United States
Environmental Protection
Agency
Robert S. Kerr Environmental Research
Laboratory
Ada OK 74820
EPA-600.-2-SG-081
May 1980
Research and Development
Evaluation of
Ammonia "Fixation"
Components in
Actual Refinery
Sour Waters
-------
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.
This document is available to the public through the National Technical Informa-
tion Service, Springfield, Virginia 22161.
-------
EPA-600/2-80-081
May 1980
EVALUATION OF AMMONIA "FIXATION" COMPONENTS
IN ACTUAL REFINERY SOUR WATERS
by
American Petroleum Institute
Committee on Refinery Environmental Control
Washington, D.C. 20037
Grant No. R 804364010
Project Officer
Fred M. Pfeffer
Source Management Branch
Robert S. Kerr Environmental Research Laboratory
Ada, Oklahoma 74820
This study was conducted by SRI International
under subcontract to the American Petroleum Institute,
Committee on Refinery Environmental Control
ROBERT S. KERR ENVIRONMENTAL RESEARCH LABORATORY
OFFICE OF RESEARCH AND DEVELOPMENT
U.S. ENVIRONMENTAL PROTECTION AGENCY
ADA, OKLAHOMA 74820
-------
DISCLAIMER
This report has been reviewed by the Roberts. Kerr Environmental
Research Laboratory, U.S. Environmental Protection Agency, and approved for
publication. Approval does not signify that the contents necessarily re-
flect the views and policies of the U.S. Environmental Protection Agency,
nor does mention of trade names or commercial products constitute en-
dorsement or recommendation for use.
ii
-------
FOREWORD
The Environmental Protection Agency was established to coordinate adminis-
tration of the major Federal programs designed to protect the quality of our
environment.
An important part of the Agency's effort involves the search for informa-
tion about environmental problems, management techniques and new technologies
through which optimum use of the Nation1a land and water resources can be
assured and the threat pollution poses to the welfare of the. American people
can be minimized. EPA's Office of Research and Development conducts this
search through a nationwide network of research facilities.
As one of these facilities, the Robert S. Kerr Environmental Research
Laboratory is responsible for the management of programs to: (a) investigate
the nature, transport, fate and management of pollutants in ground water;
(b) develop and demonstrate methods for treating wastewaters with soil and
other natural systems; (c) develop and demonstrate pollution control technol-
ogies for irrigation return flows; (d) develop and demonstrate pollution con-
trol technologies for animal production wastes; (e) develop and demonstrate
technologies to prevent, control, or abate pollution from the petroleum refin-
ing and petrochemical industries; and (f) develop and demonstrate technologies
to manage pollution resulting from combinations of industrial wastewaters or
industrial/municipal wastewaters.
The aquatic organism toxicity of ammonium compounds in solution is
attributable to the un-ionized portion—ammonia per se. The upper limit
recommended for ammonia by "Water Quality Criteria - 1972" is only 0.2 mg/1.
Since ammonia is a common component of many industrial process effluents,
high removal efficiencies are necessary to control effluent concentrations
at acceptable levels. This report contains the findings of a study to deter-
mine the technology for increasing ammonia removal efficiency of the sour
water scrubber process, a major source of the ammonia content of refinery
wastewater streams. Economics of ammonia removal technology up to 99 percent
efficiency are discussed.
W. C. -Galegar
Director
Robert S. Kerr Environmental Research Laboratory
iii
-------
ABSTRACT
Ammonia fixation is a term used to describe the occurrence of ammonia in
stripped sour waters (SSW) that is resistant to removal by means of additional
steam. Samples of refinery SSW were analyzed for organic components, oxidized
sulfur compounds, and heavy metals to determine the cause of ammonia fixation
during stripping. The heavy metals levels measured were too low to be causes
of fixation. However, acidic materials were identified that would cause fixa-
tion. These acids included thiosulfuric acid, thiocyanic acid, and weak
organic acids. In some SSW, ammonia is synthesized during analysis, resulting
in an apparent nonstrippable residual ammonia.
The amount of ammonia fixation that will occur with a particular sour
water can be predicted by a potentiometric titration. Fixation can also be
predicted by batch stripping using nitrogen gas until the pH of the stripped
water reaches 6.
Laboratory experiments showed that fixed ammonia can be freed by adding
caustic to the sour water stripper. The most effective place to add caustic
is high in the stripping tower. The amount of caustic required to free fixed
ammonia has no effect on sulfide stripping.
Laboratory stripping experiments were used to demonstrate that caustic
can substitute for steam in producing a SSW with low levels of ammonia. An
economic analysis showed that caustic use is only warranted when very high
ammonia removal (>97%) is required, and that the economic optimum steam and
caustic usage would degrade sulfide removal performance.
Laboratory stripping experiments could not be used to study the behavior
of cyanides during stripping of actual refinery sour waters. Cyanide measure-
ments were accurate in artificially prepared sour waters, and studies with
these artificial sour waters showed that average refinery stripper operation
could remove simple cyanide and convert some cyanide to thiocyanate. Metal-
complexed cyanide was largely unaffected by stripping.
This report was submitted by SRI International to the American Petroleum
Institute in partial fulfillment of Grant No. R 804364010 under the sponsor-
ship of the U.S. Environmental Protection Agency. This report covers a period
from March 15, 1976, to March 1977, and work was completed as of November 30,
1977.
iv
-------
CONTENTS
FOREWORD ii±
ABSTRACT lv
FIGURES vii
TABLES ix
ACKNOWLEDGMENT xi
1. INTRODUCTION 1
2. CONCLUSIONS 2
Sour Water Composition and Ammonia Fixation ........ 2
Eliminating Ammonia Fixation 2
Cyanides 3
Hydrogen Sulfide . 4
Organic Analysis 4
3. RECOMMENDATIONS 5
4. BACKGROUND 6
Fundamentals of Stripping 8
5. ANALYTICAL PROCEDURES 15
Sulfide Determinations 15
Ammonia Determinations 16
Cyanide Determinations 21
6. ANALYSIS OF REFINERY SOUR WATERS 24
Nitrogen and Sulfur Compounds 24
Heavy Metals and Alkali Earth Elements 26
Organic Analysis of Stripped Sour Waters 26
7. CAUSES FOR AMMONIA FIXATION 41
-------
8. EXPERIMENTAL APPARATUS FOR BENCH-SCALE STRIPPING
OF SOUR WATERS 52
9. CAUSTIC ADDITION TO FREE FIXED AMMONIA 56
10. SUBSTITUTION OF CAUSTIC FOR STEAM IN STRIPPING
SOUR WATERS WITH LOW "FIXED" AMMONIA LEVELS 66
Experimental Results 67
Economic Analysis 69
11. BEHAVIOR OF CYANIDE COMPOUNDS DURING STRIPPING 71
Cyanide Removal from Buffered Solutions ......... 71
Cyanide Removal from Artificial Feeds 73
BIBLIOGRAPHY 76
APPENDICES
A ANALYTICAL METHODS 78
B WEAK ACID TITRATION CURVES (REFINERIES A, D, AND F) . . . 87
C STRIPPING OF ARTIFICIAL FEEDS CONTAINING CYANIDES .... 90
D STRIPPING OF SOUR WATER FROM REFINERY D TO STUDY
THE EFFECT OF SUBSTITUTING CAUSTIC FOR STEAM 92
E CAUSTIC ADDITION TO FREE FIXED AMMONIA (REFINERIES A,
C, AND F) 96
F HIGH PRESSURE LIQUID CHROMATOGRAPHY OF REFINERY
STRIPPED SOUR WATERS 101
G SOURCES OF SOUR WATER IN THE REFINERIES SAMPLED 108
H CONVERSION OF ENGLISH UNITS TO METRIC UNITS 109
vi
-------
FIGURES
Number Page
1. Ammonia and Sulfide lonization as a Function of
Solution pH at 100°C H
2. Liquid-Vapor Partition Coefficients for Ammonia-
Nitrogen and Sulfide-Sulfur as a Function of pH at
100 C and 1 Atmosphere Total Pressure 12
3. Separation by XAD Resins ............ ..... 31
4. Separation by Extraction with Organic Solvents ...... 33
5. Gel Permeation Chromatographs: Molecular Weight as a
Function of UV Absorbance on Perasil Ax and Cx Columns.
Mobile Phase: 100% H20 ................. 38
6. Correlation of Ammonia Residual with Acid Content of
Sour Water ........................ 43
7. Effect of Benzoic Acid and Feed Oxidation on Ammonia
Removal from Sour Waters ................. 49
8. Schematic of Bench Scale Sour Water Stripper . . ..... 53
9. Bench Scale Sour Water Stripper .............. 55
10. Refinery A: NH3 Removal Efficiency at 1.03 Ib/gal
Stripping Steam (Fresh Feed Basis) with Caustic
Added at a Single Point in the Column .......... 59
11. Refinery C: NH3 Removal Efficiency at 1.04 Ib/gal
Stripping Steam (Fresh Feed Basis) with Caustic Added
at a Single Point in the Column .............. 60
12. Refinery F: NH3 Removal Efficiency at 1.04 Ib/gal
Stripping Steam (Fresh Feed Basis) with Caustic
Added at a Single Point in the Column ........... 61
13. Refinery A; H2S Removal Efficiency at 1.03 Ib/gal
Stripping Steam (Fresh Feed Basis) with Caustic
Added at a Single Point in the Column ........... 63
14. Refinery F; E^S Removal Efficiency at 1.03 Ib/gal
Stripping Steam (Fresh Feed Basis) with Caustic
Added at a Single Point in the, Column ........... 64
vii
-------
15. Refinery A: NH3 Removal Efficiency at 1.03 Ib/gal
Stripping Steam (Fresh Feed Basis) with Caustic
Added at Two Points in Column 65
16. Refinery D; Response Surface of Ammonia and Sulfide
Removal Efficiency Versus Steam and Caustic
Addition Rates 68
17. Refinery D: Response Surface of Ammonia and Sulfide
Removal Efficiency versus Steam and Caustic Addition
Rates. Calculation of Optimum Steam-Caustic Rates
When Steam is Inexpensive 70
18. Steam Stripping of Cyanide from Buffered Solutions
Using a 10-Plate Column 72
19. Steam Stripping of Cyanide Compounds from Artificial
Sour Waters Using a 10-Plate Column 74
B-l. Potentiometric Titration of Stripped Sour Water from
Refinery A: Sample No. A-3 87
B-2. Potentiometric Titration of Stripped Sour Water from
Refinery D 88
B-3. Potentiometric Titration of Stripped Sour Water from
Refinery F 89
D-l. Refinery D: Ammonia Removal Efficiency as a Function
of Steam and Caustic Addition Rate 94
D-2. Refinery D: Sulfide Removal Efficiency as a Function
of Steam and Caustic Addition Rate 95
F-l. HPLC Analysis of Standard Compounds 101
F-2. HPLC Analysis of Stripped Sour Water from Refinery A ... 102
F-3. HPLC Analysis of Stripped Sour Water from Refinery C . . . 103
F-4. HPLC Analysis of Stripped Sour Water from Refinery D . . . 104
F-5. HPLC Analysis of Stripped Sour Water from Refinery E . . . 105
F-6. HPLC Analysis of Stripped Sour Water from Refinery F . . . 106
F-7. HPLC Analysis of St ipped Sour Water from Refinery G . . . 107
viii
-------
TABLES
Number Page
1. Ammonia Determinations with the Ammonia Electrode —
Spike Recoveries ..................... 17
2. Determination of the Effect of Distillation Pretreatment
and Different Finishes on Ammonia Measurements ...... 19
3. Experiments to Determine Accuracy of Cyanide
Measurements ....................... 22
4. Cyanide Determinations in Refinery Stripped Sour Waters . . 23
5. Stripped Sour Waters: Analysis of Nitrogen and Sulfur
Species .......................... 25
6. Stripped Sour Waters: Inorganic Analysis ......... 27
7. pKa Values of Selected Acids at Room Temperature ..... 28
8. Separation of Phenol and Benzoic Acid by XAD Resins .... 32
9. TOG Data from XAD Resin Separation of Stripped
Sour Water from Refineries A and B ............ 32
10. Organic Content by Extraction .............. 34
11. Semiquantitative Analysis of SSW Extracts by GC/MS .... 37
12. Analysis for Total Organic Carbon, Phenols,
and Weak Acids ...................... 40
13. Acid and Ammonia Residuals in Refinery Stripped
Sour Waters ....................... 42
14. Acid and Ammonia Residuals in Laboratory Stripped
Sour Waters . . . .................... 44
15. H2S and NH3 Stripping Tests using Fresh Artificial Feed . . 47
16. H2S and NH3 Strapping Tests using Old Artificial Feed ... 48
17. Sour Water Acid Levels and Refinery Processes Feeding
Sour Water System . ........ ....... ..... ^1
18. Composition of Sour Waters Used in Studies of Caustic
Addition to Free Fixed Ammonia .............. 57
C-l. Cyanide Stripping Testa in Buffered Water ......... gg
C-2. Steam Stripping of Cyanide Compounds from Artificial Feeds . 91
ix
-------
D-l. Refinery D: Bench Scale Stripping to Determine
Effect of Varying Caustic and Steam Rates 93
E-l. Refinery A: Bench Scale Stripping Runs with Single Caustic
Injection Point 96
E-2. Refinery A: Bench Scale Stripping Runs with Multiple
Caustic Injection Points 97
E-3. Refinery C: Bench Scale Stripping Runs with Single Point
Caustic Injection ... 98
E-4. Refinery F: Bench Scale Stripping Runs with Single
Caustic Injection Point 99
E-5. Refinery F: Bench Scale Stripping Runs with
Multiple Caustic Injection Points 100
G-l. Sources of Sour Waters in Refinery Samples 108
H-l. Conversion of English Units to Metric Units 109
x
-------
ACKNOWLEDGMENT
This work was performed by David C. Bomberger and James H. Smith
at SRI International, Menlo Park, California under a subcontract to the
Committee on Refinery Environmental Control (CREC) of the American
Petroleum Institute. The U.S. Environmental Protection Agency funded
the project on a cost sharing basis with the API with Grant No. R 804364010.
xi
-------
SECTION 1
INTRODUCTION
Since 1971 the American Petroleum Institute has been investigating
the problem of stripping-resistant, or fixed, ammonia in sour waters
generated by refinery processing operations. Surveys of refinery opera-
tions conducted in 1972 indicated that the problem was severe enough to
justify a research program to develop improvements in stripping tech-
nology and stripper design. The research program has concentrated on
identifying the causes of ammonia fixation and on developing new equilibrium
data for the NHL-l^S-I^O system in order to provide an expanded data base
for stripper design.
An experimental study of sour water stripping conducted in 1974
(Bechtel, 1976) concluded that ammonia fixation was probably caused by
acidic material in the sour water. No positive identification of the
acidic material-was made in the 1974 study. In March 1976 an experimental
study of ammonia fixation was initiated to answer some questions raised
by previous studies and to study stripper operating procedures designed
to eliminate ammonia fixation. The five principal objectives of the
study were:
1. Identify the organic and inorganic compounds present in sour
waters which cause ammonia fixation.
2. Identify factors in refinery processing operations or
feedstocks that may be responsible for ammonia fixation.
3. Investigate in the laboratory the suitability of caustic
addition and pH control during stripping for eliminating
ammonia fixation.
4. Identify the cyanide compounds present in refinery sour
waters and determine their behavior during stripping.
5. Determine if there is any incentive for substituting caustic
for steam during stripping of sour waters that do not have high
ammonia residuals.
Stripped sour waters from eight refineries were analyzed to determine
the composition of refinery sour water. The refineries were chosen to
include plants processing California, Middle East, Gulf Coast and Mid-
Continent crudes. Sour waters from several of these refineries were
stripped in a laboratory scale stripping column to fulfill the other
objectives of the program.
-------
SECTION 2
CONCLUSIONS
SOUR WATER COMPOSITION AND AMMONIA FIXATION
(1) In the sour waters examined, ammonia fixation appears to
be caused by weak organic acids and acidic oxidized sulfur
compounds. Organic acids were an important acidic component
of the sour waters studied. Ammonia-metal complexes were not
a factor in the sour waters studied.
(2) The sample of 8 refineries was not large enough to identify
the source of the acidic compounds. Neither the organic
acids nor the oxidized sulfur compounds could be firmly
associated with a particular crude source or crude
processing operation.
(3) The amount of ammonia fixation can be measured in un-
stripped sour water by batch stripping tests where nitro-
gen and steam are used to strip ammonia and sulfide,
or by acid-base titrations of sulfide-free stripped
sour water.
(4) The distillation procedure for ammonia concentration measure-
ment gave high readings for some sour waters.
(5) Some ammonia may actually be synthesized during analysis.
Ammonia measurement by Nesslerization and any use of
distillation pH values greater than 7.4 should be
avoided.
(6) Phenol, cresols, and other aromatics with molecular
weights less that 300 constitute a substantial fraction
of the organic material found in sour waters.
ELIMINATING AMMONIA FIXATION
(7) Ammonia fixation can be eliminated by adding caustic to
a sour water stripper. The amount of caustic required
is the molar equivalent of the fixed ammonia.
(8) The optimum place to apply caustic to free the fixed
ammonia is at the top of the column. Even when 10%
excess is added to the top tray, sulfide stripping is
not inhibited.
-------
(9) In terms of ammonia removal efficiency, multiple-point
caustic injection is inferior to single-point injection.
(10) Caustic can be economically substituted for steam during
stripping of sour waters with no fixed ammonia. Caustic
substitution for steam is optimum when very high ammonia
removal efficiencies are required, but the level of
ammonia removal at which caustic addition becomes attrac-
tive depends on the price of steam relative to caustic.
For the sour water studied the following results were
obtained:
Ammonia Removal
Efficiency where
Caustic Addition
Price of Becomes Economically
Price of Steam Caustic Attractive
$3.00/1000 Ib $0.07/lb 97%
0.50/1000 Ib $0.11/lb 997.
When fixed ammonia is not present, caustic addition
does degrade sulfide removal efficiency.
CYANIDES
(11) Tentative ASTM methods A and F for measuring cyanide compounds
are unsatisfactory for the sour waters used in this investi-
gation.
(12) Metal-complexed cyanides are by and large unaffected by
conditions in a stripping column. A small percentage of
these break down to simple cyanide, which is stripped.
(13) Some CN~ was converted to SCN during cyanide stripping.
(14) Simple cyanides such as HCN, KCN, NaCN appear to be readily
stripped under the conditions normally encountered in
sour water strippers, even though cyanide stripping ex-
periments with buffered solutions show cyanide stripping
is severely impaired at pH 9.
*
1000 Ib - 454 kg. See page H-2 for factors used to convert
English units to metric units.
-------
HYDROGEN SULFIDE
(15) The methylene blue test used for sulfide concentration
measurement is not reproducible and it is heavily dependent on
analyst technique.
ORGANIC ANALYSIS
(16) A simple method for measuring phenol and cresol concen-
trations in sour waters was developed. The procedure
utilizes direct analysis of the sour water sample by
high pressure liquid chromatography. This method could
be readily adapted to measuring aromatic carboxilic
acids.
*
(17) XAD resins can be used to separate the organic compounds
in sour water into bas.ic, phenolic and acid fractions.
Separations using XAD resins are much sharper and quanti-
tative than separations using solvent extractions.
*
Rohm and Haas Company.
-------
SECTION 3
RECOMMENDATIONS
(1) Procedures using XAD resins and high-pressure liquid
chromatography show promise as a means of separating
phenols from organic acids so that the acids in sour
water can be identified and measured. Work in this
area should be continued.
(2) Analytical procedures for measuring cyanides in sour
waters are inadequate. Ongoing programs to validate
cyanide measurement methods should continue.
(3) Nitrogen compounds that hydrolyze to form ammonia
appear to be present in several of the sour waters
tested. The compounds could not be identified because
they could not be separated from phenolic compounds.
Separation procedures using XAD resins should be
developed.
(4) Alternate hydrogen sulfide measurement procedures
should be investigated.
(5) Additional sour waters with high ammonia residuals
should be analyzed by titration and batch stripping
tests to validate the relationships identified in this
study.
(6) A larger sample of refineries would be required to
identify crude or processing operation influences on
sour water composition and ammonia fixation. Any
future study should analyze sour waters produced by
individual processing units rather than the composite
sour water from the entire refinery.
-------
SECTION 4
BACKGROUND
Refinery sour waters contain thousands of parts per million (ppm)
of sulfides and ammonia and hundreds of ppm of various organic compounds.
Sulfides and ammonia are generated by many process operations and show
up in distillation condensates and scrubbing streams. Distillation
tower overhead condensate often contains water that entered with the
feed or was introduced as a vapor phase diluent. Because of high vapor
concentrations and water solubilities, sulfide and ammonia in the feed
streams are concentrated in the condensed water. Operations that generate
sour condensates include crude distillation and residual vacuum distilla-
tions as well as distillations following operations such as coking, gas
oil hydrotreating, and cracking. Additional sour waters result from
neutralized caustic scrubbing streams that are used to remove sulfide
and sulfur compounds as well as phenols from cracked and hydro-treated
hydrocarbon streams.
The various sour water streams are collected and steam-stripped to
remove sulfide and ammonia. Stripped sour water is amenable to biological
treatment and is usually mixed with other aqueous effluents on their way
to treatment facilities. The ammonia level in stripped sour water can
be a major determinant of whether the treated combined effluent meets
discharge requirements.
Sulfide removal from sour water is accomplished readily by stripping.
Ammonia removal is a more difficult problem. Laboratory experiments and
theoretical calculations have shown that in aqueous solutions of phenol,
hydrogen sulfide, and ammonia, almost complete removal of both sulfide and
ammonia is possible. Some refineries actually achieve this level of
sulfide and ammonia removal. Other refineries achieve good sulfide
removals but cannot achieve ammonia bottoms values below 100 ppm. This
ammonia often cannot be removed by increasing stripping steam rates.
There are a number of possible explanations for this residual or fixed
ammonia. In some cases, it can be related to acidic materials that tie
up ammonia as NH4+. Acids containing sulfur and oxygen were implicated
in laboratory work conducted for the API by Bechtel Corporation.(1976).
The Bechtel study also suggested that carboxylic acids such as fatty
acids and naphthenic acid were involved, but these compounds were not
identified positively by chemical analysis. The current project demon-
strated, however, that when acidic compounds are responsible for reten-
tion of ammonia, caustic addition to the sour water can reduce or eliminate
ammonia residuals.
-------
In some cases, however, it is more difficult to identify the
cause of ammonia retention, because pH adjustment does not release all
of the ammonia (Dobrzanski and Thompson, 1974). A possible explanation
is heavy metals that form complexes with ammonia. Alternatively, ammonia
residuals may be apparent only because they are caused by interferences
during the analytical procedure.
Metal-ammonia complexes are not strong relative to cyanide complexes
as can be seen by considering copper complexes as an example:
Cu(NH3)4 ^ Cu + + 4 NH3; Kg ^ 10"
Gu(CN)42" * Cu2+ + 4 CN~; Kj = 10
Metals form cyanide complexes preferentially and would be available to
form ammonia complexes only when the metals are present in excess of the
amount required for cyanide complexes.
Interference in analysis of ammonia may be a more plausible explana-
tion than formation of metal complexes. Organic acids, aldehydes,
ketones, and amines interfere with the direct Nesslerization test because
they react with the reagents. Nesslerization after distillation is sub-
ject to the same interferences because some of the volatile compounds
can distill over with the ammonia. Ammonia determinations by titration
are also subject to positive interference by basic organic nitrogen com-
pounds. Some compounds, such as urea, hydrolyze to form ammonia under
the distillation conditions (Hoffman, 1976). A complete understanding
of ammonia retention (apparent or real) requires identification of acidic
compounds and the various interfering substances.
i
In cases where ammonia removal can be improved by pH control, the
proper place to introduce a base into the stripper has been a point of
controversy. Introduction of caustic in the tower top is optimum for
ammonia removal since it will increase ammonia volatility in every
stripping stage. However, this is the least optimum point for sulfide
removal since sulfide volatility is decreased in all stripping stages.
It has been suggested that some point in the middle of the column would
be a reasonable compromise.
Once an appropriate caustic addition strategy has been established,
it seems proper to consider whether adding caustic would permit good
stripping of both sulfide and ammonia with less steam than is used in
7
-------
common practice. The incentive for considering this alternative is the
increasing unit cost of steam. Although in the long run caustic price
will also rise when steam cost increases, there may be some lag due to
market forces, since much caustic is produced as a coproduct of chlorine
manufacture. Stripping response will be a complex function of steam rate,
caustic addition rate, and the number of equilibrium stages in the column.
Sour water organics such as weak acids will also affect stripping response.
When adequate computational models are available, the incentive for
caustic-steam substitution can be studied without resort to experiment,
but experiments are required at present.
Some cyanide compounds are extremely toxic. Their fate during
stripping is important if the stripped sour waters are discharged to
receiving streams. Free cyanide may not be the most abundant cyanide
compound present in sour waters since many metals form strong complexes
with cyanide. Metals such as copper catalyze reactions that convert
cyanide to cyanate. Because the possibilities for cyanide reactions are
numerous, it seems most reasonable to determine how cyanides react under
stripping conditions by experiment.
FUNDAMENTALS OF STRIPPING
Since several phases of this investigation involve steam stripping
of both refinery and artificial sour waters, it is appropriate to reveiw
the fundamentals of stripper operation. A simple stripper of N equi-
librium stages with no reflux can be modeled mathematically by the ex-
pression [Smith (1963)].
where f is the fraction of a component in the feed that remains in the
bottoms. S is the stripping factor K(V/L) when V is the vapor rate, and
L is the liquid rate in the column. K is the liquid-vapor partition
coefficient for the component being considered. This simple model
assumes that liquid and vapor rates are constant along with K.
A sour water stripper has approximately constant L and V when
stripper feed is preheated,with L being the feed rate and V the steam
rate, but K need not be constant, principally because ionic equilibrium
between the species IkS - HS~, NHo - NH^, and H+ affects vapor pressure.
-------
In the case of ammonia, we express ammonia liquid-vapor equilibrium by
where x and y are mole fractions of NH3 in the liquid and vapor phases
and K'i]H3 is the liquid-vapor partition coefficient. K'jq^ is a constant
that is only a weak function of NH3 concentration. NH3 is in equilibrium
with NH^, and the relative amounts of the two species are determined by
the ionic equilibrium constant k*,:
[NH]
where [H ] is the hydrogen ion concentration. Because of this equilibrium,
Equation (2) can be rewritten as
If we identify K" / (1 + k [H+]) as 1^, Equation (4) becomes
where y and x represent the mole fraction of the sum of both ammonia
species, NH3 and NH£. This sum is measurable as ammonia nitrogen (NHg-N).
In Equation (5), liquid-vapor equilibrium has been redefined in terms
of the concentrations measured by the analytical procedures normally used
in a laboratory. Relationship (2) or (5) holds even when other ionic
species are present in the system.* The only impact of these other
species is to change k« by ionic strength effects. For ammonia, it is
KN> the liquid-vapor partition coefficient for ammonia nitrogen (which
is a strong-function of only pH) that becomes the appropriate partition
coefficient to use in the stripping factor, S.
*
Chemical reactions with carbonate and bicarbonate to form urea and
carbamide are neglected in this development.
-------
In a similar manner, we can include the pH effects on the parti-
tioning of sulfide between liquid and vapor by including the equilibrium
between sulfide species, kg = [H+] [HS~] / [H2SJ. This gives
*»
and
= Ksxs
where yg and Xg represent the mole fraction of the sum of the species
HS~ and HoS, identified as sulfide sulfur (I^S-S). Kg is a function of
pH and describes sulfide vapor-liquid equilibrium in terms of the
variable measured by analytical techniques. It is the appropriate parti-
tion coefficient to use in a stripping factor S for sulfide sulfur.
Figure 1 illustrates the effect of pH on the ionic equilibiium of
H£S and NH3. When this behavior is combined with Henry's Law constants
(Wilson, 1976), for both species, the result is Figure 2. These plots
show liquid-vapor partition coefficients KJJ and Kg over the entire pH
range at a temperature of 100°C. Figure 2 shows that H^S-S and NH^-N
are both volatile over the pH range 6 to 11. Below pH 6, only H2S-S is
volatile; above pH 11, only NH3-N is volatile. In the pH range where both
are volatile, l^S-S is the more volatile from pH 6 to pH 9, and NH^-N is
the more volatile above pH 9. It is important to note that NH^-N has a
constant KJJ for all pH values above 10.
Both Figures 1 and 2 are shown in terms of the pH 'at conditions
in the tower. These figures would look different if pH at 25°C were
used because of the way temperature effects K^, Kg, and K^, where Kr,
is the equilibrium constant for the reaction 1^0 «± IT*" + OHT When a
solution of H2S and NH-j at 100°C is cooled, the concentration of H"*"
decreases as the temperature falls. The pH of the solution increases.
The magnitude of this pH effect depends on the concentrations of HoS
and NHo present and in general is larger when concentrations are large.
Thus at concentrations typical of sour water feed, the pH at 25°C could
be almost 2 units higher than the pH at 100°C. For stripped wour water
with low concentrations of NH3 and I^S, the change in pH with temperature
would be small.
10
-------
100
789
SOLUTION pH
I I I I
10 11 12 13 14
SA-5015-1B
Figure 1. Ammonia and sulfide ionization as a function
of solution pH at 100°C (Wilson, 1976) .
11
-------
104
103
z 1C2
111
UJ
O
u
O
H
1-
IT
9 10
O
10
-1
JO
,-2
I I \l
10 11
12
13
14 I
PH
Figure 2.
SA-G01S-22
Liquid-vapor partition coefficients for ammonia-nitrogen and
sulfide-sulfur as a function of pH at 100 C and 1 atmosphere
total pressure.
12
-------
The result of the pH behavior of K and> Kg (illustrated in Figure 2)
is that any sour water containing only ammonia, sulfide, and compounds
that do not effect pH would leave a stripper at a pH of about 9, regard-
less of feed ammonia and sulfide composition. This pH of 9 will be
referred to often in the rest of this report. It is important to realize
that this pH is dependent on the choice of equilibrium and Henry's Law
constants. It is also a function of stripper temperature. An actual
stripper may converge to a different pH, but this difference does not
affect the principles involved. A feed containing a molar excess of
sulfide would start out acidic. Under this condition of Kc > K«, the
feed would lose a higher percentage of its sulfide than its ammonia on
each stripping stage. As the feed progressed down the stripping column,
it would become more basic until it approached a pH of about 9 where
Kg = K.,, and it would lose the same percentage of both components on
each stripping stage below this point in the column. This equal per-
centage loss would keep concentration ratios, and thus pH, constant during
the rest of the feed's passage through the stripper.
A feed that starts with an excess of ammonia would lose ammonia more
rapidly than sulfide and approach a constant pH of 9 from the basic side.
In cases where steam rates are high, excess sulfide or ammonia might
be lost quite rapidly and for most of the stripper Kg = 1C, = constant.
Equation (1) could be used to give an approximate model of stripping
behavior. From Figure 2 we see that at pH 9, Kg = K^j = 10. For a stripper
using 1.5 lb of steam per gallon of tower feed, S =" 2 and
The presence of species that have an effect on solution pH can
influence the operation of a stripper in a number of ways. Except for
cases where these materials are present in large amounts, we would ex-
pect the effect to be strong only in the bottom of the column where their
relative molar quantities are high. Consider, for example, a sour water
feed of 10,000 ppm l^S and 7,500 ppm NH3, which has a pH of about 9
(Beychok, 1967). The concentrations are equivalent to 294 meq/£ l^S-S
and 441 meq/A NH3~N. If this sour water also contained 500 ppm TOC from
benzole acid (733 mg/A benzoic acid), the acid would represent only
6 meq/Jl, or less than 27. of the H2S-S equivalents present. Under
these circumstances, the benzoic acid would have only a very small
effect on pH and hence the volatility of ammonia or sulfide in the top
of the column. Toward the bottom of the column, however, after most of
the sulfide and ammonia have been stripped, the benzole acid could
*1.5 Ib/gal = 179.7 kg/nr*.See p. 109 for conversion of English units
to metric units.
13
-------
represent an important ionizable species. The column bottoms would be
acidic, rather than having a pH of near 9 as expected. It can be shown
that when the moles of ammonia nitrogen present are equal to the moles
of benzoic acid present, the pH of the solution falls below 6. From
Figure 2 it can be seen that under these circumstances ammonia is no
longer volatile; it has been fixed.
>
If a basic material like sodium hydroxide is present in small
amounts, it also would not affect the pH in the top of the column. Near
the bottom of the column, however, where ammonia and sulfide concentra-
tions are low, the basic material would be an important component of the
sour water and would cause the pH to raise above the level of 9. If suf-
ficient base were present to raise the column pH above 11, sulfide would
become nonvolatile or fixed.
It is important to notice that the point in the column where ammonia
and sulfide levels become low enough to permit small amounts of acids and
bases to affect the pH is strongly dependent on the steam rate. With
high steam rates, low ammonia and sulfide levels will be achieved high
in the column. With low steam rates, acids and bases will only be able
to affect the pH near the column bottom.
14
-------
SECTION 5
ANALYTICAL PROCEDURES
This section discusses the analyses of sulfide, ammonia, and cyanide,
which significantly affected the results of this study. The routine
analytical methods, including those of heavy metals and alkaline earth
elements, all sulfur compounds, and nitrogen compounds, are included
in Appendix A.
SULFIDE DETERMINATION
Sulfides proved to be the most difficult sulfur compounds to measure.
Preliminary analyses using sour waters indicated that the methylene blue
procedure described in1 Standard Methods (American Public Health Associa-
tion, 1976) was reproducible and accurate. Known amounts of sulfide
added to sour waters could be quantitatively recovered. Measurements
of sulfide levels in refinery stripped sour waters were made by adding
the color reagents to diluted samples. Later in the project it was found
that measurements of low sulfide concentrations in laboratory stripped
sour waters could not be made by direct addition of the color reagents to
diluted samples. At low sulfide levels (10 ppm),some organic material
in the sour water interferes with proper color development. At times
it looked as though the methylene blue dye was being adsorbed by colloidal
organic material that was present in our samples.
On the basis of this finding, we suspect that our measurements on
refinery stripped samples may have underestimated the sulfide levels.
The procedure that evolved for laboratory stripped samples entailed
precipitation of the sulfide with cadmium hydroxide to separate the
sulfide from the interference. Cadmium sulfide was removed from the
sample by centrifugation.
Even with interferences removed, the method is not entirely satis-
factory because it is too dependent on analyst technique. The timing of
reagent addition and the amount of agitation used while adding reagents
has a profound effect on the ultimate color development.
On the basis of duplicate analysis of laboratory stripped bottoms,
we estimate that when concentrations were around 5 ppm the reproducibility
was ± 1 ppm. At 100 ppm the reproducibility was ± 16 ppm.
15
-------
For every stripping run the sulfide level in the feed was measured.
Feed levels were on the order of 3000 ppm and our reproducibility was ± 107».
The details of the method are described in Appendix A.
AMMONIA DETERMINATIONS
Ammonia was measured with an ammonia ion-specific electrode (Orion
Research, Ammonia Electrode 95-10). All determinations with this
electrode require that the solution measured be adjusted to a pH > 11 to
convert all ammonia to the un-ionized form. Various experiments were
conducted to determine what sample pretreatment was required to achieve
accurate and precise measurements with the electrode, and to compare
electrode results with alternate ammonia measurement procedures.
*
The major interferences claimed for the ammonia electrode are:
• Mercury ions that form strong complexes with ammonia
• Volatile organic amines
• Wetting agents such as organic acids and phenols.
The first interference could not be a problem because no mercury was
detected in any of the sour waters used in this study. Volatile amines
are a very unlikely interference in stripped sour waters, although they
may be present in the unstripped sour waters. The major interference of
concern was organic acids and phenols, components that would be present
in both stripped and unstripped sour water.
Table 1 contains a summary of a series of tests used to determine
the suitability of various procedures using the Orion ammonia electrode.
The table shows that with actual refinery-stripped sour water, the results
of spike recoveries vary.
The following procedures were investigated:
(1) Direct determinations using an Orion model 95-10 ammonia
electrode, including some measurements made after removal
of organic acids and phenols by ether extraction.
(2) Ammonia electrode determinations of condensate collected '
from a sample buffered to pH 9.5 during distillation. This
method is described in an EPA manual (1976). Phenols as a
class are not very volatile at any pH. Organic acids are
essentially non-volatile at pH 9.5. Therefore, the distillate
or condensate would not contain either of these materials.
*
Instruction Manual, Ammonia Electrode 95-10, Orion Research.
16
-------
TABLE 1. AMMONIA DETERMINATIONS WITH THE AMMONIA ELECTRODE—SPIKE RECOVERIES
Ammonia Addition
to 25 cc of Refinery A
Stripped Bottoms
Diluted to 100 cc
with Water and
Dechlorinating Agent
+ 0 ing/ A NH3-N
+ 10 mg/A NH3-N,
+ 20 mg/A NH3-N
+ 30 mg/A NH3-N
+ 40 mg/A NH3-N
Direct Measurement
with Ammonia
Electrode
3
3
13
13
22
23
32
41
rag/A NH3-N in Diluted
Direct Measurement
with Ammonia
Electrode
After Organic
Acids Removed
4
4
22
23
39
41
Sample
Measurement with
Ammonia Electrode
after Distillation at
pH 9.5 pH >11
6 13
6
16
15.5
23.5
25
41
42.5
Multiple extraction with ether under acidic conditions.
-------
(3) Ammonia electrode determinations of condensate collected from
a sample adjusted to pH > 11 before distillation.
Double entries in Table 1 represent duplicate determinations from
the same sample. The duplicate determinations suggest that the precision
of the technique is ± 1 ppm.
The data show that at sample ammonia levels below 30 ppm, good spike
recoveries are obtained when determinations are made with the ammonia elec-
trode in the condensate collected during distillation at pH 9.5. The
good spike recoveries indicate that if the total amount of ammonia in-
troduced into the distillation apparatus is carefully restricted, pH 9.5
is adequate to give 10070 ammonia recovery. For the unspiked sample,
determinations were made in condensate obtained from distillation at
pH 9.5 and pH 11. Recovery is complete at pH 9.5, and reproducibility
is ± 1 ppm. The 6 ppm increase in ammonia seen when the distillation pH
is adjusted from 9.5 to 11 indicates that free ammonia is generated
from some nitrogen-containing material present in the stripped sour water.
The data also show that good spike recoveries are obtained when the
ammonia electrode is used directly in the sample without distillation.
There is no apparent effect of concentration on spike recoveries.
Comparison of the results obtained after extraction of organic acids,
phenols, and other organic compounds with results obtained from untreated
samples indicates that with this particular stripped sour water, organic
acids do not interfere with the performance of the ammonia electrode.
In summary, the data in Table 1 suggest that:
(1) Spike; recoveries in sour water indicate that direct
determinations with the ammonia electrode without
distillation are accurate.
(2) Distillation pretreatment is unnecessary and is
harmful. Certainly, distillation at pH greater than
9.5 must be avoided. Standard Methods (1976) emphasize
the use of pH of 7.4 to reduce interferences.
(3) Organic acids and phenols did not interfere with determina-
tions made with the ammonia electrode.
EPA-approved methods for ammonia determination do not permit using
the ammonia electrode except after distillation. Therefore, additional
studies using several stripped sour waters were conducted to determine the
effects of distillation. In addition, ammonia electrode determinations
were compared to Nesslerization results. The data are shown in Table 2.
18
-------
vo
TABLE 2. DETERMINATION OF THE EFFECT OF DISTILLATION PRETREATMENT AND DIFFERENT
FINISHES ON AMMONIA MEASUREMENTS (ppm)
Refinery Sample
TKN
NH3-N
NH3-N
NH3-N
Procedure
Direct determination
with ammonia electrode
EPA distillation at pH 9.5;
Ammonia electrode finish
Nessler Finish
TKN finish
Distillation at pH 11;
Ammonia electrode finish
Nessler finish
Sample
Al A2
192 242
14 11
24 30,26*
33,28*
*
53,51
t *
56 55,50
Number
Fl
197
90
88,86
112,109
90,85
--
Gl
52
26
30,30*
40,35*
34,30
--
*
Duplicate determinations of the same sample.
t
Analysis supplied by refinery.
-------
Multiple entries are from duplicate determinations. Analysis of the
duplicates suggest that the reproducibility of our analysis was ± 2 ppm
for this series of tests.
For samples Fl and Gl and ammonia concentration in the condensate
from the EPA-approved distillation at pH 9.5 was measured by ammonia
electrode, Nesslerization, and total Kjeldahl nitrogen (TKN).
(1) The agreement between the TKN and electrode methods is
strong evidence that the electrode is accurately
measuring ammonia concentration in the condensate.
(2) The agreement between the TKN and electrode methods
indicates that samples Fl and Gl contain no volatile
amines to interfere with electrode determinations.
(3) Nesslerization is a poor choice for ammonia measurements
in sour waters. For both samples Fl and Gl, the
Nesslerization result is high, probably because of
interference from organics such as ketones, aldehydes,
or phenols.
Assuming that the electrode and TKN finishes are both correct, the
ammonia concentrations determined after distillation for Fl and Gl
are 87 ± 2 and 31 ± 2, respectively. The results can be summarized as
follows:
Fl Gl
direct determination
with ammonia electrode 90 ± 2 26 ± 2
NHo~~N determination after
distillation at pH 9.5 87 ± 2 31 ± 2
The differences (for samples Fl and Gl) between determinations with
and without distillation are not significant. Two inferences were
drawn from this result.
(1) It is another confirmation of the accuracy of direct '
measurements of ammonia with the electrode without
distillation pretreatment.
(2) It shows that distillation of these two samples at pH 9.5
does not generate ammonia.
20
-------
It is important to compare the results for samples Al and A2
with those for Fl and Gl. If direct electrode determinations of
ammonia in samples Al and A2 (and Fl and Gl as well) are correct, the
results in Table 2 show that free ammonia is being generated by the EPA
distillation procedure for sample Al and A2 but not for sample Fl and Gl.
Note the increase in ammonia generation when the pH is raised from 9.5
to 11.
It is also interesting to note that Nesslerization determinations
after distillation agree with electrode determinations after distilla-
tion for samples Al and A2. This is a sharp contrast to the results
obtained with samples Fl and Gl.
We feel that the conclusions to be drawn from these comparison
studies are:
(1) Nesslerization should not be used to measure ammonia
in sour waters.
(2) With some sour waters, the EPA distillation at pH 9.5
generates free ammonia, thus causing overestimation of
ammonia concentrations.
(3) An ammonia measurement using a distillation at pH > 9.5
should not be used.
(4) Direct determination of ammonia with an ammonia electrode
gives an accurate measurement of concentration for the
sour waters.
Even though direct electrode determinations were the preferred
technique for this study, electrode determinations after distillation
at a pH of 9.5 were chosen as the method for the bulk of the experimental
work in order to produce results that would be consistent with future
regulatory procedures.
CYANIDE DETEBMINATIONS
Total cyanide was measured using tentative ASTM Method A, and
free cyanide was measured using tentative ASTM Method F. The validity
of the procedures used in this study to measure cyanide is questionable.
Free cyanide measurements sometimes gave higher results than a total
cyanide measurement. Experiments to check the accuracy and precision
of Method F (ASTM) , which is purported to measure free cyanides, were
performed in refinery sample A2 and in distilled water. The averages of
duplicate measurements are shown in Table 3. The reproducibility was
21
-------
TABLE 3. EXPERIMENTS TO DETERMINE ACCURACY
OF CYANIDE MEASUREMENTS
I ~
CN (Method F)
Feed (ppm)
Refinery sample A2 10.75
Refinery sample A2 + 50 ppm CN 55.00
Refinery sample A2 + 50 ppm CN 10.15
from KSCN
Distilled water + 0.5 ppm CN 0.5
from KCN
Distilled water + 1.0 ppm CN 1.0
from KCN
*
ASTM Method F is purported to measure only CN in solution.
Metal-complexed cyanides and thiocyanate are not supposed to
be detected by this procedure.
on the order of one percent. The accuracy, as determined from spike
recoveries of added cyanide both in the distilled water and the sour
water, appears excellent. The data also show that added thiocyanate
does not interfere with the cyanide analysis.
In contrast to the above, cyanide measurements on eight sour water
samples obtained from various refineries were generally unsatisfactory.
These results are summarized in Table 4. The determinations for
samples Cl through Gl were duplicated and showed good reproducibility,
with measurements differing from the mean by no more than five percent.
However, for refinery samples A2, Cl, Fl, and Gl, the free cyanide deter-
mination exceeds the total cyanide determination by a significant amount,
perhaps a result of some unknown interferences. It appears that we
can measure cyanides reproducibly and accurately in distilled water but
not in actual refinery samples.
22
-------
TABLE 4. CYANIDE DETERMINATIONS IN REFINERY
STRIPPED SOUR WATERS
Total Cyanide Free Cyanide ..
Refinery (ASTM Method A) (ASTM Method F)
Sample (PP50 (ppm)
Al 18.1 5.5
A2
Bl
Cl
Dl
El
Fl
Gl
0.6
0.15
0.51
0.50
4.62
4.87
10.17
10.37
1.37
1.37
0.02
0.02
10.7
0.02
2.34
2.21
4.44
4.87
0.06
0.06
9.83
9.75
1.19
1.19
Tentative ASTM Method A measures CN and most metal-
complexed cyanide and is not supposed to measure
thiocyanate.
Tentative ASTM Method F measures CN only and is not
supposed to measure thiocyanate.
23
-------
SECTION 6
ANALYSIS OF REFINERY SOUR WATERS
Eight refineries with different crude slates and different refinery
operations feeding the sour water strippers were sampled. Most of the
stripped sour waters were collected under nitrogen and air-mailed to
SRI International, where they were placed in a refrigerator at 4°C until
analyzed. The samples were in transit from 2 to 5 days. Several refinery
samples were collected by SRI staff and transported to our facilities on
the same day.
Refinery A was sampled three times. The second sample was taken
because the first had an unexpectedly low pH (the refinery adds caustic
to the stripper). The third sample was acquired during the bench scale
stripping phase of the program.. Refinery B's sample also yielded a
pH that was too low, but it was not resampled because it contained very
low levels of the materials of interest to this study.
NITROGEN AND SULFUR COMPOUNDS
Table 5 summarizes the results obtained from the eight refineries.
The analysis on sample HI is incomplete because it arrived late in the
program. For all refineries except A and B, the two ammonia determina-
tions (direct measurement with ammonia electrode and measurement after
distillation) gave comparable results. Ammonia and TKN determinations
were close in value, indicating that most of the nitrogen in these sour
waters is ammonia. Only one, refinery A, stands out as having an excess
of non-ammonia nitrogen. This refinery also shows significant generation
of ammonia during the distillation part of the ammonia determination.
Refinery A stripped sour water shows 14 ppm NH3-N without distillation,
24 ppm NH3-N after distillation at pH 9.5 (Table 5), and 52 ppm NH3-N
after distillation at pH 11 (Table 2).
All of the refineries except B and G show significant amounts of
sulfur that is not sulfide. This can be seen by comparing the total
sulfur measurements with the sulfide measurements. For Refineries A,
C, and E, this extra sulfur is principally in the form of thiosulfate.
Although the analysis for Refinery H is incomplete, the sulfur balance
suggests that, with this refinery also, the non-sulfide sulfur is in the
form of thiosulfate. Thiosulfate will behave as a strong acid. There-
fore, thiosulfate is a more logical choice for the remaining sulfur in
the sour water from Refinery H than is free sulfur, which is neutral.
24
-------
TABLE 5, STRIPPED SOUR WATERS: ANALYSIS OF NITROGEN AND SULFUR SPECIES (ppm)
N?
Ui
Refinery Sample
PH
TO
NH3-N
After distillation*
Direct*
Total
H2S-S
SCN-S
S2°3"S
so2-s
sulfur
Free sulfur- S
Al A2
5.0 9.5
192 242
24 28
14 11
89[97] ND
<1
ND
92 -
5
ND J
Bl Cl
4.0 8.6
6 208
14 157
5 140
0 297[23l]
ND 41
60
77
1.6
52
Dl
9.0
80
30
22
45[49]
<1
23
17
1.5
6.4
El
10.2
1080
1075
1000
229[276]
116
52
109
2.1
<5
Fl
8.5
197
in
87(110)
90
137[87]
13
29
33
12
<5
Gl
8.4
52
if
30(38)
26
5.5
<1
<0.1
<2
<1
<5
HI
6.3
82
60
60
115
<1
4
ND
ND
ND
Cyanide "
Free
Total
5.5 10.7
18 0.6
0.02 2.25
0.15 0.51
4.5
4.7
0.06
10.2
9.79
1.37
1.19
0.02
Note: ND = Not determined.
*
Total Kjeldahl Nitrogen.
Ammonia measured with ammonia electrode in distillate using EPA approved distillation
method at pH 9.5.
Ammonia measured with ammonia electrode directly in sample.
^Free cyanide measured using tentative ASTM method F. Total cyanide measured using
tentative ASTM method A.
tt
( ) indicates Nessler finish
[ ] indicates total from individual determinations.
-------
The presence of these oxidized sulfur compounds indicates that the
sour waters have been exposed to oxygen (Chen, 1971, 1972). Exposure
during sampling was minimal, so we feel that the oxygen exposure occurred
in the refineries, either from oxygen in the crude or exposure to air in
the sour water system. The significance of the oxidized sulfur with re-
spect to the ammonia levels in the sour water will be discussed in
Section VII.
The cyanide measurements have been discussed previously in Section V»
the only inference that can be drawn from them is that cyanides were
present in most of the sour waters sampled.
HEAVY METALS AND ALKALI EARTH ELEMENTS
The results of metal analysis are shown in Table 6. The levels of
heavy metals in all of the refinery sour waters are low enough to eliminate
ammonia-metal complexes as a cause for ammonia fixation.
ORGANIC ANALYSIS OF STRIPPED SOUR WATERS
The objective of this phase of the project was to identify the
organic compounds responsible for high residual ammonia in stripped sour
water. We based our investigation on two general mechanisms that could
lead to residual ammonia:
(1) Protonation of dissolved ammonia by weak organic acids, thereby
reducing the ammonia vapor pressure.
(2) Presence of compounds in the stripped water that decompose
to yield ammonia under the conditions of the ammonia test.
The ability of weak acids to neutralize ammonia depends on the
pKa of the acid. Representative pKa values of selected acids are given
in Table 7. Since the pKa of ammonia (actually the pK of the conjugate
acid of ammonia, the ammonium ion NH^+) is 9.25, both aliphatic and
aromatic acids will protonate one mole of ammonia for every mole of
acid. However, phenols are weaker acids and in a solution containing
1 equivalent of phenol and 1 equivalent of ammonia, only about 32% of
the ammonia will be protonated. Thus, the reduction in the vapor pres-
sure of the ammonia will not be large, and phenols should not signifi-
cantly increase the residual ammonia.
26
-------
TABLE 6. STRIPPED SOUR WATERS: INORGANIC ANALYSIS
Refinery Sample
Nonvolatile total solids
(residual) (ppm)
*
Composition of residual (7.)
Na
Fe
Ca
Si
Mg
A!
Ni
Ba
Co
Cr
B
Mn
Pb
K
Sr
Metals in unfiltered sample
Na (ppm)
Fe (ppm)
K (ppm)
Ca (ppm)
Mg (ppm)
Cu (ppm)
Ni (ppm)
Al Bl rl
1263 <10 191
40 ND 7.5
0.5
0.12
0.07
0.025
0.02
0.015
0.003
0.002
0.001
<0.01
<0.001
<0.005
<0.5
<0.01
1
4
12.5
25
0.07
0.005
--
0.025
0.04
0.75
0.04
<0.005
2
0.15
500 1.3 11
6.3 '0.0 0.0
2.2 <1.0 2
2.0 1.6 15
ND ND 21
ND ND ND
ND ND ND
Dl
313
35
0.75
2.5
0.85
1
0.07
0.02
—
0.025
0.02
0.12
0.007
<0.005
2.5
0.04
92
1.4
1.5
5.0
1.4
ND
ND
El
20
25
0.2
1.25
12
0.5
0.08
0.75
0.004
6
0.006
2
0.06
<0.005
1.5
0.008
2.5
0.0
0.2
0.2
0.1
1.4
0.0
Fl
59
40
1
0.85
3
0.15
0.4
0.008
--
0.06
0.01
0.015
0.05
<0.005
1
0.015
18
0.0
0.5
0.5
<0.05
ND
ND
Gl HI
13 ND
20 ND
1
20
6
2.5
0.25
0.01
--
0.035
0.015
0.15
0.07
0.01
2
0.02
2.5 ND
0.0
0.3
2.3
0.6
ND
ND
Note: ND= Not determined.
*
Semiquantitative spectrofiraphic reported as oxide.
-------
TABLE 7. pKa VALUES OF SELECTED ACIDS AT ROOM TEMPERATURE
pKa
Compound °
Acetic acid
Propionic acid
Hex a no ic acid
Benzoic acid
1-Naphthoic acid
2-Naphthoic acid
Phenol
o-Cresol
m-Cresol
£-Cresol
Resorcinol
Hydrogen sulfide
Thiocyanic acid
Sulfurous acid
Thiosulfuric acid
4.75
4.87
4.88
4.19
3.70
4.17
9.89
10.20
10.01
10.17
9.81
6.9
4.0
7.2
1.7
Sources: Handbook of Chemistry and Physics, 54th Ed.
Page (1953)
Our initial intention was to identify and quantify as many specific
weak organic acids as possible in several stripped sour waters. However,
we found that the organic material in the test samples was mainly
phenolics, which made the analysis of organic acids very difficult within
the allotted time and funds. Several attempts to identify organic acids
were unsuccessful.
We were able, however, to measure the total concentration of weak
acids (organic and inorganic) in the stripped sour waters by potentio-
metric titration. The inorganic acids were identified chemically, and
the organic acids were back-calculated by difference. We also used the
technique of gel permeation chromatography to measure the molecular
weight distribution of the aromatic organic compounds in the stripped
sour waters. The results obtained with both these measurement techniques
are summarized below.
28
-------
Total Organic Carbon
The total amount of organic material was determined by measuring
the total organic carbon (TOG) and the loss on ignition. Most of the
TOC measurements were made by a Beckman total carbon analyzer. The
data for TOC and loss on ignition, as shown in Table 12, do not agree
well. Since the data for loss on ignition are based on drying at 100°C
followed by ignition in a muffle furnace, we believe that these measure-
ments are less reliable than the TOC measurements, which require little
sample pretreatment (acidification to remove CO?) prior to analysis.
Total Weak Acids
The total amount of weak acids was measured by potentiometric
titration, using a Brinkman Model E 306 automatic titration apparatus.
An aliquot of stripped sour water (SSW) was titrated with 0.1 N HC1 or
0.1 N NaOH. The pH of the solution was monitored as a function of the
volume of titrant that had been added to the solution of SSW. The
resulting titration curve was analyzed to determine the number of
equivalents of weak acids (pKa ~ 5) that were present in the solution.
Titration curves for several SSW solutions are shown in Figures B-l,
B-2, and B-3 in Appendix B. The weak acids are determined from the
equivalents of titrant used between inflections on the titration curve
between pH 9 and pH 3. The pKa is the pH at the point where the amount
of titrant used is \ of the total weak acid equivalents.
The data are reported as milliequivalents of acid per liter of
SSW (meq/liter). The titration procedure will only discriminate between
different acids as a function of pKa and not structure; weak inorganic
acids, such as I^S, thiocyanic, sulfurous, and carbonic would be titrated
and reported as weak acids. However, data in Table 5 show that the
sulfide concentrations in solutions titrated from the basic side were
very low and would contribute very little weak acid. The high sulfide
samples were titrated from the acid side and would have lost much of
the sulfide during acidification. Therefore, within the precision
of the results, sulfide probably does not contribute to the weak acids
measured. The concentrations of the weak inorganic acids containing
sulfur were reported in Table 5. There was no carbonic acid in any of
the samples.
29
-------
Separations by Polarity
XAD resins and extraction were evaluated as possible means of
separating the organic material in SSW into four class fractions--amines,
phenols, acids, and neutrals.
XAD resins are synthetic, hydrophobic, insoluble, cross linked,
polystyrene polymers. They have a macroreticular physical porosity and
a high surface area. They adsorb nonionized species and do not retard
ionized compounds in aqueous solutions. Organic acids having a pKa ~ 5
are retained if the pH is less than 3 but are eluted if the pH is
greater than about 6. Phenols, which have a pKg ~ 10, do not elute
unless the pH is above ~ 11. Neutral compounds that cannot be ionized,
such as hydrocarbons, can be eluted with methanol or ether. Thus, we
hypothesized that the organic materials in SSW could be fractionated by
the scheme shown in Figure 3.
For testing this method, test mixtures containing aniline, benzoic
acid, and phenol were prepared. These compounds were selected because
they could easily be determined directly in aqueous solution by high
pressure (or high performance) liquid chromatography (HPLC), using an
ultraviolet absorption (uv) detector. Preliminary experiments showed
that with 0.1 N HC1, aniline was eluted and phenol and benzoic acid
were retained. However, attempts to separate phenol and benzoic by
elution with water having a pH ~ 7 failed. Separation could be achieved
if the elution solvent was a pH 7 buffer. The data for a mixture of
benzoic acid and phenol are summarized in Table 8. The separation of
phenol and benzoic acid is excellent when a pH buffer is used. Samples
of SSW from refinery A were then separated by the procedure shown in
Figure 3. The same XAD column was used. (The column was regenerated
by elution with methanol, then back-washed with water.) In this case,
100-ml fractions were collected. The samples were analyzed for TOG, and
the results are summarized in Table 9. At this point, research on the
XAD resin separation technique was stopped because the sum of the in-
dividual TOG values was more than twice the TOG measured on the SSW.
This probably was due to excess methanol that had been used to prepare
the column and could have been removed by more extensive washing of
the column.
Separation of the organic material in SSW was also attempted using,
the extraction scheme shown in Figure 4. The principle of this method
is that ionized organics are more soluble in water than in an organic
*
The method is described in detail later in this section.
30
-------
. Eluted
ORGANIC BASES,
INORGANIC SALTS
Eluted
ACIDS
Eluted
PHENOLS
Eluted
NEUTRALS
SSW
Adjust to pH 10
Elute with O.IN HCI
Retarded — phenols, acids, neutrals
Elute with pH7 buffer
Retarded — phenols, neutrals
Elute with O.IN NaOH
Retarded — neutrals
Elute with methanol
SA-5015-24
Figure 3. Separation by XAD resins.
31
-------
TABLE 8. SEPARATION OF PHENOL AND BENZOIC
ACID BY XAD RESINS*
Fraction
Elutant Number
pH 7 buffer 1
2
3
4
0.1 N NaOH 5
6
7
Total eluted
Percent
Phenol
0%
0
0
0
66
24
6
96%
Eluted
Benzole Acid
5370
25
9
5
3
3
2
100%
Column contained 5 g of XAD-2 and 5 g of XAD-8, intimately
mixed.
Each fraction was 25 ml; total eluted = 175 ml.
A 10-ml aliquot of a solution containing 707 mg 1
benzoic and 301 mg 1 of phenol was eluted.
TABLE 9. TOC DATA FROM XAD RESIN SEPARATION OF
STRIPPED SOUR WATER FROM REFINERIES A AND B
Fraction
Amines (0.1N HCl)
Acids (pH 7 buffer)
Phenols (O.lN NaOH)
Phenols (?) (50 -ml H20)
Neutrals (75 ml methanol)
Total
Al
TOC
(ppm)
1,114
330
900
250
1,680
3,374*
Bl
TOC
(ppm)
120
20
970
50
20
1 , 180*
TOC values from sample prior to separation were;
for refinery A and 20 for refinery B.
1680
32
-------
ssw
Organic phase
To pH 11
Extract
Neutrals, Amines
Aqueous phase
Neutrals amines
Organic phase
NEUTRALS
To pH3
Extract
Aqueous phase
Amine Salts
To pH 11
Extract
AMINES
Acids, phenols, inorganics
Organic phase
To pH 7
Extract
Aqueous phase
Organic
PHENOLS
To pH 2
Extract
Aqueous
SA-5015-25
Figure 4. Separation by extraction with organic solvents.
33
-------
solvent. Thus, at pH 3, amines are protonated and have a positive charge
and will be more soluble in the aqueous layer. Phenols and acids are
also protonated, but are not ionized and do not have a charge. Hence,
they should be more soluble in the organic layer. After the sample is
carried through the scheme in Figure 4, the organic layers are dried, and
the solvent is removed in tared containers. The organic fraction was
determined by weighing. Data obtained are reported in Table 10. A
comparison of the total organic carbon values is also included in the
table. The total organic material extracted shows that, with the excep-:
tion of sample Al, the recoveries were poor. Some of the extracts were
subjected to analysis by gas chromatography/mass spectroscopy (GC/MS) (see
a following section for details). In all cases, the acid extracts con-
tained large quantities of phenol, cresols, and xylenols, suggesting that
the methylene chloride did not extract all of the phenols at pH 7. The
data also suggest that methylene chloride was probably a poor solvent
choice compared to ether, but we did not have the time or funds to repeat
the extractions.
TABLE 10. ORGANIC CONTENT BY EXTRACTION
Weight of Organic
Materials from Refinery:
(ppm)
Sample
Neutrals and organic
bases (pH 11)
Phenols (pH 7)
Acids (pH 2)
Total extracted
Total organic carbon
*
Al
313
239
902
1,454
1,680
A2f
150
77
95
322
2,960
A3f
337
154
131
622
Dlf
22
8
8
38
560
t
Fl
57
17
10
84
500
*
Solvent was ether.
Solvent was methylene chloride.
34
-------
These results, coupled with the GC/MS results, suggest that the
problem of separating the large concentration of phenols from the acids
deserves additional research. The preliminary studies with XAD resins
indicate that these resins could be used to accomplish the desired separa-
tion. Also, the separations are much sharper than we were able to achieve
with solvent extraction. In retrospect, the decision to stop using them
was probably an error, and it is now our opinion that the XAD resin
technique could be developed into a simple, rapid, and useful technique.
The major problem was that the TOG values measured in the eluted fractions
were high compared to the total TOG in the samples. Several publications
suggested that extensive cleanup by first washing with methanol and then
large volumes of water would eliminate the background TOG. Also, the
separations could be improved by adjusting the elution volume and rate.
It would also be possible to use preparative reverse-phase HPLC to
separate the organic acids from the other components of SSW. The pro-
cedure for separation and analysis of phenols, which is discussed in
the next section, demonstrated that aliphatic acids and benzoic acid
were eluted well before phenol. We cannot make an obvious choice between
XAD and reverse phase HPLC at this time, but both clearly deserve further
study.
Determination of Phenolic Compounds
We have developed a simple, straightforward analytical procedure
that can be used to determine quantitatively the concentration of phenol
and cresols in ppm quantities in SSWs. It should be possible to generalize
this method to other aqueous samples and to higher phenolics such as
xylenols and naphenols. The method is based on HPLC using a reverse
phase column.
The reverse phase column consists of a fine particulate silicate
support to which a hydrocarbon has been chemically bonded. In this case,
we used a Du Pont ODS column, which is a straight chain CIQ alcohol
bonded to the support (Waters, Spectre-Physics, and others supply similar
columns that would work just as well.) The column sorbs both nonpolar
and polar materials. The eluting solvent is generally an aqueous solution
of. a polar organic solvent that is miscible with water, such as methanol
or acetonitrile. Polar compounds are eluted readily when the solvent
contains a large proportion of water; nonpolar materials required higher
concentration of the organic solvent. For instance, phenols are readily
eluted in 1:1 methanol: water, but aromatic hydrocarbons are strongly
sorbed and are eluted with about 9:1 methanol: water.
35
-------
This general procedure has been further refined, using the method
of "Paired-Ion Chromatography" (Waters Associates, 1975), which, in
this case, involves the addition of tetrabutylammonium phosphate to
the HPLC solvent system. The tetrabutylammonium salt saturates the
HPLC column. The anions of acidic materials are further retained by
the column. We used this technique to retain the aromatic and aliphatic
acids on the column and to separate them from the other polar material
that may be present in the SSW. Other experiments suggest that it
may not be necessary to add the tetrabutylammonium salt to analyze for
phenolics, but that it is required to analyze for aromatic acids.
Chromatograms for all the samples are given in Appendix F. The
chromatograms for refineries A, D, E, and F show a substantial number
of peaks which are eluted before phenol. These peaks were not positively
identified but comparison with chromatograms of benzoic acid (see
Figure F-l) suggest that they are organic acids. Additional work with
different solvent mixtures would be required to resolve these peaks.
Analysis of SSW Extracts by GC/MS
Samples of SSW from Refineries Dl and Fl were adjusted to pH 11,
pH 7, and then pH 2 and were extracted three times at each pH with
methylene chloride. The extracts were analyzed by GC/MS. The results
are summarized in Table 11.
No organic acids or bases containing nitrogen were observed. We
did not attempt to show that aliphatic and aromatic acids could have
been observed if they were present. The main conclusion is that the
separation of phenolics by extraction is very poor, since phenol and
cresols were found in the extracts at all three pH levels. The data also
confirmed our suspicions, based on the HPLC analyses, the xylenols and
trimethyl phenols were also present. Quantitative data were not obtained.
Characterization, of Organics by Gel Permeation Chromatography
Gel permeation chromatography (GPC) is a method for determining the
molecular size distribution of complex organic mixtures. The GPC traces
shown in Figure 5 were obtained on Waters Porasil Ax and Cx columns in
series, using 100% water as the eluting solvent, and a 254-nm uv detector,
which will detect aromatic compounds. This column combination gives
good separations over the equivalent molecular weight range of about 50
to 500. The data confirmed that the major aromatic constituents were
phenolics. The peaks in the range of about 80 to 120 correspond to
phenol and cresols; the peaks at about 120 correspond to dimethyl phenols.
36
-------
TABLE 11. SEMIQUANTITATIVE ANALYSIS OF SSW EXTRACTS BY GC/MS (ppm)
u>
Compounds Identified
*
CoHy-Benzene
Aniline or Methyl pyridine
Phenol
£-Cresol
jj-Cresol
Xylenol (3 isomers)
Trimethyl phenols
Refinery Dl Refinery Fl
pH 11 pH 7 pH 2 pH 11 pH 7 pH 2
t
t
~30'
60 270 100 80 120 90
30 30 3 3 2 —
90 200 30 30 30 3
t t t f
~1QO ~5 ~1 ~1
t
Isopropyl benzene or triraethyl benzene or methyl ethyl benzene. Positive identification
was not attempted.
Very rough estimates; quantitative measurement was not attempted.
-------
c
*
1
I
LU
o
<
m
£t
O
oo
CD
100
200
300
400
MOLECULAR WEIGHT
Figure 5. Gel permeation chromatographs: molecular weight as a function of
UV absorbance on Porasil Ax and Cx columns. Mobile phase: 100% tiUO.
38
-------
The peaks in Dl and Fl at 200 and 250 were not identified. The signifi-
cant result of these experiments is that the amount of high molecular
weight aromatic compounds (molecular weight above 300) in the SSW is
very low.
The results of our analysis for total organic carbon, phenols, and
weak acids are given in Table 12. The data show that roughly 12 to 40%
of the organic material can be accounted for as phenol and cresols. We
also observed xylenols in small quantities in many of the samples but
no attempt was made to quantify these compounds. Specific carboxylic
acids were not identified, and their presence can only be inferred from
the potentiometric titrations. Data for weak inorganic acids (HSCN and
HoS03) from Table 5 were used to calculate concentrations for weak organic
acids from the total weak acid concentrations derived from the potentio-
metric titrations. Organic acids account for the majority of the total
weak acids for sour waters from refineries that contain measurable
amounts of weak acid.
39
-------
TABLE 12. ANALYSIS FOR TOTAL ORGANIC CARBON, PHENOLS, AND WEAK ACIDS (ppm)
Refinery
Analysis
TOG— -ppm
Loss on ignition
Potentiometric titrations
Total weak acid*
(meq/A )
Organic weak acid
(meq/A) 9
Direct determinations by
liquid chromafography
Phenol
m- and £-Cresol
£-Cresol
Total
Percent of TOC
accounted for by
phenols
Al A2 A3
1680 2960 ND
ND 2186 ND
36.0 ± 0.5 ND 39.,3 ± 0.5t
ND -- ND
ND 375 ND
ND 56 ND
ND 35 ND
466
-- 12
Bl Cl
20 260
100 589
ND <2.0f
" <0'1
4 62
3 23
3 9
10 94
38 27
Dl
560
313
2.9 ± 0.5f
2.1 ± 0.5
202
122
75
299
40
El
240
393
<2.0
<0.3
44
14
13
71
23
Fl
500
276
4.3 ± 0.5f
3.0 ± 0.5
182
21
12
215
33
Gl HI
160 ND
71 ND
<2.0 <2.0*
<2.0 <2.0
48 269
20 31
17 8
85 308
40
ND = Not determined.
*
Includes weak organic acids and the weak inorganic acids HCNS and ^803. Will also include H2S when present.
Titration run on sample as is. Since samples were generally basic, this titration will show H2S.
Titrations run on samples pretreated with acid. Much of the hydrogen sulfide would have volatilized
during pretreatment. The results probably do not include any sulfides.
ft
Calculated from total weak acids and the data in Table 5 for SCN-S and S02~S which were converted to meq/2 by
using factors of 32rag S = 1 meq S02-S and 32 mg S =• 1 meq SCN-S.
-------
SECTION 7
CAUSES FOR AMMONIA FIXATION
Organic acids, sulfur acids, heavy metals, and measurement inter-
ferences have been postulated as. causes of ammonia fixation (see Sec-
tion IV). Analysis of the refineries' sour water samples indicates that
heavy metals are probably not involved since heavy metal concentrations
were low. Measurement interferences are not a major factor except when
low effluent concentrations are required. Evidence points to weak acids,
such as HSCN, and organic acids (pka < 6), as the primary cause of ammonia
fixation. In some sour waters, strong sulfur acids such as sulfuric and
thiosulfuric acid may be involved. These acids are the result of oxida-
tion of sulfides in the refinery sour water system..
Tables 13 and 14 contain the evidence for these conclusions. Table 13
contains data from analysis of six refinery samples. Table 13 was pre-
pared from data in Table 5 and Table 12. Refinery A was excluded from
the table because caustic addition is used in that refinery to increase
ammonia removal. Refinery E was excluded because, as a matter of design,
it does not incorporate removal of all the ammonia from the sour water.
Table 13. shows that the residual ammonia level is high when the acid
level is high and low when the acid level is low. The same data are shown
in Figure 6. In the figure, a 45° line has been drawn to indicate the
results that would be expected if residual ammonia equivalents were equal
to acid equivalents. All of the refineries should fall close to this
line if they are stripping hard enough to remove all but fixed ammonia.
Some of the refineries fall above the line, which we interpret to mean
that they are not stripping hard enough. In general, however, there is a
good correspondence between acid level and residual ammonia.
Table 14 contains data generated in our laboratory from samples of
unstripped sour water. With these samples, chemical analysis was not
used to determine acid concentrations. Total acid content was determined
by batch stripping to a bottoms pH of 6 (Dobrzanski, 1974). This method
assumes that acids are the only cause of ammonia fixation. Physical-
chemical calculations of pH versus concentration of weak acid (pKa = 4.8)
and ammonia nitrogen show that when the molar concentration of the two
are equal, the pH will be near or slightly below 6. Any strong acid
present will contribute its' own equivalent of ammonia to the residual
measured at p.H 6. The only confusion introduced by a strong acid is that
it lowers the pH of the true equivalence point. This means that at pH 6
we can determine only the approximate amount of strong and weak acid
present by measuring the ammonia level.
41
-------
TABLE 13. ACID AND AMMONIA RESIDUALS
IN REFINERY STRIPPED SOUR WATERS
Refinery
B
C
D
F
G
H
it •fc
Weak Acid Weak Acid
(meq/A) (PKQ)
< 2
< 2
2.9 ± 0.5 4.9
3.9 ± 0.5 5.0
< 2
< 2
t
Strong Acid
(meq/£)
--
2.4
0.5
1.0
0.0
3.4*
Total Acid
(meq/1)
2.0
4.4
3.4
4.9
2.0
5.4
Ammonia
Residual
(meq/4)
1.0
11.0
2.1
6.2
2.1
4.2
Determined by potentiometric titration. This includes HSCN,
and organic acids.
Thiosulfuric and sulfuric acid. Sulfuric acid is not present in
significant amounts so this represents 8203.
Estimated by sulfur balance.
42
-------
Total Equivalents of Acid in Refinery - Stripped
Sour Water Determined by Titration
and Chemical Analysis — meq/C
TA-363522-4R
Figure 6. Correlation of ammonia residual with
acid content of sour water.
43
-------
TABLE 14. ACID AND AMMONIA RESIDUALS IN
LABORATORY STRIPPED SOUR WATERS
Refinery
A
C
D
F
JU
Ammonia
Residual
(meq/£)
57.0
3.3
3.14
1.3
Total Acid
(meq/£)
45.7
3.6
3.6
1.4
£
Titrated
Weak Acid
(meq / 4)
39.3
ND
ND
ND
ND = not determined.
*
Determined in a laboratory column with 8 equilibrium stages
and total steam rates in excess of 1.3 Ib/gal. This would
give better than 99% removal of ammonia if no fixation
occurred.
Determined by batch stripping with nitrogen and steam to
a bottoms pH of 6.
$
Potentiometric titration. The pKa was 4.8.
44
-------
The ammonia residual measurements shown in Table 14 were made on
samples of the same sour water after it had been severely stripped in
our laboratory stripping column. For three of the refinery samples,
Table 14 shows a close correspondence between the estimated acid content
and the ammonia residual after severe stripping. Only sample A3, with a
high ammonia residual, does not fit the pattern closely. For refinery A,
the ammonia residual exceeds the estimated total acid by a significant
amount. There are two probable explanations for this disagreement. First,
because refinery A contains strong acids, the total acid estimated, which
is made at pH 6, will be an underestimate of the true total acid content
since the true equivalence point will be below pH = 4.8. Second, and
probably the more important reason, refinery A contains significant
amounts of organic nitrogen compounds, which means that the ammonia con-
centration measured after stripping in the laboratory column will be an
overestimate of the fixed ammonia level.
Analyses of stripped sour waters from refinery A have shown that
the organic nitrogen compounds in this sour water hydrolyze to ammonia
during the distillation at pH 9.5, which is part of the ammonia measure-
ment procedure. At the temperatures in a stripping column, these hydrol-
ysis reactions occur rapidly if conditions are strongly acidic or basic,
but quite slowly when the pH is 6-8. Regardless of pH, the reactions
occur very slowly at room temperature. During stripping the laboratory
column, the sour water is hot and is between a pH of 6-8 for about 10
minutes, and significant hydrolysis would not occur. However, when the
bottoms from the column are distilled to measure ammonia, they would be
hot and at pH 9.5 for 10 minutes. Here, significant hydrolysis could
occur, and the ammonia measurement could be a significant over-estimate
of the true ammonia level.
During the batch stripping test to estimate total acid, the sour
water is hot and is at pH 6-8 for over an hour. Because of the long
residence time, significant hydrolysis could occur and the ammonia
t
generated by the hydrolysis would be stripped out. When the bottoms
from this batch stripping are distilled at pH 9.5 in preparation for an
ammonia measurement, there would be little unhydrolyzed organic nitrogen
left, and the ammonia concentration measured would be close to the true
value. When the measured ammonia level in the bottoms from the batch
stripping is converted to acid equivalents, it would be a close estimate
of the true acid concentration. No estimate of strong acid concentration
based on chemical analysis of this sample is available to calculate the
correct total acid concentration from the sum of titrated weak acid plus
strong acid, but if it were available, we would expect the two estimates
to agree closely.
45
-------
A few experimental stripping runs were conducted to confirm the
ammonia fixation potential of a weak organic acid and sulfur acids.
Sulfur acids were created by allowing ammonium sulfide solutions to sit
exposed to air for several hours. Benzoic acid (pKa = 4.2) was used as
the weak acid.
Table 15 shows the results of stripping freshly prepared ammonium
sulfide solutions with benzoic acid added. Since these solutions were
prepared from deaerated water, ammonium hydroxide, and l^S gas, we feel
certain that few sulfur oxidation products were present. At high steam
rates without benzoic acid, both ammonia and sulfide removal is practically
100%. This means that any change in removal efficiency will be due to
chemical effects alone. At the same high steam rate, addition of 100 ppm
of benzoic acid (0.82 meq/jfc), which is equivalent to 11.5 ppm NHg-N, causes
the bottoms pH to drop and the ammonia nitrogen concentration to rise
9.4 ppm (11.5 ppm expected); 500 ppm of benzoic acid causes a concentra-
tion increase of 54.4 (57.5 ppm expected). Since these concentration
measurements are only precise to ±57», this experiment gives convincing
evidence that a weak acid will fix an equivalent amount of ammonia. The
weak acid causes ammonia fixation by causing the bottoms pH to fall to the
point where ammonia is not volatile. The same effect seems to be evident
at low steam rates. At low steam rates, however, bottoms ammonia levels
reflect both fixation and insufficient steam to strip unfixed ammonia.
The fact that both effects are present is seen in the bottoms pH, which
is basic rather than approximately neutral, as in the case with high
steam rates. The pH in the bottoms tends to be lower when benxoic acid
is present, however, and the ammonia level tends to be higher. Apparently,
at the low steam rate even the change in ammonia volatility caused by
small changes in pH is sufficient to raise bottoms ammonia levels by
about the equivalent of the acid present.
Table 16 shows the results of two artificial feeds that were
allowed to oxidize. The steam rates used were sufficient to strip all
ammonia that was not fixed. The run without benzoic acid shows 25 ppm
fixed ammonia due to the effects of oxidation. Benzoic acid addition
causes an additional amount of ammonia to be fixed.
Figure 7 summarizes these results.
SOURCES OF ACIDS THAT CAUSE AMMONIA FIXATION
The operations of the eight refineries sampled in this program
were examined to determine if there was any relationship between crude
type or refinery processes and the acid levels in the refinery sour
46
-------
TABLE 15. H2S AND NH3 STRIPPING TESTS USING FRESH ARTIFICIAL FEED
Run Number
*
Steam rate
813-2
819-2
819-1
827-1
824-1
824-2
827-2
Stripping steam
(Ib/gal fresh
Total steam
Feed
H2S
NH3-N
Benzoic acid
PH
Bottoms
H2S
NH3-N
PH
NH3 Efficiency
H2S Efficiency
feed)
(ppm)
(ppm)
(ppm)
(ppm)
(ppm)
(%)
<%)
1.05
1.46
j.
2000
1100
0
7.9
f
< 0.1.
T
7.8
99
100
1.21
1617
700
0
7.3
< 0.1
1.6
8.1
100
100
1.13
1617
650
100
7.3
< 0.1
11
7.1
98
100
0.92
1.32
1843
690
500
7.5
< 0.1
56
7.5
92
100
0.38
1436
720
0
8.0
2.2
37
9.3
95
100
0.29
1397
700
100
8.0
0.5
56
9.3
92
100
0.31
0.67
1552
660
500
7.8
0.2
78
8.9
88
100
Temperatures (°F)
Reflux
Feed
185
150
205
175
212
170
190
170
161
162
160
163
157
175
Stripping steam is the amount of steam leaving the feed tray. Total steam is the steam fed
to the column minus the steam equivalent of column heat losses.
Tests using Hach colorimetric procedures for these measurements only.
-------
TABLE 16. H2S AND NH3 STRIPPING TESTS USING
OLD ARTIFICIAL FEED
*
Steam Rate
Total steam (Ib/gal)
Stripping steam
(Ib/gal)
Feed
H2S (ppm)
NH3-N (ppm)
Benzoic acid (ppm)
pH
Bottoms
H2S (ppm)
NH3-N (ppm)
pH
NH3 efficiency (70)
H2S efficiency (%)
Overhead Temper-
ature ( F)
Feed Temperature (°F)
Run
818-4
1.44
1.27
2000^
660
0
7.9
t
25
7.4
9.6
100
200
164
Number
819-3
1.89
1.18
1229
680
100
7.9
< 0.1
37
6.8
94
100
200
170
Note: All NH3 measurements are by ion specific
electrode. 100 mg/1 benzoic acid is
equivalent to 11.5 mg/1 NH3-N.
~ff
Stripping steam is the steam leaving the feed tray.
Total steam is the steam fed to the column minus
the steam equivalent top column heat losses.
Hach kit - Alka Seltzer Test.
methylene blue.
All other sulfides by
48
-------
20
10
O
o
CO
U
z
5
LU
rr
UNOXIDIZED FEED
A Unaltered
O 100 ppm benzole
acid added
D 500 ppm benzole
acid added
OXIDIZED FEED
A Unaltered
• 100 ppm benzoic
acid added
10
0.5
STRIPPING STEAM — Ib/gal
1.0
TA-363522-5
Figure 7. Effect of benzoic acid and feed oxidation
on ammonia removal from sour waters.
49
-------
water. Table 17 summarizes the results of this study for the 6 refineries
which provided the most complete operational data. More detailed data
is included in Appendix G. The refinery processes which contribute more
than 10% of the sour water to the stripper sampled have been identified
for each refinery. The major crude processed during the study period is
identified. The acid levels shown in Table 5 and 12 have been collected
and repeated on the Table to clarify the discussion.
The data in Table 17 do not show a clear pattern. There is no clear
relationship between the weak inorganic and strong sulfur acid levels and
the refinery processes contributing sour waters. The practice of adding
polysulfides to inhibit corrosion in the sour water system does not corre-
late with sulfur containing acids. There does appear to be some associa-
tion between organic acid levels and crude processed. The two refineries,
A and D, which show significant amounts of weak organic acids both are
processing crude slates where California crude oils predominate. It may
also be significant that both refineries have cokers, but any association
of weak acids with cokers and California crudes must be considered tenta-
tive. The fact that refinery C processes California crude, and has a coker,
but does not have high organic acid levels in its sour water weakens any
association of organic acids in sour water with cokers and California
crudes. In addition, Refinery A, which has a significantly higher level
of organic acids than the other refineries is the only refinery with a
gas oil vacuum still as a source of sour water.
Additional refinery sour waters would need to be sampled in order
to determine the association between organic acid levels and refinery
processing or crude slates. It may be necessary to identify the cokers
used as to type (i.e., fluid or delayed). The additional refineries
sampled would have to include several with the configuration of Refinery A
but processing different crude slates. It would be important also to
choose refineries to introduce more variation in sour water acid levels.
Finally, it would be most helpful to analyze sour wastes produced by
individual process units rather than only the composite sour water.
There are two reasons why the sample of refineries chosen for this
study is not adequate for identifying the sources of the acids which
cause ammonia fixation. First, most of the refineries sampled produce
sour water with total acid levels in the range of 3-5 meq/jfc. Therefore,
even though each refinery has a different processing scheme, they are
almost identical in terms of sour water acid level. Secondly, the sample
is probably too small to develop a good correlation.
50
-------
TABLE 17. SOUR WATER ACID LEVELS AND REFINERY PROCESSES FEEDING SOUR WATER SYSTEM
Crude
Refinery Prpcesses
Refinery
A
B
C
D
E
H
Sour Water Acid Levels
(meg/I)
X
X
X
X
X
XXX
X X
X X
X X
X
X
X
X
36
<2
<0.1
2.1
<0.3
<2
Weak Strong
Sulfur Sulfur
Acids Acid
0
0
1.9
0.8
1.7
2.8
0
2.4
0.5
3.4
3.4
Total
Acid
38.8
<2
4.3 - 4.4
3.4
5.1 - 5.4
3.4 - 5.4
-------
SECTION 8
EXPERIMENTAL APPARATUS FOR BENCH-SCALE
STRIPPING OF SOUR WATERS
Figure 8 is a schematic of the bench-scale stripping apparatus
that was used for the experimental program. Its key features are:
• Two 5-plate 1" I.D. Oldershaw columns to provide for
5- and 10-plate experiments as well as mid-column
caustic injection.
• A special reflux measuring trap at the bottom of the
two Oldershaw columns to permit measuring liquid flow
rates in the column.
• A distillation head that provides no more than one
equilibrium stage and at the same time allows direct
return of the reflux stream to the column top.
• A heater for the vapor path from the still top to the
condenser so as to eliminate premature condensation.
• Caustic injection ports at the top, middle, and bottom
of the column.
• A temperature control for the condenser which permits
reproducible control of the overhead condensate
temperature.
• A still pot (and heater) with a constant level feature
so that bottoms collection is automatic. The still pot
will have low holdup to speed system response to changes
in operating variables.
• A special column feed section that permits measurement
of temperature on the feed tray.
* A low speed peristaltic feed pump.
* A temperature-controlled feed preheater.
• An overhead gas condenser to collect any water leaving
the column with the stripped ammonia and sulfide.
• An overhead gas scrubber to eliminate dangerous emissions
of hydrogen sulfide.
52
-------
1
1
RECIRCULAT
CONSTANT
TEMPERATU
BATH
POV
VACUUM
AIR
INLET FOR CONDENSER
NG
HEA
RE
JER STAT
TING
TAPEx
— 7
— '
THERM
\
\
^v
PREHEATER p
FEED =P)=4^QQQOOPQQ-
RECIRCULA
CONSTAf
TEMPERAT
BATH
TIMG
JT
s
OMETER
OOCCOMOC N
!•>/-» MT-Of-M
THERMOMETER CW
A 1 1
K
L J
q^^^^Lifi J
/; ^^^--^TM SCRUBBEF
•^ ^--i»^REFLUX DRUM
/ ^-1 fr^V NX. OVERHEAD
,^nUI \ X>, WATER
HEATED TO
PREVENT
CRYSTALLIZATION
n .
W r^
O^J
S*r
FEED'
URE SECTION
CAUSTIC
INJECTION
PORT —
— — — —
REFLUX
MEASURING
TRAP —
CAUSTIC
\\lr
*a
w
u
I ntniviuivie i en
WE LL
5-PLATE
OLDERSHAW
—COLUMN
5-PLATE
BURETT FOR
MEASURING
REFLUX RATE
'
OLDERSHAW
—COLUMN
INJECTION 1 |~X|
PORT\ .1 K,
^\ V V
THERMOMETER ^. "rn] ,
HEATING Pi J]
MANTLE ~\^^_^/ /
POWER
STAT
— .
p. "
cw
— Vv-/ *"
™
FEED PUMP
• NITROGEN
TCCD RE°ERVOIFt
BOTTOMS COLLECTOR
TA-361583-21 R
Figure 8. Schematic of bench scale sour water stripper.
53
-------
Figure 9 is a photograph of the stripper.
All water rates in and out of the column can be measured. Except
for the mass represented by the sulfide and ammonia, complete material
balances can be calculated.
The fresh feed rate can be accurately determined by a volume dif-
ference measured over the duration of an experimental run. The bottoms
rate can be determined in the same way. Column heat loss and total heat
rate can be determined by using the special reflux trap included in the
column bottom. Column heat loss is measured first by adjusting the steam
rate to achieve steady-state operation at the point just before a reflux
stream forms. The liquid rate measured in the trap is the steam equivalent
of the column heat loss. Heat loss can be measured once and considered
constant for each experimental setup. For the setup used it was 1.2 g
steam/min. Column heating rates can be measured by operating the column
without feed (at steam rates sufficient to provide reflux) and measuring
the liquid flows in the special trap. For these measurements, the over-
head condenser is operated at 20°C to provide complete condensation. A
series of steam rate determinations were used to develop a calibration
curve for the powerstat that controls the bottom heater, so that during
the experimental program column heating rate could be determined approxi-
mately from the powerstat setting. For some runs the column heating rate
was measured to confirm the calibration curve. These confirmation runs
showed that the heating rate could not be determined accurately from
the calibration curve. Therefore, for all experimental runs the column
heating rate was determined by a mass and heat balance around the column
top using feed rate and reflux rate.
Reflux rate was measured during a run by diverting the flow from
the reflux drum into a burett. These diversions were made early in
a run so that they would not affect the concentration measured in the
bottoms at the end of a run. Diversion of the reflux stream does upset
the heat flow in the column and would interfere with accurate measurement
of the reflux rate. The maximum error is about 10% and does not interfere
significantly with interpretation of the stripping results. Steam flow
in the column below the feed tray was calculated from the reflux rate
by using heat balances around the feed tray. These calculations neglected
the heat of vaporization of the ammonia and sulfide present, but the
error does not significantly interfere with interpretation of the results.
The column was operated with liquid rates of from 7 to 20 cc/min.
With these flows, plate efficiency in the same size Oldershaw column
was found by experiment to be 50%-60% (Collins, 1946). A flow of
20 cc/min was preferred to lower flows, but some feeds foamed badly un-
less liquid rates were low.
54
-------
REFLUX DRUM
CONDENSER TO
DEWATER OVERHEAD
CASES
FEED
PREHEATER
FEED SECTION
REFLUX MEASURING
TRA»>
Figure 9. Bench-scale sour water stripper.
55
-------
SECTION 9
CAUSTIC ADDITION TO FREE FIXED AMMONIA
Three refinery sour waters were chosen for laboratory stripping
tests to determine the effectiveness of caustic addition for freeing
fixed ammonia. Refinery A water was chosen because it was known to have
a high ammonia residual. The refinery routinely adds caustic to the
stripper. Water from Refinery C was chosen because it could be charac-
terized as relatively low in dissolved organic carbon levels but highly
oxidized, as indicated by the high free sulfur concentration. Refinery
F water was chosen because it was not highly oxidized but had a relatively
high concentration of dissolved organic carbon.
Table 18 summarizes the characteristics of the feeds and shows the
results of the two methods used to predict fixed ammonia levels. These
methods were weak acid titration "of stripped sour water and batch strip-
ping of the raw feed. Also shown in the table are the results of bench
scale continuous stripping runs without caustic addition. For samples
C and F the agreement between observed and predicted fixed ammonia is
very good given that all of the estimates were not made using the
same sample. The disagreement shown by Refinery A is expected, since
this refinery's sour water contains hydrolyzable nitrogen compounds.
These hydrolyzable nitrogen compounds are not detected in the titration
or batch strip prediction of fixed ammonia. Because of short residence
times in the laboratory stripping column, these nitrogen compounds will
contribute an apparent fixed ammonia to the laboratory stripped samples
from Refinery A when caustic is not added to control pH. Table 18
shows that the feed from Refinery A contains approximately 200 ppm of
apparent fixed ammonia. High steam rates (~1 Ib/gal stripping steam or
~1.41b/gal total steam) were used in a column with approximately eight
equilibirum stages, (ten Oldershaw plates plus reboiler and reflux
drum). These conditions are probably severe enough to strip all but
fixed ammonia.
Single-Point Caustic Addition Studies
The results of the batch stripping tests were used to determine
the amount of fixed ammonia in each feed. These determinations would
exclude apparent fixed ammonia due to hydrolyzable nitrogen compounds.
The ammonia concentrations were converted into milliequivalents per
liter (meq/j&) of sodium hydroxide. This sodium hydroxide concentration
56
-------
TABLE 18. COMPOSITION OF SOUR WATERS USED IN STUDIES OF
CAUSTIC ADDITION TO FREE FIXED AMMONIA (ppm)
Predicted Fixed Ammonia-N
Refinery
A3
C2
F2
H2S
3172
3362
2954
NH3-N
3264
1942
2338
PH
8.7
8.7
9.4
Weak Acid
Titration
Plus
Strong Acids
*
540
t
95
t
68
Batch
Strip
640
50
20
Fixed
Ammonia-N
in Laboratory
Stripping
840
46
17
ft
Potentiometric titration of sample A3 stripped in our laboratory
to measure weak acids, but strong acids were estimated from analysis
of sample Al, which was stripped by the refinery.
Estimated from analysis of samples Cl and Fl.
57
-------
was defined as the stoichiometric caustic dose. This stoichiometric
dose does not include any caustic equivalent to the apparent fixed am-
monia. A series of stripping runs for each feed was made in the ten-
plate Oldershaw column at ~1 Ib/gal stripping steam. Caustic was added
at rates that were approximately 607,, 90%, and 110% of the stoichiometric
dose. The caustic was added to the feed, the middle of the column (just
above Plate 5), or the reboiler. A total of nine experimental conditions
were checked. Run conditions were chosen to be in random order, and
some runs were duplicated to check reproducibility. The run data is
shown in Appendix E.
Effects on Ammonia Residual
Figures 10, 11, and 12 show the effects of adding caustic to the
three feeds. For refineries A and C, which have the higher fixed ammonia
levels, the results fit into a clear pattern. Practically all of the
fixed ammonia can be freed by adding the 100% stoichiometric dose to
the column top. Almost all of apparent fixed ammonia is eliminated by
this caustic dose. The small pH increase due to the caustic addition
eliminates almost all of the unhydrolyzed nitrogen compounds in the
bottoms. The pattern also indicates that for maximum ammonia removal,
the proper place to introduce caustic is high in the column. For ex-
ample, with feed from Refinery A, ammonia removal achievable with a 100%
stoichiometric caustic dose increases from 857e to 99% as the caustic
application point is moved from the column bottom to the column top.
For Refinery C with a much lower fixed ammonia level, the direction of
the effect is the same; it is simply not as dramatic. Refinery F, with
a very low fixed ammonia level does not show a clear pattern. The reason
is that for Refinery F the fixed ammonia level is so low (20 ppm) that
the ammonia level measured contains a substantial variance introduced by
differences in column temperatures and steam rates from run to run.
At very low caustic dose rates and at very high dose rates, the
differences in effectiveness that are related to the caustic application
point become smaller. Some additional experimental work at high dose
rates (in excess of 120% stoichiometric) would be required to determine
if column bottom addition could ever be as effective as column top addi-
tion. Extrapolation of the curves shown in Figures 10 and 11 shows
that it could be, but consideration of the equilibrium staged nature of
the stripping process suggests that it could not be. After enough
caustic has been added to the reboiler to increase the bottoms pH to 11,
no further increase in ammonia volatility is possible and the ammonia
residual becomes a function only of the steam rate.
58
-------
No Caustic
Caustic Added to Feed
Caustic Added to Middle
Caustic Added to Bottom
40 60 80 100
PERCENT OF STOICHIOMETRIC CAUSTIC ADDED
120
SA-5015-15
Figure 10. Refinery A: NH, removal efficiency at 1.03 Ib/gal stripping
steam (fresh feed basis) with caustic added at a. single point
in the column.
59
-------
No Caustic
Caustic Added to Feed
Caustic Added to Middle
Caustic Added to Bottom
CO
I-
o
m
z
—-D
20 40 60 80 100
PERCENT OF STOICHIOMETRIC CAUSTIC ADDED
120
SA-5015-12
Figure 11. Refinery C: NH. removal efficiency at 1.04 Ib/gal stripping
steam (fresh feed basis) with caustic added at a single point
in the column.
60
-------
z.u
1.8
- 1.6
^
3
1
1 ••*
CO
^
° 1.2
I—
1-
o
m
Z 1.0
o
Z
1 1 1
—
A
- D
O
0
—
? 0.8 A-
< T
5 A
HI
x 0.6
Z
1°
Z 0.4
0.2
A
— A
0
—
—
i i i
0 20 40 60
i i i
No Caustic
Caustic Added to Feed —
Caustic Added to Middle
Caustic Added to Bottom
_
—
—
0
o 0
^^~
—
1 n I nn 1
80 100 120
PERCENT OF STOICHIOMETRIC CAUSTIC ADDED
SA-5015-19
Figure 12. Refinery F: NH3 removal efficiency at 1.04 Ib/gal stripping
steam (fresh feed basis) with caustic added at a single point
in the column.
61
-------
Effects on Sulfide Residual
Figures 13 and 14 show the effect of caustic addition on the sulfide
removal efficiency achieved with feed from refineries A and F, respec-
tively. There is no pattern to the results, which suggests that neither
caustic addition rate nor application point had any effect on the in-
dividual feeds. Comparison of the two feeds shows that they yielded the
same sulfide removal efficiency. This occurred despite the fact that
caustic doses received by these feeds differed by a factor of 40 (i.e.,
100% stoichiometric doses differed by a factor of 40).
MULTIPLE-POINT CAUSTIC INJECTION
Figure 15 shows some multiple-point caustic injection runs for
Refinery A. These runs have been shown along with the estimated responses
for single-point caustic injection. The results show that multiple-
point injection tends to be inferior to single-point injection. For
example, adding 90% of the stoichiometric dose at the feed and column
middle (45% at feed and 45% at middle) yields an ammonia residual of 8%
(92% removal). If it had all been added to the top, the ammonia residual
would have been only 2% (98% removal). Changing caustic injection from
a single point to two points had no effect on the sulfide removal effi-
ciency.
62
-------
1.2
1.0
0.8
0.6
O
z
z
0.4
CO
«
I
0.2
D
D
o 0
A No Caustic
Q Caustic Added to Feed
O Caustic Added to M.ddle
A Caustic Added to Bottom
D
20 .40 60 80 100
PERCENT OF STOICHIOMETRIC CAUSTIC ADDED
O
0 -
120
SA-501S-14
Figure 13. Refinery A: H_S removal efficiency at 1.03 Ib/gal stripping
steam (fresh feed basis) with caustic added at a single point
in the column.
63
-------
1.2
CO
5
O
O
CO
1.0
0.8
^ No Caustic
D Caustic Added to Feed
O Caustic Added to Middle
Caustic Added to Bottom
LU
DC
0.4
V)
(N
I
0.2
D
O
8
8
20 40 60 80 100
PERCENT OF STOICHIOMETRIC CAUSTIC ADDED
120
SA-5015-11
Figure 14. Refinery F: lUS removal efficiency at 1.03 ife/gal stripping
steam (fresh feed basis) with caustic added at a single point
in the column.
64
-------
28
26
24
22
20
18
(/}
re
O
I «
I «
<
1 10
z
0
i 8
D
Caustic Added to
Feed and Middle
Caustic Added to
Middle and Bottom
- \
CAUSTIC ADDED"1
TO MIDDLE ONLY
CAUSTIC ADDED S
TO FEED ONLY X
20 40 60 80 100
PERCENT OF STOICHIOMETRIC CAUSTIC ADDED
120
SA-5015-13
Figure 15. Refinery A: NH removal efficiency at 1.03 Ib/gal stripping
steam (fresh feed basis) with caustic added at two points in
column.
65
-------
SECTION 10
SUBSTITUTION OF CAUSTIC FOR STEAM IN STRIPPING
SOUR WATERS WITH LOW "FIXED" AMMONIA LFVELS
A sour water with low fixed ammonia levels contains very low levels
of acidic components other than sulfides. The pH profile in a stripping
column is usually highly dependent on the steam rate used. At high steam
^r
rates, we expect a pH of about 9 to be established high in the column.
If the steam rate is high, all of the free ammonia and sulfide will be
removed, and the pH may become acidic due to the presence of the small
amount of acid. Most of the ammonia and sulfide are stripped out near
the top of the column. At low steam rates, less sulfide and ammonia are
removed during each stage. The pH will still be approaching 9 in the
column bottom. It will approach either from the basic or acidic side,
depending on the initial feed concentration of ammonia and sulfide.
Adding caustic to such a feed will have the following general effects.
• Steam rate high: At low caustic addition rates the column
pH will be 9, except near the bottom where it will be
higher. If a small amount of fixed ammonia is present,
it will be released by the caustic. As caustic addition
rate is increased, the high pH region moves further up
into the column. Sulfide levels in the bottom will rise
as the high pH suppresses sulfide volatility on more of the
column stages. The effect on bottom ammonia levels will
be negligible. Therefore, care must be taken not to
overdose with caustic.
• Steam rate low: Column pH will depend on feed composition.
Ammonia and sulfide levels will be high and at low caustic
addition rates, pH changes due to caustic addition (and
thus stripping efficiency changes) will be small. As
caustic addition rates are increased, the column will
become basic and ammonia removal efficiency will increase
while sulfide removal will decrease. At very high caustic
addition rates, the column pH will be above 10 on every
stage, and all ammonia will be un-ionized. At pH 10
sulfide removal efficiency will be decreased. Increases
in caustic addition past this point will not change ammonia
removal but will continue to decrease sulfide removal.
*
As discussed earlier, this pH of 9 was derived theoretically from a
particular set of ionization equilibrium constants and Henry's Law
constants. If real sour water exhibits another pH, this does not
change the nature of the following discussion.
66
-------
EXPERIMENTAL RESULTS
An experimental verification of these effects was conducted with
sour water obtained from Refinery D. The raw sour water had a composi-
tion as follows:
H2S (ppm): 5710
NH3-N (ppm): 3370
pH: 9.0
Batch stripping of this sour water to pH 6.1 yielded a residual ammonia-
nitrogen level of 50 ppm, or 3.6 meq/1. The residual ammonia is 1.4%
of the feed ammonia. Fourteen stripping runs were made with caustic
addition rates varying between 0.000 and 0.023 Ib/gal of sour water feed.
Total steam rates were varied between 0.48 to 1.41 Ib/gal feed. Complete
results are shown in Appendix D. Important features of the data are shown
in Figure 16, which is a contour map of ammonia and sulfide removal effi-
ciencies. This contour map shows the following expected features:
(1) At high steam rates (total steam rate > 0.8 Ib/gal), the
first amounts of caustic added release the fixed ammonia
(which, based on the shape of the contour map is between
17a to 1.5% of the feed concentration). It is important
to note the close agreement between the fixed ammonia
level predicted by the contour map and the level
predicted by the batch stripping test. Additional
caustic cannot increase ammonia removal very much,
but it does decrease sulfide removal efficiency
dramatically.
(2) At very low steam rates, the first amounts of caustic
added have small effects on both sulfide and ammonia
removal. As caustic addition rate increases, sulfide
removal efficiency begins to fall rapidly while ammonia
removal efficiency becomes independent of caustic addi-
tion rate but highly dependent on steam rate. This is
because pH is high on all column stages and because
steam rate rather than ammonia volatility limits ammonia
removal.
67
-------
1.4
1.3
1.2
1.1
1 1-0
-a
0.8
0.6
<
UJ
0.5
0.4
0.3
0.2
0.1
Ammunia Remaining
m Bottoms
Sulfide Remaining
in Bottoms
3% 4% 5% 6% 8%
LOCUS OF
OPTIMUM STEAM AND
CAUSTIC RATES
STEAM-CAUSTIC
SUBSTITUTION RATIO
WHEN STEAM COSTS
0.3#/lb AND CAUSTIC
COSTS
J L
I I I
J I
0 0.002 0.004 0.006 0.008 0.01 0.012 0.014 0.0*16 0.018 0.020 0.022 0.024 0.026
CAUSTIC ADDITION RATE IbAjal (sour water basisl
SA-5015-10R
Figure 16. Refinery D: Response surface of ammonia and sulfide
removal efficiency versus steam and caustic addition
rates.
68
-------
ECONOMIC ANALYSIS
Figure 16 can be used to calculate economic optimum steam and
caustic rates for achieving a target ammonia removal. The line labeled
steam-caustic substitution ratio represents the combinations of steam
and caustic that can be purchased for the same expenditure. For example,
when steam costs $3.00/1000 Ib and caustic costs 7c/lb, the line shown
represents a utility expenditure of $0.0029/gal of sour water. If we
neglect changes in operating costs due to changes in cooling water usage
and capital charges that result from substituting caustic for steam, two
of the many possibilities for this fixed expenditure are 0.975 Ib/gal
of steam and no caustic, or 0.80 Ib/gal steam and 0.007 Ib/gal of caustic.
The best ammonia removal that can be obtained for this expenditure is
found where this substitution line is tangent to an ammonia level contour.
The tangency is at 0.825 Ib/gal of steam and 0.006 Ib/gal of caustic,
which yields a 0.5% ammonia level. Any other steam-caustic combination
with utility costs of $0.0029/gal of sour water yields a higher ammonia
level. In a similar manner, optimum combinations of steam and caustic
can be developed for other utility expenditures. The lines representing
other utility expenditures will be parallel to the line shown in the
figure. Each of these lines will be tangent to an ammonia level contour
at the optimum combination. The line connecting these tangency points
is a locus of optimum steam and caustic rates. For the utility prices
chosen, the locus shows that for achieving ammonia residuals of 37=,
or more (less than 97% removal), only steam should be used. If lower
ammonia residuals are required, the economic optimum requires that caustic
be used. Note, however, that the optimum utility mix of steam and caustic
does degrade sulfide removal. Thus, sulfide constraints may prohibit
using the optimum utility mix.
The steam cost used in Figure 16 is perhaps too high—especially in
a refinery where waste heat from low pressure steam is available. Fig-
ure 17 was developed to show the optimum steam-caustic usage when steam
is valued at only $0.50/1000 Ib. The price of caustic was also raised to
a more realistic ll^/lb. The change is dramatic. For this feed there
is no economic incentive to use caustic until better than 99% ammonia
removal is required.
Different feeds would yield different results with this type of
analysis, and a complete analysis would also have to include the effects
of changing the number of equilibrium stages. The results with RefineryD,
however, suggest that caustic is economically attractive only when very
high ammonia removals are required.
69
-------
1.4
1.3
1.2
1.1
V>
I i.o
TJ
03
O)
c. 0.9
c/t
a>
| 0.8
^
! 0.7
LU
t-
-------
SECTION 11
BEHAVIOR OF CYANIDE COMPOUNDS DURING STRIPPING
Analysis of stripped sour water samples collected for this study
indicated that the methodology for measuring cyanide in sour waters,
either as free cyanide or complexed with metals, was inadequate. The
principle defect was that measurements for free cyanide often showed
higher results than measurements of the sum of free plus complexed
cyanide. Some material in sour waters interferes with the analytical
procedures.
To overcome the analytical difficulty, we used two different
artificial feeds to investigate cyanide behavior during stripping. A
few runs were conducted, using buffered distilled water to investigate
the effects of pH and steam rate on free cyanide. A more extensive
series of runs was used to investigate the behavior of free and complexed
cyanide and thiocyanate in artificial sour waters prepared from ammonium
hydroxide and hydrogen sulfide gas. These artificial feeds were handled
carefully to avoid oxidation of the sulfide.
CYANIDE REMOVAL FROM BUFFERED SOLUTIONS
Phosphate and borate buffers and distilled water were used to pre-
pare solutions with pH values of 7 and 9, respectively. KCN was added
to bring the solution cyanide concentration to approximately 10 ppm CN .
Cyanide levels in the feed and bottoms were measured using a colorimetric
test sold by the Hach Chemical Company. The Hach procedure is adequate
for solutions containing no interferences. A ten-plate column was used
for these stripping runs.
Figure 18 illustrates the results. The complete run data is in
TableC-lof Appendix C. As expected, cyanide removal is sensitive both
to pH and steam rate. At pH 7 it is possible to remove all the cyanide
at realistic steam rates, because the cyanide-is almost 100% un-ionized
at this pH. At pH 9, however, cyanide removal is not complete at any
reasonable steam rate. The reason for this'is twofold. First, at
pH 9, only 50% of the cyanide in solution is in the volatile HCN form.
Second, HCN volatility is at least two orders of magnitude less than that
of sulfide. This means that at a pH like- 9, where sulfide stripping is
71
-------
Z
Z
in
tr
OJ
Q
u
100
80
60
40
20
0.4 0.6 0.8 1.0 1.2 1.4
TOTAL STEAM RATE — Ib/gal
1.6
SA-5015-26
Figure 18. Steam stripping of cyanide from buffered solutions using
a 10-plate column.
72
-------
not impaired, even though sulfide at this pH is not all in the volatile
form, cyanide stripping can be impaired. Since we expect a stripper to
operate with a pH nearer to 9 than to 7, this experiment suggests that
complete cyanide removal in a stripper will be difficult to achieve.
CYANIDE REMOVAL FROM ARTIFICIAL FEEDS
Artificial feeds were prepared from deaerated distilled water to
prevent oxidation of sulfide, since the oxidized sulfur compounds that
result are acids and would interfere with interpretation of the results.
Ammonium hydroxide was added to the deaerated water and l-^S was bubbled
through the solution until a specified weight change occurred. The
weight change was hard to measure reproducibly, so some variance occurred
in feed sulfide levels.
KCN, KSCN, and ^Fe(CM)g were added to the freshly prepared ammonium
sulfide solutions in order to obtain the equivalent of 50 ppm of CN~.
(For example, 50 ppm of CN~ requires 186.5 ppm of KSCN.) Three stripping
runs were conducted with KCN as the only source of cyanide, three with
KSCN as the only source, and three with K^FeCCN)^. Both free and com-
pLexed cyanide concentrations were determined by ASTM Method A for total
cyanides. The complete run data are given in Appendix C but the impor-
tant features are illustrated in Figure 19. Three conclusions are clear:
(1) Complex cyanides are not significantly converted to free
cyanides in the laboratory stripper and are not removed.
(2) Thiocyanate is not removed by steam stripping.
(3) Free cyanide stripping efficiencies are higher than
expected. The complete run data show that for these free
cyanide stripping experiments the feeds had pH values
of nine and the bottoms had pH values of 10. From the
results in buffered solutions, we would expect a minimum
of 70% of the feed cyanide to remain in the bottoms
because the alkaline pH values suppress cyanide volatility.
The temperature effect on the pH of ammonium sulfide solutions, which
was discussed in Section IV, is probably responsible for the high stripping
efficiency of cyanide in the artificial feeds. It is quite likely that
these feeds, which showed a pH of about 9 when measured at room tempera-
ture, became more acidic when raised to stripping column temperatures.
The pH in the top of the column was probably very close to 7. At this
73
-------
60
50
I
O
O
O
v>
O
O
a
40
30
20
10
0.4
_~ £
I * ,
NH3-N
(>•••* O Fe (CN) = - CN"
6
0.6
0.8 1.0 1.2 1.4 1.6
TOTAL STEAM RATE Ib/gal (SOUR WATER BASIS)
1.8
SA-5015-27
Figure 19. Steam stripping of cyanide compounds from
artificial sour waters using a 10-plate
column.
74
-------
pH substantial cyanide stripping would occur. The buffered solutions
would show far less pH change with temperature, than the artificial
feeds. The cyanide solution buffered to pH 9 would remain alkaline at
column temperatures and little cyanide stripping would occur.
During runs when either free or complexed cyanide was present, some
cyanide was converted to thiocyanate in the stripping column. This
probably means that our precautions to eliminate sulfide oxidation were
not entirely successful since the chemical route to thiocyanate is
through polysulfide, which is an oxidation product of hydrogen sulfide.
The conversion to thiocyanate was not sufficient to explain the higher
than expected cyanide removals.
On the basis of results from artificial feeds, we would expect the
following results for an actual stripper operating on real sour water:
(1) Good free cyanide removal efficiency if bottoms pH does
not exceed 9.
(2) Little or no removal of metal-complexed cyanides.
(3) Some conversion of feed cyanides to thiocyanate. This
will be a strong possibility if polysulfides are used in
the sour water system to control corrosion, or if the
sour water is exposed to oxygen.
75
-------
BIBLIOGRAPHY
American Public Health Association (1976). Standard Methods for
The Examination of Water and Wastewater. Fourteenth Edition
(Washington, B.C.)-
Bartlett, J. K., and D. A. Skoog (1954). "Colorimetric Determination of
Elemental Sulfur in Hydrocarbons," Analytical Chemistry, ^6 (6),
1008-1011.
Bechtel Corporation (1976). Sour Water Stripping Project, American
Petroleum Institute publication 946.
Beychok, M. R. (1967). Aqueous Wastes from Petroleum and Petrochemical
Plants (John Wiley & Sons, London).
Chen, K. Y., and S. K. Gupta (1971). "Formation of Polysulfides in
Aqueous Solution," Environmental Letters 4;(3), 187.
Chen, K. Y., and J. C. Morris (1972). "Kinetics of Oxidation of
Aqueous Sulfide by 02," Env. Sci. Tech. j5(6) , 529.
Collins, F. C. and V. Lantz (1946). Industrial and Engineering Chem.
.18(11), 673.
Dobrzanski, L. T., and W. J. Thompson (1974). "Performing Evaluation of
Sour Water Strippers," 76th Annual AIChE Meeting, Tulsa, Oklahoma,
(March 11-13, 1974).
Environmental Protection Agency (1976). Manual of Methods for
Chemical Analysis of Water and Wastes, Second Edition
(Cincinnati, Ohio).
Hoffman, E. R., and R. K. Fidler (1976). "Clean Up Urea Plant Effluent,"
Hydrocarbon Processing, August, p. 111.
Page, F. M. (1953). "The Dissociation Constants of Thiosulfuric Acid,"
J. Chem. Soc. (London), 1719-1724.
Smith, B. D. (1963). Design of Equilibrium Stage Processes,
Chapter 8 (McGraw Hill Co., New York).
Urban, P. J. (1961). "Colorimetry of Sulfur Anions," Z. Analyt. Chem.
179, 415; Z Analyt. Chem. 179. 422; Z Analyt. Chem. 180. 110.
Waters Associates (1976). "Paired Ion Chromatography:
An Alternative to Ion-Exchange," Application Note F61 (May).
76
-------
West, P. W., and G. C. Goeke (1956). "Fixation of Sulfur Dioxide as
Disulfitomercurate (11) and Subsequent Colormetric Estimation,"
Analyt. Chem. 28(12), 1816-19.
Wilson, G. M. (1976). "Preliminary Report: Sour Water Strippers,
Computer Program for Vapor-Liquid Equilibrium Calculations,"
American Petroleum Institute.
77
-------
APPENDIX A
ANALYTICAL METHODS
SULFUR COMPOUNDS
The sulfur compounds measured were free sulfur, total sulfur,
thiosulfate and polythionates, sulfur dioxide, thiocyanate, and hydrogen
sulfide. The sulfide analyses are discussed in the body of the report.
Separate determinations of sulfate and polysulfide were not made because
a reproducible method could not be developed for sour waters within the
time constraints of the project.
Free Sulfur
Free sulfur has a very low solubility in water, so if a significant
amount is present it will be as particles. The sample is filtered
(which removes all interference from dissolved sulfur compounds), and
the material on the filter is dissolved in petroleum ether. The sulfur
in the petroleum ether is determined colorimetrically by its reaction
with cyanide to form thiocyanate. The procedure is described in the
following section ("Thiosulfate and Polythionates").
Thiosulfate and Polythionates Chemistry
These compounds are detected colorimetrically with Fe after con-
version to thiocyanate. The thiosulfate reacts with cyanide at all pH
values between 4 and 14 in the presence of Cu+^. Polythionates react
with cyanide at all pH values £7 and do not require the Cu+2 catalyst.
No reactions occur when pH «4. Because of the differences in pH and
catalyst requirements, thiosulfate and polythionates can be estimated
separately. Also, since no reactions occur in highly acidic solution,
a blank prepared by changing the order of solution addition can be used
to determine the amount of thiocyanate present in the sample.
Thiosulfate
Thiosulfate can be measured without interference from polythionates
by buffering the reaction mixture to pH 5. At this pH, in the presence
of Cu"1"*",
SS03 + CN - SCN + S03 . (1)
78
-------
The SCN is determined colorimetrically by formation of a red complex
with Fe+3 from Fe(£103)3 in nitric acid. Free SCN" in solution is de-
termined by preparing another sample with the iron solution added first,
followed by the CN" and the Cu+2. Since the iron solution is very acidic,
the resultant reaction mixture has a pH «4, and only SCN~that was
originally in the solution reacts to form the red color.
Polythionate
At neutral or basic pH, all of the polythionates react with CN~ to
form SCN". For all of the polythionates except trithionate the reaction
is rapid:
SnS2 06 + (n-l)CN + 20H
(n-l)SCN~+ S203~ + 304 + H20 (2)
After this reaction proceeds to completion (12 minutes), Cu (which
causes the reaction mixture to become acidic) is added and the thiosulfate
formed by reaction (2), as well as any thiosulfate originally present in
the sample, reacts to form an additional amount of SCN." Then, correc-
tion for the thiosulfate and SCN originally present in the solution
allows an estimate of 'n1 that can be converted into polythionate (ex-
cluding trithionate) sulfur by using gravimetric factors. Alternatively,
if the correction for sample thiosulfate is not made, the procedure gives
an approximate value for total sulfur in polythionate (except trithionate)
plus thiosulfate. The approximation is an underestimate because the
804" from reaction (2) will not be counted. For samples that are mostly
thiosulfate (such as the sour waters encountered in this study), the
error is not severe.
Trithionate
Trithionate behaves exactly like the other polythionates except
that it reacts much more slowly. An overnight reaction time is required.
Trithionates can be estimated directly by comparing the results of two
determinations of polythionates: one with an overnight reaction time
and the other with only a 12-minute reaction time. Trithionates were not
detected in any of the sour waters sampled in this program.
79
-------
METHODS FOR THIOSULFATE AND THIOCYANATE
Samples are filtered through a 0.45 \jm millipore filter using
positive nitrogen pressure. This removes elemental sulfur present which
would yield high results for thiosulfate and thiocyanate. The filter
should be saved and treated with petroleum ether to dissolve the
sulfur. The resulting solution is used for free sulfur analysis.
A standard curve is prepared with each group of samples. Two
concentrations are adequate along with a reagent blank.
Thiosulfate
Standards and samples are pipetted into 150-ml beakers. After
addition of 5 ml of buffer solution, the samples are diluted to 30 ml.
The solutions are mixed thoroughly by swirling, and 5 ml of NaCN solu-
tion is added. Then 3 ml of CuCl2 solution is added immediately; the
beakers are swirled vigorously during these additions. Next, 5 ml of
Fe(N03)3 + HN03 is added rapidly with serological pipette. The GU++
and CN~ solutions are added from burettes. As quickly as possible after
the addition of the Fe4""*"4" reagent, the solutions are transferred to a
black-painted 100-ml volumetric flask and diluted to volume. These
solutions are then poured into screw-top bottles covered with aluminum
foil. The sample must be protected from light. The reagent blank is
prepared from distilled water using the same volume of reagent as for
the sample; only the order of their addition is changed. The Fe is
added first, followed by NaCN and CuCl9. The method is valid over a
2 —
concentration range of 0.0 to 20 mg 8203 .
All work should be done in a hood as HCN is evolved continuously
from these solutions.
The absorbence is measured at 460 mm.
Thiocyanate
This measurement is made by adding the Fe1'' reagent first and the
NaCN and CuCl2 reagents second as in the blank. Free thiocyanate is
measured by this procedure, and these values are subtracted from the ,
thiosulfate measurements. Absorbances can be subtracted directly to
yield a value, or concentrations can be used. A standard curve is pre-
pared from NH^SCN solution. Concentration range is from 0.05 to 1.0 mg
SCN~. No buffer is used.
80
-------
Reagents
1% (w/v) NaCN Solution. Dissolve 10 gm of NaCH in 1 liter of dis-
tilled water.
Mixed ferric nitrate-nitric acid reagent. Dissolve 300 gm of
Fe(N03)3»9H2) in distilled water, add 400 ml of (55%) HN03 solution,
and dilute to 1 liter.
0.1 M CuCl-? Solution. Dissolve 17.05 gm of CuCl2*2H20 in distilled
water and dilute to 1 liter.
Buffer Solution. 10% ammonium acetate in solution. The pH of
this solution is adjusted to 5 with ammonia or acetic acid.
METHODS FOR HYDROGEN SULFIDE
The sour water samples contain materials that interfere with the
methylene blue color method. Cadmium hydroxide is used to precipitate the
sulfide and separate it from the interferences. The solution is adjusted
to a phenolphthalein end point. Most samples are already alkaline. After
pH adjustment, 2 ml of NaOH solution and 5 ml of cadmium solution is added.
The samples are vigorously mixed and centrifuged to bring down the CdS
and Cd(OH)2 formed. The liquid is decanted from the samples and the pre-
cipitates are transferred to a 25-ml volumetric flask with 15 ml of water.
Then, 3 ml of the amine reagent is added down the side of the flask,
without stirring; 0.1 ml of the iron solution is added quickly; and the
flask is stoppered and mixed. A period of 20 minutes is allowed for color
development and the samples are read in a colorimeter at 670 ran. The
method is valid over a total H2S range of 1 (Jg in the sample.
Solutions
Mix 50 ml of concentrated H2SO^ and 30 ml of water; when the solu-
tion is cool, dissolve into it 10.5 g of n,n-dimethyl-£-phenylenediamine
oxalate. A 25-ml portion of this solution is diluted to 1 liter with
1:1 HoSOA. This solution is stable and is used for color development.
Ferric Chloride Solution—Dissolve 100 gm of FeCl3-6H20 in water and
dilute to 100 ml.
NaOH—1 gm per liter is used for pH adjustment and to make solutions
alkaline.
81
-------
Cadmium Acetate—Pis solve 3 gm of Cd^I^C^^' 2H2° in water and
dilute to 1 liter.
Standard Sulfide Solution—Dissolve 35.28 gm of Na2S'9H20 in oxygen-
free water and dilute to 1 liter. This solution is standardized by
iodometric titration.
METHODS FOR SULFUR DIOXIDE
Sulfur dioxide is measured directly in the sour water. An aliquot
of 0.5-8 ml is used depending on the concentration. The sample is added
to a 15-ml Erlenmeyer flask and water is added to increase the volume
to 8 ml. Then 1 ml of p-rosaniline solution is added. The samples are
swirled to mix and 1 ml of formaldehyde solution is added. A period of
20 minutes is allowed for color development. The sample is read against
DI water. A reagent blank is read and the absorbence is subtracted from
all sample readings. Absorbence is measured at 560 mm. Standard solu-
tions and concentrations are from 5 to 20 |jg per ml. Graduated pipettes
are used throughout.
Solutions
Standards are made up from a solution of ^2803. Solutions are
stabilized by use of sodium tetrachloromercurate.
P-Rosaniline Solution. Mix 4 ml of a 1% aqueous solution of
£-rosaniline and 6 ml of concentrated HC1 and dilute to 100 ml.
Formaldehyde Solution. Dilute 5 ml of 40% formaldehyde to 1 liter
with DI water.
Sodium Tetrachloromercurate Solution. Add 27.2 gm mercuric
chloride and 117 gm of NaCl to a 1-liter flask and dilute to volume.
METHODS FOR TOTAL SULFUR
Total sulfur is determined by an alkaline permanganate oxidation
of all sulfur species to sulfate, which is determined gravimetrically. !
The alkaline method was chosen to avoid volatilization of sulfides.
An aliquot of 50-100 ml of sour water is pipetted into a casserole
dish for concentration. Two ml of 50% NaOH is added and the sample is
evaporated in a steam bath. Then the sample is concentrated to 15 ml, and
82
-------
saturated KMnO^ is added until the purple color persists. A glass rod
is used to stir the solution. The sample is evaporated to dryness in the
steam bath and is then ignited in a muffle furnace at 600°C for 2 hours.
The casserole still contains the short glass stirring rod used to mix
the KMnC>4 and the sample.
After the ignition, the casserole is cooled and covered with a watch
glass, and concentrated HC1 is added until the brown residue is dissolved.
The solution is then digested until it is clear and pale yellow. The
watch glass is put in a vertical position, and the sample is evaporated
to dryness. The residue at dryness should be white in color. If it
is not, the dehydration from concentrated HC1 is repeated.
Concentrated HC1 (5 ml) is added to the residue, followed by 25 ml
of deionized water. This is mixed and heated to incipient boil in order
to precipitate silica that was leached from the glassware under the
alkaline conditions of the oxidations. This mixture is stirred and fil-
tered into a 400-ml beaker through Whatman 42 or equivalent filter paper.
It is then washed eight times with deionized water, after which the
silica material and filter paper are discarded. The solution is diluted
to 300 ml, and methyl orange indicator is added; color will be red. The
beaker is covered with a watch glass and heated in a steam bath until
condensation appears on the watch glass. Then 5 ml of 107. BaCl2 solution
is added and the mixture is stirred vigorously. Samples are digested
overnight on the edge of the steam bath. The mixture is then filtered
through medium-porosity Selas crucibles and washed with hot water six
times. The crucibles are ignited in a muffle furnace at 900°C for 1
hour; after which they are dessicated and weighed. The tare weight
minus the sample weight equals the weight of 83804. This weight times
the gravimetric factor 0.13735 equals the weight of the sulfur present
in the original sample.
83
-------
NITROGEN COMPOUNDS
The nitrogen compounds measured were total Kjeldahl nitrogen (TKN),
ammonia, and free and total cyanides, TKN was measured using procedures
described in Standard Methods (1976). The ammonia and cyanide methods
are discussed in the body of the report.
Total Kjeldahl Nitrogen
An unfiltered 25-ml sample of Kjeldahl nitrogen is pipetted into
50 ml of Kjeldahl digestion flash. Then 6 ml of EPA digestion solution
is added.* A blank mixture is run with 25 ml of deionized water
following the same procedure. The sample is put on a micro Kjeldahl
digestion unit and heated at full heat. If foaming occurs, the heat
should be reduced until the sample has evaporated to a minimum volume.
The sample is digested until it is clear and has no color and until 803
fumes are visible as heavy white vapor. The heat is turned off and the
sample is cooled in place. When cool, the sample is transferred to a
100-ml volumetric flask and diluted to volume. The samples are then
poured into lOoz plastic bottles with screw top lids. Each bottle contains
a magnetic stirring bar. When all samples in a series are complete to
this point, a series of standards is prepared. Concentrations of 10,
50, and 100 ppm of nitrogen, respectively, are prepared. Then 6 ml of
the digestion solution is added to each of the standards and they are
diluted to a 100-ml volume and transferred to plastic bottles. To each
of the plastic bottles is added 5 ml of a solution of 10 M NaOH and
2 N Nal. The procedure for making this solution is in the Orion NH^ spe-
cific ion book. The samples are capped tightly immediately after the base
addition. They are then swirled to mix and cooled to room temperature in
a water bath. The specific ion electrode is then calibrated with the
standards and the sample concentrations are read directly from the meter
scale. Standards are good for 8 hours when bottles are kept tightly
capped. A magnetic stirrer is'used during the measurements, and the meter
should come to a steady state in 15-20 seconds.
*
For EPA digestion solution, 267 gm of K2S04 is dissolved in 1300 ml of
distilled water, and 400 ml of concentrated 112804 is added. When the
solution is cool, sulfate solution is added and diluted to 2 liters
with water. For mercuric sulfate solution, 8 gm of red HgO is dissolved
in 50 ml of 1:5 H2S04 and the solution is diluted to 100 ml with dis-
tilled water.
84
-------
HEAVY METALS AND ALKALINE EARTH ELEMENTS
Heavy metals and alkaline earth elements were measured by a two-step
procedure. For a preliminary determination, samples were evaporated to
dryness and pyrolyzed in a muffle furnace. The residual was scanned with
spark emission spectroscopy to obtain semiquantitative estimates of the
major elements present. Detailed and quantitative estimates were obtained
by atomic absorption spectroscopy. The flame conditions and instrument
sensitivity for the elements measured are tabulated below.
Metal
Na
Ca
Mg
Cr
V
Fe
Cu
Flame
Air-
Acetylene
Nitrous
Oxide
Acetylene
Air-
Acetylene
Air-
Acetylene
Nitrous-
Oxide
Acetylene
Air-
Acetylene
Air-
Acetylene
Optimum
Working Range
(M,g/ml)
0.15-0.60
1-4
0.1-0.4
2-8
40-120
2.5-10
2-8
Sensitivity
(UK/ml)
0.003
0.021
0.003
0.055
0.88
0.062
0.04
Wavelength
(run)
589.0
422.7
285.2
357.9
318.5
248.3
324.7
85
-------
The methods used to suppress interferences with metal measurement
are tabulated below.
Metal Control of Interferences
Na Use ionization suppressant.
Ca Use ionization suppressant and release agent.
Mg Use release agent or nitrous oxide-acetylene flame.
Cr Use oxidizing air-acetylene flame or nitrous oxide-
acetylene flame.
V Enhancement of signal is stabilized by excess aluminum.
Cu Depression of signal at high Zn/Cu ratios is overcome by
lean air-acetylene or nitrous oxide-acetylene flame.
86
-------
APPENDIX B
WEAK ACID TITRATION CURVES
(REFINERIES A, D, F)
Sample Size:
Titrant Strength: O.I N
SA-5015-28
Figure B-l. Potentiometric titration of stripped sour
water from Refinery A: (sample No. A3)
87
-------
X
'o
a*
i-
cc.
i i i i
Sample Size: 500 CC
Titrant Strength: 1.0N
I I
4 5
SAMPLE pH
SA-5015-29
Figure B-2. Potentiometric titration of stripped sour
water from Refinery D.
88
-------
Sample Size: 500 CC
Titrant Strength: 1.0N
4 5
SAMPLE pH
SA-5015-30
Figure B-3. Potentiometric titration of stripped sour
water from Refinery F.
89
-------
APPENDIX C
STRIPPING OF ARTIFICIAL FEEDS CONTAINING CYANIDES
TABLE C-l. CYANIDE STRIPPING TESTS IN BUFFERED WATER
*
Steam
Stripping steam
(Ib/gal fresh feed)
Total steam
1
1.11
1.55
Run Number
2 5
0.96 0.42
1.61 0.63
6
0.41
0.64
(Ib/gal fresh feed)
Feed j.
CN (ppm)
PH
H£S (ppm)
NH3
Bottoms .
CN (ppm)
PH
Efficiency (%)
Temperatures
Reflux (°F)
Feed (°F)
9.5
7
0
0
<0.02
7
~100%
180
150
12§
98
0
0
8
9
33%
150
150
12*
7
0
0
7
7.1
42%
160
170
13§
9S
0
0
10
8.9
23%
174
172
*
Stripping steam is the amount of steam leaving the feed tray.
Total steam is the total steam fed to the column minus the steam
equivalent of column heat losses.
Measured by Hach colorimetric procedures.
Solution was 0.005 M Na H PO^, 0.005 M
^Solution was 0.005 M ^28^07.
**
See Table on p. H--2for conversion to °C.
90
-------
TABLE C-2. STEAM STRIPPING OF CYANIDE COMPOUNDS FROM ARTIFICIAL FEEDS
vo
Run Number
A
Steam
Stripping steam (Ib/gal fresh feed)
Total steam
Feed
NH3-N
H2S
CN~
SCN" as CN"
3
Fe(CN)6 as
PH
Bottoms
NH3-N
H2S
CN
SCN" as CN"
Fe(CN)f as
£ CN"
PH
Temperatures
Feed (°F)
Reflux (°F)
(Ib/gal fresh feed)
(ppm)
(ppm)
(ppm)
(ppm)
CN" (ppa)
(ppm)
(ppm)
(ppm)
(PPn»)
CN"1 (ppm)
(ppm)
1021-3
0.33
0.78
1700
1122
50
0
0
9.2
46
12
11.5
6.4
0
17.9
10.1
—
—
1022-5
0.61
1.05
1500
1496
50
0
0
9.5
<10
5
2.4
5.2
0
7.6
10.5
156
171
1022-4
1.09
1.54
1700
1356
50
0
0
9.6
<10
5
1.3
1.3
0
2.6
10.8
160
180
1028-9
0.26
0.70
-
1700
1332
0
50
0
9.8
42
4
0
49
0
49
9.9
154
155
111-11
0.60
1.06
1500
1600
0
50
0
9.8
<10
<0.1
0
50
0
50
8.7
155
163
1029-10
1.19
1.65
1600
2150
0
50
0
9.5
<10
<0.1
0
49
0
49
8.1
154
175
1027-8
0.26
0.68
1600
1519
0
0
50
9.9
38
23
—
3.6
48.
51.6
10.1
160
161
1026-6
0.72
1.19
1400
2010
0
0
50
9.5
<10
<0.2
--
6.2
43.
49.2
8.9
152
170
1027-7
1.11
1.58
1600
1402
0
0
50
9.9
<10
<0.1
—
4.1
41.
45.1
8.2
154
175
Stripping steam is the amount of steam leaving the feed tray. Fresh feed includes any water added with the
caustic. Total steam is the total steam fed to the column minus the steam equivalent of column heat losses.
Measuring by tentative ASTM Methods A and F.
-------
APPENDIX D
STRIPPING OF SOUR WATER FROM REFINERY D TO STUDY
THE EFFECT OF SUBSTITUTING CAUSTIC FOR STEAM
The data in Table D-l were analyzed by least squares techniques to
develop regressions for the percent of sulfide and ammonia remaining in
the bottoms as a function of both steam and caustic rates. The regres-
sions found were
-1 -2
%H2S-S = a± + SL2&1 + a3xl
7»NH3- N = bL + b2X1"1
where X^ is the total steam rate in Ib/gal and X2 is the caustic addi-
tion rate. The constants and their 95% confidence limits were;
aL 4.1 ± 6.5 bL 11.4 ± 6.5
a2 -8.7 ± 12.0 b2 -22.1 ± 11.9
a3 4.6 ± 5.1 b3 11.6 ± 5.2
a4 -131.5 ± 411.0 b4 -224 ± 202
a5 51414 ± 32257 b5 7052 ± 8887
Both of the regressions are significant at the 0.2% level. Since,
however, both regressions contain many coefficients whose 95% confidence
interval includes zero, a more correct treatment of the data might re-
quire a different equation form. The correlations are both defective
in that they yield negative values at some steam and caustic rates,
which results in very low ammonia and sulfide residuals. Figures D-l ,
.and D-2 show both the regression predictions and the basic data in the
region of steam and caustic rates where the correlations do not yield
negative values. In this region, the agreement between prediction
and actual data appears to be quite good. The data in these two figures
were used to generate contour maps of ammonia and sulfide removal
efficiency as a function of steam and caustic rates.
92
-------
TABLE D-l. REFINERY D: BENCH SCALE STRIPPING TO DETERMINE EFFECT OF VARYING CAUSTIC AND STEAM RATES
Run Number
at..*
Stripping ateaa (IWgal "
to
Bottom
U2S (PP»)
NHj-N (PP*>
pH
H2S remaining 7. •
HH3 remaining Z *
Temperature
Feed (°r)
Reflux (°F)
1122-2
1.01
1.37
0.0
6112
3300
8.6
22
44
7.8
0.4
1.3
158
176
1113-1.0
0.97
1.30
0.0124
5586
3300
9.0
374
2
10.9
6.5
0.1
156
178
1122-13
0.99
1.41
0.0062
6054
2900
9.1
64
2
10.3
1.1
0.1
155
178
1112-1
0.56
0.85
0.0000
5925
3600
8.9
17
65
9.0
0.3
1.9
159
173
1122-14
0.48
0.80
0.0063
5925
2800
9.1
108
9
10.3
1.9
0.3
157
160
1118-8
0.47
0.81
0.0124
5S90
3500
9.1
358
13
10.6
6.2
0.4
155
160
1118-9
0.44
0.75
0.0186
5563
3300
9.0
771
11
11.0
13.3
0.3
159
159
1117-7
0.43
0.77
0.0227
6253
--
8.8
1071
25
11.0
18.5
0.7
156
162
1119-11
0.30
0.61
0.0124
5364
3200
9.1
508
150
10.9
8.8
4.5
158
169
1119-12
0.33
0.65
0.0185
5574
3306
9.2
994
120
11.2
17.2
3.6
155
170
1115-3
0.23
0.60
0.000
5984
3700
9.2
119
900
10.1
2.1
26.7
155
160
1116-4
0.13
0.50
0.0130
5318
3500
9.2
1667
812
11.0
28.9
24.1
156
168
1116-5
0.17
0.43
0.0194
55 16
3600
9.2
1242
1200
11.0
21.4
35.5
160
154
1117-6
0.11
0.48
0.0231
6018
3600
8.7
1355
1260
10.7
23.4
37.4
155
163
Stripping tteta li the "mount of atean leaving the feed tray. Total ateam li the ateam fed to the column mlnua the itean equivalent of column heat lone*. Freeh feed
include* any water added with the cauatic.
Baaed on average Feed competition of 5791 ppn HjE.
Baaed on average feod compbittlon of 3370 ppn myv.
-------
s 5
tn
1 4
O
CD
O
Z
I 3
<
ai
I
Z
Caustic Added to Column
Top at 0.0125 Ib/gal
(Sour Water Basis)
Caustic Added to Column
Top at 0.006 Ib/gal
(Sour Water Basis)
No Caustic Added to Column
Caustic Addition
Rate-lb/gal
(Sour Water Basis)
0.5
0.6 0.7 0.8
TOTAL STEAM RATE — Ib/gal
0.9 1.0
SA-5015-16
Figure D-l. Refinery D: Ammonia removal efficiency as a function
of steam and caustic addition rate.
94
-------
O
CD
Z
z
z
I
LU
rr
n
Caustic Added to Column
Top at 0.0125 Ib/gal
(Sour Water Basis!
Caustic Added to Column
Top at 0.006 Ib/gal
(Sour Water Basis)
No Caustic Added to Column
Caustic Addition
Rate-lb/gal
(Sour Water Basis)
0.4 0.6 0.8 1.0
TOTAL STEAM RATE — Ib/gal
1.2 1.4
SA-5015-17
Figure D-2. Refinery D: Sulfide removal efficiency as a function
of steam and caustic addition rate.
95
-------
APPENDIX E
CAUSTIC ADDITION TO FREE FIXED AMMONIA
(REFINERIES A, C, AND F)
TABLE E-l. BENCH SCALE STRIPPING RUNS WITH SINGLE CAUSTIC INJECTION POINT
vo
. Caustic Injection Point
Feed
Steam*
Stripping steam (Ib/gal fresh feed)
Total steam (Ib/gal fresh feed)
Caustic |
Stolchiometrlc dose (I)
Addition rate (Ib NaOH/gal sour water)
Feed
Hj,S (ppm)
SHj-N (ppm)
pH
Bottoms
H2S (ppm)
HH3 (ppm)
PH
H2S remaining (%) 4
HHj remaining (j)
Temperatures
Feed (°F)
Reflux (°F)
Run
913-1
0.94
1.20
113
0.017
—
3250
8.6
*
--
16
10
—
0.5
168
173
Run
920-1
0.92
1.24
103
0.016
—
3200
8.3
26
40
9.6
0.8
1.2
162
160
Run
106-9
1.01
1.35
90
0.014
3428
3200
8.6
5.1
144
8.25
0.2
4.4
162
167
Run
921-2
1.15
1.48
55
0.008
—
3200
8.7
30
250
7.0
0.9
7.6
164
162
Run
914-2
1.15
1.47
50
0.008
—
3250
8.4
200
6.8
—
6.1
160
164
Middle
Run
924-4
1.28
1.63
130
0.020
3462
3550
8.5
27
18
10.7
0.9
0.6
155
166
Run
927-6
1.10
1.41
94
0.014
3550
3450
8.6
28
280
9.3
0.9
8.6
162
166
Run
107-10
0.97
1.30
88
0.013
2961
3200
8.9
15
320
8.8
0.5
9.8
160
161
Run
924-5
0.98
1.28
50
0.008
--
3550
8.5
18.5
400
7.3
0.6
12.2
168
170
Bottom
Run
921-.3
0.91
1.24
122
0.019
2784
3100
8.7
20
182
9.8
0.6
5.6
162
162
Run
106-8
1.10
1.44
98
0.015
3428
3200
8.6
17
470
9.4
0.5
14.4
156
160
Run
105-7
0.97
1.31
48
0.007
2805
3100
8.7
21
650
8.9
0.7
19.9
159
160
Runs not
Receiving
Caustic
Run
913-0
0.93
1.30
0
0
—
3250
8.6
800
6.4
—
24.5
166
170
Hun
107-11
1.02
1.37
0
0
2961
3200
8.9
0.5
880
6.5
0.0
26.9
158
168
"Stripping steam Is the amount of steam leaving the feed tray. Fresh feed Includes any water added with the caustic.
steam fed to the column minus the steam equivalent.of column heat losses.
Ammonia residual Is 640 ppm NHj-N or 45.7 meg/1. 45.7 meg/1 Is the 100% stolchlometrlc addition rate.
H2S not reported because measurement not reliable.
•(baaed on an average feed concentration of 3172 ppm H2S and 3264 pom NH3-N.
Total steam is the total
-------
TABLE E-2. REFINERY A: BENCH SCALE STRIPPING RUNS
WITH MULTIPLE CAUSTIC INJECTION POINTS
Caustic Injection Points
Feed- Feed- Feed- Middle-
Middle Middle Middle Bottom
/ Run \ I Run \ / Run \ i Run \
U013-15/ U013-14J 11012-1^) hmi-12)
*
Steam
Stripping Steam (Ib/gal fresh feed)
Total steam (Ib/gal fresh feed)
Caustic
Stoichiometric dose (%)
Addition rate (Ib NaOH/gal sour water)
Feed
H2S (ppm)
NH3-N (ppm)
PH
Bottoms
H2S (ppm)
NH3-N (ppm)
PH
H2S remaining (%)
NH3 remaining (%)
Temperatures
Feed (°F)
Reflux (°F)
1.27
1.07
94
0.0143
2727
3800
8.9
2
220
8.8
0.1
6.7
168
167
1.31
1.03
50
0.0077
2727
3700
9.0
0.5
520
7.0
0.0
15.9
165
170
1.45
1.12
52
0.0080
2493
3100
8.7
1
420
6.9
0.0
12.9
154
167
1.25
0.92
96
0.0147
3428
3600
8.9
20
420
9.6
0.6
12.9
161
168
*Stripping steam is the amount of steam leaving the feed tray. Fresh feed
includes any water added with the caustic. Total steam is the total steam
fed to the column minus the steam equivalent of column heat losses.
Ammonia residual is 640 ppm NH3-H or 45.7 me.q/1. 45.7 meq/1 is the 100%
Stoichiometric addition rate.
* Based on an average feed concentration of 3172 ppm H2S and 3264 ppm NH3-N.
97
-------
TABLE E-3. REFINERY C; BENCH SCALE STRIPPING RUNS WITH SINGLE POINT CAUSTIC INJECTION
\o
oo
Caustic Ini action
Feed
Run
. 126-11
Steam-
Stripping steam (lb/
gal fresh feed)
Total steam (Ib/gal
fresh feed)
Caustic
Stoichiometric dose (%)
Addition rate (Ib/gal
1.05
1.50
O /
^109
sour water) 0.0013
Feed
HLS (ppm)
NH.-N (ppm)
PH3
Bottoms
I^S (ppm)
NH.-N (ppm)
PH3
H_S remaining (%)-'
NH3 remaining (%)
Temperatures
Feed (°F)
Reflux (°F)
3529
1700
8.9
1.9
10
9.5
0.0
0.5
153
170
Run
126-12
1.01
1.44
90
0.0011
3529
1600
8.8
8.9
14
8.0
0.3
0.7
155
170
Run
1130-2
1.05
1.42
90
0.0011
3498
2000
8.7
7.6
16
8.7
0.2
0.8
160
173
Run
123-9
1.09
1.52
62
.0007
3182
1800
8.8
2.6
16
8.7
0.1
0.8
155
168
Points
Middle
Run
121-4
1.04
1.47
109
0.0013
3517
2000
8.5
1.7
26
9.4
0.0
1.3
155
168
Bottom
Run Run Run
122-7 122-6 121-5
1.05
1.47
90
0.0032
3471
2200
8.5
2.8
19
8.9
0.1
0.9
154
169
1.03
1.46
62
0.0007
3439
2200
8.5
2.3
29
6.4
0.1
1.5
155
167
1.05
1.46
106
Run
122-8
1.07
1.48
90
0.0013 0.
3475
2000
8.5
2.0
30
9.3
0.1
1.5
156
167
3397
2000
8.5
10.7
27
9.3
0.3
1.4
157
170
Run
121-3
1.08
1.43
61
0032 0.
3077
2000
8.7
2.5
42
9.0
0.1
2.2
165
178
Runs Not
Receiving
Caustic
Run Run
1129-1 123-10
0.95
1.32
0
0007 0
2828
2000
8.7
8.7
55
—
0.3
2.8
158
176
1.00
1.42
0
0
3405
1800
8.8
9.1
36
7.6
0.3
1.8
156
174
I/Stripping steam is the amount of steam leaving the feed tray. Total steam is the steam fed to the
column minus the steam equivalent of column heat losses. Fresh feed includes any water added with
the caustic.
^/Ammonia residual is 50 ppm NH--N or 3.57 meq/1. 3.57 meq/1 is the 100% Stoichiometric caustic addition
rate.
^/Based on an. average feed composition of 3362 ppm ELS and 1942 ppm NH--N.
-------
TABLE E-4. REFINERY F: BENCH SCALE STRIPPING RUNS WITH SINGLE CAUSTIC INJECTION POINT
Caustic Iniection Points
Steam-
Run
1223-2
Feed
Run
1229-10
Run
1228-7
Run
1223-3
Middle
Run
1227-4
Run
1229-11
Run
1228-8
Bottom
Run
1228-6
Run
1227-7
Runs Not
Receiving
Caustic
Run Run
1222-1 1229-9
Stripping steam (lb/
gal fresh feed)
Total steam (Ib/gal
fresh feed)
Caustic
Stoichiometric dose
1.03
1.46
(%)—'!]
1.04
1.45
L2 91
1.01
1.45
61
0.98
1.39
112
1.06
1.48
91
1.04
1.44
60
1.01
1.42
112
1.00
1.41
91
1.07
1.52
60
1.13
1.47
0
1.03
1.48
0
Addition rate (lb NaOH/
gal sour water)
Feed
H.S (ppm)
NH3-N (ppm)
pH
Bottoms
H-S (ppm)
NlL-N (ppm)
PH3
E S remaining (%)-3j'
Nfl remaining (%)
Temperatures
Feed (°F)
Reflux (°F)
0.0005
2968
2066
9.4
9.3
1
7.6
0.3
0.0
153
169
0.0004
3070
2400
9.1
15.2
1
9.4
0.5
0.0
154
166
0.0003
3148
2500
9.6
16.1
13
9.0
0.6
0.6
151
163
0.0005
3072
2100
9.4
8.3
1
8.9
0.3
0.0
154
161
0.0004
2952
2400
9.4
17.4
11
9.1
0.6
0.5
151
170
0.0003
3132
2400
9.1
15.8
12
8.9
0.5
0.5
154
170
0.0005
2898
2133
9.6
15.8
12
9.4
0.5
0.5
154
169
0.0004
2999
2433
9.7
15.1
15
9.5
0.5
0.6
151
164
0.0003
3034
2466
9.2
26.0
14
8.6
0.9
0.6
150
170
0
2611
2450
9.4
16.2
18
7.9
0.6
0.8
162
165
0
2653
2366
9.2
16.7
16
8.8
0.6
0.7
151
166
]./Stripping steam is the amount of steam leaving feed tray. Total steam is the steam fed to the column
minus the steam equivalent of column heat losses. Fresh feed includes any water added with the caustic.
2/Ammonia residual is 20 ppm NH--N or 1.43 meq/1. 1.43 meq/1 is the 100% Stoichiometric caustic addition
rate.
^/Based on an average feed composition of 2954 ppm H_S and 2338 Ppm NH -N.
-------
TABLE E-5. REFINERY F: BENCH SCALE STRIPPING RUNS
WITH MULTIPLE CAUSTIC INJECTION POINTS
*
Steam
Stripping Steam (Ib/gal fresh feed)
Total steam (Ib/gal fresh feed)
Caustic .
Stoichiometric dose (%)
Addition rate (Ib NaOH/gal fresh feed)
Feed
H2S (ppm)
NH3 (ppm)
PH
Bottoms
H2S (ppm)
NH-j (ppm)
PH
H2S remaining %n
NHo remaining 7,
Temperatures
Feed (°F)
Reflux (°F)
Caustic
Feed-
Middle
/ Run N ,
U3-13J (
1.10
1.57
112
0.0005 0
3116
2266
9.1
16.0
12.0
9.2
0.5
0.5
154
163
Injection
Feed-
Middle
Run \
13-13 /
1.10
1.52
91
.0004
3101
2233
9.1
16.0
10.0
9.3
0.5
0.4
154
163
Points
Feed-
Middle
/ Run v
1.10
1.45
112
0.0005
2980
2333
9.1
15.3
14.0
9.3
0.5
0.6
151
168
Stripping steam is the amount of steam leaving the feed tray. Total steam
is the steam fed to the column minus the steam equivalent of column heat
losses. Fresh feed includes any water added with the caustic.
Ammonia residual is 20 ppm NH3-N or 1.43 meq/A is the 100% Stoichiometric
caustic addition rate.
*
Based on an average feed composition of 2954 ppm HoS and 2338 ppm
100
-------
APPENDIX F
HIGH PRESSURE LIQUID CHROMATOGRAPHY OF REFINERY STRIPPED SOUR WATERS
RANGE. x2
BENZOIC
ACID
UJ
o
•z.
<
ca
oc
O
P — CRESOL
11 ppm
ELUTION VOLUME
Figure F-l. HPLC analysis of standard compounds.
101
-------
UJ
u
<
m
CC
SAMPLE DILUTION: 10:1
RANGE : X2
PHENOL
37.5 ppm
SAMPLE
INJECTION
ELUTION VOLUME
SA-5015-32
Figure F-2. HPLC analysis of stripped sour water from Refinery A:
102
-------
Range: x 16 for Phenol Peak
x 2 for Cresol Peak
PHENOL
62 ppm
p-CRESOL
23 ppm
E
c
in
HI
O
<
CD
DC
O
V3
CO
SAMPLE
INJECTION
o-CRESOL
9 ppm
ELUTION VOLUME
SA-5015-23
Figure F-3. HPLC analysis of stripped sour water from Refinery C.
103
-------
I
in
10
CN
<
UJ
CJ
CO
X
o
PHENOL
202 ppm
RANGE: x16
p-CRESOL
122 ppm
o-CRESOL
75 ppm
ELUTION VOLUME
SA-5015-33
Figure F-4. HPLC analysis of stripped sour water from Refinery D.
104
-------
E
c
8
CM
<
UJ
u
z
8
m
RANGE: x16
SAMPLE
INJECTION
PHENOL
44 ppm
p-CRESOL
14 ppm
o-CRtSOL
13 ppm
ELUTION VOLUME
SA-5015-34
Figure F-5. HPLC analysis of stripped sour water from Refinery E.
105
-------
LU
o
<
m
g
CO
as
p-CRESOL
21 ppm
' o-CRESOL
12 ppm
ELUTION VOLUME
SA-5015-35
Figure F-6. HPLC analysis of stripped sour water from Refinery F.
106
-------
6
c
Ifl
LU
O
<
CO
{£
O
ELUTION VOLUME
SA-5015-36
Figure F-7. HPLC analysis of stripped sour water from Refinery G.
107
-------
APPENDIX G
SOURCES OF SOUR WATERS IN REFINERY SAMPLES
TABLE G-l. SOURCES OF SOUR WATERS IN REFINERY SAMPLES
Flow
Refinery Sour Water Source (percent)
A Delayed coker feed fractionation 19
Delayed coker produce fractionation 26
Hydrocracker feed vacuum tower 21
Dehexanizer overhead <1
Reformer feed overhead 4
Non-phenolic waters 30
100 (58 gpm)
B Hydro cracking 100
C Cat cracking
Coking
Crude distillation
Hydrofining
Cat cracker feed hydrotreating*
Hydrocracking*
Slop tanks
Flare drums
D Cat cracking
Coking
Vacuum still
Reforming
Hydrocracking jet fuel hydrotreating
Crude distillation
Alkylation
E Crude distillation
Distillate fuel desulfurization
Coking
Cat cracking
H Cat cracking
Coking
100
37
22
17
8
8
4
4
100
17
9
45
29
100
80
20
100
(350 gpm)
(640 gpm)
(890 gpm)
(75 gpm)
*Normally sent to another stripper which was undergoing turn around at the
time of sampling of refinery stripped sour water.
108
-------
APPENDIX H
CONVERSION OF ENGLISH UNITS TO METRIC UNITS
TABLE H-l. CONVERSION OF ENGLISH UNITS TO METRIC UNITS
English Unit
Ib
gal
Ib/gal
Multiply by
0.4535
3.785
119.8
To get Metric Unit
kg
I
kg/m
Temperature Conversion
°F
160
165
170
175
180
185
190
195
200
205
210
212
°C
71.1
73.9
76.7
79.4
82.2
85.0
87.8
90.6
93.3
96.1
98.9
100.0
109
-------
TECHNICAL REPORT DATA
(Please read Instructions on the reverse before completing)
1. REPORT NO.
EPA-600/2-80-081
2.
3. RECIPIENT'S ACCESSION NO.
4. TITLE AND SUBTITLE
EVALUATION OF AMMONIA"FIXATION*COMPONENTS IN ACTUAL
REFINERY SOUR WATERS
5. REPORT DATE
May 1980 issuing date
6. PERFORMING ORGANIZATION CODE
7. AUTHOR(S)
Committee on Refinery Environmental Control
8. PERFORMING ORGANIZATION REPORT NO.
API Publication 954
9. PERFORMING ORGANIZATION NAME AND ADDRESS
American Petroleum Institute
2101 L. Street Northwest
Washington, D.C. 20037
10. PROGRAM ELEMENT NO.
1BB610 (C33B1B)
11. CONTRACT/GRANT NO.
R 804364010
12. SPONSORING AGENCY NAME AND ADDRESS
Robert S. Kerr Environmental Research Laboratory
Office of Research and Development
U.S. Environmental Protection Agency
Ada. Oklahoma 74R20
13. TYPE OF REPORT AND PERIOD COVERED
Final 3/76 - 11/77
14. SPONSORING AGENCY CODE
EPA/600/15
15. SUPPLEMENTARY NOTES
American Petroleum Institute project officer; Ron Gantz
16. ABSTRACT,,. , .- .,,..,
High ammonia concentrations (fixed ammonia) in stripped sour waters from
petroleum refining are caused by weak organic acids and both weak and strong sulfur
acids. The sulfur acids result from oxidation of sulfides present in sour water.
Fixed ammonia can be eliminated by adding its molar equivalent of caustic to the top
of the stripping column. Caustic addition does not interfere with sulfide removal.
Recommended techniques for measuring cyanide and sulfide concentrations in sour
waters are inadequate. Ammonia concentration may be overestimated when the recommended
procedure is used because ammonia can be generated from organic nitrogen compounds
which are present in refinery sour waters.
17.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
b.lDENTIFIERS/OPEN ENDED TERMS
c. COSATI Field/Group
Ammonia
Phenol
Hydrogen sulfide
Stripping
Hydrogen cyanide
Organic acids
Inorganic acids
Sour water stripping
Ammonia fixation
Sulfide oxidation
Cyanide measurement
Ammonia measurement
Sulfide measurement
7A, 7B, 7C,
13B
18. DISTRIBUTION STATEMENT
Release unlimited
19. SECURITY CLASS (This Report)
Unclassified
21. NO. OF RAGES
122
20. SECURITY CLASS (This page)
Unclassified
22. PRICE
•PA F*
U2O-1
110
U.S. GOVERNMENT HUNTING OFFICE 1980-657-146/5684
------- |