United States
Environmental Protection
Agency
Industrial Environmental Research
Laboratory
Research Triangte Park NC 27711
EPA-600 2-80-054
March 1980
Research and Development
SERA
Supercritical Fluid
Regeneration of
Activated Carbon for
Adsorption of Pesticides
-------
RESEARCH REPORTING SERIES
Research reports of the Office of Research and Development, U.S. Environmental
Protection Agency, have been grouped into nine series. These nine broad cate-
gories were established to facilitate further development and application of en-
vironmental technology. Elimination of traditional grouping was consciously
planned to foster technology transfer and a maximum interface in related fields.
The nine series are:
1. Environmental Health Effects Research
2. Environmental Protection Technology
3. Ecological Research
4. Environmental Monitoring
5. Socioeconomic Environmental Studies
6. Scientific and Technical Assessment Reports (STAR)
7. Interagency Energy-Environment Research and Development
8. "Special" Reports
9. Miscellaneous Reports
This report has been assigned to the ENVIRONMENTAL PROTECTION TECH-
NOLOGY series. This series describes research performed to develop and dem-
onstrate instrumentation, equipment, and methodology to repair or prevent en-
vironmental degradation from point and non-point sources of pollution. This work
provides the new or improved technology required for the control and treatment
of pollution sources to meet environmental quality standards.
EPA REVIEW NOTICE
This report has been reviewed by the U.S. Environmental Protection Agency, and
approved for publication. Approval does not signify that the contents necessarily
reflect the views and policy of the Agency, nor does mention of trade names or
commercial products constitute endorsement or recommendation for use.
This document is available to the public through the National Technical Informa-
tion Service, Springfield, Virginia 22161.
-------
EPA-600/2-80-054
March 1980
Supercritical Fluid Regeneration of
Activated Carbon for Adsorption
of Pesticides
by
R.P. DeFilippi, V.J. Kyukonis,
R.J. Robey, and M. Modell
Arthur D. Little, Inc.
20 Acorn Park
Cambridge, Massachusetts 02140
Grant No. R804554
Program Element No. 1BB610
EPA Project Officer: Max Samfield
Industrial Environmental Research Laboratory
Office of Environmental Engineering and Technology
Research Triangle Park, NC 27711
Prepared for
U.S. ENVIRONMENTAL PROTECTION AGENCY
Office of Research and Development
Washington, DC 20460
-------
ABSTRACT
The objective of this program was to perform laboratory-based
studies directed toward development of a new process for activated
carbon regeneration based on supercritical carbon dioxide as a desorbing
solvent. Supercritical CO^ at temperatures in the range of 30-250°C,
and pressures above about 80 atm, is a good solvent for organics,
with mass-transfer properties superior to ordinary liquids.
A series of pesticides was screened for suitability for treat-
ment by carbon adsorption and supercritical CC^ regeneration: Alachlor,
Atrazine, Carbaryl, Pentachlorophenol, Trifluralin, and Diazinon.
Alachlor and Atrazine wereselected as candidates for further study.
Both pesticide solutions permitted repeated regeneration over multiple
cycles with a low average capacity loss per cycle. Alachlor-loaded
carbon was regenerated 31 times. All pesticide candidates showed a
substantial capacity decline after the first regeneration (30+%); after
several cycles, both Alachlor and Atrazine exhibited a stable working
capacity.
Process studies showed that regeneration is rapid: a 30-minute
regeneration cycle is feasible. At least at a temperature of 120°C,
regenerability and rate of desorption was unaffected by the presence of
water in the carbon pores. Time of exposure of GAC to adsorbent in-
fluenced regenerability: initial-cycle decline was less for shorter
exposure times, even when saturation of the GAC was achieved. Desorption
rate increased with temperature; higher regeneration pressure (275 atm
vs. 150 atm) gave improved regenerability.
Through seven cycles, regeneration in a closed-loop recycle-C02
system gave Alachlor effluent levels of 0.2 ppm or less. Regeneration
of 4-ft (120-cm) long columns in desorption gave concentration-time
traces similar or slightly better (faster regeneration) than those from
1-ft (28-cm) long columns. Treatability studies carried out with a plant
sample of Atrazine manufacturing wastewater showed that a stable but
low working capacity of GAC was achievable. Depending on regeneration
pressure, working capacities of 0.05 to 0,08 g TOC/g GAC were obtained
for the pressure range of 150 to 275 atm, at 120°C.
This report was submitted in fulfillment of Grant No. 804554010
by Arthur D. Little, Inc. under the sponsorship of the U.S.-Environmental
Protection Agency. This report covers the period January 31, 1977, to
May 31, 1979.
ii
-------
TABLE OF CONTENTS
I.
II.
ABSTRACT
TABLE OF CONTENTS
LIST OF FIGURES
LIST OF TABLES
NOTATION
INTRODUCTION
CONCLUSIONS
III. RECOMMENDATIONS
IV. BACKGROUND
V. PESTICIDE SCREENING STUDIES
VI. MODEL SYSTEM STUDIES
VII. PROCESS DEVELOPMENT STUDIES
VIII. PLANT WASTEWATER TREATABILITY STUDY
IX. PROCESS DESIGN AND ECONOMIC ANALYSIS
X. REFERENCES
XI. APPENDICES
A. LOCAL EQUILIBRIUM THEORY
B. PHYSICAL PROPERTIES & DESIGN CALCULATIONS
ii
iii
iv
vii
ix
1
3
5
6
27
59
122
136
152
156
170
m
Arthur D Little, Inc
-------
FIGURES
Page.
IV-1 Reduced Pressure-Density Diagram 11
IV-2 Solubility Map of Naphthalene in SCF and NCL Carbon Dioxide 13
IV-3 Solubility of Si02 in Water 15
IV-4 P-T Projections of the Phase Diagrams and Critical Locus of
Binary C02 + Alkane Mixtures 16
IV-5 Diffusivity of Carbon Dioxide in the Near Critical Region 18
IV-6 Chromatographic Separation of n-Alkanes with Supercritical C02 23
IV-7 Chromatographic Separation of Polynuclear Aromatic Hydro-
carbons with Supercritical C02 24
IV-8 Chromatographic Separation of Some Oxygenated Compounds with
Supercritical C02 25
V-l Experimental Apparatus for Supercritical Fluid Extractions 31
V-2 Adsorption Isotherm 34
V-3 Batch Adsorption Rate Curves 36
V-4 Schematic Diagram of Adsorption Apparatus 37
V-5 Adsorption Breakthrough Curve, High Flow Rate Qu 39
H
V-6 Adsorption Breakthrough Curve, Low Flow Rate Q. 40
V-7 Solubility of Pesticides in SCF 42
V-8 Solubility of Alachlor and Carbaryl at Low Pressures 43
V-9 Adsorption Isotherm of Four Pesticides 45
V-10 Adsorption from Carbaryl Solution (Series C-l) 46
V-ll Adsorption from Carbaryl Solution (Series C-2) 47
V-l2 Diazinon Adsorption Breakthrough Curves 50
V-13 Atrazine Adsorption Breakthrough Curves 51
V-l4 Alachlor Adsorption Breakthrough Curves 52
V-l5 Comparison of Virgin Alachlor Adsorption Breakthrough for AL-1
and AL-2 Series 54
V-16 Series Al-1 Alachlor Adsorption Breakthrough Curves 55
V-l7 Series Al-2 Alachlor Adsorption Breakthrough Curves 56
V-18 Effect of Regeneration Pressure on Alachlor Breakthrough
Curves: Al-1 Series 57
V-l9 Effect of Regeneration Pressure on Alachlor Breakthrough
Curves: Al-2 Series 58
iv
-------
FIGURES (continued)
Page
VI-1 Adsorption and Desorption Apparatus 60
VI-2 Desorption of Phenol from GAC with Supercritical C02 63
VI-3 Adsorption of Phenol on Filtrasorb 300 64
VI-4 Adsorption of Phenol on CC-1230 65
VI-5 Adsorption of Phenol on GX-31 66
VI-6 Adsorption of Phenol on XE-348 67
VI-7 Prolonged Adsorption Breakthrough of Phenol on F-300:
2500 ppm Feed. 71
VI-8 Loading as a Function of Time During Prolonged Adsorption:
2500-ppm Phenol on F-300 72
VI-9 Prolonged Adsorption Breakthrough of Phenol on F-300: 120 ppm 73
VI-10 Loading as a Function of Time During Prolonged Adsorption:
120 ppm Phenol on F-300 74
VI-11 Phenol Loading as a Function of Time x Concentration 76
VI-11A Phenol Loading After 2-Day Adsorption 77
VI-12 Adsorption Breakthrough Curve for Acetic Acid on F-300 82
VI-13A Desorption Curves for Alachlor Following 1 Day of Adsorption 85
VI-13B Desorption Curves for Alachlor Following 3 Days of Adsorption 86
VI-13C Desorption Curves for Alachlor Following TO Days of Adsorption 87
VII-1 Alachlor Adsorption Breakthrough Curves 90
VII-2 Alachlor Adsorption Breakthrough Curves: Comparison of
Slow and Rapid Loading 90
VII-3 Alachlor Adsorption Breakthrough Curves with F-300 92
VII-4 Alachlor Adsorption Breakthrough Curves with F-400 93
VII-5 Alachlor Adsorption Breakthrough Curves: Effect of GAC
Mesh Size on Effluent Quality 95
VII-6 Solubility of Alachlor 96
VII-7 Schematic Diagram of Desorption and Separation Apparatus 98
VII-8 Schematic Diagram of Recycle Test Apparatus 99
VI1-9 Comparison of Complete Breakthrough Curves for Alachlor
Adsorption 101
VII-10 Adsorption of Alachlor: Closed-Loop Regeneration Series" 102
VII-11 Adsorption of Alachlor: Four-Foot Column 105
VI1-12 Desorption Curves for Small and Large Columns 106
VII-13 Reproducibility of First-Cycle Desorption Curves 110
-------
FIGURES (continued)
Paje
VII-14 Regeneration Curves for Small and Large Columns: First Cycle 112
VII-15 Regeneration Curves for Successive Cycles, 380-g Columns 114
VII-16 Breakthrough Histories at Various R Values 117
VII-17 Desorption Curve for 2500 ppm Phenol H**
VII-18 Variation of Desorption Curve with Number of Transfer Units 120
VII-19 Variation of Regeneration Curve with Number of Transfer Units 121
VIII-1 Atrazine Wastewater Adsorption Apparatus 124
VIII-2 Total Organic Carbon Analyzer (TOCA) 126
VIII-3 Adsorption Breakthrough Curve in Atrazine Wastewater 128
VIII-4 Atrazine Wastewater Regeneration Apparatus 129
VII1-5 Regeneration Apparatus with High-Pressure UV Detector 133
VII1-6 High Pressure UV Desorption Trace of Atrazine 134
IX-1 Schematic of a SCF Adsorbent Regeneration System 137
IX-2 Number of Transfer Units vs Throughput 140
IX-3 Process Flow Diagram 145
IX-4 Piping and Instrumentation Diagram 146
A-l Cylindrical Volume Element 156
A-2 Column Profile as a Function of Time 158
A-3 L.E.T. Desorption Curve 160
A-4 L.E.T. Regeneration Curve 160
A-5 Alachlor Desorption Curve 163
A-6 Best-Fit Regeneration and Desorption Curves 164
A-7 Best-Fit for 10-Day Alachlor Adsorption 166
A-8 Best-Fit for 3-Day Alachlor Adsorption 168
A-9 Best-Fit for 1-Day Alachlor Adsorption 169
-------
TABLES
Page
IV-1 Critical Conditions of Common Fluids 12
IV-2 Solubility of Organic Compounds in Liquid C02 19
V-l Characteristics of Selected Pesticides ' 28
V-2 Molecular Structure of Selcted Pesticides 29
V-3 Components of Solubility Test Apparatus 32
V-4 Characteristics of Granular Activated Carbon 33
V-5 Components of Dynamic Adsorption Apparatus 38
V-6 Adsorption Data of Carbaryl, Alachlor, Atrazine, and Diazinon 49
VI-1 Phenol: Summary of Operating Conditions and Results 62
VI-2 Prolonged Adsorption of Phenol 69
VI-3 Effect of Adsorption Period on Working Capacity 78
VI-4 Regeneration of High C0~ Flow Rate 80
VI-5 Adsorption Data for Acetic Acid 83
VII-1 Size Characteristics of F-300 and F-400 94
VII-2 Alachlor Regeneration Results 100
VII-3 Loading Data for Alachlor Closed Loop Series 103
VI1-4 Regeneration Results, Three 380-g Columns 111
VIII-1 Properties of Atrazine 123
VIII-2 Adsorption Conditions and Results 130
IX-1 Design Calculations, Small Column Case 141
IX-2 Design Calculations, Large Column Case 142
IX-3 Summary of Desorber Analysis 143
IX-4 Plant Component List 148
IX-5 Estimated Processing Costs 151
vn
-------
TABLES (continued^
Page.
B-l Physical Properties of CO 170
B-2 Properties of F-300 GAC 171
B-3 Operating Parameters 172
B-4 Adsorption Conditions 173
B-5 Mass Transfer Coefficients; Pore Diffusion 174
B-6 Mass Transfer; Film Coefficients 176
B-7 Pressure Drop in Packed Beds 180
B-8 Analysis of Breakthrough Curve 184
B-9 Analysis of R Based on 120 ppm and 2500 ppm Results 186
vm
-------
NOTATION
a Interfacial area of solid phase per volume of bed, crrr/cc
b Correction factor for kinetic coefficient
o
c Concentration of solute in fluid phase, y/cm
c Fluid phase concentration of solute at equilibrium with initial solid
loading,
Cc Fluid-phase concentration of solute in regenerant fed to column inlet,
g/cmj
c(L,t) Concentration of solute in regenerant at column outlet, g/cm
c . Solubility of solute in water, g/cm^
c . Solubility of solute in regenerant, g/cm
3
K Adsorption equilibrium constant, cm /
kf Mass transfer coefficient for fluid phase
k Mass transfer coefficient for solid phase
k Langmuir adsorption constant for aqueous solutions fo solute, pprrf
w
L Length of adsorbent bed, cm
N Number of transfer units
q Solute loading per weight of adsorbent, gr/gr
q. Solid-phase concentration of irreversibly adsorbed solute, g/g
ir reversibly adsorbed
q Maximum solid loading corresponding to a monolayer gr/gr
q Initial solid-phase concentration at the start of regeneration, g/g
qt t i Total solid-phase concentration of solute, g/g
R Dimension!ess adsorption constant, 1 + Kc
T Dimensionless throughput, UC t/q pgL
t Time, min.
t Time following the arrival of the fluid front at the column exit, min.
t Residence time for non-adsorbing fluid to elute bed, min.
tr Reduced time (t/t0), bed volumes of regenerant
tr. Reduced time for t = t, bed volumes
tr? Reduced time to complete regeneration, bed volumes
t. Time for plateau at c to elute bed, min.
t? Time to completion of regeneration, min.
IX
-------
Notation (continued)
U Superficial velocity
v Average fluid intersticial velocity, cm/min
v Characteristic velocity of a given fluid concentration
X Dimensionless concentration, c/c
x Solute mole fraction
x(L,tr) Concentration solute in regenerant at column exit, rt. fr.
X0 Fluid-phase concentration in equilibrium with qo, rt. fr.
Z Distance from column inlet, cm
Greek Symbols
e Interstittal void fraction
K Kinetic mass-transfer coefficient in the Thomas equation, (1)
PB Adsorbent bulk density
-------
I. INTRODUCTION
Granular activated carbon treatment is presently in limited use for
industrial wastewaters from the manufacture of pesticides and other hazardous
chemicals. The major factor constraining expanded use is cost, and much of
the cost is associated with carbon regeneration for reuse. The existing
thermal regeneration process for granular carbon using the multiple-hearth
furnace is capital and energy intensive, and responsible for losses that lead
to high cost for make-up carbon. The production of corrosive gases such as
HC1 is also a problem in any thermal process. Other regeneration methods
under development appear to have drawbacks also. For example, liquid solvent
extractive regeneration is expensive because of the requirements for removal
of all solvent from regenerated carbon and for purifying spent solvent for
recycle.
A novel extractive regeneration process employing supercritical fluids
(dense gases) as the extracting agent has been under development at Arthur D.
Little. A supercritical fluid is any fluid at pressures and temperatures
above the critical point. Supercritical carbon dioxide has high solubilities
for organic compounds. Moreover, solubility changes rapidly with pressure,
so that the solvent can be freed of solute by pressure changes much like
pumping and expanding a liquid. Diffusion coefficients for solutes in super-
critical fluids are about an order of magnitude higher than in liquids; thus,
the desorption rate of adsorbate from activated carbon into supercritical C02
is much more rapid than the corresponding liquid extraction.
Early tests in our laboratory using different organics adsorbed on activated
carbon confirmed that there is relatively rapid and effective regeneration using
supercritical CO-. A comparative engineering and cost analysis indicated that
both capital and operating costs for the process could be significantly less than
those for multiple-hearth furnace regeneration. Based on these results, research
and development of the supercritical C0? regeneration process were continued
under funding by the Chemical Process Branch of the Industrial Environmental
Research Laboratory, EPA/RTP.
Because of the importance of effective treatment of pesticides manufacturing
wastewater, this industry segment was chosen as the focus of the development
program. Pesticides play a major role in modern life. They are essential to
the production of food and natural fibers, the preservation of wood as a
structural material, and the control of disease. They are, however, toxic by
definition, and that toxicity can extend to man and other mammalian species.
Control of pesticide discharges into the environment represents a difficult
problem. An important step towards the solution is the control of pesticides
discharged by point sources. These sources fall into two categories: technical
pesticides manufacture, and pesticides preparations and formulations production.
Greater emphasis is placed on the reduction of wastes from pesticides manufacture,
because much higher tonnage of hazardous wastes is produced in manufacturing
compared to preparation and formulation. (Gruber, 1975).
-------
The severity of the problem of pesticides residues is underlined by EPA
action in banning the use of several of the more persistent and toxic pesticides.
While needed to eliminate the most hazardous materials, these bans place the
burden on substitute materials which are considered more acceptable, but also
have at least some degree of toxicity. The total need for pesticides will not
diminish; indeed, projections indicate that the total pesticides growth rate
is on the order of 10% per year (Anon., 1975). The great majority of pesticides
are manufactured by processes which release contaminated wastewaters. These
wastewaters are successfully treated in many cases, but in other cases they are
dealt with using methods which may not be permissable in the future, such as
deep well disposal. Additionally, the effectiveness of wastewater treatment
methods presently used for adequate reduction of hazardous effluents may be
questionable in some cases. Thus, the need exists for alternate treatment
processes which will be effective in eliminating these hazardous wastes. One
treatment process which has been successful in a number of industrial plants is
activated carbon adsorption. Wider study of adsorbent treatment of pesticide
manufacturing wastewaters is on-going, much under the sponsorship of EPA, and
it shows promise of extended applications.
Many pesticides are expensive chemicals, the average price being of the
order of $1.00/lb. Raw-material chemicals are also expensive, as indicated by
the fact that starting materials are a major cost in the manufacture of pesticides.
Manufacturing wastewaters can contain as much as several percent of starting
or product chemicals, and thereby represent an economic loss as well as a
serious pollution problem. A wastewater treatment process which would allow
recovery of chemicals from these aqueous wastes would have obvious benefit.
In addition to removing hazardous organic solutes from the wastewater, the process
would recover chemicals for recycle; in some cases, process water may be
recycled as well. Thus, the ideal of closed loop operation may be approached,
with the costs of pollution control at least partly covered by chemicals
recovery.
This rationale has been a major factor in considering an activated carbon
process which is nondestructive of the adsorbate. The ability of activated
carbon or other adsorbents to remove hazardous chemicals from wastewaters may
be combined with an adsorbent regeneration process which would permit chemicals
recovery. The development of this capability was a basic objective of this program.
The program approach included two major initial phases carried out in
parallel: screening of a series of pesticides to aid in the selection of
candidates suitable for further study; and model system studies to develop a
better understanding of the fundamental operations controlling the process. These
were followed by process development studies to determine optimum conditions
and mode of operation; treatability studies using plant samples of pesticide
manufacturing wastewater; and engineering-design and economic evaluations.
-------
II. CONCLUSIONS
A. PESTICIDE SCREENING
1. The following pesticides were screened as aqueous solutions for their
suitability for treatment by carbon adsorption and supercritical C02 regen-
eration: Alachlor, Atrazine, Carbaryl, Pentachlorophenol, TrifluraTin, and
Diazinon. Alachlor was selected as the candidate for further study, and
Atrazine was selected as a backup candidate. Both pesticide solutions per-
mitted repeated regeneration over multiple cycles with a low average capacity
loss per cycle. Alachlor-loaded carbon was regenerated 31 times.
2. All pesticide candidates showed a substantial capacity decline after
the first regeneration (30+%). After several cycles, both Alachlor and
Atrazine exhibited a stable working capacity.
3. While solubility of pesticide in SCF COp of at least several tenths
of a percent was observed in all cases, there was no correlation between
solubility level and regenerability.
B. MODEL SYSTEM STUDIES
1. Phenol as a model compound exhibited regeneration behavior similar to
the selected pesticides: an initial drop in capacity on the first cycle, followed
by constancy of working capacity over additional cycles.
2. Regeneration was rapid, as predicted. Rates depended on conditions;
a 30-minute regeneration cycle is feasible.
3. At least at a temperature of 120°C, regenerability and rate of desorp-
tion was unaffected by the presence of water in the carbon pores. It was
removed rapidly by SCF C0?.
4. Time of exposure of GAC to adsorbent influenced regenerability. Initial-
cycle decline was less for shorter exposure times (even when saturation of the
GAC was achieved). This may be caused by a slow irreversible adsorption or
chemical reaction occurring on the carbon surface.
5. Desorption rate increased with temperature. At 250 C and 15 min of
regeneration, a working capacity was obtained which was roughly equivalent
to that at 120 C for 60 min.
6. Higher regeneration pressure (275 atm vs 150 atm) gave improved
regenerability, as well as consistently higher solubility in SCF COp.
7. Acetic acid is a weakly adsorbed solute (one-third the working capacity
of phenol) which could be completely desorbed from GAC, as tested in an eight-
cycle series.
-------
C- PROCESS DEVELOPMENT
1. The Alachlor content of TOO ppm snythetic solutions could be reduced
to an effluent level below 0.2 ppm by virgin and SCF-regenerated carbon.
2. Regeneration in a closed-loop recycle CO^ system gave regenerability
results similar to once-through C0?. Effluent levels after GAC treatment
with recycle-CO?-regenerated carbon could be reduced to 0.2 ppm.
3. Regeneration of 4-ft (120-cm) long columns gave desorption concen-
tration-time traces similar or slightly better (faster regeneration) than
those from 1-ft (28-cm) long columns.
4. Local equilibrium theory was effective in modeling desorption
behavior over almost the entire desorption run length. Tailing, probably
due to mass transfer resistances, was longer than predicted by local equilibrium.
Batch adsorption isotherms measured in SCF (XL were not effective in pre-
dicting column dynamic behavior using local equilibrium assumptions.
5. The contribution of mass-transfer kinetics to SCF (XL desorption appears
to be predictable, based on known relationships such as those used in the
Thomas model. Differences in one-foot and four-foot column behavior appeared
to be rationalized on this basis.
D. PLANT-SAMPLE TREATABILITY
Treatability studies carried out with a plant sample of Atrazine manufacturing
wastewater showed that a low, stable working capacity of GAC was achievable.
Depending on regeneration pressure, working capacities of 0.05 to 0.08 g
TOC/gGAC were obtained for the pressure range of 150 to 275 atm, at 120 C.
E. PROCESS ECONOMICS
An example case of phenol treatment was used to design and cost a system
for regenerating 10,000 Ibs per day of spent GAC. The capital cost was estimated
at $800,000, and the operating cost at $0.085 per Ib of regenerated carbon.
No credit for recovered adsorbate was taken in these costs.
-------
III. RECOMMENDATIONS
1. The SCF regeneration process should be tested at larger scale. This
could be done at a carbon regeneration capacity in the range of 1-5000 Ibs
per day.
2. The process concept of adsorption and regeneration in the same small -
volume vessels should be developed and tested. This provides less operating
complexity, and eliminates losses of carbon due to transfer. It may require
small particle-size carbon to improve adsorption kinetics, but the attendant
increased pressure drop in the bed may be a worthwhile trade-off.
3. A mobile pilot plant should be built and used for treatability
studies on site. Usually, excessive sample volumes rule out testing at a
location away from the manufacturing operation producing the wastewater.
4. Bench-scale studies should be expanded to a broader range of adsorbates
and plant samples.
5. Further model studies should be performed to help verify the present
understanding of both physical and chemical processes that occur in adsorption
and desorption. These should include tests to help clarify the time-dependent
effects observed in this program.
-------
IV. BACKGROUND
The purpose of this section is to provide adequate background in-
formation in three areas directly related to this program: current
regeneration processes for activated carbon applied to wastewater treat-
ment; desorption of adsorbates from activated carbon using liquid solvents;
and other work in supercritical fluid extractions.
t
A. CURRENT REGENERATION PROCESSES FOR ACTIVATED CARBON
1. Thermal Processes
By far the predominant methods for activated carbon regeneration are
thermal processes which destroy the adsorbate. For granular carbon, the
multiple hearth furnace is in widespread use, and the rotary kiln has
been used to a limited extent. Recently, fluid-bed and radiative furnaces
have been reported to be under development. For powdered carbon, a
commercial thermal transport process has been described.
In each thermal system, the same basic steps occur. The spent carbon,
superficially drained of free water, is introduced into the furnace where the
remaining water is removed by evaporation. This water, largely occupying
the pore volume, is of the order of 40 to 50 percent by weight of wet
granular carbon, and 75% of wet powdered carbon. After drying, the tempera-
ture of the carbon is then raised to about 1500°F which causes volatilization
of lower molecular weight organics, followed by pyrolysis of the heavier
adsorbed components and volatilization of the pyrolyzed fragments. The
third step involves gasification of the carbon residue by carbon dioxide
and steam reforming reactions at temperatures up to about 1700 F. In
addition, there is some oxidation of the carbonized residue and the gaseous
volatile components.
Added fuel is required for the entire process. Evaporation of water
accounts for about 25 percent of the heat requirement; the balance goes into
sensible heat in elevating both carbon and gaseous constituents to reaction
temperature, and a small amount is needed to provide for the net endothermic
pyrolysis and gasification reactions involved. A large excess of steam is
employed to favor the steam-carbon reaction, and oxygen content is kept
low to minimize combustion of the activated carbon adsorbent.
-------
2. Non-Thermal Processes
a) Wet-Air Oxidation for Powdered Carbon
Because of the problem with losses in thermal regeneration of powdered
activated carbon, attempts have been made to apply the Zimpro wet-air
oxidation process to powdered carbon regeneration. In this process, an
aqueous slurry of spent carbon is contacted with air at temperatures
between 390 and 470 F, and pressures above 225 to 500 psi (Zimpro 1974).
A 0.8 M6PD municipal system has been tested over a period of about 50
days (Burant, 1973). In spite of carbon losses from several sources,
the make-up carbon cost was estimated to be only about 1.2<£ per thousand
gallons, which is a favorably low figure.
b) Alkaline Regeneration for Acid Adsorbates
In some industrial wastewaters, the major organic constituent is
acidic, and capable of being adsorbed by granular activated carbon. In
these cases, the adsorption equilibrium may be shifted by desorbing with
an aqueous solution at alkaline pH. This process has been applied by
Dow Chemical in Midland, Michigan, to phenol in wastewater (Himmelstein,
1974). A similar process has been used in a large-scale plant by Sherwin
Williams with p-cresol (Minor, 1974). The economic prospect for such a
process depends strongly on the relative recovery values for any given
application. An additional benefit is the ability to regenerate in situ,
avoiding the need to transport carbon from the contacting vessel to the
regeneration system.
c) Solvent Regeneration
The use of organic solvents to regenerate carbon has been considered,
although no commercial applications are presently known. There have been
pilot-scale studies on organics adsorbed from coke plant flushing liquor,
regenerated with hot benzene solution (Lovens, 1974). Process costs may
be high, at least in cases where the solvent and adsorbate cannot be
reused without some separation and/or refining step.
Because supercritical fluid desorption is a special case of solvent
regeneration, a more detailed review of past research in this area is
presented in Section IV-B.
d) Biological Regeneration
Attempts have been made to utilize the biological activity of micro-
organism cultures to regenerate carbon (Lovens, 1974; Perrotti, 1974).
Among other factors, its effectiveness depends on the degree of bio-
degradability of the adsorbed organics. While some restoration of carbon
adsorptive properties is possible, nothing approaching complete regeneration
has been observed. Since a significant amount of biological activity
exists in a carbon bed treating municipal wastes, it is probably more
appropriate to view this as combined biological and physical-chemical
-------
treatment (as in the PACT powdered activated carbon treatment process),
rather than a regeneration method. Renewed interest in biological
regeneration has accompanied recent research on activated carbon for
drinking-water treatment (Suffit, 1980).
3. Comparative Process Analysis
The multiple-hearth furnace is by far the most prevalent process
equipment in use. While the great majority of operating experience has
been in regeneration of carbon in industrial process use, its use for
wastewater treatment is growing. However, based on published information,
the multiple hearth furnace is still considered to be costly and difficult
to operate.
Costs reported over the last few years for furnace regeneration of
activated carbon show that operation of high-capacity systems run on the
order of 11-19^: per pound regenerated (Remirez, 1977). According to data
from the Lake Tahoe activated carbon plant (EPA, 1973), the regeneration
operating costs are about 70% of the total operating costs, exclusive of
amortization. For regeneration capacities ranging from 5,000 to 60,000
Ibs/day, capital costs are reported as $0.85 to $4.20 million for multiple
hearth furnace systems (Remirez, 1977). A figure for systems costs of
$1.0 million for minimum capacity furnaces have also been published
(Shuckrow, 1977).
Other thermal processes for regenerating granular carbon presently
offer no apparent advantage. Reactive regeneration systems, such as
alkaline regeneration of carbon after adsorption of organic acids, may
be economical, but only in special cases. They have the advantage of
avoiding the thermal degradation of carbon, and provide the opportunity
for recovery of the adsorbed species where desirable. However, they do
not have a broad range of application.
In contrast, solvent regeneration processes may hold substantial
promise, in that they could be capable of handling a broad spectrum of
adsorbates. At present, however, this potential advantage is outweighed
by problems such as the difficulty of solvent removal from carbon and the
attendant loss of adsorptive capacity if solvent removal is incomplete;
the cost of solvent losses in the carbon bed; the cost of solvent processing
in a separation step for recovering the solvent for recycle; and the
creation of secondary problems such as the formation of an oily condensate
if steam stripping is used.
Powdered carbon may become attractive in the future because of its low
unit cost. At present, however, the problems of handling the fine powder
and the high losses in thermal regeneration are difficult to overcome.
The latter may be solved by the Zimpro process, but there are unknowns
associated with hidden losses, such as residual carbon in the ash blow-
down, and oxidation of the adsorbent.
-------
B. DESORPTION FROM ACTIVATED CARBON USING LIQUID SOLVENTS
Although very few carefully defined experiments have been made, there
is sufficient data to conclude that organic adsorbates can be desorbed
from activated carbon by extraction with conventional liquid solvents.
The extraction is, in general, a slow and tedious process. Whether or
not adsorbates can be removed completely by liquid solvents is open to
question. The ease of desorption and the extent to which desorption
is complete appears to depend upon the nature of the carbon, solute
and solvent.
The major fraction of the carbon surface area is relatively uniform
and non-polar, with graphitic-like character. On this part of the surface,
solutes are held predominantly by physical adsorption; London dispersion
forces act as relatively weak bonds between solute and surface. A
small but significant fraction of the surface is heterogeneous and
polar. These sites have been attributed to inorganic impurities and
'surface oxides' (carboxyl, hydroxyl and carbonyl groups) that are formed
during the oxidative activation of the carbon during preparation
(Coughlin, 1968; Mattson, 1969; Snoeyink, 1967).
The number and nature of these sites vary from one carbon to another,
depending upon the procedure used for activation. On such sites, ad-
sorption of polar compounds may occur by chemisorption or by chemical
reaction of solutes with surface oxides. Thus, adsorption on activated
carbon ranges from weak to very strong.
The degree to which adsorbates can be removed by a liquid solvent
depend on the type of adsorption. At one extreme, physically adsorbed
solutes can be removed readily by solvent extraction; indeed, some low
molecular weight, volatile solutes can be removed by heating or steam-
stripping. At the other extreme, chemisorbed or chemically bound solutes
cannot be removed by solvents; they can only be removed by chemically
reacting them off the carbon (e.g., by oxidation during thermal regeneration)
While solvent extraction of many organics is technically feasible,
liquid solvent regeneration has not attained commercial viability. There
are a number of serious drawbacks to scaling up the analytical methods
to an industrial process for carbon regeneration. In particular, the
rate of desorption is very slow and the degree of removal of solutes is
far from complete. In addition, commercial solvents are relatively
expensive and must be recovered for reuse. Consequently, solutes must
be removed from the solvent after desorption. For moderately volatile
solutes, recycle of solvent will require extensive distillation. Further-
more, many conventional solvents are health hazards in themselves; thus,
they must be completely removed from the carbon before the adsorbent is
reused. If the solvent is air- or steam-stripped from the carbon, the
stripping agent may require further processing to avoid emission of an air
pollutant.
-------
For efficient and economic regeneration of activated carbon, the desired
solvent characteristics are (i) high solubility for the adsorbates, (ii)
favorable mass transfer properties for rapid desorption and (iii) high
volatility for subsequent separation of solutes. Liquid solvents exhibit
high solubility but slow diffusion. In the gas phase, mass transfer is
rapid but solubilities are very low. Fluids in the region of their critical
temperatures and pressures, called supercritical fluids (SCF), represent
a good compromise. The density is typically a third that of the normal
liquid: high enough to provide for good solubility, yet low enough to
permit high diffusivity and rapid mass transfer. We have found that
activated carbon and other adsorbents can be regenerated efficiently by
desorbing adsorbates with supercritical fluids.
C. SUPERCRITICAL FLUIDS AS EXTRACTING AGENTS
The supercritical fluid region can best be visualized with the aid
of a reduced pressure-density diagram, shown in Fig. IV-1. The dashed
curve is the focus of liquid-vapor equilibrium, which terminates at the
critical point (C.P.). In terms of reduced density,p/pCP the density of
the normal liquid is about 2.6: in other words, the critical density
is about 40% of that of the normal liquid. The critical conditions of
common materials are given in Table IV-1.
The SCF region, as we use the term, refers to reduced temperatures
in the range of 1 to 1.4 and reduced pressures from 1 to 6. We shall also
make reference by the near-critical liquid (NCL) state, which refers to
the region bounded by .95
-------
0.1
3.0
Reduced Density
FIGURE IV-1 REDUCED PRESSURE-DENSITY DIAGRAM. SUPERCRITICAL FLUID (SCF)
AND NEAR-CRITICAL LIQUID (NCL) REGIONS, AS INDICATED
(After Giddings,etal., 1968)
11
-------
Table.IV-1 Critical Conditions of Common Fluida
Tc(°C) Pc(atm) Pc(8/cm3)
Ethylene 9.9 50.5 0.23
Chlorotrifluoromethane 28.8 38.2 0.58
Carbon dioxide 31.0 72.9 0.47
Ethane 32.2 48.2 0.20
Tetrafluoroethylene 33.3 38.9 0.58
Nitrous oxide 36.5 71.7 0.46
Methyl fluoride 44.6 58.0 0.31
Sulfur hexafluoride 45.6 37.1 0.75
Propylene 91.9 45.4 0.23
Chlorodifluoromethane 96.4 48.5 0.52
Propane 96.7 42.0 0.22
Carbon disulfide 104.8 65. 0.45
Dichlorodifluoromethane 111.7 39.4 0.56
Dimethyl ether 126.9 52.6 0.26
Ammonia 132.3 111.3 0.24
n-Butane 152.0 37.5 0.23
Sulfur dioxide 157.5 77.7 0.53
Nitrogen dioxide 157.8 100. 0.56
Methyl ethyl ether 164.7 43.4 0.27
Diethyl ether 193.6 36.3 0.27
i-Pentane 196.6 33.3 0.23
n-Hexane .234.2 29.6 0.23
Isopropanol 235.3 47.0 0.27
Acetone 235.9 47. 0.28
Methanol 240.3 78.9 0.27
Ethanol 243.4 63.0 0.28
Chloroform 263.4 54. 0.50
n-Heptane 267.0 27.0 0.24
Benzene 288.9 48.3 0.30
Water 374. 218. 0.32
12
-------
10"
10"
c
o
•^
u
o
o
e
10"
Temperature (°C)
10 20 30 40
NCL —*j« SCF
50
SATURATED \
LIQUID
CRITICAL/"";
SATURATED
VAPOR
N
/6.5MPa,25°C
/ tie line
FIGURE IV-2 SOLUBILITY MAP OF NAPHTHALENE IN SCF AND NCL CARBON DIOXIDE
13
-------
We have reason to believe that the solubility map shown in Fig. IV-2
is representative of a broad range of solid solutes in supercritical
fluids. The solubility behavior shown in Fig. IV-2 results from the
usual interplay of secondary valence forces between molecules in solution.
These forces are the same ones that result in departure from ideal gas
behavior. If accurate equations of state for the mixture were available,
then it should be possible to predict the solubility behavior from
chemical potential or fugacity equality, which is the criterion of
equilibrium. The Peng-Robinson (P-R) equation of state, which is a
recent modification of the Rredlich-Kwong equation of state .is reported
to be fairly accurate in the sub- and supercritical regions (Peng, 1976).
We are in the process of using the P-R equation to predict the naphtha-
lene-£0p solubility map; results to date indicate that the method works
well. Thus, we believe that the solubility map shown in Fig. IV-2 is
not unique to the naphthalene-^ system, but is representative of the
behavior of solid solutes in supercritical fluids. For example, the
solubility of silica in supercritical water is shown in Fig. IV-3
(Kennedy, 1950). The similarities in the shapes of the isobars in
Figs. IV-2 and IV-3 are striking. In the SCF region, water is not
hydrogen-bonded to any appreciable extent; it behaves as a moderately
polar liquid solvent with dielectric constant of 2.5-10 (Franck, 1970).
Although there is scant data in the literature on solubilities of
organic solids and liquids in supercritical fluids, there is a large
body of data on solubilities in near-critical liquids. For example,
Francis (1954) reported the mutual solubilities of NCL (XL at 25 C with
each of 261 organic substances (Table IV-2 ). Each of these solubilities
represents a single data point on a solubility map. For example, Francis
reports the solubility of naphthalene in liquid C02 at 25 C as 2 wt-%.
In Fig. IV-2 the corresponding point lies on the saturated liquid curve
at 25 C, which is 0.62 mole-% or about 2 wt-%. Thus, Francis1 data can
be used as a guide for the magnitude of solubility in the critical region.
It should be noted that nearly half of the compounds studied were completely
miscible with NCL C02- The naphthalene solubility behavior discussed above
is representative of the less soluble organics in C02-
From Fig. IV-2 we see that solubility at high pressure (150-300 atm)
and supercritical temperature can be substantially higher than the sol-
ubility in saturated liquid at 25 C. At some point of higher pressures
and temperatures, solutes that are only slightly soluble at 25 C will
become completely miscible. That is, the system of solute + C02 reaches
a mixture critical point beyond which only a single, homogeneous phase
exists. Such critical loci have been reported for a number of organics
with COo (Schneider, 1970). The pressure-temperature projections of the
critical loci for mixtures of C02 and several alkalenes is shown in Fig. IV-4
(Liphard, 1975). For octane, there exists a continuous vapor-liquid
14
-------
o
CN
I
*
I
QUARTZ — + - GAS
3 PHASE REGION g \\
QUARTZ + LIQUID + GAS
160 200 240 280 320 360 400 440
Temperature °C
480 520
560 600
Figure IV-3 Solubility of SiO? in H90 (Kennedy, 1950)
15
-------
100
50
1 phase
-60
+60 +120 +180
T/K-273.15
+240 +300
'FIGURE IV-4 P-T PROJECTIONS OF THE PHASE
DIAGRAMS AND CRITICAL LOCI OF
BINARY CO2 + ALKANE MIXTURES.
FULL LINES, CRITICAL CURVES;
BATCHED LINES, VAPOR PRESSURE
CURVES OF CARBON DIOXIDE AND
OCTANE, RESPECTIVELY.
(Liphard and Schneider, 1975)
16
-------
critical locus connecting the critical points of the pure materials. There
is also a liquid-liquid critical locus that exists at low temperatures.
To the right of the L-L critical and above the L-G critical, only one phase
exists; in that region, octane and C02 are completely miscible. Note
from Table IV-2 that both heptane and dodecane are completely miscible
with liquid C02 at 25 C; thus, octane would also be completely miscible
at 25 C, which is corroborated by Fig. IV-4 . For tridecane and higher
molecular weight alkanes, the L-G and L-L critical loci merge into a
single, continuous critical locus. To the upper right side of the
critical locus, the system is completely miscible; below it and to the
left, the system exhibits two or more phases of limited miscibility.
Note that for hexadecane at 25°C, the critical point occurs at around
250 atm. At 25 C and pressures below 250 atm, the hexadecane-C02 system
exists as two phases. At 25 C and 65 atm, Table IV- 2 lists a solubility of
8 wt-% hexadecane in the C02-rich phase. Reasoning in the converse
manner, it might be anticipated that those substances exhibiting limited
solubility in liquid C02 at 25 C would be completely miscible with C02
at higher pressures and temperatures.
This brief discussion of solubility phenomena in the critical region
illustrates several important generalities:
(1) Enhanced solubility in the critical region is a
general phenomenon that arises from the normal
interplay of secondary valance forces.
(2) A large number and variety of organic compounds
are partially or completely miscible with C02
in the near critical liquid region.
(3) Partially miscible systems at NCL conditions tend
to complete miscibility at supercritical conditions.
In addition to high solubility, a good extracting agent must exhibit
high diffusivity for rapid mass transfer. There is relatively little direct
data on diffusivities in supercritical fluids, but there is a growing
body of indirect evidence for rapid mass transfer.
Several investigators have measured self-diffusivity of COp in the
sub- and supercritical region (for a compilation, see VargaftiK, 1975).
We have correlated these data and developed a map of self-diffusivity,
as given in Fig. IV-5 . In liquid C0? at 25 C, the diffusivity is about
an order of magnitude higher than that of conventional liquid solvents.
The vapor diffusivity at 25°C is another factor of 10 higher. At supexj-
critical temperatures, the diffusivities vary between 10-3 and 10-4 Cm /
sec; decreasing from the higher to lower values with increasing density.
17
-------
10
-2
CM
E
_o
>
^•fj
]>
"ws
3
10'
Saturated Li
15
20
io-5
Typical Diffusivities of
Normal Liquids
20
40 60
Temperature (°C)
80
100
FIGURE IV-5 DIFFUSIVITY OF CARBON DIOXIDE IN THE SCF AND NCL REGIONS
18
-------
TABLE IV-2
SOLUBILITIES IN LIQUID C02 AT 25°C AND ITS SATURATION PRESSURE
(Solubilities 1n Weight Percent)
M = Complete Mlsclbility
Paraffins and Naphthenes
n-Butane M
Cyclohexane M
Decahydronaphthalene 22
(decalin)
2,2-Dimethylpentane M
n-Dodecane M
Ethane , , M
n-Heptane " M
n-Hexadecane 8 .
Methylcyclobexane M
n-Octadecane (mp 28°C) 3
Paraffin wax (mp 52°C) ;1
Propane M
n-Tetradecane 16
2,2,3-Trimethyl butane (trlptane) M
Olefins
1-Decene M
1-Octadecene 10
Propylene M
Aldehydes
Acetaldehyde M
Benzaldehyde. M
n-Butyraldehyde M
Cinnamaldehyde 4
Crotonaldehyde M
1-Heptaldehyde M
Hydroclnnamaldehyde 17
Paraldehyde ' M
Proprionaldehyde M
Salicylaldehyde M
Valeraldehyde M
Aromatic
Benzene
Blbenzyl (mp 52.5°C)
Biphenyl (mp 71°C)
Chlorobenzene
a-Chloronaphthalene
Di -sec-butyl benzene
p-Dichlorobenzene (mp 53°C)
o,a-Dichlorotoluene
p-Dimethoxybenzene (mp 53°C)
Dimethyl naphthalenes (mixed)
2,4-Dinitrochlorobenzene (mp 62°C)
Dlphenyl methane (mp 27°C)
o-Hydroxyblphenyl (mp 56°C)
2-Methoxybiphenyl (mp 29°C)
a-Methoxynaphthalene
a-Methyl naphthalene
6-Methyl naphthalene (mp 35°C)
Naphthalene (mp 52°C)
Nitrobenzene
o-Nltrobiphenyl (mp 37°C)
o-N1trochlorobenzene (mp 32°C)
a-N1tronaphthalene (mp 58°C)
o-N1trotoluene
p-Nltrotoluene (mp 51 °C)
Phenylcyclohexane
Tetrahydonaphthalene (tetralin)
Toluene
Tr1 -sec-butyl benzene
a.ct.a-Trichlorotoluene
Ketones
Acetone
Acetophenone
Benzalacetone (mp = 42°C)
(4-Phenyl -3-butane-2-one)
Benzophenone (mp = 48°C)
2-Butanone
(methyl ethyl ketone)
Chloroacetone
Cyclohexanone
2,5-Hexanedlone
4-Hydroxy-4-methyl -2-pentanone
(dl acetone alcohol)
2-Octanone
N
1
2
M
1
M
M
M
M
2
1
4
1
5
1
6
9
2
M
2
21
1
M
20
8
12
M
M
2
M
5
M
M
M
M
19
-------
TABLE IV-2 (continued)
SOLUBILITIES IN LIQUID COg AT 25°C AND ITS SATURATION PRESSURE
(Solubilities In Weight Percent)
Alcohols
t-Amyl alcohol
Benzyl alcohol
sec-Butyl alcohol
t-Butyl alcohol
8-chloroethanol
Cinnamyl alcohol (mp 30°C)
Cyclohexanol
1-Decyl alcohol
Methanol
B-Ethoxyethanol (Cellosolve)
Ethyl alcohol
2-Ethylhexanol
Furfuryl alcohol
Heptyl alcohol
Hexyl alcohol
Isopropyl alcohol
Esters
Benzyl benzoate
Butyl oxalate
Butyl phthalate
Butyl stearate
S-Chloroethyl acetate
Ethyl acetate
Ethyl acetoacetate
Ethyl benzoate
Ethyl chloroacetate
Ethyl chloroformate
Ethylene diformate
Ethyl formate
Ethyl lactate
Ethyl maleate
Ethyl oxalate
Ethyl phenylacetate
Ethyl phthalate
Ethyl salicylate
Ethyl succlnate
D-Hydroxyethyl acetate
Methyl acetate
Methyl benzoate
Methyl formate
Methyl phthalate
Methyl salicylate
Phenyl phthalate (mp 70°C)
Phenyl salicylate (mp 43)
M
8
M
M
10
5
4
1
M
M
M
17
4
6.2
M
M
10
M
8
3
M
M
M
M
M
M
M
M
M
M
M
M
10
M
M
17
M
M
M
6
M
1
9
Phenols
o-Chlorophenol
p-Chlorophenol (mp 43°C)
2-Chloro-G-phenyl phenol
o-Cresol (mp 30°C)
m-Cresol
p-Cresol (mp 36°C)
2,4 Dichlorophenol {mp 45°C)
p-Ethylphenol (mp 46°C]
o-Nitrophenol (mp 45°C]
Phenol (mp 41°C)
B-Methoxyethanol
p-Methylcyclohexanol
Phenylethanol
Tetrahydrofurfuryl alcohol
Carboxylic Acids
Acetic acid
Caproic acid
Caprylic acid
Chloroacetic acid (mp 61°C)
a-Chloroproprionic acid
Formic acid
Isocaproic acid
Lactic acid
Laurie acid
Oleic acid
Phenylacetic acid (mp 77°C)
Amides
Acetamide (mp = 82°C)
N,N-Diethylacetamide
N ,N-Di ethylformami de
N,N-Dimethylacetamide
N ,N-Dimethylformamide
Formamide
M
8
1
2
4
2
14
1
M
3
M
4
3
3
M
M
M
10
26
M
M
0.£
1
2
0
1
M
M
M
M
0.5
20
-------
TABLE IV-2 (continued)
SOLUBILITIES IN LIQUID CO? AT 25°C AND ITS SATURATION PRESSURE
(Solubilities In Weight Percent)
Amines and Nitrogen Heterocycllcs
Aniline 3
o-Chloroaniline 5
m-Chloroaniline 1
N.N-Diethyl aniline 17
N,N-Dimethylaniline M
Oiphenylamine (mp 53°C) 1
N,N'-Diphenylethylene diamine
(mp 62°C) 1
N-Ethylaniline 13
N-Ethyl-N-benzylaniline 4
N-Methylaniline 20
a-Naphthylamine (mp 52°C) 1
Phenylethanolamine 1
2,5-Dimethylpyrrole 5
Pyridine M
o-Toluidine 7
m-Toluidine 15
p-Toluidine (mp 45°C) 7
NltHles
Acetonitrile
Acrylonitrile
Benzonitrile
0-Hydroxypropi oni tri1e
Phenylacetoni tri1e
Succinonitrile (mp 54.5°C)
Tolun1tr1les (mixed)
M
M
M
1
13
2
M
-------
By comparison of Figs. IV-2 and IV-5 it can be seen that the decrease in
diffusivity with density is less than the increase in solubility with
density: at 40 C, the diffusivity decreases from 80 atm (8x10 cm /sec)
to 120 atm (1.5xlO~4) by a factor of 5.3, while the solubility increases
from 80 atm (0.1 mole-%) to 120 atm (1.2 mole-%) by a factor of 12.
Thus, we see that the gain in solubility at higher density more than out-
weighs the decrease in diffusivity. Furthermore, the diffusivities in
the SCF range are at least an order of magnitude higher than that of
conventional liquid solvents.
Indirect evidence for rapid mass transfer can be gleaned from the
field of supercritical fluid chromatography, where the SCF is used as the
mobile phase for column elution with liquid or solid stationary phases
(Gouw, 1975; Rijnders, 1969). The conventional technique of discrete
peak column chromatography is closely aligned to frontal chromatography
which, in turn, is equivalent to extraction of solutes from adsorbents.
In other words, the retention time in SCF chromatography depends on the
combination of solubility and mass transfer characteristics in much the
same manner as does extraction.
An example of the decrease in retention obtainable by increasing
density of the mobile phase is shown in Fig. IV-6 (Sie, 1966). The
chromatograms were obtained with C0? as the carrier at 40 C with a
squalane stationary phase. The three curves show the effect of increasing
pressure from 1 atm (gas) to 50 atm (expanded supercritical fluid;
reduced density = 0.2) to 80 atm (supercritical fluid; reduced density =
1.2). The retention time for octane decreases by a factor of 10 in going
from 1 to 50 atm and by another factor of 3 at 80 atm. One of the technical
inconveniences in SCF chromatography is that the SCF mobile phase can
extract the stationary liquid phase. The squalane stationary phase used
in Fig. IV-6 is the C^n solute shown in Fig. IV-4. Although the
conditions used in the SCF chromatography lie below the squalane critical
locus, there is a finite solubility of squalane in the carrier at high
pressure, resulting in migration of the stationary phase.
A striking example of the ability of SCF C02 to extract non-volatile
hydrocarbons is shown in Fig. IV-7 (Jentoft, 1976). This separation was
carried out at 39 C and 122 atm. Within 40 min, methyl-substituted
benzanthracenes are eluted. Fig. IV-8 illustrates the separation of
a variety of oxygenated hydrocarbons using pressure programming of super-
critical COp from 55 to 117 atm. The Carbowax 400 stationary phase shows
significant migration under these conditions; carbowax 4000 has a
solubility of 1.7 wt-% in SCF C02 (Giddings, 1968).
22
-------
Time, Minutes
FIGURE IV-6 SEPARATIONS OF N-ALKANES ON A SQUALANE
COLUMN AT 40°C WITH CO2 AS A CARRIER AT
DIFFERENT PRESSURES AND COMPARABLE
LINEAR RATES OF MOBILE PHASE.
(Sie, et al., 1966)
180
23
-------
40
30
_J_
20
Time (Min.
10
j
0
FIGURE IV- 7
SEPARATION OF POLYISIUCLEAR AROMATIC
HYDROCARBONS ON "PERM PHASE" ETH:
CO2 MOBILE PHASE AT 39°C AND 120 atm
(Jentoft and Gouw, 1976)
24
-------
100
ro
en
Flow Rate =
1.0 Liter/Min. (STP)
20
10
0
FIGURE IV- 8 CHROMATOGRAM OF SOME OXYGENATED COMPOUNDS. COLUMN:
CARBOWAX 400 ON PORASIL. MOBILE PHASE:
PRESSURE PROGRAMMED FROM 5 TO 115 atm.
(Gouw amd Jen toft. 1975}
CO2 AT 40°C.
-------
As a result of the unique characteristics of high solubility and
rapid mass transfer, there is increasing interest in exploiting super-
critical fluids as extracting agents. A number of recent French and
German patents describe the use of SCF COo to extract caffeine from
coffee and tea (Hag, 1973 a,b), cocoa butter from pulverized kernels
(c), aroma constituents from pepper, cloves, cinnamon and vanilla beans
(dj and selective extraction of nicotine from tobacco (e). The extracts
are free of trace impurities that would stem from the use of organic
liquid solvents. Furthermore, degradation of heat-sensitive solutes
is avoided because the processes employ mild temperatures, in the range
of 30 to 60°C. Supercritical fluids have also been investigated as
potential extracting agents for crude petroleum and residual oil
(Zhuze, 1957), coal tar (Wise, 1970) and coal (Haddocks, 1979).
26
-------
V. PESTICIDE SCREENING STUDIES
A. INTRODUCTION
Six pesticides were selected for initial screening based upon considerations
such as large production volume, inclusion of a variety of organic classes in
the screening, and reasonable safety in handling in laboratory tests. The
pesticides selected for laboratory testing were Carbaryl, Alachlor, Penta-
chlorophenol, Atrazine, Trifluralin and Diazinon. Technical and commercial
information about each one is given in Table V-l. The molecular structure
of the species is given in Table V-2 in order to give a graphic illustration
of the complexity of the molecules; as was related previously, not only do
organic moledules such as simple alcohols and carboxylic acids dissolve in
SCF COp, but complex polynuclear aromatic and heterocyclics as well.
In the screening phase, the following tests were carried out:
1. Measurement of pesticide solubility in supercritical
C02 at several temperature and pressure levels.
2. Determination of the adsorption isotherm on GAC of a
solution of pesticide in water.
3. Measurement of dynamic adsorption breakthrough using a
solution of pesticide in water.
4. Measurement of the extent of regeneration by subsequent
re-adsorption tests to determine capacity and capacity
recovery of GAC.
B. APPARATUS AND TEST PROCEDURES
The apparatus for the four general classes of tests stated above, the
experimental methods of obtaining data, and the calculational methods for reducing
the data to solubility and capacity values are described below. Other tests
requiring apparatus and procedures not previously seen in the literature are
presented in subsequent sections as they arise.
1. Determination of Pesticide Solubility
Before carrying out adsorption and regeneration tests on GAC which had been
loaded with a pesticide, it was necessary to determine the solubility of the
pesticide in SCF C0?; clearly, if the pesticide were not soluble in SCF, it
could not be removed from spent GAC.
27
-------
ro
CO
TABLE V~1
PESTICIDES SELECTED FOR SCREENING STUDIES
Pesticide Class Trade Name Production Volume Toxicity
in 10& lbs/yr LDso (Mg/Kg)
Carbaryl Alkyl carbamate, Sevin, Hexavin 50 560
insecticide
Alachlor Acetanilide Lasso 30 3,000
herbicide
Atrazine Triazine, AA tram, Fenamine 200 3,080
herbicide
Pentachlorophenol Wood preservative, Dowicide, Pentachlor 70 125-210
Molluscicide
Trifluralin Nitroaromatic, Treflan 35 3,700
herbicide
Diazinon Organophosphate, Spectracide, Diazide 15 100-150
Insecticide
-------
TABLE V-2 PESTICIDES (AND CLASS) TESTED FOR SOLUBILITY IN CARBON DIOXIDE
CHo^^fio COCH Cl
X. X 2
N
ALACHLOR (Acetanilide)
CH3CH2NH
NHCH(CH3)2
Cl
ATRAZINE (Triazine)
OCONHCH,
OH
CARBARYL (Carbamate)
Ct
CI
Cl
C!
PENTACHLOROPHENOL (Chlorophenol)
N(C3H7)2
NO.
NO.,
CF3
TRIFLURALIN (Nitroaromatic)
— P(OC2H5)2
I
0
N
CH
N ' CH(CH3)2
DIAZINON (Organophosphate)
29
-------
A schematic diagram of the laboratory apparatus used for solubility
measurements is shown in Figure V-l; details of the size and rating of the
individual components are given in Table V-3. In carrying out a solubility
determination, a given amount of material is loaded into the extraction vessel,
the system pressurized to a given level by compressing C02 from the supply
SCF C0? flow rate through the sample collector, and is measured by the rota-
meter and dry test meter. The C(L flow rate fojj almost all the solubility
measurements was in the range of about 5-10 SLM. For a single solubility
determination, the flow of SCF CCL through the extraction vessel was continued
for several minutes, stopped, and the collector removed and weighed. The
increase in wieght represented the amount of pesticide collected. The
solubility, in weight percent, was calculated as
c _ /weight of pesticide \ x -|QQ
weight of C0? passed through + weight of pesticide'
2. Granular Activated Carbon
The characteristics of the granular activated carbons used are summarized
in Table V-4. The standard carbon used was FiltrasorbR-300, which is in wide
application in industry today; unless otherwise indicated, all tests were run
with Filtrasorb®-300.
3. Adsorption Isotherm in Hater
Standard methods of obtaining adsorption isotherms were used. In the exper-
imental procedure, a Wrist Action Shaker Table was normally used to agitate flasks
which contain a given volume of pesticide solution (either synthetic or real
wastewater) to which has been added a given amount of GAC. Measurement of the
concentration change over the duration of the test allows the loading, X,
in grams pesticide/gram GAC to be calculated from the experimental data as
X = ^initial - Pprnfina1
ppm GAC
Results are usually plotted logarithmically to give a Freundlich isotherm,
as shown schematically in Figure V-2. The concentration was measured by ultra-
violet spectrophotometry using a Perkin Elmer 550 doublebeam spectrophotometer.
The UV spectrophotometer was calibrated by preparing solutions of various
concentrations and determining the calibration curve at that wavelength as
Response = k (concentration)
where k is the slope of the line (millivolts/ppm)
J,
Standard liters per minutes, referenced to 25 C and 1 atm pressure.
30
-------
FIGURE V-l
EXPERIMENTAL APPARATUS FOR SUPERCRITICAL FLUID EXTRACTION
Sample
Collector
Pressure
Reduction Valve
Extraction
Vessel
Compressor
Dry Test Meter
Rotameter
-*•
COSupply
-------
TABLE V-3
EXPERIMENTAL APPARATUS FOR SOLUBILITY DETERMINATION
Item
Make
Characteristics
Compressor
American Instruments,
Model 46-13427
Working pressure,
30,000 psi
Displacement, 120 cc/min
Extraction Vessel
American Instruments
3/8" ID x 11" tubing
Rotameter
Fisher & Porter, 10A3565Y Flow rating, 100 SLPM
Dry Test Meter
Singer, DTM 200
Flow rating, 200 SCFH
Valving and Piping
American Instrument and Pressure rating,
Autoclave Engineers 10,000 psi
Sample Collector
SGA Scientific
200 mm Drying Tube
32
-------
TABLE V-4
CHARACTERISTICS OF ACTIVATED CARBON USED
Type
F-300
F-400
GX-31
XE-348
Manufacturer
Calgon
Calgon
Amoco
Rohm & Haas
Mesh Size
8-30
12-40
16-30
20-50
Surface Area
M^/g
950-1050
1050-1200
2300-2500
500
33
-------
FIGURE V-2 ADSORPTION ISOTHERM
1.0
T 1 1 1 1—TT-T
1 1 1 1—I I I I
-s
c
•o
8
0.1
0.01
I I I I I I I I I
I I I I I I I
10
Concentration (ppm)
100
34
-------
During the determination of the adsorption isotherms, the solution con-
centrations were determined periodically. A sample of about 2 ml from each
flask was pipetted into a quartz UV cell, the response obtained, and the 2
ml aliquot returned to the flask. These periodic concentration measurements
were plotted and used to determine the approach to equilibrium and ultimately
the equilibrium loading on GAC; a plot of the measurements shows the rate of
concentration change (i.e., and adsorption rate curve), and a representation
of two such rate curves for two different GAC concentrations is shown in
Figure V-3.
4. Dynamic Adsorption Breakthrough Curves
The schematic diagrams of the adsorption apparatus is shown in Figure V-4,
and the details of the individual components are listed in Table V-5.
The breakthrough concentration-vs-time curve was obtained either by
continuous on-line UV monitoring of the column effluent in a flow-through
cell or by measuring individual grab samples obtained with a fraction collector.
2
Two different flow rates0were used in the adsorption tests, 7.5 gpm/ft
(0.51 cm/sec) and 1.1 gpm/ft'' (0.075 cm sec); the high flow rate resulted in
a faster breakthrough and was used in the initial screening tests in order to
assess in a reasonably short period of time the multicycle regeneration be-
havior of GAC loaded with a given pesticide. Later in the Process Development
portion of^the program, adsorption tests were carried out at a flow rate of
1.1 gpm/ft which more nearly approximates the flow rate used in many industrial
and municipal GAC installations. Representative examples of breakthrough
curves are shown in Figures V-5 and V-6 for high and low flow rates, respectively.
In almost all adsorption tests the synthetic solution was prepared at about
80-85% of the solubility limit of pesticide in water.
Determination of the amount of pesticide adsorbed on GAC was carried out
in several parallel ways: in integration of the breakthrough curve as indicated
in Figures V-5 and V-6; by measurement of the average concentration of the total
effluent collected from which the loading value is calculated as
X = (Total volume) (Cjn - Coff)
g~GAC
and in some cases by determining the actual amount loaded by gravimetric means
Change in weight of the GAC
x = Weight of GAC
For the gravimetric determination the GAC was dried with CO^ at 1 atm and
55°C, conditions which were found to remove only water and no pesticide.
35
-------
Initial Concentration
Q.
o
I
Ǥ
m ppm GAC
n ppm GAC
n >m
0, Time (hours or days)
FIGURE V-3 BATCH ADSORPTION RATE CURVES
36
-------
FIGURE V-4
SCHEMATIC DIAGRAM OF ADSORPTION APPARATUS
Double Beam
Ultraviolet Spectrophotometer
D
-O
Collection
Tank
Adsorption
Column
Pump
Feed
Solution
Tank
37
-------
TABLE V-5
EXPERIMENTAL APPARATUS FOR DYNAMIC ADSORPTION
Item
Make
Characteristics
Feed Pump
Milton Roy, Mini Pump
Gil son, HP 4/HF Minipuls
0.2-3.2 ml/min
(for low flow)
0-100 ml/min
(for high flow)
Adsorption Column American Instruments
3/8" ID x 11" Tubing
Ultraviolet Spec- Perkin-Elmer, Model 550 Double Beam, Variable
trophotometer Wavelength
38
-------
Influent Concentration
E
a
a.
c
0)
3
LU
C
C
o
1
<-"
0)
u
o
o
o"
Amount Adsorbed =
lfe
0
(C,n-Co)
e ] . Q
H
6, Time (hours or days)
FIGURE V-5 ADSORPTION BREAKTHROUGH CURVE, HIGH FLOW RATE QH
39
-------
ADSORPTION BREAKTHROUGH CURVE LOW FLOW RATE, QL
Influent Concentration
E
Q.
Q.
03
01
c
o
'+•>
CO
O
o
o
Amount Adsorbed =
I/O (Cin-C)d0J.QL
8, Time (hours or days)
FIGURE V-6 ADSORPTION BREAKTHROUGH CURVE, LOW FLOW RATE Q.
40
-------
5. Regeneration With Supercritical C0r
The regeneration apparatus was essentially identical to the solubility
apparatus shown in Figure V-l, the extraction vessel replaced with a GAC-filled
column of the same size. Regeneration tests were carried out in situ on a
pesticide-spent GAC column in the same manner as a solubility test; a given
amount of CO- at a given pressure was passed through the column, regulated by
the flow control valve, and expanded to 1 atm. The reduction in pressure caused
the pesticide to precipitate from the C0? in the sample collector and the total
CCL passing through the collector was measured by the rotameter and dry test
meter. The amount of pesticide precipitated in the collector was weighed and
compared to the residual weight remaining on the GAC to assess the closure of
the material balance. In almost all cases, material balances closed to within
90-95%.
After each regeneration, a re-adsorption test was made with the GAC column,
again in situ, the breakthrough curve measured, and the loading obtained as
described earlier for the virgin GAC tests. The ratio of the loadings, X, for
the second and first cycle adsorption is the percent capacity recovery after
one cycle
Recovery (%) = ^2_ (100)
' ' y
xl
The average capacity recovery after n cycles is given by the expression
Average Capacity Recovery (%) = n"
— . 100
Xl
Other apparatus and test procedures are presented as they relate to specific
tests.
C. RESULTS OF SCREENING TESTS
1. Solubility in Supercritical Carbon Dioxide
The solubility of the pesticide in supercritical C02 was a primary
criterion for pesticide selection. Solubilities of each pesticide were
measured at 275 atm's (4000 psia) and two temperature levels: 70°C and
12QOC. These data are summarized in Figure V-7.
As shown, the solubilities vary over a range of more than two orders
of magnitude. Trifluralin is soluble to the extent of about 20 wt. %;
Alachlor and Diazinon have solubilities in the range of 5% to 10%; Penta-
chlorophenol on the order of 1%; Atrazine and Carbaryl on the order of
.1 x? .4 wt. %.
Because it is desirable to run the regeneration at as low a pressure
as possible, some additional testing was done for solubilities at 150 atm's,
and the results are shown in Figure V-8. Both Alachlor and Carbaryl were
tested, and both showed solubility decreases of about an order of magnitude.
41
-------
30.0
i 1 1 1 1
10.0
Diazinon ~~
Alachlor
c
-------
30.0
10.0
8
I
o
I
4-1
O
O
O
1.0
0.1
Carbaryl
150atm
0.01
I
1
I
40 50 60 70 80 90
Temperature (°C)
100 110 120
FIGURE V-8 SOLUBILITY IN SUPERCRITICAL CO2
43
-------
2. Adsorption Isotherms
Adsorption isotherms for Alachlor, Atrazine, Carbaryl, and Diazinon
are given in Figure V-9. In each case, loadings were greater than 0.1
weight of adsorbate per weight of granular activated carbon; in fact, with
the exception of Atrazine, these pesticides showed adsorptions in excess
of 0.2 grams per gram of GAC over the entire concentration range studied.
It should be noted that the organic content of wastewaters from
pesticide manufacturing will often contain only a small proportion of the
pesticide product itself. However, the pesticide is usually difficult
to treat biologically, and is often selectively adsorbed relative to the
organics present. For this reason, it is important to determine the suit-
ability of GAC for the pesticide alone, even though the total organic
carbon removed by adsorption treatment may be small. Selective adsorption
of the pesticide would allow biological treatment to effect removal of
the balance of the TOC.
Adsorption isotherm studies were attempted for Trifluralin and Penta-
chlorophenol. Trifluralin was discontinued because, at the very low
Trifluralin concentrations in aqueous solution (about 1 ppm) the weight
of adsorption was too low for adsorption in a practical time period.
Preliminary column adsorption studies with Pentachlorophenol showed
that breakthrough was very diffuse, and thus GAC adsorption did not appear
to be effective as a treatment process. Because of this, further study
of Pentachlorophenol was discontinued.
3. Regenerability
Repeated adsorption/regeneration cycles were performed for four
pesticides: Carbaryl, Diazinon, Atrazine and Alachlor. Performance in
these tests were judged as the final citerion for pesticide selection.
a. Carbaryl
Two separate series of Carbaryl adsorption/regeneration tests
were carried out. Rapid loading adsorption breakthrough curves for Series
C-l are given in Figure V-10 and for Series C-2 in Figure V-ll. C-1-A1
and C-2-A1 are the breakthrough curves forCarbaryl adsorption on virgin
GAC, and the -A2 and -A3 curves are the breakthrough curves for the second
and third adsorption tests after previous regeneration.
The conditions for regeneration of Carbaryl-loaded GAC were 275 atm
at either 120°C or 70°C and at a C02 flow rate of 10 SLM. The duration
of the regeneration period was chosen on the basis of the solubility
measurements and the loading of pesticide on the GAC. The amount of C02
required to remove the pesticide based on solubility considerations was
calculated from the relation
Pesticide on GAC \ x inn
tnrough column) x 100
44
-------
I I I I I I
I I I I I I I I
0.01
1 10
Concentration (ppm)
FIGURE V-9 ADSORPTION ISOTHERMS FROM AQUEOUS SOLUTIONS
45
-------
E
Q.
Q.
C
g
'+-*
to
l-
*-»
C
I
o
Feed Concentration — 35 ppm
A C-1-A3
• C-1-A2
• C-1-A1
12
24
36 48
Time (Hours)
60
72
84
FIGURE V-10 ADSORPTION FROM CARBARYL SOLUTION (SERIES C-1)
46
-------
35
a
o
§
8
o
Feed Concentration — 35 ppm
A C-2-A2
O C-2-A1
GAC - 7.0 g
Flow - 7.6 gpm/ft2
_L
_L
12
24
60
36 48
Time (Hours)
FIGURE V-ll ADSORPTION FROM CARBARYL SOLUTION (SERIES C-2>
72
47
-------
Specifically, with a loading of 0.2 g/g and a solubility of 0.2 wt
at 275 atm and 1200C, the minimum volume of C02 in standard liters
required is
Volume of C02 (SL) - I-Q 2 x°? 8 ] = 39°
For the initial regeneration tests, the C02 flow for regeneration
was about three to five times that calculated from the solubility and load-
ing values given previously. For example, in regenerating C-1-R1 a total
flow of about 1800 SL was used.
Adsorption data are given in Table V-6. The adsorptive capacity
decreased by about 50% per cycle; no effort was directed to the deter-
mination of the reason for the capacity decrease with Carbaryl in the
screening phase.
b. Diazinon
Adsorption breakthrough curves for three Diazinon tests are shown in
Figure V-12. As in the Carbaryl series, a large capacity drop was ex-
perienced after each regeneration using the standard regeneration conditions
of 275 atm, 120°C, and 1800 SL or C02- Loading data are tabulated in
Table V-6. The Diazinon series was stopped after three cycles.
c. Atrazine
Figure V-13 gives the breakthrough curves for seven adsorption/
regeneration cycles using an Atrazine solution of 28 ppm. Corresponding
loading data are given in Table V-6. The loading of Atrazine on regen-
erated GAC is lower than that on virgin GAC; however, the capacity value
reaches a constant level after about three or four cycles. The breakthrough
and regeneration tests in the series were carried forward for three more
cycles to verify the finding. The average capacity recovery on the seventh
adsorption test was 93%.
Despite comparable solubilities in C02» Atrazine-loaded GAC could
be regenerated well while Carbaryl could not. This finding indicated that
the solubility level per se was not the predominant factor influencing
regenerability. The equilibrium distribution of pesticide between GAC and
the C02 solution influences the duration of regeneration. This is dis-
cussed in detail in Section VII.
Because of its favorable regenerability characteristics, Atrazine
was selected as a possible candidate for further process studies.
d. Alachlor
Initial dynamic adsorption breakthrough curves with synthetic
Alachlor solution at a flow rate of 7.5 gpm/ft^ are shown in Figure V- 14
and a tabulation of the loading data is given in Table V-6. The capacity
recovery for Alachlor-loaded GAC after regeneration is much higher than
for Carbaryl; whereas the capacity for Carbaryl decreased about 50% per
cycle, adsorptive capacity for Alachlor was still about 70% of virgin GAC
after ten cycles.
48
-------
TABLE V-6
GAG LOADING AND REGENERATION DATA
(Figures in gr/gr GAC)
CARBARYL
ALACHLOR
ATRAZINE
DIAZINON
Cycle # Loaded
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
0.24
0.13
0.07
Removed
0.09
0.08
Loaded
0.20
0.17
0.17
0.13
0.15
0.13
0.12
0.14
0.15
0.12
0.15
0.15
0.14
0.13
0.10
0.12
0.11
0.11
0.11
0.09
0.10
0.11
0.10
0.11
Removed
0.18
0.15
0.23
0.12
0.14
0.18
0.14
0.17
0.15
0.20
0.14
0.14
0.15
0.14
0.11
0.14
0.12
0.12
0.12
0.06
0.17
0.20
0.20
0.22
0.21
0.20
0.22
0.20
Loaded
0.11
0.07
0.06
0.06
0.06
Removed Loaded
0.07
0.04
0.06
0.06
0.04
0.23
0.14
0.04
Removed
0.09
0.07
-------
35
Influent Concentration - 33 ppm
a
o
u
GAG - 7.0 g
Flow -7.6 gpm/ft2
16
24 32
Time (Hours)
40
48
FIGURE V-12
ADSOPRTION FROM DIAZINON SOLUTION
50
-------
30
25
_ 20
Q.
a
o
15
10
Influent Concentration — 28 ppm
16
At-1-A7
A At-1-A5
V At-1-A4
O At-1-A3
At-1-A2
At-1-A1
GAC - 7.0 g
Flow - 7.6 gpm/ft2
40
48
Figure V-13
24 32
Time (Hours)
ADSORPTION FROM ATRAZINE SOLUTION
56
51
-------
Q.
Q.
O
I
130
120
110
100
90
80
70
60
50
40
30
20
10
Influent Concentration — 118 ppm
o
o
O O'
o
Figure V-14
V A1-1-A10
O A1-1-A9
• A1-1-A4
A A1-1-A2
• A1-1-A1
GAG - 7.0 g
Flow - 7.6 gpm/ft2
I
12 16 20
Time (Hours)
24
28
ADSORPTION FROM ALACHLOR SOLUTIONS
(SERIES AI-1)
52
-------
The dynamic adsorption tests were continued with the Series
Al-1 column and another series, Al-2 was started in parallel. The first
adsorption cycle of each series is compared in Figure V-15 in order to
given an indication of the experimental accuracy of the adsorption break-
through determinations; the two breakthrough curves for virgin GAC are
in good agreement.
Both of the Alchlor series were continued; a total of 31 cycles
for Al-1 and 18 cycles for Al-2 were carried out. Figure V-16 is a com-
posite of the 31 cycles for the Al-1 series, and Figure V-17 for the
Al-2 series.
Within each Alachlor series some tests with variations in re-
generation pressure were carried out. Figure V-18 shows three breakthrough
curves for adsorption carried out on GAC regenerated at two different
pressure levels., 1950 and 275 atm in the Al-1-series. Figure V-19 shows
three adsorption curves for similar tests in the Al-2 series. The curves
given in Figrues V-16 and V-19 show that a C02 volume of 1800 SL at
150 atm did not regenerated Alachlor-loaded GAC to as great an extent
as did the same total flow at 275 atm; however, as the figures show,
a subsequent 275-atm regeneration of the adsorption which followed the
150 atm regeneration returned the capacity to its previous 275-atm level.
The average capacity recovery for the Al-1 series was 98%.
Because of the demonstration of a reasonable capacity recovery
after 31 regeneration cycles, Alachlor was selected for concerted effort
in the Process Development phases; these studies are described in
Section VII.
53
-------
I
a.
o
I
o
O
O
125
100
75
50
25
Influent Concentration — 118 ppm
•A1-2-A1
•A1-1-A1
GAC-7.0g
Flow-7.6 gpm/ft2
12
16
20
24
28
Time (Hours)
Figure V-15 COMPARISON OF VIRGIN ALACHLOR ADSORPTION
BREAKTHROUGH FOR AI-1 AND AI-2 SERIES
54
-------
125
100
1
a
3 75
o
^^
CO
l_
+•»
c
0
u
J
S 50
•*-•
0
25
1
Feed Concentration 118 ppm
X Al
X A7 "*
•1° *°
xv ••
X 1 m
x Xa*BDA« *«.D*D %OBD "D "
3^R7"Dv*v v ^^.
^^-~-*~~~
***^
^
2 a"**
Vji^B ^^
VA /
• /
LJ J
• /
A y
/
V X Test Number Column Position
^M ^
- " /
X / X
A / A
/ D
/ •
- / V
AI-1-A26 First
AI-1-A27 Second
AI-1-A28 First
AI-1-29 Second
AI-1-30 First
AI-1-31 Second
/ Flow - 7.5 gpm/ft2
/ GAC - 7.00 g
/ ,
4 8
i i i i i
12 16 20 24 28
Time (Hours)
Figure V-16 ADSORPTION CURVES FOR ALACHLOR (SERIES D
55
-------
100
E
Q.
_a
o
75
§
J
*- 50
3
O
25
Feed Concentration 1 18 ppm
A •
X
• X
A
X
V
/'
A
D
v° v
v
D
n v
v
w
v n /
/
°f
7
Test Number
• AI-2-A11
X AI-2-A12
A AI-2-A13
D AI-2-A14
• AI-2-A15
V AI-2-A16
Flow - 7.5 gpm/ft^
GAG - 7.00 g
Column Position
Second
First
Second
First
Second
First
8
12
20
24
16
Time (Hours)
Figure V-17 ADSORPTION CURVES FOR ALACHLOR (SERIES 2)
28
56
-------
130
120
110
100
a
a.
I
*-»
S
c
8
S
3
O
90
80
70
60
50
40
30
20
10
Influent Concentration - 118 ppm
A A1-1-A22 (after 275 atm regeneration)
• A1 —1 —A23 (after 150 atm regeneration)
• A1-1-A24 (after 275 atm regeneration)
GAC - 7.0 g
Flow - 7.6 gpm/ft2
12
18
Time (Hours)
20
24
28
Figure V-18 ALACHLOR ADSORPTION BREAKTHROUGH CURVES:
EFFECT OF REGENERATION PRESSURE (AI-1)
57
-------
125
100
Influent Concentration — 118 ppm
Q.
a.
c
o
c
8
0)
1=
o
50
25
A A1-2-A9 (after 4000 psi regeneration)
D A1-2-A10 (after 2200 psi regeneration)
O A1-2-A11 (after 4000 psi regeneration)
GAC - 7.0 g
Flow - 7.6 gpm/ft2
I I
4
I
8
I l
12 16
Time (Hours)
I
20
I
24 2*
Figure V-19 ALACHLOR ADSORPTION BREAKTHROUGH:
CURVES EFFECT OF REGENERATION PRESSURE (AI-2)
58
-------
VI. MODEL SYSTEM STUDIES
A, INTRODUCTION AND PROCEDURES.
The purpose of the model system studies was to obtain fundamental
information about the supercritical CCL regeneration process using
known well-characterized adsorbates. The studies were initiated in
parallel with pesticide screening, using phenol as a model adsorbate,
When Alachlor was selected from the screening studies, it was used for
further work in this phase also.
The degree to which SCF C02 can regenerate adsorbents was deter-
mined by subjecting GAC columns to repetitive adsorption and desorption
cycles. The apparatus for these tests is shown in Fig. VI-1, and is
similar to that used in the screening studies. The adsorbents were
packed into 304-stainless steel columns and maintained at ambient
temperature during adsorption. Aqueous solutions of solute were fed
at constant flow rate with a minipump (Milton Roy, model 396-31). The
column effluent was monitored with an on-line UV spectrophotometer
(Perkin-Elmer, model 550) and the total effluent was collected. The
loading on adsorption was determined by one or more of the following
methods: (i) integration of the breakthrough curves; (ii) the change
in concentration between feed and total effluent collected; (iii) the
change in weight of the column if the column was dried prior to de-
sorption.
Prior to desorption, the column was drained of interstitial water.
In some runs, the column was dried with low pressure COp at 55°C,
Liquid C0? (siphon grade, min. purity 99.5%) was pumped to operating
pressure (Aminco diaphragm-type compressor, model 46-13427) and brought
to operating temperature in a preheater, passed through the column, and
then to a pressure let-down valve. Solute was collected in a cooled
U-tube. The instantaneous flow rate was measured by a rotameter and
the total C02 flow by a dry test meter.
Results reported herein were obtained with three column sizes.
The minicolumn dimensions were 0.42 cm i.d. x 5 cm and contained 0.37 g
activated carbon. The small columns were 0.95 cm i.d. x 28 cm and con-
tained 7.0 g carbon, unless otherwise stated. The large columns were
3.3 cm i.d. x 120 cm and contained 380 g carbon. Three commercially
available granular activated carbons (Calgon Filtrasorb 300, National
Coal Board Anthrasorb CC1230EH and Amoco GX-31) and a carbonized syn-
thetic resin (Rohm and Haas Ambersorb XE-348) were used. The standard
carbon for most tests was F-300, screened to -12 + 30 mesh; when not
otherwise stated, the standard carbon was used. Table V-4 (Section V)
gives properties of these carbons.
At the outset of the program, six variables were identified as the
key parameters to be studied in column regeneration tests: type of car-
bon, regeneration temperature, pressure, flow rate of regenerant, total
59
-------
01
o
Feed
Reservoir
Effluent
Reservoir
Feed Pump
UV
Detector
Recorder
V
Adsorption
GAC
Column
Pressure
Let-down Valve
Dry Test
Meter
Column Heater
Thermocouple
SCF
Pteheater
Compressor)
Low Pressure CC^
By-pass for Drying
co2
Cylinder
V
Desorption
FIGURE VI-1 ADSORPTION AND DESORPTION APPARATUS
-------
quantity of regenerant and water content (i.e., regenerating wet or dry).
During the course of this program, an additional parameter was found to
have significant impact on the regeneration process; time of exposure of
the column to aqueous feed during adsorption.
B. PHENOL REGENERATION
1. Regenerability Studies
To demonstrate the ability of SCF C02 to regenerate activated carbon, a
series of adsorption-desorption cycles was conducted using a mini-column of
Filtrasorb 300, screened to -100 + 120 mesh. Rapid adsorption and sharp break-
throughs were attained by using relatively low flow rates (.2 to .5 ml/min)
of concentrated phenol solution (10,000 ppm). After adsorption breakthrough
occurred (15 to 25 min) the column was dried with low pressure CO,,. Desorption
was conducted at 150 to 190 atm and 55 C. In some runs, the desorption curve
was obtained by flame ionization detection of the effluent.
A series of eight adsorption-desorption cycles was run on a single carbon
column. The results are shown in Table VI-1 as experiment 1. The loading of
the virgin carbon was 0.35 g/g, which is higher than that commonly reported for
phenol owing to the high feed concentration. The loading after the first re-
generation is 0.25 g/g and is relatively steady (within experimental error)
thereafter. The drop in loading after the first regeneration has been observed
in a number of cases, as will be discussed later.
The rate 'of desorption of phenol as a function of time is shown in
Fig. VI-2. This desorption curve, which was observed during the regeneration
following the fourth adsorption cycle, is typical. The concentration of solute
in the regenerant fluid peaks soon after the onset of regeneration, decreases
rapidly until 80 to 90% of the solute has been removed and then decreases slowly
until regeneration is complete. Note that the bulk of the desorption is very
rapid; 50% of the adsorbate is removed within 20 min and 90% within 1 h.
Desorption is complete within 3 h. Unlike desorption in liquid solvents, the
desorption of phenol from activated carbon using supercritical C02 is very
rapid.
A series of experiments was made with 0.95 cm i.d. columns containing three
different carbons and a carbonized synthetic resin. The F-300 carbon was screened
to +20 mesh; the other adsorbents were used as received. A phenol concentration
of 2380 ppm was used at a moderate flow rate of 3.2 ml/min (1.1 gpm/ft ).
During adsorption, effluent concentration was monitored with an on-line UV
spectrophotometer. When the effluent concentration reached 90% or more of
the feed concentration, adsorption was terminated and the GAC was regenerated
using SCF CO, at 120°C and 150 atm for 3 h at a flow rate of 8-9 standard
liters per mTnute (SLM). Note that the columns were not dried prior to de-
sorption.
The conditions and results are shown in Table VI-1 as experiments 2-5;
the adsorption breakthrough curves for the repetitive cycles are shown in
Figs. VI-3 through VI-6. Note that the loadings reported in the first adsorptions
are less than the equilibrium loadings because the adsorptions were terminated
after 5 to 7 h. The higher loading of GX-31 is due to a surface area of about
2 to 2-1/2 times that of standard GAC.
61
-------
Table VI-1. Summary of Operating Conditions and Results
ro
Experiment No.
Type of carbon
Solute
Weight of adsorbent(g)
Mesh size
Adsorption
Feed cone, (ppm)
2
Flow rate (gpm/ft )
Desorption
Temp. (°C)
Pressure (psig)
Capacity (g/g)
Cycle No.
1
2
3
4
5
6
7
8
1
F-300
phenol
0.37
-100+120
10,000
0.3-1.0
55
2200-2800
0.35
0.25
0.24
0.24
0.25
0.26
0.23
0.22
2
F-300
phenol
7.0
+20
2,380
1.1
120
2200
0.19
0.16
0.15
0.16
3
CC-1230
phenol
7.0
N/A
2,380
1.1
120
2200
0.21
0.18
0.18
4
GX-31
phenol
4.0
-16+30
2,380
1.1
120
2200
0.50
0.44
5
XE-348
phenol
7.9
-20+50
2,380
1.1
70
2200
0.17
0.15
0.12
0.13
0.14
-------
100
50
« 2.0 f
O
.Q
(Q
O
I
I
-2? 1.0 •
o
§
30
30
60 90 120 150
Desorption Time (Min)
180
210
F-300
CO2 Flow Rate - 0.6 8 (NTP)/Min
GAC Charge - 0.37 g
60 90 120 150
Desorption Time (Min)
180
210
FIGURE VI-2 DESORPTION OF PHENOL FROM GAC WITH SUPERCRITICAL CO2
63
-------
Feed = 2380 ppm phenol
Flow Rate- 1.1 gpm/ft2
pH of Stream - 3.0
(adjusted w/HCI)
GAC Charge - 7.0 g
FIGURE VI-3
ADSORPTION OF PHENOL ON F-300
(NUMBERS ON CURVES REFER TO ADSORPTION
CYCLE)
64
-------
Feed = 2380 ppm phenol
Flow Rate-1.1 gpm/ft2
pH of Stream — 3.0 (adjusted
2/HCI)
GAC Charge - 7.0 g
FIGURE VI-4 ADSORPTION OF PHENOL ON CC-1230
(NUMBERS ON CURVES REFER TO
ADSORPTION CYCLE)
65
-------
2300
E
Q.
o
I
1840
1380
920
460
Feed = 2380 ppm phenol
GX-31
Flow Rate - 1.1 gpm/ft2
pH of Stream - 3.0
(adjusted w/H CD
GAC Charge - 4.0 g
FIGURE VI-5 ADSORPTION OF PHENOL ON GX-31
(NUMBERS ON CURVES REFER TO
ADSORPTION CYCLE)
66
-------
2500
Feed = 2380 ppm phenol
1875
a
o
1
•M
§
<§
+J
0
1250
625
XE-348
Flow Rate-1.1 gpm/ft2
pH of Stream - 3.0
(Adjusted 2/HCI)
Adsorbent Charge - 10.0 g
I I
12345
Time (hr)
FIGURE VI-6 ADSORPTION OF PHENOL ON XE-348
(NUMBERS ON CURVES REFER TO
ADSORPTION CYCLE)
67
-------
All four adsorbents responded similarly to SCF C02 regeneration. Over 80%
of the virgin carbon capacity could be attained in the second adsorption cycle.
Where more than two cycles were run (experiments 2, 3 and 5), the loading was
essentially constant at that measured during the second cycle. Thus, all of the
carbons employed in these tests could be regenerated with SCF C02-
This set of experiments illustrates a phenomenon that has been observed in
a number of tests. The drop in capacity observed after the first regeneration
of phenol cannot be recovered by longer periods of regeneration using more (XL.
The decrease in capacity might be due to the formation of immobilized species,
either chemi-sorbed or products of chemical reaction. Since the time for
adsorption and desorption was short, the process involved is probably non-
activated. Since the formation of this immobilized species is rapid and only
occurs during the first adsorption, the surface sites involved in its
formation are unavailable in subsequent adsorptions. Thus, we tentatively have
attributed this irreversible adsorption to chemisorption or complex-formation
of phenol with high-energy surface sites.
It is also to be noted that in experiments ?to 5, the column was not
dried prior to desorption with SCF C0?. At 120 C and pressures above 15
MPa, the solubility of water in regenerant is high enough to provide for
drying and solute desorption within the same time frame. Thus, a separate
drying step between adsorption and regeneration is not necessarily required.
2. Prolonged Adsorption of Phenol
In conventional industrial practice, GAC columns remain on-stream for weeks
or months. In laboratory studies of phenol, adsorption isotherms, it has been
reported that several weeks are required to reach adsorption equilibrium
(Snoeyink, et_ al_., 1969). In the phenol experiments reported above, the
adsorption step was terminated after 7 h. If additional adsorption were to
occur beyond 7 h, it would be important to establish whether the additional up-
take is reversible or irreversible with respect to SCF regeneration. Thus,
two series of adsorption-desorption tests were conducted with prolonged duration
of the adsorption step. Feed concentrations were 120 and 2500 ppm phenol with
-12+30 mesh F-3QO. After each adsorption, the column was dried with low-
pressure C02 at 55 C and then weighed to determine the total loading. Conditions
for regeneration were 55 C and 150 atm. After regeneration, the column was
again weighed to determine the amount of solute desorbed. These results are
given in Table VI-2.
The behavior^of column P-l is similar to that shown previously for experiment
2 in Table VI-1. The loading after 6 h of adsorption amounted to 0.21 g/g,
of which over 80% could be desorbed by SCF CO,,.
*
When comparing the results of Tables Vl-land VI-2, it must be kept in mind
that column weights were not measured in the earlier tests. The loadings
given in Table VI-1 are the amounts adsorbed in the i-th cycle, as determined
by integration of breakthrough curves. Those values should be compared to the
sixth column of Table VI-2.
68
-------
TABLE VI-2. PROLONGED ADSORPTION OF PHENOL
Loadings determined by weight of dried column (g solute/g GAC)
Col umn
No.
P-l
P-B
P-A
P-C
Feed
Cone. Cycle
(ppm) No.
2500 1
2500 1
2
3
4
5
6
7
120 1
2
3
4
5
6
120 1
2
Adsorption
Time
(h)
6.1
48
96
96
168
96
192
192
96
192
168
264
408
360
504
96
Loading
After
Adsorption
.209
.261
.274
.294
.300
.327
.331
.334
.137
.176
.174
.194
.199
.200
,210
.210
(g/g)
After
Regeneration
.025
.106
.146
.154
.190
.200
.217
.230
.069
.099
.111
.126
.140
.141
.140
.150
Adsorbed
in TtR
Cycle (g/g)
.209
.261
.168
.148
.146
.137
.131
.117
.137
.107
.075
.086
.073
.060
.210
.070
Desorbed
in ith
Cycle (g/g)
.184
.155
.128
.140
.110
.127
.114
.104
.068
.077
.063
.071 v
.059
.059
.070
.060
-------
The adsorption breakthrough curves for column P-B are shown in Fig. VI-7
The first 6 h of the first cycle of P-B was identical to that of P-l. By
6 h, the effluent concentration is very nearly identical to the feed con-
centration; since the total adsorption after 48 h increases to 0.26 g/g, there
is an additional up-take of 0.05 g/g between 6 and 48 h. This slow, continuing
adsorption upon prolonged exposure is consistent with results of others cited
previously.
As shown in Fig. VI-7 for column P-B, there is a decline in capacity
with successive cycles, the decline being most pronounced in the first three
cycles. The per-cycle adsorption (Table VI-2, column 6) decreases faster than
the per-cycle desorption (column 7). By the sixth or seventh cycle, the gap
between the per-cycle adsorption and desorption has narrowed considerably,
indicating that we are approaching a steady state of constant working capacity.
The difference between per-cycle adsorption and desorption is the residual
remaining after regeneration (column 5). Although this residual builds up sub-
stantially over the 7 cycles, we note that the total adsorption (column 4) also
increases. In other words, the build-up of residual does not impact the reversible
adsorption on a one-to-one basis. This phenomenon is seen more clearly in
Fig. VI-8, wherein the results of Table VI-2for the 2500 ppm feed are plotted
as a function of totaj^ time of exposure of the GAC column to aqueous feed.
Over the 7 cycles, the column was in contact with aqueous solution for more
than 35 h. From Fig. VI-8, we see the total loading increasing monotonically
with exposure, the rate of build-up decreasing with increasing time. After 2 h,
the residual loading nearly parallels the total loading so that the fraction
desorbed shows a mild decrease with increasing time.
The results of a series of prolonged adsorption at 120 ppm is given in
Table VI-2 for column P-A. The adsorption breakthrough curves are given in
Fig. VI-9and loadings as a function of exposure time in Fig. VI-10. The trends
are similar to those observed from column P-B at 120 ppm feed.
When all of these results for phenol are viewed in perspective, it appears
that there are two forms of irreversibly adsorbed species (i.e., species
not desorbed by SCF C0? under the regeneration conditions). First, there is an
irreversible species tnat forms rapidly during the first cycle, resulting in a
residual even when the adsorption exposure is of the order of 20 min (Table VI-1
experiment 1). Second, there is an irreversible species that builds up
gradually over prolonged exposure and undoubtedly forms by an activated process.
The presence of the second irreversible species decreases the amount of the
reversibly adsorbed species (i.e., the solute that is desorbed during SCF C0?
regeneration), but the decrease is not in proportion to the increase in
activated irreversible species. Thus, it appears that the activated species
does not compete directly for the surface sites upon which the reversible
species adsorbs.
The activated irreversible species could result from a slow chemisorption
or from a chemical reaction which might be catalyzed by impurities in the
carbon pores (e.g., inorganics that might have been introduced during the
70
-------
2500
2000
o
§
J
c
at
3
tu
1500
1000
500
Feed = 2500 ppm
Phenol'
Flow Rate - 1.1 gpm/ft
pH of Stream - 5.0
GAC Charge - 7.0 g
-12+ 30 Mesh
47
48
Time (hours)
FIGURE VI-7
PROLONGED ADSORPTION OF PHENOL ON F-300 (NUMBERS
ON CURVES REFER TO ADSORPTION CYCLE)
71
-------
IN3
•S
?
Total Loading After Adsorption
Residual After Desorption
Removed by Desorption
20 25
Exposure to Aqueous Feed (days)
FIGURE VI-8 LOADING AS A FUNCTION OF TIME OF EXPOSURE DURING PROLONGED
ADSORPTION; PHENOL ON F-300
-------
O.25
0.20 -
0.15
-O
D>
O
I
0.10
CO
120ppm Feed
(Series P-A)
0.05
I
4
Cycles
FIGURE VI-9 PROLONGED ADSORPTION BREAKTHROUGH OF PHENOL
ON F-300: 120 ppm FEED (NUMBERS ON CURVES REFER
TO ADSORPTION CYCLE)
-------
.20
-O
o
O
• Total Loading After Adsorption
O)
Ol
T!
CO
O
.15
.10
.05
Residual After Desorption
Removed By Desorption
10
20 30 40
Exposure to Aqueous Feed (days)
50
60
FIGUREVI-10 LOADING AS A FUNCTION OF TIME OF EXPOSURE DURING
PROLONGED ADSORPTION; 120 PPM PHENOL ON F-300
70
-------
original activation of the carbon). In either case, we might expect the
build up of residual to be proportional to the phenol concentration in
solution. To test this hypothesis, loadings for both 120 and 2500 ppm feed
were plotted as a function of "equivalent exposure," defined as the product
of concentration and time. As shown in Fig. VI-11, the residual for both feed
concentrations fall close to a single curve, while the total loadings are
clearly different. The difference between total loading and residual is
the reversible portion, which is naturally much higher for the 2500 ppm feed.
In retrospect, we see that the high concentration, prolonged adsorption
series, P-B, accentuated the build up of residual. For example, at 120 ppm,
it would take an exposure of 2 years to reach the residual level of 0.23 g/g,
which was found after 37 days at the 2500 ppm feed level.
As a further test of the hypothesis that the time of exposure during
adsorption is the key variable in determining the build up of residual,
column P-C was exposed to 120 ppm phenol for 21 days. It was then regenerated
and reloaded for a second cycle of four-days duration. The results are given
In Table VI-2 and Fig. VI-11A(»,»,A). The results, as seen in Fig. VI-HA,
are in relatively good agreement with the proposed hypothesis; the residual
build up in the first cycle of column P-C is consistent with the build up over
3 to 4 cycles of column P-A, for which the combined adsorption exposure was
20-30 days.
It appears that the residual is due to a product of chemical reaction of
phenol on activated carbon. Subsequent experiments have shown that part of
the residual can be desorbed at higher temperatures. The solute recovered
is an orange solid, the identity of which has not yet been established. The
high-temperature desorption of residual is consistent with the results of
Seewald and Juntgen (1977), who used temperature-programmed desorption of
phenol from activated carbon. They found a broad peak of moderate surface
concentration that desorbed in the range of 100 to 200 C.
The results of the prolonged adsorption tests provide a number of important
implications for the optimal design of a SCF regeneration process. For a feed
of 2500 ppm phenol on Filtrasorb 300, the pertinent results are summarized in
Table VI-3. If SCF regeneration were used with a conventional adsorber design,
relatively large adsorption columns would be used and typically left on-stream
for days to weeks. The working capacity for SCF-regenerated 6AC would be 40%
of the virgin carbon capacity. On the other hand, if short adsorption cycles
were used, then the SCF working capacity would be 65% of that of virgin carbon.
Short adsorption cycles would dictate small adsorber bed volume, with a
significant reduction in carbon inventory.
In conventional adsorber design the bed volume is much larger than the
volume of the active adsorption zone (i.e., the zone in which rapid adsorption
occurs). Carbon is inactive for most of the adsorption period, either having
reached saturation or awaiting the arrival of the adsorption zone. Large
columns are preferred mainly to minimize the frequency of carbon transfer into
and out of the bed. On the other hand, with SCF regeneration, the bed can be
75
-------
T:
CO
O
Total Adsorption After Loading, 2500 ppm Feed
Total Adsorption After Loading, 120 ppm Feed
Residual After Desorption, 2500 and 120 ppm Feed
0
0
20
30 40 50 60
Equivalent Exposure (ppm x days x 10~**)
70
80
90
FIGURE VI-11
LOADING AS A FUNCTION OF THE PRODUCT
OF TIME AND FEED CONCENTRATION
-------
.20-
Total Loading After Adsorption
Residual After Desorption
Removed By Desorption
30 40 50
Exposure to Aqueous Feed (days)
FIGURE VI-11A LOADING AS A FUNCTION OF TIME OF EXPOSURE DURING
PROLONGED ADSORPTION; 120 PPM PHENOL ON F-300
-------
TABLE VI-3 EFFECT OF ADSORPTION PERIOD ON WORKING CAPACITY
(2500 ppm phenol on F-300; regenerated at 55 C, 150 ATM)
Virgin
GAC
Ca oa city
(g/g)
Steady-
State
Working
Capacity
(g/g)
Percent
of
Virgin
Long
Cycle
Long adsorption
cycle 0.26
0.10-0.11
40
Short adsorption
cycle 0.21
0.16-0.18
65
78
-------
regenerated in situ if the column is designed to withstand high pressure If
the adsorption period is made relatively short, then the cost of small high-
pressure vessels for adsorption and jn situ desorption can be traded off for
the cost of large, low-pressure adsorbers and large carbon inventory.
3- Effect of Temperature on SCF Desorption of Phenol
A series of experiments was conducted to determine the effect of tem-
perature on SCF regeneration of phenol. Three 7-g columns of F-300 were
loaded with 2500 ppm phenol during 7 h adsorption cycles. The columns were
dried and weighed prior to regeneration. Desorption was conducted at 150
atm and at two temperatures; 120 and 250 C. In these tests, the regenerant
flow rate was 50 SLM, which is a factor of 6 higher than those used previously.
The results are given in Table VI-4 .
Column PH-3 was regenerated to 3 h. The per-cycle adsorption dropped by
10% from the first cycle loading of 0.25 g/g to subsequent cycles at 0.22 -
0.23 g/g. This result is comparable or slightly better than that observed
for 120 C without prior drying (Table VI-1, experiment 2). After the first
cycle, the per-cycle desorption is slightly less than the per-cycle adsorption,
indicating at most a slight build-up of residual.
Column PH-4 was regenerated under the same conditions as PH-3, except that
the desorption was terminated after 15 min. The fact that the first cycle
descrption of 0.16 g/g was less than that of PH-3 (0.22 g/g) does not necessarily
imply more irreversible adsorption; rather, we believe that 15 min desorption
was not quite sufficient to remove all of the solute that is capable of being
desorbed at 120°C. This interpretation is supported by the higher per-cycle
desorption and gradual decrease in residual in subsequent cycles of the PH-4
series.
Column PH-5 was regenerated for 15 min at 250°C. Although three cycles
of desorption is barely sufficient to identify any trends, it appears that
15 min at 250°C is long enough to attain a working capacity comparable to 60
min at 120°C (PH-3).
C. ACETIC ACID REGENERATION
The ease of desorption is a function of the strength of adsorption. All
solutes adsorbing from the aqueous phase must compete with water for surface
sites. Thus, weakly adsorbed species generally exhibit relatively low loadings.
Acetic acid is a typical case of weakly adsorbed solutes.
For all three columns, the virgin, first cycle adsorption is 0.25 to 0.26
g/g, which is somewhat higher than that of column P-l (see Table B). The- VI-2
difference is believed to be due to the fact that another batch of 6AC was
used for these tests; such differences from batch-to-batch are not uncommon.
79
-------
TABLE VI-4. REGENERATION AT HIGH C02 FLOW RATE
(Adsorption at 2500 ppm for 7 hrs;
Desorption at 150 ATM)
Series PH-3
desorption
at 120°C;
50 SLM;
3000 SL
Cycle
No.
1
2
3
4
After
Adsorption
.25
.26
.26
.28
Loading (g/g)
After
Regeneration
.03
.03
.06
.06
Adsorbed
in ith
Cycle (g/g)
.25
.23
.23
.22
Desorbed
in ith
Cycle (g/g)
.22
.23
.20
.21
00
o
Series PH-4
desorption
at 120°C;
50 SLM;
750 SL
1
2
3
4
.25
.27
.26
.23
.09
.08
.05
.05
.25
.18
.18
.18
.16
.19
.21
.18
Series PH-5
desorption
at 250°C
50 SLM;
750 SL
1
2
3
4
.26
.24
.2.9
.30
.02
.05
.06
.26
.22
.24
.24
.24
.19
.23
-------
A series of adsorption-desorption cycles were made with acetic
acid of F-300 GAC screened to + 20 mesh. The results are shown in Fig. VI-12
For the high feed concentration used, the adsorption front broke through
almost immediately and the effluent concentration gradually rose to the
level of the feed. On the 7-g carbon column, the effluent reached the
feed concentration within 2 hr and the loading obtained on virgin carbon
is 0.048 g/g.
The column was regenerated, without drying, with SCF C0? at 120°C and
150 atm. The desorption was essentially complete within 30 fflin. As seen
in Table VI-5 the loadings obtained upon readsorption were within experimental
error of that obtained on virgin GAC. Only one curve is shown in Fig. VI-12
because the breakthrough curves were essentially identical for all eight
adsorption cycles. Thus, for this weakly adsorbed solute, capacity recovery
is complete and the rate of regeneration is extremely rapid.
D. ALACHLOR REGENERATION
As discussed in the preceding section, Alachlor was chosen as the
pesticide for further study. In the pesticide screening studies, Alachlor
was regenerated for more than 30 cycles. The working capacity declined
gradually from 0.20 to 0.10 g/g over the first 19 cycles and then re-
mained constant from the twentieth cycle to the conclusion of the series
at the thirty-second cycle.
The behavior of Alachlor is similar in kind to that observed for
prolonged loading of phenol at low concentration (i.e., 120 ppm). In
both cases, the loading declined most during the first cycle and then
declined more gradually in subsequent cycles. On the other hand, the
rate of adsorption of phenol is significantly faster than that of
Alachlor. This herbicide has a low water solubility (140 ppm at 25°C)
and relatively low diffusivity due to its high molecular weight (MW =
269} and somewhat bulky structure. During the course of determining
the Alachlor adsorption isotherm, the rate of adsorption was measured.
The results, shown in Fig. V-13 (Section V), indicate that over the
first 5 days, adsorption is still proceeding at a modest rate.
The question arises as to whether the build up of adsorbed Alachlor
over the prolonged adsorption period is primarily irreversible species,
as was the case with phenol. If reversibly adsorbed Alachlor is taken
up rapidly and irreversible adsorption occurring slowly, then we would
expect to find no more solute desorbed after prolonged exposure than
that obtained after short exposure.
To further elucidate the nature of Alachlor adsorption a series
of experiments were conducted in which the desorption curves were
measured after several different times of exposure. The desorption
curves were obtained by trapping solute during regeneration in the
apparatus of Fig. VI-1. The procedure for regeneration was as follows:
81
-------
2000
1500
E
Q.
O
-------
TABLE VI-5
ACETIC ACID REGENERATION
120°C, 150 atm
Cycle No. GAC Capacity,
g/g
1 (virgin) 0.048
2 0.043
3 0.044
4 0.045
5 0.046
6 0.044
7 0.045
8 0.050
83
-------
a. The column, containing spent GAC, was dried at 1 atm
and 55°C and then pressurized to the desired level (typically 275 atm
for the majority of the Alachlor regeneration tests).
b. A given amount of C0? was passed through the column,
as measured with the dry test meter.
c. The amount of Alachlor removed from the column and
collected in the cold trap was weighed and the average concentration
calculated; that concentration value is plotted as a "bar" in the de-
sorption curve.
The desorption curves obtained after one, three and ten days of
adsorption are shown in Fig. VI-13. The corresponding amounts of solute
collected were 0.105 g/g (1 day), 0.131 g/g (3 days) and 0.185 g/g (10
days). The increase in desorbable solute with longer adsorption period
is evidence that the adsorption of the reversible Alachlor species
occurs by a relatively slow process. The increase in initial concen-
tration of the effluent (i.e., at the onset of regeneration) in Fig. VI-13
as adsorption period increases is consistent with a higher loading of
reversibly adsorbed solute.
The slow build up of irreversibly adsorbed Alachlor does not neces-
sarily imply that the rate is controlled by a highly activated process.
Since Alachlor feed concentration was only 120 ppm (vs. 2,000 to 10,000 ppm
for phenol) and Alachlor diffusion is relatively slow, the rate of forma-
tion of an irreversible Alachlor species may be controlled by mass trans-
fer and not reaction kinetics. During each adsorption cycle, the major
fraction of Alachlor adsorbed can be desorbed. The reversibly adsorbed
species is undoubtedly physically adsorbed, the rate of which should not
be appreciably activated in the adsorption kinetics. Since a sharp break
in adsorption rate was not observed during the 28-h adsorption period, it
is likely that the rate of adsorption was controlled by mass transfer.
As the build-up of irreversibly adsorbed species occurred in parallel with
the reversible adsorption, the rate of irreversible adsorption of Alachlor
could well have been also diffusion-controlled. Thus, we cannot distin-
guish between activated and non-activated kinetics for adsorption of the
irreversible Alachlor species.
As discussed previously (Section VI-B, Phenol Regeneration),
it may be advantageous in commercial practice to use the same column for
adsorption and in situ regeneration (i.e., no transfer of GAC between ad-
sorption and regeneration). When reversible adsorption is slow, as in
the case of Alachlor, in situ regeneration is not feasible. On the other
hand, if the rate of adsorption is diffusion-controlled, then the rate
can be increased by using a narrow cut of smaller particles. Since this
represents a departure from commercially available GAC, the cost of the
adsorbent would undoubtedly require a premium. However, a higher cost of
adsorbent would not have a significant impact on the economics of the SCF
regeneration process because GAC is not lost during regeneration. Thus,
it may be desirable to use GAC that is tailored to the SCF regeneration
process. In this manner, bulky adsorbates such as Alachlor may be made
to undergo rapid adsorption with relatively high loading of the rever-
sibly adsorbed species.
84
-------
-A-—-
/r ~ ? r? lir •' C; • TVr-
Pj
Parameters for Generated Curves:
tr1 = 33;4 bed vol. q = 0.143 g/g
tr2 = 375. bed vol.
x,, = .0022 wt. fr.
qm = 0.203 g/g
K= 1910cm3/g
180 240 300
Reduced Time, tr (bed volumes)
FIGURE VI-13A REGENERATION AND DESORPTION CURVES AFTER ONE-DAY
ADSORPTION OF 120 PPM ALACHLOR
85
-------
SJ 0
4-J2I
Reduced Time, tr (bed volumes)
Parameters for General Generated Curve:
tr1 = 6.07 bed vol. qo = 0.179 g/g
tr2 = 375 bed vol. qm = 0.203 g/g
X0 = 0.0070 wt. fr. K = 1910cm3/g
Desorption Curve
180 240 300
Reduce Time, tr (bed volumes)
360
420
438
FIGURE VI-13B
REGENERATION AND DESORPTION CURVES AFTER THREE-DAY
ADSORPTION OF 120 PPM ALACHLOR
86
-------
1.00
•I 0.00
120
180 240 300
Reduced Time, tr (bed volumes)
360
420
480
0.10-
o
X
Parameten for Generated Curve:
0.08-
tr1 = 1.60 bed vol.
tr2 = 375. bed vol.
XQ = 0.022 wt. fr.
q0 = 0.195g/g
qm = 0.203 g/g
K= 1910cm3/g
c
a>
o
c
o
o
0.05-
? 0.03
0.00
Desorption Curve
60
120 180 240 300
Reduced Time.tr (bed volumes)
360
420
480
FIGURE VI-13C REGENERATION AND DESORPTION CURVES AFTER TEN-DAY
ADSORPTION OF 120 PPM ALACHLOR
87
-------
VII. PROCESS DEVELOPMENT STUDIES
The process development studies consisted of a number of experi-
ments aimed at gathering the data required to simulate the use of SCF
regeneration in commercial practice and to obtain the necessary data
for scale-up and projection of plant-scale costs..
After the initial screening phase, concerted effort was directed
to adsorption and regeneration tests using alachlor, the pesticide
which was selected for further study; specifically, the test program
in this phase covered:
1. Examining the effluent quality by the initial breakthrough
behavior of SCF-regenerated GAC;
2. Conducting closed-loop regeneration tests with recycled
SCF C02;
3. Conducting larger scale adsorption/regeneration tests in
a 4-ft.-long column containing 380 g GAC; and
4. Modeling the dynamics of the desorption process so as to
develop a rationale for projecting capital and operating
costs for different sets of regeneration conditions.
As alachlor was thought to be representative of chemicals that are
somewhat difficult to regenerate, process development studies were also
conducted with phenol, which we believe may be representative of chemi-
cals that lie in the middle of the spectrum of ease of regeneration.
The phenol process development studies consisted of the following:
1. Conducting larger scale adsorption/regeneration tests in
a 4 ft. long column containing 380 g GAC; and
2. Modeling the dynamics of the desorption process.
A. ALACHLOR REGENERATION
1. Effluent Quality Tests; Initial Breakthrough Behavior
During the earlier screening phase, the behavior of the total
breakthrough curve had been primarily used to characterize the effi-
ciency of regeneration and the capacity recovery during subsequent ad-
sorption. The term "total" connotes the entire effluent concen-
tration-vs-time curve from 0 to the end of the test, where the end
point was chosen when the effluent concentration had reached about 85%
of the influent; for example, in Figure V-14, which shows a series of
adsorption breakthrough curves for Alachlor, the effluent concentration
at the end of A1-1-A1 is 105 ppm. Although the behavior of the total
curve was a reasonable consideration for early regeneration tests in
88
-------
the screening phase, operation of a commercial or municipal adsorber
is usually based upon considerations of an effluent quality limit with
"breakthrough" defined as an effluent concentration greater than the
allowable limit. For Alachlor, using the source data in EPA 440/1-75/
060d, Development Document for Interim Final Effluent Limitations Guide-
lines for the Pesticide Chemicals Manufacturing Point Source Category
(November 1976),an allowable discharge stream concentration of 0 22 ppm
was calculated from the following criteria:
Effluent Limitations Guideline = 0.00705 kg/1000 kg of product
Wastewater Flow = 3,840/gal/lOOO 1b. of product
and based upon the previous information,
Allowable Concentration ,of Alachlor in.Discharge Stream =
[0.00705/(3840 x 8.3 lb/gal)] x 106 = 0.22 ppm
In order to investigate the ability to achieve the effluent quality
with GAC regenerated with C02» adsorption breakthrough tests were made at
a superficial velocity of about 1 gpm/ft (rather than the 7.5 gpm ft^
value tested in the previous rapid exhaustion rate runs); the 11" x 3/8"
ID column holding 7 g GAC was used in these tests.
Adsorption tests were continued for a total length of time equal to
at least twice the time to breakthrough, i.e., if the effluent reached
0.22 ppm in, say, 16 hours, the adsorption was continued for at least
another 16 hours or longer. For the analysis of very low concentration
levels, 100 mm path-length flow cells were used in the double-beam UV
spectrophotometer which was operated at a wavelength of 225 nm, using
the long path cells the resulting accuracy in concentration measurement
was found to be about 0.02 ppm.
The initial portion of the breakthrough curves are given in
Figure VII-]. The figure shows that both on virgin and on regenerated
GAC, a 118 ppm influent Alachlor solution can be lowered to below the
allowable 0.22 ppm level. Breakthrough curves for two virgin GAC
adsorption tests and for two regenerated GAC tests which were carried
out at 1.1 and 7.5 gpm/ft2 respectively, are plotted in Figure VH-2 .
(as a function of volume throughput, rather than as a function of time)
in order to compare the breakthrough curves at the two respective flow
rates. As is seen in Figure vil-2,the effect of the higher flow rate
is a faster (and in fact, an immediate) breakthrough, in agreement, of
course, with all literature data on dynamic adsorption. The 0.22 ppm
breakthrough,which is indicated by an arrow at 2.6 liters of solution,
equates to about 350 bed volumes of wastewater passed.
89
-------
130
120 - Influent Concentration - 118 ppm
110 -
100 -
90
BO
70
60
£ 50
a. 40
I
20
10
0.5 -
GAC - 7.0g
28
ALACHLOR ADSORPTION BREAKTHROUGH CURVES
(Effluent Quality Series)
32
AI-2 Series - Rapid Loading
Al—4 Series - Slow Loading
Breakthrough at 0.22 ppm, 2.6 liters
345
Volume of Solution (liters)
FIGURE VI1-2 ALACHLOR ADSORPTION BREAKTHROUGH CURVES
COMPARISON OF SLOW AND RAPID LOADING
90
-------
Two series of adsorption/regeneration tests comparing the behavior
of F-300 and F-400 GAC were made. The effluent quality portions of the
breakthrough curves are shown in Figures VII-3 and VII-4 respectively,
and the data show that the time to reach breakthrough is greater for the
F-400 GAC, both on virgin GAC and on SCF-regenerated material. The F-300
material was screened to +20 mesh particle size, wheras the F-400 GAC was
used directly for the adsorption tests. For informational purposes, the
size characteristics of the two commercial grades are compared in Table VII-1;
the differences in the breakthrough characteristics of F-300 (+20 mesh)
and F-400 in Figures VII-3 and VII-4 then, are probably attributable to
the particle size differences in the two GAC materials.
Two effluent quality tests were next made with F-300 GAC of two
different particle size distributions; the +20 mesh cut, and a small
size range of -12+30 mesh screened from the commercial grade. The ef-
fluent quality portion of the respective breakthrough curves and the
total breakthrought curves for the two mesh sizes are compared in
Figure VII-5 On the basis of the effluent quality portion of the break-
through curve, the -12+30 mesh size demonstrated a factor of three im-
provement in capacity at breakthrough (76 hrs-22 hrs)/22 hrs), compared
to the capacity of *20 mesh GAC. Smaller mesh GAC results in sharper
breakthrough curves. For data evaluation purposes, a reasonably sharp
breakthrough is an optimum situation for studying effects of SCF regen-
eration on effluent quality,and thus it was concluded that subsequent
tests with all pesticides and with phenol were to be carried out using
only the smaller particle size fraction, -12+30 mesh, to be obtained by
screening standard F-300.
2. Closed-Loop Regeneration Tests of Alachlor-Loaded GAC
The adsorption capacity recovery data that has been reported in
previous sections was the result of regeneration of spent GAC carried
out with SCF C02 in a once-through mode; i.e., C02 from a cylinder was
passed through the column wherein dissolution of pesticide from the GAC
occurred, the C02 expanded to 1 atm to precipitate collect the pesticide,
and the C02 exhausted to the atmosphere.
A closed-loop regeneration train was assembled in order to
assess the capacity recovery of spent GAC after contact with recycled
C02- Preparatory to the start of the closed-loop regeneration series,
the efficiency of precipitating and separating pesticide from a stream
expanded to some intermediate pressure level that would be operative
in a pilot plant or full-scale facility was determined, and along with
the separation efficiency tests a more complete "solubility map" for
Alachlor was determined so that the pressure level for operating the
separator in the closed-loop series could be selected.
Figure VII-6 gives extended solubility data, for Alachlor cover-
ing a temperature range from 50°C to 120°C and a pressure range from
1300 to 4000 psi. It is seen that at the 1300 psi pressure level the
solubility behavior of Alachlor gives a minimum in solubility at about
75°C, somewhat similar to that reported for naphthalene at 80-100 atm
shown in Figure IV-6.
91
-------
10
ro
Q.
a
I
<§
c
to
LU
130
120
110
100
90
80
70
60
50
40
30
20
10
2
1.5
1
0.5
Influent Concentration - 118 ppm
(A1-4-A1
IA1-4-A2
GAC - 7.0 g
Flow - 1.1 gpm/ft2
Breakthrough at 0.22 ppm
12 16 20
Time (Hours)
24
28
32
FIGURE VII-3 ALACHLOR ADSORPTION BREAKTHROUGH CURVES WITH F-300
(Effluent Quality Series)
-------
FIGURE VII - 4
ALACHLOR ADSORPTION BREAKTHROUGH CURVES F-400 SERIES
130
120 -
110 -
100 -
90
80
70
60
50
40
30
20
10
o
o
c
0)
3
fit
Si 2.0
1.5
1.0
0.5
Influent Concentration
• A1-5-A1
Q A1-5-A3
• A1-5-A4
GAC - 7.00 g F-400
Flow-1.1 gpm/ft2
Breakthrough
@ 0.22 ppm
12
16
Time (Hours)
20
24
28
32
-------
TABLE VII-1
CHARACTERISTICS OF CALGON ACTIVATED CARBONS
TYPE NORMAL MESH SIZE SURFACE AREA
m2/g
F-300 + 20 950 - 1050
F-400 -12 + 40 1050 - 1200
94
-------
FIGURE VII-5
ALACHLOR ADSORPTION BREAKTHROUGH CURVES EFFECT OF GAC MESH SIZE ON EFFLUENT QUALITY
en
6.0
Feed -120ppm
GAC - 7.0 g
Flow - 1.1 gpm/ft2
Breakthrough in 72 hours
Breakthrough at 0.22 ppm in 14 hours
16
32 40
Time (Hours)
-------
FIGURE VII - 6
T
10.0
1.0
Solubility
in CO2,
Wt. %
x
_\ ^ _25Mpsi
0.1
o.oi
SOLUBILITY OF ALACHLOR IN CO2
1300 psi
50 60 70
80 90 100
Temperatures, °C
110 120
96
-------
The pressure level of 1300 psi was selected for closed loop separa-
tor operation and for preliminary separation tests. For these tests
the apparatus shown schematically in Figure VII-7 was used. To the high
pressure solubility set-up previously shown in Figure V-l there was added
another pressure vessel to act as a separator and a back-pressure regu-
lator as indicated. In the determination of separator efficiency the
extraction vessel was loaded with Alachlor, and operation of the system
to test separation at intermediate pressure was carried out in the follow-
ing manner: high pressure C02 was passed through the Alachlor, the
stream expanded to an intermediate level of 1300 psi as controlled by
the back-pressure regulator, and the Alachlor which precipitated at that
pressure was collected in the separator.. The 1300-psi stream was then
expanded to 1 atm across the back-pressure regulator and the Alachlor
remaining in the C02 flow-measured as before with the rotameter and Dry
Test Meter. In another test preparatory to closed loop regeneration, a
column of Alachlor-loaded GAC was regenerated in the same system shown
in Figure VII-8; a material balance on Alachlor removed from the GAC
and that collected in both filters was calculated. The pertinent data
are shown in Table VII-2 and agreement is reasonably good, the material
balances closing within 90%.
The closed-loop regeneration^ J;ests were carried out on a 7.7 g
column which was loaded using a synthetic 120 ppm Alachlor solution;
adsorption breakthrough cruves were obtained in the standard system
previously shown in Figure V-4 using on-line UV analysis. Regeneration
was carried out in the closed-loop system shown schematically in Figure VII-8
In this system the C02 stream leaving the back-pressure regulator is sent
to the inlet side of the compressor, recompressed to 4000 psi, and returned
to the GAC column.
<
The complete set of adsorption breakthrough curves for 7 cycles
are shown in Figure VII-9~ As before, there is a first-cycle capacity
loss, but for the second through seventh cycles the total capacity is
essentially constant. Figure VII-10 gives the effluent quality portions
of the same breakthrough curves on a magnified ordinate scale; loading
data are given in Table VI1-3. The breakthrough appears to fall into two
groupings, but unrelated to the progression of cycles. For example, the
second, fourth, and last cycles fall together indicating that there was
no monotonic trend in breakthrough behavior. It should be noted that
the method of plotting the data in Figure VII-10 magnifies any experimental
error; the same data plotted in Figure VII-9 previously showed that
cycles 2 through 7 coincided reasonably well.
3. 4-Ft. Column Adsorption/Regeneration
As stated previously, the rationale for using small columns
and a high feed flow rate was to obtain rapid loading of GAC in a short
time so that many cycles of adsorption/regeneration could be logged on
a single charge of GAC. Subsequently, it was found that low flow rate
tests on the small columns could provide reasonable breakthrough behavior,
but it was desirable to test the finding with a larger scale column. A
l-l/4"lD x 4-ft column adsorption/regeneration system for Alachlor was
built for this purpose.
97
-------
Regenerant Pump
CO
Controlled
Temperature
Separator
Pressure
Flow Valve Control
Separator (Filter)
.Controlled Temperature
GAC Column
(or Alachlor reservoir)
Controlled Temperature
Back Pressure Regulator
FIGURE VII-7 SCHEMATIC DIAGRAM OF DESORPTION AND SEPARATION
-------
C02 Pump
Desorption
Pressure
_ |
Separator (Filter)
Controlled Temperature
_J
GAC Column
Controlled Temperature
-Back Pressure Regulator
Fiqure VII-8 SCHEMATIC DIAGRAM OF RECYCLE TEST
-------
TABLE VII-2
ALACHLOR DESORPTION MATERIAL BALANCE DATA
ALACHLOR ALACHLOR COLLECTED IN FILTERS
REMOVED Intermediate Pressure Low Pressure
0.58 g 0.33 g 0.19 g
100
-------
FIGURE VII-9
Time (days)
COMPARISON OF COMPLETE BREAKTHROUGH CURVES
FOR ALACHLOR ADSORPTION CLOSED LOOP SERIES
-------
Feed - 120 ppm
o
ro
o.
a.
c
o
o
o
c
,XAXAX
>X
A1-CL-7
A1-CL-A6
A1-CL-A5
A1-CL-A4
A1-CL-A3
A1-CL-A2
A1-CL-A1
16
24
32
40
48
72
Time (Hours)
FIGURE VII-10 ADSOPRTION FROM SYNTHETIC ALACHLOR CLOSED LOOP REGENERATION SERIES
-------
TABLE VII-3
ALACHLOR LOADING AND EFFLUENT QUALITY
RESULTS AFTER CLOSED-LOOP REGENERATION
Loading After
Cycle # Adsorption
Residual Loading
After Regeneration
Initial
Effluent
Concentration
— *J
1
2
3
4
5
0.38 g/g
0.39 g/g
0.40 g/g
0.40 g/g
0.39 g/g
0.12 g/g
0.16 g/g
0.14 g/g
0.15 g/g
^ 0
0.1 ppm
0.4 ppm
0.3 ppm
0.3 ppm
103
-------
An adsorption test using 120 ppm Alachlor was carried out using
the 4-ft. column system, the adsorption continuing until the effluent
concentration was essentially equivalent to influent. In order to dup-
licate the previous fully-loaded tests on small columns, the flow was
continued for five days longer. The breakthrough curve for the adsorption
is shown in Figure VII-11 and integration of the curve gives an Alachlor
loading value of 0.50 g/g GAC. The 4-ft. column was regenerated at the
same conditions used for previous Alachlor tests, 275 atm and 120°C. The
same SCF supercritical velocity was used for the regeneration, and the
desorption curve was obtained described earlier. The desorption curve
is shown in Figure VII-12 and the previous desorption curve for a small
column is reproduced for comparison; the curves are normalized by the
ratio of the GAC charge in the columns, i.e., 380/7. As is evident, the
curves are nearly coincident.
4- Modeling Alachlor Desorption; Local Equilibrium Theory
Determining the optimum conditions for regenerating a column by
trial-and-error experimentation is a costly, time-consuming process.
For each desorption run, the column must be loaded by adsorption prior
to desorption. Although desorption with a supercritical regenerant is
a relatively fast process, the adsorption step is slow, especially when
one wants to attain fully-loaded conditions.
The time required to saturate the adsorbent with solute can be
determined either by batch-adsorption testing or by flow testing a
column on-line. Batch-testing is easier to do experimentally, but the
time required to reach saturation in a batch test has been found to be
shorter than that for flow testing. In other words, the batch test
vies a lower limit to the time required to saturate a column on-line.
The rate of adsorption of Alachlor from an aqueous solution was
determined by batch adsorption tests. The results, shown in Section IV,
indicate that more than seven days are required to closely approach
equilibrium coverage. In column adsorption flow tests, the time to
load the column to near equilibrium coverage is significantly longer:
for 7 g-carbon columns (20 cm length by 1 cm i.d.), over ten days were
required, while for 380 g-carbon columns (4 ft length by 1.25 inch i.d.)
over forty days were required to reach saturation.
Clearly, the number of experimental desorption runs that can be
conducted within a given time frame is limited by the slow kinetics
of adsorption; the larger the scale, the more severe is the time-constraint.
In order to minimize the number of runs required and to develop a rationale
for scale-up, an effort was undertaken to develop a theory-based model of
the desorption process. With such a theory in hand, one can develop
optimization strategies from small-scale tests, extrapolate these results
to larger scales, and then perform a small number of large-scale tests to
verify the model predictions.
104
-------
o
en
110
100
90
I 80
I 70
8 60
O
O
I 50
LU
40
Feed Concentration 120 ppm
Column 11/«" x 4'
GAC - 380 grams
Flow Rate - 1.1 gpm/ft2
30
20
10
-
0.22 ppm breakthrough 0.5 ppm
i
.»•••. •••!••• ••••••!•••• ••••••!•••• • •••••!• • • .
•
•
• •
an day 22 •
•
• i
1 •
10
15 20
Time (days)
25
30
35
40
FIGURE VII-11 ADSORPTION FROM SYNTHETIC ALACHLOR SOLUTION IN FOUR FT COLUMN
-------
2.5
o
en
(0
l_
0>
c
03
O5
a:
c
o
c
o>
2—7 g columns in series
————4 ft. column (regenerating volume normalized
by GAC ratio 380/14 = 26)
v
0.5
i
-|
I
500
1500
FIGURE VII-12
1000
Regenerant Volume (SL's)
DESORPTION TRACES OF ALACHLOR REGENERATION CONDITIONS, 120°C. 4000 psi
2000
-------
The desportion curves for Alachlor (Figure VI-13, Section VI)
resemble those predicted by the jocal equilibrium theory (LET) model,
which is based on the assumption that equilibrium exists between
adsorbent particles and the adjacent fluid at all points within the
column (Sherwood, et_ al_., 1975). Mass transfer within the SCF phase
is rapid and Alachlor solubility in SCF C0? is not very high; these
two conditions are typical of systems that follow LET behavior. Thus,
an effort was undertaken to correlate the Alachlor desorption dynamics
by the LET model. A complete description of the LET model is given in
Appendix A; a brief outline of the theory along with the significant
results are given here.
For a fixed bed wherein longitudinal diffusion is neglected
and plug flow is assumed, the material balance on fluid and solid
within a differential element is:
e * pB + " ' ° (?)
where c and q are fluid and solid concentrations, respectively, e is the
fraction of fluid-filled space outside the particles, and PB is the
bulk density of the dry adsorbent. The superficial fluid velocity is
ev, where v is the average fluid velocity in the interstices between
particles.
Before Eq. (1) can be solved, a second equation relating fluid
and solid concentrations must be introduced. In the general case, this
second equation will take the form of
P
(U-) = kaF(c,q) (2)
Bv3t
which expresses the rate of change of solid phase concentration as a
function of the interfacial mass transfer coefficient, ka, and a driving
force, F (c,q). Equations (1) and (2) can then be solved simultaneously
to obtain the function c (z,t), which is the fluid phase concentration at
any position, z, within column as a function of time, t. For example,
the effluent concentration curve is c(L, t), where L is the column length.
In tho general case, there are two typos of MUSS l.r.insfcr re-
sistances that are considered in developing Tq. (?): diffusion ol
solute out of the SCF-filled pores and interfacial mass transfer from
the external surface of the adsorbent particle into the bulk of the
SCF phase. One of the advantages of a SCF regenerant is that mass trans-
fer is relatively rapid within the SCF phase. In the limiting case where
resistance to mass transfer is negligible, Eq. (2) reduces to the equili-
brium relationship between solid phase concentration, q, and bulk fluid
concentration, c, which is just the adsorption isotherm expression:
q = f(c) (3)
107
-------
For this limiting case, local equilibrium exists at all points within
the column and at all times between particles and the adjacent fluid.
This, then, is the basis of the local equilibrium theory.
Note that Eq. (3) is the isotherm between GAC and the regenerant,
which is SCF C02- To avoid confusion with the conventional water-
6AC isotherm, the SCF-GAC isotherm will be denoted SCF-isotherm. Also
note that Eq. (3) applies to the solute that is desorbed and, hence, it
represents only the reversibly-held portion of the solute.
Although we attempted to measure several points on the SCF-isotherm,
the data obtained did not cover a wide enough range of concentrations
to accurately define the SCF-isotherm (see Appendix A for details).
As an alternative, it was assumed that the SCF-isotherm could be
described by the Langmuir expression.
where the two contants, qm and K, were varied to define a set which
best fit the desorption and regeneration curves generated by the LET
model .
Using a single set of values of qm and K, desorption and regeneration
curves were calculated from the LET model for 10, 3, and 1-day adsorption
of Alachlor. The experimentally measured desorption curves were previously
given in Fig. VI-13, along with the smooth curve generaged by the LET
model. Also shown were the experimental and computed regeneration curves,
which are the integrals of the desorption curves.
For the 10-day adsorption experiment, the results of Fig. VI-13C
show that the model describes the experimental data moderately well up
to 200 bed volumes. Beyond that point, the generated curve goes to zero
concentration faster than the experimental desorption trace represents
a departure from local equilibrium theory. The regeneration curve
accentuates the departure of theory and experiment. It may indicated
that pore diffusion out of the particles becomes rate-limiting after a
large fraction of the solute is desorbed. The results for the 3- and
1 -day adsorption experiments, as shown in Fig. VI-13B and VI-13A
agree fairly well with LET-generated curves. Taken as a whole, it
appears that the LET model is appropriate for solutes of the Alachlor
class.
108
-------
B. PHENOL REGENERATION
1- Scale-up; Comparison of Small and Large Columns
To gain insight into the dynamics of the desorption process
ai?dut™Jevelop Vat1onale for scale-up, several runs were conducted
with 380-g (4-ft) columns under conditions similar to those used for
7-g (11-in) columns. Phenol at 2500 ppm was used with F-300. In most
cases, the columns were dried prior to desorption. Desorption was con-
ducted at 55°C. The rate of desorption was followed by collecting
solute over intervals of time in traps downstream of the pressure let-
down valve.
Desorption curves for three 7-g columns on the first regeneration
cycle are shown in Fig. VII-13. The total solute collected was 10-20%
less than the weight loss of the column; for PB-RI, 0.13 g/g was collected,
whereas the column weight loss upon regeneration was 0.16 g/g. Although
the desorption curves for the three columns shown in Fig. VII-13 are
similar, it was difficult to obtain better accuracy with this procedure
because the desorption process is so rapid. The regeneration curves,
which are the integrals of the desorption curves, show less scatter
(see Fig. VI-13).
The results for regeneration of three 380-g columns are summarized
in Table VII-4. Columns Pl-48 and P2-48 were adsorbed over moderately
prolonged periods and desorbed at 150 atm; these conditions are similar
to those of column PB, the results for which were given in Table VI-2
(Section VI). There is relatively good overall agreement between the
small and large column behavior.
the regeneration curves for the small and large columns are shown
in Fig. VII-14. The difference between the curves for the small and
large columns at 150 atm regeneration pressure is related to two factors:
the small column curve is lower, in part, due to inefficient collection
of solutes in the small column traps. In addition, the large columns
contained more transfer units, which tends to enhance the rate of re-
generation when plotted on a normalized basis, as in Fig. VII-14.
Column P2-48, which was regenerated at 275 atm, exhibited a faster
desorption than the large columns regenerated at 150 atm. This behavior
is typical of the effect of pressure on regeneration. In general,
higher pressure increases solubility in the supercritical carrier. Since
the activity of the adsorbed solute is not appreciably affected by the
SCF pressure, the net effect of increasing pressure is to shift the
equilibrium adsorption isotherm to lower loading at a given fluid-phase
concentration. The desorption rate is quite sensitive to the adsorption
equilibrium because mass transfer in the SCF carrier is very rapid. As
will be discussed below, the number of transfer units in the 380-g columns
is large enough so that the desorption curve approaches that predicted by
local equilibrium theory.
109
-------
o
03
JZ
Q.
c
O
'*J
to
o
O
O
4-1
C
0)
—• UU
O
_h
.3
.2
- L,
n
.1
-11"
> * **
1
\
__ __
J.l
Tf
1
PB-R1
P5-R1
P6-R1
I U
Total Solute
Collected (g/g)
0.13
0.15
0.16
0
20
40
60 80 100
Throughput (g C02/g GAC)
120
140
160
Figure VII-13
REPRODUCIBILITY OF FIRST CYCLE DESORPTION CURVES;
2500 PPM PHENOL ON 7 g GAC, OESORBED AT 55°C, 15 MPa
-------
Table VI1-4. REGENERATION OF LARGE COLUMNS
2500 ppm Phenol on F-300 Loadings determined by i
of breakthrough curves; desorption at 55"C
Column
No.
PI -48
P3-48
P2-48
Desorption
Pressure Cycle
(MPa) No.
15.0 1
2
15.0 1
2
3
27.5 1
2
3
Adsorption
Time
(h)
63
24
52
67
51
73
56
48
ntegration
Solute
Adsorbed
in ith
Cycle (g/g)
0.26
0.25
0.26
0.24
0.19
0.26
0.19
0.19
Solute
Desorbed
in it(l
Cycle (g/g)
0.17
-wet-
0.16
0.12
0.12
0.15
0.13
0.13
-------
.20
ro
O)
.15
a
3? .10
0>
1
s
& . .05
380-g, 27.5
370-g Columns
A P1-48-R1
O P3-48-R1
03 P2-48-R1
Pressure
(MPa)
••VHMBHB
15.0
15.0
27.5
20
40
60
80
100
120
140
Throughput (g C02/g GAC)
Fig. VII-14 REGENERATION CURVES FOR SMALL AND LARGE COLUMNS.
2500 ppm PHENOL ON F-300; DESORPTION AT 55° C.
Solute
Collected
(g/g)
0.17
0.16
0.15
160
-------
Column P3-48 underwent three adsorption/desorption cycles; as
shown in Table VII-4 , the amount desorbed in the second and third
cycles was 0.12 g/g. The regeneration curves for the three cycles are
shown in Fig. VII-15- Within the accuracy of the experiment, the
desorption curves for the second and third cycles are in good agreement.
Note that the regeneration curves for the second and third cycles are
shifted down from that of the first cycle. In other words, the fraction
of the reversibly adsorbed species is relatively constant with through-
put. This results suggests that increased quantities of residual that
are present in the second and third cycles do not interfere with the
rate of desorption of the reversible species. This result, if it were
shown to be true in general, would have broad implications in experimental
testing of the applicability of SCF regeneration for other solutes. In
the phenol studies we have conducted to date, it appears that the quantity
of reversibly-held solute does not change appreciably after the first
regeneration (see Figs. VI-3 and VI-7 ). Since the reversibly-held
solute represents the working capacity for SCF regeneration, it is
possible to determine the dynamics at the working capacity by
measuring the desorption and regeneration curves on the second or
third cycles. In other words, there would be no need for conducting
a large number of repetitive adsorption/desorption cycles.
2. Modeling the Desorption Process; the Thomas Solution for
Fixed-bed Sorbers
As described in the preceeding section, the desorption and re-
generation curves for 7-g and 380-g phenol columns are not identical
when plotted on a normalized basis (see Fig. VII-14). If the local equili-
brium theory were appropriate for phenol desorption, then the normal-
ized curves for large and small columns should be superimposed on one
another. The fact that they are not identical implies that mass trans-
fer cannot be neglected in phenol desorption.
For mass transfer in packed beds, the Nusselt number can be expressed
as a function of Reynolds number, Re (Gupta and.Thodos, 1962):
Jn - X C-0100
J
- — L-UIUU - n co J
U e Re - 0 483
where e is the bed void fraction.
For pore diffusion, the mass transfer coefficient of the solid
phase is given by (Sherwood, et_ al_., 1975):
16.7Dn (9)
l> ^™ \j \ /
P clp
where D and d are the particle diffusivity and diameter, respectively.
ThetonRer is ixpressed as a function of the Knudsen diffusivity, Dk,
and the solute diffusivity in regenerant fluid, Df:
1 - I. fl + 1 ) (10)
D Y VD, D-:
un A k f
-------
o
CD
a
en
o
o
0)
^-»
D
Total Solute
Collected (g/g)
O P3-48-R1 0.16
A P3-48-R2
V P3-48-R3
80 100
Throughput (g C02/gGAC)
Fi gure V11 -15 REGENERATION CURVES FOR SUCCESSIVE CYCLES;
2500 PPM PHENOL ON 380 G GAC DESORBED AT 55°C, 1 ATM
-------
where T and x are the tortuosity factor and void fraction within the
particle, respectively. The Knudsen diffusivity is a function of par-
ticle pore size and mean free path in the fluid-filled pores. It is
commonly evaluated by the following correlation (Sherwood, et al_. , 1975):
~~
D =
Dk SgppvM (ID
where S is adsorbent surface area; PD, particle density; T, temperature;
and M, solute molecular weight.
Taken together, Eqs. (7) through (11) provide a means for
calculating kf and k which, in turn, allow us to determine K by Eq. (6).
Having evaluated K, the number of transfer units, N, can be calculated
from:
(12)
where L is the bed length and U the superficial velocity.
The Thomas solution provides a prediction of the desorption
curve in the following form:
X = f(N, R, T) (13)
where X is dimensionless outlet concentration (c/c0 for adsorption and
l-c/c0 for desorption); and T, dimensionless SCF throughput (UCpt/q0pnL).
The parameter, R,_Js related to the adsorption equilibrium constant:
where PB is adsorbent bulk density; a, solid interfacial area per unit
volume of bed; q, adsorbent loading; t, time; and c, fluid concentra-
tion. When used to model a fixed-bed desorption process, qm is the
maximum coverage corresponding to a monolayer. C0 is the fluid concentra-
tion at equilibrium with the initial adsorbent loading; and K, the ad-
sorption equilibrium constant. Note that the appropriate adsorption
equilibrium is that of reversibly held solute (i.e., the fraction of the
solute that desorbs), partitioning between adsorbent and supercritical
fluid. Although total GAC adsorption is usually correlated with Freund-
lich isotherms (see, e.g.,Giusti , et. el. , 1974), we have seen that the
Langmuir isotherm is an adequate model of reversible solute isotherms
in SCF C02 (see Section V-l). Thus, qm and K were treated as constants
in the Langmuir expression:
115
-------
When mass transfer is not negligible, then an appropriate model
of column dynamics should provide for desorption at the surface within
the pores, unsteady-state diffusion in the pore volume and/or along
the pore walls, and mass transfer to the bulk fluid outside of the
particle (Vermeulen, ejt aK , 1973). An approximate solution, based on
the mdethod of Thomas, is usually employed for the design of fixed bed
adsorbers (Sherwood, et_ aj_. , 1975). The Thomas solution, which was
originally developed for ion exchange, is based on a kinetic driving
force of the form:
PB If •
«•
(5)
The overall coefficient, K in Eq. (5), is evaluated from (Sherwood,
et a!., 1975):
1
K
1
1
k,
Co/qmPB
(6)
where k^ and kp are fluid and solid phase mass transfer coefficients,
respectively, b is a correction factor to account for the fact that
the solid and fluid phase resistances are not strictly additive. The
value of b usually lies between 1 and 2.
Values of kf, kp, b and, thence, K can be evaluated by standard
procedures from physical properties of the adsorben and SCF carrier
along with the fluid flow conditions in the bed. The fluid phase mass
transfer coefficient can be expressed as a function of the Nusselt number,
J (Sherwood, ejt al_., 1975):
kf = jDu/sc
2/3
(7)
where U is the superficial velocity and S is the Schmidt number.
R = i +KCO
One form of the solutions to Eq. (13) is given in Fig. VI1-16 (from
Vermuelen, et al., 1973).
Lacking sufficient data to accurately fit a Langmuir adsorption
expression for the reversible adsorbate-SCF C02 system, we have attempted
to use an experimental desorption curve to back-calculate R. The desorp-
tion curve for column P3-48, shown in Fig. VII-17, was used to define
the smooth curve given by the dashed line which, in turn, was put in
dimensionless form as X vs. T. Points from this curve were then plotted
on Fig. VII-16. From these results, it appears that R is the order
of 2 for phenol desorption at 55°C and 150 atm. This procedure is not
accurate enough to provide a good estimate of the number of transfer
units. The central portion of the desorption curve (which is probably
the most accurate) suggests that N is greater than 100 for the 380-g
column. An independent estimate of N was calculated from EQ. (12) using
the procedure outline above (see Appendix B). The value calculated for
the 380-g column was between 600 and 1200; the corresponding range of
N for the 7-g columns (2500 ppm phenol desorbed at 55°C and 150 atm) was
80 to 160 transfer units.
116
-------
XdtR =
XatR = IO
XatR = IOO
O.I
0.5 I 2
Throughput parameter, Z
Fianre VTT-16 BREAKTHROUGH HISTORIES AT R=2
10 AND 100 (from VERMUELEN et. at.. 1973)
X = c/c0 for adsorption and
T - C/CQ for desorption
117
-------
.5
.4
P3-48-R1
Total Solute Collected = 0.16g/g
o
a)
£
.3
\
\
00
(0
I
I
O
-------
The family of curves given in Fig. VII-17 illustrates how the de-
sorption curve varies with number of transfer units, N, and strength of
adsorption, as measured by R. For R = 2, which is representative of
relatively weak adsorption, the family of desorption curves get pro-
gressively sharper as N increases; for small N, the change in sharpness
with increasing N is rapid, while for large N, the change with increas-
ing N is less severe. To better illustrate this point, the data of
Fig. VII-16 were replotted as the normal desorption and regeneration
curves. The curves for R = 2 and N = 10, 100, and «are given in Figs.
VII-18 and VII-19. The progression of curves with increasing N is
what the model would predict when increasing column length at constant
superficial velocity, (i.e., constant U and K). These predictions
resemble the experimental results of 7- and 380-g phenol columns,
as shown in Fig. VII-18. Mote that the experimental curves in Fig. VII-19
were not normalized as in Fig. VII-18; thus, the difference between
the 7- and 380-g curves at 150 atm is accentuated.
For R = 2, the difference between N of 100 and is relatively small;
that is, the desorption curve is relatively insensitive to the flow
conditions within the column. This behavior is indicative of a system
wherein mass transfer is not a dominant factor. In the limit of
negligible mass transfer resistance (i.e., N -»•<»), the desorption dy-
namics are governed by local equilibrium theory.
The curves for R = 10 and 100 are representative of moderate-to-
strong adsorption. As shown in Fig. VII-16, as the strength of adsorption
increases, the relative shape of the desorption curve is less sensitive
to the number of transfer units; for R = 10, there is little change in
X for N = 1 to°s up to X = .95. The major impact of changing N occurs
in the tail of the desorption curve, for X >.95. For R = 100, a single
desorption curve is adequate for describing cases with N above 10.
The Thomas solution for modeling fixed-bed desorption is a powerful
tool that greatly simplifies the design procedure for SCF regeneration.
The technique has been used to advantage in evaluating alternative
desorber designs in conjunction with estimates of process economics.
Those results are described in Section IX.
119
-------
X
c"
_o
'•M
CO
C
05
o
c
o
o
UJ
0.2 L
0
0
N
oo
100
10
0.4
0.6
0.8
1.0 1.2
Throughput, T
1.4
1.6
1.8 2.0
Figure VII-18 VARIATION OF DESORPTION CURVE WITH
NUMBER OF TRANSFER UNITS (for R = 2)
-------
ro
0
0
OA
0.6
0.8
1.0 1.2
Throughput, T
1.4
1.6
1.8
2.0
Figure VI1-19 VARIATION OF REGENERATION CURVE WITH
NUMBER OF TRANSFER UNITS (for R = 2)
-------
VIII. PLANT WASTEWATER TREATABILITY STUDY
A. INTRODUCTION
The phase of the program dealing with actual pesticide manufacturing
wastewaters used sampled proved by a pesticide manufacturer. This
wastewater came from a process manufacturing the herbicide, atrazine.
Atrazine is a triazine compound, used to control weeds in corn, sorghum
and sugarcane fields. The compound, commercially available for more than
twenty years, is nonflammable, noncorrosive, and highly stable. Atrazine
is relatively nontoxic with an LDso for rats of 3080 mg/kg with no
reported cases of ill effects to humans. The compound is synthesized
by reacting cyanuaric chloride with ethylmaine and isopropylamine, in
the presence of an acid.
A characterization of the wastewater from this process is shown in
Table VIII-1. It can be seen that atrazine contributes only 10% of the
total organic carbon (TOC) present in the wastewater. Other possible
organics include the starting materials used to manufacture the herbicide
as well as other reaction products.
A series of experiments was developed to determine the effectiveness
of supercritical carbon dioxide to regenerate granular activated carbon
columns loaded with this atrazine wastewater. The plan was similar to
that used in the previous synthetic wastewater studies; GAC columns
would be cyclically adsorbed and regenerated at various process con-
ditions to determine the steady-state working capacity of the GAC.
B. EXPERIMENTAL METHODS
1. Adsorption Apparatus
A sketch of the adsorption apparatus used is shown in Figure VIII-1.
For this phase of the program, the predominant GAC used was Calgon F-400,
screened to a particle size of -12+30 mesh. The GAC columns used were
the standard, high-pressure, stainless-steel tubes, (0.45 cm id x 22.9 cm
long). Each column was loaded with 7.00 gr of GAC with plugs of glass
wool in both ends of the column to prevent spillage. The wastewater
feed was first pre-treated to remove large suspended solids by filtering
through a column packed with glass wool. This filtered feed was then
pumped up through the GAC column at a flow of 3.2 ml/min, equivalent
to 1.1 GPM/ft2 of column.
Because of the high concentration and variety of organics in the
wastewater, it was found that the concentration of atrazine could not be
directly measured by UV spectrophotometry. During earlier studies with
pure atrazine dissolved in distilled water, the breakthrough of atrazine
could be measured by passing effluent from a GAC column through a spec-
trophotometer set at 230 nm. In the case of actual wastewater, nonadsorbing
122
-------
TABLE VIII-1
ATRAZINE
CH3CH2NH —yV^ If* NHCH(CH3),
Cl
Total Organic Carbon (TOC) Concentration
Total Wastewater 1000 ppm
Atrazine 100 ppm
Other Possible Components of Wastewater
Sodium Chloride NaCI
Cyanuric Chloride
Ethylamine
Isopropylamine (CH) CH
123
-------
ro Filtered
•^ Feed
GAC
Column
I
uv
Detector
Recorder
Metering
Pump
1
Automatic Sampler
Effluent
FIGURE VIII-1 ATRAZINE WASTE WATER ADSORPTION APPARATUS
-------
species interfered with UV absorbance at this wavelength. UV scans of
effluent samples of the wastewater showed that there was a wavelength,
279 nm, where the UV absorbance value rose steadily and leveled off in
the shape of a characteristic adsorption breakthrough curve. All GAC
adsorptions were monitored at this new wavelength to determine column
performance, but this method did not necessarily trace the adsorption
of the compound atrazine, rather it indicated the adsorption of some
representative organic species in the wastewater. In order to check the
loadings on GAC, two additional methods were used. The first, described
previously, involved taking a GAC column after adsorption, heating it
to 55 C while passing low pressure carbon dioxide over it to dry the GAC,
and then weighing the dried column to determine total weight picked up
during adsorption. While this method gave a final number for total
weight gained during adsorption, it did not give any indication of the
initial breakthrough concentration nor did it provide anything like
an adsorption breakthrough curve. In order to obtain this data, it
was decided to analyze samples of GAC column effluent by means of a
Dohrmann/Envirotech DC-52D Total Organic Carbon Analyzer (TOCA).
2. TOCA Description
In the DC-52D analyzer, a combination of oxidative-reductive pyrolysis,
and a flame ionization detector are used to determine the carbon content
of a solution., The DC-52D also incorporates a carbonate bypass system
which eliminates any response to inorganic carbonates that may be present.
A sample to be analyzed is injected onto a bed of granular manganese
dioxide (MnO?) in a platinum boat as shown in Figure VIII-2. During the
first step of analysis, the boat is heated to^H5 C and all volatiles
species are swept into a bypass column by a flow of helium. This bypass
column selectively traps organics while allowing H20 and C02 to pass
through to a vent. In the second step, the helium gas flow to the bypass
column is reversed, sweeping the volatiles out to a hydrogen rich
reduction zone containing a rhodium catalyst at % 350 C. In this zone,
all the volatiles are reduced to methane gas (CH4) whi'ch is then measured
directly with a flame ionization detector. In the third and final step of
the analysis, the sample boat containing the non-volatile organics is
advanced into a zone ati,850 C where any carbonacous material reacts with
the MnOp granules, releasing carbon dioxide (COo). This C02 is then
reduced to Cfy over the rhodium catalyst and detected by the flame
ionization detector. Gas flows, temperatures, and timing sequences on the
DC-52D are calibrated such that if one injects a 30yl sample of solution
and starts the analysis, the unit will proceed automatically through all
three steps and display the Total Organic Carbon value directly in parts
per million (ppm).
125
-------
H,
ro
Flame
lonization
Detector
200°C
^
Catalyst
Reduction
Zone
350°C
—*
Pyrolysis
Zone
850°C
Vaporize
Zone
115°C
Sample
Inlet
25°C
t M A 1
1 ' He l *
Air
Bypass
Column
60-1 30°C
FIGURE VI11-2 TOTAL ORGANIC CARBON ANALYZER (TOCA)
-------
During a GAC column adsorption, effluent samples would be collected
with a fractionator (Gilson, Model #FC-80K) every 30 or 60 minutes. Each
of these samples would then be analyzed for its TOC concentration. By
plotting these TOC concentrations versus time the sample was taken, one
obtains a "TOC adsorption breakthrough curve" giving both initial break-
through concentrations and the slope of an adsorption breakthrough curve.
An example is shown in Figure VIII-3 .
3. Regeneration Apparatus
The apparatus used to regenerate these GAC columns was identical to
that used for the synthetic wastewaters as is seen in Figure VIII-4 .
Carbon dioxide from high pressure cylinders was compressed at room temper-
ature from a suction pressure of about 1000 psig to an operating pressure
in the range of 2250 to 4000 psig by means of a diaphragm compressor
(Aminco Cat. No. 46-13421). The carbon dioxide was then heated above
its critical temperature, passed down-flow through the GAC column where the
organics were dissolved off the carbon and then expanded to atmospheric
pressure where desorbed material separated from the carbon dioxide phase.
Carbon dioxide flow rates were monitored with a Fisher and Porter^ flow-
meter (FP-1/2-27-G-10/83) and total flow during a regeneration was
measured with a Singer (D7M-200), dry test meter. The extent of re-
generation was determined by direct weighing of the GAC columns before the
after desorption. Initially, the weight of material collected in the
low pressure separator was also used to determine regeneration, but due
to the volatility of some of the organic species present, this was found
not to be effective in measuring regeneration performance.
C. EXPERIMENTAL SERIES
A summary table of all the GAC column work with atrazine wastewater is
shown in Table VIII-2. The following gives a brief description of each
series, its process conditions, and results obtained.
Series ARW-1: This was the first attempt at working with actual
atrazine manufacturing wastewater. The feed for this first
adsorption was filtered through a length of glass tubing packed
with glass wool. After the first regeneration the column was
opened to reveal a brown crust-like solid on the glass wool plugs
and on the GAC surface. This glass wool plug was replaced and
the column readied for a second adsorption. However, for the
second adsorption and all subsequent adsorptions, the feed
solution was vacuum filtered through a Whatmann #42 filter paper.
Even with this additional filtering, series ARW-1 was stopped after
its third adsorption due to plugging in the column from its
exposure to inadequately filtered feed.
127
-------
Ill
_c
"E
Q.
o
ro
CD
1100
1000
900
800
700
600
500
400
300
200
100
Feed Concentration 985 ppm
Column - 7 g GAG
Flow - 1.1 gpm/ft2
_L
10
15
25
30
35
FIGURE VIII-3
20
Time (Hours)
ADSORPTION BREAKTHROUGH CURVE FOR TOC IN ATRAZINE REAL WASTEWATER
-------
J- JL
Carbon Dioxide
Supply
GAC
Column
Flow
Meter
Compressor
LJT
Vent
0
Expansion
Valve
Dry Test
Meter
Collector
FIGURE VIII-4 ATRAZINE WASTEWATER REGENERATION APPARATUS
-------
TABLE VIII-2
COLUMN PARAMETERS
ADSORPTION CONDITIONS AND RESULTS
DESORPTION CONDITIONS AND RESULTS
Column
*
ARW-1
ARW-2
ARK- 3
•*
ARW-4
— ' ARW-6
co
° ARM- 7
ARW-8
ARW-9
ARK- 10
ARM- 11
Weight of
GAC (g)
7.00
H
II
7.00
It
tl
tl
II
7.00
II
II
II
7.00
II
7.00
7.00
7.00
7.00
II
7.00
11
"
"
"
"
7.00
II
II
II
H
It
Type of No. of
GAC Cycles
F-400 2.5
11
"
F-400 5
II
II
II
"
F-400 4
H
"
"
PCB 2
-40+50
1!
F-400 1
F-400 1
F-400 1
F-400 2
II
F-400 5.5
II
tl
"
II
Ii
Fr400 6
"
»
«
H
Hours on
Stream
39
40
40
40
42
40
40
40
40
40
40
40
40
40
40
40
40
40
40
40
40
40
40
40
40
40
40
40
40
40
40
Loading aWt
(g/g GAC)
0.14
0.39
0.19
0.11
0.12
0.05
0.35
0.16
0.15
0.12
0.32
0.06
0.37
0.36
-
.
-
6.29
0.15
0.08
0.04
0.08
0.09
0.32
0.05
0.04
0.07
0.07
0.05
Loading B.T. Curve T
(g TOC/g GAC) (°C)
55
0.11
0.26 135
II
H
II
II
0.35 120
H
H
11
135
(1
135
250
250
250
11
120
tl
II
tl
II
U
- 120
U
II
i
II
"
(psig)
2,250
"
5,000
II
"
II
II
4,000
II
"
"
5,000
"
5,000
2,250
2,250
2,250
Ii
4,000
Ii
M
H
II
II
2,250
11
11
1
11
"
Total Flow
(SL)
7,875
11,830
8,350
7,050
5,900
5.250
4,870
8,400
7,650
2,420
9,680
7,900
12,150
_
750
2,500
3,230
-
1.800
-
1,500
-
.
-
_
2,000
Washed Off Desorbed
(g/g GAC) (g/g GAC)
0.09
0.14
0.12
0.07
0.09
0,07 0.07
0.14
0.11
0.09
0.09
0.09
0.04
0.16
0.11
-
_
-
0.11
0.06
0.06
0.02
0.06
-
0.04
0.06
0.04
0.05
0.05
0.03
Residual
(g/g GAC)
0.32
0.05
0.25
0.07
0.04
0.03
-0.09
0.21
0.05
0.06
0.03
0.23
0.02
0.21
0.25
0.22
0.24
0.02
0.18
0.09
0.02
0.02
0.02
-
0.28
-0.01
0.00
0.02
-0.01
0.02
-------
Series ARW-2 and ARW-3: These two series were run simultaneously,
ARW-2 at 135°C and 5000 psig, ARW-3 at the somewhat milder con-
ditions of 120°C and 4000 psig. Both of these series used feed
solution filtered with Whatmann filter paper, which aided in
removing suspended solids. Column ARW-2 was opened after three
cycles and no brown material was seen on the GAC or the glass wool.
Despite this removal of suspended solids, the loading capacity of
these two columns was less than anticipated. About 60% of the
material adsorbed in the first cycle was remaining on the GAC
after regeneration. An additional 30-40% of the material adsorbed
during subsequent adsorption was remaining on the GAC with each
additional cycle. These findings were not anticipated from
previous work with synthetic solutions. During the fifth adsorption
of column ARW-2, severe plugging and pressure drops were found in
the column, even though the feed had been pre-filtered. In an
effort to remove this plugging material, the column was backwashed
with 150 ml of water heated to 55°C. This backwashing step was
found to remove a significant amount of material from the column
(0.07 gr/gr GAC). The backwashed column was then dried and re-
generated, where a normal amount of material (0.07 g/gr GAC)
was desorbed.
Series ARW-4: In order to determine the effect of type of carbon
on steady-state capacity and residual loading, a column was pre-
pared using a narrow mesh, vapor phase carbon (Calgon PCB, -40+50
mesh). This series was only carried out to two cycles because it
was seen that the working capacity of the carbon quickly dropped
lower than any value seen for liquid phase carbons.
In order to more fully understand the plugging problem, GAC column
studies were temporarily put aside so that more effort could be put into
characterizing the wastewater itself. It was first suggested that this
solid material forming plugs in the GAC columns was water soluble species
that were deposited on the glass wool and GAC during the drying of a GAC
column. The general procedure in treating GAC columns was to dry them
at 55°C after adsorption so a determination of total weight pick-up could
be made. Any dissolved solids would be left on the glass wool and GAC
when the water evaporated. To check the nature of these solids, an
1800 ml sample of feed solution was vacuum evaporated at room temperature.
Some 40 grams of solid material was collected from this sample and tested
for solubility. Measurements of solubility in water and in an organic
solvent (acetone) revealed that about half of the solids would dissojve
in each solvent. When exposed to supercritical carbon dioxide, (135 C,
5000 psig) the solids were found to have only a 0.00635 solubility. This
data led to a hypothesis of an SCF-insoluble organic/inorganic mixture
131
-------
of solids. While further tests were being proposed to identify this
mixture, information on the characteristics of the atrazine wastewater
was received from the manufacturer. This data was collected during
research conducted at Research Triangle Institute, Research Triange
Park, N.C., and was forwarded by the herbicide manufacturer to
ADL. The data included two important pieces of information. First, the
wastewater contained-\, 8500 ppm of chloride, suggesting inorganic chloride
salts. This data confirmed a literature search which showed that in
some alternative processes to manufacture atrazine, sodium chloride
(Nad) was produced as a byproduct. This chloride could account for the
water soluble portion of the solid material. Subsequent analysis detected
-v 6000 ppm of sodium in the same feed. Secondly, it was reported that
the atrazine concentration in the wastewater was ^100 ppm, three times
its solubility limit in water. Most of the atrazine was undissolved, present
as fine crystalline rods, 5-50ym in length. This atrazine could easily
bypass the Whatmann filtration step and account for the organic fraction
of the solid material.
In order to alleviate the problems caused by fine particles of
atrazine and the chlorides, two changes were made in the experimental
procedures. First, Millipore filters (type GS), with a mean pore diameter
of 0.22ym were used to pre-fliter all wastewater solutions. Secondly,
a backwash step, using 150 ml of distilled water, was included after
each adsorption and prior to any drying or regeneration. With these steps
taken, attention returned to determining multicycle behavior of GAC columns.
Series ARW-6: In addition to being the first column using fine
filtered feed and backwashing, ARW-6 was regenerated at a mild
temperature and high pressure (135°C, 5000 psig). The results
of this regeneration were not significantly different from
columns run at lower pressures, so it was decided to halt this
series, open the column, and inspect for solids on the surface of
the glass wool or GAC. No trace of material was found anywhere
in the column.
Series ARW-7. ARW-8, and ARW-9: These three columns were used to
test another method of monitoring the desorption behavior of GAC
columns. A UV spectrophotometer, identical to that used in adsorp-
tion studies, was set up downstream of the desorbing GAC column
but upstream of the expansion valve, as shown in Figure VIII-5.
With this new configuration it was possible to measure the con-
centration of organic in the supercritical phase by means of
high pressure UV absorption cells. Regeneration of the three
columns were run at various flow rates to find the best conditions
to operate this new detector. A representative desorption trace
is presented in Figure VIII-6. As predicted by Local Equilibrium
Theory (LET), the initial concentration of organic in the super-
critical phase is quite high, and drops rapidly due to the minimal
mass transfer resistance to desorption in the supercritical phase.
This high pressure UV technique was used to monitor desorptions
for the remainder of the real waste series.
132
-------
GAC
Column
CO
GO
Compressor
Carbon Dioxide
Supply
Expansion
Valve
High Pressure
Detector
Vent
0
Dry Test
Meter
Collector
FIGURE VIII-5
Recorder
REGENERATION APPARATUS WITH HIGH PRESSURE UV DETECTOR
-------
.100
'
i
i -
i
I
i
0 ' 3
ARW-10-R
120°C
4000 psig
I :
1
' L|: ! •
_ .. . _| 5 .p .......
" ;•" "t ;
1
^ jQ1^ ""*
• • ' ' t
2 ']
i i ;
279 nm
• - • i
i
.
!
;
• 1
• I'- _,
?,
\
•
: •
'
0.'
• i
!
!
_» -^ ! , 4j
V 1.011
1 ' i | i ! ! !
, ; i .1 . ',-l.L
f\ tlCA
' 0.954
(.
~" 4.565
4857C
;
,, — - — i
t ,
1600
!! i : !
)
•» mi ' •—
! [ : '
0.54
. — •— -
I .1'
•' i'
^\
i_.
"~ — r
,. . i...
FIGURE VIII-6 HIGH PRESSURE OESORPTION TRACE OF ATRAZINE
134
-------
Series ARW-10 and ARW-11: After brief excursions of flow rate,
temperature, pressure, and the development of the high-pressure
UV technique, it was decided to carry out two extended series at
conditions representing possible large-scale operating conditions.
For series ARW-10, regenerations were done at 120 C, 4000 psig, at
a flow rate of 20 SLPM. For series ARW-11, regeneration conditions
were 120 C, 2250 psig, and again a flow rate of 20 SLPM. Highlights
of the results of these two series are shown in Table VIII-2. The
initial loading of column ARW-10 was lower than expected (0.29 gr/gr
GAC). Although the residual loading was normal for the first cycle
(0.18 gr/gr GAC), it was higher than the norm for the second cycle
(0.09 gr/gr GAC). This implied that for some as yet unknown reason,
perhaps channeling during the adsorption step, the GAC did not come
in contact with the normal volume of wastewater during the first
adsorption. The high residual loading of the second cycle of ARW-
10 was part of the irreversible loading phenomenon that is usually
seen during first cycle adsorptions, the first cycle behavior of
column ARW-11 was not unexpected. The loading of 0.32 gr/gr GAC
is more typical of the carbon, while the high residual loading
(0.28 gr/gr GAC) can be explained by the mild regeneration
conditions used. Aside from the anomally in the first two absorptions
of ARW-10, the results for both series were encouraging. The
residual loadings of both columns quickly dropped to a low level
(0.02 gr/gr GAC), a value less than that of early columns. In
addition, the amounts desorbed each cycle, which are considered
the working capacities of the carbon, leveled off (0.06 gr/gr GAC
for ARW-10 and 0.05 gr/gr GAC for ARW-11) rather than continuing to
drop, as they did in earlier work.
135
-------
IX. PROCESS DESIGN AND ECONOMIC ANALYSES
A. PROCESS DESIGN
If a solvent regeneration process is to be economically viable,
solute must be readily and efficiently separated from the carrier fluid
so that the regenerant can be recycled. One of the major benefits of
SCF regeneration is that solutes can be easily separated from the car-
rier fluid. The solubility of substances in supercritical fluids is
very sensitive to the density of the fluid. Since small changes in
temperature or pressure have a large effect on density in the critical
region, solubility can be altered dramatically by changing operating
conditions. A small change in temperature or a modest change in pres-
sure can alter solubility by orders of magnitude.
This phenomenon has been used to advantage in SCF extraction applica-
tions (see, e.g., Hag, 1973; Haddocks, ejt al_., 1979). In SCF regenera-
tion of adsorbents, such changes in solubility are used to precipitate
solutes from the carrier subsequent to desorption. In this manner, the
fluid is purified for recycle to the desorption step and the solutes
are recovered for reuse or ultimate disposal .
One mode of commercial application of the SCF regeneration process
is shown schematically in Fig.IX-l.If adsorption and desorption are not
carried out in the same vessel, then the flow of carbon is from the
spent carbon drain tank (not shown) to the pressure desorption vessel
and then to the regenerated carbon defining and storage tank (not
shown). In this mode of operation, regeneration is conducted by batch.
The flow of SCF C0? proceeds through one of the desorption vessels
where it picks up adsorbates, and is then let down in pressure. After
expansion, the temperature of the fluid is altered in a heat exchanger
to reach the conditions required for solute precipitation. The solute
is recovered from the separator and the regenerant is brought back to
the conditions required for desorption by heat exchange and recom-
pression.
In batch operation, three high-pressure desorber vessels are used.
At any one time, two vessels are off-stream for loading and unloading,
while one vessel is on-stream undergoing desorption. In this manner,
the regenerant recycle loop is operated continuously.
Early in the development program, preliminary economic analyses
were made to define the major cost items so that extra attention could
be focused on the dominant factors. The results of those analyses es-
tablished that there are two major capital costs: the high-pressure
desorber vessels and the recirculation compressor. Optimization is
simplified somewhat by the fact that the regenerant throughput required
per unit of carbon throughput is a rather weak function of column
space velocity. As discussed previously, when the mass transfer
resistance is small, desorption dynamics approach local equilibrium.
When equilibrium prevails within the column, the regenerant throughput
per unit of carbon throughput is independent of space velocity. In
136
-------
•1
I
CO
-•J
Pressure
Desorption
Vessel
V-1
Make-up C02
Pressure
Desorption
Vessel
V-2
C.W. In
Cooler
Separator
Screw Conveyor
Not Shown
Compressor Adsorbate
Out
Figure IX-1 SCHEMATIC OF A SCF ADSORBENT REGENERATION
SYSTEM
-------
actuality, the mass transfer resistance is not negligible, but it is
small enough so that column space velocity and regenerant throughput
per unit carbon throughput are only weakly coupled. Thus, column space
velocity can be chosen to minimize cycle time and, thereby, minimize
column volume, while the conditions of the recycle loop can be chosen
to minimize compression and heat exchange costs.
The desorber design was based on the method of Thomas, since that
method had been shown to adequately describe the dynamics of phenol
desorption (see Section VIIB). The method will not be illustrated for
the case of regeneration of 10,000 Ib/day of GAC, loaded with a solute
like phenol and desorbed at 55°C and 150 atm. For these conditons, the
isotherm constant, R, was found to be in the range of 2. From the
preliminary economic analyses, it was decided that a reasonable design
objective was 30 min desorption cycle time, which would conservatively
provide for 40 cycles per day. The batch size would then be 10^ Ib
GAC/day f 40 cycles/day, or 250 Ib GAC per batch. At a bulk GAC density
of 0.44 g/cnr, the volume of a desorber bed is 260 nH. Two sets of
desorber diameter and length were evaluated: case A, diam = 3.6cm (14 in),
L = 26 m (8.6 ft); case B, diam = 2cm (7.8 in), L = 85 m (28 ft).
For each case, the superficial SCF C02 velocity was treated as an
independent variable; the time required for regeneration, the amount of
CO^ passed through the column and the column pressure drop were treated
as dependent variables. The procedures will not be illustrated for the
case A column, with a superficial velocity of 7 cm/s.
From Eq. (VII-11), Dk = 3.05 x 10"3 cm2/s
From Fig. (IX-9), Df = 2.5 x 10"4 cm2/s
From Eq. (VII-10)
From Eq. (VII-9),
From Re = dpU ,
From Eq. (VI I -8),
From Eq. (VII-7),
From Eq. (VII-6),
From a = 6(1- )
d
Therefore, ap =
, D = 3.74 x 10"5 cm2/s
k = 6.24 x 10"3 cm/s
Re = 917
JD = 0.062
kf = 0.210 cm/s
= 0.138 cm/s
, a = 34.08 cm2/ cm3
4.70 s"1
From Eq. (VII-12), N = 176
Having now calculated N for the case under consideration, the
throughput of CO, required to reach any given degree of regeneration
can be determined from the Thomas solution. To aid in determining the
C02 throughput, the data of Fig VI I -16 were replotted in the following
manner. For R = 2, the throughput, T, and number of transfer units
138
-------
were read from Fig, VII-16 for outlet concentrations of X = 0.90, 0.95, and
0.98 The results were cross-plotted, as shown in Fig.iX-2 In general
the fraction of the reversible solute remaining on the column is approx- '
imately equal to (l-X)/2. In other words, X = 0.90 corresponds to 95%
regeneration, X = 0.95 corresponds to 97.5% regeneration and X = 0 98
corresponds to 99% regeneration.
At 97.5% regeneration, with a column containing 176 transfer units,
the throughput read from Fig. IX-2 is T = 1.835. The throughput and
cycle time are related by the following equation:
(1)
Uc0
where t is the time following the arrival of the fluid front at the
column exit. The throughput of CQ2 required per unit throughput of
carbon regenerated is related to t, as follows:
^
gC02 Upco2t
gGAC = LpB
For the case under consideration,
From Eq. (IX-1), t = 30.0 min
From Eq. (IX-2), gC02/gGAC = 73.8
The pressure drop of the regenerant through the bed was determined to
be 1.4 atm (see Appendix B).
This procedure was used to determine the dependent variables (cy-
cle time, regenerant throughput and pressure drop) at various superficial
velocities for each of thejtwo cases., A and B_, of column dimensions. The
results are given in Tables IX-1 and IX-2. As anticipated, we see that
cycle times are roughly inversely proportional to superficial velocity,
where the (XL/SAC ratio is only slightly affected by superficial velocity.
We began this set of calculations by assuming that the desorption
cycle time was 30 min. In the subsequent calculations, we treated cycle
time as a dependent variable and superficial velocity as the independent
variable. We see from Tables A and B that there is only one superficial
velocity, for each set of column dimensions, that is consistent with our
initial choice of a 30 min cycle. These values are summarized in Table
IX-3. We see that the case A column dimensions are clearly superior to
those of case B because A has a small pressure drop, whereas the AP of
B is prohibitive. Thus, we conclude that a 14 in i.d. x 8.6 ft bed of
GAC is capable of being regenerated in 30 min cycles for a solute like
phenol being desorbed at 55°C and 150 atm. The C02 throughput will be
in the range of 34,000 to 42,000 Ib/hr for a carbon throughput of
10,000 Ib/day.
139
-------
1.5 1.6 1.7 1.8 1.9 2.0 2.1 2.2 2.3 2.4 2.5 2.6 2.7 2.8 2.9
Throughput, T
FIGURE IX-2 NUMBER OF TRANSFER UNITS VS. THROUGHPUT
140
-------
Table IX-1
Design Calculations, Small Column Case
U(cm/s) =
Re
j
Kf(cm/s)=
X (cm/s)=
XaCS'1) -
L/U
N
P(Pa) = 7
X = .98
j _
t(min) =
gC02/gGAC=
X = .95
T
t(min) =
gC02/gGAC=
X = .90
T
t(min) =
gC02/gGAC=
5
655
.070
.167
.124
4.22
52.3
221
-4xl04
2.06
47.0
82.7
1.81
41.3
72.7
1.680
38.4
67.5
Column i.d. =
D = 35.56 cm
Ac.s. = 993.2
V = 2.597 x 1
7
917 1
.062
.210
.138
4.70
37.4
176
1.37xl05 2.
2.07
33.8
83.2
1.835
29.97
73.8
1.690
27.60
67.9
14", length of fil
cm
05cm3
10
,310
.055
.267
.152
5.18
26.2
135
73xl05
2.12
24.3
85.2
1.865
21.34
74.9
1.700
19.45
68.3
20
2,620
.044
.419
.176
6.01
13.1
78.6
1.06xl06
2.19
12.53
88.0
1.935
11.07
77.7
1.730
9.90
69.5
1 =8.58 feet
L = 261.5 cm
T = 55°C
P = 150 bar
30
3,930 5
.040
.571
.191
6.50
8.72
56.7
2.34xl06 4.
2.24
8.53
90.0
1.960
7.46
78.7
1.745
6.64
70.1
40
.256
.037
.705
.199
6.78
6.54
44.4
05x1 O6
2.28
6.51
91.6
1.980
5.65
79.5
1.755
5.01
70.5
50
6.570
.035
.833
.205
6.99
5.23
36.6
6.16xl06
2.325
5.31
93.4
2.010
4.59
80.7
1.762
4.02
70.8
141
-------
Table IX-2
Design Calculations, Large Column Case
Column i.d. = 7.75 in., length of fill = 28.0 ft.
D = 19.69 cm L = 853.4 cm
Ac.s. = 304.3 cm2 T = 55°C
V = 2.597 x 105cm3 P = 150 bar
U(cm/s) 7
Re 917
j .062
kf(cm/s) .210
X(cm/s) .138
Xa(s"]) 4.70
L/U(s) 122
N 573
X = .98
T
t(min)
gC02/gGAC
X = .95
T
t(min)
gC02/gGAC
X = .90
T
t(min)
gC02/gGAC
10
1,310
.055
.267
.152
5.18
85.3
442
1.931
71.93
77.6
1.740
64.8
69.9
1.645
61.3
66.1
20
2^620
.044
.419
.176
6.01
42.7
257
2.017
37.61
81.0
1.795
33.5
72.1
1.670
31.1
67.1
30
3,930
.040
.571
.191
6.50
28.4
185
2.061
25.56
82.8
1.830
22.7
73.5
1.688
20.9
67.8
40
5,256
.037
.705
.199
6.78
21.3
144
2.100
19.53
84.3
1.858
17.3
74.6
1.700
15.8
68.3
50
6,570
.035
.833
.205
6.99
17.1
120
142
-------
Table IX-3
Summary of Desorber Analysis
T = 55°C, P = 150 atm
Case A Case B
No. of Columns 1 1
Column i.d. (in.) 14 7.75
Length of fill (ft.) 8.6 28.0
Ac.s. (cm2) 993 304
Calculated values for 30 min. cycles,
40 cycles/day
IP4 1b GAC/day
X = C outlet @ end of cycle/C outlet in equal with loaded column
X = .02
"superficial(cm/s)
25
Ib C02/hr 42,000 40,950
P (atm) 1.5 49.4
X = .05
U -. . ,(cm/s) 7 23
superficial
Ib C02/hr 36,750 36,250
P (atm) 1-2 41.8
X = .10
U -. . , 6.5 20
superficial
Ib C02/hr 33,900 33,550
P (atm) 1-0 31.6
143
-------
B. PLAiMT DESIGN AND ECONOMICS
1. Overall Plant
These specifications were used as the basis for estimating operating
and capital costs for a 10,000 Ib/day GAC regeneration system used for
phenol. Figure IX-3 gives a process flow diagram with temperature and
pressure conditions. C02 flows through the desorber at regenerating
conditions and then through an expander which directly drives the
compressor. The expander replaces the let-down valve indicated in the
earlier general process description. Based on discussions with the
compressor/expander supplier and internal engineering evaluations, the
device is applicable at these higher C0£ flows, but is uneconomical at
the low flows that would be used in small regeneration systems of about
5,000 Ibs/day GAC or less.
The expanded C0£ stream, at 80 atm, is cooled to 45°C to further
reduce the phenol solubility for separation. This temperature was
chosen as a phenol solubility minimum based on available data. The
separator includes a mesh disengagement section to trap entrained water
and adsorbate. Solute (adsorbate) is discharged from the separator as
a solution or slurry in water which has also been stripped from the GAC
pores. C02» which is still a condensed phase near critical conditions,
is compressed to desorption pressure, heated, and recycled to the de-
sorber.
A piping and instrumentation drawing for a 10,000 Ib/day GAC-
phenol system is shown in Figure IX-4. Included is the basic C0?
circulation, plus equipment to provide the following functions:
Charging and discharging spent and
regenerated GAC, respectively;
Transfer of C02 between columns;
COp make-up.
2. GAC Charging and Discharging
Spent GAC is transferred from storage as a slurry to the available
desorber. The desorber is equipped with a plate or screen assembly to
hold the charged GAC and allow drainage of superficial water. It is
assumed that no pore water is drained.
After regeneration is complete, GAC is discharged by up-flow of
water, providing slurry flow of carbon to the adsorption columns.
The following equipment is off-plot and not included in the cost
of the regeneration plant: spent GAC storage tanks; regenerated GAC
storage tanks; slurry pump for charging; and high-water flow carbon
discharge pump.
144
-------
Steam @ 50 psig
en
Desorber
CO
Expander
Cooling Water
^
2
essor ,
v —
? *
2
W
wv
-»J
•
Compressor
. Precooler
Expander
A •f+«4*'ir-./-l j-i I/-if ^-
Cooling Water
Separator
Solute
Discharge
State Point
Location
Pressure
PSIA (MPa)
l emperature
°F l°ri
1
Desorber
|nlet
2176 (15)
248 (120)
2
Expander
Inlot
2176(15)
248(120)
3
Expander
nutlet
1161 (8)
158(70)
4
Separator
Inlet
1161 (8)
113(45)
5
Precooler
Inlet
1161 (8)
113(45)
6
Compressor
Inlet
1161 (8)
90 (32.2)
7
Steam Heater
Inlet
2176(15)
140 (60)
Fig. IX-3 PROCESS FLOW DIAGRAM (PHENOL CASE)
-------
CTl
FIGURE IX-4
PIPING AND INSTRUMENTATION DIAGRAM
-------
3. Transfer of COo Between Columns
The regeneration plant operates with one desorber on stream (for
CQ2 circulation), and two desorbers involved in either charging or
discharging carbon, or transferring C02 between them.
After a 30-minute regeneration cycle is completed in No. 1 desorber,
flow is switched from No. 1 to No. 2 desorber. No. 3 desorber has been
charged with spent carbon, and is completely filled with water to
minimize introduction of air into the system. The high-pressure metering
water pump then transfers carbonated water into No. 1 desorber at a low
flow rate (sufficiently low to keep the bed from partly fluidizing),
and slightly above bed pressure. The high-pressure water flow displaces
C02 from desorber No. 1 to desorber No. 3, thereby pressurizing No. 3
bed to desorption pressure, and displacing its interstitial water. No.
3 desorber is then ready to accept COo circulation for regeneration.
No. 1 bed, containing regenerated GAC in high-pressure water, and
with pores containing high-pressure C02, is let down to separator
pressure and held to allow expansion and release of a portion of the
pore-volume C02. That C02 is collected in the low-pressure surge tank.
No. 1 bed is then vented to atmospheric pressure, and the regenerated
carbon is discharged as described above.
The same transfer and venting operation takes place at the comple-
tion of regeneration in each of the beds in sequence. Automatic valve
operation is anticipated and accounted for in instrumenting the plant.
4. COg Make-Up
C02 make-up is provided by cylinder liquid C02 at ambient tempera-
ture ana its vapor pressure. A C02 charging tank is maintained at a
pressure slightly above the compressor suction pressure. The charging
tank is maintained full by flow from the low-pressure surge tank, and on
demand by pressure control with flow from the make-up source.
Make-up to the circulation loop is based on flow control at the
compressor suction. A short-fall on recycle flow will open the make-up
valve to allow C02 to be drawn from the make-up C02 charging tank.
5. Process Costs
Table IX-4 lists the individual equipment components. For most
components, the specification given resulted from discussions with
suppliers and identification of specific hardware. Estimates of.total
system cost were made from summing the component costs, determining
assembly costs from structure and piping layout sketches, and using
accepted installation factors- The total system cost,
including engineering, profits, and contingencies, was estimated at
about $800,000.
147
-------
Table IX-4
PLANT COMPONENT LIST
-P.
00
Regeneration of Activated Charcoal
By Supercritical Carbon Dioxide
Case: Phenol
1. Desorber Assembly
2. Carbonated Water Storage Tank
3. Metering Water Injection Pump
(for SCF C02 transfer)
4. Low Pressure C0~ Surge Tank
5. High Pressure C02 Surge Tank
6. C02 Charging Tank
7. C00 Make-up Compressor
8. Compressor/Expander Assembly
9. Expander Aftercooler
10. Separator
List of Major Components
3 x (14" ID x 8') carbon steel vessel lined with
1/8" S.S. 304 interconnected with S.S. 304 piping,
mounted with required valves and instruments
Carbon steel, 150 #, 100 gal.
10 GPM, plunger type, C.S. construction
14.7 psi to 2200 psi
Wheatley, 3000 psig design, 200 RPM
20 HP, 20" x 30" base plate, 2" piping
25 ft3, 250 # design, C.S.
16" ID x 8", 1500 # design, C.S.
16" ID x 8', 1500 # design, C.S.
Charging 24 Ib of C02/30 min. into C02 charging tank
Px = 80 bar, P2 = 100 bar
Draw CO^ from source at 250 psi (min.)
PPI
Rotoflow
Skid size: 6' x 9'
Shell and tube, S.S. 304 tube, 14 BWG,
Carbon steel shell, C02 in tube side, BEU
12" 0 x 15', 400 ft2
S.S. 304 or lined carbon steel tank
3 ft2, 18" 0x2'
(Cyclonic type, dimensions are rough, additional
design work needed)
-------
Phenol Case
Table IX-4 (cont'd.)
11. Compressor Precooler
12. Steam Heater
13. Valves
14. Instruments
Total Systems Installed Cost
Shell and tube, S.S. 304 tube, 14 BWG,
C.S, shell, C02 in tube, BEM
16" 0 x 15', 950 ft2
Shell and tube, S.S. 304 tube, 14 BWG
C.S. shell, C02 in tube, BEU
12" 0x6', 126 ft2
1 lot
1 lot
$800,000
vo
-------
Table IX-5 gives a summary of operating costs for the plant on a
daily basis. Because the plant was designed on the basis of the steady-
state GAC working capacity measured after the decline from virgin capacity,
carbon capacity losses are not a factor. Negligible destructive losses
were assumed.
The estimated operating cost is $0.085 per pound of regenerated
carbon.
150
-------
Table IX-5
ESTIMATED PROCESSING COST OF ACTIVATED CHARCOAL REGENERATION
BY SUPERCRITICAL CARBON DIOXIDE PROCESS
Plant Capacity: 10,000 Ibs/day Regenerated Charcoal
Case: Phenol
Operating Factor: 330 days/yr
Capital Investment: $812,000
'Variable Costs Unit/Day $/Unit $/Day
•
Electricity 163.5 KWH .03 4.91
Cooling Water 446.4 MGal .10 44.64
Steam 43.0 MMBtu 3.50 150.50
C00 525 Lbs .03 15.75
L 215.80
Semivariablc Costs
Operating Labor: 1/2 man/shift, 3 shifts/day @ $10/hr 120.00
Supervision: 1/2 man @ $20,000/yr 38.46
Labor Overhead: 40% Labor & Supervision 63.38
Maintenance: 2% of Capital Investment/yr 46.40
"'
Fixed Costs
Plant Overhead: 60% of Labor & Supervision ™'QA
Taxes & Insurance: 1.5% of Capital Investment/yr 34.80
Depreciation: 10% of Capital Investment/yr 232.QU
361.88
Direct Processing Costs: $845.92/day
$ .085 /lb of Regenerated
Charcoal
151
-------
REFERENCES
Am. Public Health Assn., "Standard Methods for the Examination of Water
and Wastewater," 13th ed., New York, 1971.
Anon, Chemical and Engineering News, June 9, 1975, P.
Buelow, R. W., J. K. Carswell and J. M. Symons," An Improved Method for
Determining Organics by Activated Carbon Adsorption and Solvent Extraction,"
J. Am. Hater Wks. Assn.. part I, 65(l):57-72; part II, 65(3):195-199(1973).
Burant, W. and Voelstadt, T. J., Water & Sewage Works, Nov., 1973, p. 42.
Chriswell, C. D., R. I., Ericson, G. A. Junk, K. W. Lee, J. S. Fritz and
H. J. Svec, "Comparison of Macroreticular Resin and Activated Carbon as
Sorbents," J. Am. Water Works Assn.. 69, 669 (1977).
Cohen, J. M. and English, J. N., AIChE Symp. Series No. 144, 70, 326 (1974)
Coughlin, R. W., F. S. Ezra and R. N. Tan, "Influence of Chemisorbed
Oxygen in Adsorption onto Carbon from Aqueous Solution," J. Colloid Inter-
face Sci.. 28:386(1968).
Francis, A. W., "Ternary Systems of Liquid Carbon Dioxide," J. Phys. Chem.,
58, 1099-(1954).
Franck, E. U., "Water and Aqueous Solutions at High Pressures and Temper-
atures," Pure Appl. Chem., 24, 13 (1970).
Giddings, J. C., M. N. Myers, L. McLaren and R. A. Keller, "High Pressure
Gas Chromatography of Nonvolatile Species " Science, 162:67-73(1968).
Gouw, T. H., and R. E. Jentoft, "Physical Aspects in Supercritical Fluid
Chromatography," Adv. Chromatogr., 13. 1 (1975).
Grob, K. J. Chromatogr., 84, 255 (1971).
Grob, K., and G. Grob, J. Chromatogr.. 90_, 303 (1974).
Grob, K., et al, J. Chromatogr.. 106. 299 (1975).
Gruber, G., "Assessment of Industrial Hazardous Waste Practices: Organic
Chemicals, Pesticides, and Explosives Industry," U. S. Environmental
Protection Agency.
152
-------
REFERENCES (continued)
Hag, A. 6."
a. "Extraction of caffeine from coffee," Ger. Offen. 2, 212,281
(Sept. 27, 1973).
b. "Aroma-rich decaffeinated tea," Fr. Demande 2, 140,098
(February 16, 1973).
c. "Cocoa butter," Fr. Demande 2, 140,099 (February 16, 1973).
d. "Aroma Extracts of Natural Composition," Fr. Demande 2, 140,096
(Feb. 16, 1973).
e. "Selective Extraction of Nicotine from Tobacco," Ger. Offen. 2,
142,205 (May 15, 1973).
Himmelstein, K. J., Fox, R. D.» Winter, T. H., AIChE Symp. Series No. 144,
70, 310 (1974).
Hutchins, R.A., AIChE Symp. Series No. 144, 70_, 296 (1974).
Jentoft, R. E., and T. H. Gouw, "Analysis of Polynuclear Aromatic Hydro-
carbons in Automobile Exhaust by Supercritical Fluid Chromatography,"
Anal. Chem., 48, 2195 (1976).
Kennedy, G. C., "A Portion of the System Silica-Water," Econ. Geol.,
45, 629 (1950).
Liphard, K. G., and G. M. Schneider, "Phase Equilibria and Critical
Phenomena in Fluid Mixtures of Carbon Dioxide + 2,6,10,15,19,21-hexa-
methyltetracosane up to 423 K and 100 MPa," J. Chem. Thermodyn., 7,
805 (1975).
Loven, A. W., AIChE Symp. Series No. 144, 70, 285 (1974).
Haddocks, R. R., J. Gibson and D. F. Williams, "Supercritical Extraction
of Coal," Chem. Eng. Prog., 75., 49-55 (1979).
Mattson, J. S., H. B. Mark, Jr., M. D. Malbin, W. J. Weber, Jr., and
J. C. Crittenden, "Surface Chemistry of Active CarbonrSpecific Adsorp-
tion of Phenols," J. Colloid Interface Sci.. 31:116 (1969).
Middleton, F. M., H. Braus and C. C. Ruchloft, "The Application of the
Carbon Filter and Counter-current Extraction to the Analysis of Organic
Industrial Wastes," Proc. 7th Purdue Ind. Waste Conf., 79:439 (1952).
Middleton, F. M., W. Grant and A. A. Rosen, "Drinking Water Taste and
Odor-Correlation with Organic Chemical Content," Ind. Eng. Chem.,
48:268-74 (1956).
153
-------
REFERENCES (continued)
Minor, P. S., Env. Sci. & Tech. 8. 620 (1974).
Pahl, R. H., K. G. Mayhan and G. L. Bertrand, "Organic Desorption from
Carbon-II. The Effect of Solvent in the Desorption of Phenol from Wet
Carbon," Water Research, 7:1309-22 (1973).
Paul, P.P.M., and W. S. Wise, "The Principles of Gas Extraction," Mills
and Boon, Ltd., London, 1971.
Peng, D. Y., and D. B. Robinson, "A New Two-Constant Equation of State,"
Ind. Eng. Chem. Fundam., 1_5_, 59 (1976).
Perrotti, A. E., and C. A. Rodman, AIChE Symp. Series No. 144, 70,
31 (1974).
Quinn, E. L., and C. L. Jones, "Carbon Dioxide," Rheinhold Publ. Corp.,
N. Y., 1936, pp. 109-10.
Remirez, R., "New Routes Compete for Spent Carbon Recovery," Chemical
Engineering, Sept. 12, 1977, P. 95-97.
Rijnders, G.W.A., "Supercritical Fluid Chromatography," in 5th Intern.
Symp. on Separation Methods: Column Chromatography, 1969, Chimia Suppl.,
p. 192.
Rosen, A. A., F. M. Middleton and N. W. Taylor, "Identification of Anionic
Synthetic Detergents in Foams and Surface Waters," J. Am. Water Wks,
Assn.. 48:1321-1330 (1956).
Schneider, G. M., "Phase Equilibria in Fluid Mixtures at High Pressure."
Adv. Chem. Phys., 17, 1 (1970).
Schuliger, W. G., and MacCrum, J. M,, AIChE Symp. Series No. 144, 70,
352 (1974). ~
Shell, G. L., AIChE Symp. Series No. 144, 7£, 371 (1974).
Shuckrow, A. J., and Culp, G. L., "Appraisal of Powdered Activated Carbon
Processes for Municipal Wastewater Treatment," U. S. Environmental Pro-
tection Agency Report No. EPA-600/2-77-156, 1977, P. 247.
Sie, S. T., W. Van Beersum and G.W.A. Rijnders, "High-Pressure Gas Chroma-
tography and Chromatograpny with Supercritical Fluids. I. The Effect
of Pressure on Particiation Coefficients in Gas-Liquid Chromatography
with Carbon Dioxide as a Carrier Gas," Separation Sci., 1, 459 (1966).
Smith, S. B., Chem. Eng. Prog. 71, 87 (1975).
154
-------
REFERENCES (continued)
Snoeyink, V. L., and W. J. Weber, Jr., "The Surface Chemistry of Active
Carbon - A Discussion of Structure and Surface Functional Groups,"
Environ. Sci. Tech.. 1:228 (1967).
Suffet, I. H., and McGuire, M. J., "Activated Carbon Adsorption of
Organics from the Aqueous Phase, Vol. 1 and 2, Ann Arbor Science, to be
published, 1980.
Tsekhanskaya, Yu. V., M. V. lomtev, and E. V. Mushkina, "The Solubility
of Naphthalene in Ethylene and in Carbon Dioxide Under Pressure," Russ.
J. Phys. Chem.. 38(9):1173 (1964).
U. S. Department of Health, Education and Welfare, "Public Health
Service Drinking Water Standards, 1962."
U. S. Environmental Protection Agency Technology Transfer: "Process
Design Manual for Carbon Adsorption," October, 1973.
Wise, W. S., "Solvent Extraction of Coal," Chem. Ind. (London),
1970:950.
Zhuze, T. P., and G. N. Yushkevich, "Compressed Hydrocarbon Gases as
Solvents for Crude Oils and Residuums," Izvest. Akad. Nauk SSSR, OtdejL
Tekh. Nauk. 1957 (ll):63-8.
Zimpro, "Wastewater Reclamation," commercial literature Catalog No.
2700-2.
155
-------
XI. APPENDIX
A. Local Equilibrium Theory
o o
• Fraction by
volume of voids = t
Bulk density of solid phase only = pB
• dz
FIGURE A-l
Consider a differential, cylindrical volume element of the bed, of
length dz, as shown in Fig. A-l. The material balance on the fluid and
solid phases contained within the differential volume element is
P(
el at
JB v at.
( £ )
(1)
where c and q are fluid and solid concentration, respectively, e is the
fraction of fluid-filled space outside the particles, and PB is the bulk
density of the dry adsorbent. The superficial fluid velocity is v, where
v is the average fluid velocity in the interstices between particles.
For this simple model, longitudinal diffusion is neglected and plug flow
is assumed.
Before Eq. (1) can be solved, a second equation relating fluid and
solid concentrations must be introduced. In the general case, this
second equation will take the form of
'B ( 9t I = kaF (c,q)
(2)
which expresses the rate of change of solid phase concentration as a
function of the interfacial mass transfer coefficient, ka, and a driv-
ing forces, F(c,q). Eqs. (1) and (2) can than be solved simultaneously
to obtain the function c(z,t), which is the fluid phase concentration
at any position, z, within the column as a function of time. For example,
the effluent concentration curve is c(L,t), where L is the column length.
In the general case, there are two types of mass transfer resistances
that are considered in developing Eq. (2): diffusion of solute
out of the SCF-filled pores and interfacial mass transfer from the ex-
ternal surface of the adsorbent particle into the bulk of the SCF phase.
One of the advantages of SCF regenerant is that mass transfer is rela-
tively rapid within the SCF phase. In the limiting case where resistance
to mass transfer is negligible, Eq. (2) reduces to the equilibrium relation-
ship between solid phase concentration, q, and bulk fluid concentration,
c, which is just the adsorption isotherm expression:
q =<(c)
156
-------
For this limiting case, joca! equilibrium exists at all points within the
column and at all times between particles and the adjacent fluid.
When the local equilibrium theory (LET) applies, Eg. (1) is solved
simultaneously with Eq. (3).By differentiating Eq. (3) with respect to
time, we obtain the following equation:
( M > - d[f(c)1 , 8c , ...
{ at >z ~dc ( at } (4)
or
( H \ = f'(c) < "it > (5)
where f'(c) is d[f(c)~|/dc. Substituting Eq. (5) into Eq. (1) and collecting
terms, we obtain Eq. (6):
D +-^f'(c)l (ff ) +*(|f )-0 (6)
7 t
Eq. (6) is a first-order partial differential equation that is linear
in the derivatives but has a variable coefficient because of the f'(c) term.
Such equations have relatively simple geometrical properties as expressed
by a solution procedure known as the "method of characteristics" (Sherwood,
et_ al_., 1975). In essence, each concentration, c, or coverage, q, moves
through the bed at a characteristic, velocity, v , where
vc = * C (7)
1 + (pB/e)f (C)
The characteristic velocities can be appreciated more directly by
examining the column profiles during the regeneration process. The
column profile may be represented, as shown in Fig. A-2, by the coverage
versus distance down the column at various times during the regeneration
process. Each curve in Fig. A-2 is a cross-sectional 'snap-shot1 of solute
remaining on the column at the designated time. At t = 0, q/q0 = 1 for
all x, where q0 is the initial loading. At some time, t-j, solute has
been removed from the adsorbent near the column inlet, but near the
outlet the adsorbent is still loaded to the initial value of q0. At 10
ti, a larger region near the inlet has been fully regenerated, while the
outlet has been reduced to q/q0 of .83. At 180 t-j, about 76% of the
column has been fully regenerated, while the remaining 24% (.76
-------
en
oo
Figure A-2 Column Profiles of Adsorbent Coverage as a
Function of Time Durina Reaeneration Under
Local Equilibrium Conditions
-------
In other words, this level of coverage moves through the column at a
constant velocity of z/t = .01 L/t, (where L and ti are constants that
are fixed by the nature of the adsorbent and adsorbate). A similar
analysis at q/q0 = 0 shows that the velocity of the fully regenerated
"wave" is .0042
Although Fig. A-2 is given in terms of the coverage profile, q/q0,
we could equally as well have given the profile of solute in regenerant,
c/c0, because fluid concentration and coverage are directly coupled by
the adsorption isotherm expression, as represented by Eq. (3). Thus,
given the isotherm, we could calculate a c/c0 for every q/q0, where c0
is the concentration of solute in regenerant in equilibrium with a
column at the initial loading, q0.
In general, higher values of c (or q) move through the column more
rapidly than lower concentrations. The analytical relationship between
wave velocity and concentration is given by Eq. (7) for the LET model.
The term f'(c) increases as c decreases, so that vc decreases as c
decreases.
The desorption curve is the effluent concentration as a function of
time, c(L,t). It can be obtained indirectly from the column profile,
Fig. A-2, by noting q at the outlet (z/L = 1) as a function of time,
and then converting q to c using the adsorption isotherm, Eq. (3).
Alternatively, it can be found by solving the differential equation,
Eq. (1).
A typical desorption curve predicted by the LET model is shown in
Fig. A-3. The corresponding regeneration curve, which is the fractional
regeneration, R(t), as a function of time, is shown in Fig. A-4 . The
residence time for the non-adsorbing regenerant fluid to pass through
the bed is t0. Between t0 and t-j, the effluent concentration is constant
at c0 because the adsorbent at the column outlet is still loaded with
solute at the initial value of q0 (see curve t-| in Fig. A-2). For times
greater than ti, the effluent concentration decreases with time until
the column is fully regenerated at t2 [in Fig. A-2, t2 = 237t-]]. The tail
In the desorption curve between t] and t2 is a consequence of the fact
that the characteristic velocity of the concentration wave decreases as
concentration decreases.
The regeneration curve, shown in Fig. A-4, is the fraction of the
initially adsorbed solute that has been removed from the column up to
time, t. If m is the mass of solute left on the column at any time,
t, then by material balance at the column exit,
Integrating from t to any time during desorption, t,
(8)
m - m
o
* c(L,t)dt
t_
159
-------
1.00
a. Desorption Curve
.* .25 -
0
1.0
60
90
Time (min)
120
150
180
.5
b. Regeneration Curve
30
60
90 120
Time (min)
150
180
FIGURES A-3 and A-4
L.E.T. DESORPTION AND REGENERATION CURVES
160
-------
where m0 is the initial loading:
mo = VeLAcs
Defining the fractional regeneration, R(t) as
m - m
R(t) ---
(9)
(10)
from the equations above, it follows that
R(t)
c(L,t)dt
(11)
The regeneration curve calculated from the desorption curve of Fig. A-3
is shown in Fig. A-4. It can be seen that the general shape of the desorp-
tion and regeneration curves, as predicted by the LET model, looks very
similar to the desorption curves obtained experimentally (see Fig. VI-13).
In fact, it was this similarity that led us to attempt to develop a
quantitative theoretical model to correlate our experimental results.
To apply the LET theory quantitatively, an isotherm of the form of
Eq. (3) is required. Note that Eq. (3) is not the conventional water
isotherm; rather, it is the isotherm for supercritical fluid with adsorbent.
Furthermore, the loading, q, is the surface concentration of mobile adsorbed
species rather than the total loading of solute.
Since the SCF adsorption isotherm of mobile species was not available,
we attempted to use an assumed isotherm expression for Eq. (3) and then
used an experimentally measured desorption curve to determine best-fit
isotherm constants. The assumed isotherm was taken to be the Langmuir
form,
where K and qm are constants and c is solute concentration in the regenerant
fluid. Using Eq. (12) for the function in Eq. (3), the method of charac-
teristics leads to the following solution for the desorption curve: [Note
that the time variable, t, is given in reduced form, tr - t/t0, where t0
is time for regenerant to pass through the fluid space of the bed (i.e., L/v)
In other words, t is the number of bed volumes of regenerant passed through
the bed in time t.]
c(l.tr) =0 0 ^tr <1 O3)
c(L,tr) =
|