REMOVAL AND RECOVERY OF SULFUR
DIOXIDE IN THE PULP MILL INDUSTRY
SERGIO F. GALEANO
A DISSERTATION PRESENTED TO THE GRADUATE COUNCIL OF
THE UNIVERSITY OF FLORIDA
IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE
DEGREE OF DOCTOR OF PHILOSOPHY
UNIVERSITY OF FLORIDA
MARGARITA ANDREU SANTOS
The author wishes to express his grateful appreciation to the
chairman of his supervisory committee, Dr. E. R. Hendrickson, for his
concern and guidance during this course of study. The supervision and
valuable suggestions of Dr. C. I. Harding, co-chairman of the
supervisory committee are greatly appreciated. To Dr. A. P. Black,
Professor T. deS. Furman, and Dr. M. Tyner, the author's gratitude
for their help and understanding as members of the committee.
To Professor B. Spangler, who approved and supervised structural
details of the pilot plant, the author is indebted.
Appreciation is expressed to Dr. R. S. Sholtes and Dr. J. J.
Morgan for their helpful suggestions in different phases of this work.
Appreciation is also extended to Mrs. J. Larson and Mrs. Marjorie
Turner for their typing of the dissertation.
Expressing gratitude to each one of the University of Florida
personnel who have contributed in one way or another in different aspects
of this work will be too long, thus it will suffice to mention Mr. H.
McGraw, Mr. W. Shumate, Mr. E. Warshyk, Mr. B. Crenshaw, Mr. R. Booth,
Mr. W. Murray, and Mr. J. Smith, among others.
The donation of the structural steel by the Aetna Steel Company
is duly appreciated.
This research was supported by a grant from Owens-Illinois, Forest
Product Division and Public Health Service Air Pollution Training Grant
T-l-AP-4. The pilot scrubber and blower were donated to the Air Pollution
Research Laboratory by Airetron Engineering Corporation, Ridgewood, N. J.
TABLE OF CONTENTS
ACKNOWLEDGMENTS . . . . . . . . . . . . .
LIST OF TABLES . . . . . . . . . . . . .
LIST OF FIGURES . . . . . . . . . . . . .
ABSTRACT . . . . . . . . . . . . . .
I. INTRODUCTION . . . . . . . . . . .
Industrial Development . . . .
The Air Pollution Problem . . .
Process Development in Removing Sulfur
from Flue Gases . . . . .
Chemical Recovery . . . . .
Purpose of this Investigation . .
II. FUNDAMENTAL THEORY . . . . .
Gas Absorption in General .
Theories Explaining Absorption .
Absorption with Chemical Reaction
Rate of Absorption in the Venturi
Rate of Absorption in the Spray Cha
Chemistry of the Sulfite-Bisulfite
Foaming . . . . . .
. . . . 9
. . . . 13
. . . . 17
. . . . 26
. . . . 33
. . . . 33
. . . . 39
. . . . 43
. . . . 46
mber . . ... 52
System .... 53
. . . . 63
III. PROJECT DESIGN . . . . . . . . . .
The Experiment . . . . . . . . . .
The Design . . . . . . . . . . .
The Analysis . . . . . . . . . .
IV. EXPERIMENTAL RESULTS . . . . . . . . .
Scrubbing with Water . . . . . . . .
Experiments in the Spray Chamber . . . .
Experiments in the Venturi . . . . .
Scrubbing with Sodium Carbonate . . . . .
Experiments in the Spray Chamber . . . .
Experiments in the Venturi . . . . .
Oxidation of the Weak Black Liquor . . . ... 91
Foaming . . . . . . . . . . 93
Scrubbing with the Weak Kraft Black Liquor .... 94
Economic Evaluation . . . . . . .... 97
V. CONCLUSIONS . . . . . . . . . . . 99
A. EXPERIMENTAL PILOT PLANT . . . . . . ... 102
Heating Plant No. 1 . . . . . . ... 102
Structural Details . . . . . . . ... .104
Gas Flow System . . . . . . . .... 106
Liquid Scrubbing Flow System . . . . .... 112
Recording Systems . . . . . . . ... .120
Electrical Installation . . . . . .... 129
Cooling System . . . . . . . .... 131
Steam and Air Lines . . . . . . ... .131
Transferring the Weak Kraft Black Liquor . . .. .134
B. EXPERIMENTAL AND ANALYTICAL PROCEDURES . . . ... .137
Preparation of the Scrubbing Solutions . . ... .137
Performing the Experiment . . . . .... 139
Sampling and Analysis for Sulfur Dioxide . ... 140
Analysis for Sulfur Content as Sulfates . . .. .148
Sampling and Analysis of Hydrogen Sulfide .... .148
Sulfide Determinations in Kraft Weak Black Liquor 150
Total Solids of Black Liquor . . . . .... 153
C. DATA OF THE EXPERIMENTATION WITH THE CARBONATE SOLUTION. 154
D. EXPERIMENTAL DATA OF THE WEAK BLACK LIQUOR OXIDATION 207
E. DATA OF THE EXPERIMENTATION WITH THE WEAK BLACK LIQUOR 210
F. ECONOMIC EVALUATION . . . . . . . ... .216
G. RECOVERY IN THE BLACK LIQUOR SCRUBBING SYSTEM .... .224
LIST OF REFERENCES . . . . . . . . . . . 226
BIOGRAPHICAL SKETCH . . . . . . . . ... . .237
LIST OF TABLES
1. CLASSIFICATION OF EFFORTS IN SULFUR DIOXIDE REDUCTION
FROM BOILER FLUE GASES20 . . . . . . . .... 14
2. EXPERIMENTAL FACTORS AND THEIR LEVELS . . . . .. 73
3. PAYOUT PERIOD CALCULATIONS . . . . . . .... 98
4. EXPERIMENTS IN THE VENTURI. WATER AS SCRUBBING LIQUOR.
FIELD DATA . . . . . . . . . . . . 155
5. EXPERIMENTS IN THE VENTURI. WATER AS SCRUBBING LIQUOR.
SO2 REMOVAL . . . . . . . . ... ..... 157
6. EXPERIMENTS IN THE VENTURI. WATER AS SCRUBBING LIQUOR.
BLOCK DESIGN RESULTS . . . . . . . .... .158
7. EXPERIMENTS IN THE VENTURI. WATER AS SCRUBBING LIQUOR.
BREAKDOWN ANOVA . . . . . . . .... ..... 159
8. ESTIMATES OF THE CONSTANTS AND REGRESSION ANALYSIS FOR
EQUATION (79) . . . . . . . . . . . 160
9. EXPERIMENTS IN THE VENTURI. WATER AS SCRUBBING LIQUOR.
ESTIMATES OF THE INTERFACIAL AREA . . . . .... 161
10. EXPERIMENTS IN THE VENTURI. WATER AS SCRUBBING LIQUOR.
ESTIMATES OF THE OVER-ALL MASS TRANSFER COEFFICIENTS . . 163
11. EXPERIMENTS IN THE SPRAY CHAMBER. CARBONATE AS SCRUBBING
LIQUOR AT HIGHER TEMPERATURE. FIELD DATA . . . ... 164
12. EXPERIMENTS IN THE SPRAY CHAMBER, CARBONATE SCRUBBING
LIQUOR AT HIGHER TEMPERATURE. SO2 REMOVAL . . . ... .166
13. EXPERIMENTS IN THE SPRAY CHAMBER. CARBONATE SCRUBBING
LIQUOR AT HIGHER TEMPERATURE. BLOCK DESIGN RESULTS . . 167
14. EXPERIMENTS IN THE SPRAY CHAMBER. CARBONATE SCRUBBING
LIQUOR AT HIGHER TEMPERATURE. BREAKDOWN ANOVA ...... 168
15. ESTIMATES OF THE CONSTANTS AND REGRESSION ANALYSIS FOR
EQUATION (81) . . . . . . . . . . . 169
16. EXPERIMENTS IN THE SPRAY CHAMBER. CARBONATE SCRUBBING
LIQUOR AT LOWER TEMPERATURE. FIELD DATA . . . ... 170
17. EXPERIMENTS IN THE SPRAY CHAMBER. CARBONATE SCRUBBING
LIQUOR AT LOWER TEMPERATURE. SO2 REMOVAL . . . ... .172
18. EXPERIMENTS IN THE SPRAY CHAMBER. CARBONATE SCRUBBING
LIQUOR AT LOWER TEMPERATURE. BLOCK DESIGN RESULTS .... .173
19. EXPERIMENTS IN THE SPRAY CHAMBER. CARBONATE SCRUBBING
LIQUOR AT LOWER TEMPERATURE. BREAKDOWN ANOVA . . ... 174
20. ESTIMATES OF THE CONSTANTS AND REGRESSION ANALYSIS FOR
EQUATION (84) . . . . . . . . . . . 175
21. EXPERIMENTS IN THE SPRAY CHAMBER. CARBONATE SCRUBBING
LIQUOR. BREAKDOWN ANOVA FOR TOTAL EXPERIMENT ...... .176
22. EXPERIMENTS IN THE SPRAY CHAMBER. CARBONATE SCRUBBING
LIQUOR. ESTIMATES OF THE HEIGHT OF TRANSFER UNITS .... .177
23. EXPERIMENTS IN THE VENTURI. CARBONATE SCRUBBING LIQUOR.
LOW TEMPERATURE AND HIGH RATES OF FLOW. FIELD DATA . . 178
24. EXPERIMENTS IN THE VENTURI. CARBONATE SCRUBBING LIQUOR.
LOW TEMPERATURE AND HIGH RATES OF FLOW. SO2 REMOVAL . . 180
25. EXPERIMENTS IN THE VENTURI. CARBONATE SCRUBBING LIQUOR.
LOW TEMPERATURE AND HIGH RATES OF FLOW. BLOCK DESIGN
RESULTS . . . . . . . . . . . . . 181
26. EXPERIMENTS IN THE VENTURI. CARBONATE SCRUBBING LIQUOR.
LOW TEMPERATURE AND HIGH RATES OF FLOW. BREAKDOWN ANOVA .182
27. ESTIMATES OF THE CONSTANTS AND REGRESSION ANALYSIS FOR
EQUATION (86) . . . . . . . . . . . 183
28. EXPERIMENTS IN THE VENTURI. CARBONATE SCRUBBING LIQUOR.
LOW TEMPERATURE AND HIGH RATES OF FLOW. ESTIMATES OF THE
INTERFACIAL AREA . . . . . . . . ... . .184
29. EXPERIMENTS IN THE VENTURI. CARBONATE SCRUBBING LIQUOR.
LOW TEMPERATURE AND HIGH RATES OF FLOW. ESTIMATES OF THE
OVER-ALL MASS TRANSFER COEFFICIENT . . . . . . 186
30. EXPERIMENTS IN THE VENTURI. CARBONATE SCRUBBING LIQUOR.
HIGH TEMPERATURE AND HIGH RATES OF FLOW. FIELD DATA . . 187
31. EXPERIMENTS IN THE VENTURI. CARBONATE SCRUBBING LIQUOR.
HIGH TEMPERATURE AND HIGH RATES OF FLOW . . . ... .189
32. EXPERIMENTS IN THE VENTURI. CARBONATE SCRUBBING LIQUOR.
HIGH TEMPERATURE AND HIGH RATES OF FLOW. BLOCK DESIGN
RESULTS . . . . . . . . . . . . . 190
33. EXPERIMENTS IN THE VENTURI. CARBONATE SCRUBBING LIQUOR.
HIGH TEMPERATURE AND HIGH RATES OF FLOW. BREAKDOWN ANOVA 191
34. ESTIMATES OF THE CONSTANTS AND REGRESSION ANALYSIS FOR
EQUATION (88) . . . . . . . . . . . 192
35. EXPERIMENTS IN THE VENTURI. CARBONATE SCRUBBING LIQUOR.
HIGH TEMPERATURES AND HIGH RATES OF FLOW. ESTIMATES OF
THE INTERFACIAL AREA . . . . . . . .... .193
36. EXPERIMENTS IN THE VENTURI. CARBONATE SCRUBBING LIQUOR.
HIGH TEMPERATURE AND HIGH RATES OF FLOW. ESTIMATES OF THE
OVER-ALL MASS TRANSFER COEFFICIENT . . . . . ... .195
37. EXPERIMENTATION IN THE VENTURI. COMPLETE BLOCK DESIGN
RESULTS (SO2 REMOVAL). .. . . . . . . . .196
38. EXPERIMENTS IN THE VENTURI. CARBONATE SCRUBBING LIQUOR.
BREAKDOWN ANOVA FOR TOTAL EXPERIMENT . . . . ... .197
39. EXPERIMENTS IN THE VENTURI. CARBONATE SCRUBBING LIQUOR.
HIGH TEMPERATURE AND LOW RATES OF FLOW. FIELD DATA . . 198
40. EXPERIMENTS IN THE VENTURI. CARBONATE SCRUBBING LIQUOR.
HIGH TEMPERATURE AND LOW RATES OF FLOW. SO2 REMOVAL . . 200
41. EXPERIMENTS IN THE VENTURI. CARBONATE SCRUBBING LIQUOR.
HIGH TEMPERATURE AND LOW RATES OF FLOW. BLOCK DESIGN
RESULTS . . . . . . . . . . . . . 201
42. EXPERIMENTS IN THE VENTURI. CARBONATE SCRUBBING LIQUOR.
HIGH TEMPERATURES AND LOW RATES OF FLOW. BREAKDOWN ANOVA 202
43. ESTIMATES OF THE CONSTANTS AND REGRESSION ANALYSIS FOR
EQUATION (90) . . . . . . . . . . . 203
44. EXPERIMENTS IN THE VENTURI. CARBONATE SCRUBBING LIQUOR.
HIGH TEMPERATURE AND LOW RATES OF FLOW. ESTIMATES OF THE
INTERFACIAL AREA . . . . . . . . . . .. .204
45. EXPERIMENTS IN THE VENTURI. CARBONATE SCRUBBING LIQUOR.
HIGH TEMPERATURE AND LOW RATES OF FLOW. ESTIMATES OF THE
OVER-ALL MASS TRANSFER COEFFICIENTS . . . . .... 206
46. OXIDATION OF THE WEAK BLACK LIQUOR . . . . . .. 208
47. EXPERIMENTS IN THE VENTURI. WEAK BLACK LIQUOR AS A
SCRUBBING SOLUTION. FIELD DATA . . . . . .... 211
48. EXPERIMENTS IN THE VENTURI. WEAK BLACK LIQUOR. BLOCK
DESIGN RESULTS . . . . . . . . . . . 212
49. EXPERIMENTS IN THE VENTURI. WEAK BLACK LIQUOR AS SCRUBBING
SOLUTION. BREAKDOWN ANOVA . . . . . . . .. 213
50. EXPERIMENTS IN THE VENTURI. WEAK BLACK LIQUOR AS SCRUBBING
SOLUTION. EFFECTS OF SULFATES ON SO2 ABSORPTION . . .. .214
51. SCRUBBING WITH WEAK BLACK LIQUOR. RECIRCULATION
EXPERIMENTS . . . . . . . . . . . . 215
52. GENERAL CONSIDERATIONS . . . . . . . ... 217
53. RECIRCULATION EXPERIMENTS. OPTIMUM CONDITIONS ...... 218
54. RECIRCULATION EXPERIMENTS. INDUSTRIAL CONDITIONS . .. .219
55. RECIRCULATION EXPERIMENTS AT LOW SO2 CONCENTRATION .... .220
56. EQUIPMENT AND INSTALLATION COSTS . . . . . ... .221
57. FIXED-CAPITAL INVESTMENT . . . . . . . .... 222
58. ESTIMATED MANUFACTURING COST TO PRODUCE (1000 GAL) OF
BY-PRODUCT LIQUOR 0.6 M IN Na2SO3 . . . . . .. 223
LIST OF FIGURES
1. SCHEMATIC DIAGRAM OF A TYPICAL KRAFT PROCESS5 ...... 6
2. SCHEMATIC DIAGRAM OF A TYPICAL NSSC PROCESS . . . .. 10
3. SIMPLIFIED FLOW SHEET OF THE MEAD PROCESS . . . ... .22
4. THE INSTITUTE PROCESS OF S02 RECOVERY BY DIRECT
SULFITATION . . .. . . . . . . . .... . 25
5. EQUILIBRIUM DIAGRAM FOR A SULFITE-BISULFITE SYSTEM . . 27
6. PROPOSED SYSTEM FOR SO2 REMOVAL AND RECOVERY ...... .29
7. INDUSTRIAL APPLICATIONS OF THE PROPOSED SYSTEM . . .. .31
8. EQUILIBRIUM DIAGRAM FOR SO2 ABSORPTION . . . ... .33
9. OPERATING LINE AND EQUILIBRIUM CURVE IN A GAS ABSORPTION
PROCESS . . . . . . . . . . . . . 37
10. EQUILIBRIUM DIAGRAM FOR SULFITE SPECIES . . . ... .56
11. EQUILIBRIUM DIAGRAM FOR CARBONATE SPECIES . . . ... .58
12. TITRATION CURVES pH VS. TIME. . . . . . .. .59
13. EXPERIMENTAL SET-UP FOR THE EXPERIMENTATION OF SO2
ABSORPTION . . . . . . . . . . . . 61
14. TITRATION CURVES AT DIFFERENT RATES OF SO2 INJECTION . 62
15. SCRUBBING WITH WATER IN THE VENTURI. EXPERIMENTAL RESULTS. 78
16. RESPONSE SURFACE OF EQUATION (80) . . . . . ... 80
17. SCRUBBING WITH SODIUM CARBONATE IN THE SPRAY CHAMBER AT A
HIGH TEMPERATURE. EXPERIMENTAL RESULTS . . . ... .81
18. SCRUBBING WITH SODIUM CARBONATE AT A LOW TEMPERATURE.
EXPERIMENTAL RESULTS. . . . . . . . .. .83
19. RESPONSE SURFACE FOR EQUATION (85) . . . . ... 84
20. SCRUBBING WITH SODIUM CARBONATE IN THE VENTURI AT A LOW
TEMPERATURE. EXPERIMENTAL RESULTS . . . . . . 85
21. CANONICAL REPRESENTATION OF EQUATION (87) . . . . 87
22. SCRUBBING WITH SODIUM CARBONATE IN THE VENTURI AT A HIGH
TEMPERATURE. EXPERIMENTAL RESULTS . . . . ... .88
23. CANONICAL REPRESENTATION OF EQUATION (89) . . . ... .89
24. SCRUBBING WITH SODIUM CARBONATE IN THE VENTURI, AT A HIGH
TEMPERATURE AND LOW RATES OF FLOW. EXPERIMENTAL RESULTS 90
25. COMPARISON IN REMOVAL EFFICIENCY BETWEEN THE VENTURI AND
THE SPRAY CHAMBER ................... .. 92
26. OXIDATION OF THE WEAK BLACK LIQUOR UNDER DIFFERENT
CONDITIONS . . . . . . . . . . . . 93
27. COMPLETE OXIDATION OF THE WEAK BLACK LIQUOR . . ... .94
28. FOAMING OF THE WEAK BLACK LIQUOR UPON OXIDATION . . .. .95
29. DISAPPEARANCE OF FOAM UPON ADDITION OF KEROSENE DURING
OXIDATION . . . . . . . . . . . . . 95
30. SCRUBBING WITH WEAK BLACK LIQUOR IN THE VENTURI.
EXPERIMENTAL RESULTS . . . . . . . . .. 96
31. EFFECT OF SULFATES ON SO2 REMOVAL . . . . . ... .97
32. RECORDER CHART FOR BOILING PLANT OPERATION . . ... .103
33. STRUCTURAL DETAILS OF SUPPORTS ... . . . . ... 105
34. SCALE PLOT PLAN OF ROOF . . . . . . . ... .107
35. PRESSURE BLOWER AND CONNECTIONS . . . . . ... .108
36. DETAILS OF THE VENTURI SCRUBBER . . . . . . .. 110
37. DETAILS OF THE SPRAY CHAMBER . . . . . . .. 111
38. EXPERIMENTAL PILOT PLANT. GROUND LEVEL INSTALLATIONS . 113
39. SCHEMATIC DIAGRAM OF THE LIQUID FLOW SYSTEM . . ... .114
40. OXI-HEATER TANK . . . . . . . .... ..... 115
41. SEDIMENTATION TANK . . . . . . . . . 117
42. DETAILS OF THE PUMPS AND PIPE ARRANGEMENTS AT GROUND
LEVEL . . . . . . . . . . . . . 119
43. GAS PRESSURE RECORDING SYSTEM . . . . . . ... .121
44. CALIBRATION CURVES FOR PRESSURE TRANSDUCERS . . ... 122
45. HONEYWELL RECORDER AND CONNECTIONS . . . . ... .123
46. TEMPERATURE RECORDING SYSTEM . . . . . . ... .125
47. DETAILS OF THE COPPER-CONSTANTAN THERMOCOUPLES . . .. .126
48. CALIBRATION CURVE FOR THE HONEYWELL POTENTIOMETER .... .127
49. SAMPLING SYSTEM . . . . . . . . ... . .128
50. ELECTRICAL INSTALLATION DETAILS . . . . . ... .130
51. DETAIL OF THE ELECTRICAL CONTROL PANEL AT GROUND LEVEL . 132
52. COOLING SYSTEM . . . . . . . . ... . .133
53. AIR AND STEAM SYSTEMS . . . . . . . . ... 135
54. RAILROAD CAR FOR TRANSFERRING WEAK BLACK LIQUOR . . .. .136
55. CALIBRATION CURVE FOR ORIFICE METER . . . . ... .141
56. TYPICAL TITRATION CURVE FOR SULFIDE DETERMINATION .... .152
Abstract of Dissertation Presented to the Graduate Council
in Partial Fulfillment of the Requirements for the
Degree of Doctor of Philosophy
REMOVAL AND RECOVERY OF SULFUR
DIOXIDE IN THE PULP MILL INDUSTRY
Sergio F. Galeano
Chairman: Dr. E. R. Hendrickson
Major Department: Bioenvironmental Engineering
Although pollution seems inevitable in a technological society with
a high gross national product, it has been a neglect of remedies, which
has brought about the present deterioration in the quality of the
environment. The legislation on air pollution control and the activity
of the regulatory agencies have increased in the last years in an
effort to reduce the increasing presence of pollutants in the air.
Among these pollutants, sulfur dioxide is one of the most important
not only because of its abundance but because of its detrimental effects
in animals, plants, materials, and human beings.
In a pulp mill, sulfur dioxide is present in different places along
the chemical recovery system. Sulfur dioxide emissions will take place
at the power plants of any type of pulp mill, and at the recovery
furnace of those mills engaged in the semi-chemical process.
The peculiar existing characteristics of the chemical recovery
system in the pulp mill and the fact that sulfur compounds in different
states of oxidation are required in the wood digestion process indicates
the convenience of a liquid/gas contact method to remove sulfur dioxide
from the flue gas and to recover the chemical in a proper form.
This research studies the technical and economic feasibility of
a purification system for the removal of sulfur dioxide in the pulp
mills. It comprises applications in two important types of pulping
systems. The sulfate or kraft system is the one most in use
throughout the world. The neutral semi-chemical system is gaining
popularity and improving rapidly because of wood scarcity and higher
An experimental pilot plant with a capacity of 2800 cfm consisting
of a venturi scrubber and a cyclone has been used in the different
experiments. The pilot plant was designed for its use as a purification
system for either the kraft or the semi-chemical systems.
Two different scrubbing solutions were used. For the semi-chemical
process, the use of a carbonate solution had proved technically feasible,
with sulfur dioxide removals higher than 90 per cent. For the kraft
process, the weak black liquor before entering the evaporators was used.
Removals in the same order to the ones previously mentioned has been
Two different purification units were studied with the above
mentioned scrubbing liquor: a venturi-cyclone combination, and a spray
chamber with radial inlet and lateral sprayers.
An orthogonal factorial design was conducted and the relative
significance of the operating factors was analyzed. When more than two
factors were studied a surface response equation was developed to
represent the phenomenon.
In the case of the carbonate scrubbing solution for the semi-chemical
process, the proposed system is both technically and economically feasible.
An economic evaluation shows a very small fraction of pay-out time.
In the case of the weak black liquor used as a scrubbing solution
the method is technically feasible. Its economy is a matter of more
flexibility dictated by the particular conditions of each plant.
The pulp and paper industry has experienced phenomenal progress
in the last thirty-five years. It is now the fifth largest manufacturing
industry in the United States, and predictions for even greater
development have been formulated. The actual production of paper and
paperboard in this nation, nearly 38 million tons, is expected to
increase to 50 million tons in 1975.1 Even this figure, however, will
be insufficient to meet the nation's demands.
The raw material for nearly all of the paper produced in the
United States today is wood. The history of the process of obtaining
paper from wood is somewhat vague. A decisive step towards the
manufacture of paper from pulp, made from sawdust and shavings, was
achieved by Dr. Jacob C. Schaffer, in Bavaria, around 1760. In 1800,
Matthias Koop of England disclosed his experiments in papermaking, using
wood straw and other fibers.
In the Western Hemisphere the first groundwood paper was probably
made in 1844, at Nova Scotia, by Charles Fenerty. The manufacture of
mechanical pulp moved forward slowly, gaining the acceptance of the
printers and displacing the well-used rag pulp. At the beginning,
groundwood pulp accounted for only a 20-25 per cent of the paper; rag
pulp accounted for the rest. Only after a new major development took
place in 1866 was it possible for groundwood pulp to gain complete
acceptance. This major development was the production of commercial
groundwood pulp at the Buntin Mill, Valleyfield, Quebec.
The basic principles of manufacturing groundwood or mechanical
pulp are simple, although the actual practice is quite complicated. It
is more an art than a science when compared to other methods of pulping.
A block of wood is forced by pressure against a grindstone, with water
being supplied continuously to keep the stone cool and to remove the
pulp from it. The action of the abrasive surface of the revolving stone
is such that the wood is reduced to a fibrous condition.
Sulfite or Acid Pulping
Man's continual need for improved methods of production brought
about new achievements in this field. In 1867, Benjamin Tilghman was
granted the U. S. Patent 70485, "Treating Vegetable Substances for
Making Pulp Paper." Thus a new process, afterwards named the sulfite
pulp process, was originated, which quickly began to displace the still-
new mechanical pulp process. The next decade witnessed a profusion of
investigations and works which greatly improved the basic idea of
Tilghman, and the original works of a Swedish chemist, C. D. Ekman, who
worked out a similar process, independently from Tilghman, about the
Essentially, the sulfite or acid process consists of the digestion
of wood in the form of chips at temperatures from 130 C to 1500C in an
aqueous solution containing alkaline-earth bisulfites, usually calcium
bisulfite or a mixture of calcium and magnesium bisulfites, and an
excess of sulfur dioxide. During the course of the pulping reaction,
the quantitative relationship among these components depends on the
conditions of temperature and pressure, which influence the reactions
taking place and the products of such reactions. During digestion,
lignin combines with sulfur dioxide or bisulfites and is rendered
soluble. The less resistant hemicelluloses are hydrolyzed to simpler
compounds, and a portion of the wood cellulose is degraded.
Other variants have been introduced to the above mentioned method.
It is sufficient to describe the most successful. Sulfite or acid
pulping has been improved by the use of magnesium acid sulfite, as
cooking liquor, instead of the calcium-base liquor. In this way,
economical heat and chemical recovery is possible, together with a
reduction in stream pollution problems. Today, nearly 700,000 tons,
or about 25 per cent of the total sulfite pulp manufacture in this
country, are produced annually with this process.
Kraft or Sulfate Pulping
Commercial paper in the eighteenth century was the product of
chemical pulping. Linen or cotton rags, straw and other non-wood,
fibrous materials were used, and the digestion was carried out in open
boilers. Strackan has reported that both sodium hydroxide and sodium
sulfide were used. This practice was, in effect, the beginning of a
process later to be called alkaline pulping. Because wood was less
readily delignified than straw upon alkaline digestion, the acceptance
of the alkaline pulping process for wood was retarded. New methods
for putting alkaline pulp production on a commercial basis were needed.
These methods involved higher pressure and temperature, and a relatively
large quantity of the reagents. Two major developments helped to do
this. First, in 1879, Dahl, in Danzig, introduced a modification in
the process by using a combination of caustic soda and sodium sulfide
as reagents. The latter was obtained as a reduction product of sodium
sulfate, which was previously added as make-up in the process. The
addition of this chemical gave the name of sulfate to the process. It
is also known as the "kraft" process, meaning strong, because of the
high strength of the paper manufactured from this pulp. The use of
sodium sulfate reduced considerably the cost of pulping, making the
process commercially feasible.
The second major development which helped the alkaline pulping to
attain its position was a unique combination of technical achievement
and consumer acceptance. The quality of the paper converted from kraft
pulp lacked the high opacity, excellent formation, and bulk character-
istics of the paper converted from soda pulp. The development of the
electrolytic process for the simultaneous fabrication of sodium hydroxide
and chlorine from sodium chloride made possible the commercial use of
soda pulp. For many years, mills using the sulfate process were engaged
in manufacturing a coarser paper, and those using the soda process in
manufacturing a finer paper.
The differences in the quality of paper between the two alkaline
processes existed until 1930. At that time, successful investigations
on chemical recovery and bleaching methods helped to make these two
methods quite similar. At the present time, there are only a few soda
plants as such in existence since in the majority of the plants sodium
sulfide is added as part of the process. Fine paper is no longer
produced exclusively by the soda process, since the multiple-stage
bleaching demonstrated that alkaline pulps from softwood could be
The importance of the kraft or sulfate process has been generally
recognized. Since the first kraft mill was built in 1891, the sulfate
process has gradually outgrown all the others. Since World War II,
this growth has been remarkable. Today, more than 17 million tons of
kraft pulp are produced annually in this country, which is by far more
than half of the total wood pulp in the United States.
A modern kraft pulping process could be visualized with the aid of
Figure 1. The liquor, consisting mainly of a solution of sodium sulfide
and sodium hydroxide in water is mixed with wood chips in a pressure
vessel, the digestor, and cooked for about three hours with steam at a
gauge pressure of about 110 psi. Upon completion of the cooking phase,
the bottom of the digester is opened and the digestor's contents are
forced by differential pressure into a blow tank. An important
separation then takes place. The pulp in the blow tank is diluted and
pumped to multi-stage drum filters. The spent liquor is separated from
the pulp by means of countercurrent washing with fresh water. The pulp
follows several steps which constitute the kraft pulping process, and
the spent liquor, called weak black liquor, enters the chemical recovery
process, along which several pollution sources are present.
The chemical recovery process starts when the weak black liquor is
concentrated in multi-effect evaporators to from about 15 per cent solids
to about 45 per cent solids. The make-up, sodium sulfate, and recovered
Ca(OH2 L MLUD(QCO
SLAKER LME KILN
Fig. i.- Schematic diagram of a typical kraft process.
salt-cake are added prior to the direct-heat evaporators. Further
concentration takes place in a direct contact evaporator where flue
gases from the recovery furnace are used to evaporate water. The
concentrated black liquor is forced through nozzles and sprayed into
the recovery furnace. Reducing conditions are maintained in the lower
part of the furnace to obtain a reduced form of sulfur compound. The
resulting smelt from the furnace consists essentially of sodium sulfide
and sodium carbonate. Upon dissolution in water, a green liquor results
in which the sodium sulfide is converted into sodium hydroxide and
sodium hydrosulfide. Sodium carbonate is causticized by addition of
line to form sodium hydroxide with the precipitation of calcium
carbonate. The causticized solution, called white liquor, is ready for
use in cooking.
The general practice today is the disposal of the spent liquor by
burning, thus recovering chemicals and heat. This practice transforms
the lignin content in the liquor into heat. Although it is an economical
method of disposal of the spent liquor, it is a most wasteful manner of
treating such potentially valuable chemical raw material. Alkali lignin,
very similar to the ligno-sulfates regarding chemical and physical
properties, has been studied recently by many researchers, who pointed
out.its potentiality for new products in a more scientific chemical
This process is a direct result of both the increasing demand of pulp
and the search for energy savings in pulping operation. While wood from
coniferous species is more widely used because of its long-fibered pulp
and good bleaching characteristics, the wood from broad-leaf trees is
being employed in ever increasing amounts. The production of high-yield
pulps from different types of deciduous woods has been extended to
non-wood fibrous materials such as sugar cane bagasse and cereal straws.
Further research on weeds and grasses is being conducted with a certain
degree of success.
The energy requirements for pulping have been a matter of serious
consideration for many years. From 1875 to 1890, many investigators,
such as George Marshall in the United States, Cross in England, and
Enge in Germany, made progress along similar lines to lower the energy
consumption of the conventional mechanical process by using a chemical
pretreatment. One of the greatest advantages of the semi-chemical
process is that a high-yield pulping is obtained in which much of the
hemicellulose lost in conventional chemical pulping is saved.
The semi-chemical process can be divided into two stages, chemical
and mechanical. In the first process, the fibrous raw material is
subjected to chemical action. This causes a chemical reaction with the
lignin-carbohydrate complex of the middle lamella which partially
weakens or destroys the fibrous bond. The second stage is the one
which differentiates this process from full chemical pulping. The
material from the chemical stage has been partially disintegrated and
defibered, to about 20 per cent of its original fiber bonds. During
the defibering-refining process, the combined action of friction forces
among fibers, and between them and the plates, plus the compression
created by the centrifugal forces of the discs, complete the process.
The heat generated as a result of the transformation of energy from the
action of frictional forces contributes to further weaken or partially
dissolve the fiber bond.
According to the specific composition and acid-base characteristics
of the digesting liquor, the semi-chemical process consists of various
processes: the acid sulfite semi-chemical (ASSC), the neutral sulfite
semi-chemical (NSSC), the soda semi-chemical (SSC), the cold soda
semi-chemical (CSSC), and the kraft semi-chemical (KSC) process. Each
system works differently in removing the basic constituents of the
natural fibrous materials, such as lignin, alpha cellulose, hemicellulose,
and extractive. For the different fibrous materials, the amounts
extracted vary with the specific process. It suffices to say that the
NSSC process yields fairly good extractions for all species in
comparison with the combined yields of all the other processes.
Figure 2 indicates the important features of a NSSC pulping plant.
The Air Pollution Problem
Since the beginning of the pulp and paper industry, the operators
of the mills have been aware of the air pollution problem involved.
In the last decade, vigorous efforts have been made to reduce the air
pollution emissions. These efforts are reflected by different
activities. Installation of control units at the exhaust sources,
ambient sampling programs, and process changes are the principal
activities in which the industry has engaged in order to cope with the
air pollution problem.
co i- U
In a pulp mill both particulates and odorous gases are involved.
Many investigators have reported qualitatively and quantitatively the
different air-borne wastes. Kraft pulping could be considered a
typical offender due to tie amount and variety of pollutants emitted.
Hendrickson reported that the three major sources of pollutant emissions
in kraft pulping are from liquor preparation, cooking, and chemical
recovery. In the previously outlined processes, the major sources of
sulfur dioxide, an important pollutant, result from the recovery
furnace and the boiler plant.
Since the pulp mill uses large amounts of steam, and since generally
the combustion of the black liquor does not supply all of the steam
needed, any other amount of steam needed for the process is generated
in primary fuel boilers. During the generation of this needed steam,
sulfur dioxide is emitted in those plants using coal or fuel oil.
Sulfur dioxide is of importance because, at certain concentrations,
it corrodes building materials, destroys cloth and leather, and affects
vegetation, animals and human health. The detrimental effect of
pollutants on human health is well known. When sulfur dioxide is
present at high concentrations its principal effect is irritation of
the respiratory tract. At lower concentrations it increases the possi-
bility of chronic respiratory diseases such as asthma, bronchitis and
pulmonary emphysema. The total anatomical system and the total integrated
physiological system (TIPS) of man depend upon the totality of his
environment, mainly the atmosphere. There is an evident liaison between
two of the several subsystems which constitute the TIPS. The respiratory-
metabolic subsystem (R-MS) is in direct contact with the atmosphere.
Its performance will depend on the degree of pollution of the atmosphere.
The cardiovascular subsystem (CVS) is in direct contact with the R-MS
and, through this contact, comes into indirect contact with the
atmosphere. Of the two subsystems, the R-MS is the controlling one and
the CVS the controlled. Gaseous pollutants in the inspired air can
become trapped in the alveolar air and can impede the exchange of gases
across the alveolar membrane. In that case, less oxygen would diffuse,
and hypoxia and some human cardioaccelerative responses would be
intensified by the presence of gaseous pollutants detrimental to the
system. Literature on the effects of sulfur dioxide is available in
An estimate of 25 million tons of sulfur dioxide are released
annually into the atmosphere in the United States.4 This problem will
be aggravated in the coming years. Public awareness of air pollution
has been intensified in the two decades since the episode at Donora,
Pennsylvania,15 and an increasing number of air pollution laws have
resulted. The passage of the Federal Clean Air Act of 1963 and the
broadening of the Act in 1965 indicate the national magnitude of the
problem. The presence of sulfur dioxide resulting from the combustion
of fuels was one of the two problems singled out by the Act of 1963 for
special emphasis. The determination of the regulatory agencies to solve
this problem has been stated by responsible officials.16
Coal and fuel oil No. 6 are the most common solid and liquid fuels
for high capacity boilers.17 Sulfur content in the fuel varies from
0.34 to 4.0 per cent by weight, depending on the origin of the fuel.
Ash content by weight ranges from 0.2 to 1.5 per cent. The increased
use of foreign fuels and new refinery practices have increased both the
sulfur and ash content. During combustion, virtually all of the
sulfur in the fuel is oxidized to sulfur dioxide and sulfur trioxide.
It is estimated that about 98 per cent of the sulfur in the fuel is
emitted as sulfur dioxide, 1 per cent as sulfur trioxide, and the
remaining 1 per cent is contained in the ash.19
Pollution control in the pulp mill industry seems to be a
difficult and complex problem. Not only is the number and quantity of
pollutants emitted a factor, but the more promising industrial control
methods need to be conducted within narrow limits. Paradoxically, the
very essence of pulping economy lies in the recovery process, and it
is here that air pollution occurs.
Process Development in Removing Sulfur Dioxide from Flue Gases
The reduction of the concentration of sulfur oxides in the combustion
products from boiler plants has been the object of intensive research
from various angles in recent decades. Both technical and economic
feasibilities are necessary for acceptance of any system. It means
that the process applied will operate efficiently and continuously over
long periods of time without interfering with the operation of the plant.
Thus, investment and operating costs of control units must be below the
cost of burning premium low-sulfur fuels.
Different approaches have been undertaken to achieve the desired
result. Table 1 summarizes these efforts. Their degree of success
varies according to the over-all industrial process.
Classification of Efforts in Sulfur Dioxide
Reduction from Boiler Flue Gases20
Prior to combustion Selection of fuels
After combustion Liquid scrubbing
Solid phase reaction
Gasification of coal
Hydrodesulfurization of oil
sorption with metallic
Gas-Liquid Contact Methods
Of the few commercial installations for removing sulfur dioxide
from power plant flue gases, gas-liquid contact is used almost exclusively.
The method is expensive and ground level pollution near the stack may be
increased because of the lower exit temperature of the flue gas if the
removal of the pollutant is not complete. Rees21 enumerates some of
the apparently insurmountable difficulties found in its application, the
most important two of which are the need for a high percentage of removal
of sulfur dioxide, and the recovery of sulfur in a soluble form.
Because of the availability of alkaline solutions in the pulp mill
process, gas-liquid contact removal methods might prove feasible. The
characteristics of the enriched scrubbing liquor dictate the use of a
To describe the progress in the removal of sulfur dioxide from
flue gases, it will be desirable to review the available literature
beginning with the catastrophic episode in the Meuse Valley in France
in 1930.22 Sulfur gases resulting from fuel combustion were officially
identified as responsible for the damage to vegetation, and consequently
the allowable amount of sulfur dioxide emitted was reduced.
The early work consisted mainly in the production of sulfuric acid
from the absorption and oxidation of sulfur dioxide in the presence of
catalysts in aqueous solutions. Copson and Payne started investigations
on sulfur dioxide removal using waste gases from petroleum refineries
with sulfur dioxide concentrations from 0.6 to 12 per cent volume.
Absorption rates in water using bubble washers were reported. Johnstone,
at the University of Illinois, analyzed the importance of catalytic
oxidation of sulfur dioxide using manganese and iron as catalysts.
Inhibitions of the catalysts were reported. In 1935, Grodzovskii
pointed out the acceleration of sulfur dioxide oxidation when ozone is
used in combination with manganese. Along this line, in 1945, Walthall
and co-workers developed a process for recovering sulfur dioxide.
Later in 1957, Tarbutton, et al., at the Tennessee Valley Authority,
revised the experimental cyclic method of sulfur dioxide removal with
manganese oxide and ozone, reaching the conclusion that although the
method is feasible technically, it is not so economically.
A brief survey of the use of lime as scrubbing liquor may be
summarized by saying that the method has been proved to be uneconomical.
The only cyclic process in use was the Howden process,2 9 which used
5 to 10 per cent of lime as a scrubbing solution. Conversion of the
calcium sulfate to ammonium sulfate gave a salable product.
Since 1937, the use of ammonia solution instead of lime has been
thoroughly studied, both in Europe and America. From Europe two main
systems have been patented, the Katasulf method and the Simon-Carves3
method. Johnstone333 has published complete data on the NH3-SO2
system fully proving the greater solubility of sulfur dioxide in
ammonia solution. Contributions along this line have been furnished
by Craxford, et al.,3 and Katz and Cole, among others. Later on, in
1948, Hixson and Miller36 proposed a more economical way of ammonia
regeneration than formerly proposed. Newall has reported recently
his work in the ammonia process. He indicated that the method is
limited by the cost of ammonia liquor and the price of salable
ammonium sulfate.38 The same conclusion was reached more recently by
Field, et al., in his cost analysis of SO2 removal processes.
The sodium sulfite-bisulfite system represents another line of
experimentation using gas-liquid contact methods. The system allows
for a lower percentage of removal, but favors chemical regeneration.
Johnstone, at al.,40,41,42 employed a method of regeneration based upon
addition of zinc oxide.
The description of the process development on sulfur dioxide with
specific application to the pulp industry would not be complete without
mentioning future developments. The technology of pulp processing,
related to heat and chemical recovery, has been the same for the last
thirty years. This situation is becoming critical, and two factors
will inevitably bring about a new picture in the pulp industry. The
enforcement of air pollution control regulations and the improvement
in the recovery of available products in the final cooking liquor
together will change substantially the pulp process. Actual air
pollution problems are brought about in the chemical recovery process.
When the research on better utilization of the black liquor makes
possible direct recovery from it, then it will be possible to justify
new methods on sulfur dioxide removal from flue gases. Methods
suggested by the studies initiated by Kurtzrock, et al.,43 and
Bienstock, et al., among others, would then have better justification.
Since the pulp industry utilizes sulfur compounds in the digesting
liquor, recovery of sulfur in a suitable form has been sought as a way
to reduce the cost of chemicals, and partially offset the expenditures
for any air pollution control system. When a satisfactory gas
scavenging system is developed, it might find acceptance in any of the
three major pulping systems.
In the sulfite pulping system, experimentation on the production
of cooking acid has been conducted primarily on calcium bisulfite
systems. A great amount of information on calcium oxide-sodium
dioxide-water systems has been made available by Maas and co-workers.45'46
Johnstone and Leppla47 have also given information on equilibrium of
sulfur dioxide. White, et al.,48 in 1948, gave more information on this
system. A slight variation on the sulfite method has been introduced
by which a magnesium or an ammonia base substitutes for the calcium
base. A careful study of the method brought about the conversion of
the Weyerhaeuser plant at Longview, Washington, from calcium to
magnesium base.49 Markaut, t al.,50 recently published their work on
sulfur dioxide absorption studies in a magnesium oxide-sulfur dioxide
system. A more recent work using a combined sodium-calcium base
process was reported by Schmied, et al.5
The numerous problems involved in scrubbing solutions with calcium
and magnesium bases justify experimentation with ammonia bases in
which the scaling problems could be minimized. Ammonia-base sulfite
cooking acid has been studied by Merrimer and Whitney. The use of
ammonia base introduces certain modifications in the process because
ammonia is gaseous at normal temperature and relatively volatile in
Kraft or Sulfate Pulping
Not only in the sulfite pulping system does sulfur dioxide
absorption present possibilities but also the sulfate or kraft pulping
system can benefit from scavenging sulfur dioxide. Operationally, the
method has been used to increase the concentration of the black liquor
from the multiple effect evaporators, and to recover large percentages
of chemicals from the flue gases of the recovery furnace. For this
latter application a venturi scrubber, which removes not only gases but
fumes and particulates, has been utilized primarily.
In dealing with black liquor as the scrubbing solution, many
complexities are present. Because the rate of absorption and removal
of particles is inversely proportional to the viscosity of the liquid,
a weak black liquor achieves better removal than the concentrated black
liquor, as stated by West, et al. Paradoxically, the benefits of this fact
in the past have been hampered by the foaming properties of the black
liquor. The excessive foaming from southern pine black liquor is
produced in inverse proportion to the black liquor solids concentration.
Anti-foaming devices and chemicals reportedly are expensive and
complicated to operate.
Another major item in dealing with sulfate black liquor as a
scrubbing medium is the emission of sulfur compounds during the flow
of black liquor from the digester to the recovery furnace. In the
digesters, the sulfide ion from the sodium sulfide combines with various
organic compounds of cellulose and lignin to form methyl mercaptan
(CH3HS), dimethyl sulfide (CH3)2S, dimethyl disulfide (CII3-S-CH3-S)
and other organic sulfur compounds.
The resulting liquor contains these compounds plus the original
excess of sodium hydrosulfide and sodium sulfide. In contact with the
flue gases, rich in carbon dioxide, the following reaction takes place
as indicated by Wright:5
Na2S + CO2 + 1120 -- Na2C03 + 2S (1)
Hydrogen sulfide is not the only gas emitted. Methyl mercaptan
and methyl sulfide are also released, which constitute much of the air
pollution odor problem associated with kraft pulp mills. This problem
was studied twenty years ago by Trobeck55 in Sweden and by Tomlinson
and Fergurson and Wright in Canada. These studies led to the
development of the black liquor oxidation method by which the sodium
sulfide is converted to sodium thiosulfate, and methyl mercaptan is
oxidized to dimethyl disulfide by the following reactions:
2 Na2S + H20 + 2 02 Na2 S203 + 2 NaOH (2)
4 CH3 SH + 02 -- 2(CH3)2S2 + 2 H20 (3)
The very extensive literature on black liquor oxidation deals mainly
with the problem of foaming. A good survey of the literature on this
subject is found in the work of Landry,58 but still the foaming problem,
in this country and abroad, has not been solved successfully.
Black liquor as a scrubbing agent is being used mainly in venturi
scrubbers in which the removal of gaseous products is efficiently
accompanied by the removal of particulates. Collins, at al., has
reported the use of venturi systems twenty years ago. Hendrickson and
Harding reported good results in laboratory studies with black liquor
in removing sulfur dioxide and other gases.
The application in semi-chemical pulping of gas-liquid contact
methods for SO2 scavenging looks promising. Neutral and acid sulfite
semi-chemical pulping can benefit from the chemical properties of the
resulting scrubbing liquor. This process utilizes a digesting solution
based on sulfites and bisulfites with buffering characteristics. If a
similar liquor could be produced by scrubbing sulfur dioxide with a
carbonate buffered solution, the economic feasibility of the system
will be guaranteed.
Chemical recovery in semi-chemical pulping is low. The small
amount of heat that can be recovered from the waste liquor is a result
of the high yield process. The heat in the waste liquor solids
resulting from tlle manufacture of one ton of NSSC pulp ranges between
6 and 12 million BTU, whereas in the draft process it averages nearly
20 million BTU per ton of pulp manufactured.
Recovery processes in the NSSC method are very similar to the kraft
process. In many cases the NSSC method has been put in an existing
kraft mill and the waste liquor has been used to supply the make-up
soda and sulfur to the kraft system. In each one of the three main
processes used in NSSC recovery, the waste liquor is evaporated and
burned in a kraft-type furnace. This furnace operates on a two-stage
combustion process. A primary zone or hearth operates in a reducing
atmosphere with a small portion of air admitted to it. The sulfur
compounds from this primary zone are in the sulfide form, the state of
maximum reduction. Gases leaving the primary zone are burned with
additional air. Theoretically, the furnace should be of the type used
by the acid pulping industry in which the sulfur is recovered as sulfite,
a higher level of oxidation. Nevertheless, these methods use the kraft-
The Mead Corporation Process.-6 The simplified flow sheet in Figure
3 will help to visualize the main characteristics of this process. Upon
leaving the furnace, two products are treated in different ways. The
flue gases from the furnace pass through a series of devices for purposes
of steam production. Afterwards the dust is removed, resulting in a
clean gas containing N,, C02, 02, and SO2. The gas enters a sulfiting
Fig. 3.- Simplified flow sheet of the Mead process.
tower in which SO2 is absorbed by a sodium carbonate solution. The
remaining gas is rich in CO2 and is used in the carbonation of the
green liquor in the carbonating tower, thus providing the carbonate
solution needed for scrubbing in the sulfiting tower. The role of the
carbonation tower can be understood by using some chemical equations.
A high concentration of CO2 shifts the equilibrium of equation (4) to
CO2 (aq) ; CO2 (gas) (4)
At the same time, due to this shift to the left, equations (5), (6),
and (7) also shift to the left with a resulting increase in the concen-
tration of CO3 ions. An ionic equilibrium is indicated by equations
(8) to (12).
H2CO3 (aq) H20 + CO2 (aq) (5)
2 HCO3 Z C03 + H2003 (aq) (6)
C03 + HO 2- OH + HCO3 (7)
Na2S 2 Na+ + S' (8)
Na2C03 2 Na + CO3 (9)
S + HO2 OH + HS (10)
2 HS S" S + H2S (aq) (11)
H2S (aq) H H2S (12)
An increase in CO3 ions means an increase in combination with Na
ions, thus a release in I2S will take place. The gases leaving the
tower are high in CO2 and low in H2S. This low concentration in H2S
presents difficulties for combustion, thus necessitating a precarbonation
tower in order to bring about an increase in H2S concentration. Shaffer63
gives a more detailed explanation of the chemistry of the process.
Institute Method.- Another interesting method for NSSC recovery is
the one developed by the Institute of Paper Chemistry, which is explained
in detail by Whitney. et al. This method is less complex than the Mead
method. Figure 4 illustrates the main steps. One of the main drawbacks
of the method is the production of some sodium thiosulfate, which in
spite of being inert in the cooking stage, results in greater chemical
losses and some difficulties in bleaching. The original conception by
Bradley and McKeefe introduced the idea of direct sulfitation
according to these equations.
Na2C03 + SO2 = Na2SO3 + C02 (13)
Na2S + 02 + H20 = Na2SO3 + H2S (14)
2 Na2S + 3 SO2 + Na2SO3 = 3 Na2S203 (15)
In an attempt to eliminate the formation of thiosulfate, some investi-
gators67686970 employed both carbonation and sulfitation. It seems
that this method has not been very successful. Common practice has been
the direct sulfitation method by stages, in which the amount of sulfur
dioxide is controlled to avoid excesses, thus avoiding the formation of
thiosulfate. This continuous sulfiting method called direct sulfitation
has two disadvantages: the inevitable formation of some thiosulfate,
and heavy losses of sulfur into the atmosphere. Whitney, et al., have come
with a modification of the direct sulfitation method called the bisulfite
method in which sulfur dioxide is replaced by aqueous sodium bisulfite
as the sulfiting agent. The main reactions in the system are as follows:
Na2S + NaHSO3 = Na2SO3 + NaHS (16)
NaHS + NaHSO3 = Na2SO3 + H2S (17)
to st lC
Fig. 4.- The Institute process of SO2 recovery by direct sulfitation.
Na2CO3 + NaHSO3 Na2SO3 + NaHCO3 (18)
2 NaHSO3 = Na2SO3 + H20 + SO2 (19)
The evolution of SO2 can be reduced, according to the data of Morgan,7
maintaining the bisulfite strength under 75 mole percentage, as shown
in Figure 5.
Purpose of this Investigation
Sulfur dioxide removal is being encouraged by new legislation on
air pollution control and by general public awareness. Sulfur dioxide
emissions take place at boiler plants in any type of mill, and at the
recovery furnace of pulp mills engaged in a semi-chemical method of
pulping. These investigations have been conducted to explore the
technical and economic feasibility of the implementation of a
purification system for SO2 removal.
Both technical and economic feasibility are sought in any system
proposed for the removal of sulfur dioxide. Technically speaking, it
is necessary for a purification system to give a high percentage of
removal to compensate for the loss of buoyancy and dispersion through
the cooling of the flue gas.
A flexible pilot plant with two different purification devices has
been selected for this study. It can be operated either as a simple
spray chamber with a radial inlet, or as a venturi-cyclone combination
in which atomization is achieved at the throat of the venturi.
Separation of the liquid droplets and the gas occurs at the cyclonic
demisting chamber. These two systems give a broad range of required
--5 O- C
Fig. 5.- Equilibrium diagram for a sulfite-bisulfite system.
energy over which removal of sulfur dioxide in gas-liquid contact
methods might occur.
Economic feasibility was sought according to the following
reasoning. Gas-liquid contact methods for scrubbing sulfur dioxide are
improved when a fairly rapid chemical reaction takes place. Since
sulfur dioxide is an acid gas, an alkaline solution is needed for
improving removal. It also happens that alkaline solutions are used
in the digestion process of the mill. These alkaline solutions are
rich in sulfur compounds. Thus, in principle, an immediate use of the
recovered sulfur might be possible. This chemical recovery will be
applied in two main processes, the kraft or sulfate process and the
neutral semi-chemical process. The degree of chemical recovery and the
operational characteristics of the mill will influence the merits of
Heat recovery is also possible due to the relatively high tempera-
ture of the flue gases. Many aspects can be considered in this type of
recovery whether it would be used for heating liquids, as in the
production of hot water, or increasing the concentration of chemicals
in the scrubbing liquor by evaporation of the water.
Figure 6 illustrates the proposed system. Flue gases coming from
either the boiler plant or the recovery furnace will be scrubbed in the
venturi-cyclone combination, in order to obtain a 90 per cent removal
of the original sulfur dioxide concentration. The scrubbing liquor will
be a carbonate solution in the neutral semi-chemical sulfite mill, and
-- water & NaCOC3
weak bl. liquor
U .>. MAKE- UP
SENR. to digester
pump I TANK
blacK i quorr ALTERNA TVES _
d rect con. ev ap.
Fig. 6.- Proposed system for SO2 removal and recovery.
the weak black liquor in the kraft or sulfate mill. In the first case,
a sodium carbonate solution is used to recover the sulfur as sulfite.
The corresponding stoichiometric amount of carbonate disappears as
carbon dioxide. In the second case, sulfidity is increased if the
weak black liquor is oxidized prior to the scrubbing. In this fashion,
hydrogen sulfide emissions are avoided.
The bleed-off from the purification unit is transferred by gravity
to a sedimentation tank and recirculated back to the purifying unit.
The make-up tank for the experimentation was designed to allow for the
oxidation and heating of the black liquor.
The proposed system has the following advantages:
1. It will remove high percentages of sulfur dioxide from the
flue gas at moderate energy requirements.
2. It will allow the advantageous use of a fuel with a high sulfur
content and a lower price.
3. The proposed system would be applicable in the neutral semi-
chemical sulfite mill where a sulfite-carbonate solution is required as
the digesting liquor. The bleed-off from the purification unit will be
rich in sulfites. The original carbonate concentration will be sacrificed
to obtain the sulfite. Thus, there is need for an enrichment in
carbonates and sulfites to make up for the digesting solution. Figure
7 (a) illustrates the adaptation of the system to mill operation.
4. For mills using the bisulfite method of sulfitation, the
proposed system will fit as follows. The bleed-off after several stages
Hlue gas --
PURIFICATION UNIT MAKI
E NRICHMENT0 -
(a) In the production of NSSC make-up.
CONVERSION RECIRCULATION I
Sn t t flue gas
STORAGE DILUTION RECOVERY
1 ;'ANK FURNACE
(b) Application in the Bisulfite Method of sulfitation.
Fig. 7.- Industrial applications of the proposed system.
Na CO 3
of recirculation could be made almost neutral, pH 7-8, with a
corresponding sulfite-bisulfite ratio of about 1.0. Then it will be
possible to use this bleed-off as the sulfiting solution, which upon
contact with the smelt from the furnace will produce a digesting
liquor. Figure 7 (b) illustrates the system.
5. In mills using the kraft process and interested in new
products, the bleed-off will provide a suitable compound which upon
enrichment would be used as the digesting liquor. The resulting black
liquor will be incorporated into the recovery system of the kraft mill.
6. In mills using the kraft process, it will be possible to scrub
the flue gases with oxidized weak black liquor. An increase in
sulfidity is logically expected, along with the possibility of some
concentration with the consequent increase in the evaporator's
II. FUNDAMENTAL THEORY
Gas Absorption in General
When a soluble gas such as sulfur dioxide contacts water, the
sulfur dioxide will dissolve until an equilibrium point is reached,
at which point the partial pressure of sulfur dioxide in the main gas
stream holds a certain relationship with the dissolved sulfur dioxide
in the water. Thus it is possible to state that a very soluble gas
will require a lower partial
pressure to obtain a given "
concentration of dissolved c
gas in the water than a less
soluble gas. In most cases /
dealing with a pure physical 0
absorption phenomenon, the /
solubility of the gas in the / ,/
liquid phase is inversely -
proportional to the temperature, c ,
Figure 8 gives a graphical Fig. 8.- Equilibrium diagram
for SO2 absorption.
idea of these concepts.
Absorption deals with the transfer of material between two phases.
In this case, the solute must diffuse from one fluid, the main gas stream,
into a second fluid phase, the liquid.
A mass balance will give the broadest picture of the mechanism,
from which we shall proceed into further details. If a general counter
current absorption tower is considered for the case of sulfur dioxide
in air being absorbed by water, V is the flow rate of air, L the flow
rate of liquid. Let Ll be the amount of sulfur dioxide in the inlet
water, which usually is zero. Let L2 be the amount of sulfur dioxide
in the outlet water. V2 and V1 will be the amounts of sulfur dioxide
in the inlet and outlet air. Then
L1 + V2 = L2 + V1 (20)
If X is the concentration of dissolved sulfur dioxide in the water
and Y its concentration in the air,
Ll LX1 and V1 VY1
L2 LX2 V2 = 2
X1 + VY2 = LX2 + WV
Y2 Y1 = L/V (X2 X1) (21)
This is the equation of a straight line with slope L/V. Considering
an interface at which the mass transfer is taking place, it will be
evident that all solute diffusing from the bulk gas phase to interface
must diffuse at the same rate into the bulk liquid phase from the
interface. Material is transferred from the main gas stream to the liquid
at every point along the interface, at a rate depending upon the driving
force and the resistance at each point. Because the transfer mechanism
is a combination of molecular and eddy diffusion processes, whose
individual and relative effects cannot be predicted, it is necessary
to employ the concept of a transfer coefficient. If Pai and Pag are
the partial pressures of the solute a at the interface and the bulk of
the gas stream respectively, then the rate of solute transferred from
the bulk gas phase to the interface is
N Akg (Pai Pag) (22)
If Cal and Cai are the concentrations of the solute a in the liquid
phase at the bulk and interface, then the rate of transfer from the
interface to the bulk liquid is
N = -AkI (Cal Cai) (23)
k and ki being the mass transfer coefficients for the gas and liquid
phases, A being the interfacial area.
Equation (22) determines the rate of transfer of solute from the
bulk phase to the interface, and thus the driving force is expressed in
terms of gas phase concentration units. Equation (23) expresses the
driving force in terms of liquid-phase concentration units. A relation-
ship for solute concentration at the interface, where the concentrations
of both phases are common, is needed before combination of these equations
is possible. From a general phase-equilibrium equation, the composition
of solute in the gas phase in equilibrium with a liquid of solute
concentration Ca is
Pa f(Ca) (24)
For dilute systems this function is frequently linear and becomes
Pa maC (25)
where m is a distribution factor like pure component vapor pressure in
Raoult's law, or Henry's constant, or the slope of the equilibrium curve
at Ca. After some manipulation73 with the above mentioned equations, the
following expressions can be obtained:
__L + m_
and N = -( Ca) (27)
1 + 1
Over-All Mass Transfer Coefficient
At the same time the individual mass transfer coefficients can be
substituted by the over-all mass transfer coefficients, since
__= __+ m__
AK1 Ag Akl (28)
and 1 ,_ 1
Akg mAKg Ak1 (29)
By using the over-all mass transfer coefficients, N can be
calculated without knowing the values of Pai and C ai which are
difficult to measure, at the interface. To make it clearer, P is the
partial pressure of solute in equilibrium over a solution having the
solute concentration Ca in the bulk, and C is the concentration of
solute in the bulk solution in equilibrium with the solute partial
pressure Pa. Then Pa P is the over-all driving force in the gas
phase, and Ca C is the over-all driving force in the liquid phase.
In the same fashion, the ordinate Pa can be substituted by ya, which
is the concentration of solute in the gas stream. Consequently, y
will be the concentration of solute in the gas stream in equilibrium
with a solution of solute concentration Ca.
Referring to Figure 9, A represents the conditions at any
differential surface through which diffusion is taking place. The
ordinate Pa is the average partial pressure of the diffusing gas in
the main gas stream, and the abscissa C is the average concentration of
the solute in the bulk liquid. In this case, the driving forces are
represented by the ordinate AE for the gas phase, and by FA for the
The same Figure 9 could be used to visualize the concept of the
operating line. Point C represents the conditions of the inlet gas
where both gas and liquid concentrations are high. Point D represents
the conditions of the outlet
gas, the partial pressure of
the solute in the outlet gas/ i/
having been reduced from P
F F '
to P2. For absorption to o
take place, the solute
concentration in the gas I
must be greater than in the
liquid, i.e., the operating
line must be above the Fig. 9.- Operating line and
equilibrium curve in a gas absorption
equilibrium curve, process.
As has been discussed previously, the magnitude of the slope of
the operating line equals L/V, indicating that the driving force is
dependent on how much the operating line differs from the equilibrium
line. If L is diminished, the slope diminishes to a point at which it
touches the equilibrium line, and then the driving force is zero.
Height of a Transfer Unit
Chilton and Colburn74 introduced the concept of the "height of a
transfer unit" to simplify the procedure for designing absorption towers.
This concept is still of great usefulness as a parameter for removal
efficiency. In the estimation of the height of the absorption tower,
the value of the definite integral
is always important, for it expresses the difficulty of a scrubbing
solution to absorb the solute from the gas. It is called the number
of transfer units, based on an over-all gas-phase driving force. The
equation to obtain Hog, the over-all height of a transfer unit, is
Hog = Z/Nog (31)
in which Z is the height of the tower.
From their analysis, Chilton and Colburn developed an equation for
design calculations which applies when only two gas compositions and one
liquid composition are involved.
S mVm) (Yl mX2) m
In L 2 2 Lm
Nog = I (32)
1 (mV /Lm)
Assuming that the liquid is well mixed vertically, and that
through a fixed area the liquid has a constant composition, equation
(32) can be expressed as
Nog In (1 Eg) (33)
E I eNog (34)
Eog being the removal efficiency expressed as a fraction.
The validity of this equation is based on the assumption that the
product of the transfer coefficient per unit area and the interfacial area
per unit of liquid volume is constant along the vertical path of
integration. Also, assuming mVm/Lm is equal to zero involves an
irreversible chemical reaction. According to the data on sulfur dioxide
in alkaline solutions, an equilibrium line with a slope close to zero
is acceptable within the range of actual operation. The expression for
the effective height of a transfer unit can be related to the over-all
mass transfer coefficient by the expression
Hog = V (35)
Theories Explaining Absorption
So far, operating expressions for the absorption process have been
presented in a rather simplified way. However, the theoretical principles
involved in the absorption process are complex and controversial. New
theories are in the process of resolving the apparent discrepancies.
The kinetic theory is the ultimate explanation of diffusion because
the assumed movement of the gas permits the solute to move into the bulk
of the liquid and achieve the condition of uniform distribution. Fick's
equation can be expressed as
dm = AD- dt (36)
where A is the interfacial area, D is the diffusivity coefficient,
t is the time, dc/dx is the concentration gradient.
Upon integration of this equation it is possible to ascertain the
amount of matter diffused during a given time and through a certain area,
using a diffusion coefficient, D, which depends on the gas involved.
The oldest theory regarding absorption of a liquid in a gas dates
back to 1878 when Stefan75 worked out simplified relationships deduced
from Fick's law. Stefan's theory was based on these assumptions:
1. The composition of the monomolecular gas layer at the interface
is equal to the bulk of the gas, due to the high velocities of the gas
2. The concentration of the gas in the monomolecular liquid layer
equals the concentration of the gas in the liquid, according to the
expression of Henry's law.
Stefan gave this expression for the amount of matter diffused at
N = 2 (Cgi Cl)\ / (37)
Differentiating with regard to time gives
Sd = (Cgi C1 (38)
These expressions indicate that the amount of solute diffused is
proportional to the square root of time, and the rate of diffusion is
inversely proportional to the square root of time. Cgi and C1 are the
concentrations of solute at the gas interphase and at the bulk liquid
Higbie,76 in 1935, arrived at a similar expression to Stefan's,
stating that even with a constant concentration gradient, the rate of
absorption decreases with the time of exposure. In Higbie's statement
the condition of an unsteady state mechanism was assumed.
A decade before the restatement of Stefan's law by Higbie, Lewis
and Whitman had postulated their two-film theory. This theory has
been the most widely used model for gas absorption, and perhaps the most
misunderstood. Their theory is based on a series of assumptions:
1. Steady state conditions exist in both phases.
2. The rate of transfer is proportional to the concentration
3. An equilibrium exists between liquid and vapor at the interface.
4. Retention in the interface is nil.
5. A stationary film exists at the interface.
The equations which describe Lewis' and Whitman's theory are the
same equations (26) and (27) which were used earlier to describe the
absorption process. To illustrate how the misapplication of this model
affects some systems, it will be convenient to refer to its application
in systems where chemical reaction takes place. In such a system, where
the gas is highly soluble, the over-all mass transfer coefficient is
given by the expression
Kga = 1 (39)
Since m approaches zero, Kga = kga, This analysis leads us to the
conclusion that the liquid phase resistance is negligible, which could
be wrong in many instances. Teller78 pointed out the necessity of
applying the model that fits the situation and not to restrict our
interpretation to a single model.
Surface Renewal Theory
In 1951, Danckwerts79 proposed his surface renewal theory. The
turbulence in the absorption tower creates numerous infinitesimal liquid
elements which are constantly brought to the interface. While these
elements are exposed to the opposite phase at the interface, the diffusing
solute is transported by unsteady molecular diffusion into these elements.
Danckwerts stated that the concentration gradient is a value that changes
with time. In other words, the rate of diffusion at the interface is
determined by the concentration gradient, the value of which depends upon
the time of existence of the interface and not merely on the difference
in concentration. Where there is any degree of turbulence in a liquid,
the existence of a continuous exposure of fresh surfaces to the gas is
assumed, sweeping away and mixing these elements with the surfaces which
have been in contact with the gas for a certain time. This theory assumes
an infinite depth of penetration.
More recently, in 1958, Toor and Marcello80 proposed the film-
penetration model. They showed that the two-film and the penetration
theories were not separate concepts, but only limiting cases of their
new film-penetration theory. Unlike Danckwerts' theory, they
postulated a finite value for the thickness of the surface element.
Absorption with Chemical Reaction
Most of the industrial processes involving absorption are
accompanied by a certain degree of chemical reaction. Processes for
the absorption of sulfur dioxide in aqueous solutions, which are
important in the pulping industry, in the recovery of pickle liquor
from steel manufacturing, and in the purification of flue gases, are
dealt with from the over-all mass transfer viewpoint. So far, only
one serious attempt has been made to divorce the influence of
molecular and eddy diffusion of sulfur dioxide into and out of the
solution from the diffusion of the constituents in the solution, so
that true reaction rate constants could be computed from the
In general, chemical engineers like to obtain gas-phase controlling
systems, so that the major resistance will occur in the gas-phase. To
understand this better, refer to the expression of the over-all mass
Kga I m
If m is large (low solubility), Kga approaches kl/m. If m is small
(high solubility), Kga equals kga. Since kg is larger than kl, the
process will be more rapid in the gas-phase controlling system, and
consequently the size of the absorber will be smaller. The relative
resistances of both phases are functions of the diffusivities of the
solute in the vapor and liquid phases, concentration of unreacted reagent,
rate of diffusion of the reagent, and rate of reaction. Consideration of
these factors eliminates the false assumption that the process has to
become gas-phase controlling.
The process of absorption accompanied with chemical reaction has
been subjected to many interpretations. In the pioneer work of
82,83 84 85
Hatta,8283 Hatta and Katou,84 and Hatta, et al., the two-film theory was
adopted with the assumption of a stagnant film. Sherwood and Pigford86
discussed in detail the complexities of Hatta's theory when an infinitely
rapid second-order chemical reaction, or a slow first-order irreversible
reaction takes place. In the same work they discussed the theory of
Higbie in different cases.
Because of the complexities involved in the mechanism of gas
absorption accompanied with a chemical reaction, some investigators
decided to use analytical models. In these cases, attempts were made
to predict the behavior of a system by comparison with the behavior
of another system that had been evaluated previously. The same hydro-
dynamic conditions for both systems were utilized, thus eliminating new
variations due to interfacial area, time of penetration, and degree of
surface renewal. Danckwertz and Kennedy87 arrived at some expressions
relating the value of mass transfer coefficients for physical and chemical
absorption. He assumed that no chemical reaction occurred in the
physical process, i.e., assuming absorption in water as a physical
process, which is not fully true.
In recent years many studies88 to 99 have been conducted in an
attempt to obtain a consistent physical picture of the process according
to the different systems tested. Recently, in 1963, a general mathematical
model for mass transfer accompanied by chemical reaction has been
elaborated by Huang and Kuo. In their theory, which is similar to
the renewal theory, they postulate the interface as being replaced by
infinitesimally small fresh liquid elements. The mass transfer mechanism
consists of two steps: a surface renewal by fresh liquid elements, and
a simultaneous molecular diffusion and chemical reaction within the
exposed liquid elements. When the renewal rate is slow, the surface
elements remain at the interface, and the extreme condition is the
creation of a stationary liquid layer. This is the case with the model
proposed by Hatta for chemical reaction using the two-film theory. If
the turbulence of the liquid is great, the residence time of the elements
at the interface would be small and the surface renewal theory would
prevail. Unlike Danckwert's theory, this model assumes a definite value
for the average thickness of the surface as proposed by Toor and Marcello.
Huang and Kuo prove the similarity of their equations by two different
mathematical approaches, trigonometric function series and error
function series. The proposed equations contain the dimensionless
groups C(, /9 in which
Da sr D< D
k L r sL
where sr is the surface renewal rate, L is the thickness of the surface
elements, and Da is the molecular diffusivity of the solute a.
Thus, if / approaches zero, the main equations expressed in
trigonometric or error function series can be reduced to Hatta's
equation. Furthermore, when the reaction velocity constant is zero,
an identical expression to the one derived by Toor and Marcello is
obtained from the trigonometric function series expression. Also, the
equation expressed by Danckwerts is obtained from the error function
series expression. When the dimensionless group c< approaches infinity,
the expression from the trigonometric function series is identical to
the one derived by Lewis and Whitman. In conclusion, for limiting
conditions the derived equations can be reduced to the simple
postulations already mentioned. For nonlimiting conditions, the
film-penetration theory of Toor and Marcello gives information not
attainable through any of the original theories.
Rate of Absorption in the Venturi
The rate of absorption in the venturi scrubber is determined by
the concept of transfer units. The number of transfer units can be
expressed, according to Kuznetsov and Oratovskii, as
Nog = K (40)
in which K is the over-all mass transfer coefficient, F is the interfacial
area, and Q, is the gas flow rate.
In the venturi scrubber, the absorption process takes place both
in the throat and in the diffuser section which follows the throat.
The magnitude of N should be analyzed separately for both parts of
the apparatus. The over-all coefficient of absorption accompanied by
an irreversible chemical reaction is defined differently depending on
the region in which the absorption occurs. If the rate of the chemical
reaction is much lower than the rate of diffusion,
K = K (41)
If the rate of diffusion is controlling, then
K = KoVg (42)
If the rates of reaction and diffusion equally influence the
process, the over-all mass transfer coefficient is given by an expression
of resistances in series.
K 1 (42)
The variation of the over-all mass transfer coefficient with gas
velocity has been studied by different authors.102,103 The coefficient
also varies throughout the distance from the injection point to the end
of the diffusion zone. K, and n can be determined by experimentation
from the plot of the curve of gas velocity versus over-all mass transfer
coefficient. The proportionality coefficient Ko derives from the
expression of the Nusselt number, Nnu = Kd/D, and n is a quantity which
depends upon the hydrodynamm of the process.04
depends upon the hydrodynamism of the process.
If R is the specific flow rate of the liquid or the liquid/gas
ratio, the fraction of liquid in the two-phase system is equal to
Since R amounts to about 0.001, the fraction of the apparatus
volume containing liquid may be equal to R, and the interfacial contact
surface Ft in the throat of volume Vt could be expressed as
Ft Dp (44)
To determine Dp, the average diameter of the particles dispersed,
the empirical Nukuyama-Tanasawa0506 equation is employed. Lewis107
revised this equation and found it applicable for the throat velocity
range of 200 to 600 feet per second.
The original expression for the average diameter of the aerosol is
585V' /+ 0.45 1000 Q1 1.5
Dp = + 597 ( ) Qg (45)
in which 7-, and ,4 are the density, surface tension, and viscosity
of the liquid. This expression is simplified in this form,
D A + BR1'5 (46)
A and B are values which depend on the scrubbing liquor and the gas.
For air and water, these values are 16,050 and 1.41, resulting in
Dp = 7-- + 1.41 R
In the Nukuyama equation, V is really the velocity of the gas
relative to the liquid. In the venturi used in these experiments, the
liquid overflows a weir before entering the apparatus. Since its
velocity is very small, it is therefore valid to consider the value of
the gas velocity for Vg.
Because of mathematical complications, a venturi theory that comprises
the whole tube, throat, and disperser has not been proposed. At the
present moment only a study of the separate parts is available.
Boyadzhiev108 discussed in detail the significance of Kuznetsov's
equations. Separate considerations were made for absorption taking
place in the throat and in the diffuser.
In the throat, the number of transfer units, when the absorption
is determined by the rate of chemical reactions, is given by the
600 (A + BR1.5Vg)
in which ht is the length of the throat.
When diffusion is controlling, the over-all absorption coefficient
is determined by equation (42), and the number of transfer units is
KVgn R ht
N0, = 15 (49)
Ng 600 (A + BR Vg)
When the two rates are commensurate, the following dependence
derived from equations (41) and (42) is valid.
K, Kr V n R ht
Ng 600 (A + BRI.5 Vg)(KoVgn + Kr) (50)
In the diffuser region, the gas velocity decreases from the entrance to
the discharge section. Since the diffuser is a truncated cone, the gas
velocity, Vg, and the cross section of the diffuser at any distance is
V gt (51)
g (1 + ph)
p dd dt/dthd (52)
in which dd and dt are the diameters at the diffuser and at the throat,
S = St (1 + ph)2 (53)
St is the cross section at the throat, and h is the distance from the
The interfacial area could be expressed in a differential form by
d Fd= 6 R Sdh (54)
By substituting x2 for (1 + ph)2 and grouping constant values
under q, it will be possible to have
d Fd q x 2d (55)
d p (a + x2)
and Fd = P a x2 (56)
This is a general integral of the form
2x da tan (x \a/c) (57)
Integrating over the entire length of the diffuser, that is, from
x = 1 to x = Xd, the following equation results.
Fd = (xd 1) -\ a (tan-1 a tan-1 ) (58)
The value of Nog can be ascertained by integration of the expression
Nog = d a x2dx (kinetic region) (59)
P a + x
q V d 2(1-n)
No q x 2(n dx (diffusional region) (60)
Ng P a + x
In general, the value Nog passes through a maximum as R increases
for all the cases, both in the throat and in the diffusion region. The
significance of the gas velocity, V is not discussed here. It suffices
to say that when the reaction rate is controlling, Ng decreases when
Vg increases. When either the diffusion or the intermediate state is
controlling, Nog increases with an increase of Vg.
From the field data, it will be possible to ascertain the value
of the over-all mass transfer coefficient K. Since this coefficient
is defined differently according to the controlling region or its
combination, there is need to examine carefully the gas absorption
system. By means of experimentation, it could be possible to obtain
the value of KV n. Kr could be found by rigorous experimentation and
by the analysis of the reaction kinetics of the particular system.
Rate of Absorption in the Spray Chamber
For absorption in the spray chamber, a concept similar to the one
employed in the venturi calculations is used. The concept of height
of transfer unit is normally regarded in the chemical engineering field
as the parameter to measure the behavior of the spray chamber. In the
spray chamber the flow is generally counter current. Top and lateral
spray nozzles break the liquid into small drops, thus providing the
interfacial surface across which diffusion takes place. The interfacial
surface in the spray chamber is not large. Replication of experiments
in a spray chamber seems to be difficult, probably because of the
adherence of small particles to the spray chamber walls, and some
liquid entrainment in the gas phase. This last factor is less significant.
The equation to obtain Hog, the over-all height of a transfer unit,
og = Z/Nog (31)
Since the concentration of sulfur dioxide in the gas stream is
relatively small, the following equation may be used.
Nog (Y1 Y2)/(y Y*) Im (61)
For the specific case under consideration, y is negligible in
comparison with y. Thus it is possible to write
Nog dy dl In (Y ) (62)
y y Y
This expression can be related to K., the over-all mass transfer
coefficient, by the equation
HS M (35)
Recent studies in the operation of spray towers with lateral
spray nozzles have been based upon the characteristics of the apparatus
and the hydrodynamic conditions prevailing. Mada and co-workers1
experimentally determined an expression for Hog and also for Hg and H1.
For each particular system they obtained a response function. The
response equation for the absorption of ammonia in water, for example,
0.70 -0.30 -080 012 -0.20 -0.15
og/D = 12.7 Reg 7Rel- 30n-080(Z/d) 012(d/D) 20Gag (63)
In which Re is the Reynolds number of the gas and liquid streams, d and
D are the diameters of the nozzles and tower, respectively, Ga is the
Galilei number for the gas stream, and Z is the height of the tower.
Chemistry of the Sulfite-Bisulfite System
When an acid gas as sulfur dioxide dissolves in water, the gas is
in equilibrium with the ions formed. It is possible to establish three
main equations for the ionization constants of the different species.
Latimer, from well-known data on free energies of formation, gives
the values for K1 and K2 as follows:
H2S03 __ H + HSO3 K = 1.25 x 10-2
K2 = 5.6 x 10-8
HSO3 ;= H+ + S03
Johnstone and Leppla determined the free energy of formation for
SO2 (g) : S02 (aq) as Fo = -123 cal
With this free energy data it is possible to calculate the
equilibrium constant for the following reaction:
SO2 (g) + 20O = H2S03 (aq) K = 1.2
The above three relationships can be expressed as
Kp 12S3) (66)
For a complete definition of the system, more equations are needed.
The total amount of sulfur is given by the mass equation,
St = (H2SO3) + (HS03") + (SO3 ) + (S02) aq (67)
For a sodium sulfite-bisulfite system, the ionic equilibrium
(Na) = 2 (SO3') + (HS03") + (OH-l (H') (68)
The solution of this system of equations and unknowns could be
simplified by the following reasoning. To obtain a simple expression
for PSO2, it is possible to transform equation (66) into
P ( 2S03) (69)
Substituting for the value of (H2S03) in equation (65) gives
PS02 KKp (70)
Substituting (HSO3 ) for its value in equation (64) gives
S =(H+)2 (03) (71)
By means of this reasoning, one can establish an important fact in
sulfur dioxide absorption. That is, vapor pressure or volatility of
sulfur dioxide in solution is directly proportional to the square of
the hydrogen ion concentration. This means that at lower pH values,
the vapor pressure will be greater than at higher pH values. Consequently,
the capacity of absorption for sulfur dioxide of a solution tends to
decrease with lower values of pH.
Data from Figure 10 give valuable information on the bisulfite-
sulfite system. From equation (65),
K2 =- )(SO) and K = (S03=) (72)
(HS03") (H') (HS03-)
The ratio sulfite-bisulfite can be predicted from the pH of the
solution. If pH pK2, the ratio of sulfite to bisulfite will be more
than two orders of magnitude. Thus, for practical purposes we can
eliminate (HS03") from the mass equation (67). Following the same
reasoning, from equation (64)
S(H+)(HSO3") and KI (HS ) (73)
(H2S03) (H1) (H2S03)
0 1 2 3 4 5 6 7 8 Q 1C 11 12 13
Fig. 10.- Equilibrium diagram for sulfite species.
Then is pH pK the concentration of the SO 3" species is much
greater than H2SO3, and it is possible to eliminate the term H2SO3
from equation (67). It follows that the term SO2 (aq) will be small
enough to be eliminated, leaving St = SO3 Substituting in this
fashion in equation (67)
(S03 ) = (Na/2) (74)
Buffered Scrubbing Solution
The above discussion indicates the need to have a solution which
changes pH very slowly when the solute is absorbed into it. In other
words,a buffered solution is needed. In many of its processes, the
pulp industry uses highly alkaline solutions with buffer characteristics.
If a carbonate solution is used as the scrubbing liquor, two factors
that may affect its pH need to be studied. Sulfur dioxide and carbon
dioxide are both present in the flue gas. Carbon dioxide and sulfur
dioxide are both weak acids, but sulfur dioxide is the stronger of the
two. The buffer capacity of a carbonate solution differs, depending
upon titration with a weak or a stronger acid. Weber and Stum
observed that the shift in pH caused by the weaker acid is smaller than
the one produced by the stronger acid. In an attempt to look into the
complexities of the sulfite-carbonate system, a series of titrations
of carbonate solutions with pure sulfur dioxide and pure carbon dioxide
were conducted. The theoretical relationship among carbonate species is
shown in Figure 11.
The curves in Figure 12 illustrate the titration of a 0.45 M
Na2CO3 solution with sulfur dioxide (curve A), with carbon dioxide
Fig. 11.- Equilibrium diagram for carbonate species.
(curve B) at the same rate, and with sulfur dioxide (curve C) at a
lower rate of flow. Carbon dioxide gives a typical titration curve
for a weal: acid. Sulfur dioxide seems to be a stronger acid.
The exact mechanism of the absorption of sulfur dioxide in a
carbonate solution is not clear. Apparently, when the scrubbing
solution is water, both film resistances influence the rate of
absorption. It would be necessary to go into a rigorous study of the
reaction kinetics of the system to ascertain individual contributions.
In an attempt to obtain more information on the mechanism of
sulfur dioxide absorption in a carbonate solution, a series of
experiments was conducted with the experimental set up shown in
Figure 13. Pure sulfur dioxide was bubbled into 175 ml of 0.44 M
The rates of flow for the sulfur dioxide were 0.25, 0.50, 0.75
and 1.0 SCF/H as air. The number of moles of sulfur dioxide reacting
was calculated from the variation in pH recorded during the time of
injection. The number of moles, Q, injected at any time t, is given
by the expression Q = qt, where q is the rate of injection.
Since the volume and concentration of the carbonate solution
remained the same for all the experiments, the same quantity of
reacting sulfur dioxide was needed for a certain change in pH. Thus,
Q = qltl = q2t2 (75)
This expression permits the correction for time necessary to obtain
the curves of Figure 14. Curves A, B, and C represent the conditions
25 [ .t
5 / i
65 - i t -
0 30 60 90 120 150 180 210 240 270 300 330 360 390
Time in seconds
Fig. 14.- Titration curves at different rates of SO2 injection.
for 0.25, 0.50, and 0.75 SCF/H as air. At 1.0 SCF/H as air, a similar
curve was obtained, but not plotted for the sake of clarity.
In the laboratory experiments, an instantaneous reaction occurred
within a wide range of available sulfur dioxide. The gas phase resistance
was controlling. During field conditions, the presence of carbon dioxide
will alter this picture.
Oxidation of the kraft weak black liquor represents a milestone
in the efforts to control sulfur emissions in the pulp mill. However,
particularly in Southern mills, the foaming created in the oxidation
process is possibly the main reason for delaying the implementation of
the oxidation practice as a standard procedure in all mills.
Foam is a dispersion of a gas in a liquid. It consists of gas
bubbles very close together, separated by a thin liquid layer. Kraft
weak black liquor foam is composed of a wide range of bubble sizes and
is rather long-lived if compared with ginger ale foam.
There are many theories that attempt to explain the formation of
foams. The most widely accepted is the balance layer theory.
According to this theory, two rising bubbles are subjected to a system
of forces. If brought together, the two bubbles might disappear; if
they remain apart they create a stable film at the surface. The force
which tends to separate them arises from the difference in concentration
between the surface layer and the bulk solution.
Foam stability has been the object of many studies. At the
present time, it is possible to say that it is the result of several
interacting factors. In summary, Gibbs elasticity, viscosity, and
surface area are among the most important. Kevorkian5 listed
For a solution to give rise to a stable foam, its surface tension
must always be acting to oppose any force which deforms the film.
This property of the surface tension to reduce strains is called Gibbs
elasticity. In that fashion, a mechanical or thermal shock might
decrease the concentration of solute in the film, thus increasing the
surface tension. The increased surface tension will tend to contract
the film, counteracting the extending action of the external forces on
the film. If at certain moments the increase in surface tension contracts
the film, there should be an increase in the concentration of the
surface active agent which decreases the surface tension, thus extending
the film. This "stretching" ability should be present in any local area
of the film.
Viscosity is an important factor in foam stability. If the viscosity
of the film is decreased, the attracting forces will overcome the
dispersing ones, and the bubbles will coalesce and disappear.
The third important factor in foam stability is the value of the
free surface energy. Free surface energy is the amount of work necessary
to create a fresh surface. It is equal to the product of surface tension
and surface area. Stable systems are those with a minimum free surface
energy. For the same value of surface tension, foam stability is
greater if the bubble is smaller.
An important characteristic in the lifetime of foam is drainage.
Different mechanisms of drainage exist, depending on the thickness of
the film. However, a discussion of such mechanisms is beyond the
objectives of the present study. Let it suffice to point out the
relationship of drainage with viscosity of the liquid film. A decrease
in the viscosity of the film surrounding the bubble enhances drainage,
causing the bubble to become thinner and thinner until it ruptures.
Selection of the Antifoaming Substance
An important consideration in this selection should be the
spreading characteristics of the antifoaming substance. One must be
chosen which guarantees the formation of a monomolecular layer on top
of the vessel. In general, oils, with a negative spreading coefficient,
give rise to a monolayer and a surface excess composed of lenses.
Another feature of interest for the selection of the antifoaming
agent will be its viscosity. Of all the commercial oils, kerosene
has one of the lowest viscosity values.
Antifoaming agents in general act either to prevent foam formation
or to "kill" the foam once it is formed. The former type of antifoaming
agent seems to change the existing hydrogen bonding between the film
and the substrate. The latter type is more dependent on viscosity and
Kerosene was selected as the antifoaming agent because of the
above mentioned considerations, and the fact that it was readily
available and rather inexpensive.
III. PROJECT DESIGN
It was the purpose of this experimentation to explore the
possibilities of sulfur dioxide removal from the flue gases of a
boiler plant and to adapt to the pulping industry, in the most efficient
way, any recovery process developed.
The previous discussion on the mechanism of absorption in general,
and the applicability of sulfur dioxide absorption in recovery processes,
formed a basis for selection of the variables used in this experiment.
The description of the experimental pilot plant contained in Appendix A
will help clarify many of the decisions regarding the variable levels
and experimental procedure.
The response selected to characterize the results of the experimenta-
tion is the percentage removal of sulfur dioxide. This response clearly
gives a measure of the efficiency obtained in SO2 removal plus provide
suitable figures for ascertaining recovery and air pollution reduction
potentialities. Other characteristics in the absorption process, such
as the number of transfer units for both the venturi and the spray
chamber units, can be readily obtained from the response selected.
An absorption phenomenon is complex per se, even under rigidly
controlled experimentation in a laboratory. The process is even further
complicated on a plant scale. As a result, few of the factors influencing
the absorption mechanism are purely independent. It is within this
limitation that field experimentation was conducted, and only a few
variables that were actually made independent were studied. Temperature,
total pressure, partial pressure of the solute in the feed gas stream,
hydrogen-ion concentration, liquid-gas ratio, gas velocity, and
concentration of the scrubbing solution were among the most important
variables considered. The following careful look at the system permits
the elimination of some of these variables.
Total pressure in the scrubber is constrained within narrow limits,
its value being a few inches of water above normal atmospheric pressure.
This is true, in general, of all standard gas-liquid contact units.
Partial pressure of the solute in the gas stream was beyond control
in the experimental pilot plant because the boiler plant output oscillated
continuously, as shown in Figure 32, Appendix A. Preliminary experimenta-
tion showed that similar oscillations occurred in the sulfur dioxide
concentration of the flue gas. Since it could not be controlled, the
partial pressure of sulfur dioxide was not considered as an experimental
variable. Precautions were taken to conduct the experiments within as
narrow a range of SO2 concentration as possible.
Gas velocity through the scrubber was an important factor, especially
in the venturi scrubber. However, due to the influence of gas velocity
in heat transfer processes, which introduced modification in the temperature
levels, gas velocity was not considered a variable. The influence of gas
velocity in the absorption process was studied at only two levels of
velocity. A complete block design, at the new gas velocity level, was
tested to ascertain the influence of gas velocity in the absorption
process under all existing combinations of factors.
The importance of hydrogen-ion concentration was outlined in
Chapter II. Since it was desirable to have a measure of the economic
value of the scrubbing solution, concentration of the scrubbing solution,
which is related to pH, was selected as a variable instead of pH.
After this elimination of factors or variables, only temperature,
liquid-gas ratio, and concentration of the scrubbing solution were
considered for most of the experiments. When using water or weak kraft
black liquor, concentration ceased to be a variable.
Orthogonal Factorial Design
For the study of the variation brought about by deliberate changes
in the independent variables, a useful technique is provided by a
factorial experimental design. The advantages of a factorial design are:
1. It is an efficient method, that is, a method which obtains the
desired information with the required precision for the minimum
expenditure of time and effort.
2. If interactions do exist, they are detected, and misleading
conclusions can thus be avoided.
Whenever possible, the experiment should be made orthogonal.
Orthogonality implies making the experiment symmetrical in all the
independent variables by testing all treatment combinations the same
number of times. Orthogonality insures that the main effects and
interactions can be estimated independently without entanglement.
Because no information was available about the operational levels
of the above mentioned variables, preliminary testing was necessary
to estimate the operating range for each of the variables.
To achieve a net reduction in ground-level air pollution using a
wet sulfur dioxide scavenging process, it is necessary to obtain close
to 90 per cent removal. This high percentage of removal will compensate
for the adverse effect in the dispersion pattern introduced by the
cooling of the gases.
Whenever it was possible, the levels of each variable were selected
in such a way as to reduce the possibility of overlapping values due to
the experimental error, the magnitude of which was unknown at the
beginning of the study.
The concentration level for the scrubbing solution had an upper
limit set by the common concentration of carbonates in the digesting
liquor used in the mills. This was not detrimental since the carbonate
was sacrificed to absorb the sulfur dioxide. Evidently it represented
a good starting value for an upper limit.
Liquid-gas ratio levels were selected on the basis of preliminary
experiment, since the values appearing in available literature differ
drastically. In any case, it was desirable to utilize the maximum gas
flow, because the results would resemble more closely those expected in
industrial scale operation. In the case of the spray chamber experimentation,
the rates of flow were substantially reduced because of the scrubber
flooding problems. It seems the capacity of the spray chamber is
about two thirds that of the venturi.
Temperature levels were limited by field conditions. Variations
in the flue gas temperature due to boiler plant operation, plus absence
of a temperature regulator, limited the choice of temperature levels.
A cooling system, as explained in Appendix A, was able to operate only
within certain variability. Within these constraints, the levels were
spread as much as physically possible.
Levels of the Experimental Variables
Once an idea of the maximum ranges of the different variables was
obtained, a decision was made regarding the number of levels. Only
two levels are often used in semi-industrial experimentation for both
economy and simplicity. However, it is advantageous to use three levels
of the independent variables if it is possible to do so, because
information can be obtained on both the linear and quadratic components
of each effect. A quadratic component may imply a maximum or minimum
response at some intermediate factor combination, or at a point outside
the range examined. This consideration applies especially to quantitative
factors, such as the ones involved in this project.
The levels of the independent variables were equally spaced when
using three levels. In some experiments in which only two variables at
four levels were involved, the levels were not equally spaced. In
general, the criterion used for choosing the levels was restricted
minimization of the generalized variance of the estimates, E(Y), where
Y is the response value for each different experimental condition. The
generalized variance is the determinant of the second order central
moment matrix about the regression equation.16 Table 2 indicates the
levels for the variables used in the different experiments.
It was decided after preliminary tests to allow the scrubber to
operate for a thirty-minute period, in order to stabilize the system
before taking samples. This period of time was necessary to arrive at
a fairly steady temperature. It was assumed that hydrodynamic
conditions became stabilized at the end of this period of time.
For each experiment, two observations were made. The duration of
each experiment was around 45 to 60 minutes. The general procedure for
plant operation and sampling is discussed in Appendix B. The capacity
of the make-up tank was only enough to permit running a few experiments
under the conditions already stated. Randomization also precluded the
reuse of a given batch. Thus, numerous independent batches were prepared
during the experimentation. However, the differences introduced in the
preparation of the scrubbing solutions were assumed to be negligible,
because the same technical grade of chemical from the same manufacturer
was consistently used, and the same preparation technique was rigorously
followed. It should be kept in mind that the concentration levels in
the solutions were spaced adequately to minimize inconsistencies in
successive batches of scrubbing liquor.
Factors and their Levels
System Scrubbing Factor
Liquor -L 0 +1
Water Ratio (gpm/1000 cfm) 1.40
Temperature ( F, inlet) 270
Carbonate Ratio 1.3
Carbonate Ratio 1.00
Weak Black Ratio 3.15
In statistical analysis, randomization is not absolutely necessary
for drawing conclusions from the data gathered. Nevertheless, it is
difficult, especially in the early phases of a stochastic experiment,
to know how much control can be exerted over the experiment. Randomi-
zation thus becomes the safest way to draw reliable conclusions.
Unexpected systematic effects can be cancelled out, the observations
can be made independent of each other, and any effect of uncontrolled
variables can be averaged out. The random order for each experiment
was obtained from the table of random numbers.117
Experimental errors can be of importance, especially when a new
system is under operation in a newly designed plant. Replication is
necessary to obtain a measure of precision and experimental error. Two
replications were performed for each experimental condition.
The mathematical models which describe the different experiments
can be summarized as follows:
In a 3 x 3 x 3 factorial, three factors or independent variables
are treated at three levels. The model for such an experiment is
Xijkm + Ai + Bj + ABij + k + BCjk + ACik + ABCijk + Em(ijk) (76)
in which Xijkm represents the m observation on the i, j, k treatments.
For this case,
m = 1, 2 j = 1, 2, 3
i = 1, 2, 3 k = 1, 2, 3
X1232 represents the second observation using factor A at first
level, factor B at second level, and factor C at third level, repre-
sents the average Xijkm over all populations, or E (Xijkm). Ai, Bj, and
Ck are the different treatment effects. The combinations of capital
letters stand for two-way and three-way interactions among the factors.
The error term Em(ijk) is normally considered an independently
distributed random effect. This could be expressed as NID (O,v ).
A similar mathematical model is applied to a 3 x 3 x 2 factorial.
Xjkm = + Ai + Bj + ABi + Ck + BCk + ABCijk + E(ijk) (77)
For this case,
m = 1, 2 j = 1, 2, 3
i = 1, 2 k = 1, 2, 3
Response Surface Experimental Technique
If the dependent variable (Y) and several independent variables
(X1, X2, Xk) are measurable, it is possible to express a response
surface equation for (Y) as follows:
Y = f (X, X2 . . Xk) (78)
For two independent variables, the response surface can be
represented graphically as a contour map or as equal response curves.
For three variables, a space solid representation could be used to
visualize the results.
Stochastic approximation procedures8 involve two basic
considerations: first, choosing a direction in which to search, and
then selecting the distance to travel in that direction. The main
difference between deterministic and stochastic problems is the higher
speed in approximation to the optimum of the deterministic problem.
The experimental error, or "noise," clouds the perception of what is
really happening. The steepest ascent method was selected under the
prevailing experimental conditions. It can be used in moderately
unimodal functions, and it can also work in the presence of experimental
error. The use of this method reduces the possibility of obtaining a
In industrial scale experiments, the exploration of new areas to
obtain an optimum is troublesome and expensive. For this project, the
results of a preliminary experimentation indicated an area to be
explored. This area was close enough to the optimum to be truly
representative of the response surface area.
When this is not possible, the steepest ascent PARTAN techniquell9
is indicated as an elegant and powerful tool. The methodology carefully
followed in this experiment is the same indicated by Davies and by
IV. EXPERIMENTAL RESULTS
Scrubbing with Water
The results of the experiments using water as the scrubbing solution
are in keeping with the general theory describing the process. The
experiments were conducted on both the venturi and the spray chamber.
Experiments in the Spray Chamber
Scrubbing with water in the spray chamber proved unsatisfactory.
The percentage removals were not over 5 per cent even when the liquid
rate of flow was increased to 12 gpm and the gas rate of flow was
diminished to 1800 cfm. At this same gas rate of flow, it was not
possible to increase the liquid rate of flow above 12 gpm, because
flooding conditions at the gas inlet occurred. Another unsatisfactory
condition, typical of the spray chamber, was the difficulty in obtaining
meaningful results from replications at the same conditions.
Experiments in the Venturi
Table 4, in Appendix C, indicates the field data recorded for the
different experiments. The percentages of sulfur dioxide removal from
the flue gas stream appears in Table 5. Figure 15 illustrates those
results with an average value for each treatment. A statistical
analysis was performed with the data as it appears on the block design
of Table 6. From the ANOVA of Table 7, it is possible to ascertain the
relative significance of the different factors.
S> 'EMF: LEVELS
Fig. 15.- Scrubbing with water in the venturi.
The liquid/gas ratio has the highest significance of all the factors
studied. The low point, or "sag," noticed in Figure 15 is only
justifiable by the high significance of the interaction sum of squares.
Seemingly, flow patterns along the apparatus reduced the transfer area
for the process, at certain rates of flow. In general, the experiments
in the extreme regions of experimentation were dependent on temperature,
as predicted by the traditional theory on physical absorption. If they
did not agree even more, it is because of the existence of a moderate
The breakdown of the analysis of variance shows the significance
of linear and quadratic effects. Since quadratic effects are highly
significant, it was advisable to use a second degree equation in order
to represent the experimental response. The experimental response is
the percentage of sulfur dioxide removed. Table 8 indicates the estimates
of the constants and the regression analysis for the best fitted
equation representing the experimental conditions. The best fitted
equation is expressed as:
Y 23.62 + 10.42 xl 9.92 x2 + 21.41 xl2 1.73 x22 5.06 x1x2 (79)
Differentiating with respect to xl, the ratio, and x2, temperature,
and solving simultaneously, a possible maximum value of 31.18 was
obtained. Upon a Hessian inspection of the best fitted equation, it was
determined that there was not such a maximum, but rather a saddle point.
The canonical representation of the best fitted equation, as it appears
in Figure 16, clearly indicates the saddle. The canonical expression
for the best fitted equation is
Y 31.18 = 21.74 X1 2.06 X2 (80)
where X1 and X2 are the new canonical axis.
Had the purification unit been able to perform other experiments
at higher liquid rate of flow, then it would have been advisable to
follow an optimum seeking method as described in Chapter III.
Estimates of the over-all mass transfer coefficients for the whole
purification unit appears on Table 10, based on the calculations of
Table 9. It was assumed that all the removal occurs at the venturi
in both the throat and the diffuser zone. The cyclone only provides
separation between the liquid and the gas. The throat interfacial area
on Table 9 was estimated by the same procedure as outlined in Chapter II.
The diffuser interfacial area was estimated averaging the velocities at
both end sections.
-2362 4.,- 992.i "2' 21. -,.73 5.60 .6O
a: 3 18 2 2. X2
Fig. 16.- Response surface of equation (80).
Scrubbing with Sodium Carbonate
Experiments in the Spray Chamber
For these experiments a 3 x 3 x 2 factorial was used. Only two
levels of temperature were established. For a better understanding of
the results and their significance, two completely separate analyses
were performed at both the higher and lower levels of temperature.
High Temperature.- Table 11 indicates the field data gathered for
the different treatments at the higher level of temperature. Table 12
gives the percentages of sulfur dioxide removed. Figure 17 shows the
major features of the experiments. Removal increases with an increase
of the liquid/gas ratio, but this increase is greater at the lower
level of concentration of carbonate in the liquor than it is at a
higher level. The relatively low percentages were expected for the
Fig. 17.- Scrubbing with sodium carbonate in the
spray chamber at a high temperature. Experimental results.
spray chamber. Once more, flow pattern characteristics were the reason
for a "sag" at the intermediate levels of ratio and concentration.
Figure 17 indicates the presence of a chemical reaction depending on
the concentration of the scrubbing liquor.
The same statistical procedure as the one described previously
was adopted for this experiment. Tables 11 to 15 inclusive give the
results of the analysis of variance and the determination of the best
fitted regression equation. The strong significance of concentration
in the ANOVA of Table 14 is noticeable. Linear effects of the
concentration factor are more significant than the quadratic effects.
Table 15 reveals a linear interpretation of the experiment based on the
values of the variances for the last three constants.
The best fitted equation considering quadratic effects, is given by
Y = 20.29 + 2.87 x, + 5.33 x2 + 1.36 x, 2.61 x2 1.28 x1x2 (81)
The canonical expression for the above equation is given by
Y 22.81 = 1.48 X1 2.73 X2 (82)
An inspection of the signs of these coefficients indicates the
presence of a saddle type of response surface. It seems enough to
represent the experiment by the linear expression
Y = 20.29 + 2.87 xl + 5.33 x2 1.28 xlx2 (83)
Lower Temperature.- Tables 16 to 20 inclusive give the data for the
statistical analysis of the results obtained in the experiments at the
lower level of temperature.
As Figure 18 indicates, removal increases with an increase in the
concentration of the scrubbing solution at all levels. Regarding the
liquid/gas ratio factor, removal only increases with an increase in
the liquid/gas ratio at the two extreme concentration levels. Once
more, the behavior of the spray chambers in general, and the flow
patterns in particular, explain this situation.
Table 19 indicates the significance of both factors and their
interaction. All are significant at the 1 per cent level.
The best fitted equation for the experiment is given by
Y = 15.71 + 4.41 xI + 5.82 x2 1.66 x2 0.86 x22 + 1.06 xx2 (84)
Fig. 18.- Scrubbing with sodium carbonate at a low
temperature. Experimental results.
The canonical representation in Figure 19 is given by the
Y 37.48 = -0.61 X 1.92 X2 (85)
In the same Figure 19, the relative position of the experimental
area, the response surface, and the optimum point S are shown.
Table 21 indicates the significance of the three factors and their
interactions for the total experiment. All three factors and the AB
interaction are significant at the 1 per cent level. Table 22 gives
the value of the estimates of the height of transfer units.
Experiments in the Venturi
The same procedure for the statistical analysis used in the
previous experiments was followed for the experiments in the venturi.
Fig. 19.- Response surface for equation (85).
Two levels of temperature were established and the analyses were
conducted separately for each one.
In venturi operation, the velocity of the fluid through the throat
is an important factor. This importance and the limitation of the pilot
plant in this aspect have been outlined previously. In an effort to
obtain some idea of this factor, a whole block design was conducted at
the higher level of temperature and at a lower rate of gas flow.