&EFA
United States
Environmental Protection
Agency
Robert S Kerr Environmental Research EPA-600/2-80-125
Laboratory May 1980
Ada OK 74820
REGION III
Research and Development
BGTIUH '
Cyanide Removal from
Refinery Wastewater
Using Powdered
Activated Carbon
-------
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-600/2-80-125
May 1980
CYANIDE REMOVAL FROM REFINERY UASTEUATER
USING POWDERED ACTIVATED CARBON
James E. Huff
Jeffrey M. Bigger
IIT Research Institute
Chicago, Illinois 60616
EPA Grant No. R804029-01
IITRI Project C6358
Project Officer
Thomas E. Short, Jr.
Source Management Branch
Robert S. Kerr Environmental Research Laboratory
Ada, Oklahoma 74820
This study was conducted
1n cooperation with the
Illinois Institute for Environmental Quality
Chicago, Illinois 60606
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 Robert S. Kerr Environmental Re-
search 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 endorse-
ment or recommendation for use.
11
-------
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 Nation' a 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.
Various effluent standards typically require that cyanide levels be kept
below concentrations in the microgram-per-liter range. Such levels are diffi-
cult to achieve in industrial operations like petroleum refining because of
daily and seasonal fluctuations in raw waste content. This report contains
data on a promising treatment technology for effecting significant cyanide re-
ductions with minimal capital investment. Development of such technology is
essential if EPA is to establish and enforce pollution control standards which
are reasonable, cost-effective, and which provide protection for the American
public.
W. C. Galegar
Director
Robert S. Kerr Environmental Research Laboratory
iii
-------
ABSTRACT
This research program was initiated to investigate the feasibility of
utilizing powdered activated carbon (PAC) and cupric chloride (CuCl2) for
removal of cyanide in refinery wastewaters.
A two-phase program, consisting of batch tests and continuous tests,
was conducted to determine the basic chemistry and cyanide removal efficiency
of the adsorption and catalytic oxidation of cyanide by PAC, and CuCl2. In
the first phase the operating variables of pH, copper dosage, mode of copper
addition, carbon dosage, and type of carbon were investigated. The results
indicated a pH near neutral (pH 6-8.5) was desirable to obtain low equilibri-
um cyanide in the aqueous phase while maintaining a low copper level. Cyanide
removal, greater than 95%, was readily achieved in the batch tests using 250
mg/H of powdered carbon and 1.0 to 1.5 mg/Jl of copper on solutions containing
0.5 mg/t iron cyanide. The most important factors in cyanide removal were
the copper concentration and carbon concentration in the solution.
Phase II was a series of continuous tests using two laboratory scale
activated sludge units and actual refinery wastewater. Both carbon and cop-
per were added to the aeration basin, and the organic removal as well as
cyanide removal performance was monitored. The results of these continuous
tests revealed that cyanide can be successfully removed through the addition
of PAC and CuCl2 into an activated sludge unit. The biological efficiency
did not indicate any detrimental effects from the copper addition (with the
exception of the first test where the aeration basin was slug-dosed). More
carbon is required than predicted in the batch tests due to the organics
competing with cyanide for the active sites on the carbon.
An economic evaluation indicated that this process requires little or no
capital expenditure and should provide many refineries with an economic ap-
proach for reducing effluent cyanide concentrations.
This program was made possible by the co-sponsorship of the U.S. Environ-
mental Protection Agency and the Illinois Institute for Environmental Quality.
This report covers a period from September, 1975 to February, 1977, and the
actual laboratory studies were completed in July, 1976.
IV
-------
CONTENTS
Page
Foreword i in-
Abstract jv
Figures vi
Tables vii
List of Abbreviations and Symbols viii
1. Introduction 1
2. Conclusions & Recommendations ! ! ! ! 2
3. Background Information 4
Cyanide generation in petroleum refineries 4
State and Federal regulations regarding cyanide. . . 12
Existing cyanide treatment methods 13
Adsorption with catalytic oxidation 14
Potential application of adsorption with catalytic
oxidation to refinery cyanide wastewater 17
4. Experimental Systems and Analytical Methods 21
Experimental set up 21
Sampling procedure 25
Analytical methods 27
5. Phase I - Experimental Results 30
Description & results of batch tests 30
Copper toxicity 33
Regression analysis 39
Oxidation of cyanide adsorbed on the carbon 43
Conclusions from Phase I 45
6. Phase II - Experimental Results 46
Test A - Control test 47
Test B - Low carbon/low copper test 49
Test C - High carbon/low carbon test 52
Test D - High carbon/High copper test 57
Test E - Lower organic loading with medium carbon/
high copper dosage 74
Summary of Phase II 77
7. Economic Implications of the PAC/CuCl2 System 80
Operating criteria for PAC/CuCl2 systems 80
Cost estimation of PAC/CuCl2 systems 82
Cost comparison of PAC vs 6AC cyanide removal
systems 82
References 38
Appendices 90
Glossary of Terms 97
-------
FIGURES
Number Page
1 Cyanide generation and disposal in refinery 5
2 Refinery A - Effluent total cyanide distribution 7
3 Refinery B - Effluent total cyanide distribution 8
4 Refinery A - Monthly average cyanide discharged 9
5 Refinery B - Monthly average cyanide discharged 1°
6 Iron cyanide disappearance rate 16
7 Experimental reactor 22
8 Single stage activated sludge system 24
9 Cyanide distillation apparatus 29
10 pH effect on rate of cyanide removal 31
11 Equilibrium cyanide levels as a function of copper dose 35
12 Equilibrium cyanide levels as a function of carbon concentration. . 36
13 Test C - Predicted effect of PAC on effluent cyanide levels .... 58
14 Test D - Cyanide frequency distributions 60
15 Effect of powdered carbon on BOD5 removal - Test D 61
16 Test D - Effluent cyanide levels with time 62
17 Test D - Inlet cyanide level with time 63
18 Test D - Cyanide frequency distributions - By period. 67
19 Test D - Effluent BOD5 with time 68
20 Test D - Inlet BOD5 with time 69
21 Test D - Inlet and effluent TOC with time 70
22 Test E - Effluent cyanide levels with time. 75
23 Test E - Effluent cyanide frequency distribution 78
24 Effect of carbon dose on apparent loading 81
Vi
-------
TABLES
Number Page
1 Summary of Refinery Cyanide Levels ^
2 Summary of State Standards for Cyanide IJ
3 Results of pH Effect on Ferricyanide Removal 32
4 Filtrate Copper Levels, mg/fc 32
5 Mode of Copper Addition, 1.5 mg/fc Initial Copper Dose 33
6 Mode of Copper Addition, 0.5 mg/A Initial Copper Dose 33
7 Effect of Carbon Type on Ferricyanide Removal 3Z
8 Filtrate Copper Levels 38
9 Summary of Regression Analyses 41
10 Batch Tests - Cyanide Decay Results 44
11 Test A - Summary Results 48
12 Test A - Statistical Results 50
13 Test B - Summary Results 51
14 Test B - Statistical Results 53
15 Test C - Summary Table 55
16 Test C - Statistical Results 5$
17 Test D - Summary Table 64
18 Test D - By Periods - Summary Table 65
19 Test D - Statistical Results 71
20 Cyanide Removal Across the Control Reactor 72
21 Apparent Carbon Loadings 73
22 Test E - Summary Table 76
23 Chemical Cost Summary 83
24 Cyanide Treatment - Chemical Cost Comparison 85
25 Cost Estimates for PAC/CuCl2 System 86
Vli
-------
LIST OF ABBREVIATIONS AND SYMBOLS
ABBREVIATIONS
API
BOD
COD
GAC
6PM
MGD
ML
MLSS
PAC
PPM
R2
SS
TOC
TSS
American Petroleum Institute
bi ochemi cal oxygen demand
chemical oxygen demand
granular activated carbon
gallons per minute
million gallons per day
milligrams per liter
mixed liquor
mixed liquor suspended solids
powdered activated carbon
parts per million
coefficient of determination for linear regression
suspended solids
total organic carbon
total suspended solids
SYMBOLS
CN-
CuCl2
H2S
ftl
n
s
t
Yi
cyanide
cupric chloride
hydrogen sulfide
expected difference in population means
unbiased standard deviation of MD
mean difference
degrees of freedom
sample size
standard deviation
't1 statistic with n-1 degrees of freedom
observed value of the i th sample
viii
-------
SECTION 1
INTRODUCTION
The operation of nearly all modern petroleum refineries results in the
discharge of cyanides to the receiving streams. The concentration of refin-
eries and other cyanide discharging industries along selected waterways com-
pounds the potential toxic effects in certain waterways. Various cyanide
water quality standards from 0.005 mg/5, to 0.2 mg/£ have been proposed or
adopted by state and federal agencies. A few state agencies have also adopted
effluent standards of 0.025 and 0.05 mg/X, total cyanide to assist in maintain-
ing water quality.
The formation of cyanide in an inherent result in the production of gas-
oline components where the catalytic cracking process is employed. The
cyanide produced in the cracking process ends up in the "sour water" system which-
is sent to a steam or natural gas stripping column. Some of the cyanide is
removed in the stripper, while the remaining cyanide wastewater is eventually
discharged to the sewer. Effluent cyanide levels from 0.020 mg/£ to 0.2 mg/£
are typical in the industry. Large fluctuations, both in the daily and sea-
sonal cyanide levels, makes control of the problem that much more difficult.
Other than the stripping operation and maximizing water reuse, only
minimal cyanide treatment is presently practiced in the refining industry.
Existing treatment methods for cyanide are characterized by energy-intensive
and expensive processes. A potential low-cost cyanide treatment method for
petroleum refinery wastewaters was evaluated in this study. The method in-
vestigated can be classified as an adsorption/oxidation process and utilizes
powdered activated carbon, copper salts, and dissolved oxygen. The potential
application point selected for this study was the biological activated sludge
units common in the refining industry for secondary wastewater treatment.
Utilization of the activated sludge units minimizes the capital investment
requirements and also provides biological regeneration of the carbon, thus
minimizing the operating expenses.
-------
SECTION 2
CONCLUSIONS AND RECOMMENDATIONS
1. In the absence of dissolved organic material, inlet cyanide concen-
trations of 0.5 mg/fc can be reduced to below 0.025 mg/fc with the addition of
powdered activated carbon (PAC), cupric chloride (CuCl2)» and dissolved oxygen.
2. The presence of cupric ions in solution greatly enhances cyanide
adsorption. As demonstrated in the batch tests, overall cyanide removal in-
creased from 65% to 99% when the copper level was raised from 0.5 mg/JL to
1.5 mg/JL
3. The cyanide level in refinery effluents can be reduced from current
levels through the use of PAC and CuCl2.
4. For most refineries the most economical location of a PAC/CuCl2
process is the refinery wastewater activated sludge units.
5. When there is a high organic loading in the wastewater, competition
for the adsorption sites occurs between the organics and the cyanide. As the
organic quantity increases, the amount of carbon required becomes greater.
The most economically favorable location for a PAC/CuCl2 system is the second
stage of a biological system or in a stream of low organic loading.
6. The economics of PAC/CuCl2 systems vary according to refinery flow,
cyanide level, and organic loading. For example, the costs estimated for four
refineries with single-stage activated sludge units range from $46,000 to
$500,000 per year. The location of this cyanide removal process in the
second stage of an activated sludge unit could reduce the costs to $20,000 to
$150,000 per year for the same refineries. Clearly, the second stage, where
available, is a more economical location for the PAC/CuCl2 process.
7. For refineries that currently employ the activated sludge process,
little or no capital expenditure is required to use PAC/CuCl2 for cyanide
removal. Thus the flexibility of this process is enhanced. PAC/CuCl2 can be
used for short term cyanide peaks in the effluent attributed to refinery turn-
arounds or operating changes in the catalytic cracking unit.
8. The experimental results indicate that cyanide removal is possible
with a PAC/CuCl2 process; however, the effectiveness of long term treatment
is uncertain at this time. The effect of fluctuating organic loadings and
the stability of the treatment system over an extended time period need to
be better defined for satisfactory application of this process. Sufficient
-------
data have been gathered to develop a full scale study in a refinery for deli-
neation of PAC's operational feasibility. Especially since no capital invest-
ment is required, the success of the PAC/CuCl2 process as a practical tool for
refineries can be readily demonstrated and should be considered as an impor-
tant sequel to this study.
-------
SECTION 3
BACKGROUND
Before presenting the results from this study, background information
related to the cyanide problem is provided. In this section a description of
the source of cyanide in refineries and existing state and federal laws is
included, along with the potential advantages and limitations of the proposed
treatment scheme. This background material delineates the need for an econ-
omical cyanide treatment method and the potential of the treatment scheme is
investigated in this study.
CYANIDE GENERATION IN PETROLEUM REFINERIES
Cyanides are present in the effluents of nearly all modern petroleum
refineries in the United States'. The principal source of cyanide in a refin-
ery is the catalytic cracking unit, with some contribution from the coking
unit. Over 92% of the U.S. refineries utilize the cracking process to break
heavy petroleum fractions into principally gasoline components. Two types of
cracking units are common in the industry, thermal and catalytic. As the name
implies, thermal cracking requires high temperatures (900-1100°F) and high
pressures (600-1000 psig) to break down the heavy petroleum fractions. Cata-
lytic cracking requires lower temperatures and pressures. The most common
catalytic process is the fluid catalytic cracker (FCC) unit, which contains
a powdered catalyst.
The trend during the last decade has been toward larger fluid catalytic
cracking units because of the higher gasoline yields produced. Over 57% of
the refineries in the United States use a catalytic cracking unit to meet the
high demand for gasoline (relative to fuel oils). Cyanide is a major pollu-
tant in the wastewaters from refineries employing cracking units, especially
the catalytic cracking units.
Presented in Figure 1, is the path the cyanide waste stream follows in a
"typical" refinery. As the heavy feedstocks are cracked in the reactor sec-
tion, many different compounds are formed. A large amount of nitrogen is
liberated in the form of ammonia; 'however, what appears to be a small amount
of cyanide is also generated, typically 5 to 100 ppm. From the reactor cham-
ber, all cracked material goes overhead as a gas to a distillation column,
where the cyanide and ammonia are distilled off the top of the column with
lower molecular weight organics, (butane and lighter). Because of the ex-
tremely corrosive nature of the cyanide gas, water is injected in the over-
head gas line from the fractionating column. This water is collected in one
or two accumulators or knockout drums.
-------
H20
Reactor
Catalyst
Recycle
Rgntr
I
Cooler
V-U
Vapors to
Gas Plant
Accumulator
r
Fractionating I
I Raw Oil __^ [ Column I
' Charge '
I I
j Fluid Catalytic Cracker |
Crude
Oil
I
I
Sour Water
Sour Water
from Other
Units
Sour Water Stripper Bottoms
Desalter
To
Sewer
>*-
Crude
Oil
To
Sewer
Recycle to
Units for
V.'ashwater
To
Sulfur Plant
Incinerator
Sour Water
Stripper
Figure 1. Cyanide generation and disposal in refinery.
-------
The accumulator water contains high levels of cyanide (20-200 mg/£),
phenols (500-2000 mg/£), sulfides (1,000-10,000 mgA), and ammonia (1,000-
10,000 mg/fc). This water, often termed "sour-water," is pumped to a stripping
column along with sour water from other process units. The purpose of the
sour water stripper is to remove a large portion of the pollutant load prior
to discharging to the sewer. Most strippers use steam or natural gas as the
stripping agent and send the overhead gases to an incinerator. Strippers are
typically designed to remove 99% of the H2S, and will also remove 90-98% of
the ammonia and 30-70% of the phenols. The amount of cyanide removed ranges
from 20 to 80%, depending upon the type of cyanide present and the inlet cya-
nide concentration. A 1972 American Petroleum Institute study on sour water
strippers reported an industry average cyanide removal of 37% by stripping.1
Changes in the pH, stripping rate, and other variables, have little effect on
the cyanide removal efficiency, suggesting the remaining cyanide has complexed
with some heavy metals.
From the bottom of the stripper, the water can go several places, depen-
ding upon the practices of the individual refineries. Many refineries return
part of the stripped sour water or stripper bottoms to the cracking and coker
units. The stripped sour water is also commonly used for desalting of the
crude oil and then discharged to the sewer. Another potential practice for
some refineries is to send the stripped sour water to the cooling towers.
Where reliable stripper operation exists and the trace sulfides do not inter-
fere with the corrosion inhibitors in the cooling towers, recycling the strip-
per bottoms to the cooling towers is an excellent water conservation practice.
However, most of the cyanide found in the stripped sour water eventually finds
its way to the sewer regardless of the reuse practice of the individual refin-
eries.
One of the major problems in controlling cyanide from refineries is the
large variability in the cyanide levels generated. This variability is not
only a day-to-day phenomena, but also a month-to-month problem. Shown in
Figures 2 and'3 are daily cyanide frequency distribution curves for two mod-
ern refineries in Illinois. The change in the monthly average cyanide for
these same two refineries is presented in Figures 4 and 5. The daily cyanide
distribution curves are very similar, each showing a large number of occur-
rences between 0.000 to 0.050 mq/i and a long "tail" or skewed distribution
beyond 0.25 mg/A. Thus, where standards based on daily maximum levels are
utilized, any cyanide treatment methods must be capable of handling wide
fluctuations in inlet cyanide concentrations. Figures 4 and 5, indicate that
any cyanide treatment system must also be designed to handle extended periods
of high cyanide loadings.
Typical refinery cyanide effluent levels are presented in Table 1. The
range of cyanide levels for the eight refineries included in Table 1 goes from
0.005 mg/fc to 0.176 mg/£ for the mean cyanide discharged. The maximum levels
discharged vary from 0.03 mg/fc to 0.730 mg/fc total cyanide. While cyanide is
present in all eight refineries in Table 1, the magnitude of the effluent
cyanide concentrations varies by a factor of 30 from the lowest to the high-
est discharger.
-------
56 r
u
il
U
o
32
16 -
8 -
/
/
1
1
-
^ """* -.
\
\
\
\
1
1
1
1
Total H»6 observations
1 from 7/1 /7^ - 8/29/75
, 2 observations > 0.300
| mg/£
1 Mean value = 0.07 rng/i
\ U.S. EPA test method
1 2^-hr composite samples
i ...
l
I
\
V
\
\
N
X
x
**.
^ ._
^-
*^
"^^^^_
1 ! 1 1 »-
0.025 0.05
0.10 0.15 0.20
Total Cvanide Concentration, mg
0.25
0.30
Figure 2. Refinery AEffluent total cyanide distribution.2
-------
00
56
48
40
Cfi
O
t '32
o
CJ
c
u_
o
u 24 -
_ 1
- -^
1
* (
16
s -
I
-^
\
\
V
i
i
1
I
|
1
1
1
I
| Total 213 observations
\ from 12/6/73-5/15/75
\ 5 observations > 0.350 mg/2.
\ Mean value = 0.09 tng/i
^x U.S. EPA test tnethod-a^-hr
\ composite samples
; \
" -V ^
^
~" j-^ __
0.025 0.050 0.10(1 n ,
-------
0.500-1
"-O
c
o
r-l
c
CD
O
C
Q
G)
a
r-l
C
U
0.400-
0.300-
0.200-
0.100-
0.025-
V-Hai
Dailv Maximum Effluent Standard - State of Illinois-
ill i i i 1 1 I T
"2 4 6 8 10
10 12
1973
1974
1975
Time, months
Figure 4. Refinery AMonthly average cyanide discharged.2
-------
0.25CH
tc
E
c
o
c
>
CJ
0.200-
0.150-
0.100-
0.050-
0.025
Daily Maximum Effluent Standard - State of Illinois
r
8
T
10
T T
12
1973-*
1974
Tr
ii
4 6
- 1975
-inii
8 10 12
Time, months
Figure 5. Refinery BMonthly average cyanide discharged.2
-------
TABLE 1. SUMMARY OF REFINERY CYANIDE LEVELS2
Refinery
A
B
C
D
E
F
G
H
Crude Rate
(BPSD)
173,000
154,000
78,000
92,000
102,00
245,000
87,000
37,000
Effluent
Flow Rate
(MGD)
2.6
4.2
6.5
4.7
2.0
6.8
11.4
1.2
Average
Total CN~
Discharged
(mg/Jl)
0.176
0.079
0.008
0.022
0.38
0.045
0.005
0.009
Maximum
Total CN"
Dai ly
Discharge
(mg/fc)
0.650
0.730
0.03
0.210
0.138
0.213
0.035
0.038
Average Total CM"
Discharged
pounds/day
3.8
2.8
0.4
0.9
0.6
2.6
0.5
0.1
pounds/100,000 BPD
Crude
2.2
1.8
0.5
1.0
0.6
1.1
0.6
0.3
Average %
Simple CN" / »
in Discharge
61
55
--
--
71
(*)
As measured by the "Wood River Modification" of the "Roberts-Jackson method".
-------
STATE AND FEDERAL REGULATIONS REGARDING CYANIDE
In a review of the existing state water quality and effluent standards,
several of the standards stipulated are below analytical detectability.
Table 2, which summarizes the variety of current regulations, also indicates
that many states rely on a general, all-inclusive sentence to protect waters
from toxic materials. Water quality standards relating to the bioassay tests
on fish (Louisiana, Texas, New Jersey, Mississippi) are designed specifically
to protect aquatic life.
In states where a range of values are used for different water quality
designations, drinking water is governed by the higher values of 0.1 to 0.2
mg/£ cyanide. The lower levels of 0.005 to 0.01 typically represents limita-
tions on streams which are aquatic life habitats. Only one state, New York,
recognized a different toxicity for iron cyanides. The number of states
specifying numerical values for their water quality standards is very small
compared to the number of existing regulations. Even fewer states have adopt-
ed specific effluent limitations for dischargers. Only Delaware, Illinois,
and Missouri have specified a numerical limitation on cyanide (0.05, 0.025,
and 0.05 mg/i respectively) for dischargers to their state waters.*
Thus, the states have established low values for water quality and efflu-
ent standards for cyanide in a variety of ways. While only three states have
specific effluent standards for cyanide, rigorous enforcement of the water
quality standards would require refineries in many states to reduce their
cyanide levels prior to discharging.
At the federal level, both effluent and water quality standards will be
adopted as part of the priority pollutant program. A water quality standard
of 0.005 mg/£ of cyanide was recommended in the "Redbook."3 Effluent stan-
dards, based on best available technology economically achieveable is requir-
ed by July 1, 1984.*
EXISTING CYANIDE TREATMENT METHODS
Most exiting cyanide treatment methods are very effective in removing
the simple cyanides. However, as shown previously in Table 1, refinery
wastes contain both the simple and the complex cyanides. While the complex
cyanide accounts for an average of approximately 50% of the total cyanide
discharged by refineries, the percent complex cyanide is highly variable,
ranging from 0% to 100%. Thus, any proposed treatment scheme must be capable
or removing more than just the simple cyanides from refinery wastewaters.
Common treatment methods for cyanides include the following:
Chemical oxidation
Precipitation
Ion exchange
Biological decomposition
Adsorption and catalytic oxidation
*In 1978, the State of Illinois adopted a revised cyanide effluent limitation
of 0.1 mg/£ monthly average and 0.2 mg/l daily maximum.
12
-------
TABLE 2. SUMMARY OF STATE STANDARDS FOR CYANIDE5
Water quality standard State
1. General standard of free
from toxic and harmful
materials
2. Not present greater than
1/10 96 hr TLm
3. Not present greater than
1/20 96 hr TLm
4. Not present greater than
1/10 48 hr TLM
5. 0.025 mg/H
6. None detectable
7. 0.01 - 0.02 mg/£
8. As CN < 0.01 mg/£
As Fe TCN)6£0.4 mg/A
9. 0.01 - 0.10 mg/£
10. 0.005 (free) to 0.2 mg/5,
(total)
11. 0.005 - 0.01 mg/A
12. 0.2 mg/£
Effluent Standard
1. 0.05 mg/Jl
2. 0.025 mg/H
3. No other state has specific
effluent limit on cyanide
Alabama, Arkansas, California, Colorado,
Connecticut, Delaware, Georgia, Hawaii,
Idaho, Kansas, Maine, Maryland, Massa-
chusetts, Michigan, Missouri, Montana,
Nebraska, New Mexico, North Carolina,
Rhode Island, South Carolina, Tennessee,
Utah, Vermont, Washington, Wisconsin,
Wyoming
Louisiana, Texas
New Jersey
Mississippi
Illinois, Indiana, Iowa, Kentucky,
Pennyslvania, West Virginia
Florida
Minnesota
New York
North Dakota
Ohio
Oregon
Virginia
State
Delaware, Missouri
Illinois
13
-------
Chemical oxidation, most notably alkaline chlorination, is very common in the
electroplating industries and is very effective on the simple cyanides. For
the complex cyanides, elevated temperatures, longer holding times, and more
oxidant are required.1*6 The presence of organics in refinery wastewaters
limits the practicability of chemical oxidation in the petroleum industry.
The competitive oxidation rates of the organics result in excessive chemical
consumption in order to destroy the cyanide to acceptable levels.6
Another alternative treatment method is precipitation which is limited
to high cyanide-bearing streams. This limitation is due to the solubility of
the metal cyanides formed during the precipitation reactions which is approx-
imately 4 mg/H as CN".7 Ion exchange for the removal of cyanide has limited
application in the refining industry because of the organics present in the
wastewaters. Fouling and rapid deterioration of the resin preclude the use
of ion exchange unless extensive pretreatment of the orgamcs is utilized.
While the biological wastewater treatment systems present in most refin-
eries remove some cyanide, the mechanisms involved are not well understood.
Cyanide removals by biological treatment of greater than 99% have been achiev-
ed in the laboratory utilizing acclimated seed; however, laboratory biological
treatment studies were ineffective in removing iron cyanides.
One other cyanide treatment method available is adsorption with or with-
out catalytic oxidation. Since the method investigated in this study can be
classified as an adsorption with catalytic oxidation process, the existing
technology in this area will be covered in the following section. Of the
methods discussed, chemical oxidation is the only method currently capable of
achieving cyanide levels of 0.025 to 0.1 mg/£ total cyanide. Chemical oxida-
tion of the organics present in the wastewater, however, results in very high
chemical requirements. Both the production of ozone and chlorine require
energy-intensive processes. Thus, the existing cyanide treatment methods
available to refineries are both energy and capital intensive.
If cyanide concentrations lower than those currently achieved are to be
attained, a more economical treatment needs to be developed. This study
investigates a potential low-cost method based upon adsorption with catalytic
oxidation. In the following two sections, the previous work in this area is
summarized, followed by a discussion of the potential low-cost application of
this technology to refinery wastes.
ADSORPTION UITH CATALYTIC OXIDATION
Activated carbon has been investigated over the past twelve years both
as an adsorbent and as a catalyst for the detoxification of cyanide.
Bucksteeg1* reported that passing potassium cyanide-bearing wastes over coal
or coke resulted in removals in excess of 98%. Other forms of cyanide were
also successfully removed by passing the waste over coal and coke. The mech-
anisms involved, according to Bucksteeg, were adsorption followed by catalytic
oxidation, with the coal or coke acting as the oxidation catalyst. However,
Bucksteeg concluded that removal of nickel and iron-cyanide complexes were
attributable to only adsorption and not to any oxidation.
14
-------
Activated carbon utilized as a catalyst for cyanide oxidation was patent-
ed by Kuhn in 1971.10 The process requires an alkaline pH with a mixture of
air and the cyanide-bearing waste pumped in a pulsating manner through a bed
of activated carbon. No mention is made in the patent of the type of cyanide
effectively removed, but it is probable that only simple cyanides would be
affected.
In the early seventies, the Calgon Corporation developed a cyanide detox-
ification method utilizing catalytic oxidation on granular activated carbon.11
Calgon's method is similar to Kuhn's process described previously, except that
cupric ions are added to the wastewater along with oxygen prior to passing
the cyanide-bearing waste through the granular activated carbon columns.
According to Calgon, "cupric ions are added to the water to accelerate and
increase the efficiency of the catalytic oxidation of cyanide by granular
activated carbon."11 In addition to improving the catalytic oxidation of the
cyanide, "the presence of cupric ions results in the formation of copper cya-
nides, which have a greater adsorption capacity than copper or cyanide
alone."11
Calgon's results indicated that a pH between 6.5 and 8, a constant copper
feed, and dissolved oxygen throughout the carbon bed were necessary for maxi-
mum efficiency. Adsorption isotherms by Calgon demonstrated that the capacity
of the granular carbon was limited to 2 or 3 mg of cyanide adsorbed per gram
of carbon when no copper was used. However, the addition of copper increased
the adsorption capacity of 25 mg CN~/gm carbon without any dissolved oxygen
present. When oxygen was maintained in the system, the adsorption sites were
continuously regenerated through the oxidation of the cyanide. Tests with a
100 mg/i, cyanide in the feed resulted in cyanide effluent levels of less than
0.05 mg/St and low effluent copper levels as long as the adsorption capacity
of the copper on the carbon was not exceeded.11
In another test by Calgon, ferrous iron and cyanide were added to water
and pumped through a granular activated carbon column, with high dissolved
oxygen levels. The results, shown in Figure 6, indicate that iron cyanides
can be oxidized, but at a slower rate than the copper cyanides. Calgon found
that iron cyanides were difficult to treat for the following reasons:
1. The iron cyanide complex formation competes with the
copper cyanide.
2. The iron cyanide complex is not adsorbed as well as
copper cyanide.
3. The iron cyanide complex is more stable and harder to
oxidize than the copper cyanides.
4. Iron will probably be present at low levels in most
waters.
Calgon reported the results of a two-week pilot study on a coke plant
waste, which was believed to contain iron cyanide. Over the two-week period,
the cyanide was treated from an average inlet of 28.5 mg/A to an average
15
-------
5 00-,
Column
Cu
Fe
Initial CN"
480 rag/I
500 mg/S,
400-
300-
00
E
200_
100-
Time (days)
Figure 6. Iron cyanide disappearance rate.11
Final CN
3.5
200.0
-------
effluent of 0.41 mg/fc, or an average 98.5% removal. Considerable variation in
the effluent cyanide concentration was observed, and this was attributed to
the varying iron concentration of the wastewater^11
Based upon the work of Calgon, adsorption with catalytical oxidation is
an effective method of reducing not only the simple cyanides, but potentially
the complex or iron cyanides which are found in petroleum refinery wastewaters.
Experience in this area is discussed in the following section along with the
conceptual design for the process investigated in the present study.
POTENTIAL APPLICATION OF ADSORPTION WITH CATALYTIC OXIDATION
TO REFINERY CYANIDE WASTEWATERS
The results reported by Calgon prompted one refinery to investigate ad-
sorption with catalytic oxidation on their cyanide-bearing waste streams.
The results of this study, as reported by the refinery, are presented below.
"Based upon recommendations made by Calgon Corporation, two carbon ad-
sorption pilot plants were evaluated. The first pilot plant was designed to
investigate the merits of carbon adsorption on a smaller volume, high concen-
tration cyanide bearing stream. It was conducted with the cooperation of
Calgon1s environmental engineering department. The second pilot plant, con-
ducted without direct Calgon supervision, was designed to investigate the
merits of carbon adsorption on an activated sludge clarifier effluent.
Calgon's analytic study of the first pilot plant led it to conclude that
catalytic oxidation of adsorbed cyanide in the desalter stripper bottoms using
granular activated carbon could not be achieved due to the presence of the
stable ferrocyanide complex (Fe(CN)6-") The report indicated that the ferro-
cyanide complex can be removed from the desalter stripper bottoms only if
either a very equalized wastewater flow is obtained over an extended period
of time, or a batch type treatment operation is installed to maintain proper
pH control, the iron cyanide removal being very pH dependent. The report
also emphasized that the efficiency of the catalytic oxidation cannot be
determined unless the ferrocyanide complex has been removed.
The analysis of the data from the second pilot plant program indicated
that carbon, under limited conditions, will adsorb cyanide from the final
clarifier effluent. Carbon exhaustion, however, was calculated to be approx-
imately 11.8 Ib. carbon per 1000 per gallons of water treated. Based upon a
flow rate of 3000 GPM and a total cyanide effluent concentration of 0.175
mg/A, 51,000 pounds of carbon per day would be required to remove 6.3 pounds
of cyanide per day. Further, attempts to catalytically oxidize the adsorbed
cyanide, and thus regnerate the carbon, by the presence of excess oxygen and
copper failed. Aeration actually caused an increase in carbon column cyanide
effluent concentration. It was determined, therefore, that a carbon-adsorp-
tion system could not produce sufficient cyanide removal on any practical
scale under conditions prevalent at the Union Oil, Lemont refinery."
The above results indicate that equalization and pH control are impor-
tant to the success of the system. This finding is similar to Calgon's
results on the other industrial wastes investigated.11 The increase in
17
-------
cyanide observed upon aeration suggests that the iron cyanides were being
adsorbed at a faster rate than they were being oxidized. Thus an accumulation
of cyanides in the column resulted, which, with a change in conditions, were
released into the effluent. Assuming that equalization, pH control, and a bed
of granular activated carbon adequately sized to allow sufficient time for
cyanide oxidation could be provided, the major problem then becomes one of
cost. The calculated consumption rate of 51,000 pounds per day of granular
carbon places the economics of this system in the same category, very expen-
sive, as the the high temperature chlorination and ozonation processes.
Utilizing the findings obtained to date with adsorption and catalytic
oxidation, a potential low-cost process for removing cyanides from petroleum
refineries was conceived. The process and its operational advantages are
described in this section prior to presenting the performance results in the
next chapter.
The major drawbacks with the Calgon process for refinery waste streams
are the large initial capital expenditure and the high rate of carbon exhaus-
tion because of the organic loading. The conceptualized process evaluated in
this study involves the addition of PAC and CuCl2 directly into the activated
sludge unit. The advantages of this process are a minimal capital expenditure
and potentially a lower carbon exhaustion rate because of biological regener-
ation of the carbon. There are numerous other advantages from the addition
of PAC to an activated sludge unit. If the cost of the carbon is allocated
among all of the benefits, the cost of this cyanide removal process is even
less. An excellent case history of PAC benefits was provided by Flynn and
Barry.13 A summary of the potential benefits of PAC addition to an activated
sludge plant was provided by Adams;1" and these benefits are listed below:
1. It will improve BOD and COD removals despite hydraulic and
organic overloads.
2. It will aid solids settling, decrease effluent solids, and
increased sludge solids.
3. It will adsorb dyes and toxic components that are either
not treated biologically or are poisonous to the
biological system.
4. It will prevent sludge bulking over broader food to
micro-organism ranges.
5. It will effectively increase plant capacity at little or
no additional capital investment.
6. It will provide more uniform plant operation and effluent
quality, especially during periods of widely varying
organic or hydraulic loads.
The potential advantages of the PAC/CuCl2 addition directly into the
activated sludge unit over the previous cyanide destruction methods are num-
erous. As stated previously, the capital expenditure is minimal and the
18
-------
carbon exhaustion rate should be lower, thus consuming less carbon. In addi-
tion, PAC costs approximately half of the cost of granular activated carbon
(GAC). PAC also has a greater surface area per unit weight of carbon. The
contact times between the cyanide and carbon are longer in the aeration tank
(6-12 hours) as opposed to GAC columns (15-60 minutes), although GAC columns
have a much higher carbon:wastewater ratio. In addition, the PAC remains in
the system for the time equal to the sludge age (10-25 days) which should pro-
vide a sufficiently long period for oxidation of the adsorbed cyanides. From
the refinery investigation with the GAC column, the rate of oxidation of the
cyanides may be the limiting loading factor.
Based upon the conceived process for cyanide removal, this research pro-
gram was undertaken to ascertain the potentail feasibility of the process.
There were numerous questions that had to be addressed before the overall
viability of the process could be determined. These questions were:
1. What effect will the copper have on the overall performance
of the activated sludge system?
2. Can the cyanides be reduced to effluent levels of 0.025 to
0.1 mg/& total cyanide?
3. Will the biological solids (mixed liquor suspended solids,
MLSS) and/or organics present Interfere with cyanide adsorp-
tion and/or cyanide oxidation?
4. Will the iron cyanides oxidize once adsorbed onto the carbon,
or will these compounds accumulate in the system?
5. Does the removal efficiency increase with higher initial
cyanide concentrations?
6. What fraction of the copper added to the system will be
present in the treated effluent?
7. What influence on the overall performance does each of
the following variables have:
carbon dose
cupric chloride dose
dissolved oxygen level
mode of cupric chloride addition
- PH
hydraulic residence time
sludge age
If the cyanide can not be treated to levels of 0,025 to 0.1 mg/£, and
the removal efficiency improves with increasing inlet cyanide concentrations,
then in-plant treatment may be more effective. The cyanide-bearing stripped
19
-------
sour water is the in-plant stream with the greatest potential. It could be
treated before or after the desalter, prior to discharging to the sewer.
A small activated sludge unit could be constructed, designed and operated
to maximize the removal of cyanide through the addition of PAC and CuCl2.
Directing the stripped sour water to the cooling tower and adding low concen-
trations (25-100 mg/&) of carbon and CuCl2 to the cooling water system may
also be feasible and would require minimal capital expenditure.
Until the basic chemistry is understood, the most advantageous treatment
scheme can not be determined. The experimental program described in the fol-
lowing sections was designed to provide the information needed to evaluate
the process for cyanide removal. The first phase of the program was directed
toward obtaining a basic understanding of the process. The second phase then
applied this understanding and the most promising treatment scheme to deter-
mine the overall viability of the process.
20
-------
SECTION 4
EXPERIMENTAL SYSTEMS AND ANALYTICAL METHODS
EXPERIMENTAL SET UP
The experimental development of design criteria for the use of activated
carbon and cupric ions for the removal of cyanide in an activated sludge waste
treatment system proceeded in two phases. Phase I consisted of 13 batch tests
for the determination of optimal levels of the various operational parameters.
This part of the study was conducted in batch reactors with tap water spiked
with known concentrations of cyanide and no biological activity. Phase II
tests were conducted with actual refinery wastewater in a continuous flow,
activated sludge waste treatment system. The Phase I batch tests were design-
ed to determine the effect of various parameters on cyanide removal. These
parameters included:
pH
Mode of CuCl2 addition
CuCl2 dosage
Type of carbon
Carbon concentration
Initial cyanide concentration
The reactors used for the batch tests were fabricated from plexi-glass
and had a total capacity of 12 liters, as shown in Figure 7. For Phase I the
baffle was removed and the overflow ports covered over. This provided for a
complete mixed batch reactor, the agitation being provided by the air intro-
duced through a diffuser stone located near the bottom of the reactor. The
agitation took the form of a gentle rolling of the reactor contents.
The synthetic waste utilized in the batch tests consisted of stock solu-
tions of potassium ferricyanide and ferrocyanide, (1000 mg/fc as CM"), in
distilled water. These solutions were prepared on a monthly basis. The PAC
was treated prior to these tests by first washing in distilled water, drying
in a vacuum oven, then storing in a desiccator until weighing.
The cupric ions were made up in a stock solution of CuCl2 (1000 mg/2. as
Cu) in distilled water, with 2 ml of sulfuric acid (HzSOj added per liter of
stock solution. Preparation of the reactors for the batch tests consisted of
the following procedure.
1. Charge the reactors with 11 a of tap water (pH^8.3) and
aerate for a minimum of one-half hour.
21
-------
ro
ro
Overflow ports
Figure 7. Experimental reactor.
-------
2. Adjust the pH as necessary. (NaOH or H2SOi»)
3. Pipette the required amount of cyanide from the stock solution.
(1000 mg/£ as CN~)
4. Add the CuCl2 from the stock solution (1000 mg/£ as Cu) and
the desired amount of carbon.
The timing of the tests began with the addition of the carbon and CuCl2 to the
system. The sampling and analytical methods that were used are described
later in this section.
Phase II consisted of treating actual refinery wastewater in an activated
sludge system by the addition of activated carbon and cupric ions. The single
stage system followed the pattern depicted in Figure 8 and consisted of a com-
mon feed tank for two activated sludge units that operated in parallel.
The physical equipment utilized consisted of the reactors depicted in
Figure 7. The reactor compartments were fabricated so as to maintain 8 liters
in the aeration basin and 2 liters in the clarifier section. The baffle was
adjusted for a 1.5 - 2.5 cm space between the baffle edge and the reactor
bottom. The average flow rate through the activated sludge system of 19 lit-
ers per day resulted in a detention time of approximately 10 hours in the
aeration basin and 2.5 hours in the clarifier.
To initiate biological activity for the Phase II tests, the aeration
basins were seeded with mixed liquor suspended solids obtained in part from a
refinery aerated lagoon and partially from a culture grown, with phenols as a
food supply, in the laboratory. The feed tank was a 115 liter polyethylene
tank equipped with a mixer for homogenizing the system.
The initial feed solution considered was the effluent from the American
Petroleum Institute (API) separator, which contained cyanide levels of less
than 0.05 mg/£. In order to accurately assess the potential of the process,
a higher cyanide level was desired than was found in the API separator efflu-
ent from this refinery. Initially, 42 liters of API separator effluent, 21
liters of sour water stripper bottoms, and 21 liters of tap water were charged
to the feed tank three days per week. The stripper bottoms had a cyanide
level of 1-2 mg/fc. The tap water was added to reduce the BOD and phenol con-
centrations to more "typical" refinery levels. To prevent the nutrient level
from being a limiting factor in the activated sludge units, the phosphorous
level in the feed tank was increased 2-3 mg/fc (as P). The feed composition
was modified approximately half-way through Phase II because the cyanide
removal across the control reactor was over 90% and the effluent cyanides
were thus less than 0.05 mg/X-. To increase the effluent cyanide levels the
make-up to the feed tank was changed to 21 liters of API separator effluent,
42 liters of sour water stripper bottoms, and 21 liters of tap water. This
effectively increased the inlet total cyanide level to 0.50 - 0.80 mg/fc and
a corresponding increase in the effluent cyanide concentration.
23
-------
TAP
WATER
i
i
,-\
' " "
t
AP 1
SEPARATOR
i
i
SOUR WATER
STRIPPER
BOTTOMS
i
1
1
i
J
FEED TANK
NO. 1
CUPRIC
CHLORIDE
SOLUTION
REACTOR
AIR
PAC
REACTOR!/>
y
AIR
REACTOR 1
EFFLUENT
COLLECTOR
REACTOR 2
EFFLUENT
COLLECTOR
Figure 8. Single stage activated sludge system.
24
-------
The wastewater was pumped from the feed tank to reactors 1 and 2 (control
and test reactors, respectively), via a variable speed "Master-Flex" tubing
pump with dual pump heads. The pump was controlled by a 30-minute cycle timer
set for pumping 70% of the time. The 9 minute "rest" each half-hour allowed
the pump and tubing to cool, which provided for longer tubing and pump life.
The sludge age was maintained in each reactor by wasting sludge one a
day. To waste sludge, the baffle between the aeration and clarifier section
was removed, the side walls scraped to remove any growth, and a beaker was
dipped into the reactor to remove the sludge. The sludge was poured into a
graduated cylinder for settling, and the baffle was then immediately replaced
between the clarifier and aeration section. After the solids settled in the
graduated cylinder for one-half hour, the clear supernatant was returned to
the aeration basin.
The effluent from each reactor was collected in 19 liter glass bottles
that were held in ice buckets to maintain the integrity of the samples. After
the sludge was wasted each morning, the liquid level in the clarifier was be-
low the overflow weir for 30 to 45 minutes. The composite samples were not
started each day until approximately 2 hours after the sludge was wasted, to
allow the system to equilibrate. Then, each morning, just prior to wasting
sludge the composite samples were taken for the purpose of daily analysis.
Thus, the composite samples were only 22 hour samples, with 2 hours allowed
each day for the system to settle down after the sludge was wasted.
Cupric chloride addition to the test reactor was accomplished by making
up a 300 m& solution with the desired quantity of copper in it. This solution
was then pumped into the test reactor by means of a small tubing cassette pump.
This pump was also controlled by the 30 minute cycle timer. The flow rate was
adjusted so that the CuCl2 solution was completely added during each 24 hour
period.
The last test in Phase II consisted of a two-stage activated sludge sys-
tem with cyanide removal in the second stage. The feed for the second stage
test was composed of the effluent from control reactor 1 mixed with 5 liters
per day of stripper bottoms and 15 liters per day of tap water to provide
sufficient wastewater for two second-stage units. Both the second-stage test
and control reactors were operated with a common 50 liter polyethylene feed
tank. The operation of the second-stage test was identical to the previous
tests described earlier, including flow rates, copper and carbon addition,
and sample collection.
SAMPLING PROCEDURE
This section described the procedures followed for obtaining samples.
In Phase I, or the batch tests, cyanide was the key parameter which was moni-
tored. Grab samples were taken at various time intervals and filtered. Cya-
nide analyses were then performed both on the filtrate and on the filter cake.
The filter cake was resuspended in distilled water immediately after filter-
ing and then preserved until the cyanide content was analyzed. Although three
different procedures were utilized in attempting to determine the quantity of
cyanide per unit weight of solids, none of the three proved satisfactory. The
25
-------
first technique consisted of refiltering the carbon fines from the cyanide
distilling flask after the cyanide analysis was complete. Control tests indi-
cated a loss of solids of 15-25% occurred as a result of this procedure.
Duplicate solids samples were also collected, one for cyanide analysis and one
for solids analysis. The variation in the solids level from duplicate samples
was approximately 25%. This third procedure attempted was to dry and weigh
the solids prior to cyanide analysis, but this also gave very inconsistent
results. Additional information on the measurement of the cyanide on the
carbon is included in the following section.
Phase II tests included the operation of biological waste treatment sys-
tems. The actual performances of the activated sludge units were measured by
the removal efficiences of various pollutants. All inlet samples were taken
from the feed tank as grab samples, while the effluent samples were all col-
lected from the 22-hour composite samples. The inlet and effluents were
monitored for the following pollutants:
Sample Frequency,
Parameter No. of days/week Comments
Cyanide (CN~) 7
Phenols 3 Stopped testing for phenols
midway through Phase II
Biochemical Oxygen 5
Demand (BOD5)
Total Organic Carbon 3 Filtered samples only
(TOC) during the last test
Chemical Oxygen Demand 3 Taken only during last
(COD) part of Phase II on
tered samples
Suspended Solids (SS) 3
As indicated above, only the COD and TOC samples during the last test were
filtered, all other analyses were performed on unfiltered samples.
The performance of each activated sludge unit was also monitored by
sampling for the following biological parameters.
Sampling Frequency,
Parameter No. of days/week
Mixed Liquor Suspended 3
Solids (MLSS)
Settleability 7
Through-put of treated 7
wastewater
26
-------
ANALYTICAL METHODS
The analytical methods used during this program were in accordance with
those outlined in "Methods for Chemical Analysis of Water and Wastes."15 The
cyanide test procedure was modified slightly to improve the precision at the
lower cyanide concentrations. These modifications to the cyanide procedure
are described below.
Cuprous Chloride Reagent
The EPA procedure was modified to the following:
1. Weigh 10 g of Cu2Cl2 and place in a 500 ml reagent bottle.
2. Wash by decantation four times with 250 ml portions of
dilute sulfuric acid (HaSOi*, 1+49).
3. Wash by decantation two times with 250 ml portions of
distilled water.
4. Dilute Cu2Cl2 solution to 250 ml with distilled water.
5. Add 50 ml HC1 (sp. gravity 1.19) and continue adding 1 ml
portions of the acid until solution is clear and all cyr-
stals have dissolved.
6. Add 100 ml of distilled water and enough HC1 to dissolve any
crystals that form. Dilute to 500 ml with distilled water.
Solution should be light green.
7. Store in a tightly stoppered bottle containing a few lengths
of pure copper wire.
Procedure for Cyanide Analysis
The EPA procedure is presented below, with the modifications utilized in
this study underlined.
8.1 Place 500 ml of sample, or an ali quote diluted to 500 ml in the
1 liter boiling flask. Add 50 ml of sodium hydroxide (7.1) tothe adsorbing
tube and dilute if necessary with distilled water to obtain an adequate depth
of liquid in the absorber. Connect the boiling flask, condenser, absorber
and trap in the train.
8.2 Start a slow stream of air entering the boiling flask by adjusting
the vacuum source. Adjust the vacuum so that approximately one bubble of air
per second enters the boiling flask through the air inlet tube. Caution:
The bubble rate will not remain constant after the reagents have been added
and while heat is being applied to the flask. It will be necessary to read-
just the air rate occasionally to prevent the solution in the boiling flask
from backing up into the air inlet tube.
27
-------
8.3 Slowly add 50 ml 1:1 sulfuric acid (7.5) through the air inlet tube.
Rinse the tube with distilled water and allow the airflow to mix the flask
contents for 3 min. Pour 10 ml of Cu2Cl2 reagent (7.4) into the air inlet
and wash down with a stream of water.
8.4 Heat the solution to boiling, taking care to prevent the solution
from backing up into and overflowing from the air inlet tube. Reflux for two
hours. Turn off heat and continue the air flow for at least 15 minutes.
After cooling the boiling flask, disconnect absorber and close off the vacuum
source.
8.5 Drain the solution from the absorber into a 250 ml volumetric flask
and bring up to volume with distilled water washings from the absorber tube.
8.6 Withdraw ^0 ml of the solution from the volumetric flask and trans-
fer to a _50 ml volumetric flask. Add 15 ml of sodium phosphate solution (7.6)
and 2.0 ml of Chloramine T solution (7.12) and mix. Immediately add 5.0 ml
pyridine-barbituric acid solution (7.13.1), mix and bring to mark with dis-
tilled water and mix again.
8.7 When using pyridine-barbituric acid, allow 6^ minutes for color de-
velopment than read absorbance at 578 nm in a 1.0 cm cell immediately.
8.8 - 8.11 No change in USEPA procedure.
The above modifications to the EPA procedure have been utilized by some
of the refineries in Illinois, and the refineries have found that the above
modifications yield more precise results on refinery wastewaters. Two types
of distilling glassware set-ups are recommended by the EPA test method. The
one utilized in this project is shown in Figure 9.
Other analytical equipment utilized was a Beckman Model 915 for total
organic carbon (TOC) measurements. Copper analyses were performed on a
Perkin Elmer Model 360, Atomic Adsorption Spectrophotometer.
23
-------
COOLING WATER
INLET TUBEv
HEATER-
TO LOW VACUUM
SOURCE
- ABSORBER
DISTILLING FLASK
O
Figure 9. Cyanide distillation apparatus
15
29
-------
SECTION 5
PHASE I - EXPERIMENTAL RESULTS
The initial phase of the experimental work was limited to defining the
basic chemistry of the process. A series of batch tests using tap water
spiked with potassium ferrocyanide or potassium ferricyanide were performed
to delineate the importance of the following five operating variables:
1. pH
2. copper dosage
3. mode of copper addition
4. carbon dosage
5. type of carbon
For each test, filtrate samples for cyanide analysis were collected at
times of 15 minutes, 1 hour, 6 hours, 24 hours, and for certain tests 48 hours
and/or 72 hours. Carbon samples were also periodically collected for deter-
mination of the quantity of cyanide remaining on the carbon. Of the 275 cya-
nide analyses performed during Phase I, (excluding standards), 79% of the
samples were analyzed within 24 hours, 12% were analyzed within 48 hours, and
the remaining 9% were analyzed within 72 hours. All of the filtrate samples
were analyzed withing 48 hours, thus only certain carbon samples were held 72
hours before analysis.
DESCRIPTION AND RESULTS OF BATCH TESTS
pH Effect on Cyanide Removal
The first operating variable investigated in the batch tests was pH.
Two types of cyanide, potassium ferrocyanide and potassium ferricyanide, were
utilized each at three pH levels. The results of the ferrocyanide test runs,
presented graphically in Figure 10, indicate that the rate of cyanide removal
increased with decreasing pH. The equilibrium cyanide level, which was con-
sidered to occur after 24 hours, was lower at the lower pH level. The lower
pH apparently changed the surface properties of the PAC resulting in a greater
affinity for the iron cyanides.
The results of the ferricyanide tests were found to be very similar to
the ferrocyanide tests, as expected. This second series was only carried out
for 6 hours, with the results shown in Table 3.
30
-------
0.4
0.3-
0.2 -
0.1 -
0.25
D - pH JO.2, initial CN~* 0.53 mg/f
A - pH 7.9, initial CN~ 0.48 mg/i
O - pH 2.5, initial ClT 0.4S mg/i.
Each reactor contained 250 mg/f. lignite
based carbon and 1.5 mg/£ cupric ions.
rerrocyanide, measured as CN
Time, hours (Log scale)
Figure 10. pH effect on rate of cyanide removal.
-------
TABLE 3. RESULTS OF pH EFFECT ON FERRICYANIDE REMOVAL
Time ,
hr
0
1
6
Reactor No. 1
pH 3.0
0.44
<0.01
<0.01
Reactor No. 2
pH 7.0
0.50
0.11
0.01
Reactor No. 3
pH 11.0
0.52
0.09
0.05
Thus, ferricyarn'de removal is influenced by pH in a similar manner to ferro-
cyanide; that is, the lower the pH, the more rapid is the adsorption rate,
and the higher the removal efficiency.
While the low pH is desirable for maximum cyanide removal, effluent cop-
per levels at the low pH are unacceptable. Filtrate samples collected at 6
hours for copper analysis are presented in Table 4. The low pH range of 2.5
to 3.0 resulted in effluent copper concentrations of 1.7 to 1.8 mg/&, as com-
pared to 0.02 to 0.05 mg/£ at the higher pH values.
TABLE 4. FILTRATE COPPER LEVELS, mg/fc
ferrocyanide
f err i cyanide
Reactor No. 1
pH 2.5 - 3.0
1.8
1.7
Reactor No. 2
pH 7.0 - 7.9
0.05
0.05
Reactor No. 3
pH 10.2 - 11.0
0.05
0.02
While cyanide removal is enhanced in lower pH solutions, copper levels
are not satisfactory. At near-neutral pH values, cyanide removals of 94% and
98% were obtained, while filtrate copper levels of 0.05 mg/A were also
achieved. Thus, for the remainder of Phase I, no pH adjustments were made,
and the ambient pH ranging from 8.3 to 8.4 was considered adequate for the
remaining batch tests.
Effect of Cupric Ion Concentration and Mode of Addition
on Cyanide Removal
Copper can be added by two different techniques, either directly into
the aeration basin or adsorbed onto the carbon prior to addition of the car-
bon. Two series of tests were conducted to determine if there was any sign-
ificant difference in the mode of addition. Each reactor was charged with
250 mg/& of lignite-type carbon, and ferrocyanide was used as the cyanide
source. The results are presented in Tables 5 and 6.
32
-------
TABLE 5. MODE OF COPPER ADDITION,
1.5 mg/fc INITIAL COPPER DOSE
Time,
hr
0
0.25
1.0
6.0
48
72
Cyanide concentration, mg/£
Copper
adsorbed onto
carbon
0.52
0.065
0.16
<0.01
<0.01
Copper added
directly
to basin
0.39
0.060
0.040
<0.01
0.013
Copper added
directly
to basin
0.39
0.040
0.025
0.010
<0.01
TABLE 6. MODE OF COPPER ADDITION,
0.5 mg/fc INITIAL COPPER DOSE
Time,
hr
0
0.25
1.0
6.0
48
Cyanide concentration, mg/£
Copper
adsorbed onto
carbon
0.48
0.20
0.16
0.17
0.12
Copper added
directly
to basin
0.37
0.17
0.13
0.14
0.12
33
-------
Tables 5 and 6 indicate there is little, if any, difference in cyanide
removal based upon the mode of addition. The filtrate copper levels were also
unaffected by the mode of addition. No further studies were conducted regard-
ing the mode of copper addition, and for the remainder of the tests, the cop-
per was added directly into the aeration basin.
Numerous tests were conducted with varying copper concentrations, and the
typical results are presented graphically in Figure 11. (All of the test
results are tabulated in the Appendix). Such data indicated that the copper
dose is an important factor in the cyanide removal process. For example, with
250 mg/Jl of the lignite carbon and no copper, an equilibrium cyanide level of
0.28 mg/£ was achieved; however, with 1.0 mg/£ copper added to the same system,
equilibrium cyanide levels as low as 0.010 mg/5, were achieved.
Effect of Carbon Type and Concentration on Cyanide Removal
Two types of PAC, lignin-based (Aqua Nuchar) and lignite-based (Hydro-
darco C), were evaluated. The performance of each carbon type was evaluated
throughout a range of carbon and copper concentrations. The cyanide levels
achieved with each carbon are presented in Figure 12. The lignite-based car-
bon appeared to be superior for all levels tested; however, as the copper
dosage increases, the effect of carbon type is diminshed.
Utilizing 1.0 mg/S, of the copper and ferricyanide as the new cyanide
source, the lignin and lignite carbons were compared at a dosage level of
250 mg/£ carbon. The results from this test are presented in Table 7. The
lignin carbon was found to be more effective on the ferricyanide, which is
just the opposite of the results with the ferrocyanide.
Comparing Figure 11 to Figure 12, it is apparent that the equilibrium
cyanide achieved is influenced more by the copper dosage than by the carbon
dosage. For example, a doubling in the copper level has a much greater impact
on the final cyanide concentration than does a doubling in the carbon level.
These qualitative deductions are further confirmed in the regression analysis
which follows.
The carbon dosage also influences the copper concentration found in the
filtrate, as does the initial copper dosage. While the above results suggest
that higher inital copper and lower carbon levels may be the most economical
approach for removal of cyanide, the filtrate copper level is a limiting con-
straint, which is discussed in the following section.
Effluent Copper Levels
The "acceptable" effluent copper level is an important operating con-
straint and will vary for each refinery, depending upon water usage, state
effluent standard, and water quality standards. An effluent copper level of
0.1 rng/A was assumed to be an "acceptable" level to achieve during the batch
tests, and this was used as a criterion for all filtrate samples.
The filtrate copper samples, all collected 6 hours after initiating each
batch test, are summarized in Table 8. Data for both carbon types were
34
-------
LO
in
o
en
00
c
o
H
C
O
u
c
o
u
01
-o
f-l
c
C3
>,
u
0.1--
D- 100 mg/£ lignite carbon
A- 250 mg/2. lignite carbon
Initial ferrocyanide concen-
tration, 0.5 + 0.05 mg/l as
CN~.
0.5 1.0 1.5
Initial Copper Concentration, mg/£
Figure 11. Equilibrium cyanide levels as a function of copper dose.
-------
0.5
0.4
Co
z
m
0.3- -
0.2 - -
0.1 .
O- 0.5 mg/J. Cu , lignin carbon
± i
- 0.5 mg/8. Cu , lignite carbon
ii
- 1.0 mg/J Cu , lignin carbon
1 I
A- 1-0 mg/J. Cu , lignite carbon
Initial ferrocyanide concentration
0.5 + 0.05 rag/£ as CN~
100
200
300 400 500 600
Carbon Concentration, mg/J.
700
900
1,000
Figure 12. Equilibrium cyanide levels as a function of carbon concentration.
-------
TABLE 7. EFFECT OF CARBON TYPE
ON FERRICYANIDE REMOVAL*
Time,
hr
0
0.25
1.0
6.0
24
52
72
Cyanide concentration, mg/A
Lignite carbon
0.63
0.22
0.28
0.29
0.29
0.28
0.34
Lignin Carbon
0.67
0.21
0.20
0.15
0.11
0.10
0.10
*Copper dosage 1 mg/1
Carbon dosage 250 mg/1
37
-------
combined since there was not sufficient information to distinguish any differ-
ence in copper level for the two carbons. As the carbon dosage increases, as
shown in Table 8, the effluent copper level decreases.
TABLE 8. FILTRATE COPPER LEVELS
Initial
copper
cone., rng/A
0.5
1.0
1.5
Average filtrate copper levels, mg/A
100 mg/&
carbon
0.06
0.08
0.19
250 mg/fc
carbon
0.05
0.05
0.09
1,000 mg/£
carbon
< 0.02
0.02
To maintain a low copper outlet concentration at the 1.0 and 1.5 mg/£ inlet
copper dosage, higher carbon levels are required. At the 1.5 mg/£ copper dos-
age, the 0.1 mg/fi, filtrate constraint was exceeded when 100 mg/£ carbon was
utilized. At the 250 mg/J, carbon and 1.5 mg/fc initial copper dosage, the
average filtrate copper level was below 0.1 mg/& . Thus, to insure compliance
with the 0.1 mg/£ copper constraint, 1.0 mg/£ initial copper was selected as
the maximum dosage level. As will be discussed in the following section.
1.0 mg/& copper dosage is also a limitation of this process if it is to be
utilized in the biological waste treating units.
COPPER TOXICITY
The concept of introducing a toxic substance (copper) into an activated
sludge system to enhance removals of cyanide, at first glance, does not appear
to be an acceptable technique for cyanide removal. Supervisors of refinery
wastewater treatment plants will need assurance that such a practice will not
affect the biological activity in the activated sludge units. Phase II of
this study supplies part of this assurance, but a brief review of copper
toxicity is provided to delineate the importance and aspects of the toxicity
effects of copper.
In the carbon-copper process, a uniform copper dose will be added to the
aeration basins; thus, the micro-organisms will become acclimated to the cop-
per level maintained in the system. Acclimation of the micro-organisms to
copper was shown to be important in several studies. For example, Sudo and
Aiba*6 reported that a copper dosage of 0.25 mg/fc copper added to an unaccll-
mated culture of protozoa reduced the growth rate by 50%. However, acclima-
tion for 96 hours to the copper increased the concentration at which 50%
reduction in growth rate occurred by a factor of 1.2 to 2.2
Poon and Bhayani17 also investigated the effect of slug doses of copper
on both sewage bacteria and the fungus Geotrichum candidum. The authors hypo-
thesized that metal toxicity could selectively kill off certain sewage
38
-------
organisms, permitting an overgrowth of poor settling fungus. Poon and Bhayani
found that slug doses of copper up to 25 mg/Jl did not show any obvious toxic
effects on the sewage bacteria; however, toxic effects on the fungus were ob-
served from the slug doses. Experiments performed by Lamb and Tollefson18
indicated that there can be short term impacts. Slug doses of 5 mg/H of cop-
per decreased glucose conversion rates by 75% in the first hour after which
recovery occurred. Thus, acclimation of the micro-organisms to copper appears
to reduce the manifestations of toxic effects.
The concern in utilizing the PAC-copper system will be the long-term
effects of copper addition into an activated sludge unit. Salotto, et al.,19
found no apparent effect from adding 1 mg/& copper to an activated sludge unit
over a 2-1/2 month period. At a continuous feed of 5 mg/Z over a six-month
period, the authors found the COD removal dropped from 88% (control test) at
the high organic loadings to 84%. At low organic loadings, the COD removal
efficiency dropped from 86% (control) to 67% with a 5 mg/il copper feed. It
was noted in this study that the setteability of the sludge was slightly im-
proved with the copper addition.
A follow-up study by Barth, et al.,20 concluded that, "one mg/Z in the
influent sewage is the threshold dose for the aeration phase." An increase
in effluent COD values were observed (over the control) when the influent cop-
per level was maintained at 1.2 mg/A for 60 days. This study also reported
the biological floe settled rapidly and no bulking was encountered in the
metal-fed system, whereas the control units frequently bulked. Anaerobic
digesters fed the metal-containing sludge showed no inhibitory effect, even
with total copper levels as high as 196 mg/& were measured in the sludge.
Thus, based on the literature, influent copper levels below 1 mg/& should
have no noticeable effect on the performance of activated sludge units. In
addition, the activated carbon added to the basins should adsorb the bulk of
the copper, thus reducing the quantity available to be adsorbed by the bio-
mass. As a limiting constraint for Phase II, 1 mg/& copper was considered the
maximum permissible influent dose. Based upon the batch tests reported in
Phase I, however, 1 mg/fc copper should be sufficient for high levels of cya-
nide removal.
REGRESSION ANALYSIS
*
The experimental results obtained in Tests 1 through 12 provided useful
data regarding the importance of five operating variables, pH, carbon concen-
tration and type, copper concentration, and mode of copper addition. To
quantify the effect of each variable on cyanide removal efficiencies, statis-
tical analyses were performed. Two variables, pH and the mode of copper addi-
tion, were ultimately excluded from the statistical analysis. The mode of
copper addition did not influence cyanide removal and only a limited number of
tests were performed at low and htgh pH ranges. The data set, which consisted
of 23 observations, also limited the number of variables to be tested.
39
-------
Two cyanide removal mechanisms were hypothesized using the factors con-
trolled during experimentation. These mechanisms are shown below:
[CN-]e = [C]a [Cu]6 [CNTj [CT] (1)
[CN-]e = a + a[C] + 3[Cu] + yCCN ]1 + [CT] (2)
where:
a,3,Y»o = constants
[C] = carbon concentration
[CM].. = initial cyanide concentration
[CN]g = equilibrium cyanide concentration
[Cu] = copper concentration
CT = carbon type
Eq 1 indicates that the outlet cyanide concentration is a multiplicative
function of the concentrations of inlet cyanide, copper, and carbon, plus
carbon type. In Eq 2, a linear relationship-is assumed for the range of con-
centrations considered. Each of these hypotheses was tested using multiple
regression techniques. Through statistical analysis, the effect and signifi-
cance of these operating variables was identified. These results are depicted
in Table 9.
The statistical criteria for evaluation were the t-statistic, size of
residuals, and adjusted R2 value. In Table 9, the t-statistic, which is a
measure of the significance of the variable, is found below the coefficient
in parentheses. For the 23 observations less the number of variables and at
the 5% significance level, an approximate t-value of 2.1 indicated a signifi-
cant variable. As different combinations of variables were considered, the
size of the residuals and R2 factor were monitored. The equation in which
the best fit (lowest residuals and highest adjusted R2) of the data was then
identified.
The first six rows of Table 9 illustrate linear relationship regressions
of the operating parameters. Line 4 presents the most satisfactory explana-
tion of the effects of operating variables. Line 6 is the same equation as
Line 5 in Table 9, but with the dependent variable, the cyanide concentration,
at 6 hr. Uith the exception of Line 6, all of the other regressions utilized
the equilibrium cyanide values (average of the 24, 48, and 72 hr. cyanide
samples) as the dependent variable. Line 4 presents the most satisfactory
explanation of the effects of operating variables. Initial cyanide, carbon,
and copper concentrations are very important in determining the equilibrium
cyanide concentration. The ensuing equation is a translation of the regres-
sion results of Line 4 into operating limitations.
40
-------
TABLE 9. SUMMARY OF REGRESSION ANALYSES
Dependent Variable
1. Equilibrium [CN~]
2. Equilibrium [CN~]
3. Equilibrium [CN~]
It. Equilibrium [CN~]
5. Equilibrium [CN~]
6. Six-hour CN~
Equilibrium
7. Log
[Equilibrium CN ]
8. Log
[Equilibrium CN ]
Carbon
Cone.
-0.000236
(-2.077)*
-0.000337
(-3.72)
-0.000268
(-3.16)
-0.000305
(-4.83)
-0.000272
(-4.70)
-0.000342
(-4.77)
Copper
Cone.
-0.150
(-4.93)
-0.155
(-5.09)
-0.180
(-4.78)
Ratio
Carbon/Copper
1.77 x 10~5
(1.333)
1.81 x 10"5
(1.89)
1.51 x 10~5
(1.53)
Type of
Carbon
(dummy)**
-0.0868
(-1.697)
-0.0442
(-1.22)
-0.11J 6
(-2.35)
Initial
[CN-]
0.6065
(3.080)
0.761
(4.16)
0.703
(5.04)
0.779
(6.19)
0.843
(5.39)
Log[ Carbon]
-1.558
(-5.018)
-1.353
(-4.078)
LogfCopper]
-0.403
(-3.12)
-0.363
(-2.55)
Log [Initial CN~]
2.604
(2.614)
3.632
(3.646)
Constant
0.178
(4.418)
-0.0391
(-0.306)
-0.194
(-2.05)
0.0309
(0.333)
-0.0423
(-0.589)
0.0074
(0.081)
7.9
(4.24)
6.79
(3.37)
Adjusted
R2
0.13
0.56
0.52
0.78
0.77
0.74
0.66
0.58
* t-statistic shown in ( ).
** Note: The dummy value used for lignite-based carbon was 1 and lignin-based carbon was 0. Coefficients appearing in this column are
based on these values.
-------
[CN] = 0.703[CN ]. - 0.000305[C] - 0.150[Cu] - 0.0442[CT] (3)
where:
[CN] = equilibrium cyanide concentration, mg/S,
[CN]i = initial cyanide concentration,
[C] = carbon concentration, mg/Ji
[Cu] = copper concentration, mg/£
[CT] = carbon type, lignite = 1 and lignin = 0
Eq 3 is expressed as a linear relationship of the operating variables. This
equation is appropriate only within the range of initial cyanide concentrations
tested (0.3 to 0.7 mg/£) and copper concentrations below 1.5 mg/JL Figures
11 and 12, presented earlier, did show a linear relationship between these
separate variables and cyanide over certain concentration ranges.
This analysis has shown that the effect of copper concentration, carbon
concentration, and initial cyanide concentration are all statistically signi-
ficant in influencing the equilibrium cyanide concentration. Carbon type was
not demonstrated to be a statistically significant variable at the 5% signifi-
cance level in the "best fit" equation (No. 4 in Table 9), although the t-sta-
tistic for carbon type was significant at the 5% level in Eq 7 and at the 20%
level in Eq 2. The limited number of tests with the lignin carbon could
account for the lower t-statistic associated with carbon type.
Rows 7 and 8 of Table 9 represent regressions performed assuming a rela-
tionship similar to Eq 2. The results of these analyses also support the
conclusion that carbon and copper enhance the removal of cyanide. The multi-
plicative function utilized did not explain as much of the variance in data
points as the linear function. This conclusion, however, may be due to the
limited range in which each variable was tested.
Implications of these analyses are that by increasing copper and carbon
concentrations the cyanide effluent level can be reduced. As pointed out
previously, and verified by the size of the coefficients in Table 9, the
copper has a stronger influence on the equilibrium cyanide than does carbon
concentration. From a economic point of view, this is a favorable result.
Doubling the copper dosage (say, from 0.5 to 1.0 mg/5,) is more economical
than doubling the carbon dosage (say, from 250 mg/H to 500 mg/A) , and the
resulting reduction in the equilibrium cyanide level is much greater with
the copper.
The overall statistical analysis supports the conclusions reached in the
previous section. With only 23 sets of data and testing up to five variables,
the regression results are significant. To explain 78% of the variation in
the results with the tested four variables is a strong implication that the
important operating factors have been determined. Especially considering the
scatter in the analytical procedure for cyanide, the quantitative results
42
-------
verify the importance of the initial cyanide, copper and carbon concentrations
as driving forces.
Included in the Appendix is a summary table of all batch tests performed.
OXIDATION OF CYANIDE ADSORBED ON THE CARBON
In the previous sections, the PAC-copper system was shown to be highly
effective in removing cyanide, with removals below 0.01 mg/£ achieved from
starting solutions containing approximately 0.5 mg/fc iron cyanides (as CN").
The economics of PAC/CuCl2 addition are primarily a function of the carbon
requirements. If cyanide is oxidized once adsorbed, then the carbon require-
ments are reduced and, thus, the economics enhanced. In the background sec-
tion the oxidation of iron cyanides was reported to occur, but at a slower
rate than ulher complex cyanides.
During approximately half of the batch tests an attempt was made to de-
termine the rate of cyanide oxidation. To analyze for the amount of cyanide
remaining on the carbon, the following procedure was used. The carbon samples
were obtained by filtration, then resuspended in distilled water and placed in
the cyanide distilling flask. After the analysis, the carbon was refiltered,
dried, and weighed. The cyanide adsorbed per gram of carbon was recorded.
The cyanide remaining in solution (not adsorbed) was also noted. Except for
the beginning of each test the amount of cyanide remaining in solution was
typically less than 10% of the total cyanide.
The results from these experiments are summarized in Table 10. (The data
utilized to develop Table 10 are tabulated in the Appendix.) Assuming a first
order decay reaction, the following equation was utilized to determine the
cyanide decay rate:
Coe'kt (4)
where:
CQ is the initial CN' loading
Ct is the CN~ loading at time t
t is the time in days
k is the decay rate
The decay rate, k, was computed using the least-squares fit procedure for each
test. The coefficient of determination, or R2, was also computed for each
test, and these results are included 1n Table 10. The R2 coefficients varied
considerably from test to test, and the results from four of the tests showed
an R2 value of less than 0.4, indicating considerable scatter 1n the data.
This scatter is attributed primarily to inconsistencies in individual test
results and the small sample size. The decay rates ranged from -0.01 to
-0.16 day1, with a mean of -0.06 day-1. If the four tests with R2 values
less than 0.40 are omitted, the recomputed mean decay rate 1s -0.08 day-1.
43
-------
TABLE 10. BATCH TESTS - CYANIDE DECAY RESULTS
Test
reactor no.
1-1
1-2
1-3
3-1
3-2
4-1
4-2
4-3
7-2
7-4
9-2
9-3
9-4
No. of data
points
4
4
4
5
6
7
5
6
3
3
5
6
5
Decay rate^1'
(day1)
-0.06
-0.16
-0.12
-0.08
-0.03
-0.01
-0.01
-0.05
-0.07
-0.08
-0.06
-0.07
-0.02
Intercept'2'
(C/Co)
0.90
0.96
0.96
0.83
0.97
0.95
0.96
0.96
0.97
1.00
0.90
0.96
0.80
Coefficient of
determinations, R2
0.22*
0.98
0.97
0.44
0.85
0.22*
0.16*
0.87
0.85
0.99
0.70
0.89
0.31*
kt
(i) Based on first order decay; C^ = Coe
(2) Intercept is from plotting In (C/C ) vs time.
intercept, C/CQ = CQ/CO = 1 °
* Coefficient of Determination (R2) below 0.40
For a perfect fit the
44
-------
The results of Bernardin,11 shown in Figure 6, indicated an iron-cyanide
decay rate of -0.13 day-1. However, Bernardin was working with an initial con-
centration of 500 mg/H iron cyanides as compared to the 0.5 mg/£ iron cyanide
utilized in this study. The thousand-fold difference in starting concentra-
tions could account for the difference in the observed decay rates. Perhaps
the most significant finding from these tests is that iron cyanides are des-
troyed by the PAC/CuCl2 process; however, the rate of decay appears to be
slower than that reported by Bernardin.11
CONCLUSIONS FROM PHASE I
The primary goal of Phase I was to develop a basic understanding of the
chemistry involved in the PAC-copper system for cyanide removal. The results
of the regression analysis indicate that this primary goal was achieved.
From the information collected in Phase I, the optimal location in a
refinery for treatment of the cyanides was also ascertained. Based upon the
findings in Phase I, the logical first choice for cyanide removal would be the
biological wastewater treatment (or activated sludge) unit. Cyanide treatment
at this location would require little or no capital expenditure, (unless PAC
thermal regeneration is desired). The following four possible constraints
were identified that could preclude cyanide treatment in the biological units:
1. Cyanide level below 0.1 mg/A may not be achievable because
of a concentration-limiting phenomenon.
2. A pH outside the range 6.5 - 8.5 is required for removals
of cyanide below 0.1 mg/H.
3. Copper levels in the treated effluent are greater than
0.1 mg/A.
4. The copper levels required for cyanide removal may be high
enough to be toxic to the micro-organisms.
These constraints were addressed in the Phase I research. Cyanide remov-
al does not appear to be a concentration-limiting phenomenon. In the batch
tests cyanide removals well below 0.1 mg/l were also achieved at the higher
carbon and copper dosages investigated. The experimental results also indi-
cated that pH values between 6.5 and 8.5 are satisfactory for cyanide removal.
Copper dosages up to 1.0 rng/H were effective in reducing the cyanide to below
0.01 mg/H and yet maintaining effluent (or filtrate) copper levels less than
0.1 mg/Z. Copper levels above 1.0 mg/i improved cyanide removal, but efflu-
ent copper levels also increased.
The fourth constraint regarding the potential toxic effects of copper on
the activated sludge bio-mass was partially answered by the literature review.
Confirmation of the copper toxicity literature, however, required the results
of Phase II.
45
-------
SECTION 6
PHASE II - EXPERIMENTAL RESULTS
From the results of the first phase, the PAC/CuCl2 process looked very
promising as a cyanide removal technique. There were still many unanswered
questions about the process which could not be resolved by batch tests with
synthetic wastewaters. Phase II, which utilized a continuous flow biological
treatment systems with actual refinery wastewater, was designed to provide
information pertaining to the final feasibility of the system.
The Phase I conclusion that the activated sludge system would be the most
economical treatment location for the cyanide removal system raised several
questions. First, what effect would copper have on the micro-organism? The
literature indicated that levels below 1.0 mg/Ji have little or no effect; how-
ever, none of the copper toxicity studies utilized refinery wastes. Another
area of concern was the impact of the organics present in the refinery waste-
water on the adsorption of the cyanides by carbon. Competitive adsorption
between the organics and cyanide for the active carbon sites could potentially
reduce the effectiveness of this process to an uneconomical level. Determin-
ing the answers to such questions and describing the overall feasibility of
the process were established as the goals of Phase II.
In the second phase of laboratory operation, two laboratory scale acti-
vated sludge units were set up in parallel. Reactor #1 was utilized as a
control reactor; that is, no PAC or CuCl2 was added to this reactor at any
time. With the exception of the PAC and CuCl2 addition, the control and test
reactors were operated in a like manner. In addition to monitoring the cya-
nide levels, the following parameters were measured to assess the impact of
the PAC/CuCl2 addition on the biological system.
Biochemical Oxygen Demand (BOD5)
Total Organic Carbon (TOC)
Phenols*
Suspended Solids (SS)
Chemical Oxygen Demand (COD)**
A total of five tests were carried out in Phase II. The first test was
designed as a control test, the next three tests evaluated various carbon and
copper loadings, and the last test evaluated a two-stage activated sludge pro-
cess. By investigating the different combinations of PAC and CuCl2 as
* First part of Phase II only
**Last part of Phase II only
46
-------
indicated in the test conditions summary, sufficient data could be generated
for determining the feasibility of the process.
TEST CONDITIONS SUMMARY
Test
A
B
C
D
E
Number of
Days of Test
21
19
14
41
12
Level of
CuCl2
Addition*
(nig)
0
0.5
0.5
1.0
1.0
Level of
PAC
Addition*
(ing)
0
14
39
39
25
Sludge
Age
12
12
15
15/10
15
*per liter of feed
Each of the tests is described in detail in the following sections.
These experimental results provide a general overview on the applicability
and feasibility of the PAC/CuCl2 system.
TEST A - CONTROL TEST
Refinery wastewater feed to Reactors #1 and #2 was initiated on February
8, 1976. Both reactors were initially seeded with approximately 1000 mg/Jl
mixed liquor suspended solids (MLSS). The sludge wastage rate was gradually
increased over a 10-day period until the equilibrium wastage rate was attain-
ed. The sludge age was then maintined at 12 days in both reactors during
Test A.
Test A, initiated on February 19, lasted for a 21-day interval, and
Table 11 summarizes the operating performance obtained during this period.
As would be expected, the removal performance of both reactors was very simi-
lar for all parameters. Phenol removal was very high with an average inlet
level of 41 mg/Jl and all effluent samples below 0.1 mg/£. Cyanide removal
was higher than anticipated with an average removal of greater than 90% across
both reactors. The only parameter which varied from design was feed through-
out. The flow rates of 16.0 and 16.7 I/day for Reactors 1 and 2, respectively,
were both below the design 19 I/day because of pump failures on three occa-
sions. As operating experience was gained, the frequency of the pump failures
was minimized reducing the fluctuation in flow rates.
Statistical analysis of the experimental data from Test A provided con-
firmation of the operational similarity of the control and test unit prior to
initiation of the test program. The mean cyanide values of the two units were
compared using two statistical tools, a t-test and the paired observations.
These tests also provided a measure of the comparability of the biological
47
-------
TABLE 11. TEST A - SUMMARY RESULTS
Parameter
Flow rate, I/day
Cyanide, mg/H
BOD, mg/SL
TOC, mg/£
Phenol , mg/H
Copper, mg/Z
Suspended Solids,
mg/l
Mean Values
Inlet
_
0.17 ± 0.04
250 ± 120
111
41
<0.02
25
Ef fl uent
Reactor #1
16.0 ± 5.0
0.016 ± 0.021
13 ± 5
27
<0.10
<0.02
22
Effluent
Reactor #2
16.7 ± 4.3
0.014 ± 0.016
14 ± 10
31
<0.10
<0.02
21
Number of
Data Points
21
14
16
10
4
8
11
The purpose of Test A was to demonstrate that the two reactors performed in a
similar manner prior to beginning the carbon/copper addition tests.
Sludge age - 12 days, both reactors
Length of test - 21 days (after an 11 day start-up period)
Carbon dosage - None to either reactor
Cupric Chloride dosage - None to either reactor
48
-------
activity of the two units. The first statistical test was comparing the cya-
nide levels of Reactors 1 and 2. The following t-test was used for the paired
observation tests:
Mn - E(Mn)
T = -2 - D (5)
est aMD
where:
T = is the 't1 statistic with n-1 degrees of freedom
n = is the sample size
M = Z(Y - Y. )
Y. = Observed value of the ith sample from reactor 1
E(Mg) = Expected difference in population means
est aMD = unbiased standard deviation of MD
With the aid of this test, various hypotheses were formulated and the
confidence levels of rejection and/or acceptance determined. Using the 10%
confidence level for rejecting the hypothesis, the similarity of the two units
was tested. The statistical results from Test A for cyanide and BOD5 removals
are presented in Table 12. Statistical analyses were limited to the cyanide
and BOD results because of the small number of samples for the other parame-
ters. However, from Table 12, the results indicate that the effluent cyanide
and BOD levels from the two reactors are indeed very similar. Thus, even
though there was wide dispersion around each sample mean, the performance of
the two reactors was considered to be the same based upon the statistical
analysis.
TEST B - LOW CARBON/LOW COPPER TEST
The first test for evaluating the PAC/CuCl2 system utilized a low carbon
and copper dose. Then, in subsequent tests the carbon and/or copper dosages
could be increased and the new test initiated quickly. This technique mini-
mized the time interval between tests since several complete turnovers of
solids would be necessary to flush high concentrations out of the system.
Cyanide removal through just a biological system was shown to be greater
than 90% in Test A. Therefore, the apparent cyanide loading on the activated
carbon was expected to be quite small. Apparent loading used in this context
is considered the difference in cyanide removed between the control and test
unit divided by the carbon dosage. A copper dose of 0.5 mg as Cu per liter
of feed was selected for Test B accompanied by a carbon dose of 14 mg per
49
-------
tn
o
Parameter
Effluent Cyanide levels
Effluent BOD,, levels
TABLE 12. TEST A - STATISTICAL RESULTS
Sample means rng/A
MI M,
0.0158 0.0139
13.4 14.2
Hypothesis
MI = MZ
MI = M2
Paired observations and 10% significance level utilized.
'T'
Statistic
0.45
0.63
Hypothesis is
accepted or rejected
Accepted
Accepted
-------
TABLE 13. TEST B - SUMMARY RESULTS
Parameter
Flow rate, I/day
Cyanide, mg/A
BOD, mg/A
TOC, mg/A
Phenol , mg/A
Copper, mg/£
Suspended solids,
mg/fc
Inlet
_
0.31 ± 0.13
180 ± 180
145
57
<0.02
26
Effluent
reactor
n- Control
17.9 ± 3.8
0.017 ± 0.010
16 ± 11
24
<0.10
<0.02
8
Effluent
reactor
#2-Test
18.4 ± 2.8
0.020 ± 0.010
30 ± 13
25
<0.10
0.15
15
Number
of data
Points
19
19
13
8
6
8
9
Sludge age - 12 days, both reactors
Length of test - 19 days (at start of test, slugged dosed the aeration basin
of reactor #2 with 2.5 gm Carbon plus 9.5 mg Cu.)
Carbon dosage - Reactor #2 only - 14 mg/fc
Cupric Chloride - Reactor #2 only - 0.5 mg as Cu/fc dosage
51
-------
liter of feed. The sludge age was maintained at 12 days, thus allowing a
theoretical PAC concentration of approximately 250 mg/£ in the aeration basin
of the test reactor.
At the beginning of Test B, the test reactor (#2) was slug-dosed with
2,500 mg of carbon (lignite type) and 9.5 mg of copper to bring the operating
conditions close to the design levels. Test B was then carried out over a
19 day period. A summary of performance results during Test B is presented
in Table 12. The BOD5 removal across the test reactor was 89% as compared to
94% for the control reactor. The difference in BOD5 removal efficiency was
attributed to the initial slug-dose of copper and perhaps carbon. The BODs
removal dropped off the first day in the test unit and never recovered during
this test interval. In retrospect, the carbon and copper should have been
added gradually over several days so as to not upset the system.
The results of Test B didnot indicate any substantial advantage of the
test unit over the control. Cyanide removal across the test reactor averaged
93.5% as compared to 94.5% across the control unit. Comparing the total quan-
tity of cyanide removed across both reactors (taking into account the slightly
different flow rates), yields of 5.34 mg/day for the test unit and 5.24 mg/day
for the control unit were obtained. Thus, the apparent loading on the carbon
was the following:
5.34 mg CN" 5.24 mg CN-
day day _ 0.37 mg CN"
0.250 qm Carbon ~ gm Carbon (6)
day
This apparent loading was quite low when compared to the batch test load-
ings of typically 1.5 - 2.0 mg CN~/gm Carbon. The results from Test B indi-
cate that because of the competitive adsorption between the organics and the
cyanide for the active carbon sites greater quantitites of PAC would be re-
quired to satisfactorily remove the cyanides.
By comparing the effluents from the two reactors in a statistical manner,
the effect of the PAC/CuCl2 system can be determined. In Table 14 the sta-
tistical test results are summarized. The cyanide removal efficiency of the
control and test unit appears to be the same in this test. The means are
close together, and allowing for the standard deviation and analytical preci-
sion these means cannot be considered distinctive. Thus, on a statistical
basis, no additional cyanide removal could be attributed to the PAC/CuCl2
addition. The slug dose of copper did significantly reduce biological activ-
ity, and in Table 14 the mean effluent BOD from the control is significantly
better at the 10% confidence level. Thus, toxic effects of copper occurred
when copper amounts were slug-dosed and the biological system did not recover.
TEST C - HIGH CARBON/LOW COPPER TEST
The results of Test B suggested that a higher carbon dosage would be nec-
essary to improve the cyanide removal. At the end of Test B, the carbon dos-
age was increased from 250 mg/day to 750 mg/day, or an average of 39 mg carbon
per liter of wastewater. The sludge age was also increased from 12 days to 15
52
-------
TABLE 14. TEST B - STATISTICAL RESULTS
Parameter
Effluent Cyanide levels
Effluent BOD levels
Sample mean, mg/H
MI Ma
0.0167 0.0196
16.5 30.1
Hypothesis
MI = M2
MI = M2
T
Statistic
-1.65
5.85
Hypothesis is
accepted or rejected
Accepted
Rejected
en
Paired observations and 10% significance level utilized.
-------
days to allow a longer holding time for cyanide oxidation on the carbon.
This resulted in an equilibrium carbon level in the aeration basin of approx-
imately 1000 mg/& for Test C. After a 2-day period for acclimation of the
system to the higher carbon make-up rate, Test C was initiated and continued
for a 14-day period.
The results of Test C are summarized in Table 15. Unlike the previous
test, no detrimental effects on the organic removal (BOD and TOC) were ob-
served in the test reactor. The mean effluent copper level was also lower in
Test C (0.10 mg/£) as compared to Test B (0.15 mg/£), and this was considered
the result of the higher carbon level in Test C. The PAC/CuCl2 addition re-
duced the mean effluent cyanide level over the control unit by 0.019 mg/Jl, or
28%. The apparent cyanide loading for Test C is computed below:
11.63 mg CM" 11.39 mq CN"
day day _ 0.32 mg CN"
0.750 gm Carbon ~ gm Carbon (7)
day
This apparent loading is lower than the 0.37 mg CN~/gm carbon determined
for Test B, as would be expected. (Removal efficiency increases as the load-
ing on carbon decreases in a completely mixed system.) As compared to the
loadings reported in the batch tests (1.5 - 2.0 mg CN~/gm carbon), the load-
ing in Test C indicates that competitive adsorption by the organics is a real
phenomenon that consumes most of the active sites on the carbon. However,
sufficient carbon was available for the removal of an additional 0.019 mg/S,
cyanide over the control reactor.
The statistical analyses confirm the influence of the PAC/CuCl2 addition.
In Table 16 the computed statistical results are summarized. That the PAC/
CuCl2 system improved cyanide removal over the control can be stated at the
2% significance level. At the 10% significance level, the effluent cyanide
was reduced at least an incremental 0.01 mg/fc due to the addition of the PAC/
CuCl2. Thus, the increase in carbon concentration improved the cyanide remo-
val.
Linear regression analyses were performed for the test and control units,
comparing the effluent cyanide level to the inlet cyanide concentration. The
results from both the test and control reactors were combined, and a dummy
variable utilized for the use of carbon/CuC!2. The following equation is the
result of this technique.
[Effluent CIT] = 0.14 [Inlet CN~] - 0.019 [PAC]
T-Statistic (2.84) (-2.81) (8)
where:
PAC = 1 for test reactor
= 0 for control reactor
54
-------
TABLE 15. TEST C - SUMMARY TABLE
Parameter
Flow rate, I/day
Cyanide, mg/fc
BOD, mg/H
TOC, mg/£
Phenol , mg/£
Copper, mg/A
Suspended solids, mg/H
Inlet
_
0.66 ± 0.071
380 ± 90
130
80
<0.02
10
Effluent
Reactor #1
19.2 ± 3.8
0.067 ± 0.024
27 ± 9
27
<0.10
<0.02
10
Effluent
Reactor #2
19.0 ± 3.7
0.048 ± 0.016
28 ± 7
26
<0.10
0.10
10
Sludge age - 15 days
Length of test - 14 days (after a 2 day transition from Test B)
Carbon dosage - Reactor #2 only - 39 mg/A
Cupric chloride dosage - Reactor 12 only - 0.5 mg as Cu/A
55
-------
en
Parameter
Effluent Cyanide levels
Effluent Cyanide levels
Effluent BOD levels
TABLE 16. TEST C - STATISTICAL RESULTS
Sample means, mg/H
Mi M2
0.0670 0.0478
0.0670 0.0478
26.7 27.6
Hypothesis
Mi = M2
M2 + 0.011 <
Mi = M2
Paired observations and 10% significance level utilized.
'T1
Statistic
3.01
1.28
-0.408
Hypothesis is
accepted or rejected
Rejected
Accepted
Accepted
-------
Both parameters, the inlet cyanide concentrations and the presence of the PAC/
CuCl2 are significant factors at the 1% level. The R2 (adjusted) for this
regression was 0.34, which indicates that there are other variables not includ-
ed in this regression that influence effluent cyanide concentration, and these
other variables are discussed in a latter section. Basically, this statisti-
cal regression illustrates the importance of PAC/CuCl2 in removing cyanide,
holding all other operational variables constant.
The significance of the PAC/CuCl2 is depicted graphically in Figure 13.
The lines are separated by the difference in the effluent cyanide means,
0.019 mg/A which is the most probable difference. Because of the use of the
dummy variable to indicate the effects of PAC/CuC12» the lines in Figure 13
are parallel. Not enough data were collected to discern the true relative
slopes of the two lines; although carbon adsorption theory would predict that
the lines would converge at the lower inlet cyanide concentrations.
The various statistical analyses of the results from Test C clearly show
that the addition of PAC and CuCl2 into a biological wastewater treatment
system significantly improves cyanide removal. The results also indicate that
there was no deterioration in the removal of other organics (as measured by
the BOD and TOC tests) as a result of the PAC/CuCl2 addition. The average
cyanide removal of 28% over the control reactor was achieved with a relatively
high carbon level and a low copper level. In the following test, the copper
dose was increased to ascertain if cyanide removal could be further enhanced
via more copper.
TEST D - HIGH CARBON/HIGH COPPER TEST
The final test with the single stage activated sludge units utilized a
high carbon and copper dosage to determine the long term effects of the PAC/
CuCl2 addition. The copper level for this test was increased to 1.0 mg/& for
a feed rate of 19 mg of copper per day. The carbon level was maintained at
the same level as the previous test, 750 mg per day for a dosage of 38 mg of
carbon per liter of wastewater.
The transition from Test C to Test D was accomplished by allowing the
test reactor to adjust to the higher copper feed rate for one day. After this
acclimation period Test D was initiated and lasted for 43 days. Any potential
detrimental effects of PAC/CuCl2 addition to the activated sludge system were
expected to be evident by the end of the 43-day test.
During the first 30 days of Test D the sludge age was maintained at 15
days after which it was reduced to 10 days for the remainder of the test.
The reason for the change in sludge age is included with the results.
In Test C with the lower copper dose, the PAC/CuCl2 addition reduced the
mean cyanide effluent level by 28% over the control. In Test D the copper
dose was double that utilized in Test C, which based upon the batch tests,
should have resulted in effluent cyanide levels over 50% lower than the con-
trol reactor. The results of Test D, summarized in Table 12, show that the
test reactor effluent cyanide level averaged only 9% lower than the control
57
-------
W>
-o
H
0.12 -
0.11 -
0.10 -
0.09 .
0.08 .
0.07 .
0.06 .
oi 0.05
3
r-t
M-l
w 0.04
0.03
0.02 -
0.01 -
vith PAC/CuCl2
Effluent =0.14 [Inlet CN ~] -.01 91 PAC]
Cyanide
T-Statistic(2.84)
(-2.81)
where PAC =
1 for test unit
0 for control unit
0.4
0.5
0.6
0.7
0.8
0.9
Inlet Cyanide, mg/2.
Figure 13. Test C-Predicted effect of PAC on effluent cyanide levels.
58
-------
reactor during the 43 day test. In addition to the cyanide removal being
lower than expected, the dampening effect observed in Test C in the effluent
cyanide variation as a result of the carbon addition was not apparent in this
test. The standard deviation of the test unit effluent cyanide level during
Test C was 0.016 mg/£ as compared to 0.024 mg/Jl in the control reactor. In
Test D, the standard deviations were 0.051 and 0.034 mg/fc for the test unit
and control unit effluent cyanide levels, respectively. Frequency distribu-
tions for the effluent cyanide levels, shown in Figure 14, also reflect this
greater variation in the test units's effluent.
Also contained in the summary Table 17 are the other chemical character-
istics used to describe the effluent and monitored during Test D. The BOD
removal was improved by the addition of PAC/CuCl2 by 26% over the control unit.
This effect on BOD removal is graphically depicted in Figure 15. The larger
mean effluent COD and TOC concentration from the test reactor could be attri-
buted to the carbon fines in the effluent, as all of the analyses were per-
formed on unfiltered samples. The removal of over 99% of the phenol across
both reactors continued as in previous tests.
The test reactor effluent copper level averaged 0.20 mg/A indicating that
80% of the copper was being adsorbed onto the carbon and removed from the
system in the waste sludge.
Cyanide Removal Performance
Analysis of the effluent cyanide concentrations from the two reactors
indicate that twice during Test D major performance changes occurred in the
test reactor. These changes in operating performance divide Test D into 3
intervals, which for discussion purposes, are designated Dl, D2, and D3.
The change in effluent cyanide concentration with time is presented for
the test and control reactor in Figure 16. Interestingly enough the inlet
cyanide concentrations illustrated in Figure 17 do not indicate this varia-
tion from period to period. The effluent characteristics of the reactors in
Test D are depicted by time intervals in Table 18, and the performance in
each interval is discussed herein.
The first phase, Dl, lasted for 16 days, during which time the control
reactor effluent cyanide averaged 0.073 mg/SL compared to 0.045 mg/A for the
test reactor. The PAC/CuCl2 addition thus resulted in a mean cyanide level
38% lower than the control reactor during the first 16 days of Test D. The
17th day of the test marked the beginning of a 2 week period during which the
test reactor performed poorly compared to the control unit. The control unit
effluent during this second phase, D2, averaged 18% less cyanide with a mean
effluent cyanide concentration of 0.115 mg/A while the test unit averaged
0.140 mg/Jl. After 30 days the sludge wastage rate was increased (sludge age
decreased) in an attempt to reduce the cyanide equilibrium concentration in
solution by lowering the total quantity of cyanide maintained in the aeration
tank.
59
-------
i't
12 -
U)
01
c 10-
(U
u
M
3 0
o o
u
O
^ 6 -
^ CD H-" 1 1
o ro f^
ii i i i i i
test
uni
t
Mean 0.091 tug
Std. Dev. ±.0
I 1
.025 .050 .073 .100 .125 .150.175 .200.225 .250.275
Cyanide, mg/1
Figure 14. Test D-Cyan1de frequency distributions,
60
-------
130-1
120
^ 1101
60
6
- 100-
s~\
in
Q
§ 90
-a
C
m
E
Q)
Q
CJ
bO
^
X
o
u
H
6
QJ
(U
r-l
80-
70-
60-
50-
u
.2 ^o
30
20
10
O - Control Unit
A - Test Unit,
25 50 75
Percent of Time Less Than or Equal To
100
Figure 15. Effect of powdered carbon on BOD5 removal-Test D.
61
-------
c
o
c
-,
u
0.20
0.15
0.10
0.05
o- Control
A- Test
10
.* m _._..
20 30 40 1
Time , Days
)* DJ »|
Figure 16. Test D-Effluent cyanide levels with time.
62
-------
1.0 .,
0.9 -
0.6 '
ii
e 0.7 '
2 0.6 1
u
CO
Wi
c 0.5 1
a
o
3 0.4 H
0.3
U
0.2 -
0.1
-*- Dl
10
20
D2
30 40
Time, days
Figure 17. Test D-Inlet cyanide level with time.
63
-------
TABLE 17. TEST D - SUMMARY TABLE
Parameter
Flow rate, I/day
Cyanide, mg/£
BOD, mg/£
COD, mgA1
TOC, mg/a
Phenol, mg/H
Copper, mg/£
Suspended solids,
mg/A
Inlet
0.56 ± 0.12
280 ± 110
262
111
71
<0.02
24
Effluent
reactor #1
19.4 ± 2.2
0.100 ± 0.034
38 ± 20
70
24
<0.10
<0.02
10 ± 3
Effluent
reactor #2
19.5 ± 2.3
0.091 ± 0.051
28 ± 21
84
29
<0.10
0.20
13.5 ± 10
Number
of
Samp! es
tt
41
23
9
17
4
19
19
Sludge Age - 15 days for first 30 days, sludge age 10 days for last 13 days
Length of Test - 43 days (after a 1 day transition from Test C)
Carbon dosage - Reactor #2 only - 38 mg/Jl
Cupric Chloride dosage - Reactor #2 only - 1.0 mg as Cu
1 COD samples taken only during the last 18 days of Test D.
64
-------
TABLE 18. TEST D - BY PERIODS - SUMMARY TABLE
Parameter
Cyanide, mg/H
BOD, mg/fc
TOC, mg/£
Suspended solids,
mg/A
Parameter
Cyanide, mg/&
BOD, mg/H
TOC, mg/fc
Suspended solids,
Parameter
Cyanide, mg/&
BOD, mg/fc
TOC, mg/H
#1 - Days
Inlet
0.61 ± 0.1
330
106
13
#2 - Days
Inlet
#1-16, Sludge Age -
Effluent
reactor #1
1 0.073 ± 0.022
34
24
10
#17-31, Sludge Age -
Effluent
reactor #1
0.56 ± 0.11 0.115 ± 0.018
270
141
22
#3 - Days
Inlet
0.51 ± 0.
220
74
37
28
10
#32-43, Sludge Age -
Effluent
reactor #1
11 0.123 ± 0.038
49
19
15 days
Effluent
reactor #2
0.045 ± 0.017
34
23
9
15 days
Effluent
reactor #2
0.140 ± 0.028
15
38
22
10 days
Effluent
reactor #2
0.097 ± 0.048
40
21
Number
of
sampl es
16
9
5
7
Number
of
sampl es
15
9
7
6
Number
of
sampl es
11
5
5
Suspended solids,
38
27
12
65
-------
After 31 days the test reactor effluent cyanide concentration decreased
rapidly and remained lower than the control for the remainder of the test.
During this period, D3, the test reactor effluent cyanide level averaged 0.097
mg/fc and the control unit averaged 0.123 mg/Jl for a 26% improvement attributed
to the PAC/CuCl2. Frequency distributions of the effluent cyanide concentra-
tions for the three phases of Test D are illustrated in Figure 18. The fre-
quency distributions in Figure 18 show the increasing effluent cyanide concen-
tration from the control unit with time. (This same trend was also observed
in Figure 16.) An analysis of these phenomena is discussed later in this
section.
BOD Removal Performance
Daring Test D the BOD level was also monitored in the inlet and both ef-
fluents. The variation in the BOD concentrations with time are presented in
Figures 19 and 20 for the effluents and inlet, respectively. The first phase
consisted of steadily rising BOD's in the effluents whereas the inlet BOD was
actually dropping during this period. After the first 10 days of the test the
inlet BOD concentration rapidly increased but due to analytical difficulties
(contaminated dilution water), no BOD data for the last 6 days of this period
were obtained.
Although the BOD data are extreme1/ limited near the end of Dl, and the
beginning of D2, Figure 20 indicates that the organic loading was considerably
higher than during the rest of Test D. A higher organic loading could result
in the displacement of some adsorbed cyanide on the carbon due to the higher
driving force of the organics. While the data are too limited to draw any
conclusions concerning the reason for the higher cyanide levels from the test
reactor during D2, the higher organic loading during the later portion of Dl
is a plausible explanation. One other interesting observations from Figure 19
is the lower effluent BOD's during period D2 in the test reactor effluent.
Whether this is related to the higher effluent cyanide levels during the same
period could not be determined. Although the test reactor effluent BOD was
significantly lower than the control reactor during period D2, (59% lower),
the total organic carbon results did not exhibit the same trend. The TOC
results are presented in Figure 21.
Statistical Analysis
The statistical analyses of Test D for each time period are summarized in
Table 19. The effluent cyanide concentrations for all of Test D were signi-
ficantly lower in the test reactor than in the control reactor and so were
the effluent BOD levels. To further clarify the results of Test D statisti-
cal analyses of the results during each period were also performed, as summar-
ized in Table 19.
During period Dl, the effluent cyanide level from the test reactor was
consistently lower than the control, and this difference in performance was
verified statistically at the 10% significance level. In period D2 an
increase in the test unit's effluent cyanide level occurred which resulted
in higher cyanide concentrations for the test unit rather than the control.
66
-------
Number of Occurrences
0 V f 7* J» 5
MH
^M
101
Mean 0.073 mg/i
Std Dev. ± .022 mg/i
8
in
«
V
i_
u-
0
'°1
Mean 0.11 A mg/«.
Std Dev. ± .018 mg/i ^
in
u
S 6-
t_
u
<4-
o
E
z o
Mean 0.120 mg/i
Std Dev. ± .038 mg/i
Tfti
Cyanide, mg/Jl
Control Unit during Test D-1
Cyanide, mg/i
Control Unit during Test D-2
0.00 0.05 0.10 0.15 0.20 0.25
Cyanide, mg/i
Control Unit during Test D-3
in
a>
u
c
t)
i.
8<
6.
4-
2l
0
0
w«i
.00
^^^
0.
'
Mean 0.045 mg/i
Atd Dev. ± .017 mg/i g.
4)
1.
0
"1 u 2-
4)
_^ ^i
1 ' °
05 0.10 0.15 0.20 0.25 c
1.00 0.05 0.
«
10
0.
10-
Mean 0.139 mg/i
Std Dev. ± .028 mg/i
8-
i/i
4>
|6-
3
f
o
i 2-
-g
z 0
15 0.20 0.25 0
Mean 0.094 mg/i
Std Dev. ± .048 mg/Jl
"1
.00 0.05 0.10 0.15 0.20 0.25
Cyanide, mg/H
Test Unit during Test D-l
Cyanide, mg/i
Test Unit during Test D-2
Effluent Cyanide, mg/i
Test Unit during Test D-3
Figure 18. Test D - Cyanide frequency distributions by period.
-------
100 -i
90
80 -
70
60 -
50
Q
§ 40 -
30 -
20 '
10 '
10
Dl
o - Control
- Test
i
20
30
D2
Time, days
i
40
D3
Figure 19. Test D-Effleunt BOD5 with time.
68
-------
600
500
400 -
cp
£
g° 300 I
200 -
100 -
1 1 1 ,
10
-, Dl »u
20 30 40
0
*"
* D3 *-
Time, days
Figure 20. Test D-Inlet BODS with time.
69
-------
o
H
140
120
100
80
60-
20 -
10
Dl
20
30
Time, days
40
Figure 21. Test D-Inlet and effluent TOC with time.
70
-------
TABLE 19. TEST D - STATISTICAL RESULTS
Parameter
Effluent Cyanide - D
Effluent BOD - D
Effluent Cyanide 0-1
Effluent Cyanide D-l
Effluent Cyanide D-2
Effluent Cyanide D-3
Sample means, mg/£
Ml Ma
0.0997 0.0905
38.3 27.6
0.0726 0.0452
0.0726 0.0452
0.1142 0.1392
0.122 0.097
Hypothesis
MI = Ma
MI = Ma
MI = Ma
M2 + 0.023£Mi
MI = Ma
Ma = MI
T
Statistic
2.29
2.45
7.85
1.25
-2.50
4.41
Hypothesis is
accepted or rejected
Rejected
Rejected
Rejected
Accepted
Rejected
Rejected
Paired observations and 10% significance level utilized.
-------
The performance difference was statistically analyzed and the two means were
significantly different at the 10% level. During the last test interval, D3,
the removal efficiencies of the two reactors reversed again with the bet-
ter performance being achieved in the test unit. The statistical test used
indicated that this performance change was significant at the 10% level.
Thus, it can be confirmed with some statistical support that there occur-
red significant differences in performance both between the two units, and
between the different time periods. However it should be noted that the
sample dispersion and sample size both influence the validity of the conclu-
sions offered and should be considered limitations to the analysis.
Discussion of Test D
The phenomenon of decreasing removal efficiency was one which was ob-
served not only in Test D but throughout the continuous tests. The control
reactor exhibited a generally decreasing cyanide removal efficiency during
the course of these tests as summarized below in Table 20.
TABLE 20. CYANIDE REMOVAL ACROSS THE CONTROL REACTOR
Test
A
B
C
Dl
D2
D3
Mean inlet
cyanide concentration, mg/£
0.17
0.31
0.66
0.61
0.56
0.50
% Cyanide
removed
91
94
90
88
80
76
A satisfactory explanation for the decreasing removal efficiency in the
control could not be readily determined. The change with time could be attri-
buted to a gradually increasing fraction of complex cyanides in the feed. The
possibility of an increasing fraction of complex cyanides in the refinery
wastewater arises due to the greater demand for gasoline during the summer
months. As the summer approaches, refineries generally increase their feed
rates and cracking severity in the cracking units. The very severe cracking
conditions promote the formation of more cyanide. The more cyanide produced,
the more corrosion that will occur, and this in turn effects the quantity of
iron cyanide generated. The sour water stripper removes all but a small frac-
tion of the simple cyanides (stripping appears to be a concentration limiting
phenomena) and none of the iron cyanides. Thus, one would expect a relatively
constant simple cyanide concentration in refinery wastewaters and increased
complex cyanide concentration during periods of maximum gasoline production.
Iron cyanide removals across biological treatment systems are lower than the
72
-------
simple cyanide removals, as discussed in Section 3. The postulated higher
complex cyanide concentrations would explain the reduced cyanide removal effi-
ciency across the control reactor observed with time.
The increasing percentage of complex cyanides in the inlet would also
have an effect on the test unit due to the difference in rates of oxidation
of simple and complex cyanides. As explained in Section 3, the rate of oxi-
dation of complex cyanides has been found to be much slower than for simple
cyanides. Thus, the test unit would be affected by an increasing equilibrium
cyanide concentration in the carbon sytem. As the fraction of complex cya-
nides in inlet increased, so would the amount adsorbed onto the carbon, but
due to the slower oxidation rate, more cyanide adsorbed on the carbon would
in turn increase the equilibrium cyanide concentration remaining in solution
or in the effluent.
A comparison of the apparent cyanide loadings on the PAC for the various
tests is shown in Table 21. This loading, which is defined as the difference
between the quantity of cyanide removed in the test and control units divided
by the carbon dosage, changed significantly during the test program. The
apparent loading at the beginning of Test D was more than 2% times greater
than that for Test C, even though the carbon dose and inlet cyanide concentra-
tion were approximately the same. (The copper dosage for Test D was twice
that of Test C, and this accounts for the greater removal and thus higher
apparent loading.) However, the apparent cyanide loading for Test Dl was
still below the 1.5 to 2.0 mg CN"/gm carbon measured in the batch tests.
TABLE 21. APPARENT CARBON LOADINGS
Carbon dose,
Test rng/A
B
C
Dl
D2
D3
14
39
38
38
38
Apparent loading,
mg CN~/gm carbon
0.37
0.32
0.80
(-0.
0.73
59)
The results of Test D confirm the findings of the previous continuous
tests, that is, competitive adsorption from the organics reduces the quantity
of cyanide adsorbed by the PAC/CuCl2 addition. The higher outlet cyanide lev-
els during period D2 were attributed to overloading of the carbon from high
organic levels in the feed, although the data are lacking to confirm this
statement. To increase the quantity of cyanide removed, the following two
options are available:
73
-------
1. Reduce the organic loadings
2. Increase the carbon dosage.
The economics of this process depend almost entirely on the carbon dosage.
Instead of increasing the carbon dosage further, the first option of a reduced
organic loading was evaluated in the final test. Some refineries utilize a
two-stage biological treatment scheme. By adding the PAC/CuCl2 to the second-
stage, the organic loading on the carbon should be 50-80% lower than In- a one-
stage system. In the last continuous test, Test E, a two-stage biological
treatment system was utilized to check the hypothesis that a lower organic
loading should increase the quantity of cyanide removed.
TEST E - LOWER ORGANIC LOADING WITH MEDIUM CARBON/HIGH COPPER DOSAGE
In order to evaluate the effect of lower organic loadings on cyanide
removal, a two-stage activated sludge system was used in Test E. The effluent
from a common first stage was split into two streams; one for a second-stage
control and the other a second-stage test reactor. In the second-stage test
reactor a carbon dose of 25 mg/A, or 500 mg/day, and a copper level of 1.0
mg/5, or 19 mg (as Cu) per day were utilized. The carbon and copper were both
added for twelve days prior to initiating Test E to acclimate the micro-orga-
nisms and to build up the carbon level in the aeration basin.
The actual test period lasted 12 days at a sludge age of 15 days. As in
the previous tests, the efficiency of the biological treatment process was
monitored by measuring the BOD, COD, and TOC levels in the influent and efflu-
ents. The average performance characteristics of the test and control reac-
tors are summarized in Table 22.
As Table 22 indicates, the effluent cyanide concentration from the test
reactor was 53% lower than that of the control reactor. The effluent cyanide
level from the control reactor averaged 0.141 mg/jl, as compared to the mean
inlet cyanide level of 0.140 mg/£. The lack of any cyanide removal across the
second stage control reactor supports the previous hypothesis that primarily
simple cyanides are removed in the first stage activated sludge unit, leaving
the stable complex cyanides for the second stage. As in Test C and part of
Test D, the carbon addition dampened the day to day cyanide variation as mea-
sured by the standard deviation, (0.024 mg/A to 0.060 mg/fc). This dampening
effect in the effluent cyanide level is also evident in Figure 22, which
compares the two effluent cyanide levels with respect to time.
The biological performances of the second-stage control and test reactor
also varied but to a much smaller extent. The organic loading on the second-
stage, which averaged 82 mg/Jl BOD (unfiltered), 92 mg/A COD (filtered), and
25 mg/ TOC (filtered), was much lower than that for Tests A, B, C, and D.
The effluent from the test reactor contained, on the average, less BOD (6 mg/
A), COD (7 mg/£), and TOC (1 mg/A) than the effluent from the control reactor.
However, the suspended solids in the test effluent were usually 5 mg/A higher
than the control effluent.
74
-------
0.250 1
o - Control
- Test
0.200
, 0.150
-------
TABLE 22. TEST E - SUMMARY TABLE
Parameter
Flow rate, Jl/day
Cyanide, mgfi
BOD, mg/2.
COD, mg/sX1)
TOC, mg/J^1)
Copper
Suspended solids,
Inlet
-_-
0.140 ± 0.045
82
92
25
<0.02
Test
effluent
Reactor #3
20.0 ± 1.9
0.066 ± 0.024
21
43
10
0.20
17
Control
effluent
Reactor #4
20.2 ± 3.6
0.141 ± 0.060
27
50
11
<0.02
12
Sludge age - 15 days
Length of test - 12 days (after a 12 day acclimation period with carbon
and copper added to system) .
Carbon dosage - Reactor #3 only - 25 mg/£
Cupric chloride dosage - Reactor #3 only 1.0 mg as Cu/£
Test E only all COD and TOC analyses were performed on filtered
samples.
76
-------
Only a limited amount of data was collected during this short test, and
to statistically evaluate the results, these data were pooled using a dummy
variable for reactor type. With 20 data points the difference in cyanide re-
moval efficiencies between reactors was considered statistically significant
at the 0.5% level. Or, in other words, the probability that the control and
test reactor performed differently is greater than 0.995. This can be ob-
served visually by inspecting Figure 23. Not only are the means quite differ-
ent but also the dispersion and location of the frequency distribution.
The large reduction in cyanide (53%) attributed to the PAC/CuCl2 addition
tends to support the hypothesis that a lower organic loading would increase
the removal of cyanide. The apparent cyanide loading (the difference in quan-
tity of cyanide removed by the test and control reactors, divided by the car-
bon dose) for Test E was 2.74 mg CN~ per gm carbon. The apparent cyanide
loadings for all the one-stage tests ranged from 0.32 to 0.80 mg CN~ per gram
of carbon (see Table 21). Such a variation in cyanide loading clearly indi-
cates that lower organic loadings increased the quantity of cyanide removed
per gram of carbon.
SUMMARY OF PHASE II
Continuous tests were performed over a four-month period to evaluate PAC/
CuCl2 addition into an activated sludge system for cyanide removal. Batch
tests, utilizing iron cyanides added to Chicago tap water, achieved cyanide
removals in excess of 95%. The continuous tests were performed with actual
refinery wastewaters utilizing two reactors; one as a control unit and the
other for evaluating the PAC/CuCl2 system. A series of five tests were per-
formed, which are listed below:
Test A - no carbon - no copper
Test B - low carbon - low copper
Test C - high carbon - low copper
Test D - high carbon - high copper
Test E - medium carbon - high copper (two-stage system)
The results from these five tests demonstrated that activated sludge
units remove as much as 90% of the cyanide present in refinery wastewaters.
The addition of PAC/CuCl2 directly into the aeration basin can remove addi-
tional quantities of cyanide without any apparent effect on the micro-organ-
isms. With the exception of the initial slug dose in Test B, no inhibitory
effects on the biota were observed in the activated sludge performance.
The addition of PAC may improve the removals of BOD, COD, and TOC. While
a significant improvement in BOD removal was observed in the longest test
(Test D) as a result of the PAC addition, Tests B and C did not confirm this
trend and the data in Test E were too limited for analysis. The TOC and COD
analysis were performed on unfiltered samples (except in Test E) and the car-
bon fines in the effluent samples contributed to both the COD and TOC values
observed. Again in Test E, the data were too limiting to make any definite
statement regarding the removals of COD and TOC.
77
-------
5-,
(/I
/I
A) ^
U
|
I 3
u
u
o
1-
OJ
Mean 0.066mg/2.
Std. Dev. ± .024 mg/Jl
TEST UNIT
0.050 0.100 0.150 ' 0.200
Cyanide, mg/H
0.250
Figure 23. Test E-Effluent cyanide frequency distribution.
78
-------
The organic loading was found to be a major factor in the removal of
cyanide by the PAC/CuCl2 system. Test E supported the hypothesis reached after
Test D, namely, decreasing the organic loading improves the cyanide removal
efficiency. This is probably due to the effect of reducing the competition
between the smaller amount of organics and the cyanide for the available ad-
sorption sites on the carbon. The increased availability of adsorption sites
to cyanide resulted in more cyanide being adsorbed and, therefore, correspond-
ing higher apparent cyanide loadings. The greater apparent cyanide loading
of 2.74 mg CN" per gm of carbon, is a more efficient use of the carbon, there-
by decreasing costs for the system. Therefore, the use of the PAC/CuCl2 system
is most attractive in a low organic load waste stream, such as in the second-
stage of a two-stage activated sludge system.
79
-------
SECTION 7
ECONOMIC IMPLICATIONS OF THE PAC/CuCl2 SYSTEM
The technological feasibility of utilizing the PAC/CuCl2 system in an
activated sludge unit to remove cyanides has been demonstrated by the results
of Phase II experimentaion. Reduction of cyanides can be achieved; however,
the competitive adsorption of organic compounds on the activated carbon re-
duces the quantity of cyanide adsorbed per gram of carbon. Without prefer-
ential adsorption of the cyanides, this system will require larger carbon
dosages when the organic concentration is high.
Cost estimates for operating this PAC/CuCl2 system in a refinery provide
additional insight on its overall feasibility and applicability. Based on the
data collected from the continuous tests in Phase II, which most nearly simu-
lated actual refinery conditions, operating conditions of a PAC/CuCl2 system
are projected. The costs of operating and installing such a system are then
developed for single-stage and two-stage activated sludge units. A comparison
of these costs with that of a granular carbon system emphasizes the economic
advantages derived from the PAC/CuCl2 treatment technique. Thus, the operat-
ing range of economic feasibility is clearly demonstrated for this experiment.
OPERATING CRITERIA FOR PAC/CuCl2 SYSTEMS
The continuous tests of Phase II provided basic information regarding the
operating performance of a PAC/CuCl2 cyanide removal system. Important cri-
teria in designing a system or estimating costs are the apparent cyanide load-
ing, existing effluent cyanide level, and copper dosage required. Data
gathered experimentally were used to describe these factors.
To determine the quantity of carbon required to remove a specified amount
of cyanide, the apparent loading must be estimated. The apparent cyanide
loading is defined as the difference in cyanide removed between the test and
control reactors, divided by the carbon dosage. According to adsorption the-
ory, the loading of a contaminant on carbon decreases as the carbon dosage
increases. Or, in other words, for each additional increment of carbon added,
a progressively smaller increment of the contaminant will be removed. (This
can be readily seen in the batch test results, as shown previously in Figure
12.)
To ascertain the effect of carbon dosage on apparent loading Tests B and
C were both run at a constant copper dosage. The apparent loadings for Tests
B and C, plotted in Figure 24, illustrates the effect of carbon dosage.
80
-------
50-,
00
40-
OJ
CD
(C
o
JD
ea
o
30.
20.
10.
1-stage
activat
sludge,
0.5 mg/
copper
1-stage activated
sludge, 1 mg/2. copper
, B
2-stage activated
sludge, 1 mg/I copper
0.5
2.0
2.5
3.0
Apparent loading, mg CN /g Carbon
Figure 24. Effect of carbon dose on apparent loading.
-------
From Figure 24, increased carbon dosage had only a slightly negative
influence on the apparent cyanide loading.
Also shown in Figure 24, are extrapolated loading lines for the conditions
utilized in Test D and E, under the assumption that these lines would be par-
allel to the line passing through Test B and C. The apparent cyanide loading
is only slightly dependent on carbon levels, which means that the quantity of
cyanide removed is approximately proportional to the carbon dose. Using Fig-
ure 24, the quantitites of carbon required to achieve specified effluent
levels can be computed for both a one-stage and a two-stage activated sludge
system.
The copper concentration to be used in a PAC/CuCl2 system was specified
at 1 mg/i as a result of the batch and continuous tests. No detrimental
effect on the performance of the activated sludge units were found at 1 mg/£
copper; however, the literature indicates that above 1 mg/Jl, the performance
of the activated sludge units may begin to decrease.
COST ESTIMATION OF PAC/CuCl2 SYSTEMS
The costs of utilizing a PAC/CuCl2 cyanide removal system were based upon
the apparent loading values of Figure 24 and copper concentration of 1 mg/5,
with the goal of achieving an effluent cyanide level of 0.025 mg/i. In Table
23 the chemical costs are presented for single-stage and two-stage activated
sludge units. The chemical costs were developed under the assumption that the
PAC would not be recovered after use. By eliminating PAC regeneration, this
cyanide treatment method can be applied without any capital expenditure or
additional plant operators. For larger carbon users regeneration would be
more practical and economical. Thus, the estimated chemical costs presented
Table 23 are actually close to the total cost of this system. Powdered carbon
is the primary cost of the system (85-95%). Thus the required carbon dose
determines the economic feasibility of the system.
While the PAC/CuCl2 process will not solve the cyanide problem for all
refineries, this process has considerable potential for some refineries. For
refineries with two-stage activated sludge units operated in series, the costs
as shown in Table 23, are approximately one-fourth of those for a single-stage
system. To place the costs of operation of this proposed cyanide removal sys-
tem in perspective, a comparision with another potential system is made in the
following section.
COST COMPARISION OF PAC VS 6AC CYANIDE REMOVAL SYSTEMS
The cost advantages of the PAC/CuCl2 system are most aptly illustrated
by comparison with other cyanide treatment methods. As stated in Section 3,
all existing cyanide treatment methods are considered expensive by the refin-
ing industry. The economics of this cyanide treatment system are compared
with the GAC catalytic oxidation system described in Section 3 as an example
of the projected cost savings to be realized.
The GAC/CuCl2 system was investigated by one refinery in Illinois.12
Utilizing the refinery cost estimates and cyanide levels, a cost comparison
82
-------
TABLE 23. CHEMICAL COST SUMMARY
1-Stage Activated Sludge
To Achieve 0.025 mg/Jl Total Cyanide
Present cyanide ef-
fluent concentration,
mg/A
Required carbon
dose,
mg/J,
Chemical cost
$/million gallons
0.05
0.10
36
108
106
286
2-Stage Activated Sludge
To Achieve 0.025 mg/fc Total Cyanide
Present cyanide ef-
fluent concentration,
Required carbon
dose,
mg/fc
Chemical cost
$/million gallons
0.05
0.10
9.4
28.0
40
86
Based upon 1 mg/SL CuCl2 (as Cu+2) and carbon cost of $.30/pound
and CuCl2 cost of $1.91/pound (as Cu+2).
83
-------
was prepared. Placement of granular activated sludge system (with CuCl2 and
oxygen addition), would bring this refinery into compliance with the Illinois
effluent standard of 0.025 rag/A. Based on pilot plant data and an effluent
flow of 4.2 MGD, the refinery estimated 51,000 pounds per day of carbon would
be exhausted. Assuming a carbon regeneration furnace is purchased and an
attrition loss of 10%, fresh carbon make-up would be 5,100 pounds per day.
Ignoring the large capital investment required for the granular carbon system,
the chemical costs for both systems are compared in Table 24.
Even with the thermal regeneration of the granular carbon, the make-up
carbon alone would cost $3,000 a day, as compared with $210 or $810 per day
for the powdered carbon cost without any regeneration. Taking into consider-
ation the differences in capital investment, the PAC system is even more
advantageous.
To provide some indication of the range of operating costs expected,
chemical expenditures for the other refineries described in Table 1 were
calculated. The values of Table 25 are based on an effluent level of 0.025
mg/i total cyanide and a copper concentration of 1.0 mg/£. Depending upon
the effluent characteristics of the refinery and the number of activated sludge
stages, the costs varied from $20,000 to $530,000 per year. For any of the
refineries listed in Table 25, the costs are much lower than the alternative
cyanide treatment methods.
QUALITATIVE ECONOMIC ASPECTS
There are other aspects of the PAC/CuCl2 system with offer economic ad-
vantanges that are more difficult to quantify. The following list of benefits
from the addition of PAC into an activated sludge unit should reduce the oper-
ating costs assessed to cyanide removal even further:
1. The fluctuation in daily effluent cyanide levels is reduced,
(except during periods of high organic loadings).
2. Although not observed in the laboratory activated sludge
reactors, PAC is reported to lower effluent suspended
solids.
3. Lower effluent BOD's are achieved.
4. Sludge dewatering characteristics are improved.
While PAC/CuCl2 addition offers some economic advantages over other pro-
cesses, this process is not without its drawbacks. Where effluent cyanide
standards are based on daily maximums as opposed to monthly averages, the
value of this system is reduced. As occurred in Test D, slug doses of high-
strength organic wastes can result in a desorption of cyanide from the carbon,
resulting in significantly higher effluent cyanide levels. Thus, where wide
fluctuations in the organic loading exist, high cyanide effluent peaks will
occur for one or more days. Where a two-stage sludge system is used, the
organic fluctuations should be reduced.
84
-------
TABLE 24. CYANIDE TREATMENT - CHEMICAL COST COMPARISON
System
Granular carbon - with
thermal regeneration
Powdered carbon - no
regeneration
1-stage activated
sludge system1
Powdered carbon - no
regeneration
2-stage activated
sludge system2
Carbon cost*
$/day
3,000
810
210
Copper cost.
$/day
67
67
67
Total
Chemical cost,
$/day
3,067
877
277
Design Basis:
Flow 4.2 MGD
Existing cyanide effluent level - 0.079 mg/fc
Design cyanide effluent level - 0.025 mg/£
Granular activated carbon - $0.60/lb
Powdered activated carbon - $0.30/lb
Cupric chloride - $1.91/lb (as Cu+2)
Apparent cyanide loading 0.7 mg CN"/gm carbon
Apparent cyanide loading 2.7 mg CN"/gm carbon
85
-------
TABLE 25. COST ESTIMATES FOR PAC/CuCl2 SYSTEM
Refinery
A
A
B
B
E
E
F
F
No. Of
stages 1n
activated
sludge
1
2
1
2
1
2
1
2
Effluent
flow rate,
MGD
2.6
2.6
4.2
4.2
2.0
2.0
6.8
6.8
Existing
effluent
cyanide,
rag/A
0.176
0.176
0.079
0.079
0.038
0.038
0.045
0.045
Annual
operating
costs,
$
530,000
150,000
320,000
100,000
46,000
20,000
220,000
86,000
Design basis:
Cyanide effluent level - 0.025 mg/A monthly average
PAC - $.30/lb, CuCl2 - $1.91/15 (as Cu+2)
Apparent CN" Loadings: 1-stage - 0.7 mg CN'/gm carbon
2-stage - 2.7 mg CN'/gm carbon
86
-------
In summary, the PAC/CuCl2 system has definite economic advantages over
existing cyanide treatment methods available to the refining industry. This
system is perhaps best suited for refineries requiring less than a 60% reduc-
tion in effluent cyanide levels and especially for refineries with two-stage
activated sludge units. Because there is no capital expenditure required to
utilize the PAC/CuCl2 system, its use is extremely flexible. For example,
some refineries experience high effluent cyanide levels for one to two months
after bringing the catalytic cracking unit back into operation after a major
maintenance shutdown. The PAC/CuCl2 system is Ideally suited for short-term
use in such instances where high cyanide levels are projected. As discussed
in Section 3, many refineries also experience one to three months each year
of very high effluent cyanide levels. The PAC/CuCl2 system provides an econ-
omic approach to reducing these cyanide levels. When the effluent cyanide
levels are in compliance, no carbon or copper needs to be added. For refin-
eries that require high degrees of cyanide reduction on a yearly basis, Table
25 provides an indication that over $500,000 per year of powdered carbon may
be required. While not included in this economic evaluation, PAC regeneration
by wet air oxidation would reduce the overall cost of treatment for the larger
PAC users.
-------
REFERENCES
1. American Petroleum Institute, Report WBWC 3064, 1972 Sour Water Stripping
Evaluation, Publication 927, Washington D.C., 1973, pp. 6, and 20-28.
2. Huff, L.L., and J.E. Huff. Analysis of the Benefits and Costs of Alter-
native Cyanide Standards in Illinois. Illinois Institute for Environmen-
tal Quality. Report No. 75-24, 1975.
3. U.S. EPA, Quality Criteria for Water, 1976, p. 65.
4. Public Law 95-217, The Clean Water Act, Section 307 (a)(2), December,
1977.
5. Environmental Reporter. McGraw-Hill Publishing Company, 1975.
6. Cyanide Destruction Using Ozone and Ultraviolet Light. Mobil Oil
Bimonthly Variance Report on Cyanide to the Illinois EPA, Springfield,
Illinois, August, 1975.
7. Irvin, J.W., and H.D. Tomlinson. Testimony presented in Regulatory
Hearing R74-15, 16 before the Illinois Pollution Control Board, 1975.
8. Pettet, A.E.J., and E.V. Mills. Biological Treatment of Cyanide With
and Without Sewage. J. Appl. Chem. 4:434-444, 1954.
9. Bucksteeg, W. Decontamination of Cyanide Wastes by Methods of Catalytic
Oxidation and Adsorption. Presented at the 21st Annual Purdue Industrial
Waste Conference, Purdue University, Lafayette, Indiana, May 3, 4, and 5,
1966.
10. Kuhn, R.G. Process for Detoxification of Cyanide Containing Aqueous
Solutions. U.S. Patent 3,586,623, June 22, 1971.
11. Bernardin, F.E. Cyanide Detoxification Using Adsorption and Catalytic
Oxidation on Granular Activated Carbon. JWPCF 45:221-231, 1973.
12. Bernickas, J.V. Testimony presented in Regulatory Hearing R74-15, 16,
before the Illinois Pollution Control Board, 1975.
13. Flynn, B.P., and L.T. Barry. Finding a Home for the Carbon: Aerator
(Powdered) or Column (Granular). Presented at the 31st Annual Purdue
Industrial Waste Conference, Purdue University, Lafayette, Indiana,
May 1-3, 1973.
-------
14. Adams, A.D. Improving Activated Sludge Treatment with Powdered Activated
Carbon. Presented at the 28th Annual Purdue Industrial Waste Conference,
Purdue University, Lafayette, Indiana, May 1-3, 1973.
15. U.S. EPA. Methods for Chemical Analysis of Water and Wastes. 1974.
16. Sudo, R., and S. Aiba. Effect of Copper and Hexavalent Chromium on the
Specific Growth Rate of Ciliate Isolated from Activated Sludge. Water
Resources 7:1301, G.B., 1973.
17. Poon, C.P.C., and K.H. Bhayani. Metal Toxicity to Sewage Organisms.
Proceedings ASCE, J. Sanitary Engr. Division 97:S.A.2:161, 1973.
18. Lamb, A., and E.L. Tollefson. Toxic Effects of Cupric, Chromate and
Chromic Ions on Biological Oxidation. Water Resources 7:599, G.B., 1973.
19. Salotto, B.V., E.F. Barth, W.E. Tolliver, and M.B. Ettinger. Organic
Load and the Toxicity of Copper to the Activated Sludge Process.
Presented at the 19th Purdue Industrial Waste Conference, Purdue Univer-
sity, Lafayette, Indiana, 1964.
20. Barth, E.F., M.B.Ettinger, B.V. Salotto, and G.N. McDermott. Summary
Report on the Effect of Heavy Metals on the Biological Treatment
Processes. JWPCF 37:86, 1965.
21. Winker. R.L., and W.L. Hays. Statistics: Probability, Inference, and
Decision. Holt, Rinehart and Winston, New York. 2nd Ed., 1975.
89
-------
APPENDIX
TABLE A-l. TABULATION OF ALL DATA FROM PHASE I.
Test &
Reactor
No.
1-1
-2
-3
2-1
-2
-3
3-1
-2
-3
4-1
-2
-3
5-1
-2
-3
-It
6
1-2
-3
-4
8-1
-2
-3
-It
9-2
-3
-4
10-1
-2
-3
-4
11-3
-4
12-3
-4
13-3
-4
6-hr CN
Cone. , mg/J.
< 0.01
0.027
0.060
< 0.01
0.01
0.05
0.30
< 0.01
<0.01
0.168
0.145
0.010
<0.01
0.040
0.010
0.010
Equilibrium
CN~ Cone. , mg/£
< 0.01
0.024
0.038
0.28
<0.01
0.013
0.120
0.115
< 0.01
< 0.01
0.025
0.010
<0.01
Carbon
Cone. ,
ng/*
250
250
250-
250
250
250
250
250
250
250
250
250
250
250
250
250
Type
of
Carbon1
H.D.
H.D.
H.D.
H.D.
H.D.
H.D.
H.D.
H.D.
H.D.
H.D.
H.D.
H.D.
H.D.
H.D.
H.D.
H.D.
Initial
Copper
Cone . ,
mg/i
1.5
1.5
1.5
1.5
1.5
1.5
0.0
1.5
1.5
0.5
0.5
1.5
10.0
0.75
1.0
1.25
Initial
CN Cone . ,
mg/i
0.48
0.48
0.53
0.44
0.50
0.52
0.53
0.52
0.39
0.48
0.37
0.39
0.46
0.42
0.46
0.46
Type
of
CN"
Ferro
Ferro
Ferro
Ferri
Ferri
Ferri
Ferro
Ferro
Ferro
Ferro
Ferro
Ferro
Ferro
Ferro
Ferro
Ferro
Filtrate
Copper
Cone, at 6 hr,
mg/i
1.8
0.05
0.05
1.7
0.05
0.02
< 0.02
0.10
0.10
0.05
0.05
0.15
0.4
0.05
0.05
0.05
Used in
Regression
Analysis
No2
Yes,
Yes
No3
No3
No3
Yes
Yes
Yes
Yes
Yes
Yes
No"
Yes
Yes
Yes
Test aborted after six hours, reactor split open;
repeated as Test 7.
0.16
0.12
0.05
0.23
0.46
0.39
0.023
0.29
0.29
0.015
_
0.32
0.50
0.26
0.29
0.15
._
0.025
0.13
0.11
0.39
0.29
0.023
0.12
0.16
0.01
0.41
0.26
0.45
0.27
0.30
0.10
0.09
0.24
0.63
0.51
100
250
1,000
0
100
250
1,000
100
250
1,000
0
250
100
100
250
250
100
100
0
0
H.D.
H.D.
H.D.
L.D.
L.D.
L.D.
L.D.
L.D.
L.D.
L.D.
H.D.
H.D.
H.D.
L.D.
H.D.
H.D.
1.0
1.0
1.0
0.5
0.5
0.5
0.5
1.0
1.0
1.0
0.0
1.5
0.5
1.5
1.0
1.0
1.0
0.5
0.0
0.5
0.34
0.44
0.38
0.53
0.53
0.66
0.57
0.41
0.67
0.50
0.67
0.69
0.66
0.70
0.63
0.67
0.40
0.41
0.66
0.65
Ferro
Ferro
Ferro
Ferro
Ferro
Ferro
Ferro
Ferro
Ferro
Ferro
Ferro
Ferro
Ferro
Ferro
Ferri
Ferri
Ferro
Ferro
Ferro
Ferro
0.05
0.05
<0.02
0.12
0.08
0.05
< 0.02
0.12
0.05
0.02
0.18
0.04
0.19
< 0.02
0.17
Yes
No5
No6
No7
Yes
Yes
Yes
Yes
Yes
Yes
No'
Yes
No8
No8
No3
No3
No8
No8
No'
No'
'H.D. High Density, lignite-based; L.D. Low Density, lignin-based carbon.
2pH too low, did not have enough pH variation to analyze statistically.
3Ferricyanlde data not used, too few tests.
"Copper dosage beyond concentration range.
5Lost equilibrium cyanide sample during analysis boiled over.
'Equilibrium value appears erroneous, does not agree with 6-hr cyanide sample.
'Suspect reactor contaminated with carbon particles.
"Final data collected after regression analysis completed.
'Without carbon in the system, any removal would be by a different mechanism.
90
-------
vo
APPENDIX
TABLE A-2. DATA FROM PHASE II TEST A - BASELINE OPERATION
1
.*d
2-19
20
21
22
23
25
26
27
2S
2*
VI
2
3
4
5
6
7
ft
9
10
Hun
Cyanlda Cone.
titUt tfflu.nl 1 Effluent 2
0.23 0.059 0.063
--
1
0.19 0.067 0.012
0.21 0.020 0.020
0.15 O.OIS 0.012
-- "
o.ii o.oo4 o.ooa
0.13 0.002 0.008
0.14 0.002 3.006
0.13 0.006 O.OOt
0.13 0.004 0.002
0.24 0.001 0.001
0.15 o.ooa O.OM
0.21 0.002 n.OU
0.20 0.01(1 0.026
0.19 0.010 0.012
WO
310 15 »
130 14 10
540 7 14
120 9 S
210 19 19
340 19 22
270 21 1*
360 16 16
10* 12 3
m »
242 l> 10
168 9 10
219 19 23
127 11 U
130 « »
250 13 **
Total Organic Carbon
89 22 28
91 37 31
106 24 27
84 27 25
6* 20 19
90 25 20
79 22 22
100 26 26
97 2) 23
III 27 31
Flow
I/day
19 24
24 17
II 14
12 13
165 22
21 16
18 14
15 22
15 1*
5
14 11
U U
11 12
16 14
17 21
19.5 18
18.5 19
8 11.5
23 22.5
23 22
16.0 16.7
Suspended Solid*
14 5
30 52
<5 <5
7 31 28
24 23
25 10
41 8 3
16 10
8 10
27 9 7
25 22 21
MLSS«».
I/I
2968 2805
2359 3013
2982 3159
2411 2717
2622 2377
2194 2126
1*31 1870
1795 1981
1667 1792
1642 1646
2257 2M9
Copper Cone.**
-./' .
._
__
--
_-
.-
* loth r«»ctor« operated without Copper or PAC addition
** Copepr m>t Monitored during T«*t A
.» Slud«. u. -..ted « . r.t. of 800 .l/d.y fro. «rh ..ration ba.ln
-------
APPENDIX
TABLE A-3. DATA FROM PHASE II TEST B LOW CARBON/LOW COPPER*
(-11
12
1 1
14
15
16
17
IS
19
20
21
21
11
It
2}
11,
II
28
21
Neon
Cyan Ida Com- .
»''
11.21 0.008 0.016
o.iB 0.014 a.nin
n.17 0.006 O.O08
0.19 0.014 0.010
0.20 0.008 0.010
0.20 0.008 0.010
0.21 0.004 0.012
0.2) 0.006 0.012
0.021 0.012 0.010
0.027 0.008 0.012
0.049 0.026 0.020
0.62 0.028 0.0)7
0.11 0.022 0.028
0.1) 0.014 0.0)2
0.57 0.020 0.022
0.41 0.040 0.028
0.41 0.0)2 0.0211
0.40 0.020 0.0)7
0.)) 0.028 O.OM
O.JI 0.017 0.020
POD
=»/'
220 10 14
210 15 24
4iO 6 29
SO II 20
:.'4 IJ 2)
2)0 10 M
51(1 52 60
160 16 JO
2ID 17 22
190 11 48
-.
'50 14 )t
;>o 21 16
-.
j<» 14 ;-4
280 16.5 )0.«
Ti>La) Organic Carbon
08/1
97 23 2)
104 26 JO
~
-.
If 18 IB
..
4)0 :i 36
.-
154 )) 21.)
..
109 21 26.)
..
IJft 21. S 2*. )
..
112 11.5 21
145 24 25
Flov
^...'(X.... i
21.5 21.)
:2.5 17.5
16 20
18 18
18 18
» 9
17 21
8 20.)
19.) 19
18 17.9
17 18
17. J 17.)
Id.) 16.5
18 17.)
19 17
20 22
22 20
21 11
21.) 20
17.9 18.4
Suapendcd Solids
**/t
5 It
.-
.-
6 8
6 I)
7 2)
9 )7
.-
..
14 1 )
--
6 2
4 15
..
..
8 8
8 IS
KLSS"
mg/l
1861 16)3
_
_
2020 1964
1117 1MI
. 669 1207
1970 1196
_ _
1)73 14)9
--
1898 1719
_
2167 18)5
_
1705 1871
Copper Cone .
8/1
<.OI <.OI 0.05
0.0) <.OI 0.20
-
<.OI <.01 0.16
«.01 <.01 0.18
<.OI <.01 0.1)
<.02 <.02 0.20
_
<.02 0.12
_ _
<.02 <.02 0.17
<.02 <.02 0.15
--I
VO
ro
* Reactor I Is a mntrol with no copper or PAC
functor 2 hA» 25O oft/dor PAC and O. 5 ««/' .-pp«r
Sludge ngir 12 days In both rtvi-lorn
* Sludfif vnntcd nt a r..l<- ..f HCWl al/day fr<« hnth reart-
-------
APPENDIX
TABLE A-4. DATA FROM PHASE II TEST C HIGH CARBON/LOW COPPER*
.
10
co
_ 1
Dace
4-1
t
3
4
5
6
7
8
9
10
11
12
13
U
Mean
Cyanide Cone.
c/t
tnl.t Effluent i Effluent 2
0.1.7 0.0*0 0.065
0.59 0.062 0.52
0.60 0.059 »-037
0.67 0.0*9 0.047
0.5J 0.062 0.04*
0.72 0.067 0.042
0.68 0.056 0.015
0.60 0.028 0.027
0.59 0.0*9 0.032
0.72 0.090 0.010
0.80 0.110 0.05«
0.72 0.103 0.065
0.69 0.096 0.056
0.65 0.067 0.081
0.66 0.067 0.0*8
, :
.
BOD
«/l
Inlet Effluent 1 Effluent 2
390 29 32
336 16 18
273 13 23
500 30 30
550 35 37
305 *2 37
349 20 28
400 JO l«
120 25 25
380 27 2'
_
Total Organic Carbon
M/t
170 30 10
121 23 25
117 30 29
136 35 36
124 28 22
110 1* '2
130 27 26
^__^
Flov
8 H
21 17
20 20
18 22
15.5 15.5
18.5 19
19 20
20.5 19.5
21 21
22 22
22 2*
21 21
21.5 21.5
21.5 12.5
19.25 1»-0
Suspended Solids
»B/l
15 11
6 13
16 1*
13 13
7 6
3 4
10 10
MLSS**
g/t
teactor 1 Reactor 2
2030 2054
1349 2149
1957 2812
1639 2878
1663 2492
1924 3166
^ ^
Copper Cone .
.g/t
Inlet Effluent 1 Effluent 2
<.02 0.15
<.02 0.12
<.02 «.02
-------
APPENDIX
TABLE A-5. DATA FROM PHASE II TEST D HIGH CARBON/HIGH COPPER*
to
-p.
Cyanide Cone. 1 |OD
«/' 1 mt/t
"Bate ' Inlet Effluent T Effluent I Inl«t Effluent 1 tffluent 1
r*rtod 1,
4-16 j 0.7* 0.0i2 0.028
17 0.74 0.0*9 O.OJO
18
19
20
21
22
23
2*
25
26
27
28
2*
30
5-1
Hun
reeled 2
5-2
3
10
II
12
1}
1*
IS
16
Nun
Period 3
J-17
18
0.5* 0.0** 0.024
O.S3 0.052 0.035
O.S3 0.032 0.022
0.53 O.O*5 0.030
0.57 0. IDS 0.0**
0.5* 0.08* 0.0*1
0.*9 0.067 0.0*9
0.67 0.070 0.0**
0.53 0.084 0.0*5
0.49 0.087 0.059
0.56 0.078 0.0*2
0.85 0.096 0.061
0.67 0.0«6 0.08*
0.72 0.100 0.0*9
0.61 0.073 0.0*5
0.37 0.090 0.127
0.41 0. 110 0.202
0.54 0.096 0.135
0.50 0.100 0.120
0.5* 0.170 0.110
0.47 0.100 0.1*0
0.3* 0.103 0.103
O.53 0.110 0.140
0.53 0.127 0.191
0.67 0.120 0.127
0.6! 0.100 o. 113
0. 70 0. 1 35 0. 160
0.70 0.1*0 0. U*
0.70 0.1*8 0. 127
0.56 0.11* 0.139
0.70 0.175 0.103
O.t-5 0.151 0.072
19
20 1 0.51 O.CM* 0.032
21 '
22
23
2*
25
26
27
0.51 0.116 0.0*2
0.5* O.O93 0.0*7
0.5* 0.106 0.093
0.32 0.100 0.08*
O.«0 0.191 0.135
O.*0 0.115 0.155
0.45 0.078 0.090
28 j 0.45 0.09ft 0.186
Ne.ii> 1 0.50 .-'.130 0.094
310 U 20
**0 22 1J
230 29 Tl
310 18 30
210 30 3]
230 45 30
2*0 19 37
580 83 97
*OO ** J3
S30 3* 3*
__
*20 50 31
220 «6 *
2*0 51 8
180 3* 12
320 60 21
II II II
330 24 9
280 1* 12
140 *0 18
320 16 17
272 37 15
II II II
100 55 5*
270 35 66
300 60 39
230 18 15
182 75 22
216 *8.6 39.6
M/t
Inlet ('fluent 1 Effluent 2
120 25 2*. 5
106 25.5 20.5
86.6 25 29.5
99 ?., 21
-- ..
117 16 19.5
106 2* 23
7* 27 32
10* 11 U
137 23 tO
130 31.5 18
13* 20 39
127.5 28 37
283.5 3*. 5 46
1*1 28 38
-
11.5 14.5 17.5
70.1 U.I 12.*
-*
I-" 17.8 21
47.4 19.2 19.3
58 31 31
74.2 19.3 20. 2
Flow
I/day
I»»ctor 1 K««ctor 2
21.5 21
21.5 21
21 19.5
22 19
14.5 15.5
19 21
18.5 21.5
18.5 20.5
19 22
18 19.5
19 20.5
21.5 21
21.5 20.5
20.5 21.5
19.5 21.5
20 20
19.7 20.3
17 18
17 16
16 16
18 t7
15 5 14 5
18 17
20 19
18 19
17.5 18
16 5 18
18 18
17.5 16.5
19.5 20
21 20
18 18
17.8 17.7
18 20
19.5 22
23.5 25
21.5 22
22° 21
23.5 23
23 18.5
22 22
2O 19
2* 2*
22 21
18 16
21.4 21.1
Suspended Solid*
6 4
6 5
7 5
7 4
«
21 8
9 7
13 28
99 9
12 1*
9 9
10 26
_
9 32
10 29
13 24
10.5 22.3
10 7
10 10
10 9
11 5
U 12
II 27
MLSS««
2019 3214
. __
1619 3160
2231 3312
1621 3O65
2051 3126
2027 2979
1368 1593
-_ .
1/38 1201
991 9OO
705 996
156O 2325
1390 2597
1110 1784
_
1780 27*0
7*1 1180
12*6 1960
« "
1797 1378
1707 1651
1172 2376
Copper Cone.
mm/ 1
l.l.t Effluent 1 Effluent 2
- .02 0.08
«. ..
<.02 0.12
<.02 <.02 O.I*
< .02 0.1*
<.02 0.18
<.02 <.02 0.13
<.02 0.28
<.02 <.02 0.26
<.02 0.29
<.02 0.28
<.02 <.02 0.28
<.02 0.28
<.02 <.02 0.28
<.02 0.16
<.02 <.02 0.15
<.02 <.02 0.18
__
<.02 0.19
<.02 <-02 0.20
<.02
-------
APPENDIX
TABLE A-6. DATA FROM PHASE II TEST E TWO STAGE TEST, MEDIUM CARBON/HIGH COPPER*
IW
5-30
31
6-01
2
3
4
5
6
7
1
9
10
11
Mean
Cyanide Con1:.
Inlet Effluent 3 Effluent *
0.087 0.087 0.197
0.0*0 0.0*4 0.22?
0.1SO 0.100 0. 1*8
0.180 0.103 0.151
0.113 0.062 0.127
0.180 0.062 0.170
0.081 0.0*4 0.191
0.130 0.032 0.0*0
0.155 0.056 0.070
0.202 0.065 0.087
0.1*0 0.066 0.1*1
BOD
Inlet Effluent 3 Effluent *
_
_
ISO 19 70
88 27 36
62 17 28
«7 13 17
41 8 10
97 IB 14
89 13 13
82 21 27
Total Organic Carbon
13 10 11
_
20 6 7
*2 10 10
40 10 11
17 12 1*
19 12 10
24 13 12
25 10 11
Flow
17 10
20 21
20 21
18 20
20 21
19 21
20 21
18 21
22 21
22 22
22 22
22 22
20 20
Suspended Solids
5 4.6
10 3.4
.-
* 5.4
* 5.*
~
6 5
6 6
5.8 5.0
MLSS"
-111
636 205
1156 2*3
494 238
646 21*
835 *2*
790 402
-
Copper Cone.
-/I
<.02 0.21 <.02
<.02
<.02 0.11 '-02
<.02 0.20 <-02
c.02 0.23 <-02
""
__
<.02 0.19 '-02
~
<.02 0.11 ''02
0.35 <-02
<.02 0.20 «-02
tn
Reactor 4 1» a control unit with no copper or PAC added
Reactor 3 has 500 «g/day PAC and 1.0 mg/l of copper added
Sludge age was 15 days
Sludge WHS wasted «t a rate of 650 at/day fro« both »er«tlon basins
-------
APPENDIX
TABLE A-7. BATCH TEST CYANIDE DECAY DATA
Test and
Reactor No.
1-1
1-2
1-3
3-1
3-2
4-1
4-2
4-3
7-2
7-4
9-2
9-3
9-4
Initial
1.83
1.81
2.02
2.20
1.99
1.81
1.70
1.78
3.74
0.53
5.29
2.56
0.67
me
0.01 days
--
--
1.91
1.72
--
2.38
~
of CN" per gram of Carbon at:
1 day
1.77
1.71
1.99
1.45
1.80
1.58
1.57
1.74
--
--
3.86
2.12
0.58
2 days
1.01
1.33
1.67
1.33
1.77
1.68
1.57
1.74
2.95
0.45
4.10
2.25
0.56
3 days
_ _
1.49
1.76
1.63
1.47
1.64
4 days
--
--
1.43
1.69
1.76
1.64
1.65
2.84
0.39
4.01
1.98
0.50
5 days
1.37
0.84
1.17
__
__
__
__
__
__
_ _
-_
6 days
_
«~
_ _
__
_ .
__
_
__
3.16
1.60
0.58
7 days
__
_ _
_ _
1.58
_ _
1.28
.
. _
. _
CT>
-------
GLOSSARY
acclimation: Period during which the nvicroogranisms become accustomed to
their new environment and substrate.
activated sludge process: A method of secondary wastewater treatment in which
a mixture of wastewater and activated sludge is agitated and aerated.
The activated sludge is subsequently separated from the treated waste-
water (mixed liquor) by sedimentation, and wasted or returned to the
process as needed.
aeration: The bringing about of intimate contact between air and liquid by
one of the following methods: spraying the liquid in the air, bubbling
air through the liquid, or agitation of the liquid to promote surface
absorption of air.
apparent loading: The amount of soluble organic material [as measured by any
suitable test such as TOC) removed by a Powdered Activated Carbon (PAC)
system, less the amount of removed by a biological system at the same
sludge age divided by the applied carbon dosage.
Aav,0n+ i«a,Hr,« - AOrganlcs CPAC) - AOrganlcs (BIO)
Apparent Loading -- * App1i -
App1ie >n osage
average daily flow: The average quantity of wastewater passing through the
system in a 24 hour period.
biochemical oxygen demand (BOD): A measure of the oxygen necessary to satisfy
the requirements for the aerobic decomposition of the decomposable orga-
nic matter 1na liquid by bacteria. The standard (BOD5) is five days at
20° C.
biological fouling: A decrease 1n the real loading on the Powdered Activated
Carbon (PAC) as a result of micro-organtsms and their by-products on the
carbon surface.
biological seed: Sludge which is mixed with composed wastewater for the
purpose of Introducing acclimated organisms, thereby accelerating the
initial stages of biological growth and the decomposition of organtcs
in the wastewater.
chemical oxygen demand (COD) : A measure of the oxygen required to approach
total oxidation of the organic matter 1n the waste.
97
-------
clarifier: A tank or basin, in which wastewater is retained for a sufficient
time, and in which the velocity of flow is sufficiently low to remove
by gravity a part of the suspended matter.
coefficient of determination (R2): Statistical device used to measure the
"goodness of fit" of the regression analysis defined by the following
equation:
/\
R2 _ regression sum of squares _
total sum of squares
where:
Y-j = observed y
^
Y-J = y calculated from regression coefficients
y = mean value of observed y
completely mixed activate sludge: Treatment system in which the untreated
wastewater is instantly mixed through the entire aeration basin.
composite sample: Integrated sample collected as a portion of the total flow.
complex cyanides: Compounds containing the cyanide group, CN~, in combination
with other elements forming a complex radical, such as ferricyanide and
ferrocyanide, Fe(CN);3 and Fe(CN)6", respectively.
CuCl2: Cupric chloride.
cyanides: Compounds containing the CN" group, in both the simple and/or comp-
lex modes,unless otherwise noted.
detention time: Period of time required for a liquid to flow through a tank
or unit.
dissolved oxygen (DO): Free or uncombined oxygen in a liquid.
effluent: Liquid flowing out of the treatment system.
filtrate: Liquid passing through a filtering medium with a specified pore
size.
floe: Small gelatinous masses, formed in water or wastewater by biochemical
processes, or by agglomeration.
flocculation: The bringing together of flocculating particles by hydraulic
or mechanical means.
influent: Liquid flowing into a basin or treatment plant.
mean: Average value of all observations defined x = _j!l
n = number of observations n
Xj = value of "ith" observation
98
-------
milligrams per liter (mg/A): The weight of material in one liter of liquid.
mixed liquor (ML): A mixture of sludge and wastewater in a biological reac-
tion tank undergoing biological degradation in an activated sludge system.
nutrient: Any substance absorbed by organisms which promotes growth and re-
placement of cellular parts, such a phosphorous.
PAC: Powdered activated carbon.
pH: The negative logarithim of the hydrogen ion concentration. It is used
to express the intensity of the acid or alkaline condition of a solution.
settleable solids: Suspended solids which will settle in sedimentation basins
(clarifiers) in normal detention times.
simple cyanides: Cyanide that may be determined by the Wood River modified
Roberts-Jackson procedure. This procedure measures the amount of free
cyanide as HCN, which includes readily hydrolyzed metal cyanide com-
plexes such as nickel and copper cyanides. Cyanides not included are
bound in complex form, such as ferricyanide, and ferrocyanide.
sludge: The accumulated settled solids separated from wastewater in clari-
fiers, and containing more or less water to form a semi-liquid mass.
sludge age: The average total time of detention of a suspended solids parti-
cle in a system. It is defined as the total weight of suspended solids
in the aeration basin divided by the total weight of suspended solids in
the effluent and otherwise wasted per day.
sludge volume index (SVI): The volume of sludge occupied by one gram of
activated sludge after 30 minutes of quiescent settling in a 1,000 ml
graduated cylinder.
standard deviation(s): Measure of the deviation of the x's form the sample
mean. The square root of the variance, which measures the dispersion
of the distribution.
For paired observations the sample standard deviation is:
s = /£(xi - x2) - (mean difference)'
n
Then the unbiased standard deviation is:
s
/ n-1
substrate: Raw waste feed on which a microorganism grows or is placed to
grow by decomposing the waste material.
99
-------
substrate removal: The total BOD in plant influent, minus the soluble BOD in
plant effluent, divided by the total influent BOD.
supernatant liquor: The liquid overlying settled solids.
suspended solids (SS or TSS): The quantity of material deposited when a quan-
tity of wastewater is filtered through an asbestos mat in a Gooch crucible
or equal method.
t-test: To test a hypothesis assuming a normally distributed variable in the
t-distribution can be used to determine the probability level of occur-
rence. It is defined as follows:
T = (x-yo) n
Vl ~ §
where:
x = mean sample value
y0 = true mean hypothesized
n = number of observations
n-1 = degrees of freedom
s = standard deviation
total organic carbon (TOC): The amount of carbon, measured as C, in the form
of organic compounds, dissolved in an aqueous solution.
total solids (TS): The solids in the wastewater, both suspended and dissolved.
volatile suspended solids (VSS): The quantity of suspended solids in waste-
water that are lost on ignition of the total suspended solids.
waste sludge: Sludge which is no longer needed in the treatment system and
therefore is removed and disposed of.
100
-------
TECHNICAL REPORT DATA
(Please read Instructions on the reverse before completing)
1. REPORT NO.
EPA-600/2-80-125
2.
3. RECIPIENT'S ACCESSION-NO.
4. TITLE ANDSUBTITLE
Cyanide Removal From Refinery Wastewater Using Powderec
Activated Carbon
5. REPORT DATE
_ May 1980 .
6. PERFORMING ORGANIZATION CODE
7. AUTHOR(S)
James E Huff
Tpffrpy Mr B-i ,
8. PERFORMING ORGANIZATION REPORT NO.
9. PERFORMING ORGARfZATION NAME AND ADDRESS
IIT Research Institute
10 West 35th Street
Chicago, IL 60616
10. PROGRAM ELEMENT NO.
1BB610
11. CONTRACT/GRANT NO.
EPA R8QAQ2-9-01
12. SPONSORING AGENCY NAME AND ADDRESS
Robert S. Kerr Environmental Research Laboratory
Office of Research and Development
U.S. Environmental Protection Agency
Ada, Oklahoma 74820
13. TYPE OF REPORT AND PERIOD COVERED
Final 9/19752/1977
14. SPONSORING AGENCY CODE
EPA/600/15
15. SUPPLEMENTARY NOTES
16. ABSTRACT
The objective of this project was to evaluate the removal of low level
cyanide in petroleum refinery wastewater by the addition of powdered activated
carbon and cupric chloride to an activated sludge unit. The activated carbon
and cupric chloride act as catalyst in the oxidation and destruction of the
cyanides.
A two-phase study was carried out to develop the process. The first phase
consisted of a bench-scale study using solutions of metal-cyanide complexes
in order to determine the mechanics of cyanide destruction. The second phase
consisted of bench-scale tests using actual refinery wastes.
17.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
b.IDENTIFIERS/OPEN ENDED TERMS
COSATl Field/Group
Activated Carbon Treatment
Cyanides
Activated Sludge Process
Petroleum Refining
Cupric Chloride
68D
18. DISTRIBUTION STATEMENT
Release to Public
19. SECURITY CLASS (This Report)
Unclassified
21. NO. OF PAGES
109
20. SECURITY CLASS (Thispage}
Unclassified
22. PRICE
EPA Form 2220-1 (9-73)
101
US GOVERNMENT PRINTING OFFICE: 19(0-657-146/5701
------- |