed system (ref. Takuma)
Figure 1-4: Schematic diagram of rotary kiln system (ref. MHI)
8
Figure 1-5: Schematic Diagram of Reverse Acting Grate (Martin Grate)
RB1
FB1
Outlet of Ash
FBn: fixed bar (step)
RBn: reciprocating bar
(step)
26
Inlet of Solid Waste
1-3. Traveling grate systems in WTE chambers
1-3-1. Reverse acting grate (Martin type)
A typical Martin reverse acting grate is 7 meters long, consists of a total of 15
bars and is inclined 26 degrees to the horizontal. The bars are positioned at an angle
of 13 degrees from the traveling bed. In a common reverse acting grate configuration,
eight of the 15 bars are reciprocating and move a distance of 0.42 m (420 mm). The
frequency of motion of the reciprocating bars can be adjusted from 10 to 60 double
strokes per hour. A typical operating frequency in a WTE unit is 12 strokes per hour
but the frequency used depends greatly on the composition of the MSW feed.

9
A typical feed rate of MSW into the combustion chamber is 18.1 metric tons
per hour (WTE of 480 short tons per day capacity). The average density of MSW is
500 lb per yd
3
or 173.2 kg per m
3
(Tchobanoglous et al. 1993) .Therefore, the
corresponding volumetric feed rate of MSW is 104.8 m
3
per hour. A typical ratio of
the downward volumetric flow rate of MSW feed to the upwards flow of waste due to
the motion of the reciprocating bars (volume of material pushed upward by the bar
motion) is approximately 5/1.

Figure 3-3: Schematic Diagram of Forward Acting Grate (Von Roll type)
Figure 1-6: Fixed and moving bars of Reverse acting grate (Martin): The reciprocating bars
move against the direction of the feed by gravity: a) geometry of fixed and reciprocating bars,
b) primary air (arrows) injected through the grate openings.
10
1-3-2. Forward acting grate (Von Roll type)
Forward acting grate is made up of step grate type hearth as shown in Figure
1-7, in which the moving bars and fixed bars are arranged alternately. Moving bars
reciprocate the same direction with the inlet flow of MSW.
1-3-3. Roller grate (Riley type)
The roller grate (also called the Duesseldorf/Babcock Grate) consists of 6
rollers of 2 m diameter that are inclined at 30
o
to the horizontal as shown in Figure 1-
8. As the rollers rotate slowly, the solid waste moves downward. Each roller “knead”
and turn the solid waste. The rate of revolution for each roller can be varied from 0 to
6 rph.
Figure 1-7: Schematic Diagram of Forward Acting Grate (Von Roll type)
RB1
FB1
Outlet of Ash
FBn: fixed bars
RBn: reciprocating bars
30
Inlet of Solid Waste
11
Characteristics of these three types of traveling grates are summarized shown
in Table 1-2. Forward and reverse acting grate systems are able to push MSW
particles directly due to the motion of their reciprocating bars, whereas a roller grate
conveys MSW particles because of the rotation of the rollers.
Action on MSW
bed
Grate Types
Reverse acting grate Forward acting grate Roller grate
Mechanical Push reverse forward ----
Friction force
(shear stress)
both directions both directions
forward
direction

Table 1-2: Comparison of reverse acting grate, forward acting grate, and roller grate systems

Figure 1-8: Schematic Diagram of Roller Grate (Duesseldorf/Babcock Grate)
30
o

Inlet of Solid
Waste

Outlet of Ash

12
1-4. Physical and chemical properties of Municipal Solid Waste (MSW) and
combustion residues (ashes)
1-4-1. Physical properties
Physical properties of MSW and WTE combustion ashes include density,
particle size, particle shape, and compressibility. These are dominant factors that
determine how solid waste particles travel on the moving grate, transforming during
the combustion process, and finally turning into ashes. The effect of sizes and shapes
of solid waste particles, however, have been less studied, even though they govern
movement and mixing on the grate as well as drying, volatilization, and, combustion
rates. There is no study defining the shape factor of solid waste particles. The only
comprehensive study of particle sizes (length, width and height) and size distributions
was by Ruf, shown as Figure 1-9 (Ruf 1974), but it did not discuss the effect of shape
or shape factor of MSW particles. In this study, physical properties of MSW particles
were measured in order to develop more efficient mass-burn chambers (see Section 3-
2). These are different from
designs of coal chambers, since
MSW, unlike coal, is a non-
homogeneous fuel, of which
combustion process is more
complicated.
Figure 1-9: Particle size distribution in FL (J. Ruf 1974)
13
INPUT (Solid)
Municipal Solid
Wastes (MSW)
OUTPUT (Solid)
Ash
(Fly and Bottom)
OUTPUT (Gas)
Gas Emission

INPUT (Gas)
Air inlet

WTE
Combustion
Chamber
Figure 1-10: Waste-To-Energy Combustion
System
Besides these physical properties of particles, there are some important
physical/physicomechanical characteristics that include agglomeration, aggregation,
bulk density, cohesivity, combustion, compressibility explosion, flowability, packing,
permeability, porosity, reactivity, segregation, and shear behavior. These properties
influence a particle flow through a WTE combustion chamber. It is a material flow
system and can be thought as a black box shown in Figure 1-10. MSW is pushed by
feeders, heated, combusted, and, eventually, pushed though the outlet of chamber,
changing into ashes. In addition to this solid flow, there is the air and emission gas
flows during the MSW combustion.
1-4-2. Proximate and ultimate analysis
Proximate Analysis of MSW is used to determine the percent of four
components (moisture, volatile matter, fixed carbon, and ash) of MSW and consists of
14
the following tests:
• Moisture (loss of moisture when heated to 105
o
C for 1 hour)
• Volatile matter (additional loss of weight on ignition at 950
o
C in covered
crucible)
• Fixed carbon (combustible residue left after volatile matter is removed)
• Ash (residue after combustion in an open crucible)
Table 1-3 shows typical proximate analysis data of MSW (Tchobanoglous 1993) and
coal (Smoot and Smith 1985).
Ultimate analysis is used to determine the elements C (carbon), H (hydrogen),
O (Oxygen), N (Nitrogen), S (Sulfur), and inorganic ash of a sample. To measure
weight percent of C, H, and N, a few mg of sample is heated to a temperature of 1000
o
C and completely combusted in the furnace and N is measured using thermal
conductivity detection. O in the sample is measured by the infrared detector in the
pyrolysis furnace heated to 1300
o
C (LECO 1999). S is also measured by the infrared
detector but heated from 600 to 1450
o
C. These detected data are used to characterize
the composition of the organic matter in MSW. Data of ultimate analysis of MSW
and coal are presented in Table 1-4.


15




Types of Waste Proximate analysis, % by weight
Moisture Volatile matter Fixed carbon Ash (non-
combustible)
Food and food products
Fats 2.0 95.3 2.5 0.2
Food wastes (mixed) 70.0 21.4 3.6 5.0
Fruit wastes 78.7 16.6 4.0 0.7
Meat wastes 38.8 56.4 1.8 3.1
Paper products
Cardboard 5.2 77.5 12.3 5.0
Magazines 4.1 66.4 7.0 22.5
Newsprint 6.0 81.1 11.5 1.4
Paper (mixed) 10.2 75.9 8.4 5.4
Waxed cartons 3.4 90.9 4.5 1.2
Plastics
Plastics 0.2 95.8 2.0 2.0
Polyethylene 0.2 98.5 <0.1 1.2
Polystyrene 0.2 98.7 0.7 0.5
Polyurethane 0.2 87.1 8.3 4.4
Polyvinyl chloride 0.2 86.9 10.8 2.1
Textiles, rubber, leather
Textiles 10.0 66.0 17.5 6.5
Rubber 1.2 83.9 4.9 9.9
Leather 10.0 68.5 12.5 9.0
Wood, trees, etc.
Yard wastes 60.0 30.0 9.5 0.5
Wood (green timber) 50.0 42.3 7.3 0.4
Hardwood 12.0 75.1 12.4 0.5
Wood (mixed) 20.0 68.1 11.3 0.6
Glass, Metals, etc.
Glass and mineral 2.0 - - 96-99
Metal, tin cans 5.0 - - 94-99
Metal, ferrous 2.0 - - 96-99
Metal, nonferrous 2.0 - - 94-99
Residential MSW 21.0 (15-40) 52.0 (40-60) 7.0 (4-15) 20.0 (10-30)
Coal
Pittsburgh bituminous (high volatile) 2.0 36.6 55.4 6.0
Illinois coal bituminous 10.1 35.9 46.7 7.3
North Dakota lignite 29.9 29.5 33.4 7.2
Wyoming subbituminous 27.8 32.9 34.3 5.0
Kentucky bituminous 8.6 35.2 41.5 23.3
Table 1-3: Typical proximate analysis data of MSW and Coal (Tchobanoglous 1993, Smoot and Smith 1985)
16







Types of Waste Ultimate analysis, % by weight
Carbon Hydrogen Oxygen Nitrogen Sulfur Ash
Food and food products
Fats 73.0 11.5 14.8 0.4 0.1 0.2
Food wastes (mixed) 48.0 6.4 37.6 2.6 0.4 5.0
Fruit wastes 48.5 6.2 39.5 1.4 0.2 4.2
Meat wastes 59.6 9.4 24.7 1.2 0.2 4.9
Paper products
Cardboard 43.0 5.9 44.8 0.3 0.2 5.0
Magazines 32.9 5.0 38.6 0.1 0.1 23.3
Newsprint 49.1 6.1 43.0 <0.1 0.2 1.5
Paper (mixed) 43.4 5.8 44.3 0.3 0.2 6.0
Waxed cartons 59.2 9.3 30.1 0.1 0.1 1.2
Plastics
Plastics 60.0 7.2 22.8 - - 10.0
Polyethylene 85.2 14.2 - <0.1 <0.1 0.4
Polystyrene 87.1 8.4 4.0 0.2 - 0.3
Polyurethane 63.3 6.3 17.6 6.0 <0.1 4.3
Polyvinyl chloride 45.2 5.6 1.6 0.1 0.1 2.0
Textiles, rubber, leather
Textiles 48.0 6.4 40.0 2.2 0.2 3.2
Rubber 69.7 8.7 - - 1.6 20.0
Leather 60.0 8.0 11.6 10.0 0.4 10.0
Wood, trees, etc.
Yard wastes 46.0 6.0 38.0 3.4 0.3 6.3
Wood (green timber) 50.1 6.4 42.3 0.1 0.1 1.0
Hardwood 49.6 6.1 43.2 0.1 <0.1 0.9
Wood (mixed) 49.5 6.0 42.7 0.2 <0.1 1.5
Wood chips 48.1 5.8 45.5 0.1 <0.1 0.4
Glass, Metals, etc.
Glass and mineral 0.5 0.1 0.4 <0.1 - 98.9
Metal, (mixed) 4.5 0.6 4.3 <0.1 - 90.5

Coal
Pittsburgh bituminous (high volatile) 77.5 5.3 8.5 1.5 1.2 6.0
Illinois coal bituminous 68.5 5.0 13.8 1.3 3.5 8.1
North Dakota lignite 69.7 3.8 13.2 1.9 1.1 10.3
Wyoming subbituminous 76.3 4.4 10.8 1.1 0.5 6.9
Kentucky bituminous 61.0 4.4 5.6 1.4 4.3 23.3
Table 1-4: Typical ultimate analysis data of MSW and Coal (Tchobanoglous 1993, Smoot and Smith 1985)
17
1-4-3. Mass reduction
The mass and volume of MSW particles are reduced during the combustion
process. The effects of heating rate and sample size in the thermal process are shown
Figure 1-11 and 1-12 (Gaur and Reed 1998). Figure 1-12 represents the size effect of
mass reduction of 5.8mg and 180mg pelletized baker cellulose. These effects are
dependent on heat transfer from the gas film surrounding the particle into the particle.














Big Small Mass %
Temperature
o
C
a) Effect of particle size
High
Low Mass %
Temperature
o
C
b) Effect of heating rate
Particle Particle Heating Rate
Heating Rate
Figure 1-11: Effects of sample size and heating rate
18















1-4-4. Calorific values of different materials

The calorific value (or heating value) of the organic compounds in MSW are
the amounts of energy released when the compounds are burned completely. It can be
determined using an oxygen bomb calorimeter. The colorific value is dependent on
the phase of water or steam in the combustion products. When vapor H
2
O is the
product of combustion, the calorific value is called Lower Heating Value (LHV). If
condensed H
2
O is the product of combustion, it is called Higher Heating Value
(HHV). Themelis et. al. found that composition of organic matter in MSW can be
represented by the molecular formula C
6
H
10
O
4
(Themelis et. al. 2002) and compared
experimental heating values of various waste materials with the percentage of H
2
O
(moisture) content shown in Figure. 1-13.
Figure 1-12: Sample Size effect of pelletized Baker Cellulose, 5.8mg and 180 mg (S. Gaur
and T. B. Reed 1998)
0
20
40
60
80
100
120
0 200 400 600 800
Temp (C)
T
G

%
180 mg
5.8 mg

The calorific value of the MSW compounds
elemental composition quantified by ultimate analysis. There are several
calculate calorific values using the elemental compo
Dulong’s equation had been used by engineers working in
stoker facilities (Cho et. al 1995):
Figure 1-13: Comparison of experimental heating values of various waste materials
2002)
The calorific value of the MSW compounds can be determined using
elemental composition quantified by ultimate analysis. There are several
calculate calorific values using the elemental composition of MSW.
s equation had been used by engineers working in coal and wood
et. al 1995):
: Comparison of experimental heating values of various waste materials (Themelis
19
can be determined using
elemental composition quantified by ultimate analysis. There are several formulae to
coal and wood
(Themelis et. al.
20
H
h
= 81·C +342.5·(H -1/8·O) +22.5S -6·(9·H +W) (Eq. 1-1)
where H
h
is higher heating value, C is carbon fraction (%), H is hydrogen (%), O is
Oxygen (%), S is sulfur (%), W is water (%). This equation provides good
approximated heating values for fuels that mainly contain carbon, such as coal and
wood.
For calorific calculation of MSW, Steuer’s equation has been used (Cho et. al 1995):
H
h
= 81·(C-3/8·O) + 57·3/8·O + 342.5·H + 22.5·S + 57·3/4·O – 6·(9·H+W)
(Eq. 1-2)
The heating value of RDF is expressed as follows:
H
h
= 18400·X
comb
-2636·X
H2O
-628·X
glass
-544·X
metal
kJ/kg (Eq. 1-3)
where X
comb,
X
H2O,
X
glass
, and X
metal
are the fractions of RDF components.





21
1-5. Physical and chemical transformations of MSW in the bed
The MSW physicochemical transformations in the WTE combustion chamber,
as shown 1-14, are composed of three main processes: drying of the surface of solid
waste particles, volatilization as the particles may soften and undergo internal
transformation, and char combustion processes. In addition to these three processes,
the volatiles released from heated solid particles react with oxygen, which is called
volatile combustion.
1-5-1. Moisture evaporation in the drying process
The drying process occurs at the surface of solid waste particles just after
MSW is fed into the combustion chamber. Heat transferred by convection of the
gases surrounding the waste particles and of the primary air injected from the
2. Devolatilization
1. Drying (moisture evaporation)
3. Combustion (char oxidization)
MSW particle
Ash particle
Figure 1-14: Combustion and transport phenomena of one particle in a MSW bed on the traveling grate
Ash particle size distribution
MSW particle size distribution
22
traveling grate into the MSW bed. The rate of solid waste evaporation can be
expressed as:
) (
, ,
2 2
g O H s O H s s drying
C C h A R − = , when T
s
< 100C
o
(Eq. 1-4)
evp ccr drying
H Q R / = , when T
s
> 100C
o
(Eq. 1-5)
)) ( ) ( (
4 4
s env s s g s s ccr
T T T T h A Q − + − = σ ε (Eq. 1-6)
1-5-2. Devolatilization in the gasification process
Volatilization model of first-order reaction process (Badzioch and Hawksley 1970)
) ( v v k
dt
dv
− =

(Eq. 1-7)
) exp(
RT
E
A k − = (Eq. 1-8)
p c
v v Q v ) 1 ( − =

(Eq. 1-9)
where v is the fractional yield of volatiles (mass of volatiles per mass of daf coal), and
v

is the hypothetical ultimate yield.
23
(a) k = 3.12x105 exp[-8961/T] (Badzioch and Hawksley 1970) (Eq. 1-10)
(b) k = 4.3x1014 exp[-27,544/T] (Solomon, et al. 1976) (Eq. 1-11)
A continuous Gaussian distribution of activation energy is another approach to treat
pyrolysis (Anthony et al. 1975):
)
`
¹
¹
´
¦
(
¸
(

¸

− =

∫ ∫
∞ ∞


0 0
2
1
) ( ) ( exp
) 2 (
1
dE E f kdt
v
v v
π σ
(Eq. 1-12)
|
|
¹
|

\
| − −
=
2
2
0
2
1
2
) (
exp
) 2 (
1
) (
σ
π σ
E E
E f (Eq. 1-13)
Tchobanoglous et al. reported data about the mounts of volatiles during
decomposition processes of the MSW. Table 1-5 and 1-6 shows volatile products and
components of MSW with temperature gradient (Tchobanoglous et al. 1985).
Temperature (K) Volatiles (kg/kg waste)
755 0.123
921 0.186
1088 0.237
1199 0.244

Table 1-5: Volatile products of MSW




24




Component (%
vol.)
Temperature (K)
755 921 1088 1199
H
2

CH
4

CO
CO
2

C
2
H
4

C
2
H
6

5.56
12.43
33.5
44.77
0.45
3.03
16.58
15.91
30.49
31.78
2.18
3.06
28.55
13.73
34.12
20.59
2.24
0.77
9.26
10.45
35.25
18.31
2.43
1.07

Table 1-6: Composition of volatile matters

1-5-3. Combustion of volatile matter
In the volatilization process, a wide range of different volatile matters are
generated. This complex process cannot be treated separately. Therefore, assuming a
representative hydrocarbon to be the only combustible product from the volatilization
process, the simplified reaction can be presented as follows:
O H
n
mCO O
n m
H C
m m 2 2
2 4 2
+ → |
¹
|

\
|
+ + (Eq. 1-14)
2 2
2
1
CO O CO → + (Eq. 1-15)
5 . 0 5 . 0 11
2 2
10 3 . 1
O O H CO CO
C C C R × = (Eq. 1-16)
|
|
¹
|

\
|

=
g
O O H g H C
T
C C P T R
n m
12200
exp 8 . 59
2 2
5 . 0 3 . 0
(Eq. 1-17)
25
1-5-4. Combustion of fixed carbon within the bed
Fixed carbon in a solid waste particle is combusted and described as following
equations:
( )
2
) 1 2 ( 1 2 2 ) ( CO CO O solid C − + − → + α α α (Eq. 1-18)
)
6420
exp( 2500
2
T CO
CO −
⋅ = (Eq. 1-19)
|
|
¹
|

\
|
+
=
d r
O
solid C
k k
C
R
1 1
2
) (
(Eq. 1-20)
Eq. 1-20 is char consumption rate expressed by Smoot and Pratt (Smoot and Pratt
1979)
1-6. Mathematical modeling for development of WTE technologies
Further development of WTE technologies will depend on a better
understanding and application of mass and heat transfer phenomena in the WTE
combustion chamber. The transport and combustion phenomena of solid waste
particles traveling on the grate systems are controlled by a large number of
parameters that depend on the design of the combustion chamber (as explained in
Section 1-2 and 1-3) and on the properties of the MSW particles (as described in
Section 1-4 and 1-5). Especially, as mentioned in Section 1-4, MSW is a very non-
26
homogeneous material and the properties of the particles vary widely in terms of size,
shape, specific weight, moisture content, energy content (heating value), and
chemical composition (Carbon, Hydrogen, Oxygen, Nitrogen, Sulfur, and Ash). In
addition, there is large amount of interactions among MSW particles reacting primary
air in a high temperature zone of combustion chamber. As the MSW particles travel
over the moving grate, they interact with each other and with the flow of primary air
though the bed and their temperature, size and composition changes with time and
location on the bed (e.g., whether a particle is located near the surface or the bottom
of the bed)
One of the most effective ways to understand those transport phenomena and
chemical rate reactions is by mathematical modeling of fluid flow though the bed and
above the bed, heat transfer between the furnace and particle/gas, mass transfer
between particles and air flow, and chemical reactions (i.e. mostly combustion)
within the bed and in the chamber above the bed. Simulating these phenomena in a
computer is less costly than building physical models and is helpful to develop a
better chamber design.
The combustion chamber simulation consists of two major components: 1) a
bed model of the drying, volatilization and chemical reactions within the bed, and 2)
CFD modeling of the gas flow, heat transfer and gas-phase reactions in the chamber
above the grate. These models have been developed separately. Generally, there have

been more studies on CFD
and corrosion of the chamber
WTE facilities and developers
1-7. Shortcomings of existing
There have been several different approaches to model WTE combustion
systems in the past (see
widely accepted and used in the WTE industry, the solid waste bed modeling still
Figure 1

CFD chamber modeling because the behavior of emission gases
chamber walls have been the main concerns of both
developers of WTE technologies.

xisting WTE models
here have been several different approaches to model WTE combustion
(see Chapter 2 in detail). Whereas CFD chamber modeling is
widely accepted and used in the WTE industry, the solid waste bed modeling still
Figure 1-15: Trend of current bed models
27
the behavior of emission gases
both operators of
here have been several different approaches to model WTE combustion
). Whereas CFD chamber modeling is
widely accepted and used in the WTE industry, the solid waste bed modeling still
28
needs to include an algorithm for overcoming the heterogeneity of MSW gasification
and combustion.
Figure 1-15 shows previous studies on bed modeling that can be categorized
in two classes: empirical modeling and theoretical modeling.
Empirical models are mainly used for practical basis such as training operators
of WTE facilities, predicting approximate results for combustion chamber
maintenance, and calculating initial or boundary conditions, like bed surface
temperature, as a pre-process of CFD modeling. These models are practical and
sometimes called “industry-friendly” because of their simplicity, without long
calculation time and need to use extensive computer resources.
Theoretical models are used for academic studies such as establishing new
theories, verifying them with experimental results, and sometimes developing new
grate systems. These models include complicated MSW flow and combustion. Hence,
they need a lot of run-time of computer calculation. Theoretical models, in general,
express bed transport and combustion phenomena in more detail than empirical
models, but do not represent the heterogeneity of MSW. The problem here is that
complicated simulation results do not always agree with the real chamber situation.
For example, the Discrete Element Method (DEM) applied in Peters’ model is a new
approach in order to predict particle movements, but he considers one single spherical
particle type such as density (700kg/m3), Young’s modulus (10MPa), and shear
29
modulus (3 MPa) for heterogeneous MSW. This “ideal MSW particle” does not
match the diversity of MSW in a real combustion chamber. On the other hand,
empirical bed models are too simple to simulate solid waste flow and combustion
phenomena. In Shin’s model, for example, the interaction between sections of MSW
bed, which is agitated by a traveling grate, was not taken into account. Only the
height of each section reduced by thermal conversion was discussed in this model.
This discrepancy in bed modeling research is due mainly to the heterogeneous
characteristics of MSW and the complexity of its physical and thermochemical
conversion in a mass-burn combustion chamber.
In addition to those shortcomings of both empirical and theoretical modeling,
the current bed models (see the discussion in Chapter 2) are not equipped to predict
other effects during the combustion processes on the grates:
1. Continuous variation of solid flow (inlet through outlet, limitation of steady-
state bed models)
2. Transient phenomena (channeling through the bed and break-up effects of
MSW particles)
3. Mixing of the MSW particles is not taken into account adequately (even
though the solid mixing process is one of the principal factors for promoting
combustion)
30
The main reason for these shortcomings is that past studies have considered
MSW as one layer, not as particles. Addressing these issues in a novel fashion, i.e.
particle-based analysis, can help alleviate some of the problems experienced in
current combustor design. Examples of the major problems that need to be addressed
are:
1. Worldwide, there are 700 WTE plants with different types of traveling grate
systems, such as reverse acting, forward acting, and roller grate. However,
there are no methods for evaluating the advantages or disadvantages of
various traveling grate systems for mass combustion of non-homogeneous
MSW, in terms of throughput, energy, etc.
2. The lack of mixing results in incomplete combustion of solid waste.
3. Unstable combustion caused by the inlet flow of highly non-homogeneous
MSW results in unstable burnout point.
4. The channeling of airflow through the grate results in a high excess air
requirement and lower temperatures in the chamber. Yang et.al. have been
studying this effect though both theoretical and experimental analyses (Yang
et. al. 2003).

The relationship among mixing phenomena, chamber designs, and chamber
operation on the mass-burn traveling bed is not fully understood and is not fully
included in existing bed models. These shortcomings result from lack of quantitative
analysis of the movement of MSW particles during the combustion process. In order
31
to quantify the motion of MSW particles on the various types of grates, some studies
using small-scale physical models were carried out. For example, particle tracking
tests using 1/15 scale models developed by Lim et al clarified mixing phenomena
enhanced by the traveling grates (Lim et al 2001). Residence time of each wooden
and ceramic particle was measured using a scale model of forward and reverse acting
grate systems (Beckmann et al 2000). Even though these studies were worthwhile,
they did not accurately replicate a real combustion chamber and grate using physical
scale modeling. One of the difficulties of physical small-scale models has been that
the range of particle sizes, densities and shape factors encountered in the MSW fed to
WTE combustion chambers is impossible to simulate in small physical models. Also,
the physical small-scale models cannot simulate the volatilization and combustion
phenomena on the grate that change the density of solids by an order of magnitude.
1-8. Objectives of study
The objectives of this doctoral research are mainly to provide a better tool for
understanding effects of solid flow and mixing phenomena enhanced by the motion of
traveling grate:
1. Developing a mathematical model to characterize and quantify flow and
mixing processes of the highly non-homogeneous MSW.
2. Analyzing the relationship between the behavior of solid waste particles and
operation parameters (especially, reciprocation speed of moving bars) as well
as chamber design parameters (traveling grate)

3. Calibrating and validating
model of reverse acting grate with camera
particle movements)
analysis of NYC-MSW),
In order to develop
current bed models described in the
incorporate two mathematical
mixing model under a CFD model,
Figure 1
Calibrating and validating mathematical models using full-scale
model of reverse acting grate with camera observation (analysis of
particle movements) and “as-collected” MSW packed bed (collection and
MSW),
develop an integrated bed model to address the limitations of
described in the previous section (Section 1-7), it is necessary to
two mathematical models based: a bed combustion model, and a
under a CFD model, as shown in Figure 1-16.
e 1-16: Schematic Diagram of Integrated Modeling
32
scale physical
observation (analysis of MSW
(collection and
model to address the limitations of
is necessary to
a bed combustion model, and a bed

33

The bed modeling results have been expressed in terms of two parameters: a)
the residence time of particles with different sizes, and b) the mixing diffusion
(dispersion) coefficients along the length (D
l
) of the bed on the grate. These two
parameters will be correlated to equipment and operating parameters that can be
changed either during design or operation (e.g., diameter, number and speed of
rotation of roller grates; number, height, frequency/speed of Martin moving grates).

Figure 1-17: MSW flow and mixing in the reverse acting grate: The MSW particles are subjected to
different temperatures and their size and composition change with time and location (e.g.,
whether a particle is located at the surface or bottom of the bed)

V1=102.74 m
3
/h
V2=97.71 m
3
/h
V3=83.65 m
3
/h
V4=54.52 m
3
/h
V8=6.31 m
3
/h
Have
FB2
RB2
RB1
FB1
Inlet of Solid Waste
(104.75 m
3
/h)
[18.14 metric ton/h]
Outlet of Ash
(4.30 m
3
/h)
[1.87 metric ton/h]
L=7 m
FBn: fixed bar (step)
RBn: reciprocating bar
(step)
0.2 m
0.04 m
0.19 m
13
o

26
V5=27.40 m
3
V6=16.35 m
3
/h
26
V7=9.23 m
3
/h
26.68 m
3
/h/bar
Zone
1 Zone
2
Zone
4 Zone
5
Zone
6
Zone
7
Zone
8
Zone
3
L: Effective Length of the grate: 7 m
Vn: Velocity of Solid Wastes
Have: Average height of solid wastes: 0.5 m
Volumetric flow rate of solids downward: 104.75 m
3
/h
Volumetric flow rate of solids upward: 26.68 m
3
/h/bar


34
Figure 1-17 shows the MSW flow and mixing on the reverse acting grate
system in a WTE facility. The forces to be considered in the mathematical model are:
a) the gravity force acting on the MSW deposited on the grate; for example, in
the absence of any motion of the grate, material deposited on a horizontal
plane would form a solids cone of a predictable angle of repose. In the case of
an inclined plane (e.g., 26
o
in the case of the Martin grate), the downward
angle of the cone would be proportionally greater.
b) the motion of the grate imposes an additional shear force on the bottom
layer of the solids that is in contact with the grate surface. Depending on the
“viscosity” of the solids on the grate, this shear force is transmitted to upward
layers, similarly to what would happen in a liquid. The combination of gravity
and shear forces due to the motion of the grate results in the downward motion
of the MSW particles.
The study is based on the dimensions and operating characteristics of an
actual industrial WTE plant. The flow rate of MSW fed in the inlet of combustion
chamber is 480 short tons per day or 18 metric tons per hour. The specific density of
MSW is 500 lb/yd
3
or 0.173 tons per cubic meter. Hence, the total volumetric rate of
MSW fed by the pusher is 105 m
3
/hr. If the height of the MSW bed is assumed to be
1.0 meter at any position, the total volume of the MSW bed is 56 cubic meters (8 m in
width, 7m in length, and 1 m in height). Therefore, the calculated residence time is
35
0.53 hour or 32 minutes. However, this residence time calculation of the MSW flow
does not include combustion and mixing phenomena: a) volume reduction of MSW
during the drying, gasification, combustion processes and b) mixing of MSW due to
the motion of the traveling grate.
Regarding MSW volume reduction, as noted earlier, there are mainly three
processes: drying, volatilization, and combustion (char oxidization). As shown in
Figure 1-18, the mass balance of this WTE chamber system is described as follows,
m
inlet
= m
drying
+ m
VM
+ m
FC
+ m
outlet
(Eq. 1-21)
minlet moutlet
mdrying mVM
mFC
Zone 1 Zone 2 Zone 3
Figure 1-18: Mass balance of a MSW bed in the combustion chamber
Zone 1
Zone 2
Zone 3
MSW
ASH
Combustion
Chamber
36
where, m
inlet
, m
drying
, m
VM
, m
FC
, m
outlet
are weight % of raw MSW, moisture, volatile
matters, fix carbon, and ash, respectively. A typical residential MSW contains about
21% of moisture, 52% of volatile matters, 7% of fixed carbon, 20% of ash (see
Section 1-4-2). Since each mass reduction corresponds to a volume reduction, the
total volume reduction of MSW in the chamber will be in accordance with this
proximate analysis of MSW.
In terms of MSW mixing on a common Martin grate, there are 8 moving bars
that reciprocate 12 strokes per hour. Volumetric flow rate of bar motions is 19.7 m
3
/hr
(see Chapter 2 for detail). The MSW flow is described by the dispersion model:
x
c
u
x
c
D
t
c
e





=


2
2
for 0 < x < L (Eq. 1-22)
where D
e
is eddy diffusivity, u is velocity in flow direction, L is length of the
combustion chamber, x is distance along reactor, in direction of flow. In
dimensionless form where z = (ut+x) / L and θ=t / t = tu /L, equation 1-2 becomes
z
c
z
c
uL
D c
e





|
¹
|

\
|
=


2
2
θ
(Eq. 1-23)
where
|
¹
|

\
|
uL
D
e
is the inverse of the dimensionless Peclet number, P
e
. If P
e
is smaller
than 0.01 (close to zero), dispersion in the MSW flow is negligible and called plug
37
flow. When P
e
is larger than 0.01 the MSW flow has large dispersion and called
mixed flow.
The motion of the traveling grate is important in mixing MSW in the bed
since the major source of heat transfer to the MSW bed is radiation from flame of
volatile matter combustion and chamber walls. If there is no grate motion and the
MSW flow has no dispersion (i.e. P
e
< 0.01), only particles located near the surface of
the bed will be heated up and reach the drying and devolatizing temperatures. The
inside of the bed will be heated by conduction and then will slowly reach these
temperatures (A flame front will move from top to bottom, as it can be seen in a batch
reactor). Therefore, for estimating the rate of reactions or heat transfer, the Peclet
number P
e
of the MSW flow is one of the most important parameters. The projections
of the mathematical bed models will be validated by a) numerical analysis of physical
characteristics, and b) observation of the MSW particles in the bed using full-scaled
physical section model of traveling grate (reverse acting grate) (see Chapter 3 and 5).
In addition to these objectives, this doctoral research hopes to advance the
design and operation of WTE combustion chambers for high energy efficiency. This
includes assessing of current WTE combustion chamber systems and suggesting the
design of a new traveling grate for promoting combustion.
38

Radiation from the flame and the furnace wall
Air
Ash
Waste
Figure 2-1: Discretization of the fuel bed (Ahmed et al. 1989)
H
∆x
L
B

Chapter 2: Mathematical Modeling


2-1. Prior models of grate transport and combustion phenomena
2-1-1. Bed modeling (solid waste combustion)
A bed model developed by Ahmed et al. (1989) dealt with only pyrolysis
reactions of MSW that was simplified to be composed of only cellulosic material. As
shown in Figure 2-1, the volume reduction process in the MSW bed is simply
proportional to the decomposition rate of cellulose. The MSW bed is ideally well-
mixed in the vertical direction. The flame zone above the bed has a constant
temperature and it is the heat source for the bed. The calculation method of this bed
39

modeling incorporated the pyrolysis kinetics, enthalpy of pyrolysis reactions, heat
transfer by convection and radiation, kinetics and thermodynamics of volatile matter
combustion, and material transport by the traveling grate. In order to examine the
effects of the solid feed rates, several parameters are considered such as the air flow
rates, the bed and gas temperature profiles, and the height and length of the
combustion chamber required for MSW conversion. This study showed simplified
trends of these effects, but a detailed quantitative analysis for bed modeling was not
undertaken.
40


Beckmann et al. (Beckmann et al. 1995) considered a combustion chamber as
combination of several continuously stirred reactors (CSRs) shown in Figure 2-2. In
this bed model they defined an effective reaction coefficient determined by data from
a batch-stoker test plant. For both steady state and non steady state conditions
modeling results were compared with experimental results. This research concluded
Radiation from the flame and the furnace wall
Air and recycled flue gas etc.
Ash
Waste
Cascade of Continuous
Stirred Reactor Elements
(CSR)
Grate
Waste
Gas Gas Gas
Air Air
Air
Xj1 Xj2
Figure 2-2: Combustion Steps of a Stoker Firing Process and Model Idea (Beckman 1995)
41

Ash zone
Combustion zone
Pyrolysis zone
Heat-up zone
Temperature
Air
Radiation Heat
Initial Bed Height
Grate Level
Combustion Gas
Fuel Temperature
Gas Temperature
Radiation from the flame and the
furnace wall
New Fuel
Waste Moving Direction (Time)
Ash
Raw Fuel
Figure 2-3: Schematic view of the Combustion in a Solid Fuel Bed (Shin et el., 2000)
Grate
Reaction Zone
Ash
that the model was successful for describing non steady combustion processes of
MSW.









42

Figure 2-4: The volume distribution of the various defined components in the bed (Yang et al.
1999, 2002)
Drying, pyrolysis and gasification
Gasification only
Time
Key
Gas space
Fixed carbon
Volatiles
Moisture
Internal pore space
Free ash
Bound ash

DRYING PYROLYSIS GASIFICATION
Initial Waste (B):
Moisture
Volatiles
Carbon
Ash
Dried Waste (C):
Volatiles
Carbon
Ash
Dried and pyrolysed
waste (D):
Carbon
Ash
Dried, pyrolysed and
gasified waste (A):
Ash
V
B
V
C
V
D
V
A

An unsteady one-dimensional bed model was developed by Shin et al. (Shin et
al 2000). This bed model consists of a waste shrinkage sub-model and a two-flux
radiation sub-model shown in Figure 2-3. Because of heat transfer by radiation from
the flame and furnace walls, the temperature on the surface of a MSW bed increases.
This transport phenomenon was described in this model and compared with
experimental results. The three stages of material conversion such as raw MSW,
reacting zone, and ash in the bed were identified. This model results were
43

incorporated with a CFD model by Ryu et al. (Ryu et al. 2001).
Yang et al. (Yang et al. 1999, 2002) developed a comprehensive bed model
called FLIC shown in Figure 2-4. This two-dimensional bed model was incorporated
with various sub-process models and solved the governing equations for mass,
momentum, and heat transfer for both gases and solids. FLIC includes several
characteristics: Gaussian distribution of activation energies for waste devolatilization,
mixing effect between volatile matter and oxidant in the volatile combustion rates,
and the momentum equation for the gaseous phase. Validation of this bed model was
achieved by means of a bench-top batch furnace used in this study. However, the sub-
models are too detailed to determine some coefficients such as the central value and
variation of waste activation energy. The FLIC simulation is mainly used by
academic researchers when all parameters are known or assumed because MSW is a
much more heterogeneous materials than any other material that scientists have ever
studied.
Peters et al. (Peters 1994, Peters et al. 2005) present a model for the
calculation of an unsteady, three-dimensional flow and combustion phenomena in a
packed bed of a solid waste combustion chamber. As shown in Figure 2-5, this bed
model simulated a bed comprised of wood particles that mechanically interacted with
neighbor particles. On the bases of the results of this study, they developed a new
44

model for mixing MSW particles on a forward acting grate using the Discrete
Element Method (DEM), but it did not include any combustion phenomena.





Other numerical bed models have been developed for simulating solid waste
combustion processes (chemical reactions and mass/heat transfer). CUTEC-Institut
has developed a dynamic model for optimizing MSW combustion (Reindorf et al).
Wolf has simulated mixing phenomena on a reverse acting grate in waste combustion
systems (Wolf, C., 2005). Empirical models have also been developed for training
facility operators rather than for research or development. Program codes of them are
much simpler than the codes of phenomenal models.

Figure 2-5: The particle-based model using Discrete Element Method (Peters 1994, Peters et al. 2005)
45

Figure 2-10: A grid structure of CFD modeling
(Nakamura et al.2002)


MSW
Ash
Flue gas
Air/recirculated
flue gas injection
Air injection
Gas flow
(Computed from a bed model)

2-1-2. Chamber modeling (volatile combustion)
Chamber modeling that uses data computed by a bed model is used to
investigate volatile matter combustion above the bed. Volatile matter combustion
generated during the thermal conversion process of MSW bed is dependent on the
flow and mixing of gases and air. The chemical reactions are carried out between
volatile matter and secondary air injected by arrays of nozzles on the chamber walls.
The flow of gas and air in the chamber results in turbulence and mixing, depending
on temperatures, velocity, and secondary air distribution. In order to calculate these
complicated phenomena in the
combustion chamber, Computational
Fluid Dynamics (CFD) is widely used
for chamber modeling.
Computational Fluid Dynamics
(CFD) is concerned with obtaining
numerical solutions to fluid flow
problems by using the continuity
equation (conservation of mass), the
Navier-Stokes equation (conservation
of momentum), and the energy
conservation equations. CFD is one of
46

the best numerical tools for WTE engineers, and is used for geometric analysis of a
new chamber design to calculate air and gas flows as well as for the chemical
reactions of volatiles (C
m
H
n
) in the chamber. Figure 1-9 shows the grid structure used
for Fluent CDF modeling of the WTE mass-burn combustion chamber in Union
County Facility (Nakamura et al. 2002). Commercial CFD software packages such as
CFX (ANSYS, Inc.) and Fluent (Fluent Inc.) usually need an additional program sub-
code, since CFD does not consider solid waste combustion in a bed.


2-2. Solids mixing theories
Solids mixing is widely used in the manufacture of ceramics, glass, plastics,
processed food, sand and fertilizers because mixing processes in these industries are
necessary for controlling rates of mass and heat transfer and chemical reactions.
Mixing processes can be classified in many ways based on the modes of solids flow
mixing mechanisms that are involved. There are some types of mixing systems: free-
flow systems and fluidized systems. Free-flow systems carry out a process of
randomization of particles induced in mechanisms such as diffusion, confection, or
both, depending on the type of mixer or traveling grate. Reverse acting grate, forward
acting grate, and Roller grate in Mass-burn facilities can be categorized in free-flow
mixing. Whereas, fluidized bed systems for MSW can be in fluidized systems that
have several effects of bubbles: the fluidized bed is considered to comprise a
discontinuous phase of bubbles and a continuous phase of gas and solids. In this
47

section (2-2), theories of mixing processes in free-flow systems, to which reverse
acting grate belongs, have been explained.
Mixing in Free-flow system can be modeled by four types of models: a)
diffusion model (Section 2-2-1), b) diffusion-convection model (Section 2-2-2), c)
convection model (Section 2-2-3), and d) stochastic model (Section 2-2-4).
2-2-1. Diffusion model
Diffusion model for molecules is a possible mixing mechanism of solids. This
mechanism is described by Fick’s equation:
2
2
x
c
D
t
c
e


=


(Eq. 2-1)
The mixing process can be well characterized by the diffusion model when the
mean diameter of binary component particles is same. In order to describe movement
of particles in the axial direction, the following equation was applied to the
experimental data:
2
2
) , ( ) , (
x
n x p
D
n
n x p
a


=


(Eq. 2-2)
where n is number of revolutions of the mixer, p(x, n) is relative concentration of the
particles which is described as follows:
48


=
L
dx n x c
n x c
n x p
0
) , (
) , (
) , ( (Eq. 2-3)
where L is the length of mixer and c(x, n) is particle concentration that is a function of
distance and number of revolutions. The solution of Eq. 2-3 is obtained as
(
¸
(

¸

− =
n D
x x
n D
x n x p
a
a
4
) (
exp
4
1
) 0 , | , (
2
0
0
π
(Eq. 2-4)
where x
0
is the initial location of the tracer. This equation can be rewritten in the form
of Gaussian distribution as
(
¸
(

¸

− =
) ( 2
) (
exp
) ( 2
1
) 0 , | , (
2
2
0
2
0
n
x x
n
x n x p
x
x
σ
πσ
(Eq. 2-5)
Compared with Eq. 3-3, this equation has variance σ
x
2
(n) that related to the
diffusivity D
a
, through Einstein’s equation, as
|
|
¹
|

\
|
=
n
L
dn
n d
D
x
a
2 2
) (
2
1 σ
(Eq. 2-6)
The diffusivity D
a
*
is obtained by
49

N D D
a a
=
*
(Eq. 2-7)
where N is the rotating speed of the mixer. The experimental data shows that the
diffusion coefficients of the particles of different densities are larger than those of the
particle of the same density.
2-2-2. Diffusion-convection model
Diffusion-convection model, as concisely shown in Chapter 1, is described by the
diffusion equation (Eq. 1-22) with an additional term including drift velocity. Hwang
and Hogg (Hwang and Hogg 1980) examined the mixing of powders flowing over a
inclined surface in order to validate the diffusion-confection model. Under steady
state conditions, Eq. 1-22 is simplified followings:
x
c
u
x
c
D
e


=


2
2
(Eq. 2-8)
Dispersion coefficient of particles, D
e
, is dependent on shear in the flowing material.
From their experiment results, the diffusion coefficient varies linearly with the
velocity gradient:
|
|
¹
|

\
|


+ =
y
u
L D D
e e
1
*
(Eq. 2-9)
50

D
*
is the “intrinsic” dispersion coefficient due to random oscillations or fluctuations
of the individual particles during flow. The L includes interparticle collisions
enhanced by the velocity gradients. These parameters are dependent on the
characteristics of the flowing powder and the flow system.
2-2-3. Convection model
A convection model is developed to represent convective mixing by a process
of forming a striated mixture. This process is a repetitive series of division and
recombination. The process of forming a striated RR-mixture (regular packing
arrangement mixture with regular pattern), for example, shown in Figure 3-1. There
are two striate of a rectangular block. The block is compressed to one-half of its
original height, cut into halves, and recombined. Let x
k
(N)
denote average
concentration of a component in cell k (k = 1,2,3,4) after N cycles of the mixing
operation. The average concentration distribution after N cycle is expressed by
) 0 ( ) 1 ( ) 0 ( ) ( ) (
] [ x P x P x
N N N
= = (Eq. 2-10)
where
| |
T
N N N N N
x x x x x
) (
4
) (
3
) (
2
) (
1
) (
= (Eq. 2-11)
51

(
(
(
(
(
¸
(

¸

=
) (
44
) (
43
) (
42
) (
41
) (
34
) (
33
) (
32
) (
31
) (
24
) (
23
) (
22
) (
21
) (
14
) (
13
) (
12
) (
11
) (
N N N N
N N N N
N N N N
N N N N
N
p p p p
p p p p
p p p p
p p p p
p (Eq. 2-12)
For the case shown in Figure 2-11,
(
(
(
(
¸
(

¸

=
2 / 1 2 / 1 0 0
0 0 2 / 1 2 / 1
2 / 1 2 / 1 0 0
0 0 2 / 1 2 / 1
) 1 (
p (Eq. 2-13)
| |
T
x 1 1 0 0
) 0 (
= (Eq. 2-14)
Hence,
(
(
(
(
¸
(

¸

= =
1
0
1
0
) 0 ( ) 1 ( ) 1 (
x P x (Eq. 2-15)
(
(
(
(
¸
(

¸

= =
2 / 1
2 / 1
2 / 1
2 / 1
] [
) 0 ( 2 ) 1 ( ) 2 (
x P x (Eq. 2-16)
52

This means that after two cycles, the average concentrations in all the cells becomes
uniform.
2-2-4. Stochastic model
A non-stationary Markov chain model was applied to the mixing and
segregation occurring within vibrated poly-dispersed particulate bed (Chou et. al.
1977). Boss and Dabrowska proposed a stochastic model for the mixing process,
during discharge of particles from bins, of multi-component homogeneous solids
N=0
N=1
N=2
1
4
3
2
1
4
3
2
1
4
3
2
1
4
3
2
1
4
3
2
1
4
3
2
1
4
3
2
1
Figure 2-11: Formation of a striated mixture (Lai et. al. 1978)

Cut from here
53

(Boss and Dabrowska 1985, 1985). The concentration distribution and the degree of
mixing were calculated with the transition matrix in a Markov chain.
The advantages of the stochastic approach are a) that particle distributions and
particle motilities can be studied simultaneously; b) the process of mixing can be
examined in detail, c) particle movement simulation can be quickly carried out more
than Newtonian approach based on the distinct element method (DEM) or particle
element method (PEM). In the consideration of these advantages, stochastic approach,
based on Markov chains, is chosen in this study (see Chapter 4) for simulating the
process of MSW mixing.
2-3. Mass and volume reduction model
2-3-1. Progressive-Conversion Model (PCM)
Progressive-Conversion Model (PCM), also called Continuous-Reaction
Model (CRM), is used when solid reactant is converted progressively throughout the
solid particle. In this case the diffusion of gaseous reactant into the particle is rapid
more than chemical reaction (Kunii and Levenspiel 1991).
54


Figure 2-12: Progressive-Conversion Model (Levenspiel 1999).
3-3-2. Shrinking-Core Model (SCM)
Shrinking-Core Model (SCM), or Unreacted Core model, is used when the
diffusion into the reactant particle is slow. The reaction occurs at the outer surface of
the particle first. Thus, an unreacted core shrinks in size during reaction, as illustrated
in Figure 2-13 (Levenspiel 1999).

Figure 2-13: Shrinking-Core Model (Levenspiel 1999).
R R
Concentration of
solid reactant

Original
concentration

Time

Time

R
R R 0
R
Unreacted Core

Ash
R R
0
Concentration of
solid reactant

R R
0
R R
0
Original
concentration

Time

Time

55

A gas contacts a solid particle, reacts with in the particle, and transforms it
into product. This kind of heterogeneous reaction may be expressed by the following
equation:
A (fluid) +bB (solid) → fluid and solid products. (Eq. 2-17)
In this heterogeneous reaction there are three steps: 1) Diffusion through gas film
controls, 2) Diffusion through ash layer controls, 3) Chemical reaction control.
1) Diffusion through gas film controls
The rate of reaction of Eq. 2-17 is
A B
bdN dN = (Eq. 2-18)
Where N
A
and N
B
are amount of A and B respectively, b is stoichiometric coefficient
for B. Thus,
. ) (
1
const C bk C C bk
dt
dN
S
b
dt
dN
S
Ag g As Ag g
A
ex
B
ex
= = − = − = − (Eq. 2-19)
R S
ex
π 4 = (Eq. 2-20)
where S
ex
is unchanging exterior surface of a particle, k
g
is the mass transfer
coefficient, C
Ag
is the concentration of gaseous reactant A, C
As
is the concentration of
reactant A on the surface of the particle. C
As
= 0 because no gaseous reactant is
present at the particle surface.
56

The decrease in radius of unreacted core is given by
c c B c B B B
dr r r d dV dN
2 3
4
3
4
πρ π ρ ρ − = |
¹
|

\
|
= = − (Eq. 2-21)
where ρ
B
is molar density of B in the solid and V is the volume of the particle. From
Eq. 2-19 and 2-21, the rate of reaction in terms of shrinking radius of unreacted core
is given by
Ag g
c B B
ex
C bk
dt
dr
R
r
dt
dN
S
= − = −
2
1 ρ
(Eq. 2-22)
2) Diffusion through ash layer controls
Using the same idea of the gas film control, the rate of reaction is described
followings:
. 4 4 4
2 2 2
const Q r Q R Q r Q S
dt
dN
Ac c As A A ex
B
= = = = = − π π π (Eq. 2-23)
dr
dC
D Q
A
e A
= (Eq. 2-24)
A
C
C C
e
r
R
B
C d D dr
r dt
dN Ac
As Ag
c
∫ ∫
=
=
= −
0
2
4
1
π (Eq. 2-25)
Ag e
c
A
C D
R r dt
dN
π 4
1 1
=
|
|
¹
|

\
|
− − (Eq. 2-26)
57

3) Chemical reaction controls
The rate of reaction for the Eq. 2-18 is given by

Ag
B
ex
C bk
dt
dN
S
' '
1
= − (Eq. 2-27)
where k” is the first-order rate constant for the surface reaction. Replacing Eq. 2-22 in
Eq.2-28, the shrinking radius of core is given by
Ag
c
B c B
c
C bk
dt
dr
dt
dr
r
r
' ' 2
2
4
4
1
= |
¹
|

\
|
− = − ρ πρ
π
(Eq. 2-28)
Shrinking-Core Model (SCM) without ash formation: If a particle burns without ash
formation such as pure carbon combustion in air, the shrinking particle will keep
getting smaller and finally disappearing. Because there are no ash nor residue layers,
the core of the particle is surrounded by gas film illustrated in Figure 2-14. In this
case there are two control steps considered: 1) diffusion through gas film controls,
and 2) chemical reaction control. The mechanism of chemical reaction controls of
SCM without ash formation is the same as that of SCM with ash formation.
58

Figure 2-14: Shrinking-Core Model without ash formation

1) Diffusion through gas film controls: Mass transfer in a hot air steam to free-falling
solids studied by Frossling (1938) gives
333 . 0 5 . 0
Re 552 . 0 2 Sc Sh
p
+ = (Eq. 2-29)
AB
d
D
L k
Sh = (Eq. 2-30)
µ
ρ Lu
= Re (Eq. 2-31)
AB
D
Sc
ρ
µ
= (Eq. 2-32)
Study on mass transfer in order to determine the evaporation of water and
benzene drops in air was carried out by Renz and Marshall. Their experimental results
were expressed by the following equation:
Concentration of
gas-phase reactant
and product

Time

Time

R R
Shrinking particle

Gas film
C
As
=C
Ac

C
Ag

0
Radial position

CAg: Concentration of fluid A

CAs: Concentration of fluid A on the
surface of the particle

CAc: Concentration of fluid A on the
surface of the core
59

333 . 0 5 . 0
Re 6 . 0 2 Sc Sh
p
+ = (Eq. 2-33)
This equation is called the Renz-Marshall equation for mass transfer to spherical
particles.
p
g
d
k
1
~ (Eq. 3-34)
2 / 1
2 / 1
~
p
g
d
u
k (Eq. 3-35)
Eq. 3-35 expresses particles in the Stokes law regime.
2-4. Heat transfer through the bed
All processes in the thermal conversion of backed bed of MSW are governed
largely by mass and heat transfer. The MSW bed is an aggregation of particles shown
in Figure 2-15. In this section heat transfer, (radiation, convection, and conduction)
of a single particle has been described.
As a first step of modeling, heat and mass transfer of the MSW bed is
calculated using FLIC (see Figure A-1 in Appendix A). For calculation of volatile
combustion CFD modeling is used (see Figure B-2, B-3, and B-4 in Appendix B).
60

Q
rad, w-p

Q
conv, a-p

Q
conv, g-p

Q
cond, p-p
Surface
Wall
Grate
Flame
Q
rad, g-p

Figure 2-15: Combustion and transport phenomena of one particle in a MSW
bed on the traveling grate
2-4-1. Radiation among walls and particles
As shown in Figure 2-16, at high temperatures, a particle i emits a radiative
flux and absorb a flux
j rad
q
,
& from all neighboring particles j. The total radiative flux a
particle i emits is expressed as:
4
, , s j rad
j
j i p p rad
T q F q εσ α − =
∑ → −
& & (Eq. 2-36)
where α and ε are absorption and the emission coefficient, respectively. The <Ts> is
the surface temperature of a particle i that absorbs a flux
rad
q& from all neighboring
particles j and a wall j with the view factor F
i→j
.
61

2-4-2. Radiation between a gas and a particle
The radiative flux a particle i emits is expressed as:
) ( '
4
,
4
, s s g g g s g p rad
T T q α ε σ ε − =

& (Eq. 2-37)
where ε’
s
is the effective emissivity of MSW and approximated by
2
1
'
+
=
s
s
ε
ε (Eq. 2-38)
where ε
s
is surface emissivity.
s g,
α is effective gas absorptivity.
Particle i
Particle j,4
Particle j, 5
Particle j, 3
Particle j, 2
Particle j, 1
Figure 2-16: Radiation from particle i to all neighboring particles j

Wall j, 1
62

Note: For radiation within the fluid phase and from combusting volatile
matters will be calculated using CFD software. Fluent has several models for
calculating volatile combustion with radiation. The radiative transfer
equation (RTE) for an absorbing, emitting, and scattering medium is
described as:

Ω Ω |
¹
|

\
|
+ + + − =
π
φ θ
π
4
0
) , ( ) (
4
) ( d P I
K
I K I K K
dl
dI
s
b a b s a
(Eq. 2-39)
The radiation intensity I is a function of position and direction. The
first term on the right-hand side of Eq. 2-36 is absorption and out-scattering
loss. The second term shows emission of media. The third one is in-scattering.
This integral-differential equation is the third term that is in integral form and
should be reduced to differential form to simplify the solving problem.
2-4-3. Convection between a particle and the primary air though the bed
Heat transfer by convection between the primary air and particles is expressed
as:
) (
s g conv
T T h q − = & (Eq. 2-40)
where h is heat transfer coefficient. T
s
is the surface temperature of a particle and T
g
is
the temperature of the primary air.
63

2-4-4. Conduction with two particles
The conductive heat flux between two neighboring particles in contact is
estimated by
2 , 1 ,
2 , 1 ,
2 1 2 1
/ 1 / 1
1
/ 1 / 1
1
S S
S S
cond
r r
T T
k k r
T
k k
q
∆ − ∆

+
− ≈


+
− = &
(Eq. 2-41)
where k
1
and k
2
are conductivities of the two particles. The temperature gradient
between two particles is approximated by the surface temperature difference over
distance
i S
r
,
∆ from outer particle surface. The contact area is assumed to be quadratic
and determined by the contact angles γ
1
and γ
2
as shown in Figure 2-17.

R
2

R
1

γ
1

γ
2

A
1
A
2

Figure 2-17: Conduction between two neighboring particles (Peters, B. 2005)
64

2-4-5. Equation describing mass and heat balance
Governing Equation of the MSW bed: The governing equations for mass and heat are
summarized as follows. For energy (heat) balance of the inside of a single particle,
ω ρ
ρ
& + −




=


) (
1
,
2 2
2
T C v r
r
T
D r
r r t
T C
g p g
p
(Eq. 2-42)
Energy balance of the outside of a single particle is describe as:
cond rad conv
R
q q q
r
T
D & & & + + =


(Eq. 2-43)
where v is the gas velocity, ρ is the particle density ρ
g
is the gas density, C
p
and C
p,g
are the heat capacity of the particle and heat capacity of the gas, respectively. D is
heat diffusivity and <T> is temperature of the particle surface.
Mass balance of the inside of a single particle:
ci i
i
i
i
i
c v r
r
c
D r
r r t
c
ω& + −




=


) (
1
2 2
2
(Eq. 2-44)
Mass balance of the outside of a single particle:
65

) , (
,
,
, ∞
− =


− c c
r
c
D
i
R i G
R
i
i G
eff i
β (Eq. 2-45)
where v is the gas velocity, c
i
is the concentration of component i,
ci
ω& is the
conversion rate of component i.
eff i
D
,
is the effective diffusivity of component i. β is
the mass transfer coefficient.

c stands for the ambient gas concentration.
Governing Equation of Gas: The governing equations of gas for mass, momentum,
energy and species are summarized as follows:
Continuity:
g g g
g
S v
t
= ⋅ ∇ +


) (
r
ρ
ρ
(Eq. 2-46)
where
g
ρ and
g
v
r
are the gas density and velocity at a position and time t, respectively.
Sg includes mass transfer rates between the solid and the gaseous phase due to
decomposition processes inside of the particle.
Momentum:
) ( ) (
) (
g g g g g
g g
v F p v v
t
v
r r r
r
+ ⋅ ∇ = ⋅ ∇ +


ρ
ρ
(Eq. 2-47)
66

where p
g
is gas pressure. The function ) (
g
v F
r
is defined by the following equation:
g g
v
k
v F
r r µ
− = ) ( if Re <10 (Darcy) (Eq. 2-48)
g g g g g
v v C v
k
v F
r r r r
ρ
µ
− − = ) ( if Re >10 (Forchheimer) (Eq. 2-49)
where k is the permeability and K and C are constants of the form:
2
3 2
) 1 ( 150 P
P d
K
− ⋅
= (Eq. 2-50)
3
) 1 ( 75 . 1
dP
P
C
− ⋅
= (Eq. 2-51)
Species transport:
g i g i
Y Y g i g g
g i g
S Y v
t
Y
, ,
) (
) (
,
,
ω ρ
ρ
&
r
+ = ⋅ ∇ +


(Eq. 2-52)
where
g i
Y
,
are mass fraction of H
2
O, CO, CO
2
, and C
m
H
n
.
g i
Y
S
,
is the source term
including mass sources due to drying and devolatilization of particles released into
the gas phase.
g i
Y
,
ω& is the gaseous conversion rate of a species i:
67

)) exp( , min(
,
, , ,
,

=
g
k a
k k i k i m k i Y
RT
E
k Y
g i
ν τ ν ω& (Eq. 2-53)
3
1 1
v
L
C
t
m
m
m
r
ν
τ = = (Eq. 2-54)
where t
m
is mixing time scale, C
m
is an empirical constant, L is a characteristic length,
υ is the viscosity. k indicates summation of all reactions for a species i.
Energy:

+ Γ + − ⋅ ∇ − ⋅ ∇ ⋅ ∇ = ⋅ ∇ +


k
i m Y g g g g g g g g g
g i g
H v v F v T e v
t
e
g i
,
,
,
) ( ) ( ) (
) (
ω ρ λ ρ
ρ
&
r r r r
(Eq. 2-55)
where Г is the heat from walls and the particles. H
m,i
is the heat generation due to
chemical reactions. λ
g
is the thermal dispersion coefficient. e
i,g
is the specific internal
energy equal to C
v


68
Chapter 3: Experimental Work



3-1. Introduction
3-1-1. Study of particle size and shape
Methods of particle size measurement have been much studied particularly in
powder technology. These methods are applicable to measurements of MSW particles.
This section describes two kinds of particle size analysis: a) sieve analysis and b)
image analysis.
a) Sieve analysis
The American Society for Testing and Materials (ASTM) provides several
standard test methods for sieve analysis, such as C136-05: “Standard Test Method for
Sieve Analysis of Fine and Coarse Aggregates” and E-389: “Test for Particle Size or
Screen Analysis at No. 4 Sieve and Coarser for Metal Bearing Ores and Related
Figure 3-1: Particle diameter and sieve size

Sieve
d
A

dA: Sieve diameter
Particle

69
Materials.” These sieve analysis methods have been adopted for MSW (Ruf 1972) in
order to measure particle sizes and size distributions defining the sieve diameter (d
A
)
as the diameter of a particle that can pass though sieve size A (Figure 3-1).
b) Image analysis
In image analysis, the length, width, perimeter, and area of each particle are
measured from digital camera images. An advantage of image analysis is that, it
measures both size and shape of particle. It is used mainly for small particles of sands,
powders, and grains such as coals, cements, and various chemical compounds. In this
study, image analysis has been used because of the above advantage.
3-1-2. Study of mixing and flow
There have been two types of modeling of the flow of solids on a combustion
grates: small-scale modeling of the full grate or a section of the grate. Lim et al. built
1/15-scale models of several grate systems and measured the mixing process due to
the movement of the grate using small cubic particles (Lim et al. 2001). Figure 3-2
shows the model of the roller grate system and its residence time distribution in the
direction of movement of the particles. At time 0 there was a narrow distribution of
tracer particles close to the feed end of the grate. The distribution flattened out, that is,
there was mixing of tracer particles with the rest, as the material flowed downward.
One of the advantages of small scale models is that they are less costly to
build than full scale models, even though substitutional particles made of plastics,

70
ceramics, or woods must be used instead of real MSW particles. Beckmann and
Scholz used clay, wood and ceramic spheres in their scale model of the reverse acting
and forward acting grates (Beckmann and Scholz 2000).
On the other hand, section models can be used to reproduce parts of full grate
system and examine the mixing process using real MSW particles. However, no one,
except Yang et al (Yang et al 2005) has ever such as full scale tests for the study of
solid waste mixing on the grate. They measured several mixing diffusion coefficients
with different densities (see Table 5-2 in detail)
Because of the size limitation, “as-collected” MSW cannot be used in small-
scale models. On the other hand, full-scale grate models have been less applicable for
the purpose of solid waste mixing, because of difficulties in construction. However,
full-scale models would allow investigating packed beds of real “black bag” MSW
collected from local communities. This is an important consideration since “as-
collected” MSW samples on a full-scale grate can represent similar geometries and
actual combustion chamber conditions of commercial WTE plants. For these reasons,
in this study, a full-scale section model of the WTE reverse acting grate was built and
actual NYC-MSW particles were used in controlled tests of solids mixing in the
section model.

71

3-2. Measurement of particle size and Shape Factor (SF) distributions
3-2-1. Methodology
a) Sample collection by field trip in New York City and by WTE facility visit
Black bags were collected as typical MSW samples in residential areas of the
five boroughs (Bronx, Brooklyn, Manhattan, Queens and Staten Island) of New York
City. Based on the map shown in Figure 3-3, residential areas of five boroughs were
picked as random, where per capita income is the same range. After several black
Figure 3-2: Physical Scale Modeling (Lim et al. 2001).

72
Figure 3-3: New York City map with per capita income (Bowen 2004). X marks are the
locations that black bags were collected

X
X
X
X
X
bags were collected in the five boroughs, 350 particles of MSW were randomly
picked up from the black bags.

Samples of combustion residues were randomly collected at the Union County
WTE facility of Covanta Energy. Particles of combustion residues, or bottom ash
particles, were collected after the bottom ash discharger. Hence, particles are

73
influenced by this process, such as cracked inside the discharger (water cools down +
pressure of the ramp). The residues were sieved using a 1/2 inch (1.27 cm) screen to
eliminate fine powder residues and 210 particles were selected.
b) Sample measurement by image analysis
Particles of the residue samples were also measured in terms of height, width,
area and perimeter in this study. Size and surface area of each particle were measured
using the image analyzer (digital camera: Sony PCG-C1VR/BP and software: Able
Image Analyzer
TM
version 2.1) shown in Figure 3-4. There are two steps in image
analysis: 1) capturing particle images from the digital camera with color inversion
and 2) outlining the shapes of particles regarding length, width, perimeter, and area.
Able Image Analyzer
TM
(Mu Labs 2004) is one of the powerful image analysis
software packages and was used in order to measure particle sizes from captured
image. It supports image analysis functions that include dimensional, gray scale and
Image Analyzers
Digital Camera
Sample
Figure 3-4: The image analyzer system, outlining the shapes of particles regarding length, width,
perimeter, and area

Able Image Analyzer
TM
version 2.1 (Mu Labs 2004)

74
24-bits color measurements: distance, area, angle, point, line, pixel profile, histogram
etc. (from images or selections) with statistics that calculate basic statistics and
frequencies. From the image analysis of MSW and ash particles, shape factors for
MSW and ash particles were determined.
Cluster analysis was carried out using Clustan Graphics
TM
version 7.05
developed by Clustan Ltd. (Wishart 2004). This program can be used for a)
calculating Tree diagram to investigate similarities of a group compared with other
groups and b) drawing scatter plots (multidimensional scaling) to classify MSW and
residue particles by shape factors, including aspect ratio, roundness, and spheresity.


Figure 3-5: Digital camera images of MSW samples (left) and ash samples (right)

5 cm
5 cm
15 cm
15 cm

75


3-2-2. Definitions of particle size and shape
Particle Size (diameter): There are many definitions of particle size. Table 3-1 shows
various particle diameters. For example, the difference among the Feret, Martin, and
Sieve diameters is illustrated in Figure 3-6 (Allen 1997).





d
M

1) dF: Feret diameter

Figure 3-6: Various types of diameters (Allen 1997)

Particle
d
F

2) dM: Martin diameter

d
A

3) dA: Sieve diameter


76






Symbol Name (diameter) Definition Formula
d
v

Volume Diameter of a sphere having the same
volume as the particle
3
6
v
d V
π
=
d
s

Surface Diameter of a sphere having the same
surface as the particle
2
s
d S π =
d
sv

Surface-Volume Diameter of a sphere having the same
external surface to volume ratio as the
particle
) / (
2 3
s v sv
d d d =
d
d

Drag Diameter of a sphere having the same
resistance to motion as the particle in a
fluid of the same viscosity and the same
velocity and at the same velocity (d
d

approaches d
s
when Re is small)
v d F
d d
η π 3 =
d
f

Free-falling Diameter of sphere having the same free –
falling speed as a particle of the same
density and viscosity

d
St

Stokes Free-falling diameter in the laminar flow
region
) / (
2 3
s v St
d d d =
d
a

Projected area Diameter of a circle having the same
projected area as the particle in stable
orientation

d
p

Projected area Diameter of a circle having the same
projected area as the particle in random
orientation (mean value of d
p
= d
s
for
convex particles

d
c

Perimeter Diameter of a circle having the same
perimeter as the projected outline of the
particle
c
d P π =
d
A

Sieve Width of the minimum square aperture
through which the particle will pass

d
F

Feret The distance between pairs of parallel
tangents to the projected outline of the
particle in some fixed direction

d
M

Martin Chord length, parallel to some fixed
direction, which divides the particle
projected outline into two equal areas

d
R

Unrolled Chord length through the centroid of the
particle outline




Table 3-1: Various definitions of particle diameters (Allen 1997)


77


Shape factors: The shape of particulate matters is mainly described by shape factors,
aspect ratio (AR), roundness (circularity), and sphericity. Aspect Ratio (AR) is
described as the following expression (Schneiderhohn 1954):

˓sp˥cˮ ˞oˮ˩o =
Icngth
wìdth
(Eq. 3-1)
Roundness (Circularity shape factor) is defined as the following equation (Cox 1927):
˞o˯nˤn˥ss =
(pc¡ìmctc¡)
2
4n·(su¡]ucc u¡cu)
(Eq. 3-2)
Sphericity is defined as the following equation (Wadell 1932):
Spheiicity u =
(surIacc arca oI a sphcrc oI thc samc voIumc as thc partIcIc)
(actuaI surIacc arca oI thc partIcIc)

(Eq. 3-3)


78
3-3. Physical Modeling
3-3-1. Methodology
a) Geometry of a physical model
The full-scale physical section model of the reverse acting grate used in this
study had the dimensions of 121 cm (4 ft) in length, 91 cm (3 ft) in height, and 61 cm
(2 ft) in width and the angle of bed declination to the horizontal was 15 degrees.
Figure 3-7 shows the geometry of the full-scale section model. The MSW bed was
divided into 16 cells (4 layers × 4 sections), each having a size of 20 × 20 cm. The
four bars (2 fixed and 2 reciprocating) have a height of 13 cm and an angle of 15
degrees to the bed (also 45, 30, and 0 degrees can be selected. Martin grate has an
angle of 26. Because of weight limitation, 14 degree was chosen in this test). The
reciprocating bars traveled 42 cm from the top to the bottom positions, so that the
horizontal distance (ŸŶ cos ŵŸ
o
= ŸŴ.Żc˭) was approximately the same as the length
of two cells (40cm). Two digital cameras (model: Sony PCG-C1VR/BP and Sabrent
SBT-WCCK), positioned above and on the side of the apparatus, were used to
monitor the particle movement.

Note: An ideal full-scale physical model for examining only the effect of grate motions should
have an infinite length or, at least, the same length of real combustion chambers. A section
model always encounters the limitation of the particle behavior. To minimize the effect of the
boundary of the box, the first and last sections located close to the wall of the box were not
fed in the stochastic model.



Figure 3-7: Full-scale physical section model of a combustion chamber: positions of CCD cameras
(top) and geometries of reverse acting grate (bott


scale physical section model of a combustion chamber: positions of CCD cameras
(top) and geometries of reverse acting grate (bottom)
79

scale physical section model of a combustion chamber: positions of CCD cameras


b) Making tracers in different size
Particle tracers were prepared
small (6 cm), medium (14 cm), and large (22 cm) sizes were made
insulating foam. These tracers were based on the
MSW particles: the mean value µ of the size distribution was found to be 14 cm,
where µ-σ ≈ 5.8 cm, and µ+
This value was more representative of the precompacted MSW an
typical value of about 297 kg/m
after compression during the collection process.






cers in different size
article tracers were prepared as shown in Figure 3-8. Spherical tracers of
small (6 cm), medium (14 cm), and large (22 cm) sizes were made using
. These tracers were based on the particle size distribution of NYC
MSW particles: the mean value µ of the size distribution was found to be 14 cm,
≈ 5.8 cm, and µ+σ ≈ 22.8 cm. The density of the tracers was
This value was more representative of the precompacted MSW and is lower than the
typical value of about 297 kg/m
3
(500 lb/yd
3
) (Tchobanoglous 1993), as measured
after compression during the collection process.
80
Spherical tracers of
using gap filling
particle size distribution of NYC-
MSW particles: the mean value µ of the size distribution was found to be 14 cm,
ity of the tracers was 221 kg/m
3
.
is lower than the
) (Tchobanoglous 1993), as measured

81

Figure 3-8: Spherical tracers of different sizes (small 6cm, medium 14cm, large 22cm) (left) and
NYC-MSW particle size distributions
Side view

Height:
20 cm

Height:
40 cm

Height:
60 cm

Height:
80 cm

Figure 3-9: Side views and top views of NYC-MSW beds with 20, 40, 60, and 80 cm
heights and angles of chamber bed decline: 15 degree

Top view


82
MSW bed height
20 cm
MSW bed height
40 cm
MSW bed height
60 cm
MSW bed height
80+ cm
Figure 3-10: Cell patterns of 20, 40, 60, 80 cm beds

The bed height was measured at the lowest
position of reciprocating bars

83
c) Loading NYC-MSW particles
The MSW particles used in the tests were obtained from black bags collected
in NYC, as discussed in Section 3-2-1 a) (Figure 3-3). They were loaded into the
physical model, along with tracers. Figure 3-9 shows pictures of side and top views
that a NYC-MSW bed that was loaded in the full scale section model. The bed
heights were 20, 40, 60, and 80 cm and the angle of camber grate declination was 15
degrees to the horizontal, as mentioned earlier. Figure 3-10 shows cell pattern of 20,
40, 60, 80 cm beds.
d) Operation conditions
Since this physical model was not motorized, the operation of reciprocating
bars was conducted manually. This apparatus enabled the determination of the
probabilities of particle movement in each vertical and horizontal position of the
packed bed of actual NYC-MSW. After each reciprocation of the moving bars, the
tracer positions were recorded by the cameras. Therefore, particles were traced from
the original position to the current position after n reciprocations. Because this
physical model has neither an inlet nor a feeder, there is no MSW flow pushed into
the chamber by means of a feeder. Therefore, a purely mixing phenomena
characterization, caused by the reciprocating bars, was experimentally examined. In
this “no-feeding” physical model, the motion of the reciprocating bars and gravity
were the driving forces that caused the particles to move within the MSW packed bed.

Chapter Four
84
Chapter 4: Mathematical Work

4-1. Mass and volumetric flow of solids and air on the grate
4-1-1. MSW flow in the chamber
Unstable combustion of a MSW packed bed flows in a mass-burn chamber is
one of the major problems caused by the variability of municipal solid waste (MSW).
One reason for the instability of combustion is the extreme variability of the
properties of the feedstock. MSW is more complex and non-homogeneous than any
other fuel. The physical and chemical properties of MSW directly affect its
combustion. Another reason for unstable combustion of MSW is transient phenomena
such as the break-up of waste particles and the channeling of combustion air. The
resulting unstable combustion of MSW introduces operating difficulties such as
fluctuating chamber temperature, low thermal efficiency of the combustion chamber
and formation of undesirable compounds (CO, NO
x
, and dioxin) in the process gas of
waste-to-energy facilities.
Some advanced automatic combustion control systems for mass-burn plants
have been developed worldwide to stabilize the operating conditions during
combustion. For example, Infrared Radiation (IR) sensors are provided in the
advanced control systems for image processing and detection of burn out lines. In this
system, an auto-regressive model with periodic functions (fuzzy logic) is used in
chaos analysis for continuous variation of MSW in the control system. It provides

Chapter Four
85
good control performance for reducing concentration of CO, NO
x
and dioxins in the
combustion gases (Takatsudo et al. 1999).
Although these new tools have resulted in improved control and operating
performance, the continuous variation of the MSW feed into the combustion chamber
still makes control of the combustion process difficult. The existing combustion
models for mass-burn MSW chambers have used mean values for physical and
chemical properties of MSW components, such as proximate analysis and Lower
Heating Value (LHV). In this study, a mathematical model for time series analysis of
continuous variation of MSW was developed using the Monte Carlo method to
simulate stochastic combustion. Additionally, the percolation theory is applied in
order to simulate transient phenomena with the bed.
4-1-2. Monte Carlo Modeling for MSW feeding and flow
This study presents a time series model for continuous variation of MSW
using the Monte Carlo Method. Table 4-1 shows percentages of MSW components
and the approximate analysis of various types of combustible dry waste materials in
New York City. They are used in this model as an initial distribution. The probability
distribution of components of MSW for a uniform distribution U
k
~U (0, 1) is
expressed by
X
k
~
|
¹
|

\
|
046 . 0 048 . 0 05 . 0 022 . 0 002 . 0 047 . 0 089 . 0 047 . 0 266 . 0
& Other Metals Glass Wood Leather Rubber Textiles Plastics Cardboard Paper
.

Chapter Four
86
(Eq. 4-1)



In order to carry out the computation for the components of MSW, the
MATLAB 6 program was used to generate 100 random numbers (k = 100) for each of
100 samples (n = 100).
The FLIC two-dimensional bed program (Yang et al. 2002) developed by
Sheffield University was used, in conjunction with our model for continuous variation
of MSW, in order to simulate the combustion process. The FLIC program is used to
calculate the fluid flow, heat transfer and combustion reactions in both gaseous and
solid phases.
Component of
waste stream
% in NYC Weight (metric
tons day)
Carbon
(% by
Weight)
Oxygen
(% by
Weight)
Nitrogen
(% by
Weight)
Sulfur
(% by
Weight)
Ash
(% by
Weight)
Paper 26.6 3144 43.5 6 44 0.3 0.2
Cardboard 4.7 555 44 5.9 44.6 0.3 0.2
Plastics 8.9 1052 60 7.2 22.8 - -
Textiles 4.7 555 55 6.6 31.2 4.6 0.2
Rubber & Leather 0.2 24 69 9 5.8 6 0.2
Wood 2.2 260 49.5 6 42.7 0.2 0.1
Glass 5 591 0.5 0.1 0.4 <0.1 -
Metals 4.8 567 4.5 0.6 4.3 <0.1 -
Other 4.6 544 26.3 3 2 0.5 0.2
Tons per day 7292 2865 372 2194 42 10

Table 4-1 Ultimate analysis of dry stream municipal solid wastes in New York City (Themelis
et.al 2002, Tchobanoglous et al. 1993)

Chapter Four
87
Also a model for transient phenomena was developed in this study, based on
the percolation theory. The first percolation model, by Broadbent and Hammersley
(Broadbent and Hammersley 1957), was primarily concerned with the existence of
the ‘open path’ in a system, in which a large porous stone is immersed in a bucket of
water. It has been claimed that percolation theory is a cornerstone of the theory of
disordered media, and provides a reasonable model for a disordered medium such as
MSW. Although there are two types of percolations (bond percolation and site
percolation), we considered site percolation for transient phenomena in combustion in
Figure 4-1. Two-dimensional square lattice site percolation model used in the solid waste on the
combustion grate
P=0 P=0.2 P=0.4
P=0.6
P=0.8 P=1.0

Chapter Four
88
this study. There are a number of network models such as Honeycomb, Square,
Kagome, Triangular, Diamond, Simple Cubic, BCC and FCC. A network of MSW on
the traveling bed of combustion chamber is specified in the model as a two-
dimensional square lattice (Figure 4-1) in order to simplify the combustion process.
The lattice size L was defined as 40 (40 x 40) for the same reason. We designate each
vertex of the lattice ℤ
2
open with probability p and closed otherwise. p
c
is the critical
probability and is defined by
p
c
= sup {p : Ө(p) = 0} (Eq.4-2)


The current estimate of this critical probability p
c
for site percolation in the
two-dimensional square lattice is 0.5927 (Grimmett 1991). We assume that
(a) Channeling (b) Break-up (Subsidence)

Figure 4-2: Channeling and break-up of MSW bed in the percolation model on the grate

Chapter Four
89
combustion in the lattice proceeds uniformly. The program code for this model was
developed using Visual Basic. Figure 4-2 shows the channeling and break-up of
MSW on the combustion grate. Channeling usually occurs when particles of solid
waste combust and create an empty space vertically through the bed as shown in
Figure 4-2 (a). This means at least one cluster reaches both vertical edges of the
lattice. Break-up, on the other hand, usually occurs when particles in the middle of
the solid waste combust and create a space horizontally as shown in Figure 4-2 (b).
This means that at least one cluster reaches both horizontal edges of the lattice in the
percolation model.


Figure 4-3: Percolation Model in the combustion chamber

Chapter Four
90
In our model, we specified the use of the Martin reverse acting grate (Figure
4-3), one of the most popular WTE grates used in the United States. The MSW fed on
this grate travels approximately 7.0 m in an hour. We divided the grate into 8 zones,
through which MSW runs continuously. Mixing coefficients were assumed to be zero
in order to simplify motion of the solid waste. The simulations were performed on a
Linux computer (CPU: 1.8MHz, Memory: 256MB) with VMware for Windows 98
environment. It takes approximately two hours to complete one set of the calculations
for this model.

4-2. Stochastic model for MSW particles mixing on the traveling grate
Simulation of the physical and chemical processes in moving beds is used
widely to study combustion and other chemical reaction phenomena in gas-solid
systems. Three general types of mathematical models are used for investigating the
transport and chemical reaction phenomena in mass burn combustion chambers:
Computational Fluid Dynamics (CFD), bed models, and stochastic models. CFD
models simulate the fluid flow, heat and mass transfer, and reaction phenomena in the
combustion chamber above the traveling grate by solving numerically the continuity
and energy conservation equations and the Navier-Stokes equations (conservation of
momentum). Numerical bed models of solid waste combustion have been developed
since the early 1970s (Essenhigh, R. H. and Kuo, T. J. 1970). Yang, et al. (Yang. et al.
2002) developed a two-dimensional program, called the Fluid Dynamic Incinerator

Chapter Four
91
Code (FLIC). FLIC is graphically interactive and is widely used by WTE engineers to
simulate the physical and chemical transformations involved in the drying,
volatilization and combustion processes on the grate; however, the combustion
process is modeled using typical or average data such as composition, particle size,
density and heating value, even though MSW is a very non-homogeneous fuel.
Stochastic models, such as the one described in this paper are relatively new,
although some researchers have suggested combining mixing models of the traveling
bed with experimental work (Lim et al. 2001). Markov chain models have been used
in the past to estimate mixing of powders in hoppers (Aoun-Habbache et al. 2002)
and mixing in fluidized bed reactors. The present study is the first to apply the
Markov chain model to the grate of a mass-burn chamber model in order to determine
how solid particles move and mix on a moving grate.
4-2-1. One-dimensional Stochastic model for reverse acting grate
A model, based on the Markov chain model (Bremaud 1991), was developed
to simulate particle flow and degree of mixing as the MSW particles travel over the
grate. The first procedural step in the design of the Markov chain model was to divide
the solid waste on the bed into several cells (“state” of the system). Then, the
transition of particles between these cells with time is examined. In the model, the
entire grate was represented by 16 cells and by their respective boundary conditions
(Figure 4-4). The computed data were compiled in the form of a transition matrix P

Chapter Four
92
and a state vector S(n) that describes the particle distribution over the states after n
transitions.

Figure 4-4: Illustration of the spatial disposition of cells on the reverse acting grate: Each cell
represents either the reciprocating bar or the fixed bar.

The interchange of particles between successive cells (“Transition graph”) is
illustrated in Figure 4-5.

Chapter Four
93
Figure 4-5: Interchange of particles between successive cells (transition graph): The final cell of
this model is cell 17(ash bin); all particles in the MSW feed after combustion eventually arrive at
cell 17, although the paths of individual particles differ considerably.



The Markov chain model assumes that, at each motion of the reciprocating
bars (transition), the transition probabilities of the waste particles between adjacent
cells do not depend on the previous state in time. The rule governing the particle
migration (travel) of the system is expressed by the following equation:
S(n) = S(0) P
n
(Eq. 4-3)
where S(n) represents the profile of MSW traveling on the chamber bed after n times
of double strokes of the reciprocating bars and S(0) is the initial profile of MSW feed
at the inlet (prior to any stroke of the reciprocating bars, i.e., n=0); P
n
is the nth
power of P (called the n-step transition matrix) of the matrix that contains the
probabilities of the solid particle movement. The latter is controlled by feed rate,

Chapter Four
94
particle size and density, geometry of the grate and frequency of reciprocating bars.
The transition matrix P for the movement of the solid waste particles corresponding
to the series of cells shown in Figure 4-5 is defined as follows:









P =









where p is the probability that MSW particles will remain at the same grate position
(cell) and r the probability that the particles transit the next grate position (cell). The
subscripts f and r in the following expressions represent fixed bars and reciprocating
bars, respectively; the subscripts i and o represent inlet and outlet, respectively.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
¹
|

\
|
− −
− −
− −
− −
− −

1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
1 0 0 0 0 0 0 0 0 0 0 0 0 0 0
0 1 0 0 0 0 0 0 0 0 0 0 0 0 0
0 0 .. .. .. 0 0 0 0 0 0 0 0 0 0 0 0
0 0 0 .. .. .. 0 0 0 0 0 0 0 0 0 0 0
0 0 0 0 .. .. .. 0 0 0 0 0 0 0 0 0 0
0 0 0 0 0 .. .. .. 0 0 0 0 0 0 0 0 0
0 0 0 0 0 0 .. .. .. 0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 .. .. .. 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0 .. .. .. 0 0 0 0 0 0
0 0 0 0 0 0 0 0 0 .. .. .. 0 0 0 0 0
0 0 0 0 0 0 0 0 0 0 .. .. .. 0 0 0 0
0 0 0 0 0 0 0 0 0 0 0 .. .. .. 0 0 0
0 0 0 0 0 0 0 0 0 0 0 0 1 0 0
0 0 0 0 0 0 0 0 0 0 0 0 0 1 0
0 0 0 0 0 0 0 0 0 0 0 0 0 0 1
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1
r o o r
f f f f
r r r r
f f f f
r r r r
i i
r p p r
r p p r
r p p r
r p p r
r p p r
p p
(Eq. 4-4)


95
The movement of the reciprocating bars causes an upward motion of the solid
waste particles against the downward direction of the feed. Thus, there is a transition
from position/cell k (k is the cell number and is a positive integer, 1 < k < 16) to the
two adjacent compartments k - 1 and k + 1 (Figure 4-5). If k is an even number (i.e..
cells #2, 4, 6, 8, 10,…, 14), cell k represents a reciprocating bar. If k is an odd number
(i.e., cells #3, 5, 7, 9, …, 15), cell k represents a fixed bar. In order to simulate the
mixing phenomena on the actual grate, six probabilities are defined at each cell:
- The probability p
i
of particle remaining in state 1 (cell #1).
- The probability p
r
of particle remaining on one of the reciprocating bars (cell #2,
4, 6, ….14).
- The probability r
r
of transiting to a reciprocating bar (cell #2, 4, 6, ….,14).
- The probability p
f
of remaining on a fixed bar (cell #3, 5, ….,15).
- The probability r
f
of transiting to a fixed bar (cell #3, 5, ….,15).
- The probability p
o
of remaining in outlet state (cell #16)
It should be noted that the probabilities p and r are always positive fractions
and the sum of the elements within each row in the matrix P equals one. The values
of p and r depend on the feed rate R
f
,, particle size of MSW components S, particle
density of MSW components S, state of combustion, frequency of reciprocating bars
R
r
, the length of the reciprocating bars travel L, angle α of chamber bed decline, the


96
height H and the angle θ of the reciprocating bars. The relationships between
probabilities and operating parameters can be expressed as follows:
- For the reciprocating bars (cells #2, 4, 6,….,14),
p
r
~ f
r
(R
f
, S, ρ, R
r
, L, α, H, θ) (Eq. 4-5)
r
r
~ g
r
(R
f
, S, ρ, R
r
, L, α, H, θ) (Eq. 4-6)
p
r
+ r
r
< 1 (Eq. 4-7)
and for the fixed bars (cells #3, 5, ….,15),
p
f
~ f
f
(R
f
, S, ρ, R
r
, L, α, H, θ) (Eq. 4-8)
r
f
~ g
f
(R
f
, S, ρ, R
r
, L, α, H, θ) (Eq. 4-9)
p
f
+ r
f
< 1 (Eq. 4-10)
The relationships between probabilities are:
- For cell #1, solid waste particles are disposed to move only downward to cell #2,
because cell #1 is next to the inlet and new feed enters all the time pushing the
older particles downward the grate. Therefore,
p
i
< 1- p
i
(Eq. 4-11)
- For cell #2 (also, cell #4, 6,…., 14), although the reciprocating bar moves up and
down, particles that ride this bar have less chance for transition from cell #3 to


97
cell #2 and are disposed to remain in cell #3 (same as cells #5, 7, 9,….,15).
Therefore,
r
r
< 1- p
r
- r
r
< p
r
(Eq. 4-12)

- For cell #3 that represents a fixed bar (same as cells #5, 7, 9,…., 15) particles
have a high probability to move to the higher cell #2 (same as cells #4, 6,
8,….,14) because the motion of the reciprocating bar below cell #4 pushes
particles upward. Therefore,
r
f ≤
p
f
< 1- p
f
- r
f
(Eq. 4-13)
The assumed probabilities for satisfying Eqs (4-11), (4-12) and (4-13) are
shown in Table 4-2. At this stage of the study it was not possible to define the
functions f
r
, f
f
, g
r
and g
f
in Eqs. (4-5), (4-6), (4-8), and (4-9), because there has not
been any experimental study of the mixing phenomena. Further work will be required
to estimate values of these parameters, define Eqs (3-21) to (3-26), and then validate
the model projections by field tests in an industrial WTE. Since all the solid wastes
enter at the inlet of the combustion chamber, the initial state vector S(0) can be
formulated as:
PROBABILITY
ASSUMED
VALUE
PROBABILITY
ASSUMED
VALUE
PROBABILITY
ASSUMED
VALUE
p
i
0.4 p
r
0.5 p
f
0.3
p
o
0.3 r
r
0.2 r
f
0.3
Table4- 2: Assumed values of probabilities satisfying Eqs (4-10), (4-11) and (4-12) for the transition
matrix used in this calculation


98
S(0) =| | 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1
(Eq. 4-14)
This initial state vector represents the state of solid waste in cell #1, the
position next to the inlet just after the solid waste is fed on the grate by the pusher at
the bottom of the hopper. As stated earlier, the present model examined only one
layer of MSW, that in contact with the grate surfaces.
MATLAB 6, an interactive matrix-based program for engineering calculations,
was used on a Windows 2000 PC to carry out the matrix calculations required in the
Markov chain simulation.
4-2-2. Expansion to two-dimensional stochastic model
To develop the two-
dimensional Markov chain
model of the mass-burn
combustion chamber, we
defined 5 layers of the MSW
fuel above the grate and 32
zones for representing the
reciprocating bars and fixed bars of the reverse acting grate. Figure 4-6 shows the
total of 160 cells that were defined. The cells express the position where the MSW
particles might be located. The average height of the MSW fuel on the grate is
assumed as 1 meter. Therefore, the assumed height of each layer is 0.2 meter. The
Figure 4-6: Illustration of the spatial disposition of cells
on the reverse acting grate
5 layers
32 zones
0.20 m

0.20 m

H=1.0m

Lg = 6.4m

Divided into 32 zones (Lg /32) Total = 32x5 = 160 cells (mesh)


99
MSW Particle
Reciprocating Bar Fixed Bar
Figure 4-7: Particle movement by reciprocating and fixed bars
rule governing the particle migration on the grate system is expressed by the same
equation as Eq. 4-3:
S(n) = S(0) P
n
(Eq. 4-3)
where S(n) represents the profile of MSW particles traveling on the grate after n times
of double strokes of the reciprocating bars; S(0) is the initial profile of MSW feed at
the inlet (prior to any stroke of the reciprocating bars, i.e., n = 0); P
n
is the nth power
of the matrix P (called the n-step transition matrix) that contains the probabilities of
the solid particle movement; P and S(0) are represented by a 162x162 matrix and a
162x1 matrix, respectively:

P
(
(
(
(
¸
(

¸

=
162 , 162 , 162
162 , 11
... ...
... ... ... ...
... ... ... ...
... ...
p p
p p
k
j
(Eq. 4-15)


162
162
162


100
S(0) | | 0 .......... .......... 0 1 = (Eq. 4-16)
where p
j,k
are the probabilities to be measured in future field tests on a Martin grate
WTE.. In this section we used assumed values for the probabilities. The rules for
determining these probabilities are the same as those for the one-dimension Markov
model. This simulation was carried out using the MATLAB 6 program on a
Windows 2000 PC. To simplify the complexity of the movement of MSW particles,
we simulated the movements without the effect of MSW properties and continuous
feeding, i.e., the driving forces in this system are gravity and the movement of the
bars.
4-3. Size (mass and volume) reduction model
4-3-1. Introduction
The size and volume of municipal solid waste (MSW) particles is reduced
during the combustion process in the waste-to-energy (WTE) combustion chamber.
As they travel over the length of the grate, particles are subjected to drying,
volatilization, and char oxidation and finally turn into ash (Figure 4-8). The Martin
reverse-acting traveling grate depicted in Figure 1 controls the rate of flow and also
enhances mixing. The grate operation is necessary because MSW is much more
heterogeneous than coal and other fuels.
Major factors in the size reduction of MSW particles are the physical and
chemical transformations occurring on the grate, including drying, volatilization, and
combustion. In addition, because of the motion of the moving bars of the Martin grate,
some MSW particles break up and their size is reduced. This breakage process can be


101
numerically expressed as a breakage matrix based on experimental data. For example,
Campbell and Webb analyzed roller milling performance and developed a breakage
equation (Campbell and Webb, 2001). Before the MSW enters into the inlet of the
combustion chamber, the MSW particles are accumulated in the feed hopper from the
bottom of which they are pushed into the combustion chamber by the ram feeder. The
resulting compaction process can be estimated by means of a ram pressure-density
curve, because MSW varies greatly in both densities and compressibility. After the
drying, volatilization, and combustion processes in the chamber, MSW particles
become ash consisting mostly of non-combustible residues. The particles that are on
the surface on the bed, where the temperature rises above 1100
o
C, are subjected to
some fusion and agglomeration, and they form larger particles (clinker). Due to the
motion of the traveling grate, clinker particles of ash can break again and their size is
finally reduced to the PSD of ash at the outlet of the combustion chamber (Figure 4-
8).


102
In order to simulate the size reduction of MSW particles, we modeled the PSD
of MSW particles and ash particles and applied the shrinking core model (SCM) to
these distributions. Also, Image Analysis was used to determine the PSD of MSW
and ash samples. Finally, thermogravimetric analysis (TGA) data was applied on
particles of the main constituents of MSW in order to examine the effect of thermal
decomposition on particle size.

2. Devolatilization
1. Drying (moisture evaporation)
3. Combustion (char oxidization)
MSW particle
Ash particle
Figure 4-8: Combustion and transport phenomena of one particle in a MSW bed on the traveling grate
Ash particle size distribution
MSW particle size distribution


103
Gamma Distribution
0
0.04
0.08
0.12
0.16
0.2
0 5 10 15 20 25 30
D
f
p
(
D
)
α= 1 β= 5
α= 2
α= 3
α= 4
α= 5
Figure 4-9: Particle size density function (Gamma
distribution) for different values of α (=1, 2, 3, 4, 5)
and constant β (= 5)
4-3-2. Modeling
Particle Size Distributions (PSDs):
PSDs of MSW and ash were described in this study by using the Gamma
function distribution that is expressed as follows:
β α
α
α β
D
p
e D a D f


Γ
⋅ =
1
) (
1
) (
(Eq. 4-17)

where a is constant, D is the particle size, and α, β are positive parameters,
respectively. Figure 4-9 shows the gamma distributions for different values of α but
constant β.
Combustion of MSW particles:
The shrinking core model
(SCM) is used to simulate the
thermal decomposition of MSW
particles. The SCM was originally
developed by Yagi and Kunii for
describing solid and fluid reactions
for non-porous particles (Yagi and Kunii 1955). Gbor and Jia developed a SCM
combined with PSDs (Gbor and Jia, 2004) The unreacted mass fraction in the core of
a particle can be defined as follows:
Unreacted Fraction ( ) ( )


⋅ =
0
) ( ) , ( dD D f t D g
p
(Eq. 4-18)


104
where g(D, t) = 1 - X
D
is the unreacted fraction and X
D
depends on the type of the
controlling regime, such as diffusion through gas film, diffusion through ash layer, or
chemical reaction control.
g(D, t) = 0 for 0 < D < D
t

and g(D, t) = 1 - X
D
for D
t
< D < D
max


Therefore, the overall conversion for all particle sizes (X) is



Γ
⋅ ⋅ − =
max
1
) (
1
) , ( 1
D
D
D
t
dD e D a t D g X
β α
α
α β
(Eq. 4-19)
A large part of the size reduction of MSW is due to thermal decomposition that
occurs at temperatures below 400
o
C and can be expressed as follows:
MSW → moisture + volatile matters + char (Eq. 4-20)
For this reaction,
3
1 1 ) , ( |
¹
|

\
|
− = − =
D
t k
X t D g
rn
D
(Eq. 4-21)
where k
rm
=k
r
D, D
t
= k
rn
t, and k
r
= k
1
C
Ag
/ ρR. R is radius of particles (R = D/2).
Other possible size reduction phenomena that are not included in this model:
(1) Compaction of MSW particles: The size of MSW particles can be reduced due to
compaction. When MSW particles are fed into the chamber, the feeder compresses
them and they become smaller than the particles in black bags as collected.


(2) Breakage of MSW particles:
some of them break due to movement of
interaction with neighbor
chamber may break into hundreds of small pieces.

4-4. Stochastic model for
The first procedural step in the design of the stochastic model was to divide
the MSW bed into several cells (
a commercial WTE plant, a reverse acting grate, whose total length is 6.4m, has 8
moving bars and 8 fixed bars and the
bed into 32 sections along the axial direction and 5 layers along the horizontal
direction, resulting in a total of 162 cells including the inlet and ashbin cells. The
particle movement probabilities were determined from experimental data
formulate a transition matrix


Figure 4-10: MSW bed on the reverse acting grate of a
stochastic simulation
akage of MSW particles: As the particles travel on the reverse acting grate,
break due to movement of the traveling grate and mechanical
interaction with neighboring particles. For example, a glass bottle fed into the
chamber may break into hundreds of small pieces.
model for MSW mixing combined with a flow
The first procedural step in the design of the stochastic model was to divide
the MSW bed into several cells (mesh grid shown in Figure 4-10, same as
a commercial WTE plant, a reverse acting grate, whose total length is 6.4m, has 8
moving bars and 8 fixed bars and the initial MSW bed height is 1m. We divided the
bed into 32 sections along the axial direction and 5 layers along the horizontal
tion, resulting in a total of 162 cells including the inlet and ashbin cells. The
particle movement probabilities were determined from experimental data
a transition matrix P.
: MSW bed on the reverse acting grate of a WTE plant showing divided cells (mesh) for the
stochastic simulation
105
the particles travel on the reverse acting grate,
traveling grate and mechanical
a glass bottle fed into the
The first procedural step in the design of the stochastic model was to divide
, same as in 4-6). In
a commercial WTE plant, a reverse acting grate, whose total length is 6.4m, has 8
MSW bed height is 1m. We divided the
bed into 32 sections along the axial direction and 5 layers along the horizontal
tion, resulting in a total of 162 cells including the inlet and ashbin cells. The
particle movement probabilities were determined from experimental data and used to

WTE plant showing divided cells (mesh) for the


106
The stochastic process employed in the mathematical model is called a
Markov Chain and assumes that, at each reciprocation of the moving bars, the
transition probabilities of the waste particles between adjacent cells are independent
of the previous state in time. The rule governing the particle migration of the system
is expressed by the following equation:

S(n) = ( F
k
· P )
n
· S(0) (Eq. 4-22)

where S(n) represents the profile of MSW traveling on the chamber bed after n times
of reciprocations of the moving bars and S(0) is the initial profile of MSW feed at the
inlet (prior to any reciprocation of the moving bars, i.e., n = 0). F
k
is the k th power
of matrix F that controls the MSW flow in the packed bed pushed by the feeder at the
combustion chamber inlet. k is the ratio between MSW feed flow rate and the
frequency of the reciprocating bars. If there is no inlet feeder flow, then k = 0 and F
becomes the identity matrix that neither affects P or S(0). The dimension of the flow
matrix F is 162-by-162. The elements of F that we used are shown in Figure 4-11 and
their corresponding cells and directions are shown in Figure 4-12. There are several
different types of feeding systems, such as feeding bars and screws that are currently
employed in mass-burn combustion chambers. In order to simplify a MSW flow
caused by a feeder, we considered it as a plug flow. In this type of flow system, since
particles travel only from section i to the next section i+1 in a given layer, all
transition probabilities p
i,i+1
= 1, except for the exit location probabilities p
33,34
, p
65,66
,
p
97,98
, and p
129,130
(= 0). This is because particles at the exit position in each layer (the


107
32nd section) cannot transition back to the inlet position for these 4 cells. All other p
i,j

are equal to zero. Since the MSW flow greatly depends on the type of feeding system
and chamber geometry, further study is needed to determine the elements of flow
matrix F, specific to the particular operational conditions.




Figure 4





















Figure 4-11: The elements of the flow matrix F
108


Figure 4-12: Diagram of probabilities and directions of the flow for each cell of the MSW bed on the grate

P is the transition matrix that contains the probabilities predicting the solid particle
movement due to the motion of the reciprocation bars. As shown in Figur
transition matrix is 162-by-162 including a total of 26,244 probabilities,
matrix F. The main diagonal ( p
that the MSW particles remain in the same cell.
probability that the particle transits from cell
and p
i,i-1
is the probability from cell
probability that the particle moves to the cell in the layer directly below the current cell
location(from cell i to cell i+32
neighboring layer above (from cell
transition matrix P are always positive fractions and the sum of the elements within each row of
the matrix P equals one.

Diagram of probabilities and directions of the flow for each cell of the MSW bed on the grate
is the transition matrix that contains the probabilities predicting the solid particle
movement due to the motion of the reciprocation bars. As shown in Figure 4-13
162 including a total of 26,244 probabilities, as the same size of flow
p
1,1,
p
2,2 ,…..,
p
i,i, …,
p
162,162
) elements represent the probabilities
main in the same cell. As also shown in Figure 4
probability that the particle transits from cell i to the neighbor cell i+1 (along the flow direction)
is the probability from cell i to cell i-1 (opposite to the flow direction).
probability that the particle moves to the cell in the layer directly below the current cell
i+32) and p
i,i-32
is the probability that it moves to the cell in the
yer above (from cell i to cell i-32). It should be noted that the probabilities in the
are always positive fractions and the sum of the elements within each row of
109

Diagram of probabilities and directions of the flow for each cell of the MSW bed on the grate
is the transition matrix that contains the probabilities predicting the solid particle
13, the size of the
same size of flow
) elements represent the probabilities
4-12, p
i,i+1
is the
(along the flow direction)
(opposite to the flow direction). p
i,i+32
is the
probability that the particle moves to the cell in the layer directly below the current cell
is the probability that it moves to the cell in the
It should be noted that the probabilities in the
are always positive fractions and the sum of the elements within each row of



Figure 4-13





















13: The elements of the transition matrix P
110


111
The transition probabilities were determined using measured data from the full-scale
experimental cross-section model. The grate system of an actual combustion chamber has 8
moving bars as shown in Figure 4-10, whereas the physical section model employs 2 moving
bars as shown in Figure 3-7. The predictive capability of the stochastic model was expanded by
employing 8 moving bars to match a full-length grate system. Based on the fact that the
residence time of a real combustion chamber roughly ranges from 30 to 120 min (sometimes a
particle stays shorter or longer than this range), a typical value for the MSW feed in this model
was set at a time of 64 min for traveling the total chamber length of 6.4m (32cells): i.e. 2min/cell.
Since all of the solid waste enters at the inlet of the combustion chamber, the initial state
vector S(0) can be expressed as:

S(0) ={
ŵ Ŵ Ŵ Ŵ
{ (Eq. 4-16)

This initial state vector, whose size is 1×162, represents the state of the solid waste in the
inlet cell, whose position is at the bottom of the hopper adjacent to the feeder bar.

In order to carry out the matrix calculations required in this stochastic simulation,
MATLAB 7.1 was used on a Windows XP PC.

112

Chapter Five
Chapter 5: Results and Discussion

5-1. Results from experimental work
5-1-1. Particle size and shape distributions
Figure 5-1 shows the particle size distributions (PSDs), sphericity
distributions, aspect ratio distributions (ARDs) and roundness distributions of MSW
and ash particles. As shown in Figure 5-1-a, the peak of MSW PSD, about 10 cm,
shifts to the peak of the ash PSD, about 2 cm, during the combustion process. Also,
the tail of the MSW PSD is larger than that of the ash PSD. In Sphericity
Distributions (Fig 5-1-b), two peaks of MSW and ash particles are located between
1.5 and 2. The two peaks are ranging from 1.2 to 1.4 in Aspect Ratio Distributions
(Figure 5-1-c) and from 1.3 to 1.4 in Roundness Distributions (Figure 5-1-d),
respectively. Either ash distribution has a higher peak than MSW distribution. This
means ash particles are more homogeneous in size, sphericity, aspect ratio, and
roundness than the MSW particles.




In the combustion chamber
the acting grate and heated up.
shape change processes is
start generating volatile matter and
particle shrinkage results
the MSW particles distributions
a) Size Distributions
Ash

Aspect Ratio Distribution
0
2
4
6
8
10
12
14
16
0.5 1.5 2.5
Aspect Ratio
R
e
l
a
t
i
v
i

F
r
e
q
u
e
n
c
y

%

/

c
m

(
b
y

N
u
m
b
e
r

o
f

P
a
r
t
i
c
l
e
s
)
MSW

Ash

c) Aspect Ratio Distributions
Figure 5-1: Distributions by particle
size distributions (PSDs), b) Sphericity distribution
distributions




In the combustion chamber, the MSW particles are mixed by the movement
acting grate and heated up. The most important phenomenon during
is volatilization, since around 300
o
C MSW particles usually
tart generating volatile matter and their mass and volume decrease rapidly
shrinkage results in size reduction and shape change of MSW particles
distributions get closer to ash distributions.
Sphericity Distribution
0
2
4
6
8
10
12
0.5 1 1.5 2 2.5
Sphericity
R
e
l
a
t
i
v
i

F
r
e
q
u
e
n
c
y

%

/

c
m

(
b
y

N
u
m
b
e
r

o
f

P
a
r
t
i
c
l
e
s
)
a) Size Distributions b) Sphericity Distributions
MSW

Ash

MSW

Aspect Ratio Distribution
3.5 4.5
Aspect Ratio
Roundness Distribution
0
5
10
15
20
25
0.5 1.5 2.5
Roundness
R
e
l
a
t
i
v
i

F
r
e
q
u
e
n
c
y

%

/

c
m

(
b
y

N
u
m
b
e
r

o
f

P
a
r
t
i
c
l
e
s
)
MSW

Ash

Aspect Ratio Distributions
d) Roundness Distributions
: Distributions by particle numbers of residential MSW and combined ashes in NYC a) Particle
), b) Sphericity distributions, c) Aspect Raito distributions (ARD
113
Chapter Five
the movement of
during size and
C MSW particles usually
rapidly. This
size reduction and shape change of MSW particles. Then
to ash distributions. Finally, the
Sphericity Distribution
3 3.5 4 4.5
Distributions

Roundness Distribution
3.5 4.5
Roundness Distributions
residential MSW and combined ashes in NYC a) Particle
(ARDs), d) Roundness
114

Chapter Five
combustible MSW particles become ash during the char gasification stage (fixed
carbon oxidization).

The particle size distributions (PSDs), sphericity distributions, aspect ratio
distributions (ARDs) and roundness distributions of MSW and ash particles were
analyzed. The peak of MSW PSD is around 10 cm and shifts to the peak of ash PSD,
around 2 cm, during the combustion process. Also, the “tail” of the MSW PSD is
larger than that of ash PSD. In sphericity distributions, the peak of the MSW particles
was found to be at 1.5 and of the ash particles at 2. The Aspect Ratio Distributions are
1.2 and 1.4 and the Roundness Distributions 1.3 and 1.4, respectively.
115

Chapter Five

5-1-2. Probabilities from full-scale physical model
Figures 5-2 to 13 show particle movement and the state diagrams (transition
graph) of small, medium, and large tracer particles at different heights (20, 40, 60,
and 80 cm) of the MSW bed. A tracer particle in each cell has a different behavior.
For example, a tracer located in the cells at layer E moves very quickly because it is
pushed by a reciprocating bar.
Figure 5-14 shows the measured probabilities of small, medium, and large
particles that remain in the same cell location of a NYC-MSW packed bed after one
reciprocation of its moving bars. These probabilities in layer E, at the bottom of the
bed, are lowest for small and medium particles. For large particles, no experimental
data could be obtained because the small and medium particles occupied layer E and
tended to remain at the bottom so that large particles could not transit there. These
probabilities showed that 22% of small particles and 8% of medium particles stay in
layer E while the rest of them (78% of small and 92% of medium particles) move to a
neighboring cell because of the motion of a reciprocating bar.

Layer B is where small and medium particles have the second lowest
probabilities and large tracers have the lowest. Approximately 50% for any one of the
three sizes remain in layer B while the rest of them move to other cells. Retention
probabilities for the middle of the MSW bed (layers C and D) are much higher with
values up to 81%. Since the MSW was loaded up at the height of 80 cm, layer B is at
116

Chapter Five
the top of the bed and a tracer on the free surface moves more because there are fewer
particles acting on it to constrain its motion. Particles in this top layer were
sometimes observed to be easily rolling along the free surface, a type of motion that is
unavailable to particles in the lower layers.

117

Chapter Five

15 degree
Grate inclination: 15
o

Bed solid: Mixed MSW
Tracer: 6 cm sphere tracer
Original bed height: 20 cm
A0 A1 A2 A3 A4 A5
B0 B1 B2 B3 B4 B5
C0 C1 C2 C3 C4 C5
D0 D1 D2 D3 D4 D5
E0 E1 E2 E3 E4 E5
E1 E2 E3 E4 E5 E0
D1 D2 D3 D4 D5 D0
C1 C2 C3 C4 C5 C0
B1 B2 B3 B4 B5 B0
A1 A2 A3 A4 A5 A0
20 cm
Figure 5-2: Observed particle movements and corresponding state diagram

Corresponding state diagram

Observed particle movements: Number of the circles in a cell
means how many a tracer particle stayed in the cell per
reciprocating motion of moving bars. One arrow indicates a
movement and direction of a particle tracer into a neighbor
cell during a reciprocating motion of moving bars.
Solid bed divided by20cm×20cm cells
CCD image from Side view of the bed
118

Chapter Five

15 degree
Grate inclination: 15
o

Bed solid: Mixed MSW
Tracer: 14 cm sphere tracer
Original bed height: 20 cm
20 cm
Figure 5-3: Observed particle movements and corresponding state diagram

Corresponding state diagram

Observed particle movements: Number of the circles in a cell
means how many a tracer particle stayed in the cell per
reciprocating motion of moving bars. One arrow indicates a
movement and direction of a particle tracer into a neighbor
cell during a reciprocating motion of moving bars.
Solid bed divided by20cm×20cm cells
CCD image from Side view of the bed
A0 A1 A2 A3 A4 A5
B0 B1 B2 B3 B4 B5
C0 C1 C2 C3 C4 C5
D0 D1 D2 D3 D4 D5
E0 E1 E2 E3 E4 E5
E1 E2 E3 E4 E5 E0
D1 D2 D3 D4 D5 D0
C1 C2 C3 C4 C5 C0
B1 B2 B3 B4 B5 B0
A1 A2 A3 A4 A5 A0
119

Chapter Five

15 degree
Grate inclination: 15
o

Bed solid: Mixed MSW
Tracer: 22 cm sphere tracer
Original bed height: 20 cm
20 cm
Figure 5-4: Observed particle movements and corresponding state diagram

Corresponding state diagram

Observed particle movements: Number of the circles in a cell
means how many a tracer particle stayed in the cell per
reciprocating motion of moving bars. One arrow indicates a
movement and direction of a particle tracer into a neighbor
cell during a reciprocating motion of moving bars.
Solid bed divided by20cm×20cm cells
CCD image from Side view of the bed
A0 A1 A2 A3 A4 A5
B0 B1 B2 B3 B4 B5
C0 C1 C2 C3 C4 C5
D0 D1
D2
D3 D4 D5
E0 E1 E2 E3 E4 E5
E1 E2 E3 E4 E5 E0
D1 D2 D3 D4 D5 D0
C1 C2 C3 C4 C5 C0
B1 B2 B3 B4 B5 B0
A1 A2 A3 A4 A5 A0
120

Chapter Five

40 cm
Grate inclination: 15
o

Bed solid: Mixed MSW
Tracer: 6 cm sphere tracer
Original bed height: 40 cm
Figure 5-5: Observed particle movements and corresponding state diagram

Corresponding state diagram

Observed particle movements: Number of the circles in a cell
means how many a tracer particle stayed in the cell per
reciprocating motion of moving bars. One arrow indicates a
movement and direction of a particle tracer into a neighbor
cell during a reciprocating motion of moving bars.
Solid bed divided by20cm×20cm cells
CCD image from Side view of the bed
15 degree
A0 A1 A2 A3 A4 A5
B0 B1 B2 B3 B4 B5
C0 C1 C2 C3 C4 C5
D0 D1 D2 D3 D4 D5
E0 E1 E2 E3 E4 E5
E1 E2 E3 E4 E5 E0
D1 D2 D3 D4 D5 D0
C1 C2 C3 C4 C5 C0
B1 B2 B3 B4 B5 B0
A1 A2 A3 A4 A5 A0
121

Chapter Five

40 cm
Grate inclination: 15
o

Bed solid: Mixed MSW
Tracer: 14 cm sphere tracer
Original bed height: 40 cm
Figure 5-6: Observed particle movements and corresponding state diagram

Corresponding state diagram

Observed particle movements: Number of the circles in a cell
means how many a tracer particle stayed in the cell per
reciprocating motion of moving bars. One arrow indicates a
movement and direction of a particle tracer into a neighbor
cell during a reciprocating motion of moving bars.
Solid bed divided by20cm×20cm cells
CCD image from Side view of the bed
15 degree
A0 A1 A2 A3 A4 A5
B0 B1 B2 B3 B4 B5
C0 C1 C2 C3 C4 C5
D0 D1 D2 D3 D4 D5
E0 E1 E2 E3 E4 E5 E1 E2
E1 E2 E3 E4 E5 E0
D1 D2 D3 D4 D5 D0
C1 C2 C3 C4 C5 C0
B1 B2 B3 B4 B5 B0
A1 A2 A3 A4 A5 A0
122

Chapter Five


40 cm
Grate inclination: 15
o

Bed solid: Mixed MSW
Tracer: 22 cm sphere tracer
Original bed height: 40 cm
Figure 5-7: Observed particle movements and corresponding state diagram

Corresponding state diagram

Observed particle movements: Number of the circles in a cell
means how many a tracer particle stayed in the cell per
reciprocating motion of moving bars. One arrow indicates a
movement and direction of a particle tracer into a neighbor
cell during a reciprocating motion of moving bars.
Solid bed divided by20cm×20cm cells
CCD image from Side view of the bed
15 degree
E1 E2 E3 E4 E5 E0
D1 D2 D3 D4 D5 D0
C1 C2 C3 C4 C5 C0
B1 B2 B3 B4 B5 B0
A1 A2 A3 A4 A5 A0
A0 A1 A2 A3 A4 A5
B0 B1 B2 B3 B4 B5
C0 C1 C2 C3 C4 C5
D0 D1 D2 D3
D4
D5
E0 E1 E2 E3 E4 E5 E1 E2
D1 D3
E3 E4
123

Chapter Five

60 cm
Grate inclination: 15
o

Bed solid: Mixed MSW
Tracer: 6 cm sphere tracer
Original bed height: 60 cm
Figure 5-8: Observed particle movements and corresponding state diagram

Corresponding state diagram

Observed particle movements: Number of the circles in a cell
means how many a tracer particle stayed in the cell per
reciprocating motion of moving bars. One arrow indicates a
movement and direction of a particle tracer into a neighbor
cell during a reciprocating motion of moving bars.
Solid bed divided by20cm×20cm cells
CCD image from Side view of the bed
15 degree
A0 A1 A2 A3 A4 A5
B0 B1 B2 B3 B4 B5
C0 C1 C2 C3 C4 C5
D0 D1 D2 D3 D4 D5
E0
E1 E2 E3 E4
E5
E1 E2 E3 E4 E5 E0
D1 D2 D3 D4 D5 D0
C1 C2 C3 C4 C5 C0
B1 B2 B3 B4 B5 B0
A1 A2 A3 A4 A5 A0
124

Chapter Five

60 cm
Grate inclination: 15
o

Bed solid: Mixed MSW
Tracer: 14 cm sphere tracer
Original bed height: 60 cm
Figure 5-9: Observed particle movements and corresponding state diagram

Corresponding state diagram

Observed particle movements: Number of the circles in a cell
means how many a tracer particle stayed in the cell per
reciprocating motion of moving bars. One arrow indicates a
movement and direction of a particle tracer into a neighbor
cell during a reciprocating motion of moving bars.
Solid bed divided by20cm×20cm cells
CCD image from Side view of the bed
15 degree
A0 A1 A2 A3 A4 A5
B0 B1 B2 B3 B4 B5
C0 C1 C2 C3 C4 C5
D0 D1
D2
D3 D4 D5
E0
E1 E2
E3 E4
E5 E0
E1 E2 E3 E4 E5 E0
D1 D2 D3 D4 D5 D0
C1 C2 C3 C4 C5 C0
B1 B2 B3 B4 B5 B0
A1 A2 A3 A4 A5 A0
125

Chapter Five


60 cm
Grate inclination: 15
o

Bed solid: Mixed MSW
Tracer: 22 cm sphere tracer
Original bed height: 60 cm
Figure 5-10: Observed particle movements and corresponding state diagram

Corresponding state diagram

Observed particle movements: Number of the circles in a cell
means how many a tracer particle stayed in the cell per
reciprocating motion of moving bars. One arrow indicates a
movement and direction of a particle tracer into a neighbor
cell during a reciprocating motion of moving bars.
Solid bed divided by20cm×20cm cells
CCD image from Side view of the bed
15 degree
A0 A1 A2 A3 A4 A5
B0 B1 B2 B3 B4 B5
C0 C1 C2 C3 C4 C5
D0 D1 D2 D3 D4 D5
E0
E1 E2
E4
E5 E0
E3
E1 E2 E3 E4 E5 E0
D1 D2 D3 D4 D5 D0
C1 C2 C3 C4 C5 C0
B1 B2 B3 B4 B5 B0
A1 A2 A3 A4 A5 A0
126

Chapter Five


80 cm
Grate inclination: 15
o

Bed solid: Mixed MSW
Tracer: 6 cm sphere tracer
Original bed height: 80 cm
Figure 5-11: Observed particle movements and corresponding state diagram

Corresponding state diagram

Observed particle movements: Number of the circles in a cell
means how many a tracer particle stayed in the cell per
reciprocating motion of moving bars. One arrow indicates a
movement and direction of a particle tracer into a neighbor
cell during a reciprocating motion of moving bars.
Solid bed divided by20cm×20cm cells
CCD image from Side view of the bed
15 degree
A0 A1 A2 A3 A4 A5
B0 B1 B2 B3 B4 B5
C0
C1 C2 C3
C4
C5
D0
D1 D2 D3 D4 D5
E0
E1 E2
E4
E5 E0
E3
E1 E2 E3 E4 E5 E0
D1 D2 D3 D4 D5 D0
C1 C2 C3 C4 C5 C0
B1 B2 B3 B4 B5 B0
A1 A2 A3 A4 A5 A0
127

Chapter Five

80 cm
Grate inclination: 15
o

Bed solid: Mixed MSW
Tracer: 14 cm sphere tracer
Original bed height: 80 cm
Figure 5-12: Observed particle movements and corresponding state diagram

Corresponding state diagram

Observed particle movements: Number of the circles in a cell
means how many a tracer particle stayed in the cell per
reciprocating motion of moving bars. One arrow indicates a
movement and direction of a particle tracer into a neighbor
cell during a reciprocating motion of moving bars.
Solid bed divided by20cm×20cm cells
CCD image from Side view of the bed
15 degree
A0 A1 A2 A3 A4 A5
B0 B1 B2 B3 B4 B5
C0
C1 C2 C3
C4
C5
D0
D1 D2 D3 D4 D5
E0
E1 E2
E4
E5 E0
E3
D1 D3
E0
E1 E2
E4
E0
E1 E2 E3 E4 E5 E0
D1 D2 D3 D4 D5 D0
C1 C2 C3 C4 C5 C0
B1 B2 B3 B4 B5 B0
A1 A2 A3 A4 A5 A0
128

Chapter Five


80 cm
Grate inclination: 15
o

Bed solid: Mixed MSW
Tracer: 22 cm sphere tracer
Original bed height: 80 cm
Figure 5-13: Observed particle movements and corresponding state diagram

Corresponding state diagram

Observed particle movements: Number of the circles in a cell
means how many a tracer particle stayed in the cell per
reciprocating motion of moving bars. One arrow indicates a
movement and direction of a particle tracer into a neighbor
cell during a reciprocating motion of moving bars.
Solid bed divided by20cm×20cm cells
CCD image from Side view of the bed
15 degree
E1 E2 E3 E4 E5 E0
D1 D2 D3 D4 D5 D0
C1 C2 C3 C4 C5 C0
B1 B2 B3 B4 B5 B0
A1 A2 A3 A4 A5 A0
A0 A1 A2 A3 A4 A5
B0 B1 B2 B3 B4 B5
C0 C1
C2 C3
C4
C5
D0
D1 D2 D3 D4 D5
E0
E1 E2
E4
E5 E0
E3
D1 D3
E0
E1 E2
E4
E0
D1 D2 D3 D4
129

Chapter Five


Figure 5-14: Measured probabilities that particles stay in the same position (cell) after one
reciprocation of the moving bars


0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
20 40 60 80
P
r
o
b
a
b
i
l
i
t
i
e
s


t
h
a
t

p
a
r
t
i
c
l
e
s

s
t
a
y

i
n

t
h
e

s
a
m
e

p
o
s
i
t
i
o
n

(
c
e
l
l
)
MSW bed height (cm)
Small
Medium
Large
20 cm (Layer E)
(bottom)
40 cm (Layer D) 60 cm (Layer C) 80 cm (Layer B)
(top)
130

Chapter Five
5-2. Results from mathematical work
5-2-1. Mass and volumetric flow of solids (Monte Carlo simulation results)
Figure 5-15 shows the simulation results for continuous variation of the
components of MSW generated in New York City. The abscissa shows the MSW
sample number. Each sample has several components and, together, they define a
time series for continuous variation of MSW. The first 20 of the 100 samples used are
shown in Figure 5-15. The time series made by these samples specifies one of the
initial conditions to simulate the combustion process.
Figure 2. Percolation Model in the Combustion Chamber
Figure 5-15: Time series of continuous variation of MSW components by Monte Carlo simulation

0
0.1
0.2
0.3
0.4
0 5 10 15 20
C
o
m
p
o
n
e
n
t

p
r
o
b
a
b
l
i
l
t
y

(
x

1
0
0

W
e
i
g
h
t

%
)





Sample Number
paper
cardboard
plastic
textiles
rubber & leather
wood
glass
metals
other
131

Chapter Five
The simulation results using the above samples are shown in Figure 5-16. The
solid temperature is highest in the middle of the grate which is located around Zone 4,
as shown in Figure 5-16 (a) (see Appendices in detail). Near the inlet of the
combustion chamber, temperatures of solid waste remain under 371 K because
moisture evaporation occurs in this stage and combustion of most components of
MSW is limited. Figure 5-16 (b) shows the residual ash in solid (mass %) (see
Appendices in detail). From the middle of the grate to the ash discharge end of the
grate, the MSW burns well and generates ash until it reaches the burn-out line, which
is in Zone 8. Figure 5-16 (c) shows the simulation results of percolation model for
transient phenomena (break-up and channeling). It is based on the calculation result
of residual ash shown in Figure 5-16 (b). From Zone 5 to Zone 8, each combustion
probability is nearly equal to or exceeds the critical probability P
c
. This indicates that
the possibility of break-up and channeling of solid waste after Zone 5 is high and
channeling is observed in the lattice of Zone 5.
132

Chapter Five


(c) Result of percolation model for transient phenomena (break-up and channeling)

Figure 5-16: Calculation results of the MSW combustion on the Martin grate

(a) Solid temperature (K) (b) Residual ash in solid (mass %)
133

Chapter Five
Figure 5-17 shows how much combustion is completed by volume % in the
each zone. We assumed that MSW from inlet of the chamber consists of 20 %
volumetric voids, i.e., P
v
= 0.2. Therefore, for example, solid waste in Zone 1 burns
only 13 volume % of total solid waste. The most important combustion phenomena
and their most probable occurrences are shown in Figure 5-18. In this study, we
assumed P
v
= 0.2. P
c
is 0.5927 because we specified a square lattice for this
percolation model. P
ch
is the channeling probability and, between P
c
and P
ch
,
channeling is generated frequently. P
a
is the ash probability. This means the burn-out
line which means the remaining solid waste does not contain combustible matter.
Break-up occurs between P
ch
and P
a
.
The simulation results of this combustion model with continuous variation of
MSW have provided the temperatures and percentage of residual ash by weight. From
that, the volume change (%) from waste to ash has been calculated and the
combustion probabilities for each zone have been estimated. The percolation model
has shown the mechanism of transient phenomena such as channeling and break-up of
solid waste. The governing parameters such as void probability P
v
, channeling
probability P
ch
, critical probability P
c
, and ash probability P
a
were found by this
model.
134

Chapter Five

Figure 5-17: Combustion probability in each zone
Figure 5-18: Transient phenomena during the combustion process
Zone 1
Zone 2
Zone 3
Zone 4
Zone 5
Zone 6
Zone 7
Zone 8
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
1 2 3 4 5 6 7 8
P
r
o
b
a
b
i
l
i
t
y

(
C
o
m
b
u
s
t
i
o
n

[
V
o
l
u
m
e

%
]
)
Zone Number
0
135

Chapter Five
5-2-2. Simulation results from the stochastic model for MSW mixing
a) One-dimensional simulation
Figure 5-19 shows a simulated “random walk” that represents the particle
movement based on a calculation over 46 steps (the required number of calculations
varies between different time series because of the stochastic nature of the simulation
results). Cell #1 represents the feed end of the grate and cell #16 the end of the grate
where the ash falls to the ash pit below. Cell #17 denotes the state of bottom ash. This
graph shows how a hypothetical solid particle travels over the reverse acting grate. At
each motion of the reciprocating bar, the particle may remain in the same cell or
move to the adjacent cell, downward or upward, due to the upward motion of the
bars. At the #2 and #5 cells, the particle seems not to move for a while, which, in a
multi-layer model of the bed on the grate, would result in mixing with the next layer
up (see next section 5-2-2-b, Two-dimensional simulation). The lower graph in
Figure 5-19 provides a visual illustration of the behavior of a particle on the moving
grate, based on the results of the matrix calculation. The particle actually moves back
and forth between cells as indicated by the solid line. For example, between cells #11
and #12, the particle travels back and forth because of the action of the reciprocating
bar.
136





Figure 5-19: Simulation results of the movement of a solid waste particle on the Reverse Acting
Grate: Top graph: tracing a behavior of one particle on the bed. In cells #2 and 5 the particle
seems to stay put for a while. In cells #11 and 12, the particle goes back and forth because of the
movement of the reciprocating bar. Bottom graph: Visualization of the particle travel based on the
results of the calculations shown on top graph.

Figure 5-20 shows the residence time distributions of solid waste particles
with successive motions of the reciprocating bars. Under the combination of
downward gravity and upward motion of the reciprocating bars, all particles travel
gradually to the outlet. The peak of each particle distribution profile is displaced
toward the outlet and the profile becomes much flatter. The solid particles that move
slower (particles in the left part of each profile in Figure 5-20 have been pushed
137


upward by the reciprocating bars. This action contributes to mixing partly combusted
material with newly fed solid wastes.
It is evident that solid wastes that are more mixed have a better chance of
being combusted completely. On the other hand, solid particles that travel faster than
the bulk of the solid wastes (particles in the right part of the each profile in Figure 5-
20 will not burn completely. Without additional feed coming in, the probability of
particles staying in a specific cell, after an infinite number of movements (step n = ∞,
approaches zero, because all solid particles travel toward the outlet and eventually
reach it.
0
0.1
0.2
0.3
0.4
0.5
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
Cell Number (posit ion in t he grat e)
P
r
o
b
a
b
i
l
i
t
y




Inlet Outlet
n=5
n=10
n=40
n=80
Figure 5-20. Change in tracer particles distribution profiles over grate with number of movements
of reciprocating bars (n=5, 10, 40, 80). The regular perturbations in the n=40 and n=80 profiles
are due to the effect of alternatively passing over stationary and reciprocating bars.

138


In summery of this section, a new method for investigating mixing on the
grate of a WTE combustion chamber, using the one dimensional stochastic model,
was presented. This method may provide a useful for predicting the actual path of
particles, for a certain grate configuration (frequency of reciprocation bar motion,
length of travel, etc.). This information would contribute to the characterization and
quantification of the mixing process. In the first test of the one-dimensional stochastic
model presented in this section, the bottom layer of the bed, which is in contact with
the bars of a reverse acting Martin grate, was simulated. Based on these calculation
results, a two-dimensional stochastic model (multi-layer model) for MSW mixing was
developed in order to expand the single layer of MSW bed to the multi-layer. The
following section (5-2-2-b) shows the results of a two-dimensional stochastic
simulation of MSW particle mixing.
b) Two-dimensional simulation
Figure 5-21 shows the MSW particle distributions with successive motions of
the reciprocating bars. Since the reciprocating bars move at a constant frequency (a
typical value is around 11-13 strokes/hour), Figure 5-21 represents residence time
distributions (RTD) over the grate. The peak of each distribution is moved toward the
outlet of the chamber because all solid waste particles travel from the inlet to the
outlet, affected by the motion of the reciprocating bars in combination with the
downward force of gravity. The effect of particles passing alternatively over fixed and
139


reciprocating bars results in the regular perturbations eminently shown in the n = 40
and 80 profiles.

140



Figure 5-21: Change in particle distribution profiles over grate with number of movements of
reciprocating bars (a) n=5, (b) n = 10, (c) = 20,(d) n = 40,(e )n= 80
(a) n = 5
(b) n = 10
(c) n = 20
(d) n = 40
(e) n = 80
141


Figure 5-22 shows the motion (paths) of traveling MSW particles in layers of
the MSW fuel above the reverse acting grate for various numbers of reciprocating
motions. According to the assumptions made in defining probabilities within the
matrix P, particles in layer A exhibit the highest movement toward the outlet because
layer A is near the surface of the MSW fuel and some particles can slip on the surface
or roll down along the surface. The particle path in Figure 5-22 (a) shows that this
particle travels mainly in the layer A and B where the probabilities of the movement
forward to the outlet direction are higher. In the case of Figure 5-22 (a), the step
number n is smallest (n = 67) than in the other cases (i.e., 5-22 (b), (c) and (d)) and
the particle has the shortest residence time on the grate. The particle can be located in
the surface layer where chemical reaction rates are high, although it has the shortest
residence time and path. The path of a particle shown in Figure 5-22 (b) indicates that
this particle remained for a long period of time in zones 25-30.
These zones include the burnout point and the particle must be almost ash in
an actual combustion chamber. The case shown in Figure 5-22 (c) expresses mixing
of the particle in the first half of the chamber (zones 1 to 16). On the other hand, the
rest of the path along the second half of the grate (zones 17 to 32) seems smoother.
The bottom layer E is the nearest layer to the reciprocating bars and particles in this
layer are exposed to the driving force of the reciprocating bars against the feed
direction. Figure 5-22 (d) shows a path of a MSW particle that reached ash bin when
142


n = 401. This indicates mixing of this particle near zone 10 and between zone 20 and
25, due to the movement of the reciprocating bars.
143



Figure 5-22: Movements (paths) of MSW particles on the Reverse Acting Grate:
(a) n=67, (b) n=106, (c) n=213 and (d) n=401
(a) n =67
(b) n=106
(c) n=213
(d) n=401
144


In summary, a novel two-dimensional model simulating the mixing of the
MSW fuel on the Martin reverse acting grate has been developed using the stochastic
simulation. This model can be as a tool for charactering and quantifying the mixing
process, although the properties of MSW particles, such as size, shape, and density,
was not taken into account. The probabilities in the transition matrix P were assumed
based on the successful results of the one-dimensional model. However, these
assumed probabilities were found to be relatively low. Lower probabilities results in
longer residence time and more strokes to reach ash bin (for example, as shown in
Figure 5-22-d, 401 strokes of reciprocating bars is too high, even though, the particle
remained same zones many times in this case). In order to determine these
probabilities in transition matrix P, they are measured as shown in section 3-3 and 5-
1-2, using full-scale physical section model.
5-2-3. Results from mass reduction model
Figure 5-23 shows the simulation results of the gamma distribution discussed
in section 4-3-2. Dots represent experimental data and lines are the predicted
distribution. As summarized in Table 5-1, the PSD of MSW has an “a” constant value
of 2.5 x 100, an α parameter of 4, a β parameter of 3.2, a mean µ of 12.8, a standard
deviation, σ, of 6.4, and a covariance, CV, of 0.5. The PSD of ash resulted in having a
constant a value of 0.3 x 100, an αparameter of 12, a βparameter of 0.17, a mean μ
of 2.04, a standard deviation, σ, of 0.5889, and a covariance, CV, of 0.289.
Compared with the measured PSDs shown in Figure 5-23, the predicted PDSs have a
145


Particle Size Distribution
0
5
10
15
20
25
1 10 100
Particle Size (cm)
R
e
l
a
t
i
v
i

F
r
e
q
u
e
n
c
y

%

/

c
m

(
b
y

N
u
m
b
e
r

o
f

P
a
r
t
i
c
l
e
s
)
MSW

Ash
Figure 5-23: Particle size distributions (PSD) by particle numbers of residential MSW and
combined ashes in NYC: (lines: estimated gamma distributions, dots: experimental data)
similar shape except for the asymptotic tail of the MSW PSD (30-40 cm). This tail
group of MSW particles contains a small peak. That is primarily because some of
physical phenomena such as compaction and breakage have not been included in this
combustion model asyet.








146



Ash particle size
distribution
MSW particle size
distribution
Parameter Predicted Experimental Predicted Experimental
Mean µ 2.04 2.32 12.8 14.3
Standard
dev. σ
0.589 0.821 6.4 8.508
Covariance
CV
0.289 0.354 0.5 0.594
a 0.3 X 100 - 2.5 X 100 -
α 12 7.97 4 2.83
β 0.17 0.29 3.2 5.045


5-2-4. Integrating model for flow and mixing with data from the full-scale
physical model
Visual representation of the computed particle movement behavior, that is
difficult impossible to measure experimentally can be very useful understanding and
characterizing the mixing phenomena on the grate. Figure 5-24 shows the calculated
movements for 12 particles traveling from section 1 through 32. Each particle travels
along different paths in the MSW bed, but trends are consistent for a given size
Table 5-1: Particle size parameters obtained from residential MSW and combined ashes in NYC
147


especially in the vertical (depth) direction. The general trend indicated by the
simulation is that small particles tend to migrate downward (toward the grate surface)
and large particles tend to migrate upward (toward the free surface) as they move
along the bed. This size segregation has been called the Brazil Nut Effect (BNE)
(Mohabuth, N., and Miles, N. 2005) and results from the slight differences in vertical
probabilities as a function of size. The segregation mechanism we observed in this
study is illustrated in Figure 5-25. The moving grates create empty space after one
stroke from the top to the bottom position. Immediately, because of the unstable
conditions, particles whose size is less than the grate height, can easily drop into the
empty space. Larger particles are less likely to drop into this newly created vacancy
because their comparatively larger size offers more opposition to easily slipping into
this limited empty space. After several reciprocations of moving bars, the size
segregation appears as shown in Figure 5-25. This mechanism of size segregation can
be observed in rock, sand, powder, and granular movement. Some studies have been
carried out employing this vertical selectivity mechanism for electronic waste
(Mohabuth, N., and Miles, N. 2005), and for applying separation and recycling
technologies to the processing of industrial waste and residential MSW.


Figure 5-24: Calculated
particles (S),



: Calculated particle path over the grate as a function of size: small
particles (S), medium particles (M), and large (L) particles
148

as a function of size: small



Figure 5-25: Brazil Nut Effect (BNE) in

Figure 5-26 shows
concentration versus residence time
grate speed ranging from 15 to
a flow material were origi
motion of the reverse acting grate increases
medium particles to 106 min
that of large particles to
distributions for small particles
exceeded 30 recip./hr. As we discussed
mean residence times with size comes from the following reasons:
medium particles are pushed by
nearly the same or smaller than the height
are less likely to be caught by the grate so that their motion is less likely to be in
: Brazil Nut Effect (BNE) in a packed MSW bed
shows the residence time distributions (RTD, dimensionless
versus residence time t) for small, medium, and large particles
from 15 to 90 reciprocation/hour. Residence time distributions of
a flow material were originally defined by Levenspiel (Levenspiel, O. 1999)
reverse acting grate increases the mean residence time
106 min (68%) and 69 min (9%), respectively, while decre
that of large particles to 51 min (17%). In addition, two peaks of residence time
distributions for small particles prominently appeared when the reciprocation speed
. As we discussed earlier, it is reasonable that the difference of
mean residence times with size comes from the following reasons: (1)
are pushed by the reverse acting grate because their diameters are
smaller than the height of the moving bars, h. (2) Larger particles
likely to be caught by the grate so that their motion is less likely to be in
149


(RTD, dimensionless
) for small, medium, and large particles for a
Residence time distributions of
nally defined by Levenspiel (Levenspiel, O. 1999). The
mean residence time of small and
ctively, while decreasing
. In addition, two peaks of residence time
eciprocation speed
earlier, it is reasonable that the difference of
(1) Small and
their diameters are
Larger particles
likely to be caught by the grate so that their motion is less likely to be in
150


opposition to the flow direction. (3) A special mode of transport, rolling down the top
surface along the flow direction, is available exclusively to those particles near the
free surface of the bed. These two opposing motions at the bottom and the surface of
the bed enhance MSW particle mixing and are responsible for the different residence
time distributions. Due to the particle motion behavior (1)-(3), a vertical selectivity
mechanism known as the Brasil Nut Effect (BNE) develops that tends to keep small
particles in the bottom of the MSW bed. This effect is manifested by the presence of
two peaks that appear only in the residence time exit concentration distributions of the
small particles. The full data sets of RTDs (C and F diagrams) for grate speed ranging
from 15 to 90 recip./hr are shown in Figure 5-27, 5-28, and 5-29.


Figure 5-26: C (top) and F diagrams (bottom) for small, medium, and large particles with a


20 40
0
0.5
1
1.5
2
2.5
3
3.5
4
D
i
m
e
n
s
i
o
n
l
e
s
s

E
x
i
t

C
o
n
c
e
n
t
r
a
t
i
o
n
C

20 40
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
D
i
m
e
n
s
i
o
n
l
e
s
s

C
u
m
u
l
a
t
i
v
e

C
o
n
c
e
n
t
r
a
t
i
o
n


F
t
large-90
=51
: C (top) and F diagrams (bottom) for small, medium, and large particles with a
reciprocation speed of 90 recip./hr.
40 60 80 100 120 140
C Diagram
Residence Time t (minutes)

t
large-90
=51mi n.
t
medium-90
=69min.
t
small-90
=106min.
40 60 80 100 120 140
F Diagram
Residence Time t (minutes)
t
small-90
=106min.
t
medium-90
=69min.
=51min.
151


: C (top) and F diagrams (bottom) for small, medium, and large particles with a




Figure 5-27: C and F diagrams for small particles: Dimensionless exit concentration C (top) and
Dimensionless cumulative
different reciprocation speeds ranging from15 to 90 recip./hr

40 50
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
D
i
m
e
n
s
i
o
n
l
e
s
s

C
u
m
u
l
a
t
i
v
e

C
o
n
c
e
n
t
r
a
t
i
o
n


F
: C and F diagrams for small particles: Dimensionless exit concentration C (top) and
Dimensionless cumulative concentration F (bottom) versus residence time (min) with
different reciprocation speeds ranging from15 to 90 recip./hr
60 70 80 90 100 110 120
F Diagram
Residence Time t (minutes)
t
30
=68min.
t
15
=64min.
t
0
=63min.
t
60
=82min.
t
90
=106min.
152

: C and F diagrams for small particles: Dimensionless exit concentration C (top) and
concentration F (bottom) versus residence time (min) with
153





Figure 5-28: C and F diagrams for medium particles: Dimensionless exit concentration C (top) and
Dimensionless cumulative concentration F (bottom) versus residence time (min) with
different reciprocation speeds ranging from15 to 90 recip./hr
40 50 60 70 80 90 100
0
0.5
1
1.5
2
2.5
3
3.5
4
4.5
5
C Diagram
Residence Time t (minutes)
D
i
m
e
n
s
i
o
n
l
e
s
s

E
x
i
t

C
o
n
c
e
n
t
r
a
t
i
o
n
C
t
30
=65min.
t
15
=62min.
t
0
=63min.
t
60
=67min.
t
90
=69min.
40 50 60 70 80 90 100
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
D
i
m
e
n
s
i
o
n
l
e
s
s

C
u
m
u
l
a
t
i
v
e

C
o
n
c
e
n
t
r
a
t
i
o
n

F
Residence Time t (minutes)
F Diagram
t
90
=69min.
t
60
=67min.
t
0
=63min.
t
15
=62mi n.
t
30
=65min.
154




Figure. 5-29: C and F diagrams for large particles: Dimensionless exit concentration C (top) and
Dimensionless cumulative concentration F (bottom) versus residence time (min) with
different reciprocation speeds ranging from15 to 90 recip./hr


20 30 40 50 60 70 80 90 100
0
0.5
1
1.5
2
2.5
3
3.5
4
4.5
5
C Diagram
Residence Time t (minutes)
D
i
m
e
n
s
i
o
n
l
e
s
s

E
x
i
t

C
o
n
c
e
n
t
r
a
t
i
o
n
C
t
90
=51min.
t
60
=54min.
t
0
=63min.
t
15
=61min.
t
30
=58min.
20 30 40 50 60 70 80 90 100
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
F Diagram
D
i
m
e
n
s
i
o
n
l
e
s
s

C
u
m
u
l
a
t
i
v
e

C
o
n
c
e
n
t
r
a
t
i
o
n

F
Residence Time t (minutes)
t
30
=58min.
t
15
=61min.
t
0
=63min.
t
60
=54min.
t
90
=51mi n.
155


Mixing Diffusion Coefficient
Figure 5-30 shows the mixing diffusion coefficient D
e
for different particle
sizes and grate reciprocation speeds R
r
. The mixing diffusion coefficient D
e
for all
particle sizes increases linearly for speeds up to 30 recip./hr. The D
e
for medium
particles has a weak linear relationship with R
r
, throughout the entire speed range
from 15 - 90 recip./hr. In contrast, the coefficient D
e
for large particles increases at a
much slower rate and reaches a value of about 45 cm
2
/min when the reciprocation
speed reaches 90 recip./hr. For small particles, D
e
increases exponentially to 192
cm
2
/min. For R
r
< 30 recip./hr, these relationships of grate reciprocation speeds R
r
and
mixing diffusion coefficient D
e
can be described as follows:
˖
c

¡
, ˤ, ˨) =
`
1
1
1
1
o Ә
d
h
ә
2
˞
¡
Ә
h
d+b
ә
˩˦ ˤ < ˨
o Ә
d
h
ә
2
˞
¡
Ә
h
d
ә
˩˦ ˤ ¸ ˨

(Eq. 5-1)
where a is a constant ( ≈ 1.1) and can be considered to be a function of the number of
moving bars. The constant b has an approximate value is of 2.8. The parameters d and
h are the particle size and height of the moving bars, respectively, whose ratio d/h
controls the relationship between D
e
and R
r
. The mixing diffusion coefficients range
from 11 to 99 cm
2
/min and monotonically increase for all particle sizes that with
increasing reciprocation speed from 10 to 60 recip./hr. These numbers represent the
range at which the reverse acting grate is operated in commercial WTE plants. This
range of D
e
is in good agreement with the results of full-scale furnace tests in another
study (Yang Y.B.et. al., 2005), also shown in Table 5-2. In this full-scale test D
e



varies from 27 to 109 cm
traveling grate are not specified. Though no mention was made of the tracer size, their
densities were given as values 225 and 770 kg/m
bed is highly dependent on the experimental conditions such as the tracer size and
density as well as physical properties of the bed itself.
Figure 5-30: Mixing Diffusion Coefficients versus grate reciprocation speed




to 109 cm
2
/min, though the type and operational conditions of the
traveling grate are not specified. Though no mention was made of the tracer size, their
densities were given as values 225 and 770 kg/m
3
. Particle motion along the MSW
t on the experimental conditions such as the tracer size and
density as well as physical properties of the bed itself.

: Mixing Diffusion Coefficients versus grate reciprocation speed
for different particle sizes
156
/min, though the type and operational conditions of the
traveling grate are not specified. Though no mention was made of the tracer size, their
. Particle motion along the MSW
t on the experimental conditions such as the tracer size and

: Mixing Diffusion Coefficients versus grate reciprocation speed
157



Material Density (kg/m
3
) D
e
(cm
2
/min)
Kaowool boards 225 109
Insulation bricks 480 88
Refractory bricks 770 27

Table 5-2: Mixing Diffusion coefficient with different densities [full-scale test] (Yang et al 2005)

As noted earlier, in real combustion chambers, the particle size is
reduced because of volatilization and combustion processes, and the motion can be
characterized as a reacting flow. Pushed by the reverse acting grate, small particles,
stay in the combustion chamber longer than medium and large particles, although
they require a shorter time for complete combustion. According to a study of wood
particle size versus conversion time, the 95% reaction time for small, medium and
large sized particles at temperature of 900
o
C is about 12 min, 46 min, 95 min,
respectively (coversion time ≈ d
1.62
) (Petek J., 1998) as shown in Figure 5-31. The
mixing diffusion coefficient D
e
for small particles is higher and they are easily
dispersed along the horizontal direction. The advantage of this effect is that small
particles that are mostly burned out can transfer their heat to large particles which
need more heat and take more time to be thermally processed, especially during the
drying and gasification processes. Primary air at room temperature comes from the
bottom of the MSW packed bed. Due to the burn-out of the small particles pushed by
the reciprocating bars and due to the high residence times at the bottom of the MSW
bed, the primary air is warmed up by these small particles. The burn-out particles that


turn into ash have a higher heat capacity than MSW and so they transfer heat to the
large particles and the primary air.
observed that, within the range of 15 to 90 recip./hour,
bars does not look that it
that this is due to the fact that the grate motion effects only small/medium particles of
diameters less than the bar height, and whose mass volume is
typical MSW (Nakamura
model is combined with particle size change due to combustion, the stochastic model
simulation will become much more complicated, because the Markov property will be
applicable. In this case, the calculation should be carried out for each single time step
(reciprocation), since the initial condition, instead of only the previous step.

Figure 5-31: Wood particle conversion time vs particle diameter
turn into ash have a higher heat capacity than MSW and so they transfer heat to the
large particles and the primary air. Operators of the reverse acting grate have
e range of 15 to 90 recip./hour, “the speed of the reciprocating
look that it affects burnout of the entire MSW bed.” This study showed
that this is due to the fact that the grate motion effects only small/medium particles of
diameters less than the bar height, and whose mass volume is less than 20% of a
Nakamura, et al 2008). In the future research, when the stochastic
model is combined with particle size change due to combustion, the stochastic model
simulation will become much more complicated, because the Markov property will be
n this case, the calculation should be carried out for each single time step
(reciprocation), since the initial condition, instead of only the previous step.
Wood particle conversion time vs particle diameter (Petek J, 1998)
158
turn into ash have a higher heat capacity than MSW and so they transfer heat to the
Operators of the reverse acting grate have
“the speed of the reciprocating
burnout of the entire MSW bed.” This study showed
that this is due to the fact that the grate motion effects only small/medium particles of
less than 20% of a
In the future research, when the stochastic
model is combined with particle size change due to combustion, the stochastic model
simulation will become much more complicated, because the Markov property will be
n this case, the calculation should be carried out for each single time step
(reciprocation), since the initial condition, instead of only the previous step.
, 1998)


159
Chapter 6: Conclusions


The principal findings of this study can be summarized as follows:

1. In sphericity distributions, the peak of the MSW particles was found to be at
1.5 and of the ash particles at 2. The Aspect Ratio Distributions are 1.2 and
1.4 and the Roundness Distributions 1.3 and 1.4, respectively.
2. The movement and state diagram (transition graph) of three particle sizes
(small, medium, and large) at different MSW bed heights (20, 40, 60, and 80
cm) were estimated by means of a full scale physical model of the reverse
acting grate. Tracer particles exhibit a different behavior in different cells. For
example, a tracer particle located in the bottom of the bed cells (layer E)
moves quickly because it is pushed by a reciprocating bar.
3. The probability of particle movement with bed height (layer) of MSW was
calculated. All sizes of tracers have the highest probabilities in the bottom
layer E; the next highest probability of particle movement is in the top layer B
(60-80 cm in the physical model, when the MSW in the physical model was
loaded over 80 cm). In the top layer, the tracer particles can slide or roll easily
along the surface of the bed because there are no particles resting on them.


160
4. The

percolation model has shown the mechanism of transient phenomena such
as channeling and break-up of solid waste. The governing parameters such as
void probability P
v
, channeling probability P
ch
, critical probability P
c
, and ash
probability P
a
were determined by using this model.
Also a novel two-dimensional model simulating the flow and mixing of the MSW
fuel on the Martin reverse acting grate was developed using the stochastic simulation.
This model can be a useful tool in characterizing and quantifying the flow and mixing
phenomena on a combustion grate:
5. Different particle sizes result in different residence times according to the
Brazil Nut Effect (BNE): Larger particles rise to the surface while smaller
particles migrate to lower depths of the bed where the reciprocating bars push
them backwards against the main direction of flow due to the gravity force.
6. The motion of the reverse acting grate, at speeds ranging from 15 to 90
recip./hr, resulted in the net effect of increasing the mean residence time of
small and medium sized particles by 68 % and 9%, respectively, while
decreasing that of large particles by 17%.
7. The calculations of the mixing diffusion coefficient D
e
showed that it was
affected principally by the ratio of the height of reciprocating bars, h, to the
diameter of the tracer particle. Although the physical model tests were all


161
carried out at one h value (13 cm), the results showed conclusively that the
diffusion coefficient D
e
for a given particle size distribution will increase with
grate bar height.
8. A grate speed does not affect burnout of the entire MSW bed. This is due to
the fact that the grate motion effects only small/medium particles of diameters
less than the bar height, and whose mass volume is only 18% of the total
NYC-MSW. Pushed by the reverse acting grate, small particles, stay in the
combustion chamber longer than medium and large particles, although they
require a shorter time for complete combustion (coversion time ≈ d
1.62
). The
advantage of this effect is that small particles that mostly burned out can
easily transfer their heat to large particles which need more heat and take more
time to be thermally processed, especially during the drying and gasification
processes.

Suggestions for Future Work
This study developed a quantitative analytical tool for examining the mixing
of solid waste particles of different sizes during the combustion process in a WTE
chamber. The results can be used to assist in the evaluation of operational parameters
as well as geometric parameters of a reverse acting grate. The combination of
stochastic and full-scale physical modeling can also be used for comparing and


162
evaluating various types of traveling grate systems and designing the next generation
of combustion chambers. Suggestions for future work are as follows:
1. The forward acting and roller grate systems can be investigated by means of
full-scale physical models and stochastic simulation, as was done for the
reverse acting grate in this study.
2. A combination of traveling grate systems such as the reverse acting grate and
roller grate, or of parts of the reverse acting grate moving at different
reciprocating speeds can be investigated using full-scale physical models and
stochastic simulations, as illustrated in Figure 6-1
3. Analysis of the mixing phenomena for different shapes and densities of MSW
particles, along with the size effect investigated in this study, is necessary to
more accurately simulate the actual particle motion in the combustion
chamber of an MSW packed bed.
4. A combustion model can be combined with this stochastic mixing model. The
effect of mixing phenomena for different operation conditions (reciprocation
speeds) on the drying, gasification, and combustion stages, can be analyzed. In
this case, the stochastic model simulation will become more complicated,
because losing its Markov property. The calculation should be carried out in
each single time step (reciprocation).


5. The Discrete Element Method (DEM), based on Newtonian equations,
popularly used for
when much faster computers become available
has started being used
(Simsek et. al. 2008)






Figure 6-1: Stochastic models
novel traveling grate system

Element Method (DEM), based on Newtonian equations,
used for mathematical simulation of MSW mixing and
when much faster computers become available in the future. Currently
being used for solid fuel conversion on forward acting grate
(Simsek et. al. 2008).
s, calibrated by full scale grate tests, can simulate concept design
traveling grate systems
163
Element Method (DEM), based on Newtonian equations, may be
on of MSW mixing and combustion,
Currently DEM
n forward acting grate
concept design for
164
References


ANSYS, Inc., CFX software for thermal and fluid simulation
http://www.ansys.com/default.asp
Allen, Terence, “Particle Size Measurement Volume 1, Powder Sampling and Particle
Size Measurement,” Fifth Edition, Powder Technology Series, Chapman & Hall
(1997)
Aoun-Habbache, M., M. Aoun, H. Berthiaux, and V. Mizonov (2002), “An
experimental method and Markov chain model to describe axial and radial mixing in
a hoop mixer.” Powder Technology 128, pp. 159-167.
Anthony, D. B., Howard, J. B., Hottel, H. C., and Meissner, H. P. (1975). "Rapid
devolatilization of pulverized coal." Fifteenth Symposium (International) on
Combustion, The Combustion Institute, Pittsburgh,.
Beckmann, M. Scholz, R. “Residence Time Behaviour of Solid Material in Grate
Systems,” Proc. INFUB 2000, 5th European Conference on Industrial Furnaces and
Boilers, Porto, Portugal, 11.-14. 04. 2000.
Bowen, William A. Digital Atlas of NYC. California Geographical Survey (1997),
http://130.166.124.2/ny_3.html (accessed June 6, 2004)
Beckmann, M. and Scholz, R., Simplified mathematical model of combustion in
stoker systems. Third European Conference Industrial Furnaces Boilers 18/21 (1995),
pp. 61–70.
165
Beckmann, M., Scholz, R., “Residence Time Behaviour of Solid Material at Grate
Systems.” Proc. INFUB 2000, 5th European Conference on Industrial Furnaces and
Boilers, Porto, Portugal, 11.-14. 04. (2000).
Bremaud, P. (1991): “Markov Chains Gibbs Fields, Monte Carlo Simulation, and
Queues.” Springer.
Broaddbent, S. R., Hammersley, J. M. “Percolation Processes I. Crystals and Mazes,”
Proceedings of the Cambrigde Philosophical Society 53 (1957), 629-641
Badzioch, S., and Hawksley, P. G. W. (1970). "Kinetics of Thermal Decomposition
of Pulverized Coal Particles." Ind. Eng. Chem, 9, 521.
Bruch, C., Peters, B., and Nussbaumer, T. (2003). "modelling wood combustion
under fixed bed conditions." Fuel, 82, 729-738.
Campbell, G., M., Webb, C. (2001). On predicting roller milling performance Part I:
the Breakage equation. Powder Technology 115 pp. 234-242
Chejne, F., Hernandez, J. P., Florez, W. F., and Hill, A. F. J. (2000). "Modelling and
simulation of time-dependent coal combustion processes in stacks." Fuel, 79, 987-997.
Chern, J. S., and Hayhurst, A. N. (2004). "Does large coal particle in a hot fluidised
bed lose its volatile content according to the shrinking core model?" Combustion and
Flame, 139, 208-221.
Clustan Ltd. Clustan software for cluster analysis for solving classification problems
http://www.clustan.com/
166
Cho, K. W., Park, H. S., Kim, K. H., Lee, Y. K., and Lee, K. H. (1995) “Estimation
of the heating value of oily mill sludges from steel plant, ” Fuel Vol. 74 No. 12, pp.
1918-1921
Essenhigh, R. H. and Kuo, T. J. (1970): “Combustion and Emission Phenomena in
Incinerators: Development of Physical and Mathematical Models of Incinerators,
Part1: Statement of the problem.” Proceedings of the National Incinerator Conference,
ASME, New York, pp. 261-271.
ECOMB AB, The Ecotube system for optimizing combustion processes and
principally, http://www.ecomb.se/
Everson, R., Heomagus, H., and Kaitano, R. (2005). "The modeling of the
combustion of high-ash coal-char particles suitable for pressurised fluidized bed
combustion: shrinking reacted core model." Fuel, 84, 1136-1143.
Fluent Inc., Fluent software, commercial computational fluid dynamics (CFD)
http://www.fluent.com/
Gaur, S., and Reed, T. B. (1998). Thermal Data for Natural and Synthetic Fuels,
Marcel Deeker, Inc.IWSA (International Solid Wastes Association), Energy from
Waste, State-of-the Art Report, www.wte.org.
Gohlke, O., and Busch, M., “Reduction of combustion by-products in WTE plants:
O2 enrichment of underfire air in the MARTIN SYNCOM process,” Chemosphere 42
(2001) 545-550
167
Gbor, P. K. Jia, C. Q. (2004). Critical evaluation of coupling particle size distribution
with the shrinking core model, Chemical Engineering Science 59 pp. 1979-1987
Grimmett G., “Percolation” Springer (1991)
Heidenreich, C. A., Yan, H. M., and Zhang, D. K. (1998). "Mathematical modelling
of pyrolysis of large coal particles - estimation of kinetic." Fuel, 78, 557-566.
Homma, S., Ogata, S., Koga, J., and Matsumoto, S. (2005). "Gas-solid reaction model
for a shrinking spherical particle with unreacted shrinking core." Chemical
Engineering Science, 60, 4971-4980.
Hwang, C. L., and Hogg, R. (1980). "Diffusive mixing in flowing powders." Powder
Technology, 26(1), 93-101.
Japanese Advanced Environment Equipment: Mitsuhishi Martin Refuse Incinerator:
http://nett21.gec.jp/JSIM_DATA/WASTE/WASTE_3/html/Doc_435.html
Jentsch, T., Beckmann, M., Davidovic, M., and Biollaz, S. (2003). "Investigation of
Heavy Metal Release during Thermal Waste Treatment on a Forward-Acting Grate by
Means of Radiotracers." Chemical Engineering & Technology, 26(6), 691 - 696.
Kaufman, S. M., Glodstein, N., Millarth K., Themelis, N.J., “The State of Garbage in
America,” BIOCYCLE, January (2004), 31-41
Kira et al., Development of New Stoker Incinerator for Municipal Solid Wastes Using
Oxygen enrichment, Mitsubishi Heavy Industiries Technical Review, Vol.38 No.2
(2001) p.78
168
Klasen, T. and Goerner, K. “Numerical Calculation and Optimization of a Large
Municipal Solid Waste Incinerator Plant,” IFRF Combustion Journal, February (2002)
ISSN 1562-479X
Kobayashi, H., J. B. Howard, and Sarofim, A. F. (1977). "Coal Devolatilization at
High Temperature." Int. Symp. on Combust., Combustion Institute, Pittsburgh, 411
Kunii, D., and Levenspiel, O. (1991). Fluidization Engineering, John Wiley & Sons,
Inc., Butterworth, Boston, MA.
Levenspiel, O. (1999). Chemical reaction engineering, John Wiley & Sons, Inc.
LECO Co. Ultimate analysis of the elements C, H, N, S and O
Downloadable literature http://www.leco.com/organic/elementalanalyzers/chns.htm
Lim, C. N., Goh, Y. R., Nasserzadeh, V., Swithenbank, J., Riccius, O, “The
modelling of solid mixing in municipal solid incinerators,” Powder Technology 114
(2001) 89-95
Martin GmbH: The MARTIN reverse-acting grate
http://martingmbh.de/englisch/index2.htm
Mohabuth, N., and Miles, N. The recovery of recyclable materials from Waste
Electrical and Electronic Equipment (WEEE) by using vertical vibration separation,
Resources Conservation & Recycling 45, 60-69 (2005)

169
Nakamura, M., Zhang, H., Millrath, K., and Themelis, N. J. ( 2003). "Modeling of
Waste-to-Energy Combustion with Continuous Variation of the Solid Waste Fuel."
2003 ASME ICMEE, Washington, D.C.
Nakamura, M., and N. J. Themelis, "Modeling of Solid Waste Flow and Mixing on
the Traveling Grate of Waste- to-energy Combustion Chambers," Proc. 12
th
annual
North American Waste To Energy Conference (NAWTEC12), Georgia. (2004)
Nakamura, M., Castaldi, M.J., and Themelis, N.J. (2005). Measurement of Particle
Size and Shape of New York City Municipal Solid Waste and Combustion Residues
Using Image Analysis," Proc. 16th Japan Society of Waste Management Experts
(JSMWE) Fall Conference, pp. 1-3, Sendai, Japan
Nakamura, M., Castaldi, M.J., and Themelis, N.J., (2008)"A 2-dimensional stochastic
modeling for Municipal Solid Waste (MSW) particle mixing within a waste-to-energy
(WTE) combustion bed" submitted to International Journal of Thermal Sciences
Petek J., Experimentelle Untersuchung der Pyrolyse in inerter und reaktiver
Atmosphare unter den Bedingungen der Wurfbeschickung, PhD thesis, TU Graz,
(1998)
Peters, B., Dziugys, A., Hans Hunsinger, and Krebs, L. (2005). "An approach to
qualify the intensity of mixing on a forward acting grate." Chemical Engineering
Science, 60(6), 1649-1659.
Peters, B. (1994). "A model for numerical simulation of devolatilization and
combustion of waste material in packed beds." KERNFORSCHUNGSZENTRUM
KARLSRUHE, KARLSRUHE (GERMANY).
170
Peters, B., Schroder, E., Bruch, C., and Nussbaumer, T. (2002). "Meaturements and
particle resolved modelling of heat-up and drying of a packed bed." Biomass and
Bioenergy, 23, 291-306.
Reindorf, T., and Vodegel, S., “Dynamic Grate-Model as a Tool for the Optimization
of Thermal Waste Treatment Plants,” CUTEC-Institut GmbH, Germany
Ruf, John Adam, “Particle Size Spectrum and Compressibility of Raw and Shredded
Municipal Solid Waste,” The University of Florida, Ph.D. Thesis (1974)
Simsek, E., Brosch, B., Wirtz S., Scherer, V., Krüll, F., “Numerical Simulation of
Solid Fuel Conversion on a Forward Acting Grate with a Coupled CFD / Discrete
Element Method (DEM)” INFUB-8th European Conference on Industrial Burners and
Furnaces, Vilamoura, Portugal, March 25-28, 2008-03-27
Smoot, L. Douglas, and Smith, Philip J., “Coal combustion and gasification,” The
Plenum Chemical Engineering Series, Plenum (1985)
Sohn, H. Y. (1973). "On the maximum temperature rise in gas-solid reactions."
AIChE Journal, 19(1), 191
Tchobanoglous, G., Theisen, H. and Vigil, S. “Integrated Solid Waste Management,”
McGraw-Hill, New York (1993)
Themelis, N. J., Kim, Y. H. and Brady, M. H. “Energy Recovery from New York
City Solid Wastes.” ISWA Journal of Waste Management and Research (2002): 20:
223-233
171
Takatsudo, Y., Nakamura, N., Ono, H., Mitsuhashi, M., Kira, M. “Advanced
Automatic Combustion Control System for Refuse Incineration Plant.” MHI
Technique Vol. 39, No. 2 (1999-5) in Japanese
Tchobanoglous, G., H. Theisen, and S. Vigil, S (1993): “Integrated Solid Waste
Management.” McGraw-Hill, New York, NY.
Von Roll, Inc., Brochure of Waste-to-energy Combustion System,
http://www.vonrollinc.com/
Wolf, C., “Erstellung eines Modells der Vermischung von Abfall auf Rostsystemen
unter besonderer Berucksichtigung der Vermischung - ein Beitrag zur Simulation von
Abfallverbrennungsanlagen”, Doctoral Dissertation, the Fraunhofer Institute for
Environmental, Safety and Energy Technology (UMSICHT) (2005)
Yagi, S., Kunii, D., Studies on combustion of carbon particles in flames and fluidized
beds. In: Fifth Symposium (International) on Combustion Reinhold, New York, pp.
231-244 (1955).
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.; Nasserzadeh, V.; Goodfellow, J.; Swithenbank, J. Simulation of channel
growth in a burning bed of solids. Chemical Engineering Research and Design, v 81,
n 2, p 221-232, 2003.
172
Yang Y.B., Lim C.N., Goodfellow J., Sharifi V.N., Swithenbank J., A diffusion
model for particle mixing in a packed bed of burning solids, Fuel, 84, 213-225.
(2005).
Yang, Y. B., Sharifi, V. N., Goh, Y. R., and Swithenbank, J. (1999). "User's Manual,
The Fluid Dynamic Incinerator Code (FLIC) For Modelling Incinerator Bed
Combustion."




173
Appendices


Appendix A: Bed calculation by FLIC*


* Data from a commercial reverse acting grate indicates the peak temperature is located in the beginning of grate
more than FILC calculation results, because of the effect of mixing agitated by the motion of the reverse acting
grate.
Thu Nov 14 11:31:52 2002
FLIC (Version1.1) - SUWIC
PROJECT
NewProject
Solid Temperature [K]
20 hr 00 min 00 sec 000 msec
359
295
423
487
551
615
679
743
807
871
935
1000
1.06e+003
1.13e+003
1.19e+003
1.26e+003
1.32e+003
1.38e+003
1.45e+003
1.51e+003
0 866.75 1733.5 2600.3 3467. 4333.8 5200.5 6067.3 6934.
678.28
( in mm )
1356.6
2034.8
2713.1
3391.4
4069.7
Temp, K
Thu Nov 14 11:33:26 2002
FLIC (Version1.1) - SUWIC
PROJECT
NewProject
Combustion Heat Release [kW/m^3]
20 hr 00 min 00 sec 000 msec
479
0.
958
1.44e+003
1.92e+003
2.4e+003
2.87e+003
3.35e+003
3.83e+003
4.31e+003
4.79e+003
5.27e+003
5.75e+003
6.23e+003
6.71e+003
7.19e+003
7.67e+003
8.15e+003
8.62e+003
9.1e+003
0 866.75 1733.5 2600.3 3467. 4333.8 5200.5 6067.3 6934.
678.28
( in mm )
1356.6
2034.8
2713.1
3391.4
4069.7
c) Combustion
d) Temperature
Figure A-1: Calculation results of Bed modeling (Nakamura et al.2002)
a) Water evaporation
b) Volatilization
Thu Nov 14 11:32:25 2002
FLIC (Version1.1) - SUWIC
PROJECT
NewProject
Residual Water in Solid (mass %)
20 hr 00 min 00 sec 000 msec
1.17
0.
2.35
3.52
4.69
5.87
7.04
8.22
9.39
10.6
11.7
12.9
14.1
15.3
16.4
17.6
18.8
20.
21.1
22.3
0 866.75 1733.5 2600.3 3467. 4333.8 5200.5 6067.3 6934.
678.28
( in mm )
1356.6
2034.8
2713.1
3391.4
4069.7
Thu Nov 14 11:35:31 2002
FLIC (Version1.1) - SUWIC
PROJECT
NewProject
Volatile Release [kg/m^3.hr]
20 hr 00 min 00 sec 000 msec
106
0.
213
319
425
532
638
745
851
957
1.06e+003
1.17e+003
1.28e+003
1.38e+003
1.49e+003
1.6e+003
1.7e+003
1.81e+003
1.91e+003
2.02e+003
0 866.75 1733.5 2600.3 3467. 4333.8 5200.5 6067.3 6934.
678.28
( in mm )
1356.6
2034.8
2713.1
3391.4
4069.7


174
Appendix B: Chamber Calculation by CFD
Figure B-1: A grid structure of CFD modeling (Nakamura et al.2002)



MSW
Ash
Flue
gas
Air/recirculated flue gas
injection
Air injection
Gas flow
(Computed from a bed model)


175

Figure B-2: Calculated contours of Velocity (m/s) (top) and Total temperature (K) (bottom)


176

Figure B-3: Calculated contours of Turbulence intensity (top) and mass fraction of hydrocarbon concentration (bottom)
((bottom)


177
Figure B-4: Calculated mass fraction of oxygen concentration (top) and of CO2 concentration (bottom)

1
7
8
A
p
p
e
n
d
i
x

C
:

C
o
m
p
a
r
i
s
o
n

o
f

c
o
m
b
u
s
t
i
o
n

m
o
d
e
l
s


T
a
b
l
e

C
-
1
:

C
o
m
p
a
r
i
s
o
n

o
f

c
o
m
b
u
s
t
i
o
n

m
o
d
e
l
s
D
e
v
e
l
o
p
e
r

s

N
a
m
e

M
o
d
e
l

N
a
m
e

F
u
e
l

T
y
p
e

(
M
S
W
,

C
o
a
l
,

W
o
o
d
)

S
o
l
i
d

R
e
a
c
t
i
o
n
,

(
D
r
y
i
n
g
,

D
e
v
o
l
a
t
l
i
z
a
t
i
o
n
,

C
h
a
r

o
x
i
d
a
t
i
o
n

)

G
a
s

R
e
a
c
t
i
o
n

V
o
l
a
t
i
l
e

c
o
m
b
u
s
t
i
o
n

H
e
a
t

/
M
a
s
s

T
r
a
n
s
f
e
r

R
e
m
a
r
k
s

R
e
f
e
r
e
n
c
e
s

A
h
m
e
d

e
t

a
l
.



C
e
l
l
u
l
o
s
e

p
y
r
o
l
y
s
i
s


[
C
e
l
l
u
l
o
s
e



v
o
l
a
t
i
l
e
s
]


H
e
a
t

(
c
o
n
v
e
c
t
i
o
n
,

r
a
d
i
a
t
i
o
n

)

t
r
a
n
s
f
e
r

1
-
d
e
m
e
n
s
i
o
n


(
A
h
m
e
d

e
t

a
l
.

1
9
8
9
)

B
e
c
k
m
a
n
n

e
t

a
l
.



M
S
W

G
e
n
e
r
a
l

G
a
s
-
s
o
l
i
d

r
e
a
c
t
i
o
n


H
e
a
t

a
n
d

M
a
s
s

t
r
a
n
s
f
e
r

C
o
m
b
i
n
a
t
i
o
n

o
f

s
e
v
e
r
a
l

c
o
n
t
i
n
u
o
u
s
l
y

s
t
i
r
r
e
d

r
e
a
c
t
o
r
s

(
B
e
c
k
m
a
n
n

e
t

a
l
.

1
9
9
5
)

S
h
i
n

e
t

a
l



W
o
o
d
,

C
o
k
e

D
r
y
i
n
g
,

p
y
r
o
l
y
s
i
s
,

o
x
i
d
a
t
i
o
n

[
W
o
o
d



c
h
a
r

+

v
o
l
a
t
i
l
e
s
]


H
e
a
t

(
r
a
d
i
a
t
i
o
n
)

a
n
d

M
a
s
s

t
r
a
n
s
f
e
r


F
u
e
l

s
h
r
i
n
k
a
g
e

s
u
b
-
m
o
d
e
l

a
n
d

a

t
w
o
-
f
l
u
x

r
a
d
i
a
t
i
o
n

s
u
b
-
m
o
d
e
l

(
S
h
i
n

e
t

a
l

2
0
0
0
)

Y
a
n
g

e
t

a
l
.


F
L
I
C

M
S
W

G
a
u
s
s
i
a
n

d
i
s
t
r
i
b
u
t
i
o
n

o
f

a
c
t
i
v
a
t
i
o
n

e
n
e
r
g
i
e
s

f
o
r

w
a
s
t
e

d
e
v
o
l
a
t
i
l
i
z
a
t
i
o
n
,



m
i
x
i
n
g

e
f
f
e
c
t

b
e
t
w
e
e
n

v
o
l
a
t
i
l
e

m
a
t
t
e
r

a
n
d

o
x
i
d
a
n
t

i
n

t
h
e

v
o
l
a
t
i
l
e

c
o
m
b
u
s
t
i
o
n

r
a
t
e
s


H
e
a
t

(
c
o
n
v
e
c
t
i
o
n
,

r
a
d
i
a
t
i
o
n
)

a
n
d

M
a
s
s

t
r
a
n
s
f
e
r

(
d
i
f
f
u
s
i
o
n
)

T
h
e

g
o
v
e
r
n
i
n
g

e
q
u
a
t
i
o
n
s

f
o
r

m
a
s
s
,

m
o
m
e
n
t
u
m
,

a
n
d

h
e
a
t

t
r
a
n
s
f
e
r

f
o
r

b
o
t
h

g
a
s
e
s

a
n
d

s
o
l
i
d
s
.

(
Y
a
n
g

e
t

a
l
.

1
9
9
9
,

2
0
0
2
)

P
e
t
e
r
s
,

P
e
t
e
r
s

e
t

a
l
.



W
o
o
d

D
u
r
i
n
g

d
r
y
i
n
g


a
n
d

d
e
v
o
l
a
t
i
l
i
z
a
t
i
o
n

t
h
e

d
i
a
m
e
t
e
r

i
s

a
s
s
u
m
e
d

t
o

r
e
m
a
i
n

c
o
n
s
t
a
n
t
.

=

=
3
1
,
)
e
x
p
(
k
k
i
a
o
c
R
T
E
k
r
&

N
/
A

H
e
a
t

t
r
a
n
s
f
e
r

(
c
o
n
v
e
c
t
i
o
n
,

r
a
d
i
a
t
i
o
n
,

c
o
n
d
u
c
t
i
o
n
)

M
a
s
s

t
r
a
n
s
f
e
r

(
d
i
f
f
u
s
i
o
n
)

S
i
n
g
l
e
-
p
a
r
t
i
c
l
e

m
o
d
e
l

P
a
c
k
e
d
-
b
e
d

m
o
d
e
l

(
i
n
c
l
u
d
i
n
g

s
h
a
p
e

f
a
c
t
o
r
)

S
I
M
P
L
E
R

a
l
g
o
r
i
t
h
m

C
o
n
n
e
c
t
i
v
i
t
y

w
i
t
h

o
t
h
e
r

p
a
r
t
i
c
l
e
s
,


T
h
i
e
l
e

m
o
d
u
l
u
s

(
P
e
t
e
r
s

1
9
9
4
,

P
e
t
e
r
s

e
t

a
l
.

2
0
0
5
)

S
o
h
n


G
e
n
e
r
a
l

G
e
n
e
r
a
l

G
a
s
-
s
o
l
i
d

r
e
a
c
t
i
o
n


H
e
a
t

t
r
a
n
s
f
e
r


(
S
o
h
n

1
9
7
3
)

C
h
e
r
n

a
n
d

H
a
y
h
u
r
s
t


C
o
a
l

D
e
v
o
l
a
t
l
i
z
a
t
i
o
n

[
C
o
a
l



c
h
a
r

+

v
o
l
a
t
i
l
e
s
]

N
/
A

H
e
a
t

t
r
a
n
s
f
e
r

(
E
x
a
m
i
n
e
d

f
r
o
m

7
5
0

t
o

9
5
0

C
)

S
h
r
i
n
k
i
n
g
-
c
o
r
e

M
o
d
e
l

(
C
h
e
r
n

a
n
d

H
a
y
h
u
r
s
t

2
0
0
4
)

C
h
e
i
j
n
e

e
t

a
l
.


C
o
a
l

D
r
y
i
n
g

[
W
e
t

c
o
a
l



c
o
a
l

+

H
2
O
]

D
e
v
o
l
a
t
i
l
i
z
a
t
i
o
n

[
V
o
l
a
t
i
l
e
s



B
1
C
O
2

+

B
2
C
O

+

B
3
O
2

+

B
4
N
2

+

B
5
H
2
O

+

B
6
H
2

+

B
7
C
H
4

+

B
8
S
O
2

+

B
9
N
O

+

B
1
0
C
2
H
6

+

B
1
1
H
2
S

+

B
1
2
N
H
3

+

B
1
3
t
a
r
]

C
h
a
r

O
x
i
d
a
t
i
o
n

[
C
o
a
l

+
α
O
2



(
2
β
-
1
)
C
O
2

+

(
2
-
2
β
)
C
O

+
a
1
H
2
O

+

a
2
N
O

+

a
3
S
O
2
]

C
O

+

H
2
O



C
O
2

+

H
2
O

2
C
O

+

O
2


2
C
O
2

2
H
2

+
O
2



2
C
O
2

C
H
4

+

2
O
2



4
C
O
2

+

6
H
2
O

4
N
H
3

+

5
O
2


4
N
O

+

6
H
2
O

H
e
a
t

a
n
d

M
a
s
s

t
r
a
n
s
f
e
r

P
a
c
k
e
d

b
e
d

m
o
d
e
l

a
n
d

S
h
r
i
n
k
i
n
g
-
c
o
r
e

M
o
d
e
l

(
o
n
l
y

f
o
r

c
h
a
r

o
x
i
d
a
t
i
o
n
)

(
C
h
e
j
n
e

e
t

a
l
.

2
0
0
0
)

H
e
i
d
e
n
r
e
i
c
h

e
t

a
l
.

D
i
s
t
r
i
b
u
t
e
d

A
c
t
i
v
a
t
i
o
n

E
n
e
r
g
y

M
o
d
e
l

(
D
E
A
M
)

C
o
a
l

(
l
a
r
g
e

p
a
r
t
i
c
l
e
)

V
o
l
a
t
i
l
e
s

g
e
n
e
r
a
t
i
o
n

b
y

D
E
A
M

(
)
| ¹ |
\ |
( ¸ (
¸
| ¹ |
\ |


×
=
| | ¹ |
\ |






=
d
r
r
d
E
E
f
d
t
t
r
R
T
E
k
X
R
V
V
V
R
t
o
j
j
n j
a
v
g
e
i
i
2
0
0
0
1
3 0
*
*
0
)
(
)
,
(
/
e
x
p
e
x
p
3
w
h
e
r
e

π
σ
σ
2
)
2
/
2
)
(
e
x
p
(
)
(
2
0
j
j
E
E
E
f


=

N
/
A

H
e
a
t

t
r
a
n
s
f
e
r

(
c
o
n
v
e
c
t
i
o
n
,

r
a
d
i
a
t
i
o
n
,

c
o
n
d
u
c
t
i
o
n
)

f
r
o
m

4
0
0

t
o

1
0
0
0
C

S
h
r
i
n
k
i
n
g
-
c
o
r
e

M
o
d
e
l

(
o
n
l
y

f
o
r

d
e
v
o
l
a
t
i
l
i
z
a
t
i
o
n
)

(
H
e
i
d
e
n
r
e
i
c
h

e
t

a
l
.

1
9
9
8
)

E
v
e
r
s
o
n

e
t

a
l
.


C
o
a
l

(
h
i
g
h

a
s
h

p
a
r
t
i
c
l
e
)

C
h
a
r

O
x
i
d
a
t
i
o
n

C
o
a
l

+

O
2


A
s
h

N
/
A

N
/
A
(
E
s
t
i
m
a
t
e
d

a
t

7
5
0
,

8
0
0
,

8
5
0
,

9
0
0
,

a
n
d

9
5
0

C
)

S
h
r
i
n
k
i
n
g
-
c
o
r
e

M
o
d
e
l

(
o
n
l
y

f
o
r


c
h
a
r

o
x
i
d
a
t
i
o
n
)

T
h
i
e
l
e

m
o
d
u
l
u
s

(
E
v
e
r
s
o
n

e
t

a
l
.

2
0
0
5
)



179
Appendix D: Movie files on CD-ROM

On the attached CD-ROM, there are 5 movie files: comb_chamber2_mpeg1,
sideview80_mpeg1, plugflow, large8bars, and small8bar. comb_chamber2_mpeg1 shows
the view from the end of a commercial WTE combustion chamber. sideview80_mpeg1 is
the movie taken from the side of the full-scale physical section model. plugflow shows
the simulated plugflow using the stochastic model. large8bar is the result of simulated
particle density profile for large particles. small8bar is the movie file that shows
simulated particle density profile for small particles.

Ⓒ2008 Masato Nakamura All Rights Reserved

Mathematical and Physical Modeling of Mixing and Flow Phenomena of Municipal Solid Waste Particles on a Reverse Acting Grate Masato Nakamura

ABSTRACT The dominant process for the recovery of energy from municipal solid wastes (MSW) is mass burning on a moving grate. Also, the leading grate technology is the Martin reverse acting grate that was modeled in this study. The drying, volatilization, and combustion phenomena during the travel of the solid waste fuel over the grate depend on the physical and chemical properties of the diverse particles in MSW, the design of the grate and combustion chamber, and the distribution of air flowing through the bed of solids. On the average, it takes one hour for particles to traverse a 10-meter long grate. Therefore, the bed behaves essentially as a fixed bed moving downwards, by gravity because of the inclination of the grate. In the case of the Martin grate, the reverse action motion of the grate results in some upward flow and consequent mixing of the materials within the bed.

This study of the flow and mixing phenomena on a reverse acting grate included both theoretical and experimental research. In order to characterize the heterogeneous particle behavior, a two-dimensional stochastic model of particle mixing within the bed on the reciprocating grate was developed. The model was calibrated and validated by means of a full-scale physical model of the reverse acting grate, using tracer particles ranging from 6 – 22 cm in diameter. This size range was

The motion of the reverse acting grate (whose speed ranged from 15 to 90 reciprocations/hour) increases the mean residence time of small and medium particles by 69 % and 8%. De at 60/hr. The ratio of particle diameter to the height of moving bar.46 to 1. Accordingly..4. In the future research. within this speed range. larger particles rise to the surface while smaller particles migrate to lower depths of the bed where the reciprocating bars push them backward against the main direction of the MSW flow. the traveling grate operation and moving bar height can be optimized in a more efficient way. the mixing diffusion coefficient. and the residence time. De. t.69. It was found that different particle sizes result in different residence times because of the Brazil Nut Effect (BNE). usually the mean diameter of MSW particles. decreases from 96 to 38. the mixing diffusion coefficient of each particle size was determined. the stochastic model simulation will become much . was found to be a major parameter for affecting the mixing diffusion coefficient. d/h. respectively. d/h. based on these quantitative results and the local MSW particle size distribution.based on a quantitative analysis of the size and shape factor distributions of MSW samples collected from different locations in New York City. Also. at reciprocation speeds above 30/hr. while decreasing the mean residence time of large particles by 19%. increases from 0. In the BNE. This indicates that the height of the moving bars should be grater than the diameter of targeted particles. when the stochastic model is combined with particle size change due to combustion. When the ratio.

the calculation should be carried out for each single time step (reciprocation).more complicated. . instead of only the previous step. because the Markov property will be applicable. In this case. since the initial condition.

vi …………………………………………………….... Combustion of volatile matter ………………………. Proximate and ultimate analysis ……..……………………..... Mathematical modeling for development of WTE technologies 1-7.………………………..………………………...v …………………………………………………………….. Physical and chemical properties of Municipal Solid Waste (MSW) and combustion residues (ash) ………………………………..……………………………….. Moisture evaporation in the drying process ……….xiv …………………………………………………1 Chapter 1: Introduction 1-1.…... The mass-burn system ……….….. Roller grate (Riley type) ……………………….. Sustainable waste management and current WTE technologies …………………..……………………………………………………. Other combustion systems ..………………………..…. The RDF combustion system ..21 1-5-1... Traveling grate systems in WTE chamber …………..10 1-3-3..14 1-4-3....…………………..18 1-5..27 i .……….8 1-3-1.Table of Contents List of Tables List of Figures Nomenclature Acknowledgment ………………………………………………………………...4 1-2-2..………..17 1-4-4.. Physical properties ..……….12 1-4-2....5 1-3..10 1-4. Outline of the dissertation …………………………3 1-2. Combustion of fixed carbon within the bed ………..21 1-5-2.22 1-5-3..xii . Shortcomings of existing bed models ..12 1-4-1.8 1-3-2.1 1-1-2. Volatilization in the gasification process ………………... Overview of waste-to-energy (WTE) technologies ……..24 1-5-4. Calorific values of different materials ……………….. Background …………………………………………………………1 1-1-1.25 . Forward acting grate (Von Roll type) ………………. Mass reduction ………………………………………....25 1-6.4 1-2-1. Physical and chemical transformations of MSW in the bed ……….5 1-2-3.………………….. Reverse acting grate (Martin type) ……….

.59 2-4-1.68 b) Image analysis ……………………………………….…………………………………………….……………68 a) Sieve analysis ………………………………………. Stochastic model ………………………………………. Equation describing mass and heat balance ………. Prior models of grate transport and combustion phenomena ..……71 a) Sample collection by field trip in New York City and by WTE facility visit ……………………….. Definitions of particle size and shape ………. Radiation between a gas and a particle ………………..61 2-4-3. Chamber modeling (volatile combustion) . Diffusion model ………………………………………... Introduction ... Heat transfer through the bed ……………………………….………68 3-1-1...38 2-1-2. Convection between a particle and the primary air though the bed ………………………………………………..69 3-2. Measurement of particle size and Shape Factor (SF) distributions ……………………………………….……. Shrinking-Core Model (SCM) ………………………. Diffusion-convection model ………………………...……63 2-4-5. Objectives of this study .1-8.. Methodology …………………….. Solids mixing theories ……………………………………….47 2-2-2.…….49 2-2-3..71 b) Sample measurement by image analysis ……….71 3-2-1...69 3-1-2..Study of particle size and shape …………..………………………………………31 Chapter 2: Mathematical Models .……..45 2-2. Bed modeling (solid waste combustion) .64 Chapter 3: Experimental Work ………………………………………68 3-1..50 2-2-4..……………….... Study of mixing and flow …………………....53 2-3-1.38 2-1-1.73 3-2-2. Radiation among walls and particles ……………….. Conduction with two particles ………………….. Convection model ………………………………………...62 2-4-4.46 2-2-1..……………….…….60 2-4-2. Progressive-Conversion Model (PCM) ………………..………………38 2-1......75 ii ..53 2-3-2.52 2-3..……………. Mass and volume reduction model ………………………………..………….54 2-4..

Probabilities from full-scale physical model ………115 5-2.………………………………83 Chapter 4: Mathematical Work ………………………………………. Mass and volumetric flow of solid (Monte Carlo simulation results) . Mass and volumetric flow of solids and air on the grate .... Introduction ...……...………………………………84 4-1-2..112 5-1-1. Particle size and shape distributions ………………112 5-1-2. MSW flow in the chamber . Results from size reduction model …..146 Chapter 6: Conclusions ………………………………………………………159 iii ..144 5-2-4.. Stochastic model for MSW mixing combined with a flow …..………….85 4-2.. One-dimensional stochastic model for the reverse acting grate ……………….105 Chapter 5: Results and Discussion .. Methodology ………………..………...……..84 4-1.………….. Stochastic model for MSW particles mixing on the traveling grate ……………………………………….132 5-2-2.....……... Size (mass and volume) reduction model ………………………100 4-3-1..…..... Expansion to a two-dimensional stochastic mode . Monte Carlo modeling for MSW feeding and flow . Integrating model for mixing and flow with data from the full-scale physical model ………..……………………………..……78 3-3-1.. Physical modeling ………………………………………. Modeling .………..…………….……………. Simulation results from the stochastic model for MSW mixing ………………………………………………135 a) One-dimensional simulation ………………135 b) Two-dimensional simulation ………………138 5-2-3.…………………………….98 4-3.………. Results from mathematical work ……………….3-3.……78 a) Geometry of physical model ………………………......130 5-2-1..…….. 90 4-2-1..80 c) Loading NYC-MSW particles ……………………….100 4-3-2..84 4-1-1....103 4-4..112 5-1. Results from experimental work …………….91 4-2-2.78 b) Making tracers in different size ………………………..………….……………………………………..……..83 d) Operation conditions .

References Appendices ………………………………………………………………164 ………………………………………………………………173 ………………………………173 ………………………174 ………………178 ………………179 Appendix A: Bed calculation by FLIC Appendix B: Chamber calculation by CFD Appendix C: Comparison of combustion models Appendix D: Movie files on CD-ROM iv .

Tchobanoglous et al. Smoot and Smith 1985) Table 1-5: Volatile products of MSW Table 1-6: Composition of volatile matters Table 3-1: Various definitions of particle diameters (Allen 1997) Table 4-1 Ultimate analysis of dry stream municipal solid wastes in New York City (Themelis et. Smoot and Smith 1985) Table 1-4: Typical ultimate analysis data of MSW and Coal (Tchobanoglous 1993. and roller grate systems Table 1-3: Typical proximate analysis data of MSW and Coal (Tchobanoglous 1993. 1993) Table 4-2: Assumed values of probabilities satisfying Eqs (4-10).al 2002.List of Tables Table 1-1: Reported WTE Capacity in Europe (IWSA 2003) Table 1-2: Comparison of reverse acting grate. forward acting grate. (4-11) and (4-12) for the transition matrix used in this calculation Table 5-1: Particle size parameters obtained from residential MSW and combined ashes in NYC Table 5-2: Mixing Diffusion coefficient with different densities [full-scale test] (Yang et al 2005) Table C-1: Comparison of combustion models v .

B. 5. Takuma) Figure 1-4: Schematic diagram of rotary kiln system (ref. whether a particle is located at the surface or bottom of the bed) Figure 1-18: Mass balance of a MSW bed in the combustion chamber vi . b) primary air (arrows) injected through the grate openings.g.8mg and 180 mg (S. Gaur and T. 2002) Figure 1-14: Combustion and transport phenomena of one particle in a MSW bed on the traveling grate Figure 1-15: Trend of current bed models Figure 1-16: Schematic Diagram of Integrated Modeling Figure 1-17: MSW flow and mixing in the reverse acting grate: The MSW particles are subjected to different temperatures and their size and composition change with time and location (e. al. Reed 1998) Figure 1-13: Comparison of experimental heating values of various waste materials (Themelis et..List of Figures Figure 1-1: Mass-burn grate combustion system (Von Roll. MHI) Figure 1-5: Schematic Diagram of Reverse Acting Grate (Martin Grate) Figure 1-6: Fixed and moving bars of Reverse acting grate (Martin): The reciprocating bars move against the direction of the feed by gravity: a) geometry of fixed and reciprocating bars. Ruf 1974) Figure 1-10: Waste-To-Energy Combustion System Figure 1-11: Effects of sample size and heating rate Figure 1-12: Sample Size effect of pelletized Baker Cellulose. Brochure of Combustion System) Figure 1-2: Combustion Chamber in a RDF system (Themelis 2003) Figure 1-3: Schematic diagram of fluidized bed system (ref. Figure 1-7: Schematic Diagram of Forward Acting Grate (Von Roll type) Figure 1-8: Schematic Diagram of Roller Grate (Duesseldorf/Babcock Grate) Figure 1-9: Particle size distribution in FL (J.

1989) Figure 2-2: Combustion Steps of a Stoker Firing Process and Model Idea (Beckman 1995) Figure 2-3: Schematic view of the Combustion in a Solid Fuel Bed (Shin et el. X marks are the locations that black bags were collected Figure 3-4: The image analyzer system. 2000) Figure 2-4: The volume distribution of the various defined components in the bed (Yang et al. 1999. 2002) Figure 2-5: The particle-based model using Discrete Element Method (Peters 1994. 1978) Figure 2-12: Progressive-Conversion Model (Levenspiel 1999) Figure 2-13: Shrinking-Core Model (Levenspiel 1999) Figure 2-14: Shrinking-Core Model without ash formation Figure 2-15: Combustion and transport phenomena of one particle in a MSW bed on the traveling grate Figure 2-16: Radiation from particle i to all neighboring particles j Figure 2-17: Conduction between two neighboring particles (Peters. large 22cm) (left) and NYC-MSW particle size distributions vii .Figure 2-1: Discretization of the fuel bed (Ahmed et al. 2005) Figure 2-10: A grid structure of CFD modeling (Nakamura et al. 2005) Figure 3-1: Particle diameter and sieve size Figure 3-2: Physical Scale Modeling (Lim et al.2002) Figure 2-11: Formation of a striated mixture (Lai et.. 2001) Figure 3-3: New York City map with per capita income (Bowen 2004). width. Peters et al. perimeter. medium 14cm. B. al. outlining the shapes of particles regarding length. and area Figure 3-5: Digital camera images of MSW samples (left) and ash samples (right) Figure 3-6: Various types of diameters (Allen 1997) Figure 3-7: Full-scale physical section model of a combustion chamber: positions of CCD cameras (top) and geometries of reverse acting grate (bottom) Figure 3-8: Spherical tracers of different sizes (small 6cm.

although the paths of individual particles differ considerably. Figure 4-6: Illustration of the spatial disposition of cells on the reverse acting grate Figure 4-7: Particle movement by reciprocating and fixed bars Figure 4-8: Combustion and transport phenomena of one particle in a MSW bed on the traveling grate Figure 4-9: Particle size density function (Gamma distribution) for different values of α (=1. 40. 40. b) Sphericity viii . 4. 5) and constant β (= 5) Figure 4-10: MSW bed on the reverse acting grate of a WTE plant showing divided cells (mesh) for the stochastic simulation Figure 4-11: The elements of the flow matrix F Figure 4-12: Diagram of probabilities and directions of the flow for each cell of the MSW bed on the grate Figure 4-13: The elements of the transition matrix P Figure 5-1: Distributions by particle numbers of residential MSW and combined ashes in NYC a) Particle size distributions (PSDs). 60. 3. all particles in the MSW feed after combustion eventually arrive at cell 17. and 80 cm heights and angles of chamber bed decline: 15 degree Figure 3-10: Cell patterns of 20. Figure 4-5: Interchange of particles between successive cells (transition graph): The final cell of this model is cell 17(ash bin). 60.Figure 3-9: Side views and top views of NYC-MSW beds with 20. 2. 80 cm beds Figure 4-1: Two-dimensional square lattice site percolation model used in the solid waste on the combustion grate Figure 4-2: Channeling and break-up of MSW bed in the percolation model on the grate Figure 4-3: Percolation Model in the combustion chamber Figure 4-4: Illustration of the spatial disposition of cells on the reverse acting grate: Each cell represents either the reciprocating bar or the fixed bar.

d) Roundness distributions Figure 5-2: Observed particle movements and corresponding state diagram Figure 5-3: Observed particle movements and corresponding state diagram Figure 5-4: Observed particle movements and corresponding state diagram Figure 5-5: Observed particle movements and corresponding state diagram Figure 5-6: Observed particle movements and corresponding state diagram Figure 5-7: Observed particle movements and corresponding state diagram Figure 5-8: Observed particle movements and corresponding state diagram Figure 5-9: Observed particle movements and corresponding state diagram Figure 5-10: Observed particle movements and corresponding state diagram Figure 5-11: Observed particle movements and corresponding state diagram Figure 5-12: Observed particle movements and corresponding state diagram Figure 5-13: Observed particle movements and corresponding state diagram Figure 5-14: Measured probabilities that particles stay in the same position (cell) after one reciprocation of the moving bars Figure 5-15: Time series of continuous variation of MSW components by Monte Carlo simulation Figure 5-16: Calculation results of the MSW combustion on the Martin grate Figure 5-17: Combustion probability in each zone Figure 5-18: Transient phenomena during the combustion process Figure 5-19: Simulation results of the movement of a solid waste particle on the Reverse Acting Grate: Top graph: tracing a behavior of one particle on the bed. c) Aspect Raito distributions (ARDs). ix .distributions. the particle goes back and forth because of the movement of the reciprocating bar. In cells #11 and 12. Bottom graph: Visualization of the particle travel based on the results of the calculations shown on top graph. In cells #2 and 5 the particle seems to stay put for a while.

Change in tracer particles distribution profiles over grate with number of movements of reciprocating bars (n=5./hr x . dots: experimental data) Figure 5-24: Calculated particle path over the grate as a function of size: small particles (S). 80). medium. 5-29: C and F diagrams for large particles: Dimensionless exit concentration C (top) and Dimensionless cumulative concentration F (bottom) versus residence time (min) with different reciprocation speeds ranging from15 to 90 recip. (b) n = 10.(d) n = 40. Figure 5-27: C and F diagrams for small particles: Dimensionless exit concentration C (top) and Dimensionless cumulative concentration F (bottom) versus residence time (min) with different reciprocation speeds ranging from15 to 90 recip. 40. medium particles (M). Figure 5-21: Change in particle distribution profiles over grate with number of movements of reciprocating bars (a) n=5. (c) n=213 and (d) n=401 Figure 5-23: Particle size distributions (PSD) by particle numbers of residential MSW and combined ashes in NYC: (lines: estimated gamma distributions./hr Figure. and large particles with a reciprocation speed of 90 recip. (b) n=106./hr Figure 5-28: C and F diagrams for medium particles: Dimensionless exit concentration C (top) and Dimensionless cumulative concentration F (bottom) versus residence time (min) with different reciprocation speeds ranging from15 to 90 recip. 10.Figure 5-20.(e )n= 80 Figure 5-22: Movements (paths) of MSW particles on the Reverse Acting Grate: (a) n=67. (c) = 20. and large (L) particles Figure 5-25: Brazil Nut Effect (BNE) in a packed MSW bed Figure 5-26: C (top) and F diagrams (bottom) for small./hr. The regular perturbations in the n=40 and n=80 profiles are due to the effect of alternatively passing over stationary and reciprocating bars.

can simulate concept design for novel traveling grate systems Figure A-1: Calculation results of Bed modeling (Nakamura et al.2002) Figure B-2: Calculated contours of Velocity (m/s) (top) and Total temperature (K) (bottom) Figure B-3: Calculated contours of Turbulence intensity (top) and mass fraction of hydrocarbon concentration (bottom) ((bottom) Figure B-4: Calculated mass fraction of oxygen concentration (top) and of CO2 concentration (bottom) xi .2002) Figure B-1: A grid structure of CFD modeling (Nakamura et al. 1998) Figure 6-1: Stochastic models.Figure 5-30: Mixing Diffusion Coefficients versus grate reciprocation speed for different particle sizes Figure 5-31: Wood particle conversion time vs particle diameter (Petek J. calibrated by full scale grate tests.

Rr . α. L. ρ. L. H and θ to determine rf function of Rf ./hour) number of transition (number of the reciprocation of moving bars) ratio between feed rate of MSW and frequency of reciprocating bars (cells / reciprocation) = uf /Rr (how many cells MSW travels during one reciprocation) t residence time (min) mean residence time (min) h pi pr rr pf rf po ff gf fr gr Rf height of moving bars (cm) probability of remaining in state1 after one transition near the inlet probability of remaining in a state after one transition on the reciprocating bar probability of transiting to the adjacent state on the reciprocating bar probability of remaining in a state after one transition on the fixed bar probability of transiting to the adjacent state on the fixed bar probability of remaining in state16 after one transition near the outlet function of Rf . S. Rr . S. H and θ to determine pr function of Rf . Rr . H and θ to determine pf function of Rf . Rr . α. S. α. ρ. ρ.Nomenclature S(0) n S(n) initial state vector (initial distribution of MSW) number of transition (number of the stroke of reciprocating bars) state vector after n transitions (distribution of MSW after nth stroke of reciprocating bars) P F uf d De Rr n k transition matrix flow matrix feed rate of MSW in the inlet of the mass-burn WTE chamber (cm/min) diameter of MSW particle (cm) mixing diffusion coefficient (cm2/min) reciprocation speed of moving bars (recip. ρ. α. S. H and θ to determine rr feed rate of MSW in the inlet of the mass-burn WTE chamber xii . L. L.

S ρ Rr L H θ α particle size of MSW components particle density of MSW components frequency of reciprocating bars the length of the reciprocating bars travel height of the reciprocating bars angle of the reciprocating bars angle of chamber bed decline xiii .

Fabiola Brusciotti. I would like to acknowledge Professors Upmanu Lall. and my mother. I would like to express my deepest gratitude to Professors Nickolas J. Also. Dionel Albina. Dr. Dr. Ling. the Earth Institute (EI). and the Solid Waste Processing Division (SWPD) of the American Society of Mechanical Engineers (ASME). Rennee. I have special appreciation for Anuta Belova. Werner Sunk. Mark White. Finally. my father Kazuo Ryuwa Nakamura. and Vijay Modi for helping me as committee members of my dissertation defense.Acknowledgements First and foremost. Peter J. I would also like to recognize Liliana Themelis. Hoe I. Toyoko Nakamura for helping me along the whole way through. and Dr. Hanwei Zhang. thanks to my wife. Karsten Millrath. Castaldi for their support and advice throughout this research and guiding me to the right direction. Themelis and Marco J. Heidi C. Gary S. Eilhann Kwon. and Yue Zhou. Dr. Shang-Hsiu Lee. Frank S. Butterman. and Greg Epelbaum greatly helped me regarding experimental and modeling work. Dr. I am deeply indebted to these organizations. especially Dr. the Waste-to-Energy Research and Technology Council (WTERT). Scott Kaufman. Zeman. as well as my research colleagues: Dr. Discussion with the engineers and researchers of Covanta Energy Corporation. the Earth Engineering Center (EEC). Hill . Joseph Di Dio III. xiv . Ralf Koralewska from Martin GmbH introduced me useful theoretical research carried out by European WTE industry and academics. who organized a doctoral study group and encouraged me during my dissertation writing process. Yoshie Tomozumi-Nakamura. Federico Barrai. The financial support of several organizations made this research possible: Columbia University’s Department of Earth and Environmental Engineering (Henry Krumb School of Mines).

Recovery of energy. Effective collection and transportation of MSW from residential areas to waste management facilities or landfills are required for reducing the transportation cost. while minimizing the environmental impact due to final deposition of residual MSW such as landfilling. 2002) and the daily generation of MSW is estimated as 3. Landfilling is the disposal of MSW or the residuals from recycling and combustion processes. is the conversion of post-recycling MSW into energy though the combustion or gasification processes. WTE capacities in Europe are . transport. metals and glass. by means of recycling of paper. and processing. known as waste-to-energy (WTE).31 short ton per capita of Municipal Solid Waste (MSW) annually (Kaufman et al. Sustainable Waste Management attempts to optimize the energy and resource recovery from collection.1 Chapter 1: Introduction 1-1.6 kg per capita. These technologies are utilized in Sustainable Waste Management. Sustainable waste management and current WTE technologies The United States generates approximately 1. For example. Current WTE technologies are mainly used for the transformation of waste materials though combustion which reduces the volume and weight of residues for landfilling and recovers energy and metals. Materials separation and processing technologies are used in the recovery of materials. These advantages of WTE have increased WTE capacities throughout the world. Background 1-1-1. plastics.

000 2.472.000 322.303.000 109.000 1.484.000 558.000 22.895.000 2.000 Electric energy (gigajoules) 131.000 3. See Section 1-2-2 for detail.164. Country Austria Denmark France Germany Hungary Italy Netherlands Norway Portugal Spain Sweden Switzerland UK Total reported Tonnes/year (in 1999) 450.000 1.000 40.000 2.000 10.169.005.354.000 2.000 352.000 3.5 (average) Thermal energy (gigajoules) 3.934.000 2.039.000 1. etc. Total capacity in 1999 was more than 40 million tons annually and the generation of electrical and thermal energy were 41 million GJ and 110 GJ.000 2.. to what is called refuse-derived fuel (RDF) units. .000 399. by magnetic separation.000 12.2 shown in Table 1-1.053. respectively.000 Table 1-1 Reported WTE Capacity in Europe (IWSA 2003) Combustion of MSW is carried out using WTE units that are broadly divided into two classes: Those that combust bulk MSW as received (“mass burn”) and those that pre-process MSW.000 12.000 40.338.000 Kilograms /capita 56 477 180 157 6 137 482 49 32 26 225 164 18 154.000 32.000 1.698.000 27.000 4.761.074. shredding.000 1.311.000 1.130.000 1.000 2.000 1.409.000 10.000 220.853.360.000 27.984.000 8.550.996.636.000 9.562.000 4.543.042.190.818.

WTE technologies have advanced rapidly.” has been developed by ECOMB company in order to reduce NOx. 2001). which injects industrial oxygen in the primary air. These advanced technologies have started to be applied to practical use in current WTE facilities. A combustion improvement system with air injection tubes. and objectives of this research. This chapter introduces a background of the WTE technologies and MSW properties. Outline of the dissertation This dissertation consists of six chapters. Chapter 3 is about experimental work of this study. called “Ecotube. 2002).3 In recent years. like the thermal plasma torch ash melting system. have been developed to vitrify fly ashes and eliminate the chance of future contamination. CO. has been developed by Martin GmbH and Mitsubishi Heavy Industry in order to generate a better grade of ash and reduce NOx and dioxin concentrations (Kira et al. the oxygen-enriched combustion system. This technology was tested at a demonstration plant in Germany and is combined with a flue gas recirculation system in the Martin Syncom process (Gohlke and Busch 2001). 1-1-2. High temperature furnaces. shortcomings of existing bed models. and Volatile Organic Compounds (VOCs). For example. using image analysis for a . Chapter 2 gives mathematical models of WTE combustion including literature review of other related theoretical work. The “prism” system developed by IBB Engineering GmbH provides for additional secondary injection ports in the chamber with the objective of stabilizing stable temperature and gas velocity distributions in WTE units (Klasen et al.

Brochure of Combustion System) . Chapter 5 is results and discussion of both experimental and mathematical work. The mass-burn system The mass-burn system is used by most existing WTE facilities in the world. Chapter 4 describes stochastic modeling of the flow and mixing of particles on the traveling grate. A typical mass-burn WTE system is shown in Figure 1-1. Overview of WTE combustion systems (WTE units) 1-2-1. The final chapter concludes this research with its major finding and limitation. Abstract of this study is shown at the beginning of this dissertation. It consists of the following Figure 1-1: Mass-burn grate combustion system (Von Roll. including suggestions for future work.4 measurement of MSW properties and a full-scale physical section model of the reverse acting grate system. 1-2.

A typical combustion chamber of a . separation. and compaction before it is fed into the combustion chamber of a WTE facility.5 sections: 1) combustion chamber. The RDF combustion system Refuse-Derived Fuel (RDF) is produced from MSW by means of several processes. it is loaded into the charging hopper by a crane. MSW is brought in by collection vehicles and is “tipped” on the unloading bay in an enclosed building. There are several types of traveling grates and will be described in Section 1-3. and is pushed by feeders into the Figure 1-2: Combustion Chamber in a RDF system (Themelis 2003) combustion chamber. 1-2-2. From there. such as shredding. because the heating value of RDF is higher than that of MSW as-collected. 3) air pollution control systems (APC). The RDF system has a combustion chamber that is physically smaller than a mass-burn chamber. Each massburn combustion chamber is equipped with a traveling grate the motion of which slowly propels the MSW feed gradually down the grate. 2) energy recovery system (boiler).

which in this case is equipped with a horizontal traveling grate. This system is operated on a variety of fuels such as MSW. sludge. A fluidized bed system consists of a vertical steel cylinder and a sand bed shown in Figure 1-3. The rotary kiln. RDF is fed by a fuel distributor into the combustion chamber. 1-2-3. Other combustion systems There are other combustion systems that are less common in WTE: Fluidized beds and rotary kiln. and chemical wastes. . ranging from solid to liquid wastes. coal.6 RDF system is shown in Figure 1-2. shown in Figure 1-4. is often used to process medical wastes as well as various industrial wastes. Rotary kiln systems are also versatile and can treat a range of materials.

Takuma) Figure 1-4: Schematic diagram of rotary kiln system (ref.7 Figure1. MHI) .3: Schematic diagram of fluidized bed system (ref.

A typical operating frequency in a WTE unit is 12 strokes per hour but the frequency used depends greatly on the composition of the MSW feed. consists of a total of 15 bars and is inclined 26 degrees to the horizontal. Reverse acting grate (Martin type) A typical Martin reverse acting grate is 7 meters long. .42 m (420 mm). The bars are positioned at an angle of 13 degrees from the traveling bed.8 1-3. Inlet of Solid Waste RBn: reciprocating bar (step) FBn: fixed bar (step) FB1 RB1 26 Outlet of Ash Figure 1-5: Schematic Diagram of Reverse Acting Grate (Martin Grate) eight of the 15 bars are reciprocating and move a distance of 0. In a common reverse acting grate configuration. The frequency of motion of the reciprocating bars can be adjusted from 10 to 60 double strokes per hour. Traveling grate systems in WTE chambers 1-3-1.

2 kg per m3 (Tchobanoglous et al. 1993) .Therefore. The average density of MSW is 500 lb per yd3 or 173.1 metric tons per hour (WTE of 480 short tons per day capacity). the corresponding volumetric feed rate of MSW is 104. A typical ratio of the downward volumetric flow rate of MSW feed to the upwards flow of waste due to the motion of the reciprocating bars (volume of material pushed upward by the bar motion) is approximately 5/1.8 m3 per hour.9 Figure 3-3: Schematic Diagram of Forward Acting Grate (Von Roll type) Figure 1-6: Fixed and moving bars of Reverse acting grate (Martin): The reciprocating bars move against the direction of the feed by gravity: a) geometry of fixed and reciprocating bars. b) primary air (arrows) injected through the grate openings. . A typical feed rate of MSW into the combustion chamber is 18.

. Forward acting grate (Von Roll type) Inlet of Solid Waste RBn: reciprocating bars FBn: fixed bars FB1 RB1 30 Outlet of Ash Figure 1-7: Schematic Diagram of Forward Acting Grate (Von Roll type) Forward acting grate is made up of step grate type hearth as shown in Figure 1-7. 1-3-3. in which the moving bars and fixed bars are arranged alternately. The rate of revolution for each roller can be varied from 0 to 6 rph. Roller grate (Riley type) The roller grate (also called the Duesseldorf/Babcock Grate) consists of 6 rollers of 2 m diameter that are inclined at 30o to the horizontal as shown in Figure 18. As the rollers rotate slowly. the solid waste moves downward.10 1-3-2. Each roller “knead” and turn the solid waste. Moving bars reciprocate the same direction with the inlet flow of MSW.

Action on MSW bed Mechanical Push Friction force (shear stress) Reverse acting grate reverse Grate Types Forward acting grate forward Roller grate ---forward direction both directions both directions Table 1-2: Comparison of reverse acting grate. and roller grate systems . Forward and reverse acting grate systems are able to push MSW particles directly due to the motion of their reciprocating bars.11 Inlet of Solid Waste 30o Outlet of Ash Figure 1-8: Schematic Diagram of Roller Grate (Duesseldorf/Babcock Grate) Characteristics of these three types of traveling grates are summarized shown in Table 1-2. forward acting grate. whereas a roller grate conveys MSW particles because of the rotation of the rollers.

is a nonhomogeneous fuel. particle size. and compressibility. however. transforming during the combustion process. shown as Figure 1-9 (Ruf 1974). of which combustion process is more complicated. width and height) and size distributions was by Ruf. but it did not discuss the effect of shape or shape factor of MSW particles. There is no study defining the shape factor of solid waste particles. physical properties of MSW particles were measured in order to develop more efficient mass-burn chambers (see Section 32). particle shape. These are dominant factors that determine how solid waste particles travel on the moving grate. and. In this study. Figure 1-9: Particle size distribution in FL (J. since MSW. These are different from designs of coal chambers. Ruf 1974) . even though they govern movement and mixing on the grate as well as drying. The effect of sizes and shapes of solid waste particles. unlike coal. Physical properties Physical properties of MSW and WTE combustion ashes include density. The only comprehensive study of particle sizes (length. Physical and chemical properties of Municipal Solid Waste (MSW) and combustion residues (ashes) 1-4-1. have been less studied. and finally turning into ashes.12 1-4. combustion rates. volatilization.

permeability. segregation. Proximate and ultimate analysis Proximate Analysis of MSW is used to determine the percent of four components (moisture. These properties influence a particle flow through a WTE combustion chamber. heated. there are some important physical/physicomechanical characteristics that include agglomeration. and. packing. compressibility explosion. changing into ashes. pushed though the outlet of chamber.13 Besides these physical properties of particles. It is a material flow system and can be thought as a black box shown in Figure 1-10. bulk density. reactivity. flowability. eventually. In addition to this solid flow. 1-4-2. combustion. aggregation. fixed carbon. combusted. porosity. MSW is pushed by feeders. and ash) of MSW and consists of OUTPUT (Gas) Gas Emission WTE Combustion Chamber INPUT (Solid) Municipal Solid Wastes (MSW) INPUT (Gas) Air inlet OUTPUT (Solid) Ash (Fly and Bottom) Figure 1-10: Waste-To-Energy Combustion System . and shear behavior. there is the air and emission gas flows during the MSW combustion. cohesivity. volatile matter.

H. and inorganic ash of a sample. N (Nitrogen). a few mg of sample is heated to a temperature of 1000 o C and completely combusted in the furnace and N is measured using thermal conductivity detection. O (Oxygen). S (Sulfur). S is also measured by the infrared detector but heated from 600 to 1450 oC. and N. H (hydrogen). Data of ultimate analysis of MSW and coal are presented in Table 1-4. To measure weight percent of C. O in the sample is measured by the infrared detector in the pyrolysis furnace heated to 1300 oC (LECO 1999). Ultimate analysis is used to determine the elements C (carbon). These detected data are used to characterize the composition of the organic matter in MSW. .14 the following tests: • • Moisture (loss of moisture when heated to 105 oC for 1 hour) Volatile matter (additional loss of weight on ignition at 950 oC in covered crucible) • • Fixed carbon (combustible residue left after volatile matter is removed) Ash (residue after combustion in an open crucible) Table 1-3 shows typical proximate analysis data of MSW (Tchobanoglous 1993) and coal (Smoot and Smith 1985).

0 42.8 12.2 10.9 68.1 68.4 77.4 5. Metals.0 0.7 38.4 16.0 22.0 60.0 2.6 4. tin cans Metal.4 34.5 9.7 3.2 5.4 11.5 2.5 4.2 5. etc.2 2.5 7.7 33.9 90.7 87.0 12.0 <0.0 (40-60) 36.2 4.9 29.1 75. etc.5 98.1 86.0 1. Yard wastes Wood (green timber) Hardwood Wood (mixed) Glass.2 0.4 1.1 0.2 2.5 1.9 9.5 8.1 5.0 10.3 41. trees.7 8. nonferrous Residential MSW Coal Pittsburgh bituminous (high volatile) Illinois coal bituminous North Dakota lignite Wyoming subbituminous Kentucky bituminous Proximate analysis.8 5. % by weight Volatile matter Fixed carbon Ash (noncombustible) 0.3 2.0 0.1 6.9 66.0 7.8 98.2 10.0 50.5 3. Glass and mineral Metal.9 35.1 52.0 (4-15) 55.0 23.9 95.6 95. rubber.4 81.9 12.9 27.5 0.4 2.2 0.3 21.2 3.0 20.4 0.6 35.0 70.6 96-99 94-99 96-99 94-99 20.3 7.0 1.1 29.3 7.3 12.2 0.5 9.5 4.2 0.3 10.0 83.15 Types of Waste Moisture Food and food products Fats Food wastes (mixed) Fruit wastes Meat wastes Paper products Cardboard Magazines Newsprint Paper (mixed) Waxed cartons Plastics Plastics Polyethylene Polystyrene Polyurethane Polyvinyl chloride Textiles.0 78.0 (15-40) 2.0 10. ferrous Metal.0 11.3 7.2 0.0 2.8 8.4 0.5 0.4 4.0 5.5 30.8 17.5 Table 1-3: Typical proximate analysis data of MSW and Coal (Tchobanoglous 1993.0 1.0 (10-30) 6.5 66.5 32.1 6.0 2.0 21.4 46.6 56. Smoot and Smith 1985) . leather Textiles Rubber Leather Wood.3 75.

5 68.8 0.0 5.4 59.2 0.8 38.2 1.1 0.2 20.2 4.2 60.1 5.0 44.2 10.1 0.0 4.0 5.8 5. Metals.0 3.1 0. etc.1 0.3 1.0 11.2 14.5 6.2 8.5 59.1 - Ash 0.3 6.8 13.9 90.3 4.7 8.3 Table 1-4: Typical ultimate analysis data of MSW and Coal (Tchobanoglous 1993.7 76.4 8.9 49.2 0. Glass and mineral Metal.3 43.4 1.1 0.1 43.3 0.1 <0.6 0.2 3.0 3.0 69.6 39.2 0.0 48.9 5.1 0.3 0.1 0.8 4.0 46.3 5.7 45.9 5.3 7. Smoot and Smith 1985) .4 98.6 1.3 61.0 3.1 0.0 32.3 2.2 10.5 4.6 49.2 6.5 4.3 45.2 5.1 <0. leather Textiles Rubber Leather Wood.9 1.6 14.4 8.1 1.0 0.0 17.1 <0.2 42.1 2.1 0.7 44. (mixed) Coal Pittsburgh bituminous (high volatile) Illinois coal bituminous North Dakota lignite Wyoming subbituminous Kentucky bituminous 73.1 10.3 1.1 <0.5 0.2 9.6 43.1 63.1 0.0 0.1 22.5 Ultimate analysis.9 23.0 10.2 0.6 1.2 0.3 1.8 9.0 42.2 10.3 0.2 48.1 49.1 0.3 5.0 8.1 6.16 Types of Waste Carbon Food and food products Fats Food wastes (mixed) Fruit wastes Meat wastes Paper products Cardboard Magazines Newsprint Paper (mixed) Waxed cartons Plastics Plastics Polyethylene Polystyrene Polyurethane Polyvinyl chloride Textiles.3 30.4 0.4 4.4 0.0 50.0 85.5 0. rubber.7 60.2 0.2 0.6 40.2 87.0 1.5 77.0 6.6 38.4 4.4 6.3 6.8 37.0 48.0 6. trees. % by weight Hydrogen Oxygen Nitrogen Sulfur 11.3 0.5 1.5 48.0 6.1 <0.5 24.4 5.5 13.1 0.4 0.5 69.4 1.5 6.5 1.4 0.4 2.0 23.1 <0.9 1.4 6.0 6.1 <0.0 0.1 <0.4 6. Yard wastes Wood (green timber) Hardwood Wood (mixed) Wood chips Glass.6 1.6 43.6 6.1 <0. etc.8 4.

8mg and 180mg pelletized baker cellulose. These effects are dependent on heat transfer from the gas film surrounding the particle into the particle. Mass % Small Particle Big Particle Mass % Low Heating Rate High Heating Rate Temperature oC a) Effect of particle size Temperature oC b) Effect of heating rate Figure 1-11: Effects of sample size and heating rate . Figure 1-12 represents the size effect of mass reduction of 5. Mass reduction The mass and volume of MSW particles are reduced during the combustion process. The effects of heating rate and sample size in the thermal process are shown Figure 1-11 and 1-12 (Gaur and Reed 1998).17 1-4-3.

Themelis et. 1-13.18 120 100 80 TG % 60 40 20 0 0 200 400 Temp (C) 600 800 5. Calorific values of different materials The calorific value (or heating value) of the organic compounds in MSW are the amounts of energy released when the compounds are burned completely.8 mg 180 mg Figure 1-12: Sample Size effect of pelletized Baker Cellulose. the calorific value is called Lower Heating Value (LHV). 5. . B.8mg and 180 mg (S. al. al. 2002) and compared experimental heating values of various waste materials with the percentage of H2O (moisture) content shown in Figure. It can be determined using an oxygen bomb calorimeter. Reed 1998) 1-4-4. found that composition of organic matter in MSW can be represented by the molecular formula C6H10O4 (Themelis et. When vapor H2O is the product of combustion. it is called Higher Heating Value (HHV). If condensed H2O is the product of combustion. Gaur and T. The colorific value is dependent on the phase of water or steam in the combustion products.

al 1995): Figure 1-13: Comparison of experimental heating values of various waste materials (Themelis et.19 The calorific value of the MSW compounds can be determined using elemental composition quantified by ultimate analysis. : 2002) . al. Dulong’s equation had been used by engineers working in coal and wood s stoker facilities (Cho et. There are several formulae to calculate calorific values using the elemental compo composition of MSW.

al 1995): Hh = 81·(C-3/8·O) + 57·3/8·O + 342.5·(H -1/8·O) +22. W is water (%). 1-3) where Xcomb. For calorific calculation of MSW. Steuer’s equation has been used (Cho et. S is sulfur (%). C is carbon fraction (%). such as coal and wood.5·H + 22. and Xmetal are the fractions of RDF components. O is Oxygen (%). H is hydrogen (%). XH2O. 1-2) The heating value of RDF is expressed as follows: Hh = 18400·Xcomb -2636·XH2O -628·Xglass -544·Xmetal kJ/kg (Eq. Xglass.5·S + 57·3/4·O – 6·(9·H+W) (Eq.20 Hh = 81·C +342. . 1-1) where Hh is higher heating value. This equation provides good approximated heating values for fuels that mainly contain carbon.5S -6·(9·H +W) (Eq.

Combustion (char oxidization) Ash particle Ash particle size distribution Figure 1-14: Combustion and transport phenomena of one particle in a MSW bed on the traveling grate 1-5-1.21 1-5. as shown 1-14. are composed of three main processes: drying of the surface of solid waste particles. Physical and chemical transformations of MSW in the bed The MSW physicochemical transformations in the WTE combustion chamber. Devolatilization 3. Heat transferred by convection of the gases surrounding the waste particles and of the primary air injected from the . Drying (moisture evaporation) MSW particle size distribution 2. Moisture evaporation in the drying process The drying process occurs at the surface of solid waste particles just after MSW is fed into the combustion chamber. volatilization as the particles may soften and undergo internal transformation. MSW particle 1. the volatiles released from heated solid particles react with oxygen. and char combustion processes. In addition to these three processes. which is called volatile combustion.

The rate of solid waste evaporation can be expressed as: Rdrying = As hs (C H 2O . 1-9) where v is the fractional yield of volatiles (mass of volatiles per mass of daf coal). 1-8) v∞ = Q(1 − vc )v p (Eq.22 traveling grate into the MSW bed. Devolatilization in the gasification process Volatilization model of first-order reaction process (Badzioch and Hawksley 1970) dv = k ( v∞ − v ) dt (Eq. 1-7) k = A exp( − E ) RT (Eq. 1-4) Rdrying = Qccr / H evp . 1-5) 4 Qccr = As ( hs (T g − Ts ) + ε sσ (Tenv − Ts4 )) (Eq. when Ts > 100Co (Eq. when Ts < 100Co (Eq. s − C H 2O. 1-6) 1-5-2. . and v∞ is the hypothetical ultimate yield. g ) .

12x105 exp[-8961/T] (Badzioch and Hawksley 1970) (Eq. Temperature (K) 755 921 1088 1199 Volatiles (kg/kg waste) 0. 1976) (Eq.23 (a) k = 3. et al.237 0. reported data about the mounts of volatiles during decomposition processes of the MSW. 1985). 1-13) Tchobanoglous et al. 1-12) f (E) =  − ( E − E0 ) 2   exp 1   2σ 2   2 σ (2π ) 1 (Eq.244 Table 1-5: Volatile products of MSW . 1975): v∞ − v = v∞ 1 σ (2π ) 1 2  ∞ exp − ( ∞ kdt ) f ( E )dE   ∫0  ∫0     (Eq. 1-11) A continuous Gaussian distribution of activation energy is another approach to treat pyrolysis (Anthony et al.544/T] (Solomon.3x1014 exp[-27.186 0. 1-10) (b) k = 4.123 0. Table 1-5 and 1-6 shows volatile products and components of MSW with temperature gradient (Tchobanoglous et al.

24 Component (% vol. Combustion of volatile matter In the volatilization process.59 2.18 2.91 13.24 3.45 35. This complex process cannot be treated separately. assuming a representative hydrocarbon to be the only combustible product from the volatilization process.25 18.58 28. 1-14) 1 CO + O2 → CO 2 2 (Eq.03 Temperature (K) 921 1088 16.31 2.43 1. 1-15) 0 0 RCO = 1.77 0.12 31.06 0.26 10.49 34. 1-17) .) H2 CH4 CO CO2 C2H4 C2H6 755 5.25O C O.77 1199 9. the simplified reaction can be presented as follows: n m n C m H m +  + O2 → mCO + H 2 O 2  2 4 (Eq.8Tg P 0.07 Table 1-6: Composition of volatile matters 1-5-3.5 44.25 (Eq.25O CO2 exp  T  g   (Eq. a wide range of different volatile matters are generated. 1-16)  − 12200  0  RCm H n = 59. Therefore.3C H.45 3.43 33.3 × 10 11 C CO C H.73 30.78 20.56 12.55 15.

1-20) Eq. 1-18) − 6420 CO = 2500 ⋅ exp( ) CO 2 T (Eq. Combustion of fixed carbon within the bed Fixed carbon in a solid waste particle is combusted and described as following equations: C(solid) + αO2 → 2(1 − α )CO + (2α − 1)CO2 (Eq. 1-20 is char consumption rate expressed by Smoot and Pratt (Smoot and Pratt 1979) 1-6. as mentioned in Section 1-4. The transport and combustion phenomena of solid waste particles traveling on the grate systems are controlled by a large number of parameters that depend on the design of the combustion chamber (as explained in Section 1-2 and 1-3) and on the properties of the MSW particles (as described in Section 1-4 and 1-5). Mathematical modeling for development of WTE technologies Further development of WTE technologies will depend on a better understanding and application of mass and heat transfer phenomena in the WTE combustion chamber. 1-19) RC ( solid ) = C O2  1 1   +   k  r kd  (Eq. MSW is a very non- . Especially.25 1-5-4.

e. and Ash). specific weight. Nitrogen. and chemical composition (Carbon. The combustion chamber simulation consists of two major components: 1) a bed model of the drying. heat transfer between the furnace and particle/gas. Simulating these phenomena in a computer is less costly than building physical models and is helpful to develop a better chamber design. Oxygen. These models have been developed separately. Hydrogen. size and composition changes with time and location on the bed (e. and chemical reactions (i.g. energy content (heating value). Generally. volatilization and chemical reactions within the bed. there is large amount of interactions among MSW particles reacting primary air in a high temperature zone of combustion chamber. Sulfur. they interact with each other and with the flow of primary air though the bed and their temperature. mass transfer between particles and air flow. shape. there have . moisture content.26 homogeneous material and the properties of the particles vary widely in terms of size.. mostly combustion) within the bed and in the chamber above the bed. whether a particle is located near the surface or the bottom of the bed) One of the most effective ways to understand those transport phenomena and chemical rate reactions is by mathematical modeling of fluid flow though the bed and above the bed. heat transfer and gas-phase reactions in the chamber above the grate. and 2) CFD modeling of the gas flow. In addition. As the MSW particles travel over the moving grate.

the solid waste bed modeling still Figure 1 1-15: Trend of current bed models . 1-7. Whereas CFD chamber modeling is ). widely accepted and used in the WTE industry. Shortcomings of existing WTE models xisting There have been several different approaches to model WTE combustion here systems in the past (see Chapter 2 in detail).27 been more studies on CFD chamber modeling because the behavior of emission gases and corrosion of the chamber walls have been the main concerns of both operators of WTE facilities and developers of WTE technologies.

but he considers one single spherical particle type such as density (700kg/m3). the Discrete Element Method (DEM) applied in Peters’ model is a new approach in order to predict particle movements. Theoretical models. as a pre-process of CFD modeling. The problem here is that complicated simulation results do not always agree with the real chamber situation. and shear . but do not represent the heterogeneity of MSW. predicting approximate results for combustion chamber maintenance. they need a lot of run-time of computer calculation. in general. verifying them with experimental results. These models are practical and sometimes called “industry-friendly” because of their simplicity. Theoretical models are used for academic studies such as establishing new theories. and sometimes developing new grate systems. Hence. like bed surface temperature. For example. and calculating initial or boundary conditions. Figure 1-15 shows previous studies on bed modeling that can be categorized in two classes: empirical modeling and theoretical modeling. without long calculation time and need to use extensive computer resources. These models include complicated MSW flow and combustion. express bed transport and combustion phenomena in more detail than empirical models. Young’s modulus (10MPa). Empirical models are mainly used for practical basis such as training operators of WTE facilities.28 needs to include an algorithm for overcoming the heterogeneity of MSW gasification and combustion.

This discrepancy in bed modeling research is due mainly to the heterogeneous characteristics of MSW and the complexity of its physical and thermochemical conversion in a mass-burn combustion chamber. the current bed models (see the discussion in Chapter 2) are not equipped to predict other effects during the combustion processes on the grates: 1. In addition to those shortcomings of both empirical and theoretical modeling. In Shin’s model. Mixing of the MSW particles is not taken into account adequately (even though the solid mixing process is one of the principal factors for promoting combustion) .29 modulus (3 MPa) for heterogeneous MSW. Only the height of each section reduced by thermal conversion was discussed in this model. the interaction between sections of MSW bed. empirical bed models are too simple to simulate solid waste flow and combustion phenomena. This “ideal MSW particle” does not match the diversity of MSW in a real combustion chamber. limitation of steadystate bed models) 2. for example. which is agitated by a traveling grate. was not taken into account. Continuous variation of solid flow (inlet through outlet. On the other hand. Transient phenomena (channeling through the bed and break-up effects of MSW particles) 3.

Yang et. and roller grate. 2003). there are no methods for evaluating the advantages or disadvantages of various traveling grate systems for mass combustion of non-homogeneous MSW. not as particles. 2. In order . there are 700 WTE plants with different types of traveling grate systems. These shortcomings result from lack of quantitative analysis of the movement of MSW particles during the combustion process.e. i. have been studying this effect though both theoretical and experimental analyses (Yang et. etc. chamber designs. such as reverse acting.30 The main reason for these shortcomings is that past studies have considered MSW as one layer. Examples of the major problems that need to be addressed are: 1. particle-based analysis. in terms of throughput. al. The channeling of airflow through the grate results in a high excess air requirement and lower temperatures in the chamber. Addressing these issues in a novel fashion. and chamber operation on the mass-burn traveling bed is not fully understood and is not fully included in existing bed models.al. forward acting. The lack of mixing results in incomplete combustion of solid waste. Unstable combustion caused by the inlet flow of highly non-homogeneous MSW results in unstable burnout point. 3. can help alleviate some of the problems experienced in current combustor design. However. Worldwide. energy. 4. The relationship among mixing phenomena.

Even though these studies were worthwhile. 1-8. some studies using small-scale physical models were carried out. Residence time of each wooden and ceramic particle was measured using a scale model of forward and reverse acting grate systems (Beckmann et al 2000). Also. the physical small-scale models cannot simulate the volatilization and combustion phenomena on the grate that change the density of solids by an order of magnitude. particle tracking tests using 1/15 scale models developed by Lim et al clarified mixing phenomena enhanced by the traveling grates (Lim et al 2001). One of the difficulties of physical small-scale models has been that the range of particle sizes. they did not accurately replicate a real combustion chamber and grate using physical scale modeling. Developing a mathematical model to characterize and quantify flow and mixing processes of the highly non-homogeneous MSW. For example. Objectives of study The objectives of this doctoral research are mainly to provide a better tool for understanding effects of solid flow and mixing phenomena enhanced by the motion of traveling grate: 1. reciprocation speed of moving bars) as well as chamber design parameters (traveling grate) .31 to quantify the motion of MSW particles on the various types of grates. 2. Analyzing the relationship between the behavior of solid waste particles and operation parameters (especially. densities and shape factors encountered in the MSW fed to WTE combustion chambers is impossible to simulate in small physical models.

Figure 1 e 1-16: Schematic Diagram of Integrated Modeling . Calibrating and validating mathematical models using full-scale physical scale model of reverse acting grate with camera observation (analysis of MSW particle movements) and “as-collected” MSW packed bed (collection and analysis of NYC-MSW). it is necessary to incorporate two mathematical models based: a bed combustion model. and a bed mixing model under a CFD model.32 3. MSW). as shown in Figure 1-16. In order to develop an integrated bed model to address the limitations of current bed models described in the previous section (Section 1-7).

31 m3 /h Zone 5 Zone 6 Have Zone 7 Zone 8 L=7 m L: Effective Length of the grate: 7 m Vn: Velocity of Solid Wastes Have: Average height of solid wastes: 0.52 m3 /h V5=27..19 m V1=102.75 m3 /h) [18.40 m3 FB2 Zone 2 FBn: fixed bar (step) V6=16.30 m3 /h) [1. These two parameters will be correlated to equipment and operating parameters that can be changed either during design or operation (e..35 m3 /h RB2 Zone 3 Zone 4 V7=9.68 m3 /h/bar FB1 RB1 Zone 1 3 13o 0.33 The bed modeling results have been expressed in terms of two parameters: a) the residence time of particles with different sizes.75 m3 /h Volumetric flow rate of solids upward: 26.65 m3 /h 26.g. diameter.g.87 metric ton/h] 26 Figure 1-17: MSW flow and mixing in the reverse acting grate: The MSW particles are subjected to different temperatures and their size and composition change with time and location (e. and b) the mixing diffusion (dispersion) coefficients along the length (Dl) of the bed on the grate.68 m3 /h/bar Outlet of Ash (4.71 m /h V3=83. 0.23 m3 /h V8=6.2 m Inlet of Solid Waste (104.04 m RBn: reciprocating bar (step) 26 V4=54.74 m3 /h V2=97. height. number and speed of rotation of roller grates.5 m Volumetric flow rate of solids downward: 104. frequency/speed of Martin moving grates).14 metric ton/h] 0. number. whether a particle is located at the surface or bottom of the bed) .

Hence. The forces to be considered in the mathematical model are: a) the gravity force acting on the MSW deposited on the grate. 26o in the case of the Martin grate). the calculated residence time is . Depending on the “viscosity” of the solids on the grate. The combination of gravity and shear forces due to the motion of the grate results in the downward motion of the MSW particles. 7m in length.g. similarly to what would happen in a liquid. this shear force is transmitted to upward layers. Therefore. and 1 m in height). in the absence of any motion of the grate. The flow rate of MSW fed in the inlet of combustion chamber is 480 short tons per day or 18 metric tons per hour. the total volumetric rate of MSW fed by the pusher is 105 m3/hr. In the case of an inclined plane (e.173 tons per cubic meter. the downward angle of the cone would be proportionally greater. b) the motion of the grate imposes an additional shear force on the bottom layer of the solids that is in contact with the grate surface. material deposited on a horizontal plane would form a solids cone of a predictable angle of repose.0 meter at any position. for example..34 Figure 1-17 shows the MSW flow and mixing on the reverse acting grate system in a WTE facility. If the height of the MSW bed is assumed to be 1. the total volume of the MSW bed is 56 cubic meters (8 m in width. The study is based on the dimensions and operating characteristics of an actual industrial WTE plant. The specific density of MSW is 500 lb/yd3 or 0.

as noted earlier. However. and combustion (char oxidization). minlet = mdrying + mVM + mFC + moutlet (Eq. 1-21) Combustion Chamber MSW Zone 1 Zone 2 Zone 3 mdrying mVM ASH mFC minlet Zone 1 Zone 2 Zone 3 moutlet Figure 1-18: Mass balance of a MSW bed in the combustion chamber . combustion processes and b) mixing of MSW due to the motion of the traveling grate. the mass balance of this WTE chamber system is described as follows.53 hour or 32 minutes. Regarding MSW volume reduction. this residence time calculation of the MSW flow does not include combustion and mixing phenomena: a) volume reduction of MSW during the drying.35 0. there are mainly three processes: drying. As shown in Figure 1-18. volatilization. gasification.

1-22) where De is eddy diffusivity. fix carbon. respectively. x is distance along reactor. there are 8 moving bars that reciprocate 12 strokes per hour. moutlet are weight % of raw MSW.01 (close to zero). L is length of the combustion chamber. equation 1-2 becomes ∂c  De  ∂ 2 c ∂c = −  ∂θ  uL  ∂z 2 ∂z (Eq. minlet. the total volume reduction of MSW in the chamber will be in accordance with this proximate analysis of MSW. dispersion in the MSW flow is negligible and called plug . Volumetric flow rate of bar motions is 19. and ash.7 m3/hr (see Chapter 2 for detail). u is velocity in flow direction. in direction of flow. 52% of volatile matters. A typical residential MSW contains about 21% of moisture. moisture. The MSW flow is described by the dispersion model: ∂c ∂ 2c ∂c = De 2 − u ∂t ∂x ∂x for 0 < x < L (Eq. mdrying. 1-23) D  where  e  is the inverse of the dimensionless Peclet number. 20% of ash (see Section 1-4-2). 7% of fixed carbon. Pe. If Pe is smaller  uL  than 0. mFC. volatile matters. mVM. In terms of MSW mixing on a common Martin grate. In dimensionless form where z = (ut+x) / L and θ=t / t = tu /L. Since each mass reduction corresponds to a volume reduction.36 where.

and b) observation of the MSW particles in the bed using full-scaled physical section model of traveling grate (reverse acting grate) (see Chapter 3 and 5). as it can be seen in a batch reactor). In addition to these objectives. When Pe is larger than 0. The projections of the mathematical bed models will be validated by a) numerical analysis of physical characteristics. only particles located near the surface of the bed will be heated up and reach the drying and devolatizing temperatures. for estimating the rate of reactions or heat transfer. If there is no grate motion and the MSW flow has no dispersion (i. The motion of the traveling grate is important in mixing MSW in the bed since the major source of heat transfer to the MSW bed is radiation from flame of volatile matter combustion and chamber walls. Therefore. Pe < 0.01 the MSW flow has large dispersion and called mixed flow. this doctoral research hopes to advance the design and operation of WTE combustion chambers for high energy efficiency.37 flow.e. . the Peclet number Pe of the MSW flow is one of the most important parameters. The inside of the bed will be heated by conduction and then will slowly reach these temperatures (A flame front will move from top to bottom.01). This includes assessing of current WTE combustion chamber systems and suggesting the design of a new traveling grate for promoting combustion.

38 Chapter 2: Mathematical Modeling 2-1. 1989) . The MSW bed is ideally wellmixed in the vertical direction. the volume reduction process in the MSW bed is simply proportional to the decomposition rate of cellulose. Bed modeling (solid waste combustion) A bed model developed by Ahmed et al. The calculation method of this bed Radiation from the flame and the furnace wall Waste H ∆x B L Ash Air Figure 2-1: Discretization of the fuel bed (Ahmed et al. The flame zone above the bed has a constant temperature and it is the heat source for the bed. (1989) dealt with only pyrolysis reactions of MSW that was simplified to be composed of only cellulosic material. Prior models of grate transport and combustion phenomena 2-1-1. As shown in Figure 2-1.

the bed and gas temperature profiles. In order to examine the effects of the solid feed rates. This study showed simplified trends of these effects. enthalpy of pyrolysis reactions.39 modeling incorporated the pyrolysis kinetics. . kinetics and thermodynamics of volatile matter combustion. but a detailed quantitative analysis for bed modeling was not undertaken. and material transport by the traveling grate. several parameters are considered such as the air flow rates. heat transfer by convection and radiation. and the height and length of the combustion chamber required for MSW conversion.

In this bed model they defined an effective reaction coefficient determined by data from a batch-stoker test plant.40 Waste Radiation from the flame and the furnace wall Grate Cascade of Continuous Stirred Reactor Elements (CSR) Air and recycled flue gas etc. 1995) considered a combustion chamber as combination of several continuously stirred reactors (CSRs) shown in Figure 2-2. Ash Gas Waste Xj1 Gas Gas Xj2 Air Air Air Figure 2-2: Combustion Steps of a Stoker Firing Process and Model Idea (Beckman 1995) Beckmann et al. For both steady state and non steady state conditions modeling results were compared with experimental results. This research concluded . (Beckmann et al.

2000) .41 that the model was successful for describing non steady combustion processes of MSW.. Radiation from the flame and the furnace wall Ash Reaction Zone Raw Fuel New Fuel Grate Waste Moving Direction (Time) Radiation Heat Gas Temperature Fuel Temperature Ash Initial Bed Height Combustion Gas Ash zone Combustion zone Pyrolysis zone Heat-up zone Grate Level Temperature Air Figure 2-3: Schematic view of the Combustion in a Solid Fuel Bed (Shin et el.

Because of heat transfer by radiation from the flame and furnace walls. the temperature on the surface of a MSW bed increases. and ash in the bed were identified. This bed model consists of a waste shrinkage sub-model and a two-flux radiation sub-model shown in Figure 2-3. 1999. pyrolysed and gasified waste (A): Ash Figure 2-4: The volume distribution of the various defined components in the bed (Yang et al. 2002) . This transport phenomenon was described in this model and compared with experimental results. pyrolysis and gasification Gasification only Fixed carbon Volatiles M oisture Internal pore space Free ash Bound ash Time DRYING PYROLYSIS GASIFICATION VB Initial Waste (B): Moisture Volatiles Carbon Ash VC Dried Waste (C): Volatiles Carbon Ash VD Dried and pyrolysed waste (D): Carbon Ash VA Dried. reacting zone. (Shin et al 2000).42 An unsteady one-dimensional bed model was developed by Shin et al. The three stages of material conversion such as raw MSW. This model results were Key Gas space Drying.

(Peters 1994. Peters et al. This two-dimensional bed model was incorporated with various sub-process models and solved the governing equations for mass. The FLIC simulation is mainly used by academic researchers when all parameters are known or assumed because MSW is a much more heterogeneous materials than any other material that scientists have ever studied. 1999. the submodels are too detailed to determine some coefficients such as the central value and variation of waste activation energy.43 incorporated with a CFD model by Ryu et al. mixing effect between volatile matter and oxidant in the volatile combustion rates. On the bases of the results of this study. Peters et al. However. they developed a new . 2005) present a model for the calculation of an unsteady. Validation of this bed model was achieved by means of a bench-top batch furnace used in this study. momentum. this bed model simulated a bed comprised of wood particles that mechanically interacted with neighbor particles. and the momentum equation for the gaseous phase. FLIC includes several characteristics: Gaussian distribution of activation energies for waste devolatilization. Yang et al. (Yang et al. and heat transfer for both gases and solids. (Ryu et al. three-dimensional flow and combustion phenomena in a packed bed of a solid waste combustion chamber. As shown in Figure 2-5. 2002) developed a comprehensive bed model called FLIC shown in Figure 2-4. 2001).

44 model for mixing MSW particles on a forward acting grate using the Discrete Element Method (DEM). Empirical models have also been developed for training facility operators rather than for research or development. Figure 2-5: The particle-based model using Discrete Element Method (Peters 1994. Peters et al.. but it did not include any combustion phenomena. C. CUTEC-Institut has developed a dynamic model for optimizing MSW combustion (Reindorf et al). 2005). . Wolf has simulated mixing phenomena on a reverse acting grate in waste combustion systems (Wolf. 2005) Other numerical bed models have been developed for simulating solid waste combustion processes (chemical reactions and mass/heat transfer). Program codes of them are much simpler than the codes of phenomenal models.

Chamber modeling (volatile combustion) Chamber modeling that uses data computed by a bed model is used to investigate volatile matter combustion above the bed. the Gas flow (Computed from a bed model) Ash Figure 2-10: A grid structure of CFD modeling (Nakamura et al. and secondary air distribution. velocity. The flow of gas and air in the chamber results in turbulence and mixing. and the energy conservation equations. Computational Fluid Dynamics (CFD) is concerned with obtaining Air injection MSW Air/recirculated flue gas injection numerical solutions to fluid flow problems by using the continuity equation (conservation of mass). In order to calculate these complicated phenomena in the combustion chamber. depending on temperatures.2002) Navier-Stokes equation (conservation of momentum). CFD is one of .45 2-1-2. The chemical reactions are carried out between volatile matter and secondary air injected by arrays of nozzles on the chamber walls. Volatile matter combustion generated during the thermal conversion process of MSW bed is dependent on the flow and mixing of gases and air. Computational Flue gas Fluid Dynamics (CFD) is widely used for chamber modeling.

since CFD does not consider solid waste combustion in a bed. Whereas. and is used for geometric analysis of a new chamber design to calculate air and gas flows as well as for the chemical reactions of volatiles (CmHn) in the chamber. There are some types of mixing systems: freeflow systems and fluidized systems. forward acting grate. processed food.) and Fluent (Fluent Inc. Mixing processes can be classified in many ways based on the modes of solids flow mixing mechanisms that are involved. plastics. Inc.46 the best numerical tools for WTE engineers. and Roller grate in Mass-burn facilities can be categorized in free-flow mixing. or both. confection. glass. Figure 1-9 shows the grid structure used for Fluent CDF modeling of the WTE mass-burn combustion chamber in Union County Facility (Nakamura et al. fluidized bed systems for MSW can be in fluidized systems that have several effects of bubbles: the fluidized bed is considered to comprise a discontinuous phase of bubbles and a continuous phase of gas and solids. 2-2. sand and fertilizers because mixing processes in these industries are necessary for controlling rates of mass and heat transfer and chemical reactions. Reverse acting grate. Free-flow systems carry out a process of randomization of particles induced in mechanisms such as diffusion.) usually need an additional program subcode. Solids mixing theories Solids mixing is widely used in the manufacture of ceramics. Commercial CFD software packages such as CFX (ANSYS. In this . 2002). depending on the type of mixer or traveling grate.

theories of mixing processes in free-flow systems. p(x. 2-1) The mixing process can be well characterized by the diffusion model when the mean diameter of binary component particles is same.47 section (2-2). 2-2) where n is number of revolutions of the mixer. have been explained. b) diffusion-convection model (Section 2-2-2). 2-2-1. In order to describe movement of particles in the axial direction. to which reverse acting grate belongs. and d) stochastic model (Section 2-2-4). n) ∂ 2 p ( x. the following equation was applied to the experimental data: ∂p ( x. n) = Da ∂n ∂x 2 (Eq. c) convection model (Section 2-2-3). Diffusion model Diffusion model for molecules is a possible mixing mechanism of solids. Mixing in Free-flow system can be modeled by four types of models: a) diffusion model (Section 2-2-1). This mechanism is described by Fick’s equation: ∂c ∂ 2c = De 2 ∂t ∂x (Eq. n) is relative concentration of the particles which is described as follows: .

this equation has variance σx2(n) that related to the diffusivity Da. 2-3) 0 c( x. 3-3. 2-5) Compared with Eq. n) is particle concentration that is a function of distance and number of revolutions. n | x0 . n ) ∫ L (Eq. n | x0 . n) = c ( x.48 p( x. n)dx where L is the length of mixer and c(x. as Da = 2 1 dσ x (n)  L2  2 dn  n      (Eq.0) =  ( x − x0 ) 2  exp −  4 Da n  4πDa n  1 (Eq. through Einstein’s equation. 2-3 is obtained as p( x. 2-6) The diffusivity Da* is obtained by .0) =  ( x − x0 ) 2  exp−  2 2 2πσ x (n)  2σ x (n)  1 (Eq. The solution of Eq. 2-4) where x0 is the initial location of the tracer. This equation can be rewritten in the form of Gaussian distribution as p( x.

2-8) Dispersion coefficient of particles. Hwang and Hogg (Hwang and Hogg 1980) examined the mixing of powders flowing over a inclined surface in order to validate the diffusion-confection model. 2-2-2. the diffusion coefficient varies linearly with the velocity gradient: ∂u  * De = De 1 + L   ∂y    (Eq.49 * D a = Da N (Eq. Diffusion-convection model Diffusion-convection model. 2-9) . is described by the diffusion equation (Eq. 1-22 is simplified followings: De ∂c ∂ 2c =u 2 ∂x ∂x (Eq. 1-22) with an additional term including drift velocity. is dependent on shear in the flowing material. as concisely shown in Chapter 1. From their experiment results. Eq. De. Under steady state conditions. 2-7) where N is the rotating speed of the mixer. The experimental data shows that the diffusion coefficients of the particles of different densities are larger than those of the particle of the same density.

cut into halves. These parameters are dependent on the characteristics of the flowing powder and the flow system.3. shown in Figure 3-1. Let xk(N) denote average concentration of a component in cell k (k = 1. and recombined.2.4) after N cycles of the mixing operation. Convection model A convection model is developed to represent convective mixing by a process of forming a striated mixture. The process of forming a striated RR-mixture (regular packing arrangement mixture with regular pattern). The L includes interparticle collisions enhanced by the velocity gradients. 2-11) . 2-2-3. The average concentration distribution after N cycle is expressed by x ( N ) = P ( N ) x ( 0 ) = [ P (1) ] N x ( 0) (Eq. There are two striate of a rectangular block. for example. This process is a repetitive series of division and recombination. 2-10) where x ( N ) = x1( N ) [ ( x 2N ) ( x3N ) ( x 4N ) ] T (Eq.50 D* is the “intrinsic” dispersion coefficient due to random oscillations or fluctuations of the individual particles during flow. The block is compressed to one-half of its original height.

2-12) For the case shown in Figure 2-11. x (1) = P (1) x ( 0) 0  1  =  0    1  (Eq.51 p( N ) (N  p11 )  (N ) p =  21 ) (N  p31  (N )  p41  (N p12 ) (N p22 ) (N p32 ) (N p42 ) (N p13 ) (N p23 ) (N p33 ) (N p43 ) (N p14 )  (N  p24 )  (N p34 )  (N  p44 )   (Eq. 2-15) x ( 2) = [ P (1) ] 2 x ( 0 ) 1 / 2 1 / 2 =  1 / 2   1 / 2 (Eq. 2-13) x ( 0 ) = [0 0 1 1] T (Eq. 2-16) . p (1) 0  1 / 2 1 / 2 0  0 0 1 / 2 1 / 2   = 1 / 2 1 / 2 0 0    0 1 / 2 1 / 2  0 (Eq. 2-14) Hence.

Boss and Dabrowska proposed a stochastic model for the mixing process. 2-2-4. Stochastic model A non-stationary Markov chain model was applied to the mixing and segregation occurring within vibrated poly-dispersed particulate bed (Chou et. of multi-component homogeneous solids . 1977). al. the average concentrations in all the cells becomes uniform.52 Cut from here N=0 4 3 2 1 N=1 4 3 2 1 4 3 2 1 N=2 1 4 3 2 1 4 3 2 1 4 3 2 1 4 3 2 1 Figure 2-11: Formation of a striated mixture (Lai et. al. during discharge of particles from bins. 1978) This means that after two cycles.

The advantages of the stochastic approach are a) that particle distributions and particle motilities can be studied simultaneously. stochastic approach. b) the process of mixing can be examined in detail. is used when solid reactant is converted progressively throughout the solid particle. based on Markov chains. In this case the diffusion of gaseous reactant into the particle is rapid more than chemical reaction (Kunii and Levenspiel 1991). Progressive-Conversion Model (PCM) Progressive-Conversion Model (PCM). . is chosen in this study (see Chapter 4) for simulating the process of MSW mixing. 1985). The concentration distribution and the degree of mixing were calculated with the transition matrix in a Markov chain. 2-3. also called Continuous-Reaction Model (CRM). Mass and volume reduction model 2-3-1. In the consideration of these advantages.53 (Boss and Dabrowska 1985. c) particle movement simulation can be quickly carried out more than Newtonian approach based on the distinct element method (DEM) or particle element method (PEM).

as illustrated in Figure 2-13 (Levenspiel 1999). an unreacted core shrinks in size during reaction. . or Unreacted Core model. is used when the diffusion into the reactant particle is slow. Ash Time Time Concentration of solid reactant Unreacted Core Original concentration R R R R R 0 R Figure 2-13: Shrinking-Core Model (Levenspiel 1999). 3-3-2. The reaction occurs at the outer surface of the particle first.54 Time Time Concentration of solid reactant Original concentration R 0 R R 0 R R 0 R Figure 2-12: Progressive-Conversion Model (Levenspiel 1999). Thus. Shrinking-Core Model (SCM) Shrinking-Core Model (SCM).

kg is the mass transfer coefficient. S ex dt S ex dt (Eq. b is stoichiometric coefficient for B. and transforms it into product. 3) Chemical reaction control. 2-20) where Sex is unchanging exterior surface of a particle. 2) Diffusion through ash layer controls. − 1 dN B b dN A =− = bk g (C Ag − C As ) = bk g C Ag = const. CAs = 0 because no gaseous reactant is present at the particle surface. 2-17 is dN B = bdN A (Eq. . (Eq. 2-17) In this heterogeneous reaction there are three steps: 1) Diffusion through gas film controls. This kind of heterogeneous reaction may be expressed by the following equation: A (fluid) +bB (solid) → fluid and solid products. 1) Diffusion through gas film controls The rate of reaction of Eq. CAs is the concentration of reactant A on the surface of the particle. Thus. reacts with in the particle. 2-19) S ex = 4πR (Eq. CAg is the concentration of gaseous reactant A. 2-18) Where NA and NB are amount of A and B respectively.55 A gas contacts a solid particle.

56

The decrease in radius of unreacted core is given by
4  − dN B = ρ B dV = ρ B d  πrc3  = −4πρ B rc2 drc 3 

(Eq. 2-21)

where ρB is molar density of B in the solid and V is the volume of the particle. From Eq. 2-19 and 2-21, the rate of reaction in terms of shrinking radius of unreacted core is given by

ρ B rc2 dr 1 dN B − =− = bk g C Ag S ex dt R dt
2) Diffusion through ash layer controls

(Eq. 2-22)

Using the same idea of the gas film control, the rate of reaction is described followings:

dN B = S ex Q A = 4πr 2 Q A = 4πR 2 Q As = 4πrc2 Q Ac = const. dt dC A dr
C Ac = 0 1 dr = 4πDe ∫ dC A 2 C Ag = C As r

(Eq. 2-23)

Q A = De

(Eq. 2-24)

dN B dt

rc

R

(Eq. 2-25)

dN A  1 1   −  = 4πDeC Ag dt  rc R   

(Eq. 2-26)

57

3) Chemical reaction controls The rate of reaction for the Eq. 2-18 is given by

1 dN B = bk '' C Ag S ex dt

(Eq. 2-27)

where k” is the first-order rate constant for the surface reaction. Replacing Eq. 2-22 in Eq.2-28, the shrinking radius of core is given by

dr  1 dr  4πρ B rc2  = − ρ B c  = bk '' C Ag 2 dt  dt  4πrc

(Eq. 2-28)

Shrinking-Core Model (SCM) without ash formation: If a particle burns without ash formation such as pure carbon combustion in air, the shrinking particle will keep getting smaller and finally disappearing. Because there are no ash nor residue layers, the core of the particle is surrounded by gas film illustrated in Figure 2-14. In this case there are two control steps considered: 1) diffusion through gas film controls, and 2) chemical reaction control. The mechanism of chemical reaction controls of SCM without ash formation is the same as that of SCM with ash formation.

58

Time

Time

Gas film

Concentration of gas-phase reactant and product

CAg: Concentration of fluid A

CAg CAs=CAc R0R
Radial position

CAs: Concentration of fluid A on the surface of the particle CAc: Concentration of fluid A on the surface of the core

Shrinking particle

Figure 2-14: Shrinking-Core Model without ash formation

1) Diffusion through gas film controls: Mass transfer in a hot air steam to free-falling solids studied by Frossling (1938) gives

Sh = 2 + 0.552 Re 0.5 Sc 0.333 p

(Eq. 2-29)

Sh =

kd L D AB

(Eq. 2-30)

Re =

Luρ

µ

(Eq. 2-31)

Sc =

µ ρD AB

(Eq. 2-32)

Study on mass transfer in order to determine the evaporation of water and benzene drops in air was carried out by Renz and Marshall. Their experimental results were expressed by the following equation:

59

Sh = 2 + 0.6 Re 0.5 Sc 0.333 p

(Eq. 2-33)

This equation is called the Renz-Marshall equation for mass transfer to spherical particles. 1 dp

kg ~

(Eq. 3-34)

kg ~

u1 / 2 d 1/ 2 p

(Eq. 3-35)

Eq. 3-35 expresses particles in the Stokes law regime.

2-4. Heat transfer through the bed
All processes in the thermal conversion of backed bed of MSW are governed largely by mass and heat transfer. The MSW bed is an aggregation of particles shown in Figure 2-15. In this section heat transfer, (radiation, convection, and conduction) of a single particle has been described.

As a first step of modeling, heat and mass transfer of the MSW bed is calculated using FLIC (see Figure A-1 in Appendix A). For calculation of volatile combustion CFD modeling is used (see Figure B-2, B-3, and B-4 in Appendix B).

60

Wall Flame Qrad, g-p

Qconv, g-p

Qrad, w-p

Qcond, p-p Surface Grate

Qconv, a-p

Figure 2-15: Combustion and transport phenomena of one particle in a MSW bed on the traveling grate

2-4-1. Radiation among walls and particles
As shown in Figure 2-16, at high temperatures, a particle i emits a radiative
& flux and absorb a flux qrad , j from all neighboring particles j. The total radiative flux a

particle i emits is expressed as:

& & q rad , p − p = ∑ Fi → jα q rad , j − εσ Ts
j

4

(Eq. 2-36)

where α and ε are absorption and the emission coefficient, respectively. The <Ts> is

& the surface temperature of a particle i that absorbs a flux qrad from all neighboring
particles j and a wall j with the view factor Fi→j.

61

2-4-2. Radiation between a gas and a particle
The radiative flux a particle i emits is expressed as:

4 & q rad , p − g = ε ' s σ (ε g T g − α g , s Ts

4

)

(Eq. 2-37)

where ε’s is the effective emissivity of MSW and approximated by

ε 's =

εs +1
2

(Eq. 2-38)

where εs is surface emissivity. α g , s is effective gas absorptivity.

Wall j, 1

Particle i

Particle j, 1 Particle j, 2 Particle j, 5 Particle j,4 Particle j, 3

Figure 2-16: Radiation from particle i to all neighboring particles j

emitting. 2-36 is absorption and out-scattering loss. 2-4-3. φ )dΩ dl  4π  0 (Eq.62 Note: For radiation within the fluid phase and from combusting volatile matters will be calculated using CFD software. 2-40) where h is heat transfer coefficient. 2-39) The radiation intensity I is a function of position and direction. The third one is in-scattering. Convection between a particle and the primary air though the bed Heat transfer by convection between the primary air and particles is expressed as: & qconv = h(Tg − Ts ) (Eq. Fluent has several models for calculating volatile combustion with radiation. The radiative transfer equation (RTE) for an absorbing. This integral-differential equation is the third term that is in integral form and should be reduced to differential form to simplify the solving problem. The first term on the right-hand side of Eq. . and scattering medium is described as: dI  K  4π = −( K a + K s ) I b + K a I b +  s  ∫ I (Ω) P(θ . The second term shows emission of media. Ts is the surface temperature of a particle and Tg is the temperature of the primary air.

Conduction with two particles The conductive heat flux between two neighboring particles in contact is estimated by & q cond = − TS . The temperature gradient between two particles is approximated by the surface temperature difference over A1 A2 γ1 γ2 R2 R1 Figure 2-17: Conduction between two neighboring particles (Peters.i from outer particle surface. 2 1 ∂T 1 ≈− 1 / k1 + 1 / k 2 ∂r 1 / k1 + 1 / k 2 ∆rS . 2005) distance ∆rS . The contact area is assumed to be quadratic and determined by the contact angles γ1 and γ2 as shown in Figure 2-17. 2-41) where k1 and k2 are conductivities of the two particles. B. 2 (Eq.1 − ∆rS . .1 − TS .63 2-4-4.

2-43) where v is the gas velocity. D is heat diffusivity and <T> is temperature of the particle surface. Cp and Cp.64 2-4-5. 2-42) Energy balance of the outside of a single particle is describe as: D ∂T ∂r R & & & = q conv + q rad + q cond (Eq. Equation describing mass and heat balance Governing Equation of the MSW bed: The governing equations for mass and heat are summarized as follows. respectively. 2-44) Mass balance of the outside of a single particle: . ∂ ρC pT ∂t = 1 ∂ 2 ∂T & − r 2 vρ g C p. g T ) + ω (r D 2 ∂r r ∂r (Eq. For energy (heat) balance of the inside of a single particle. Mass balance of the inside of a single particle: ∂ ci 1 ∂ 2 ∂ ci = 2 (r Di ∂r ∂t r ∂r i & − r 2 vci ) + ωci (Eq. ρ is the particle density ρg is the gas density.g are the heat capacity of the particle and heat capacity of the gas.

Governing Equation of Gas: The governing equations of gas for mass. Di . g Sg includes mass transfer rates between the solid and the gaseous phase due to decomposition processes inside of the particle. 2-47) . Momentum: r ∂( ρ g v g ) ∂t r r r + ∇ ⋅ ( ρ g v g v g ) = ∇ ⋅ p g + F (v g ) (Eq.eff is the effective diffusivity of component i.eff ∂ c G . momentum.i .65 − Di . 2-45) & where v is the gas velocity. β is the mass transfer coefficient. ω ci is the conversion rate of component i. c∞ stands for the ambient gas concentration. 2-46) r where ρ g and v are the gas density and velocity at a position and time t. respectively. ci is the concentration of component i.i ∂r i = β ( c G . R R i − c∞ ) (Eq. energy and species are summarized as follows: Continuity: ∂ρ g ∂t r + ∇ ⋅ (ρ g vg ) = S g (Eq.

2-49) where k is the permeability and K and C are constants of the form: K= d 2 P3 150 ⋅ (1 − P) 2 (Eq. 2-48) r r r µr F (v g ) = − v g − ρ g C v g v g k if Re >10 (Forchheimer) (Eq. SYi . 2-50) C= 1. g are mass fraction of H2O. CO2. The function F (v g ) is defined by the following equation: r µr F (v g ) = − v g k if Re <10 (Darcy) (Eq. 2-51) Species transport: ∂ ( ρ g Yi . and CmHn. g ) ∂t r & + ∇ ⋅ ( ρ g v g Yi . CO. ωYi . g + ωYi .g is the gaseous conversion rate of a species i: . 2-52) where Yi .66 r where pg is gas pressure.75 ⋅ (1 − P ) dP 3 (Eq. g (Eq.g is the source term including mass sources due to drying and devolatilization of particles released into & the gas phase. g ) = S Yi .

i k (Eq. k k k exp( i. k indicates summation of all reactions for a species i.k RTg )) (Eq. g H m.g E a . Cm is an empirical constant. ∑ν i . 2-55) where Г is the heat from walls and the particles. k . L is a characteristic length.67 & ωY = min(ν i . Energy: ∂ ( ρ g ei . υ is the viscosity. ei.g is the specific internal energy equal to Cv . 2-54) where tm is mixing time scale.kτ mYi . λg is the thermal dispersion coefficient.i is the heat generation due to chemical reactions. Hm. 2-53) τm = 1 = tm 1 Cm νL r v3 (Eq. g ) ∂t r r r r & + ∇ ⋅ ( ρ g v g e g ) = ∇ ⋅ (λ g ∇ ⋅ Tg ) − ∇ ⋅ ρ g v g − F (v g )v g + Γ + ∑ ωYi .

a) Sieve analysis The American Society for Testing and Materials (ASTM) provides several standard test methods for sieve analysis. This section describes two kinds of particle size analysis: a) sieve analysis and b) image analysis. 4 Sieve and Coarser for Metal Bearing Ores and Related Sieve dA Particle dA: Sieve diameter Figure 3-1: Particle diameter and sieve size . such as C136-05: “Standard Test Method for Sieve Analysis of Fine and Coarse Aggregates” and E-389: “Test for Particle Size or Screen Analysis at No. Introduction 3-1-1. These methods are applicable to measurements of MSW particles. Study of particle size and shape Methods of particle size measurement have been much studied particularly in powder technology.68 Chapter 3: Experimental Work 3-1.

it measures both size and shape of particle. built 1/15-scale models of several grate systems and measured the mixing process due to the movement of the grate using small cubic particles (Lim et al. Figure 3-2 shows the model of the roller grate system and its residence time distribution in the direction of movement of the particles. as the material flowed downward. In this study. . An advantage of image analysis is that. Study of mixing and flow There have been two types of modeling of the flow of solids on a combustion grates: small-scale modeling of the full grate or a section of the grate. 2001). It is used mainly for small particles of sands. At time 0 there was a narrow distribution of tracer particles close to the feed end of the grate. powders. that is. image analysis has been used because of the above advantage. there was mixing of tracer particles with the rest. The distribution flattened out.69 Materials. and grains such as coals. width. One of the advantages of small scale models is that they are less costly to build than full scale models.” These sieve analysis methods have been adopted for MSW (Ruf 1972) in order to measure particle sizes and size distributions defining the sieve diameter (dA) as the diameter of a particle that can pass though sieve size A (Figure 3-1). Lim et al. and area of each particle are measured from digital camera images. cements. b) Image analysis In image analysis. the length. and various chemical compounds. perimeter. 3-1-2. even though substitutional particles made of plastics.

section models can be used to reproduce parts of full grate system and examine the mixing process using real MSW particles. full-scale grate models have been less applicable for the purpose of solid waste mixing. a full-scale section model of the WTE reverse acting grate was built and actual NYC-MSW particles were used in controlled tests of solids mixing in the section model. For these reasons. . in this study. This is an important consideration since “ascollected” MSW samples on a full-scale grate can represent similar geometries and actual combustion chamber conditions of commercial WTE plants. full-scale models would allow investigating packed beds of real “black bag” MSW collected from local communities. because of difficulties in construction. On the other hand. or woods must be used instead of real MSW particles. However. Beckmann and Scholz used clay. wood and ceramic spheres in their scale model of the reverse acting and forward acting grates (Beckmann and Scholz 2000).70 ceramics. except Yang et al (Yang et al 2005) has ever such as full scale tests for the study of solid waste mixing on the grate. They measured several mixing diffusion coefficients with different densities (see Table 5-2 in detail) Because of the size limitation. However. no one. “as-collected” MSW cannot be used in smallscale models. On the other hand.

residential areas of five boroughs were picked as random. Manhattan. Brooklyn. 2001). 3-2. where per capita income is the same range. Methodology a) Sample collection by field trip in New York City and by WTE facility visit Black bags were collected as typical MSW samples in residential areas of the five boroughs (Bronx. Queens and Staten Island) of New York City. Based on the map shown in Figure 3-3. After several black .71 Figure 3-2: Physical Scale Modeling (Lim et al. Measurement of particle size and Shape Factor (SF) distributions 3-2-1.

350 particles of MSW were randomly picked up from the black bags. particles are . Particles of combustion residues. Hence. X marks are the locations that black bags were collected Samples of combustion residues were randomly collected at the Union County WTE facility of Covanta Energy. were collected after the bottom ash discharger. X X X X X Figure 3-3: New York City map with per capita income (Bowen 2004).72 bags were collected in the five boroughs. or bottom ash particles.

Size and surface area of each particle were measured using the image analyzer (digital camera: Sony PCG-C1VR/BP and software: Able Image AnalyzerTM version 2.1) shown in Figure 3-4. It supports image analysis functions that include dimensional. width. such as cracked inside the discharger (water cools down + pressure of the ramp). gray scale and Able Image AnalyzerTM version 2. and area . perimeter. Able Image AnalyzerTM (Mu Labs 2004) is one of the powerful image analysis software packages and was used in order to measure particle sizes from captured image. The residues were sieved using a 1/2 inch (1. area and perimeter in this study.73 influenced by this process. width. and area. perimeter. outlining the shapes of particles regarding length. There are two steps in image analysis: 1) capturing particle images from the digital camera with color inversion and 2) outlining the shapes of particles regarding length.27 cm) screen to eliminate fine powder residues and 210 particles were selected.1 (Mu Labs 2004) Digital Camera Image Analyzers Sample Figure 3-4: The image analyzer system. b) Sample measurement by image analysis Particles of the residue samples were also measured in terms of height. width.

74

24-bits color measurements: distance, area, angle, point, line, pixel profile, histogram etc. (from images or selections) with statistics that calculate basic statistics and frequencies. From the image analysis of MSW and ash particles, shape factors for MSW and ash particles were determined.

Cluster analysis was carried out using Clustan GraphicsTM version 7.05 developed by Clustan Ltd. (Wishart 2004). This program can be used for a) calculating Tree diagram to investigate similarities of a group compared with other groups and b) drawing scatter plots (multidimensional scaling) to classify MSW and residue particles by shape factors, including aspect ratio, roundness, and spheresity.

15 cm

5 cm

15 cm

5 cm

Figure 3-5: Digital camera images of MSW samples (left) and ash samples (right)

75

3-2-2. Definitions of particle size and shape Particle Size (diameter): There are many definitions of particle size. Table 3-1 shows various particle diameters. For example, the difference among the Feret, Martin, and Sieve diameters is illustrated in Figure 3-6 (Allen 1997).

Particle

dF
1) dF: Feret diameter

dM
2) dM: Martin diameter

dA
3) dA: Sieve diameter

Figure 3-6: Various types of diameters (Allen 1997)

76

Symbol

Name (diameter) Volume Surface Surface-Volume

Definition Diameter of a sphere having the same volume as the particle Diameter of a sphere having the same surface as the particle Diameter of a sphere having the same external surface to volume ratio as the particle Diameter of a sphere having the same resistance to motion as the particle in a fluid of the same viscosity and the same velocity and at the same velocity (dd approaches ds when Re is small) Diameter of sphere having the same free – falling speed as a particle of the same density and viscosity Free-falling diameter in the laminar flow region Diameter of a circle having the same projected area as the particle in stable orientation Diameter of a circle having the same projected area as the particle in random orientation (mean value of dp = ds for convex particles Diameter of a circle having the same perimeter as the projected outline of the particle Width of the minimum square aperture through which the particle will pass The distance between pairs of parallel tangents to the projected outline of the particle in some fixed direction Chord length, parallel to some fixed direction, which divides the particle projected outline into two equal areas Chord length through the centroid of the particle outline

Formula

dv ds dsv dd

V=

π

6 S = πd s2

d v3

d sv = (d v3 / d s2 )

Drag

Fd = 3πd dηv

df dSt da dp dc dA dF dM dR

Free-falling

Stokes Projected area

d St = (d v3 / d s2 )

Projected area

Perimeter

P = πd c

Sieve Feret

Martin

Unrolled

Table 3-1: Various definitions of particle diameters (Allen 1997)

and sphericity.77 Shape factors: The shape of particulate matters is mainly described by shape factors. 3-2) Sphericity is defined as the following equation (Wadell 1932): ’Š‡”‹…‹–› {GIF@ =? F? @ G B?F? @ HB? G ? J I ? G HB? { =HI GIF@ =? F? @ HB? FHC= ?{ FHC= ?{ (Eq. aspect ratio (AR). Aspect Ratio (AR) is described as the following expression (Schneiderhohn 1954): JJ I I J (Eq. roundness (circularity). 3-1) Roundness (Circularity shape factor) is defined as the following equation (Cox 1927): J J J JJ { &Ñ{ { { (Eq. 3-3) .

30. were used to monitor the particle movement. the same length of real combustion chambers. The reciprocating bars traveled 42 cm from the top to the bottom positions. The MSW bed was divided into 16 cells (4 layers × 4 sections). at least. positioned above and on the side of the apparatus. . and 0 degrees can be selected. To minimize the effect of the boundary of the box. 14 degree was chosen in this test). Physical Modeling 3-3-1. the first and last sections located close to the wall of the box were not fed in the stochastic model. so that the horizontal distance ( …‘• I ) was approximately the same as the length of two cells (40cm). 91 cm (3 ft) in height. A section model always encounters the limitation of the particle behavior.78 3-3. each having a size of 20 × 20 cm. Figure 3-7 shows the geometry of the full-scale section model. Martin grate has an angle of 26. Two digital cameras (model: Sony PCG-C1VR/BP and Sabrent SBT-WCCK). The four bars (2 fixed and 2 reciprocating) have a height of 13 cm and an angle of 15 degrees to the bed (also 45. Because of weight limitation. Note: An ideal full-scale physical model for examining only the effect of grate motions should have an infinite length or. Methodology a) Geometry of a physical model The full-scale physical section model of the reverse acting grate used in this study had the dimensions of 121 cm (4 ft) in length. and 61 cm (2 ft) in width and the angle of bed declination to the horizontal was 15 degrees.

79 Figure 3-7: Full-scale physical section model of a combustion chamber: positions of CCD cameras scale (top) and geometries of reverse acting grate (bott (bottom) .

medium (14 cm).8 cm. where -σ ≈ 5. and + ≈ 22. The density of the tracers was 221 kg/m3.8 cm. and large (22 cm) sizes were made using gap filling insulating foam. NYCMSW particles: the mean value of the size distribution was found to be 14 cm. as measured after compression during the collection process. . +σ ity This value was more representative of the precompacted MSW an is lower than the and typical value of about 297 kg/m3 (500 lb/yd3) (Tchobanoglous 1993). These tracers were based on the particle size distribution of NYC . Spherical tracers of article small (6 cm).80 b) Making tracers in different size cers Particle tracers were prepared as shown in Figure 3-8.

40.81 Figure 3-8: Spherical tracers of different sizes (small 6cm. large 22cm) (left) and NYC-MSW particle size distributions Side view Top view Height: 20 cm Height: 40 cm Height: 60 cm Height: 80 cm Figure 3-9: Side views and top views of NYC-MSW beds with 20. medium 14cm. and 80 cm heights and angles of chamber bed decline: 15 degree . 60.

60. 80 cm beds .82 MSW bed height 20 cm The bed height was measured at the lowest position of reciprocating bars MSW bed height 40 cm MSW bed height 60 cm MSW bed height 80+ cm Figure 3-10: Cell patterns of 20. 40.

This apparatus enabled the determination of the probabilities of particle movement in each vertical and horizontal position of the packed bed of actual NYC-MSW. In this “no-feeding” physical model. d) Operation conditions Since this physical model was not motorized. and 80 cm and the angle of camber grate declination was 15 degrees to the horizontal. After each reciprocation of the moving bars. . 40. caused by the reciprocating bars. They were loaded into the physical model. along with tracers. 60. there is no MSW flow pushed into the chamber by means of a feeder. Because this physical model has neither an inlet nor a feeder. Therefore. a purely mixing phenomena characterization. 60. Therefore. 40. The bed heights were 20. the tracer positions were recorded by the cameras. was experimentally examined. as mentioned earlier.83 c) Loading NYC-MSW particles The MSW particles used in the tests were obtained from black bags collected in NYC. the operation of reciprocating bars was conducted manually. 80 cm beds. as discussed in Section 3-2-1 a) (Figure 3-3). the motion of the reciprocating bars and gravity were the driving forces that caused the particles to move within the MSW packed bed. particles were traced from the original position to the current position after n reciprocations. Figure 3-10 shows cell pattern of 20. Figure 3-9 shows pictures of side and top views that a NYC-MSW bed that was loaded in the full scale section model.

low thermal efficiency of the combustion chamber and formation of undesirable compounds (CO. MSW is more complex and non-homogeneous than any other fuel. In this system. The physical and chemical properties of MSW directly affect its combustion. Another reason for unstable combustion of MSW is transient phenomena such as the break-up of waste particles and the channeling of combustion air. Mass and volumetric flow of solids and air on the grate 4-1-1. MSW flow in the chamber Unstable combustion of a MSW packed bed flows in a mass-burn chamber is one of the major problems caused by the variability of municipal solid waste (MSW).84 Chapter 4: Mathematical Work 4-1. Some advanced automatic combustion control systems for mass-burn plants have been developed worldwide to stabilize the operating conditions during combustion. an auto-regressive model with periodic functions (fuzzy logic) is used in chaos analysis for continuous variation of MSW in the control system. The resulting unstable combustion of MSW introduces operating difficulties such as fluctuating chamber temperature. Infrared Radiation (IR) sensors are provided in the advanced control systems for image processing and detection of burn out lines. One reason for the instability of combustion is the extreme variability of the properties of the feedstock. NOx. and dioxin) in the process gas of waste-to-energy facilities. It provides Chapter Four . For example.

Monte Carlo Modeling for MSW feeding and flow This study presents a time series model for continuous variation of MSW using the Monte Carlo Method. Table 4-1 shows percentages of MSW components and the approximate analysis of various types of combustible dry waste materials in New York City. a mathematical model for time series analysis of continuous variation of MSW was developed using the Monte Carlo method to simulate stochastic combustion. NOx and dioxins in the combustion gases (Takatsudo et al.85 good control performance for reducing concentration of CO. the percolation theory is applied in order to simulate transient phenomena with the bed.047 Rubber & Leather Wood 0. The existing combustion models for mass-burn MSW chambers have used mean values for physical and chemical properties of MSW components.089 0. 1) is expressed by Xk ~  Paper  0. Additionally. Chapter Four . In this study. such as proximate analysis and Lower Heating Value (LHV). Although these new tools have resulted in improved control and operating performance.05 Metals Other  0.048 0. the continuous variation of the MSW feed into the combustion chamber still makes control of the combustion process difficult.047 Plastics Textiles 0.002 0. The probability distribution of components of MSW for a uniform distribution Uk ~U (0.266  Cardboard 0. They are used in this model as an initial distribution. 1999).046  .022 Glass 0. 4-1-2.

1 0. heat transfer and combustion reactions in both gaseous and solid phases.3 4.2 <0.2 10 Table 4-1 Ultimate analysis of dry stream municipal solid wastes in New York City (Themelis et.2 5.2 0.1 0. 1993) The FLIC two-dimensional bed program (Yang et al.6 9 6 0.5 0.6 3144 555 1052 555 24 260 591 567 544 7292 Carbon (% by Weight) 43.6 3 372 Nitrogen (% by Weight) 44 44. in order to simulate the combustion process. Tchobanoglous et al.7 0.5 42 Ash (% by Weight) 0.8 42.1 0.4 4.2 0.8 4.2 0.8 31.2 6.6 4.2 2.2 0. 4-1) In order to carry out the computation for the components of MSW.86 (Eq. Component of waste stream Paper Cardboard Plastics Textiles Rubber & Leather Wood Glass Metals Other Tons per day % in NYC Weight (metric tons day) 26. in conjunction with our model for continuous variation of MSW.5 4.9 7. 2002) developed by Sheffield University was used.1 <0.3 2865 Oxygen (% by Weight) 6 5. Chapter Four .2 5 4.5 26.6 6 0.3 2 2194 Sulfur (% by Weight) 0.7 0. the MATLAB 6 program was used to generate 100 random numbers (k = 100) for each of 100 samples (n = 100).9 4. The FLIC program is used to calculate the fluid flow.7 8.6 22.3 0.al 2002.5 44 60 55 69 49.

6 P=0.0 Figure 4-1.87 P=0 P=0. was primarily concerned with the existence of the ‘open path’ in a system. we considered site percolation for transient phenomena in combustion in Chapter Four . Although there are two types of percolations (bond percolation and site percolation).8 P=1. by Broadbent and Hammersley (Broadbent and Hammersley 1957).4 P=0. and provides a reasonable model for a disordered medium such as MSW.2 P=0. The first percolation model. It has been claimed that percolation theory is a cornerstone of the theory of disordered media. based on the percolation theory. Two-dimensional square lattice site percolation model used in the solid waste on the combustion grate Also a model for transient phenomena was developed in this study. in which a large porous stone is immersed in a bucket of water.

Simple Cubic. Triangular. pc is the critical probability and is defined by pc = sup {p : Ө(p) = 0} (Eq. Diamond. BCC and FCC. We designate each vertex of the lattice ℤ2 open with probability p and closed otherwise.4-2) (a) Channeling (b) Break-up (Subsidence) Figure 4-2: Channeling and break-up of MSW bed in the percolation model on the grate The current estimate of this critical probability pc for site percolation in the two-dimensional square lattice is 0. We assume that Chapter Four . A network of MSW on the traveling bed of combustion chamber is specified in the model as a twodimensional square lattice (Figure 4-1) in order to simplify the combustion process. Square.5927 (Grimmett 1991). Kagome. There are a number of network models such as Honeycomb. The lattice size L was defined as 40 (40 x 40) for the same reason.88 this study.

Break-up. This means that at least one cluster reaches both horizontal edges of the lattice in the percolation model. Figure 4-2 shows the channeling and break-up of MSW on the combustion grate. usually occurs when particles in the middle of the solid waste combust and create a space horizontally as shown in Figure 4-2 (b). on the other hand.89 combustion in the lattice proceeds uniformly. Figure 4-3: Percolation Model in the combustion chamber Chapter Four . This means at least one cluster reaches both vertical edges of the lattice. The program code for this model was developed using Visual Basic. Channeling usually occurs when particles of solid waste combust and create an empty space vertically through the bed as shown in Figure 4-2 (a).

90

In our model, we specified the use of the Martin reverse acting grate (Figure 4-3), one of the most popular WTE grates used in the United States. The MSW fed on this grate travels approximately 7.0 m in an hour. We divided the grate into 8 zones, through which MSW runs continuously. Mixing coefficients were assumed to be zero in order to simplify motion of the solid waste. The simulations were performed on a Linux computer (CPU: 1.8MHz, Memory: 256MB) with VMware for Windows 98 environment. It takes approximately two hours to complete one set of the calculations for this model.

4-2. Stochastic model for MSW particles mixing on the traveling grate Simulation of the physical and chemical processes in moving beds is used widely to study combustion and other chemical reaction phenomena in gas-solid systems. Three general types of mathematical models are used for investigating the transport and chemical reaction phenomena in mass burn combustion chambers: Computational Fluid Dynamics (CFD), bed models, and stochastic models. CFD models simulate the fluid flow, heat and mass transfer, and reaction phenomena in the combustion chamber above the traveling grate by solving numerically the continuity and energy conservation equations and the Navier-Stokes equations (conservation of momentum). Numerical bed models of solid waste combustion have been developed since the early 1970s (Essenhigh, R. H. and Kuo, T. J. 1970). Yang, et al. (Yang. et al. 2002) developed a two-dimensional program, called the Fluid Dynamic Incinerator

Chapter Four

91

Code (FLIC). FLIC is graphically interactive and is widely used by WTE engineers to simulate the physical and chemical transformations involved in the drying, volatilization and combustion processes on the grate; however, the combustion process is modeled using typical or average data such as composition, particle size, density and heating value, even though MSW is a very non-homogeneous fuel. Stochastic models, such as the one described in this paper are relatively new, although some researchers have suggested combining mixing models of the traveling bed with experimental work (Lim et al. 2001). Markov chain models have been used in the past to estimate mixing of powders in hoppers (Aoun-Habbache et al. 2002) and mixing in fluidized bed reactors. The present study is the first to apply the Markov chain model to the grate of a mass-burn chamber model in order to determine how solid particles move and mix on a moving grate.

4-2-1. One-dimensional Stochastic model for reverse acting grate A model, based on the Markov chain model (Bremaud 1991), was developed to simulate particle flow and degree of mixing as the MSW particles travel over the grate. The first procedural step in the design of the Markov chain model was to divide the solid waste on the bed into several cells (“state” of the system). Then, the transition of particles between these cells with time is examined. In the model, the entire grate was represented by 16 cells and by their respective boundary conditions (Figure 4-4). The computed data were compiled in the form of a transition matrix P

Chapter Four

92

and a state vector S(n) that describes the particle distribution over the states after n transitions.

Figure 4-4: Illustration of the spatial disposition of cells on the reverse acting grate: Each cell represents either the reciprocating bar or the fixed bar.

The interchange of particles between successive cells (“Transition graph”) is illustrated in Figure 4-5.

Chapter Four

93 Figure 4-5: Interchange of particles between successive cells (transition graph): The final cell of this model is cell 17(ash bin). although the paths of individual particles differ considerably.e. at each motion of the reciprocating bars (transition). n=0). Chapter Four . all particles in the MSW feed after combustion eventually arrive at cell 17. The rule governing the particle migration (travel) of the system is expressed by the following equation: S(n) = S(0) P n (Eq. The Markov chain model assumes that. Pn is the nth power of P (called the n-step transition matrix) of the matrix that contains the probabilities of the solid particle movement. 4-3) where S(n) represents the profile of MSW traveling on the chamber bed after n times of double strokes of the reciprocating bars and S(0) is the initial profile of MSW feed at the inlet (prior to any stroke of the reciprocating bars.. The latter is controlled by feed rate. i. the transition probabilities of the waste particles between adjacent cells do not depend on the previous state in time.

. . 0 . respectively. . . . 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 . ... .... .. . The subscripts f and r in the following expressions represent fixed bars and reciprocating bars.. 0 0 0 0 0 0 0 0 0 0 0 0 0 . The transition matrix P for the movement of the solid waste particles corresponding to the series of cells shown in Figure 4-5 is defined as follows: P=  pi 1 − pi  pr 1−  rr 0 rf  0 0 0 0  0 0  0 0  0 0  0 0  0 0  0 0 0 0  0 0  0 0  0 0  0 0  0 0  0 pr − rr pf 1− rr 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 p f − rf pr . ... . 0 0 0 0 0 0 0 0 0 0 0 0 0 . ... 0 0 0 0 0 0 0 0 0 0 0 0 0 ... respectively. 0 rf 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0        0   0  0   0  0   0  0   0   0  0   0  1 − po − rr   1  0 0 0 0 (Eq. ... 0 0 0 0 0 0 0 0 0 0 0 0 0 .... 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 .. 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 − pr − rr . 0 0 0 0 0 0 0 0 0 0 0 0 0 .. .. the subscripts i and o represent inlet and outlet. ....94 particle size and density. 4-4) where p is the probability that MSW particles will remain at the same grate position (cell) and r the probability that the particles transit the next grate position (cell). .. . 0 0 0 0 0 0 0 0 0 0 0 0 0 .. pf rr 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 − p f − rf po 0 0 0 0 0 .. Chapter Four . .. geometry of the grate and frequency of reciprocating bars.. .

14).. …. six probabilities are defined at each cell: .15). 5. the length of the reciprocating bars travel L.The probability rf of transiting to a fixed bar (cell #3. 6. 4. Thus. 15). 9. frequency of reciprocating bars Rr. 14). there is a transition from position/cell k (k is the cell number and is a positive integer. .. …. the .. state of combustion. 7.…. . ….e.14).The probability pf of remaining on a fixed bar (cell #3. . 1 < k < 16) to the two adjacent compartments k . …. . In order to simulate the mixing phenomena on the actual grate.The probability rr of transiting to a reciprocating bar (cell #2. cell k represents a fixed bar. angle α of chamber bed decline. cells #2.. . particle density of MSW components S. 4. …..The probability pi of particle remaining in state 1 (cell #1).e. 6. 6. The values of p and r depend on the feed rate Rf. 5. 8. If k is an odd number (i.1 and k + 1 (Figure 4-5).95 The movement of the reciprocating bars causes an upward motion of the solid waste particles against the downward direction of the feed.The probability po of remaining in outlet state (cell #16) It should be noted that the probabilities p and r are always positive fractions and the sum of the elements within each row in the matrix P equals one.. 4. 5. If k is an even number (i. 10.15). particle size of MSW components S.The probability pr of particle remaining on one of the reciprocating bars (cell #2. cell k represents a reciprocating bar. cells #3.

4-9) (Eq. pr ~ fr (Rf . θ) rf ~ gf (Rf . ρ.. H. solid waste particles are disposed to move only downward to cell #2.For the reciprocating bars (cells #2. Therefore. …. ρ.For cell #2 (also. 6. 4-6) (Eq. 4-8) (Eq. The relationships between probabilities and operating parameters can be expressed as follows: . 6. 4-11) (Eq.14)..pi (Eq. θ) pf + rf < 1 The relationships between probabilities are: . H. cell #4. ρ. H. L. S. L. θ) rr ~ gr (Rf . α. 4-10) (Eq. L. H. α. S.96 height H and the angle θ of the reciprocating bars. α. 14).For cell #1.…. 4-7) . 5. α. ρ.. pi < 1. although the reciprocating bar moves up and down. Rr . because cell #1 is next to the inlet and new feed enters all the time pushing the older particles downward the grate. S. 4. 4-5) (Eq. Rr .15). S.…. pf ~ ff (Rf . Rr . Rr . L. θ) pr + rr < 1 and for the fixed bars (cells #3. particles that ride this bar have less chance for transition from cell #3 to .

Since all the solid wastes enter at the inlet of the combustion chamber..15). 9. (4-12) and (4-13) are shown in Table 4-2.…. (4-8).14) because the motion of the reciprocating bar below cell #4 pushes particles upward. gr and gf in Eqs.2: Assumed values of probabilities satisfying Eqs (4-10). and then validate the model projections by field tests in an industrial WTE.3 Table4.rr < pr (Eq.5 0. (4-5). the initial state vector S(0) can be formulated as: .…. rf ≤ pf < 1. 9. Further work will be required to estimate values of these parameters. 4-13) The assumed probabilities for satisfying Eqs (4-11). 15) particles ASSUMED VALUE ASSUMED VALUE ASSUMED VALUE PROBABILITY PROBABILITY PROBABILITY pi po 0..97 cell #2 and are disposed to remain in cell #3 (same as cells #5.pf .2 pf rf 0. 6. (4-11) and (4-12) for the transition matrix used in this calculation have a high probability to move to the higher cell #2 (same as cells #4.pr .rf (Eq. ff . because there has not been any experimental study of the mixing phenomena.…. 4-12) . rr < 1. 7. 8.. At this stage of the study it was not possible to define the functions fr . and (4-9). define Eqs (3-21) to (3-26).4 0.3 pr rr 0.3 0. Therefore. (4-6). 7.For cell #3 that represents a fixed bar (same as cells #5. Therefore.

As stated earlier. 4-2-2. the position next to the inlet just after the solid waste is fed on the grate by the pusher at the bottom of the hopper.20 m 0. Figure 4-6 shows the total of 160 cells that were defined. we 5 layers defined 5 layers of the MSW 0. The .98 S(0) = [1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0] (Eq. an interactive matrix-based program for engineering calculations. Therefore. 4-14) This initial state vector represents the state of solid waste in cell #1. was used on a Windows 2000 PC to carry out the matrix calculations required in the Markov chain simulation.4m dimensional model of Markov the chain H=1. the assumed height of each layer is 0. that in contact with the grate surfaces. Expansion to two-dimensional stochastic model To develop the twoLg = 6.2 meter. MATLAB 6.20 m Divided into 32 zones (Lg /32) Total = 32x5 = 160 cells (mesh) fuel above the grate and 32 zones for representing the Figure 4-6: Illustration of the spatial disposition of cells on the reverse acting grate reciprocating bars and fixed bars of the reverse acting grate.0m mass-burn 32 zones combustion chamber. The cells express the position where the MSW particles might be located. the present model examined only one layer of MSW. The average height of the MSW fuel on the grate is assumed as 1 meter.

....k 162 . P and S(0) are represented by a 162x162 matrix and a 162x1 matrix.. 4-3: S(n) = S(0) P n (Eq.   162 ....e.. p j .162  .. S(0) is the initial profile of MSW feed at MSW Particle Reciprocating Bar Fixed Bar Figure 4-7: Particle movement by reciprocating and fixed bars the inlet (prior to any stroke of the reciprocating bars.162  (Eq..... P n is the nth power of the matrix P (called the n-step transition matrix) that contains the probabilities of the solid particle movement. n = 0).99 rule governing the particle migration on the grate system is expressed by the same equation as Eq.. respectively: P  p11  . i... =  . . 4-3) where S(n) represents the profile of MSW particles traveling on the grate after n times of double strokes of the reciprocating bars..   p162. 4-15) 162 . .. .. .   p162.. . ......

. including drying. 0] (Eq. volatilization. As they travel over the length of the grate. Major factors in the size reduction of MSW particles are the physical and chemical transformations occurring on the grate.e..100 S(0) = [1 0 . because of the motion of the moving bars of the Martin grate... and char oxidation and finally turn into ash (Figure 4-8). 4-16) where pj. the driving forces in this system are gravity and the movement of the bars. Size (mass and volume) reduction model 4-3-1. particles are subjected to drying.. i..k are the probabilities to be measured in future field tests on a Martin grate WTE.. The rules for determining these probabilities are the same as those for the one-dimension Markov model. Introduction The size and volume of municipal solid waste (MSW) particles is reduced during the combustion process in the waste-to-energy (WTE) combustion chamber.. we simulated the movements without the effect of MSW properties and continuous feeding... This simulation was carried out using the MATLAB 6 program on a Windows 2000 PC.. some MSW particles break up and their size is reduced.... .. volatilization. To simplify the complexity of the movement of MSW particles... 4-3. The grate operation is necessary because MSW is much more heterogeneous than coal and other fuels. The Martin reverse-acting traveling grate depicted in Figure 1 controls the rate of flow and also enhances mixing.... In addition. In this section we used assumed values for the probabilities. This breakage process can be . and combustion.

because MSW varies greatly in both densities and compressibility. MSW particles become ash consisting mostly of non-combustible residues. After the drying. Before the MSW enters into the inlet of the combustion chamber. and they form larger particles (clinker). For example. Campbell and Webb analyzed roller milling performance and developed a breakage equation (Campbell and Webb. Due to the motion of the traveling grate. the MSW particles are accumulated in the feed hopper from the bottom of which they are pushed into the combustion chamber by the ram feeder. where the temperature rises above 1100 oC. The resulting compaction process can be estimated by means of a ram pressure-density curve. and combustion processes in the chamber.101 numerically expressed as a breakage matrix based on experimental data. The particles that are on the surface on the bed. 2001). . are subjected to some fusion and agglomeration. clinker particles of ash can break again and their size is finally reduced to the PSD of ash at the outlet of the combustion chamber (Figure 48). volatilization.

Devolatilization 3.102 MSW particle 1. . Combustion (char oxidization) Ash particle Ash particle size distribution Figure 4-8: Combustion and transport phenomena of one particle in a MSW bed on the traveling grate In order to simulate the size reduction of MSW particles. Image Analysis was used to determine the PSD of MSW and ash samples. Drying (moisture evaporation) MSW particle size distribution 2. thermogravimetric analysis (TGA) data was applied on particles of the main constituents of MSW in order to examine the effect of thermal decomposition on particle size. Also. we modeled the PSD of MSW particles and ash particles and applied the shrinking core model (SCM) to these distributions. Finally.

08 0. 4-17) where a is constant. 2. The SCM was originally developed by Yagi and Kunii for describing solid and fluid reactions 0 0 5 10 15 D 20 25 30 Figure 4-9: Particle size density function (Gamma distribution) for different values of α (=1. Gbor and Jia developed a SCM combined with PSDs (Gbor and Jia. D is the particle size.04 α= 3 α= 4 α= 5 particles. β are positive parameters. 4-18) .103 4-3-2. t ) ) ⋅ ( f p ( D) )dD 0 (Eq.12 β= 5 α= 2 0. Modeling Particle Size Distributions (PSDs): PSDs of MSW and ash were described in this study by using the Gamma function distribution that is expressed as follows: D f p ( D) = a ⋅ − 1 D α −1 e β α β Γ(α ) (Eq. Figure 4-9 shows the gamma distributions for different values of α but constant β. and α . 3. 2004) The unreacted mass fraction in the core of a particle can be defined as follows: ∞ Unreacted Fraction = ∫ ( g ( D. 4. respectively. 5) and constant β (= 5) for non-porous particles (Yagi and Kunii 1955).2 α= 1 0. Gamma Distribution Combustion of MSW particles: The shrinking core model (SCM) is used to simulate the thermal decomposition of MSW fp(D) 0.16 0.

g ( D.104 where g(D. 4-19) A large part of the size reduction of MSW is due to thermal decomposition that occurs at temperatures below 400 oC and can be expressed as follows: MSW → moisture + volatile matters + char 3 (Eq. t ) = 1 − X D = 1 − rn  D   (Eq. 4-21) where krm =krD. R is radius of particles (R = D/2). When MSW particles are fed into the chamber. t) = 1 . or chemical reaction control.XD is the unreacted fraction and XD depends on the type of the controlling regime. . t) = 0 for 0 < D < Dt and g(D. such as diffusion through gas film. t) = 1 . diffusion through ash layer. Dt = krnt. t ) ⋅ a ⋅ − 1 D α −1 e β dD α β Γ(α ) (Eq. Other possible size reduction phenomena that are not included in this model: (1) Compaction of MSW particles: The size of MSW particles can be reduced due to compaction. the overall conversion for all particle sizes (X) is Dmax D X =1− Dt ∫ g ( D. and kr= k1CAg/ ρR. 4-20)  k t For this reaction. the feeder compresses them and they become smaller than the particles in black bags as collected.XD for Dt < D < Dmax Therefore. g(D.

Stochastic model for MSW mixing combined with a flow The first procedural step in the design of the stochastic model was to divide the MSW bed into several cells ( (mesh grid shown in Figure 4-10. a commercial WTE plant. same as in 4-6). akage some of them break due to movement of the traveling grate and mechanical interaction with neighbor neighboring particles.105 (2) Breakage of MSW particles: As the particles travel on the reverse acting grate. We divided the bed into 32 sections along the axial direction and 5 layers along the horizontal direction.4m. particle movement probabilities were determined from experimental data and used to formulate a transition matrix P. whose total length is 6. The tion. resulting in a total of 162 cells including the inlet and ashbin cells. In . 4-4. For example. has 8 moving bars and 8 fixed bars and the initial MSW bed height is 1m. a reverse acting grate. Figure 4-10: MSW bed on the reverse acting grate of a WTE plant showing divided cells (mesh) for the : stochastic simulation . a glass bottle fed into the chamber may break into hundreds of small pieces.

130 (= 0). k is the ratio between MSW feed flow rate and the frequency of the reciprocating bars.106 The stochastic process employed in the mathematical model is called a Markov Chain and assumes that.34. This is because particles at the exit position in each layer (the .. p97. such as feeding bars and screws that are currently employed in mass-burn combustion chambers. and p129.98. at each reciprocation of the moving bars. The rule governing the particle migration of the system is expressed by the following equation: S(n) = ( F k · P ) n · S(0) (Eq. n = 0). since particles travel only from section i to the next section i+1 in a given layer. the transition probabilities of the waste particles between adjacent cells are independent of the previous state in time. If there is no inlet feeder flow. all transition probabilities pi. F k is the k th power of matrix F that controls the MSW flow in the packed bed pushed by the feeder at the combustion chamber inlet. p65.e. In this type of flow system. except for the exit location probabilities p33. 4-22) where S(n) represents the profile of MSW traveling on the chamber bed after n times of reciprocations of the moving bars and S(0) is the initial profile of MSW feed at the inlet (prior to any reciprocation of the moving bars. There are several different types of feeding systems. we considered it as a plug flow.66. i. The elements of F that we used are shown in Figure 4-11 and their corresponding cells and directions are shown in Figure 4-12. The dimension of the flow matrix F is 162-by-162. In order to simplify a MSW flow caused by a feeder. then k = 0 and F becomes the identity matrix that neither affects P or S(0).i+1 = 1.

specific to the particular operational conditions. Since the MSW flow greatly depends on the type of feeding system and chamber geometry. .107 32nd section) cannot transition back to the inlet position for these 4 cells.j are equal to zero. All other pi. further study is needed to determine the elements of flow matrix F.

108 Figure 4 4-11: The elements of the flow matrix F .

244 probabilities.1. pi. pi.i-1 is the probability from cell i to cell i-1 (opposite to the flow direction).i+1 is the probability that the particle transits from cell i to the neighbor cell i+1 (along the flow direction) and pi. p162. pi. As shown in Figur 4-13 the size of the Figure 13.i-32 is the probability that it moves to the cell in the i+32) neighboring layer above (from cell i to cell i-32). ….i. As also shown in Figure 4 main 4-12. It should be noted that the probabilities in the yer transition matrix P are always positive fractions and the sum of the elements within each row of the matrix P equals one.i+32 is the probability that the particle moves to the cell in the layer directly below the current cell location(from cell i to cell i+32 and pi. p2..109 Figure 4-12: Diagram of probabilities and directions of the flow for each cell of the MSW bed on the grate P is the transition matrix that contains the probabilities predicting the solid particle movement due to the motion of the reciprocation bars.2 .…. transition matrix is 162-by-162 including a total of 26.162 ) elements represent the probabilities that the MSW particles remain in the same cell. as the same size of flow 162 matrix F. The main diagonal ( p1.. .

110 Figure 4-13 The elements of the transition matrix P 13: .

represents the state of the solid waste in the inlet cell. Since all of the solid waste enters at the inlet of the combustion chamber. In order to carry out the matrix calculations required in this stochastic simulation. whose size is 1×162. a typical value for the MSW feed in this model was set at a time of 64 min for traveling the total chamber length of 6. The predictive capability of the stochastic model was expanded by employing 8 moving bars to match a full-length grate system. . The grate system of an actual combustion chamber has 8 moving bars as shown in Figure 4-10.1 was used on a Windows XP PC. Based on the fact that the residence time of a real combustion chamber roughly ranges from 30 to 120 min (sometimes a particle stays shorter or longer than this range).111 The transition probabilities were determined using measured data from the full-scale experimental cross-section model. 4-16) This initial state vector.e.4m (32cells): i. MATLAB 7. whereas the physical section model employs 2 moving bars as shown in Figure 3-7. the initial state vector S(0) can be expressed as: S(0) ={ { (Eq. 2min/cell. whose position is at the bottom of the hopper adjacent to the feeder bar.

4 in Roundness Distributions (Figure 5-1-d). about 10 cm. Chapter Five . shifts to the peak of the ash PSD. This means ash particles are more homogeneous in size. As shown in Figure 5-1-a. and roundness than the MSW particles.5 and 2. aspect ratio. about 2 cm. Results from experimental work 5-1-1.4 in Aspect Ratio Distributions (Figure 5-1-c) and from 1. Particle size and shape distributions Figure 5-1 shows the particle size distributions (PSDs). Either ash distribution has a higher peak than MSW distribution. the tail of the MSW PSD is larger than that of the ash PSD. during the combustion process. sphericity distributions.112 Chapter 5: Results and Discussion 5-1. respectively. aspect ratio distributions (ARDs) and roundness distributions of MSW and ash particles. sphericity. The two peaks are ranging from 1.3 to 1. two peaks of MSW and ash particles are located between 1. Also. In Sphericity Distributions (Fig 5-1-b). the peak of MSW PSD.2 to 1.

(ARDs).5 1.5 4. b) Sphericity distribution c) Aspect Raito distributions (ARD d) Roundness ).5 Figure 5-1: Distributions by particle numbers of residential MSW and combined ashes in NYC a) Particle : size distributions (PSDs).5 4.5 4 4. the MSW particles distributions get closer to ash distributions.5 c) Aspect Ratio Distributions Aspect Ratio Distribution 16 14 Relativi Frequency % / cm (by Number of Particles) Relativi Frequency % / cm (by Number of Particles) 25 Roundness Distribution d) Roundness Distributions 12 10 8 6 4 2 0 0. distributions. the Chapter Five .5 0. particle shrinkage results in size reduction and shape change of MSW particles Then particles.5 Ash MSW 20 15 Ash 10 MSW 5 0 2.5 1. distributions start generating volatile matter and their mass and volume decrease rapidly This tart rapidly.5 2 2.5 Sphericity 3 3.5 1 1.5 Roundness 3. the acting grate and heated up. The most important phenomenon during size and shape change processes is volatilization. since around 300 oC MSW particles usually a) Size Distributions 12 Sphericity Distributions b) Sphericity Distribution Ash Relativi Frequency % / cm (by Number of Particles) 10 8 MSW MSW Ash 6 4 2 0 0.5 Aspect Ratio 3.5 2.113 In the combustion chamber the MSW particles are mixed by the movement of chamber. Finally.

The Aspect Ratio Distributions are 1.3 and 1. sphericity distributions. The particle size distributions (PSDs). around 2 cm.2 and 1. Chapter Five .114 combustible MSW particles become ash during the char gasification stage (fixed carbon oxidization). respectively. In sphericity distributions.4.4 and the Roundness Distributions 1.5 and of the ash particles at 2. The peak of MSW PSD is around 10 cm and shifts to the peak of ash PSD. the “tail” of the MSW PSD is larger than that of ash PSD. the peak of the MSW particles was found to be at 1. aspect ratio distributions (ARDs) and roundness distributions of MSW and ash particles were analyzed. Also. during the combustion process.

These probabilities in layer E. Approximately 50% for any one of the three sizes remain in layer B while the rest of them move to other cells. For example. Probabilities from full-scale physical model Figures 5-2 to 13 show particle movement and the state diagrams (transition graph) of small. a tracer located in the cells at layer E moves very quickly because it is pushed by a reciprocating bar. and large tracer particles at different heights (20. medium. These probabilities showed that 22% of small particles and 8% of medium particles stay in layer E while the rest of them (78% of small and 92% of medium particles) move to a neighboring cell because of the motion of a reciprocating bar. medium. no experimental data could be obtained because the small and medium particles occupied layer E and tended to remain at the bottom so that large particles could not transit there. are lowest for small and medium particles. Since the MSW was loaded up at the height of 80 cm. A tracer particle in each cell has a different behavior. Figure 5-14 shows the measured probabilities of small. For large particles. and large particles that remain in the same cell location of a NYC-MSW packed bed after one reciprocation of its moving bars. Layer B is where small and medium particles have the second lowest probabilities and large tracers have the lowest. 40. Retention probabilities for the middle of the MSW bed (layers C and D) are much higher with values up to 81%.115 5-1-2. layer B is at Chapter Five . at the bottom of the bed. and 80 cm) of the MSW bed. 60.

a type of motion that is unavailable to particles in the lower layers.116 the top of the bed and a tracer on the free surface moves more because there are fewer particles acting on it to constrain its motion. Particles in this top layer were sometimes observed to be easily rolling along the free surface. Chapter Five .

20 cm A0 A1 A2 A3 A4 A5 15 degree B0 B1 B2 B3 B4 B5 Solid bed divided by20cm×20cm cells C0 C1 C2 C3 C4 C5 D0 D1 D2 D3 D4 D5 E0 E1 E2 E3 E4 E5 Grate inclination: 15o Bed solid: Mixed MSW Tracer: 6 cm sphere tracer Original bed height: 20 cm Corresponding state diagram Figure 5-2: Observed particle movements and corresponding state diagram Chapter Five . One arrow indicates a movement and direction of a particle tracer into a neighbor cell during a reciprocating motion of moving bars.117 A0 A1 A2 A3 A4 A5 B0 B1 B2 B3 B4 B5 C0 C1 C2 C3 C4 C5 D0 D1 D2 D3 D4 D5 CCD image from Side view of the bed E0 E1 E2 E3 E4 E5 Observed particle movements: Number of the circles in a cell means how many a tracer particle stayed in the cell per reciprocating motion of moving bars.

118 A0 A1 A2 A3 A4 A5 B0 B1 B2 B3 B4 B5 C0 C1 C2 C3 C4 C5 D0 D1 D2 D3 D4 D5 CCD image from Side view of the bed E0 E1 E2 E3 E4 E5 Observed particle movements: Number of the circles in a cell means how many a tracer particle stayed in the cell per reciprocating motion of moving bars. One arrow indicates a movement and direction of a particle tracer into a neighbor cell during a reciprocating motion of moving bars. 20 cm A0 A1 A2 A3 A4 A5 15 degree B0 B1 B2 B3 B4 B5 Solid bed divided by20cm×20cm cells C0 C1 C2 C3 C4 C5 D0 D1 D2 D3 D4 D5 E0 E1 E2 E3 E4 E5 Corresponding state diagram Grate inclination: 15o Bed solid: Mixed MSW Tracer: 14 cm sphere tracer Original bed height: 20 cm Figure 5-3: Observed particle movements and corresponding state diagram Chapter Five .

One arrow indicates a movement and direction of a particle tracer into a neighbor cell during a reciprocating motion of moving bars. 20 cm A0 A1 A2 A3 A4 A5 15 degree B0 B1 B2 B3 B4 B5 Solid bed divided by20cm×20cm cells C0 C1 C2 C3 C4 C5 D0 D1 D2 D3 D4 D5 E0 E1 E2 E3 E4 E5 Corresponding state diagram Grate inclination: 15o Bed solid: Mixed MSW Tracer: 22 cm sphere tracer Original bed height: 20 cm Figure 5-4: Observed particle movements and corresponding state diagram Chapter Five .119 A0 A1 A2 A3 A4 A5 B0 B1 B2 B3 B4 B5 C0 C1 C2 C3 C4 C5 D0 D1 D2 D3 D4 D5 CCD image from Side view of the bed E0 E1 E2 E3 E4 E5 Observed particle movements: Number of the circles in a cell means how many a tracer particle stayed in the cell per reciprocating motion of moving bars.

40 cm A0 A1 A2 A3 A4 A5 15 degree B0 B1 B2 B3 B4 B5 Solid bed divided by20cm×20cm cells C0 C1 C2 C3 C4 C5 D0 D1 D2 D3 D4 D5 E0 E1 E2 E3 E4 E5 Corresponding state diagram Grate inclination: 15o Bed solid: Mixed MSW Tracer: 6 cm sphere tracer Original bed height: 40 cm Figure 5-5: Observed particle movements and corresponding state diagram Chapter Five .120 A0 A1 A2 A3 A4 A5 B0 B1 B2 B3 B4 B5 C0 C1 C2 C3 C4 C5 D0 D1 D2 D3 D4 D5 CCD image from Side view of the bed E0 E1 E2 E3 E4 E5 Observed particle movements: Number of the circles in a cell means how many a tracer particle stayed in the cell per reciprocating motion of moving bars. One arrow indicates a movement and direction of a particle tracer into a neighbor cell during a reciprocating motion of moving bars.

121 A0 A1 A2 A3 A4 A5 B0 B1 B2 B3 B4 B5 C0 C1 C2 C3 C4 C5 D0 D1 D2 D3 D4 D5 CCD image from Side view of the bed E0 E1 E2 E3 E4 E5 Observed particle movements: Number of the circles in a cell means how many a tracer particle stayed in the cell per reciprocating motion of moving bars. 40 cm A0 A1 A2 A3 A4 A5 15 degree B0 B1 B2 B3 B4 B5 Solid bed divided by20cm×20cm cells C0 C1 C2 C3 C4 C5 D0 D1 D2 D3 D4 D5 E0 E1 E2 E3 E4 E5 Corresponding state diagram Grate inclination: 15o Bed solid: Mixed MSW Tracer: 14 cm sphere tracer Original bed height: 40 cm Figure 5-6: Observed particle movements and corresponding state diagram Chapter Five . One arrow indicates a movement and direction of a particle tracer into a neighbor cell during a reciprocating motion of moving bars.

122 A0 A1 A2 A3 A4 A5 B0 B1 B2 B3 B4 B5 C0 C1 C2 C3 C4 C5 D0 D1 D2 D3 D4 D5 CCD image from Side view of the bed E0 E1 E2 E3 E4 E5 Observed particle movements: Number of the circles in a cell means how many a tracer particle stayed in the cell per reciprocating motion of moving bars. 40 cm A0 A1 A2 A3 A4 A5 15 degree B0 B1 B2 B3 B4 B5 Solid bed divided by20cm×20cm cells C0 C1 C2 C3 C4 C5 D0 D1 D2 D3 D4 D5 E0 E1 E2 E3 E4 E5 Corresponding state diagram Grate inclination: 15o Bed solid: Mixed MSW Tracer: 22 cm sphere tracer Original bed height: 40 cm Figure 5-7: Observed particle movements and corresponding state diagram Chapter Five . One arrow indicates a movement and direction of a particle tracer into a neighbor cell during a reciprocating motion of moving bars.

60 cm A0 A1 A2 A3 A4 A5 15 degree B0 B1 B2 B3 B4 B5 Solid bed divided by20cm×20cm cells C0 C1 C2 C3 C4 C5 D0 D1 D2 D3 D4 D5 E0 E1 E2 E3 E4 E5 Corresponding state diagram Grate inclination: 15o Bed solid: Mixed MSW Tracer: 6 cm sphere tracer Original bed height: 60 cm Figure 5-8: Observed particle movements and corresponding state diagram Chapter Five .123 A0 A1 A2 A3 A4 A5 B0 B1 B2 B3 B4 B5 C0 C1 C2 C3 C4 C5 D0 D1 D2 D3 D4 D5 CCD image from Side view of the bed E0 E1 E2 E3 E4 E5 Observed particle movements: Number of the circles in a cell means how many a tracer particle stayed in the cell per reciprocating motion of moving bars. One arrow indicates a movement and direction of a particle tracer into a neighbor cell during a reciprocating motion of moving bars.

One arrow indicates a movement and direction of a particle tracer into a neighbor cell during a reciprocating motion of moving bars.124 A0 A1 A2 A3 A4 A5 B0 B1 B2 B3 B4 B5 C0 C1 C2 C3 C4 C5 D0 D1 D2 D3 D4 D5 CCD image from Side view of the bed E0 E1 E2 E3 E4 E5 Observed particle movements: Number of the circles in a cell means how many a tracer particle stayed in the cell per reciprocating motion of moving bars. 60 cm A0 A1 A2 A3 A4 A5 15 degree B0 B1 B2 B3 B4 B5 Solid bed divided by20cm×20cm cells C0 C1 C2 C3 C4 C5 D0 D1 D2 D3 D4 D5 E0 E1 E2 E3 E4 E5 Corresponding state diagram Grate inclination: 15o Bed solid: Mixed MSW Tracer: 14 cm sphere tracer Original bed height: 60 cm Figure 5-9: Observed particle movements and corresponding state diagram Chapter Five .

60 cm A0 A1 A2 A3 A4 A5 15 degree B0 B1 B2 B3 B4 B5 Solid bed divided by20cm×20cm cells C0 C1 C2 C3 C4 C5 D0 D1 D2 D3 D4 D5 E0 E1 E2 E3 E4 E5 Corresponding state diagram Grate inclination: 15o Bed solid: Mixed MSW Tracer: 22 cm sphere tracer Original bed height: 60 cm Figure 5-10: Observed particle movements and corresponding state diagram Chapter Five .125 A0 A1 A2 A3 A4 A5 B0 B1 B2 B3 B4 B5 C0 C1 C2 C3 C4 C5 D0 D1 D2 D3 D4 D5 CCD image from Side view of the bed E0 E1 E2 E3 E4 E5 Observed particle movements: Number of the circles in a cell means how many a tracer particle stayed in the cell per reciprocating motion of moving bars. One arrow indicates a movement and direction of a particle tracer into a neighbor cell during a reciprocating motion of moving bars.

126 A0 A1 A2 A3 A4 A5 B0 B1 B2 B3 B4 B5 C0 C1 C2 C3 C4 C5 D0 D1 D2 D3 D4 D5 CCD image from Side view of the bed E0 E1 E2 E3 E4 E5 Observed particle movements: Number of the circles in a cell means how many a tracer particle stayed in the cell per reciprocating motion of moving bars. One arrow indicates a movement and direction of a particle tracer into a neighbor cell during a reciprocating motion of moving bars. 80 cm A0 A1 A2 A3 A4 A5 15 degree B0 B1 B2 B3 B4 B5 Solid bed divided by20cm×20cm cells C0 C1 C2 C3 C4 C5 D0 D1 D2 D3 D4 D5 E0 E1 E2 E3 E4 E5 Corresponding state diagram Grate inclination: 15o Bed solid: Mixed MSW Tracer: 6 cm sphere tracer Original bed height: 80 cm Figure 5-11: Observed particle movements and corresponding state diagram Chapter Five .

One arrow indicates a movement and direction of a particle tracer into a neighbor cell during a reciprocating motion of moving bars.127 A0 A1 A2 A3 A4 A5 B0 B1 B2 B3 B4 B5 C0 C1 C2 C3 C4 C5 D0 D1 D2 D3 D4 D5 CCD image from Side view of the bed E0 E1 E2 E3 E4 E5 Observed particle movements: Number of the circles in a cell means how many a tracer particle stayed in the cell per reciprocating motion of moving bars. 80 cm A0 A1 A2 A3 A4 A5 15 degree B0 B1 B2 B3 B4 B5 Solid bed divided by20cm×20cm cells C0 C1 C2 C3 C4 C5 D0 D1 D2 D3 D4 D5 E0 E1 E2 E3 E4 E5 Corresponding state diagram Grate inclination: 15o Bed solid: Mixed MSW Tracer: 14 cm sphere tracer Original bed height: 80 cm Figure 5-12: Observed particle movements and corresponding state diagram Chapter Five .

80 cm A0 A1 A2 A3 A4 A5 15 degree B0 B1 B2 B3 B4 B5 Solid bed divided by20cm×20cm cells C0 C1 C2 C3 C4 C5 D0 D1 D2 D3 D4 D5 E0 E1 E2 E3 E4 E5 Corresponding state diagram Grate inclination: 15o Bed solid: Mixed MSW Tracer: 22 cm sphere tracer Original bed height: 80 cm Figure 5-13: Observed particle movements and corresponding state diagram Chapter Five .128 A0 A1 A2 A3 A4 A5 B0 B1 B2 B3 B4 B5 C0 C1 C2 C3 C4 C5 D0 D1 D2 D3 D4 D5 CCD image from Side view of the bed E0 E1 E2 E3 E4 E5 Observed particle movements: Number of the circles in a cell means how many a tracer particle stayed in the cell per reciprocating motion of moving bars. One arrow indicates a movement and direction of a particle tracer into a neighbor cell during a reciprocating motion of moving bars.

6 0.8 0.2 0.7 0.1 0 20 cm (Layer E) 20 (bottom) 40 cm (Layer D) Small Medium Large 40 60 cm (Layer C) 60 80 cm80 (Layer B) (top) MSW bed height (cm) Figure 5-14: Measured probabilities that particles stay in the same position (cell) after one reciprocation of the moving bars Chapter Five .129 Probabilities that particles stay in the same position (cell) 0.5 0.4 0.9 0.3 0.

The first 20 of the 100 samples used are shown in Figure 5-15. Results from mathematical work 5-2-1. Mass and volumetric flow of solids (Monte Carlo simulation results) Figure 5-15 shows the simulation results for continuous variation of the components of MSW generated in New York City.1 metals other 0 0 5 10 Sample Number Figure 5-15: Time series of continuous variation of MSW components by Monte Carlo simulation 15 20 Figure 2.3 cardboard plastic textiles 0.2 rubber & leather wood glass 0. Each sample has several components and. The time series made by these samples specifies one of the initial conditions to simulate the combustion process. The abscissa shows the MSW sample number.130 5-2. Chapter Five .4 paper Component probablilty (x 100 Weight %) 0. Percolation Model in the Combustion Chamber time series for continuous variation of MSW. they define a 0. together.

each combustion probability is nearly equal to or exceeds the critical probability Pc. Figure 5-16 (c) shows the simulation results of percolation model for transient phenomena (break-up and channeling). which is in Zone 8. Near the inlet of the combustion chamber. From Zone 5 to Zone 8. The solid temperature is highest in the middle of the grate which is located around Zone 4. Chapter Five . Figure 5-16 (b) shows the residual ash in solid (mass %) (see Appendices in detail). It is based on the calculation result of residual ash shown in Figure 5-16 (b). From the middle of the grate to the ash discharge end of the grate. the MSW burns well and generates ash until it reaches the burn-out line. as shown in Figure 5-16 (a) (see Appendices in detail). temperatures of solid waste remain under 371 K because moisture evaporation occurs in this stage and combustion of most components of MSW is limited. This indicates that the possibility of break-up and channeling of solid waste after Zone 5 is high and channeling is observed in the lattice of Zone 5.131 The simulation results using the above samples are shown in Figure 5-16.

132 (a) Solid temperature (K) (b) Residual ash in solid (mass %) (c) Result of percolation model for transient phenomena (break-up and channeling) Figure 5-16: Calculation results of the MSW combustion on the Martin grate Chapter Five .

The governing parameters such as void probability Pv.133 Figure 5-17 shows how much combustion is completed by volume % in the each zone. This means the burn-out line which means the remaining solid waste does not contain combustible matter. The percolation model has shown the mechanism of transient phenomena such as channeling and break-up of solid waste. between Pc and Pch. From that. We assumed that MSW from inlet of the chamber consists of 20 % volumetric voids. Therefore.e. Pch is the channeling probability and.2. channeling is generated frequently. we assumed Pv = 0.5927 because we specified a square lattice for this percolation model. i. channeling probability Pch. Pc is 0. Pa is the ash probability.. solid waste in Zone 1 burns only 13 volume % of total solid waste. the volume change (%) from waste to ash has been calculated and the combustion probabilities for each zone have been estimated. The simulation results of this combustion model with continuous variation of MSW have provided the temperatures and percentage of residual ash by weight. The most important combustion phenomena and their most probable occurrences are shown in Figure 5-18. and ash probability Pa were found by this model. In this study. Break-up occurs between Pch and Pa. for example.2. Chapter Five . critical probability Pc. Pv = 0.

4 0.134 1 Probability (Combustion [Volume %]) 0.2 0.8 0.1 0 1 2 3 4 5 Zone Number 6 Zone 1 Zone 2 Zone 3 Zone 4 Zone 5 Zone 8 Zone 7 Zone 6 7 8 Figure 5-17: Combustion probability in each zone 0 Figure 5-18: Transient phenomena during the combustion process Chapter Five .5 0.3 0.9 0.7 0.6 0.

the particle seems not to move for a while. the particle may remain in the same cell or move to the adjacent cell. The particle actually moves back and forth between cells as indicated by the solid line. This graph shows how a hypothetical solid particle travels over the reverse acting grate.135 5-2-2. which. due to the upward motion of the bars. would result in mixing with the next layer up (see next section 5-2-2-b. the particle travels back and forth because of the action of the reciprocating bar. Two-dimensional simulation). between cells #11 and #12. At each motion of the reciprocating bar. in a multi-layer model of the bed on the grate. Simulation results from the stochastic model for MSW mixing a) One-dimensional simulation Figure 5-19 shows a simulated “random walk” that represents the particle movement based on a calculation over 46 steps (the required number of calculations varies between different time series because of the stochastic nature of the simulation results). For example. At the #2 and #5 cells. based on the results of the matrix calculation. Cell #17 denotes the state of bottom ash. The lower graph in Figure 5-19 provides a visual illustration of the behavior of a particle on the moving grate. Chapter Five . downward or upward. Cell #1 represents the feed end of the grate and cell #16 the end of the grate where the ash falls to the ash pit below.

Bottom graph: Visualization of the particle travel based on the results of the calculations shown on top graph. The peak of each particle distribution profile is displaced toward the outlet and the profile becomes much flatter. In cells #11 and 12. the particle goes back and forth because of the movement of the reciprocating bar. Figure 5-20 shows the residence time distributions of solid waste particles with successive motions of the reciprocating bars.136 Figure 5-19: Simulation results of the movement of a solid waste particle on the Reverse Acting Grate: Top graph: tracing a behavior of one particle on the bed. The solid particles that move slower (particles in the left part of each profile in Figure 5-20 have been pushed . In cells #2 and 5 the particle seems to stay put for a while. Under the combination of downward gravity and upward motion of the reciprocating bars. all particles travel gradually to the outlet.

3 n=10 0. It is evident that solid wastes that are more mixed have a better chance of Inlet 0. 1 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 Cell Number (posit ion in t he grat e) n=80 Figure 5-20. 10. being combusted completely. On the other hand. . 40. Without additional feed coming in. the probability of particles staying in a specific cell. approaches zero. 80). because all solid particles travel toward the outlet and eventually reach it. The regular perturbations in the n=40 and n=80 profiles are due to the effect of alternatively passing over stationary and reciprocating bars. Change in tracer particles distribution profiles over grate with number of movements of reciprocating bars (n=5. solid particles that travel faster than the bulk of the solid wastes (particles in the right part of the each profile in Figure 520 will not burn completely. 5 0. This action contributes to mixing partly combusted material with newly fed solid wastes. after an infinite number of movements (step n = ∞. 4 Outlet n=5 Probability 0.137 upward by the reciprocating bars. 2 n=40 0.

a two-dimensional stochastic model (multi-layer model) for MSW mixing was developed in order to expand the single layer of MSW bed to the multi-layer. a new method for investigating mixing on the grate of a WTE combustion chamber. The following section (5-2-2-b) shows the results of a two-dimensional stochastic simulation of MSW particle mixing. Based on these calculation results. In the first test of the one-dimensional stochastic model presented in this section. b) Two-dimensional simulation Figure 5-21 shows the MSW particle distributions with successive motions of the reciprocating bars. was presented. This information would contribute to the characterization and quantification of the mixing process. The effect of particles passing alternatively over fixed and .). was simulated. etc. using the one dimensional stochastic model. Figure 5-21 represents residence time distributions (RTD) over the grate. The peak of each distribution is moved toward the outlet of the chamber because all solid waste particles travel from the inlet to the outlet. Since the reciprocating bars move at a constant frequency (a typical value is around 11-13 strokes/hour). affected by the motion of the reciprocating bars in combination with the downward force of gravity. This method may provide a useful for predicting the actual path of particles. the bottom layer of the bed. for a certain grate configuration (frequency of reciprocation bar motion. which is in contact with the bars of a reverse acting Martin grate. length of travel.138 In summery of this section.

139 reciprocating bars results in the regular perturbations eminently shown in the n = 40 and 80 profiles. .

(d) n = 40. (b) n = 10. (c) = 20.(e )n= 80 .140 (a) n = 5 (b) n = 10 (c) n = 20 (d) n = 40 (e) n = 80 Figure 5-21: Change in particle distribution profiles over grate with number of movements of reciprocating bars (a) n=5.

particles in layer A exhibit the highest movement toward the outlet because layer A is near the surface of the MSW fuel and some particles can slip on the surface or roll down along the surface. although it has the shortest residence time and path.e. According to the assumptions made in defining probabilities within the matrix P. The bottom layer E is the nearest layer to the reciprocating bars and particles in this layer are exposed to the driving force of the reciprocating bars against the feed direction. 5-22 (b). the rest of the path along the second half of the grate (zones 17 to 32) seems smoother. The case shown in Figure 5-22 (c) expresses mixing of the particle in the first half of the chamber (zones 1 to 16).. The particle path in Figure 5-22 (a) shows that this particle travels mainly in the layer A and B where the probabilities of the movement forward to the outlet direction are higher. On the other hand. the step number n is smallest (n = 67) than in the other cases (i. In the case of Figure 5-22 (a). The particle can be located in the surface layer where chemical reaction rates are high.141 Figure 5-22 shows the motion (paths) of traveling MSW particles in layers of the MSW fuel above the reverse acting grate for various numbers of reciprocating motions. (c) and (d)) and the particle has the shortest residence time on the grate. The path of a particle shown in Figure 5-22 (b) indicates that this particle remained for a long period of time in zones 25-30. These zones include the burnout point and the particle must be almost ash in an actual combustion chamber. Figure 5-22 (d) shows a path of a MSW particle that reached ash bin when .

. due to the movement of the reciprocating bars. This indicates mixing of this particle near zone 10 and between zone 20 and 25.142 n = 401.

(c) n=213 and (d) n=401 .143 (a) n =67 (b) n=106 (c) n=213 (d) n=401 Figure 5-22: Movements (paths) of MSW particles on the Reverse Acting Grate: (a) n=67. (b) n=106.

of 0. they are measured as shown in section 3-3 and 51-2. a standard deviation. these assumed probabilities were found to be relatively low. a standard deviation. a βparameter of 0. an αparameter of 12. The probabilities in the transition matrix P were assumed based on the successful results of the one-dimensional model. such as size. σ . 5-2-3. In order to determine these probabilities in transition matrix P. and a covariance. The PSD of ash resulted in having a constant a value of 0.04. and a covariance. shape.17. This model can be as a tool for charactering and quantifying the mixing process.144 In summary.8. the PSD of MSW has an “a” constant value of 2. using full-scale physical section model. as shown in Figure 5-22-d. As summarized in Table 5-1. and density. CV. However. Dots represent experimental data and lines are the predicted distribution. Compared with the measured PSDs shown in Figure 5-23.5889. a mean μ of 2. σ. CV. 401 strokes of reciprocating bars is too high.289.3 x 100. even though. an α parameter of 4. of 0. Lower probabilities results in longer residence time and more strokes to reach ash bin (for example. Results from mass reduction model Figure 5-23 shows the simulation results of the gamma distribution discussed in section 4-3-2. was not taken into account. a β parameter of 3. a mean µ of 12. although the properties of MSW particles. the predicted PDSs have a . of 6.5. the particle remained same zones many times in this case).2.4. of 0. a novel two-dimensional model simulating the mixing of the MSW fuel on the Martin reverse acting grate has been developed using the stochastic simulation.5 x 100.

dots: experimental data) . Particle Size Distribution 25 Ash Relativi Frequency % / cm (by Number of Particles) 20 MSW 15 10 5 0 1 10 Particle Size (cm) 100 Figure 5-23: Particle size distributions (PSD) by particle numbers of residential MSW and combined ashes in NYC: (lines: estimated gamma distributions. This tail group of MSW particles contains a small peak.145 similar shape except for the asymptotic tail of the MSW PSD (30-40 cm). That is primarily because some of physical phenomena such as compaction and breakage have not been included in this combustion model asyet.

that is difficult impossible to measure experimentally can be very useful understanding and characterizing the mixing phenomena on the grate.5 0.2 2.4 8.83 5. Each particle travels along different paths in the MSW bed. σ Covariance 0.3 X 100 12 0.045 Table 5-1: Particle size parameters obtained from residential MSW and combined ashes in NYC 5-2-4.8 Experimental 14.5 X 100 4 3.289 CV a α β 0.3 6.594 2.354 0.04 Experimental 2.17 7.97 0.589 dev. Figure 5-24 shows the calculated movements for 12 particles traveling from section 1 through 32.146 Ash particle size distribution Parameter Mean µ Standard 0. but trends are consistent for a given size .32 MSW particle size distribution Predicted 12.821 Predicted 2.508 0. Integrating model for flow and mixing with data from the full-scale physical model Visual representation of the computed particle movement behavior.29 0.

can easily drop into the empty space. and for applying separation and recycling technologies to the processing of industrial waste and residential MSW.147 especially in the vertical (depth) direction.. Immediately. 2005). 2005) and results from the slight differences in vertical probabilities as a function of size. This mechanism of size segregation can be observed in rock. N. and Miles. and granular movement. The moving grates create empty space after one stroke from the top to the bottom position. powder. and Miles. the size segregation appears as shown in Figure 5-25. This size segregation has been called the Brazil Nut Effect (BNE) (Mohabuth. The general trend indicated by the simulation is that small particles tend to migrate downward (toward the grate surface) and large particles tend to migrate upward (toward the free surface) as they move along the bed. After several reciprocations of moving bars. sand. Larger particles are less likely to drop into this newly created vacancy because their comparatively larger size offers more opposition to easily slipping into this limited empty space. The segregation mechanism we observed in this study is illustrated in Figure 5-25.. N. because of the unstable conditions. N. Some studies have been carried out employing this vertical selectivity mechanism for electronic waste (Mohabuth. N. . particles whose size is less than the grate height.

medium particles (M).148 : Figure 5-24: Calculated particle path over the grate as a function of size: small particles (S). and large (L) particles .

/hr. Residence time distributions of a flow material were originally defined by Levenspiel (Levenspiel. As we discussed earlier. dimensionless concentration versus residence time t) for small. 1999) The originally 1999). while decre ctively. it is reasonable that the difference of . distributions for small particles prominently appeared when the reciprocation speed eciprocation exceeded 30 recip. respectively. two peaks of residence time . mean residence times with size comes from the following reasons: (1) Small and medium particles are pushed by the reverse acting grate because their diameters are nearly the same or smaller than the height of the moving bars. medium.149 Figure 5-25: Brazil Nut Effect (BNE) in a packed MSW bed : Figure 5-26 shows the residence time distributions (RTD. h. motion of the reverse acting grate increases the mean residence time of small and medium particles to 106 min (68%) and 69 min (9%). (2) Larger particles are less likely to be caught by the grate so that their motion is less likely to be in . and large particles for a ) grate speed ranging from 15 to 90 reciprocation/hour. In addition. decreasing that of large particles to 51 min (17%). O.

This effect is manifested by the presence of two peaks that appear only in the residence time exit concentration distributions of the small particles. The full data sets of RTDs (C and F diagrams) for grate speed ranging from 15 to 90 recip.150 opposition to the flow direction. 5-28. . rolling down the top surface along the flow direction. (3) A special mode of transport. Due to the particle motion behavior (1)-(3). a vertical selectivity mechanism known as the Brasil Nut Effect (BNE) develops that tends to keep small particles in the bottom of the MSW bed./hr are shown in Figure 5-27. is available exclusively to those particles near the free surface of the bed. These two opposing motions at the bottom and the surface of the bed enhance MSW particle mixing and are responsible for the different residence time distributions. and 5-29.

5 1 0. and large particles with a : reciprocation speed of 90 recip.5 large-90 =51min. 40 60 80 100 Residence Time t (minutes) 120 140 Figure 5-26: C (top) and F diagrams (bottom) for small.6 0.5 2 t 1.9 0.3 0.4 t 0.2 0.7 0. t small-90 =106min.151 C Diagram 4 3.5 Dimensionless Exit ConcentrationC 3 2. =51 t medium-90 =69min. t small-90 =106min. medium./hr.1 0 20 large-90 =51min.5 0. . t medium-90 =69min.8 0. 0 20 40 60 80 100 Residence Time t (minutes) 120 140 F Diagram 1 Dimensionless Cumulative ConcentrationF 0.

30 60 70 80 90 100 Residence Time t (minutes) 110 120 Figure 5-27: C and F diagrams for small particles: Dimensionless exit concentration C (top) and : Dimensionless cumulative concentration F (bottom) versus residence time (min) with different reciprocation speeds ranging from15 to 90 recip./hr .152 F Diagram 1 0.2 0.8 0.3 0.1 0 40 50 60 t =63min.4 0. 90 t =82min. t =68min.7 0.9 Dimensionless Cumulative Concentration F 0. 0 t =64min. 15 t =106min.5 0.6 0.

0 t =65min./hr .5 0.7 0.5 3 2. 15 0 t =65min. 30 t =67min. 30 t =67min. 15 t =63min. 60 t =69min.4 0. 90 90 100 1 Dimensionless Cumulative Concentration F 0.3 t =63min. 0.5 Dimensionless Exit ConcentrationC 4 3.5 1 0.2 0.8 0.153 C Diagram 5 4.5 0 40 50 60 70 80 Residence Time t (minutes) F Diagram t =62min.1 0 40 t =62min.6 0. 90 50 60 70 80 Residence Time t (minutes) 90 100 Figure 5-28: C and F diagrams for medium particles: Dimensionless exit concentration C (top) and Dimensionless cumulative concentration F (bottom) versus residence time (min) with different reciprocation speeds ranging from15 to 90 recip. 60 t =69min.9 0.5 2 1.

5 0 20 60 t =63min. 0.5 2 1. t =58min.4 0. 0 40 50 60 70 Residence Time t (minutes) 80 90 100 Figure. 90 0 30 40 50 60 70 Residence Time t (minutes) F Diagram 80 90 100 1 Dimensionless Cumulative Concentration F 0.6 t =58min.7 0. 5-29: C and F diagrams for large particles: Dimensionless exit concentration C (top) and Dimensionless cumulative concentration F (bottom) versus residence time (min) with different reciprocation speeds ranging from15 to 90 recip. t =63min. 30 t =61min.5 3 2.5 t =54min. 15 t =51min.1 0 20 30 90 60 30 t =54min. t =51min.5 Dimensionless Exit ConcentrationC 4 3.9 0. 15 1 0.2 0.154 C Diagram 5 4.8 0./hr .5 0.3 0. t =61min.

/hr. For Rr < 30 recip.et.1) and can be considered to be a function of the number of moving bars. For small particles. whose ratio d/h controls the relationship between De and Rr . al. These numbers represent the range at which the reverse acting grate is operated in commercial WTE plants. The De for medium particles has a weak linear relationship with Rr . The constant b has an approximate value is of 2.155 Mixing Diffusion Coefficient Figure 5-30 shows the mixing diffusion coefficient De for different particle sizes and grate reciprocation speeds Rr . This range of De is in good agreement with the results of full-scale furnace tests in another study (Yang Y. these relationships of grate reciprocation speeds Rr and mixing diffusion coefficient De can be described as follows: { { I I $ $ $ 4 (Eq. The mixing diffusion coefficient De for all particle sizes increases linearly for speeds up to 30 recip. In this full-scale test De . throughout the entire speed range from 15 .90 recip.. The parameters d and h are the particle size and height of the moving bars. De increases exponentially to 192 cm2/min. The mixing diffusion coefficients range from 11 to 99 cm2/min and monotonically increase for all particle sizes that with increasing reciprocation speed from 10 to 60 recip./hr./hr. 2005).B. the coefficient De for large particles increases at a much slower rate and reaches a value of about 45 cm2/min when the reciprocation speed reaches 90 recip. respectively. In contrast. 5-1) where a is a constant ( ≈ 1. also shown in Table 5-2./hr./hr.8.

Though no mention was made of the tracer size. their densities were given as values 225 and 770 kg/m3. though the type and operational conditions of the traveling grate are not specified. Figure 5-30: Mixing Diffusion Coefficients versus grate reciprocation speed : for different particle sizes .156 varies from 27 to 109 cm2/min. Particle motion along the MSW bed is highly dependent on the experimental conditions such as the tracer size and t density as well as physical properties of the bed itself.

The advantage of this effect is that small particles that are mostly burned out can transfer their heat to large particles which need more heat and take more time to be thermally processed. 95 min. although they require a shorter time for complete combustion. Pushed by the reverse acting grate. in real combustion chambers. The burn-out particles that . small particles. Primary air at room temperature comes from the bottom of the MSW packed bed. According to a study of wood particle size versus conversion time. respectively (coversion time ≈ d 1. and the motion can be characterized as a reacting flow. medium and large sized particles at temperature of 900oC is about 12 min. the particle size is reduced because of volatilization and combustion processes. the primary air is warmed up by these small particles.. especially during the drying and gasification processes.157 Material Kaowool boards Insulation bricks Refractory bricks Density (kg/m3) 225 480 770 De (cm2/min) 109 88 27 Table 5-2: Mixing Diffusion coefficient with different densities [full-scale test] (Yang et al 2005) As noted earlier. 1998) as shown in Figure 5-31. 46 min. Due to the burn-out of the small particles pushed by the reciprocating bars and due to the high residence times at the bottom of the MSW bed. the 95% reaction time for small. stay in the combustion chamber longer than medium and large particles. The mixing diffusion coefficient De for small particles is higher and they are easily dispersed along the horizontal direction.62 ) (Petek J.

1998) .” This study showed that this is due to the fact that the grate motion effects only small/medium particles of diameters less than the bar height. and whose mass volume is less than 20% of a typical MSW (Nakamura et al 2008). within the range of 15 to 90 recip. the calculation should be carried out for each single time step n (reciprocation). since the initial condition. instead of only the previous step. “the speed of the reciprocating e bars does not look that it affects burnout of the entire MSW bed.158 turn into ash have a higher heat capacity than MSW and so they transfer heat to the large particles and the primary air. when the stochastic Nakamura. because the Markov property will be applicable. model is combined with particle size change due to combustion. Figure 5-31: Wood particle conversion time vs particle diameter (Petek J. In this case. Operators of the reverse acting grate have observed that. the stochastic model simulation will become much more complicated. . In the future research./hour.

The probability of particle movement with bed height (layer) of MSW was calculated.2 and 1. All sizes of tracers have the highest probabilities in the bottom layer E. when the MSW in the physical model was loaded over 80 cm).3 and 1. For example. 60. The Aspect Ratio Distributions are 1. the tracer particles can slide or roll easily along the surface of the bed because there are no particles resting on them. Tracer particles exhibit a different behavior in different cells. medium. .4 and the Roundness Distributions 1. 2. respectively.5 and of the ash particles at 2.159 Chapter 6: Conclusions The principal findings of this study can be summarized as follows: 1. the next highest probability of particle movement is in the top layer B (60-80 cm in the physical model. In sphericity distributions. and 80 cm) were estimated by means of a full scale physical model of the reverse acting grate. In the top layer.4. the peak of the MSW particles was found to be at 1. a tracer particle located in the bottom of the bed cells (layer E) moves quickly because it is pushed by a reciprocating bar. 40. 3. The movement and state diagram (transition graph) of three particle sizes (small. and large) at different MSW bed heights (20.

channeling probability Pch./hr. The calculations of the mixing diffusion coefficient De showed that it was affected principally by the ratio of the height of reciprocating bars. The motion of the reverse acting grate. Different particle sizes result in different residence times according to the Brazil Nut Effect (BNE): Larger particles rise to the surface while smaller particles migrate to lower depths of the bed where the reciprocating bars push them backwards against the main direction of flow due to the gravity force. respectively. critical probability Pc.160 4. This model can be a useful tool in characterizing and quantifying the flow and mixing phenomena on a combustion grate: 5. h. The governing parameters such as void probability Pv. resulted in the net effect of increasing the mean residence time of small and medium sized particles by 68 % and 9%. and ash probability Pa were determined by using this model. at speeds ranging from 15 to 90 recip. while decreasing that of large particles by 17%. The percolation model has shown the mechanism of transient phenomena such as channeling and break-up of solid waste. 6. Also a novel two-dimensional model simulating the flow and mixing of the MSW fuel on the Martin reverse acting grate was developed using the stochastic simulation. to the diameter of the tracer particle. Although the physical model tests were all . 7.

A grate speed does not affect burnout of the entire MSW bed. stay in the combustion chamber longer than medium and large particles. Suggestions for Future Work This study developed a quantitative analytical tool for examining the mixing of solid waste particles of different sizes during the combustion process in a WTE chamber. small particles. The combination of stochastic and full-scale physical modeling can also be used for comparing and .161 carried out at one h value (13 cm). especially during the drying and gasification processes. the results showed conclusively that the diffusion coefficient De for a given particle size distribution will increase with grate bar height. Pushed by the reverse acting grate. and whose mass volume is only 18% of the total NYC-MSW. The advantage of this effect is that small particles that mostly burned out can easily transfer their heat to large particles which need more heat and take more time to be thermally processed. although they require a shorter time for complete combustion (coversion time ≈ d 1. This is due to the fact that the grate motion effects only small/medium particles of diameters less than the bar height. The results can be used to assist in the evaluation of operational parameters as well as geometric parameters of a reverse acting grate.62 ). 8.

Analysis of the mixing phenomena for different shapes and densities of MSW particles. as illustrated in Figure 6-1 3. because losing its Markov property. The calculation should be carried out in each single time step (reciprocation). as was done for the reverse acting grate in this study. gasification. or of parts of the reverse acting grate moving at different reciprocating speeds can be investigated using full-scale physical models and stochastic simulations.162 evaluating various types of traveling grate systems and designing the next generation of combustion chambers. The effect of mixing phenomena for different operation conditions (reciprocation speeds) on the drying. 2. A combustion model can be combined with this stochastic mixing model. . is necessary to more accurately simulate the actual particle motion in the combustion chamber of an MSW packed bed. can be analyzed. The forward acting and roller grate systems can be investigated by means of full-scale physical models and stochastic simulation. In this case. and combustion stages. along with the size effect investigated in this study. 4. the stochastic model simulation will become more complicated. A combination of traveling grate systems such as the reverse acting grate and roller grate. Suggestions for future work are as follows: 1.

The Discrete Element Method (DEM). novel traveling grate system systems . may be popularly used for mathematical simulation of MSW mixing and combustion. can simulate concept design for s. Figure 6-1: Stochastic models calibrated by full scale grate tests. Currently DEM has started being used for solid fuel conversion on forward acting grate n (Simsek et.163 5. al. on when much faster computers become available in the future. based on Newtonian equations. 2008) 2008).

” Fifth Edition. http://130. and Meissner. Mizonov (2002).” Proc. Simplified mathematical model of combustion in stoker systems. Inc.-14. California Geographical Survey (1997). Howard. Berthiaux. 2004) Beckmann.. pp.ansys.asp Allen. INFUB 2000. Digital Atlas of NYC. 2000. 61–70. M. Porto. H. pp.com/default. Terence. 5th European Conference on Industrial Furnaces and Boilers. and Scholz. CFX software for thermal and fluid simulation http://www. Portugal. Beckmann. M. and V.. M. P..2/ny_3. R.. “An experimental method and Markov chain model to describe axial and radial mixing in a hoop mixer.. Pittsburgh. B. Scholz. Anthony. 159-167. Aoun.166. Powder Sampling and Particle Size Measurement. B.html (accessed June 6. "Rapid devolatilization of pulverized coal. H. . R.” Powder Technology 128. D. (1975). “Residence Time Behaviour of Solid Material in Grate Systems." Fifteenth Symposium (International) on Combustion. J. Chapman & Hall (1997) Aoun-Habbache.. 11. William A.164 References ANSYS. Bowen. H. The Combustion Institute. “Particle Size Measurement Volume 1. C. 04. Third European Conference Industrial Furnaces Boilers 18/21 (1995). Powder Technology Series. M. Hottel.124..

“Residence Time Behaviour of Solid Material at Grate Systems. (2001). T. S.clustan.. Hernandez. 04. C. On predicting roller milling performance Part I: the Breakage equation. J. Hammersley. 729-738. J. and Hill. B. “Percolation Processes I.. A. 9.. 629-641 Badzioch. 139. and Nussbaumer. Clustan Ltd. (1970). 208-221. P.” Proceedings of the Cambrigde Philosophical Society 53 (1957).. "Modelling and simulation of time-dependent coal combustion processes in stacks. R.." Fuel.” Proc. and Hawksley.-14. 79.. (1991): “Markov Chains Gibbs Fields. Clustan software for cluster analysis for solving classification problems http://www." Ind. (2000). Porto. P. C. and Queues. (2000). R.. G. F. Portugal. Powder Technology 115 pp. M. "modelling wood combustion under fixed bed conditions. 82. Bruch. Campbell. "Kinetics of Thermal Decomposition of Pulverized Coal Particles. M. Peters. J. 521. W. Crystals and Mazes. 11. F. P. N. F. Chern. Bremaud.. INFUB 2000. Monte Carlo Simulation... 987-997. 234-242 Chejne. Chem.” Springer. Webb. (2004). and Hayhurst. Broaddbent. S. J. M. Scholz.165 Beckmann. W. "Does large coal particle in a hot fluidised bed lose its volatile content according to the shrinking core model?" Combustion and Flame. 5th European Conference on Industrial Furnaces and Boilers." Fuel. Eng. A.. Florez.. (2003). G. S.com/ .

K. http://www. and Busch.org. K. K. 84.” Proceedings of the National Incinerator Conference. Inc. S. M.... and Kaitano.” Chemosphere 42 (2001) 545-550 . B. Y. commercial computational fluid dynamics (CFD) http://www. ECOMB AB. W. 74 No. and Kuo.fluent. The Ecotube system for optimizing combustion processes and principally. "The modeling of the combustion of high-ash coal-char particles suitable for pressurised fluidized bed combustion: shrinking reacted core model. (1995) “Estimation of the heating value of oily mill sludges from steel plant. 1136-1143. Marcel Deeker. R.. O. Heomagus. 1918-1921 Essenhigh.. pp.. R. State-of-the Art Report. 261-271.com/ Gaur. K.ecomb. Energy from Waste. Part1: Statement of the problem. H. H. www. (1970): “Combustion and Emission Phenomena in Incinerators: Development of Physical and Mathematical Models of Incinerators. H. 12. T. Kim. (2005). T. H.. Fluent software. ” Fuel Vol.se/ Everson. Park. New York.166 Cho. R. Fluent Inc." Fuel. ASME. pp. and Reed. and Lee. “Reduction of combustion by-products in WTE plants: O2 enrichment of underfire air in the MARTIN SYNCOM process.. Gohlke. Lee.IWSA (International Solid Wastes Association). S..wte. Thermal Data for Natural and Synthetic Fuels.. (1998). J. H.

M.gec. S.. 691 . “The State of Garbage in America. "Mathematical modelling of pyrolysis of large coal particles ..78 .. (1980). 78. Jia. and Matsumoto. Themelis.J. S. Koga. and Zhang." Chemical Engineering & Technology.. "Diffusive mixing in flowing powders. 93-101. H. A. 60. N. S. J. 1979-1987 Grimmett G. C.jp/JSIM_DATA/WASTE/WASTE_3/html/Doc_435. N. "Investigation of Heavy Metal Release during Thermal Waste Treatment on a Forward-Acting Grate by Means of Radiotracers.html Jentsch. S. Millarth K. 557-566. "Gas-solid reaction model for a shrinking spherical particle with unreacted shrinking core. L. P. Homma. K..." Powder Technology. Glodstein.estimation of kinetic. and Biollaz.. Vol. “Percolation” Springer (1991) Heidenreich.. C.. R. Mitsubishi Heavy Industiries Technical Review. 26(1).. 31-41 Kira et al. 4971-4980.696. M. Chemical Engineering Science 59 pp. M." Chemical Engineering Science. Japanese Advanced Environment Equipment: Mitsuhishi Martin Refuse Incinerator: http://nett21. (2005).” BIOCYCLE. S.167 Gbor. (1998). (2004). Yan. 26(6)... Beckmann. D.. M. (2003). C. and Hogg. Davidovic.. Ogata. January (2004). Q.38 No. Development of New Stoker Incinerator for Municipal Solid Wastes Using Oxygen enrichment. T. Hwang.. Critical evaluation of coupling particle size distribution with the shrinking core model. Kaufman." Fuel. K.2 (2001) p.

(1999).” Powder Technology 114 (2001) 89-95 Martin GmbH: The MARTIN reverse-acting grate http://martingmbh. J. Boston...168 Klasen. 411 Kunii..” IFRF Combustion Journal. John Wiley & Sons. February (2002) ISSN 1562-479X Kobayashi.de/englisch/index2. Symp. Goh. Riccius. and Sarofim." Int.. N. A.htm Lim. The recovery of recyclable materials from Waste Electrical and Electronic Equipment (WEEE) by using vertical vibration separation. "Coal Devolatilization at High Temperature. Chemical reaction engineering. Swithenbank. N. Inc. Ultimate analysis of the elements C. S and O Downloadable literature http://www. Levenspiel. on Combust.com/organic/elementalanalyzers/chns. T. Fluidization Engineering. F. B. H. Inc. Butterworth.. Pittsburgh..leco. H. K. R. Y. “The modelling of solid mixing in municipal solid incinerators. John Wiley & Sons. and Miles.. V. J. N.. and Goerner.. (1991). 60-69 (2005) . C. and Levenspiel.htm Mohabuth. LECO Co. Resources Conservation & Recycling 45. Nasserzadeh. Howard. “Numerical Calculation and Optimization of a Large Municipal Solid Waste Incinerator Plant. N. (1977). O. MA. Combustion Institute. O. D. O.

. Dziugys. (2005). Castaldi. Washington. Nakamura." Proc. Castaldi.." KERNFORSCHUNGSZENTRUM KARLSRUHE. M.J. KARLSRUHE (GERMANY). J.. and Krebs.. ( 2003).. B.. Hans Hunsinger. "An approach to qualify the intensity of mixing on a forward acting grate. M. (1998) Peters. N. 12th annual North American Waste To Energy Conference (NAWTEC12). . and N." Proc. Zhang.. J. and Themelis. K. Millrath.. (2004) Nakamura. PhD thesis. H.to-energy Combustion Chambers. pp. Japan Nakamura. Georgia.J.169 Nakamura.." 2003 ASME ICMEE.. "Modeling of Waste-to-Energy Combustion with Continuous Variation of the Solid Waste Fuel. L. M. M. D. Themelis.. Peters.J." Chemical Engineering Science. "A model for numerical simulation of devolatilization and combustion of waste material in packed beds. A. 60(6). Measurement of Particle Size and Shape of New York City Municipal Solid Waste and Combustion Residues Using Image Analysis. (1994). N. N. B. and Themelis. Sendai.. and Themelis.C. M. M. (2008)"A 2-dimensional stochastic modeling for Municipal Solid Waste (MSW) particle mixing within a waste-to-energy (WTE) combustion bed" submitted to International Journal of Thermal Sciences Petek J. 16th Japan Society of Waste Management Experts (JSMWE) Fall Conference.J. Experimentelle Untersuchung der Pyrolyse in inerter und reaktiver Atmosphare unter den Bedingungen der Wurfbeschickung. (2005). 1649-1659. TU Graz. "Modeling of Solid Waste Flow and Mixing on the Traveling Grate of Waste. 1-3.

. N.. Y. Philip J.” ISWA Journal of Waste Management and Research (2002): 20: 223-233 ... Ph. 23. L. 19(1). (1973).. Portugal. 191 Tchobanoglous. and Nussbaumer. John Adam.. E. B. and Vigil. M. Krüll.. “Particle Size Spectrum and Compressibility of Raw and Shredded Municipal Solid Waste. Germany Ruf. Brosch... "Meaturements and particle resolved modelling of heat-up and drying of a packed bed. H. 2008-03-27 Smoot.” CUTEC-Institut GmbH. Plenum (1985) Sohn. and Brady. “Numerical Simulation of Solid Fuel Conversion on a Forward Acting Grate with a Coupled CFD / Discrete Element Method (DEM)” INFUB-8th European Conference on Industrial Burners and Furnaces. Reindorf. S.” The University of Florida. and Vodegel. New York (1993) Themelis." AIChE Journal. J. Vilamoura. 291-306. "On the maximum temperature rise in gas-solid reactions.” McGraw-Hill. V." Biomass and Bioenergy. G. “Integrated Solid Waste Management. T. F. (2002). H. Thesis (1974) Simsek. and Smith. H. S.. Scherer. “Energy Recovery from New York City Solid Wastes.. E. “Coal combustion and gasification. B. Douglas. Theisen. March 25-28.. Bruch. Y.170 Peters.” The Plenum Chemical Engineering Series. C. “Dynamic Grate-Model as a Tool for the Optimization of Thermal Waste Treatment Plants. H. Wirtz S. Schroder.. T. Kim.D.

. the Fraunhofer Institute for Environmental. No. G. Swithenbank. Chemical Engineering Research and Design. C. “Mathematical modelling of MSW incineration on a travelling bed. Brochure of Waste-to-energy Combustion System. v 81. New York. NY. B. Doctoral Dissertation.” McGraw-Hill. Studies on combustion of carbon particles in flames and fluidized beds... Nasserzadeh. M. Theisen. S (1993): “Integrated Solid Waste Management. “Erstellung eines Modells der Vermischung von Abfall auf Rostsystemen unter besonderer Berucksichtigung der Vermischung .com/ Wolf. Nasserzadeh. H. Simulation of channel growth in a burning bed of solids. N. Nakamura.. Inc.. “Advanced Automatic Combustion Control System for Refuse Incineration Plant.. Goh.. 231-244 (1955). D. Swithenbank..... Zakaria. Y.B. and S.ein Beitrag zur Simulation von Abfallverbrennungsanlagen”.. R. Goodfellow.. http://www. pp. Kira. V. p 221-232. Ono. Yang. n 2. Safety and Energy Technology (UMSICHT) (2005) Yagi.” MHI Technique Vol. Mitsuhashi. J.vonrollinc. Y. Y. Kunii. Vigil. Von Roll. S.171 Takatsudo. Y. H. M.” Waste Management 22 (2002) 369-380 Yang. V. 2003... 2 (1999-5) in Japanese Tchobanoglous. J..R. 39. In: Fifth Symposium (International) on Combustion Reinhold. New York. J. .

(2005).. N..N. B. J. R.172 Yang Y.. Sharifi V. "User's Manual. and Swithenbank. Swithenbank J. A diffusion model for particle mixing in a packed bed of burning solids. V." . 84.. The Fluid Dynamic Incinerator Code (FLIC) For Modelling Incinerator Bed Combustion. Fuel. Yang. Sharifi. Y. Goh..N.. Goodfellow J.. (1999).. Y. Lim C. 213-225.B.

5 6067. 4069.5 2600.91e+003 1.87e+003 2.SUWIC PROJECT NewProject Volatile Release [kg/m^3.75e+003 5.6 1356.7e+003 1.6e+003 1.06e+003 957 851 745 638 532 425 319 213 106 0.hr] 20 hr 00 min 00 sec 000 msec Thu Nov 14 11:35:31 2002 FLIC (Version1.17 0. K 1.75 1733.81e+003 1.8 2034.3 6934.5 6067. 866.13e+003 1.22 7. .1) .92e+003 1. 4333.2002) * Data from a commercial reverse acting grate indicates the peak temperature is located in the beginning of grate more than FILC calculation results.28 ( in mm ) 0 866.87 4.04 5.71e+003 6.62e+003 8.44e+003 958 479 0.28 ( in mm ) 0 678.4 3391.3 14.6 1356.3 3467.1 20.7 9.35 1.7 Temp.SUWIC PROJECT NewProject Solid Temperature [K] 20 hr 00 min 00 sec 000 msec Thu Nov 14 11:31:52 2002 FLIC (Version1.173 Appendices Appendix A: Bed calculation by FLIC* 4069.1 2713.5 6067.79e+003 4.75 1733.15e+003 7.28 ( in mm ) 0 678.6 16.7 2.6 678.5 6067.32e+003 1.3 3467.4 15.1) . because of the effect of mixing agitated by the motion of the reverse acting grate.6 678.75 1733.67e+003 7. PROJECT NewProject Residual Water in Solid (mass %) 20 hr 00 min 00 sec 000 msec Thu Nov 14 11:32:25 2002 FLIC (Version1. 4333.8 5200.1 2034.1 2713.26e+003 1.38e+003 1.3 21.1 2034. 4069.3 6934.1 12.SUWIC c) Combustion d) Temperature Figure A-1: Calculation results of Bed modeling (Nakamura et al.9 11.4 2713. 3391.28 ( in mm ) 0 866.45e+003 1.3 6934.75 1733.27e+003 4.8 5200.06e+003 1000 935 871 807 743 679 615 551 487 423 359 295 3391.5 2600.5 2600.02e+003 1.1) .4 2713.7 22.51e+003 1. 866.6 9.19e+003 6.8 1356.19e+003 1.31e+003 3.8 5200.83e+003 3.39 8.49e+003 1.5 2600. 4333.8 2034. 4333.8 1356.3 3467.69 3.8 5200.38e+003 1.52 2.4 3391.17e+003 1.3 6934. PROJECT NewProject Combustion Heat Release [kW/m^3] 20 hr 00 min 00 sec 000 msec Thu Nov 14 11:33:26 2002 FLIC (Version1.7 10.3 3467.8 17.4e+003 1. 18.28e+003 1.35e+003 2.23e+003 5.SUWIC a) Water evaporation b) Volatilization 4069.1e+003 8.1) .

174 Appendix B: Chamber Calculation by CFD Flue gas Air injection MSW Air/recirculated flue gas injection Gas flow (Computed from a bed model) Ash Figure B-1: A grid structure of CFD modeling (Nakamura et al.2002) .

175 Figure B-2: Calculated contours of Velocity (m/s) (top) and Total temperature (K) (bottom) .

176 Figure B-3: Calculated contours of Turbulence intensity (top) and mass fraction of hydrocarbon concentration (bottom) ((bottom) .

177 Figure B-4: Calculated mass fraction of oxygen concentration (top) and of CO2 concentration (bottom) .

where f (E ) = exp(−( E − E0 j )2 / 2σ 2 ) j Distributed Activation Energy Model (DEAM) Coal (large particle) σ 2π N/A Heat transfer (convection. t ) )dt  f ( E ) dE r 2 dr      0    Heidenreich et al. 1995) (Shin et al 2000) Beckmann et al. Thiele modulus Heat transfer Heat transfer (Examined from 750 to 950 C) Shrinking-core Model (Peters 1994. FLIC MSW Gaussian distribution of activation energies for waste devolatilization. radiation. 1998) Everson et al. 900. and heat transfer for both gases and solids. Coal. radiation) and Mass transfer (diffusion) The governing equations for mass. (Drying. (Ahmed et al. 2005) General Coal Sohn Chern and Hayhurst (Sohn 1973) (Chern and Hayhurst 2004) Cheijne et al. Coke Drying. Wood Peters. Peters et al. Coal General Gas-solid reaction Devolatlization [Coal → char + volatiles] Drying [Wet coal → coal + H2O] Devolatilization [Volatiles → B1CO2 + B2CO + B3O2 + B4N2 + B5H2O + B6H2 + B7CH4 + B8SO2 + B9NO + B10C2H6 + B11H2S + B12NH3 + B13tar] Char Oxidation [Coal +αO2 → (2β-1)CO2 + (2-2β)CO +a1H2O + a2NO + a3SO2] Volatiles generation by DEAM CO + H2O → CO2 + H2O 2CO + O2 →2CO2 2H2 +O2 → 2CO2 CH4 + 2O2 → 4CO2 + 6H2O 4NH3 + 5O2 →4NO + 6H2O n ∞ R0 j Heat and Mass transfer Packed bed model and Shrinking-core Model (only for char oxidation) (Chejne et al. Wood) Solid Reaction. Devolatlization. Char oxidation ) Gas Reaction Volatile combustion Heat /Mass Transfer References Ahmed et al. 850. conduction) from 400 to 1000C Shrinking-core Model (only for devolatilization) (Heidenreich et al.k RT k =1 N/A Heat transfer (convection. 2005) Table C-1: Comparison of combustion models . pyrolysis. 2002) During drying and devolatilization the diameter is assumed to remain constant. radiation ) transfer Model Name Fuel Type (MSW. radiation. and 950 C) Shrinking-core Model (only for char oxidation) Thiele modulus (Everson et al. 800.178 Appendix C: Comparison of combustion models Developer’s Name Remarks Cellulose MSW Wood. oxidation [Wood → char + volatiles] Heat (radiation) and Mass transfer General Gas-solid reaction Heat and Mass transfer pyrolysis [Cellulose → volatiles] 1-demension Heat (convection. 1989) (Beckmann et al. momentum. conduction) Mass transfer (diffusion) Single-particle model Packed-bed model (including shape factor) SIMPLER algorithm Connectivity with other particles. 1999. mixing effect between volatile matter and oxidant in the volatile combustion rates Heat (convection. Peters et al. Coal (high ash particle) Char Oxidation Coal + O2→ Ash N/A N/A(Estimated at 750. Shin et al Combination of several continuously stirred reactors Fuel shrinkage sub-model and a two-flux radiation sub-model Yang et al. 2000) V * −V  3  i *  = 3  V  i   avge R0  ∑ × X ∫  ∫    j =1 0 0 t  exp  − ∫ k oj exp (− E / RT ( r . & r = k o exp(− N/A Ea 3 )∏ ci . (Yang et al.

large8bar is the result of simulated particle density profile for large particles. there are 5 movie files: comb_chamber2_mpeg1. plugflow. and small8bar. . sideview80_mpeg1 is the movie taken from the side of the full-scale physical section model. sideview80_mpeg1.179 Appendix D: Movie files on CD-ROM On the attached CD-ROM. small8bar is the movie file that shows simulated particle density profile for small particles. large8bars. plugflow shows the simulated plugflow using the stochastic model. comb_chamber2_mpeg1 shows the view from the end of a commercial WTE combustion chamber.

Sign up to vote on this title
UsefulNot useful