PB-223 626
DESIGN AND  CONTROL OF INCINERATORS
VOLUME  I
MASSACHUSETTS  INSTITUTE OF TECHNOLOGY
PREPARED  FOR
ENVIRONMENTAL PROTECTION AGENCY
PUBLIC HEALTH SERVICE
SEPTEMBER  1973
                         DISTRIBUTED BY:
                         National Technical Information Service
                         U. S. DEPARTMENT  OF COMMERCE

-------
 BIBLIOGRAPHIC DATA
 SHEET
1. Hcrport No.
    EPA-670/2-73-089A
4. Tide nni] Suhi itlc-
         DESIGN  AND  CONTROL  OF INCINERATORS
                        VOLUME  I
                                               5. Report Date
                                                1973-Issuing date
7. Authors)  A.  F.  Sarofim,  G.  C. Williams,  J.  E.  Howard,
  	and J.  E.  L. Rogers	
                                               8- Performing Organization Kept.
                                                 No.
9. Performing Organization Name and Address
Fuels  Research  Laboratory
Chemical Engineering  Department
Massachusetts  Institute of  Technology
  ambridge,  Massachusetts
                                                K Project/Task/Work Unit No.
                                               11. Contract/Grant No.

                                                EC-00330-03
 12. Sponsoring Organization Name and Address
 U.S.  Environmental Protection  Agency
 National  Environmental Research Center
 Office of  Research &  Development
 Cincinnati,  Ohio   4526.8
                                               13. Type of Report & Period
                                                  Covered
                                                 Final
                                               14.
   Supplementary Notes
 16. Abstracts
  A versatile batch incinerator wee designed and built  in order to determine  the effect of
  operational variables on the ignition and burning  rates in a fuel bed  under conditions
  similar to those encountered in municipal incinerators.  Detailed concentration and
  temperature profiles wl.thin and above the fuel bed were obtained for different distribu-
  tions of overfire and underfire air rates.  These  results were used to evaluate both the
  rate controlling processes and control strategies.  It was concluded that for the syn-
  thetic refuse beds studied, internal diffusion within the burning elements was limiting
  •nd that drying, pyrolysis, and gasification were  contemporaneous through most of the
  runs giving rise to'burning rates and overfire gas compositions that were constant over
  most of a run.  From the oxygen concentrations in  the bed, it Is concluded that low under-
  fire air rates are desirable aarly and late in a burn, and that for a  fully ignited bed
  the maximum air rate is determined more by consideration of ash carryover and channelling
  than by the bed's ability to consume the oxygen. •  The fraction of the  heat  released above.
  the bed, which determines the overfire air requirements, is governed by the energy re-
  quirements of the bed, and decreases with bed moisture and with heat loss from the bed.
  Towards the end of a burn, the carbon dioxide concentration above the  bed falls off
  dramatically and provides a parameter that could possibly be used for  control purposed.
 17. Key Words and Document Analysis.  17o. Descriptors
  Waste disposal,  Wastes, Refuse disposal,  Tests,  *Incinerators,  *Igni-
  tion, *Burning  rate,"  Concentration  (composition),  Temperature,  Control,
  Diffusion, Drying, Pyrolysis,  Gasification,  Oxygen, Gases,  Heat,
  Moisture,  Carbon dioxide,  *Fuels
 17b. Identifiers/Open-Ended Terms
   Solid  waste  disposal,  *0perational variables,  *Fuel bed,  Air  rates,
   Rate  controlling,  Synthetic  refuse,  *Batch incinerator
17c. COSATI Field/Group  13-B
                  Reproduced by
                  NATIONAL TECHNICAL
                  INFORMATION  SERVICE
                    US D«p«r1m»nt of Commerco
                      Sprirgfiold, VA. 22151
18. Availability Statement

     Release  to  public
                                    19. Security Class (This
                                      Report)
                                         UNCLASSlFIFn
                                    •20v Security Class (This
                                                           Pa
                                                                 ASSIFIED
21. No. of Pages
                                                                              22. Price
FORM NTIS-35 (REV. 3-72)
                                                                              USCOMM-DC M952-P72

-------
                                EPA-670/2-73-089A
                                September  1973
  DESIGN AND CONTROL OF INCINERATORS

               VOLUME I
                  By

   A. F. Sarofim, G. C. Williams,
  J. B. Howard, and J. E. L.  Rogers
      Fuels Research Laboratory
   Chemical Engineering Department
Massachusetts Institute of Technology
      Cambridge, Massachusetts
        Grant No. EC-00330-03
      Program Element No.  1D2063
           Proj ect Officer

           Donald Oberacker
   Solid Waste Research  Laboratory
National Environmental Research  Center
       Cincinnati, Ohio   45268
             Prepared  for
  OFFICE OF RESEARCH AND DEVELOPMENT
 U.S. ENVIRONMENTAL PROTECTION  AGENCY
       WASHINGTON, D.C.  20460
                    i -

-------
                     REVIEW NOTICE







     The Solid Waste Researbh Laboratory  of  the




National Environmental Research Center  -  Cincinnati,




U.S. Environmental Protection Agency, has  reviewed




this report and approved  its publication.  Approval




does not signify that the  contents  necessarily  re-




flect the views and policies of this  laboratory  of




of the U.S. Environmental  Protection  Agency,  nor




does mention of trade names or commercial  products




constitute endorsement or  recommendation  for  use.




     The text of this report is reproduced by the




National Environmental Research Center  -  Cincinnati




in the form received from  the Grantee;  new prelimi-




nary pages have been supplied.
                            ii

-------
                         FOREWORD
     Man and his environment  must  be  protected from the
adverse effects of pesticides,  radiation,  noise and other
forms of pollution,  and  the unwise management of solid
waste.  Efforts to protect  the  environment require a
focus that recognizes  the interplay between the com-
ponents of our physical  environment—air,  water, and
land.  The National  Environmental  Research Centers
provide this multidisciplinary  focus  through programs
engaged in

       9  studies on  the effects of environmental
          contaminants on man and  the biosphere, and

       ®  a search for ways to  prevent contamina-
          tion  and  to recycle  valuable resources.

     In an attempt to  solve the problems  involved in
solid waste disposal,  this  study investigated a batch
incinerator that was  designed and  built to determine
the effect of operational variables on the ignition
and burning rates in  a fuel bed.   The test conditions
were similar to those  encountered  in  municipal in-
cinerators .
                             A.  W.  Breidenbach, Ph.D.
                             Director
                             National Environmental
                             Research Center, Cincinnati
                           iii

-------
                    TABLE OF  CONTENTS

                                                         Page
ABSTRACT	   i

TABLE OF  CONTENTS	ii

LIST OF FIGURES	iii-viii

LIST OF TABLES	:	ix-X

CONCLUSIONS	xi

RECOMMENDATIONS   	  xi

INTRODUCTION  ....  	   1

SURVEY OF EXPERIMENTAL AND  THEORETICAL TREATMENT OF
FUEL BED  COMBUSTION	21

PROCESSES OCCURRING WITHIN  A  FUEL BED	130

EXPERIMENTAL  EQUIPMENT AND  PROCEDURE  	 152

RESULTS AND DISCUSSION  	 184

ACKNOWLEDGMENTS	262

REFERENCES	263

NOMENCLATURE  	 278

APPENDIX  A	   1

APPENDIX  B	  .  16

APPENDIX  C	21

APPENDIX  D	45

APPENDIX  E .  .  .	64

APPENDIX  F	84

APPENDIX  G	92

APPENDIX  H	112

-------
                        ABSTRACT

A versatile batch incinerator was designed and built in
order to determine the effect of operational variables on
the ignition and burning rates in a fuel bed under con-
ditions similar to those encountered in municipal incin-
erators.  Detailed concentration and temperature profiles
within and above the fuel bed were obtained for different
distributions of over- and underfire air rates.  These
results were used to evaluate both the rate controlling
processes and control strategies.  It was concluded that
for the synthetic refuse beds studied, internal diffusion
within the burning elements was limiting and that drying,
pyrolysis, and gasification were contemporaneous through
most of the runs giving rise to burning rates and overfire
gas compositions that were constant over most of a run.
From the oxygen concentrations in the bed it is concluded
that low underfire air rates are desirable early and late
in a burn, and that for a fully ignited bed the maximum
air rate is determined more by consideration of ash carry-
over and channelling than by the bed's ability to consume
the oxygen.  The fraction of the heat released above the
bed, which determines the overfire air requirements, is
governed by the energy requirements of the bed, and de-
creases with bed moisture and with heat loss from the bed.
Towards the end of a bum, the carbon dioxide concentra-
tion above the bed falls off dramatically and provides a
parameter that could possibly be used for control purposes
                          iv

-------
                     LIST OF FIGURES

Figure                                                  Page

  1    Different Modes of Operating a Fuel Bed	16

  2    Common Incinerator Types  	  19

  3    Deissler-Eian Correlation for the Thermal
       Conductivity of Packed Beds	29

  4    Schematic of Apparatus Used by Kreisinger et al..  41

  5    Equipment Used by Kolodtsev	42

  6    Gas Compositions and Temperature Profiles in
       an Overfeed Fuel Bed	43

  7    Overall Reac-cion Rate Coefficient for the
       Combustion of a Pure Carbon Particle	45

  8    Schematic of Nicholls1 Apparatus for
       Underfeed Burning Tests 	  47

  9    Underfeed Burning? High Temperature Coke:  rate
       of ignition and rate of burning with rate of
       primary air and size of coke as variables ....  49

 10    Rate of Ignition of Different Fuels, 1-1 1/2 in.
       size, 80°F air temperature	52

 11    Temperature below Ignition Plane for
       Different Fuels 	  53

 12    Gas Compositions and Temperature Profile
       within an Underfeed Bed	55

 13    Underfeed Burning, Action through Fuel Bed
       Expressed as Weight of Fuel Products Carried per
       Pound of Dry Air Supplied, 3/4 - 1 in. Illinois
       Coal	57

 14    Volatile Matter and Ash Content of Fuel Bed
       Layers at Different Air Flow Rates	58

 15    Representation of Mayers' Simplified Fuel Bed . .  62

 16    Gas Analysis in a Fuel Bed	66

 17    Temperature in Underfeed Fuel Bed	67
                                 •
 18    Mayers' Correlation of yl/G as a Function of the
       Underfire Air Flow Rate	70

                          vii

-------
                    LIST OF FIGURES

                      (Continued)
Figure

 19    Comparison of Mass Transfer Correlation and
       Data of Kreisinger, Ovitz and Augustine for
       Oxygen Consumption in Fuel Beds
 20    Mayers' Theoretical Relationship between Combus-
       tion and Ignition Rates and Air Flow Rates  ...  74

 21    Theoretical Gas Composition Profiles Calculated
       Using the Three-Zone Theory of M. W. Thring ...  78

 22    Comparison between Theoretical and Experimental
       Relative Carbon Saturation Factors for Air-Coal
       Beds   ......................  79

 23    CO2 Reduction Rates in a Bed of 1 - 1 1/2 in.
       Coke Particles with Various Preheat Temperatures.  81

 24    Gas Composition and Temperature in a Fuel Bed  . .  83

 25    Gas Composition within a Fuel Bed, Calculated  by
       Silver ......................  87

 26    Apparatus Used by the U.S. Bureau of Mines for
       Studying Combustion Characteristics of Refuse  . .  91

 27    Bed Thermocouple History, Incineration of Syn-
       thetic Feed with 50% Moisture ..........  92

 28    Travel of Evaporation and Burning Fronts and
       Total Weight LOBS as Functions of Test Time in
       a Typical Bed of Synthetic Refuse Having a
       Moisture Content of 50% and a Bed Depth of 18  in.  93

 29    Schematic of Test Incinerator Used by Essenhigh
       et al ..... . .......... .......  96

 30    Comparison of Predicted and Experimental Pro-
       files through the Solid Bed (normalized axes)  . . 102

 31    Sketch of Dimensionless Heat Generation  and  Eeat
       Loss Curves .........  . ......... 115

 32    Variation of Gas Composition,  Temperature and
       Reaction Rate in the Vicinity of Ignition Front . 119

 33    Comparison of Vortmeyer's Ignition Theory with
       Experimental Data of P. Nicholls  ........ 123
                          viii

-------
                    LIST OF FIGURES

Figure                (Continued)                        Page

 34    Comparison of Ignition Theories with Experi-      127
       mental Data of P. Nicholls  	

 35    Surface Temperature and Depth of Penetration of
       the Vaporization Plane in a Semi-Infinite Slab
       for Convective Heating of Surface by Gas at T^. . 133

 36    Surface Temperature and Depth of Penetration of
       the Vaporization Plane in a Semi-Infinite Slab
       for Constant Heat Flux Condition	134

 37    Drying Times for Spherical Particles of Dif-
       ferent Moisture Content 	 138

 38    Simplified Schematic of Processes Occurring in
       the Fuel Bed on a Traveling Grate	142

 39    Equilibrium Constant of Common Fuel Bed Reactions 146

 40    Equilibrium Gas Compositions at Various Assumed
       Fuel Bed Temperatures	149

 41    Variation of Burning Rates with Temperature
       for Different Fuels	150

 42    Diagraromatic Representation of Direction of
       Ignition Wave Propagation Relative to the Direc-
       tion of Grate Travel and Underfire Air Flow .  . .155

 43    Sketch of Experimental Incinerator  	 158

 44    Experimental Incinerator, Gas Analysis Train
       and Peripheral Equipment  	 159

 45    Interior of Fuel Bed Section with Thermo-
       couple Probes Installed 	 160

 46    Schematic of Overfire and Underfire Air Supply- . 164

 47    Gas Sampling and Analysis Train	169

 48    (a) Stack Gas Moisture Content Measurement;
       (b) Details of Sampling Probe 	 170

 49    Instrument to Data Acquisition Connections  .   . .173

 50    Fuel Bed Temperatures (Run 10)	193
                           ix

-------
                    LIST OF FIGURES

_.                    (Continued)                       «,,.-
Figure                                                  Page

 51    Fuel Bed Temperature Profiles at Different
       Times throughout Run.  Data from Run 18	197

 52    Fuel Bed Temperatures (Run 5)	   .  .  . 199

 53    Fuel Bed Temperatures (Run 7)	 200

 54    Fuel Bed Temperatures (Run 9)	 201

 55    Fuel Bed Temperatures Below the Ignition Front
       for Different Fuel Types and Under fire Air Rates. 207

 56    Ignition Wave Propagation (Run 10)........ 210

 57    Ignition Wave Propagation (Runs 17 and 18). .  .  .211

 58    Burning Rates (Run 10)	 .  .  . 214

 59    Burning Rates (Run 17)	  . 215

 60    Burning Rates (Run 18)	216

 61    Ignition and Burning Rates as a Function of
       Underfire Air Rate	 218

 62    Estimation of Fuel Bed Thickness and Accumulated
       Inert Layer (Run 17)	225

 63    Estimation of Fuel Bed Thickness and Accumulated
       Inert Layer (Run 18)	.226

 64    Burning and Ignition Rates (Run 17)	228

 65    Burning and Ignition Rates (Run 18)  ....... 229

 66    Accumulated Carbon, Hydrogen and Oxygen Weight
       Loss of Fuel (Run 17)	234

 67    Accumulated Carbon, Hydrogen and Oxygen Weight
       Loss of Fuel (Run 18)	235

 68    Hydrogen/Oxygen and Carbon/Hydrogen Ratio of
       Fuel Burnt as a Function of Fraction of Combust-
       ible Consumed (Runs 17 and 18)	  .236

 69    Gas Compositions from Fuel Bed Probe 1 (Run 17)  .238

 70    Gas Compositions from Fuel Bed Probe 2 (Run 17)  .239

-------
                   LIST OF FIGURES

Figure               (Continued)                        page

 71   Gas Compositions from Fuel Bed Probe 1 (Run 18)  .  240

 72   Gas Compositions from Fuel Bed Probe 2 (Run 18)  .  241

 73   Gas Compositions as a Function of Distance
      Above Ignition Front (Run 17)	242

 74   Gas Compositions as a Function of Distance
      Above Ignition Front (Run 18)	243

 75   Gas Compositions as a Function of Distance
      Above Ignition Front (Run 18)	244

 76   Oxygen Concentration as a Function of Distance
      Below Ignition Front	246

 77   Stack Gas Compositions (Run 17)	251

 78   Stack Gas Compositions (Run 18)	252
                     List of Figures for Appendices
A.I   Cross Section and Plan of Fuel Bed Section  ...    2

A.2   (a) Schematic of Locating Probe;
      (b) Support Plate for Fuel Bed Section	    5

A.3   Bearing Housing and Movable Base for Fuel Bed
      Section 	    7

A.4   Cross Section and Plan of Overfire Section  ...    9

A.5   Thermocouple and Gas Burner Locations in the
      Overfire Section  .	11

A. 6   Sliding Refractory Shield	13

B.I   Estimate of Accumulative Heat Loss Through
      Fuel Bed Walls	19

C.I   Load Cell Circuit Diagram	21

C.2   Distribution of Forces in Weighing System ....   25

C.3   Schematic of Apparatus for Load Cell Calibration   27

D.I   Schematic of Orifice Plate Installation 	   45

F.I   Schematic of Incinerator Showing Material
      Inputs and Outputs	84
                          xi

-------
                  LIST OF FIGURES

Figure              (Continued)                        Page

H.I   Fuel Bed Temperatures (Run 17)	144

H.2   Fuel Bed Temperature (Run 17)	145

H.3   Fuel Bed Heat Loss Probe Temperatures (Run 17).  . 146

H.4   Overfire Refractory Temperatures (Run 17) .... 147

H.5   Overfire Shell,  Inlet Air and Refractory
      Shield Temperatures (Run 17)	 148

H.6   Fuel Bed Weight Loss (Run 17)	149
                          xii

-------
                   LIST OF TABLES

Table                                                  Page

  1   Parameters of Design for Refuse Incinerators. . .   5

  2   Air Requirements of Complete Combustion for
      Different Fuels on a Dry Basis	   6

  3   Classification of Refuse Components .......  10

  4   Composition and Ultimate Analysis of Refuse ...  11

  5   Analysis of a Composite Municipal Refuse  ....  12

  6   Projected Average Generated Refuse Composition,
      Heating Value and Quantity, 1970-2000 	  15

  7   Comparison of Predicted and Experimental Values
      of Maximum Bed Temperatures and Combustion Zone
      Depth	101

  8   Common Fuel Bed Reactions and Their Heats of
      Reaction  .	145

  9   Relative Rates of Common Fuel Bed Reactions . . . 147

 10   Synthetic Fuel Composition and Analysis 	 185

 11   Summary of Experimental Conditions  ....... 186

 12   Maximum Temperature Observed at Ignition Front
      with Varying Underfire Air Rates and Fuel Com-
      positions	205

 13   Material Balance Calculations (Run 17)	231

 14   Material Balance Calculations (Run 18)	232

 15   Check on Water Gas Shift Equilibrium within
      Fuel Bed (Run 17)	249

 16   Check on Water Gas Shift Equilibrium within
      Fuel Bed (Run 18)	250
                           List  of  Tables  for  Appendices
C.I   Specifications for BLH Precision C3P1 Load Cell  .  24

C.II  Load Cell Test 1:  Diaphragm Disconnected;
      Weights Added.  (a)  Least squares fit to data;
      (b) Comparison of calculated and actual weights
      using least squares calibration curve;  (c) Com-
      parison of calculated and actual weights using
      calibration curve from load cell test alone ...  30-32

                          xiii

-------
                   LIST OF TABLES

                    (Continued)
Table                                                  Page

C.III  Load Cell Test 2:  Diaphragm Disconnected;
       Weights Subtracted.  (a) Least squares fit t©
       data; (b) Comparison of calculated and actual
       weights using least squares calibration curve;
       (c) Comparison of calculated and actual weights
       using calibration curve from load cell test
       alone. . .......... ..........  33-3*

C.IV   Load Cell Test 3:  Diaphragm Connected; Weights
       Subtracted.  (a) Least squares fit to data;  (b)
       Comparison of calculated and actual weights
       using least squares Calibration curve; (c) Com-
       parison of calculated and actual weights using
       calibration curve from load cell test alone * . .  39-44

D.I    Factors for Orifice Plate Calculations .....  50

D.II   Calculation of Flow Rates through Orifice Plate
       as a Function of Orifice Pressure Drop and Air
       Temperature and Static Upstream Pressure
       (B » 0*4129) ..................  52

D.III  Calculation of Flow Rates through Orifice Plate
       as a Function of Orifice Pressure Drop and Air
       Temperature and Static Upstream Pressure
       (3 = 0.6347)  ..................  56

D.IV   Compressibility Correction Factor (y) for
       Orifice Plates .................  62

E.I    Instrument Connections .............  67

H.I    Instrument to Data Acquisition System
       Connections for Run 17 ............. 113
H.II   Output Voltages for Run 17
H.III  Listing of Temperatures, Gas Compositions
       and Fuel Bed Weight Loss for Run 17  ...... 118
                          xiv

-------
               CONCLUSIONS AND RECOMMENDATIONS
The major effort expended under the grant was in developing
a versatile test incinerator.  The conclusions enumerated be-
low were derived from the preliminary twenty runs made on syn-
thetic refuse mixtures,,  The conclusions are being tested and
added to using shredded refuse in a follow-on grant.  The
study demonstrated that the major features of refuse burning
on a travelling grate can be simulated in a batch incinerator
in which time in a run corresponds to transit time of an ele-
ment of refuse in the prototype.  The extensive instrumen-
tation available in the batch incinerator permit measurements
to be made which would be extremely costly or impossible to
duplicate in the field.

      (a)  The burning of refuse is not ignition limited over
     the air flow rates of interest.  Burning rate increases
     with increasing air rate being limited only by the ten-
     dency to excessive channelling at high air flow-rates.
     When channelling occurred the fuel bed could be consi-
     dered to be burning in two parallel modes, with the core
     behaving as a gasifier and the edges  as a by-pass for
     underfire air.

      (b)  The major combustible gases emerging from the top
     of the fuel bed are carbon monoxide and hydrogen although
     significant amounts of methane are evolved within the bed
     near the ignition front.  The gases evolved are relatively
     easy to burn and complete combustion was achieved above
     the bed in all cases by using overfire air jets with high
     momentum.  The results suggest that the volume of the com-
     bustion chambers of incinerators may be reduced consider-
     ably if good overfire mixing is provided.

      (c)  The gases evolved by the bed satisfied the water-gas
     shift equilibrium.  This permits the development of sim-
     ple models for the burning rate.

      (d)  For refuse, the burning rate increases with factors
     that increase the CO/C02 ratio and H^/H^O ratios.  Both
     ratios are increased by decreasing moisture content of
     the refuse and decreasing heat losses from the bed.

      (e)  Drying, pyrolysis, gasification and burnout reactions
     occurred contemporaneously within the bed.  With the ex-
     ception of very short periods at the beginning and end
     of a run the ratio of H^O/CO- in the combustion products
     remained constant.
                          XV

-------
(f)  During the initial and final periods oxygen from
the underfire air broken through the bed indicating that
the oxygen consumption was limited by the mass transfer
rate to the burning bed.  Underfire air rates should
therefore be kept low at the ignition and discharge ends
ofva grate and should be raised as high as practicable
once full ignition is achieved.

(g)  The carbon dipxide concentration above the bed drops
sharply as complete burn-out is approached.  This obser-
vation might provide a means of monitoring bed burn-out.

(h)  Material balances using stack gas composition provide
good measures of burning rate and are a better indicator of
short-range transients than the load-cell used to weigh
the bed in the present system.
                    xvi

-------
                      INTRODUCTION
Definition of the Solid Waste Problem and the Role of Incinera-
tion     ~~~~

The Nature of the Solid Waste Problem.  Solid wastes are gener-
ated in five major areas:industrial, agricultural, commercial,
household and institutional.   The health and safety of the com-
munity in highly populated urban areas, where the disposal of
the commercial, institutional and household waste is of major
concern, have necessitated the development of municipal dis-
posal systems.  The general aversion to spending money on what
is discarded as worthless and the apparent "out of sight, out
of mind" philosophy of urban residents have made it evident to
local governments that the first criteria for any waste dis-
posal program would be low cost and ease of operation.  Rising
sanitary and environmental ^-candards and increased public inter-
est in the quality of the environment have added to these
criteria of acceptability.  Public concern over air and water
pollution problems and the depletion of natural resources,
among other considerations, has handed technology a serious
challenge to improve the methods of solid waste disposal.

Three hundred billion pounds of solid wastes are collected by
municipalities throughout the United States annually and the
problem of how to dispose of them in an acceptable way has
been receiving a surge of attention over the past few years.
Considerable impetus and financial backing for this long-
needed attack on the mounting problem of solid waste were pro-
vided by the Solid Waste Disposal Act, PL 89-272, of October
1965.  Consequently? the solid waste management area has
attracted the attention of a number of individuals and com-
panies and several ways to deal with the old problem have been
suggested.  These have included high-temperature incineration,
compaction followed by either landfill or offshore dumping,
shredding followed by either burning or composting, as well as
numerous ways of sorting out and classifying the reusable
portions of solid wastes.

Traditionally, there have been three main methods of disposing
of solid waste—-open dumping, landfill and incineration, with
dumping and landfill being the more widely practiced.  The
disposal of refuse by open dumping is an old concept.  In the
Middle Ages the streets of European cities were used as a
dumping ground, and even in the Golden Age of ancient Greece
"the narrow crooked streets of Athens were heaped with refuse"
(1).  The concept of modern sanitary landfill dates back to
Biblical times, when wastes were disposed of by burial  (2_) .
Furnaces designed to bum refuse were not built until the end
of the nineteenth century, with Nottingham, England, being
                            1 -

-------
credited as the birthplace of the first municipal incinerator
in 1874
Of the refuse collected in the United States in  1967,  86  per-
cent was disposed of at landfill sites, 8 percent was. burned
in municipal Incinerators arid 6 percent was disposed  of in
sanitary landfill operations (4).  Much of the refuse disposed
of in landfill sites other than by sanitary landfill  is ulti-
mately burned vender poorly controlled open-burning conditions.
The quantity of solid waste disposed of by each  method has
been traditionally determined primarily on the basis  of cost.
Concern, however, over the quality of the environment has
generated a v.xllingness to spend higher sums of  money on  solid
waste disposal — in particular on methods that  involve re-
cycling reusable portions of solid waste.  Act PL 89-272  was
instrumental i:? promoting recycling, as it specifically put-
lined that attention bs given to the conservation of  natural
resources by reducing waste and unsalvageable material.

It is expected that tlie portion of solid waste that is re-
cycled will increase over the next decade or so, as it becomes
apparent that the "acokiomic rents" associated with environ-
mental deca^" c_id the depletion of natural resources are deemed
to be greater t'u^ri the costs of other disposal methods.   Re-
cycling of solid waste is not a new idea, but the low resale
value of the salvageable itsias  (rags, paper, glass, rubber
and tin cans/ rs^cie these operations economically unsuccessful
except as aidelinea for o'char methods  of disposal.   Concep-
tually, the recycling of all the reusable portions of solid
waste is the most attractive method of disposal  but attainment
of this halcyca position ^ill take the time required  for  re-
search effort s>.rui public education and acceptance.  It is
unlikely that recycling will be considered competitive with
conventional methods of disposal for at least a  decade, and
until it is practiced on a wide basis these methods will  need
to be used and disproved to meet the challenge of legislated
environmental standards.  Even when recycling has become  a
major method c/.' disposal„  thare will be significant quantities
of waste that vill titill be: unseparable, unusable, or which
will require w^atly pretreatment for recycling,  and these will
have to be dit^^d of by uh® other conventional methods.  For
example, of tkreo tons of refuse collected from  three loca-
tions in Saia i'rancisco,only between 29.2 and 32.2 percent was
recyclable (5_) .

Of the conveiriioaai Methods presently employed,  sanitary  land-
fill has becera-i r,oro popular: over the past years in response
to public ir*djuaction over poor landfill and dumping  practices
which produce urxyightiy, odorous areas often infested ""with
rats and veraair.;*   The method of disposal is not  without prob-
lems — specifically, accidental explosions, pollution of well
                          - 2  -

-------
water, and the increasing cost of the earth cover required.
The anaerobic decomposition of the organic portions of the
waste produce, among other gases, methane, whose concentra-
tion may build up in a confined enclosure to produce an ex-
plosive mixture.  Anderson and Colliman (£) report that "dan-
gerous explosions have occurred or persons have been seriously
burned as a result of accidental ignition of methane gas which
had accumulated near old landfills or refuse dumps."  These
problems detract from the most attractive aspect of sanitary
landfill, which is the eventual use of previously unusable
land after it has been filled in.  In many major metropolitan
areas, available landfill and dumping sites are becoming
scarce and this dilemma has prompted the shipping of metro-
politan waste to outlying communities.  This practice, in turn,
has come under sharp criticism from communities that under-
standably resent having to act as a dumping ground for others'
problems.  The pressure for more landfill sites will continue
to mount.  Many metropolitan areas are experiencing rapid
growth and the per capita daily waste in the U. S., a function
of the rising standard of living, is steadily increasing.  Both
of these facts suggest that disposal practices requiring large
land acreages and costly earth cover may become uneconomical
and impractical.  The only practical alternative in these areas
may be incineration.

The Role of Incineration in Solid Waste Management.  Incinera-
tion at present xs inefficiently practiced, as indicated by
Vaughan:s (7) analysis of a national solid wastes survey which
revealed that of the 300 incinerators in the United States,
75 percent were inadequate^ as far as air pollution control
and effective residue burnout were concerned.  Such statistics
are not surprising, since 70 to 75 percent of the incinerators
in the United States were built before 1960, with some dating
back as far as the 1940"s.  Poor incinerator operation has
caused the public to regard an incinerator as a bad neighbor
and, although emissions from the burning of all solid refuse
contribute less than 10 percent of the 11 million tons of
particulates emitted to the atmosphere annually, they are the
frequent cause of local pollution complaints by citizens (£) .
In addition to the air pollution problem, the poor control of
municipal units,, which results in incomplete combustion,
yields neither full volume reduction of the refuse nor an
inert residue.  Potentially,incineration should offer the most
significant volume reduction of all solid waste disposal
methods, as according to a U. S. Bureau of Mines report (9)
poor incineration will reduce the volume of refuse by 75 per-
cent,  while efficient burning can reduce it by 95 percent.
The residue from an incinerator eventually must be landfilled
but, if it has been rendered completely inert by good inciner-
ator operation, none of the landfill problems mentioned above
should occur, and the life of present landfill sites could be
                          - 3  -

-------
dramatically extended.  Ultimately incinerator residue may
have a utility other than landfill, especially its ferrous
content (10),  but investigations into this possibility have
only just started; however,  problems have been .encountered in
the case of tin cans with the alloying of the protective metal
coating with the ferrous body at the high temperatures encoun-
tered in a fuel bed.

Inefficient incinerator design and operation mitigates against
incinerators being a suitable solution for the urban solid
waste problem.  However, good incinerator design and the de-
velopment and implementation of reliable control instruments
and procedures present formidable problems as the processes
occurring in an incinerator are very complex, as exemplified
by the varied nature of the material being burnt.  It is not
surprising, therefore,-, to find that designers have had to rely
on experience and empirical relationships rather than quantita-
tive descriptions of the chemical kinetics and heat,mass and
momentum transfer processes.

Typically, designs are based on gross overall heat and material
balances, rule-of-thumb guidelines on the achievable burning "'
rates per unit area of grate surface with different refuse
types, and on allowable corabustion intensities in the overfire
volume (Table 1}.  Of these three requirements the most dif-
ficult to estimate has been the achievable burning rate.  The
heat and mass balances are straightforward to calculate, par-
ticularly as t.he air requirements for most solid fuels are
remarkably uniform when expressed on a basis of energy liber-
ated (see Table 2}„   The achievable burning rates per unit of
grate are estimated from guidelines such as those provided by
Table 1 or from the rule of thumb that the heat release rate
within the fuel bed should be about 300,000 Btu hr"1 ".ft~* of
grate area.  For a typical as-fired heating value for refuse
(5000 Btu Ib"1}f the burning rate for the latter criterion
would be 60 Xb kr"^ ft~^ of grate area which agrees with the
Incinerator Institute of ?jaerica's guideline of 60-65 Ib hr"1
ft"2 for a Class V incinerator and the values given in Table
1.  The establishment of these guidelines has ignored the
fundamental problem in many engineering operations, namely the
calculation of the rate of the process, and has perhaps estab-
lished a precedent of perpetuating inefficient incinerator
designs.

The maximum allowable heat release rate within a fuel bed
should be determined frora consideration of the maximum bed
temperature that would prevail and therefore must take heat
losses frail the bed into account; temperatures that are too
high cause difficulties with clinkering and problems with
grates cloggeci with molten glass and aluminum.  There are
indications that successful operation has been achieved at
                         - 4 -

-------
                            TABLE 1:
        Parameters of Design for Refuse Incinerators  (11)

of Values
QJ
1


Average
Values

Type of
Refuse*

M
R
C
M
R
C
Grate Loadings in
Lbs . of Refuse per
Hour of Operation
per square foot
of Grate Area

58 to 109
50 to 72
54 to 98
77
58
77
Volume in Cubic Feet per Ton
of Refuse per
Furnace
(Primary)
Chamber

8.5 to 25.0
13.4 to 14.5
9.9 to 13.8
12.7
13.6
11.5
24 Hours
Combustion
(Secondary)
Chamber

12.1 to 28.0
26.6 to 31.8
17.2 to 28.3
18.5
29.9
21.3
*M - Mixed refuse made up of garbage, rubbish, and noncombustibles.
 R - Refuse comprised of burnable rubbish only.
 C - Refuse containing combustibles only, such as garbage and
     burnable rubbish.
                          - 5  -

-------
                         TABLE  2 S
            Air Requirements of Complete Combustion

          For Different Fuels on a Dry Basis  (12, 13)
              Fuel
  Atmospheric Air
  Required
  (Ib.710,000 B.t.u.)
                                         Range
              Avg.
Anthracite:
   New Mexico	
   Colorado	
   Pennsylvania	
SemiAnthracite	
Bituminous coal:
   Low-volatile	
   High-volatile A	,
   High-volatile B	
   High-volatile C	,
Subbituminous coal	
Lignite:
   North Dakota	
   Texas	
Coke:
   High-temperature	
   Low-temperature	
   Beehive	
   By-product	
Gasworks coke	
Petroleum coke	
Pitch coke	
Wood:
   Softwoods	
   Hardwoods	
   Bagasse	
Petroleum oils:
   Gasoline  (60°A.P.I.).
   Kerosene  (45° A.P.I.)
   Gas oil  (30°A.P.I.)..
   Fuel oil  (15°A.P.I.).
Gaseous fuels:
   Natural gas	
   Refinery  and oil gas.
   Blast-furnace gas....
   Coke-oven gas	
Miscellaneous:
   Cellulose	
   Glucose	
   Glycol dipalmitate.. . .
   Methyl alcohol	
7.81-7.93
7.68-7.82

7.62-7.76
7.51-7.73
7.56-7.73
7.54-7.67
7.56-7.57
8.02-8.10
7.02-7.22
7.09-7.28
6.25-6.99
7.32-7.41
6.52-7.38
5.73-6.27
6.66-7.02
7.83
7.85
6.87
7.74

7.69
7.63
7.66
7.60
7.56

7.47
7.52

7.96
7.63
8.05
8.01
8.06
7.73
8.13

7.11
7.15
6.59

7.46
7.42
7.45
7.58

7.37
7.44
5.82
6.80

6.80
•6.90
7.40
6.70

-------
heat release rates up to three times the suggested maximum
value of 300,000 Btu hr"1 ft"2, hinting at the tempting pros-
pect of reduced investment cost per ton of refuse processed.

A similar situation to that encountered when selecting grate
sizes is found in the overfire region, where few design cri-
teria are available.  The only specifications given are for
furnace volumes based on guidelines such as those given in
Table 1, or on the rule of. thumb that the volumetric heat re-
lease should be around 20^,000 Btu hr~^ ft"2, and on allowable
gas velocities at different points in the incinerator.  The
value of the maximum combustion intensity is given without any
regard for the amount of ccaibustible that has to be burnt in
the overfire region,,  No guidelines are given as to the desir-
able ratio of primary air introduced through the fuel bed to
secondary air injected into the overfire region.  There is
general agreement in the literature (14, 15) that the overfire
air must be supplied with sufficient momentum to provide ade-
quate mixing with the combustible gases.  There are, however,
no reliable methods presently available for determining how
this should be accomplished.

Designs based on these methods may have been satisfactory in
the past, but, with the effects of the new legislation con-
cerning acceptable levels of gaseous and particulate emissions
as well as ash and residue quality beginning to be felt, in-
creasing numbers of practitioners are becoming interested in
developing more fundamental ways of designing their incinera-
tors.  For example, Hollander (16) has pointed out the need
for indicators for determining the probable burning character-
istics of different fuels? and the selection of ths size,
number and location of overfire jet systems and the ratios of
primary and secondary air.  In addition, the long-term trends
in refuse quality as predicted by Niessen and Alsobrook (17)
and Niessen and Chansky (jjjj have suggested that the volatTle
content of refuse, which is a measure of quantity of overfire
air that is required to complete the combustion of the volatile
products distilling from the fuel bed, will increase over the
years (Table 6).  The projected increase of this component in
refuse is also expected to require that more significance be
placed on the successful operation of the overfire air jets.
This will require a more sophisticated approach to the design
of these jets, along the lines taken by Niessen et al.  (19).

The problems of operating an incinerator effectively are enor-
mous when one realizes the tremendous variation in feed mater-
ial that is handled from day to day.  The variation in the
feedstock quality of an incinerator is orders of magnitude
greater than that in a pulverized coal fired utility boiler
(or its chain grate stoker predecessor), yet incinerator con-
trols are barely existent compared to the sophisticated
                          - 7  -

-------
controls of a modern utility boiler.  A  few of the-many oper-
ating problems connected with just the fuel bed combustion
covered below.

Discussions with incinerator operators have shown the need  for
operational guidelines for times when the incoming refuse has
a higher than normal moisture content; however, a recent paper
 (20) in this area could do no better than summarize the opin-
ion of a number of operators on what operational adjustments
they made when confronted with this type of situation.  Very
little is understood about the burning behavior of shredded
refuse on a grate system? an attempt to burn shredded refuse
in New York City's East Side incinerator met with complete
failure when the fire was quenched (21) , due primarily to .the
difficulty in supplying underfire air through the shredded
refuse.  There are no easy and quick tests to determine if
complete combustion has been achieved prior to dumping the  ash
in the quench tank and it is not an uncommon sight to see
actively burning refuse being discharged to the quench tank.

The operational problems are further heightened by the gen-
erally low level of technical training of incinerator opera-
tors compared to their counterparts in utility boilers.  The
rationale behind  decisions to run incinerators on what
appears to be a less than professional basis seems ill .founded
as modem incinerators are fairly capital intensive.  Koenig
 (22) reported that the cost for a basic incinerator, without
extensive pollution control and heat recovery features, is
approximately $12 per ton per year capacity, although the
costs vary widely.  On the basis of capital costs, therefore,
a basic incinerator ranks well down among the spectrum of
chemical and process plants.  For example, a suifurijc .acid
plant costs roughly $26 per ton per year capacity while a
butadiene plant costs as much as $1500 per ton per year capac-
ity.  However, & modern unit with the presently required ad-
vanced pollution control equipment, waste heat recovery and
aesthetic exterior would result in an appreciably higher unit
investment of around $30-40 per ton per day, depending on size.


                    Project Objectives

The increasing investment costs for suitable incinerators,
coupled with the challenges posed by more stringent pollution
codes,  the difficulty of finding skilled labor, and the finan-
cial pressures on municipalities, bring out the basic need  to
improve the designs and to develop inexpensive and reliable
methods of controlling them.,   These improvements will only be
forthcoming as a result of extensive research on a laboratory
scale and careful experimentation on full-scale unit%

Detailed testing in large-scale units is both difficult and
                          - 8 -

-------
costly and it is therefore desirable to perform the basic re-
search on small-scale units which retain some of the important
features of the prototype.  This approach was used in the stud-
ies of solid fuel-fired stokers where the definitive experi-
ments on the effect of fuel size and quality and underfire air
rate and preheat were carried out on small-scale units by
Kreisinger, Ovitz, and Augustine (23) and Nicholls (2A) .  Studies
of a similar nature on small or pilot-scale incinerators in-
clude the pioneering work at Los Angeles (25) and the studies
by the U. S. Bureau of Mines  (26, 2_7, 28), the Public Health
Service  (29) , and Pennsylvania State University (30_-4_7) .  The
incinerator studies have provided some qualitative information
on the physical and chemical processes occurring in refuse
beds during drying, pyrolysis and combustion.  Incinerator
research has yet to produce an adequate quantitative descrip-
tion of these processes, and consequently the major thrust of
this project was directed towards developing a better under-
standing of combustion processes in refuse beds.  A prime
objective was the design of a test incinerator with sufficient
instrumentation to obtain the data required to unravel the
processes occurring within the bed.  The experimental study
was limited to a. synthetic refuse which could be easily char-
acterized but which contained many of the traits of refuse.


Background Information on Incineration and Fuel Bed Processes

The Nature of Solid Waste.  Table 3 gives the generally ac-
cepted breakdown of the constituents of refuse.  The domestic
and commercial refuse which comprises the major portion of the
refuse burnt in municipal incinerators is composed primarily
of paper and paper products with quantities of garbage, grass
clippings, plastic, glass, metal and various other materials.
A number of analyses of refuse components have been made in
th6 past few years (17, 18, 4j3, 49) and although, not surpris-
ingly, each study difTers slightly* in its findings. The data of
Kaiser (49) provide a satisfactory basis from which to work.
Table 4 presents Kaiser's data for a composite municipal ref-
use.  The composite was selected from the proportions of the
ingredients that were reported in the literature (51, 52) and
obtained from private sources.  The moisture content was ad-
justed to a total of 20 percent by the addition of water to
bring the moisture content into the range experienced by in-
cinerators.

A proximate and ultimata analysis of the composite is shown in
Table 5, from which it is possible to calculate the theoreti-
cal combustion air requirements, assuming all the sulfur con-
tent of the fuel is converted to SO2 and that none of the
metal is oxidized,  using the relationship:

-------
              TABLE 3 S
Classification of Refuse Components (5£)
Kind
Refuse
Garbage
Rubbish
Ashes
Street
Refuse
Dead
Animals
Abandoned
Vehicles
industrial
Wastes
Demolition
Wastes
Construction
Wastes
Special
VVastes
i
i
Sewage
Treatment
Residue
Composition
Wastes from preparation,
cooking, and serving of food;
market wastes; wastes from
handling, storage, and sale of
produce
Combustible: paper, cartons,
boxes, barrels, wood, excel-
sior, tree branches, yard
trimmings, wood furniture,
bedding, dunnage
Noncombustible: metals, tin
cans, metal furniture, dirt,
glass, crockery, minerals
Residue from fires used for
cooking and heating and
from on-site incineration
Sweepings, dirt, leaves, catch
basin dirt, contents of litter
receptacles
Cats, dogs, horses, cows
Unwanted cars and trucks
left on public property
Food processing wastes, boil-
er house cinders, lumber
scraps, metal scraps, shav-
ings
Lumber, pipes, brick, mason-
ry, and other construction
materials from razed build-
ings and other structures
Scrap lumber, pipe, other
construction materials
Hazardous solids and liquids:
explosives, pathological
wastes, radioactive materials
Solids from coarse screening
and from grit chambers;
septic tank sludge
Sources
Households, restau-
rants, institutions,
stores, markets
Streets, sidewalks, al-
leys, vacant lots
Factories, power plants
Demolition sites to be
used for new build-
ings, renewal projects,
expressways
New construction, re-
modeling
Households, hotels,
hospitals, institutions,
stores, industry
Sewage treatment
plants; septic tanks*
              - 10  -

-------
                                       TABLE 4 t
                     Composition and Ultimate  Analysis of Refuse  (49)
Refuse
Component
Corrug. paper boxes
Newspaper
Magazine paper
Brown paper
Mail
Paper food cartons
Tissue paper
Plastic coated paper
Wax cartons
Vegetable food wastes
Citrus rinds and saeds
Meat scraps, cooked
Fried fats
Wood
Ripe tree leaves
Flower garden plants
Lawn grass, green
Evergreens
Plastics
Rags
Leather goods
Rubber composition
Paints and oils
Vacuum cleaner catch
Dirt
Metals
Glass, ceramics ash
Adjusted moisture

TOTAL
(Dry
Per
cent
23.36
S.40
S.80
S.57
2.75
2. 06
1.98
0.76
0.76
; 2.29
is 1.53
2.29
2.23
2.23
2.29
1.53
1.53
1.53
0.76
0.76
0.39
0.38
0.76
0.76
1.53
6.85
7.73
9.05
Basis in

Carbon
«3.73
49.14
32.91
-3-3 . SO
37.87
44 , 74
-
45.30
55.13
49. OS
47. 9S
59.59
73.14
53.30
52.15
46. S5
48.18
48.51
-
-
42.01
53.22
-
35.69


-

Percent by

Hydrogen
5.70
6.10
4.95
6.08
5.41
6.10
-
6.17
9.25
6.62
5.68
9.47
11.54
6.66
6.11
6.61
5.96
6.54
-
-
5.32
7.09
-
4.73
-
-
-
"
Weight)

Oxygen
44.93
41.92
38.55
47.84
42.74
41.92
-
45.50
30.13
37.55
41.67
24.65
14.82
35.17
30.34
40.18
36.43
40.44
-
-
22.83
7.76
-
20.08
-
-
-



Nitrogen
0.09
0.05
0.07
0.00
0.17
0.15
-
0.1G
0.12
1.68
1.11
1.02
0.43
1.49
6.99
1.21
4.46
1.71

-
5.98
0.50

6.26
-





Sulfur
0.21
0.16
0.09
0.11
0.09
0.16
-
0.08
0.10
0.20
0.12
0.19
0.09
0.20
0.16
0.26
0.42
0.19
-
-
1.00
1.34
-
1.15
-
-
-



Ash
5.34
1.52
23.43
1.07
13.72
6.93
-
2.77
1.22
4.89
3.46
5.08
0.00
3.18
4.25
5.09
6.55
2.61
-
-
22.86
30.09
-
32.09
-
-
—

100.00
                                 -  II -

-------
                      TABLE  5 8

      Analysis of a Composite Municipal Refuse  (49)



      Proximate Analysis (in percent by weight)

      Moisture                   20.00
      Volatile matter            52.70
      Fixed carbon                7.30
      Ash and metal              20.00
      Total                     100.00
      Ultimate Analysis (in percent by weight)
                                           Dry Ash and Inert
               Wet Basis     Dry Basis     Free Basis
Moisture       20.00
Carbon         29.83         37.29         49.72
Hydrogen        3.99          4.99          6.65
Oxygen         25.69         32.11         42.81
Nitrogen        0.37          0.46          0.61
Sulfur          0.12          0.15          0.20
Ash & Metal    20.00         25.00

Total         100.00        100.00         100.00
                         - 12  -

-------
          Wa = 11.5C  -4-   34.5 (H -  )  +  4.32S              (1)
           a.                       a


where W  is the pounds of air per  pound of  fuel required for
complete combustion,and CP H, and  O are the weight  fractions
of carbon, hydrogen and oxygen in  the fuel  respectively.  For
the compositef the air requirements for complete combustion
are approximately 3,71 pounds of air per pound of refuse
(50.5 cubic feet at 70°F  and  1 atmosphere).

The higher heating value  of refuse must be  determined experi-
mentally or estimated from the heating values of its individ-
ual components.  For Kaiser's composite the higher  heating
value is 8766 Btu/ib, on  a dry ash-free basis excluding the
small exothermic effect of partial oxidation of the metal con-
tent.  On the bcisis of this heating value,  7.0 pounds of
theoretical air are required par 10,000 Btu's liberated during
combustion, a figure which compares well with the data pre-
sented in Table 2„

Although the composition  of refuse is complex and will obvi-
ously vary with the season of the  year (3/7) , locality (17, 18)
and from day to day,as a  rough working approximation for
theoretical Calculations, the combustible portion of refuse
can be assumed, en a dry  basis, to have the composition of
cellulose  (CgH^QGs).  On  a weight  basis the composition of
cellulose is carbon 44.4  percent,  oxygen 49.4 percent, and
hydrogen 6.2 percent, and requires 5=12 pounds of air for com-
plete combustionf, or 6.8  pounds per 10,000 Btu's liberated.
Using the composition of  cellulose as the average composition
of refuse has one disadvantage in  that, unlike refuse, it con-
tains no net hydrogen.  Net hydrogen is the quantity of hydro-
gen in the fuel that cannot be reacted to water using the oxy-
gen content of the fuel.

Kaiser's data suggest that a suitable synthetic refuse, espe-
cially for experimental purposes,  can be made out of a mixture
of wood blocks, water, and inerts.  An ultimate analysis of
spruce, which is a typical scft wood, shows a content of 45.5
percent oxygen, 6.5 percent; hydrogen, and 48.1 percent carbon.
This analysis as con ba seen from  Table 5 is very close to the
dry ash and inert free composition of Kaiser's composite ref-
use.  The higher heating  value of  this wood (approximately
9300 Btu/lb)  and the theoretical air requirements per 10,000
Btu's liberated (7.1 Ib)   are close to that of Kaiser's com-
posite refuse.

The other problem in the  simulation of refuse for experimental
purposes is to select the required size of the wood blocks so
that their thermal thickness is close to that of refuse and to
introduce the moisture and inert content in a way similar to
                           - 13 -

-------
that experienced in practice.  The inert may be conveniently
introduced by adding tin cans of suitable dimensions, while
moisture content can be built up to the desired level by a
ing some of the wood blocks in water for a period of a few
(moisture levels of around 60 percent can be achieved this way),
The major difficulty associated with introducing the water con--
tent in this manner is that the drying behavior of "the wet wodq
blocks may be somewhat different from that of many constituent*
of refuse, for example the grass, leaves and vegetable matter.

The final consideration in the characterization of solid refuse
is to examine its expected trend in composition over the next
decade or so.  Miessen and Alsobrook (17) and Nlessen and
Chansky (18) have made extensive studies of the composition of
refuse from various cities throughout the United States and
made some projections of &a average refuse composition from
1970 through 2000 which are shown in Table 6.  These results
do not provide for the effect of recycle on composition and as
such probably only provide a good estimate of incinerator feed
up to about 1980.  Their projections show that moisture content
is expected to decrease while volatile carbon and plastic con-
tent will increase.  The decrease in moisture content coupled
with the increase in plastic content will significantly raise
the higher heating value of the refuse; this will adversely
affect the capacity of incinerators because of the greater flue
gas volume to be handled.  The projected increase in the fuel
volatile content is expected to require that more emphasis be
placed on the successful operation of the overfire air jets.
This point has been discussed in the previous section.

Types of Fuel Beds.  There are relatively few different princi^
pies embodied in the numerous different ways in which solid
fuel can be buroit in a fuel bed.  The purpose of this section
is to review the three basic types of fuel beds (fluidized bed
combustion is excluded from consideration) and to indicate the
principles they embody.  This section will also serve to in-
troduce a few of the common terms used in discussing4fuel beds.
Nicholls (24)  made the first attempt to clarify the principles
involved in the different types of fuel beds and this discus-
sion is based on his paper.
                                                    •>v
The type of fuel bed is fixed by the relative direction of the
flow of fuel and air.  The two major modes of f iring: a fuel bed
are termed overfeed and underfeed.  A fuel bed can be con-
sidered to be operating in the overfeed mode if the fuel and
combustion air flow through the bed in opposite senses.  In
this arrangement the combustion gases are used to preheat and
ignite the virgin fuel; a possible arrangement of this is shown
in Figure I(a).  The underfeed mode occurs when the fuel and
combustion air flow through the bed in the same sense and, in
this case, the virgin fuel is ignited only by radiation and
                         - 14  -

-------
                           TABLE 6;

          Projected Average Generated Refuse Composition

            Heating Value and Quantity, 1970-2000 (17)
                                   1970   1975   1980   1990
2000
Composition

   (weight %, as discarded)

   Paper
   Yard Wastes
   Food Wastes
   Glass
   Metal
   Wood
   Textiles
   Leather and Rubber
   Plastics
   Miscellaneous

   (weight %, as burned)

   Moisture
   Volatile Carbon
   Total Ash
   Ash  (excluding glass and metal)
37.4
13.9
20.0
9.0
8.4
3.1
2.2
1.2
1.4
3.4
25.1
19.6
22.7
6.5
39.2
13.3
17.8
9.9
8.6
2.7
2.3
1.2
2.1
3.0
23.3
20.1
23.4
6.2
40.1
12.9
16.1
10.2
8.9
2.4
2.3
1.2
3.0
2.7
22.0
20.6
23.9
6.1
43.4
12.3
14.0
9.5
8.6
2.0
2.7
1.2
3.9
2.4
20.5
21.8
22.8
6.0
48.0
11.9
12.1
8.1
7.1
1.6
3.1
1.3
4.7
2.1
19.9
23.4
20.1
6.0
                             - 15  -

-------
                                     (a)
                                              IGNITION
                                              FRONT
                                    (b)

                                      »!•     \//A
                                    (c)
Fig. 1.     Different Modes of Operating a  Fuel Bed: (a)  Overfeed;
                  (b)  Underfeed;  (c) Traveling Grate
                            -  16 -

-------
conduction from the combustion section located just behind the
ignition plane.  In this mode of combustion the underfire air
(air supplied through the fuel bed and sometimes called prim-
ary air) tends to hinder ignition as it convectively cools the
virgin fuel ahead of the ignition plane.  Figure 1(b) shows
what Nicholls  (24) termed the "unrestricted-ignition underfeed
principle," where the ignition rate is free to be  fixed by the
fuel type and size,and the underfire air rate,and degree of
preheat.  For this condition to exist the fuel feed rate must
be equal to or greater than the ignition rate.  If the fuel
feed rate were lower than the ignition rate, the ignition
plane would eventually reach the gratis; at this point ignition
would be restricted and the burning behavior of the bed would
change to reflect this.

It is perhaps confusing to talk about overfeed and underfeed
burning, as the major difference in the burning rate character-
istics between the two types of fuel bed lies in the methods
of igniting the fresh fuel.  The rate of ignition can obviously
provide a limiting step in the overall rate of combustion.  In
the overfeed bed, because the fuel is being preheated by the
combustion gasea, ignition does not limit the burning rate and
essentially burning rates can be increased by increasing the
underfire air flows up to the point where the fuel is blown
out of the bed.  In the miderfeed bed (unrestricted ignition)
the maximum rate of burning is fixed for each underfire air
rate by the rate of ignition.

There are other differences in the two types of bed in addi-
tion to the burning rate characteristics.  In the overfeed bed
the drying and pyrolysis of the fuel take place in a region
where the underfire air has little or no oxygen content and
primarily consists of carbon monoxide and carbon dioxide.
Under these circumstances it can be expected that fairly high
quantities of pyrolysis products will be evolved from the top
of the fuel bed.  In the underfeed system the drying and pyro-
lysis of the fuel take place in a region where the underfire
air is still rich in oxygen which will partially react with
the pyrolysis products and any char formed; the reaction
products and any remaining pyrolysis products can react with
the char forming the rest of the fuel bed as they flow upward
through it.

The final type of fuel bed to be discussed here is that which
occurs on a traveling-grate stoker and is shown in Figure 1(c).
Nicholls (24) has suggested that in this type of fuel bed
length H is burning on the unrestricted ignition underfeed
principle and length O on the overfeed principle since the
ignition plane has reached the grate at this point and there
are no further ignition constraints on burning.  The burning
rate in this section of the grate may well follow  the type of
behavior experienced in an overfeed bed, but there are obvi-
ous differences in, for example, location of pyrolysis regimes,
                         - 17 -

-------
ash build-up and volatile generation between the fuel bed
shown in Figure 1C a) and the length O of Figure KG).  Nxcholis
(24) also suggested that there is a length, P, located between
lengths H and O, in a change-over state where the burning  aa-
justs itself because of the cessation of ignition.

The last term which will be introduced here is "overfire air",
or what is sometimes called secondary air.  This is air intro-
duced above the fuel bed in order to complete the Combustion
of gasification and pyrolysis products issuing frtfm the fuel
bed.  These gsses can be quite rich in carbon monoxide and
hydrogen as well as in pyrolysis products, and adequate mixing
of the overfire air with these gases at high enough tempera-
tures is necessary to achieve complete burn-out in the resi-
dence time available in the furnace.  In addition to the task
of providing the air required to complete combustion, cooling
air may also be brought in through the overfire air jets;  this
is the accepted practice in refractory lined furnaces where it
is necessary to keep wall temperatures down to xfeaslShable
levels.

Description of Incinerator Types.  Municipal incinerators
operate on both batch and continuous bases and the average
capacity of the 254 incinerators operating in the United States
in 1966 was 300 tons per day (S3) .  The larger units' operate
on a continuous basis and many of them are of the traveling-
grate or reciprocating-grate variety.  Refractory lined fur-
naces are the raost commonly used for refuse incineration in
the United State-3 partly because of their low construction
cost and the f&st. that licensed personnel are not required for
operation, which results in a savings in labor costs.  These
units are limited to & maximum size of around 300' tons per day,
as at this siae operating and maintenance costs offset capital
costs (54_) .  Large quantities of excess air are required in
these units to 'liaep the walls from overheating; all this air
must pass through the air pollution devices, which increases
the cost of this equipment.  A typical refractory lined in-
cinerator is shown in Figure 2(a).

The large excetu; air problem can be obviated with the use  of
water-walled furnaces, where water-filled tubes are used to
form the furnace walls.  In these furnaces the excess air  can
be kept at tha Liinimuin required for complete combustion in the
residence time available, as the extra air is not required for
cooling purposes.   This type of construction has been used
widely in Europe,  where the cost of fuel has prompted utiliza-
tion of the w&ate heat to produce steam.  Although the use of
water walls reduces the cost of the pollution control equipment,
the labor.charges for operating these units are higher, as
licensed, power plant operators are needed.  Problems of corro-
sion of the steel tubes by acid fumes and con den sate- and
                         - 18 -

-------
                         TYPICAL REFRACTORY LINED INCINERATOR

                                    (a)
                            TYPICAL  WATERWALL FURNACE
                            WITH WASTE HEAT BOILER
                                    (b)
                             TYPICAL ROTARY KILN INCINERATOR
                               (KITH WASTE HEAT BOILER
                                    (c)
Fig.  2.     Common Incinerator Types: (a)  Typical Refractory
      Lined Incinerator;  (b)  Typical  Waterwall Incinerator with
      Waste Heat Boiler;  (c)  Typical  Rotary Kiln Incinerator with
      Waste Heat Boiler
                              - 19  -

-------
erosion by fly-ash are other factors which have to be con-
sidered.  A typical water-wall furnace with waste heat boiler
is shown in Figure 2(b).

The only other type of incinerator commonly used is the rotary
kiln, which is of European extraction.  Only one company is
licensed to sell and manufacture these units in the United
States; presently they are only offered in a limited size
range, around 250 tons per day, and are coupled with a refrac-
tory furnace.   The refractory furnace acts as the drying and
initial combustion stage,  while the rotary kiln is used for
the final burnout stage.  The slow rotary action provides bed
agitation which is supposed to aid rapid and complete combus-
tion.  A typical rotary kiln is shown in Figure 2(c).

The three incinerator types discussed above are the only ones
currently being used in the United States.  There are a number
of companies that, in the  last few years, have developed their
own units, the most notable of which is the high temperature
incinerator (55) ,   These and other designs are still in the
development and demonstration stage and are a few years away
from being in  widespread use.
                        - 20 -

-------
         SURVEY OF EXPERIMENTAL AND THEORETICAL TREATMENT
                      OF FUEL BED COMBUSTION

                   Outline of Material Covered

This survey will start with the development of the general equa-
tions describing the processes of heat and mass transfer within
a fuel bed, along with a number of the simplifications typically
used by various authors to permit these equations to be cast
into tractable forms.  This material will serve as a background
to the discussion of a number of the theoretical treatments of
fuel bed combustion that have appeared in the literature.  In
this section, some points will be raised on the estimation of
the suitable physical properties of fuel beds necessary for use
in any mathematical model.  The large degree of uncertainty in
the physical properties provides a rationale for the simplifi-
cations to these equations.  Following this, a short survey of
the more important experimental work in coal combustion will be
given, which in turn will be followed by a survey of the most
interesting theoretical treatments of combustion and ignition
within coal beds.  Both the theoretical and experimental stud-
ies on coal beds, which cam be considered to be idealized ref-
use beds, provide some valuable insights into the behavior of
fuel beds despite the fact that they lack the heterogeneity
and complexity of a refuse bed.

The experimental and theoretical work on simulated refuse burn-
ing that has been reported in the literature will then be
covered and the section will close with a discussion of igni-
tion rates and combustion stability in fuel beds.

The literature in the field of coke and coal combustion, espe-
cially on the experimental side, is voluminous but somewhat
dated.  The survey presented here makes no pretense to be a
complete literature survey and covers only the areas that the
author finds pertinent„  Likewise, there is a great deal pub-
lished in the literature concerning experimental investigations
on large-scale incinerators; no attempt is made here to cover
this area, as it is not directly applicable to this work.
Hence, the survey will cover only those studies that closely
parallel and complement the work presented here.

To previse the reader of some of the results of this survey,
there are no theories that can adequately predict the rate of
ignition within a fuel bed and how it is affected by underfire
air rate and fuel particle size and type.  Similarly, the sur-
vey shows that there are no reliable a priori methods for cal-
culating gas compositions and temperature profiles within fuel
beds.  These comments apply to the simple case of a coal bed
as well as to a refuse bed.
                         - 21 -

-------
The calculation of burning rates for deep  coal  beds under most
conditions  (except where ignition limited) can  be  done  with rea-
sonable accuracy since the compositions of the  gases  issuing
from the top of the bed are not markedly altered by changes in
particle sise and underfire air rate and the gas compositions
closely approach their equilibrium values  at the prevailing bed
temperature.  The burning rate for these cases  is  thus  found to
vary almost linearly with the air flow rate.  There are,  at
present, no theories that can be used to predict the  burning-
rate of a refuse bed,, given the physical and chemical char-
acteristics of she fuel arid the underfire  air rate.   There are
therefore no methods currently available for testing  the  valid-
ity of the empirical guidelines discussed  on pages 3  -  8.   It
is possible„ howeverf to put an approximate upper  bound on
achievable burning rates under non ignition controlled  condi-
tions.
      Mathematical Treatment of Fuel Bed Combustion
Development _of ^Biigral Equations.  A general mathematical  treat-
ment of a fuel bad would involve material balances on the  solid
and gas phases for each reacting component, a continuity equa-
tion, an energy balance, aad a set of appropriate boundary con-
ditions.  The solution to this set of equations would permit
the prediction c-f temperatures, pressures, velocities,  and com-
position change:?' within the bed as a function of time and  posi-
tion.  This cjeivoral problem has not yet been attempted  and can-
not be justified with the present state of knowledge of the
physical properties of fuel beds  and the various chemical pro-
cesses taking place.  Consequently, simplifications have to be
made in order -';.o obtain tractable models.  The simplifications
used in many cU»Vcslop;nnant:3 ares

       (i)  coilstant total mass flow
      (ii)  constant fluid density; this eliminates the neces-
            sity for the continuity equation
     (iii)  no velocity variation in the gas flow
      (iv)  no pressure changes in the system; this eliminates
            the necessity for the momentum equation
       (y)  kxneuic and potential energy changes are neglected
      (vi)  r•£,".<.£{;ion is confined to the solid phase
     (vii)  all model parameters are considered constant

With these simplifications the general conservation equations
for heat and mass are:

Gas Phase

       (i)   Heat Balance
                         - 22 -

-------
          -7) , Yagi, Kunii and Wakao (58) and  Luikov et al.  (59) .

V.(GCpgTg) is  the difference in heat  flow into and out of a
unit Volume of bed per unit time by means of mass flow through
the established fluid temperature gradient.  The parallel term
for the solid  heat balance represents a similar heat-transport
mechanism due  to  bulk movement  of the solid particles.
                         - 23 -

-------
n (T -T )  represents the rate of heat transport per unit volume
aXd sti$e by convection from the solid to the gas.  Values of
hv, the volumetric heat transfer coefficient, can be obtained
from available correlations for heat transfer in packed beds.
The extensive data in this field have been reviewed recently by
Barker (60) and Handley and Heggs (61) .
 R

 4^ r.eff H± represents the net heat liberated by the R chemical
 i-1 x reactions of gas phase  species with the solid particles;
r^eff is the effective mass rate of reaction per unit volume of
bed of the ith reaction and AHj_ is its heat of reaction.  AHf
is considered positive for exothermic reactions and negative for
endothermic reactions.  rieff must be calculated from considera-
tion of the true heterogeneous chemical rate and the rate of
mass transport to the solid-gas interface.

* (Ts) is the net heat absorbed by all the endothermic reactions
occurring within the bed exclusive of the gas-solid reactions-
considered aboiye»  This term contains the heat effects of dry-
ing and pyrolysis of the fuel and will be a complex function of
the solid temperature .
      8T
6pgCp grjrfr represents the heat absorbed in a local rise in the
gas temperature weighted in proportion to the volume fraction
(6) of the bed that contains gas.  The parallel term in the
solid phase heat balance has equivalent significance.
R
         eff
    viiri    is the mass rate of appearance of specie j per unit
          voltsaie of the bed and per unit time through the R
simultaneous g&s-solid reactions taking place within the bed.
The Vj^ are the stoichiometric coefficients of specie j in re-
action i and are written positive for products and negative for
reactants.

¥j (Ts) is the mass rate of appearance in the gas phase of specie
j per unit volume of the bed and per unit time through drying
and pyrolysis of the fuel and will be a complex function of the
solid temperature,  y>  ^(Tj is the loss of weight of the
solid caused by the 4^,    lots of all species 1 through J during
the processes of drying and pyrolysis.

In form, these equations are similar to those used in studies of
fixed bed reactors, but they are somewhat more complex due to
the reaction behavior of the solid.  The solid phase in a refuse
bed undergoes drying, pyrolysis and char combustion and is con-
tinuously changing in shape and physical structure as well as
chemical composition.  The solution to a set of equations such
as these with their appropriate boundary conditions is impos-
sible because it is extremely difficult to adequately describe
                         - 24 -

-------
the various solid phase reactions.  In addition, the various
parameters of the equations  (reaction rates, heat transfer
coefficients, thermal conductivities, etc.) are poorly known.
The following section discusses briefly how some of the more
important physical properties of a fuel bed may be estimated.

Notes on the Selection of Physical Properties.  Caution must
be exercised in selecting appropriate values of K| and hv or
hs.  The direct experimental measurement of average and local
heat transfer coefficients for single particles is only practi-
cal in a bed of large spheres .  Examples of methods for calcu-
lating such heat transfer coefficients are provided by the work
of Furnas  (62) and Gillespie, Crandell and Carberry (63) .  Furnas
used a suction pyrometer to measure the temperature of the in-
terstitial gas and the same pyrometer to measure the solids
temperature, but; without drawing any gas through it.  It is
likely that this technique of measuring the solids temperature
did not give a true solid temperature, but rather a weighted
average of the gas and solids temperatures.  Gillespie et al.
measured local and average heat transfer rates from an inter-
nally heated sphere (1 -in. -diameter) in a bed one foot square.

The experimental difficulties associated with the direct ex-
perimental determination of local and average values of hv have
led researchers to determine overall or integral heat transfer
coefficients by indirect means.  In one of the more widely used
techniques of this type, hv is calculated from a suitable
mathematical model of the modes of heat transfer within the
packed bed and the time variation of the heating or cooling gas
exit temperature obtained from unsteady -state experiments.  For
these experiments the mathematical model must take into account
all important modes of heat transfer in the bed as well as the
effects of internal resistance to heat transfer within the par-
ticles; in addition, the experimental equipment must be care-
fully designed to minimise end effects and radial velocity
gradients.  Some of the scatter observed in the various cor-
relations of overall heat transfer coefficients may be partly
due to (a) inadequate representation of the modes of heat
transfer and neglect of the internal resistance to heat trans-
fer, (b)  the use of different methods and assumptions when cal-
culating the coefficients and (c) the effect of channeling and
gross variations in packing arrangements within "random" packed
beds.

Under conditions in which the Biot number is large, the inter-
nal resistance to heat transfer may be accounted for by cal-
culating an effective heat transfer coefficient (hg) using the
following expression developed by Bradshaw et al. (64) :
               ,      ,      D         6K  (1-6)

                         - 25 -

-------
where hs  is  the  "true" heat-transfer coefficient based  on  the
heat-transfer  area per unit volume of bed and  is given,  for
hard spherical particles, as:
                           h
        h D
         v P
        6(1-6)
                            (7)
A  similar expression for the effective heat-transfer  coeffi-
cient was developed by Kitaev  (65) from the results of  a  study
of the heating times of thermally thick elements using  the
hydraulic equivalent of an analog computer.  Kitaev's expres-
sion did not include the term  associated with the effective gas
thermal conductivity.

The data of Furnas (62) would  be the best for predicting  heat
transfer rates within fuel beds as he worked with the widest
range of materials of differing surface characteristics and
with particle sizes comparable to those found in fuel beds.
However, Kitaev et ai. (£7) have concluded from an analysis of
Furnas' data that the correlation presented by him, namely
                       Aw"
'.7T0.3
10
                                   1.686-3.566"
                 v
        .0.9
                                                             (8)
where w = g&s flow rate in standard cubic feet per hour per
foot of cross-sectional area of the bed

     T  =  average gas temperature (°F)

     D  =  particle diameter (ft)
      P
     6  =  voidage fraction

     A  =  constant for each material

is in error because the internal resistance to heat transfer
was not taken into account.  According to Kitaev et al., the
anomalous behavior noticed by Furnas in his experiments on the
cooling of coke can be substantially explained by the neglect
of this factor.  Kitaev at al. suggest that a modified form of
Furnas' equation be used, based on their recalculations of his
data, using the relationship given in equation (8).  The equa-
tion is:
              h
                  =  0.0044
                          D
                           QT75

where M  is a function of the voidage and an average value of
the factor A has been used.  Kitaev et al. suggest that the
                         - 26 -

-------
overall heat -transfer coefficient will be a strong function of
M , but quote only one value of M   (M  =0.5 for a voidage of
20 percent) .  This mitigates against the usefulness of equa-
tion  (9) .   In addition, the dependence of hv on the particle
size  and the mass flow rate of gas  is different from most re-
cent  correlations (see equations  (10) and (11) ; and the method
of taking into account the temperature at which transfer takes
place is only approximate.  Furthermore, the rather low ratio
of the diameter of the packed bed to the particle diameter, in
many  cases  5 to 8, probably meant that serious channeling took
place near  the walls, giving a high estimate of the overall
heat  transfer coefficient because of the greater possibility
of heat loss to the containing walls.  The literature (68, 69)
suggests that the ratio of bed diameter to particle diameter
does  not affect the correlation of  heat transfer at ratios
greater than 18:1.

With  the shortcomings indicated above , it is suggested that the
standard correlations of heat transfer be used rather than the
modified form of the Furnas equation.  Typically, these may
take  the form of the empirical Chilton-Colburn j -factor cor-
relation, a suitable form being given by Yoshida, Ramaswami
and Hougen  (66) :


             JH = 0.91 W~°-51           N^ < 50          (10)
             JH = 0.61 VN'             NRe > 50          (11)

In these equations N   is defined as
             NRe  '  AT¥

where A  is the interfacial area per unit volume of bed and 4*
is a shlpe factor  (Y = 1 for spheres) .  It should be realized
that estimates based on these correlations may be no better
than - 50 percent and may be worse; in addition the presence
of chemical reaction within a bed may affect the heat transfer
rate but in the absence of any experimental data under these
conditions the standard correlations must suffice.

The effective thermal conductivity of a packed bed can be de-
termined for stagnant conditions by using any of the methods
suggested on page  23   .  The method of Schotte (57) appears
to fit most of the available experimental data and is rela-
tively easy to use.  However, the available data, primarily
that of Yagi and Kunii ( 70 ) , are only for particles    up to
about 11 mm diameter and bed temperatures    up to 850°C.
Schotte allowed for the effect of radiation with the following
                          - 27  -

-------
relationship:
       K
        s
          stagnant
           K
           _j
           k
                        g
     h  D
     r P
                                            1-5
              DE
(13)
where (K_Ag) n- is taken from the semi-empirical correlation
of Deissler aK& Eian (71)  shown in Figure 3, and
             h.   =  0.692eT3/108 Btu hr i ft"2
                                                (14)
The correlation works best with regularly shaped and evenly
sized particles and must be considered to give only approxi-
mate results for large particles of random shapes and sizes.

For nonstagnant conditions, Yagi, Kunii and Wakao (72) have
proposed that the effective thermal conductivity can be cal-
culated from
I 'V
i  J J flowing
                            K
.E
l!L
Cg
                                           °-80 NReNPr
                                                (15)
                                stagnant
Equation (15) was determined from measurements of the steady-
state axial temperature profile established under flowing con-
ditions in a packed bed which was heated at one end.  In the
calculation used to determine the effective bed conductivity
it was assumed that Ta=Ts (i.e., the heat transfer coefficient
hv = °°)  and that the effective conductivity was not a function
of temperature,

The above discussion has served to indicate that many of the
published results depend on the models that the particular
authors chose v/han calculating the physical property of inter-
est.  Over and above these problems are questions concerning
the effective heat transfer area per unit volume of the bed
and how this changes with void fraction, particle shape and
size, and surface characteristics.  There is no definitive
answer to these problems and any estimates of the physical
properties for a bed as heterogeneous as a refuse bed can only
be considered approximate.  For a burning bed the combustion
process adds another degree of complexity and additional un-
certainty into these estimates.

The conductive term in the gas phase equation for the conser-
vation of heat is small under most practical conditions in
comparison to the convective term, and therefore the methods
                          -  28  -

-------
     6000
      IOOO
(A
  O
  O
  (A

  O
       100
            FRACTION «  !O.6f0.5
                        9EO CONDUCTIVITY
                                                              IOOO
                        GAS CONDUCTIVITY
                 l.'eissler -Eian Correlation  tor the Thermal
                 Conductivity of Packed Beds  (71)
                          -  29 -

-------
 for  estimating the effective gas conductivity, previously  cited,
 will not be  elaborated upon.  The last physical properties,
 therefore, that need to be estimated are the mass transfer co-
 efficients for the various species undergoing heterogeneous
 reactions.

 Mass transfer coefficients depend on the flow patterns close
 to the  solid-gas interfaces and on diffusion coefficients.  The
 latter  can be reasonably predicted to within ^5 percent, but
 the  effect of the fluid flow is more difficult to determine.
 Generally, masa transfer coefficients are correlated in terms
 of the  following:
Strictly, correlations of this form are only valid at  low mass
transfer rateo and when the transferring species are present
in  low concentrations.  If this is not the case, the concentra-
tion  levels will appear in the functional dependence of the
Sherwood nuniber.  High transfer rates and rapid gas temperature
changes caused by exothermic reactions can conceivably affect
the flow patterns close to the interphase boundaries,  and this
in  turn will affect the functional form of equation (16) .  The
calculation of laass transfer coefficients under these  condi-
tions is extremely complex.

The influence of the transfer rate on the magnitude of the mass
transfer coefficient can ba estimated following Spalding  (73) .
For the idealised case of a constant property turbulent bound-
ary layer, the relationship between the ratio of the mass trans
fer conductance (g)  to the mass transfer conductance at zero
transfer rate {g*)  in terms of the mass transfer driving force
B is given by Kays (74) as

                         2  =  ln(l + B)
                         g*        B                „
For pure carbon burning in air, B is readily shown to  be equal
to 0.174 and therefore the ratio g/g* = 0.92.  The small change
in g is typical for solid fuels, where the mass transfer driv-
ing force is always low.  However, the ratio g/g* for  actual
fuels will be somewhat less than that calculated above because
of the presence of volatile matter distilling from the fuel
particles.

This section has served to point out the difficulties  inherent
in selecting good physical properties for use in any mathemat-
ical model and provides a rationale for simplifying the general
conservation equations formulated in the previous section.  In
addition,  although the methods outlined above will provide use-
ful guidelines f  the above discussion clearly shows the need
for determining  these physical properties from in situ tests
using representative bed materials.
                         - 30 -

-------
Criteria for Particle Isothermalit'y.  Equations  (1)  and  (4)
are written under the assumption  that there is no variation  in
temperature within the  fuel particles; for large particles of
finite conductivity, this may not be true.   (In this case, a
rigorous analysis is still possible, following Amundsen  (75) ,
but analytical solutions are only possible when the  reaction
rate is considered to have & simple linear dependence on tem-
perature.  Even with this major simplification, the  solutions
are very complex and require tabulations of special  functions.)
The extent to which temperature gradients will be present
within a particle will  depend on  the Biot number  (Nr>£=hsDp/ks) .
Chukhanov  (77) has shown, using a simple theoretical model,
that particles can be considered  isothermal if

                         ~   > 2                           (18)
                         NBi

Saunders and Ford  (76)  have shown experimentally that tempera-
ture gradients within" particles may be neglected if the fol-
lowing condition is met :

                       (GC  D /k ) < 4                      (19)
                         Pg P

This £ondition is similar in form to equation (18) but the
term GCp  gives only an approximate measure of the heat trans-
fer       coefficient.  Saunders  and Ford worked with relatively
small particles  (the maximum diameter of particles used was
0.25 in.) and as the criterion has a poor theoretical basis  it
must be regarded as only approximate for larger particles.

Handle y and Heggs  (78)  have suggested that the particles can
be considered isothermal if

                        3k L(l-6)
                        — § - 5 —  > 60                     (20)
where L is the bed depth  and the other symbols have the mean-
ings given to them in the Nomenclature (Appendix I) .  It should
be noted that the group on the left-hand side does not contain
the heat transfer coefficient and therefore this criterion has
the same shortcoming as that of Saunders and Ford's.

These criteria are based on the convective heating or cooling
of particles and do not consider the effect of radiant heat
transfer to the particle or the effect of chemical reaction
within the particle.  As such, they must also be considered
approximate for many fuel bed situations.  It should be noted
here that exothermic reactions within the particle would help
                         - 31 -

-------
lower the internal resistance to heat transfer,  while endother-
mic reactions would have the opposite effect.

Simplification of the eonservation Equations.   For the purposes
of the following development, the bed particles are considered
to be isothermal.  Many systems can be suitably approximated as
being one-dimensional and in this case equations (1)  tnrougn
(5) simplify to:
Gas Phase     0
             32T
             32Cb     •    3C?         ,     „    «»<£
                                                           (21)
°E^-  '-r^vX"*-^       ,«
              3z      p   32


Solid Phase

                    3T
                      s
                    _^w«_

                     3z
                            3T
                          S
                    E*           °^
-------
simplified equations of the conservation of  heat.   Schumann
neglected the thermal conductivity  of  the  solid and gas  phases
and assumed that there was no bulk  flow of the  solid and non-
reacting conditions; equations  (21)  and  (23)  then  reduce to:
3T
- GC
        V W  -
                                              3T
                                       6pgcP
                                   (25)
                                             3T
                  - wv  -
with the boundary conditions
                      T  (0,0) = T,
                      T  (0,0) =  0
                       5

                                            (26)



                                            (27)


                                            (28)
                      Tg(0,t)  =
                                            (29)
The formal  solution  to  this  set  of  simplified  equations is
relatively  complex,  involving  sums  of  Bessel's series:
       =  e
                                  n=l
                (yx)
                                                            (30)
where,
           -y-x  V   n
T /T   -  e Y    /^  X'M  (yx)
 9               n=0
                   MQ(yx)
              JQ(2i/yx)
                                                            (31)
                                   (32)
                   M(yx)
                    n
                               d(yx)
                                   n
                                            (33)
                   y  =  hvz/GCp
                                            (34)
                   x  =
hv/(l-6)psCp
                                              z6p
                                t -
                                            (35)
                         - 33 -

-------
The solution to this simplified problem is given for complete-
ness, as it has been widely used in many studies on heat trans-
fer within packed beds.  Furnas (62_) was the first to realize
that Schumann's solution provided a new approach to calculating
heat transfer coefficients rather than direct determination  isee
page 25) .

Subsequent to Schumann's solution, a number of approximate solu-
tions to the rigorous problem have been proposed and the reader
is referred to Klinkehberg (80) for a further discussion.  More
complete analyses of the conservation of heat equations have
been developed by Singer and Wilhelm (8_1) , Brinkley (82) ,
Wilhelra, Johnson and Acton (83) and Amundsen et al. (7f[, 84,
85, 86) .  Ail of these authors had to make some simplifications
to the  equations before they could develop analytical solutions,
the most common assumptions being a linear variation of the re-
action  rate with temperature and negligible axial conduction
through the solid .  All of these solutions are not strictly
applicable to fuel beds and a somewhat different approach was
suggested by Stewart and Saville (87) .  This method suggests a
way of  determining the effective thermal conductivity of the
fuel bed.

The development below follows Stewart and Saville (87) , and
only those equations dealing with the conservation oT heat will
be considered i it is therefore assumed that the reaction rate
term ^± Rj can be determined independently of the equations of
conservation of mass.  The two equations considered are:

            3T                           3T

                                                           (36>
                  3T                V~^ *                3T
                                                            (37)

where the conductive term in the gas phase equation has been
eliminated, as under most practical situations it is small in
comparison with the convective term.  The temperature of the
gas Tg can be eliminated from these equations and the..,equa-
tions combined in a number of different ways, depending on the
physical model selected.  It is commonly assumed, particularly
when catalytic reactors are being modeled, that the heat trans-
fer rate is so high that  dTs  _ dTg .  Equations (36)  and  (37)
then reduce to:            3z  ~  3z
 S 3z'
"CPJ  t£ + ZXAHj =  [apgcpg+  C1~6)p
             3                            (38)

        -  34  -

-------
where, as suggested by Stewart and Saville  (87_) , the  subscript
on the temperature has been dropped and now can be  considered
to represent the temperature measured in the bed with a  thermo-
couple.  As pointed out by Stewart and Saville  (87) ,  the above
assumption can be a poor one under conditions where bed  tem-
perature gradients are large  (e.g., near the ignition front of
a fuel bed) .
By assuming that all reactions take place on the  solid  and  that
32T
             , a better approximation   first proposed by Woods
 3z2    3z*
and Harris  (88)  is obtained.  Equations  (36)  and  (37) then  can
be combined to give:
      ^2^2
                     GC   + UC
                       pg     ps
                                  3T
           *
           .
           3
                             C
J
                                  3t
(39)
Except near  the  ignition  zone  in the  underfeed bed, where there
will be  large  quantities  of  oxygen  available to react in the
gas phase with pyrolysis  products and carbon monoxide, the
assumption that  reactions are  confined to  the solid phase is
probably quite good.   (There is strong evidence  (89) suggesting
that carbon  monoxide  is the  primary product of the carbon-
oxygen reaction  at high temperatures.)  Stewart and Saville
propose  that the second assumption  may be  abrogated by using  a
truncation factor, 3, which  compensates for any difference be-
tween the second derivatives of gas and solid temperature with
respect  to distance.  The truncation  factor was found as fol-
lows.  Neglecting the heat capacity of the gas, rearrangement
of equation  (36) gives:
                          3T
                                                            (40)
Continuous differentiation of equation  (40) gives  —g-*- in  terms
of a series expansion of derivatives of Ts with  respect to
distance.
           dz
                                                           .(41)
                                               dz"
                         - 35 -

-------
where y
Substituting (41)  back into (40)  gives
                               >
                               'P,
                       T
                    -g
           3T
           3T                       -^
where T1 = ^ .   The series  given in equation  (42) can be
shown  s       to be equal  to
  Ti .
                                -Z/Y

'/
                                                           (43)
and, using this relationship,  (42)  can be  rewritten as

                                          2   „
                          dTs
                   - GV -af  +  3
where
                     3
                                              's
                                                           (44)
                                             dz-
                                                           (45)
Combining equation (44)  with equation (37)  gives the Stewart
and Saville equation,
              2
                  2T   r-
                  Hc
                                     3T
                            (i-«)pscp
                                     S
                                       ]3T
                                       Jat
                      (46)
From the development, B must be a constant for equation (46)
to hold.  For a simple exponential dependence of temperature
with distance such as T = a + becz this is true/ kvit for more
complex relationships it would be purely fortuitous if 3 were
found to be constant.  Consequently, this modification of
Stewart and Saville's must be used with cautipn.  Developments
                         - 36 -

-------
along these  lines  can  take us^no further and it is suggested
that under conditions  where 2^  R* is an unknown function of
temperature  that either      J  3 equation (38) or equation
(39) be used.
          Eft
        R. is  a loiown  function of distance, another method be-
comes j   ^aviable  for  eliminating Tq; this method also takes
into account gas phase reaction.  By differentiating equation
(37), substituting  ^£g_  front equation (36)  and eliminating
hy(Tg-T ) from    3z~  che resulting equation by using equa-
tion  (37) , the following is obtained:
KEGC
P
y
h
V
d3T


dz3
r
5 ^
* \ Tf
s
uc
• P,.
s.-
h
                           '  ]
                            i-\-, I
                            -   UC
                                              + GC
                                                 dT
                                                 dz
                                fir1
                  R. AH. -8-
                   33
                                  p
                          >
                                           AH.

                    (47)
It will  be noted that this approach involves no approximations
Mayers  (90)  used this method in his analysis, which will be
discussed in a later-, section.
Equation  (47)  can also be rendily solved using the method of
variation  of parameters to give
       m,z
 S
Ae
 + Be
                        .
                            \  e

 *         n^2
R.AH.dzdz-e
 D  D
z      z
  -nuz
 e
 z
7
where
m.
m
 2   12
              v
              g
                            -^—^ &
                  > R.AH.dzdz
                     J   3
                                           +  C
GC,
                          - 37

-------
A, B and C are constants which must be determined from  the
boundary conditions (two on the solid phase and one on  the  gas
phase) .  Even for the very simplest of reaction rate  terms  the
evaluation of the integration constants presents a teaioua
algebraic task.

The approximate equations (38) and (39) can be written, for
steady-state conditions, in the general form:


         A      dT
which can be integrated directly to give:

                            z       z

                 yz _ e-Yz J  eYZ j  S(2)dzdz  =   0       (52)
where A and B are integration constants which can be determined
from the appropriate boundary conditions.

Equation (52) may be used to determine the maximum temperature
that will be reached in a fuel bed.  With the boundary  con-
ditions,

             at  z - 0,  T. = T     and      -  *
                              ulciX
             at  z » ~,  T = 0                              (53)

both A and B can be found, the result being
                           «5c
where Qnet is tne net total heat released from  z=°°  to z=0.
Equation  (54) is, of course, directly obtainable  from a simple
heat balance.  The difficulty in applying equation  (52)  in  a
more general way lies in evaluating the double  integral for
which Q"(z) must be known explicitly.  Stewart and Saville  (87)
have used equation (52) along with some approximate methods
of evaluating the double integral to obtain some  interesting
results for a sinter bed.  .The application of equation (52)
to sinter beds is much easier than for a fuel bed,  since in
the former the separation of the exothermic and endothermic
zones can be readily achieved and the total heat  released,  or
absorbed, in each zone can be well-established; the picture
                         - 38 -

-------
 is not  so well-dafined in a fuel bed,

 For  unsteady-state conditions such as  experienced in a combus-
 tion pot,  equations (38)  and (39)  may  be written in the general
 form,
                                               3T           (55)

 If  the  temperature-time data from a combustion pot experiment
 are plotted up as temperatore-distance curves at different
 times,  it  would be possible to evaluate 3^T , 3JT and 3JT .   For
 a region where thermal  reactions are not7~~2~   3z     3t
 important,  the value of >c could be estimated from

                         ,   ,     3T       3T
                         vpp'ave 3¥  +  V 3z"               (56)
                   j^  _    	:	
Similarly,  the  net  reaction rate Q(z)  could be determined from
the same  data obtained frc\i a "reacting"  zone using this value
of Jc.  Admittedly thi-a extrapolation of Ic to the higher tem-
perature  region contains soi;:e uncertainties,  but a correction
for the mean  temperature difference in each zone could be
crudely estimated using equation (13).  Q(z)  could then be
calculated  from
                           lptVave  5¥       3z              (57)
                 Q(z)  =   	—	
 Experimental  Investigations  of Combustion in Coal  Beds

Overfeed Combustion;   The  Work  of  H.  Kreisinger et  al.  (23)
and K. I. Kolodtsev' CSJLJ „   The  purpose  of  these investigations
was to study the  rate  of combustion  of  an  overfeed  bed  and to
measure gas compositions arid  temperatures  within the burning
fuel bed.  In  both  these works,  water-cooled probes were used
to collect the gas  samples.   Kreisinger et al.  measured the
concentrations of C^T  CO2»  CH'4,  N2,  tar and soot, while
Kolodtsev only measured CO, CO?, and  02-  Kolodtsev  used plati-
num/pi atinum-rhodiura thermocouples and  an  optical pyrometer
(for temperatures greater  than  1500°C)  for the  temperature
measurements,  while Kreisinger  et  al. measured  temperature
                         - 39 -

-------
only with an optical pyrometer.  The apparatus used inthe *****
vestigations were somewhat .different in design and mode of ope*
ation.

Kreisinger et al. used the equipment shown in Figure 4, «",
arrangement that is commonly referred to as a combustion pot.
Three different fuels were used — Pittsburgh coal, anthracite
and coke, with the Pittsburgh coal being the richest in vola-
tile matter ( 34%) , while the anthracite and coke had similar
compositions and low Volatile matter contents (5% for the
anthracite and 1.5% for the coke).  Another difference between
the coke and the anthracite was the moisture content (2.8% for
the anthracite and 0.7% for the coke) .  The moisture content
of the Pittsburgh coal was around 2.3%.  All the fuel was
graded to the same size by passing it over 1 in. square-mesh
screen and through 1 1/2 ill. screen of the same type.  Kreisinger
et al."operated the fuel bed under "steady-state" conditions,
adding fresh fuel, at regular intervals, to the top. of the bed
during the test to keep th« depth of the fuel bed constant.
Two different bed depths were studied — 6 in. and 12 in*  In
this study, gas compositions were taken at a few fixed posi-
tions above the grate and temperature measurements were taken
sequentially through the same holes that were used to introduce
the gas sample probes into the bed.  The temperatures were read
immediately after the sample probe was pulled from the bed, the
optical pyrometer being sighted down the tunnels left by the
probe in the fuel bed.

In Kolodtsev's experiments a fixed amount of fuel was charged  '
to the apparatus (see Figure 5) and the fuel bed was burnt in  :
an unsteady-state mode; the ignition and burning "front" prop-
agated down through the bed in the direction of the imposed gas
flow.  Because of the unsteady-state nature of the experiment
and the fact that the gas samples were taken at a fixed loca-
tion in the bed, the distance of the probe from the ignition
front had to be calculated from knowledge of the prevailing
ignition rate and the time the sample was taken.  The tempera-
ture measurements were carried out in parallel with the gas
sampling, but at a different position in the bed.  Kolodtsev
used electrode carbon of 98.6% purity with 0.01% ash and 0.05%
moisture for his fuel.  Experiments were conducted with fuel
graded to four different sizes':  2.6 - 3.7 mm, 4.8 - 6.0 mm,
6.0• - 7.2 mm,  7.2 - 9.0 mm.  In the majority of the experiments,
the original height of the bed above the gas sampling port was
70 mm, while the total depth of the bed was 190 mm. u

Kolodtsev's results will be discussed first because-athey rep-
resent the best data available on the most idealized fuel
possible.  Figure 6 shows the compositional change with par-
ticle size and air velocity.  In Figure 6(a), (b) and (c) ,
the particle size was held constant and the air velocity
changed over more than an order of magnitude.  The composition

-------
                       f- ''•'.'•%     Gas samplers
            -       ..  ..,.•
         •>v jFuel bed.-;-4xvx :
                                                    SECTION A B
                                Manometer connection
                                        Air
                                Screen
                                                ^Onfice
                                                                    Ajr from
                                                                     blower
                                                                 • Screen
                                                     18"-
                                                                    -12--
Fig.   4.      Schematic of Apparatus Used by Kreisinger  et al

                (23)

-------
Bojdyx
  Fig. 5.     Equipment Used ;by  Kolodtsev (ML)
  (1) Combustion chamber,  (;2J  grate,  (3)  water jacket,  (4)fuel
  bed, (5) blast compartment,  (6)  connecting pipes,  (7) mixer,
  (8) electric furnace,  (9) siit of gas sampling tube,  (JU»
  holder,  (1,1) hot junctipn s sampling ^battery, :(17) ijiaiio-
  meters,  (18) orifice
                           -  42  -

-------
   THE  BURNING  RATE  OF  A  SOLID  FUEL
                         750
                        •
                                  0-3
                                  0-2
                                 0-1
                                         i    i
                                                   I	I
                                                             1,750
                                                             r°c
                                                             1,500
                   V    5cm
                                                           5cm
                                  0-3
                                  0-2
                                  0-1
              3
1    2   3
                                                           5cm
Particle diameter  cm
Air velocity  cm/sec
                                              2   3    V    5cm
                              a     b    c    d   e    f
                            0rf2  0-J2  fl-J?  0-5f  0-66  0-81
                             11     ¥9   150  150  150  150
Fiq.  6.     •'••as Compositions -ind Temperature Profiles in nn
             Overfeed Fuel Bed.  Data of Kolodtsev (91)

-------
profiles were only marginally affected by  this  change.   The
calculations of Yagi and Kunii  (92) , which are  based on the
experimental data of Parker and Hottel  (93_) ,  show that  the rate
of  the  reaction C * O2 •* CO?, under the  conditions of Kolodtsev s
experiments, would be expected to be mass  transfer controlled
 (see  Figure 7) .  Thus the oxygen profile would  not be expected
to  shift significantly as a consequence  of the  small Reynolds
number  dependence on mass transfer.   (Note that the 'Reynolds
number  will not change directly with the gas  velocity,  as the
viscosity  also changes due to the higher temperatures encoun-
tered.)  The rate of the Boudouard reaction,  C  -I- CO2 -*•  2CO,  is
known not  to be controlled solely by mass  transfer considera-
tions;  this fact explains the slight increase in the concentra-
tion  of CO at o. fixed distance from the  ignition zone,  since
the rate increased with the higher temperature.   "In .addition,
the equilibrium constant shifts at higher  temperatures  to favor
larger  CO/CO? radios„  However, for pure carbon gasified by air,
the maximum concentration of CO is 34.5%,  which is reached at
temperatures above 900°Cf and the equilibrium concentration  of
CO  is greater than 30% at temperatures above  800°C.   Thus equi-
librium did not appear to be a constraint  in  these experiments.
The temperature level in the bed rose as the  gas velocity in-
creased, sines the heat lost from the fuel bed  became a smaller
fraction of tha total heat liberated.  The temperature  peaked
at  the  pl&ne of sero oxygen concentration.  This was as ex-
pected  since the oxidation reactions were  the only exothermic
ones  taking place.  After the point of zero oxygen concentration,
the temperatures fall because the Boudouard reaction occurring
in  this region is endothermic.  Presumably if the fuel  bed was
deep  enough/ the heat loss from the bed coupled with the endo-
thermic reaction would lead to the "freezing" of the Boudouard
reaction..  In Figure 6(c)» (d) , (e) and  (f) ,  the air velocity
was held constant rfhilti the particle size  was increased by a
factor  of  2.5.  These figures show that the oxygen consumption
distance (ZQ)  Increasscl with particle size and  appeared to have
scaled  almost in proportion, to the particle size.   Mass trans-
fer considerations suggest that ZQ should  scale approximately
as  Dp--4 (see pages 75-82} .  The CO and CO2  curves shifted to
the   right as tha particle size increased,  but  not in direct
proportion to uae particle size.  Since the Boudouard reaction
is  not  truly iaass transfer controlled, there  is no simple scal-
ing law which can be applied.

Kolodtsev1s results show clearly that if a fuel  bed is  being
operated for the purposes of heat generation, it should be kept
thin  (i.e., not much greater than the oxygen  consumption dis-
tance) .   If it .is not kept thin, then overfire  air needs to  be
added to burn out the carbon monoxide issuing from the  top of
the bed.   if,  oii the other hand, the fuel  bed is being  run to
generate producer gas,  which is rich in hydrogen and carbon
monoxide,  then the bed should be kept as deep as possible.
Finally, Kolodtsev indicated that the ratio of  particle size to

-------
                        1000
                         100
                     o _-
                     *• Wl
                       £  10
                          1000   1700  1100
                                    Temperature l°K ]

                                            1800
                                        1600
                                                 2000  2100
Fig.  7.      Overall Reaction Rate  Coefficient for the  Combustion
              of  a Pure Carbon Particle (92).   Data of A.  S.
              Parker and  H.  C. Hottel  (93)

-------
diameter of his caaba^iion chamber  had to be greater than 1:18
for his results not to be substantially affected by this ratio.

The results of Kreisinger et al.  closely parallel those of
Kolodtsev, even though the fuel type  and size were quite dif-
ferent.  These authors concluded  that their fuel beds, whxch
were quite deep, acted as gas producers with the flue gases
being rich in hydrogen a:ad carbon monoxide.   They also found
that the oxygea was" rapidly consumed  within approximately 2 to
4 particle diameters frortt the ignition zone.   Kreisinger et al.
varied the air .,;low rate through  the  bed over almos.t as wide- a
range as did Zelocttiev.,  They found that within wide limits of''
the air flow r.-v-c the^u was practically no effect on the cant-
position of gases within the bed.   Thus,  the  measured burning
rate found by 'ihese authors scaled  directly with the air rate..
They also policed out that, since the variation of the air
flow rate did net r-\i.ft.ot tie composition of the gas issuing
from the top oir the bed, it i^as essential for secondary air to
be added above the bed -to effect  complete burnout of combust-
ibles.  The ^bove cc-uclu^ia:.^, which  appeared in 1916, seem to
be the first dufi^itive statements  backed by  experimental
proof of the necessity for secondary  air for  complete combus-
tion.

The Underfet-i ;.• -_•.•!:  The -,'crh of P.  Kicholls (24  and W. C.
                                 Along with  the work of
K rex singer ec ?1. ? Hi at oils* experiments  have become classic
in the field ' oz combustion.  His experiments  were the first to
give combustion ^r.c.ineers a clear picture of  the processes of
ignition and ccrubat;ii,oB at Different positions on a traveling-
grate stoker,  ::>J. theog, h lyucholls also  studied the effect of
underfire air on
-------
 ,TS sampling hole



Sieht and Campling hub
               Tar and eoot
          i£    sampling hole
Fig.  8.       Schematic of Nicholls' Apparatus for Underfeed
               Burning Tests (24)

-------
temperature &istory at their respective  positions within tne
bed, but rather for determining-the  rate of  travel of the ig-
nition front„  For each test, the  fuel was charged to the com-
bustion pot and the thermocouples  placed in  position.  The
fuel was ignited with a fixed weight of  charcoal and petroleum
coke wetted with kerosene and spread over the  top of the fuel
bed.  After tho fuel was observed  to have been ignited satis-
factorily,, th-3 water-cooled top  was  put  on and gas 'Samples
were taken ecritinuou^l^ from the stack so that the instantane-
ous burning rates could be calculated.   The  time at which the
ignition plan*:; p^s^d the different  thermocouples was noted
and from this data the ignition  rate was calculated. ^ Only 1for
a few tests ye re detailed measurements of gas  comprositions and
temperatures within tLe bed made.  In these  tests, -the gas
samples were b,-.j:«ri with standard water-cooled  gas probes and
analyzed for CO, CO:,, CH4; H^, tar and soot.   Nicholls tested
for channeling of the underfire  air  close to the side walls toy
comparing go-s coeoosItioiiss obtained  from the center of the 't>ed
with average cor^ositio-ns from the stack.  There was strong
evidence that       a large amount of the underfire air flowed
   preferentially up tho side walls  but  probably not enough sfeo
seriously detract fro;*. the validity  of the general trend of
the results«  Guarding against the occurrence  of channeling -is
almost impoiusiblo,, but this channeling could have been .kept to
a minimum, by Nichollf;, by u^ing a grate with  a  high' pressure
drop  instead uf the simple bar  grate he employed.  In any
case, the very deep beds used certainly  aggravated the problem.

Nicholls studied four different  sizes of fuel  (2-2 1/2 in.,
1-1 1/2 in.,, and .1/2 - I in.)    of     different fuel types
(coke, both liigh and low temperature; anthracite;  petroleum
coke, an ashj.es.-s f^tl? or.d four  different bituminous coals) .
This array of fuels covered most of  the  different .fuel types
with respect -uo volatile matter/  moisture content, and the fuels'1
proclivity to cake end :?cr;c solid  barriers to  the flow of
underfire air.  The proximate analyses of the  fuels .can be con-
sidered to be close to those for similar fuel  types'7in the
study of Krsisiuger at ai.

Some typcial caUi cjotaiitec. by Nicholls on the  underfeed com-
bustion of hich temperature coke are shown in  Figure 9.  These
results will hi. discussed ±i\ detail  and  the  results'-'on the
other fuels referred, to only as  necessary.   Figure 9 indicates
that at low underfire air flow rates the rate  of ignition was
greater than the burning rate and  consequently the depth of
the live fuel bee' continuously increased until the ignition
plane reached the grate0  A'j, the underfire air flow rate in-
creased, the rate of bi rninc? increased more  rapidly than the
rate of igniticr  and eventually  the  two  rates  became equal
(point c for the 1 1/2 - 2 in. size  coke).   At air flows higher
than this critical air flow rate,  the rate of  burning was

-------
                              MB DEFICIENCY, PEBCBWT
                              GO        SB       0
                              ! i I I I  I I I / / /
                 ISO    SM    few    iOO    SODSo
                PRIMARY AIR PER SQ. FT. PER HOUR, POUNDS
Fig.  9.      Underfeed Burning, High-Temperature Coke:  rate  of
              ignition &nd rate of  burning with  rate of primary
              air and  size of coke  as variables.   Data of P. Nicholls
              (24)

-------
controlled by the rate of ignition, since the  fuel  could
viously not burn faster than it was ignited.   The heavy solid
lines define the region in which there was, to use  Nicholls1
term, equilibrium burning, under which condition  the  depth of
the live fuel bed remains constant.  As  the underfire air was
increased further„ the ignition rate started to decrease and
eventually a point was reached where ignition  could not be
sustained.  The results on the different fuel  sizes show that
the ignition rate increased as the particle size  decreased,
increasing by a factor of about 1.5 for  a halving in  the par-
ticle size.  The decrease in the ignition rate at the higher
air flow rates is expected for the following reason.   The fuel
is ignited by conduction and radiation of heat from the hot
combustion soae immediately behind  the  ignition  front, while
at the same ti;:.t2 it is convectiveiy cooled by  the underfire
air; as the underfire'air flow increases, the  cooling effect
becomes dominant and eventually will completely hinder igni-
tion.  The increase in the ignition rate with  a decrease in
the fuel size would not be expected from the correlation of
the dependence given i& equation (13) of the effective thermal
conductivity of a bed on the particle size.  For  smaller'-par-
ticles, it would be expected that the effective thermal con-
ductivity would decrease and likewise the ignition  rate.  This
was not observed and the reason probably lies  in -the  'criterion
necessary for !:he ignition of a particle.  It  is  common knowl-
edge that it i£; far more difficult to light a  large log of
wood than a sraall ona principally because there is  less ther-
mal inertial to overcomes.  For a piece of wood to ignite,  it
needs to have been heated        to the  extent that the rate
of generation of pyrolysis products is great enough to provide
a combustib.la gas in the vicinity of the particle surface.
This condition it; achieved more quickly  with a small  particle
than a large one with the time scaling roughly as "the square
of characteristic radius.

The above discussion has centered on the situation  observed in
a fuel bed with no restriction to the ignition plane.  The
curves shown are, however, not restricted to this sole con-
dition and in I act for the case of restricted  ignition the
fuel bed can be operated at any point within the  envelope of
the burning and ignition curves (i.e., anywhere inside curve
ace  for the I 1/2 - 2 in. size fuel particle).   An equilibri-
um bed will resi-.lt at any point in this  envelope.   An increase
or decrease in the feed rate, provided it does not  cross ab,
will result in a gradual change in the fuel bed thickness cor-
responding to th-a -new feed rate.  If the increase in  feed rate
crosses ab the thickness of the live fuel bed will  increase
continuously;  if it crosses bd the thickness will ncft increase
and the ignition plane will sTowly be forced up by  the fresh
fuel below.
                         - 50 -

-------
Changes in fuel bed thickness at different underfire air rates
are reflected in the position of the burning rate with respect
to the air deficiency or air excess lines.  When the burning
rate is not restricted by the ignition rate, Figure 9 shows
that the burning rate is almost at a maximum.  Following curve
ce it can be seen that1" the thickness of the bed must have be-
come   less  as the air flow rate was increased, arsd fpr burn-
ing rates to the right of line OP, oxygen had penetrated through
the bed.  For the 1 1/2 - 2 .in. size particles the thickness of
the bed at an air flow rate of 350 Ib hr-1ft~2 was about 10 in.
and it is therefore surprising to find oxygen breaking through
the bed at this point„  It is possible that the lower bed tem-
peratures probably associated with this point caused a signif-
icant shift in the Controlling mechanism of the carbon-oxygen
reaction, causing the oxygen consumption distance to increase
markedly over that predicted from mass transfer theory.  It
will be shown later that it is probably not possible for fuel
bed temperatures to drop low enough for this to happen without
having the fire die out and it is more likely that this is a
manifestation of the degree of channeling that occurred near
the side walls.

Figure 9 brings out the fundamental differences between the
underfeed and overfeed modes of operation.  For the underfeed
mode, the maximum burning rate of any particular size of fuel
is fixed, while for the overfeed mode any size of fuel can be
burnt at almost any rate desired by adjusting the air rate and
the bed depth.

The ignition rates for all the fuels studied by Nicholls at
various air flow rates are presented in Figure 10.  All the
fuels were of the 1-1 1/2 in. size and the underfire air was
kept at 80°F (the degree of preheat of the underfire air was
found to markedly affect the rate of ignition).  The dotted
portions for the bituminous coals are extrapolations, since
these fuels did not bum satisfactorily at these low air rates
due to serious caking.  All of the fuels, except the low tem-
perature coke and the petroleum coke, reached a maximum rate of
ignition; it car. be presumed that for these cases further in-
creases in the underfire air rate would have eventully caused
the fire to be blown out.  The situation with the petroleum
coke and low temperature coke is not so clear, but, from the
nature of the ignition process discussed above, it is likely
that they would follow the same behavior.

Nicholls also presented a series of smoothed and everaged plots
for the temperature profiles below the plane of ignition for
various fuels of the 1-1 1/2 in. size and at a fixed under-
fire air rate.   These profiles are depicted in Figure 11.  The
shape of the profile depends;in a complex way on the effective
thermal conductivity of the bed, the heat transfer coefficient
                         - 51 -

-------
     100   2TO
















































i
f

/
/
/
/j
//
C//J
y
7
V






/






^
/^



/
\f
/
*

/


/
/s















f
1
1


/
/

/


/


Xx
/














f
/




/

V
\
t_


' A
i-
X

x^
'














/





/


gh-t.


X
nthi





^
/k





x'



cmpi



•acifa
'1,
\"
X L^H
x1^












/
/^
Pet





^
""Lo


:ratu

»<

53
^
*-^—
Illino


\
\






An1
\
\


B



/


rota






w-tc


ire o
4


/



imc




*— •

mpe


oke
—
5plir
Hw
^



• ••
a coal


hra<


• -



ites






^




oke




^ ••

ratu



• •.
tea

X









• •

ren



il
esbnorelh
X










^^^

>ke




KKJ

-i_
L >1
\
^


tin





h=k
Mtiburg
fcoal
^^^
\
\

d2

























HWl
I"
N
















IV
20
25
,
15
10
5
                                                                       §
                      PRJ^iABY AJS PER SQ. FT. PER HOUR, POUNDS
Fig,  10.     Rate of Ignition  of Different Fuels, 1-1*5  in.  size,
              80°F air temperature.  Data of  P.  Nicholls  (24)
                          -  52

-------
woo
1200
                               1.  High-temperaturr coke
                               2.  Low-temperature coke
                               3.  Splint coal
                               4.  Illinois coal
                               5.  Pittsburgh coal
    0        0.5       1.0        1.5       2.0       2.5       3.0
          DISTANCE BELOW PLANE OF IGNITION, INCHES
    Fie;.  11     Temperature Below Ignition Plane for Different
                Fuels.  Data of P. Nicholls (24)
                        -  53 -

-------
and the sensible heat fluxes of the air and fuel.  The rela-
tionship, using equation (43) with the reaction terms set to
zero and the boundary conditions T=Tign at z=0 and T=0 at z=«,
is
                      T
Tigne
                                                            (58)
where
          T.    =  ignition temperature

          z     =  distance from the ignition plane
          a
On the assumption that     Nicholls1 study (24) the ignition
temperature was the s$ma for all the fuels, all the"terms in
equation  (50) except U were approximately constant for the dif-
ferent tests.  On this basis, the higher ignition rates should
have been associated with the steeper temperature profiles.
Comparison of the data of Figure 10 shows that this ,was not
always the case.  The discrepancy probably lies mainly in the
assumption of equal ignition temperatures for the fuels.  The
ignition temperature for a fuel will depend on its chemical
and thermal properties as well as its size.  Ignition tempera-
tures were not determined by Nicholls and, as Figure 10 shows,
were all arbitrarily considered to be 1378°F.  For those fuels
which had a lower ignition temperature than 1378°F, a certain
amount of reaction would have taken place below the '"ignition"
front; these additional energy terms were not considered in
the development of equation (58).  For a true comparison of
theory and experiment, ignition temperatures would need to be
determined and T/T.    plotted (ordinat® against z (abscissa).
It should also     g be recognized that equation  (58) holds
only for isothermal particles, and that in this region of the
bed the shailowar the temperature profile shown in Figure 11
the more closely this stipulation will be met.

The final piece of Hicholls" experimental work to be covered
is his study of the temperature and gas composition profiles
within an underfeed bed of high temperature coke.  These are
shown in Figure 12.   The underfeed bed profiles were much the
                         - 54 -

-------
 30  2400
                                   1   I	1  I
                                   Temperature of fuel surface
                                         I
                   +2     4    6    8    10    12    14
                      HEIGHT ABOVE PLANE OF IGNITION. INCHES
                                        16
                                             18
                                                  20
Fig.  12.
Gas Compositions and  Temperature Profile  within
an Underfeed Fuel Bed.   High-temperature coke, 1
- lh in.  size, underfire air 160 Ib h
Data of P.  Nicholls  (24)
                        - 55  -

-------
same as those for the overfeed bed, but the former were more
spread out near the ignition front.  In this experiment,  the
bed was deep enough for the CO/C02 ratio to reach equilibrium,
as shown by the almost constant CO and C02 compositions.   A
similar test was performed for Illinois coal  (Figure 13),
which is rich in volatile matter  (35%).  In this test, the oxy-
gen was much more rapidly consumed than in the test with  high
temperature coke and disappeared in a distance of about 1 1/2
in. from the ignition plane.  This was presumably dtie to  the
greater possibility of reaction of oxygen with the larger
amount of pyrolysis products generated by the fuel.  Despite
the large volatile content of the fuel, the concentrations of
tars, soot and raethane are quite small and appear mainly  near
the ignition sone.  It is interesting to note that in this ex-
periment only one third of the total heat of combustion was
released within the bed-  The other two thirds would\have  been
released in the overfire section if the off-gases had been
combusted with overfire air.

Marskell and Miller (94) studied the change in structure  of a
coal bed during combustion for varying combustion rates.   The
data were obtained from a modified underfeed combustion pot
segmented into layers 1 in. deep which could be removed sep-
arately, allowing tha coal contained therein to be analyzed.
For the tests, the combustion process was stopped by cutting
off the air at various intervals after lighting the fuel;  the
layers of burning coal were then quickly removed and quenched
by placing the live coals into a metal container immersed  in
water.  In these experiments the fuel used was a bituminous
type of size 1/2 - 1/4 in. and the underfire air was preheated
to 200°F.

Figure 14 shows the volatile and ash content of the six layers
of the bed (the bottom layer is numbered "1")  for two dif-
ferent air flow rates (190 and 750 Ib hr^ft"2) .  For both air
rates, the volatile content fell from the initial value of 36%
to zero, with the top layer being devolatized first.  :For  the
low flow rate (190 Ib hr~ift-2), the ash compositions in  the
layers rose to a steady content until all the volatile' matter
was driven off.   At this stage,  the ash content in the bottom
layer started to rise and only after this layer was almost
burnt away did the higher layers commence to burn out ;in  quick
succession from the Bottom upwards.  For the high air ..flow
rate (750 Ib hr^ff"-) ,  the picture was different.  In this
case, the ash content in each layer rose as the volatile  con-
tent fell;  thus the bed burnt out from the top down.
                                                      t~
No burning or ignition rates were quoted by Marskell and Miller
for these tests.   However, for the degree of underfire; air pre-
heat (200°F)  and the type and size of fuel used, Nicholls1  (24)
experiments suggest that for the low underfire air rate the
ignition rate was greater than the burning rate; and for  the
                         - 56  -

-------
 0.02

  .01

   0

  .02

  .01
D
Z
DC

>
Q
55

I
OS
£
H

U
  .01








	


) i







0 	 	






Ifl


6 	

4
)r»

0













T











R
w
.s?
o>
c;
_«







T

f




t




JQ


a


„-«-


I
S?
J|
; j
1
4-
: t

!•
t\
'. 1
1
1
1
1
\!J


•~™O-4-










s





1









\ ^
— 5»
{,*- ^
/
|/'
fc'

O
"•^ .



J
— " -,
1




V




r "







i


!
*rfe



'•

]
	 ni
-


. 	 ©












(^





O1
''




"•o





^


1
-4
i
1















/
f





. ^


I


)
o


M> 	






1




•flL
r*<^
	 1


X
c:
x^






i.°

r
I


in
V


— i




i

"ar £









^







, c
( ir





H2O


r
, 	 Sr












CH4









1
1

i CO,

Fre
— —

|





K>t
£




1










X







O

e Hz


,
o


.


1


1
"1




I

a



x.



:



i
I-— T
H>-\





'o1^

,
i
•















•x


?
Cf
c . . .
c



	 o-


































i .
|


































~*


    -2
                    2       4      6       8       10
                 HEIOKT ABOVE PLANE OF IGNITION, INCHES
     Fig.  13.     Underfeed Burning, Action through Fuel Bed
     Expressed as Weight of Fuel Products  Carried Per Pound
     of Dry Air Supplied, 3/4 - 1  in. Illinois Coal.  Data
     of P. Nicholls  (24)
                         -  57 -

-------
          '/.VolMiles
           o
9  IO
                      • 45 M i ft u l s >i
                                                          •/.Volatile*
                                                                                          (a)
                   O    -S   -2
                                                                                          9  IO
                                                                        23 Minutes
ui
00
           )    I   2   3  -4  -5  6  7  '8  -9  KD

                   Frectlpn Of Combustion Time


            Volatile Matter and  Ash Contents of
            Fuel Bed Layers.  Low Air Rate
                           23456769  IO

                           Froclion of Combustion Time
                    Volatile Matter and Ash Contents of Fuel
                    Bed Layers.  High Air Rate
            Fig. 14   Volatile Matter and Ash Contents  of Fuel Bed  Layers at  Different Air Flow
                      Rates.  Data of Marskell and Miller (94)

-------
high underfire air rate equilibrium burning  (equal  ignition  and
burning rates) was achieved.  If this  is  correct, the burning
zone depth increased with time  in the  low underfire air experi-
ments while it remained of  constant thickness  for the high
underfire air rate experiments.  This  hypothesis is consistent
with the experimental observation that, at the low  underfire
air rates, active burning continued for some time after the
volatile content had fallen to  zero at the bottom of the bed,
while for the high underfire  air rate  this condition was only
achieved close to the end of  active combustion.  Using the ex-
perimental observations of  Kicholls (24)  and Marskell and
Miller  (94) it seems reasonable to conclude the following:   (a)
For low underfire air rates where the  ignition rate is greater
than the burning rate the burning depth will increase up to  the
point that the ignition front reaches  the grate.  From this
time until the end of the run the bed  will "burn out" from the
bottom  (where th® oxygen concentration is the  greatest) up-
wards, with the bed slowly  settling towards the grate; (b) For
high underfire air rates where  equilibrium burning  is achieved
the burning will take place in  a. zone  of  constant thickness
which moves slowly through  the  bed; there will be a gradual
increase in the ash content of  this zone  as it moves towards
the grate.  It should be pointed out that although  coal beds
exhibit different behavior  at these extremes of underfire air
rate, for refuse burning high underfire air rates would be un-
acceptable because of the possibility  of  excessive  particulate
entrainment.

This discussion concludes the survey on the experimental work
with coal, the purpose of which was to provide the  necessary back-
ground for the following discourse on  the various theoretical
treatments of coal bed combustions.  Although  there is as yet
no direct evidence to say there  refuse  beds will behave in a
similar way to coal bads? the work of  Kreisinger, Augustine
and Harpster  (95) with lignite,, which  contained 40% moisture
and only 25% fixed carbon,  showed results closely parallel to
those for coal and suggests that this  may well be the case.


       Theoretical Treatment  of Fuel Bed  Combustion

Background Information.  There  have been  very  few papers  (nota-
bly the work of Mayers (90, 9_S_, 97) , Thring  (98, 99) and Silver
(100, 101) concerned with developing theoretical treatments  of
fuel bed combustion.  This  situation is partly due  to the ex-
tremely complex nature of the process  and partly due to lack
of real interest.  The impetus  for the first theoretical treat-
ments came from attempts to improve the operation of the chain
grate stokers which were used for power generation  up until
the 1940's, when the larger units began to be  superseded by
pulverized coal-fired plants.  The gas producer is  the only
                         - 59 -

-------
other process using solid fuel combustion or gasification  in
which there has been a theoretical interest; however, this
process is not widely used.  Incineration, then, has been  the
main unit operation over the past forty or fifty years in
which solid fuel bed combustion is used.  For many years,
there was no incentive to improve incineration, but the real-
ization that it will provide an important platform in the
solid waste program and that effective design and control
criteria are required in view of increasingly more stringent
environmental constraints has caused a renaissance of interest
in this field.

The development of fuel bed combustion models can start with
three separate objectives:  (a)  a description of the burning
rates;  (b)  a description of the ignition rates, or (c) a  com-
bination of both.  Concern over the description of ignition
rates is confined to the underfeed bed, where, as previously
discussed, the ignition process may limit burning rates.
Strictly, therefore, it is not possible to separate the burning
and ignition rates in an underfeed system.  In general, the
ignition rate will be partly dependent on the relative pos-
ition of various exothermic and endothermic reactions taking
place within the fuel bed which must be determined by burn-
ing rate models.  The processes are often separated, however,
to make the problem more tractable.

Almost all of the fuel bed models presented in the literature
to date can be placed in two general classes:  those based in
part on empirical relationships concerning the gross fuel  bed
behavior and those based on a description of the rates of
material and energy transfer and of the fundamental reaction
rates.  Both classes have their drawbacks.  The former suffers
from the paucity of empirical information on fuel beds and the
uncertainty associated with using these relationships in situa-
tions different from those under which they were derived.  The
latter, although a more rigorous approach despite the many
simplifications which need to be invoked, depends heavily  on
an accurate knowledge of the physical and chemical properties
of the fuel bed and of the chemical kinetics of the reactions
involved.  In addition,, realistic boundary conditions need to
be selected for the governing differential equations.  The
discussion on pages 25 - 30 has highlighted the difficulty of
obtaining accurate data on the physical properties of fuel
beds.   The extreme difficulty in obtaining the requisite infor-
mation on kinetics and the chemical and physical properties of
fuel beds has forced the proponents of this type of model  to
select the values of these model parameters that make their
calculated theoretical results fit particular experimental re-
sults.  This approach obviously reduces the generality of  the
model  and severely limits its applicability in systems other
than the one to which it was originally fitted.
                       - 60 -

-------
The fuel bed combustion models that have been presented in the
literature and that are most applicable to this study will be
discussed below, and some of their advantages and  shortcomings
pointed out.  The work of Mayers  (90, 96, 97) will be discussed,
first, as illustrative of those models which utilize a fairly
rigorous description of the material and energy transport
within the fuel bed, coupled with kinetics to predict ignition
and burning rates, fuel bed temperatures and gas compositions.
Following this, the work of Thring  (98, 99) and Silver (100,
101) will be covered.  Thring"s paper is limited to an attempt
to calculate gas composition profiles in a fuel bed and the
distance required for complete oxygen consumption  in an under-
feed coal bed from a knowledge of mass transfer rates in packed
beds.  Silver proposed that the concentrations of  oxygen, carbon
dioxide and carbon monoxide at different depths through the
fuel bed could be calculated from a knowledge of the pressure
drop across the fuel bed and argued that this pressure drop
could be used to determine a friction factor which in turn
could be related to the rate of mass transfer.  In addition to
his work on predicting gas composition within a fuel bed, Sil-
ver developed a criterion for the combustion stability of a
fuel bed.  Thring and Silver both attempted in their work to
avoid the semi-empirical nature of Mayers' approach by relying
on correlations of mass and momentum transfer within packed
beds.

The M. A. Mayers Theory (90, 96, 97_) .  The first attempt to
develop a theoretical modelT^of a fuel bed, to explain and com-
plement the experimental investigations of Kreisinger,
Augustine and Ovitz  (23} and P. Nicholls (24), was by M.  A.
Mayers (9_0_) .  The simplified fuel bed proposed by Mayers,
shown in Figure 15, was thought of as a continuous solid, of
a porous nature such that air could be blown through it.   Thus,
the bed was considered as a whole rather than as a collection
of discrete pieces, an assumption that eliminated  the neces-
sity of dealing with temperature variations in the lumps of
fuel.  Variation of the properties of the bed was only con-
sidered for the vertical (z) direction; radial and angular
variations were neglected.   There was a flow of fuel in the
bed, taken as positive in the z direction, and ignition was
supposed to take place at some plane z=Zi which could have
any value from z=0 to z=£ (the depth of the bed).  The dif-
ferential equation, developed from an energy balance on a
differential element of the fuel bed, describing the tempera-
ture of the bed as a function of z was stated as:
                         - WV
                                                           (59)

                         -  61  -

-------
                      Combustion  Zone
   Ignition Front _L
    —'               -i j
           Grate
     Air in  ( G )
                            Fuel  in (u)
Fig.  15.   Representation of Mayers'  (90) Simplified Fuel Bed
                     - 62 -

-------
As written, equation  (59) rigorously describes the various modes
of energy transfer within the bed given the three assumptions
used.  However, the equation had to be substantially simplified
in order to obtain an analytical solution.  The assumptions
necessary were:  steady-state operation; heat effects of all
endothermic reactions were small; the solid was uniform and
isotropic allowing constancy of Cp  and hv; the mass flow of
fuel was constant throughout the depth of the fuel bed; a suit-
able average value of K| „ the effective bed thermal conductiv-
ity, could be found.  Equation (59) then simplified to
E
Ks

d2T
— f-S*
dz *s
2
\~^ ef
                   r?rrAH. = 0
  v  s
                                                           (60)
 2
2
 i=l
           r   AH.   given by


                I 0 when 0 < z 1 z

(z,
                              when
                                                        ± z < H
where PI and y2 represent t'f-3 rates of reaction C + 02 •*• CO2
and C + CC>2 •* 2CO respectively, at unit concentration of the
reacting gases and on the basis of a unit volume of fuel bed;
AHi and AH2 are the quantities of heat liberated or absorbed
in the two reactions , referred to unit weight of reacting gas,
and pi and P2 represent the fractional concentrations of -02
and C02 respectively.  Representing the function   V1 rf   AH.
in this way introduces two further assumptions:    jd,       1
the two reactions C 4- 02 -*• CO-? and C + CO7 ->• 2CO are "the only
ones occurring within the fuel bed; and all the reactions take
place in the solid phase.

Four additional equations plus suitable boundary conditions
were required to specify Mayers' problem.  A differential
energy balance on the gas phase , excluding the contribution of
axial conduction and assuming no gas phase reactions and
steady-state operation,, gave the following equation for the
gas temperature:
                           VTs - V  ' °
where G is the mass rate of gas flow per unit area of bed,
assumed constant throughout the depth of the fuel bed, and Cp
its mean specific heat.  The equations describing the change ^
in concentration with fuel bed depth of the various constitu-
ents in the gas phase were given by Mayers as
                         - 63 -

-------
                                                          (63)
                                  wnen z_5 z £ A
     CO  :
where P3 is the ir.
These equation;- &
trolled or first
suitable average
by equilibrium.
controlling mecha
neglect any axiiii
tions are formula
stancy of the gas
    fraction of CO in the combustion gases.
.ssuir.e that each reaction is mass transfer  don-
crcier with respect to the reactant with  a
rate constant and that they are not constrained
li'i addition to the assumption about the  rate
riism of the two reactions, these equations
 diffusion of the three species.  The equa-
ted for steady-state operation and imply con-
 flow rate.
Mayers was abl^ to solve the above equations analytically with
the boundary condLtions described below.  In setting up these
boundary conditions the ambient temperature was taken as zero,
and thus the calculated Ve lues of Ts represented temperature
elevations abov:^ c-hibient conditions.
for equation (60)  were given as
                     The boundary conditions
            dT
    + °v
                                  =  0
at z = 0
                                       (64)
            dT
                   -4- I  T
                 =  0
at z =
                                                       (65)
where 3 and y ere ^quiyalent radiation heat-transfer coeffici
ents at the bottom and top of the bed, respectively.  The
quantity UCpg  v:as included as a coefficient in equation  (64)
because  ——  o:" cJie necessity of providing sufficient heat
at the   K     bottom of the bed to raise the temperature  of
the incoming fresh fuel from the ambient temperature to that
of the bed at 2=0.  These boundary conditions introduce
further assumptions as they are not exact, since radiant heat
losses are proportional to the fourth power of the absolute

-------
temperature.  Mayers compensated  for  this  error through the
appropriate choice of  the heat transfer  coefficient.   Although
Mayers did not explicitly state that  the values of  K*jj in equa-
tions  (64) and  (65) were different, it must be  assumed that
values were chosen which were representative of conditions at
the top and bottom of  the beda  In  addition,  the value of K^
for the bottom boundary condition would  be expected to be a
function of the position of  the ignition front  when the igni-
tion front is close to the bottom of  the fuel bed.

Only one boundary condition  was needed for equation (62)  and
Mayers chose

                T   =  0  at z   =  0                      (66)


Finally, the three boundary  conditions needed for equation set
(63) were

              PI  =  P^   for  z  <   z±


              P2  =  0    for  z  -   z^                     (67)


              p.,  =  0    for  z  -   z,

       *
where p, is the mass fraction of  oxygen  in the  underfeed air.

The gas temperature vas eliminated from  equations(60)  and (62)
as described on pages  32-39, to give  a third order  inhomogene-
ous differential equation (v/ith constant coefficients) ;  equa-
tion (63) was readily  solved permitting£]rf  AH^ to be evalu-
ated.  Mayers" soluticu to the third         order inhomogene-
ous equation is algebraically very complex and  the  interested
reader is referred to  the original paper (90).

Figures 16 and 17 show now the gas compositions calculated from
the solution of equation set. (S3)  and how  the temperature of
the solid phase of the 3;>el  bed,  calculated from the  solution
of the third order d:,f;:e re.nti.il equation,  compare to  the ex-
perimental data of Michclli.  -!"_4) -  Considering  the  complex
nature of a fuel bed;, the agreement,  at  least superficially,
is quite good.  However,, a few points need to be raised as to
how the agreement was  achieved.

The experimental temperature profile  shown in Figure  17 (solid
line)  is for an underfeed bed with unrestricted ignition for
high temperature coke of slse 1-1 1/2  in.  with an underfire
air rate of 160 lb hr~^ft~^0  The gas compositions  of Figure
16 shown by the solid and dashed  lines are taken from Mayers'
                         - 65 -

-------
                10    12
H£KsMT A8OVE GRATE,
                                     14    16    16
                                 INCHES
Fir. lio.      Gas Analysis in a Fuel Bed.   Solid
curve plotted from experimental values for over-
feed combustion of coke (24).  Dotted curve plotted
frcT, experimental values for  underfeed combustion
of coke (24)•  dashed curve plotted from calculated
values (90).
               - 66  -

-------
  Q      4
£MSTANC£
                                i      12      83      20
                              f»LAN£ OF KiWITION  SNCHES
Fig.  17.  Temperature in Underfeed  Fuel  Bed.   Solid curve
           plotted from experimental values (24).   Dashed
           curve plotted from values calculated" for unre-
           stricted ignition  (90).   Dot and dash curve plot-
           ted from values calculated for restricted igni-
           tion  (90).  Air flow  rate =_160 Ib  hr  ft  ,  com-
           bustion rate = 25 Ib  hr~  ft"  ,  and  rate of ad-
           vance of ignition zone =  26.1  Ib hr  ft  .
                       -  57  -

-------
original paper  (90) , which implies that they were used  to cal-2" "
culate the temperature profiles of Figure 17.  The experimental
gas compositions shown by the solid line in Figure 16 were,  in
fact, taken from Nicholls" data (24) for an overfeed, bed of  1 -
1 1/2 in. coke with an underfire air rate of 149 Ib  hr~*ff* and
preheated to 80°F.  Nicholls  (24)  points out, in his study of
underfeed combustion, that the gas composition profiles through
the fuel bed are more spread out in this mode of burning than
in the overfeed mode (sea pages 46-59).  The gas compositions
taken from the asaae experiment as the temperature profile of
Figure 17 was obtained are shown in Figure 16 by the dotted
   es.  The GAS composition profiles define the function
   *+.£ £      _ „  _   .    ._  .   .    _       «__ 	•*_!_• 	 .ft. 1. —.
ft
	rf"AHj_, which gives the heat release rates within the  bed.
1 Using the gae; compositions from one set of data to calculate
temperature profiles for comparison with another set, of data
is therefore not justified.  The agreement between the theo-
retical and experimental temperature profiles is not particu-
larly good (for most of the bed the theoretical temperatures
are at least 20 percent too high); in addition, the heat  losses
from the top of the bed were over-estimated, probably in  an
attempt to try to lower the profiles.  Any one-dimensional
theory would be expected to predict temperatures that are
higher than experimental because it would not take into account
radial heat losses from the bed.  The author feels that tem-
perature measurements taken in the center of a combustion pot
and near the ignition zone are to a certain extent insulated
from side wall losses as a result of the larger quantities of
air flowing up near the walls.  The greater mass flux of  oxy-
gen at the wails causes higher temperatures at this location
than at the center of the bed and helps minimize heat losses to
the cold refractory walls.
                                                     4-
For the calculation of the gas composition profiles Mayers
needed to have estimates of the two quantities, Uj/6 and y.,/6.
H® found ©aieslfetsd values of ii-/<3 at different underfire
air flow rates from           the data of Kreisinger et al.
(23)  on overfeed beds.  The results are shown in Figure 18.
The scatter in the results is appreciable but this must be ex-
pected from the complexity of the processes occurring.  How-
ever, there are soae fundamental erroneous assumptions lying
behind the presentation of the data in this way.
                   The two-zone theory of Mayers suggests that
Hi  can be related to a mass transfer coefficient if. it is
~~<*  assumed that the reaction is very fast at the fuel bed
temperatures encountered.  The two assumptions necessary  for
this relationship to hold for all fuel beds are that there is
no other sink for oxygen consumption other than reaction  at
the fuel surface (i.e., combustion with CO or pyrolysis prod-
ucts)  and that the effective surface area for mass transfer
per unit volume of bed for all fuel beds of the same particle
size is the same.  The first assumption is more nearly correct
                         - 68 -

-------
for an overfeed bed than for an underfeed bed, except that
there will always be CO present, which is a primary product of
the carbon-oxygen reaction at high temperatures  (89).  A third
assumption necessary for the relationship Mayers presented is
that the reaction is not hindered by ash.  In the data of
Kreisinger et al., a plot of the log of     oxygen concentra-
tion within the bed as a function of     distance from the
ignition front  (suggested by equation  (63)) often deviates
from a straight line, indicating that there was in fact ash
build-up at the bottom of the grate, against which insufficient
precaution was taken.  This fact probably accounts for some of
the scatter in Figure 18.

Recognizing the assumptions above, another attempt was made,
by this author, to recalculate the data of Kreisinger et al.
(23) using a form more appropriate to calculating mass trans-
fer data in packed beds than that suggested by Mayers (97).
The results of this study are shown in Figure 19.  The same dif-
ficulty with the nonlinear serailog plot of oxygen concentration
versus distance .as experienced by Mayers was found when analyz-
ing the data.  The problem of selecting a best straight line
through the points plotted in the manner above accounts for
much of the scatter in Figure 19.

Figure 19 stresses some of the inherent difficulties in analyz-
ing any fuel bed data:  namely,  (a) the selection of an appro-
priate Reynolds Number complicated by the ambiguities associ-
ated with the values of Dp and y;  (b) the evaluation of the
effective mass transfer area per unit volume of the bed, As;
and (c) the determination of an average Pg to use in calculat-
ing the JD factor.  Figure 19 shows that the effective mass
transfer rate for 02 is considerably higher than that predicted
by a typical mass transfer correlation, but somewhat lower than
the line suggested by Mayers (97).  From the earlier discus-
sion, it would be expected that, because of the lesser possi-
bility of oxygen reacting with pyrolysis products, the oxygen
consumption rates in coke beds would be lower than those in
bituminous coal beds.  This was not observed under all cir-
cumstances, although the coke data on the average fell below
the bituminous coal data.  The pattern was different when com-
parison of the anthracite and bituminous coal was made; the
anthracite data appeared to lie, on the average, above that of
the bituminous coal.,  There could be a number of plausible
reasons for this discrepancy, a likely one being varying CO/C02
ratios        at the surface of the different fuels.

Returning to Mayers' work, the disturbing feature of Figure 18
is that the line Mayers drew through the data is weighted
heavily in favor of data taken in situations where the oxygen
concentration might be expected to fall much faster than pre-
dicted from mass transfer theory because of secondary gas
phase reactions (the circles shown in Figure 18).  The squares
                         - 69 -

-------
   1-0 J-
  30-
  '20S
O  15 h
        o  o c>
        S?  WJ 
-------
         1.O
         0.5
 u
 I/)

^
 u
   •0
         0.1
       0.05
                                       M.A. Mayers (97)


                                             26.1
             DcAcetis * Thodos (102)
        0.01
             Pittsburgh Bituminous Coal
             D 6" Fuel Bad
           _ S 12" Fuel Bed
              Anthracits
             A 6" Fuel Bed
             A12" Fuel Bed
             C oke
             O 6" Fuel Bed
             «12" Fuel Bed
                                Mass  Transfer Coefficient Calc.
                                Using
                                        «xp
                                with As - 33 FFT
                                               6
                                               -3
L<-.>j




   I
           10
                           50
                            NReS
5OO
        1000
Fig.   19
           Comparison  of Mass Transfer Correlation and
           Data of Kreisinger, Ovitz and Augustine
           (123) for Oxygen Consumption in a Fuel Bed.
           (Analysis of Fuel:  Pitt. Bituminous,  2.28%
           moist., 34.27% volat.,  52.63% fixed  C, 10.82%
           ash; Anthracite, 2.76%  moist., 5.28% volat.,
           81.62% fixed C, 10.34%  ash; Coke, 0.71% moist.,
           1.42% volat., 81.26%  fixed C, 16.61% ash.
           Mean particle diameter  = 1.25 in.)
                        -  71  -

-------
                                •                  11'
on the plot shown at a value of G about 150 Ib hr'-'-ft"^  are  for
coke, while the starred point is taken from Nicholls1  (24) data
from an overfeed bed, depicted as the solid line iri Figure 16,
which are, of course, the results Mayers' theory is attempting
to predict.  In the calculation of the parameter V2/G similar
fitting techniques were used.  For the CO and CO2   concentra-
tion curves shown, in Figure 16, two different values of  this
parameter, effective in different portions of the  bed, were
used to make the theoretical curves agree with experiment.   It
is hardly surprising then that the agreement between experi-
mental and theoretical gas composition was so close.

The above discussion raises the question as to how Mayers
elected to select the other average values of the  various co-
efficients in his equations,,' The methods he employed will be
briefly covered.  Mayers (97) discussed how the othe,r model
coefficients were estimated in a second paper and  pointed out
that they were chosen to make theory agree with experiment.
The effective thermal conductivity of the fuel bed was esti-
mated using an extension of an equation developed  by, Terres  est
al.  (103) which proposed that the effective thermal conductiv-
ity across any plane in the fuel bed was equal to  the average
of the thermal conductivities of the air and fuel, weighted  in
proportion to the part of the plane covered by each and  an
equivalent conductivity due to black body radiation across the
voids.
                                                    :C
         K^  =  (1  -  6)ka  +   6k  +   40T^d x  10"8       .     (68)
where d is an effective mean radiation beam length.  Terres et
al. developed an experimental correlation of d with the bulk
and true density of the particle and the bulk density of a dust
that showed ivo increment in conductivity due to radiation.  This
correlation developed by Terres was a function of particle size*
Terres et al. or.Iy worked with particles of 6 mm in -size and
Mayers had to extend the correlation of the mean beam length so
that he could estimate d for the size coke particles. that were
used by Nichollr, ., which ranged from 1/2 in. to 2 1/2; in. diam-
eter.  Mayers used the correlation he developed and "equation
(68) to obtain an estimate of the bed thermal conductivity for
the different .size particles at different temperatures.  Com-
parison between his results and those predicted from equation
(13) shows that the two estimates differ by upwards "of 50 per-
cent and could both be quite badly in error because of the
rather scanty data on the thermal conductivity of coke and coal
particles o  Mayers appears to have used a value of ks of around
0.1 Btu hr-ift" - op"1,, while the author favors a value of at
least an order of magnitude greater.  Short of a detailed ex-
perimental investigation, there is no answer to this problem
and the discussion, serves only to point out the difficulties
always encountered in trying to model fuel bed combustion.
                         -  72  -

-------
For Kg to be a constant in Mayers' energy balance equation, he
Had to select an average value of Ts that would give a value
applicable throughout the fuel bed; the value to pick is not
intuitively obvious particularly as the fuel bed temperature
rises sharply after the ignition front.  Mayers circumvented
this problem by balancing the average effective thezmai,con-
ductivity and the average value for the solid-gas heat trans-
fer coefficient.  For his solution to fit Nicholls1 data, he
found that two quantities, each a function of K| , h  and the ratio
of the specific rate fp^) of the reaction C + O2 - CO2 to the
rate of flow of gas (G) through the fuel bed, had to assume
certain values.  The ratio Hj/G was assumed to b@ accurately
known from measurements of    the rate of decay of the oxygen
concentration in fuel beds»  From these two independent ex-
pressions he was able to determine K| and hv.  The value of Kg
chosen in this manner provided the clue to the temperature
level that should be selected in evaluating the mean bed con-
ductivity; this temperature, around 1800°F, is somewhat lower
than the peak bed temperature for restricted ignition (2300°F)
or the temperature of most of the bed (2200°F).  The heat
transfer coefficient (hv) calculated by this method was 6000
Btu ft'^hr"1 °F~^s this appears very high, and on the basis of
most heat transfer work, including that of Furnas (62), a value
at least an order of magnitude lower would seem more appro-
priate.  Mayers (97) points out this anomaly and suggests that
gas radiation accounts for the difference.  This is unlikely.
Furnas1 data were obtained by heating packed beds to tempera-
tures up to I100°C with combustion products having the approx-
imate composition 5% CO^, 10% H2O, 10% C>2 and 75% N2; the ef-
fect, therefore, on heat transfer by gas radiation would be
expected to be about the same in both Furnas' experiments and
a fuel bed.  The high temperature data obtained by Furnas were
for packed beds of iron ore.  It could be argued that the ex-
ternal resistance to heat transfer of this material is differ-
ent from that of coke.  Recalculating Furnas1 data using equa-
tion (6)  still gives an estimated effective heat transfer
coefficient for coke of around 200 Btu hr~1ft~3 °F  , well
below the value of 6000 Btu hr~1ft~3 °F~1 used by Mayers.

Mayers used his solution to calculate burning rates and igni-
tion rates in fuel beds as a function of the underfire «Ar rate.
The results of these calculations are shown in Figure 20.  The
burning rates followed experimental results quite well because
of the force-fitting of the theoretical and experimental gas
composition profiles.   As has been already mentioned, the cal-
culation of the burning rate for a coal bed in a non-ignition
controlled regime is relatively simple because equilibrium is
often closely reached and variations in gas flow hardly affect
the gas compositions.   (In a refuse bed this is not so, a
point which will be returned to later.)   Mayers' ignition cal-
culations were markedly different from Nicholls' results (see
                         - 73 -

-------
             0     IOO    ZOO   300   4OO   5OO
                AIR FLOW LBS. PER MR . PER SQ.FT.
Fig. 20.     Mayers' Theoretical Relationship between
     Combustion and Ignition Rates  and  Air Flow Rates
     (90).  The solid curves give ignition rates for
     different particle sizes.  The dashed curves give
     combustion rates for different bed depths in feet.
                  - 74  -

-------
Figure 9)  although they showed a similar  trend with particle
size.  One reason the ignition rate showed a much  sharper peak
and a maximum at a lower air flow rate than observed experi-
mentally was that in the formulation of the rate termsfl no
temperature dependence was considered.  At both high arid low
air flow rates, the change in temperature level between the
ignition plane and the plane of zero oxygen concentration will
cause the average oxygen-carbon reaction  rate to fall more into
the chemical controlled regime, leading to a broadening of the
exothermic reaction zone.  This broadening will in turn lead
to a lower ignition rate.  Another reason may be that  tive igni-
tion problem is an "internal" one, meaning that it raay depend
on the unsteady-state heating of the fuel particle.  Such a
phenomenon cannot be explained by an isothermal model.  These
points will be discussed in more detail on pages 94-106.

The M. W. Thring Theory {9__8, 99) .  Thring' s theoretical work
was concerned with calculating reaction rates in coal beds as-
asuming an idealized fuel comprising only carbon.  Thring as-
sumed that the physical processes of diffusion and mixing
rather than chemical reaction were the rate-determining steps
in the combustion process,,  The fuel bed was considered as a
series of inter-connecting passages between particles  through
which the gas flowed and was rapidly mixed by turbulent eddies
in the core of the void spaces.  The chemical reactions taking
place within the core of the voids were considered to  occur at
an infinite rate.  Each particle was assumed to be surrounded
by a boundary layer where the flow was laminar and through
which the gases had to diffuse before they could react with
the solid surface.  The controlling resistance was therefore
assumed to be the rate of diffusion across the boundary, layer.
The reactions, considered to be taking place as postulated by
Bangham and Thring (104), were in three zones:         '

     Zone I   Particle surface C + 1/202 * CO               (69)
              accessible to 02

     Zone II  Gas phase region CO + l/2O2-»-  CO2             (70)
              accessible to £>2

     Zone III Particle surface CO2 + C  •»•  2CO              (71)
              accessible to C02

This three-zone theory was in contrast to Mayers'  two-zone
theory, where, it will be recalled, the reactions  were con-
sidered to be

                C + 02 + C02                                (72)


                C02 + C -»• 2CO                               (73)
The change in the molar flow rate of oxygen at any point within
                         - 75 -

-------
the bed from consideration of the reaction  scheme  stoichiometry
is

     O.Q34G -22. = -   N
                        co
where pQ  is the partial pressure of oxygen, NCQ  the number of
moles   2 of carbon monoxide diffusing away  from  the fuel sur-
face, and -NQ,, the number of moles of oxygen diffusing  toward
the fuel surface.  The reaction scheme proposed implies that


                       Nco - 2(No2+

and that as long as any oxygen remains

                        p    = 0.21  - p^                   (76)
                         C02             2

The flux of A through the boundary layer is given approximately
by

                           h     i            —1  —2
                  N, = k,, (p, - p,)   moles  hr  ft            (77)
                   A    G  A    A

where kG is a suitable mass transfer coefficient.  Using the
fact that the concentration of the various reacting  species
at the solid interface is small in comparison  with the  bulk
concentration, equation (74)  can be rewritten  as

                        . dpo,
                   0.034G —,-=- =   -k_(p,, + 0.21)             (78)
z=0, p_ = 0.21, gives
       2
Thus the oxygen concentration falls to zero in a  finite  dis-
tance ZQ and the rate of consumption of oxygen is greater
than that predicted by the two-zone theory of Mayers  (90)
because of the allowance for the reaction of oxygen with car-
bon monoxide.  The oxygen consumption distance from equation
(79) is given by

                     Zo = 0.024G                             (80)


Using reasoning similar to that above, Thring also developed
equations for the changes in concentration of carbon  dioxide
                          - 76 -
which on integration, using the boundary condition  that at
                   es

                    = °-42

-------
and carbon monoxide after.all the oxygen had disappeared.  The
results are shown in Figure 21, which also shows the rate of
disappearance of oxygen calculated from the two-zone theory.
The general shape of the gas composition curves followed that
measured within fuel beds (see Figure 10), but there were ob-
vious differences, namely that in the real bed some carbon
monoxide appears before all the oxygen is consumed, showing
that the mixing in the voids is not perfect, as assumed by
Thring.  However, this approach does have some merit in that
the curves are calculated from basic principles and are not
derived empirically from the data that the theory is required
to explain, as was the case with Mayers.

Using the correlation of Gamson  (105), which is similar to
equation (13), for the rate of mass transfer within packed
beds, Thring calculated the length ZQ of the oxygen consumption
distance.  This distance was found to be of the order of a few
particle diameters, a situation which fits well with experi-
mental observation.  The variation of ZQ with gas velocity,
particle diameter, void fraction and temperature can be found
using equations  (80) and  (13) under the assumption that jp *JH-
The result, with some approximation in the coefficient of the
temperature dependence, is
(G)°-41  (Dp)1'41

 (1_fi)1.41*0.37
                                                          (81)
Equation (81) shows the weak dependence of ZQ on the air rate
and temperature and its somewhat larger dependence on the par-
ticle size; this result was anticipated in the discussion on
pages 39-46.

Thring also presented a plot of various experimental relative
carbon saturation factors  (RCS factors) as a function of dis-
tance from the ignition front, normalized with particle size,
and compared them to his theoretical calculations.  These are
shown in Figure 22.

The RCS factor was developed by Thring  (109) as an aid in pre-
dicting coal bed gasification rates.  The factor is the ratio
of the amount of carbon gasified  (as determined from the bed
gas compositions) to the theoretical maximum amount that could
be gasified per mole of underfire air.  The maximum amount of
carbon that can be gasified is limited by consideration of the
carbon, carbon dioxide and carbon monoxide equilibrium at tem-
peratures that can be practically achieved.  At around 2000°F
in an air-carbon combustion system, the equilibrium carbon
monoxide content is approximately 34.6 percent, and does not
increase significantly with further increases in temperature.
On this basis, the RCS factor FRCS can be readily shown to be
                          - 77  -

-------
    0-30,
            i?n  /P calculated
             v* ' f
                for some
             on two -zone theory
Fig. 21.  Theoretical Gas Compositions Profiles Calculated
          Using the Three-Zone  Theory of M.W. Thring  (98).
          Also shown is the  corresponding oxygen profile"
          calculated using the  two-zone theory of Mayers
          (90).
                     - 78 -

-------
                      Particle s/ze
Fig. 22.  Comparison between Theoretical and Experimental
          Relative Carbon Saturation Factors for Air-Coal
          Beds (98).  Curve A, P. Nicholls, 600°F pre-
          heat (24); Curve B, P. Nicholls, 80°F preheat
          (24) Curve C, L.H.F. Nichols, 60 ft /min  (106);
          Curve E, L.H.F. Nichols, 20 ft /min  (106); Curve
          F, -R.A. Mott and R.V. Wheeler (107); Curve G,
          M.W. Thring, theoretical curve for D  = 3 cm,
          air flow rate, G = 20 Ib hr  ft  usiRg Lof and
          Hawley (108) mass transfer correlation; Curve H,
          same as G but based on Gamson (105) mass trans-
          fer correlation with fuel bed voidage, 6 = 0.40;
          Curve I,  same as H but with 6 = 0.50.
                     - 79 -

-------
given by


                                               [°2J
                                                           (82)
     RCS    TT75 |      ! + o.Ol] [C02] +  [02]

where C02 and 02 represent the measured concentrations of  car-
bon dioxide and oxygen respectively in volume percent.

The RCS factor can be developed as a function of any two of  the
three combustion gases:  Oxygen, carbon dioxide and carbon
monoxide.  The use of oxygen as one of the gases seems to  be m.
general choice (36, 110$  as it disappears rapidly within the
fuel bed and Kiakes computation of the RCS factor easier.   The
rapid disappearance of oxygen leads to uncertainty in the  con-
centration profile within the bed and can cause some errors  if
the RCS factor is calculated, for the gas close to the ignition
front; however, r.he errors decrease once all of the oxygen has
been consumed.  Two points should be noted about the RCS fac-
tor.  First, the convenience of casting the RCS factor into
this particular form of equation (82)  is only possible with  a
pure carbon-air combustion system.  For a more typical inciner-
ator fuel, such as wood or paper, it is not strictly possible
to calculate a RCS factor from the carbon dioxide anel oxygen
concentrations because the excess hydrogen in the fuel will
compete for part of the oxygen in the underfire air.  Second,
the RCS factor does not take account of any gas phase reactions
(i.e., CO + 1/2O2 -> CO2;  and as such cannot be used to give  a
measure of the heat release rates within the fuel bed.

Figure 22 shows that the experimental RCS factor increased
initially at about the rate predicted by Thring's theory,  but
that at greater, distances froia the ignition front the theo-
retical RCS factor increased faster than the experimental
curves.  This is as expected since the theory, as shown above,
fits experimental observations quite well up to the distance
ZQ.  After this point, the discrepancy is readily explained,
as the theory assumed that all the reactions taking' place  were
mass transfer controlled and this is definitely not true for
the C + C02 -<" 2 CO reaction at fuel bed temperatures.  Thus the
CO concentration calculated under the assumption that the  re-
action was mass transfer controlled would increase more rapidly
than if the reaction was partially kinetically controlled.

Some quite clear evidence, shown in Figure 23, that the C  +
CO2 •* 2CO reaction is not raass transfer controlled was supplied
by Thring (99) f who calculated CO2 reduction rates from P.
Nicholls| data (24).  The dotted line  (R.R.aT2) is shown for a
typical increase in reaction rate with temperature under mass
transfer controlled conditions.  Obviously the actual tempera-
ture dependence of the Boudouard reaction shown by Figure  23
is more in line with a kinetics controlled reaction rate.  The
                          -  80  -

-------
  so
    I3OO
liOO
Fig. 23.  CO2 Reduction Rates  in a  Bed  of  1  -  1-1/2
          in. Coke Particles with Various  Preheat
          Temperatures  (99).   Data  of P. Nicholls
          (24).

-------
experimentally observed jump in apparent activation energy at
1540°C has not bean fully explained, but Essenhigh  (110) has
suggested that it was either a peculiarity of the experimental
conditions or that large pores opened up, helping to lower any
internal diffusional resistance that was present.  Wheeler
(141) and Weisz and Prater (142) have shown that the apparent
activation energy under conditions where internal diffusion is
important is one half that where these effects are absent.  The
jump in activation energy   shown in Figure 23 follows this
pattern and it is possible that, at the higher temperatures,
the rate was controlled solely by chemical kinetics,, while at
the lower temperatures the rate was partly controlled by in-
ternal diffusion.  However, experimental data taken under
laboratory conditions with pure carbon (111) have indicated
that the temperature dependence of reaction rates of to
Boudouard reaction was even greater (50-58 kcal/g.male"-1-) tfcan
that shown in Figure 23.  This suggests that the rate observed
in a fuel bed may be controlled by physical as well as chemical
considerations  (i.e., the hindrance of the reaction by the
presence of ash) .

The D. B. Spaldir.g Theory (7_3J .  Spalding utilized his mass
transfer theory to predict gas compositions and temperature
profiles within fuel beds.  He compared his results to the ex-
perimental results of Kolodtsev (91) .  His theoretical calcu-
lations and Kolodtsev ' s experimental data are shown in Figure
24.  The agreement was quite good up to the point of complete
oxygen consumption, and then, as expected for a mass transfer
based theory, the concentrations of CO increased far more rap-
idly than observed experimentally.  This point has been ade-
quately discussed on pages 75-82 .  The theoretical temperature
profile was quite close to the experimental one but peaked at
a higher value, probably because of heat losses from the bed
unaccounted for by the theory „  The general agreement of the
temperatures was not too surprising, since good agreement be-
tween the gas compositions,  at least up to the point of zero
oxygen concentration, was obtained.  The theoretical tempera-
ture, after the plane of oxygen concentration, decreased far
more rapidly than experimentally observed, in line with the
larger CO concentrations that were calculated.

Unfortunately, the agreement of the CO and C02 compositions up
to the plane of zero oxygen was based on a completely arbitrary
condition which arises in the following way.  The Transfer
number (B)  can only be calculated from knowledge of the gas
composition at the solid surface and in the bulk of the gas
stream.  In general, the solid surface composition is only
easily calculated if it is assumed that the reaction is mass
transfer controlled; the surface concentration is then close
to zero.   For carbon combustion, it can be postulated that the
oxygen concentration at the surface is zero and that the CO2
                         - 82 -

-------
                               Theoretical volumetric
                               concentrations and
                               temperatures
                                                      2,000
                                                       °C
                                                           I
                                                      woo
                                                          I
                                                          a
  o-m
  0-150
  0-100
 \
  0-050
Fig. 24.
                                                   cm.
Gas Compositions  and Temperature in a  Fuel  Bed
(7_3_) ,. "Experimental  data for bed of particles
of diameter  (K32  cm, air velocity 150  cm/sec.
Solid curves,  theoretical volumetric concen-
trations and temperatures; dashed curves, ex-
perimental,  from  Kolodtsev (91).

-------
concentration is close to zero  (if it is assumed that the
Boudouard reaction is mass transfer controlled).  However,  the
CO concentration at the surface is not zero.  To get  around
this difficulty, a suitable Transfer number  (a  conserved prop-
erty of the system) can be found by elimination of the CO  con-
centration by appropriate addition of the oxygen and  carbon
atom;balance.  This puts a constraint on the calculations,
since they are now independent of the CO concentration,  and an
assumption has to be made concerning the CO/CO2 ratio J»efore
gas composition plots can be made.  Spalding found that if  he
assumed that the CO and CO2 concentrations were equal, a good
fit between experiment and theory could be obtained.   Spaldistg
has shown that the Transfer number can be related to  Thritig'^S
relative carbon saturation factor.  Spalding's  analysis ob-
viously has the same shortcomings as were pointed ©ut in the
discussion of Thring's RCS factor, however, and therefore -can-
not be considered as a completely predictive theory.

The R. S. Silver Theory (100, 101).  A different method df
approach from that of Thring and Spalding was taken by R. S.
Silver (100), which invoked the analogies between mass and
momentum transfer in order to predict gas composition through-
out the bed.  The basic assumptions of the model are  listed
below:

     (i)   The fuel bed was considered to be made up of  a num-
ber of parallel irregular channels.  It was assumed that it
was legitimate to discuss an average channel and to consider
it a tube with carbon walls.

    (ii)   All surface reactions were mass transfer controlled
and gas phase reactions mixing controlled.

   (iii)   The fraction 3 of the total number of molecules flow-
ing through the channel which struck an element of wall  was
proportional to the axial length of the element.  The fraction
of any particular species which struck the wall was assumed to
be the same fraction of the local number of molecules which
struck the wall.

    (iv)   Every oxygen molecule which struck the surface re-
acted with the carbon, a    portion x of the oxygen molecules
giving carbon monoxide and the remainder (1-x)  giving carbon
dioxide.   Every carbon dioxide molecule which struck  the car-
bon reacted to give two molecules of carbon monoxide.  Oxygen
and carbon monoxide reacted in the gas phase to form  carbon
dioxide;  this reaction was confined to an annulus close to  the
surface of the channel.

There are two distinctions between Thring's model and Silver's
model.   First,  the introduction of the variable x generalizes
                          -  84  -

-------
Tnring's model to fit the experimental  picture  of
oxygen reaction for which Arthur  (112)  and  others  have shown
that the primary CO/C02 ratio  can be expressed  approxlm&tely
by                                                      <


                =  103-V12<00°/RT                     '    (83)
Second, Silver 's model utilised  all  possible  reactions  at all
possible positions in the bed  and did not  restrict it 90$$ to
the two-zone burning raodel of  Taring.   It  will  be  refill Add
that in Thring's development CO2 was assumed  not to Jtf$*£fc with
the carbon until ail the 02 was  consumed.

If NO^ / Nco and ^QQ? are ^ie mo^ar flaxes  of  oxygen,  ^carbon
monoxide and carbon dioxi.de,, and y0? , yg0  an(^ y^Qj fc^^f re~
spective concentrations at any point 1  in  a channel,  then the
assumptions above,, coupled with  the  reaction  stoichionttetry ,
gives  the following differential equations:
     dNc°2                  r          i
     -an— = -6Nco9 + 3No9  M- - x +  7 yco(1 + x)
                  2      2  L

     dN
                               1 + x) + 2BxN
                  ,      ,                  o0
                  2      z                   2

Now, since the overall reaction  scheme  is not mole-conserving,



                                                             <87)
where N1 is the initial number  of moles  present.   Equations
(84, 85, 86) can then be rewritten  in  terms  of  yA.-.,  yA^  ,  and
 t    .                                           \~\s    \+\J+±
y' , where                                              2

  2
                                                            (88)
The resulting equations are  intractable  as  they staled,  bwt by
making an approximation that y'  =  yco Silver obtained  a
rather algebraically complex solution; the  interested reader
is referred to the original  paper  (100) .  From his solution
                          -  35  -

-------
to these equations Silver calculated the gas composition    pro-
files through a bed as a function of u = 01.  These compositions,
for the case «*»0>. are 'shown in Figure 25.  Physically, u  repre-
sents the fraction of the total number of molecules flowing
through the channel that have struck the carbon surface by  the
time point 1 is reached.  Figure 25 agrees in general shape
with experimental Observation, but the maximum CX>2 content
appears to be rather low.  For values of x greater than O,  the
CO2 curve is depressed and the carbon monoxide curve rises
more steeply.  The asymptotic value of the CO concentration is,
of course, 35% since there was no equilibrium constraint  in
Silver's model? fehi^ value is theoretically onlf possible at
high temperatilgfes but is a good approximation for most fuel bed
conditions.  ''£h« other interesting feature of Figure 25 is  the
exponential depay with distance of the oxygen concentration re-
sulting from the solution of equation (84).  The oxygen con-
centration ia this theory, like Mayers', will never reach zero
in a finite distance.  This arises in Silver's model as a con-
sequence of the assumption that the reaction CO * l/2<>2 •*• CO2
takes place 'ih a thin shell close to the carbon surface and
competes for the BNO-> molecules of oxygen supposedly about  to
strike the carbon surface.  This points out a flaw in the for-
mulation of eqmtion (84) , for by definition 0NO2 is the  num-
ber of moles of oxygen striking the carbon wall, not just pass-
ing into a boundary layer.  The proper formulation of the oxy-
gen balance equation (84) using 0 as defined by the list  of
assumptions woul
-------
1-0    I'D    3-0
  Valuta tfu

   Figure /a
                                      30
                                     l#
                                       10
                                         V
                                               \
                                                    i   i   i
1-0     £•&    3-0
   Valun tfu.

 Figure /b
                                                               *0
Fig.  25.  Gas  Composition within a  Fuel Bed Calculated
           by Silver  (100) .
                        - 87  -

-------
mass transfer., which takes place at  surfaces.   Silver attempts
to make some implicit correlation  for  the  amount of form drag
but the correction is only very approximate.   When values of
3 where calculated from experimental results on pressure drops
across a cold simulated fuel bed,  the  depth of the bed re-
quired to give a value of u=i was  found to be  4 1/2 in.  for a
fuel of the I - 1 1/2 in. size.  This  distance doubled for a
fuel of 1 1/2 - 2 in, size, an increase far greater than that
experimentally observed.  The gas  compositions of Figure 25 at
u=l are approximately correct for  this depth from the ignition
zone for a fuel of the 1-1 1/2 in. size, but the oxygen con-
centration i£ much higher than typically observed.   The  calcu-
lation of 3? tossed on a correlation  of mass transfer in  packed
beds as a function of pressure drop, is a  weak link in this
theory because of the difficulty of  adequately correcting for
form drag.  Even the usual jD correlations of  mass transfer in
packed beds are rather suspect when  applied directly to  a fuel
bed because of the rapid acceleration  of the underfire air in
the vicinity o-f the ignition sone, which probably causes hy-
drodynamic conditions different from those occurring in  an
isothermal system.

From the foundations of his first  paper, Silver (101)  developed
a theoretical treatment of the temperatures attained in  combus-
tion and a condition for sustained ignition in a fuel bed.
Only the latter development will be  covered here as it is one
of the few theories which deal with  combustion stability in a
fuel bed.  From ths previous development,  it as assumed  that
all molecules; striking the carbon  wall react.   For this  assump-
tion to have been valid, the chemical  rate must have been
greater than the mass transfer rate; on a  unit area basis this
can be expressed as


                      n   —E/RT      2
                r ^ Av   e Kj/KJ- >    *                    (on)
                r = AYQ^ e           jjg                    
-------
     hg  =\ heat transfer coefficient, assumed equal to both
            gas and solid                                 /

The rate of heat generated by reaction, assuming chemical kinet-
ics control,  is,

                 Qgen £ ndAyD e-E/RTAHn                      (92)
                            U2        °2

where AHQ2 is the heat of reaction of oxygen with carbon mon-
oxide per unit weight of 02.  For sustained combustion


                    Qgen  >  Qloss                          (93)

or


           AyS2AH02e"E/RT - hs(Tf - V + hs(Tf - V       (94)


Silver showed that the term on §.he right-hand side of equation
(94) was in fact equal to $YQ N1- and concluded that for a fire
to exist the temperatures    2	 had to be great enough for
the chemical kinetics not ~ ITd   to be a controlling factor.
This appears to have been thu first quantitative statement
that has been published concerning the phenomenon of combustion
stability in a fuel bed.  Silver's development will not be
covered in detail here as a discussion (which is somewhat more
general than this approach) on fuel bed combustion stability
will follow on pages 113-118. Briefly, the development required
that hs be related to 3 using the Reynolds analogy for'the re-
lationship between hs and the friction factor, and that T
could be approximated as a linear function of Ts.

  Experimental and Theoretical Work on Refuse Combustion

The Experimental Studies at the Bureau of Mines  (26, 27, 28).
The Bureau of Mines pioneered some of the first small-scale
experimental work in incineration about fifteen years ago.
Their work has been concerned primarily with the testing and
development of the vortex incinerator  (27, 28), so called be-
cause the combustion air (almost all supplied* above the bed)
is fed tangentially into the overfire region.  The main con-
cern of the Bureau of Mines was the complete combustion of
the volatiles distilling off the bed, and thus the major ex-
perimental effort was concentrated in the gas phase overfire
region.  From their most recent data (28), it appears that
burning rates of         16 - 30 Ib hr=^"ft~2 of bed area are
achievable with this system when using a standard synthetic
refuse containing 29% moisture,, 56% volatile matter and 12%
fixed carbon.  Burning rates at the lower end of this range

-------
                                   I
were those most commonly reported.  In these tests a very  small
fraction of the total air requirements for combustion  (approx-
imately 5%) was supplied as underfire air, but from the  com-
position and temperature profiles above the bed it was likely
that a portion of the cold overfire air, still with some oxy-
gen content, spiralled down the side of the unit and then
flowed back up through the center of the bed.  The extent  to
which this will occur can be determined from a momentum  and
buoyancy force balance and is roughly proportional to L1/* and
to     Tc  1/2 where T  is the cold gas temperature at the
   1 ~ xT  walls, Th iS the hot gas temperature at the center
of the unit, and L is a characteristic height.

The only work that the Bureau of Mines has done on fuel  bed
combustion has been with the equipment shown in Figure 26.  The
unit was divided into two sections which were joined together
with a water seal.  The bottom section, which contained  the
fuel bed, could be weighed  throughout the course of combustion
with the scale,,  The temperature profiles throughout the bed
were measured with Chromel-AIumel thermocouples inserted through
the insulating walls ct regular intervals above the grate.  The
underfire air  (sometimes not used) was supplied by a blower
through the grate supporting the fuel bed.  The overfire air
was supplied tawgentially front ports at a fixed height above
the bed.^ To perform a test, the two sections were separated;
the bottom section was      charged with fuel and the thermo-
couples inserted into the bed.  During this time, the top  sec-
tion was preheated with the gas burners.  When the temperatures
in the top section refractories had stabilized, the two  sec-
tions were joiaed together arid the progress of combustion  was
monitored by noting the change in weight of the bottom section
and taking gas samples from the stack.  No auxiliary fuel  was
used in any of the tests.

The fuel used in these experiments was composed of wood  chips
(3/4 in. sg-uaife) , newspaper, corrugated carboard (1/4 -  2  in.
strips), and leafy vegetables such as lettuce and cabbage  (3/4
in. square),  The moisture content of the fuel was adjusted to
either 25 or 50 percent moisture by varying the proportion of
leafy vegetables.

Typical  teaperature-tirae profiles from one of their tests  are
illustrated In Figure 27, which shows distinct plateaus  when
the temperature &t eacl'i thermocouple reached 212°F, correspond-
ing to the v&porisatiori of theroisture.  When vaporization was
nearly completed, the temperature rose quickly and the fuel
ignited  rapidly.   From this type of data the rate of travel of
the evaporation front and the ignition could be calculated; a
typical  result is shown in Figure 28, along with the corre-
sponding weight loss.  For these calculations, the rate  of
travel of the drying front was determined from the position of
the leading edge of the drying front and the rate of travel of
                         - 90 -

-------
Tangential air -
      Gas
Gas
                                              Gas sample
                                                 Thermocouples
                                                 Gas ports
                                                 Tangential airs lot
       Hood in heating position
                  Hood in test position
  Fig.   26.   Apparatus  Used  by the U.S.  Bureau of  Mines
               (26)  for Studying Combustion Characteristics,*
               of  Refuse.
                           - 91  -

-------
   1,800
   1,6001-
   1,400 h~
   1,200 I—
j 1,000 \-
OE


§
cc
LJ
Q.
   400
   200
                                           Thermocouple  Distance from
                                              number   top, inches
   800h-
   600 h-
                          5      8      10

                              TIME, minutes
   Fig.  27.
Bed  Thermocouple History,  Incineration of  Syn-
thetic Feed  with 50  percent  moisture, total
air  rate 61  cfm, roof temperature  1800°Ff  (27),
                         - 92  -

-------
201	1	1	1	r
    Travel of evaporation front
                                10
                          TEST TIME, min
                                                15
Fig. 28.   Travel of Evaporation and Burning  Fronts and
           Total Weight  Loss as Functions of  Test Time
           in a Typical  Bed  of Synthetic Refuse Having a
           Moisture Content  of 50 Percent and a Bed Depth
           of 18 Inches  (27).
                       - 93  -

-------
the ignition front from the propagation rate of the 500°F plane
through the bed.  From Figure 28, it appears that the burning
rate was at a maximum at the beginning of the experiment and
     continuously decreased throughout the duration of the
run.  However, this high initial burning rate is not possible
on a physical basis, for the thermal wave could not have pene-
trated very deeply into the bed initially and thus pyrolysis
and "burning could not have occurred to any appreciable extent.
The data show that the distance between the drying and ignition
fronts increased initially and then remained almost constant.
This is roughly consistent with a drying limited combustion
process.

This study provided useful data on a specialized vortex unit
and some valuable qualitative insights into the combustion of a
high moisture content fuel, but the experiments do not lend
themselves to 'quantitative analysis.  The major deficiency of
the study was tha inability to measure the quantity of air that
spiralled down the cold walls and up through the bed.  The
quantity of air that is drawn through the bed in this manner
will depend in a complex way on the refractory temperatures,
the firing rate of the tangential air, the geometry of the
system, and on the burning rate itself.

Research of £gsejrifoigh et a 1.  Essenhigh and his co-workers at
Pennsylvania Sti/Ee University have presented a number of papers
(30-47) in the general incineration area and have attempted in
some  of them to develop theoretical equations for design pur-
poses  (32, 36, 37., 3J5_, 41 „ 43, 47) .  Only those papers and sec-
tions of~~papers which deal with 'the solid bed combustion pro-
cesses will be discussed here? many of the papers are also
concerned with modelling conditions in the overfire region.

Some of the p&p-ars (37_,, 3_3) which dealt with the combustion of
the solid fuel developed a simplified Mayers-type model for the
rate of propagation of the Ignition wave through a fuel bed
simulated by a eat of dowel rods placed vertically in a com-
bustion pot.  The analysis encountered the same problems as
Mayers' did in estimating the physical properties of the fuel
bed and was essentially a semi-empirical approach.  The authors
concluded, as did Mayers, that radiation played an important
part in the ignrtion process, since the observed ignition
rates were much higher than those predicted from their theory
on the assumption that all of the heat fed forward for ignition
was by conduction alone0  When a contribution for radiation was
added into the tliamal conductivity term using a mean radiating
length of a reasonable magnitude, the theory could be made to
fit the experimental results adequately.  The results from such
experiments must be taken as being only qualitative as this
analysis ha® even less merit than Mayers', since the experi-
mental set-up could not be considered a model of any real sys-
tem.
                         - 94 -

-------
The other papers by Essenhigh et al. that  deal  with  solid  bed
combustion attempt to develop theory directed towards  predict-
ing burning rates and maximum temperature  levels  in  beds as
well as the effect of inert content on the burning rates.
Shieh and Essenhigh (4_7J tried to unify much of the  previous
theoretical work of the group and tested the theory  against
experimental data taken on their test incinerator.   As this
paper covered nearly all the group's work  connected  with the
fuel bed combustion processes, it will be  the only one dis-
cussed here in detail <,

Shieh and Essenhigh (47) proposed that the burning rate of a
fuel bed per unit area~(F^J could be given by an  expression of
the form

           FA =  (0.75FRCS) rn     (1_v)  (g.A.M)              (95)
where F^g is the relative carbon  saturation  factor  (see pages
75-82), mQ2 the mass fraction of air in the underfire  air, 6
the air supply rate per unit, area, V the volatile  fraction of
the fuel on a dry-ash-free basis,  and A and M the  ash  and
moisture fractions of the fuel, respectively.  The derivation
of this equation was based on the  simplified  physical  model of
an incinerator shown in Figure 29.  It was assumed that the
fuel was fed in an over-feed mode  and that the material was
fully pyrolyzed (i.,e.- it consisted only of carbon) before it
entered the combustion and cj&sification zone.  The rate of sup-
ply of oxygen tp the combustion zone by the underfire  air was
mQ-G Ib hr~lft~^0  Assuming all the carbon dioxide formed in
the combustion zone is reduced to  carbon monoxide  by the time
the gases reach the top of sone I(A) of Figure 29, the carbon
burning rate will be (2M 5ra  G  where M  and  M   are the
molecular weights of    	2   carbon and     2 oxygen, re-
spectively.  If the     MQ      bed is not deep enough to per-
mit complete reduction    2     of carbon dioxideeto carbon
monoxide, the carbon burning rate  becomes 0.75m0?GF c  Ib hr~^
ft" .  TO calculate the total fuel burning rate from rhis car-
bon burning rate, the volatile, moisture and  ash content of
the fuel must be considered.  If the volatile fraction of the
fuel on a dry-ash-free basis is V, each pound of carbon is
produced from   1  pounds of dry and inert-free fuel;  and one
pound of char (1-V}  is produced from  	1	  pounds of as-
fired fuel.                          (1-V)(1-A-M)


Essenhigh and Shieh related the RCS factor to the  combustion
bed depth Lr by the equation
"I
J
                T  -                                     (96)
                Lc -
                          - 95 -

-------
     INCINERATION PROCESS
                               INCINERATOR
             BURN-OUT
          SECTION
   .OVERFED
   IGSJIYIOM
   SECTION


     -4-
SOLID PYROLYSIS
   SECTION
                    OVERFIRE
         SOLID BED
  (COMBUSTION &ND GASIFICATION)
          SECTION
               UNDERFIRE
                 AIR
                                 n IB)
OVERFIRE
  AIR
Fig.   29.     Schematic of Test Incinerator Used by
                Essenhicjh et al.  (47)
                    - 96  -

-------
where Ln was defined by the authors as being a characteristic
reaction depth and K a semi-empirical constant of approximately
1.10.  Equation  (96) was derived in an earlier paper by Kuwata,
Kuo and Essenhigh (31) f where the RCS factor was related  to
depth assuming that the rates of the two reactions C + 02 ->• 2CO
and C + CC>2 -*• 2CO were controlled by the diffusion of reactants
to the solid char surface and that the diffusivities of the
reactants and the temperature dependence of the diffusivities
were the same.   In the case of the carbon-oxygen reaction, there
is a good body of evidence to suggest that it is diffusion con-
trolled under the conditions in a fuel bed; but there is  strong
evidence to suggest that the Boudouard reaction is kinetically
controlled under fuel bed conditions unless ash build-up hinders
the reaction.  These points have been fully discussed earlier
on pages 75-82.  The form of equation (96) is very similar to
the empirical expression suggested by Thring  (110), namely

                 FRCS  '  *«-al/D>                        (">

where A and a are empirical constants, L the fuel bed height at
which the RCS factor is required and D the mean diameter of the
particles in the fuel bed*  The RCS factor    calculated from
equation (97) with suitable values of A and a increases far
more rapidly with depth than that experimentally observed? as
discussed on pages      .  Equation  (96) would be expected to
have the same deficiency as it is not materially different from
equation (97).   The simple exponential dependence of the RCS
factor can only  ba obtained using the above assumptions of
Essenhigh and Shieh and, as it is established that the Boudouard
reaction is much slower (about five orders of magnitude) than
the carbon-oxygen reaction at the same temperature and reactant
concentration,, it is not surprising to find that the calculated
RCS factor increases more rapidly with depth than that experi-
mentally observed.

The Kuwata, Kuo  and Essenhigh development of the variation of
the RCS factor with fuel bed depth does suggest bounds that can
be placed on the constant K but it provides very little insight
into the selection of LQ, which, although it has some theo-
retical basis (given the assumptions involved), cannot be con-
fidently calculated, from first principles.  Although LQ can be
estimated from experimental data, the underlying assumptions of
the theory are unsound and the use of a value of LQ determined
from one set of  experimental results in another system would be
unwise.  In addition to the fuel bed material balance, the
authors wrote a  simple energy balance for a differential element
of the fuel bed  and used the equation so developed to predict
the temperature  profile within the fuel bed.  This energy bal-
ance was decoupled from the mass balance equation because the
RCS factor could not be used to give an indication of the heat
release rate in  the fuel bed as a function of height.  The
                          -  97  -

-------
equation used to describe the temperature variation  within the
fuel bed was stated as
         K
          .£ d2T
            az
where for zone I
QR .
G - F 
-------
major assumptions.  First, the authors assumed that  all  reactions
in this zone were exothermic; this is not  strictly correct,  as
near the top of this zone carbon monoxide  would be produced  via
the endothermic Boudouard reaction.  Second, no account  was
taken of changes in gas composition  (given by r) with  depth  as
r was assumed to be a constant.  Obviously the heat  release  at
different points within the bed was  fixed  by the local gas com-
position and temperature.  The authors have completely neglected
this and have essentially assumed a  fixed  heat release rate
scaled with depth by the factor .1 exp(-z/LQ) .  There is  no
physical significance to this   L approach.  Equation  (99) can
be broken down into three parts:   (a) 3/4GmQ , which,  as shown
before, gives the pounds of carbon carried   by the combustion
gases assuming carbon monoxide is the only product of  combus-
tion; (b) (l-6.72r)AHc represents the proportion of  char which
reacted to give carbon monoxide.  The net  heat release is then
given by
                           c
                   AH =    - +  (1  - r)AHC                 (104)
                         n

        Q
where AH  is the heat of combustion of carbon to carbon dioxide,
and n is the ratio of the heats of combustion of carbon to car-
bon dioxide and to carbon monoxide (approximately 94/26.4 =
3.55).  Part (c) , <5exp(-z/L0) f is ^te factor that scales the
heat release rate     L_      with depth;  it is highest when
z=0 and decays exponentially with height up to z=Lc.  This term
gives only a crude approximation to the heat release as it is
based neither on the appropriate kinetics  of the reactions in-
volved, nor on any experimental gas composition profiles within
the bed.

Equation (100)  may be considered reasonably valid for the
simplified sequential pyrolysis-combustion system of Figure 29.
However, in the experimental set-up used in this research, con-
siderable quantities of oxygen were known  to have spiralled
down into the fuel bed from  the overfire region  (46) and it
must be concluded that some  exothermic reactions dTd take place
in zone I (B) .

The three boundary conditions used are open to serious critic-
ism.  The first boundary condition states  that the temperature
of the fuel bed at z=0 is the same as the  ambient, yet the
heat release as given from equation (99) is at a maximum; it
was physically impossible for the temperature at z=0 to be
held at To in the experimental apparatus employed.  Using the
boundary condition that d_T = 0 at z=L  forces the temperature
profile to peak at z=L  dz   and in essence also forces the
profile to fit the experimentally observed form, once a suit-
able temperature at the top of the bed is  specified.  The tem-
perature at the top of the bed was calculated from the remaining
                          - 99  -

-------
boundary condition given in equation (103).  This equation  is
a heat balance taken on a differential section at the top of
the fuel bed; the effect of convective heating of the incom-
ing fuel by the combustion gases has been neglected and no-
where in the paper do the authors say how Qr, the net radiant
heat flux to the top of the bed, was obtained.

The experimental results, against which the above theory was
tested, were obtained on a batch incinerator fully discussed
in another pape^ by Shieh and Essenhigh (46) and Figure 29  will
suffice as a rough schematic of the apparatus.  Briefly, the
unit consists of a refractory box of 2-ft-square cross-section
standing about 10 ft high with 9-in. heavy-duty refractory
walls.  Computer cards were used as the fuel and were fed from
an aperture in one wall near the top of the unit so that they
fell onto the grate, which was made of a perforated steel
plate.  The bed was normally maintained about 10 in. deep with
a fuel feed rate of 120 Ib hr"1 (30 Ib hr-1ft~2).  Air was
divided between overfire air and underfire air in varying pro-
portions, with the total air running from deficient to excess.
When overfire air was used, it was injected through two sets
of four nozzles located so that two vortices rotating in
opposite senses were produced above the bed.  This produced a
stirred section — zone I (A) — roughly 10 in. to 20 in. above
the top of the fuel bed.  Above that, another 50 in. was avail-
able for the (plug flow) burn-out region — zone II(B) .  Ex-
perimental data included temperature and oxygen measurements
in the fuel bed; wall temperatures; gas temperatures, and
measurements of CO, CO2, 02 and H20 in the overbed region.  The
procedure adopted for comparing theory and experiment was by
way of the temperature profile through the bed.  As indicated
earlier, the bed temperatures were found to increase to a peak,
Tmax» at a height above the grate designated by the authors as
Lc.  Normalisation of the temperature profile was based on
these two parameters which were obtained both experimentally
and through calculations based on the theory previously dis-
cussed.  Table 7 shows the comparison of predicted and experi-
mental values of Tmaxand Lc and Figure 30 shows the theoretical
and experimental normalized temperature through the fuel bed.
The calculations were performed for a value of LQ = 0.5 ft.
No values of the other model parameters Cp, r, BQE, Qp and  Q
were given.  Tha theoretical value of Lc Was calculated usinf
equations (955  and (96) as FA was known from the rate of fuel
addition to the test apparatus that was required to keep the
bed depth constant during an experiment.

The variation of Lp with underfire air rate, as calculated  by
Shieh and Essenhigh and reproduced in Table 7, is unusal.   As
the underfire air rate increased (runs 1 to 4) the calculated
value of Lc decreased, as expected from the equations.  How-
ever, with the underfire air flow rate still increasing, but
                        - 100 -

-------
Comparison of Predicted and Experimental Values
      of iMmximiurri C~->d Temperature
and Combt'stiO'- Zone 'C-i^n !i-c) in the Solid B«d  (47)
. i r ._ . ^ n


Run
No.


1
2
3
4
. 5
6
Air ii'p|.i|y '/jts
Clill 1
Gver-IUriUCT-
tire I fire
Lx.ess
I'UX

''re
Air ',!•:< ;d
fits Air 1 r/'

80
80
80
80
40
40
7 I 40
8 I 40

10 29 1230
55
20
25
30
40
50
60
36 MO!
43 i<>:0
50 1&07
0 ',035
14 998
29
973
43 953

{•"•, geri-
mcntal


900
950
980
iOOO
940
950
960
980
i L ,.!i)
1
H ,
1 Pro- I 1'xpen-
! dieted mentaJ
i j
'
0.67 i) 58
O.t>3 ' 058
(159 ' 058
! 0 56 : 0.58
; 0.90 i 0.60
[076 ' 0 60
j 0.67 j (173
| 060 1 073

-------
    1.2
    1.0
                       Predicted Totol
                         Bed  Depth

-------
this time coupled with a step decrease in the overfire air rate,
the calculated value of LC was reported to rise suddenly  (the
overfire air rate was not considered in developing either equa-
tion (95) or (96).  This effect can be seen by comparing the
values of Lc calculated for runs 4 and 5.  With the overfire
air maintained at the higher value, the value of L^ again de-
creased with an increasing underfire air rate.  This jump in
the calculated value .of Lc is anomalous, because equations (95)
and (96) do not take into account the effect of the overfire
air flow rate since they constitute only a simple material bal-
ance on the fuel bed; the authors offer no explanation for this
behavior.  The experimental values of Lc were found to be
approximately constant at 0.58 ft for the high overfire air
rate and did not depend on the underfire air rate.  For the low
overfire air rate the value of Lc increased with underfire air
rate, contrary to the theory presented by the authors.

Another astonishing result emerges from the air supply data of
Table 7.  The burning rate of the cards was not increased by
augmenting the underfire air rate.  On the assumption that the
composition of the computer cards used was approximately the
same as cellulose, the burning rate for runs 1 through 8 can be
calculated; the calculations show that the burning rate for all
the runs was the same, about 16.5 Ib hr  ft"2.  This value is
mucn lower than the value of 30 Ib hr~^ft~2, which the authors
quoted as being typical.  Previous experience with coal beds
burning in the overfeed mode has shown (23) that the burning
rate should be roughly proportional to the underfire air rate.
Departures from strict proportionality are caused by changes of
the RCS factor; out over a wide range of conditions it was
found that the higher fuel bed temperatures, and therefore the
greater reaction rates associated with increased underfire air
flow rates, compensated for the shorter residence time of the
gases in the fuel bed.

The major reason the theory predicts the opposite effect of
underfire air rate on the value of Lc from that experimentally
observed stems from the fact that the area firing rate, FA,
appears to have remained constant with increasing underfire
air; this forced the value of Lc, as predicted from equations
(95) and (96), to decrease.  Physically, this means that a
given value of the RCS factor was reached in a progressively
shorter distance as the underfire air was increased.  This is
completely contrary to all other experiences with overfeed fuel
beds, as well as to the authors' own theory.  Shieh and
Essenhigh state that the characteristic depth L  can be given
as

                     LQ  =  v6d/4D0_N                     (105)
                         - 103  -

-------
where v is the average gas velocity through the bed, d is the
average pore or channel diameter in the bed, D is the coef-
efficient of oxygen diffusing through nitrogen, and 6 is an
effective diffusion path length or diffusion-layer thickness.
The analysis leading to equation (105) is discussed by Kuwata,
Kuo and Essenhigh (37j and contains, like the development of
equation (96), a number of approximations.  However, given
that equation (105)  is valid, it would follow that LQ would
increase as the underfire air flow rate increased for the en-
suing reasons.  The film thickness 6 can be expressed as (113)
                               0.5
                         L
where Dp is the diameter of the fuel particle and M and p the
viscosity and density of the combustion gases.  Assuming that
d and Dp do not vary with underfire air rate
               L  a  .JL-       v-     --£               <107>

                0   LP*DJ

Equation (107) indicates that LQ should increase slowly with
the underfire air rate.  An offsetting factor is the rise in
temperature associated with the increase of underfire air.
The Schmidt number!   _u__ 1  is a weak function of temperature
since y a /T,p a I  p D   , and D a T1'75; hence
                 frL  9  j

                    N   =    V  aT'0-25                    (108)
                     be     p D

so that LO a /G .   Using the data from runs 1 and 4 on the
maximum bed V T  temperature and the amount of underfire air,
LO would be expected, on the basis of equations (107) and
(108), to increase about 70% from run 1 to run 4;  thus the
depth required to reach a given value of the RCS factor would
increase, a result compatible with experiment.

In summary, the two problems with the Essenhigh and Shieh
theory rest with the unvarying firing rate FA with underfire
air and the assumed constancy of LQ.  It is likely that chan-
neling may have caused the burning rate not to increase with
underfire air.  The method used in this study, of feeding the
computer cards onto the fuel bed, and the low pressure drop
grate, almost definitely would lead to bad channeling, espe-
cially near the walls.   The problem was probably aggravated by
the use as a fuel  of computer cards, which would tend to pack
together in layers inaccessible to the underfire air.
                         - 104 -

-------
this time coupled with a step decrease in the overfire air rate,
the calculated value of LC was reported to rise suddenly  (the
overfire air rate was not considered in developing either equa-
tion (95) or (96).  This effect can be seen by comparing the
values of Lc calculated for runs 4 and 5.  With the overfire
air maintained at the higher value, the value of L,, again de-
creased with an increasing underfire air rate.  This jump in
the calculated value of Lc is anomalous, because equations  (95)
and (96) do not take into account the effect of the overfire
air flow rate since they constitute only a simple material bal-
ance on the fuel bed; the authors offer no explanation for this
behavior.  The experimental values of Lc were found to be
approximately constant at 0.58 ft for the high overfire air
rate and did riot depend on the underfire air rate.  For the low
overfire air rate the value of Lc increased with underfire air
rate, contrary to the theory presented by the authors.

Another astonishing result emerges from the air supply data of
Table 7.  The burning rate of the cards was not increased by
augmenting the underfire air rate.  On the assumption that the
composition of the computer cards used was approximately the
same as cellulose, the burning rate for runs 1 through 8 can be
calculated; the calculations show that the burning rate for all
the runs was the same, about 16.5 Ib hr  ft~2.  This value is
mucn lower than the value of 30 Ib hr~^ft~2, which the authors
quoted as being typical.  Previous experience with coal beds
burning in the overfeed mode has shown (23) that the burning
rate should be roughly proportional to the underfire air rate.
Departures from strict proportionality are caused by changes of
the RCS factor; out over a wide range of conditions it was
found that the higher fuel bed temperatures, and therefore the
greater reaction rates associated with increased underfire air
flow rates, compensated for the shorter residence time of the
gases in the fuel bed.

The major reason the theory predicts the opposite effect of
underfire air rate on the value of Lc from that experimentally
observed stems from the fact that the area firing rate, FA,
appears to have remained constant with increasing underfire
air; this forced the value of Lc, as predicted from equations
(95) and (96), to decrease.  Physically, this means that a
given value of the RCS factor was reached in a progressively
shorter distance as the underfire air was increased.  This is
completely contrary to ail other experiences with overfeed fuel
beds, as well as to the authors' own theory-  Shieh and
Essenhigh state that the characteristic depth L  can be given
as

                     LQ  =  v6d/4D0_N                     (105)
                         - 103  -

-------
where v is the average gas velocity through the bed, d  is  the
average pore or channel diameter in the bed, D is the coef-
efficient of oxygen diffusing through nitrogen, and 6 is an
effective diffusion path length or diffusion-layer thickness.
The analysis leading" to equation (105) is discussed by  Kuwata,
Kuo and Essenhigh {_3?) and contains, like the development  of
equation  (96}, a number of approximations.  However, given
that equation (105)  is valid, it would follow that LQ would
increase as the underfire air flow rate increased for the  en-
suing reasons.  The film tliickness 6 can be expressed as (113)
                             -,,0.5
                                                         (10€)
where Dp is the diameter of the fuel particle and n and  p the
viscosity and density of the combustion gases.  Assuming that
d and Dp do not vary with imderfire air rate  ;
                                                          (107)
Equation (107) indicates that LQ should increase slowly with
the underfice sir rate.  Ar, offsetting factor is the rise  in
temperature associated with the increase of underfire air.
The Schmidt m'aruberf _j£  1 is a weak function of temperature
since p a /T~,p a Is P D   ? and D a T1*75; hence
                 i  L  9  J
so that Lo a /G .  Using the data from runs 1 and 4 on the
maximum bed \? f  temperature and the amount of underfire air,
LQ would be expected, on the basis of equations  (107) and
(108), to increase about 70% from run 1 to run 4; thus the
depth required to reach a given value of the RCS factor would
increase, a result compatible with experiment.

In summary, the. two problems with the Essenhigh and Shieh
theory rest with the unvarying firing rate F» with underfire
air and the assumed constancy of LQ.  It is likely that chan-
neling may have caused the burning rate not to increase with
underfire air.  The method used in this study, of feeding the
computer cards onto the fuel bed, and the low pressure drop
grate, almost definitely would lead to bad channeling, espe-
cially near the walls .   The problem was probably aggravated by
the use as a fuel of computer cards, which would tend to pack
together in layers inaccessible to the underfire air.
                         - 104 -

-------
Another point which can be raised in discussing the material
balance equations is the validity of comparing the calculated
value of Lc with the value found from experiment.  First, it
will be recalled that the experimental LC was taken as the
depth where the fuel bed temperature peaked; this was also the
point where the oxygen concentration reached its minimum.  This
point does not compare physically with the value of LC as cal-
culated in the theory*  The experimental value of LC reflects
the degree of back-mixing of the overfire air into the fuel bed
and Che rate of reaction of the oxygen in this air with the
pyrolysis products in zone 1(3) of Figure 29.  The amount of
back-mixing will decrease as the underfire air rate is in-
creased and therefore Lc will  increase, a conclusion in accord
with the experimental observations.  Lc as used in equation
 (96) has a different physical  significance, namely the depth
of fuel bed required to reach  a given RCS factor.  As shown
earlier, Lc will be expected to increase as the underfire air
increases, but not for the same reasons that the experimental
values increased.  Second, it  was assumed that there was sub-
stantially no net hydrogen in  the char throughout the entire
depth of the so-called combustion zone.  This assumes that char
combustion is the controlling  step in the burning process; this
may not be correct as the pyrolysis and drying stages may be
much slower than the combustion stage, being heat transfer
limited in many cases.  It is  possible that the computer cards
are thin enough for this assumption to hold, but this assump-
tion would need to be abrogated for a fuel with a larger char-
acteristic dimension.

The heat balance results, for  the value of Tmax, exhibit the
same anomalous results as those for Lc, as they also show a
dependence on the overfire air rate  (compare the calculated
values of Tmax for runs 4 and  5 in Table 7).  It is possible
that the overfire air rate has some effect on Qr, the net
radiant heat flux to the top of the bed, but as the authors do
not state how this value was calculated it is impossible to
comment on it.  However, if one assumes that Qr is dependent
on the amount of excess air, runs I and 7 and 3 and 8 would
have equivalent heat fluxes to the top of the fuel-bed and
hence equal temperatures TL, since FA is a constant.  The tem-
perature at the top of the bed will directly influence Tmax
but as can be seen from Table  7 the values of Tmax for runs 1
and 7 and 3 and 8 are very different.  In addition to this
anomalous behavior, the calculated value of Tmax progressively
decreased with increasing underfire air rate (at constant over-
fire air rate).  This is in direct contrast to the experimental
values, which increased, in agreement with other experimental
evidence (23).

Table 7 also shows that the experimental values of Tj.-  were
affected by the overfire air rate and decreased by 60*C between
                        - 105 -

-------
runs 4 and 5.  It is felt that this reflects the quantity of
air that back-mixed into the fuel bed, which would diminish  as
the overfire air was decreased and with increasing underfire
air.  The reason that the predicted values of Tmax exhibited
the opposite behavior to the experimental values probably lay
with the unchanging burning rate with underfire air rate.   In-
creases in G would on the basis of equation (99) increase the
heat release within the bed (Qr) , but the major effect will be
felt near the grate at z=0 where the boundary condition of
equation (101| forced the temperature to T0.  At the same time,
increases in G will increase convective heat losses without
compensation from the convective heating of the fuel^flow,  and
thus the predicted Tmax will drop with increases in G.  In
short, despits the superficially good agreement between theory
and experiment, as shown in Figure 30, the heat balance has
many shortcomings and it can only be assumed that the theory
was forced to fit experiment.

The authors conclude in their paper that their theory has been
adequately substantiated by experiment and needs minor refine-
ments for it to have a broader range of applicability.  The
above discussion has highlighted the shortcomings of the theory
and suggests that the authors" conclusion is perhaps premature.
In addition, it; may be questioned if this theory will be valid
for fuel bed conditions where large diffusional resistances to
heat transfer are encountered within the fuel elements.

The Theory of Jf 0 R. Jliessen, A. F. Sarofim et al. (19) .  The
authors have p'resenticT~ir~s imp;li fled global picture of re fuse
pyrolysis and gasification on the assumption that the water
gas shift reaction (CO2 * ^2 + co + H2°) is equilibrated at
the top of the fuel bed.  This assumption was based on Kaiser's
data (114)  from the Oceanside Incinerator on Long Island, New
York.  The assumption of the equilibration of the water gas
shift reaction is often made in calculations on coal gasifica-
tion (115)  when temperatures above 2000°F are encountered.  The
water gas shift reaction is primarily considered as a hetero-
geneous phenomenon, although it can also take place in the  gas
spaces in the fuel bed via a free radical mechanism,

              ,H0O  +  H  *  H_  +  OH                     (109)
               iff             £


              CO  +  OH  t  CO-  +  H                     (110)

There is no direct experimental evidence concerning the rate of
the water gas shift reaction,  but it will obviously depend  on
temperature and may well be strongly affected by fuel reactiv-
ity and catalytic effects of the fuel ash.  There is, however,
a good body of experimental evidence (115) which suggests that
the reaction is rapidly equilibrated above 2000°F.  The authors
                        - 106 -

-------
assume that the gasification reactions occurring within  the
bed produce mainly CO, CC»2, ^2 • an<* H2° witn only  small  quan-
tities of CH4, tars and soot.  Nicholls'  (24)  results  show
that this is a reasonable assumption  (see Figure 13) and
Kaiser's data (114) also substantiate  this assumption.

The refuse was assumed, on a dry/inert and ash free basis, to
have the composition of cellulose.  The composition could then
be represented by C(H2O)n where n had the value 5/6 for  dry
cellulose.  The following overall material balance, which
assumes no ignition restrictions, was written  for  the  gasifica-
tion of the process

00 + 3.76N0 + xC(H00)  ->• aCO + bC00 + cH0 + dH00 + eC  +  3.76N0
 2        2       2  n            222              2

where x is the number of moles of C(H20)n gasified per mole of
oxygen in the underfire air.  Equation (111) contains  six un-
knowns.  Five relationships between these unknowns can be
found from consideration of the three element  balances,  an
overall energy balance, and by assuming that the water gas
shift reaction is equilibrated.  The energy balance, of  course,
introduces one more unknown T, the temperature of  the  gases
leaving the bed.  These relationships are:

                    Element Balances

       Carbon:     a •*- b + e = x                          (112)

       Hydrogen:   c + d = nx                             (113)

       Oxygen:     b + (a+d)/2 = 1 + nx/2                 (114)

               Water Gas Shift Equilibrium

       [C02] [H23     bc
       [H20] [COT  =  da"  ~  'StfGS                          (115)

The amount of energy released in the bed per Ib mole of  oxygen
(QR)  is given by the difference between the heats  of formation
of the products arid react an ts

             products                 reactant  moisture^vaporization

QR = '47,560a + 169,290b + 140,2406* -  320,940(|) -  104,240 (^^Ox


       (CO)        (C02)         (H20)   (C6(H20)5)        (H20)


                                                           (116)
                        - 107  -

-------
where the last term represents the endothermic vaporization of
the moisture in the fuel exclusive of the 5/6 mole per  C  atom
that is chemically bound.  The energy lost from the bed in  the
sensible heat of the off gases (Qp) assuming that the temper-
ature of these gases is 2000°F is given by
                                                          (117)

Qn -  (aC   +bC    +cC   +dC    +3.76C   ) (1940) + (18,000) (n  - |)
 P      PCO   PC02   PH2   PH20      ?N2                     6


where the average heat capacities  (Btu mole"1 "F"1) between
60°F and 20QO°F are C     = 13.8, C    - 8.3, C^ Q - 11.0,


C    =7.6 and C    =8.2.

 P"2
With the rate of heat loss from the bed through side-wall and
radiant-heat losses being QL/ the following balance is obtained


                    QR  =  Qp  +  QL                      (1U)


which provides the fifth independent equation relating the six
unknowns.  The temperature of 2000°F was chosen by Niessen et
al. (19) as it was representative of the data of Nicholls1
(24) and Kaiser's (114)  results on the Oceanside Incinerator.
Although a somewhat arbitrary temperature, the 2000°F figure
may not be unreasonable, as the endothermic reactions which
occur after all the oxygen has been consumed may well be ef-
fectively quenched at temperatures below this value.  The only
other way the temperatures in the bed could drop below this
level would be through radiative and side-wall losses.

Another assumption had to be made in order for the set of
equations (112) and (118) to be solved.  This assumption con-
cerned the amount of char produced per mole of oxygen (e),
which will be a function of the rates of ignition and burning
and as such will depend on the under fire air rate.  From the
previous,discussion on pages 46-59, it can be concluded that
at high air rates, e will be zero and will increase as the
air rate is decreased.

Results were calculated for different moisture contents of the
fuel,  fraction carbon gasified and heat lost or gained by the
bed.  With all other parameters held constant, increases in
moisture content were shown to decrease the amount of CO and
H2 formed; this resulted in a decrease in the percentage of
the total energy release that was given off above the bed (for
                       - 108 -

-------
0% moisture, 745 of total heat release occurred above the bed,
while for 33% moisture this value dropped to 44%).  The same
trend was observed when the amount of heat lost by the bed
increased (all other parameters being held constant).  The
explanation of these results is evident when gasification is
considered to proceed via a net exothermic step  (reaction of
the cellulose to give char, OC>2 and H2O and the vaporization
of any free moisture) followed by an endothermic  step con-
sisting of the reactions C -5- H20 -»• CO + H2 and C  + CO2 •*• 2CO.
With the assumption of a fixed final gas temperature the
smaller the quantity of heat that is available for the endo-
thermic reactions, the lower the concentrations of CO and H2
above the bed.  The amount of heat available for  these re-
actions depends on the moisture content of the fuel and the
heat lost from the bed.  Heat losses from the bed will depend
in part on the furnace configuration and would be greater
with a water-walled unit than with a conventional refractory-
lined furnace.

This simple model showed the necessity of overfire air for
the burn-out of the combustibles issuing from the bed and how
the air requirements would vary with changes in the feed mois-
ture content and heat loss froia the bed.  Much of the combust-
ible would be expected to be mainly CO and H2, with tar and
soot and 014 only present in any quantity near the inlet of
the furnace where ignition was achieved.  The model further
showed that for conditions where the fuel moisture content
increased the burning rate could be kept up if the underfire
air rate was increased; the air helps provide more energy for
water vaporization.,    This function would obviously be helped
if the air was preheatedo  This result concurs with what is
often done in practice and gives a quantitative basis for
a guideline which has been disputed by some incinerator oper-
ators.

The major shortcoming of the model is the simple  one-step char
production assumption of equation (111) and the somewhat
arbitrary assumption about the magnitude of e.  Obviously in
a fuel bed there will be ss&jor diffusional resistances to
heat transfer in many particles and therefore pyrolysis will
continue long after the ignition front has passed over the
particle.  The modification of th,is model, to take into
account events in a real bed, wilfL have to await  a better
understanding of the bed burning processes if the model is to
be truly!predictive.  The development of a more realistic
model, followed by a set of arbitrary assumptions that would
have to be made, at this stage of our understanding of bed-
burning processes, concerning the varying C/H/0 ratios in the
fuel as a function of combustion extent would negate the attractive
                        - 109 -

-------
features of the model as it now stands.

It is possible, on an order-of -magnitude basis, to check the
validity of -tfce 2000°F quench temperature by considering tne
rates of the two reactions C + H2O * CO + H? and C + C02 -*
CO? which are the predominant reactions taking place in tne
endothewic *eg&» of the fuel bed.  A simple heat balance
on a differential section of the bed, assuming equality ot
the gas and solid temperatures and negligible heat capacity
of the solid phase, gives,


                                   -
where c* is the molar concentration of A(lb mole ft  ) and
A0A the*heat of reaction (Btu Ib mole"-1) .  Equation  (119) can
be conveniently changed to a time base by substituting
t = zpg/G , giving,
           St
                         C   p
                         Pg
                                                        (120)
For the reaction C + CO2 -»• 2CO, the data of Wu  (111)  and
Lewis, Gillllsnd and McBride  (116) for electrode  carbon of
30 - 40 raesH sis®, as indicated by von Predersdorff and
Elliott (115) i suggest a reaction rate of the form
 co
For the reaction C -t- H20 * C02 + H2, the rate and mechanism
of which i& understood less well than the Boudouard reaction,
the rate is given by von Fredersdorff and Elliott  (115) for
coconut sheii charcoal as,
                      -62,300'1
                  exp -  — P                    _

r    = - ,- - — • — j - ^Ai' , A^«; - 9  mole  g  min   atm
               "                    ^     y
6.9 x IOSexp
1 + 0'. 014exp
-4?,60'0
RT j
^5,000 ^
RT
Pco2
PCO + 0-21exp -^r-
g mole
"I (g) (miri) (atm)
        i  co
        1.58  x
 "H20
                 -16    lOexp
                                                         (122)
 This  rate  is exclusive  of the water gas shift reaction, CO +
 H2O * CO2
                         - 110 -

-------
The rates given in equations  (121) and  (122) are on the basis
of unit weight of carbon and  it is necessary to convert these
rates into rates based on a unit volume of bed for use in
equation (120).  There is no  exact way of doing this as there
is no experimental evidence given by von Fredersdorff and
Elliott as to the effective surface areas of the different
carbon types used.  For simplicity the weight to unit volume
of bed ratio will be taken as  (1 - <5)ps although the treat-
ment of the data in this manner may grossly misrepresent the
kinetics under fuel bed conditions.  The rate can then be
given by
x 60 x (1-6)
                               lb mole ft~3hr~1atm~1
                                                          (123)
                          = °-1 atm* PHoO =0.15 atm, 6 = 0.4
    ps =  0 lb     ft ~   for temperatures of 2500°F and 2000°
F the rates of the two reactions are found to be
Assuming pco ~ Pc02 = §H
and ps = 30 lb     ft ~3
       dc
         co
               T=2500°F  = 39 lb mole
       dc
         CO,
        dt

       dcH
               T=2000UF
                         2 2 lb mole hr  ft
                                                         (124)
        dt
               T=2500VF  =
                         1 13
                                lb mole hr xft
           o
        dt
               T=2000 F
                                        -
                         ^ 42 lb mole hr xft
Equation (124) indicates that the rates of both reactions
drop by more than an order of magnitude from the maximum ex-
pected rate at the end of the exothermic zone to the 2000°F
temperature level.  At temperatures a little below 2000°F
the reactions will be slow enough that they can be considered
quenched.  Moreover, the above calculations show that at tem-
peratures above 2000°F the steam-carbon reaction may well be
mass transfer controlled.

The rate at which the temperature of the gases falls can be
calculated using equations (120) and (124) .  As the steam-
carbon reaction is very fast at 2500.°F, it will be assumed
that it is mass transfer controlled and that the rate of
steam conversion is 40 lb mole hr^ft"3.  On this basis, a
rough estimate of the possible rate of change of temperature
is
                        -  Ill  -

-------
   AT     -v,  JlJLj[1^02_L-i° x ^6:5°°  £ 4500°F sec"1  (125)
   It max =  ~  ~3bl5D x CT2 x 0.4
This calculation has neglected the heat effects  associated
with the solid? but since generally UCpg < GC   , the  order of
magnitude of (AT/At)i:flax will not be           g  greatly
affected.  This result suggests that the temperature  drop in
the endothermic £1300G - 350°F/sec) , confirming that -the rates
of the eadothsrmie reactions at this point in the bed are
indeed quite high.


      Ignition anJ Combustion Stability of a Fuel Bed

Background.  The problem of ignition and combustion stability
of a fuel "bed can be sonvsniently divided into three  parts:
the ignition of the top layers of fuel in the virgin  bed by
radiation received from the hot combustion gases &ad  the re-
fractory lining of the furnace; the propagation  of the  igni-
tion wave down into the fuel? and the stability  of the  fully
ignited bed*.  The i'irst of these problems has been extensively
studied in t!ie literature on fire research and flame  spread.
Under most operating conditions incipient ignition will cause
little difficulty is an incinerator as long as the underfire
air rate at the front of the grate, where the fuel is ignited,
is kept low enough.  The reason for this is covered below.
Simras and Lav  i±±^'i have skoun that radiation densities of
the order of 1. - 1.7 cal &.a~* sec"1 (13 - 22 x 103 Btu  hr"1
ft~^)  are sufficient -'cor the spontaneous ignition of  typical
woods (= I 1/2 OM in thickness) containing 20 to 60 percent
weight of moist-ore,  ita ignition criterion based solely on a
flux density saust be used with discretion, as other factors
(thermal properties of tho material and size, for example)
can markedly affect the amount of heat that must be trans-
ferred to the ;.;peciriisii to achieve ignition.  However, the
ignition of the types of material used by Simms  and Law would
probably be acre difficult than that of most objects  in refuse
because of their high moisture contents and thickness.   Radi-
ant flux densities under siost conditions within  an incinerator
would b£ in excess of 50,000 Btu hr"1 ft"2, showing"that in-
cipient ignition Is not likely to be a problem unless the con-
vective cooling of the fuel by the underfire air is excessive.
                        - 112 -

-------
The remainder of this section will be devoted to studying the
process of ignition through the bed and the stability of the
fully ignited bed.  The latter problem will be discussed first,
as it is the leas complicated of the two and more studies have
been addressed to it.  These studies have been concerned with
the stability of catalytic packed-bed reactors but the basic
principles developed can be expected to apply to a fuel bed; a
catalytic packed-bed reactor can be considered to be a highly
idealized refuse bed.  The literature in this field  (e.g.,
143, 144, 145, 146, 147 and references contained therein) has
shown that a multiplicity of steady states can be achieved in
a reactor.  Two different physical phenomena are apparently
responsible for this multiplicity of steady states.  First,
the fact that the heat must be conducted in the opposite di-
rection to the fluid flow gives rise to multiple steady states
for the reactor system, and second, because reaction takes
place on the surface of or inside a catalyst pellet  (fuel
element) each pellet may be in one of two steady states.  For
a fuel bed, at least two steady states are immediately recog-
nizable:  (a) the virgin fuel bed at ambient temperature, (b)
active combustion? 
-------
where it has been assumed that all the heat is generated
within the Solifi end that the rate of heat generation may be
controlled by both mass transfer and kinetic factors.  Equa-
tion (126) m&y £»e made dimeasionless by dividing through by
the maximum possible heat generation rate giving

                             T  "                       (127)
                          k m_ AHC
                           m °2
A simple heat balance ? on an element of the fuel bed con-
sidered as a tube with carbon walls for adiabatic steady-
state operation, gives
      T   - T  =   M- —-— I   IT  ,  - T^l                  (128)
      Xg   A0
                  L   '"°2j

where To is the entering air temperature, mQ2 the mass frac-
tion of oxygen in the entering air and Ta
-------
                                               2/3
                                        Tad
Fig. 31.  Sketch of Dimensionless Heat Generation and Heat
          Lost Curves Q     represents heat  lost curve  for
          no radiative losses.  Q,....   represents heat lost
          curve with radiative losses.  Points  (•) are
          stable points; points  ( 0); are unstable).
                     - 115 -

-------
(101) .   Inspection of Figure 31 shows that once a  coal  par-
ticle has started to bum it is inherently very stable  as  a
consequence of the Lewis number being so close to  unity (N^g
for an 03-N 2 mixture is about 1.15).  The stability of  the
system increases with the gas temperature (Tg); T~ will, of
course, increase as the gases pass through the bed.  As Tg
increases, point A moves downward and the temperature of the
upper stable intersection points moves towards Ta(j.  The re-
action can 'be quenched by increasing the gas-flow, which
shifts the heat,generation curves to the right.  The critical
gas flow rate (Gc) and temperature (Tc) for this condition to
occur can be found by using equation (130) and the fact that
the slopes of the two carves must be equal at this point/  the
latter condition is given by
       kmEexp(E/RTc)
                K

                          k Eexp(E/RT
                       r\j   in         v
                     2 =         2
                                 c
                             N
            2/3
            Le
                                        (131)
                                               ad
                                                   _  T
where the approximation can be made since Figure 31 indicates
that at Tc the chemical kinetics will be faster than the mass
transfer.  Squations (130) and (131) can be used to eliminate
km, giving the following approximate equation for Tc
fP  _ fp *T1  ^.
 c    g c
E(T
                      ,
                    a
                       - T  )
                   RN
                     Le
Tc -
ETad 
-------
          rp  — £	
           c   2R
                          4R  i 12AHC
E   ) I6C
                                  pg
                                                   U3I)
in which 12AHC/16C   rti   + T    was quoted to bti of the Order
       o           ^   2    9-
of 1720 K.  Spalding*9 analysis, although based on a
approach, uses much the same reasoning given above.  The author
feels that although Spalding reaches the same general conclu-
sion as above, he overstated the importance of posaibi*
quenching of the reaction by high gas flow rate* as » bMifie-
quence of a poor resumption used in deriving equation (133) ;
the value of T  calculated using this equation is too lov.
Employing a value of T  calculated from equation  (133) Spal-
ding suggested that the minimum diameter particle whioh could
burn in air was 0.256 cm; and as this phenomenon is ftever ob-
served experimentally he suggested that some of the cttettlcal
and physical constants used in deriving this figure iNM?4 pro-
bably wrong.  The result derived above, that TC i« of tft» order
of T ,, indicates that the decrease in size is not nearly so
important as Spaldiog'a analysis claims.
The treatment given here covers a very restricted CaSft where
no radiant heat los-ees are taken into account.  The affect of
considering radiant heat losses is shown in Figure 31 with
the heat loss curve- Qj°;8s.  For this case, PQVJt A 4* the same
as before but point B has moved up to B1 = NEe  j 1 * **Md/
h  (Tad- Tg) }  .  The slope of the line, instead Of bein^f
linear, is now proportional to T|.  Without resorting to working
out the mathematics, it can be readily deduced that beat losses
by radiation tend to make the fuel bed less stable art* th4 bed
could be quenched with a somewhat lower G than that required
in the absence of radiation heat losses.  Radiant h*4fc *nd side-
wall losses account for the quenching of fuel beds at
values of air rate; this behavior cannot be explail»*tf-
on the basis of cOnvective h£at losses.

In concluding this section, it will be shown that the Solid
temperature, in the case where convection heat losses dominate,
is handly affected by the degree of conversion, and i* ftOt
dependent on the gge flow rate G. Since the combustion ttte
is entirely controlled- by ma.ss transfer and assumiftg
                    - *g '
                            S
                      - 117 -

-------
                                                        (134)
                        Le
Eliminating Tg from equation (134) , using equation  (128)  gives

       Ts - T0 =  (Tad - V
which shows that (T§-TO) is hardly affected by the degree  of
conversion and is of the order  (Ta2.  However, the heat  of  re-
action AHCmay £>@ adjusted to compensate for endothermic
effects.   An assumption has to be made concerning the func-
tional dependence of the reaction term r§|f on distance  and
of  the oxygen consumption distance ZQ         on   the  im-
posed gas flow rate (G).

Vortmeyer assutaed that ZQ could be given by an equation  de-
veloped by Wicke (122)  for mass transfer controlled condi-
tions,
                        - 118 -

-------
            =Q.2 3
                                                        Tad
Fig. 32.  Variation of Gas Composition, Temperature and
          Reaction Rate in the Vicinity of Ignition Front.
          After Vortmeyer (119).
                      - 119 -

-------
            0
              *  A - 0.185
                                                      (137)
where A is the distance that the oxygen concentration has
dropped by e factor of e (called "Abklinglange" by Wicke).
The deveiopiJisat by Wicke for A follows  Mayer's   two-zone
theory discussed on pages 61-75 and results in an infinite
depth being required for complete oxygen consumption.  As
shown on pages 75-82, the three-zone theory is required to
predict the correct order of zo, but Vortmeyer (119) implic-
itly corrected for this by assuming ZQ = A.

To obtain a reaction term insensitive to changes in gas flow
rate and oxygen concentration .Vortmeyf retook an average heat
generation rate defined by (r^*)
function
                                 ave
                                            and defined the
 f(z)  =
          eff

          °2
              ave
                            ef f
                  Where
dz
                                 ave
= z  in which z
is an
average
value of the oxygen consumption length and  is  assumed to be
independent of G and mg    Introducing the  dimensionless
variables              *
              T  - T
           9 =
                   O
              Tign   T0
                     (T
                       ign
                 P + % >
                 *g    *s
                                                       (138)
                                 (T.
                                 v  ign
into equation

          •<22e
                        .  %
                        (C) = 0
                                                       (139)
The appropriate boundary conditions are,
                        - 120 -

-------
       9 = 0 at  ? =  -«»


                                                        (140)

          + 0 as C - C
where Co is defined as
         J
          03  eff
(rFV
v  O0  'ave
                                             (141)
This is an eigenvalue problem, but,. with  the  transformation
given above, C is now dependent on G  and  Dp as  well  as the
initial mass fraction of oxygen in the  air (m§2) .  Using the
relationships for K| proposed on pages  25-31,   the dependence
of t on G and Dp is found to be very  weak [Ca(DpG)       ]•
Assuming that   m   does not change,  and  that only saall vari-
ations in G and     Dp are considered,  for a given  f(C)
                                s  00
      •       •           II 4  a        2  clVG               / T A n \
      GC    +  UC    * X(mJ: )  /   	r^	_ T t             (142)
       Pg      ps        2 y     v ign    O'


                             E
Using the relationships  for Ks, equation  (142)  may be written
approximately as

         c                                       ~~
           pg
                             -1.55 (Pa6h

                             ^  A^R?
where A is a constant, dependent upon mn
                                       U2
            as a function of
A  plot  of    G +
                        - 121 ,-

-------
is shown in Figure 33 for the data of P. Nicholls  (24) on the
underfeed combustion and ignition of different sizes of high
temperature coke.  The expected linear behavior is quite
closely followed, but the effect of particle size is not fully
taken into account by the theory-  It should be noticed that
this method of plotting the data is not very, sensitive to
values of U but the deviations from linearity shewn for any
particle sise in Figure 33 can probably be attributed more to
experimental arror than to insensitivity of the plot.

Given the many assumptions that went into the development of
equation (143), the general agreement between theory and ex-
periment shown by Figure 33 is quite good; but a major short-
coming is the inability of the theory to predict the upper
and lower ignition extinction limits.  The theory suggests
that there is no upper limit to extinction, for it is possi-
£le to have the ignition aon^ travel in the same direction as
G (i.e., negative values of U) .  The only limit would be
reached when the ignition zone was blown out of the reactor
or fuel bedo  This type of behavior was noticed by Vortmeyer
(119)  in his sxperiraents.  This behavior has never been ob-
served in a fuel bed and points to an essential difference
between the two systems.

Vortmeyer's experiments were conducted with 0.4 cm diameter
activated charcoal particles burning in O2-N2 mixtures; the
oxygen concentration of the mixing was kept in the range 5% -
9% to minimize the amount of combustion that occurred.  The
particles were contained in a quartz tube which was surrounded
by a heating mantle to keep heat losses from the burning zone
small.  The ...particles ^are ignited in an underfeed mode.  The
low oxygen concentrations used in this study explain why it
was possible to observe the ignition front traveling with the
gas flow; at nigh gas velocities, there was enough fuel re-
maining in the "burnt™ bed to sustain ignition as the ignition
front reversed direction and traveled back up the bed.  This
behavior would not be expected to be observed in a fuel bed
because there would be very little fuel left behind the igni-
tion plane.

An attempt was ma
-------
U
U
       60O
       550r-
       500K
       450K
       400r-
       350r-
       300J—
       250 r-
       200h-
       150
       100
        50
             iots are  for '             /
             ^gs*»r           -1  -2  /
             I/'T^ \ .   * 4.4 ib hr ft  v'
        4O   80
                        Data  for High Temperature
                        Coke ; P  Nicholls  (24J
                         • Dp = 2"-2i"

                         £i Dp = 17" - 2 "
              I    ^  I
                             t60   200   240   280  32O
                       .1.55
                         0.45  ( 0.8
   Fig.  33. Comparison of Vortmeyer's  Q.19)  Ignition
              Theory with the  Experimentar~Data of P.
              Nicholls (24).

-------
 cannot be described,  as  it  can  in  laminar flame propagation
 theory, by a simple  function of temperature;  thus the varia-
 tion  of the rate with distance  through  the bed must also be
 considered.  This makes  the problem  rather intractable, but
 an  approximate  solution  can be  found; the development follows.

              dT
 Assuming that -T-* can be represented as  f
              dz
              -dm.
 D
 i?    *-»i      ^  0
«  	™_ ,  and that the only exothermic reaction
 is C + 02 •* CO2, then for steady-state conditions,

      d1
                                           H   =  0         (144)
                              'g ~"    "2
and
                   dm
                 * -£ - -*S"                          (i«>
where
                   m                    c*
      eff        As °2            ,      Ps
     r0o  = I—•	1	 r and a = ^-2.              (146)
             ~ + pKexp(-E/RT)            Pg


            T             0
Putting Q => T    , m   = mh  (1-C)  gives from equations  (144)
             max    U2     2
and (145)
                                   AHCm°  (I-,) ,
                                       °2
                               - G 	^	 > = 0     (147)
                                       max     I
and
                                                         (148)
                       - 124 -

-------
where

                 pKexp(-E/RTmaxe)
                                                          (149)
Using the boundary condition that
gives


    de
              -»• 0 as  6 -»•  1  and C -*- 1
(Ua + G)C   (6  -  1)  + GAH m~ (l-£)/T
         t>                 U«-*       IllcLA
	12	?_	
                 K!
                                                 (150)
                                                          (151)
Equation  (151)  cannot be integrated directly  as
But
d; _ dc;   dz
d9 ~ d^ '    ~
                         (1 - O
                                (e .
                                                          (152)
                                           max
                                                          •p
which can  be  integrated if a mean value  of  the product KgE is
taXen.  Putting,
                    1  -   C
                                                            (153)
and
           C  T
                                  (6 - 1)
                                                  (154)
               "O.
and taking  Kgk at,its maximum value, when 6=1 gives
            dw
            a?
                    max
                          w
                                                  (155)
                        - 125 -

-------
The solution to equation (155) ,  which passes through w
is given by
            w =
                               + 1
                                                              0,
                                                           (156)
 Using equation  {156)  in  equation (152)  gives,  upon integrating
 the resulting expression for dC/dO.from O to 1,  the following
 result for the  ignition  velocity (U)
        u =
            G j    (Ksk)max
a \ 2
i 6 c
? p
( 9
1 1
/ f R
-------
LJ
a:
O
0
    70
    60
    50
    40
    3O
    20
    10
          *
Particle  Size = l"- 1 y"
Fuel Type : High Temperature Coke
E = 54,000 Btu Ib mole'1 (93)
K .  4.13 x10ft ft  hr"1 (93)
km=Mayers  Empirical  Eq. (90)
A Hc= 520O Btu  Ib '  Oxygen
    "\   .^Mayers'  Theoretical Relationship (90)


        \ ^--Present  Theory Equation ( HI -156)
                                   Experimental  Observation
                                   of P Nicholls ( 24 )
            100   200    300   400   500  600
             UNDERFIRE  AIR RATE , G ( Ib hr ft )
 Fig. 34.   Comparison of Ignition Theories with  Exper-
            imental Data of P.  Nicholls  (24).
                     - 127  -

-------
implicitly corrected by the use of Mayers' empirical  relation-
ships; this forced the reaction zone length to  be of  the right
order of magnitude.  Using a standard mass transfer correla-
tion would, with this two-sone theory, broaden  the  reaction
zone to such ait extent that only very small ignition  veloci-
ties  (= 5 Ib hr"1£t~2} would be obtained.

Figure 34 shows that the present theory is not  much of  an
improvement ovar Mayers' results, but the trend in  the  correct
direction is encouraging.  In addition, all physical  and
chemical constant a, except for km, were chosen  from what are
known to be representative values; this is unlike Mayers'
theory which used, for exaaaple, a heat transfer coefficient
many times greater than physically reasonable.   The curve in
Figure 34 is therefore in no sense a best fit.   The present
theory predicts shifts in the ignition rates with changes in
particle diair.etez' iTi the correct direction but  of a magnitude
more of the ordar showa by Mayers' results than those ex-
perimentally oosarved.  The reason for this is  simply that
Mayers' empirical correlations had to be used to calculate,
km.  The theory also predicts a much sharper variation  of U
with G than experimentally observed, and there  is nothing in
the mathematical development which prevents U from  becoming
negative.  This is, of course, physically impossible  in a
fuel bed.

Both Vortmeye.;:8 s (119} theory and the present theory  fail to
predict for the fueTTbed (although, in fairness, Vortmeyer's
theory was not developed for this application)  the  upper and
lower limit to ignition.  The lower limit probably  depends
on the magnitude of the heat losses from the bed and  as such
will depend on the type of apparatus used in the experimental
study.  Such h'-itit losses are not taken into account by  either
of the theories presented here.  The failure to predict the
upper extinction point is, this author believes, a  result  of
the internal nature of the problem.  This has been  hinted at
in the discussion or4 pages 46-59 on Nicholls' (24)  work.   As
the cooling effect of the underfire air increases,  it becomes
more difficult to obtain the mean temperature level within
the particle ineeesjsry for ignition to occur.   Presumably,
for coke or coal particles,, ignition is quantized to  each
particle b^cau.-ja of tha large thermal sink of the particle
and"its good theraal conductivity.  Thus it is  postulated
that the localized ignition of one spot on the  particle is
an inherently unstable condition.  For ignition to  proceed,
the whole iayar of virgin fuel particles must ignite  and this
condition will only be reached when the whole layer reaches  a
mean temperature level high enough for this to  occur.   Once
the cooling effect of the underfire air reaches a level where
                        -  128  -

-------
this mean temperature cannot be reached, ignition stops.  The
ignition plane cannot now be driven backwards and the fuel
that has been ignited simply burns away and eventually dies
out.

For the case of refuse, similar trends would be expected even
though the characteristic dimension of the fuel elements will
vary greatly and thus the amount of heat that is required to
be transferred to each element for it to reach an ignition
temperature will also vary.  Those elements (i.e., paper,
cardboard, etc.) which ignite easily will tend to propagate
flame ahead of the main ignition plane.  It is likely that the
effect of the thermal sinks for both conductive and radiative
heat transfer surrounding these tongues of flame will cause
them to be damped out and the ignition front will in all like-
lihood propagate relatively evenly.  The effect of tumbling
the refuse (which has not been considered in this discussion
up to now, although it commonly occurs in practice)  will help
the ignition processes if the tumbling is mild enough to pre-
vent ignited material being plunged into an environment where
it can lose enough heat to colder materials to quench active
burning.  The tumbling action can be considered an effective
way of increasing the heat transfer in the bed to the cold
unignited parts from the active burning part.

     If, as suggested above, the ignition problem is an in-
ternal one, the simple theories presented here will not be
able to adequately predict ignition behavior.  A far more
complex model would have to be developed, taking into account
the thermal history of each layer of particles.  In a simplis-
tic way, some progress may be made by considering the unsteady-
state heat transfer into semi-infinite slabs of material with
suitable boundary conditions.
                       - 129  -

-------
          PROCESSES OCCURRING WITHIN A FUEL BED

                      Introduction

The conditions within the bed of a traveling grate have been
qualitatively determined for coal-firing furnaces  (123, 124,
94) .  The processes within a refuse bed are markedly more
complex than those in a coal or coke bed because of the higher
moisture and volatile content, as well as the greater hetero-
geneity, of the fuel? the various processes occurring at dif-
ferent positions on the traveling grate of an incinerator have
not been adequately described.  The purpose of this section is
to provide a description of these processes as a reference
frame for the discussion of the experimental results which will
be presented on pages 184-261. The discussion given here will
draw heavily on the results of the survey on pages 21-130.

The one area of refuse combustion not covered in the survey on
pages 21-130  is that concerning the role of drying and pyrolysis
of the fueli to date there has been no work in this field of
direct application to incinerator conditions.  This section
will therefore begin with a description of the effect of mois-
ture content on the tisKe required for ignition and on the rate
of combustion of the fuel particles.


     Effect of Moisture on Ignition and Combustion Times

Outline of Problem Considered.  The heat effects associated
with drying are greater than those associated with pyrolysis
and, therefore, for a high moisture content fuel such as ref-
use, the drying heat load will be assumed to be the only one
of importance..  The moisture content of a fuel particle can
obviously affect the time required for the particle to ignite,
and the drying of the particle may provide the limiting step
in the overall combustion process.  The drying behavior of a
particular particle will depend on many factors (amount of
moisture, both free and bound? micro- and macroscopic struc-
ture of the particle; its diathermicity; and the heating rate
itself).  None of these factors has been studied in enough
detail to permit anything but a crude picture to be established
for a "typical" refuse particle.  There will, naturally, be a
wide range of behavior in the drying characteristics of refuse
components;  only the two extremes of this range will be dis-
cussed here.  These limits correspond to the cases where (1)
the rate of heat transfer to the particle is much higher than
the rate of internal diffusion of water and (2)  the rate of
internal diffusion of water is high enough for the surface
temperature to remain in the vicinity of the vaporization
                         - 130 -

-------
temperature of water (212°F).

For the first case it is assumed that, at the high heating
rates associated with a fuel bed, the solid exhibits drying
characteristics that resemble the "falling rate period" in dry-
ing operations, and that a vaporization plane will retreat into
the material as it is heated.  This model, at least for wet
wood irradiated with intensities from 2.0 to 3.1 cal cm"2 sec"1
(2.7 to 4.2 x 104 Btu hr~1ft~2), has been observed to fit ex-
perimental results reasonably well  (125, 126).  The progress
of the drying front is controlled by the conduction of heat to
the vaporization plane, and other mechanisms of moisture trans-
port (diffusion of bound water through cell walls and the
migration of free water under capillary effects) are slow in
comparison to the rate of travel of the vaporization plane.
The effect of the intertracheid motion of the steam has been
shown by both Williams (125) and Garden (126) to have little
effect on the drying process.  This model of the drying be-
havior of the fuel particle reduces mathematically to the so-
called Stefan problem, which despite its conceptual simplicity
is mathematically complex because of the nonlinearity intro-
duced by the boundary condition at the vaporization plane.

For the second extreme of behavior, it is assumed that the in-
ternal diffusion of water is very rapid.  In this case, the
drying time can be calculated from the amount of heat that
must be transferred to the particle to supply all the heat for
drying under the assumption that the surface remains at 212°F.
This type of behavior might be exhibited by, for example,
oranges and watermelons.

Development of Drying Models.  For the first type of limiting
behavior, the ignition time may be calculated using the fol-
lowing mathematical description of the drying process.  The
particle is considered semi-infinite, and initially at the
vaporization temperature of the liquid (T^); all thermal prop-
erties of the material are uniform and constant.  The equations
describing the process of vaporization, assuming that the con-
vective flow of the steam has a negligible effect on the tem-
perature distribution within the solid, are then as follows.
Within the dry phase, the temperature is determined by

             2       1  9Ts
            v TS  - -a- air                               (160)
                   as

where a^ is the thermal diffusivity of the dry solid.  The
required boundary conditions are
                        - 131 -

-------
      (i)    T  = Tv  at x = X, the dry-wet interface
             S    £>
     (ii)    -ks
              *   e
              d   s
x , x  - ps AHv ar
                                                           (161)
                f"^ rn  1
    t • • • »     i ^"*   *«5 i         — V» / rn  ^ rn \
    (111)    'ks "dx" I  x = 0  ~ hs(Ts   T»J
              d dTs
or   (iv)   "ks -35F I  x = 0  = Q
where Tg is the temperature at the drying front; kg the thermal
conductivity of the dry solid; hs the gas-solid heat transfer
coefficient per unit area of solid surface; T^ the temperature
of the gases flowing past the solid (assumed constant); Q the
net heat flux per unit area of solid surface; and
	       / pw     \
AH  = AH  1 -4  - 1 I , where AH  is the latent heat of vaporiza-
  v     v \  d      /          v
          Us     '„
                   d      w
tion of water and ps and ps the dry and wet density of the
solid.  The boundary condition at the dry-wet interface is
obtained by a heat balance which equates the rate at which
heat is absorbed at the interphase by vaporization of the water
to the rate at which heat is transferred to the interface by
conduction.  The, boundary condition at the surface can be ex-
pressed in two forms,  by postulating either radiant heat flux
or convective heating of the solid.  Kreith and Romie  (127)
and Goodman (128)  have presented graphical results of the
solution to the above set of equations.  These are shown in
Figures 35 and 36.

A rough check on this model may be made by calculating ignition
times for different moisture contents and comparing them to
the data of Sirams and Law (117) .  Using Q = 0.5 cal cm"2 sec"1
(6^6 x 103 Btu hr'^-ft"2) ,  ~^ pg = 50 Ib ft"3, Cp  =0.5 Btu
Ib"1 °F~1, ks = 0,2 Btu hr""1ft~2 <>F-1^ -^3 g^    ^s ignition
temperature suitable for piloted ignition, (T°).   = 644°F
gives, from Figure 36, the time required for igASSion of wood
with a moisture content of 20% as 2.5 min.  The data of Sirams
and Law show that, for the above conditions, times of between
4-5 min were needed for the piloted ignition of European
                         - 132  -

-------
CO
UJ
               OOQC!   000)
                           OO
                            dh2 +
                          <*s M
                                              Jpa
                             (kds>2
ilk
T.O-V
                                                               o5
                                                                     01

               Fig. 35.  Surface  Temperature (T °)  and Depth of Penetration  (x) of  the Vapor-
                         ization  Plane  in a Semi-infinite Slab for Convective Heating  of
                         Surface  by  Gas  at
                         Results  of  Kreith and Romie (127) ,
                                                                      AH
                                                                 C   (T^ + Too)
                                                                  ps

-------
I

u>
            10.0 CT
        Ix*
       U
      O
       L.
       O
(SI
        if
       U
             0.1 -
            0.01
                O.01
                       0.1
1.0
1O.O
1OO.O
                                                   or
                                                              Q(t)
1000.0
              Fig.  36
                  Surface Temperature  (T  )  and Depth of Penetration (X)  of the
                  Vaporization Plane in Semi-infinite Slab for Constant Heat Flux
                  (Q) Surface Condition.   Results of Kreith and  Romie (127) and
                  Goodman  (128).

-------
whitewood.  For Q = 0.7 cal cm~2 sec"1  (9.4 x 103 Btu hr~1ft~ )
and a moisture content of 60%, the data of Simras and Law indi-
cate piloted ignition times of about 4.5-5 min.  Using Figure
36, an ignition time of about 3 min is obtained.  The agreement
is quite reasonable considering the approximate nature of the
thermal properties used.  The data of Simras and Law show that
the ignition time increased with decreasing heat flux at a rate
greater than that predicted by the theory presented here, which
indicates that the ignition time is proportional to Q"2.  The
explanation for this discrepancy probably lies in the increas-
ing importance of convective cooling of the surface of the
material at the lower radiant heat flux densities; this cooling
effect was not considered in the above theory.

The flux level in a fuel bed in the vicinity of the ignition
plane can be readily evaluated using


              ^ ~ ~ s dx i x = ignition front

       E
where Ks can be calculated using equation (13) .  With the fol-
lowing approximate values s  DD = 0.125 ft, ks * 0.2 Btu hr"1
ft'1 °F~1, 6 = 0.4, e = 1.0, T = 1800°F (2260°R) , K| is found
to be approximately =4.2 Btu hr  ft"1 °F~1.  The temperature
gradients in the vicinity of the ignition front were found in
this study (see pages 203-208} to be in the range 6-20 x 103 °F
ft"1, while in Nicholis" experiments temperature gradients in
the range 10-30 x 103 °F ft"1 were observed.  The flux level
in the vicinity of the ignition zone is therefore expected to
be of the order of 25-100 x 103 Btu hr^ft"2.  On this basis,
ignition times from a few seconds (for the high fluxes) to
about half a minute (for the low fluxes) can be expected within
a fuel bed even with particle moisture contents as high as 60%-
70% by weight.

The flux level within the bed will depend on the size of the
particles, the bed voidage and the temperature level of the
bed.  Following Rosseland8s (129) treatment of radiation as a
diffusion process

                       160T3 dT
                 q = -- jj- g-

                                                     — 8
where a is the Stefan Boltzraann constant (0.1713 x 10   Btu
         °R~4) and A the absorption coefficient  (I/A is equiv-
alent to a mean free path or mean beam length) .  Comparing
equations (162) and  (163) and using Schotte's  relationship,
equation (13) , for K|J gives A =   3   .  Thus  as the particle
                        - 135 -

-------
size  (D ) or the voidage  (6) increases, the mean  beam length
(1/X) will increase and the unignited fuel will be  exposed to
a greater heat flux as it will "see" more of the  hot  combus-
tion  zone.
                                  r>
The most important dependence of Ks will be on the  temperature,
but the results of this study (see pages 204-206)and  Nicholls1
(24)  data show that the peak temperature at the ignition  front
is not strongly affected either by changes in underfire air
rate  or by fairly wide variations of bed composition.  For fuels
which approximate refuse^ temperatures of the order of 1800°F
are typically observed near the ignition front (26, 27, 47)  and
this  finding has been substantiated by this study.  Evaluating
the thermal conductivity on the basis of a temperature of 1800°F
therefore appears reasonable for incinerator fuel bed conditions.

The next consideration is how the moisture content  of the fuel
hinders the combustion rate.  This may be estimated as follows.
Assuming a fixed surface temperature (T|) and a linear tempera-
ture  distribution in the particle (which will be valid for
high  moisture contents), the progress of the drying front can
be readily shown to be given by
Equation  (164) shows that the rate of propagation of the vapor-
ization plane decreases as t""^* , as expected, since the poten-
tial for heat transfer CdT/dx) decreases as t"*->.  Further as-
suming that the pyrolysis reactions occur instantaneously above
a critical temperature T§, the rate of generation of pyrolysis
products is
                                                           (165)
where Wp is the mass of pyrolysis products per unit volume of
fuel.  Bamford, Crank and Malan (130) indicated that a pyroly-
sis product generation rate greater than 1.84 Ib hr'^-ft"2 is
necessary for spontaneous combustion to occur.  For the con-
ditions in a fuel bed, where some flame is present, this
pyrolysis product generation rate should provide an upper
bound for the rate required to sustain combustion.  Inserting
                        - 136 -

-------
typical values into equation  (174) —  p^=  50  Ib  ft"3, k^~  0.2

Btu hr^ft"1 °F~1, T°=< 1800°F, TV* 212°F,  T°=  600°F, w  = 30
     -3  —         s   _i      s           s          P
Ib ft  , AHv= 600 Btu Ib    (for 60% moisture  content) —

gives R = — - - .  Using the Bamford, Crank and Malan criter-
       P   /t
ia for sustaining combustion, the rate of  pyrolysis product
generation would fall below the threshold  level  only after
a period of 0.2 hr.  It can be hypothesized,  therefore, that
drying may limit the pyrolysis rate to the extent that  it
will hinder active flaming.  Some further  elementary calcul-
ations show that the heat load associated  with the movement
of the drying wave will not act as a severe local heat  sink,
as the enthalpy required can be readily supplied by convective
transfer from the combustion gas and radiative heat transfer
from surrounding elements.

For elements whose drying characteristics  are  of the first
limiting type, the time required to dry a  particle of known
moisture content and size can be estimated following Tao
(131) .  The equations solved in this case  are  the same  as
equations  (160) and  (161) , except they are written for  spher-
ical geometry and for a constant surface temperature, so that
                       s
becomes

                2  3T    3 T    1  3T
and
                T° = a   at  r = r                       (168)
                 s                o
This model assumes that there are no reactions taking place
within the solid.  Wear the ignition zone the reaction of
oxygen on the surface of the particle will have a strong exo-
thermic effect while, when all the oxygen has been depleted,
the reactions C + C02 •*• 2CO and C + H2
-------
U,'
CO
                    or
                   U
,">
                           7


                           6


                           5
1
<  3
                               I   I   I
                                                      I  I   I
                                                             I   I  I
                                    Extrapolation Based  on Hnal
                                    Slope  Being  That  Predicted
                                                                   /  ~
                                   From  Unsteady - State Heat    /
                                                               //
                                                       Conduction in  a Sphere
Numerical  Solution

of Too (13D
                                                                   Straight  Line  Extrapolation"
                                                                   Based  on Last  Two Points
                                                                   Calculated byTao(131)    _j
                                                                            ll
                                                                I   I
                                                                   0.4
                                                            0.5
                                                                      0.6
                                                                         *w
                                                                           -1
                             Fig.  37.   Drying Times for Spherical Particles  of Different Moisture
                                        Content.  Numerical results of Tao (131)

-------
where TF is the final dyring time  (dimensionless) and y the
dimensionless moisture loading.  Tao's results for the rate
of propagation of the drying wave  appear to agree  (within  5%)
with the analytical solution to equations  (160)  and  (161)  for
short times, but there is no known analytical solution that
can be used to check his numerical values  of Tp\

As attempt was made in this study  to develop an  approximate
analytical method for calculating  drying times for spherical
particles.  The first method follow that of Shih and Chou  (132),
who indicated that their iterative integral technique for
spherical geometry produced results very similar to those  re-
ported by Tao  (131).  From these authors'  study  of the Shih and
Chou method, which was originally  proposed by Siegel and Sa-
vino  (133), it is concluded that,  although the method gave
results which agreed closely  (within 5% -  10%) with Tao's  cal-
culation for nearly all times in the range 0 < TD < T$, the
method failed at times close to T^.  It was found that, in
the limit as TD approached T?, the method  gave times, when the
drying front reached a certain radius„ which started to de-
crease as the drying front moved towards the center of th~e~ par-
ticle.  This is obviously impossible on a  physical basis.  Shih
and Chou made no comment about this behavior in  their paper.
Furthermore, this author also found that even after four it-
erations had been used  (about the  maximum  that could be ob-
tained without an enormous amount  of algebraic manipulation),
the solution gave a linear dependence of the drying time  (tp)
on the initial moisture content  (y).  A linear dependence  is
not shown by Tao's calculations and would  not be expected  on
physical grounds.  These authors suggest on the  basis of this
study, that the usefulness of this iterative integral technique,
at least in spherical geometry, is limited.

A second approximate integral technique suggested by Goodman
(128) was tried.  This technique has been  shown  to give good
results with the semi-infinite slab with various boundary  con-
ditions, but it has not, to these  authors' knowledge, been used
before in spherical geometry.  The spherical heat conduction
equation  (167? was transformed into the linear form using  6 =
Tr, on the suggestion of Goodman  (134) that the  method worked
best for the. Laplace equation.  The results showed that at
zero moisture content,, the method  gave the correct dimension-
less drying time of 1/6 (see Figure 37) but that, as the mois-
ture content increased, the drying time increased exponentially
and approached infinite at values  of the dimensionless heat
load greater than 5»  The only explanation for the failure of
this method that can be offered lies in the different form
of the boundary conditions at the  interface, which results
when the transformation 9 = Tr is  used, to that  form for the
semi-infinite slab given in equation (161).  On  the basis  of
                         •- 139 -

-------
this study and a brief literature review it appears that at
present there are no approximate analytical techniques suitable
for the Stefan problem in spherical geometry.

Tao's calculations therefore represent the only generalized
method of caluclating drying times for spherical particles.
Using his results, the drying time for a 3-in.-diameter
sphere of wood containing 60% moisture is found to be about
25-30 min.  The drying time scales as the radius squared
(r2) so that for a particle of 6 in. diameter     containing
60$ moisture the drying time would be greater than a typical
residence time for   traveling-grate incinerators, which is
about one hour.

On the basis of the foregoing discussion the following con-
clusions can be drawn for combustible elements which exhibit
the first type of drying behavior (i.e., where the rate of
heat transfer to the particle is much higher than the rate of
internal diffusion of water).  The presence of moisture in-
creases the time required for ignition over that required
for a dry element; however, at the flux levels typically
obtained within a fuel bed, the ignition delay is not expec-
ted to be more than about half a minute even for particles
containing 60-70% by weight of water.  From Tao's generalized
calculation of drying times, active drying and pyrolysis will,
for large particles, continue for extended periods after the
particles have been ignited.  Using equation  (165) and typical
values of the physical properties of wood, it may be hypo-
thesized that the drying rate may, with high moisture content
fuel elements, provide the rate-limiting step for combustion.
Finally, the heat load associated with the propagation of the
drying wave is not expected to act as a severe local heat
sink, and therefore the drying process will not affect the
combustion behavior of the drying particles' neighbors.

For the other case of limiting behavior, where the internal
diffusion of moisture is high, the calculation of an ignition
time is somewhat more complex because of the effect of the
local heat sink on the fuel bed surrounding the particle.
A simple calculation shows that, at the heat transfer rates
expected in a fuel bed, particles of up to a few inches  (say,
the size of an orange) would be expected to dry in times well
under one minute.  However, it is common experience that ma-
terials of this type can remain uncharred in a fuel bed for
much longer periods.            The explanation lies in the
strong local heat sink created by this type of particle because
of its low surface temperature.  It can easily be shown that
the strength of this heat sink can be expected to be far great-
er than any possible local heat generation  (except very close
to the ignition front), and it can therefore be concluded
                      - 140 -

-------
that these particles essentially quench all the surrounding
materials  (see pages 113-118 and the discussion of radiant
heat loss on ignition stability).  Fuel elements of this
type are dried mainly by the convective heating of the hot
gases flowing past them.  The drying time, based on the con-
vective heating of the particle with gases at 2000°F,' is about
10-15 min. for a particle the size of an orange.

The conclusion that, for refuse materials which exhibit this
type of drying behavior, the fuel bed is locally quenched is
a strong argument for effecting some tumbling action on the
grate so that these particles will be exposed to the flame and
refractory above the bed.  For this type of refuse particle,
the drying time plays the dominant role in the overall com-
bustion time, as the combustion of the fully dried material
(which will only be a small fraction of the original weight
of the particle) will be a very small fraction of the total
"combustion" time.
    Description of Fuel Bed Conditions on a Traveling Grate
Ignition of the Fuel Bed.  Traveling grates are used in many
incinerators, and therefore the ensuing discussion will focus
mainly on conditions in the fuel bed of such a system.  Des-
pite the heterogeneity of the bed and the different drying
and combustion characteristics of the various fuel elements,
it is convenient forpurposes of discussion to subdivide the
burning fuel bed on the traveling grate into well-defined
zones as shown in Figure 38.

A combustible element in a fuel bed may receive the thermal
energy required for drying, pyrolysis and ignition via a
number of different mechanisms.  In the overfeed bed the ther-
mal energy is supplied by radiation from the overfire region
(from both the hot combustion gases and the refractory walls)
by the convective heating of the combustion gases flowing up
through the fuel bed and by radiative heating from the com-
bustion zone of the fuel bed.  In the underfeed be<3 the major
portion of the thermal energy is supplied via radiation from
the combustion zone directly behind the ignition front, while
the remainder is supplied by conduction through the fuel  (97).
In the case of the traveling-grate stoker the fuel at the top
of the bed near the feed end of the grate is heated solely by
radiation from the overfire region.  Once the ignition plane
progresses down into the fuel bed, as shown in Figure l(c),
the thermal energy
                         - 141 -

-------
RAW
REFUSE-t—
             UNDERFEED
             — BURNING  -
        I       i
       CHANGE
        'OVER '
OVERFEED
               I
               I
•EVOLUTION  OF COMBUSTIBLE—H
           Drying
           Pyrolysis
           Ignition

          Zero Free  O2
                                ttt
                           Underfire Air
                                            I

                                  Discharge End
       Fig. 38.   Simplified Schematic of Processes Occurring in the Fuel
                 Bed on a Traveling Grate

-------
required is transferred to the virgin fuel in a manner anal-
ogous to the underfeed case.

As discussed on pages 112-113 and pages 118-129 , it is  im-
portant to keep the underfire air rate low at the entrance
to the traveling-grate stoker in order to insure that igni-
tion is achieved for the larger particles and for those  with
high moisture contents.  Subsequently, as the ignition plane
propagates through the bed, it is again necessary to supply
the right amount of air through the grate so as not to hin-
der the progress of the ignition wave.  Ideally the under-
fire air should be set at the value that gives the maximum
possible ignition rate.  Practically this is impossible, be-
cause not enough is known about .the processes of ignition
propagation  for this value to be determined on an a priori
basis.  Field tests would be the only method of selecting
the correct underfire air rate.  Qualitatively it can be ex-
pected that the ignition rate will be a function of under-
fire air preheat (24) and supply rate  (£4_) , particle size
(24_) , fuel type (23T and moisture content (26,27).  For  re-
fuse, Kaiser's  (iXT) tests on the Oceans ideTncTnerator  in-
dicated that the ignition rate varied from 0.3 ft/jnin for wet
material to 0.5 ft/min for average material.  The Ignition
rate would be expected to decrease as the moisture content of
the fuel increased since there will be a decrease in the to-
tal heat liberated within the bed that is available for  heat-
ing up fresh fuel  (see pages 118-129 ) .  Present theories  -
do not give an indication of how the underfire air rate  should
be varied to take into account the different fuel moisture
contents; it would seem reasonable to expect that lower  air
rates might be necessary with increasing fuel moisture con-
tents.  Finally, the hindrance of ignition by having too little
underfire air may be of little practical importance; since
in reality the underfire air is likely to be augmented by
air induced through the bed by temperature gradients between
the center of the furnace and the "cold" walls.

Combustion of the Fuel Bed.  After the ignition plane has
passed over a fuel particle, the course of combustion will be
dictated by the nature of the particle and its moisture  con-
tent.  Two types of limiting behavior, controlled by the dry-
ing characteristics of the fuel, were discussed on pages 131-
141.  When the internal rate of diffusion of moisture is low,
the surface temperature of the particle will increase slowly
until ignition is achieved.  Depending on the particle's size
and moisutre content, this may take place over a period  of
a few minutes after the main ignition front has passed by.
After the surface reaches an ignition temperature, the par-
ticle will continue to pyrolyze and burn until either all the
                       - 143 -

-------
oxygen surrounding the element is consumed or the rate of
generation of pyrolysis products falls below the level re-
quired to sustain ignition; the latter condition would be
achieved if the heat supplied to the fuel element was not
great enough to drive the vaporization plane at a rate suf-
ficient to supply the required amount of dry pyrolyzable
material.  This point was discussed on pages 131-141  .  The
surface of a fuel element, shortly after the ignition plane
has passed it, will begin to char; it will then be oxidized
by any available 0- or will react with C02 and H20, accord-
ing to the reactions C + CO- -»• 2CO and C + H2O •> H- + CO.
Only the very small particles will burn out completely at this
stage, and the larger ones will still be drying and pyroly-
zing when all oxygen in the underfire air is consumed.

For fuel elements in which the rate of diffusion of moisture
is rapid, the heat sink will cause, as discussed on pages
131-141 , local quenching of ignition and combustion; these
particles and those surrounding them will then remain in this
quenched state until all the moisture from the fuel element
has been removed and there is enough oxygen and heat supplied
to the element for it to ignite and burn.  If the residence
time in the furnace is not long enough, these elements will
be discharged incompletely combusted and may cause some of
the surrounding particles to be discharged in a similar state.

The important reactions occurring within the fuel bed, ex-
cluding pyrolysis reactions and the combustion of tarry py-
rolysis products, are given in Table 8 along with their heats
of reaction.  The equilibrium constants of these reactions
are plotted as a function of temperature in Figure 39 and their
relative reaction rates are compared in Table 9.  Table 8
shows that, except for the almost neutral water gas shift
reaction (vi) and the slow and thermodynamically unfavora-
ble reaction (vii)„ the only exothermic reactions taking place
within the bed will consist of the reaction of oxygen with
char and pyrolysis products.  The very high rate of reaction
of oxygen with both char and pyrolysis products, as indicated
by Table 9 suggests that all the oxygen in the underfire air
will be quickly consumed within a small zone just behind
the ignition front; this behavior was observed in the coal
bed studies of Nicholls (24) and Kreisinger et al. (23) —
see pages 9 - 14  and pages 18-21 .  The heat released with-
in this zone provides the only source of energy within the
fuel bed to dry and pyrolyze the fuel and to sustain ignition.

Active burning will take place in a zone                de-
termined by the rate of consumption of underfire oxygen in
both volatile and char combustion.  Burning rates in coal
beds were found by Nicholls (24) to be approximately propor-
tional to underfire air supply rate, unless the burning rate


                       - 144 -

-------
                            TABLE 8
    Common Fuel Bed Reactions and Their Heats of Reaction  (135)
Reaction
(i) CffU + °? •*• C05
(*' (g) (g)
(ii) C + ^0 + CO
(6) 2 2(g) (g)
l
(iii) CO , N + -p. •* CO
(g) 2 2(g) 2(g)
(iv) C, . + CO., •* 2 CO, v
(6) 2 (g)
(g)
(v) C, . + HO. . -* CO, . + II
(6) 2 (g) (g) 2
(vi) CO, . + HO. . ->• CO^ + H_
(g) 2 (g) 2(g) 2(g)
(vii) c/0. +2 H -*" CH
(6) 2(g) 4(g)
(viii) H +|o - no
(g) (g)
(ix) en +20,, •*• con + 2 ii o
42 22
(x) C II 0 + 2 On -»• 2 H 0_ N + 2 CO,
Standard Heat of
Reaction* (Btu Ib
mole"1 x 10~3)
-169.29
-47.55
-121.74
74.19
56.48
-17-71
-32.20
-104.04
-345.17
-358.16
              (g)        (g)
          (acetic acid)
                                   2(g)       2
(g)
* Negative sign indicates exothermicity.   3-qraph.itG assumed to have
zero heat of formation; various types of amorphous carbon arc ronorted
to have positive heats of formation ranging from 3.06 to 4.68 Rtu Ib
mole"1 (136)
                           - 145  -

-------
                    TEMPERATURE (° F)
   2600    2000    1600
                         1200      100O
3.0
Fin.   39.
                                               70
 4.0          5.0          6.0

RECIPROCAL  TEMPERATURE (
           Equilibrium Constants of Common Fuel Bed Reactions
           (92) .   (All compounds as gases; carbon as 8-
           graphite.)
                 - 146  -

-------
                            TABLE 9
           Relative Rates, of Common Fuel  Bed  Reactions
     Reaction
Approximate Relative Rate at 1800°F
and 0.1 Atm. Reactant Pressure
C + O     •*  CO
                  10"
C + HO   -+  CO + H
                   3-10
C + CO    •*  2 CO
C + 2 H   •*  CH
                  10
                                                     -3
                           - 147  -

-------
was restricted by the ignition rate  (see pages  46  -  59  )•
Roughly the same behavior can be expected in a  refuse bed
if the model of Niessen et a_l. (1!9)  is correct  (see  pages
106-112 )•  Beyond the point of oxygen depletion,  additional
release of volatiles will occur from the larger particles
and the char will be gasified by the C02 and H20 rising from
the burning zone.  Up to the point where the ignition wave
reaches the grate, the burning action will be of the unres-
tricted underfeed type.  After the ignition plane  has reached
the grate, combustion of the residual char will be limited
by the supply of oxygen; the CO- released on combustion in
this region will partially react with the remaining  char to
yield CO.

Burning rates may be estimated from  the results of Niessen
et al.  (19) or by considering that the reactions C + C02 •*
3
-------
               I	1	1
              CH4«0.012 at 5OO*F
          8OO 1OOO  12OO  MOO 16OO 18OO 2OOO 220O
           eoo 1000 1200  1400 ieoo woo 2000  2200
           80O XXX) 12OO  14OO 16OO 18OO  2OOO 22OO

                    TEMPERATURE CF )
Fig.  40.   Equilibrium Gas Compositions at Various Assumed
           Fuel Bed  Temperatures.   Different  fuels gasified
           with air  at 1 atm pressure:  (a)   pure cellulose,
           (b)  Kaiser's (49) approximate composition  (dry),
           (c)  Kaiser's (4"!T) approximate composition
           (15% moisture).
                -  149 -

-------
    12O -
ct
0:
U-
CE
u
o
z
z>
u
-J
O
CD
a
u
a.
a
LJ
1OO -
ui
;D
u
to
CO
     20
      O
       8OO
               1200        16OO        2CXX)
                       TEMPERATURE  (°F)
2400
        Fig,  41.   Variation of Burning Rates with Temperature for
                   Different Fuels.  1 - dry cellulose;  2  - Kaiser's
                   composite, dry ash sulphur and  nitrogen free;
                   3   Kaiser's composite,  15% moisture;  4 - Kai-
                   ser's composite.  30% moisture.
                         -  150  -

-------
the greater "saturation" of the underfire air with respect
to carbon and hydrogen when enough energy can be supplied
to the bed so that H_ and CO are the only products of com-
bustion.  Figure 41 shows an interesting characteristic of a
fuel such as refuse, which contains relatively high amounts
of oxygen and hydrogen.  As the moisture content of the fuel
is increased, more of the oxygen required to convert the fuel
carbon to CO can be supplied by the fuel and less is needed
from the underfire air.  Figure 41 shows that for Kaiser's
composite refuse (see pages 9-14  ) with 30% moisture,
         the gasification rate becomes infinite as the tem-
perature increases, so favoring only H~ and CO as products;
obviously, under these conditions, all the energy require-
ments would have to be supplied from an external source.

This qualitative description of the four major processes —
heating, drying, pyrolysis, and the gas reactions of carbon
and carbonaceous materials —- occuring within a fuel bed will
provide a framework for the discussion in subsequent sections
of this author's experimental work.
                       - 151  -

-------
             EXPERIMENTAL EQUIPMENT AND PROCEDURE


               Choice of Experimental Equipment
 Brief Outline of Equipment Selected.  The experimental equip-
 ment chosen for this study was an extension of the underfeed
 combustion pot equipment used by the  early workers [e.g.,
 Kreisinger et eO_.  (23)  and Nicholls (24) ]  in the field of
 solid fuel Bed combustion.  The simple combustion pot was mod-
 ified so that the  radiant heat flux condition which would be
 experienced by the top  of the fuel bed in  a traveling grate
 unit could be simulated under these batch  conditions and so
 that burning rates could be determined by  measuring the
 weight loss of the fuel bed as a function  of time.  In prin-
 ciple an instantaneous  fuel bed burning rate can be calcul-
 ated if it is possible  to close a complete material balance
 around the unit at any  particular time.  However, the cum-
 ulative errors associated with this approach are greater
 than those connected with a weighing  technique and the latter
 method is therefore preferable.

 The  modification adoped here follows  the approach used by
 Weintraub et al.  (26),  in which the apparatus is constructed
 in two sections, tHe overfire combustion region,  or top sec-
 tion,  and the fuel bed  section,  or bottom  section.  Overfire
 air  was supplied at a number of different  positions in the
 top  section,  while the  underfire air  was fed through a grate
 which  supported  the  fuel  bed in the bottom section.   The de-
 sign allowed  the top section to be isolated  from the  bottom
 section and  then preheated with gas burners.   When the tem-
 perature gradients  in the  top section  refractories had stab-
 ilized,  the  two  sections were joined  together and the fuel bed
 was  ignited by exposure to the  radiant heat  flux  from the hot
 brickwork of  the top section.   The  initial radiant heat flux
 density used  to  ignite  the fuel  was controlled by the final
 temperature level of  the refractories  in the  top  section at
 the  end  of the preheat  stage.   During  a run  the radiant heat
 tiux to  the top of the  fuel  bed  could  be roughly  controlled
 by adjusting  the overfire  air  flow  rate, which in turn changed
 the  temperature level of the  refractories.  The course of
 combustion was followed as a  function  of time by  measuring
 the change in weight of the bottom  section.


This type of apparatus provided a satisfactory model  of the
conditions in the fuel bed on a traveling grate,  in that time
                       - 152 -

-------
in this batch-type experiment was equivalent to position on
a traveling grate  (94, 137, 138). The major criterion that
needed to be satisfied for adequancy of this model was the
similarity of the fuels in the two systems.  This analogy has
had certain limitations which are discussed below.

Discussion on Limitations of Batch Systems.  In the moving-
grate system a continuous belt carries the burning bed in one
direction.  The bed is ignited at the feed end of the grate
by the absorption of radiant heat and the ignition zone grad-
ually penetrates the bed until it reaches the grate further
towards the discharge end of the grate.  Referring back to
Figure l(c), it will be recalled that H is the length qf the
fuel bed burning dn what Nicholls (24) called "the unrestric-
ted ignition underfeed principle," and 0 is the lengthjof the
fuel bed burning in a manner similar to the overfeed principle,
It should again be stressed that length 0 of the fuel bed is
burning only in a way similar to the overfeed principle.  The
burning rate in this section will increase roughly in propor-
tion to the quantity of underfire air, but there will be no
pyrolysis products issuing from the top of the bed, unlike
a continuous overfeed operation.  The actual process of com-
bustion can be further complicated since the bed is in many
cases agitated by a reciprocating grate.  This has the advan-
tageous effect of mixing and breaking up agglomerates in the
bed, but the back-mixing can also lead to increased losses
in unburnt material to the ash pit and in increased fly-ash
carry-over.

The development of methematical and experimental models of the
combustion processes within the fuel bed can proceed from the
stand-point of an external or internal observer.  From any
external position, the conditions on a grate would appear
to be    steady-state, resulting in a steady-state, two-dim-
ensional  (length and depth) problem.  The justification for
only considering two of the spatial directions in any math-
ematical representation lies in the fact that many traveling
grates are wider than 6 feet and therefore property variations
across the width of the grate can be ignored.  From any fixed
position on the grate the conditions in the fuel bed would
be changing with time, and, if one were to model a small
section of this fuel bed as it traveled across the grate at
the prevailing grate speed, an unsteady-state, one-dimensional
(depth) problem would result.  The property variations across
the width of the grate can also be ignored in this formulation
for the same reason as stated above.  In this model, the time
coordinate is proportional to distance traveled across the
grate, the proportionality constant being the grate speed.
These considerations indicate that a small batch incinerator
                      - 153 -

-------
with no property variations in either the length or breadth
 (or radial or angular) directions (i.e., one-dimensional in
depth) could be used to simulate a small section of the bed
as it travels across the length of the incinerator.  A cri-
ticism of this formulation is that the inclination of the
ignition zone in the traveling grate creates both horizontal
and vertical components for the ignition wave propagation
velocity as indicated in Figure 42.   The angle of inclination
of the ignition zone to the grate will be a function of both
the grate velocity, V , the underfeed ignition propagation
velocity, V , and the crossfeed ignition propagation velocity,
V .  The relationship being


                 tan 3 = VN/(VG - V )                   (168)


or

                 tan 6 = V../V- for V_ » V_             (169)
                          N  la      G     P

where 3 is the angle of inclination of the ignition front to
the grate.  Typical values of VN and Vp (4 in/min) and V
 (1 ft/rain) indicate that 3 will in general be a small angle
 (< 20 ) and suggest that under normal operating conditions
a small segment of the bed, taken perpendicular to the grate,
could be adequately treated as being one-dimensional in depth.

Further complications with this model formulation ensue, since,
during the course of conbustion, the resistance of the bed
varies and as a result, on a traveling grate, there will be
a tendency for greater underfire air flow through the portions
of the bed with low resistance  (i.e., near the discharge end).
This problem is partly circumvented in practice by dividing
the wind box into compartments and controlling the flow of
air to each compartment.  In the batch system this effect
can be simulated by varying the underfire air as a function
of time.  A drawback with the batch mode of operation is that,
as the fuel bed burns down, cold walls are constantly being
uncovered in the bottom section and thus the isotherms in this
type of bed cannot be expected to follow those in a traveling
grate.  This cooling effect can be minimized in the experimen-
tal model by carefully insulating the bottom section, or by
providing a movable grate which can be used to hold the ig-
nition front at the same relative position in the apparatus.
The latter expedient is difficult to accomplish experimentally
and all investigators using the combustion pot type of appa-
ratus have opted for the former approach.   A somewhat simpli-
fied analysis of the cooling effect expected when using care-
fully insulated side walls is given in Appendix B and shows
                      - 154 -

-------
                             IGNITION
                             FRONT
                       UNBURNT FUEL
                                                             BURNING FUEL
en
en
                                              UNDER FIRE AIR
                    Fig. 42.   Diagrammatic Representation of Direction of Ignition Wave
                              Propagation Relative  to Directions of Grate Travel  and
                              Underfire Air Flow.   V  = ignition velocity parallel to
                              grate; VN = ignition  velocity normal to grate; V =  result-
                              ing ignition velocity; V  = grate velocity;   =  angle of
                              ignition front to grate

-------
that the heat losses are small compared with the heat gener-
ated during combustion.

Up to this point the discussion has focused on the modeling
of the fuel bad itself,  but it is also .necessary to consider
the similarities and dissimilarities in the processes occur-
ring in the overfire region of the continuous and batch systems
and any advantages or disadvantages that the batch system
might have.  Generally,  the flue gas compositions in batch
experiments would not be expected to compare with those found
in continuous units, unless a period of "pseudo steady-state"
operation were achieved.  In fact, the gas composition from a
continuous unit would more closely agree with the gas compos-
ition found from a batch experiment if all the flue gases
from such an experiment were collected and mixed together.
Further difficulties are also apparent, since it would be
very unlikely that the degree of mixing of the overfire air and
combustion products would be alike in the two types of ex-
periments, as the geometry of the system plays an important
role in this process.  The gas temperatures in the batch sys-
tem would tend to be lower than those in a continuous system,
since the latter would be physically larger and heat losses
less signifleant.  Consequently, it is impossible to match
weight-averaged temperature histories of the gas in the dif-
ferent apparatus, and thus difficult to develop quantitative
theories which would be equally applicable to both.  However,
as serious as these shortcomings appear, the batch system pro-
vides a very convenient and compact way of studying the qual-
itative effects of overfire mixing and its effect on the burn-
out of soot particles, smoke and carbon monoxide.  Further,
in a batch system,, the changes in composition of the flue
gases may be studied during the different burning regimes of
the fuel bed, and effects detected which may well be dis-
guised by gross mixing in a continuous unit.

The above discussion has shown that many of the shortcomings
of this approach at simulating a traveling grate can be over-
come by careful design of the experiment.  The advantages of
using a batch system over a scaled-down traveling grate far
override the disadvantages for the following reasons:

     (a)   The distribution of air, both above and below the
          bed, can be closely controlled in a small experi-
          ment by the use of a high-pressure drop grid below
          the bed and by careful selection of jet size and
          air momentum above the bed.

     (b)   A unidimensional system is approached so that a
          fairly complete map of temperature and composition
          distribution may be obtained both within and above
                      - 156 -

-------
         the fuel bed.  The problem of obtaining gas and tem-
         perature profiles is greatly reduced.

    (c)  Instantaneous values of the rate of incineration may
         be obtained.  The measurement of a burning rate at a
         particular time corresponds to that at a given posi-
         tion on a gratei the start of a run corresponds to
         the discharge from the feed-hopper, and the end of a
         run to the point of discharge into the ash-quench
         tank.

         Description of Experimental Apparatus and
               Peripheral Support Equipment

Incinerator Hardware.  A brief description of the apparatus
is given in this sectione a fuller description with complete
engineering drawings being left to Appendix A.  The descrip-
tion given here covers all the modifications that have been
made to the equipment since it was first completed in May
1971.  Many of the experimental runs were not performed with
all the system modifications described below; the evolutionary
development will be fully covered in the section on Results
and Discussion (see pages 184-261).

A sketch of the apparatus indicating all the salient features
is shown in Figure 43.  Figures 44 and 45 are photographs of,
respectively, the apparatus with some of its peripheral equip-
ment and the interior of the fuel bed section.  The principle
of design for the apparatus closely followed that used by
Weintraub et al. (26), the top section (overfire region)
was      joined to~~the bottom section (fuel bed region) so
that the bottom section could be weighted independently.  A
flexible steel diaphragm, mounted in the "no flex" position
(which exerted the minimum drag on the bottom section), pro-
vided a seal between the two sections.  The top section was
suspended from the ceiling of the test cell, while the fuel
bed section was supported on a load cell which measured this
section's weight using the strain gauge principle.  As shown
in Figure 43, the fuel bed section was mounted on a base
plate which had a central shaft protruding from the under-
surface.  This shaft, made from special case hardened steel,
passed through two linear ball bushings and rested on the
load cell.

There were a number of possible arrangements for weighing the
fuel bed section using load cell techniques, but the above
method was considered the most satisfactory as it eliminated
the effect of eccentric loads and kept inherent load cell
errors to a minimum.  A more detailed discussion of this
                        - 157 -

-------
1  Levelmg  Jocks (or Grate  Support
2 Movable  Base
3 Hydraulic Jack
4 Support  Jack
5 Load Cell
G Load Cell Housing
7 Support  Shaft
8 Linear Ball  Bearing  Bushing
9 Spirit  Level
1O Track
11 Support  Brace s
12 Support  Jack  Locating  Rings
13 Probe Locating Hole
14 Protec live Fin
15 Thermocouple  DAS  Junction  Bo*
16 Gra te
17 Thermocouple  Probe
18 Flexible   Diaphragm  Seal
19 Heat  Shield
20 Movable Refractory Shield
21 Movable Shield Support Structure
22 Roller  Bearing Housing
23 Gas Burner
24 Gos Burner  Pilot
25 Eclipse  Zero  Governor
26 Main Gas Supply Line
27 Pilot Gas Supply Line
28 Air Ports  ( Capped )
29 Air Ports with Nozzle
3O Cooling Jacket
 31 Cooling Coils
 32 Stack
 33 Gos , Smoke  and Temperature  Probes

 Jsffiffl  Castable  Ceramic
 [JS53  Lightweight Agglomerate
 L^HJ  Insulating  fir*  Bnc*
          FigQ   43.     Sketch  of   Experimental  Incinerator.
                                          -  158   -

-------
cn
vo
                      Fig.   44.   Experimental  Incinerator,  Gas Analysis
                                  Train  and Peripheral  Equipment.

-------
Fig.   45.   Interior of Fuel Bed Section  with Thermocouple
                            Probes  Installed.
                                    This page is reproduced at the
                                    back of the report by a different
                                    reproduction method  to provide
                          — 160  -  better detail.

-------
point and details of the load cell are given in Appendix C.

When the apparatus was not in use, the dead load of the fuel
bed section was lifted off the load cell using the support
jacks.  Before using the weighing system the bottom section
was accurately leveled with the leveling jacks and the fuel
bed section was then lowered onto the load cell.

The two sections of the incinerator were separated by a slid-
ing refractory shield which was mounted just above the dia-
phragm seal.  The shield moved in a plane perpendicular to
the rest of the incinerator and was constructed in two
refractory-lined sections,,  One half had a cooling coil set
into its base while the other half had an 18-inch-diameter hole
cut through the middle of the refractory lining.  When the
shield was in the "off" or closed position, the section con-
taining the cooling coils sealed the top section from the bot-
tom section and thus permitted the top section to be preheated
without any excess heat lealiage to the bottom section.  In the
"on" or open position, the hole     in the shield aligned with
the top and bottom section combustion areas, thus connecting
the two sections and permitting the radiant heat flux from the
top section to ignite the fuel bed.

Overfire air was supplied to the top section via any combina-
tion of twelve nozzles at various vertical and radial posi-
tions.  The nozzle ports permitted the installation of a wide
range of nozzle configurations.  The fuel bed was built on a
grate supported by a hydraulic jack which was used to set the
grate at different vertical positions so that the burning
characteristics of beds of different depths could be studied.
The underfire air was supplied through the grate, the design of
which insured an even air distribution.       The pressure drop
across the grata was kept large in comparison to the pressure
drop across the fuel bed, thus minimizing the probability of
having the underfire air channel through the fuel bed.

The bottom section of the apparatus had three sets of nine
sample ports spaced at 120-degree intervals around its cir-
cumference.  The vertical arrangement of the twenty-seven
sample locations allowed measurements to be made at one-inch
intervals throughout the bed.  The maximum bed depth that could
be studied was thirty inches, equivalent to a fuel bed volume
of 4.2 ft^.  The thermocouples (Chromel-Alumel) and the gas
probes were designed so that they were mutually compatible with
the sample port fittings.  Three "heat loss" probes, each con-
sisting of three thermocouples at different radial positions,
were located at different vertical and angular positions with-
in the insulating walls of the fuel bed section.  The tempera-
ture histories of the three thermocouples in each probe gave
                        - 161 -

-------
an indication of the rate of heat leakage through the insul-
ating walls of the fule bed.  The top section had eight
thermocouples imbedded in the refractory lining near the sur-
face and three thermocouples on the outside steel shell.
These were used to monitor the temperature and temperature
gradients of the rafractory, both during the preheat stage
and during an experimental test.  Gas and smoke samples were
taken from the flue gases in the stack, and the temperature
of the flue gas was recorded using a shielded Chromel-Alumel
thermocouple.

For the purposes of calculating material balances, it was
essential to know all air flow rates into the equipment.
Since it was very difficult to seal the whole unit against
air leaks, particularly around the sliding refractory shield,
the leakage rate was determined by feeding a known flow rate
of helium into the overfire air.  Batch samples were taken
during a run from the stack gas probe and the helium concen-
tration in the sample subsequently measured using a Beckman
gas chromatograph.

The measurements taken during a typical experiment are dis-
cussed below.  Twelve thermocouples were used within the fuel
bed, ten located at the center, the other two near the edges
of the fuel bed.  The thermocouples were roughly evenly
distributed throughout the depth of the fuel bed.  Two gas
probes were used in the bed, one 20 inches and the other 12
inches above the grate.  These probes were used to analyze
for oxygen, carbon monoxide, carbon dioxide, nitrogen, meth-
ane and hydrogen.  All three fuel bed heat loss probes were
employed, as were all the thermocouples in the top section.
The flue gas temperature and its water, carbon dioxide, car-
bon monoxide and oxygen content were recorded, and the con-
centration of the helium leak tracer was measured.  The fuel
bed weight loss was followed by recording the load cell output,
and the overfire and underfire air flow rates were measured
at frequent intervals.  The ambient conditions of temperature,
pressure and relative humidity were noted both prior to and
immediately following the experiment.  Fuller details of the
measurements and the measurement techniques are given in sub-
sequent sections and appendices.  All of the measurements
made with equipment and instruments that gave response sig-
nals in a voltage form were recorded on punched paper tape
using a Hewlett-Packard 2014B Data Acquisition System.  The
data collection and analysis procedures are covered fully in
a later section.

Air Supply.  The overfire and underfire air, as well as the
air necessary for running the gas burners and pilots, was
                          - 169 -

-------
supplied by a Hoffman blower capable of delivering  470  cfm
at 2 psig.  A back-up air supply was available  from an  Inger-
soll-Rand 125 psig oil-free compressor capable  of delivering
up to 450 cfm.  The air  supply system is  shown schematically
in Figure 46.  The air from the Hoffman blower  and  the  125
psig compressor was fed to a distribution  plenum 18 inches
in diameter and 8 feet high, the air pressure from  the  com-
pressor having been previously reduced to  2 psig by a large
capacity pressure regulator.  The air from the  above blower
was introduced into the plenum,, a filter screen which helped
eliminate extraneous matter.  The air flow rate to  the  plenum
from the compressor was controlled by globe valve V]_, while
that from the flower was controlled pneumatically by valve
V2«  From the plenum, the air was distributed to the gas bur-
ners and the lines supplying the overfire  and underfire air.

For general operation, the air for the gas burners  and  their
pilots was taken from the plenum below the filter screen; it
was then fed to the two burners via two 2-inch  flexible hoses
and to the pilots via one 1-inch flexible  hose.  The two
globe valves, V3 and V4, provided an air shut-off for the gas
burners in addition to the butterfly valves on  the  gas  burners.
An air shut-off for the pilots was supplied at  the  pilots
with a gas cock.  In special situations, when it was impera-
tive to have detiled knowledge of the air  flow  rates to the
two burners, the air was fed through one of the lines avail-
able for supplying the overfire and underfire air,  and  the flow
rate was measured using an orifice plate.

Six lines were available for supplying the overfire and under-
fire air,the flow rates being measured with orifice plates.
The orifice plates were mounted between 300-lb  flanges  at the
ends of two lengths of nominal 2-inch-diameter  shcedule-40
mild steel pipe that were mounted horizontally  along one wall
of the test cell.  The pipes extended six  feet  upstream of the
orifices, with four feet of pipe downstream of  the  orifice
plates before the flow control valves, vs  to v10, as shown
in Figure 46.  Downstream of each flow control  valve there
was a short length of pipe followed by an  elbow and nipple.
Flexible hose was used as a connector between these nipples
and the overfire air nozzles and the grate.  At the far up-
stream end of each 6-foot length of pipe,  a 6-inch  section of
honeycomb was placed inside the mild steel pipe to  help
straighten and calm the air flow.  The pipes were connected
to suitable 2-inch mild steel nipples at different  vertical
and radial positions in the distribution plenum with flexible
hoses.

Flange taps, set in the 300-lb flanges, were employed for
measuring the pressure drop across the orifice  plates.  The
pressure drops (inches of ^0) were measured on a bank  of man-
ometers while the upstream static pressures were recorded in
mm of Hg on another manometer bank.  The temperature of the
                        - 163 -

-------
                                                                        6 Stove
 I
H
                Line 6 -C^b
Line  5-&h;


Line  4-C&J


Line
                Line 2-tJj
                  •2" Globe Valves
r -

-------
air inside the plenum was measured with a Chromel-Alumel ther-
mocouple.

Two orifice sizes were employed.  The smaller size  (0.8535-
inch diameter orifice) was used for measuring the overfire
air rates_^nd_^he underfire air rate when it was set below
100 Ib hr  ft  .  The larger size  (1.312-inch diameter ori-
fice) was used when the underfire air rate was set above this
value and for measuring the air supply to the two burners,
when this was necessary.  In the latter case, the one orifice
sufficed for both burners, the air supply to the burners be-
ing split downstream of the orifice plate.  The smaller ori-
fice gave a pressure differential of 24 inches of H-O for a
flow of 50 cfm of air measured at 14.70 psia and 60  °F, while
the larger orifice gave the same pressure differential for a
flow of 120 cfm of air measured under the same conditions.

The orifice plates were used as a primary measuring device and
therefore great care was taken in installing them in hydro-
dynamic conditions which met American Gas Association Standards
 (139).  A fuller description of the orifice plate installation
and of the procedures and computations undertaken to calcul-
ate the air flow rates is given in Appendix D.

Gas Burners and Gas Supply. The overfire combustion section
was preheated with the two Eclipse 838-36 PMP sealed tunnel
burners shown in Figure 43j they were located 90 degrees a-
part and just above the sliding refractory shield.  The bur-
ners consisted of a combustion block cemented to a flanged
cast-iron block holder, and were mounted to the outer shell of
the overfire section using the four corner mounting holes in
the block holder.  Each burner had its own gas-air mixing system
and was supplied with low pressure air from the distribution
plenum and city gas at atmospheric pressure from an Eclipse
Zero Governor.  The low pressure air was directed into a ven-
turi-shaped burner throat producing a pressure low enough to
draw in the gas from the Zero Governor.  The gas-air ratio could
be varied, using the adjustable gas orifice located in the mix-
ing tee.  Each burner was provided with a peep sight and a high-
pressure, manually ignited blast pilot.  The air for the pilots
was taken from the air distribution plenum.  The gas line to
the Zero Governor had an emergency "quick shut-off" gate valve
installed in it so that both burners could be rapidly shut down
in case of a flame-out.  A pressure tap was located at the air
entrance side of each burner to measure the air delivery pres-
sure.  The pressure was read on the manometer bank in inches
of H-O.  This allowed both burners to be run under balanced
conditions (i.e., the heat output from both were equal).  The
burners had a dual purpose in that they could also have been
used as after-burners when it was not possible to burn out the
smoke using overfire air alone.

Cooling Water Supply.  Cooling water was used for the top
                        -  165  -

-------
section cooling coils shown in Figure 43, the cooling coils
in the sliding refractory shield, the stack gas sample probe,
and the fuel bed gas sampling probes.  Each cooling circuit,
except for the fuel line probes, had a mercury pressure switch
installed in it to indicate when water was flowing through the
circuit and as a warning system to indicate a loss of water
main pressure.  Visual checks were used to verify water flow
through the fuel bed probes.  Provision was made for flow-
ing the water out of the top section cooling coils and the
stack gas probe after an experiment.  This prevented the lines
from rupturing during winter months.

Fuel Bed Grate and Support System.  The grate measured approx-
imately 3 Tnches~"deep~ and 17.5 inches on its outside diameter
at the top.  The side walls were conical in shape and narrowed
down the cross section of the grate to roughly 15 inches OD
at its base.  This method of construction minimized the chance
of binding between the grate and the fuel bed walls.

The inside of the grate was divided into two layers, each 4
inches deep.  The first layer acted as a distribution plenum
for the air that yras supplied through the base of the grate.
A baffle was placed over the entry port to the plenum to help
distribution.  The upper layer consisted of a 1/2 inch-thick
piece of high-temperature felt placed across the entire dia-
meter of the grate and carefully sealed at the edges, with the
remaining depth of the layer filled with small granite stones.
The stones prevented the felt, which supplied the major pres-
sure dropf from being blown out of the grate and were an ef-
fective packing for good air distribution.  Under typical
operating conditions., the pressure drop across the grate was
approximately 5 inches H^O, while that through the bed was
of the order of 0.1 inch H20.  The high pressure drop across
the grate relative to that across the fuel bed gave  (at lease
in the lower portions of the fuel bed) good air distribution
across the entire fuel bed.  The grate was mounted on a hy-
draulic jack-               An automobile  power-steering pump
utilizing a 3/4-horsepower,  1750 rpm electric motor was the
driving source.  A four-way valve controlled the direction
of the flow of the hydraulic fluid to the jack, allowing the
grate to be ei-cher raised or lowered.  A pump bypass regulated
the flow of fluid to the jack and was used to control the
speed at which the jack was moved.

Miscellaneous Equipment.  Two audio warning systems helped to
provide ease and safety in operating the equipment.  One in-
dicated when tha Data Acquisition System was about to run out
of paper tape while the other registered a low air pressure
in the large air distribution plenum.  All the thermocouples
located in the top section of the apparatus were wired in
parallel with a Honeywell Electronik 16 recorder and the Data

                      - 166  -

-------
Acquisition System.  The visual display of the recorder was
useful in controlling the refractory temperatures.

One of the channels through to the Data Acquisition System
was reserved as an "event indication" channel.  By connecting
this channel to different known voltage sources, the time and
duration of a particular event  (such as the  start of a run
or when the underfire air was set at its  final value) could
be accurately recorded.


                 Gas Analysis System

Previous Experimental Techniques.  Apart  from the notable
early work of Kreisinger et al7~(23) in 1916, Nicholls  (24)
in 1934, and the later work of Kolodtsev  (91) in 1945, there
have been relatively few studies reported HT the literature
of experimental measurements of gas concentrations within a
solid fuel bed.  The three  studies cited  dealt with fairly
homogeneous fuel beds of coke, coal or carbon particles of
well-defined sizes and attempted to measure  the oxygen, car-
bon dioxide, carbon monoxide and, in the  case of Nicholls
and Kreisinger et al., hydrogen, methane, tar and soot com-
position at different depths throughout the  bed.  The samples
were collected on a batch basis during an experimental run
by drawing samples under vacuum with water-cooled probes
from different positions in the fuel bed  into collection ves-
sels.  The samples were analyzed using wet chemistry tech-
niques.  The probes used were of large  (approximately 0.3 inch)
internal bore, which helped prevent problems of clogging with
tarry pyrolysis products.

Analysis of the data obtained from these  fuel beds indicates
that there was a fair degree of error associated with them,
a shortcoming which is understandable, considering the exper-
imental difficulties inherent in this type of study and the
pioneering aspects of these investigations.  For example, the
oxygen concentration fell off very rapidly through the fuel
bed and once the ignition plane was reached  it was difficult
to locate accurately the point  of oxygen  extinction; dif-
ficulties were also encountered in freezing  the CO + 1/2 02 -*•
C02 reaction in the probes.  Even in these relatively well
defined systems, the experimental difficulties associated
with obtaining good data were great.  The experimental chal-
lenge offered in the study  of a heterogeneous refuse bed
where the fuel is undergoing drying, pyrolysis and combus-
tion, and where relatively  large concentrations of tarry
                         -  167  -

-------
pyrolysis products may be expected, is enormous.  Only one
previous investigation (46) has been reported on gas compos-
itions with a simulated refuse bed, all other investigations
on refuse systems having been concerned exclusively with the
gas-phase combustion regime, where experimental techniques
have been developed.

Gas Sample Traip.  Figures 47 and 48 show, schematically, the
arrangement used" for gas sampling and analysis.  Gas samples
were taken from three positions in the apparatus; two water-
cooled probes were employed in the fuel bed section and one
water-cooled probe in the stack.  Figure 48(b) shows the me-
thod of construction of the water-cooled probes.  The gas sample
was drawn through the center tube while water passed down the
inside annulus formed by the other two tubes and out through
the outer annulus.  The plug at the top of the probe between
the center and outside tubes was made with silver solder.  The
fittings at the other end were made from 1/4-inch Sawgelok el-
bows .

The samples from the fuel bed were analyzed for oxygen, carbon
monoxide and carbon dioxide at frequent intervals throughout
the run with on-line instruments.  Batch samples were also
taken from the fuel bed frequently during the run and analyzed
afterwards using a gas chroinatograph for hydrogen, oxygen, ni-
trogen, carbon dioxide, carbon monoxide and methane.  The
samples from the water-cooled probe in the stack were also
analyzed with the on-line instruments at frequent intervals
for oxygen, carbon dioxide and carbon monoxide; in addition,
batch samples were taken regularly for the helium tracer
analysis (see Figure 47).  The sample from this probe, which
was made from a 1/4-inch stainless-steel tube, was split
into two portions.  One portion was cooled, the condensate re-
moved, and the dry gas tested with a Von Brank filtering re-
corder to obtain  a rough  estimate of the amount of particu-
late and smoke emissions.  At regular intervals throughout
a run, a known volume of the second portion was drawn through
a steam-traced, line to a U-tube trap placed in an acetone-Dry
Ice mixture where the water content was removed by freezing.
The U-tubes used for these measurements were made from 6 mm
Pyrex tubing and measured about six inches from the top of the
legs to the bottom of the U.  The volume of gas drawn through
the U-tube was measured with a wet test meter, and the amount
of condensate was found from the difference between the initial
and final weights of the U-tube.  The weighings were performed
only after the U-tiabes had equilibrated with the ambient con-
ditions in the balance room.  The details of the computational
procedure used to calculate the water content of the stack
gases from the above measurements are given in Appendix F.

The on-line instrument tain was set up to measure carbon
                        - 168 -

-------
H
cn
                              Cooling Water
                              CftttQLttt
                                                                           Solenoid  Valves  for Sock  Flushing
                                                                               .Sotenoid  Voives  tor  Controlling Ftaw to
                                                                                     Anolysrs  Trotn
                                                                                                                  Both
                                                                 Bleed
                                                                      r-
Chamber



_j
                                                                                                         To  Atmosphere
                                                                                              Vacuum Pump ( P-2)




Oxygen
Analyzer
( Paramagnetic)



\


Carbon
Monoxide
Analyzer
NDIR)






Carbon
Dioxide
Analyier
(NDIR)






                                                                                    CaSO4
                                                                                    Drying
                                                                                    Tube
                                                                                               Diaphragm
                                                                                               Pump (P-1)
                                                      •*• To  Atmosphere

                                               Gas  Sample  Bottle
On-Line  Instrument  Tram
Thermostatically  Controlled
at  100 *F
                               vent
                               Valve( V-5 )
                                             Fig.  47    Gas  Sampling  and  Analysis  Train

-------
                           To  Smoke  Meter
 i
H
O
 I
                                   1/2  P'pe
                                   1/4 Copper  Tuning
                                   Steam Inlet
                                   Rubber Tubing Wound
                                   With  Electrical Tape
                                   Rubber  Tubing
                                    Pyrex Tube  Packed
                                    With steel  Wool
                                      Acetone and
                                      Dry  Ice
                 / V-• To Atmosphere
                   'X
                       Wet  Test
               Water
VI: Flow  Control Valve
V? -- Pressure Beliel v
-------
monoxide  (0-50%) , carbon dioxide  (0-30%) using nondispersive
infrared analyzers, and oxygen  (0-21%), using a paramagnetic
analyzer.  The analyzer sections  of the oxygen, carbon mono-
xide andQcarbon dioxide units were situated  in a closed box
in a 100 F thermostated environment.  The temperature was •
held constant in order to minimize errors in the temperature-
compensating systems of the analytical instruments.

The batch samples were collected  by attaching sample bottles
to the downstream end of the instrument train on the oxygen
bypass line.  The samples taken from the fuel bed were passed
through a glass wool filter to remove the major portion of the
high boiling pyrolysis products,  and then passed on to two banks
of three-way solenoid valves.  The first bank was set up so
that the probes could be cleaned  at regular  intervals with
pulses of high pressure nitrogen.  When a button was depressed
on the operator's control box, one solenoid valve was activated;
the  valve would      shut off the line leading to the instru-
ment train and open up the probe  to the high pressure nitrogen
line.  Under most operating situations, this technique of back-
flushing the probe was adequate in keeping the probes clean
if performed at regular intervals.  On release of the button,
the solenoid valve would return to its normal open position.
After the first bank of solenoid  valves, the fuel bed probe
sample lines joined the stack probe sample line and the three
lines entered the second bank of  solenoid valves.  These valves
were situated as close as possible to the instruments and were
activated from the operator's box.  In the closed position,
the gas sample flow was directed  to a "dump chamber" connected
to a vacuum pump.  The dump chamber pressure was controlled by
adjusting a bleed valve leading into it.  In the open position,
the valve directed the flow through to the instrument train and
under operation conditions, therefore, only one valve was open
at a time.  As a safeguard against opening two valves simultan-
eously, a light next to the control switch for each valve was
activated whenever the valve was  opened.  From the opened sole-
noid valve, the gas stream passed first through a water trap
placed in a freezing ice-water mixture and then through a small
diaphragm pump.  The pump pushed  the sample through a drying
tube filled with calcium sulphate and then through the instru-
ment train.  Two calcium sulphate drying tubes were mounted in
parallel in the instrument train, permitting a tube with expen-
ded calcium sulphate to be removed from the circuit and refilled
without disrupting the flow to the instruments.  The bypass
to the oxygen meter was manually  controlled to give the re-
guired flow through the sensor.
                        -  171  -

-------
The pressure inside the instrument train  (generally around 5
to 6 inches of H?O) was recorded on a manometer.  The mano-
meter had a dual purpose:  to indicate blockage of a probe and
to act as a warning device when the batch samples were being
taken.  The latter purpose needs some explanation.  When the
sample bottles were attached to the end of the instrument train,
sloppy operation could result in a pressure surge through the
train.  All the instruments measured the absolute number of
molecules in a sample cell and pressure fluctuations could cause
spurious results.  The most critical instrument in this respect
 is the oxygen meter, where a surge in pressure in the bypass
line could send too much of the sample gas through the sensor,
causing erratic swinging of the delicately suspended magnetic
dumbbells.

The length of line between the solenoid valve and the outlet
of the last instrument was kept as short as possible to mini-
mize the "dead, volume" that had to be swept out every time a
different probe was switched through to the instrument train.
Operating experience showed that it usually took about 45 se-
conds to flush out this "dead volume" adequately.
                       Data Collection
Almost all of the data taken during a run was  collected on
punched paper tape with a Data Acquisition System, the only
data not being recorded this way being the stack gas water con-
centrations and, of course, the gas compositions determined by
gas chromatography.  The Data Acquisition System  (Hewlett-
Packard System model 2014B) had lyV resolution and an input
impedance of 10 megohms.  Eight readings could be recorded
per second and the system provided five voltage ranges from
100 yV through    1000V with five significant figures recorded
on each range.  The paper tape was punched in standard IBM
8 level code.  Some data, for which a visual display was
required for control purposes, were collected on a Honeywell
Electronik 16 multichannel recorder.  These data consisted
of the top section temperatures, stack gas temperatures, in-
let air temperature in the distribution plenum, and inlet and
outlet cooling water temperatures.

Figure 49 shows, schematically, the connections from the various
recording devices         to the Data Acquisition System ;
Appendix E contains a table that shows the order in which the
instruments were connected to the Data Acquisition System
and the multichannel recorder.  The switchboard shown in the
                      - 172 -

-------
u>
i
  DATA
  ACQUISITION
  SYSTEM
COOLING WATER,
REFRACTORY,
AND STACK
THERMOCOUPLES
 HONEYWELL
 RECORDER
 CONNECTION
 BOX
 HONEYWELL
 RECORDER
                                                          FUEL BED
                                                       THERMOCOUPLES [
                                   FUEL BED
                                   THERMOCOUPLES
                                   CONNECTION
                                   PANEL
»A a «^T -£-j «£a =&> «A •>?«> «t j ^> •£* "Aii =.?=
4- 4=V 4-4-S-4-444-44-
•f I- 4 4- 4= 4- -i
4» 4- 4J- 4- 4" 4- 4> -{5 -!-' 4- 4- 4>
                                           O 0/0 00000/000
                                           O6OOOO Oy6 O O O O
                                           O O O O O O €T O O O O O
                                           oooooooooooo
                    7
INSTRUMENTS
IN THE GAS
ANALYSIS
TRAIN



GAS ANALYSIS
CONTROL
BOX

                                                   f ACQUISITION
                                                          MASTER
                                                          SWITCH
                                                          BOARD
                                                        INDICATION
                                                        SIGNAL
                                                        SYSTEM
           Fir.  49.  Instrument to Data Acouisition System. Connections

-------
center of Figure 49 provided a useful device for grouping sets
of data together.  The top section, or upper switchboard,
contained sixty permanent connections to the Data Acquisition
System storage registers 140 through 199-  The bottom section,
or lower switchboard, held lengths of cable with appropriate
electrical connections which were wired to the various re-
cording devices.  This arrangement allowed the output from
any instrument or thermocouple to be connected to any of the
sixty storage locations available on the Data Acquisition System.
About forty separate thermocouple and instrument outputs were
monitored during any one run and measurements of these, outputs
were taken at ten-second intervals during the course of the
experiment.

The primary reason for taking readings so frequently was to
permit the recording of an adequate number of readings of
the gas compositions at the different probe locations.  As
indicated in the previous section, it was only possible to
analyze the oxygen, carbon monoxide and carbon dioxide composi-
tions at one probe at a time with the on-line instrument train,
and the operator, therefore, had to switch each probe through
to the instrument train in sequence.  After switching the sam-
ple flow from a new probe to the train there was a period of
approximately 45 seconds before a "pseudo steadystate" sample
concentration was achieved.  The pseudo steady-state condition
applied after the instrument train had been adequately flushed
off the gas from the previous probe; true steady-state condi-
tions were never achieved in this type of batch operation.
To check the pseudo steady-state condition it was necessary to
obtain at least three readings of the sample oxygen, carbon
monoxide and carbon dioxide concentrations before switching
onto the next probe.  With  the Data Acquisition System taking
readings every ten seconds, the complete oepration of record-
ing the sample composition from one probe required seventy to
ninety seconds.   This was about the maximum time permitted
if one were to map out a semicontineous plot of gas composi-
tions as a function of time throughout a run.

It was desirable to store the three sets of data (oxygen, car-
bon dioxide and  carbon monoxide compositions) obtained from
each gas probe in separate storage registers in the Data Ac-
quisition System.  A control box performed the task of direc-
ting the three outputs of the on-line instruments to three
sets of separate storage locations allocated.  Each probe was
assigned one set of three storage locations.  For exampl^, the
lower switchboard addresses 50, 51 and 52 held the outputs
from the oxygen, carbon dioxide and carbon monoxide analyzers
for samples taken from the probe located in the stack; addresses
53, 54 and 55 held the oxygen, carbon dioxide and carbon monoxide
                         - 174 -

-------
outputs for samples taken  from  fuel bed probe number one, et
cetera.  It was of course  still possible  to  store the  infor-
mation held in the lower switchboard  addresses  50 through 55
on any of the Data Acquisition  System channels, simply by
making the appropriate connections through    the upper  switch-
board.

The control box had one master  switch for each  probe.  When
the switch was in the "on" position,  the  signal from the on-
line gas analyzers was switched       to  the Data Acquisition
System; in the "off" position the switch  short-circuited the
signal to the Data Acquisition  System.  At any  particular time
during a run only the switch that was associated with  the probe
being sampled was in the "on" position.   This arrangement
made it very easy to tell  from  the data which probe was being
used during any time period .

One channel through to the Data Acquisition System was kept
as an "indication channel" and was used to indicate the start
of various events.  For example, when the sliding refractory
shield was opened at the start of an  experimental run, a
switch was closed„ completing an electrical circuit which sent
a four-volt signal through the  indication channel.  Voltages
of different magnitudes were used to  signify the beginning
of other operations.
     Sequence of Operations During an Experimental Run
The sequence of operations undertaken in preparing for, and
during, an actual test is outlined below.  Full details on
the operation of all the equipment are presented in Appendix
E.

Feedstock Preparation.  The feedstock was made up to the spec-
ifications required on moisture and inert content by combin-
ing the necessary weights of wet and air-dried wood and tin
cans.  The moisture content of the air-dried wood was typically
around 7 percent, while that of the wet wood was around 50 to
60 percent.  The individual constituents of the feedstock
were weighed and stored in separate containers, care being
taken to cover the wet blocks so that they did not lose any
moisture content during the period prior to the fuel bed being
charged.  A priori it was difficult to determine what the bulk
density of the feedstock would be.  Experience showed that
roughly 60 pounds of material were required for filling the
                      - 175 -

-------
fuel bed section when the maximum fuel bed depth of 30 inches
was used.  This3corresponded to a bulk density of approximately
14 to 15 Ib ft" .

Top Section Preheat.  After the feedstock had been weighed, the
overfire section was prepared for preheating.  The fuel bed
section was pulled out along its tracks from underneath the
top section; this facilitated working on both sections and
was essential for charging the fuel.  The four overfire jets
were then installed in their required positions.  For all the
runs, four overfire jets in the second level of ports were
used.  This  configuration proved very satisfactory and the
jets were therefore installed semipermanently in these pos-
itions.  With the jets installed, the sliding refractory
shield was pushed closed and the counterweight placed on the
hook of the draw wire used to pull open the refractory shield
at the start of the run.

The cooling water for the refractory shield, the top section
and the stack gas probe was then turned on.  The overfire and
underfire air line valves were shut off, as were the butterfly
valves to the burners and the pneumatic valve on the inlet
to the Hoffman blower; this procedure minimized the load on the
blower  motor during start-up.  The oil level in the two bear-
ings on the blower  were checked before the blower was
started and, once the impeller had reached its operating rpmf
the pneumatic valve was opened slightly.  The control valves
on the overfire air lines in use were opened    and the air
flow rates through the jets were set at the approximate values
to be used during the run.  The valves for one of the burners
and its pilot burner were opened slightly and the gas line
valves on both the main supply line and the pilot line were
opened.  The pilot burner was then ignited.  After the pilot
flame had stabilized, the main burner was ignited, the heat
output being kept low until the combustion block warmed up.
Once this stage was reached, the major work in the preheat
stage had been done and the top section was slowly heated
over a period-of four to five hours by increasing the heat
output of the one burner until the refractories had attained a
temperature of around 1000 F, and then by using the other burner
to help boost the temperature    to 2000 -2200 F.  The re-
fractories were maintained at this temperature for at least
half an hour before the run was started so that temperature
gradients within tile refractories would have stabilized.

Fuel Bed Section preparation.  While the top section was being
preheated,the fuel bed section was prepared.  The section
was first cleaned by removing any unburnt remains from the
previous run with a vacuum cleaner.  Concurrently, the condi-
tion of the fuel bed thermocouples was checked, badly corrod-
ed ones removed, new ones installed, and bent ones straightened.
                        17*;  -

-------
If possible, the thermocouples were pulled out  so  that  just
the tips were protuding  into  the  fuel bed; this  facilitated
charging the feedstock.  Sometimes extensive  carburi-
zation of the sheath over the thermocouple did  not permit  the
thermocouple to be pulled out.  In this case, the  thermocouple
was left in place and care was taken in loading  the  fuel bed
not to bend it out of position.

After the thermocouples  had been  checked out, the  fuel  bed
gas probes were removed  and disconnected from the  water-cool-
ing lines and sample lines? they  were then thoroughly cleaned,
both inside and out.  The cleaning of the probe  sample  tubes
reduced the chance of blocking during a run and  was  per-
formed by drawing a thin wire back and forth  through each
probe and flushing it out, at frequent intervals,  with  ace-
tone.  The outside of the probe was easily cleaned with a
wire brush.  After cleaning,  the  probes were  reconnected to
their associated cooling-water and gas sample lines.

At this stage the grate  was set at the required  height  using
the hydraulic pump, and  a seal was made between  the  bottom of
the grate and the fuel bed section to prevent air  leakage
between the grate and the containing walls of the  fuel  bed.
The seal was made from suitable pieces of heavy-duty alumunum
foil held in place with  electrical tape.

Gas Analysis Train Preparation.   Following the  preparation of
the fuel bed section, the instruments in the  gas analysis
train were checked out and calibrated using the  manufacturer's
standard procedures.  In addition to following  the manufac-
turer's procedures, each on-line  instrument was  checked against
at least one other gas of known concentration;  this  provided
a useful double check on instrument operation.   THe  procedure
adopted for these checks was  as follows.  The tip  of one of the
fuel bed probes was pushed through one hole of  a two-hole
rubber stopper placed in the  mouth of a side-arm flask.  A
1/4-inch line from the requisite  gas source was  pushed  through
the other hole in the stopper and a hose from the  side  arm
dipped into a few inches of water.  The calibration  gas supply
was opened; the solenoid valve which directed the  flow  from
the probe being used to  the instrument train was opened; and
the gas sample pump turned on.  The gas supply  rate  was ad-
justed and balanced against the sampling rate so that just
a small gas flow was seen to  bubble through the  water trap.
The procedure insured that only the desire gas was sampled,
that a minimum of the gas sample  was used in  the process,  and
that thte pressure in the system  was as close as possible  to
that found under actual  operating conditions.

Miscellaneous Equipment  Preparation.  Once the  instrument  train
                         - 177 -

-------
was checked, the cold junctions for the thermocouples—five
cold junctions were used for the fuel bed thermocouples and
one for all the other thermocouples—were placed in Dewar
flasks filled with ice-water mixtures.  The final step before
the fuel bed section was charged with fuel was the preparation
of the stack gas water determination equipment and the connec-
tions to the Data Acquisition System.  For the water deter-
mination the dry U-tube traps were weighed (they were dried in
a vacuum over night) with rubber stoppers covering each arm
of the U's.  The weights were noted and the tubes numbered
and laid out in order next to the cold trap, which was a large
vacuum flask filled with a Dry Ice — acetone mixture.  The
wet test meter was then leveled and zeroes and the steam to
the sample line was turned on.

The connections for the Data Acquisition System were made
after the number of fuel bed thermocouples to be used had been
decided.  The connections were made using the switchboard
arrangement shewn in Figure 49 so that sets of data were con-
veniently grouped together on adjacent registers on the Data
Acquisition System.  The only guideline for this process was
that all the thermocouple registers had to be grouped together,
as the computer programs developed for analyzing the data
were written on this basis.  The switchboard set-up typically
used was as follows:  event indication channel, fuel bed ther-
mocouples arranged so that the thermocouple that was closest
to the top of the bed was first and the lowest one last, heat
loss probes, top section thermocouples, other thermocouple
outputs, load cell output and input, and finally gas analysis
outputs.  Once tha connections were made, the Data Acquisition
System clock was synchronized with the clock in the test cell
and the Data Acquisition System was set up to take readings
every ten seconds through all the registers in use.  The data
Acquisition System was then left in a stand-by mode of opera-
tion.

Loading the Feedstock. The first step in loading the fuel was
to obtain a reacting of the weight of the fuel bed section
without the fuel and with all thermocouple and gas probe con-
nections made.  To do this, the section was leveled with the
appropriate jacks and then the support jacks were slowly lo-
wered until the full weight of the fuel bed section was on the
load cell.

A two-volt signal was switched to the event indication channel
and nitrogen zero gas was passed through the oxygen meter.
When the zero of the oxygen meter had been adjusted and
stabilized, five to ten readings were taken with the Data
Acquisition System.,  The support jacks were raised until the
                         - 178 -

-------
the bottom section weight was completely off the load cell,
and the fuel was loaded into the fuel bed section.  The fuel
bed was loaded in a pseudo-random fashion with care taken to
distribute the wet and dry blocks of wood and tin cans evenly
throughout the depth of the fuel bed.  The various thermocouple
and gas probes were pulled into place as the fuel bed was built
up, care again taken to spread the fuel evenly across the fuel
bed, as otherwise large gaps tended to develop at the sides.
The fuel bed was charged so that the top of the bed was even
with the top fo the brick doughnut of the bottom section;
fuel was added or subtracted in order to do this and these
additions or subtractions were carefully weighted and taken
into account in determining the final weight and composition
of the fuel.

With all the fuel charged, the support jacks were again lowered
carefully and a five-volt signal was switched to the event
indication channel.  Air was passed through the oxygen meter
and the up-scale reading adjusted; five to ten more readings
were taken with the Data Acquisition System.  The weight dif-
ferential betwen the empty and full weights of the fuel bed
section was calculated from the load cell outputs recorded
by the Data Acquisition System and compared with the known
weight of fuel added.  The largest error between the calculated
and actual weight of the fuel was always less than one pound.

The support jacks were then raised and the bottom section
moved into place underneath the top section.  The bottom sec-
tion was carefully leveled with the leveling jacks and then
raised using these jacks until the flange on the fuel bed
section just mated with the flexible diaphragm; the mating
was indicated by locating probes positioned on the fuel bed
section flange (see Appendix A).  The section was raised by
rotating all three leveling jacks simultaneously one half-turn
at a time.  The close fit between the two mating doughnuts
of the top and bottom sections (see Appendix A) made it
imperative that the bottom section be raised in this manner,
as the doughnuts would have bound together if the bottom sec-
tion were not vertical.  Once this operation was performed
all electrical connections from the fuel bed thermocouples
to the control switchboard were made.  The support jacks were
lowered and the weight taken on the load cell.  At this point
nothing was allowed to touch or bump the bottom section, as
spurious readings from the load cell would have resulted.

At this juncture, another set of readings of all instrument
outputs was taken with the Data Acquisition System.  The weight
of the fuel bed was computed from the load cell output obtained
                         -  179  -

-------
from this set of readings and compared with the previous value.
This gave a good check that there was no binding between the
top and bottom sections.  This set of readings was also used
to check the outputs of all the thermocouples to make sure
that they were all operating correctly.  Once all systems
were operating properly, the run could be started.

The Start and_jgperation of a Run.  A run was commenced by
switching the solenoid valves for the gas analysis train so
that a gas sample stream was taken from the stack gas probe.
The Data Acquisition System was then switched on to a mode
where it took readings through the total set of inputs every
10 seconds.  The underfire air flow was turned on slightly
(about 1/10 its final value) and the gas and air supply to the
gas burners and pilots was shut off.  Immediately after the
burners were shut off, the sliding refractory shield was
gently pulled open using the draw wire.  This latter opera-
tion needed to be done carefully as the shield could cause
considerable vibrations (with subsequent load cell errors) if
it hit the restraining stops too hard.  The shield was pulled
out far enough to activate a switch which sent a six-volt
signal into the event indication channel and activated a light
on the instrument panel.  This signal indicated the start of
the run.

Shortly after the run had begun, stack gas water concentration
measurements were begun and the underfire air rate was slowly
raised to its final value.  About three to four minutes were
taken to raise the underfire air to the desired level, at which
time a twelve-volt signal was sent to   the event indication
channel.  Readings were taken of the orifice-plate pressure
drops and upstream air pressure on all the overfire air lines
and on the underfire air line every time they were changed.
The times at which these changes occurred were noted on a
log sheet.

Samples were taken from the three gas probes in a cyclic
fashion.  However, samples were not taken from the fuel bed
until the ignition front had reached the first gas probe.  Up
to that time, samples were taken continuously from the stack
with only brief breaks to check conditions at the first fuel
bed probe.  Once the ignition front reached the first fuel
bed probe, as indicated by a dramatic increase in carbon dio-
xide level, a sampling sequence of gas probe /fuel bed probe
was followed as rapidly as this operation permitted.  Typi-
cally it took one and a half minutes to obtain a sample, as it
took 45 seconds to attain a :s^oady-state reading after switch-
ing over the probes and at j&tst^ three readings were taken
before making the next switch.  Every so often, the second fuel
                      — i <
                        JL i

-------
bed probe was tested and once active burning was detected
at this level a sampling sequence of stack/fuel bed probe
one/stack/fuel bed probe two/stack was followed.  Near the
end of the run, when the carbon monoxide concentration in the
fuel bed began to decay and the oxygen content started to
increase, samples were again taken solely from the stack.
During these sampling operations batch samples were taken
from all the probes at those intervals throughout the run
when the probes were switched to the gas analysis train.  The
probe location and the time at which the sample was taken were
noted on the sample bottles which were analyzed after th run
(hydrogen, oxygen, carbon monoxide, carbon dioxide, methane and
nitrogen for the stack samples).

Stack gas water determinations were performed at regular in-
tervals throughout the run, about every three to four minutes.
The procedure for this was as follows.  One of the U-tubes
was connected between the two lengths of rubber hose shown in
Figure 48(a) and then the vacuum flask filled with the Dry
Ice — acetone mixture was raised so that the top level of
the liquid was as close as possible to the rubber/glass joint.
After a period of a few seconds to let the glass and steel-
wool packing cool, the valve V]_ of Figure 48 (a) was opened
slightly  (the pump had been switched on and left running be-
fore the run was started) and a low flow rate of air was
drawn from the probe.  After 0.3 cubic feet of air had been
pulled through the U-tube  (three revolutions in the dial of
the wet test meter) valve v^ was rapidly closed and valve V2
quickly opened to equalize the pressure on both sides of the
U-tube.  This procedure prevented the partial vacuum created
on the pump side by the high pressure drop across the ice-
packed U-tube from allowing any unaccounted-for air to be
pulled through the U-tube.  Once the pressure had equalized,
the vacuum flask was lowered, the U-tube rapidly removed and
both ends stoppered.  The time of the sample, U-tube number
and exact volume of gas passed through the U-tube were noted,
as well as the time taken to collect the sample, generally
one to two minutes.  A new tube was mounted in position as
soon as possible and the process repeated.  Water determina-
tions were started as soon as the run had begun and were con-
tinued until the end of the run.

Other than working the gas analysis train, taking batch gas
samples, doing the water determination and changing the punch
paper tape on the Data Acquisition System, the only additional
operation necessary during a run was adjusting the overfire air
rates in an attempt to keep the top section refractory
                           - 181 -

-------
temperatures around 1400°-1500°F.  This method of operation
was necessary in this batch system if the net heat flux to
the top of the fuel bed was to be kept approximately constant.
In practice, it had been found that the best way to achieve
this goal was to keep the oxygen content in the stack gases
around 10 to 15 percent.  The large thermal sink effect of
the refractories and the slow recording speed of the multi-
point recorder meant that there was a large response time asso-
ciated with any control system based on the refractory tempera-
tures .

The runs was ended when the concentration of oxygen in the
stack gases reached around 20^5 percent.  At this stage the
Data Acquisition System was switched off, the support jacks
were raised until all weight was off the load cell and the
load cell power supply was shut off.  The Hoffman blower
was switched off, all city gas lines closed at the main valves,
the instrument train and control system switched .off and the
fuel bed probes removed„  All cooling water circuits were turned
off except those for the roof and for the stack gas probe.
The cooling water circuits to these sections were left on for
a period of at least four to six hours.
                      Data Processing
Three computer programs were used to process the raw data,
which had been stored on punched paper tape.  The first step
in the data processing was to transfer the Data Acquisition
System output from paper tape to magnetic tape.  The first
program then read the magnetic tape data and decoded the IBM
8 level code format to give an output in punched card form.
This output deck could be checked for any errors which might
have been caused by malfunctions of the high-speed punch on
the Data Acquisition System.  The deck also provided a con-
venient form for storing the raw data.  The card output from
the first progra, coupled with a few control cards, was used
as input for the second program, which performed the task of
converting the voltage readings into corresponding tempera-
tures, gas compositions, and weight loss using conversion
tables and calibration curves.  The card output from the second
program, which contained arrays of temperature, gas comppsition,
probe number,
                         - 182 -

-------
and weight loss at the corresponding elapsed time from the start
of a run formed the input data to the third program. This
program plotted the data as a function of elapsed time using
a Stromberg Carlson 4020 plotter.  Complete details of the
three programs are given in Appendix G and sample outputs from
the programs are included in Appendix H.

In addition to the programs for handling and processing the
raw data a program was developed for calculating instantan-
eous material balances around the unit.  This program used the
stack gas measurements, the overfire and underfire air flow
rates and the helium tracer flow rate as input data.  The
development of the equations used for this program and a list-
ing are given in Appendix F.
                        - 183 -

-------
                    RESULTS  AND DISCUSSION
                       Introduction
 Summary of  Experiments Performed.   The  experiments  that were
 conducted can  be  conveniently divided into  four  groups  of
 runs:   Runs 1-2,  Runs 3-10,  Runs  11-16,  Runs  17-18.   A  break-
 down of the fuel  type and composition used  in all the tests
 is  given in Table 10; the conditions used in  the experiments
 and the measurements taken are  summarized in  Table  11.   Runs
 1 and  2 were of a preliminary nature, designed to check the
 overall operation of the test incinerator.  The  fuel  bed grate
 used in these  tests consisted of  an 18-in.-diameter piece  of
 sheet  metal drilled with 1/8-in.-diameter holes  on  3/16in.
 centers.  No effort was made to control  the underfire air, the
 air supply  being  set by the  prevailing natural draft.   The
 results from these two experiments  showed that the overall
 operation of the  equipment was  satisfactory,  but that in order
 to  obtain the  detailed information  desired, a number  of changes
 would  have  to  be  made.  These improvements included
 removing intermittent short  circuits in  the fuel bed  thermo-
 couple electrical system and improving the method of  joining
 the two sections  of the apparatus in the interest  of minimi-
 zing drag on the  weighing system.

 For Runs  3  through 10, an improved  grate was  added to the  sys-
 tem,  which     permitted a controlled amount of underfire
 air to be introduced through the  fuel bed.  In these  experi-
 ments,  the  fuel composition was maintained approximately con-
 stant  (17-15%  inerts, 28-33% moisture, and the remainder com-
 bustible) and  the  underfire air was varied from  127 to  275
 Ib  hr  Ift~Z.

 In  most of  this second group of experiments,  difficulty was
 encountered  in igniting the fuel bed properly, although for
 the  majority of experiments the underfire air  was slowly in-
 creased to  its final level over a period of a  few minutes
 after  the start of the run.   Under  these circumstances,  the
 bed  initially burnt vigorously but  then slowly died down; this
 resulted  in a severe decrease in the temperature level  of the
 refractory  lining of the overfire section and  consequently a
decrease  in the radiant heat flux to the top  surface  of  the
bed.  The fuel bed temperatures showed that in the experiments
where ignition was not properly achieved, the  temperatures rose
above the ignition temperature in the first few  inches  (%5-7
in.) of the bed and then the ignition process  stagnated.  The
                      -  184  -

-------
                                                                                              TMU   10

                                                                               Synthetic fuel Composition and Analysis
 I
H
00
tn
 I
Run
No.
1

2
3
4
5
6

7

8

9
10
11
12
13
14

15
16
17
18
Hood Analysis
ultiaate Analysis
(Percent by Height
-Bono Dry Hood)
tH


tc


to


NOT ANALYZED





6.08










5.96








52.24










51.30








41.68










42.74



As-Received
Moisture
Content
of Hood


"*8





7.96










6.90



Sites of Fuel Constituents
(Dimensions in Inches)
Hood
2>ixm
[] 3/8




2Hr.m
X3/4
2>sx.m
i! 3/8
1 3/4
ilx3/4



-}!„•> 1
*Vti%
xl",
ISxlS
xu




Psper1
_

-
-


5»iK2

6:;3

4!1H6






S
ul
z>
R



Metal
Cans
h-height;
-

-
-


h=4,
d»3 5/8
h»3.
d«2 3/8
n-5.
d-4





h«3 1/8,
d«2 3/4




Glass
Jars3
dKi2i6fteter
_

-
-


h=l 7/B,
d-2 1/8

h»4,
d-2(i






3
Ul
3
R



Conposition of Fuel
(Percent by Height)
Moisture
~-B

~17
-
27.0
33.0
29.9

27.8

31.2

30.7
31.8
24.0
25.9
23.9
14.7

26.4
24.9
25.4
25.1
Coffibuotible
Hood
(Bone Dry)
"92

•^83
-
48.4
41.9
46.5

47.9

44.9

46.0
47.6
60. 5
61.3
62.3
70.2

73.6
58.9
59.4
59.9
Paper
-

-
-
8.8
7.3
8.4

8.7

8.5

8.3
3.6
-
-
-
.

-
-
-
-
Inert
-

-
-
15.8
17.8
15.2

15.6

15.4

15.0
17.1
15.5
12.8
13.8
15.1

0.0
16.2
15.2
15.0
Ultimate Analysis2
of Puel
(Percent
bv Height)
tc
-

-
-
29.2
25.2
28.0

28.9

27.2

27.7
26.4
31.0
31.5
32.0
36.0

37.8
30.2
30.5
30.7
tH
-

-
-
6.5
6.6
6.7

6.5

6.7

6.7
6.6
6.3
6.5
6.4
S.8

7.3
6.6
«.3
6.4
to
-

-
-
4S.5
50.4
50.1

49.0

50.7

50.6
49.9
47.2
49.2
47.8
43.1

54.9
47.0
48.0
47.9
                                                         1   Random configurations -- cut up, crumpled and twisted

                                                         2   Assuming  analysis  for paper was the same as for cellulose, which from Kaiser's data  ( 49)
                                                            was  a  good  approximation of the average composition of the mixture of papers used

                                                         3   When glass  was used the ine^c was made up of approximately 55% glass and 45% tin cans.

-------
             TABLE  11




Summary of Experimental Conditions5
Run
No.
b,P
la, r
,b,p
^q-r
a^ff ,P
q.r
4b,f
r,o
5b,f
r,q
6b,f
r,q
?b,f,r
a,s
8b'£
r,o
Fuel Analysis
(% by weight)
Combus-
tible
-92
-83
-75
57.2
49.2
54.9
56.6
53.4
Inert
-
-
-
15.8
17.8
15.2
15.6
15.4
Mois-
ture
-8
-17
-25
27.0
33.0
29.9
27.8
31.2
Actual
Amount
of Fuel
Charged
(Ib)


60.0
57.0
68.3
59.4
57.6
58.6
Amount
of Fuel
Charged ,
from Load
Cell Read-
ino (Ib)
Determined
o
53.0
69.9
X
58.0
61.5
Dnderfire
Air Ratedb
hr'lff2)
at (elap-
sed time,
sec)
Natural
Draft
137C
146°
93(1600-
2200)
145(2200-
4000)
127(4000-)
X
146°
210(0-120)
130(3400-
5260)
150(5260-
6220)
210(6220-)
Burning
Rate(lb
hr-ift"2)
at (elap-
sed time,
sec)
68° (960)
57c,d
_JJL200J__
d
359(1000)
64 (1900)
62(3800)
X
289(2000)
67 (2750)
d
Ignition
Ratedb
hr-lff2)
at (elap-
sed time,
sec)
110C (960)
e
e
m
125h(3800)
X
79h(1200)
y
Max.
Temp.
Obser
ved
(°F)
e
e
e
m
19001
X
2300:
y
Comments
ixploratory experiments. Data Acquisi-
sition System malfunction in Run 2 pre-
i/ented detailed analysis of data.
Demonstration experiment.
Burning appeared normal.
Bed initially ignited, then smoldered
for first 2/3 of run. At 2600 sec burn-
ina began to improve. At 3600 sec act-
ive burning started and bed burnt rapid-
ly for 800 sec until all fuel was con-
sumed.
Bed ignited but could not sustain burn-
ing, despite varying underfire air and
reigniting gas burners. Bed smoldered
for 5 hr and almost complete combustion
achieved. No data taken after first
20-25 min of runT
Behavior of bed similar to that obser-
ved in Run 5. Data indicated probability
that bed had not been allowed to ignite.
properly before underfire air was
raised to final level.
Bed did not ignite properly using under-
fire air at 210 Ibhr-lft"2. Air rate lo-
wered and then increased slowly back to
210 Ibhr'^ft"2 over 5 min period. Fire
still did not improve and underfire air
rate again decreased. Bed eventually
•sfrarl-iart to hum.

-------
                                          Table 11 (Cont'd)
I

H-
00


I
*.•
r,q








10b,f,r
q,s

HT7kTT
q,t,s
„, ^ 	
i-p-.q
t, s

!—_____
^pf-







I4i,q
t,s
IS1'"
t,s
16l'Q-
t,s
1?l,u
v,w
IB1'"



54.3







	 ____
51.2


60.5


61,3


62.3







70,2

73.«

58.9
59.4

59.9



1
15:.0







•^^a^**™.
17.1


15.5


12.8


13.8







15,1

0.0

16.2
15.2

15.0



30.7







-onuau-^aun ._;
31,8


24.0


25.9


23.9







14,7

26.4

24.9
25.4

25.1



60.0







t— •--••— wuau
70.1


69.3


68.0


64.9







59.7

82.3

61.6
62.6

66.5


-i
58.4








69.7


70.8


69.5


65.6







60.5

81.7

63.1
62.7

67.2



275(180-
900)
88(900-
1020)
175(2400-
4200)
175(5000-)



275(120-
600)
175(600-)
155C


155C(600-)


240(0-900)
varying
(900-2400)
150(2400-
4200)
175(4200-)


160C

156C

118C
140C

86C



5(1250)
27(3200)
27(5200)







22(2200)
35(3600)

40C


n


n







n

n

n
39C

30(0-
1700)
41(1700-
3200)
64^(4600)








34e|256b'j


50C


y


y







48C

54C

41C ,
38°

43C



1800








22001


24001


2000


22001







24001

21001

2400i
23001

23501



Underfire air raised over 180 sec to
275 lbhr~1ft~2. Fire died down, under-
fire air was decreased, then fire caught
and underfire air increased back to 175
lbhr~1ft-2. Again fire died down, air was
lowered, and, when fire caught again, rais-
ed to 175 Ibhr-iff2 for rest of run.
Data only taken during later stages of
run when active burning was visually
observed.
At high initial underfire air ra'te.fir'e
began to go out; under fire air rate was
decreased aRd burning improved.
Gas probas plugged during test. Poor
electrical connection between oxygen
raster apd D . A . S .
Bed ignited but fire died down quickly.
Under fire air decreased and then, after
bed caught. was increased to final value.
Underfire air slowly raised to 2401b hf-1
ft" 2 over 15 min period. Fire died down.
Underfire air rate lowered and then, when
fire had caught, raised to 2101bhr~1ft~2.
Fire again died down. Underfire air rate
decreased and, when fire caught, was set
at constant level. Top of gas sampling
probe blew off during run twice .



»
Burning appeared normal .







-------
FOOTNOTES:
     (a) All bed depths studied w.re 30 in. deep (4.2 ft3).
     (b) Operating difficulties with Nondispersive Infrared CO meter and Sectarian 715 amperornetric oxygen sensor.
     (c) Rate constant over the major portion of test
     (d) Considerable scatter in data because diaphragm tightened too much.
     (e) Intermittent short circuits in the thermocouple electrical system gave poor fuel bed temperature data.
     (f) Runs 3 through 10 inclusive were performed with the low pressure drop grate.
     (g) Oxygen present at both fuel bed probe positions
     (h) Ignition rate increased with time.
     (i) Observed towards end of run
     (j) Runs 11 through 18 inclusive were performed with the high pressure drop grate.
     (k) From Run 11 through Run 18 two temperatures were recorded at the edge of the bed in addition to the  ten
         measurements taken in the center.
     (1) Paramagnetic oxygen analyzer used instead of amperometric sensor
     (m) Problem with fuel bed thermocouple electrical system gave spurious results.
     (n) Building vibrations and other difficulties with the load cell gave very poor weight loss data.
     (o) Binding between top and bottom sections caused spurious results from weighing system.
     (p) No fuel bed gas samples taken
     (q) No air leakage rate measurements taken
     (r) No measurements of moisture content in stack gas
     (s) Some batch samples analyzed for CO, C0_, 02/ CH., and N,
     (t) Measurement of moisture content in stack gas but readings all low because of poor water trap design
     (u) Complete analysis made for fuel bed batch gas sample - CO, C02/ H_, 02, CH4, N
     (w) Air leakage rate measurement satisfactory
     (x) Incomplete data precluded analysis.
     (y) Difficulties in igniting bed negated value of these measurements.

-------
maximum bed temperature noted under these conditions  (1000-
1400 F) was  very much less than the maximum observed when the
bed was burning properly  (1800 -2200°F).  When this behavior
was obseryed^the underfire air was decreased to around 60-
70 Ib hr  ft   and held at this level until the fire started
to burn vigorously again.  At this time the underfire air was
raised back up to the desired level.  On some occasions the
underfire air was pulsed  in an attempt to revive the fire,
and the refractory brickwork reheated with the gas burners
so that the heat flux to  the top of the bed would be increased
Runs 6,9 and 10).  None of these methods proved to be very
satisfactory, although they only failed once  (Run 6) to re-
vive the combustion process.

In this second group of experiments  (Runs 3-10) data collec-
tion was extended to include gas samples from the fuel bed,
as well as from the stack.  These samples were taken from two
positions in the bed and  were analyzed using the on-line
gas analysis train.  Some batch samples were taken from the
feul bed, and a limited analysis was performed on a gas chroma-
tograph for O?, CO, C0?f  N? and CH..  A considerable amount
of operational difficulty was encountered in this new proce-
dure; most of the difficulty concerned the poor operation and
response time of the Beckman 715 Amperometric oxygen Analy-
zer originally used for the oxygen measurements.  Additional
equipment problems were related to the electrical circuitry
of the non-dispersive infra-red CO analyzer.  Finally, the
distortion of the flexible diaphragm, where the top and bottom
portion   of the equipment were connected together, continued
to cause weighing errors„  The  data from this group of runs
are therefore rather sketchy, although they did provide some
insights into the ignition behavior of the bed.  These runs
further showed (1) that the underfire air tended to channel
through the fuel bed along the walls, (2) that the concept
of drying and pyrolysis waves propagating through the fuel
bed was unreasonable for  the size and type of fuel used in
the tests, and (3) that the closure of material balances
coupled with the measurement of the concentration of water
and hydrogen within tha fuel bed would provide a valuable
insight into the progress of drying and gasification.

For all experiments subsequent to Run 10, the glass content
of the fuel was omitted.  It was found that, when used, the
glass readily melted at the high temperatures of the bed,
coating the grate and insulating walls of the fuel bed. Great
difficulty was encountered in cleaning out the fuel bed, be-
cause the solidified glass clogged the holes in the grate and
covered the thermocouples.  It was felt that, although remov-
ing the glass meant eliminating an element characteristic
of refuse, the tin cans would suffice to simulate the inert
content.
                      - 189 -

-------
For the remainder of the experiments (Runs 11-18), a para-
magnetic oxygen analyzer replaced the amperometric sensor
and the gas analysis train was modified to decrease the time
required to take a sample and to make its operation easier.
The method previously used to join the two sections of the
apparatus together (i.e., clamping the two sections together)
was abandoned and the sections were joined by butting them
together and relying on the rubber gasket between them to ef-
fect a seal.  The rubber gaskets on the flange of the bottom
section were built up so as to minimize gaps between the flange
and the flexible diaphragm when two sections were butted to-
gether.  For calculation of material balances ,the stack-gas-
water content had to be measured.  The equipment necessary for
this measurement was built and installed for Run 11.  A des-
cription of this equipment is given (see pages 168-172 ).

Runs 11 through 16 attempted to fill the information gap sug-
gested by Runs 3 through 10.  For these runs (Runs 11-16),
both the underfire air and fuel composition were varied around
a base case.  For the base case  (Runs 11-12), the fuel compos-
ition was - 25% moisture, 15% inert, and 60%_combustible, and
the underfire air was fixed at 155-160 Ib hr  ft  .  In Runs
13_and_16 the underfire air was varied from 118 to 240 Ib
hr  ft   while the fuel composition was held constant; in
Runs 14 and 15 the fuel compositions were 0% inert, 25% mois-
ture, 75% combustible and 15% inert, 15% moisutre, 70% com-
bustible, respectively, and the underfire air was held to the
base case value of 155-160 Ib hr  ft" .

Problems were encountered in measuring the moisture content
in the stack (from which the moisture content in the bed was
calculated)  and with the weighing system.  The first problem
was attributable to faulty design of the measuring system used
and was readily corrected.  The latter difficulty was caused
by excessive vibrations set up by other operating equipment
elsewhere in the building and could not be cured; however,
by scheduling runs during relatively quiet times, the pro-
blem was largely, circumvented.

Starting with Run 11, the grate design was modified so that
the pressure drop across the grate was markedly higher than
the drop in Runs 3 through 10; the modification appeared to
decrease the amount of channeling that occurred through the
fuel bed.  Attempts to close material balances for Runs 11-16
showed that there was considerable air leakage into the test
incinerator, probably through the spaces around the sliding
refractory shield.  This air leakage did not affect the mea-
sured burning rates because this air did not pass through
the fuel bed.   For material balance purposes, however, the air

-------
leak had to be measured and a helium tracer method, described
on pages 168-172, was selected  for this purpose.

The improved operation of the gas analysis train made it pos-
sible to obtain more gas samples from the fuel bed; in addi-
tion a greater number of batch  samples were taken and analyzed
more completely than previously (i.e., samples were analyzed
for H2 in addition to CO, C02,  02, N2, and CE^).  Although
the additional measurement of hydrogen provided more detail
of fuel bed conditions, it was  not possible to check whether
the water-gas-shift reaction was equilibrated at the top of
the bed because of the poor stack-gas-water measurement and
the unknown quantity of air leakage into the equipment.  (The
moisture content of the gases at the top of the bed could have
been calculated from the known  underfire and the overfire air
flow rates and the stack gas moisture measurement.)

Before progressing with the final group of experiments (Runs
17 and 18), the apparatus, the  gas measuring equipment, and
the leak measuring technique were all thoroughly tested by
closing material balances around the unit while burning a known
amount of city gas.  For these  tests, the two sections of the
apparatus were joined together  in the manner used for a regu-
lar run, and overfire and underfire air was supplied at about
the ratios and rates used in a  typical test.  The combustion
air supplied to the gas burners was metered with a suitable
orifice plate.  The amount of city gas used was measured by
a dry test meter specially installed in the main supply line
to the Fuels Research Laboratory for this purpose.

Runs 17 and 18 provided the last set of results obtained
and were the most complete runs performed.  For both of these
runs, all equipment worked well and good closure of material
balances was achieved.  The material balance gave an inde-
pendent check on the load cell  results and provided insight
into the C/H and O/H ratios of  the fuel that was being con-
sumed.  Using the material balances and the measurements of
the concentration of hydrogen within the fuel bed made it pos-
sible to check the validity of  the equilibrium of the water-
gas-shift reaction at the top of the fuel bed.  The base case
fuel composition was used in both of these runs and the under-
fire air was varied from 140 Ib hr~lft   in Run 17 to 86
Ib hr~1ft~2 in Run 18.

Outline of Discussion of Results.  The ensuing discussion
will be based primarily on the  results obtained from Runs 5
through 18, as the results prior to Run 5 were of a prelimin-
ary nature and experimental conditions were ill-defined.  The
                       - 191 -

-------
 complete data  set  for one  experiment  (Run  17)  is  contained in
 Appendix H.  It was not considered  feasible  for easons  of
 space  to include the complete data  sets  for  all experiments;
 thus,  the  following discussion will be based on selected data
 from typical runs.  Rather than dealing  with the  results
 from each  of the experiments in sequence,  the  following sec-
 tions  will each be devoted to considering  a  different aspect
 of  the results (for example, Bed Temperatures, Material Bal-
 ances,  Fuel Bed Gas Compositions, etc.).
                   Fuel Bed Temperatures
Fuel Bed Temperature Distributions for Active Burning  Runs.
The temperature"at the center of the fuel bed was  measured
at ten different vertical positions spaced, on the average,
three inches apart.  For the later runs  (Runs 11 through 18)
two thermocouples were used to measure the bed temperatures
close to the side wall.  The measurements taken in Run 10 at
five positions  in the bottom half of the bed are shown in
Figure 50.  It  should be emphasized that these measurements
were made with  a thermocouple at the center of the bed, and
that they represent a weighted mean  of the temperature of
the gas that flowed over the couple and the temperatures of
the surfaces with which the couple was in radiative exchange.
The recorded temperature therefore depended on the drying
characteristics of the fuel surrounding the termocouple to
the extent that these characteristics affected the surface
temperature of  the drying fuel element.

Despite the above qualifications, the temperatures in most
runs showed, as in Run 10, a regular progression of the burn-
ing front from  the top surface of the bed to the grate.  In
a few experiments the temperatures were observed to plateau
at 212 F ang about 500 F.  Peak temperatures were typically
around 2200 F,,  These maximum temperatures give a measure of
the degree of melting, clinkering and ash fusion that could
be expected on a traveling grate.  The temperatures observed
here,  which appear typical for most refuse beds, are high enough
for all soft glass to melt (as was experimentally observed)
and were high enough to suggest that aluminum would also melt.
The temperatures were never high enough to melt the tin cans,
but the bright metal coating disappeared as a probable' conse-
quence of both vaporization and diffusion into the substrata;
    the steel was badly carburized.   Figure 50 shows, from the
temperature registered by thermocouple No. 5, that the ignition
                        192 -

-------
    240O
    20OO
    16OO
IL


O
LJ
Q


U
cr
o:
LJ
Q.

2
LJ
12OO —
 BOO
     40O  —
            3OOO
                           40OO
5000
                     ELAPSED  TIME, SECONDS
       Fig.  50.  Fuel Bed Temperatures (Run 10).  Thermocouples

                  1, 2, 3,4 and 5 were located 11,  9, 7, 2, 0

                  in. from the grate.   Initial fuel  bed depth 30

                  in.; underfire air rate 175 Ib hr~l ft~2; bulk

                  density of fuel 16.7 Ib 
-------
plane reached the grate 4400 sec after the start of the run.
The temperature recorded by this thermocouple can be considered
to be the grate temperature and as such gave a measure of the
degree to which the grate was heated.

Care has to be taken when interpreting the fuel bed thermo-
couple data because of the continuously changing nature of
the fuel bed,.,  In this regard, two problems were encountered.
First, at the ei\d of a few runs, some thermocouples were found
to be severely bent out of position.  The following tentative
explanation for this observation can be offered as no direct
evidence was available from the data.  The consumption of the
combustible resulted in a continuous change in the density
of fuel bedf and therefore the ash and inert content of the
fuel had to settle slowly towards the grate over the course
of an experiment.  The mechanics of the settling of the bed
would have depended on the difference between the ignition
and burning rates,.  When the ignition rate was not much grea-
ter than the burning rate of the combustible content of the
fuel, it can be hypothesized that the fuel was consumed in a
thin reaction zone which moved through the bed.  Under these
circumstances, the inert and ash content of the fuel would
have moved with the burning zone, and within this burning
zone the inert and ash content would have increased with
time.  The settling of the bed in this fashion would have
caused little drag on the fuel bed thermocouples.  When the
ignition rate was much faster than the burning rate, it can
be conjectured that most of the fuel was consumed after the
ignition front reached the grate; after this time, the major
portion  of the remaining fuel would have been consumed by
the oxygen in the underfire air in the space of a few in. a-
bove the grata,  Under these conditions, it can be expected
that the whole of ;he remaining bed would have slowly set-
tled towards the grate as the combustible was consumed; the
resulting drag on the fuel bed thermocouples under these con-
ditions would have been considerable.

An attempt was made to minimize the effect of the settling
fuel bed on the thermocouples by supporting the thermocouples
inside quartz tubing held in place by metal sheaths cemented
into the side wall of the fuel bed.  These efforts were largely
successful but there were occasions when the quartz tubing
broke and the thermocouple casing was rapidly carburized
and weakened,, allowing the thermocouple to be dragged down
by the settling bed.  (The rapid carburization of steel within
a fuel bed is one of the primary difficulties associated with
obtaining good fuel bed temperature data with thermocouples.
It was found that 304 Stainless was rapidly carburized at the
temperature i-avels of the bed; and of the more common high
nickel content alloys only Inconcl 600 appeared to have a
                       -  194  -

-------
a reasonable lifetime.)  A further difficulty raised by the
consumption of the fuel was the relative position of the ther-
mocouples within the bed as a function of time.  As the bed
burned down, the thermocouples were, since their position
with respect to the apparatus was fixed, eventually exposed
above the bed;  At this point, the thermocouples were no long-
er  recording bed temperatures but instead recorded a weighted
mean of the temperatures of the refractory lining, the over-
fire combustion gases, and the gases issuing from the top of
the bed.

The second problem assumed to have been encountered, but never
proved explicitly, concerned the shielding of the thermocou-
ples by the inert material (tin cans).  It was possible that
some of the thermocouples were covered by the inert material when
the bed was being built up prior to the start of the run; in
this case, it is conceivable that the affected thermocouples
did not respond as rapidly as the unscreened thermocouples
when the ignition front passed over them.  In addition, it was
likely that some thermocouples were shielded during the run as
the inert material settled towards the grate.  (This may, in part,
be the reason for the erratic behavior of thermocouple No. 2
in Figure 50).

The thermocouple history can, on the basis of the above dis-
cussion, be expected to give a reasonable representation of
the temperatures within the fuel bed up to the point when the
ignition plane reached the grate.  After this time, the inter-
pretation of the temperatures becomes difficult because of the
uncertainty introduced by the bed movement.

From the temperature histories of the twelve fuel bed thermo-
couples, bed temperature profiles were constructed.  Figure
51 shows the temperature profiles throughout the bed at dif-
ferent times during Run 18.  The plot is typical of the re-
sults of those runs in which sustained active burning was
achieved.  The heterogeneous nature of the fuel bed created
a rather poorly defined ignition front, but this restriction
aside, Figure 51 shows that an ignition front of approximately
constant shape propagated through the bed from top to bottom.
Near the end of the run, the bed temperatures increased (typ-
ically by 400-500 F)„  It is thought that this increase in
temperature occurred subsequent to the drying and pyrolysis
of the major portion of the bed (all but the last layer or so
of fuel particles), and did so because at this late stage of
the run more of the energy liberated by chemical reaction
was available to raise the sensible heat of the bed and less
was required to provide the energy requirements of drying and
pyrolysis.-  This result indicates that the grate links of a
traveling-grate stoker could receive their major punishment
                       - 195 -

-------
towards the discharge end of the furnace; in practice the under-
fire air rate to the last grate is high and provides the neces-
sary convective cooling of the links.  Some care needs to be
exercised in interpreting the profiles shown in Figure 51,
as the location of the top of the fuel bed at different times
was not determined experimentally-  The profiles indicated
are strictly, therefore, composite profiles of the entire
fuel bed section measuring temperature both in and above the
bed.  The initial and final depths of the fuel bed are indi-
cated in Figure 51; the final depth is based on measurement
of the volume occupied by the quantity of inert material pre-
sent at the end of the experiment.  The rate of recession of
the top of the bed was not measured experimentally and any
theoretical estimate would have to depend on a somewhat crude
model of the mechanics of fuel burn-out.  This question will
be reconsidered on pages 213-230.  This drawback affects
neither the above discussion of the temperature profiles at
the ignition front nor that concerning the temperature increase
at the end of the run; however, the shortcoming does not permit
any reasonable estimate of the heat lost or gained by the bed
to be obtained from the profiles of Figure 51.  From the earlier
discussion of the mechanics of the settling of the bed as it
burnt out, it may be hypothesized that under the conditions
of Run 18 (approximately equal burning and ignition rates) the
inert content probably accumulated above the burning zone as
it moved through the bed, insulating the bed from radiative
heating from the overfire section.

The slight temperature decline from the peak temperature at
the ignition front to that measured at the top of the bed
observed in this study over most of the duration of the runs
parallels the results of Nicholls (24).  Nicholls1results also
indicated that the temperature decline in an underfeed bed
was rather less than observed in an overfeed bed.  The de-
crease in temperature can be explained by the occurence of the
endothermic reactions C + C02 -> 2CO and C + H_O -> H2 + CO in
in this region of the bed.  Figure 51 shows tnat the tempera-
ture drop through the bed increased near the end of the run,
probably due to the augmented rates of the endothermic reac-
tions as a consequence of the general temperature increase
through the bed at this time.

In the early experiments  (Runs 1-10) it was noticed that the
temperatures at the side wall appeared to be higher than at
the center  of the fuel bed.  Although side wall temperatures
were not measured directly, this trend was deduced from the dis-
coloration of the thermocouple sheaths extending
       for about two inches from the side wall towards the cen-
ter of the bed.  The higher side wall temperatures were attri-
buted to channeling of the underfire air, resulting in a
localized      heat release.  For the later runs (Runs 11-18) ,

                      - 196 -

-------
L-
o

u
QL
15
DC
LJ
Q-
Z
Ld
    1800 -
    1600 -
    1400
    12OO -
                            Approximate  Location
                            of  Top  of Bed at End
                            of  Test
                                        Top  of Fuel
                                        Bed at Start
                                        of  Test
                                          16OO sees
                                        V 1800 sees
                                        ® 2200 sees
                                        o 2300 sees
                                        © 2800 sees
                                        A 3200 sees
1OOO —
     200 -
         0   2   4  6  8  10 12  14 16  18  20 22  24 26 28 30

               DISTANCE  ABOVE   GRATE  (IN)
      Fig.   51.   Fuel  Bed Temperature Profiles at Different
                 Times throughout Run.  Data from Run 18.
                      -  197 -

-------
where the improved high pressure drop grate was used, the occur-
rence of channeling as manifested by high side wall tempera-
tures was not noticed; this was confirmed by the two thermo-
couples used in these later experiments to measure the fuel
bed temperatures at two different vertical and angular loca-
tions about one inch from the side wall (see Appendix H, Fig-
ures H.I and H.2).  These edge temperature measurements also
showed that the ignition front appeared to propagate evenly
across the width of the bed (see pages 210-211 , Figures 56
and 57).

Fuel Bed Temperature Distributions under Poor Ignition Con-
ditions!In those runs where difficulty was experienced in
igniting the bed at the start of the experiment, the bed tem-
perature profiles were markedly different from those shown in
Figure 50.  Although this problem concerns mainly those runs
in the second group of experiments (namely, Runs 5-9), similar
experiences were found in Runs 12 and 13.   The temperature pro-
files observed depended on the methods used in promoting ac-
tive burning.  A discussion will follow on the bed tempera-
tures observed in Runs 5, 6, 7 and 9, as they represent the
spectrum of behavior.

The fuel bed temperature data for Run 5, shown in Figure 52,
indicated that the bed     smoldered until enough of the mois-
ture content of the bed had been removed for satisfactory burn-
ing to proceed.  This was shown by the initial rise in the first
6-7 in. of the bed to 900°-1400°F by 900 sec, followed by a
slow decrease in the temperature at this position in the bed
and a slow rise of  the temperatures through most of the remain-
der of the bed.  The temperatures in the bed about 6 in.
below the ignited portion of the bed plateaued around 200 F.
During this period (0-2800 sec) the underfire air had been
raised in an attempt to kindle the fire.  After 2800 sec the
temperatures started to rise slowly throughout the bed with
the temperatures toward  the top of the bed leading those
lower down.  By 4000 sec the whole bed had ignited and the
temperatures throughout the bed steadily increased until they
reached a peak of 2000 F at 4600 sec.  At this time combustion
was nearly completed and the temperatures started to fall.  The
underfire air was decreased at 4000 sec because visually the
bed did not appear to be burning satisfactorily at this time.
The data indicated that burning would have proceeded satis-
factorily if the underfire air had been left at its previous
level of 145 Ib hr xft  .

No satisfactory ignition of the bed could be obtained in Run
6 despite a number of underfire air changes and reignition of
the gas burners so as to increase the heat flux to the top
of the bed.  Visual observation of the fire showed that it
                      - 198 -

-------
  2000
I

H
VO
   160O-
    1200 -
cc
ID
o:
    800 -
    400 -
0



Fig.
                                                                                                       5600
               52.
                         Fuel Bed Temperatures (Run 5).  Thermocouples 1, 2, 3, 4. 5, 6, 7, and 8 were located

                         25, 23, 19, 16, 11, 9, 7, and 2 in. from the grate.  See Table VI.2 for experimental

                         conditions.

-------
1
w
o
I
Ld
CC
ID
t—
<
CK
U
CL
2
LJ
    2400
    2000(—
     ieooh-
    1200h
          eool-
     400h
         Fig. 53,
                     400
                                     800
                                                                      2000
                                                                                      2400
2800
                        1200         1600

                   ELAPSED   TIME  (SECONDS)
Fuel Bed Temperatures  (Run 7).   Thermocouples 1, 2, 3, 4, 5, 6,  7,  and  8 were located
23, 19,  18,  13,  11, 9, 7, and 0 in. from the grate.  See Table VI.2 for experimental
conditions.

-------
I

o

I
        1800 -
        1600 -
        14OO -
     -  12OO-
LJ
CC
ID
I-
<
ce
LJ
Q.
5
LJ
   100O -
         800 -
         600 -
         400 -
        ' 2OO
                  400   8OO
                        12OO  14OO  2000  240O  280O  32OO  3600 4OOO 440O 48OO 5 2OO 56OO 6OOO

                                      ELAPSED  TIME  (SECONDS)
             Fig.  54.   Fuel Bed Temperatures  (Run 9)
                         Thermocouples 1, 2,  3,  4,  5,  6, 7, 8, 9, and  10  were located 25,  23,
                         19,  18, 13, 11, 9, 7,  2,  and  0 inches from the grate.  See Table  11,
                         for  experimental conditions.

-------
appeared to be almost totally quenched about 25 min after
the run had started.  Data taking was stopped at this time
and a complete record does not exist for this run.  It was
observed that the bed continued to smolder for about 5 hr
after the underfire air was cut off, and almost complete burn-
out of the fuel was obtained.  This result provided striking
evidence of the tendency for air to be drawn down through the
bed by natural draft.  The bottom of the bed had been completely
sealed by the cfrate and the combustion air could only have
been supplied by air sinking down through the bed at the "cold"
wall.  The burning rate under these conditions was very low.

The data for Run 7 depicted a different picture of the bed
temperatures,, which are shown in Figure 53.  In the first
few inches of the bed, the temperatures rose quickly to well
above ignition temperature, but at depths below 10 in. the
temperatures rose slowly over a period of about 10 min to
1000 F. About halfway through the run (1500 sec) the whole bed
had reached a temperature of around 1000 F; from this time un-
til complete burn-out was achieved, the temperatures roseQ
slowly throughout the bed over a period of 10 min to 2200 F
(the maximum temperature observed,       was achieved about
10 min before complete burn-out).  The burning rate of the bed
increased raarkely shortly after the bed temperatures reached
their maxiim.ua and remained relatively constant until final
burn-out.  The underfira air washeld-constant for the major
portion of this run at 146 Ib hr  ft   since, visually, the
bed appeared to be burning reasonably well.

For Run 9, yet another type of behavior was observed.  The
bed temperature profiles for this experiment are shown in
Figure 54.  The bed ignited rapidly down to 5 in. (about 400
sec,into the run) despite the underfire air rate of 275 Ib
hr  ft  , and then the temperatures in this ignited sec-
tion of the bed fell to around 600 F (at 1200 sec).  The gas
burners were raignited at 1500 sec to keep the top section
refractories above 1400 F, and the underfire air was decreased
since it was visually observed at this time that there had
been poor ignition.  From this time up to 2400 sec into the run,
the Jemperature in the top few inches of the bed rose slowly to
1800 F, after which the ignition plane started to move through
the bed until about two thirds of the bed was ignited.  Vis-
ual observation showed that the bed appeared to be burning
properly at th|s time and the underfire air was raised to
175 Ib hr  ft  .  The temperature data indicated that the
underfire air rate must have been increased too quickly, for
at 3000 sec the bed temperatures in the ignited portion of
the ^ed began to decrease, dropping to a minimum of 800°-
1400 F at 4400 sec.  The temperature profile was still pro-
pagating into the virgin fuel at this time but temperatures
                      - 202 -

-------
did not reach the ignition level.  The underfire air was again
decreased and at 4400 sec the temperatures  in the top two
thirds of the bed began to increase.  The remainder of the bed
began to ignite, slowly at first and then more rapidly.  The
bed was fully ignited at 5000 sec, at wbich time the under-
fire air was again raised to 175 Ib hr^ff2 since the bed
appeared to be burning satisfactorily.  The burning rate was
observed to increase in the later stages of the run but not
to the extent noticed in Run 4.

Drying Front.  As discussed on pages 192-208 , the drying rate
of a fuel element is limited by the rate of transfer of
energy from the burning front and by the rate of conduction
of energy in the object being dried.  When the latter resis-
tance is low, for example in thin or porous objects, the tem-
perature will plateau at 212°F and the drying wave will pro-
pagate ahead of the ignition wave.  This was the case in the
Bureau of Mines study (2J>_,2_7_) , where the moisture was intro-
duced with leafy vegetable material.  In the present study
the moisture was introduced in blocks of wood, in sizes such
that the heating of the center of a block lagged behind the
surface heating by times of the order of 1000 sec.  For this
case only a slight plateau was observed when the temperature
measured within the bed reached 212 F and drying continued
within the wood blocks even after the surface had ignited.
The peak temperatures within the fuel bed observed toward
the end of  a  run probably occurred subsequent to the dry-
ing of the entire bed.

Additional evidence for the protracted drying of elements of
the fuel bed is provided by the data from runs in which dif-
ficulty was experienced igniting the bed (e.g., Run 5).  In
this case the temperature wave from the ignition front dis-
sipated slowly enough that marked plateaus were observed when
the temperature ahead of the ignited zone reached the vicin-
ity of 212 F (see Figure 52).  Under normal burning conditions
this 212 F plateau would not be expected because of the much
higher heating rates to which the fuel bed elements would be
subjected (on pages 192-208 ) .  Under the normal burning con-
ditions encountered in this study, the quenching effect of the
wet wood blocks, where the internal diffusion of water was
low relative to the heating rate, would not have been a hin-
drance to ignition and combustion.

Pyrolysis Front.  In a number of runs a pl?.teau at about 500 F
was observed, corresponding roughly to the temperature of py-
rolysis of cellulosic materials.  It is difficult to deter-
mine if this was a real effect or was caused by temporary
shielding of the thermocouples by inert material.  The heat
                      - 203 -

-------
effects of pyrolysis will generally be rather lower than
the drying heat effects and at the higher fuel bed heating
rates it is unlikely that the pyrolysis "wave" would have
been noticed.  However, as in the case of drying, the pyroly-
sis of larger wood blocks was limited by the rate of pro-
pagation of energy into the center of the blocks.  Pyrolysis
would therefore have been initiated as the surface tempera-
ture of a block approached 500 F and continued internally
within the block for extended periods.

Temperatures at the ignition Front.  The maximum temperatures
observed at the ignition front for the different runs in
which it was possible to obtain reasonable values are shown
in Table 12,-there was considerable scatter in the data, and
selecting a range of values was often arbitrary.  There was
surprisingly little variation in the ignition front tempera-
tures despite fairly considerable changes in the underfire
air flow rate (86-175 hr~lft~2) and the moisture content of
the fuel (15-32%).  The lowest temperatures observed occurred
in Run 1, but these temperatures were quite possibly erroneous
as they seem very low for the type of active burning usually
observed.  It appeared that the lower temperatures were asso-
ciated with the lower air flow rates.  A simple explanation
for this is the greater effect of heat losses at these lower
heat generation rates.  There were no distinct trends with
increasing moisture content although the theory developed ear-
lier (a simple overall heat balance) suggested that the maxi-
mum temperature should decrease with moisture content, accor-
ding to

                             net
                 T    = 	s	                     (54)
                  max   GC   + UC
                          P      P
                          rg     *s

This result may be attributed to the drying characteristics
of the material used.  The wood was known to dry over a pro-
tracted period of time, and the amount of moisture released
up to the point when the temperature at the ignition front
reached a maximum may have remained roughly constant, regard-
less of the moisture content.  For material where the inter-
nal diffusion of water is high  the effect of moisture on the
maximum temperature will be much more pronounced.  This will
have a strong adverse effect on the ignition rate.


On the basis of  equation (54) ,  the maximum temperature for Run
15,   would have been expected to be lower than in Runs 11 and
12 because of the higher ignition rate (based on the rate of
                      - 204 -

-------
                                 TABLE  12




            Maximum Temperature  Observed  at  Ignition Front with




             Varying. Underfire Air Rates  and Fuel Compositions
Run
No.
1
10
11
12
14
15
16
17
18
Fuel Analysis
» Inert
-
17.10
15.51
12.79
15.08
-
16.23
15.18
15.04
% Moisture
8
31.76
23.99
25.89
14.73
26.40
24.85
25.38
25.06
% Com-
bustible
92
51.14
60.50
61.32
70.19
73.60
58.92
59.49
59.90
Underfire
Air Rate
(Ib hr~1ft~2)
Natural Draft
175
155
155
160
156
118
140
86
Maximum Temper-
ature Observed at
Ignition Front <°F)
1100 - 1400
1600 - 1800*
1800 - 2000
1800 - 2000
1800 - 2000*
1800 - 2200
1800 - 2100
1600 - 2000
1600 1800*
1600 - 2200
1700 1900*
1600 - 1900
1500 - 1700*
1600 - 1900
(a)   Observed in top two thirds of bed




(b)   Observed in bottom third of bed
                          - 205 -

-------
 ignition of both  fuel and combustible).  The  range  of  maxi-
 mum  temperatures  was observed  to be  slightly  lower  than  in
 Runs 11 and 12 but  the difference was  hardly  significant.

 Temperature Profiles Below Ignition  Front.  From the temper-
 ature history of  the different thermocouples  in the bed  and
 a  knowledge of the  ignition rate, the  fuel bed temperature
 could be plotted  as a function of distance below the igni-
 tion plane.  The  results for one test  were not all  the same
 because of the heterogeneous nature  of the bed surrounding
 the  thermocouples;however, it was possible to construct  an
 average curve for the temperature-distance profile.  These
 profiles are shown  in Figure 55.

 From equation 58  the temperature below the ignition plane  is
 given by

                              - (a +  3)z
                         ign
                                                          (58)
where
            a =
            6 =
Under most practical conditions h  - 200 Btu hr  ft   F  ,

             ~1~1°~1                        ~lo~1
   - 4 Btu hr
                          pg
   = 0.5 Btu lb
 50  Ib  hr'1  ft~2 and G = 100 Ib hr~1ft~2
                                                       U *
                                        giving
                  T ^ T.   e
                    ~  ign
K
 v
 E
(170)
Equation  (170) predicts that as the volumetric heat transfer
coefficient, h , igcreases or the effective thermal conducti-
vity of the bed, K ,  decreases the profile below the ignition
front will become steeper; conversely, as the heat transfer
coefficient decreases or the effective thermal conductivity
increases, the profile below the ignition front will become
shallower.  This prediction is confirmed by the results of
Runs 18, 16 and 11 shown in Figure 55, in which progressively
                      - 206 -

-------
M
O
•z.
O
CC
U.

Z
              CD

              LJ
              CL
              Ct
              UJ
              Q_
              Z
              LJ
                  600
                  500
              =    40O
              O
              Q   300
Run 16
  Run 18
                                                  Run
                                                  No
	1	

Underfire
Air  Rate
Lb  HrVt?
	1	

Ignition
Rate
Lb Hr Ft2
                                           	1	
                                            Fuel
                                            Composition
     200
     100
                     0
                             I
                       0
                       Fig.  55.
                           0.1
                    DISTANCE
                          0.2                 0.3
               BELOW  IGNITION  FRONT  (FT)
                    Fuel Bed Temperatures Below the  Ignition Front for Different Fuel
                    Types and Underfire Air Rates

-------
greater underfire air flow rates were used (h  scales approxi-
mately as G0-63).  The temperature profiles of Run 15 also
follow the trends indicated by equation (170).  In this run
the absence of the inert material (the components of which were
slightly larger than those of the combustible portion of the
fuel) very likely decreased the effective mean beam radiating
length through the bed (see pages 198-203 ) and thus decreased
the effective thermal conductivity.

Using the heat transfer correlation of Bradshaw fit &1.  (64)
and parameter values as follows:  Cp - 0.4 Btu Ib"10?"1 ,
N2/3 = 0.81, y = 0.1 Ib hr^ft"1, D  - 0.125 ft, 6 = 0.4,
the heat transfer coefficient based on a unit volume of bed
is found to be,    = 10<460.625       -1-30      Figure
55 shows theoretical curves, calculated fgom equation (170)
using the above relationship for h  and K^ = 4 Btu hr-lft"1
 F'1  (see pages 198-203 } , for two different values of G
corresponding to conditions for Runs 18 and 11.  The theore-
tical profiles are of the correct order but the variation with
changing underfire air rate is not as large as that observed
experimentally.  However,  the theory is only approximate and
the assumptions of constancy of effective thermal conductivity
and particle isothrrmality, which are known not to be strictly
true for many fuel bed conditions, will mitigate against
precise prediction of the temperature profile below the igni-
tion front.

From the results of the theory presented above, the tempera-
ture profiles of Run 14 should have been close to those of
Run 11 as the only difference between the two runs lay in
the moisture content of the fuel.  The fact that the data
fell between those obtained for Runs 16 and 18 is anomalous
and no satisfactory explanation can be offered.
                      Ignition Rates
The ignition of the bed can be conveniently divided into two
stages — the initial ignition of the bed at the start of the
run by the radiant heat from the hot refractory brickwork and
the propagation of the ignition front through the bed.  For
all the runs performed, the fuel bed ignited at the start of
the run, even when the top surface contained more than a pro-
portionate amount of wet wood blocks, based on the total mois-
ture content of the fuel.  The initial radiant heat fluxes
                      - 208 -

-------
to the top of the bed were of  the order  10   Btu  hr   ft   ,
very much greater than  required  for  ignition,  even  for  blocks
of wood containing  60%  moisture  (117) , which was the maximum
moisture content of any of the wet wood  blocks used in  this
study.

The major difficulty with the  ignition process lay  in adjust-
ing the underfire air rate so  that the ignition  front would
propagate through the bed; this  has  been described  on pages
198-203 for the cases of Runs  5-7 and 9.  At the time these
runs were conducted, it was  felt that burning  rates typical-
ly observed in a traveling-grate incinerator (60 Ib hr   ft~2)
could be obtained with  underfire air rates  around 150 Ib hr~l
ft~2 and it was thought that at  these underfire  air rates
the burning rate would  be close  to its maximum.   It was
for this reason that underfire air rates varying from 120 to
275 Ib hr  ft~2 were employed.   The  range of underfire  air
rates was weighted  to rates  higher than  150  Ib hr-1ft~2  be-
cause of the desire to  map out the ignition-limited regime
of burning.

Runs 5-9 clearly showed that raising the underfire  air  too
quickly at the beginning of  a  run  (on a  traveling grate  this
would be equivalent of  having  too high an underfire air  rate
through the first few windboxes) seriously  hampered the  igni-
tion rate; this behavior was not observed by Nicholls (24)
in his studies on coke  and coal  as in his experiments tEe"
underfire air was raised to  its  final level  very quickly
after the fuel bed had  been  ignited.  In addition,  once  the
ignition front had propagated  a  few  inches  into  the bed, the
effect of changing  the  radiant heat  flux to  the  top of  the bed
was found to be minimal because  of the insulating effect of
the inert and charred fuel.  All the energy  required to  drive
the ignition front  forward had to come solely  from  the heat
liberated by the reaction of oxygen  with the char and pyrolv-
sis products in the partially  ignited bed.   Once the tempera-
tures in the ignited zone dropped low enough to  provide  a
significant chemical resistance  to combustion, it was very dif-
ficult to increase  the  heat  generation rate  to get  the bed
back into an active burning  state because of heat losses to
the cold fuel bed wall  and the unignited fuel.

As discussed on pages 192-208, ignition  temperatures differ
between materials and with experimental  conditions.  For cel-
luosic materials, piloted ignition occurs at temperatures of
550° to 600°F.  In this study, a temperature of  600  F was se-
lected for the derivation of curves, such as that shown  in
Figures 56 and 57 for the propagation of the ignition front.
In active burning runs  the temperature gradient  near the ig-
nition front was sufficiently  steep  that the values  of  the
                       - 209 -

-------
     26
LJ
X
(J
(L
UJ
O
     2O
16
     12
      8
      4
                                2000        3OOO

                      ELAPSED  TIME, SECONDS
                                                     4OOO
        Fig.  56.   Ignition Wave Propagation  (Run 10).   Ignition
                   temperature assumed equal  to 600° F;  underfire
                   air flow 175 Ib hr"1 ft"2; bulk density  of fuel

                   16.7 Ib ft""3; initial bed  depth 30 in.  $ meas-
                   ured at center of bed;   gj  measured at edge
                   of bed.

-------
I
ro
                               D Measured at Edge
                                 of Bed
                               O Measured at Center
                                 of Bed
                                 Measured  at Edge
                                 of Bed
                                 Measured  at Center
                                 of Bed
               u   20 —
en
(D

UJ
>
O
CD
UJ
U
(f)
Q
                    0
                                   1000          20OO          3OOO

                                  ELAPSED  TIME (SECONDS)

                      Fig. 57.  Ignition Wave Propagation  (Runs  17 and 18)
                                                                 4OOO

-------
 ignition  rate were  found  to  be  insensitive  to the value of
 ignition  temperature  selected.   The  ignition rates were com-
 puted  as  the product  of the  rate of  travel  of the ignition front
 and  the initial bed density.  For active  burning runs the rate
 of travel of the  ignition front was  found to remain approxi-
 mately constant over  the  duration of the  experiment;  the varia-
 tion of the data  shown in Figures 56 and  57 was  typical of the
 deviation from linearity  in  the plot of depth of ignition
 front  penetration versus  elapsed time.  Ignition rates were
 found  to  vary from  125 Ib hr'^-ft"2 for an underfire air rate
 of 127 Ib hr~1ft~  (Run 5) to 34 Ib  hr-1ft"  for an underfire
 air  rate  of 175 Ib  hr~1ft~2  (Run 10).  A  summary of the igni-
 tion rates found  in the experiments,  the  underfire air rates,
 and  fuel  compositions in  given  in Table 51.

 The  high  ignition rates observed in  the later stages  of the
 runs where difficulty had been  experienced  in igniting the
 bed  (e.g., Runs 5,  7  and  9)  were undoubtedly much too high
 for  the initial fuel  composition employed.   This descrepancy
 occurred  because,, at  the  times  when  these ignition rates
 were calculated,  the  bed  had been partially dried (see page
 203  and Figures 52, 53 and 54).   The high ignition rate ob-
 served in Run 1  (110  Ib hr~-*-ft~2)  can be  attributed to the
 low  moisture content  of the  fuel.  This ignition rate was pro-
 bably  close to the  maximum achievable for the inert-free
 fuel used in this study-   As clearly demonstrated in  Nicholls1
 (24) experiments  arid  by the  embryonic ignition theory developed
 on pages  118-129, the ignition  rate  would be expected to pass
 through a maximum as  the  underfire air rate is increased.
 It is  hypothesized  that under natural draft conditions the
 underfire air rate  increases to a level which gives an igni-
 tion rate close to  its maximum.   If  this  hypothesis is cor-
 rect it lends credence to the belief that the ignition rate
 for  Run 5 was measured for a fuel which had been almost com-
 pletely dried.  For Runs  11  through  18, where the calculated
 ignition  rates were more  truly  representative of the  initial
 fuel composition  and  of the  conditions of the test, the ob-
 served ignition rates lay in the range 34-50 Ib  hr~1ft~2.

 The  ignition rates  calculated for Runs 10,  11, 16, 17 and 18
 which  were all conducted  with a fuel of approximately constant
 composition, show (with the  exception of  Run 11)  a monotonic
.increase  with decreasing  underfire air rate (see Table 11).
 This result follows the general trend in  ignition rates ob-
 served by Nicholls  (24) in the  equilibrium  burning regime (e-
 qual ignition and burning rates).  It may be postulated, there-
 fore,  that if underfire air  rates below 86  Ib hr~1ft"2 had
 been employed the ignition rate would have  been  observed to
 reach  a maximum and then  to  decrease as the underfire air
                         — •  O 1 *>

-------
was further decreased.  The ignition behavior observed  in  Run
11 is anomalous and no satisfactory explanation can be  offered.
More experiments are required to extend  the range of under-
fire air rates and to duplicate the results presented here
before a more quantitative picture of the  ignition behavior
can be drawn.

Camparison of the ignition rates obtained  in Runs 11, 14 and
15 indicates that fuel composition changes had very little
effect on the ignition rates.  Within the  limits of experi-
mental error in determining the ignition rates, the rates
were the same for the base case composition  (15% inert, 25%
moisture and 60% combustible) and for the  cases where the  inert
and moisture content were both 15%, and  where no inert  was
present and the moisture content was 25%.  This is a some-
what surprising result, as decreasing both the inert and
moisture fraction of the fuel would be expected to enhance
ignition propagation.  Further experiments are needed cover-
ing a wider range of inert  (0-30%) and moisture 0-40%)  con-
tents before trends can be determined.

It is thought that the maximum underfire flow rate at which
ignition will still be sustained for the typeof fuel used
in this study is in the vicinity of 275  Ib hr  ft".  In Run
10 the underfire air rate was increased  over a period of around
2 min to the 275 Ib hr-1ft~2 level only  after the bed had  ig-
nited and started to burn vigorously-  About 10 min after
the air rate had reached 275 Ib hr~^-ft~2 the bed appeared  to
have been almost totally quenched.  The  underfire air rate
was then reduced to 175 Ib hr-1ft~2 and  visual observation
of the fuel bed indicated that, within the space of a few
minutes, the ignition and burning processes had started again.
This behavior was borne out by the fuel  bed temperature pro-
files, the first thermocouple  (5 in. from  the top of the bed)
not reaching the ignition temperature until after 900 sec
into the run; typically the time for the ignition plane to
reach this level was around 300-400 sec.   Similar experiences
with early quenching were obtained from  Run 9, where the un-
derfire air rate was also increased to 275 Ib hr~1ft""1^  over
the first few minutes of the run.
                    Burning  Rates
Plots of weight  loss of the  fuel  as  a  function  of  elapsed
time for Runs 10,  17 and  18  are shown  in  Figures 58,  59  and
60.  These plots were obtained using the  load cell weighing
                       -  213  -

-------
n   4O
CO
O

I—
X
O
LU
    2O
    1O
             1OOO    2OOO     3OOO     4OOO

                ELAPSED  TIME ,  SECONDS
                                              5OOO   6OOO
      Fig.  58.  Weight Loss as a Function of time  (Run 10)
                Underfire air rate 175 Ib hr"1 ft"2 and
                bulk density of the fuel 16.7 Ib ft"3;
                initial fuel bed depth 30 in.
                     - 214  -

-------
tn

I
                          Burning  Rate Over Constant
                          Burning  Period = 39 Ibs./Hr"1  Ft
                         4OO   8OO   12OO
1600  2OOO  24OO  28OO

     TIME  IN SECONDS
3200   36OO  4OOO  44OO
                   Fig.  59.  Weight Loss as a Function of Time (Run 17)
                             Underfire air rate 140 Ib hr"1 ft"1; bulk density of fuel
                             15  Ib ff3; initial fuel bed depth 30 in.

-------
CD
   60
   50
   40
to
to
P  30
   20
ui
    10
	 Burning Rate
   30 Lb HrVt"2
                                  11      I
                                                                               -1   -2
                                                         Burning Rate = 41 Lb Hr  Ft
                                                   I   I   I    I
                                                             I	I    I   I    I
                     1000
                                2OOO            3000

                                       TIME  (SEC)
40OO
      Fig. 60.  Weight Loss as a Function of Time  (Run 18)
                Underfire Air Rate  86  Ib hr"1 ft~2, bulk density of fuel 16  Ib ft"  ,
                initial bed depth 30 in.

-------
system and are representative of the burning rate curves
found for those runs where active burning was observed for
the major portion of the test.  These weight loss curves
have been smoothed; a representative unsmoothed curve is shown
in Figure H.7 in Appendix H.  For Runs 17 and 18 independent
checks on the burning rates were available from the material
balance calculations (see pages  230-237 ) .  The slight offset
in the weight loss curve at zero time in Run 17 was caused
by slight differences in the recorded weight of the fuel bed
section before and after it was butted up to the top section
of the apparatus.  This offset did not affect the weight loss
results and the "weight loss" of 1.5 Ib at time zero may
simply be taken as the origin for the ordinate.  The burning
rates were obtained for the different runs by taking the slopes
of the weight loss curve during the time periods that the burn-
ing rates were constant.

For Run 14 no load cell data were available because of equip-
ment difficulties, but the burning rate was estimated from
closure of material balances at different times during the
run.  Since no accurate measurements were available for the
stack gas water content and the air leakage rate was not mea-
sured, the material balance was forced to close by selec-
ting air leakage rates and stack gas moisture content correc-
tion factors that closed the carbon and hydrogen balances.
The burning rates for Runs 15 and 16 were estimated by assum-
ing that almost complete combustion had been achieved by the
time the ignition front had reached the grate and by further
assuming that the burning rate was constant over this period
of time.  This estimate is expected to be a little low.  For
comparison, the same technique was used to,calculate the burn-
ing rate for Run 14j the result, 43 Ib hr  ft" , agreed close-
ly with that calculated from the material balances.

The burning rates for all the runs in which representative
values could be obtained are shown as a function of underfire
air rate in Figure 61,-  for comparison, the ignition rates cal-
culated-for these runs  (on the basis of Ib of combustible
hr  ft  ) are also indicated.  For burning rates lying on the
solid line AB, the underfire air and fuel consumption rates
were in stolchiometric proportions for a fuel of a composition
approximating the average fuel composition used in Runs 11-
13 and 16-18.  Although the rates were in stoichiometric pro-
portions the compositions of the gases leaving the top of the
bed were not restricted to being only CO- and H20; the off-
gases could have contained any mixture of 0~, CO, CO_, H_,
H_0, N« and trace pyrolysis products as long as the 6- and
combustibles were in stoichiometric proportions.  For burning
rates to the left of this line, secondary air was required for
complete combustion; for burning rates to the right of
                      - 217 -

-------
                      AIR  DEFICIENT (PERCENT)
'cc
 i

 ui
 CD
 2
 O
 u

 CO
 LU
 (-
 <
 CC

 z
 O
 I—
 I/)
 3
 CO
 z
 O
 CJ

 ce
 O


 z
 O
                 75
                         ,50
                          .25
             75
                      50
                     25
          75
                  50
                25
                                              /
     80
     70
60
50
   D Burning Rate

   O'Q^ition Rate

   ^Burning Rate Estimated

     From CO and H2 Cone. At

     Top of  Bed    stoichiometric Line

                  Runs 11 -13 , 16-18


—    Stoichiometric Line

     Run 10
30 -
2O —
      10 —
             /50
           / 25
/°/
                            Air
                            Deficient
                                               /25
                                                  /
                                              25
                                                /
                                   25
                                                        /25
                                            (2) (3)
                     m
                     x
                     O
                     m
                     x
                     o
                     m
                     z
                                 Combustible
                                            30
                              56
                                  25
           60
               20
                                                65
                                                    15
70
                    Stoichiometic  Line Based on the

                    Following  Analysis of Combustibles

                             ( Wt "/„ )

Stoichiometric  Line Run 14       C-042

                               H = 0.06

                               O= 0.52
                -Stoichiometric Line Run 15


                  _J	I	I	L_
        O           100         2OO         3OO

                                      -1  -2
         UNDERFIRE AIR  RATE  ( LB  HR  FT )
         Fig.  61.   Ignition and Burning Rates as a Function
                     of Underfire Air Rate
                           - 218  -

-------
the line, excess oxygen passed  through  the  bed and ftccoft4*ry
air was not strictly required if  adequate mixing wa» Achieved
above the bed.  The scales on the top and sides of Fi>
the burning bed depth was not sufficient for all >th
-------
This finding was confirmed by an estimation of the burning
rate from the air deficiency calculated from the measured H2
and CO concentrations within the fuel bed; this ppint is shown
as a triangle in Figure 61.

For Run 18, the carbon burning rates calculated from fuel bed
and stack gas measurements gave a different picture.  From
the material balance and the stack gas C02 content, an aver-
age carbon burning rate (from 2400 - 3200 sec) of 1.4 Ib
moles hr-lft-2 was calculated,, while the carbon burning rate
estimated from the fuel bed gas composition was about 1.45
Ib moles hr-lff2.  In this run, therefore, channeling did
not appear to be a problem.  Further confirmation that no
channeling occurred came from an estimation of the burning
rate by calculating the air deficiency from the fuel bed gas
compositions.  The point shown as the triangle in Figure 61
was calculated from the gas compositions obtained from the top-
most fuel bed probe at 2800 sec, at which time it was esti-
mated that the probe was close to the top of the bed.  The re-
sult agreed closely with the burning rate obtained from the
load cell, which is an average rate for the whole bed.  Nic-
holl's data (24) indicated that in his experiments channeling
became more serious as the underfire air rate was increased.
The results of this study indicated at similar trend.

For all runs the early burning rates increased as the run
progressed.  The shape of weight loss/time plots took on two
basic forms as shown by Run 17, Figure 59 and Runs 10 and 18,
Figures 58 and 60.  For Run 17, the burning rate slowly in-
creased with time until a constant rate was attained which
was then maintained up to about the time the ignition front
reached the grate.  Similar behavior was seen in Runs 1 and
2.  For Run 18 the burning rate started at a constant level
and about halfway through the run suddenly increased to ano-
ther constant level.  This sudden break in the burning rate
was confirmed by the material balance calculations.  Similar
trends were noticed in Run 10, but the burning rate appeared
not to make such an abrupt change.

For Runs 10, 14, and 15-18, the data showed that, during the
early stages of these runs, the burning rates were less than
the ignition rate of the combustible content of the fuel,1
while during the later stages the burning rates were greater
than the ignition rates.  These findings were in marked con-
trast to Nicholls' (24) results, where equilibrium burning
(equal ignition and burning rates) was observed at high under-
fire rates over the major portion of an underfeed burning
test.  Nicholls did not report any data on the developing thick-
ness of the burning zone as burning rates were only measured
(from stack gas compositions)  after equilibrium burning was
established.  The results presented here, by contrast, were taken
                          -220-

-------
during the period when the depth of the active fuel bed was
developing.  It is obvious that a situation where the igni-
tion rate is lower than the combustion rate is unstable
since under these conditions, the active bed thickness will
decrease until the burning and ignition rates equilibrate.

Before discussing these trends it will be worthwhile to indi-
cate exactly what constituted the measured burning rate.  The
burning rate was made up of contributions from the drying and
pyrolysis  (or carbonization) of the combustible, char oxi-
dation  (C + 1/2O2 + CO and C + 02 -»• CC>2) and char gasification
(C + H20 -» H2 + CO and C + CO^ -> 2 CO) throughout a depth
slightly greater than the ignited zone  (because some drying
and to a lesser extent pyrolysis takes place ahead of the
ignition front).  Vaporization of metal coating and oxidation
of steel are small enough to be neglected.

The observed increase in the burning rate as the runs pro-
gressed was in accord with the development of a burning depth.
As the distance between the ignition front and the top of the
bed increased, more time was available for the various gas
phase and gas-solid reactions to occur.  In addition, more
depth was provided in which drying and pyrolysis could take
place.   (As indicated above, the observed burning rate was
the sum of the weight losses associated with drying, pyrolysis
and solid char consumption.)

As the burning zone developed, an increasing amount of the oxy-
gen in the underfire air was consumed and in most experiments
the burning zone eventually increased to the extent that all
the oxygen was consumed.  The depth required depended on the
rate of mass transfer of the oxygen to the char surface and
the rate of reaction of the oxygen with pyrolysis products.
The oxygen consumption depth may be theoretically calculated
from the correlation presented on pages 61 - 75, Figure 19,
although the calculation must only be considered approximate.
For this study the Reynolds numbers DpG/u used lay in the range
80-200 and from Figure 19 the "effective" jD factor was there-
fore about 0.15-0.20=  On the assumption that the effective
mass transfer area was 30 hr~2ff3, the oxygen consumption dis-
tance (z0), where the oxygen concentration fell to 0.1%, was
of the order of 0.5 ft.  This value of ZQ is of the same
order of magnitude as that observed experimentally - see pages
75 - 82  •  As the depth of the burning zone increased, and
more oxygen was consumed, the burning rate would be expected
to increase partly as a result of increased char oxidation
and partly as a consequence of the greater heat release
within the bed.   (From the analysis on pagesm        increas-
ing the underfire air rate increases ZQ  (zo G°-41) and
                           -221-

-------
therefore low underfire air rates are preferable while the
burning depth develops.)

Once enough depth had become available for all the oxygen to
be consumed, further increases in depth provided a region in
which the gasification reactions could proceed.  The degree
to which the gasification reactions occurred depended on there
being adequate carbon left for reaction and high enough tem-
peratures in the fuel bed (ca. 2000 F) for the reactions to
have proceeded at reasonable rates.  Burning rates in this
regime would, on the basis of the theory of Niessen et al.
 (19) , and on the assumption of no ignition rate limitations,
be~~expected to increase proportionately with increases in the
underfire air rate and would increase with decreasing heat
losses from the bed  (e.g., side wall losses or radiant heat
losses from the top of the bed). The burning rate will obvious-
ly be dependant on the amount of channeling that takes placef
but an effective burning rate could be calculated on the hy-
pothesis that there are regions of the bed that act as gasi-
fyers and  regions where bypass of the underfire air occurs.
For the fuel particles used in this study, where there was
a major resistance to heat transfer within the elements, the
weight loss that was associated with the drying and pyroly-
sis of each of the combustible fuel elements would have de-
pended on the size and moisture content of the element and
the length of time it had been in the ignited zone.

Towards the end of a run, after the ignition front had reached
the grate the thickness of the burning zone started to de-
crease.  Eventually the burning zone became thin enough for
oxygen to break through the bed; the' depth at which this occurred
can be estimated using the methods outlined above for calcul-
lating z , but the effective mass transfer coefficient would
be expected to be rather less in this case because of the lo-
wer probability of any reaction between oxygen and pyrolysis
products.  The rate of oxygen breakthrough and the burning
rates at this stage in the combustion process may be esti-
mated as follows.  Assume that the oxygen concentration at
the top fo the bed can be given by


                                 -k
           f " m  T    1
= 0.21 exp   -^51 z^t) J                 (171)
       T
where PQ (t) is the mole fraction of oxygen at the top of the

bed; km the effective mass transfer coefficient, lb hr~1ft~3;
G the underfire air flow rate, lb hr-ift'2; and zT(t) the height
of the bed at any time t, ft.  The burning rate, assuming a
fuel comprising only carbon and inerts and that the reaction
                      -  222  -

-------
         C02 predominates,  is  then  given  by
         Burning  Rate  =  i| G   0.21  -  P£  (t)
                               (172)
The rate of recession  of  the  top of  the  bed  can  be  approxi-
mately expressed  as,
-Ap
                       12  -
                       2~9"  G
6 To.
21 -
(t)
(173)
where Ap is the density  difference  between  the  carbon  and
inert fuel and the  inert alone.   Substituting equation (172)
into equation  (173) ,   T   integrating  and  substituting  the re-
sulting equation  for  z  as  a  function of  time back  into equa-
tion  (171) gives, with some approximation in the  final an-
swer, which becomes more correct  as complete burn-out  is
achieved.
       = o.21 exp
                   -k
                     m
                    2G
            log!1-expj
T
-Vo)
G j
Yl-0.0869Gt/l
/J Ap
(174)
(J
where ZQ is the height of  the bed when oxygen  just starts  to
break through; since  the oxygen  concentration  never reaches
zero using the representation of equation  (171),  ZQ is taken  as
when the oxygen concentration reaches an arbitrarily small
number.  From equation  (174) the time required for the oxygen
concentration to reach 21% after incipient breakthrough  (t )
is given by
             t  «
             T1   0.0869G
                               (175)
                                   -3    T
For typical values of Ap   -  3 Ib  ft    ,  z    -  0.7 ft  , G
 - 100 Ib hr-lff2  and k    - 560 Ib hr-lff3  , t  is approxi-
mately of the order of the1 times  experimentally observed for
the burning rates to decrease to  zero  from their "steady
state" values prior to oxygen breakthrough  (Figures 58 - 60)
and for the times it took the fuel bed oxygen  concentrations
to increase from 0% to 21% at the end  of a test — see pages
237-253 ; from the assumptions involved  in this simplistic
theory, the estimate of t  would be expected to be too high.
                      - 223 -

-------
 (The above analysis indicates that underfire air rates should
be decreased once the bed thickness has decreased to the point
tht oxygen breakthrough occurs, for the following reasons:
firstly decreasing the underfire air will decrease the excess
air above the bed and secondly will help keep bed temperatures
high and will, therefore, prevent premature quenching of the
bed.)

The rate at which the active fuel bed depth (defined as the
distance from the ignition front to the plane of zero combust-
ible) increased was not strictly proportional to the differ-
ence between the ignition and burning rates as the measured
weight loss (somewhat loosely called the burning rate here)
included, as mentioned above, weight loss that did not affect
the depth of the active bed (i.e., drying and pyrolysis).
A distinction must be drawn here between the active fuel bed
depth as defined above and the distance from the ignition
front to the top of the bed.  This latter distance will,
under certain circumstances, be rather greater than the former
because of the accumulation of inert material as the bed burns.
Inert accumulation will depend on the relative rates of igni-
tion and burning and has already been qualitatively discussed
in regard to the work of Marskell and Miller (94) and on pages
192-198 .  The inert content at the top of the~^ed insulated
the burning zone from the radiant heat flux from the overfire
flame and the hot refractory brickwork of the top section and
helped occlude the oxygen present in the overfire section
from the combustible in the bed.  Presumably gas phase reac-
tions took place in the void spaces in this "dead zone" of
the bed; for instance, the oxygen in the underfire air could
have reacted with CO, H9, and pyrolysis products in this region
of the bed.            ^

No experimental determinations were made of the depth of either
the active fuel bed or the accumulated inert.  However, an
approximate calculation was used to provide an estimate of
the depth of the active burning zone and the rate of accumu-
lation of inert above this zone; these calculations will be
discussed below.  The depth of the active fuel bed can only
be calculated reasonable accurately if it is assumed that no
volume change of the fuel particles takes place during the
time required to completely dry and pyrolyze them; and if
it is possible to estimate from the fuel bed gas composition
what fraction of their carbon content can be attributed to
char combustion.  For a fuel, such as that used in this study
with a large volatile content (- 75%) in comparison to fixed
char content (= 25%); this latter estimation is particularly
difficult.  Paced with these difficulties more approximate
methods were employed.

Figures 62 and 63 show the calculated depth of the active
                        224 -

-------
NJ
ro
U
\-
<
o:
(D


LJ
>
o
CO
          LJ
          U
          Ul

          Q
                                                     Depth  of Accumulated
                                                          Inert
Location  of
   ition Front
                                                          Location  of  Top

                                                              of  Bed
                       Active  Burning

                           Depth
     0.8 —
              0.4 —
                0
                              1000
                                  2000         3000

                               TIME ( SEC.)
                                                   4000
                 Fig. 62.  Estimate of Fuel Bed Thickness and Accumulated Inert Layer
                           (Run 17)

-------
to
K>
           UJ
           I-
           <
           cr
LU
O
CD
           LJ
           U
           Z
           <

           —
           Q
                                                    I       I

                                        Location  of Top  of Bed
                    Location  of
                  — Ignition Front
                                                 th  of  Accumulated
                                                     n
-------
burning zone and the depth of accumulated  inert over the course
of Runs 17 and 18, respectively.  For these calculations it
was assumed that the fuel maintained its original density  (po)
until it was completely consumed, at which time its density
dropped to that of the inert material  (pj  - 9  Ib ft"3) .  For
the case where the density of the active fuel bed varies
linearly from pQ to pj over the depth of active burning zone
and where the inert fraction within this zone  is constant,
active burning depths about 30% greater than those shown in
Figures 62 and 63 would have been calculated.  For an  alter-
native case, where the inert fraction  (X^) changes as  the
density changes, if it were assumed that the desnity of the non-
inerts  (i.e., wood blocks) remains unchanged during the course
of combustion, the calculated active burning depths would
have been about twice those shown in Figures 62 and 63.  For
this  case the depth of the inert layer  (where xj - 1) would
be much less than those indicated for early times in Figure
62 and 63.  These models are expected to provide reasonable
bounds on the active burning depth.

Figures 62 and 63 indicate how the depth of the active burning
layer changed with variations in the burning and ignition
rates.  Figure 62 indicates that because of the varying igni-
tion rate, the depth of the active fuel bed changed markedly
during the course of the experiment.

For Run 18, Figure 63 shows that the active fuel bed depth
changed gradually over the course of the test? the depth
reached a maximum at the time the burning  rate increased.
In addition, a comparison between Figures  62 and 63 indicates
that the active fuel bed depth in Run 17 was greater than
that for Run 18.  This result coupled with the fact that the
underfire air rate used in Run 17 was greater  than that used
in Run 18 suggests that the burning rate for Run 17 should
have been greater than that for Run 18.  The observed  aver-
age burning rate for Run 17  (39 Ib combustible hr-1ft~2)
was actually lower than that observed in the later stages of
Run 18  (41 Ib combustible hr-lft~2).  This apparent anomaly
was caused by channeling in Run 17 and the burning rates
based on the gas measurements at the centerof  the bed  for
this test were rather higher (53 Ib combustible hr~lft~2)
than the average burning rate.

Figures 64 and 65 present ignition rates and burning rates
obtained from both the load cell and material  balances  (see
pages 192-198) for Runs 17 and 18 respectively-  The varying
depth of the burning zone can be deduced from  the different
areas under the burning and ignition rate  curves.

The increase in the burning rate in Runs 10 and 18 cannot be
                        -  227  -

-------
K)
K)
00
                   80
                    40
                ICC
                I
                HI
                _J
                UJ
                    30
o:

O
z
z   2O
cr
ID
CD

LL
O
                    10 -
           Burning
           Rate
         (Load Cell )
                                              Burning  Rate
                                              ( Material Balance)
                                                       I
                                  10OO
                               20OO         30OO

                              TIME  ( SEC)
                                                           1      ^1
40OO
                           Fig.  64.   Burning and  Ignition Rates  (Run 17)

-------
KJ
NJ
VO
             o
             (D
                                                                  Burning Rate
                                                                   Load  Cell)
 Burning Rate
/(Material  Balance)
                              1O 00         2OOO         3OOO

                                             TIME ( SEC.)

                        Fig. 65.  Burning and Ignition Rates (Run 18)
    4OOO
5OOO

-------
explained strictly by the increase in fuel bed temperature
observed towards the end of each run, as originally proposed
by Rogers, Sarofim and Howard  (140)> since the increase in
temperature always followed the increase in burning rate.
No satisfactory explanation can be offered at this time to
explain these results,,  It is likely that the degree of chan-
neling changes as the bed burns and      will give rise to
different burning rates.  Local changes in the way the fuel
and inert were charged to the bed may also have affected the
degree of channeling.  This effect was probably aggravated
by the larger size of the inert with respect to the fuel but
could not be readily altered as the tin cans used were the
closest to the fuel in size that were easily obtainable in
bulk quantities.  However, these effects would need to be
very non-linear to explain the magnitude of the changes in
burning rate observed.  Although they probably did affect
the burning rate to a certain extent it is very unlikely
that they were the main cause of the observed increase.  It
is also possible that the burning rates changed because of
changes in the rate of heat loss from the bed (e.g., by accu-
mulation of inert on top of the bed) but this cannot be exper-
imentally proven.
                    Material Balances
Complete material balances for Runs 17 and 18 are given in
Tables 13 and 14.  The symbols used in Tables 13 and 14 have
the^following meanings:  F = total fuel burning rate, Ib
hr  ; FC = carbon burning rate, Ib hr"1; FH = hydrogen burning
rate, Ib hr"1; FO = oxygen burning rate, Ib hr~l; ALPHA =
Ib carbon/lb oxygen burnt;  BETA = Ib hydrogen/lb oxygen
burnt;  GAMMA = Ib carbon/lb hydrogen burnt;  G = stack gas
flow rate, Ib moles hr"1; LEAK = air leakage rate, Ib moles
hr"1; SUM = sum of CO., and H20 concentrations within the
stack gases, mole fraction; C = cumulative carbon loss, Ib;
0 = cumulative oxygen loss, Ib; H = cumulative hydrogen loss,
Ib; TLOSS = total cumulative loss, Ib.

Tables 13 and 14 show good agreement on the overall material
balance as indicated by the calculated carbon, hydrogen, and
oxygen loss compared to the know weights of these materials
in the fuel.  The largest discrepancy is shown in the carbon
balance for Run 17, and can be explained by either loss of
carbon as particulates in the stack gases or by incomplete
combustion.  The former is very unlikely considering previous
                     - 230  -

-------
Table  13.    Material Balance  Calculations   (Run  17)
                       BURNING SATES AND INTEGRATED HEIGHT  LOSt FOR  RUN   17
TOTAL FUEL CHARGED • 62.J.30     WT. FRACTION MOISTURE  . 0.2517  WT. FRACTION INERT • 0.1516
A^BUNT CONDITIONS.    Y02A » 0.20T6  YN2A . 0.7008 YH20A • 0.0115
INTEGRATION STARTED 200.0 SECONDS INTO RU«4
TIHF
200.0
100.0
400.0
900.0
600.0
700.0
800.0
900.0
1000.0
1100.0
1200.0
1100.0
1400.0
110C.O
1150.0
17IC.O
1*00.0
1-3C.C
?w"»P.r
21T. )
22C PA'IO •  0.114
 OLCIILATFT  FllFL C/1 PATIO .  0.180
ACTUAL FUEL  C/H RATIO •  4.793
ACTUAL FUEL  H/0 RATIO >  0.112
ACTUAL FUEL  C/0 RATIO •  0.635
 BU»NING RATF OVER CONSTANT  BURNING PERIOD IFROM 1300.0 TO 3700.0 SECONDS! • 62.6*7 LBS/HR
                                 -  231  -

-------
Table   14.    Material  Balance  Calculations   (Run  18)
                        OURMIN6 RATES MO JMTE6RATEO XC16MT  LOSS FO* RUN   II


TOTAL FUEL CHARGED • 66.900     BT.FRACTION MOISTURE • 0.2306   WT.FRACTION INERT • O.UOJ
AMBIENT CONDITIONS    Y02A ° 0.2099   YN2A » 0.77*7  YH20A • 0.0194
INTEGRATION STARTED 100.0 SECONDS INTO RUN
TIME
100.0
200.0
300.0
400.0
300.0
600.0
700.0
600.0
900.0
1000.0
1100.0
1200. C
1300.0
1400.0
! f 30 .0
iuoo.o
noo. )
1FOO.O
19PO.O
7000.0
2 100.0
27PO.P
2300.0
24HP.P
2100.0
?600.P
7700.0
2S1C.C
2900. C
icro.o
nor.c
32"0.0
'33P.;:
3400.?
? ^19.1
1n">P.O
1'OO.C
1H30O
?93C .'J
4010.0
4 i?p . p
42or.c
43CC.O
4400.0
43CO.C
4630.0
4700.0
T

49.441
30.923
34.721
96.90*
96.367
37.866
32.400
S3.B14
90.4R9
34.1)4
53.288
33.247
32.906
34.086
53.332
34.63B
6C.170
64.964
63.271
66.786
69.349
72.124
77.729
77.921
71.P75
68.432
70.326
66.364
66.872
63.842
93.904
48.02C
43.291
34.73B
23.392
12.597
9.216
7.269
3.84P
4.908
3.196
2.112
1.073
2.738
2.714
2.692
FC
11.370
14.273
16.695
16.939
16.612
17.707
17.369
17.606
17.330
19.99))
16. 4U,
16.346
15.890
15.776
15.988
16.87.!)

19.269
20.394
71.677
27.874
2S.770
24.904
24.S79
24.541
75.137
24.709
?5.40»
25.646
25.374
24.491
21.810
27.. 310
21.303
18.539
15.097
11.60?
".908
7.21S
.638
.629
.367
.736
.724
.412
.376
.398
FM FO
2.4Si« 25. 330
3.JM 32. OSS
S.»S2 30.975
3.487 S*oO'»S
4.007 34.234
4«:80 34.479
4.0'Ji 30.204
S.835 30.90B
1.849 10.614
3.717 30.779
S.46S 3'J. OSS
jcSOO 33.359
1.629 S3.7JS
3.7'>1 )3.389
3. 964 34.113
4.234 1J,S'J3
4.6S3 31.934
4.S16 3&.085
3.076 39.333
5»")63 38.230
3»98
0.3323
0.49VO
0.342B
0.31)3
0.4S92
0,5691,
1.1672
0.5194
0. 4619
0.4900
0.4711
0.4724
0.4636
0.5200
0.3776
0.5339
0.3223
0.5471
0.6026
Oo699*>
0.9963
0.3930
0.977)
0.6093
0.6463
0.6446
0.7263
0.6342
0.7075
0.7729
0.9S43
1.1140
1.3223
2.21&2
3D. 9133
-6.6722
•73.2(33
4.8793
3.1640
3.3037
5.4341
1.3117
1.0781
1.0997
1.0*42
BETA
0*0973
0.0963
0. 1 il*»
0.1093
0.1U6
0.1213
o.usp
0.1237
0.1263
0.1507
0.1076
C.1073
0.1079
0.1120
0.1166
0.1301
0.1393
0.1S34
0.1290
0.1403
0.1440
0.1327
0.1310
0.12Q6
0.1296
0.1321
0.1437
0.1396
0.1331
3.U07
0.1376
0.1391
0.1632
0.1476
C.1336
O.J779
3.3397
-0.4619
-1.1&66
0.2635
0.1236
0.1206
0.1377
0.0269
0.0266
0.02C.S
0.0339
GAMMA
4.6108
4.3237
4.7781
4.6075
4.6*42
4.2361
4.2922
4.3314
9.3610
4.30U6
4.&7R 1
4.5663
4.3793
4.2167
4.0123
4.0021
4.1463
4.0305
4.0491
4.0413
4.1834
4.S169
4.350V
4.6399
4.4403
4.6127
4.4973
4.6174
4.7417
4.9633
9.1417
3.9339
6.0293
7.5476
8.4986
9.7241
11.6316
13.8444
J9.9600
19.3119
2J. 7699
27.409)
50.6122
*§.7447
40.9027
40.S220
40.2269
G
29.64)3
30.4917
30.9269
31.30)2
31.64*3
12.16*)
32.3129
32.4762
31.6*09
32.7)6)
32.8233
32.6394
32.3U02
32.3370
32.182B
31.8692
11.7030
31.6493
31.6*92
31. 4674
31.3216
31.2200
31.3332
31.2698
31.3047
31.2073
11.1243
31.1636
31.0118
30.9VO)
30.6308
30.6*38
30.1676
29.7015
29.2234
28.3846
27.6922
27.3621
26.8902
26.4047
29.7669
23.4042
24.6877
34.4*67
24.0291
23.411)
23.0970
LEAK
10.3*1*
10.9619
11.9972
11.6169
12.072*
17.3631
li.6996
12.840)
13.0212
13.1*6*
13.1*64
1J.02S2
12.840)
12.6396
13.6120
14.49U1
14.2207
13.9389
13.772)
13.7932
14.6110
14.9974
14.9039
14.4307
14.4907
14.3976
14.3976
14.3976
14.3976
14.1976
14.4507
14.4907
14.3976
14.2)96
14.0634
13.8268
13.4793
13.0643
12.6110
12.1938
11.9396
11.201)
10.7104
10.2182
9.820*
9.201*
6.869*
SUM C
0.211 0.197
0.219 0.914
0.211 0.9*1
0.212 1.406
0.210 1.9UU
0.210 2.40!.
0.212 <6Vi
0.209 .38)
0.208 .669
0.210 .132
0.212 .782
0.212 .217
0.21) .689
0.212 .129
0.210 .366
0.206 .-122
U.206 .311
0.207 .03*
0.207 .307
0.20* .17)
0.20) .792
0.201 10.440
0.207 11.116
3.207 11.808
0.20B 12.494
0.207 13.U4
3.20) 13.876
0.204 14.57}
0.202 19.282
0.203 13.VU6
0.206 16.674
0.206 17.149
0.203 17.981
0.206 18.981
0.206 19.136
0.203 19.60)
0.201 19.97)
0.203 20.298
0.236 20.412
0.206 20.6*7
0.209 20.762
0.206 20.8*9
0.203 20. 9J)
0.210 20.931
0.210 20.999
0.210 21.0)1
0.210 21*071
0 H TLOSS
0.391 0.01* 0.3*)
1.1*6 0.112 1.77)
2.021 0.20* 3.167
2.92* 0.303 0.61*
3.871 0.41U 6.184
4.826 0.32) 7.796
9.810 0.6)8 V.)**
6.742 0.7*9 10.676
7.17* 0.82V 12.07}
8.00} U.9U6 11.2**
6.909 1.009 14.697
9.9*2
10.77)
11.706
12.6*)
11.566
14.436
19.400
16.447
17.323
16.S6)
19.610
20.691
21.894
2). «27
24.190
23.294
26.572
27.410
28.40V
29.399
JO. 307
31.0*3
31.622
32.002
32.371
)2.*6V
32.433
32.4)2
32.441
12.470
12.496
12.310
)2.3))
12.370
.109 16.169
.209 17.669
•112 19.1*3
.419 20.629
.9)) 22.121
.69) 23.621
.761 23.216
.919 26.V14
.06* 29.763
.21* 30.990
«)»6 32.418
•31V 34.327
•670 )6.))2
.621 )».)4)
.972 *0.)*7
.12* 42.295
.277 44.22)
.429 46.121
.376 *7.»72
.713 49.767
.8)9 31.492
.9*9 32.V77
.0)9 3*. 2*3
.109 35.326
.160 36.1)9
. 1V6 56.»)V
.218 »t.9)3
.232 97.1*6
.241 97.3)0
.246 97.479
.250 97.992
.131 37.663
.2*2 97.7)7
.21) 37.816
12.606 4.234 37.894
12.6*2 4.233 37.970
 IHTfr.RATFD C«SBOM LOSS   » 21.071 LBS
 I'!TF.F«ATfP HYOROGFM LOSS  •  4.253 LBS
 INTEGRATFC OXYiE'i LOSS   • 32.642 LSS
 I'iTEGRATEP '.vEISHT LOSS
                         47.970 LBS
 THEORETICAL CARBON CHARGED   • 20.439 LBS
 THEORETICAL HYDROGEN CHARGED •  4.226 LBS
 THEORETICAL OXYGEN CHARGED   • 3U6*o LBS
                                        TOTAL  COMHuSTlBLE  CHARGED
                                                                   •  36.499 LBS
 CALCULATED FUFL C/H <5aT10  «  4.931
 fALCULATFP FUEL H/0 RATIO  •  0.130
 CALCULATED FUFL C/0 RATIO  •  0.643

 PUNNING RATE  OVER CON5TA1T BURNING PERIOD

 1U9NINC, RATE  OVER CONSTANT BURNING PERIOD
 ACTUAL FUEL C/H RATIO •  4.8)9
 ACTUAL FUEL H/O RATIO •  0.1)2
 ACTUAL FUEL C/0 RATIO •  0.6*1

(FROM  100.0 TO  1700.0 SECONDS) • 51.4)9 LBS/Hd

(FROM 1700.0 TO  3200.0 SECONDS! • 66.749 LB4/HR
                                   -  232  -

-------
experience with the  smoke meter  (not  used  in  this  test)  which
showed only trace participates in  the stack gases.   The  latter
explanation is more  likely  and is  confirmed by  the weight  loss
plot of Figure 59, which shows a final weight loss of  close
to 51 Ib.  The incomplete combustion  can be explained  by the
fact that small pieces of wood tended to fall inside the tin
cans, where they burnt slowly and  were incompletely burnt
by the end of the run when  fuel bed temperatures fell  low
enough for the carbon-oxygen reaction to be quenched.

The accumulative carbon, hydrogen  and oxygen  losses and  their
sum, TLOSS, for Runs 17 and 18 are plotted in Figures  66
and 67.  Comparing the values of TLOSS versus the  load cell
data of Figures 59 and 60 gives an independent  check on  the
accuracy of the weight loss determination; the  agreement be-
tween the material balance calculations and the load cell
data was very close  for both these experiments.  For Run 17
the average burning  rate over the  constant burning  period
(1300 to 3700 sec) calculated from the material balance  is
38 Ib hr'-'-ft" , which is within 3  percent  of  the load cell
result of 39 Ib hr-1ft~2»  For Run 18 two  constant  burning
rate periods were observed, one lasting from  about  100 sec
to 1700 sec into the run and the other from 1700 sec to  3200
sec.  For the former constant burning,period, the  load cell
data gave a burning  rate of 30 Ib  hr   ft~2 and  the  material
balance 31 Ib hr~1ft~2 while for the  latter the load cell
data gave a burning  rate of 41 Ib  hr~lft~2 and  the  material
balance 40 Ib hr  ft

The material balance calculations  were sensitive to the  O
and CO2 stack concentrations because  they  involved  the compu-
tation of differences between large numbers;  this  sensitivity
was most marked near the beginning and end of a test.  The
anomalous behavior indicated in both  Tables 13  and  14 near
the end of Runs 17 and 18 can be attributed to  inaccurate
data on these concentrations.  Small  changes  in the measur-
ing instrument calibration could have accounted for these
errors.  A check on  the O.., and CC>2 compositions measured in
the stack was given  by the SUM column;  for the  complete  com-
bustion of cellulose to CO, and H-0 the mole  fraction of
these two components in the stack  should have equaled the
mole fraction of O2  in the combustion air  (0.21).   For all
the points taken for the material  balance  this  was  found; the
worst  point at 2200 sec in Run 18 being less than  5 percent
in error.  (Actually, the fuel used contained net hydrogen so
that the value of SUM should have  been slightly less than
0.21.)

Figure 68 shows the  values of the  ratio of hydrogen to oxy-
gen (BETA) and the ratio of carbon to hydrogen  (GAMMA) in
                       -  233  -

-------
to
u>
    CO
    CO
    (A
    to
    O
'    i-
    I
    O
    u
    £
                                           HYDROGEN  LOSS
                     TOTAL WEIGHT  LOSS
         0
                         1OOO
                                                                        4OOO
                               2000           3000

                       ELAPSED  TIME (SEC)

Fig.  66.  Accumulated Carbon, Hydrogen and Oxygen Weight Loss of Fuel (Run 17)

-------
CO
OJ
en
                                            HYDROGEN
                                            LOSS
                                                                 CARBON  LOSS
           TOTAL
           WEIGHT
           LOSS
                                                       OXYGEN  LOSS
    LJ
                                                                         4OOO
              1000           2OOO           30OO

                              TIME  (SECONDS)

Fig.  67.  Accumulated Carbon, Hydrogen  and Oxygen Weight Loss of Fuel  (Run 18)

-------
to
u>

Q
LJ
0 ID
|8
_j
U LL.
^U
15
10
5
0
	 ^ y Ignition Front |
— —18 at Grate Runs |
~ 17 and 18 — |
vC/H Ratio of Fuel (4.8)
X^ "'' ''
^/ >•'' '
s'
I I I I I I I I I
            O
            P
            O

            I b-
Q   O.3
u
                  0.2
                   0.1
     0
                            • 17

                            •18
              ,O/H  Ratio  of  Fuel (  0.13)
                 I
I
1
                     0
   0.2          0.4          0.6          0.8

FRACTION  OF COMBUSTIBLE  CONSUMED
                      Fig. 68.  Hydrogen/Oxygen and Carbon/Hydrogen Ratio of Fuel
                                Burnt as a Function of Fraction of Combustible
                                Consumed (Runs 17 and 18)
                                                                      .•I

                                                                      I
                                                                     1.O

-------
the fuel burnt for Runs  17 and  18;  in this  figure the regions
where there was the most uncertainty in the calculated values
are shown as dashed lines.  For Run 18 some anomalous values
of the H/O ratio were calculated  for the times between 3700
and 3900 sec; these inconsistencies were probably caused by
incorrect stack gas composition measurements.

The constancy of the H/O ratio  over the major portion of
these experiments confirms the  view that drying and pyrolysis
occur over protracted time periods.
                     Gas Compositions
Fuel Bed Gas Compositions.   Representative  compositions of
the gases withdrawn  from the center of  the  bed  at  two posi-
tions  20 in. and  12  in  above the  grate  are  shown,  for Run 17,
in Figures  69  and 70, and for Run 18 in Figures 71 and 72.
Measurements of fuel bed gas compositions were  taken, as des-
cribed on pages 152-184, on an intermittent basis.   The smooth
curves joining the data points must therefore be considered
only to represent the average trend in  gas  compositions with
time.

The bias introduced  by  radial concentration gradients was
checked for a  number of runs by comparing the carbon contents
of the gases at the  sample points with  those in the stack.
For the gas samples  from Run 18 there was little bias, but
for other runs where larger quantities  of underfire air were
used,  the degree  of  channeling increased; this  has been fully
discussed on pages 213-230.

Figures 73, 74 and 75 show the smooth gas composition curves
replotted versus  distance above the ignition front. These
plots  were  obtained  in  a manner analogous to that  used in cal-
culating the temperature profiles below the ignition front
and the reader is referred to on  pages  206-208  for a more com-
plete  description of the method used.  The  gas  compositions
measured during Run  17  from Fuel  Bed Probe  2 were  not plot-
ted in the  above  manner because of the  incompleteness  of the
data on H2  and CH4 concentrations.  (For Figures 70, 71 and
72, note the Fuel Bed Probe 1 and Fuel  Bed  Probe 2 were 20
in.  (1.67 ft)  and 12 in. (1 ft) above the grate; this was
the maximum that  these  probes could be  above the ignition
front.)
                         - 237 -

-------
to
U)
CO
                 in
                 CD
                 cc
                 Q
                o
                o
                o
                CL
                2
                O
                U
           I    1
Time Ignition Front

Reached Probe
                                                  at Grate     "*|
                     0
                     1400
     2000
       30OO

TIME (SEC)
400O
                      Fig. 69.  Gas Compositions from Fuel Bed Probe 1 (Run 17)

-------
(0
u>
vo
                    (f)
                    CO
                    CC
                    Q
                    o
                    >
                    Z
                    O
                    cr
                    u
                    u
                    z
                    O
                    U
I    I    I    I    I
 Time Ignition Front
 Reached Probe
              4
  Ignition  Front
f at  Grate
                        8  —
                        O
                          2000
           3000

          TIME (SEC)
        40OO
4500
                          Fig.  70.   Gas Compositions from Fuel Bed Probe 2 (Run 17)

-------
CO
CO
<
CO
o
>
CO
O
CL
2
O
u
    2O P"**^  Time Ignition Front
            \xReachcd Probe
     16 -
       1000
                         Ignition Front     S_
                          at Grate       •
2000             3000

   ELAPSED   TIME (SEC)
4OOO
                                                CO
                                             2  2
                                             1   3
                                                U
       Fig. 71.  Gas Compositions from Fuel Bed Probe  1 (Run  18)

-------
                                                           C/)
CD
o:
Q
Z
O
to
O
CL


O
u
4  -
0
                                       Ignition Front

                                       at  Grate   f
      Time Ignition Front

      '"Reached  Probe
  2000
   Fig. 72
                                                           or
                                                           O
                                                           O
                                                           >
                          30OO              4000


                      ELAPSED   TIME(SEC)


                 Gas Compositions from Fuel Bed Probe 2  (Run 18)
0  O
   u

-------
V
ft.
          (ft
          <
          CD
          o:
          Q
          z
          Q

          *
          oc.
          UJ
          o
          z
          o
          u
               0
                 O
O.2
 O.4      0.6     Q8      1;O      1.2      1.4

DISTANCE  ABOVE IGNITION  FRONT (FT)
                 Fig. 73.  Gas Compositions as a Function of Distance  Above Ignition
                           Front (Run 17)

-------
I

NJ
CD
Q
O
O
H
<
or
K
Z
u
u
z
O
u
                O.2      0.4     0.6     0.8      1.0     1.2      1.4

                 DISTANCE  ABOVE IGNITION  FRONT (FT.)
                                                                                   1.6
         Fig.  74.   Gas Compositions as a Function of  Distance Above Ignition
                   Front  (Run 18)

-------
<
or
Q
o
z
O
o:
8-
Z
UJ
u
z
o
u
                 0.2
   0.6     0.8     1.0


8GNITION  FRONT (FT)
         Fig.  75.  Gas Compositions as a Function of

                  Distance Above Ignition Front  (Run 18)

                  (Calculated using ignition velocity =

                  0.083 ft sec  ); fuel bed probe located
                  12.n. above grate.
                      - 244 -

-------
For2these runs where low underfire air rates  (150 Ib fer
ft  ) were employed, the oxygen concentration fell rapidly to
zero as the ignition plane passed the probe and typically was
consumed in a distance of 0.25 - 0.8 ft.  The point at 2300
sec in Figure 68 appears to be spurious as there was no de-
crease in combustibles  (CO and H2) at this time.  The decay
of the oxygen concentration with distance below the ignition
front for Runs 15, 17 and 18, where fairly complete records
were available, is shown in Figure 76.  Some of the variation
of the shape of these profiles can be attributed to inaccur-
ate determination of the time the ignition front reached the
sampling probe.  In addition, local variations of tfte igni-
tion velocity could have affected the shape of the profile
as the distance-time conversion was performed using a con-
stant average ignition velocity.

As the O, concentration decreased,the concentrations of CO2,
CO, H2 and CH  increased.  The increase in C02 concentration
always led the increases in concentration of the other gases;
the same behavior was observed by Nicholls (24).  Peak concen-
trations in both C02 and CH. were observed near the point of
oxygen extinction, which aglin parallels Nicholls1 (24) ex-
perimental results.  Peak concentration of CO- were E^pically
16-20%; while those for CH4 were 3-4%.

After all of the O« had been consumed, the conditions within
the fuel bed appeared to differ between experiments.  From
the earlier discussions on the various processes occurring
within a fuel bed, it would be postulated that due to the Bou-
douard reaction (C + C0~ -*• 2CO) the C02 concentration would
decrease, while the CO concentration would increase with dis-
tance from the plane of zero oxygen concentration.  Similarly
the H_ and CO concentrations would be expected to increase,
at the expense of the H2O because of the occurrence of the
reaction C + H,0 -»• H2 + CO.  In addition, if the assumption of
the equilibration of the water-gas-shift reaction (except in
the vicinity of the ignition front) is correct, increases in
the CO concentration should be followed by increases in the H2
concentration and vice versa.

The above behavior was exhibited in Run 18 (Figures 71 and 72)
with the exception of the H? concentration measured with Fuel
Bed Probe 1, at times greater than 2800 sec.   (Figure 71 shows
that, after this time, the H2 concentration decreased while
the CO concentration increased.)  However, the exact location
of the H2 concentration profile was not known because of the
paucity of the data for this species and it is possible that
the H2 concentration profile could have followed a similar
shape to the CO profile.  The gas compositions obtained from
Run 17 exhibited behavior different from that observed in Run
18, as the CO and H2 concentrations decreased with tine (or,
                      - 245 -

-------
I

to

c\

I
z
UJ
u
o:
UJ
O
O
Z
O
O
LU
O
>-
X
O
            o
        0      0.1      0.2      0.3     0.4     0.5      0.6     0.7     0.8


             DISTANCE   BELOW   IGNITION  FRONT  (FT)

        Fig. 76.  Oxygen  Concentration as a Function  of Distance Below Ignition Front

-------
equivalently, distance above the ignition front).  This var-
iation of gas composition cannot be explained by bed tempera-
ture changes as these temperatures increased from 2000 sec
to 2800 sec by about 200°F and would have favored the forma-
tion of higher concentrations of CO and H20.  The decrease
in CO and H2 in this run could be explained by mixing of by-
pass oxygen back into the core of the bed; at the higher tem-
peratures which prevailed in this region, the oxygen would
have reacted rapidly with the combustible and no oxygen would
have been observed.  The high amount of bypass oxygen that
apparently channeled up the side wall during this run (see
pages 213-230) makes this explanation plausible.  Further
confirmation of this hypothesis is supplied by the lower
than normal CO and H2 concentrations measured at position
12 in. above the grate.  The CH4 concentrations measured with
this probe were also very small  (<0.1%).

For both Runs 17 and 18, as well as others, the CO2 concen-
tration within the bed fell off quite sharply near the end
of active combustion  (3800 sec for Run 17 and 3600 sec for Run
18 — see Figures 59 and 60, which show that the major por-
tion of the total weight loss was achieved by these times).
Similar results were obtained from the stack gas measure-
ments for these two experiments; this is discussed on pages
248-253 •  The rapid decrease in CO-> at the time when bed
burnout is nearly complete may provide a relatively simple
criterion for the burnout of the fuel bed on a travling grate
incinerator as long as there is no gross back-mixing of the
overfire combustion gas at the discharge end of the grate.

The gas compositions obtained from Fuel Bed Probe 2 in Run
18 (Figure 72) indicated that, near the end of the run, the
C02 concentration started to increase as the CO concentra-
tion began to decrease.  The CO2 concentration continued to
increase up to the time the CO concentration fell to zero
at which point it started to fall.

From Figure 63 it is seen that even with a generous estimate
for the depth of the active bed, the gas sampling probe was
probably well above the top of the bed at the time the CO
concentration started to decrease.  This observation may be
explained by the presence of a diffusion flame above the bed;
the fuel to the flame was supplied by the combustible rising
from the top of the bed and the oxidant was supplied by the
oxygen from the overfire region.  As the top of the bed re-
ceded from the probe, samples were first taken from the fuel
regime of the diffusion flame, and then as the probe passed
                           -247-

-------
through the flame envelope from the oxidant side of the flame.
The turbulence above the bed would make the location of the
flame envelope ill-defined.  Similar behavior would have been
expected from the other gas compositions measured, but the
intermittent sampling procedure may have obscured this trend.

Tables 15 and 16 (a) and (b)  show the calculated ratio of
the product of the CO and H20 concentrations to the product
of the C02 and H2 concentrations within the fuel bed at var-
ious times during Runs 17 and 18.  For both these runs, batch
samples were taken from the fuel bed probes and analyzed for
H2.  The H20 content within the bed was back-calculated from
the measured t^O content in the stack gases.  Table 15 shows
that the water gas shift equilibrium was closely followed
when oxygen began to penetrate through the bed  (at 3760 sec).
The two points, at 2080 sec and 2240 sec, where the calculated
values are lower than would be predicted from equilibration
of the water gas shift reactions could well be caused by inter-
ference from pyrolysis products  (see Figure 66 or 70).  Similar
results were obtained from the two different sampling posi-
tions in Run 18, as shown in Table 16 (a) and (b) for two dif-
ferent locations within the fuel bed.

From the method of calculating the H20 content in the bed the
estimates of these concentrations were probably higher than
those representative of conditions at the top of the bed for
those cases where the sample was taken some distance from the
top of the bed? this the estimate of (CO)(H2O)/(CO2)(H2) at
these positions would have been higher than that observed at
the top of the bed.  These results although they must be con-
sidered of a preliminary nature tend to support the theore-
tical model of Niessen et al. (19) (see pages 106-112), and
give a rough guideline for determining overfire air require-
ments.

Stack Gas Compositions.  Figures 77 and 78 show the stack
gas compositions for Runs 17 and 18.  Although the burning
rate was constant over the major portion of Run 17, there was
some variation in the stack gas concentrations.  This was
caused by the varying air leakage into the top section.  The
air leakage would be expected to be proportional to the
temperature differential between the ambient and the inside
of the test incinerator.  Figure 77 shows that the air leak-
age reached a maximum at around 3000 sec; at which time the
inside refractory temperatures were at a maximum  (see Appendix
H, Figure H.4).  The line marked "Start of Integration" at
200 sec in Figure 77 indicates the time when the first mater-
ial balance was calculated — see pages 230-237^t
                           - 248 -

-------
                                                          TABLE 15
Clieck  oa • Water -XJaa- . JEa.ulJ.JJuJ.um .wl.t±Lin
vo
 I
                                           (Fuel Bed Probe  1,  20 in. above grate)
                                                                                      (.Run -17)
Elapsed
Time
(sec)
1570
1930
2080
2240
2540
2930
3350
3760
Temperature
(°F)
just at ignition
1700
1700
1750
1800
1900
1750
1800
0=2}
PJ
0.2
0.4
1.0
1.0
0.75
0.52
0.59
0.09
[H20]
[H2]
77.0
6.6
0.3
0.66
2.9
4.3
1.7
OB
M C«2°]
Cco J f HJ
(Experimental )
15.0
2.6
0.3
0.66
2.2
2.2
1.0
00
[co]fi2o]
[C°21 EH21
(Calculated assuming
water gas shift
equilibrium)
> 0.10
1.42
1.42
1.54
1.65
* 1.86
1.65
1.65
                                [A]  denotes concentration of A

-------
                         TABLK 16  (a)




Check on Water Gas Shift Equilibrium within Fuel Bed (Run 18)




            (Fuel Bed Probe 1, 20 in.  above grate)
Elapsed
Time
(sec)


1450
1600
1850
2830
3550

Temperature
(°F)


?
7
•v 1400
1500
1650

[col



0.1
0.4
0.7
1.0
0.57

[H20]
KT



90
59
4.7
1.2
2.1
[C0l fa O]
I J L 2 J

PJ [H2]
(Experimental )


9.0
24.0
3.3
1.2
1.2
[CO] [H20]

[C°J [»2]
(Calculated assuming
water gas shift
equilibrium)
—
—
0.8
1.0
1.3
                         TABLE  16 (b)




 Check on Water Gas Shift Equilibrium  within Fuel Bed (Run 18)




            (Fuel Bed Probe 2,  12  in.  above grate)
Elapsed
Time
(sec)



2550
3320

Temperature



1100-1500
1700

[CO]
Kl



0.74
2.0

[H20]
[H2]



4.1
1.8

[CO] [H20]
On 1 fu T
LC02] 1H2 J
(Experimental )


3.1
3.6

[CO] [H20]
[C02]H2]
(Calculated assuming
water gas shift
equilibrium)
~ 0.5
1.6
     [A]  denotes  concentration of  A
                   - 250  -

-------
NJ
cn
        CO
        CE
        Q
        Z
        O
        O
        o_
        2
        O
        u
                       Carbon   Dioxide
4 —
           0
                  400   800   12OO   1600   2000   2400  2800  320O   3600  4000

                                        TIME   IN  SECONDS

                          Fig. 77.   Stack Gas Compositions (Run  17)
                                                                          4400

-------
K)
Ol
K)
         to
         CD
        cr
        Q
        O
        z
        O
        CO
        O
        Q.
        2
        O
        u
             16
12
                   hUO
                     I
                    400   80O   12OO  1600  2000   2400  2800  3200  36OO  40OO  44OO  46OO

                                            TIME  ( SEC)
                             Fig.  78.   Stack Gas Compositions  (Run  18)

-------
   For Run 18 the temperatures inside the overfire section did
not change as markedly as those for Run 17; the leak rate was
23Qr?f?re relativelY constant during this test  (see pages
   ~^J7  ) and the CO-, O  and H.,0 concentrations did not
cnange appreciably during periods when the burning rate was
steady.  The increase in the CO9 concentration that was ob-
served around 1600-1800 sec was consistent with the increase
in the burning  rate observed at this time  (see Figure 60).

Figures  77 and  78 show the same rapid decrease of
tne concentrations of CO- towards the end of tne run as in-
dicated  by Figure 66 through 72.  For Run 17 a comparison be-
tween Figures 77 and 59 shows that at the point where the con-
centrations of  C02 and H0O began to fall  (3600 sec) about 95
percent  of the  combustible fuel content had been consumed.
Similar  results were obtained for Run 18.

For both Runs 17 and 18 the H70 concentration in the stack
decreased at the same point that the CO- concentration de-
creased.  This  result confirmed the hypothesis of protracted
drying and pyrolysis of the fuel.
                     Energy Balances
Differential energy balances taken at different times dur-
ing an experiment can provide  information on the total heat
release rate  (the sum of the heat release rates within and
above the bed) at those times, while an integral heat balance
taken over the duration of active burning can give an esti-
mate of the total heat released which can be checked against
that calculated from the known fuel composition and amount
consumed.  Attempts were made  to calculate both types of
balance but only integral heat balances could be calculated
with any certainty because of  the unsteady-state nature of the
combustion processes and the "thermal flywheel" provided
by the top section refractory  lining.  The difficulties en-
countered with the energy balance calculations will be dis-
cussed below.

Prior to the start of an experiment a steady temperature pro-
file was established in the refractory brickwork of thg over-
fire section; typically, inside temperatures were 2100 F and
outside temperatures 275°F.  At the start of an experiment,
a few seconds after the refractory shield was opened the tem-
perature dropped sharply (generally from 2100 F to 1700 F
in about 200 sec) as the energy stored in the refractory sec-
tion was transferred to heat    the fuel bed surface and the
                       -  253  -

-------
under fire and overfire air.  Stack gas temperatures typically
fell a few hundred degrees Fahrenheit during this "ignition"
period.  Once active burning was established the temperatures
of the refractories changed slowly to an "equilibrium" level.
For those runs where adjustments were made in the amount of
overfire air employed or where the burning rate varied,
the temperatures changed continually during the course of the
experiment.

For the purposes of calculating differential energy balances
it was important to be able to estimate the effective heat
sources and sinks associated with the changes in temperature
level of the refractories.  This was particularly important
at the beginning and end of a run if heat release rates were
to be determined for these times but no reasonable method for
calculating this effect was available.  Assuming a linear
temperature profile within the top section refractories.
the change in total energy content of this section per  F
change in the temperature of the inside surface AO/ATS was
calculated to be

                          - 90 Btu °F-1                  (176)
This equation can be used to calculate the integral heat
source or sink effects but clearly would give an erroneous
result under conditions of rapid heat transfer.  (When the
temperature of the top section dropped rapidly at the start
of a test, the temperature profile within the refractories
reached a maximum some distance from the inside surface.)
With the temperature measurements taken, there was no simple
method available which could be used to improve the calcula-
tion of the source sink effect of the refractories and there-
fore limited the use of differential energy balances.

The other difficulty encountered concerned the estimation of
the heat loss through the feul bed wall.  Heat losses through
the fuel bed wall were determined from the temperatures mea-
sured with the heat loss probes and from the theoretical cal-
culations of heat loss through the fuel bed wall (see Appen-
dix B) .  From Figure B.I the maximum heat loss (at a -short
time after the wall was exposed to the burning fuel) was
estimated to be of the order of 10 x 103 Btu hr~r per foot
of bed height, while at times greater than about 0.6 hr the
heat loss dropped to approximately 6 x 103 Btu hr"1 per foot
of bed heigiVc,,  From the heat loss probe3data for. Run 18
the heat losses were estimated as 8 x 10  Btu hr   per foot
of-jbed height. _^For calculational purposes a value of 8 x
10  Btu hr  ft   was chosen as representing a mean heat loss
                      - 254 -

-------
rate.  The heat  loss depended  on  the  amount  of wall  exposed
to the ignited solid.   From  the ignition  rate and  the  estimates
?lv®jj above, a mean-jheat  loss  for Run 18  was calculated  to
be 10 x  1CP Btu  hr~  .   This  value was considered to  be typi-
cal  for  the majority of experiments conducted in this  study.

For  simplicity,  heat losses  were  considered  constant through-
out  the  duration of a  run? estimates  of the  fuel bed heat
losses would be  most markedly  affected by this assumption but
the  effects proved to  be  fairly small in  relation  to the
other heat losses, so  this assumption was reasonable.  Heat
losses were considered to take place  in three places:  over-
fire region, top section  cooling  coils, and  fuel bed.  Heat
losses to the grate were  neglected because,  for the  run  in
which heat release rates  were  calculated,  the ignition front
reached  the grate at a time  close to  the  end of the  experi-
ment.  Heat losses from the  top section  (assumed to  include
the  refractory tray) were estimated using

                  2Lk  (T1  - T.°)

            Ql = -THTrfT^T-   BtU  hr                (177)

where L  was the  length of the  top section area  (416  ft) ; k
the  mean thermal conductivity  of  the  refractory brickwork S
(0.18 Btu hr-ift"1 °F) ; T^ and T° the inside and outside
refractory temperature respectively;  and r  and r,  the  inside
and  outside radius of the  overfire section respectively (r  =
0.75 ft; r1 = 1.3 ft).                                   °

The  heat loss to the top  section  cooling  plate was calculated
from the mass flow rate of the cooling water used  (450 Ib
hr   ) and the temperature rise (^ 30  F max.).  These heat
losses were therefore  about  14 x  103  Btu  hr~ and were roughly
constant over the duration of  a run and between different ex-
periments.

For  the  specific case  of  Run 18,  the, heat loss from  the  top
section, using equation (177)  with T^ = 1600UF and T"  =  275 F,
was  about 12 x 103 Btu hr"1...  Total Keat  losses for  this test
were therefore about 36 x 10  Btu hr"1.   For this run  the tem-
perature of the  top section  refractories  remained practical-
ly constant from 1800  to  3000  sec.  During this time the^aver-
age  rate of enthalpy loss via  the stack gas  was 256  x  10
Btu  hr   .  (This value  was calculated using  the molar  flow
rates determined from  the material balanc^p^ogram,  a  mean
flue gas specific heat  of 8.1  Btu Ib  mole   F *•, the  rgcord-
ed stack gas temperatures and  a datum temperature of 80  F.)
From the burning rate  of  68.5  Ib  combustible hr"1  (41  Ib
combustible hr-lff2)  the,average heating value of the com-
bustible was 4300 Btu  Ib  .
                      - 255 -

-------
The above calculation was repeated for times from 200-1600
sec.  In this case, however, the temperatures of the top sec-
tion fell an average of 200 F over this time period.  The
total heat generated from 200-1600 sec (82 x 10  Btu) was
computed by summing the integrated enthalpy losses to the stack,
fuel bed wall and cooling coils and subtracting the net en-
thalpy  change in the refractory brickwork given by
equation (176).  The total weight loss over this period was
approximately 20 lb, giving an average heating value of the
combustible of 4100 Btu Ib"1.  The calculated heating values
for the two different periods of the experiment agree closely.

Using a heating value for the dry wood of 8090 Btu lb   (cal-
culated from tfhe Du Long formula, AH = 14600C + 620QO (H -
0/8), Btu Ib"1) and subtracting from this the amount of ener-
gy required to vaporize any water.present (the heat of absorp-
tion term is small;  ^ 30 Btu lb~ ), the heating value of
the fuel used in this experiment was calculated to be 5450
Btu lb~l.  The agreement with the experimental values is
reasonable considering the approximate nature of the calcula-
tions.

Energy balances were not caluclated for Run 17 because of
the difficulties encountered in correcting for the changes
in temperature level of the refractories which occurred dur-
ing this experiment.
                     Overfire Air Regime
From the fuel bed gas concentrations shown in Figures 69
through 72, it is apparent that the oxygen required to com-
plete the combustion of the gases leaving the fuel bed varied
with operating conditions and with time.  For a traveling
grate incinerator this means that overfire air requirements
will vary with position along the grate as well as with opera-
ting conditions.

For the experiments conducted in this study the average over-
fire (or secondary) air requirements for the burning rates
observed could be readily evaluated from Figure 61.  However,
this calculation could not be performed on an a priori basis
because there were no theories available that could be used
to determine the achievable burning rate for a given under-
fire air rate.  For cases where the burning rate was known as
a function of underfire air rate (i.e., from experimental
                       -  256  -

-------
                                                      use in
observation) a plot similar to Figure 61 could be of u=c j.,,
 etermining the appropriate secondary air requirements.  For
example, this technique could be used to evaluate the secon-
t-h^ air re<3uirements for a traveling grate incinerator if
p   ^rra9e burning rate and the underfire air rate were known,
 orK     Particular example, however, some care would have
to be exercised, as the total oxygen feed rate through the
IS  K 5  would need to be known.  Oxygen can be supplied to
the bed both in the underfire air and in any overfire air
induced through the bed by temperature gradients; for the
latter case, the overfire air  (which has a lower mass fraction
of oxygen and a lower density than the underfire air) will
tend to sink down at the "cold" walls of the furnace and to
rise up through the "hot" core of the bed.  The oxygen in
this air would be expected to be rapidly consumed near the
edges of the bed.  The estimation of the quantity of air being
induced into the bed in this manner is difficult but a rough
estimate of the expected magnitude of this effect can be found
as follows.

                                                       9  9
If only the significant forces are those of momentum (U pL )
and buoyancy  (L3Apg), the entrainment velocity U is found
to be
                                                       (178)
where L  is  a  characteristic length given by the distance
from the top  of the bed to the roof of the incinerator;
Ap the density difference of the gas at the hot and cold
temperatures;  pc the density of the cold gas;  and g the
gravitational  constant.  The mass flux of oxygen through  a
unit area of  the bed by this natural draft (GNE))will be
where  (M^ )  is the mass  fraction  of  oxygen  in  the overfire
        °2 c
air.  The mass flux of oxygen  through a  unit of bed  in  the
underfire air  (GFD) is given by
                                                \

               dFD = VPa  (MoJa                        U80)
                      -  257  -

-------
where V is the superficial velocity of the underfire air and
p  and (M_ )   the density and mass fraction of oxygen in the
 a       On a
underfire air, respectively.

If the effect of the natural draft is to be negligible
and using p  =
           C
               P.T
                     and pH =
                          ri

                                __
                                     where T  , T  and T  are
                                            aw      n
the ambient cold Snd hot air and gas temperatures, gives
               V  (M
                   Q
                                                        (182)
Putting typical values into
0.5 ft/sec (- 120 Ib hr"1ft~
                             quation (177)  as follows
                                    >a = 0.23; (MQ ) c
                                                        V =
0.08;  Ta = 520°R;  TC = 2000°R;  TR = 2500°R, gives


                     0.12 » 0.053-y/TT1
                                                         (183)
For many furnaces, L, one of the driving forces for the
buoyancy term, is of the order of 10-20 ft.  Substituting
this value into equation  (183) indicates that the oxygen mass
flux induced by natural draft may well be greater than that
supplied in the underfire air.  The uncertainty associated
with estimating the unknown amount of air which is entrained
through the bed imposes a serious constraint upon the above
method of calculating the secondary air requirements.

The theoretical model of Niessen e_t al.  (19) may also be
used to determine secondary air recfuTFemervts if the average
burning rate and fuel composition have been determined.  This
model assumes that relatively small amounts of higher hydro-
carbon pyrolysis products are present in the gases at the top
of the bed and that these gases are equilibrated with res-
pect to the water gas shift reaction at the temperature of
the bed.  The results of this study indicated that this assump-
tion was relatively good for the simulated refuse used; re-
sults from full scale operating units  have also suggested that
this assumption holds for a real refuse  (149) .  Details of
the Niessen ejb al. (19) model are given on pages 106-112  .

For this study the mixing above the bed supplied  by the overfire
                      - 258 -

-------
}ets was sufficient to provide nearly complete combustion of
all combustibles.  Carbon monoxide was never detected in the
stack for any of the runs; for the later runs the limit of
detection was only about 0.1 - 0.2% but earlier runs where
the detection limit was about 0.05% also failed to show the
presence of CO.  Trace quantities of soot and smoke were re-
corded, but only during periods of smoldering was any smoke
visually noticeable.  Although this synthetic refuse was not
inherently a "smoky" fuel, these results are encouraging and
suggest that if, in operating units, adequate mixing is a-
chieved and high enough temperatures are maintained in the
overfire region, clean combustion of the volatiles and com-
bustible can be achieved.

As will be discussed below, for the very low degree of tur-
bulence encountered in all incinerators, and in the test in-
cinerator of this study, the rate of mixing of the combusti-
ble and oxygen is generally expected to provide the limiting
step in the combustion process.  The ensuing discussion fol-
lows that of Sarofim  (150).  From the studies (e.g., 151,
152, 153) in well-stirred reactors the following general con
clusions can be drawn:

     (a)  The kinetics of oxidation  for a large number of
          fuels are so fast that volumetric heat release
          rates of 106 to 108 Btu hr~1ft~3 are achievable
          over the temperature range 2000° to 3000°F.

     (b)  The reactions of the hydrocarbons and ketones stu-
          died are very fast relative to that of the carbon
          monoxide intermediate formed in the reactions.

     (c)  The rate of burnout of the carbon monoxide, which
          provides the rate limiting step, in the overall
          combustion process, is given by


- ^£ - 1.8 x lo".xp

          g/mole  sec                                    (184)

where f.^, fn , fH n are the mole fractions of CO, 02 and
       i—vJ   \_/ *}   ti«*-/                  ,~
H~0; T is the absolute temperature in  K, and (P/RT) is in
g moles/liter.  Using equation  (184) and representative values
of temperature, oxygen and water vapor concentrations found
under most incinerator conditions, the time required to re-
duce the carbon monoxide concentration to less than  .01 per-
cent is of the order of a millisecond.  Thus it can be con-
cluded that if the turbulence above the fuel bed is high
enough to provide perfect mixing and if there is no equilibrium
                      - 259 -

-------
constraint imposed by the reaction C02 + H2 -*• CO 4- H20 no
CO would be expected in the flue gases and complete reaction
of the pyrolysis products could be achieved.  Moreover,  the
very rapid rate of chemical reaction suggests that if the
mixing were greatly improved above the fuel bed, furnace  vol-
umes could be considerably reduced.

The rate of soot burn-out can be estimated from the expres-
sion for mass transfer and chemical reaction.  The rate  of
soot oxidation can be determined from the semi-empirical for-
mula proposed by Nagle and Strickland-Constable (154)to
correlate the measurements of the oxidation rate of polycry-
stalline pyrographite.  Appleton (155) has summarized various
experimental evidence on surface structure to support the
assumption that the kinetics of soot oxidation are similar
to those for pyrographite.  Park and Appleton (156) have pro-
vided experimental evidence that at high temperatures
 (> 1500 K) the rate of soot oxidation is closely predicted
by the theory for the rate of oxidation of pyrolyzable graph-
ite.  Assuming that the Nagle and Strickland-Constable re-
lationship holds for soot oxidation at the low temperatures
typical of incinerator operation, the rate of soot oxidation
can be given approximately by:
           K
            chem
= 12
                       20 exp  j-15,000/TJ;
                         02
      1 -!- 21.3 exp|2060/T[P0
                                             '2J
    -2    -1
g cm   sec

        (185)
where T is in  K and PQ2 in atmospheres.  From equation  (185)
it can be concluded tha  for all conceivable conditions within
an incinerator the chemical rate will provide the rate limit-
ing step for the oxidation of soot.  Therefore the time to
burn a soot particle (t ) of original diameter  (d  cm) is
given by               "                         °

                  p d      d
                                                         (186)
                  2k
                    chem
          ''chem
On the assumption that the density of the soot particle  is
about 2 g. cm"3, equation  (186) shows that the time  for  burn-
out of the particles will be weakly dependent on the oxygen
partial pressure (P  ), will increase with particle  size d,
                    2
and will decrease with increasing temperature.  Using equa-
tion (186) and typical incinerator conditions, P  =  0.1  atm

and T - 1500°F  {- 1090°K), the time to burn out a2l-micron
particle is 61 sec and a 200R particle is 1.2 sec.
                      -  260  -

-------
Jh iS ®vldent from the kinetic control of soot oxidation
^nat the temperatures in the overfire region should be kept
as nigh as possible.  This result suggests that the addition
or air too high above the fuel bed may cool the gases and
  •JUC2 the reaction' since the rate of reaction faU$ markedly
with decreasing temperature  (at incinerator temperatures the
rate changes an order of magnitude with 200°K change in tem-
perature) .  The importance of the proper design and place-
ment of the overfire air jets has been emphasized in the early
studies at Battelle under Engdahl (157).

Finally, in this study, the combustion intensities for active
burning runs were of the order of 40,000 Btu ft  , assuming
that all the overfire volume and 50% of the fuel bed section
volume were needed for complete combustion.  The complete
oxidation of the combustibles was probably achieved in a
volume rather less than that used in this calculation.  This
combustion intensity is twice that typically used in most
incinerators.  No difficulties were encountered in using a
combustion intensity of this magnitude, even for those runs
where paper was used as a constituent of the fuel? there
have been suggestions that for light material, such as paper,
high combustion intensities aggravate the problem of entrain-
ment.  Extrapolating the combustion intensity results from
this study to full-scale units would be unjustified, but the
results suggest that the possibility of using higher combus-
tion intensities  (which would allow furnace volumes to be
decreased) should be investigated further.
Fuel Bed Conductivity and Rate of Heat Release within Bed
An attempt was made to calculate the effective thermal con-
ductivity of the fuel bed and the net rate of heat release
within the bed following the methods proposed by Stewart and
Saville  (8T) •  These methods have been outlined on pages 34 -
39   and can be seen to rely upon accurate measurement of
the temperature profiles within the bed, as they require the
evaluation of the second derivative of temperature with dis-
tance  —*—  .  The temperature measurements obtained in this
       3z?
study were not of a high enough quality to permit meaningful
estimates of either the effective thermal conductivity or
the net heat release rate using the Stewart and Saville
technique.
                      - 261 -

-------
                      ACKNOWLEDGEMENTS
Financial support from the Solid Waste Research Division of
the Environmental Protection Agency is gratefully acknowledged.
The principal investigators are particularly grateful to Mr.
Louis W. Lefke and Mr. Daniel J. Keller for their guidance
and patience; during the early stages of design and construc-
tion.  The authors are indebted to Mr. Robert C. Thurnau for his
encouragement and advice in the later phases of the program.

The report is based largely on the Sc.D. thesis of Joseph L.
Rogers.

The building of the experimental equipment could have been
achieved only with a great deal more trial and tribulation if
it were not for the outstanding contributions of Steven R. Le-
Mott and Charles T. Johnson.  Grateful acknowledgement goes
also to Reed C. Fulton, for his many helpful suggestions and
practical assistance in constructing much of the heavy equip-
ment; and to Paul W. Bletzer, Allan H. Merrill and Arthur Clif-
ford, for the generous donation of their time in completing
many smaller but no less important tasks.  Special thanks are
due Stanley R. Mitchell, who not only participated in numerous
aspects of building the equipment, but who also lent enthus-
iastic assistance with its operation.  The guidance of Dr.
Anwar E.Z. Wissa of the Department of Civil Engineering with
the selection of the load cell weighing system is much appre-
ciated.

Credits for typing and assembling the final manuscript are due
to Mrs. Alice Biladeau and Mr. Wayne Wendler.  We are grate-
ful to Mr. Donald C. Aldrich for his careful proofreading.
                       - 262  -

-------
                          REFERENCES
 1.   Wallabark,  j.w.r  and  Taylor, A.M.   "Civilization Past
      and  Present/'  Rev.  Ed., vol. 1, p.  148, Chicago, Scott
      Foresman  and Co.  (1949).

 2.   American  Public Works Association,  Public Administration
      Service,  "Municipal Refuse Disposal,"  2nd Edition, p. 7
       (1966) .

 3.   Bering, R. , and Greely, s.A.   "Collection and Disposal
      of Municipal Refuse," McGraw-Hill,  New York  (1921).

 4.   Black, R.J., Munich,  A.T., Klee, A.J., Hickman, H.L.,
      Jr.,  and  Vaughan,  R.D0  "The National  Solid Wastes
      Survey, an  Interim Report," presented  at the 1968 Annual
      Meeting of  the Institute of Solid Wastes of the American
      Public Works Association, Miami Beach, Florida  (October
      24,  1968) .

 5.   Solid Wastes Management/Refuse Removal Journal, Vol. 14
       (4), No.  16, p. 24  (1971).

 6.   Anderson, D.R., and Colliman, J.P.   "Gas Generation and
      Movement  in Landfills/1 Proceedings National Industrial
      Solid Wastes Management Conference,  Houston, Texas,
      March 24-26, pp.  311-316  (1970).

 7.   Vaughan,  R.D.   "National Goals for  Solid Waste Manage-
      ment ," Proceedings  National Industrial Solid Wastes
      Management Conference, Houston, Texas, March 24-2(>,
.nagem
.. 14-
      pp. 14-16  (1970).

 8.   Prindle, R.A.   "Health Aspects of  Solid Waste Disposal,"
      Proceedings The Surgeon General's  Conference on Solid
      Waste Management, Washington, D.C., Public Health Serv-
      ice, PHS-Pub-1729, pp. 15-20  (1967).

 9.   Kenahan, C.B.,  Sullivan,  P.M., Ruppert, J.A., and
      Spano, E.F.   "Composition and Characterization of
      Municipal Incinerator Residues," Report No. 7204, Bureau
      of Mines, U.S.  Department of Interior  (1968).

10.   Ostrowski, E.J.  "Recycling of Ferrous Scrap from
      Incinerator Residue in Iron and Steel Making,"
      Proceedings National Incinerator Conference, ASME,
      New York, pp. 87-96 (1972).
                         - 263 -

-------
11.  "Municipal Refuse Disposal,"  loc.  cit.,  p.  159, (1966).

12.  Johnson, A.J., and Auth,  G.H., Eds. "Fuel and Combus-
     tion Handbook,"  p. 355,   McGraw-Hill, New York, (1951).

13.  Orning, A. A. "Principles  of Combustion" in "Principles
     and Practices of Incineration," Corey, R.C., ed., p. 20,
     Wiley Interscience, New Yok (1969).

14.  Meissner, H.G. "Central Incineration of Community Wastes,"
     in "Principles and Practices of Incineration," loc. cit. ,
     p. 195.

15.  "Municipal Refuse Disposal," loc. cit., p. 164,  (1966).

16.  Hollander? H.I.  "Developing Thermal Energy Release In-
     tensity Parameters for Solid Waste," Proceedings Nation-
     al Industrial Solid Wastes Management Conference, Hous-
     ton, Texas, March 24-26,  pp. 129-132,  (1970).

17.  Niessen, W.R., and Alsobrook, A.F.   "Municipal and
     Industrial Refuse: Composition and Rates,"  Proceedings
     National Incinerator Conference, ASME, New York , pp .
     319-337,  (1972).

18.  Niessen, W.R., and Chansky, S.H.  "The Nature of Munici-
     pal Solid Waste," American Society of Mechanical Engin-
     eers Incinerator Division, May 15,  1969,  New York, New
     York.

19.  Neissen, W.R., Sarofim, A.F., Mohr , C.M., and Moore, R.W.
     "An Approach to Incinerator Combustible Pollutant Control,"
     Proceedings National Incinerator Conference, ASME , New
     York, pp. 248-259, (1972).

20.  Stephenson, J.W. "Burning Wet Refuse," Proceedings Na-
     tional Incinerator Conference, ASME , New York, pp. 260-
     ~    (1972) .
21.  Thurnau, R.C., Project Manager, Thermal Degradation Pro-
     ject, Solid Waste Research Laboratory, U.S. Environment-
     al Protection Agency, National Environmental Research
     Center, Cincinnati, Ohio, personal communication, March,
     1972.

22.  Koenig, L. "The Cost of Municipal Research," Proceedings
     National Industrial Solid Wastes Management Conference,
     Houston, Texas, March 24-26, pp. 436-442,  (1970).
                        - 264 -

-------
*") "5
   .  Kreisinger, H., Ovitz, F.K., Augustine, C.E.  "Combus-
     tion in the Fuel Bed of Hand-Fired Furnaces," U. S.
     Bureau of Mines, Technical Paper No. 137,  (1916).

24.  Nicholls, p.   "Underfeed Combustion, Effect of Preheat
     and Distribution of Ash in Fuel Beds,"  U.S. Bureau of
     Mines, Bulletin No. 387,  (1934).

25.  Rose, A.H., and Crabaugh, H.R.  "Incinerator Design
     Standards: Research Findings," Publications of the Los
     Angeles County Air Pollution Control District, No. 60.

26.  Weintraub, M., Orning, A.A., and Schwartz, C.H.  "Experi-
     mental Studies of Incineration in a Cylindrical Combus-
     tion Chamber," U. s. Dept. of the Interior, Bureau of
     Mines Investigation 6908,  (1967).

27.  Orning, A.A.,  Pfeiffer, J.J., Harold, W.C., and Schultz,
     J.F.  "Experimental Study of Incineration," 51st Annual
     Meeting of APCA, Philadelphia, Pa., May 25-29, paper 58-
     12,  (1958).

28.  Schwartz, C.H., Orning, A.A., Snedden, R.B., Demeter,
     J.J. and Bienstock, D.  "Development of a Vortex Incin-
     erator with Continuous Feed," Proceedings National In-
     cinerator Conference, ASME, New York, pp. 145-154, TT972) ,

29.  Stenburg, R.L., Hangerbrauck, R.P., VonLehunden, D.J.,
     and Rose, A,H.  "Effects of High Volatile Fuel on In-
     cinerator Effluents,"  53rd Annual Meeting of APCA,
     Cincinnati, Ohio, May 22-26,(1960).

30.  Essenhigh, R.H.  "Incineration-A Practical and Scientific
     Approach," Environmental Sci. and Tech., Vol. 2, p. 524,
     (1968).

31.  Stumbar, J.P., and Essenhigh, R.H.  "Flame Spread Down
     Cardboard Sticks and Paper Cylinders," Fire Research
     Abstracts and  Reviews, Vol. 10, p. 112,  (1968).

32.  Essenhigh, R.H.  "Burning Rates in Incinerators, Parts I
     and II," Proceedings National Incinerator Conference,
     ASME, New York, p. 87, (1968).              ~~~~~~~~~

33.  Rao, S.T.R., Gelernter, G., and Essenhigh, R.H.  "Scale-
     up Factors for Combustion Pot Behavior in Predicting
     Incinerator Performance," Proceedings Third National
     Incinerator Conference, New York, p. 232,(1968).
                       - 265 -

-------
34.  Biswas, B.K., and Essenhigh, R.H.  "Heat Value of Refuse,"
     Mech. Eng.f Vol. 91, p. 68, (1969).

35.  Kuwata, M., Stumbar, J.P., and Essenhigh, R.H.  "Combus-
     tion Behavior of Suspended Paper Spheres," Proc. Twelfth
     Symposium (International)  on Combustion, Poitiers, France,
     July, 1968, The Combustion Institute Pittsburgh, pp. 663-
     675, (1969).

36.  Essenhigh, R.H., and Kuo,  T.J.  "Combustion and Emission
     Phenomena in Incinerators: Development of Physical and
     Mathematical Models of Incinerators, Part I: Statement
     of the Problem,"  Proceedings National Incinerator Con-
     ference, ASME, New York, pp. 261-271,(1970).

37.  Kuwata, J. , Kuo,, T.J., and Essenhigh, R.H.  "Combustion
     and Emission Phenomena in Incinerators:  Burning Rates
     and Operational Limits in a Solid Fuel Bed," Proceedings
     National Incinerator Conference, ASME, New York, pp. 272-
     287, (1970).

38.  Stumbar, J.P., Kuwata, M., Kuo, T.J., and Essenhigh, R.H.
     "Combustion and Emission Phenomena in Incinerators:
     Propagation Velocity and Flame Spread Through Simulated
     Fuel Beds, Part I:  Flame Spread Down a Simulated Fuel
     Bed, Part II, The Mechanism of Flame Spread Over a Cellu-
     losic Surface,"  Proceedings National Incinerator Confer-
     ence, ASME, New York, pp.  288-293, 293-303,(1970).

39.  Biswas, B.K., Kuo, T.J., and Essenhigh,  R.H.  "Combus-
     tion and Emission Phenomena in Incinerators: Studies on
     Combustion Behavior and Extinction Limits of Smoke Flames,"
     Proceedings National Incinerator Conference, ASME, New
     York, pp. 304-313,(1970).

40.  Rao, S.T.R., Kuo, T.J., and Essenhigh, R.H.  "Combustion
     and Emission Phenomena in Incinerators:  Characteriza-
     tion of Stirring Factors by Cold Model Simulation,"
     Proceedings National Incinerator Conference, ASME, New
     York, pp. 314-326, (1970).

41.  Kuo, T.J., Kuwata, M., Shieh, W.S., and Essenhigh, R.H.
     "Combustion and Emission Phenomena in Incinerators:
     Physical and Mathematical Model of a Simple Incinerator:
     Part II, The  Mechanistic Analysis," Proceedings Nation-
     al Incinerator Conference, ASME, New York, pp. 327-330,
     (1970) .
                       - 266 -

-------
   •  Biswas, B.K., and Essenhigh, R.H.  "Some Characteristics
     or stabilized Smoke Flames,"  Combustion and Flame,
     vol. 15, pp. 93-96,  (1970).   	

43.  Essenhigh, R.H., and Kuo, T.J.  "Development of Funda-
     mental Basis for Incinerator Design Equations and
     Standards,"   Proceedings Third Mid-Atlantic Industrial
     Waste Conference. University nf Marylandr MOV. lQfiQr
     PP. lOb-146,  (1970).

44.  Rao, S^T.R, and Essenhigh, R.H.  "Experimental Determina-
     tion of Stirring Factors Generated by Straight and Swirl-
     ing Jets in Isothermal Combustion Chamber Models,"  Pro-
     ceedings Thirteenth Symposium  (International) on Combus-
     tion, University of Utah, August iQ7n, Th^ rornhng-Hon
     Institute, pp. 603-614,  (1971).

45.  Biswas, B.K., and Essenhigh, R.H.  "The Problem of
     Smoke Formation  and Control," Paper No. 39f presented
     at A.I.Ch.E. 70th National Meeting, Atlantic City,
     Aug/Sept.  1971.

46.  Shieh, W.S., and Essenhigh, R.H.  "Experimental Behavior
     of Waste Combustion in a Test Incinerator,"  Fourth Mid-
     Atlantic Industrial Waste Conference, University of Dela-
     ware , November,(1970).

47.  Shieh, W.S., and Essenhigh, R.H.  "Combustion of Computer
     Cards in a Continuous Test Incinerator: A Comparison of
     Theory and Experiment," Proceedings National Incinerator
     Conference, ASME, New York, pp. 120-134, (1972).

48.  "Municipal Refuse Disposal,"  loc. cit., pp. 45-54, (1966)

49.  Kaiser, E.R. "Chemical Analyses of Refuse Components,"
     publication 65-WA/PID-9, ASME, New York, 1965.

50.  "Municipal Refuse Disposal," loc. cit., p. 12,  (1966).

51.  Kaiser, E.R. "Refuse Composition and Flue Gas Analyses
     from Municipal Incinerators," Proceedings National In-
     cinerator Conference, ASME, New York, p. 180,  (1964) .

52.  "Municipal Refuse Disposal," loc. cit., p. 47,  (1966).

53.  "Control Techniques for Particulate Air Pollutants,"
     U.S. Dept.  of HEW, PHS, Consumer Protection and Environ-
     mental Health Service, National Air Pollution Control
                       - 267 -

-------
     administration Publication No. AP-51.

54.  "Proposals for a Refuse Disposal System in Oakland Coun-
     ty, Michigan/' Report SW-7d prepared by Jones and Henry
     Engineers Limited under Grant D01-U1-00068 from the Bu-
     reau of Solid Waste Management, Public Health Service
     Publication No. 1960 (1970).

55.  Bortz, S.A., and Pincus, A.G.  "High Temperature Inciner-
     ation," Proceedings National Industrial Solid Wastes
     Management Conference, University of Houston, March 24-
     26,  (1970).

56.  Bernard, R.A., and Wilhelm, R.H.  "Turbulent Diffusion
     in Fixed Beds of Packed Solids," Chem. Eng. Prog., Vol.
     46, No. 5, pp. 233-244, (1950).

57.  Schotte, W.  "Thermal Conductivity of Packed Beds,"
     A.I.Ch.E. Journal, Vol. 6, No. 1, p. 63 (1960).

58.  Yagi, S., Kunii, D., and Wakao, N.  "Internal Develop-
     ments in Heat Transfer," Part IV, p. 742, ASME, New York,
     (1961).

59.  Luikov, A.V., Shashkov, A.G., Vasiliev, L.L., and Fraiman,
     Yu.E. "Thermal Conductivity of Porous Systems," Int. Jour-
     nal of Heat and Mass Transfer, Vol. 11, pp. 117-137^  TT9~6"8)

60.  Barker, J.J.  "Heat Transfer in Packed Beds," Ind. and
     Eng. Chem., Vol. 57, No. 4, p. 43 (1965).

61.  Heggs, P.J., and Handley,  D.  "Momentum and Heat Trans-
     fer Mechanisms in Regular Shaped Packings,"  Trans. Instn.
     Chem. Engrs, Vol. 46, T251  (1968).

62.  Furnas, C.C.  "Heat Transfer from a Gas Stream to a Bed
     of Broken Solids," U.S. Bureau of Mines Bulletin 261
     (1932) .

63.  Gillespie, B.M., Crandall, E.D., and Carberry, J.J. "Lo-
     cal and Average Interface Heat Transfer Coefficients in
     a Randomly Packed Bed of Spheres,"  A.I.Ch.E. Journal,
     Vol. 14, pp. 483-490, (1968).

64.  Bradshaw, A.V., Johnson, A., McLachlan, N.H. and Chin,
     Y-T.  "Heat Transfer Between Air and Nitrogen and Packed
     Beds of Non-reacting Solids," Trans. Instr. Chem. Engrs.,
     Vol. 48, pp. T77-T84, (1970).

65.  Kitaev, B.I.  "Heat Exchange in Shaft Furnaces, Metallur-
     gizdat, (1945).                                   	
                       - 268 -

-------
      °   lda" F- i  Ramaswami,  D. , and  Hougen,  D.A.   "Temper-
      atures  and  Partial  Pressures  of  the  Surfaces  of  Catalyst
      ^articles," A.I.Ch.E.  Journal, Vol.  8, pp.  5-11,  (1962).

   "
            '  B-z-»  Yaroshenko,  Yu.  G. ,  and  Suchkov, V.I.
            Exchange in  Shaft  Furnaces,"  p.  90,  Pergammon  Press,
     Oxford,  (1967).

68.  Baumeister,  E.B.,  and  Bennett,  C.O.   "Fluid-Particle Heat
     Transfer m  Packed Beds,"  A.I.Ch.E. Journal, Vol.  4, No.
     1,  pp.  69-74,  (1958).      — -

69.  Denton,  w.H.   "Proceedings of  the  General  Discussion on
           >"  Inst°  °f Mech.  Eng.  (London), p. 370
70.  Yagi,  s.f  and  Kunii,  D.  "Studies on Effective Thermal
     Conductivities in  Packed  Beds," A.I.Ch.E. Journal, Vol.
     3, No.  3,  pp.  373-381 (1957).   -

71.  Deissler,  L.G.f  and Eian,  G.S.  Natl. Advisory Comm. ,
     Aeronaut . ,  RME. 52005  (1952).    - -

72.  Yagi,  S. ,  Kunii, D. ,  and  Wakao, N.  "Studies on Axial Ef-
     fective Thermal  Conductivities in Packed Beds," A.I.Ch.E.
     Journal , Vol.  6, No.  4, pp.  543-546 (1960).

73.  Spalding,,  D.B.   "Some Fundamentals of Combustion,"  pp.
     53-81, Gas  Turbine Series, Vol. 2, Butterworths Scien-
     tific  Publications, London (1955) .

74.  Kays,  W.M.   "Convective Heat and Mass Transfer,"  McGraw-
     Hill,  New York,  pp. 327-329  (1966).

75.  Amundson, N.R.   "Solid-Fluid Interactions in Fixed and Mo-
     ving Beds:   Part 2, Fixed Beds with Large Particles,"
     Ind. and Eng.  Chem. ,  Vol.  48, p. 35 (1956).

76.  Sauders, O.K., and Ford,  H.   "Heat Transfer in the Flow
     of Gas Through a Bed  of Solid Particles," J. Iron and
     Steel  Inst. , Vol.  1,  pp.  292-328  (1940).

77.  Chukhanov,  Z.F.  "Thermal Regime of Semi Coking, Burning
     and Gasification of Particles in a Solid Fuel,"  Dokl.
     Akad. Nauk.  SSSR,  Vol. 47, No. 3, 1945. Clearing House
     for Scientific and Technical Information, Washington,
     D.C. 20230  Translation No. 62-10785.

78.  Handley, D. , and Heggs, P.J.  "The Effect of Thermal Con-
     ductivity of the Packing  Material on Transient Heat
                        - 269  -

-------
     Transfer in a Fixed Bed," Int. J. Heat Mass Transfer,
     Vol. 12, pp. 549-570, Pergammon Press (1969).

79.  Schumann, T.E.W., "Heat Transfer:  A Liquid Flowing Through
     a Porous Prism,"  J. Franklin Inst., Vol. 208, No. 3,
     p. 405  (1929).

80.  Klinkenberg, A.  "Numerical Evaluation of Equations Des-
     cribing Transient Heat and Mass Transfer in Packed Solids,"
     Ind. and Eng. Chem., Vol. 40, Pt. 2, No. 10, p. 1992
     (1948)"

81.  Singer, E., and Wilhelm, R.H.  "Heat Transfer in Packed
     Beds:  Analytical Solution and Design Method; Fluid Flow,
     Solids Flow and Chemical Reaction," Chem. Eng. Prog.,
     Vol. 46, No. 7, p.  343,  (1950).

82.  Brinkley, S.R.  "Heat Transfer Between Fluid and Porous
     Solid Generating Heat," J. App. Physics, Vol. 18, No. 6,
     p. 582  (1947).

83.  Wilhelm, R.H., Johnson, W.C., and Acton, F.S.  "Conduc-
     tion, Convection and Heat Release in Catalytic Converters,"
     Ind. and Eng. Chem., Vol. 35, No. 5, p.  562 (1963).

84.  Kasten, P.R., Lapidus, L., and Amundson, N.R.  "Mathema-
     tics of Absorption in Beds:  Part V, Effect of Intrapar-
     ticle Diffusion in Flow Systems in Fixed Beds,"  J. Phy s.
     Chem., Vol. 56, No. 6, P. 683, (1952).

85.  Munro, W.D., and Amundson, N.R.  "Solid-Fluid Exchange
     in Moving Beds,"  Ind. and Eng. Chem., Vol. 42, No. 8,
     p. 1481  (1950).

86.  Amundson, N.R.  "Solid-Fluid Interactions in Fixed and
     Moving Beds:  Part 1, Fixed Beds with Small Particles,"
     Ind, and Eng. Chem,. , Vol. 48, No. 1, P.  26  (1956).

87.  Stewart, M.C., and Saville, J.  "A Simplified Heat Trans-
     fer Equation for Sinter Beds,"  Trans. Australian Mech.
     and Chem. Eng., p.  135  (1968).

88.  Woods, S.E., and Harris, C.F.  "Heat Transfer in Sinter
     Roasting,"  Proceedings Symposium on Chemical Engineer-
     ing in the Metallurgical Industries, Edinburgh 25-26,
     Sept. 1963, London I. Chem. E., p. 77 (1963) .

89.  Walker, P.L., Rusinko, R., and Austin, L.G.  "Gas  Reac-
     tions of Carbon," in "Advances in Catalysis,"  Vol. XI,
                       - 270  -

-------
      P- 135, Academic Press  (1959).

   .  Mayers, M.A.  "Temperatures and Combustion Rates in a
       nQ^7?ed'"  Trans° ASME, Vol. 59, Paper No. FSP-59-8
       \ ^- -? J / / .


 91.  Kolodtsev, K.I.  "Dynamics of Gas Formation in Carbon
      Beds'  J- phy. Chem. U.S.S.R., Vol. 19, p. 417  (1945).
 92.  Yagi, s., and Kunii, D.   "Studies on Combustion of Car-
      bon Particles in Flames and Fluidized Beds,"  5th Sym-
      posium  (International) on Combustion, Reinhold, New York,
      p. 231  (1955)": -   -

 93.  Parker, A.S., and Hottel, H.C.   "Combustion Rate of Car-
      bon:  Study of  Gas-Film Structure by Micro sampl ing, "
      ^;^fnd Eng° Cheiru° v°l- 28, No. 11, pp. 1334-1341
      (1936) .

 94.  Marskell, W.C0,and  Miller, J.M.  "Mode of Combustion of
      Coal on a Chain Grate  Stoker,"   Fuel, Vol. 25, No. 1,
      pp. 4-12  (1946).                 -

 95.  Kreisinger, H., Augustine, C.E., and Harpster, W.C.
      "Combustion Experiments with North Dakota Lignite,"
      U.S. Bureau of  Mines,  Technical  Paper 207 (1919).

 96.  Mayers, M.A.  "Combustion Characteristics of Coke,"
      Fuel in Science and Practice, Vol. 21, No. 1, p. 4 (1941),

 97.  Mayers, M.A.  "Some Factors Affecting Combustion in  Fuel
      Beds,"  ASME Tech.  Publication,  No. 77  (1937).

 98.  Thring, M.W.  "Physics of Fuel Bed Combustion,"  Fuel,
      London, Vol. 31, p.  355  (1952).

 99.  Thring, M.W.  "Physical Processes in a Bed of Fuel,"
      Inst. of Fuel Journal, Vol. 24,  No. 75,  pp. 47-65  (1940).

100.  Silver, R.S.  "Theoretical Treatment of  Combustion in
      Fuel Beds I, Gas Compositions and Heat Release,"  Fuel,
      Vol. 32, p. 121,  (1953) .

101.  Silver, R.S.  "Theoretical Treatment of  Combustion in
      Fuel Beds II -  Temperature Attained in Combustion,"
      Fuel, Vol. 32,  p. 138  (1953).

102.  De Acetis, j. ,  and  Thodos, G.   "Flow of  Gases Through
      Spherical Packings,"   Ind. and Eng. Chem. , Vol.  52, p.
      1003 (1960) .           -
                         - 271 -

-------
103.  Terres, E., Tjia, Hie A., Herrmann, W., Johswich, F.,
      Patscheke, G., Pfeiffer, J., and Schwarzmann, H.
      Beitrage zur Kenntris der thermischen grundlagen der
      Verschwelung und Verkokung von Braunkohlen  (Ausang),"
      Angew. Chm., Vol. 48, pp. 17-21  (1935).

104.  Bangham, D.H., and Thring, M.W.  Coal Research, p.  48
      (Sept., 1944).

105.  Gamson, B.W.  "Heat and Mass Transfer:  Fluid Solid Sys-
      tems,"  Chem. Eng. Prog., Vol. 47f No. 1, pp. 19-28
      (1951).

106.  Nichols, L.H.F.  "Gas Sampling and Temperature Measure-
      ment in Fuel Beds,"  Inst. of Fuel Journal, Vol. 14,
      No. 75, pp. 71-73, (1940).

107.  Mott,  R.Ao, and Wheeler, R.V.  "Coke for Blast Furnaces,"
      Iron and Steel Industry Research Council  Report No. 1,
      Midland Coke Research Committee  (1930).

108.  Lof, G.O.G., and Hawley, R,W.  "Unsteady State Heat Trans-
      fer Between Air and Loose Solids,"  Ind. and Eng. Chem.,
      Vol. 40, p. 1061  (1948).

109.  Thring, M.W.  Coal Research, p. 70  (Sept. 1944).

110.  Essenhigh, R.H.  "The Speed of Combustion of Solid  Fuel
      on Beds,"  in "The Science of Flames and Furnaces,"
      Thring, M.W., p. 184, 2nd Edition, John Wiley & Sons,
      Inc., New York (1962) .

111.  Wu, P.C.   "The Kinetics of the Reaction of Carbon with
      Carbon Dioxide,"  Sc.D. Thesis, Dept. of Chem. Eng.,
      M.I.T.  (1949).

112.  Arthur, J.R.  "Reactions Between Carbon and Oxygen,"
      Trans. Faraday Soc.,  Vol. 47, pp. 164-178  (1951).

113.  Rice, C.W., "Forced Convection of Heat in Gases and Li-
      quids - II,"  Ind. and Eng. Chem., Vol. 16, No. 5,  pp.
      460-467 (1924).

114.  Kaiser, E.L., personal communication to Niessen, W.R.,
      Mohr, C.M., and Sarofim, A.F.  (1970).

115.  von Fredersdorff, C.G., and Elliott, M.A.  "Coal Gasi-
      fication,"  Chapter 20 in "Chemistry of Coal Utilization,"
      Lowry, H.H.,  ed„, Supplemental volume, John Wiley and
      Sons, New York (1945).
                        - 272 -

-------
116.  Lewis  W.K., Gilliland, E.R., McBride, G.T., Sr.   "Gas-
      irication of Carbon by Carbon Dioxide in Fluidized Pow-
      1226  (19"   Ind' and En?- Chem., Vol. 41, No. 6, pp. 1213-


 17.  Simms, D.L., and Law, M.   "The Ignition of Wet and Dry
      wood  Ignited by Radiation,"  Combustion and Flame, Vol.
      11, p. 377-388  (1967).       	

118.  Spalding, D.B.  "Conditions  for the Stable Burning of
      Carbon in an Air Stream,"  Inst. of Fuel Journal, Vol.
      26, No.  155, pp. 289-294  (1953) .	  	

119.  Vortmeyer,  D.   "Das Verhalten von  Zvindzonen in exotherm
      reagierenden gasdurchstromten Kornerschuttungen,"  Zeit-
      schrift  fur Elektrochemie. Vol. 65, pp. 282-289 (196TT7

120.  Eigenberger, G.  "On  the dynamic behavior of the cataly-
      tic fixed-bed  reactor in the region of multiple steady
      states - I.  The influence of heat conduction in two
      phase models,"  Chem., Eng. Sci., Vol. 27, pp. 1909-1915
      (1972).         	~	

121.  Vortmeyer,  D.,  and Jahnel, W.  "Moving reaction zones in
      fixed bed reactors under the influence of various para-
      meters," Chem. Eng.  Sci., Vol. 27, pp. 1485-1496  (1972).

122.  Wicke, E.   "Einfluss  des Stofftransportes auf den Verlauf
      heterogener Gasreaktionen,"  Chem. Eng. Sci., Vol. 8,
      p. 61  (1958).

123.  Wekmeister, H.  "Verbrennungsverlauf bei Steinkolen mitt-
      lerer Korngrossen,"   Arch, fur Warmewirtschaft, Vol. 12,
      pp. 225-232  (1931).

124.  Maughan, J.D.,  Spalding, H.B., and Thornton, B.M.  "A
      Method of Controlling the  Temperatures of Travelling
      Grate Stokers,"  Engineering, Vol. 137, pp. 587-589,
      656-658, 669-671  (1934).

125.  Williams, C.C.  "Damage Initiation in Organic Materials
      Exposed  to  High Intensity  Thermal  Radiation,"  Technical
      Report No.  2, Fuels Research Laboratory, Massachusetts
      Institute of Technology, Cambridge, Mass.  (1953).

126.  Garden,  R.  "Temperature Obtained  in Wood Exposed  to High
      Intensity Thermal Radiation,"  Technical Report No. 3,
      Fuels Research Laboratory, Massachusetts Institute of
      Technology, Cambridge, Mass.  (1953).

127.  Kreith,  F., and Romie, F.E.  "A Study of the Thermal
                         - 273 -

-------
      Diffusion Equation with Boundary Conditions Correspon-
      ding to Solidification or Melting of Materials Initially
      at the Fusion Temperature,"  Proc. Phys. Soc. B., Vol.
      68, pp. 277-291 (1955).

128.  Goodman, T.R.  "The Heat-Balance Integral and Its Appli-
      cation to Problems Involving a Change of Phase,"  Trans.
      ASME, Vol. 8, No.  2, pp. 335-342 (1958).

129.  Rosseland, S.  "Astrophysik und Atom-Theoretisches
      Grundlage,"  Springer Verlag, Berlin, 1931, pp. 41-44.

130.  Bamford, C.H., Crank, J., and Malan, D.H.  "The Combus-
      tion of Wood.  Part I,"  Proc. Camb. Phil. Soc., Vol.
      42, pp. 146-181,  (1946).

131.  Tao, C.H.j  "Generalized Numerical Solutions of Freezing
      a Saturated Liquid in Cylinders and Spheres,"  A.I.Ch.E.
      Journal, Vol. 13,  No. 1, pp. 165-169 (1967).

132.  Shih, Yen-Ping, and Chou, Tse-Chuan, "Analytical Solu-
      tions for Freezing a Saturated Liquid Inside or Outside
      Spheres,"  Chem. Eng. Sci., Vol. 26, pp. 1787-1793 (1971).

133.  Siegel, R., and Savino, J.M.  "An Analytical Solution for
      Solidification of a Moving Warm Liquid onto an Isother-
      mal Cold Wall,"  Int. J. Heat Mass Transfer, Vol. 12,
      p. 803 (1969).

134.  Goodman, T.R.  "Integral Methods for Non-Linear Heat
      Transfer,"  Advan. Heat Transfer, Vol.  1, pp. 51-122
      (1964).

135.  Rossini, F.D., Pitzer, K.S., Arnett, R.L., Braun, R.M.,
      and Pimental, G.C.  "Selected Values of Physical and Ther-
      modynamic Properties of Hydrocarbons and Related Compounds,"
      Carnegie Press, Pittsburgh  (1953) .

136.  Hougen, B.A., and Watson, K.M.  "Chemical Process Prin-
      ciples,"  Part I.  p. 254, Wiley and Sons, New York, 1950.

137-  Dunningham, A.C.,  Grummell, E.S.  "Contribution to the
      Study of the Ignition of Fuel Beds,"  Inst. of Fuel Jour-
      nal, Vol. 11, pp.  117-122, 129-133  (19T7Ti

138.  Johnstone, R.E., and Thring, M.W.   "Pilot Plants, Models
      and Scale-up Methods in Chemical Engineering,"  McGraw-
      Hill, New YOrk, 1957, p. 214.

139.  Spink, L.K.  "Principles and Practices of Flow Meter
                        - 274 -

-------

       Engineering,"   The Foxboro Company,  March 1967,  Chapter
               i'E-L"  Sarofim,  A.F.,  and Howard,  J.B.   "The
            n 2  Underfire Air Rate on a Burning Simulated Re-
             « '    Proceedings National Incinerator Conference,
           ,  New York,  pp. 135-144 (1972). - - : -

 141.   Wheeler, A.   "Reaction Rates and Selectivity of  Catalyst
       n   ?Jo  ChaPter in "Advances in Catalysis,"  Vol.  3,
       n     o
       PP.  <249-327,  Academic Press,  Inc.,  New York,  1951.

 142.   Weisz, _P.B.,  and Prater,  C.D.   "Interpretation of Measure-
       ments in Experimental Catalysis,"   Chapter in "Advances
       in Catalysis,"  Vol.  6,  pp.  143-196,  Academic Press,  Inc.,
       New York, 1954.

 143.   Van Heerden,  C.   "The Character of  the Stationary State  of
       Exothermic Processes,"  Chem.  Eng.  Sci". ,  Vol.  8,  p. 133


 144.   Liu,  S.f and Amundson, N.R.   "Stability of Adiabatic  Packed
       Bed Reactors:  An Elementary Treatment,"   Ind.  Eng. Chem.
       Fundamentals, Vol. 1, p.  200  (1962).              -

 145.   Aris, R.  "Some Problems in  the Analysis  of Transient Be-
       havior  and Stability of Chemical Reactors,"  Advances
       in Chemistry, No. 109, p.  578,  American Chemical  Society
       (1972) .

 146.   Aris, R. and  Schruben, D.L.   "Transients  in Distributed
       Chemical Reactors, Part I:   A Simplified"  Model* "' Chem.
       Eng.  J., Vol. 2, p. 179 (1971).      - -~-^**^^— -• -

 147.   Aris, R. , and Fatina, I.H.   "Transients .in Distributed
       Chemical Reactors, Part 2:   Influence of  Diffusion in the
       Simplified Model,"  Chem.  Eng.  J.,  Vol. 4, p.  149 (1972).

 148.   Rhee, Hyun-Ka,  Foley, D.,  and Amundson, N.R.   "Creeping
       Reaction zone in a Catalytic,  Fixed-Bed Reactor:  A
       Cell  Model Approach,"  Chem.  Eng. Sci., Vol.  28,  pp.
       607-615  (1973).

149.   Sarofim,  A.F., personal  communication, March  5, 1973.

150.   Sarofim,  A.S.   "Combustion Reactions  in Overfire  Volumes,"
       Presented to  the Grates  and  Combustion Committee  ASME
       Incinerator Division, Thursday, Jan.  15,  1970.

151.   Longwell,  J.P. and Weiss, M.A. , "High Temperature Reac-
       tion  Rates in Hydrocarbon Combustion," Ind.  Eng. Chem. ,


                         - 275 -

-------
      Vol. 47, p. 1534 (1955).

152.  Hottel, B.C., Williams, G.C., Nerheim, M.N. , and Schneider
      G.R. ,  "Kinetic Studies in Stirred Reactors:  Combustion
      of Carbon Monoxide and Propane,"  Tenth Symposium (In-
      ternational) on Combustion, pp. 111-121, The Combustion
      Institute  (1965) .

153.  Williams, G.C. , Hottel, H.C., and Morgan, A.C.  "The Com-
      bustion of Methane in a Jet Mixed Reactor , "  Twelfth Sym-
      posium  ( International ) on Combustion , pp. 913-925, The
      Combustion Institute
154.  Nagle, J. and Strickland-Constable, R.F.  "Oxidation of
      Carbon Between 1000-2000 C,"  Proceedings of Fifth Car-
      bon Conference, Vol.  1,  p.  154 (1962) .

155.  Appleton, J.P.  "Soot Oxidation Kinetics at Combustion
      Temperatures,"  Presented at AGARD Propulsion and Ener-
      getics Panel, 41st Meeting on "Atmospheric Pollution by
      Aircraft Engines,"  London, England, 9-13 April (1973).

156.  Park, C., and Appleton,  J.P.  "Shock-Tube Measurements
      of Soot Oxidation Rates,"  Combustion and Flame, (in
      Press), (1973).

157.  Engdahl, R.B., and Major, W.S.  Bituminous Coal Research,
      Inc., Tech.  Paper No. VII,  1957.

158.  American Gas Association Report No. 3 (1954).
                        - 276 -

-------
                   Biographical Note
 n SaVx   Edward Lloyd Rogers was born on January 30. 1945,
              Rhodesia. and is the son of Dr. and Mrs. Edward
          o            .                     .        .
 no    •  Salisbury, Rhodesia.  He attended Ruzawi Preparatory
Sali*£  in Marandellas, Rhodesia, and Prince Edward High School,
1962  hY' Rhodesia-  After completing high school in December
      h  WaS awarded a four year Student Engineer Scholarship
     ?ne.Angl0 American Corporation of South Africa and worked
     Anglo American at the Rhokana mine in Zambia until enter-
        University of Edinburgh, Scotland, in the fall of
    .   He graduated with honors from the University of Edin-
burgh in June 1967.  m the summer of 1967 he was awarded a
Fulbright Travel Grant and he entered M.I.T. in the fall of
that year as a teaching assistant.  He attended the School of
Chemical Engineering Practice in the fall of 1968 and received
the degree of Master of Science in Chemical Engineering Prac-
tice in February 1969.  He then accepted a one-semester appoint-
ment as the Assistant  Director of the Bound Brook Practice
School  in the spring of 1969.  In the summer of 1969 he returned
to M.I.T. as a Research Assistant and commenced work on his doc-
torate.

     Joseph Rogers is  a member of the honorary society of the
Sigma Xi, the American Society of. Chemical Engineers, the In-
stitution of Chemical  Engineers  (London) , the Combustion Insti-
tute, and is an honorary citizen of the State of Tennesse.  He
is co-author of

     "The Effect of Underfire Air Rate on a Burning Simulated
     Refuse Bed,"  Proceedings National Incinerator Conference,
     ASME, New York, pp. 135-144,  (1972).

     "Combustion Characteristics of a Simulated Refuse Bed,"
     presented at the  1973 Technical Session of the Central
     States Section of the Combustion Institute, March 27,
     28, Champaign, Illinois.

     He was married to the former Sally Ann Doonan in August
1972 and has accepted  a position with Halcon International,
Inc.,  in New York City.
                         - 277 -

-------
                     NOMENCLATURE
Dimensional units are expressed in terms of length  (L) , mass
(M) , quantity of heat (H) , time (t) and temperature  (T) .
Dimensional quantities are given in terms of Btu, hr, ft, °F
unless otherwise specified.
Symbol

English

A
B
D
D
E

E
Meaning
Frequency factor in kinetic
expression


Ash fraction of fuel, equation
(95)

Specific heat ratio, C  /C
                      ps  pg

Interfacial area per unit
volume of bed

Mass transfer driving force

Concentration of specie j

Heat capacity of gas

Heat capacity of solid

Coefficient of molecular
diffusion

Effective coefficient of
diffusion

Particle diameter

Mean radiating beam length,
equation (65)

Turbulent eddy diffusion

Activation energy
                                   Dimensions
                    Dimensions of
                    reaction rate
                    constant
                                           L2lT3
                                           ML
                                             ~3
                                           HM"1!'1

                                           LV1
                                           L

                                           L
                    H lb mole
                       - 278  -

-------
 A         Area firing rate, equation
            (95)
p
 RCS       Relative Carbon Saturation
           factor, defined by equation
            (82)
•

G          Mass velocity of fluid          ML"2t"1

9          Mass transfer conductance       Mt"1!,"2
 A                                            1  O
9          Mass transfer conductance       Mt  L
           for very small mass transfer
           rates

Hv         Effective latent heat of vapor- HM"1
           ization,
                       AH    ps
                        v   —.	1
HV          Latent  heat  of vaporization     HM
 c                                           -1
H           Overall heat of  combustion      HM

H.          Heat  of reaction per mole of    H mole
 -1          j  - endothermic  negative,
            exothermic positive
                                             -1 -2 -1
h           Radiation contribution  to       Ht  L  T
 r          effective thermal conductivity
            of porous solid  - defined by
            equation (14)
                                             -1 -2 -1
h           Heat  transfer coefficient       Ht  L  T
 S          based on unit area
                                             -1 -2 -1
h           Effective heat transfer coef-   Ht  L  T
 S          ficient based on unit area,
            defined by equation  (6)

hv          Heat  transfer coefficient       Ht  L  T
            based on unit volume

•j_          k  p    (N,,
JD          eg     Sc
                       -  279 -

-------
                  
-------
NBi        Biot Number /h D
NLe        Lewis Number /BL, \
                        l-^£\
                        VW

NPr        Prandtl Number /C V
                               \

                               '
                             g

NRe         Modified  Reynolds Number /D G'
NSC         Schmidt  Number / y
Nc.         Sherwood  Number /k D
 Sh                        let
Qne         New total  heat release from     Ht  L
            z=0 to  z=°°

Q(z)        Net heat release                Ht"1!,""3

—                                            -2
Q(z)        Conductivity normalized net     TL
            heat release
                                             2 -2 -1    -1
R           Gas constant                    ML t  T  mole

R           Radius  of  particle              L

r.          Rate of reaction  i                    ML  t

r.          Effective  rate of reaction i          ML  t
                                                   -3 -
R*          Rate of reaction

T           Ambient temperature             T

T d         Adiabatic  temperature           T

T           Gas  temperature                 T
                       - 281 -

-------
t

U

V



V
x

Y

z
Symbol

Greek

B

Y
<5

e
Solid temperature

Time

Mass velocity of solid

Volatile fraction of fuel,
equation (95)
UC
       GC
              •8
                                T

                                t
                                L

                                L
Weight fraction of pyrolyzable
material

Defined by equation  (35)

Defined by equation  (34)

Vertical position

Oxygen consumption distance


          Meaning



Defined by equation  (45)

V/k", equation  (51)

GC
  pg, equation (41)
Void fraction

Emissivity

Stoichiometric coefficient of com-
ponent i in reaction j
                                   Dimensions
           Gas density

           Solid density
                                ML
                                ML
                                             -3
                       - 282  -

-------
*
           Shape factor (unity for
           spheres), equation  (10)

           Gas viscosity

           Stefan-Boltzman constant

           Net heat absorbed in endo-
           thermic reactions, equa-
           tion  (4)

           Mass  rate of appearance of
           specie  j, equations  (3)
           and  (5)
Symbol

Superscript

b

d

i

o

v

w

Subscript

j
Meaning



Bulk

Dry

Interface

Initial

Vaporization

Wet



Specie j
                     ML'V1
                      - 283 -

-------
      THE  FOLLOWING PAGES ARE DUPLICATES OF






  ILLUSTRATIONS APPEARING ELSEWHERE IN THIS





 REPORT.  THEY HAVE BEEN REPRODUCED HERE BY






A DIFFERENT METHOD TO PROVIDE BETTER DETAIL

-------
I
u" 3 cr H
£-6 £ 2?
* 3 *"
III
  0-0
  2 -PS
  Ib
                      Fig.   44.   Experimental Incinerator,  Gas  Analysis
                                 Train and Peripheral Equipment.

-------
Fig.   45.   Interior of  Fuel  Bed  Section with  Thermocouple
                             Probes  Installed.
                                     This page is reproduced  at the
                                     hack of the report by a different
                                     reproduction method  to provide
                           - 1 6 0 -  better detail.

-------