EPA/600/2-86/099
LIMESTONE BED CONTACTORS FOR
CONTROL OF CORROSION AT SMALL
WATER UTILITIES
by
Raymond D. Letterman
Charles T. Driscoll, Jr.
Marwan Haddad
H. Alan Hsu
Syracuse University
Syracuse, New York 13210
Cooperative Agreement No. CR-809979-01-3
Project Officer
Gary S. Logsdon
Drinking Water Research Division
Water Engineering Research Laboratory
Cincinnati, Ohio 45268
WATER ENGINEERING RESEARCH LABORATORY
OFFICE OF RESEARCH AND DEVELOPMENT
U.S. ENVIRONMENTAL PROTECTION AGENCY
CINCINNATI, OHIO 45268
PB87-112058
-------
FOREWORD
The U.S. Environmental Protection Agency is charged by Congress with
protecting the Nation's land, air, and water systems. Under a mandate of
national environmental laws, the agency strives to formulate and implement
actions leading to a compatible balance between human activities and the
ability of natural systems to support and nurture life. The Clean Water
Act, the Safe Drinking Water Act, and the Toxic Substances Control Act are
three of the major congressional laws that provide the framework for restoring
and maintaining the integrity of our Nation's water, for preserving and en-
hancing the water we drink, and for protecting the environment from toxic
substances. These laws direct the EPA to perform research to define our
environmental problems, measure the impacts, and search for solutions.
The Water Engineering Research Laboratory is that component of EPA's
Research and Development program concerned with preventing, treating, and
managing municipal and industrial wastewater discharges; establishing prac-
tices to control and remove contaminants from drinking water and to prevent
its deterioration during storage and distribution; arid assessing the nature
and controllability of releases of toxic substances to the air, water, and
land from manufacturing processes and subsequent product uses. This publi-
cation is one of the products of that research and provides a vital communi-
cation link between the researcher and the user community.
Use of limestone contactors to raise the pH, calcium content, and alkalinity
of low pH, soft water was evaluated in this project. Studies were conducted
in pilot plant cplumns at Syracuse University, and field evaluations of three
types of contactors were carried out at Big Moose Lake in the Adirondacks.
The limestone contactors were shown to be capable of reducing the corrosive
tendency of water, as measured by copper and lead concentration increases
in water held in plumbing of cottages at Big Moose Lake. The model for water
quality changes in a limestone column hold promise for application to design
of limestone beds placed in upwelling zones (submerged springs) on the beds
of acidified lakes.
11
-------
DISCLAIMER
The information in this document has been funded in part by the United States
Environmental Protection Agency under assistance agreement number
CR-809979-01-3 to Syracuse University. It has been subject to the Agency's
peer and administrative review, and it has been approved for publication
as an EPA document. Mention of trade names or commercial products does not
constitute endorsement or recommendation for use.
iii
-------
ABSTRACT
A study was conducted to investigate the use of limestone contactors
as a technique for mitigating corrosion in small water supply systems that
use dilute acidic water. As water is transported through a packed bed of
crushed limestone, calcium carbonate dissolves and the pH, calcium ion concen-
tration, and alkalinity increase.
A mathematical model was derived for use in contactor design. The model
is based on the interfacial transport of calcium ion and relates the depth
of limestone required in the contactor to the desired effluent water chemistry,
influent water chemistry, limestone particle size and shape, limestone bed
porosity, and water temperature, and superficial velocity. The model was
calibrated and tested using laboratory column experiments.
In a contactor monitored for 2.5 years (except for the initial few months)
the water quality following treatment was essentially constant. No gradual,
long-term degradation in performance was noted. After several months of
operation, however,the rate of CaC03 dissolution was not as high as that
observed in the laboratory using fresh limestone. The rate of dissolution
is possibly reduced by an alumino-silicate residue that remains after the
CaC03 is dissolved from the limestone matrix. A microbiological film may
also have been a limiting factor.
Field study results indicated that limestone contactors can be used
to effectively reduce the tendency of water to take up corrosion byproducts
(copper, lead, and zinc) from surfaces in piping systems. Copper and lead
concentrations in first-flush samples of cottage tapwater receiving untreated
spring water were 1.9 + 0.31 mg Cu/L and 0.046 +_ 0.004 mg Pb/L, respectivley.
Contactor-treated water at Bay Side cottage contained copper concentrations
of 0.030 + 0.037 mg Cu/L and lead concentrations of 0.0084 + 0.0084 mg Pb/L.
This report was submitted in fulfillment of Cooperative Agreement CR-809979-01-3
by Syracuse University under the sponsorship of the U.S. Environmental Protec-
tion Agency. This report covers the period August 1, 1982 to July 31, 1985
and work was completed.as of July 31, 1985.
iv
-------
CONTENTS
Foreward ii
Abstract iv
I List of Figures vii
List of Tables xii
Acknowledgments xiv
Section 1
Introduction 1
Statement of Problem 1
Study Objective 1
Section 2
Conclusions 3
Section 3
Recommendat ions 5
Section 4
Literature Review 6
Introduction 6
Limestone Properties 6
Kinetics of Limestone Dissolution .\. 8
Packed Bed Reactors 22
Metal Release from Pipes 24
Section 5
Methods and Materials 32
Apparatuses - Laboratory and Field Contactor Units 32
Laboratory contactors 32
Field Contactors 34
Limestone Characteristics 40
Limestone Bed Characteristics 45
Pipesection Procedures 55
Sampling and Analytical Procedures 55
General Procedures 55
Laboratory contactors 57
Quality Assurance/Quality Control Information Data 57
Computative Analysis 65
Section 6
Derivation of Contactor Design Equations 69
Equilibrium Calcium Concentration 71
Section 7
Results and Discussion 74
Model Verification 74
Equilibrium Calcium Concentration 74
Contactor Design Equations 81
v
-------
Section 7 (con't)
Field Study Results 94
Baffled Box Contactor 94
Bayside Cottage Wound Fiberglass Column 116
Culligan (Cullneu®) Contactor 119
Evaluation of Contactor Design Equations Using
Field Measurements 127
Sensitivity Analysis - Design Equations 133
Thermodynamic Calculations of Trace Metal Chemistry 138
Pipe Leaching Experiments 156
Metal Release from Field Site 164
Spring Contactor Treatment 164
Lake Contactor Treatment 167
References 177
Appendices
A - Chemical Equilibrium Model used in Contactor Design Equation.... 184
B - Dissolution Rate Data from Column Experiments 199
C - Estimates of Limestone Contactor Costs 205
VI
-------
FIGURES
Number Page
1 Locations of major chalk and limestone deposits in the
continental United States 9
2 Schematic representation of the calcium carbonate dissolution
process 13
3 Initial rate of calcite dissolution as a function of the bulk
solution pH from Sjoberg and Rickard (1984a) 18
4 Initial rate of calcite dissolution as a function of bulk
solution pH and partial pressure of carbon dioxide (Plummer
et al. , 1975) 19
5 Initial rate of calcite dissolution as a function of the
square root of rotating disk rotational speed 21
6 Laboratory columns with water supply and flow control system...33
7 Baffled-box contactor used in the field study 35
8 Wound-fiberglass and Culligan contactors used in the field
study 36
9 Map of the Covewood lodge property located near Old Forge,
N.Y. Site of the field study 38
10 Diagram showing the installation of the baffled-box contactor
in the spring at Covewood 39
11 Measured porosity plotted as a func tion of container surface
area to volume ratio for four limestone particle effective
diameters 46
12 Measured effluent tracer concentration plotted as a function
of time elapsed after tracer injection for four values of the
superficial velocity 50
13 Mean residence time calculated using the superficial velocity
and measured porosity plotted as a function of the mean
residence time from the tracer experiments 53
14 Measured effluent tracer concentration plotted as a function
of the time elapsed after tracer injection for the baffled-box
contactor (Figure 7) 54
15 Measured and calculated alkalinity for field measurements 61
16 pH plotted as a function of the axial distance to the sampling
port and influent pH, pHo 75
VII
-------
Number Page
17 Calcium concentration plotted as a function of the axial
distance to the sampling port and influent pH, pHo 76
18 Dissolved inorganic carbon concentration plotted as a
function of the axial distance to the sampling port and
influent pH, pHo 77
19 Equilibrium pH, dissolved inorganic carbon and calcium concentrations
plotted as a function of the influent pH and the
following conditions; Curve A - closed systen and C^Q = 0;
Curve B closed system and C^Q = 28 mgCa/L; Curve C - closed/
open system and Cfco = 0; Curve D - closed/open systen and
Cbo = 28 mgCa/L 79
20 Influent calcium concentration plotted as a function of the
influent dissolved inorganic cargon concentration and the
equilibrium pH for an influent pH of 6.0 80
21 Sum of the square of the difference between the observed and
the model predicted calcium concentration plotted as a
function of the dissolution rate constant for run number 32,
Appendix B 83
22 Model predicted and measured calcium concentrations plotted as »
a function of the axial distance to the sampling port for run
number 32 and Ko = 0.032 cm/min 84
23 In [CbL - Ceq)/(Cb0 - Ceq)] plotted as a function of the
axial distance to the sampling port for runs 29, 31 and 32 85
24 Dissolution rate constant determined by the least squares
method (Method II) plotted as a function of the value obtained
using plots such as Figure 23 (Method I) 86
25 Mass transfer factor, JQ, plotted as a function of a modified
Reynold's number using the equations derived by Chu and
Khalil (1953) 88
26 Values of the dissolution rate constant calculated using the
model equations plotted as a function of the experimental
(best-fit) values listed in Appendix B 90
27 Observed calcium concentration plotted as a function of the
model predicted value. The points include all sampling port
locations for the runs listed in Appendix B 91
28 Model predicted and measured pH plotted as a function of the
axial distance to the sampling port for run number 32 and Ko =
0.032 cm/min 93
29 Observed pH plotted as a function of the model predicted values
for all sampling port locations for the runs listed in
Appendix B 95
viii
-------
Number Page
31 Model predicted and measured alkalinity plotted as a function
of the axial distance to the sampling port for run number
32 and Ko = 0.032 cm/min 98
32 Measured change in alkalinity within the laboratory contactors
plotted as a function of the model predicted change 99
33 Water temperature plotted as a function of time for the
baffled-box contactor 100
34 Influent and effluent pH plotted as a function of time for the
baffled-box contactor 101
35 Influent and effluent calcium concentration plotted as a
function of time for the baffled-box contactor 103
36 Influent and effluent alkalinity plotted as a function of time
for the baffled-box contactor 104
37 Influent and effluent dissolved inorganic carbon concentration
plotted as a function of time for the baffled-box contactor.... 105
38 Influent and effluent standard plate count bacteria concentration
plotted as a function of time for the baffled-box
contactor 106
39 Influent and effluent total coliform bacteria concentration
plotted as a function of time for the baffled-box contactor.... 107
40 Calculated partial pressure of carbon dioxide plotted as a
function of time for the influent and effluent of the
baffled-box contactor Ill
41 X-ray energy spectra for the following samples: A - fresh
limestone, B - limestone after prolonged dissolution in the
baffled-box contactor, compartment 1, C - same as B except
compartment 5, D - limestone after prolonged dissolution
in the laboratory 117
42 Influent and effluent pH plotted as a function of time for the
wound-fiberglass contactor in Bayside Cottage 120
43 Influent and effluent calcium concentration plotted as a
function of time for the wound-fiberglass contactor in
Bayside Cottage 121
44 Influent and effluent alkalinity plotted as a function of
time for the wound-fiberglass contactor in Bayside Cottage 123
45 Influent and effluent dissolved inorganic carbon concentration
plotted as a function of time for the wound-fiberglass contactor
in Bayside Cottage 123
ix
-------
Number Page
46 X-ray energy spectrum for a limestone sample taken from the
wound-fiberglass contactor at the end of the experiment 125
47 X-ray energy spectra for fresh Cullneu® medium (A) and Cullneu®
used in the Culligan contactor for 9 months (B) 128
48 Total depth of limestone required to obtain an effluent pH
of 8.5 plotted as a function of the ionic strength 140
49 Predominance area diagram for the stability of lead passivation
films over a range of pH and dissolved inorganic carbon
concentrations at 25°C p*Kso = -8.15 142
50 Predominance area diagram for the stability of lead passivation
films over a range of pH and dissolved inorganic carbon
concentrations at 250°C p*Kso = -13.07 143
51 Lead concentrations calculated with the chemical equilibrium
model MINEQL as a function of pH for several concentrations of
dissolved inorganic carbon 144
52 Predominance area diagram for the stability of lead passivation
films over a range of pH and partial pressures of CC>2 at
25°C p*Kso = 8.15 146
53 Predominance area diagram for the stability of lead passivation
films over a range of pH and partial pressures of CC>2 at
25°C p*Kso = 13 . 07 147
54 Lead concentrations calculated with the chemical equilibrium
model MINEQL as a function of pH for several partial pressures
of C02 148
55 Predominance 'area diagram for the stability of copper passivation
films over ranges of pH and dissolved inorganic carbon
concentrations at 25°C 149
56 Copper concentrations calculated with the chemical equilibrium
model MINEQL as a function of pH for several concentrations
of dissolved inorganic carbon 150
57 Copper concentrations calculated with the chemical equilibrium
model MINEQL as a function of pH for several partial pressures
of C02 151
58 Predominance area diagram for the stability of zinc passivation
films over a range of pH and dissolved inorganic carbon concen-
trations at 25°C 152
-------
Number Pagt
59 Zinc concentrations calculated with the chemical equilibrium
model MINEQL as a function of pH for several dissolved inorganic
carbon concentrations 153
60 Predominance area diagram for the stability of zinc passivation
films over ranges of pH and partial pressures of CC>2
at 25°C 154
61 Zinc concentrations calculated with the chemical equilibrium
model MINEQL as a function of pH for several partial pressures
of C02 155
62 Lead concentrations from lead pipe sections leaching
experiments 157
63 Zinc concentrations from galvanized steel pipe section
leaching experiments 158
64 Total and filtered concentrations of lead from lead pipe section
leaching experiments 159
65 Variations in pH (a), dissolved inorganic carbon (DIG) concen
tration (b) and measured copper (c) and lead (d) concentrations
from pipe section leaching experiments as a function of column
treatment by CaCC>3 160
66 Copper concentrations from copper pipe section leaching
experiments at various levels of CaCO^ treatment (variations
in pH) 162
67 Lead concentrations from copper pipe section with lead-tin
solder leaching experiments at various levels of laboratory
CaCC>3 treatment (variations in pH) 163
68 The probability of copper concentrations in untreated and
CaCC>3 treated lake and spring waters exceeding a given
concentration 168
69 The probability of lead concentrations in untreated and CaCC>3
treated lake and springwaters exceeding a given concentration..169
70 A comparison of measured copper concentrations from first flush
tapwater derived from CaCC>3 treated lakewater and calculated
values from the chemical equilibrium model MIEQL as a function
of pH 173
71 A comparison of measured copper concentrations from first flush
tapwater derived from CaC03 treated lakewater and calculaated
values from the chemical equilibrium model MINEQL 174
72 A comparison of measured lead concentrations from first flush
tapwater derived from CaC03 treatment and calculated values
from the chemical equilibrium model MINEQL as a function
of pH 176
xi
-------
TABLES
Number Page
1 Representative Chemical Analysis of Different Types of Limestone
(from Boynton, 1980) 7
2 Major Impurities in High Calcium Limestone (45 U. S. Samples
(from Murray et al., 1954) 10
3 Minor Impurities in High Calcium Limestone (25 U. S. Samples)
(from Murray" et al. , 1954) 11
4 Oxidation Potential of Metallic Materials 26
5 Passivation Film Minerals That May Be Important In Regulating
Metal Solubility to Water Distribution Systems 29
6 Effective Solubility of Crushed Limestone Experimental
Results 42
7 Limestone Particle Size and Sphericity Analysis Results 44
8 Bed Porosity and Limestone Particle Surface Area Per Unit
Volume of Interstitial Water 48
9 Results of Tracer Response Measurements Obtained Using Laboratory
Columns (Figure 6) 51
10 Analytical Methods 56
11 Summary of Sampling and Analytical Precision from Sample
Triplicate Program 59
12 Estimates of Sample Collection and Analytical Precision from
4x4 Analysis for Big Moose Lake 60
13 Summary of Blind Sample Analysis Obtained from USEPA Clinic
Municipal Environmental Research Laboratory 62 & 63
14 Summary of USEPA CERL of Blind Audit Analysis. All Values in
eq 1 Except Where Indicated 64
15 Equilibrium Constants at 25°C for the Solids Considered in the
MINEQL Calculations . . 66
16 Reactions and Equilibrium Constants at 25°C for the Aqueous
Complexes Considered in the MINEQL Calculations 67
17 Summary of Baffled-Box Contactor Results Field Measurements . . . 109
xii
-------
Number Page
18 Baffled-Box Contactor - Limestone Dissolution June 28, 1982
September 26, 1983 114
19 Summary of Bay Side Cottage Wound Fiberglass Column Results. . . . 124
20 Culligan Contactor - Summary of Results November 3, 1983 July 31,
1984 126
21 Baffled-Box and Wound Fiberglass Contactors - Special Test of
Model Equations. Experimental Conditions and Results 129
22 Special Test of Model Equations Calculated Equlibrium pH and
Calcium Concentration 130
23 Results of Field Test of Model Equations 131
24 Results of Chemical Equilibrium Model Calculations 135
25 Sensitivity Analysis Results 137
26 Effect of Ionic Strength on the Equilibrium and Contactor
Effluent (pH = 8.5) Calcium Concentrations 139
27 Comparison of Trace Metal Concentration (as mg/1) in Spring Water
and from the First Flush of Treated (Hillside, Bay Side) and
Untreated (Covewood) Cottages 165
28 Comparison of Copper and Lead Concentrations (Mean ± Std. Dev.
as mg/1) from First Flush and Three Minutes of Flowing Tapwater
Derived from the Box Contactor Treated Spring 166
29 Metal Concentrations (as mg/1) in Lake Influent, Untreated and
Treated First Flush Tapwater at Bay Side 171
30 Comparison of Copper and Lead Concentrations (Mean ± Std. Dev,.
as mg/1) from First Flush and After Three Minutes of Flowing
Tapwater Derived from Both CaCCL Treated and Untreated
Lakewater 172
xiii
-------
ACKNOWLEDGMENTS
The field study part of this project was conducted with the help and
cooperation of C.V. "Major" Bowes, proprietor of Covewood Lodge on Big Moose
Lake. The first contactor at Covewood was built and installed by Major Bowes
and it was his interest in water quality that led to our first measurements
and eventually to this research project. We would also like to acknowledge
Dr. Gary S. Logsdon of the U.S. Environmental Protection Agency for his con-
tinuous assistance, review comments and patience.
xiv
-------
SECTION 1
INTRODUCTION
STATEMENT OF PROBLEM
In many areas of the United States individual homeowners and small public
and private water supply systems use water that is potentially corrosive
to the materials used in the water distribution system. Corrosion is a concern
to the owners and users of small water supply systems because of the potential
health problems associated with the ingestion of corrosion byproducts, the
degradation of the esthetic quality of the water and the significant economic
consequences of piping system deterioration.
Corrosion and the contamination of the water by corrosion by-products
may be caused by the use of dilute acidic waters that generally have low
pH, alkalinity and concentrations of dissolved solids. Dilute acidic ground
and surface waters are found in a number of regions of the country, particu-
larly in regions underlain by siliceous bedrock. These waters are naturally
low in buffering capacity and they are corrosive. They are also prone to
acidification by atmospheric deposition of strong acids (acid precipitation)
or other factors such as changes in land use. In some areas (for example
the Adirondack Region of New York State) it is possible that the corrosivity
of water has been increased by acidic deposition. In any event, until recent
concern developed about acidic deposition and the deterioration of water
quality as a result of acidification residents and visitors tolerated or
ignored the problems caused by the use of corrosive water. Now this indiffer-
ence has changed to a significant concern, and many home and resort owners
as well as those responsible for village water supplies have begun to adopt
techniques designed to mitigate drinking water corrosivity.
Low cost is a very important criterion in establishing the feasibility
of a corrosion mitigation technique for a small water supply system. Also
the maintenance required should be minimal, and the technique should present
a low potential for public health hazard resulting from improper construction,
installation or maintenance. Limestone contactors are water treatment devices
that tend to meet these requirements.
In a limestone contactor water flows through under a closed-to-the-atmo-
sphere condition and dissolves a packed bed of crushed limestone. The chemistry
of the water is altered as the limestone dissolves. Sources of CaCO, other
than high calcium limestone (eg., seashells,) are sometimes used. Limestone
contactors are simple but effective devices with low capital cost and minimum
maintenance requirements. They have been used for the neutralization of
acid mine drainage, acidic industrial wastes, and dilute acidic surface waters.
STUDY OBJECTIVES
The overall objective of this project was to investigate the use of
limestone contactors as a technique for the mitigating corrosion in small
-------
water supply systems that use dilute acidic water. The research plan included
the development and testing of a rational method for contactor design and
the evaluation of the field operation of a contactor with respect to corrosion
control and operation and maintenance problems. The study had the following
specific objectives:
(1) to derive and test (using laboratory, column-type reactors) a mathe-
matical model for limestone contactor design,
(2) to develop design objectives by experimentally determining the
relationship between contactor-treated water quality and metal
release from pipes, and
(3) to evaluate the practical application of the design equations and
objectives by monitoring the field performance of full-scale contac-
tors and to determine the feasibility of long-term operation and
the type and frequency of maintenance required.
-------
SECTION 2
CONCLUSIONS
As dilute acidic water is transported through a packed bed of crushed
limestone, calcium carbonate in the limestone dissolves, the pH, calcium ion
concentration, and alkalinity increase, and the tendency may decrease for
water to dissolve corrosion by-products from surfaces in piping systems.
The depth of limestone, L, required to achieve a given level of treatment
can be calculated using a mathematical model based on interfacial transport
of calcium ion,
ln[(Ceq - CbL)/(Ceq - Cbo)]
L = K°ae OH rK°ae~i 2
- 2 d —
Us
where a is the interfacial area of limestone per unit volume o_f interstitial
water, e is the bed porosity, Us is the superficial velocity, d is the effective
diameter of the limestone particles, and KQ is the overall CaC03 dissolution
rate constant. C^Q is the influent calcium concentration. The results of
this study indicate that KQ can be estimated using an existing correlation
of dimensionless mass transfer parameters. The quantity "a" can be estimated
using d and the particle sphericity. The equilibrium and effluent calcium
concentrations, Ceq and C^L > are determined using a chemical equilibrium
model. The magnitudes of these parameters are a function of the characteristics
of the influent solution, particularly the temperature, pH, and calcium and
dissolved inorganic carbon (DIG) concentrations. As the influent calcium
and/or DIG concentrations increase the maximum pH (pHeq) that can be attained
in a contactor decreases, and the depth of limestone required to reach a given
effluent pH « pHeq) increases. The depth of limestone required to achieve
a given treatment objective also increases with decreasing influent pH, increasing
superficial velocity, and increasing limestone particle size.
An evaluation of a limestone contactor in the field suggests that except
for the initial few months, water quality following treatment was constant
through the 2.5-year study period. There was no evidence of a gradual, long-
term reduction in performance. However, after 3 or 4 months of continuous
operation, the rate of CaC03 dissolution was not as high as that predicted
by the laboratory results obtained with fresh limestone. Analysis of the
limestone surfaces by x-ray energy spectrometry indicated that prolonged opera-
tion altered the surface of the limestone; the relative abundance of calcium
on the surface decreased, and aluminum and silicon increased. Apparently,
alumino-silicate impurities in the limestone remained as a thin "residue"
after the CaC03 was leached from the limestone surface matrix. This residue
may have slowed the dissolution rate. Also possible is that the dissolution
process was adversely affected by a microbiological film on the limestone.
The model developed for contactor design assumes that the water flows
through the limestone under a closed-to-gaseous carbon dioxide condition.
-------
Equilibration of the column effluent with atmospheric carbon dioxide can have
a significant effect on the pH of the solution and hence on the tendency to
dissolve corrosion by products. When the influent DIG is high, (e.g., greater
than 10 mg C/L), equilibration with the atmosphere may cause the pH to increase.
When the influent DIG concentration is less than several mg C/L, the pH tends
to decrease.
Results of the study suggest that dilute acidic waters facilitate the
release of elevated concentrations of trace metals from metal piping systems.
Passivation films of most significance include Cu2(OH)2 CC-3 and Cu(OH)2 for
copper, PbC03, Pb3(OH)2(C03>2 or Pb(OH)2 for lead and ZnsCOH^CCC^^ for zinc.
Because of the pH and inorganic carbon-dependent solubility of these minerals,
metal corrosion can generally be mitigated by increases in pH and dissolved
inorganic carbon concentrations. However, elevated inorganic carbon concentrations
coupled with high pH values can facilitate the solubilization of trace metals
through the formation of soluble metal carbonate complexes. This problem
is most significant for lead, as copper and zinc do not form strong aqueous
complexes with carbonate.
Laboratory pipe section experiments using copper pipe with lead-tin solder
indicate that limestone contactor treatment reduces copper, and to a smaller
extent, lead leaching. Theoretical thermodynamic calculations were consistent
with measured copper concentrations in the neutral pH (pH 6.5 to 7.5) region.
However, copper concentrations in acidic waters (pH _< 6.0) were substantially
undersaturated with respect to theoretical metal solubility. Lead derived
from lead-tin solder in pipe section experiments was highly undersaturated
with respect to the solubility of lead passivation films.
Trace metal field results were generally consistent with laboratory obser-
vations. Spring and lake waters with and without limestone contactor treatment
were corrosive. Elevated metal concentrations were observed in first-flush
tapwater from both treated and untreated cottages. Running tapwater (3 minutes)
significantly reduced copper, lead, and zinc concentrations. Although treated
waters were generally corrosive, trace metal concentrations were significantly
reduced in both treated spring and lake water, relative to untreated water.
For example, first-flush copper and lead concentrations in cottage tapwater
receiving untreated spring water were 1.9 + 0.31 mg Cu/L and 0.0046 + 0.004
Pb/L, respectively. While treated spring water at Bay Side cottage contained
copper concentrations of 0.030 +_ 0.037 mg Cu/L and lead concentrations of
0.0084 +_ 0.0084 mg Pb/L. Likewise, CaCC>3 treatment of acidic lakewater at
Bayside cottage significantly reduced copper concentrations in first-flush
tapwater from 1.9 + 0.35 mg Cu/L to 0.54+0.30 mg Cu/L and reduced lead concen-
trations from 0.033+ 0.009 mg Pb/L to 0.015 + 0.014 mg Pb/L. Limestone treat-
ment greatly reduced the probability of metal concentrations exceeding the
secondary MCL of 1.0 mg Cu/L from greater than 75% to less than 15%. The
probability of consuming elevated lead concentrations in first-flush tapwater
was also reduced by CaC03 treatment.
Measured trace metal concentrations from first-flush tapwater were compared
with theoretical calculations from the chemical equilibrium model. Generally
untreated lakewater was highly acidic (pH 4.6), and measured copper concentra-
tions were highly undersaturated with respect to the theoretical solubility
of Cu2(OH)2CC>3. However, following treatment, measured copper values
were in close agreement to thermodynamic predictions. Concentrations of lead
largely derived from lead-tin solder were highly undersaturated with respect
to the solubility of lead passivation films.
-------
SECTION 3
RECOMMENDATIONS
The results of this study suggest that as calcium carbonate is dissolved
from the limestone particle matrix a layer of residue forms. It appears that
transport across this layer eventually limits the overall rate of dissolution.
This has important implications for the design of a contactor for long term
use. In this study there was limited evidence that the performance of a contac-
tor operated for two years in the field was influenced by the formation of
a residue layer.
It seems reasonable to assume that the rate of build-up of a residue
layer will be a function of the level of insoluble impurities in the limestone.
Additional research is needed to determine this relationship. Long-term exper-
iments should be conducted using contactor columns filled with limestones
of varying purity. The rotating disk apparatus has been used to effectively
study the kinetics of calcite dissolution and should be considered for use
in measuring the effect of limestone purity on the rate of dissolution. Until
the significance of the residue layer has been determined the results of this
study should be used with an awareness that limestone purity may be an important
variable.
-------
SECTION 4
LITERATURE REVIEW
INTRODUCTION
This review of the literature is divided into three parts. In the first
section the characteristics of limestone are discussed. Limestone is, in
most parts of the country, a readily available and inexpensive source of
CaCO~. However because it is a natural material, its physical and chemical
characteristics are variable and this variability may affect its use as a
neutralizing substance.
The engineering design of a limestone contactor requires an understanding
of the kinetics of the neutralization (CaCO~ dissolution) reaction. This
topic is covered in the second part of the literature review.
In the third part of the literature review the effect of water chemistry
on the release of corrosion by-products such as lead from lead-tin solder
and copper from copper pipes is discussed.
LIMESTONE PROPERTIES
Limestone is a general term used to describe sedimentary rock composed
primarily of calcium carbonate or combinations of calcium and magnesium carbon-
ate with varying amounts of impurities, the most common of which are silica
and alumina. There are numerous forms and types of limestone, varying in
chemical composition, mineralogy, crystallinity, color, texture and hardness.
Next to sand and gravel, limestone, including all of its carbonate forms,
is the second greatest tonnage material produced in the United States.
The two most fundamental types of limestone are high calcium and dolomitic
limestone. Pure high calcium limestone is 100 percent calcium carbonate
(calcite or aragonite). Pure dolomite is 54.37, CaC03 and 45.7% MgC03. High-
quality, high calcium limestone is 97-99% CaCO~. (54-567, CaO). Chemical
analyses for a number of U.S. limestones are summarized in Table 1.
High calcium limestone was used exclusively in this study. Since there
is considerable evidence to suggest that the dissolution rate of dolomitic
limestone is substantially less than high calcium limestone (Pearson and
McDonnell 1975a, 1975b) the results of this study should therefore, only
-------
Table 1 Representative chemical analyses (percentage composition)
of different types of limestone (from Boynton, 19SO)
Limestone Sample*
CaO
MgO
C02
SiOo
A1203
Fe203
so3
P205
Na20
K20
H20
Other
*
1 =
2 =
3 =
4 =
5 =
6 =
7 =
8 =
123456
54.54 38.90 41.84 31.20 29.45 45.65
0.59 2.72 1.94 20.45 21.12 7.07
42.90 33.10 32.94 47.87 46.15 43.60
0.70 19.82 13.44 0.11 0.14 2.55
0.68 5.40 4.55 0.30 0.04 0.23
* 0.08 1.60 0.56 0.19 0.10 0.20
0.31 -- 0.33 -- -- 0.33
0.22 — 0.05 0.04
0.16 -- 0.31 0.06 0.01 0.04
0.72 -- 0.01 0.03
1.55 -- 0.16 0.23
0.29 -- 0.01 0.06
Indiana high calcium stone.
Lehigh Valley, Pa. "cement rock."
Pennsylvania "cement rock."
Illinois Niagaran dolomitic stone.
Northwestern Ohio Niagaran dolomitic stone.
New York magnesium stone.
Virginia high calcium stone.
Kansas cretaceous high calcium (chalk).
7 3
55.28 52.48
0.46 0.59
43.73 41.35
0.42 2.38
0.13 1.57
0.05 0.56
0.01
__
--
__
n. d.
0.08 0.20
-------
be applied to the use of high calcium stone. Active sources (quarries and
mines) of high calcium limestone are present in essentially every state (See
Figure 1).
Care must be taken in selecting a high calcium stone for use in a lime-
stone contactor. Some states have either high calcium or dolomitic or abundant
quantities of both types. The distribution of these materials is however
without a predictable pattern, in some cases they occur in separate broad
expanses, while in other cases both types may be present in close proximity,
for example, on opposite sides of a quarry.
Limestone may contain a number of impurities. Clay, silt and sand (or
other forms of silica) may have become incorporated in the stone when it
was first deposited or material may have collected later in crevices and
between strata. These mineral contaminants are the sources of the major
impurities, silica and alumina. Other impurities, in a rough order of relative
amounts are iron, phosphorus and sulfur. Trace substances such as manganese,
copper, titanium, sodium, potassium, fluorine, arsenic and strontium may
be present.
Murray et al. (1954) analyzed 45 different high calcium limestones from
the United States. The principal impurities are listed in Table 2. All
stone analyzed contained measurable amounts of silica, alumina and magnesium
oxide. Potassium, sodium and sulfur were present in some samples.
Murray et al. (1954) also examined 25 high calcium limestones spectro-
graphically for 25 metallic elements. The findings obtained for elements
other than calcium, magnesium, silicon and iron are listed in Table 3. Alumi-
num, barium, manganese,phosphorus, potassium, sodium, strontium and tin were
detected in all 25 samples. Titanium, zinc and chromium were detected in
20 of the samples. The metals present at a concentration greater than 1000 ppm
in at least one sample are aluminum, manganese, potassium, sodium, strontium,
titanium and zinc.
KINETICS OF LIMESTONE DISSOLUTION
The engineering design of a limestone contactor requires an understanding
of the kinetics of the CaCO, dissolution process and the effect of this dis-
solution on the chemistry of the bulk solution. The overall neutralization/
dissolution reaction is given by
-------
Legend
Chalk deposits
I 1 Limestone deposits
Figure 1. Locations of major chalk and limestone deposits in the continental
United States.
-------
Table 2 Major Impurities in High Calcium Limestone
(45 U.S. Samples) (from Murray et al., 1954)
Si02 0.10 - 2.897o
A1203 0.13 - 0.927o
K0 0.00 - 0.2l7o
Na20 0.00 - 0.167,
S03 0.00 - 0.567,
MgO 0.12 - 3.117,
10
-------
Table 3 Minor Impurities in High Calcium Limestone
(25 U.S. Samples) (from Murray et al., 1954)
Element
Al
Ba
B
Cr
Number of Samples With
Detectable Amount
25
25
3*
20
Co
Pb
Mn
Hg
Mo
Ni
P
K
Ru
Ag
Na
Sr
Sn
Ti
Zn
*trace amounts only
15*
25
4
8*
22
25
25
17*
13*
25
25
25*
23
23
Maximum
Amount
0.35-0.607,
(5 samples)
not given
10 ppm+
(3 samples)
10 ppm+
(2 samples
(1 sample)
0.017,
(1 sample)
0.001-0.017o
(2 samples)
0.27o
(1 sample)
O.l7o+
(3 samples)
0.01-0.17,
(all samples)
O.l7o+
(1 sample)
0. 17,+
(1 sample)
11
-------
CaC03 + H+ > Ca"*"*" + DIG
where, DIG, the dissolved inorganic carbon, includes the species, H CO-
(C02 + H2C03), HC03" and C03=.
The dissolution reaction at the solid surface is influenced by the trans-
fer of the reactants (e.g., hydrogen ion) to the interface and the products
(calcium and DIG species) away from it. In addition, if the objective is
to understand the effect of dissolution on the chemistry of the bulk solution,
the rates of homogeneous reactions involving dissolution products in the
solution and, if a gas phase is present, the rate of transport of inorganic
carbon to or from the aqueous phase must be considered.
A schematic diagram illustrating the overall dissolution process in
a system which includes a CaCO_ solid phase, the aqueous solution and a gas
phase which may contain carbon dioxide or may act as an infinite sink for
CCL released from the aqueous phase is presented in Figure 2. The rate of
change in bulk solution chemistry is affected by one or more of the reactions
shown.
Reaction A in Figure 2 represents the decomposition of the solid phase,
i.e., the net release of calcium and carbonate to the solution. This step
might be a combination of reactant adsorption (e.g., H or HO) on the CaCO,
surface, chemical reaction with the surface and desorption of reaction products.
The rate of decomposition of the surface (and the rate of change in
the bulk solution chemistry) may be controlled by the transport of hydrogen
t t
ions to the surface (reaction C) or the transport of reaction products (Ca ,
CO,, , HCO_ , and H C0_) away from the surface (reaction B). If a gas phase
is present, as shown in Figure 2, the bulk solution chemistry may be affected
by the transport of reaction products or gas phase components to or from
the bulk solution (reactions D and E). It is also possible, as indicated
by reaction E, that a homogeneous solution phase reaction such as the proton-
ation of the bicarbonate ion or the dehydration of carbon dioxide may effect
the time varying chemical characteristics of the bulk solution and the solution
within the boundary layers.
12
-------
DIFFUSION BOUNDARY
LAYERS
HtCO, HlCO.
completely
mixed bulk
solution
Gas _J
Phase I
Figure 2. Schematic representation of the calcium carbonate dissolution
process.
-------
A significant amount of research has been conducted on mineral dissolution
kinetics. The dissolution of calcite and limestone has been investigated
for applications such as the formation of antacids (Lund et al., 1975), the
neutralization of pickling acids (Eden and Truesdale, 1950; Gehm, 1944; Hoak
et al., 1944, 1945, 1947; Reidl, 1947; Galloway and Colville, 1970), the
neutralization of acid mine drainage (Pearson and McDonnell, 1975a, 1975b;
Jarret, 1966, Mihok et al., 1968; Vatanatham, 1975), the effect of CaCCL
sediments on the pH of sea water (Morse, 1978; Morse, 1974; Morse and Berner,
1972; Berner and Morse, 1974), the neutralization of CO -saturated waters
(Frear and Johnson, 1929; Erga and Terjesen, 1956; Terjesen et al., 1961;
Plummer et al., 1978; Plummer and Wigley, 1976), the neutralization of dilute
acidic ground and surface waters (Bjerle and Rochelle, 1982, Vaillancourt,
1981; Sverdrup and Bjerle, 1982; Driscoll et al., 1982; Haddad, 1983), the
neutralization of nitric acid solutions (Wentzler, 1971), sulfuric acid solu-
tions (Vatanatham, 1975) and hydrochoric acid solutions (Lund et al., 1975;
Tominaga, 1939).
It has been recognized for a long time that mass transport to or from
the dissolving CaC03 surface has at least some effect on the kinetics of
the process and therefore most recent,investigators have been careful to
control (to some extent) the hydrodynamic conditions in their experimental
reactor. A number of experimenters controlled the mixing intensity in mechan-
ically agitated batch reactors containing suspensions of powdered calcite
(Erga and Terjesen, 1956; Terjesen et al., 1961; Berner and Morse, 1974;
Sjoberg, 1976; Sverdrup and Bjerle, 1982; Rickard and Sjoberg, 1983). Others
have mounted rotating cylinders (King and Liu, 1933) or rotating disks (Wentzler,
1972; Lund et al., 1975; Rickard and Sjoberg, 1983; Sjoberg and Rickard,
1983) made of CaCO_ in batch reactors. A few investigators have studied
the dissolution reaction using flow-through packed-bed reactors (Pearson
and KcDonnell, 1975a, 1975b; Vaillancourt, 1981; Haddad, 1983). In one case
(Weyl, 1958) a fluid jet was directed against a calcite crystal.
In the cases where a batch reactor is used the rate of CaCO, dissolution
is usually monitored by either a "pH stat" or "free drift" technique. The
pH stat technique involves maintaining the bulk solution at a set-point pH
by the controlled addition of mineral acid. The rate of CaCO, dissolution
14
-------
is then related to the rate of hydrogen ion addition. The free drift technique
involves measuring the pH and/or calcium ion concentration as a function
of time as the suspended particles, rotating disk, etc. dissolve in the batch
reactor.
Most batch reactor studies have been conducted using an "open" system,
where the solution is in contact with a gas phase with a carbon dioxide partial
pressure ranging from 0 to 1007o. The packed column is usually operated as
a closed system; inorganic carbon does not enter or leave the solution during
the course of the dissolution reaction. The closed system is less complicated
than the open system to model because, as noted in regard to Figure 2, the
open system model may require an understanding of the rates of transport
and reaction at the gas-solution interface (reactions D and E in Figure 2).
There is no gas/liquid interface in an ideal closed system.
A review of the literature suggests that many investigators have recog-
nized the complexity of the CaCO- dissolution process. Most have attempted
to simplify the modeling of this process by delineating the rate limiting
steps. It is, however, apparent that in making assumptions and interpre-
tations of experimental data the various investigators have often been limited
by the type of apparatus used and the experimental conditions. Consequently,
it is difficult to generalize results.
Various processes have been proposed to regulate the dissolution of CaCC>3:
The diffusion of hydrogen ion to the solid surface (King and Liu,
1933, Tominaga et al., 1939, Kaye, 1957, Gortikova and Panteeva,
1937, Neirode and Williams, 1971, Berner and Morse, 1974, Wentzler,
1972, Vaillancourt, 1981, Haddad, 1983).
A heterogeneous "dissolution" reaction at the solid surface (Erga
and Terjesen, 1956, Terjesen et al., 1961, Plummer and Wigley, 1976,
Plummer et al., 1978, Berner and Morse, 1974, Sjoberg, 1976).
Mixed kinetics in which transport and a heterogeneous reaction at
the surface acting in series are important (Pearson and Mcdonnell,
1975, Rickard and Sjoberg, 1983, Lund et al., 1975, Berner and Morse,
1974, Plummer et al., 1978, 1976).
t 1
The diffusion of reaction by-products, e.g. Ca , away from the solid
surface (Weyl, 1958, Bjerle and Rocheele, 1982, Berner and Morse,
1974, Haddad, 1983).
15
-------
The dissolution and/or exsolution of carbon dioxide in or from the
solution (Volpicelli et al., 1981).
Recent papers by Sjoberg and Rickard (Sjoberg and Rickard, 1983; Sjoberg
and Rickard, 1984a; Sjoberg and Rickard, 1984b, Rickard and Sjoberg, 1983)
provide a detailed analysis of the dissolution process. Sjoberg and Rickard
used a rotating-disk/batch reactor apparatus and determined the initial rate
of calcite dissolution using the pH-stat technique.
Rickard and Sjoberg (1983) concluded that in neutral to alkaline solutions
at ambient temperature the dissolution of calcite was controlled by a mass
transfer resistance and a surface reaction acting in series. In this scheme
the observed rate of dissolution is a function of a transport rate, RL, where
"L = *L (cs - V (2)
and a first order surface reaction rate, Rc, where,
R = K (C - C ) (3)
c c eq s
K, and K are the mass transfer and surface reaction rate constants and C ,
L c s
C, and C are the molar calcium concentrations at the calcite surface, in
b eq
the bulk solution and at equilibrium, respectively. The equations for R^
and R can be combined by assuming a steady state condition near the inter-
face. The result is an expression for the overall rate of dissolution, R,
i.e.,
R = K (C - C, ) (4)
o eq b
where the overall rate constant, Ko, is given by
K IL
According to Eq. (5), when K » 1C the dissolution rate is controlled by
mass transfer and when K^ » K the surface reaction controls.
16
-------
Rickard and Sjoberg (1983) concluded that at low pH the initial rate
of calcite dissolution was controlled entirely by the rate of mass transfer
of the hydrogen ion to the calcite surface. They determined that for pH < 4,
R = K^ [H+]°'9, (6)
+ i
where [H ] is the bulk solution hydrogen ion concentration and 1C. is an
"apparent" mass transfer coefficient for the hydrogen ion. It is not clear
exactly why Rickard and Sjoberg found it necessary to change from calcium
ion transport control at neutral pH values and above to hydrogen ion transport
at low pH. In any case, their assumptions and rate expression for low pH,
Eq. (6), are generally consistent with low pH rate equations presented by
a number of other investigators (Miadokova and Bednarova ; Lund et al. , 1975;
Berner and Morse, 1974; Plummer et al., 1975a; Nierode and Williams, 1971).
A plot from Sjoberg and Rickard (1984a) of the initial rate of calcite
dissolution as a function of the bulk solution pH for a batch reactor/rotating
disk/pH-stat system operating at 25°C and a disk rotational speed of 1000 rpm
is presented in Figure 3. Note, the initial rate of dissolution was highest
at low bulk solution pH. The rate decreased with increasing pH and approached
a minimum (asymptotically) at pH > 5. Under the conditions used to obtain
the data of Figure 3 the minimum initial rate of dissolution was approximately
6.3 x 10 moles calcium cm s . Sjoberg and Richard (1984a)concluded
that the magnitude of this minimum rate was determined by both the mass transfer
and surface reaction rate constants (Eq. 5).
Plummer et al. (1978) obtained the initial calcite dissolution rate
as a function of the bulk solution pH. The results are presented in Figure
4. The pH-stat technique was used in conjunction with a mechanically agitated
batch reactor containing crushed calcite (Iceland Spar). The CO partial
pressure was a controlled parameter and the temperature was 25°C.
The results plotted in Figure 4 are similar to those obtained by Sjoberg
and Rickard, (1984a). As the bulk solution pH increased above pH = 4 the
initial rate of dissolution asymptomatically approached a minimum value.
The minimum rate (for pCO = 0.00) was approximately 3 x 10 moles cm s
17
-------
-7
V)
2
g
15
o
-8
I -9
•o
"o
'c
0»
o
-10
5
PH
8
Figure 3. Initial race of calcite dissolution as a function of the bulk
solution pH from Sjoberg and Rickard (1984a) . System was closed
to atmospheric CO.,.
18
-------
rvj
i
§
V)
-------
Sjoberg and Rickard (1984b) plotted the initial rate of dissolution
as a function of the square root of the disk rotational velocity, u2, to
illustrate the effect of the mass transfer coefficient, 1C, on the dissolution
rate. An example for carrara marble, a bulk solution pH of 8.4, a background
electrolyte of 0.7 M KC1 and temperatures of 1 and 25°C is illustrated in
Figure 5. In the rotating disk system 1C was directly proportional to u2
and, therefore, if mass transfer was the controlling step, the dissolution
rate, R, would be directly proportional to u) 2. At a bulk solution pH of
i" i-
8.4 the relationship between R and u) 2 is linear only at low values of GO 2.
i,
As u) 2 increased, the surface reaction apparently became an increasingly impor-
tant factor in determining the initial rate of dissolution.
For the results obtained at 25°C (Figure 5)) the relationship between
i, 3^-1
R and w2 was linear and mass transfer apparently controlled up to w2 = 5 s
where R was approximately equal to 7 x 10 moles cm s . At 1°C the effect
of the surface reaction on R seemed to be even greater than at 25°C. The
upper limit for mass transfer control was R = 1 x 10 moles cm s at
It will be indicated in a subsequent section that in packed bed limestone
contactors operated under the conditions used in this study, the maximum
rate of dissolution for bulk solution pH values greater than 4 is generally
less than 1 x 10 moles cm s . It therefore seems reasonable to assume
that, based on Sjoberg and Rickard's results (Figure 3)), the rate of transport
of calcium ion away from the interface controlled the dissolution rate through-
out the entire depth of the packed columns used in this study.
It has been shown that the presence of certain substances can reduce
the rate of calcite dissolution. This effect has been noted for ferric and
chromic ions, (King and Liu, 1933), copper (Erga and Terjesen, 1956), aluminum
(Volpicelli et al., 1981), scandium (Nestaas and Terjesen, 1969), organic
matter, magnesium and orthophosphate (Morse 1974a, 1974b; Berner and Morse,
1974). The effect of contaminants on the rate of dissolution can be significant
at very low contaminant concentrations. Nestaas and Terjesen (1969) concluded
that metal ions adsorb at active spots or kinks on the surface of the dissolving
crystal, blocking the dissolution process at that location. At the present
time there are no methods available for quantifying the effect of contaminants
20
-------
20
g
O
X
«> 15
(Nl I w
'E
o
en
ju
O
O
or
c
O
o
a
[o
"E
0
10
cul/2(s-|/2)
20
Figure 5. Initial rate of calcice dissolution as a function of the square
root of the rotating disk rotational speed. Bulk solution pH was
constant at 8.4 and the system was closed to atmospheric CCL
(Sjoberg and Rickard, 1984b). " 2
21
-------
on the dissolution rate, particularly the dissolution rate of limestone.
Eventually, for example, relationships between the contaminant concentration
and K , the surface reaction constant, may be developed.
PACKED-BED REACTORS
Only a few studies have involved attempts to model the kinetics of lime-
stone dissolution in continuous flow, packed-bed reactors (Pearson and McDon-
nell,1975a, 1975b; Barton and Vatanatham, 1976; and Vaillancourt, 1981).
Pearson and McDonnell (1975a, 1975b) studied the neutralization of acidic
drainage from coal mines using packed columns and in-stream barriers of large
(6.4 to 10 cm. effective diameter) limestone particles. Their open-to-the-
atmosphere experiments were conducted at ambient temperature and in the pre-
sence and absence of dissolved metal ions.
Pearson and McDonnell (1975a) indicated that a rate equation based on
hydrogen ion transport coupled with a surface reaction can be used to describe
limestone dissolution kinetics. The proposed model is given by:
dC
T TT2 = K C n = K C.n = K, (C - C.) (7)
Adt o ai doi
where V/A is the inverse of the interfacial area per unit volume of fluid
in the column, K is an overall rate constant, K is a surface reaction rate
cl
constant, K is the mass transfer coefficient, C is the hydrogen ion con-
centration in the bulk solution, C. is the hydrogen ion concentration at
the limestone/water interface and n1 and n are exponents.
Pearson and McDonnell (1975a, 1975b) did not use their experimental
data to test the proposed rate equation, Eq. 7, but instead developed an
empirical expression which related the rate of limestone dissolution to water
temperature, pH, solution ionic strength, the bicarbonate ion concentration
and the hydraulic shear stress. Their overall model included an expression
for predicting the rate of CO exolution at the air/water interface above
the packed bed. The experimental conditions used by Pearson and McDonnell
(1975a) to develop their empirical equations are appropriate for the treatment
22
-------
of acidic drainage from coal mines but not the dilute acidic surface waters
used as potable supplies.
Two groups of investigators, Barton and Vatanatham (1976) and Vaillancourt
(1981) assumed that the rate of limestone dissolution in closed and open-to-
the atmosphere, packed-columns is controlled by the rate of hydrogen ion
transport from the bulk solution to the limestone surface. Vaillancourt
(1981) used the conventional relationship,
U dHK
-K a (< - H+ ) (8 )
E dx b eq
where U is the superficial velocity, e is the bed porosity, K is the mass
s
transfer coefficient for hydrogen ion, a is the surface area of limestone
per unit volume of interstitial water, x is distance in the axial direction,
H, is the hydrogen ion concentration in the bulk solution and H is the
hydrogen ion concentration when the solution and limestone solid phase are
at equilibrium (under a closed-to-the-atmosphere condition). Vaillancourt
(1981) correlated experimentally determined mass transfer coefficients with
the limestone particle diameter, superficial velocity and fluid properties
using dimensionless parameters.
Unfortunately, Vaillancourt (1981) used very short packed-columns with
high Reynolds numbers in his experiments and did not consider the adverse
effect these conditions had on his assumption of plug flow. He also did
not consider the effect of raw water chemistry on the magnitude of H
A constant value was incorrectly used for all solutions studied.
Vatanatham (1975) and Barton and Vatanatham (1976) studied the kinetics
of limestone dissolution in acidic solutions using an open-to-the-atmosphere
batch reactor and a recycle-downf low, packed-bed reactor system. Several
models were tested including, zero order reaction controls, carbon dioxide
transport controls, surface reaction controls and hydrogen ion transport
controls. Barton and Vatanatham (1976) concluded that in the pH range of
2 to 6 hydrogen ion transport controls. They did not determine the rate
limiting step outside this range but assumed that the lack of agreement
between the experimental data and the hydrogen ion transport model was due
to experimental error or the increasing importance of other transport or
rate limiting mechanisms.
23
-------
The kinetic equation used by Barton and Vatanatham (1976) for pH values
between 2 and 6 is given by,
-H], (9)
dt 6 p D0 T) eq
where W and WQ are the mols of CaC03 present at any time, t, and at t = 0,
M is the molecular weight of CaC03 , Do is the initial diameter of the lime-
stone particles, p is the mass density of limestone, H is the hydrogen
ion concentration in the bulk solution at equilibrium and H, is the hydro-
gen ion concentration in the bulk solution at any time. Eq. 9 is essentially
a first order (film) transport equation modified to include the change in
interfacial area as the particles dissolve and decrease in size. Unfortunately
Barton and Vatanatham (1976) made an error in deriving Eq. 9. The number
six should appear in the numerator and not in the denominator and therefore
all their model calculated results were in error by a factor of 36.
METAL RELEASE FROM PIPES
There is considerable concern over the corrosion of water distribution
systems. Elevated corrosion rates may substantially reduce the service period
of piping systems resulting in increased operation and maintenance expenses
(Anderson and Berry, 1981). Metal release from water distribution systems
may also cause water supplies to exceed the U.S. Environmental Protection
Agency (U.S. EPA) Standards for maximum contaminant levels (MCL) or secondary
maximum contaminant levels (SMCL). Maximum contaminant levels (MCL) are
established for concentrations of compounds that may result in human health
problems, while SMCL are primarily established for esthetic criteria.
Metal release may occur from copper, galvanized steel, iron and lead
pipes, and from lead-tin solder coated on copper piping materials. Human
health concerns are largely associated with the leaching of lead from lead
pipe or lead-tin solder coated on copper pipe. The toxic effects of lead
are well established (NAS 1977; Waldbott, 1978). Lead is an active and
cumulative toxicant which alters neurological and metabolic functions. It
has been associated with hyperactivity , hypertension, mental retardation
and motor disfunctions (NAS 1977; Patterson and O'Brien, 1979). Several studies
24
-------
have established a link between high concentrations of lead in drinking water,
and elevated concentrations of lead in blood and subsequent health problems
(Beeners et al, 1976; Campbell et al., 1977; Cameron and Wunderlich, 1976).
Because of human health concerns, the U.S. EPA established a MCL for lead
of 0.05 mg Pb- IT1.
Although copper is an essential trace metal, at elevated concentrations
it has been implicated as a gastrointestinal poison (Doull et al., 1980).
The U.S. EPA Secondary MCL for copper is 1.0 mg Cu-L~l. This standard has
largely been established for esthetic considerations, such as the taste and
staining characteristics associated with elevated concentrations.
Elevated corrosion rates have been reported for a number of regions
(Hudson and Gilcreas, 1976; Dansel, 1976; Patterson and O'Brien, 1979; Kara-
lekas, et al., 1983; Maessen et al., 1985). Of particular concern are soft-
water supplies such as in the northeastern, southeastern and northwestern
United States (Patterson and O'Brien, 1979).
Corrosion is a deterioration of a metal which usually occurs as a result
of an electrochemical reaction. For corrosion to occur, an electrochemical
cell must be established including an anode, a cathode, an electrolyte solu-
tion, and an electrical (metal) connection between the anode and cathode.
As an electrochemical reaction proceeds oxidation occurs at the anode releasing
electrons which are transmitted through the electrical connection to the
cathode. These electrons are accepted at the cathode through a reduction
reaction. The tendency for a metal to oxidize (and subsequently exhibit
corrosion) is measured through its oxidation potential (E°). Some values
of oxidation potential for some relevant reactions are summarized in Table
4. Note the reaction with the highest oxidation potential will have the
greatest tendency to undergo oxidation in an electrochemical reaction. For
example, if copper and lead form an electrochemical cell at a copper-lead
solder joint, lead would be oxidized (corroded) while copper would be reduced
by virtue of their values of oxidation potential.
There are two conditions by which corrosion may be restricted. First,
the electrochemical (redox) potential and pH may not thermodynamically favor
oxidation. This condition is termed immunity. The second condition involves
the formation of a sparingly soluble solid phase with an oxidation by-product,
25
-------
TABLE 4 Oxidation Potential of Metallic Materials
Anode Anodic Reactions Potential
E° (volts)
Zinc Zn(s) --> Zn2+ + 2e~ 0.76
Iron
Soft Solder Fe(s) --> Fe2+ + 2e~ 0.44
Tin Sn(s) --> Sn2 + 2e~ 0.136
Lead Pb(s) --> Pb2+ + 2e~ 0.126
Copper Cu(s) --> Cu2+ + 2e" -0.345
26
-------
such as an oxide, hydroxide or salt. If this solid adheres as a film on
the metal surface, then it may mitigate corrosion. This process is referred
to as passivation. The effectiveness of passivation films is highly variable,
and depends on the affinity of the solid phase for the metal and whether
coverage is complete or partial.
Hilburn (1983) developed two conceptual models for uniform corrosion.
The direct-dissolution model is applicable when the metal is oxidized and
directly released to solution. Under these conditions the corrosion rate
is controlled by either the kinetics of the reaction, or the transport of
reactants and products to and from the metal surface through solution. The
dissolution-and-film-growth-model applies to metals which form a passivation
film. The overall corrosion rate may be regulated by reaction kinetics,
transport through the passivation film or solution transport, whichever is
the rate-limiting process.
Corrosion is an extremely complicated process. For example factors
such as pipe age, pipe length, impurities in the pipe material, interval
of solder joints, temperature, turbulence and water chemistry can all contri-
bute to corrosion (Herrera et al., 1982; Hilburn, 1983, 1983; Schock, 1984;
Maesson et al. , 1985; Treweek et al., 1985). As, a result, it is often diffi-
cult to evaluate factors regulating metal release from piping systems. Maessen
et al. (1984) studied metal mobilization in home well-water systems in Nova
Scotia. They assessed bedrock type (e.g. granite, limestone), proximity
to the coast, well-type (e.g. dug, drilled) and depth, plumbing data (e.g.
length of piping, age of piping, type of piping), as well as solution chemistry
on the extent of metal release from water distribution systems. They found
significant leaching of copper, lead and zinc occurred in some systems.
Concentrations of metals were elevated in water that had been in contact
with piping material for a prolonged periods of time (e.g. overnight, standing)
relative to running water samples. Although a wide range of bedrock, water
chemistry and plumbing conditions were evaluated, no factor could be found
to systematically predict the extent of metal leaching. Moreover, indexes
commonly used to assess the corrosive tendency of a water (Langelier, Ryznar,
f\
Aggressiveness indexes, and the ratio of SO, and Cl~l to alkalinity) and
pH were poor predictors of metal release.
27
-------
Meranger et al. (1983) evaluated metal leaching from cottage piping
systems that contacted acidic lakewater in northern Ontario. They found ele-
vated leaching of cadmium, copper, lead and zinc. Mobilization rates were
greatest during the first two hours of contact time with the pipe, but concen-
trations continued to increase for a period of up to 10 days. Highest metal
concentrations were again obtained with the first sample collected and concen-
trations decreased by up to 977» in the third liter of water collected. Al-
though the authors were concerned that acidic deposition to the region resulted
in surface water acidification and enhanced the corrosivity of lake water,
because these waters are naturally soft and corrosive this effect is not
clear.
Although it is often difficult to interpret field data because of all
the physical and chemical factors which contribute to corrosion, considerable
progress has been made in recent years through controlled laboratory experi-
ments in evaluating the chemistry of passivation films and processes regulating
the formation of films. Housing and building systems frequently have sections
of pipe that remain stagnant for prolonged periods of time. Initially metal
release is regulated by mass-transport reactions, however over time concentra-
tions can approach and reach saturation with respect to mineral phase solubi-
lity (Schock, 1984). Therefore, solubility calculations may be used as worst-
case assessment of metal leaching.
In recent years thermodynamic calculations have been used as a tool
to assess trace metal chemistry and the stability of passivation films within
water distribution systems. Several types of passivation films may form
on metal pipe depending on the chemical characteristics of the water supply
(Table 5). Patterson and O'Brien (1979) discussed the role of inorganic
carbon in regulating the release of lead from lead pipe. Using thermodynamic
calculations, they found that the solubility of lead decreases with increasing
inorganic carbon concentrations. Moreover, they suggested that elevated
inorganic carbon concentrations result in the formation of an insoluble lead
carbonate passivation film. This film not only reduces lead solubility but
also strongly adheres to the pipe surface, limiting the release of particulate
lead to water. Their calculations suggest that reduced inorganic carbon
28
-------
TABLE 5
Passivation film minerals that may be important in
regulating metal solubility in water distribution systems
COMPOUND
Lead Pipe
Lead hydroxide
Lead carbonate (cerussite)
Basic lead carbonate (hydrocerussite)
Lead Sulfate
STOICHIOMETRY
Pb(OH)2
Pb C03
Pb S04
Copper Pipe
Copper hydroxide
Copper carbonate
Basic copper carbonate (malachite)
Basic copper carbonate (azurite)
Copper sulfate
Basic copper sulfate(brochantite)
Basic copper chloride(atacamite)
Cu(OH)2
Cu2(OH)2C03
Cu3(OH)2(C03)2
Cu2(OH)2Cl
Galvanized Steel Pipe
Zinc hydroxide
Zinc carbonate
Basic zinc carbonate(hydrozincite)
Zinc sulfate
Basic zinc silicate(hemimorphite)
Zn(OH)2
ZnC03
Zn5(OH)6(C03)2
Si2
• H20
29
-------
concentrations facilitate the formation of a lead hydroxide film which does
not adhere strongly to the pipe surface and periodically is released to the
water supply as particulate lead.
In a series of papers, Schock (1980), Schock and Gardels (1983), and
Schock (1984) greatly elaborate on the role of inorganic carbon in controlling
trace metal concentrations in water distribution systems. Schock (1980)
suggested that the thermodynamic analysis by Patterson and O'Brien (1979)
was incorrect due to a failure to consider soluble lead-carbonate complexes.
Lead forms strong aqueous complexes with carbonate and therefore elevated
dissolved inorganic carbon concentrations can significantly enhance aqueous
lead concentrations. As a result, the contention by Patterson and O'Brien
(1979) that increases in dissolved inorganic carbon concentration reduce
aqueous lead concentrations is incorrect and may suggest counter-productive
water treatment strategies.
Due to the relatively high solubility of lead at low pH and the potential
to form lead-carbonate complexes at higher pH values the conditions under
which the theoretical solubility of lead is below the U.S. EPA MCL of 0.05
mg Pb-L"^- are limited to relatively high pH values (8.0 - 10) and low dissolved
inorganic carbon concentrations. Schock (1984) indicated that under these
conditions the concentrations of lead would be generally regulated by the
solubility of hydrocerussite (Pb3(C03)2(OH2), a tightly adhering passivation
film. These conditions would limit the release of particulate lead to water
supplies.
In addition to lead solubility, Schock (1984) evaluated the theoretical
solubility of passivation films from copper and galvanized steel pipe. Because
both zinc and copper are hydrolyzing metals and form soluble complexes with
carbonate, it is reasonable to expect their solubility to mimic lead. Copper
exhibits a considerable variation in solubility over a range of pH and dis-
solved inorganic carbon concentrations. Generally the solubility of copper
in the pH range 7 to 11 is well below the U.S. EPA secondary MCL of 1 mg
Cu-L~l. It is, therefore not as difficult to meet the U.S. EPA secondary
MCL for copper as it is to meet the MCL for lead. Like lead, the minimum
theoretical solubility occurs at elevated pH values (9-10) and the solubility
is enhanced at high pH values due to the formation of soluble carbonate com-
30
-------
plexes. Schock (1984) indicates that in the pH range of 9-10 the theoretical
solubility of copper is regulated by tenorite (CuO).
The theoretical solubility of zinc has a minimum value near pH 9 and
in this pH range is thought to be regulated by the solubility of hydrozincite
(Zn5(C03)2(OH)g) (Schock 1984). Hydrozincite is not a strongly adhering
passivation film. Therefore, zinc concentrations in water supplies using
galvanized steel pipe may be significantly enhanced by the release of particu-
late Zn. Unlike lead and to a lesser extent copper, zinc does not form strong
soluble complexes with carbonate. Therefore, pH is the major factor regulating
variations in the solubility of dissolved zinc from galvanized steel pipe.
However within the pH range 7 to 11, the theoretical solubility of dissolved
zinc is well below the USEPA SMCL of 5 mg Zn-L"1 (Schock 1984).
While thermodynamic calculations represent an important tool to assess
trace metal solubility and the stability of passivation films, they clearly
have many limitations. As indicated previously, thermodynamic calculations
should be viewed as an upper-limit of dissolved metal concentrations. Under
many conditions, particularly when water has been in contact with piping
material for a short period of time, the release of corrosion by-products
will be controlled by mass-transport reactions. Physical factors such as
the poor adherence of passivation films to piping material and/or erosion
of these films due to turbulance can significantly increase metal concentra-
tions through the release of particulate metal. Moreover, our understanding
of the temperature dependence (standard enthalpy values) of metal complexation
and solubility reactions is limited. So it is difficult to make thermodynamic
solubility calculations at temperatures other than 25°C with confidence.
Finally, while thermodynamic calculations provide theoretical values of dis-
solved metal concentration which may be useful in evaluating compliance with
U.S. EPA drinking water standards, no information is obtained on the destruc-
v tion of the metal pipe. While an insoluble passivation film may restrict
the release of metal to solution, if it is not impervious to molecular oxygen
then oxidation may continue and substantially diminish the lifetime of the
metallic piping material.
31
-------
SECTION 5
METHODS AND MATERIALS
APPARATUSES - LABORATORY AND FIELD CONTACTOR UNITS
Laboratory Contactors
Four downflow, packed-column contactors were used in the laboratory
study. Each column contained a different limestone particle size. The column
diameters were chosen to yield a column-to-particle diameter ratio of at
least ten to minimize the effect of the higher porosity at the wall on the
flow through the bed . The four columns and the water feed system are illus-
trated in Figure 6.
Column A in Figure 6 was constructed of clear acrylic plastic and con-
tained limestone particles with a 0.96 cm mean diameter. The column inside
diameter was 15.2 cm and the length was 3.5m. Columns B, C and D were con-
structed of polyvinyl chloride pipe. Columns B and C both had inside diameters
of 15.2 cm and Column D had an inside diameter of 38.1 cm. The stone sizes
(mean diameter) in these columns were 0.54 cm, 1.5 cm, and 3.2 cm for Columns
B, C, and D respectively. All three columns were 2.1 m. long..
All four columns in Figure 6 were equipped with through-the-wall sampling
tubes. The tubes were spaced in the axial direction at 15.2 cm intervals
at the influent end and at 30.4 cm intervals over the remaining portion of
each column.
Each sampling tube (0.6 cm diameter acrylic plastic) extended to the
center of the column. Five 0.25 cm diameter holes were drilled in the upper
part of each tube. Each tube was cemented to a plastic adaptor which was
threaded into the column wall. A short length of plastic tubing with a hose
clamp was attached to the plastic adaptor. A drawing of a typical sampling
tube is included within Figure 6.
The water supply and flow control system used with the four laboratory
columns is shown in Figure 6. The raw water was pumped from a 200 L plastic
tank to a constant head tank located above Column A. Overflow from the con-
stant head tank returned to the plastic tank. Flow control for each column
effluent was accomplished using a flowmeter with a micrometer controlled
32
-------
Figure 6. Laboratory columns with water supply and flow control system.
Insert is a drawing of a typical through-Che-wall sampling tube,
33
-------
valve assembly. From the flowmeter the water discharged to a small open
chamber and from this unit to a floor drain. The flowmeter calibration was
checked frequently using a volumetric cylinder and a stopwatch.
The limestone was washed with tap water and placed in each column layer
by-layer to facilitate installation of the sampling tubes and to minimize
later compaction of the bed. Gentle tapping and shaking of the column were
used to consolidate the bed as it was installed.
Field Contactors
Three devices were studied in the field investigation. These included
a large baffled-box device which was submerged in a mountainside spring at
the head end of a rural resort water supply system and two small column-type
units which were used for individual resort cabins. One of these small units
was obtained from Culligan, Inc.*
The baffled-box contactor is described in Figure 7. The unit was con-
structed several years ago at Syracuse University using 1.9 cm thick marine-
grade plywood covered with 2 mm thick plexiglass sheets. The overall dimen-
sions are 0.6 x 0.6 x 1.2 m. Sampling cells which also serve as baffles
to direct the flow along the bottom of the chambers were constructed of 0.6
and 1.3 cm plexiglass (each is 12.7 cm x 12.7 cm x 0.6 m). The sides and
lid were braced with fiberglass resin coated aluminum angles. Fiberglass
resin was also used to coat small areas of the contactor not covered by plexi-
glass sheets. The unit contained approximately 479 Kg. of 0.96 cm mean dia-
meter limestone particles and the length of the flow path through the limestone
was approximately 354 cm. The cross-sectional area perpendicular to the
direction of flow was approximately 915 cm^.
The two smaller column type units used in the field study are shown
in Figure 8. Column 1 had an inside diameter of 20.2 cm and an overall length
of 130 cm. Flow entered this column at the top, passed down through the
bed and exited through a cylindrical plastic strainer connected to a 2 cm,
inside diameter plastic pipe which passed up through the center of the bed.
Column 1 was constructed of wound fiberglass and contained 60 Kg of crushed
limestone (0.96 mean diameter particle size). The overall depth of limestone
was 122 cm. Column 1 was a slightly modified version of a container used
in ion exchange systems.
*Mention of trade names or commercial products does not constitute endorsement
or recommendation for use.
34
-------
Plexiglass
sampling Cell
Crushed Limestone
Exterior Q Interior Walls of the
Chamber (3/4 "marine plywood
coveted with 0.08'
plexiglass sheets on
both sides)
Lid of the
contactor
Inlet of the
contactor
LO
Ul
COVE WOOD LODGE
LIMESTONE CONTACTOR
Outlet of the.
contactor
Figure 7. Baffled-box contactor used in the field study.
submerged in a mountain-side spring.
Entire unit was
-------
outlet
inlet
4"'
48"
<
TD
l
Valves for
Backwashing medium
, Wound
Fiberglass
1 ^-Limestone
Medium
PVC pipe
8"
-i.5"0 Section
of PVC Pipe wifh
1/4" 0 holes
2U!L
32"
15"
,
£
.'/
J
.
j
r,«
i 1 1^ —
I II in (of
i
ILL
V//'
• i .
^ \
' _ '
!
-------
Column 2 illustrated in Figure 8, was rented from Culligan, Inc. It
had an inside diameter of 23 cm and a total length of 127 cm. It was con-
structed of galvanized steel coated with a heat treated epoxy resin.
A granular, calcium carbonate medium (Cullneu , neutralizing medium,
catalog number 1600-10) sold by Culligan, Inc.* was used in place of limestone
in Column 2. The column was filled to a depth of 40 cm with Cullneu". The
flow conditions within Column 2 were very similar to those in Column 1, however,
Column 2 was equipped with a valve arrangement at the top which allowed one
to direct water into the effluent pipe to backwash the medium by upflow through
the bed.
The baffled-box contactor and the two column units were installed at
the Covewood Lodge, a resort with housekeeping cottages and a rustic lodge
located in the Adirondack Region of New York State near Old Forge. A map
illustrating the layout of the gravity-fed supply system is presented in
Figure 9. The baffled-box contactor was installed, completely submerged,
.in the spring which serves the seven cottages on the west side of the resort.
The spring water elevation was approximately fifteen meters above the ground
floors of the cottages. Water flowed for a distance of approximately 20
ft. (6 m) into two, 400 gallon, (1600 L) galvanized steel storage tanks.
Flow to the cottages from the storage tanks was through a 3.8 cm diameter
plastic pipe. The plumbing in each cottage was copper pipe soldered with
50/50 lead-tin solder. The installation of the baffled-box contactor installed
within the spring is illustrated in Figure 10. Bay Side and Hillside cottages
contained approximately 30 m (100 ft) and 15 m (50 ft) of 1.3 cm (^ in.)
diameter copper pipe, respectively with approximately forty 50/50 lead-tin
solder joints per cottage or two joints per meter of copper pipe.
The wound fiberglass column with limestone particles (Column 1, Figure
8) was installed in the heated basement of Bay Side cottage (see Figure 9).
The unit was used during the months of January, February, March and April
1984 when the plastic line from the spring and baffled-box contactor became
frozen and it was necessary for the resort owner to supply water to the winter-
ized cottages by pumping water directly from Big Moose Lake. The contactor
in Bay Side Cottage was installed on the pressure side of a pressure switch
activated supply pump. The cottage contains two small living units, each
*Mention of trade names or commercial products does not constitute endorsement
or recommendation for use.
37
-------
BIG MOOSE LAKE
00
Figure 9. Map of the Covewood Lodge property located near Old Forge, NY.
Site of the field study.
-------
PLAN
7' 6'
li'PV.C.
Inlet
. l^-to 2"
Adapter
S%. Galvanized
CLPi*
Limestone
Contactor.
Chamber
|/'r IQNUI f—
f 'fh '
1 ] '
li 1
1 • •
1 1
I 1
1 r
i '
i i
j. i
.|4_.L
tp 4,
•• *
5'
Section A-A
«i'
M
S
4
I
,''*
5 i
In
»i
ui
i
2 Galvanized
,pipe
Limestone
Contacttf
Chamber
t
L
3L-
E
I
f
Y<
f
Figure 10.
Diagram showing the installation of the baffled-box contactor in
the spring at Covewood.
39
-------
has a kitchen and a bathroom with a toilet, sink and shower. Normal total
occupancy during the winter (most but not all guests limit their stay to
a weekend) is four adults.
The Culligan column with Cullneu medium was installed in the basement
of a cottage (Henry Covey, Figure 9) on the east side of the resort. The
east side of the resort receives water from a spring, which at the time of
this study, contained a marginally effective limestone contactor installed
by the resort owner. The unit in this spring was somewhat ineffective because
of significant short-circuiting. The Henry Covey cottage is winterized and
has a kitchen and a bathroom with sink, toilet and shower. Normal occupancy
is two persons.
Estimates of limestone contactor cost are given in Appendix C.
Limestone Characteristics
The limestone used throughout the study was obtained from a quarry in
Boonville, New York. The limestone was analyzed in the laboratory to determine
its physical and chemical characteristics.
Chemical Characteristics - A sample of limestone was ground to a powder (parti-
cle diameter less than 0.29 mm) and then washed with tap water and dried
24 hours at 105°C.
Three 0.2 gram portions of the powdered limestone were dissolved in
50 ml 1:1 HCL/HN03- After dilution with deionized water elemental analysis
was conducted by atomic absorption spectrophotometry.
It was determined that the cation content of the Boonville limestone
is (by mass) 85 percent calcium, 12.3 percent aluminum and 2.4 percent magne-
sium. Iron, Mn, Zn, Cu and Cd were present at less than 0.1 percent and
Pb, K and Na were not detected. These results indicate that the Boonville
limestone is essentially a "high calcium" limestone.
A supplemental experiment was conducted in which a measured quantity
of Boonville limestone was dissolved in concentrated hydrochloric acid in
a closed system. The CC>2 evolved was captured and its amount measured.
This result combined with the calcium measurement indicates that the Boonville
stone contains 79% CaCC>3 by mass. Therefore, although it can be labeled
a high calcium stone it is not of high purity.
40
-------
The effective CaCC>3 solubility product for the limestone was determined
by placing 0.10 gram samples of the powdered limestone in twelve open flasks
containing 100 ml of deionized water. Different amounts of acid were added
to each flask (between 0.25 to 1000 ueq/L using IN HC1) so that the initial
pH of the samples was between 3.00 and 6.60. The flasks were agitated on
a shaker table in the 20°C room for one week. At the end of the equilibration
period samples were filtered using 0.45 vim millipore membrane filter and
analyzed to determine Ca, Mg, DIG and pH.
The molar concentrations of these constituents for each sample were
input to the MINEQL chemical equilibrium computer program (Westall et al.,
1976) using the following conditions:
a - Fixed carbon dioxide partial pressure of 10~3-5 atm.
b - Fixed pH (measured final value for each sample)
c - Total hydrogen ion concentration equal to the molar concentration
of acid initially added to the sample.
A solubility product of CaC03 for the limestone of each sample was calcu-
lated as the product of the equilibrium activities of calcium and carbonate
computed by the computer program. The average effective solubility product
of CaCC>3 in Boonville limestone was found to be 10"8-71 (20°C). The experi-
mental results and the computed values of the effective solubility product
are listed in Table 6.
Physical Characteristics - The four size fractions of limestone particles
obtained from the Boonville quarry were analyzed to determine particle size,
sphericity and mass density.
The median particle size for each size fraction was determined using
a standard ASTM (ASTM Manual 447-4) sieve analysis. The percent by weight
finer than a given sieve opening was plotted as a function of the size of
the sieve opening on arithmetic probability graph paper. The median particle
size was determined by interpolation from this graph. In the case of the
0.96 cm median size fraction, 90 percent of the particles were between 0.7
and 1.3 cm.
The volume-weighted mean particle diameter was determined by measuring
the volume of at least 1200 particles in each size fraction. Particle volume
41
-------
Table 6 Effective Solubility of Crushed Limestone
Experimental Results
Sample
Number
1
2
3
4
5
6
7
8
9
10
11
12
Acidity
Added
eq/1 HC1
0.25 -
20
60
100
140
200
260
300
340
400
500
1000
Initial
PH
pHo
6.6
4.7
4.22
4.0
3.85
3.7
3.59
3.52
3.4
3.4
3.3
3.0
Final
PH
pHf
7.54
7.85
7.88
7.92
7.89
7.86
7.92
7.94
8.09
8.11
8.11
8.02
Final
DIG
mgC/L
11.9
12.38
12.38
12.38
10.24
10.95
9.29
9.52
10.71
11.43
10.48
9.29
Final
Calcium
Cone.
mgCa/L
23.08
23.87
25.61
24.41
21.88
24.57
23.7
24.24
28.3
29.95
31.81
44.4
Final
Mag.
Cone .
mgMg/L
0.18
0.18
0.2
0.18
0.18
0.18
0.19
0.19
0.21
0.21
0.23
0.33
Computed
pKsp
9.663
8.998
8.94
8.881
8.988
8.844
8.894
9.029
8.48
8.416
8.389
8.422
42
-------
was measured by drying a random sample of particles at 105°C for 24 hours.
Each particle in the sample was weighed and numbered and then carefully sus-
pended in a small volumetric cylinder filled with water. The volume displaced
was accurately measured with a 1 ml pipet. The volume-weighted mean diameter,
dp, for each fraction was calculated using
6
_£
n ir
(10)
where Vp is the total measured volume and n is the number of particles included
in the measurement. In the case of the size fraction with a 1.01 cm median
diameter (sieve analysis) the volume weighted mean diameter was 0.93 cm.
The results of the particle size measurements for the four fractions are
given in Table 7. The diameter used for a given fraction in model calculations
was the average of the value obtained by the sieve analysis and the value
obtained by fluid displacement. The sieve analysis results were approximately
normally distributed and therefore it is reaonable to assume that the median
size from the sieving/weighing procedure and the mean size from the fluid
displacment measurements should be nearly the same since the particles all
have the same density.
The sphericity of a particle is equal to the surface area of a sphere
with the same volume as the particle divided by the measured surface area
of the particle. The sphericity of each particle, ^, was determined by
(6 Vj/TT )2/3 ( TT/4)
*i Ai
where V^ is the volume of the particle measured by fluid displacment and
A^ is the actual surface area measured planimetrically. The sphericity listed
in Table 7 for each size fraction is the average value for the particles
in the sample. The sample size for each size fraction was approximately
fifty particles.
The average sphericity ranged from 0.83 for the 3.20 cm size fraction
to 0.78 for the 1.50 cm fraction. In the case of the 0.96 cm fraction the
measured sphericities ranged from 0.50 to 0.98, with an average value of
0.79.
43
-------
TABLE 7 Limestone Particle Size and Sphericity Analysis Results
Mean Diameter Volume Weighted Diameter Used Particle
Size
Fraction
I
II
III
IV
Sieve Analysis
(cm)
3.20
1.45
1.01
0.55
. Mean Diameter,
dp (cm)
1.55
0.93
0.54
in Design
Calculations, d(cm)
3.20
1.50
0.97
0.54
Sphericity
(dimensionless)
0.83
0.78
0.79
0.81
-------
The mean density of the particles was determined by dividing the sum
of the particle weights by the sum of their measured volumes. The calculated
density was 2.64 g/cnH.
Cullneu is described by the manufacturer, Culligan, Inc., as "a specially
graded calcium carbonate compound for neutralizing acid waters which provides
consistent dissolving rate for treatment." The particle size is 6-30 mesh
or a mean effective diameter of approximately 2.2 mm. The bulk density is
approximately 1.5 g/cnH. No other information is available on the Cullneu
material.
Limestone Bed Characteristics
A number of tests were conducted to measure pertinent physical character-
istics of the packed-bed contactors used in the study. The bed porosity
was measured and used with the mean particle diameter and particle sphericity
to calculate the area of limestone particle surface per unit volume of inter-
stitial water. This quantity is important in modeling dissolution kinetics.
Tracer studies were conducted to measure fluid residence time and axial disper-
sion in the contactor.
The porosity of a,packed bed is the ratio of the void space and the
total enclosed volume of the bed. The porosity of each column was determined
by measuring the volume of fluid required to displace all the air from the
bed. This volume was divided by the total volume of the column to obtain
the porosity. The complete procedure was repeated five times.
To evaluate the effect of the column wall on the bed porosity a series
of special tests were conducted. Beakers of various sizes and hence various
wall plus bottom area to volume ratios were filled with each of four limestone
particle sizes and the porosity was measured. The measured porosities have
been plotted as a function of the vessel contact area to volume ratio (A/V
in cm"^-) in Figure 11.
The measured porosity for the column which contained the 0.96 cm limestone
particles was 0.41 and the vessel contact area to volume ratio was 2.25 cm~l.
From the least square regression line fitted to the 0.96 cm particle size
data points in Figure 11, the expected porosity for a vessel contact area
to volume ratio of 0.25 cm"-'- is 0.43 +_ 0.04. The measured porosity of 0.41
45
-------
0.60
•5 0-50
o
0.
0.40
0
OA
0.54cm
I
I
0.5 1.0 1.5
Container surface area to volume ratio (cm"1)
2.0
Figure 11. Measured porosity plotted as a function of container surface area
to volume ratio for four limestone particle effective diameters.
Lines were fitted to the data by the method of least squares.
-------
falls within this range. This result also suggests that under these conditions
the column wall has a negligible effect on the overall bed porosity.
The effect of the vessel contact area to volume ratio on the porosity
increases with increasing particle size (Figure 11). For example using the
four least squares regression lines, for A/V = 1 cm~l, the overall porosity
is 0.62 for 3.2 cm limestone, 0.55 for 1.5 cm limestone, 0.49 for 0.96 cm
limestone and 0.43 for 0.54 cm limestone. The porosities measured (or esti-
mated using Figure 11) for each of the columns used in this study, except
the Culligan unit are listed in Table 8.
The limestone particle surface area per unit volume of interstitial
water (a, cm"-'-), which was used in modeling the dissolution process, is also
listed in Table 8. This quantity was calculated for each column using the
measured or estimated porosity ( e ) and the measured mean particle size (d)
and sphericity ( t|; ) . The equation used is
(12)
The contactors described in Figure 6 were used in a set of experiments
designed to determine the effect of limestone particle size, flowrate and
the depth of the packed-bed on axial dispersion and mean fluid residence
time. Axial dispersion may be an important factor in modeling the effect
of limestone dissolution on effluent chemistry. Tracer studies were conducted
to evaluate axial dispersion and to test calculated values of mean fluid
detention time within the bed.
Lithium chloride was used as a tracer salt. Lithium is easily detected
(by atomic absorption spectrophotometry) , it does not react with nor is it
significantly adsorbed by the contents of the columns and the background
concentration of lithium was negligible in the tap water used in the tracer
experiments.
In most experiments a 200 mg quantity of LiCl dissolved in 10 mL of
deionized water (20g Li/L) was injected with a syringe into the feed port
at the top of the column. Samples from the effluent port were collected
every 15-30 seconds around the peak concentration of the tracer curves and
every minute for the remainder of the test. The tracer study was repeated
47
-------
TABLE 8 Bed porosity and Limestone Particle Surface Area
per unit volume of Interstitial Water
Limestone Particle
Diameters d
Figure 8
Baffled-Box
Figure 8
0.96
0.44*
Limestone Particle
Surface Area
Per Unit Volume of
Interstitial Water,
A,
B,
c,
D,
Column
Figure
Figure
Figure
Figure
(cm)
6
6
6
6
Wound Fiberglass
0.
0.
1.
3.
0.
96
54
50
20
96
Porosity
0
0
0
0
0
.41
.43
.49
.49
.44*
a
11
18
5
2
9
(cm'1)
.4
.2
.3
.6
.7
9.7
*Estimated Using Figure 11 and measured vessel contact area to volume ratios.
Wound Fiberglass Column, A/V = 0.21 cm"!;
Baffled-Box Contactor, A/V = 0.19 cm"1.
48
-------
three times for each flowrate. The results of four experiments are plotted
in Figure 12.
The results from each tracer test were analyzed to determine the total
mass of lithium injected passing the sampling port using the following equa-
tion:
n
[Mass of Lithium Recovered] = Q £ C± t^ (13)
1=1
The quantity £ C^ t^ is the area under the tracer response curve and Q is
i=l
the volumetric flowrate.
The mean fluid residence time, t, was determined using the first moment
of the tracer response curve, i.e.,
The axial dispersion number was determined by the second moment matching
procedure described by Levenspiel and Smith (1957). For low levels of axial
dispersion
i C,
ND = ~~— - - — - 0.5 . (15)
2t2ZCiAti
where NQ is the dimensionless axial dispersion number.
The axial dispersion number and mean fluid residence time were determined
for ranges of superficial velocity, limestone particle size and depth of
packed-bed. The results obtained are listed in Table 9. Note that the axial
dispersion number was less than 0.02 in all cases and therefore the use of
Eq. 15 was reasonably appropriate.
A number of investigators including Edwards and Richardson (1968) and
Wilhelm (1962) have compiled data from various researchers and noted that
for axial dispersion in liquids in packed beds the Peclet number, i.e.,
Peclet number = = — • — , (16)
1) £ Nn JL
49
-------
o
40
30
5 20
o
8
10
0
p5cm/min
I6.5cm/min
ll.Ocm/min
A
i i r
limestone size-0.96cm
Temperature - I6°C
Superficial velocity-as
indicated
5.5cm/min
O
Figure 12.
10 15 20 25 30
Time after tracer injection , mm
35 40
Measured effluent tracer concentration plotted as a function of
time elapsed after tracer injection for four values of the super-
ficial velocity. Results were obtained using Column A, Figure 6,
-------
TABLE 9 RESULTS OF TRACER RESPONSE MEASUREMENTS OBTAINED USING
LABORATORY COLUMNS (FIGURE 6)
F.xp .
Ho.
1
2
1
tt
5
6
1
H
9
10
11
12
n
14
15
16
17
1R
I1)
20
21
22
21
2/i
25
26
27
2R
29
JO
11
12
13
Depth
L, cm
305
152.
335
315
335
335
335
335
335
335
335
335
315
335
315
335
315
315
315
61
152
213
211
213
213
213
213
213
213
213
211
211
213
Pnrtlcle
_Sl*e
d, cm
0.96
0.96
0.96
0.96
0.96
0.96
0.96
0.96
0.96
0.96
0.96
0.96
0.94
0.96
0.96
0.96
0.96
0.96
0.96
0.54
0.5'.
0.54
0.54
0.54
0.54
1.50
1.50
1.50
1.50
3.20
3.20
3.20
3.20
Porosity
c
0.41
0.41
0.41
0.41
0.41
0.41
0.41
0.41
0.41
0.41
0.41
0.41
0.41
0.41
0.41
0.41
0.41
0.41
0.41
0.43
0.43
0,43
0.43
0.43
0.43
0.49
0.49
0.49
0.49
0.49
0.49
0.49
0.49
Superficial
Velocity
U , cm/mln
B
12.2
18.2
18.2
6.1
6.1
14.7
22.0
22.0
22.0
29.3
29.3
29.3
37.5
36.7
36.7
5.4
21.4
37.5
53.5
5.4
5.4
5.4
21.5
32.2
53.7
5.3
16.0
37.4
48.1
0.3
1.1
1.9
2.7
Observations
Dispersion
Number,
\
0.0181
0.0075
0.0034
0.0078
0.0062
0.0106
0.0046
0.0034
0.0069
0.0040
0.0043
0.0065
0.0118
0.0072
0.0045
0.0088
0.0063
0.0051
0.0047
0.0183
0.0079
0.0085
0.0127
0.0085
0.0065
0.0149
0.0183
0.0089
0.0082
0.0200
0.0125
0.0096
0.0073
Peclet
Number,
Pe
0.17
O.H4
0.84
0.37
0.46
0.27
0.62
0.84
0.42
0.72
0.67
0.44
0.24
0.40
0.64
0.33
0.45
0.56
0.61
0.48
0.45
0.10
0.20
0.30
0.39
C.47
0.38
0.79
0.86
0.75
0.20
1.56
2.05
Mean Res-
idence
Time, t, mln
12.5
4.0
8.6
26.6
26.7
14.3
8.8
8.6
8.0
7.4
6.9
7.2
5.6
5.7
5.3
25.9
6.8
3.5
3.0
4.2
12.0
21 ,1
3.3
3.4
2.2
34.0
4.4
3.4
2.6
111.7
31.3
16.7
13.1
Tracer Mass
Recovered
%
98
96
99
104
105
113
96
99
99
102
127
106
US
101
104
104
108
97
109
119
79
102
81
114
97
108
96
110
108
109
88
76
69
-------
is essentially a constant over a wide range of Reynolds numbers and in addi-
tion, is only slightly affected by variation in the size of the packing mater-
ial. For the Reynolds number range of this study (1 < Re < 100) all the
literature values of the Peclet number analyzed by Wilhelm (1962) and Edwards
and Richardson (1968) fall in the interval 0.2 to 2. The range of Peclet
numbers for the results listed in Table 7 fall in the range 0.2 to 2 and
are therefore consistent with published values.
The mean and standard deviation of the Peclet numbers derived from the
quantities listed in Table 7 are 0.50 and 0.21, respectively. These values
suggest that a reasonable estimate of the dispersion number for the range
of conditions used in this study can be obtained from the following expression,
ND = (Pl)"1(d/L) = 2.0 (d/L) (17)
where, Pe, is the mean value of the Peclet number. Given the standard devia-
tion of 0.21 and the expected value of 2.0, ND, should fall in the interval
1.4 (d/L) to 3.3 (d/L).
The mean fluid residence time in the columns was calculated using the
measured bed porosity, e, (Table 8), the depth of the packed-bed, L, superfi-
cial velocity, Us, and the relationship
Ec = ^ (18)
The mean fluid residence time determined using the tracer response curve,
t, plotted as a function of the value calculated using Eq. 18 is given in
Figure 13. The agreement obtained is reasonable, a result which tends to
support the method used to measure bed porosity and the quality of the tracer
response data.
Before it was installed in the field the baffled-box contactor was sub-
jected to a pulse input, lithium chloride tracer response test. The results
of this test are plotted in Figure 14.
According to the dimensions of the contactor, the porosity of the bed
and the flowrate used in the test (13.6 L/min) the mean residence time should
52
-------
Ul
CJ
c
1
o
B
XI
c
o
I
"o
-E o
v»
e
VI 0)
.- 2
«T
E
o>
0
c
10
01
or
c
o
5 10 15 20 25 30 35
T from MRT calculations [ min ]
Mean Residence Time From First Moment of TrocerResponse Curve,min
Figure 13. Mean residence time calculated using the superficial velocity and
measured porosity plotted as a function of the mean residence time
from the tracer experiments.
-------
12
10
.
c
o
o
I 4
_J J-
I T
1 1 i T
1 1
I T
Flowrate = 13.6 L/min
Totol lithium
Input = ||50mg
Recovered = 1180 mg
8 16 24 32 40 48
Time After Tracer Injection (min)
56
64
Figure 14. Measured effluent tracer concentration plotted as a function of
the tine elapsed after tracer injection for the baffled-box
contactor (Figure 7).
54
-------
be 10.4 min in the limestone and a total of 6.5 min in the nine sampling/baffle
chambers (see Figure 7). The sum of these two quantities is approximately
17 min, a value which is in reasonable agreement with the mean residence
time of 18 min determined using the tracer response data and Eq. 18. Also
there was no evidence of significant short-circuiting or dead space.
PIPE SECTION PROCEDURES
To evaluate metal corrosion prior to and following limestone treatment,
pipe section leaching studies were conducted in both laboratory and field
experiments. Most pipe section experiments were conducted with 1 m (3.3
ft) lengths of 1.27 cm (% in.Hnside diameter copper pipe. Copper pipe was
amended with 2.54 cm (1 in.) of 50-50 percent lead-tin solder at both ends
of a given section, to simulate Pb corrosion from Pb solder joints. A limited
number of additional experiments were conducted with 1 m (3.3 ft) lengths
of 1.59 cm (5/8 in.) lead and galvanized steel pipe.
The pipe cleaning procedure used was a modified version of the ASTM
procedure. Pipe sections were soaked in 5% HC1 for two minutes. These sec-
tions were then drained and rinsed with 0.1 N NaHC03 to neutralize any acidic
solution adhering to the pipe. Finally, pipe sections were rinsed copiously
with distilled deionized water.
During metal leaching studies, aliquots of solution were placed in pipe
sections and the openings covered with parafilm. Solutions were equilibrated
with pipe sections at room temperature (22°C), for a given period of time,
generally 10 hours. Both pH and metal concentrations of leachate were measured
after equilibrium.
SAMPLING AND ANALYTICAL PROCEDURES
General Procedures
The analytical methods used in this study are summarized in Table 10.
Samples were collected in air-tight polyethylene containers for major solute
and trace metal analysis, in a sterilized glass bottle for bacteriological
analysis, and in biochemical oxygen demand bottles for oxygen analysis.
Temperature was measured and dissolved oxygen samples were fixed in the field.
Samples were transported in a cooler to the water quality laboratory at Syracuse
55
-------
Table 10 Analytical Methods
METHOD
PROCEDURE
REFERENCE
PH
alkalinity
Ca2+, Mg2+, Na+, K+
Al, Fe, Mn, Zn, Ca, Pb
potentiometrically with
glass electrode
strong acid titration with
Gran plot analysis
atomic absorption
spectrophotometry (AAS)
filtration 0.4 um polycar-
bonate filter, acidifica
tion (pH 1 with HN03 for
1 hr) analysis by AAS
graphic furnace
Standard Methods,
1975
Gran, 1952
Slavin, 1968
Slavin, 1968
2-
S04
dissolved inorganic
carbon (DIG)
dissolved organic
carbon (DOC)
NH4
+1
ion chromatography
ion chromotrography ;
turbidimetric method
syringe stripping of C02
and detection by gas
chromatography
filtration, ampoulation,
persulfate oxidation,
syringe stripping of C02
and detection by gas
chromatography
phenate colorimetry,
autoanalyzer
dissolved oxygen (D.O.) Winkler titration
Small et al., 1975
Small et al., 1975;
Standard Methods,
1975
Stainton, 1973
Menzel & Vaccaro,
1964
USEPA, 1983
Standard Methods,
1985
specific conductance
conductivity bridge
Standard Methods,
1985
standard plate count
Standard Methods,
1985
coliform
membrane filter
Standard Methods,
1985
turbidity
nephelometry
Standard Methods,
1985
temperature
thermometer
56
-------
University where they were measured for pH, alkalinity, specific conductance,
dissolved inorganic carbon, dissolved oxygen, turbidity, coliform and standard
plate count, and ampulated for the analysis of dissolved organic carbon within
8 hours of collection. Samples were stored at 4°C and analysis were completed
within one week of collection.
Laboratory Contactors
Samples were collected starting at the top sampling point of the column
and moving down the column using all the sampling ports provided. Samples
were withdrawn by gravity flow and collected in 500 mL polyethylene bottles.
To minimize C02 exchange, the bottles were completely filled and closed immedi-
ately. To minimize disturbance of the flow in the column during sampling,
a period of time equal to twice the distance between two sampling ports divided
by the interstitial flow velocity was allowed to elapse before the next sample
was taken.
The column experiments were conducted at room temperature (15° - 22°C).
To minimize microbial growth, the columns were initially rinsed with chlorin-
ated water followed by deionized water. The clear acrylic column was covered
with black plastic sheets to reduce exposure to light.
Field Contactors
Water samples were collected from the baffled box contactor and the
housekeeping cottages connected to this unit for a period of 2.5 years.
The sampling frequency was monthly except when weather conditions restricted
access. Samples were also collected from the spring and cottages on the
eastern side of the resort. This program included sampling at the cottage
with the Culligan unit. A more frequent, sampling schedule was employed
when the wound fiberglass unit was installed in Bay Side Cottage to treat
Big Moose Lake water during January - April, 1984.
Two types of tap water samples were collected in the field, a flowing
grab sample taken when the faucet was first opened and a grab sample obtained
after 3 minutes of continuous flow.
Quality Assurance/Quality Control Information Data
An assessment of field data requires an understanding of the precision
and accuracy associated with analytical determinations. In this study, both
sampling and analytical precision were evaluated. Triplicate samples were
57
-------
collected for analysis on a minimum of five percent of the total samples
collected, and triplicate determinations were performed on a minimum of five
percent of the samples collected. A summary of the range and coefficient
of variation from the triplicate sampling (an estimate of sampling and analy-
tical precision) program for a variety of water chemistry parameters is provi-
ded in Table 11. Moreover, we periodically performed a 4 by 4 analysis in
which four samples were collected and split four ways. The resulting 16
solutions were analyzed for major solutes. By a two-way analysis of variance,
(Barr et al. 1976) the sampling and analytical precision were evaluated (Table
12).
To evaluate analytical accuracy we performed charge balances, conductivity
checks, and alkalinity checks (Figure 15). Also we periodically evaluated
blind samples obtained from the USEPA Municipal Environmental Research Labor-
atory at Cincinnati, Ohio; the USGS Standard Reference Water Sample Program,
Denver, Colorado; and the USEPA Long-Term Monitoring Program through Radian
Inc. Results of some blind audit samples obtained from the USEPA Municipal
Environmental Research Laboratory are summarized in Table 13. Generally
the analyses of audit samples from this program were in agreement with reported
values. However, this audit program was not designed to evaluate analytical
f
accuracy of the low concentration ranges generally observed in dilute waters.
A more reasonable depiction of the accuracy of our analytical methods is
available through the analysis of dilute audit samples from the USEPA Long-Term
Monitoring Program conducted in May 1985 (Table 14). Although the percent
differences between the theoretical and values obtained by Radian Compared
to the values reported by Syracuse University were high for some determin-
ations, the actual magnitude of these discrepancies were generally low.
These relatively high percent differences may be attributed to the low solute
concentrations in this particular sample. Note some decrease in pH and increase
in DIG is evident between determinations made by Radian and analyses conducted
by Syracuse University, however ANC values were similar. These trends suggest
that when this synthetic sample was made-up it was undersaturated with respect
to the solubility of atmospheric CC>2. Over storage time, C02 equilibration
evidently served to depress pH values while increasing DIG concentrations.
Some discrepancy in DOC concentrations are also evident, however, given that
the source of this synthetic DOC is unknown, this trend is difficult to explain.
58
-------
TABLE 11
Summary of Sampling and Analytical Precision from Sample Triplicate Program
Range of
Range of Coefficient of
Parameter Range of Mean Standard Deviation Variation
PH
alkalinity
(mg CaC03-L'1)
Sp. Cond . (pmho • cm"^)
DIG (mg C-L'1)
DOC (mg C-L'1)
Turbidity (NTU)
DO (mg 02'L"1)
Standard Plate Count
(#•100 mL'1)
Total Coliform (#-100 mL"1)
Ca (mg Ca-L'1)
Mg (mg Mg-L"1)
Na (mg Na-L'1)
K (mg K-L'1)
S04 (mg 504'L'1)
Al (ug Al-L'1)
Cu (ug Ca-L'1)
Pb (ug Pb-L'1)
Zn (ug Zn-L-1)
6.01
7.7
50
4.9
0.76
0.31
7.0
3.7
0
5.3
0.65
2.4
0.66
4.0
0
0
0
13
- 7.68
- 34
- 107
- 7.7
- 2.3
- 0.53
- 7.3
- 195
- 64
- 12.2
- 0.89
- 7.2
- 2.6
- 4.3
- 33
- 1
- 123
- 42
0.006
0.29
0.12
0.06
0
0.035
0.1
1.1
0
0.08
0
0.012
0
0.21
3
0
0
20
- 0.231
- 1.0
- 6.2
- 0.60
- 0.30
- 0.10
- 0.5
- 36
- 16
- 0.67
- 0.016
- 1.36
- 0.25
-0.40
- 16
- 2
- 7
- 32
0.079
0.85 .
0.12
0.84
5.0
11
1.4
20
0
0.94
0
0.46
0
5
29
0
0
60
- 3.0
- 3.6
- 5.8
- 9.1
- 10
- 20
- 7.6
- 31
- 25
- 5.5
- 1.8
- 19
- 9.4
- 10
- 48
- 43
- 55
- 76
59
-------
TABLE 12 Estimates of sample collection and analytical precision
from 4x4 analysis for Big Moose Lake
Parameter
Samolins Precision
Analytical Precision
field pH
air equilibrated pH
ANC (neq-L"1)
Spec. Cond(vnnho-cnT^)
Ca(umol-L~l)
Mg(umol-L-l)
NaCpmol-L"1)
KCumol-L'1)
monomeric Al
( iinio 1 ' L ~" )
SO/ ~ ( UQO 1 * L~ )
NO 3 ~ ( UIQO 1 * L ~ )
Cl'Cmol-L"1)
H2S04(Mmol-L-l)
DOCCumol-L'1)
DICCumol'L'1)
Free F(umol'L~^)
Total FCumol-L'1)
Std. Dev.
0.020
0.028
2.6
2.1
0.49
0.02
0.77
0.34
0.40
2.9
1.3
1.6
5.1
54
1.6
0.018
0.14
C.V.
0.39
0.54
44
5.9
1.0
1.5
2.8
3.2
9.6
4.3
6.9
19
6.9
14
6.8
3.4
3.2
Std. Dev.
0.0088
0.012
5.1
0.3
0.36
0.003
0.14
0.13
0.19
0.61
0.76
0.83
1.7
21
1.4
0.0095
0.059
C.V.
0.17
0.23
86
0.84
0.78
0.79
0.52
1.2
' 4.9
0.90
4.0
9.5
2.3
5.4
1.9
1.8
1.4
60
-------
3CO
250
•^ 200
o-
O)
s
>: 150
"c
"a
a
-------
TABLE 13
Summary of Blind Sample Analysis Obtained from USEPA
Municipal Environmental Research Laboratory
Date Parameter True Value
7/1/82 Turbidity (NTU) 1.35
Turbidity (NTU) 5.50
N03"(mg N-L"1) 0.42
N03~(mg N-L'1) 7.3
F'(mg F-L"1) 0.12
F~(mg F-L'1) 1.1
10/1/82 Pb(ug Pb-L-1) 25
Zn(ug Zn-L'1) 15
AHug Al-L'1) 78
Mn(ug Mn-L-1) 15
Mn(ug Mn-L'1) 75
Fe(ug Fe-L'1) 80
Fe(ug Fe-L'1) 900
7/5/83 Turbidity (NTU) 5.9
Turbidity (NTU) 0.42
Pb(ug Pb-L-1) 22
Pb(ug Pb-L-1) 56
Cd(ug Cd-L'1) 1.2
Cd(ug Cd-L'1) 22
Measured Value
1.30
5.35
0.42
7.3
0.12
1.1
25
16
78
16
68
81
890
5.6
0.31
21
43
2.3
22
% Difference
6.7
2.7
0
0
0
0
0
-6.7
0
-6.7
9.3
-1.2
1.1
5.1
26
4.5
23
-92
0
62
-------
Date
Parameter
TABLE 13 (con't)
True Value Measured Value
Difference
12/14/83 pH
pH
Sp. Condtumho-cm"1)
Sp. Cond(umho•cm"1)
Ca(mg Ca-L"1)
Ca(mg Ca-L""1)
Mg(mg Mg-L"1)
Mg(mg Mg-L"1)
Na(mg Na-L"1)
Na(mg Na-L"1)
K(mg K-L"1)
K(mg K-L"1)
Cu(ug Ca-L"1)
Cu(ug Cu-L"1)
Pb(ug Pb-L'1)
Pb(ug Pb-L'1)
1/13/84 Turbidity (NTU)
Turbidity (NTU)
Pb(ug Pb-L"1)
Pb(ug Pb-L'1)
7/16/84 Cd(ug Cd-L'1)
Cd(uCd-L-1)
Pb(ug Pb-L'1)
Pb(ug Pb-L"1)
6.87
8.60
215
616
4.8
32
1.26
9.46
33.3
68.5
0.62
12.3
78.0
5.20
158
11.7
6.0
0.7
30
90.1
2.1
10.8
37.6
105
6.84
8.45
235
685
4.8
32
1.20
9.41
35.6
77
1.52
17.0
73.0
5.80
170
'30
6.8
1.2
45
94
1.8
8.8
29
86
-9.3
-11
0
0
4.7
0.5
-6.9
1.3
-145
-38
6.4
-11
-7.5
-156
-13
-71
-50
-4.3
14
18
23
18
^•Difference = (True Value - Reported Value)/(True Value) x 100
63
-------
TABLE 14 Summary of USEPA Corvallis Environmental Research Laboratory
Blind Audit Analysis.
All values in ueq-L~^ except where indicated.
Parameter Theoretical
pH
alkalinity — -- —
S04 48
N03 7 . 4
F- 2.2
Ca2+ 9 . 8
Mg2+ 37
Na+ 121
K+ 5.2
NH4+ 9 . 3
DOCCumol-L"1) 83
u L u v umo L L, j — - - -
Si02(umol-L"1) 18
5p. Cond .
(umol- cm~M
Cal. op. Cond . _--- —
sum of cations ------
sum of anions
Cal. HC03.
Radian
Value
7.31
108
49
7.9
2.2
11.1
35
117
4.8
8.3
98
98
18
17.3
18.6
176
177
an
^Difference
Syracuse Univ.
839
6.67
113
44
7.0
2.6
13
39
118
5
12.1
140
138
20
20
19.3
187
180
OR
840
6.96
110
45
8.2
2.4
12
39
109
4
10.7
152
138
20
20
18.5
175
179
QQ
Theoretical Radian
839
-9
15
25
5
-3
-4
23
41
10
840
-7
8
18
5
-11
-30
13
45
10
839
4
-11
15
15
10
1
4
31
30
29
10
6
840
2
-9
8
8
10
-7
-10
22
36
29
10
6
64
-------
Samples were routinely split with other researchers that analyze low
ionic strength solutions. Analytical checks on dilute solutions have been
made with investigators from Cornell University, McMaster University, Univer-
sity of Virginia, University of California at Los Angeles, the Insitute of
Ecosystem Studies Gary Arboretum, and Rensselaer Polytechnic Institute.
COMPUTATIVE ANALYSIS
Thermodynamic calculations involving trace metal solubility were conducted
with a modified version of the chemical equilibrium model MINEQL (Westall
et al., 1976). Calculations were corrected for the effects of ionic strength
using the Davies equations (Stumm and Morgan, 1981) and temperature. The
solubility and complexation constants used in our analysis are summarized
in Tables 15 and 16, respectively. The results obtained from chemical equilib-
rium calculations are highly dependent on the thermochemical data used.
Data analysis is complicated by inconsistencies in the literature. In this
regard, we conducted a thorough review to evaluate if there was consensus
among researchers in the use of thermodynamic data relevent to our study
(Tables 15 and 16). The results of this literature search suggest that
generally there is consensus in the use of thermochemical data. However,
some inconsistencies were evident in trace metal reactions.
There is considerable uncertainty in the stability constant for Cu(OH)2(aq)
(Vacenta, 1976). This uncertainty is significant because predictions of
total Cu in the neutral pH range are very sensitive to this stability constant.
The stability constant for Cu(OH)2(aq) was evaluated potentiometrically by
Quintin (1937), obtaining a value of log* 82 = -13.7; while Spivakovski and
Makouskaya (1968) used a precipitation method to obtain log* 62 = 13.2.
However, Mesmer and Baes (1974) estimated log* 82 = "17.3, almost four orders
of magnitude lower than previous estimates. Vacenta (1976) noted the magnitude
and significance in this discrepancy. She evaluated the Cu(OH)2(aq) stability
constant potentiometrically with a Cu ion selective electrode and obtained
results consistent with log* ^2 = ~13.7. Therefore we followed her lead
and used this value in our study.
Another perplexing inconsistency in thermodynamic data involves the
solubility of Pb(OH)2(s). Wagman et al. (1968) reported a value log* Kso
65
-------
TABLE 15 Equilibrium Constants at 25°C for the Solids Considered in the MINEQL Calculations.
REACTIONS
REFERENCE
1. Cu(OH)2(s) + 2H+
2. CuC03(s)
3. Cu2(OH)2C03 4- 3H+
4. Cu3(OH)2(C03)2 + 4H+
5. CuSC»4
6. Pb(OH2(s) + 2H+
7. PbC03(s)
8. Pb3(C03)2(OH)2(s)
9. PbS04(s)
10. Zn(OH)2(s) + 2H+
11. ZnC03(s)
12. Zn5(OH)6(C03)2(s)
13. ZnS04(s)
6H
2H20
Cu+2 4- C03-2
2Cu+2 + HC03~ + 2H20
3Cu+2 + 2HC03~ + 2H20
Cu+2 + S04"2
Pb+2 + 2H20
Pb+2 + C03~2
3Pb+2 4- 2C03~2
Pb+2
Zn+2
Zn+2
5Zn+2 + 2C0'2 + 6H20
804" 2
2H20
C03"2
2C03'2
Zn
+2
S04
-2
- 8.64 Baes and Mesmer 1976
- 9.63 Smith and Martell 1976
5.15 Baes and Mesmer 1976
3.75 Baes and Mesmer 1976
3.01 Wagman et al 1969
8.15-13.07 Wagman et al 1969
Topelman 129
-13.13 Hem 1976
-17.46 Sillen and Martell 1964
- 7.79 Smith and Martell 1976
12.45 Baes and mesmer 1976
-10.00 Smith and Martell 1976
9.65 Sillen and Martell 1964
3.01 Wagraann et al 1969
-------
TABLE 16 REACTIONS AND EQUILIBRIUM CONSTANTS, AT 25°C FOR THE AQUEOUS COMPLEXES CONSIDERED IN THE
MINEQL CALCULATIONS
log K (Ball et al., 1980)
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
M+2 +
M+2 +
M+2 +
M+2 +
2H+2 H
2M+2 H
3M+2 H
4M+2 ^
6M+2 ^
M+2 +
M+2 +
M+2 +
M+2 +
H+2 +
H+2 +
M+2 +
REACTIONS
H20 ,
2H20
3H20 1
4H20
- H20
- 2H20
i- 4!I20
h 4H20
h 8II2O
C03'2 + H+
C03'2
2C03"2
Cl"
2CI"
3C1~
4C1~
SO/ ~" ^ — •*-
Pb
»- M01I+ + }\+
Q
y. M(OH) + 2ll+
*• H( 011)3+ 3H+
*- M(OH)2"2 + 411+
*• M2OH+3 + 11+
*. M2(OH)2 + 211*
*• M3(OJ)/,+2 + 21I+
M4(OH)8+4 + ^+
*- M6(011)8"'
-• MI(C03+
-*• M(C03)2"2
-*. MCL+
-»- MC12°
->. MC1.J-
-*• MCl4~
- 7
-17
-28
-39
- 6
-21
-20
7
10
1
1
1
2
.71
.12
.06
.70
.36
.88
.88
.24
.64
.60
.80
.70
.75
Cu
- 8
-13
-26
-39
-10
-22
13
6
9
0
0
- 2
- 4
2
.00
.68
.90
.60
.36
.05
.03
.73 '
.83
.43
.16
.29
.59
.31
Zn
- 8.96
-16.90
-28.40
-41.20
- 9.00
12.43
5.30
9.63
0.43
0.45
0.30
0.20
2.37
-------
= 8.15. This value has been used throughout the literature in studies of
Pb chemistry (e.g. Hem and Durum, 1973; Ball et al., 1980; Faust and Aly,
1981). Topelmann (1929) obtained log* Kso = 13.07 for "freshly precipitated"
Pb(OH)2. This latter value has been cited by Feithnecht and Schindler (1963),
and ultimately used by Patterson et al. (1977), and Schock (1980, 1984) in
studies of Pb corrosion in water distribution systems. Schock (1980) indicated
that the discrepancy between the two solubility values represents the differ-
ence between "fresh" and "aged" precipitates. However given the magnitude
of this discrepancy (5 orders of magnitude), it is doubtful that crystallinity
of the precipitate explains the variation. Note that the value obtained
by Wagman et al. (1968) was calculated, not experimental. While Topelmann's
(1929) work was experimental the magnitude of experimental error in his study
in unclear. Therefore one value is not obviously superior to the others;
in fact the validity of both values could be challenged. The solubility
of Pb(OH)2 is a classic example of thermochemical data finding its way in
the literature and gaining acceptance over years of use without the benefit
of a critical review. Clearly if we are to improve our understanding of
Pb corrosion, better information on the solubility of Pb(OH)2 is desperately
needed.
In this study statistical analysis was facilitated by the use of the
Statistical Analysis System (SAS; Barr et al., 1976).
68
-------
SECTION 6
DERIVATION OF CONTACTOR DESIGN EQUATIONS
A set of equations was developed for use in predicting the effect of
design and operating variables on the chemistry of the limestone contactor
effluent. The following assumptions were made in formulating the model:
The contactor is a closed system, i.e., as the water passes through
the unit there is no exchange of carbon dioxide with the atmosphere,
The rate of limestone dissolution at any axial location, z, within
the column, Rz , is controlled by a mass transfer resistance (calcium
ion transport) and a surface reaction acting in series. Eqs . (4)
and (5), SECTION 4, were assumed to apply, i.e.,
Rz • K0(Ceq - Cbz) (19)
and
Kn = ^ . (20)
0 KC+ KL
where CQ(, is the calcium ion concentration when the influent solution
and limestone have reached equilibrium, Cbz is the bulk solution
calcium ion concentration at axial location z, Kc is the first
order surface reaction rate constant and KL is the first order
mass transfer rate constant for calcium ion.
Steady state conditions apply, i.e., the rate of limestone particle
shrinkage is negligible, and
The contactor is essentially a plug flow reactor with limited axial
dispersion.
The dispersion (dispersed plug flow) model of Levenspiel (1972) was
used to derive the principal design equation. The governing differential
equation is
where NQ is the dimensionless dispersion number, C is the reactant concentra-
tion, Z is the dimensionless axial distance (Z = z/L), L is the overall depth
of limestone in the column, t is the mean fluid detention time, e is the
bed porosity and r is the reaction rate expression.
69
-------
For steady state dissolution of limestone particles in a packed bed,
the reactant concentration, C, in Eq. 21 can be replaced by the quantity
(Ceq - Ct,z) and the reaction rate, r, by
r = Rz a C = K0 a (Ceq - Cbz) (22)
where a is the interfacial area of limestone particles per unit volume of
interstitial fluid in the column. Since the reaction rate expression, Eq.
22, is first order, the solution of Eq. 21 by Wehner and Wilhelm (1956) can
be used. The solution is given by;
Ceq - CbL _ 4n exp (1/2RD) _ (23)
ceq " cbo (1+ n)2exp(n/2 ND)-(1- n)2exp(-n/2ND)'
and
n - (1 + 4 K0 a ND e L/Ug) . (24)
where, in this case Cbo and C^L are the influent and effluent calcium ion
concentrations for a column of depth L and Us is the superficial or approach
velocity for flow through the column. Note that e L/US = t .
In packed bed reactors of an overall length which is much greater than
the size of the packing the amount of axial dispersion is small (NQ < 0.01)
and Eqs . 23 and 24 reduce to
cea ~ cbL KQaLe Ko a L e 7
= exp [" -§- + ( u - } ND]- (25)
us us
Therefore to determine the effluent calcium concentration C^L (Eq. 25),
for a column of depth, L, one must know the equilibrium and influent calcium
ion concentrations, Ceq and Cbo, the rate constant for the overall rate of
dissolution, Ko, the interfacial area of limestone per unit volume of inter-
stitial water, a, the bed porosity, e, the superficial velocity, Us , and
the axial dispersion number, NQ. In this study the porosities listed in
Table 8 were used and a was determined using Eq. 12,
70
-------
(12)
where d and <|> are the mean limestone particle size and sphericity. The mag-
nitude of Nj) was estimated using Us , d, L and Eq. 17,
ND = 2.0 (d/L) (17)
It will be shown in the next section that at least for column-type reac-
tors operating within the range of conditions used in this study the magnitude
of Ko can be estimated using well known dimensionless correlations from the
mass transfer literature. The equilibrium calcium ion concentration, Ceq ,
was determined using the chemical equilibrium model described below and in
Appendix A.
Equilibrium Calcium Concentration, Ceq
The equilibrium concentration of calcium ion at the limestone surface,
Ceq, was determined as a function of the raw water chemistry and temperature.
The calculations were based on chemical equilibrium principles which were
used to derive the following set of equations*
Charge Balance:
2(Cbo+ S) + Cc + [H+] = ((DIC)0 + S) (ai + 2a2) + Ca + KW[H+] (26)
Solubility Product Relationship for CaC03 :
(Cbo + S) ((DIC)0 + S) a2 = Ksp (27)
Inorganic Carbon lonization Fractions:
ai. = {([H+]/Kal) + Ka2/[H+] } "I (28)
and
ct2 = {([H+]2/Kal Ka2 + ([H+]/Ka2) + 1}-1 (29)
where Cbo is the initial (raw water calcium ion concentration, (DIC)O is
the initial dissolved inorganic carbon concentration, S is the amount of
71
-------
CaC03 dissolved from the limestone, Cc represents the total concentration
of the non-calcium and hydrogen cations, Ca represents the total concentration
of the non-inorganic carbon plus hydroxyl anions, Ka^ and Ka2 are the first
and second ionization constants for carbonic acid, Kw is the ion product
of water and Kgp is the effective solubility product for the calcium carbonate
in limestone.
The magnitude of Ceq was determined for each set of initial conditions
using a computational procedure in which the pH interval 6 to 10.5 was system-
atically searched to find the pH and the corresponding value of S at which
both the charge balance and CaC03 solubility product relationships, Eqs.
26 and 27, were satisfied. At equilibrium Ceq = C^o + S.
At each pH tested in the search procedure the ionic strength was cal-
culated and the activity coefficients were determined using the equations
given in Appendix A. The activity coefficients were used to correct the
equilibrium constants for changes in ionic strength. These calculations
were repeated at each pH until the ionic strength converged to an essentially
constant value.
The equilibrium constants at infinite dilution used in this analysis
are listed in Appendix A. The enthalpies listed in Appendix A and equations
from Plummer and Bussenberg (1982) were used to correct the equilibrium con-
stants for temperature. The effective solubility product for calcium carbonate
in limestone was determined experimentally. (See Section 5 and Table 6.)
Calculations were also made to determine the effect of equilibrating
the contactor effluent with atmospheric CC>2 on the pH and dissolved inorganic
carbon concentration.
The following equations were used:
Charge Balance:
2(Cbo + S) + Cc + [H+] =
-------
CT is the dissolved inorganic carbon concentration after the contactor effluent
has equilibrated with atmospheric carbon dioxide, Kg is Henry's Law constant
for C02, and pC02 is the atmospheric partial pressure of carbon dioxide.
c^ and 0(2 were determined using Eqs. 28 and 29.
The charge balance expression, Eq. 26, was solved for the equilibrium
pH using the search procedure described previously. The equilibrium dissolved
inorganic carbon concentration was then determined using Eqs. 31 and 32.
In many cases the concentrations of the ions which determine Cc and
Ca in Eqs. 26 and 30 are unknown and it is necessary to estimate the effect
of Cc and Ca on the total ionic strength of the solution. It was found in
this study that for dilute acidic waters the contribution of the Cc and Ca
ions to the total ionic strength was usually small and essentially constant
with pH and the dissolution of CaC03. A method for estimating the contribution
of the Cc and Ca ions to the total ionic strength using the measured specific
conductivity and the pH and the calcium and DIG concentrations is described
in Appendix A.
73
-------
SECTION 7
RESULTS AND DISCUSSION
MODEL VERIFICATION
Equilibrium Calcium Concentration
A series of laboratory experiments was conducted to test the predictive
capability of the chemical equilibrium model. Water of known chemical charac-
teristics was treated using a column containing d = 0.96 cm limestone particles
^
and a very low flowrate (Re =1). At this flowrate and limestone particle
size, equilibrium conditions were reached or closely approximated in at least
the bottom half of the downflow laboratory column.
The water used in these experiments was prepared by adding HC1, CaCl2
and NaCl to deionized water. The raw water pH ranged from 2.3 to 4.5. While
most experiments were conducted with no added calcium ion in the raw water
five were made with 28 mg Ca/L.
The results obtained using raw water with no added calcium are plotted
in Figures 16 to 18. A plot of pH as a function of the distance to the sam-
pling port is indicated in Figure 16 and corresponding plots for [Ca"1"1"] and
(DIG) are depicted in Figure 17 and 18, respectively. Using each of these
figures the equilibrium values of pH, -logtCa"*"*"] and -log(DIC) were determined
for each raw water pH by estimating the magnitude of the asymptotic limit
for each parameter at sampling port depths greater than approximately four
feet.
pHeq, -logtCa"*"1"] and -log(DIC) plotted as a function of the raw water
pH, pHo are shown in Figure 19. The data points were determined by the loca-
tions of the asymptotes in Figures 16, 17 and 18 and similar plots. The
lines shown in Figure 19 were plotted using the chemical equilibrium model
and the constants listed in Appendix A. In general the agreement between
the model predictions and the column data is good for pH and calcium concen-
tration. In the cases of DIG, for pHo > 3.5 the measured values of DIG are
somewhat greater than those predicted by the model. It is possible that
some carbon dioxide entered the solutions after the samples had been drawn
from the column.
74
-------
-o-
-O-
-o-
Curve
A
B
C
0
E
F
8
-o F
0
468
8ED DEPTH,L (ft)
10
12
Fi<»vre 16. pH plotted as a function of the axial distance to the sampling
port and influent pH, pH . Results were obtained using Column A,
Figure 6.
75
-------
200 _
4.50
4.10
3.60
3.20
3, CO
2.90
2.70
2.60
2.30
468
BED DEPTH ,L(ft )
/0
12
14
Figure 17. Calcium concentration plotted as a function of the axial distance
to the sampling port and influent pH, pH . Results were obtained
using Column A, Figure 6.
76
-------
3.00
2.90
2.70
2.60
2.30
6 8
8£0 DEPTH,
14
16
(fr)
Figure 18. Dissolved inorganic carbon concentration plotted as a function of
the axial distance to the sampling port and influent pH, pH .
Results were obtained using Column A, Figure 6.
77
-------
The CaC03 chemical equilibrium model was also used to plot the set of
curves in Figure 19 which give the pH, and equilibrium calcium and DIG concen-
trations for the case when the contactor effluent was equilibrated with atmos-
pheric carbon dioxide. The measured equilibrium DIG concentrations for pHo
> 3.5 were less than the values predicted by the model indicating that if
C02 uptake by the sample solutions inadvertently occurred, equilibrium with
the atmosphere was not reached.
In general, as the raw water pH decreased the maximum effluent pH decre-
ased, from a value which was greater than 9.5 when the raw water pH was greater
than 4.5 to approximately 8 when the raw water pH was 3.2. If the effluent
pH was 9.5 or greater, equilibrating the contactor effluent with atmospheric
carbon dioxide reduced the pH of the solution to slightly less than 7.6.
Decreasing the raw water pH also increased the amount of CaC03 dissolved
at equilibrium. The amount dissolved increased exponentially as the raw
water pH was reduced below 4.
The presence of calcium ion in the raw water tended to reduce the dis-
solution of the CaC03 by the common ion effect. With 28 mg Ca/L in the inf-
luent, the maximum effluent pH was approximately 9.5. For raw water pH values
greater than 3 the amount of DIG at equilibrium was also reduced.
An experiment was conducted in which the raw water DIG concentration
was adjusted by the addition of sodium bicarbonate. A plastic sheet was
used to cover the raw water reservoir. Unfortunately the raw water DIG con-
centration decreased significantly during the course of the experiment, appar-
ently through the release of carbon dioxide to small pockets of gas which
remained under the plastic cover. As the result of this experimental problem
the laboratory column data could not be used to test model predictions for
variable raw water DIG.
The CaC03 chemical equilibrium model was used to plot pHeq isopleths
on a graph of influent calcium concentration as a function of influent DIG.
An influent pH of 6, and a total ionic strength of 4 x 10"^ were used in
these calculations. The results are presented in Figure 20.
The influent DIG concentration may have a significant effect on the
equilibrium pH (Figure 20). For example, for this initial pH (pHo = 6.0)
and a calcium concentration of 10 mg Ca/L, the equilibrium pH is slightly
78
-------
K10
ao
2" ao
0.
r.o
6 JO
z.o
o»
E 3.0
o
Q
O
O
4.0
2.0
' 4.0
j I
I I l
C
A _
I I i I !
A ~
j i
I ! I
20 3.0 4.0 SO
initial pH, pH0
6.0
Figure 19. Equilibrium pH, dissolved inorganic carbon and calcium concentra-
tions plotted as a function of the influent pH and the following
conditions: Curve A - closed system and C, =0; Curve B - closec
system and C.QO = 28 mgCa/L; Curve C - closea/open system and
C^Q = 0; Curve D - closed/open system and C, = 28 mgCa/L. The
lines were drawn using the chemical equilibrium model described
in Appendix A.
79
-------
OO
o
Raw Water
pHo =6.0
2 4 6 8 JO 12
Influent Dissolved Inorganic Carbon (mgC/L)
Figure 20. Influent calcium concentration plotted as a function of the in-
fluent dissolved inorganic carbon concentration and the equilibrium
pH for an influent pH of 6.0.
-------
less than 9.0 when the influent DIG concentration is 1 mg C/L. Increasing
the influent DIG concentration to 5.5 mg C/L causes the equilibrium pH to
decrease to 8.0. Since the DIG concentration in dilute acidified water usually
ranges from 0.5 to 5 mg C/L these results indicate that the raw water dis-
solved inorganic carbon concentration can be an important parameter in contac-
tor design.
Contactor Design Equations
The contactor design equations and the laboratory column data were used
to test the assumption that under the conditions of this study the overall
dissolution rate constant, Ko, is equal to the mass transfer coefficient,
KL, i.e.,
KL « Kc
and
* K (33)
where Kc is the surface reaction rate constant. •
A best-fit value of Ko was determined for each laboratory column exper-
iment. For each measured calcium concentration at a given column depth,
L, a corresponding model predicted value was determined using L, an assumed
value of KQ ancj the basic design equation, Eq. 25, rearranged to give,
^bL Ceq -^
us
- Ko a L £ + rKo a L £ )2
(34)
Ceq was determined using the chemical equilibrium model and the raw water
characteristics. NQ and a were calculated using Eqs. 17 and 12 and the known
quantities, d, \l>, L and e .
The best-fit value of Ko was determined by minimizing the sum of the
square of the difference between the measured and the calculated value of
the calcium concentration for each depth, L, i.e.,
- CbL')2 (35)
81
-------
where n is the number of data points for each run. In most cases the total
number of data points per run was between 10 and 14.
n
A plot of E (cbL " cbl/)2 as a function of Ko for a typical experiment
i=l
(number 32) is illustrated in Figure 21. As indicated by the minimum in
the curve, the best-fit value of Ko in this case was approximately 0.032
cm/min. The experimental conditions and the best-fit value of Ko for each
run in this series are listed in Appendix B.
Calculated and measured values of the calcium concentration for this
typical experiment are plotted as a function of the column depth in Figure
22. The best-fit value of Ko (0.032 cm/min) was used with Eq 34 to plot
the curve.
The best-fit values of Ko were compared with values of K0 determined
by plotting
-in c*q '
ceq ~ cbo
as a function of the depth of the sampling port, L, for each value of
Only the experimental runs in which there were at least three values of
(from the top of the column) where Cb < 0.9 Ceq were used in this comparison.
The slope was determined for each set of data by fitting a straight line
through the 0.0 point and as many data points as possible. Examples of these
plots and fitted lines are given in Figure 23 for superficial velocities
of 5.5, 22 and 55 cm/min. Each slope was converted to a value of Ko using
(slope)Us
K0 = - (36)
a e
and the known quantities Us, a and e.
The correlation coefficient for the comparison of the best-fit values
of K0 and the values determined by fitting a straight line to the In [(Ct>L-
Ceq)/(Ceq - CbL)] points is 0.85 (Figure 24).
Many studies have been conducted in which the object was to measure
mass transfer coefficients in packed beds and to formulate predictive relation-
ships using dimensionless parameters (Roberts et al., 1985). Numerous data
from a number of investigations have been correlated by plotting the Chilton-
Colburn mass transfer factor,
82
-------
00
RUN 32, APPENDIX 8
10
20 30 4O 50 60 70
DISSOLUTION RAT£ CONSTANT, Kflx I03 (cm/mln)
60
Figure 21. Sum of the square of the difference between the observed and the
model predicted calcium concentration plotted as a function of
the dissolution rate constant for run number 32, Appendix B.
Plot illustrates how best fit values of K were determined.
-------
CXI
o
CJ
o>
a
u
§
cc
i
o
o
u
O
O
RUN 32 APPENDIX B
K0 = 32 x I0~3cm/min
I
50
100
150 200
BED DEPTH, (cm)
250
300
350
Figure 22. Model predicted and measured calcium concentrations plotted as a
function of the axial distance to the sampling port for run num-
ber 32 and K = 0.032 cm/min.
-------
8£0 OEPTH.Ucm)
0 3O
60
120
150
ISO
210
RUN 86 APPENDIX 0
Figure 23. In [(C. - C )/(C, - C )] plotced as a function of Che axial
oL eq DO eq
distance to the sampling port for runs 29, 31 and 32.
85
-------
0.15
0.10
•o
o
_c
E
0.05
0.05 0.10
K0 (cm/min) Mefhod I
0.15
Figure 24. Dissolution race constant determined by the least squares method
(Method II) plotted as a function of the value obtained using
plots such as Figure 23 (Method I).
86
-------
KL 2/3
(37)
as a function of a modified Reynolds number
d Us
MRe = —,-. r (38)
v(1-e)
Chu and Khalil (1953) found that the following expressions gave a reason-
able fit of data compiled from the literature,
jD = 5.70 (MRe)~°-78 1 < MRe < 30 (39)
jD = 1.77 (MRe) -°-44 30 < MRe < 10,000 (40)
JD is plotted as a function of MRe in Figure 25.
Eqs. 38, 39 and 40 were used in this study to estimate a value of the
mass transfer coefficient, KL, for each experimental run. For example at
low values of the modified Reynolds numbers,
KL = 5.70 (MRe)'0-78 (Us) (v/D)'2/3 (41)
In computing a value of KL for each experimental run the approximate
calcium ion diffusivity at 20°C, D20°C> was corrected for the temperature,
T°C , by,
D = n T + 273 v20°C (42)
TO uor\°(-i ' —^— \^f-/
T° 20 C 293 v T°
A corrected value of the kinematic viscosity, VTO, was determined for each
temperature using equations derived by Blackwell (1984).
Values of Ko (calculated) were calculated using Eqs. 20, 38, 39, 40,
and 42 and assumed values of D2Q°c anc^ ^c an<* compared with the values of
KQ determined by fitting the experimental data, Ko (best-fit) (see Appendix
B). A simple variable step grid, search procedure was used to find the values
of D2Q°c an<* KG which maximized the correlation coefficient for the comparison
of Ko (calculated) with Ko (best-fit). The values determined were
87
-------
CO
CO
or
1.0
o
2
S
U.
CO
<
or
in
c/j
0.1
0.01
1 I I I I 11II 1 .
I i i i i i ill i i i i i 1111 ii i Mini L
10 100 1000
MODIFIED REYNOLDS NO. 3*Us/i/ (I-*)
10,000
Figure 25. Mass transfer factor, j , plotted as a function of a modified
Reynolds number using the equations derived by Chu and Khalil
(1953).
-------
D20°C = 1>2 x 10~5 Cm2/s
and
Kc = 0.85 cm/min .
The corresponding maximum correlation coefficient was 0.73. KQ (calculated)
is plotted as a function of Ko (best-fit) in Figure 26 using Kc = 0.85 cm/min
and D2o°c = 1-2 x 10~5 cm2/s.
The magnitude of Kc, the surface reaction rate constant determined using
the optimization procedure was significantly greater than all calculated
values of KL and therefore supports the assumption that under the conditions
of this study the effect of the surface reaction was negligible and mass
transfer (of calcium ion) controlled the rate of the dissolution process,
i.e.,
a.
It should be noted that the use of a constant calcium ion diffusivity,
e.g., 1.2 x 10~5 cm2/s (20°C), in these calculations is an approximation.
The diffusivity of a cation such as calcium depends on the nature of the
associated anion and the presence of other electrolytes. Diffusion to or
from dissolving calcium carbonate in limestone is a complicated process which,
in many cases, probably involves changes in ion speciation (particularly
the inorganic carbon species) with distance from the surface. Therefore,
the principal anion associated with the calcium may change as it diffuses
from the surface, and hence, the diffusivity of the calcium ion may be affec-
ted.
A plot of measured as a function of model predicted values of the calcium
ion concentration is presented in Figure 27 for the runs listed in Appendix
B. The points plotted represent measurements and calculations for all sampling
port locations (see Figure 6). The model predicted values were determined
using the assumption that Ko ^ KL and that KL is given by Eq. 39 or 40.
The agreement between the measured and model predicted calcium ion concentra-
tions is good except at high calcium concentrations (> 8 mg Ca/L) where all
the calculated values tend to be larger than the measured concentrations.
89
-------
o
a
a-
o
E
E
o
,E
0.001
0.01
0.001
0.01
KQ (cm/min , from dofo)
0.1
1.0
Figure 26. Values of the dissolution race constant calculated using the model
equations plotted as a function of the experisental (best-fit)
values listed in Appendix B.
90
-------
o
CJ
0>
E
-Q
CJ
2
O
5
O
z
o
o
o
Q
liJ
cc
Uj
to
CO
o
0 5 10
MODEL CALCULATED CALCIUM CONCENTRATION,
Figure 21. Observed calcium concentration plotted as a function of the model
predicted value. The points include all sampling port locations
for the runs listed in Appendix B. The model equations were used
to determine K for each run.
-------
It will be shown in later sections that the pH of the contactor effluent
is an important parameter in determining the effect of contactor treatment
on metal release. Since the contactor design equations are based on the
transport of the calcium ion, the calculated effluent calcium concentration,
CbL, must be used with the chemical equilibrium model to determine the corre-
sponding effluent pH. To test the efficacy of this approach, effluent pH
values were calculated for each sampling port in a number of experimental
runs and then compared with the measured pH values.
The pH at each sampling port was calculated using the charge balance
equation and the theoretical amount of CaC03 dissolved at that port. For
examp.le, since the calculated molar concentration of calcium ion at a given
depth, L, is CbL, the following substitutions are made in Eq. 26;
cbo + S = cbL
and
(DIC)0 + S = (DIC)0 + (CbL - Cbo). (44)
The magnitude of the quantity (Ca - Cc) in Eq. 26 was estimated using
the raw water conditions (Cbo, (DIC)O, pH and temperature) and the basic
charge balance assumption, i.e.,
(Ca - Cc) = 2 Cbo + [H+] - (DIC)0
where a^ and a2 are given by Eqs . 28 and 29. The specific conductivity can
be used, as suggested in Appendix A, to make approximate ionic strength correc-
tions, if necessary. Using Eqs. 43 and 44 and the estimated quantity (Ca
_ c )> Eqs. 26, 28 and 29 were solved for the pH which corresponds to the
transport model calculated value of CbL.
A plot of the measured pH as a function of column depth for experimental
run number 32 is illustrated in Figure 28. In this run the experimental
conditions were pHo = 4.0, Cbo = 0.2 mg Ca/L, (DIC)O = 0 and T = 10°C. The
initial ionic strength was approximately 4 x 10~4M. The calculated equilibrium
concentration of calcium ion and the equilibrium pH are 6.0 and 9.6, respec-
tively. The model calculated effluent pH values plot somewhat above the
measured values.
92
-------
I
O.
Q
UJ
in
<<
UJ
2
Q
<
O
UJ
O
UJ
Q
O
10
8
40-
t
0
O
O
50
Calculated using Model predicted calcium concentration*
O
O
O
RUN 32, APPENOIXB
I
100 150 200
BED DEPTH,L(cm)
250
300
350
Figure 28. Model predicted and measured pH plotted as a function of the
axial distance to the sampling port for run number 32 and
K = 0.032 cm/min.
-------
A plot of the measured pH as a function of the calculated pH for the
runs listed in Appendix B is shown in Figure 29. The poor agreement obtained
between the measured and the model calculated pH values may be due to the
inadvertent uptake of carbon dioxide in samples during collection and pH
measurement. The uptake of carbon dioxide would result in pH values less
than those predicted using the mathematical model with the closed-to-the-
atmosphere assumption.
The conclusion that carbon dioxide entered at least some of the samples
is supported by the results plotted in Figure 30 where the dissolved inorganic
carbon concentration has been plotted as a function of column depth for experi-
mental run number 32. The equilibrium concentration of DIG for this run
is approximately 1.8 mg C/L. A number of points are 20 percent greater than
the equilibrium value (Figure 30).
The model calculated calcium concentrations were used to determine the
alkalinity for each sampling port location using
Alkalinity (ueq/L) = influent alkalinity (ueq/L) + (46)
(50, ueq/mg) (CbL> mg/L)
The model calculated and measured alkalinities for run number 32, Appen-
dix B, are plotted in Figure 31.
Measured alkalinities plotted as a function of the model calculated
values for the runs listed in Appendix B are illustrated in Figure 32. The
alkalinities are given as the change in alkalinity between the influent and
each sampling port. The agreement obtained between the model calculated
and measured alkalinities supports the use of the model as a predictive tool.
It is apparent, however, that alkalinity and calcium ion concentration are
significantly better parameters for model calibration than pH.
FIELD STUDY RESULTS
Baffled Box Contactor
The baffled-box contactor was sampled at regular intervals during the
period June 1982 to October 1984. Twenty-three sampling visits were made
to the unit. During the period January to April 1984, the main pipe between
the contactor and the cottages became frozen and no samples were collected.
94
-------
8
o
UJ
o:
ui
to
S 6
678
MODEL CALCULATED, pH
10
Figure 29. Observed pH plotted as a function of the model predicted values
for all sampling port locations for the runs listed in Appendix B,
The model equations were used to determine K for each run.
-------
The first samples in this study were collected on June 24, 1982. The
first data point in each plot of the field study results is for the samples
obtained on this date. At that time the contactor contained limestone which
has been in place approximately one year. On June 28, 1982 the limestone
in the contactor was replaced with fresh stone and the results plotted for
6/28/82 (the second point on each graph) are for this condition.
On September 26, 1983 the contactor was opened to obtain samples of
limestone and to measure the amount of the limestone that had dissolved during
the 455 day period since June 28, 1982. Before the unit was put back into
operation, each compartment was filled to the top with fresh stone. The
partially dissolved stone was not replaced.
Results obtained from the baffled box unit are plotted in Figures 33
to 38. In plotting these data points "influent" depicts sample location
1, the sampling compartment at the inlet, Figure 6, and "effluent" represents
sample location 6, the compartment at the outlet.
The influent water temperature is plotted in Figure 33. The temperature
during the sampling period ranged from 2.5°C in February to 13°C in August.
The average water temperature for the 24 month period, June 1982 to July
1984 was 7.5°C.
The baffled-box contactor increased the pH on average by about 1 unit.
An effluent pH of 8.5 was measured immediately after fresh limestone was
installed in June, 1982. During the next month the pH decreased to about
7. The average effluent pH for the study period was 7.33 and the average
influent pH was 6.34. The pH values are plotted in Figure 34.
A significant amount of the variability in the effluent pH (and in other
effluent characteristics such as the calcium concentration) can be attributed
to the variability in the influent chemistry and temperature as well as the
flowrate through the units. It was not possible to routinely measure the
instantaneous flowrate at the baffled-box contactor. The estimated average
daytime flowrate is approximately 10 L/min. It is possible the flowrate
ranges from near zero at times in the fall and spring to approximately 20
L/min during the summer months when the seven cottages served by the unit
were occupied by families. In the winter the resort owner attempts to maintain
a constant minimum flowrate through the system to minimize the chance that
the pipeline will freeze.
96
-------
MOOELCALCULAteDEQUJUeRIUM DlC
50
IOO
ISO 200
B£0 OEPfH, L fern)
250
300
350
Figure 30. Model predicted and measured dissolved inorganic carbon concentra-
tions plotted as a function of the axial distance to the sampling
port for run number 32 and K =* 0.032 cra/min.
-------
CO
MODEL PREDICTED
50
100 150 200
BED DEPTH, L (cm)
250
3OO
350
Figure 31.
Model predicted and measured alkalinity plotted as a function of
the axial distance to the sampling port for run number 32 and
K = 0.032 cm/min.
o
-------
4OO
cr
•>
Q
Uj
or
CO
<
UJ
5
300
200
/CO
o o
0 /OO 330 300
A ALKALINITY MOCEL CALCULATED(ueq/L)
Figure 32. Measured change in alkalinity? within the laborator3.T contactors
plotted as a function of the model predicted change. The data
obtained for the runs liscad in Appendi:-: 3 (all sampling ports)
were used.
each run.
The model equations were used to determine K for
99
-------
y 9
-------
I
Q.
j i
• Treated
o Untreated
J A 0 D
IS82
F A J A 0 0
1983
Ti me
F A J A 0 D
1984
Figure 34. Influent and effluenc pH plocted as a function of tine for the
baffled-box contactor.
101
-------
The calcium concentration results are plotted in Figure 35. The average
influent and effluent calcium concentrations for the study period were 4.1
and 11.3 mg Ca/L, respectively. The highest effluent calcium concentration,
24 mg Ca/L, was measured in January 1984 when flow through the contactor
was at a minimum.
The average influent and effluent magnesium concentrations were 0.67
and 0.80 mg Mg/L, respectively. The average increase in the magnesium concen-
tration across the contactor was 0.13 mg Mg/L, or 5.4 x 10~^M magnesium.
The average dissolved aluminum concentrations in the influent and effluent
were 0.071 and 0.066 mg Al/L, respectively. Given the variation in the measured
aluminum concentrations the difference between these values is not significant.
The alkalinity is plotted in Figure 36. The average influent and effluent
alkalinities for the study period were 0.15 and 0.57 meq/L (7.5 and 28.5
mg CaC03/L), respectively.
The average influent and effluent DIG concentrations were 3.7 and 7.0
mg C/L, respectively. The DIG data are plotted in Figure 37. There are
no obvious trends in this data.
The influent and effluent standard plate count bacteria results are
plotted in Figure 38. The SPG results correlate with water temperature,
the highest counts were obtained during the summer months. While there was
considerable variability in both the influent and effluent measurements it
appears that the effluent counts were somewhat higher than those of the influ-
ent. The average values for the influent and effluent for the study period
were 89 and 170 per 100 ml, however, the statistical significance of this
difference is negligible.
Total coliform measurements were initiated in November 1982 in response
to a request by the resort owner and the New York State Department of Health.
High coliform densities, (Figure 39), correspond as expected to periods with
high surface runoff, i.e., late fall 1982 and spring snowmelt, 1983. In
general, transport through the contactor seemed to have no effect on the
measured total coliform density. In early 1984 the resort owner installed
an on-line ultra-violet light disinfection unit downstream of the contactor
unit.
102
-------
• Treated
o Untreated
J A 0 D
1984
Figure 35. Influent and effluent calcium concentration plotted as a function
of time for the baffled-box contactor.
103
-------
.8
.6
to
-5
o- 4
-------
12
8
O
c
o>
o
c
o
o
c
o
JQ e
6
o
'c
o
o»
O)
I 2
• Treated
o Untreated
J_
J A 0 0
IS82
F A J A 00
1983
Time
A J A 0 D
f984
Figure 37. Influent and effluent dissolved inorganic carbon concentration
plotted as a function of time for the baffled-box contactor.
105
-------
I I
Treated
o Untreated
CO
J A 0 D
1982
A J A 0 0
1984
Figure 38. Influent and effluent standard plate count bacteria concentration
plotted as a function of time for the baffled-box contactor.
106
-------
o
-g
70
1 60
o
o
6 50
2
o
'£ 40
o
CD
E
o
30
o 20
o
0
0
1 I
I I
I I
I I 1 I
I I
o Untreated
• Treated
JJASONDIJ
1982
Time
Figure 39. Influent and effluent total coliform bacteria concentration
plotted as a function of time for the baffled-box contactor.
-------
The results obtained in monitoring the baffled box contactor are sum-
marized in Table 17. The mean, standard deviation and total number of data
points are listed for each measured quantity. The average change in the
calcium concentration through the contactor (7.2 mg Ca/L or 1.8 x 10~^M cal-
cium) is in reasonable agreement with the average change in the alkalinity.
(0.42 meq/1 or, in terms of the calcium ion molar concentration, 2.1 x 10~^M).
The average increment in the DIG, 3.3 mg C/L, is equivalent to a 2.8 x 10"^
molar increment in the calcium concentration. This lack of agreement between
the average incremental change in the DIG and changes in calcium and alkalin-
ity suggests that DIG may be produced within the contactor by microbial res-
piration or that gaseous carbon dioxide entered the samples after they were
collected.
The results listed in Table 17 indicate that the fluid contact time
within the box contactor during the sampling period was apparently long enough
to produce an effluent which was essentially in equilibrium with the CaC03
in the limestone. If the average influent water chemistry is used with the
chemical equilibrium model (with a closed-to-the-atmosphere assumption) to
predict the equilibrium calcium concentration the result obtained (10.9 mg
Ca/L) is close to the average measured effluent calcium concentration (11.3
mg Ca/L).
The chemical equilibrium model, however, predicts an equilibrium pH
(pHeq = 9.18) which is significantly greater than the average measured value
for the contactor effluent (7.33). This discrepancy is apparently the result
of the uptake of gaseous carbon dioxide by the effluent samples or possibly
by the production of carbon dioxide within the sample bottles by microbial
respiration.
The negative logarithm of the carbon dioxide partial pressure, pCC>2,
was calculated for each measured pH value and alkalinity. The following
equations were used:
ao ,,^
pc°2= ^- ' (47)
and
DIG = alkp -10(-PKW - PH m) + KTPHm (4g)
108
-------
Table 17 Summary of Baffled-Box Contactor Results
Field Measurements
Parameter
pH
Calciumdng Ca/L)
Magnesium
Aluminum
Alkalinity(Meq/L)
Alkalinity(mgCaC03/L) 7.5
Dissolved Inorganic
Carbon(mg C/L) 3.7
Specific. Conductivity
(umhos/cm) 56
Standard Plate Count 89
(No/ml)
Dissolved Organic
Carbon (mg C/L) 1.4
Dissolved Oxygen
(mg 02/L) 7.2
Turbidity (NTU) 0.6
Mean
6.34
4.1
0.67
0.071
0.15
7.5
Influent
Std. Dev.
0.26
1.0
0.12
0.087
0.06
30
n
23
20
21
10
18
18
Mean
7.33
11.3
0.80
0.066
0.57
28.5
Effluent
Std. Dev.
0.46
3.8
0.19
0.066
0.20
10.0
n
23
20
22
10
18
18
1.1
9
142
0.5
0.6
0.5
19
12
15
8
9
7.0
87
170
1.5
7.4
0.7
1.6
16
365
0.6
0.8
0.2
20
13
16
9
9
109
-------
where alkm and pH are the measured alkalinity and pH, KH is Henry's Law con-
stant and ao, aj_, and 02 are the ionization fractions for H2C03 (see Eqs.
28, 29 and 32). The calculated values of pCC>2 are plotted in Figure 40.
The CC>2 partial pressure values plotted in Figure 40 were all signifi-
cantly greater than the partial pressure of atmospheric carbon dioxide, 10"^ -5
or pC02 = 3.5 (20°C). The average pCC>2 values for the influent and effluent
were 2.55 adn 2.90, respectively. Both values are consistent with CC>2 partial
pressures determined for soil-water systems and indicate that the water flowing
through the contactor is supersaturated with respect to atmospheric carbon
dioxide.
The pH calculated using the equilibrium model for contactor effluent
equilibrated with atmospheric CC>2 is 7.87, a value which is greater than
the average observed value of 7.33. This result supports the contention
that the effluent samples were not maintained as intended, in a closed-to-the-
atmosphere condition and that microbial respiration within the bed may have
been an additional factor.
The influent and effluent dissolved oxygen concentrations, 7.2 and 7.4
mg/L, were essentially the same (Table 17). The influent and effluent dis-
solved organic carbon and turbidity measurements also showed essentially
no change across the contactor.
On September 26, 1983, after 455 days of continuous operation the spring
was partially drained and the lid was removed from the box contactor unit.
Measurements were carefully made with a ruler to determine how much of the
total volume of each compartment was still occupied by the limestone bed.
(On June 26, 1982, each of the five compartments had been completely filled
with fresh stone.) Representative samples of stone were removed from each
compartment to determine the mean particle diameter by the water displacement
technique (see Section 5). Two additional samples from the first and second
compartments were obtained for size analysis by sieving. Representative
samples were removed from the first and last compartments for particle surface
analysis by scanning electron microscopy (SEM) and x-ray energy spectrometry
(XES).
The general physical condition of the limestone was similar to that
of fresh limestone, however, some differences were noted. In the first several
110
-------
&
o
UJ
QC
ID
CO
en
UJ
o:
a.
h-
cr
o
o
UJ
o
UJ
3.5
2.5
Effluent
/nfluenf
SO NO
1982
JF
MA MJ JA
1983
i ! I
SO NO
I I I
JF MA MJ JA
1984
TIME (months)
Figure 40. Calculated partial pressure of carbon dioxide plotted as a function
of time for the influent and effluent of the baffled-box contactor.
Ill
-------
compartments a brown, humous-like material was evident in the interstitial
spaces. The bed did not appear to be clogged by this material but when the
stone layer was disturbed the water in the compartment became turbid with
coarse, brown particulate matter. It seems reasonable to assume that because
the spring is in a hardwood forest most of the material was decomposing leaf
litter.
The used, wet limestone from the contactor seemed to have a slightly
"slimy" feel when rubbed between the fingers. A cursory microscopic examina-
tion of the surface (SEM and light microscope) did not yield any obvious
indications of microbiological contamination. The XES analysis (which will
be discussed in more detail later) did not indicate the presence of major
amounts of metal hydroxide precipitates on the particle surfaces. A reasonable
assumption seems to be that although the microscopic examination was essen-
tially negative, the sliminess was due to some microbiological contamination
and possibly deposited particles of soil material.
The amount of limestone dissolved during the 455 day period was sufficient
to cause a measurable decrease in the volume occupied by the bed and a decrease
in the average size of the stones, particularly the material in the first
two compartments. The percent of the total compartment volume occupied by
the limestone bed and the mean particle diameter for each of the five com-
partments is given in Table 18.
The final volume occupied by the bed ranged from 83 percent in the com-
partment at the inlet to 93 percent of the total compartment volume in the
fifth compartment (Table 18). The mean limestone particle diameter varied
from a low value of 0.78 cm in Compartment 1 at the inlet to 0.98 cm in Com-
partment 5. The initial mean particle diameter was 0.97 cm in all compart-
ments.
The measured particle diameters indicate that very little dissolution
occurred in the fourth and fifth compartments. However, during the test
period there was apparently some consolidation of the stone in these com-
partments. The final volumes occupied by the limestone bed in the fourth
and fifth compartments were 92 and 93 percent, respectively, of the total
volume occupied when the compartments were filled.
112
-------
The calculated final porosity of the limestone bed in each of the compart-
ments is listed in the fourth column of Table 18. The final porosity, e',
was calculated using the initial porosity, e, the initial and final limestone
particle diameters, d and d', and the initial and final volumes filled by
the limestone bed, V and V. It was assumed that the number of limestone
particles in each compartment remained constant during the dissolution process,
i.e. ,
(1-e) V _ (1- e') v1
£(d>3
o
(49)
or
e1 = 1 - (1 - e) (V/V) (d'/d)3 (50)
The quantities V/V1 and (d'/d)3 were determined using the percent volume
occupied and mean particle diameter values listed in Table 18. For example,
in the case of Compartment 1, (V/V1) = 1/0.83 = 1.20, (d'/d)3 = (0.78/0.97)3
= 0.52, e = 0.44 and therefore, according to Eq. 50, e1 = 0.66.
The final porosity values for the first two compartments in the contactor
were significantly greater than the initial porosity of 0.44. This trend
suggests that while under these conditions the limestone bed tended to consol-
idate somewhat as the limestone particles became smaller (the bed volume
decreased in every compartment), the porosity increased in the first two
compartments. It is possible that the brown humous-like deposit which was
found in the compartments near the inlet tended to prevent the bed from col-
lapsing as the particles dissolved.
The amount of limestone dissolved from the first compartment of the
box contactor was estimated using the measurements listed in Table 18. This
value was compared with an amount determined using a mass balance calculation
and measured increases in the calcium ion concentration across the chamber.
According to the results listed in Table 18 the mass of limestone dis-
solved from the first compartment during the 455 day test period was approx-
imately 25 kg. This amount was determined using the following relationships:
113
-------
Table 18 Baffled-Box Contactor - Limestone Dissolution
June 23, 1982 - September 26, 1983
Percent of Total
Compartment Volume
Mean Limestone
Final Porosity
Compartment
Number
1
2
3
4
5
Occupied by the
Limestone Bed
83
84
87
92
93
Particle Diameter
(cm)2
0.784
0.794
0.95
0.97
0.98
of the
Limestone Bed-'
0.66
0.66
0.42
0.42
0.42
NOTES:
1. Original Volume Occupied was 100 percent in all compartments.
2. Original Mean Diameter was 0.95 cm.
3. Original Porosity was 0.44.
4. A sieve analysis of the stone in Compartments 1 and 2 yielded median
stone diameters of 0.71 and 0.84 cm., respectively.
114
-------
initial mass of
limestone in the = (compartment volume)(1-porosity)(limestone density)
compartment
(36,203 cm3)(l-0.44)(2.64 g/cm3)(l(T3 kg/g)
52 kg
final mass of ... , , . ,
limestone in the = finalb?O?S£Bg£*ied (1-porosity) (limestone density)
compartment
(36,203 cm3)(0.827)(l-0.66)(2.64 g/cm3)(10-3 kg/g)
27 kg
[Mass of Limestone Dissolved] = 52 - 27 = 25 kg.
In a special series of constant flowrate experiments using the expected
range of operating flowrates (1-4 gpm, 4-15 liters/min) the increase in calcium
ion concentration across the first compartment ranged from 3 mg Ca/L at 4
liters/min to 1 mg Ca/L at 15 liters/min. Since the limestone in this study
contained 79 percent CaCC>3 and CaC03 is 40 percent calcium (by mass) the
amount of limestone dissolved for a given increase in calcium ion concentra-
tion, Ca, and flowrate, Q, is given by:
Mass of Limestone = (ACa, g/L)(Q, L/min)(1440 min/day)(455 days) x
Dissolved (1Q3 kg/g)(3-13 kg iimestone/kg Ca)
For ACa = 0.003 g/L and Q = 4 L/min, the mass of limestone dissolved
during the 455 day period is 25 kg. For ACa = 0.001 g/L and Q = 15 L/min,
the mass dissolved is 31 kg. Given the assumptions upon which these calcu-
lations are based the agreement obtained between the mass dissolved calculated
by the two methods (25 kg and 25 to 31 kg) is not unreasonable.
A qualitative analysis was made of the surface chemical characteristics
of limestone samples using an ISI scanning electron microscope with a Kevex
x-ray energy spectrometer attachment. In this instrument the electron beam
is used to provide an image of the sample through electron scattering and
to cause characteristic x-rays to emanate from the surface layer of the sample.
The measurement of the energy of the characteristic x-rays indicates the
presence of certain elements and their relative abundance on the sample sur-
face.
115
-------
Four elements were detected and identified in one or more of the four
samples; calcium, silicon, aluminum and iron. The x-ray energy spectra for
the four samples are given in Figure 41, A-D.
Calcium was the dominant identifiable element in the fresh limestone
sample. A lesser peak, characteristic of silicon was also apparent. It
is possible that aluminum is present in the fresh limestone surface (Figure
41A), however, the peak which is characteristic of aluminum is partially
obscured by the silicon peak.
Prolonged contact of the limestone with the spring water in the box
contactor reduced the amount of calcium and increased the amount of silicon
and aluminum on the limestone particle surface (Figures 40B and 40C). A
small peak attributable to iron also appears after prolonged use in the con-
tactor. A comparison of the results obtained for samples from the first
and fifth compartments (Figures 41B and 41C) shows that the decrease in the
prominence of the calcium peak is much greater in the case of the first com-
partment where more of the limestone was dissolved.
The results obtained using the SEM/XES system suggest that as the calcium
carbonate was dissolved from the limestone particles the relative abundance
of aluminum and silicon on the surface increased. Apparently alumino-silicate
(clay) impurities in the limestone remained as a thin "residue" coating after
the CaCC>3 was leached from the limestone matrix.
There is no evidence that the source of the aluminum, silicon and iron
was the spring water. The x-ray energy spectrum obtained for the sample
from the laboratory column (treating a high purity acidified water; Figure
41D) was very similar to that obtained for the first compartment of the box
contactor unit (Figure 41A).
Bay Side Cottage Wound Fiberglass Column
In December 1983, the water line between the box contactor and the winter-
ized cottages became frozen and the resort owner had to pump water from Big
Moose Lake directly to the cottages. Column 1, a wound fiberglass, ion ex-
change type column, Figure 8, was installed in the heated basement of Bay
Side Cottage (Figure 9). The unit was operated during the months of January
to April 1984. The temperature of the water was 3 to 4°C during this period.
116
-------
liiji!1
B
.Figure 41. X-ray energy spectra for the following samples; A - fresh
limestone, B - limestone after prolonged dissolution in the
baffled-box contactor, compartment 1, C - same as B except
compartment 5, D - limestone after prolonged dissolution in
the laboratory.
117
-------
The results obtained from the wound fiberglass column are plotted in
Figures 42 to 45. In these figures influent means water drawn from a sample
tap located near the inlet to the unit and effluent means a sample taken
from the cold water tap in the kitchen of Bay Side Cottage.
The pH results are plotted in Figure 42. The wound fiberglass column
increased the pH from 4.6 to approximately 7.0. Some of the variability
which was evident in the effluent results is probably due to our inability
to maintain a constant flowrate through the unit during routine operation.
Inadvertent uptake of CC>2 by some of the water samples may also have affected
the results.
A plot of influent and effluent calcium concentrations for the 86 day
sampling period is presented in Figure 43. The average increase in calcium
ion concentration across the column was 5.5 mg Ca/L. On average the calcium
ion concentration increased form 1.8 to 7.3 mg Ca/L.
The alkalinity was increased from an average influent value of -0.03
meq/L (-1.5 mg CaC03/L) to an average influent value of 0.26 meq/L (13 mg
CaC03/L). The results are plotted in Figure 44.
The dissolved inorganic carbon concentration showed significant vari-
ability, probably due to the uptake or (in the case of the influent samples),
release of carbon dioxide. The average influent DIG concentration was 1.0
mg C/L and the average effluent concentration was 3.2 mg C/L. The DIG results
are plotted in Figure 45. The results obtained in monitoring the wound fiber-
glass column are summarized in Table 19.
The chemical equilibrium model (with the closed-to-the-atmosphere assump-
tion and using the average influent water chemistry (Table 19)) predicts
an equilibrium pH of 9.8 and an equilibrium calcium concentration of 9.1
mg Ca/L. It is apparent that the wound fiberglass unit did not produce an
effluent which was in equilibrium with the CaC03 in the limestone. The average
effluent calcium concentration was 7.3 mg Ca/L and the average effluent pH
was 6.9.
The low magnitude of the average effluent pH suggests that in this case,
as in the case of the box contactor unit, the effluent pH was depressed below
the value associated with a truly closed system, apparently through the uptake
of gaseous carbon dioxide.
118
-------
The average increase in alkalinity across the wound fiberglass column
was 0.30 meq/L (Table 19). This trend corresponds to a calculated average
increase in DIG of 1.8 rag C/L. The observed increase in DIG according to
Table 19 was 2.2 mg C/L, a value which was larger than the calculated increment
and, hence, in support of the assumption that the measured effluent pH values
were depressed by the uptake of carbon dioxide.
A sample of limestone was taken from the wound fiberglass unit at the
conclusion of the experiment in April. The chemistry of the surface layer
was analyzed, as in the box contactor case, using x-ray energy spectrometry.
The energy spectrum for one analysis is shown in Figure 46. The spectrum
is very similar to that obtained with fresh (undissolved) limestone, Figure
41A. Apparently the 3 month period of operation in the wound fiberglass
unit was insufficient to alter the elemental make-up of the surface layer
to a measureable extent.
Culligan (Cullneu) Contactor
Access to the basement of Henry Covey cottage was limited due to use
by tourists and therefore the influent and effluent of the Culligan contactor
were sampled infrequently. The results obtained are listed in Table 20.
The average increase in the calcium ion across the Culligan unit was
3.2 mg Ca/L and the average increase in the alkalinity was 240 ueq/L. The
average influent pH was 6.52 and the average effluent pH was 7.1.
The depth of Cullneu medium in the Culligan contactor was 15 in. (40
cm.). The mean particle size of the Cullneu material was approximately 2
mm (sieve analysis) and therefore it was estimated that the particle surface
area per unit volume of interstitial water (a) was approximately 50 cm"-'-.
This value is about 5 times greater than the value for the 0.97 cm diameter
limestone particles used in the box contactor and wound fiberglass units.
The higher specific surface area of the Cullneu medium explains why the incre-
mental change in water chemistry was as high as it was given that the depth
of the medium was only 15 inches (versus 48 inches in the wound fiberglass
unit).
The Culligan contactor was monitored for 9 months. During this period
there was no evidence of fouling or other operational problems. In the fall
119
-------
8
x
Q.
I I I I I I
o Effluent
a Influent
I I i
0 10 20 30 40 50 60 70 80 90 100
IIME (days)
Figure 42. Influent and effluent pH plotted as a function of time for the
wound-fiberglass contactor in Bayside Cottage.
120
-------
a
o
CO
i
h-
I
I
8
2
o
o
12
10
8
Effluent
Influent
10 20 30
4O 50 60
TIME (days)
70 80 90 100
Figure 43. Influent and effluent calcium concentration plotted as a function
of time for the wound-fiberglass contactor in Bayside Cottage.
121
-------
0.6
0.5
£ 0.4
I 0.3
"5
Or
-------
o
CD 4.
CE *
<
O
O
13
o
or
o
o
to
CO
Q
1 \ i T
o Effluent
^ Influent
Figure 45.
0 10 20 30 40 50 60 70 80 90 100
TIME (days)
Influent and effluent dissolved inorganic carbon concentration
plotted as a function of time for the wound-fiberglass contactor
in Bayside Cottage.
123
-------
Table 19 Summary of Bay Side Cottage
Wound Fiberglass Column Results
Parameter
pH
calcium
(mg Ca/L)
magnesium
(mg Mg/L)
alkalinity
(meq/L)
alkalinity
(mg CaCC>3/L)
dissolved
inorganic
carbon
(mg C/L)
Jan. 25, 1984 to April 20, 1984
Influent (Big Moose Lake) Effluent
Mean Std. Dev.
4.64
1.82
0.37
-0.03
-1.5
0.99
0.17
0.33
0.03
0.02
1.0
0.52
n
28
28
28
25
25
25
Mean Std. Dev.
6.93
7.27
0.49
0.26
13.0
2.33
0.32
1.37
0.11
0.07
3.5
0.85
n
27
25
26
23
23
25
124
-------
Si CA
Figure 46. X-ray energy spectrum for a limestone sample taken from the wound-
fiberglass contactor at the end of the e:q3eriment.
125
-------
TABLE 20 Culligan Contactor - Summary of Results
November 3, 1983 - July 31, 1984
Date Influent
Nov.
Dec.
Jan.
Feb.
June
July
pH Calcium
(mgCa/L)
3, 1984 6.76 9.95
30 6.42 7.64
25, 1984 6.25 8.64
10
13
31 6.64 11.40
Alkalinity
(ueq/L)
376 7
312 6
270 6
7
7
590 7
PH
.16
.79
.57
.53
.40
.08
Effluent
Calcium
MgCa/L
13.
10.
10.
11.
19.
15.
3
6
3
10
10
2
Alkalinity
eq/L
811
458
439
834
--
726
to
-------
of 1984 the unit was drained and a sample of medium was extracted for a sieve
analysis and for particle surface analysis by x-ray energy spectrometry (XES).
The sieve analysis result was not significantly different than that obtained
before the material was used. The results of the XES analysis are presented
in Figure 47.
The XES scan for unused Cullneu (Figure 47A) shows a significant calcium
peak, but unlike the fresh limestone, no evidence of silicon or aluminum.
The used Cullneu (Figure 47B) has XES peaks which indicate the presence of
aluminum, calcium and copper on the particle surface. It is not known whether
the aluminum came from the influent water or was a contaminate in the Cullneu
material. The copper peak suggests that copper released from several short
sections of copper tubing upstream of the contactor unit was adsorbed on
the medium.
EVALUATION OF THE CONTACTOR DESIGN EQUATIONS USING FIELD MEASUREMENTS
The contactor design equations (Section 6) were evaluated using data
obtained in experiments conducted in the field. The wound fiberglass and
box contactors were disconnected from the effluent piping and valves were
installed to control the flowrate. Samples of the influent and effluent
were collected after the units had been operating at constant flowrate for
at least fifteen minutes. In the case of the box contactor, effluent samples
were obtained for limestone depths of 39 and 78 cm. The depth of limestone
in the wound fiberglass unit was 122 cm. The experimental conditions and
the results are listed in Tables 21 to 23.
The overall dissolution rate constant, Ko, was calculated for each experi-
mental run using a simplified version of Eq. 25, Section 6, i.e.,
J> "ln[(CbL - Ceq)/(Cbo - Ceq)]Us
K° L a e
where C^o and C^L are the measured influent and effluent calcium concentra-
tions, Ceq is the model calculated equilibrium calcium concentration, Us
is the superficial velocity, L is the depth of limestone, a is the interfacial
area of limestone per unit volume of bed and e is the bed porosity. The
magnitude of a was determined using the limestone particle mean diameter,
d, and sphericity, fy , and Eq. 12 Section 6.
127
-------
B
AL CA.
Figure 47. X-ray energy speccra for fresh cullneu medium (A) and Cullneu used
in the Culligan contaccor for 9 months (3).
128
-------
Table 21 Baffled-Box and Wound Fiberglass Contactors
Special Test of Model Equations
Experimental Conditions and Results
INFLUENT CHARACTERISTICS
Wound Fiberglass Contactor
Big Moose Lake
4.7
1.8
1.0
PH, pH0
Calcium, Cbo(mg Ca/L)
Dissolved Inorganic Carbon,
DIC0 (mg C/L)
Temperature (°C)
Ca - Cc(moles/L) 1.1 x 10
CONTACTOR DESIGN AND OPERATING CONDITIONS
-4
Limestone diameter, d(cm)
Limestone particle
sphericity, ^
Bed Porosity,e
Superficial flow velocity,
Us(cm)
Bed depth, L(cm)
Kinematic viscosity, v(cm2/s)
Big Moose Lake
0.97
0.79
0.44
3.0, 12.3
122
1.62 x 10'2
MEASURED EFFLUENT CALCIUM CONCENTRATIONS
Big Moose Lake
Us = 3.0 cm/min
Us = 12.3 cm/min
Covewood Spring
Us = 5.4 cm/min
U0 = 10.8
Ug = 16.2
Uc
21.5
Box Contactor
Covewood Spring
6.4
4.0
3.6
10
5.1 x 10'5
Covewood Spring
0.97
0.79
0.44
5.4, 10.8,
16.2, 21.5, 26.9
39, 78
1.31 x 10"2
Calcium Concentration
(mg Ca/L)
Us = 26.9
L = 122 cm
8
6
L = 39 cm
7.0
6.6
5.0
5.7
5.8
.7
.9
L =
8
6
5
6
5
78 cm
.2
.7
.6
.3
.5
129
-------
TABLE 22 Special Test of Model Equations
Calculated Equilibrium pH and Calcium Concentration
MODEL CALCULATED EQUILIBRIUM (Closed-to-the-Atmosphere)CONCENTRATIONS
Big Moose Lake Covewood Spring
pH, pHeq 9.8 9.2
calcium concentration, 9.1 10.9
Ceq(mg Ca/L)
130
-------
TABLE 23 Results of Special Test
of Model Equations
BIG MOOSE LAKE RESULTS - WOUND FIBERGLASS CONTACTOR
Superficial Velocity
Us(cm/min)
K0 x 10-3(cm/min)
(from experimental results)
K0 x 10~3(cm/rain)
(from contactor design equations)
COVEWOOD SPRING RESULTS - BOX CONTACTOR
3.0
15
19
12.3
25
26
Superficial Velocity
Us(cm/min)
5.4 10.8 16.2 21.5 26.9
K0 x 103(cm/min) 17 23 14-31 38
(from experimental results)*
K0 x 103(cm/min) 31 36 41 47 54
(from contactor design equations)
*Average values for L = 39 and 78 cm.
131
-------
The use of Eq. 51 instead of Eq. 25, section 6, is appropriate because
in the case of these field experiments, the dispersion number, NQ, was small,
us us
and therefore ideal plug flow can be assumed.
The equilibrium calcium concentration, Cec,, was determined for a closed-
to-the-atmosphere condition using the influent characteristics listed in
Table 15 and the equations and thermodynamic constants discussed in Section
6 and Appendix A. The results, pH6q and Ceq, are given in Table 22. Only
the equilibrium calcium concentration, Ceq, is used in the calculation of
K0.
Model calculated values of Ko were determined for each set of experimental
conditions using Eqs . 38 to 40, Section 7. For example for modified Reynold's
numbers (MRe) less than 30, i.e.,
d Us
< 30,
v(l-e)
The magnitude of KL is given by Eq. 39,
KL = 5.70(MRe)-°-78 Us(v/D)-2/3 (41)
The calcium ion diffusivity determined in the laboratory experiments,
= 1.2 x 10~5cm2/5, was used with Eq. 42, Section 7 to estimate D at the in-
fluent water temperature.
To determine the overall dissolution rate constant, Ko, it was assumed
that the surface reaction rate constant, Kc, was significantly larger than
KL (as was observed in the laboratory experiments) and therefore Ko = KL-
The model calculated values of Ko are listed in Table 23 next to those deter-
mined using the experimental data.
The agreement between the values of Ko calculated using the experimental
results and those determined using the design equations is reasonable in
the case of the wound fiberglass unit treating water from Big Moose Lake.
In the case of the box contactor results the agreement is less satisfactory;
132
-------
the model calculated values of Ko are essentially two times the values derived
from the experimental results. An exact reason for this discrepancy has
not been determined, however, one possible explanation will be discussed.
It is possible that the presence of a microbial film on the surface
of the limestone reduced the interfacial area available for mass transfer
and consequently, reduced the overall dissolution rate. To explain the average
difference between the values of Ko, the interfacial area per unit volume
of interstitial water, a, would have to be reduced by a factor of about two.
As noted earlier, a limited examination of the surface with light and scanning
electron microscopes did not give a positive indication of significant bio-
logical fouling of the surfaces. However, the "sliminess" of the limestone
particles which was noted when the unit was opened after one year, and the
higher than expected dissolved inorganic carbon concentrations in the effluent,
suggest that there was significant biological activity within the box contac-
tor.
The box contactor and the wound fiberglass units had both been in oper-
ation about 3 months when the experiments were conducted to determine Ko.
A biological film may not have formed on the limestone in the wound fiberglass
unit because of the low temperature (3°C) of the lake water influent. The
box contactor was operating during the summer months (June to September)
with a water temperature of about 12°C immediately before the test was con-
ducted. This fact combined with the proximity of the unit to the soil may
have enhanced the formation of a microbiological film.
SENSITIVITY ANALYSIS - DESIGN EQUATIONS
The design equations (transport and chemical equilibrium) were used
to determine the effect of a number of physical and chemical parameters on
the depth of limestone required to achieve an effluent pH of 8.5. The "aver-
age" conditions above and below which each parameter was incremented (one
at a time) are listed below:
Influent water characteristics:
pH, PH0 5.5
Calcium Concentration, C^Q 3.0 mg Ca/L
Dissolved Inorganic Carbon 3.0 mg C/L
concentration, DICO
133
-------
Temperature 10°C
Contactor Design Parameters:
Superficial velocity 20.4 cm/min
Limestone particle size 0.96 cm
The effect of pHo, C^QJ °ICO and temperature on the magnitude of the
equilibrium calcium concentration, C6q, and the calcium concentration, C^L,
which corresponds to an effluent pH of 8.5 was determined for the average
and the high and low parameter values using the chemical equilibrium model
(Appendix A). The results are listed in Table 24. Items A and B in Table
24 were used in the following equations to calculate the required depth of
limestone. Line C, the effluent pH which would be obtained if the closed
system effluent (initially at pH = 8.5) was equilibrated with the atmosphere,
shows the significant effect that opening the effluent to the atmosphere
can have on the pH. In general as the effluent equilibrates with the atmo-
sphere the pH decreases from 8.5 to a value in the range 7.5 to 8.2.
The calculation of the limestone depth involved the following equation
from Section 5,
,
~ln
-"eq
C
°
Kn a L e
K0 a L £ 2
ND
(25)
It was assumed based on the laboratory experiments that
ND = 2(d /L), (17)
K0 = KL, (52)
and
D20°(calcium ion) = 1.2 x 10"^ cm^/s
D2Q° was corrected for temperature using Eq. 42.
The magnitude of KL was determined using the Chu and Khalil (1953) equa-
tions (Eqs. 37 to 40) and a, the interfacial area of limestone per unit volume
of interstitial water and e, the bed porosity, were assumed to be equal to
the measured values listed in Table 8.
The depth of limestone, L, was calculated by combining Eqs. 25, 17 and
52, i.e.,
-ln
_eq
- CbL
- cbo
(53)
. 2 d
134
-------
Table 24 Results of Chemical Equilibrium Model Calculations
A = calcium concentration at equilibrium, Ceq (mg Ca/L)
B = calcium concentration in contactor effluent when pH = 8.5 (mg Ca/L)
C = effluent pH if effluent was equilibrated with atmospheric CC-2
A
B
C
A
B
C
A
B
C
A
B
C
PH0
4.0
17.5
17.1
7.9
0
10.0
9.3
7.9
0.5
6.9
4.8
7.5
2
13.8
12.4
7.9
5.5
12.9
12.3
7.9
Cbo (mg Ca/L)
3.0
12.9
12.3
7.9
DIC0 (mg C/L)
3.0
12.3
12.3
7.9
Temperature °C
10
12.9
12.3
7.9
7.0
6.8
5.4
7.8
10.0
19.5
19.3
8.2
6.0
21.6
*
7.9
20
12.2
11.8
7.9
*PHeq < 8.5
135
-------
The results of the sensitivity analysis calculations are listed in Table
25. The depth of the limestone bed required to achieve a pH of 8.5 increases
with;
(1) decreasing influent pH
(2) increasing influent calcium concentration
(3) increasing influent dissolved inorganic carbon concentration
(4) increasing superficial velocity and
(5) increasing limestone particle size.
The effect of temperature on L is complex. With increasing temperature
between 2 and 10°C, L increases slightly and between 10 and 20°C it decreases.
This complexity is due to the opposing effects of temperature on the mass
transfer coefficient, KL, and the equilibrium and effluent calcium ion concen-
trations (Ceq and C^L)-
When the influent dissolved inorganic carbon concentration is increased
to 6.0 mg C/L, the equilibrium pH is 8.26 and the target pH of 8.5 can not
be reached.
Increasing the influent calcium ion concentration can have an effect
similar to that 'of increasing the influent DIG. For example, if DICO is
3.0 mg C/L and the calcium concentration is increased to values greater than
approximately 5 mg Ca/L the equilibrium pH becomes less than 8.5 and a con-
tactor which will meet the pH = 8.5 objective is infeasible. (See Figure
20, Section 6).
The effect of ionic strength of the influent on the depth of limestone
required to achieve an effluent pH of 8.5 was evaluated using the previously
described average conditions. The ionic strength was adjusted by assuming
that the background electrolyte (Ca and Cc, Eq. A-10, Appendix A and Eq.
26, Section 6) is NaCl. The results of the chemical equilibrium model cal-
culations are listed in Table 26. In general, as the amount of NaCl is increa-
sed the equilibrium and effluent (at pH = 8.5) calcium concentrations (Ceq
and C^L) increase. The ionic strength of the column effluent (pH = 8.5)
is also listed in Table 26.
136
-------
Table 25 Sensitivity Analysis Results
L is the Depth of Limestone in meters required to obtain an effluent
pH of 8.5.
PH, PH0
L(m)
Calcium, C^Q (mg Ca/L)
L(m)
Dissolved Inorganic Carbon
DIC0(mg C/L)
L(m)
Temperature (°C)
L(m)
Superficial Velocity, Us
(cm/min)
L(m)
Limestone Particle Size
d (cm)
L(m)
Low
Value
4.0
3.5
0
2.4
0.5
0.5
2
2.2
8.2
1.3
0.54
1.0
Average
Value
5
2
3
2
3
2
10
t
2
20
2
0
2
.5
.4
.0
.4
.0
.4
.4
.4
.4
.96
.4
High
Value
7.0
0.8
10.0
3.9
6.0
20
2.1
40.8
3.3
3.2
14.8
137
-------
The results listed in Table 26 were used with Eq. 53 to calculate the
required depth of limestone. It was assumed that the ionic strength does
hot affect the mass transfer coefficient, KL. The depth of limestone bed
is plotted as a function of the influent ionic strength in Figure 48. The
plotted results show that when the influent ionic strength is attributable
to a simple 1:1 electrolyte such as NaCl and is less than approximately 2
x 10~-^M the effect on the depth of limestone needed to reach an effluent
pH of 8.5 is negligible.
It was noted in Section 5 that the mean value of the dispersion number
based on the results of the tracer experiments was equal to 2 d/L. The upper
and lower limits on this quantity, based on the standard deviation of the
Peclet number, were approximately 3.3 d/L and 1.4 d/L, respectively. The
"average" conditions were used with the chemical equilibrium model and Eq.
53 to calculate the effect of the variability in the dispersion number on
the depth of limestone required to achieve an effluent pH of 8.5. The fol-
lowing results were obtained.
ND=0 L=2.2m
ND = 1.4 d/L L = 2.2 m
ND = 2.0 d/L L = 2.3 m
ND = 3.3 d/L L = 2.3 m
The variability in the dispersion number has an essentially negligible
effect on the depth of limestone required to reach pH = 8.5. For many cases,
particularly when the limestone particle diameter is less than 1 cm, it is
reasonable to assume that NQ = 0 and plug flow exists.
THERMODYNAMIC CALCULATIONS OF TRACE METAL CHEMISTRY
A series of thermodynamic calculations was performed with the chemical
equilibrium model, MINEQL, to evaluate the solubility of trace metals and
the stability of passivation films. Calculations were made over a range
of pH values and dissolved inorganic carbon concentration; the latter including
both constant dissolved inorganic carbon concentrations (closed-to-atmospheric
C02) and equilibrium (open) with gaseous C02- To facilitate an evaluation
of the conditions for which passivation films are stable, a series of predomin-
138
-------
Table 26 Effect of Ionic Strength on the Equilibrium and
Contactor Effluent (pH = 8.5) Calcium Concentrations
Influent Equilibrium Effluent Calcium
Ionic Calcium Concentration, Concentration when
Strength, M Ceq pH = 8.5
(mg Ca/L) CfcL mg Ca/L
8.6 x 10~4 12.9 12.4
1.9 x 10~3 13.0 14.4
5.9 x 10"3 13.1 12.5
1.1 x 10'2 13.4 12.5
5.1 x 10~2 15.0 . 12.8
139
-------
1
_l
*
fE 3.0
CL
UJ
Q
Q
LU
CD 2.0
UJ
O
H
2 1.0
o
i | i i 1 1 i ji t i I ' i i i i j t I I i i i l > i i i | i i m t
*s^
\L. x _
X
N
^^ ^ ^*m
1 ' 1 I tiill t 1 1 1 i till f 1 1 M Ml 1 i l 1 1 nil l
10 I0'4 I0"3 I0~2 IO"1
IONIC STRENGTH.1 (M)
Figure 48. Total depth of Limestone required to obtain an effluent pH of 8.5
plotted as a function of the ionic strength. The influent was
assumed to have the average conditions used in the sensitivity
analysis calculations.
140
-------
ance area diagrams were made. These diagrams were constructed over a range
of pH values and either various dissolved inorganic carbon concentrations
(closed to gasesous CC^) or partial pressures of CC>2 (open system).
As discussed previously, there is considerable uncertainty in the thermo-
dynamic solubility of Pb(OH)2- As a result, predominant area diagrams were
constructed for both proposed values of the solubility constant (log*Kso
= 8.15, Wagman et al., 1969; log*Kso = 13.07, Topelman, 1929). For constant
dissolved inorganic carbon systems (closed atmospheric), PbS04 is the stable
mineral phase under acidic (pH < 6) conditions with low dissolved inorganic
carbon concentrations (Figure 49 and 50). Note that the solubility of lead
increases substantially with decreases in pH. It is unlikely that equilibrium
with PbSC>4 would ever be obtained below pH 6. Therefore under these condi-
tions, concentrations of lead are likely controlled by dissolution kinetics.
In the neutral pH range PbCC>3 becomes the stable lead-controlling solid
phase at modest dissolved inorganic carbon concentrations (10~^ mol-l~l).
With increases in dissolved inorganic carbon concentrations, the pH range
over which PbC03 controls solubility increases. In the alkaline pH range,
the lead regulating solid phase is in doubt. If a Pb(OH)2 solubility constant
(log*Kso) of 13.07, is assumed then Pb3(OH)2(C03)2 is the thermodynamically
stable solid phase. This condition is similar to that proposed by Schock
(1980). However, if a Pb(OH)2 solubility constant of 8.15 (log*Kso) is assumed
then this lower solubility constant suggests that Pb(OH)2 is the thermodynam-
ically stable passivation film under high pH, low dissolved inorganic carbon
conditions. Unfortunately, because of this uncertainty in thermochemical
data it is impossible to evaluate the lead controlling solid phase by thermo-
dynamic calculations.
Both variations in pH and dissolved inorganic carbon concentrations
have a profound influence on the solubility of lead (Figure 51). Lead con-
centrations are highest under low pH conditions and generally decrease with
increasing pH. When a Pb(OH)2 log*Kso of 8.15 is assumed, the theoretical
solubility of Pb can be reduced below the MCL above the pH range 7.5 to 8.5
(Figure 51). Variations in dissolved inorganic carbon concentrations alter
the pH-dependent solubility trend of lead. Note that under acidic conditions
(pH <7), however, dissolved inorganic carbon acts to enhance lead solubility
through the formation of soluble lead carbonate complexes. i ,i i\ ~j\
141
-------
-2.8
-3.2
-3.6
•R -4.0
~ -4.4
o»
o
-4.8
-5.2
25°C
Pb(OH)2 logK=8.!5
Pb(OH);
PbS04
8
10
PH
Figure 49. Predominance area diagram for the stability of lead passivation
films over a range of pH and dissolved inorganic carbon con-
centrations at 25°C. A p*Kso of -8.15 for the solubility of
Pb(OH)n was assumed.
-------
LJ
•
o
"o
O
Q
-2.5
-3.0
-3.5
-4.0
-4.5
-5.0
-5.5
- PbSO,
25°C
Pb(OH)2 log«= 13.07
PbCO,
1
Pb3(OH)2(C03)2
7
pH
8
10
Figure 50. Predominance area diagram for the stability of lead passivation
films over a range of pH and dissolved inorganic carbon concen-
trations at 25°C. A p*Kso of -13.07 for the solubility of Pb(OH>2
was assumed.
-------
o I.OmgC/L(83uM)
• .fVOmgC/U250/iM)
a /p.OmgC/Ua30uM)
A 30.0mgC/U2500/iM)
PH
Figure 51. Lead concentrations calculated with the chemical equilibrium model
MINEQL as a function of pH for several concentrations of dissolved
inorganic carbon. The maximum contaminant level (MCL) for lead is
indicated.
-------
In systems that are in equilibrium with gaseous C02 the formation of
PbC(>3 is the predominant stable mineral phase under essentially all but very
acidic conditions (pH < 5.5; Figures 52 and 53). However, note that when
the solubility of Pb(OH)2 (log*Kso) is assumed to be 8.15, then Pb(OH)2 becomes
the solubility controlling mineral phase under low partial pressures of CC>2
Figure 52). Like under conditions of constant dissolved inorganic carbon
concentrations, variations in the partial pressure of C02 alter the pH de-
pendent solubility of lead (Figure 54). At pH values below 7 elevated partial
pressures of CC>2 serve to reduce lead concentrations through the formation
of PbCC>3, while at pH values above 7 increased partial pressure of C02 enhances
lead solubility through the formation of soluble lead carbonate complexes.
Note that even at atmospheric levels of CC>2 (10~^*^ atm), the solubility
of lead in equilibrium with gaseous CC>2 exceeds the MCL (Figure 54).
Variations in water chemistry also influence the stability of copper
passivation films (Figure 55). Under low pH and high dissolved inorganic
carbon concentrations Cu2(OH)2C03 is the thermodynamically stable passivation
.film, while under high pH, low dissolved inorganic carbon concentrations
Cu(OH)2 appears to regulate copper solubility. Variations in dissolved inor-
ganic carbon concentrations do not alter the solubility of copper to the
same extent as lead (Figure 54). At pH values below 7 to 8, increased con-
centrations of inorganic carbon, either under constant dissolved inorganic
carbon (closed atmosphere) or through the solubility of gaseous CC>2, act
to reduce copper concentrations by the formation of Cu2(OH)2CC>3. Unlike
lead, copper does not form strong carbonate complexes. Therefore copper
solubility at elevated pH is only enhanced at extremely-high carbonate con-
centrations associated with high partial pressures'of CC>2 (Figure 57). Note
that reductions in copper below the MCL can be accomplished by increasing
pH values above 7.
As discussed by Schock (1984), under most conditions the solubility
of zinc is regulated by Zn5(OH)6(C03)2 (Figure 58 and 59). Only with high
inorganic carbon concentrations can ZnCC>3 become the stable passivation film.
Trends in the pH-dependent solubility of Zn at different dissolved inorganic
carbon concentrations (Figure 60) or partial pressure of CC>2 (Figure 61)
are similar to copper. The solubility of zinc generally decreases with in-
145
-------
-1.5
25 °C
Pb(OH)2 logK=8.(5
-2.0
•«—
o
(VJ
o
o
Q.
-2.5
PbC03
-3.0
PbSO,
-3.5
Pb(OH)2
if
7
PH
8
10
Figure 52. Predominance area diagram for the stability of lead passivation
films over a range of pH and partial pressures of CO at 25°C.
A p*Kso of -8.15 for the solubility of Pb(OH)2 was assumed.
-------
-1.5
-2.0
CM
8-2.5
cn
o
-3.0
-3.5
25°C '
Pb(OH)2/ogK=l3.07
PbC03
PbSQ,
8
10
pH
Figure 53. Predominance area diagram for the stability of lead passivation
films over a range of pH and partial pressures of CO- at 25°C.
A p*Kso of -13.07 for the solubility of Pb(OH)2 was assumed.
-------
-3
-4
oo
o
.0
Q.
Cn
o
-6
MCL
n r
A jo~3 9 ofm
o IO-30
PH
8
10
Figure 54. Lead concentrations calculated with the chemical equilibrium model
MINEQL as a function of pH for several partial pressures of C0_.
Calculations are in equilibrium with gaseous C0?. The maximum
contaminant level (MCL) for lead is indicated.
-------
-2.6
-3.1
25°C
-J
\
o
"o
e
o
Q
o>
o
-3.6
-4.1
Cu2(OH)2C03
Cu(OH)2
-4.6
-5.1
I
6
7
PH
8
10
Figure 55. Predominance area diagram for the stability of copper passivation
films over ranges of pH and dissolved inorganic carbon concentra-
tions at 25°C.
-------
Ul
o
I
0
_o
o -3
E
3 -4
o>
-5
-6
-7
SMCL'
O.I mgC/U8.3^M)
0.3mgC/L(25/iM)
3.0
o IO.OmgC/U830/iM)
30.0mgC/l(250O/iM)
8
10
Figure 56. Copper concentrations calculated with the chemical equilibrium
model MINEQL as a function of pll for several concentrations of
dissolved inorganic carbon. The secondary maximum contaminant
level (SMCL) for copper is indicated.
-------
Figvire 57. Copper concantrations calculated with the chemical equilibrium
model HINEQL as a function of pH for several partial pressures
of CO-2- Equilibrium with gaseous CC>2 is assumed. The secondary
maximum contaminant level (SMCL) for copper is indicated.
-------
-2.6
-3.1
o -3.6
"o
o
Q -4.1
o>
o
-4.6
-.5.1
ZnCO-.
25°C
7
PH
8
10
Figure 58. Predominance area diagram for the stability of zinc passivation
films over a range of pH and dissolved inorganic carbon con-
centrations at 25°C.
-------
-1.5
ZnCO,
25°C
SMCL
-2.0
-2.5
-3.0
-3.51
8
pH
Figure 59. Zinc concentrations calculated with the chemical equilibrium model
MINEQL as a function of pH for several dissolved inorganic carbon
concentrations. The secondary maximum contaminant level (SMCL)
for zinc is indicated.
-------
3
2
I
0
N ~3
o>
- .4
-5
-6
SMCL
\\ 1
Q. I mgC/U8.3/zM)
p.3mgC/L(25uM)
'
olO
• 30 mgC/l-(2500/iM)
8
10
pH
Figure 60. Predominance area diagram for the stability of zinc passivation
films over ranges of pH and partial pressures of CO™ at 25°C.
-------
I
0
-I
^ -2
C
Nl
-4
-5
* IO"J 3 oJm
o JO'30 otm
• /0"2'a otm
o 10-20
• 10"'-*
SMCL
8
10
pH
Figure 61. Zinc concentrations calculated with the chemical equilibrium model
MINEQL as a function of pit for several partial pressures of CO'^.
Calculations are in equilibrium with gaseous CC>2. The secondary
maximum contaminant level (SMCL) for zinc is indicated.
-------
creasing pH. If pH values are above 7 to 8, zinc concentrations generally
fall below the MCL. Increases in dissolved inorganic carbon concentrations
(or partial pressure of CC>2) generally result in a decrease in the solubility
of Zn. Like copper, zinc forms relatively weak aqueous complexes with car-
bonate so the solubility of zinc is only enhanced when the carbonate concen-
trations are extremely high due to high pH values and elevated partial pressure
of CC>2.
PIPE LEACHING EXPERIMENTS
A series of experiments was conducted to evaluate the extent to which
calcium carbonate treatment could reduce the corrosivity of dilute acidic
waters. Aliquots of water obtained from the ports of the.laboratory limestone
column were sealed in one meter pipe sections for ten hours and analyzed
for trace metals. Results of lead (Figure 62) and zinc leaching experiments
(Figure 63) with lead and galvanized steel pipe sections, respectively, yielded
inconsistent results. Concentrations of both metals were highly variable
and demonstrated no systematic trends with the level of treatment of pH.
As mentioned previously, both lead and zinc may form non-adhering passivation
films. The extremely high scattered concentrations observed for these experi-
ments may be a reflection of this condition. The lead pipe experiments,
were repeated (Figure 64) and aliquots of leachate were analyzed for both
total and filtered (Filtration through 0.40 urn polycarbonate filter) lead.
In some samples considerable discrepancy was evident between total and filtered
lead concentrations. Note that very fine particulate lead, capable of passing
a 0.40 urn filter, may be released from lead pipe. It appears that the release
of particulate metal was a complicating factor in the lead and galvanized
steel pipe leaching experiments.
Additional leaching experiments were conducted by applying aliquots
of water from various stages of laboratory contactor treatment with sections
of copper pipe and copper pipe with lead-tin solder joints. To illustrate
results from these experiments measured water chemistry parameters as a func-
tion of limestone column depth (0.96 diameter particle) at a flow rate of
5 liters/minute are plotted in Figure 65. As discussed earlier with increasing
contact with calcium carbonate pH (Figure 65a) and dissolved inorganic carbon
156
-------
I\_/W
a
o. 8°
a.
CO
8 60
cc
h-
§ 40
§
O
Q
LU 20
0
1 1 1 I 1
— • —
o
0 0 0
0
o o o o ° o ~
O Oo°
° a ®
° o o ° %
0 0 0°° 0 (P CU
o 5> ° °o
O _ rt O « **»
O u x^ O O
0 • n °
0 0 _
0 °
0 0
1 1 1 II
456 7 S 9 1C
pH
Figure 62. Lead concentrations from lead pipe section leaching experiments.
-------
00
40
c
N
co
H
O
30
20
LU
O
O
O
O
N
10
0
O
O O
°o
O
o
6
0000
o
pH
8
D
Figure 63. Zinc concentrations from galvanized steel pipe section leaching
experiment.
-------
3
£
O>
O
h-
Q-
CONCENT!
Q
-------
(el
• Laboratory fltiuflt
O MlKOL CoeulartcJ
Volu*
234567890 II
DEPTH OP COLUMN (ft)
Figure 65. Variations in pH (a), dissolved inorganic carbon (DIG) concen-
trations (b), and measured copper (c) and lead (d) concentrations
from pipe section leaching experiments as a function of column
treatment by CaC03. The experimental conditions were 0.96 cm
diameter CaCC>3 and a flow rate of 5 liters/min. Calculated values
of copper (c) and lead (d) obtained from MINEQL calculations are
plotted for comparison. The secondary maximum contaminant level
(SMCL) for copper and the maximum contaminant level (MCL) for lead
are indicated.
160
-------
concentrations (Figure 65b) increased resulting in a pronounced decrease
in copper concentrations from pipe section leachates (Figure 65c). Unlike
copper, lead concentrations from lead-tin solder joints did not exhibit marked
variations (Figure 65d), although lead concentrations did decrease somewhat
with increased contactor treatment. Superimposed on the results of trace
metal leaching experiments (Figures 65c, 65d) are predicted concentrations
from measured water chemistry (e.g. pH: Figure 65a, dissolved inorganic carbon;
Figure 65b) using the chemical equilibrium model MINEQL. Note that generally
the MINEQL calculations followed measured copper concentrations (Figure 65c).
However, some discrepancy was evident under acidic conditions, associated
with minimum calcium carbonate treatment, and under the higher pH conditions
(pH 7.5 to 9.5), associated with greater contactor treatment. The former
may be attributed to non-equilibrium conditions. Acidic water chemistry
resulted in measured copper concentrations that were highly undersaturated
with respect to the solubility of Cu2(OH)2C03, probably due to insufficient
contact time with the copper pipe. Under these conditions, aqueous copper
concentrations were probably controlled by dissolution kinetics. There is
also an apparent deviation between measured predicted concentrations associated
with the higher level of treatment. Again laboratory results were highly
undersaturated with respect to anticipated mineral phase solubility. This
discrepancy is most likely due to one of two considerations. First, the
deviation coincides with the shift from Cu2(OH)2C03 to Cu(OH)2 passivation
films. Second, the predominant form of aqueous copper under these conditions
is Cu(OH)2(aq). Uncertainity in thermodynamic data of one or both of these
copper forms may be responsible for the deviation between measured and predic-
ted values.
There was poor agreement between measured values of lead from lead-tin
solder leaching experiments and values predicted from MINEQL (Figure 65d)
(MINEQL calculations were made by assuming log*Kso = 8.15). Note that under
low pH conditions lead concentrations were highly undersaturated with respect
to anticipated lead mineral phase solubility. These results are not surprising
given the relatively small contact area associated with lead-tin solder joints.
Results of all copper pipe and lead-tin solder leaching experiments
from limestone contactor treated water are summarized in Figure 66 and 67,
161
-------
M
1
— J
X
o
en
P
LW
to
p
§
1^—
2
UJ
O
2
0
o
o
c..v^ c
1.75
1.5
1.25
1
0.75
050
0.25
O O
1 '•! 1 i
. o LABORATORY OBSERVED VALUE
• MINEOL CALCULATED VALUE
o
_ o '•
o
o
o
o •
oo SMCL
•
0
0 0
o
0 °^ » •» MM«MMHM
O ° •
%• • *
o • V*
o o _
O O Q
o o OQ Q Q
1 1 °i on 1 o 8ffi?tetorJ^
UU4 5 6 7 8 9 ~ ~ 10
PH
Figure 66. Copper concentrations from copper pipe section leaching experiments
at various levels of CaCC>3 treatment (variations in plO. The
corresponding values of copper calculated with the chemical equili-
brium model MINEQL are indicated. The secondary maximum contaminant
level (SMCL) for copper is indicated.
-------
10
o>
I I0"
o:
h-
UJ in-2
O
O
UJ
ia4
• LA BORATOR YOB SERVED VAUJE
o MINEOL CALCULATED RESULTS
8
10
PH
Figure 67. Lead concentrations from copper pipe section with lead-tin solder
leaching experiments at various levels of laboratory CaCO_ treatment
(variations in pH). The corresponding values of lead calculated
with the chemical equilibrium model MINEQL are indicated.
-------
respectively. Again measured copper concentrations systematically increased
with decreasing pH. Measured results were qualitatively consistent with
MINEQL predictions however again a discrepancy was evident under low pH (pH
< 6; low CaCC-3 treatment) and under higher pH (pH > 7.5, greater CaCC>3 treat-
ment) conditions. As discussed previously unlike MINEQL predictions, measured
lead concentrations were relatively insensitive to changes in pH and calcium
carbonate treatment (Figure 67).
METAL RELEASE FROM FIELD SITE
Trace metal concentrations were monitored in inlet spring and lake water,
as well as tapwater from two cabins, Hillside and Bay Side (Figure 9). During
most of the study period, water to these cottages was obtained from the spring
and was treated with the box contactor. During certain times it was possible
to collect untreated tapwater at Covewood (Figure 9). Therefore, trace metal
concentrations from Covewood tapwater served as reference values for treated
spring tapwater at Hillside and Bay Side. During the spring 1984, lake water
was used as a water supply to Bay Side. This water was treated with a wound
fiberglass column contactor. During this latter study both treated and refer-
ence (untreated) samples were collected from Bay Side tapwater.
Spring Contactor Treatment
Copper, lead and zinc concentrations were elevated (significant at the
0.05 level; two tailed t-test) in tapwater samples relative to untreated
spring water (Table 27). The source of zinc was probably largely leaching
from two 400 gallon water storage tanks, made of galvanized steel, located
immediately down flow of the spring with the box contactor. Although fewer
samples were analyzed for concentrations of other trace metals, there was
no statistically significant evidence of leaching (or deposition) of cadmium,
manganese-iron or aluminum within the water distribution system relative
to the influent spring water (Table 27).
As reported in other studies (e.g. Meranger et al., 1984) trace metal
concentrations were highest with the "first-flush" after tapwater had been
in contact with the distribution system overnight. Copper and lead concen-
trations were significantly reduced (at the 0.05 level; two tailed t-test)
in tapwater which had been flushed for three minutes (Table 28).
164
-------
TABLE 27 Comparison of trace metal concentration (as mg/L) in spring water and
from the first flush of treated (Hillside, Bay Side) and untreated
(Covewood) cottages.
;r\
r\
Metal
Copper
Lead
Cadmium
Zinc
Manganese
Iron
Aluminum
n
19
13
6
8
8
10
10
Spring
mean + std. dev.
0
0
0
0
0
0
0
.0047 +
.0027 +
.0014 +
.025 +
.0044 +
.07 + 0
.064 +
0.0087
0.0043
0.0012
0.017
0.006
.11
0.074
• Hillside
n mean -t- std.
dev.
8 0.
7 0.
2 0.
2 <0
4 0.
4 0.
087 + 0.049
018 + 0.024
26
.001
11 + 0.15
016 + 0.014
n
14
13
4
6
6
7
7
Bay Side Covewood
mean + std. n mean + std.
dev. dev.
0
0
0
0
0
0
0
.030 + 0.37 4 1.9 + 0.31
.0084+; 0.0084 3 0.046 + 0.0040
.0010 + 0.0010
.26 + 0.18
.0043 + 0.005
.13 + 0.19 1 0.11
.018 + 0.018 1 0.056
-------
TABLE 28 Comparison of copper and lead concentrations (mean +_ std. dev. as mg/L)
from first flush and three minutes of flowing tapwater derived from the
box contactor treated spring.
Copper
Lead
Zinc
n
(7)*
(6)
(2)
Hillside
First Flush
0.091 + 0.052
0.021 + 0.025
0.26
Three Minutes
0.007 + 0.011
0.0074 + 0.0096
0.07
n
(10)*
(8)*
(6)*
Bayside
First Flush
0.32 + 0.38
0.0069 + 0.0084
0.26 + 0.16
Three Minutes
0.031 + 0.050
0.0049 + 0.0045
0.061 + 0.041
*Indicates three minute flowing samples were significantly lower than first flush samples at 0.05
level (two tailed t-test).
-------
Contactor treatment appeared to diminish the corrosivity of the spring
water (Table 27). Concentrations of copper and lead at both Hillside and
Bay Side were significantly lower (at the 0.05 level; two-tailed t-test)
than the reference tapwater at Covewood. The apparent decrease in both copper
and lead solubility may be attributed to the increase in pH and dissolved
inorganic carbon concentrations associated with CaC03 treatment (Table 17).
These trends are consistent with the theoretical solubility of copper and
lead passivations films discussed previously (e.g Figures 56 and 51, respec-
tively) .
Substantial variation was evident in trace metal concentrations. To
illustrate this variability the probability distribution of copper and lead
concentrations in first-flush tapwater were plotted for both treated and
untreated springwater (Figures 68 and 69, respectively). The probability
of treated springwater exceeding the secondary MCI for copper was low (^ 4%),
particularly in comparison with the untreated spring supply. However, note
that relatively few observations were available for untreated first-flush
.tapwater.
Our results suggest that the probability of first-flush tapwater, derived
from the treated spring, exceeding the MCL for lead was extremely low (Figure
69). (None of the first-flush tapwater samples from the treated spring supply
exceeded 0.05 mg Pb/L). Note however, that some of our untreated spring
tapwater samples (2 out of 3 collected) exceeded th MCL for lead.
The variability of trace metal concentrations in this field phase of
study is not surprising. Cottages were generally in use during sample col-
lection. While we attempted to collect tap water samples in the early morning
to obtain maximum metal concentration from an overnight leaching period,
this controlled collection was not always possible. Moreover, lead in tapwater
samples was largely derived from lead-tin solder joints. Given that two
joints are present on the average at every 2 m of pipe, it is not surprising
that the concentrations were so variable.
Lake Contactor Treatment
Both treated and untreated tapwater at Bay Side were also greatly enriched
in both copper and lead concentrations relative to the lake water supply
167
-------
J
H
O
UJ
O
2
O
O
cr
UJ
o_
Q.
O
O
3.0
1.0
0.3
OJ
0.03
0.0 f
O Untreated Lake Tapvvater
C> Treated Lake
• Untreated Spring
O Treated Spring
Secondary
MCL
10 20304050607080 90
PERCENT EXCEEDANCE
Figure 68. The probability of copper concentrations in untreated and CaC03
treated lake and spring waters exceeding a given concentration.
The secondary maximum contaminant level (SMCL) for copper is
indicated.
168
-------
-Q
a
O
O
Ld
0.1
0.05
003
^ 0.01
o
< acos
K
§ O.OC3
o
0001
0.005
O.C03
I i i i r
_O Untreated Lake Tapwater
A Treated Lake
• Untreated Spring
o Treated Spring
Detection Limit
i i
10 20 30 40506070 80 90
PERCENT EXCEEDANCE
Figure 69. The probabilicy of lead concentrations in untreated and CaCO_
treated lake and spring waters exceeding a given concentration.
The maximum contaminant level (MCL) for lead is indicated.
169
-------
(Table 29). However, again tapwater treated with the limestone contactor
were significantly lower (at the 0.05 level; two tailed t-test) than untreated
lakewater samples, for both copper and lead. As we observed from the spring
water supply, metal concentrations were generally greatly reduced in tapwater
that had been flowing for three minutes relative to the first flush (Table
30). The exception to this trend was that no statistically significant differ-
ence in lead concentrations were evident between first flush and three minute
flowing samples in the treatment cottage.
To evaluate the applicability of chemical equilibrium modeling to the
field observations, theoretical concentrations of both copper and lead were
calculated using measured water chemistry (e.g. pH, DIG) with the equilibrium
model MINEQL. These calculations were then compared to first-flush tapwater
concentrations obtained from the lake contactor equipment. Results from
the lake contactor experiment were better suited to evaluate the chemical
equilibrium model than the spring contactor experiments because sample col-
lection was conducted under more controlled conditions. We have confidence
that the first-flush tapwater collected from this phase of the study had
been in contact with the cottage piping system for a prolonged period of
time (.e.g overnight) and therefore may be used to depict equilibrium con-
ditions .
Untreated lakewater samples were highly acidic (pH = 4.60 +_ 0.1) and
therefore it is not surprising that measured concentrations of both copper
and lead were highly undersaturated with respect to mineral phase solubility.
(The theoretical solubility of copper and lead for untreated lake water were
370 +_ 1500 mg Cu/L and 69 + 1.0 mg Pb/L, respectively). These results are
similar to the metal pipe experiments discussed previously and suggest that
under highly acidic conditions metal concentrations were regulated by kinetics
rather than equilibrium solubility.
Copper concentrations of treated lakewater, however, were qualitatively
consistent with MINEQL calculations (Figures 70 and 71). Although measured
and calculated copper concentrations were similar for the treated lakewater,
there were considerable scatter in the measured values. Unlike MINEQL pre-
dictions, measured first-flush copper concentrations did not demonstrate
a systematic increase in copper concentration with decrease in pH over the
pH range measured (pH 6.3 to 7.5).
170
-------
TABLE 29 Metal Concentrations (as mg/L) in lake influent, untreated and treated
first flush tapwater at Bay Side
Lakewater Untreated Treated
n mean + std. dev. n mean + std. dev. n mean + std. dev.
Copper 27 <0.0005 3 1.9 + 0.35 25 0.54 + 0.30
Lead 27 0.0034 + 0.0049 3 0.033 + 0.009 26 0.015 + 0.014
171
-------
Table 30 Comparison of copper and lead concentrations (mean +_ std. dev. as mg/L)
from first flush and after three minutes of flowing tapwater derived
from both CaCC>3 treated and untreated lakewater.
Untreated Treated
n First Flush Three Minutes n First Flush Three Minutes
Copper 3* 1.92 + 0.35 0.082 + 0.072 5* 0.63 + 0.099 0.027 + 0.040
Lead 3* 0.033 + 0.009 0.013 + 0.007 5 0.005 + 0.0009 0.011 + 0.012
*Indicates three minute flowing samples were significantly lower than first
flush samples at 0.05 level (two tailed t-test).
172
-------
3
V.
O
C«
2
0
i
f£
2
UJ
2
O
O
cr
LU
CL
Q.
o
o
1.75
1.50
1.25
1.0
0.75
0.50
0.25
0.0
iiill
• Field Observed Value
o MINEQL Calculated Value
— —
a
D SMCL
y •
• •
• * •* •••
•• •• ••
& .
LIT 0
Rjpj '
D Dt4bDD
1 1 • l" t 1
4 5 6 7 8 9 1C
pH
Figure 70. A comparison of measured copper concentrations from first flush
tapwater derived from CaC03 treated lakewater and calculated values
from the chemical equilibrium model HINEQ1 as a function of p».
The secondary maximum contaminant level (SMCL) for copper is
indicated.
-------
X
CJ
0»
150
1.25
1.0
£
8: 0.75
o
o
Q
0.50
0.25
0
0
o
.25 50 .75 1.0
CALCULATED COPPER (mgCu/LJ
o
1.25
150
Figure 71. A comparison of measured copper concentrations from first flush
tapwater derived from CaCO treated lakewater and calculated values
from the chemical equilibrium model MINEQL. One-to-one line is
indicated.
-------
As observed in laboratory pipe section experiments, measured lead concentra-
tions were highly undersaturated with respect to the solubility of anticipated
lead passivation films (Figure 72). Again these results are not surprising
in view of the fact that tapwater lead was largely derived from lead-tin
solder on joints.
As we reported for tapwater derived from spring supplies, the probability
of copper exceeding the SMCL was high for untreated lakewater while it was
low for treated lakewater (Figure 68). It is also evident that concentrations
of copper at Bay Side tapwater were generally higher for treated lake water
than treated spring water (significant at the 0.1 level; two tailed t-test).
Although the reason for this discrepancy is not clear, differences in the
level of treatment of the two water supplies may have contributed. The in-
fluent lakewater was considerably more acidic and had a lower dissolved inor-
ganic carbon concentration than the spring water (Tables 17 and 18). These
influent chemical characteristics coupled with the longer path length of
the spring box contactor resulted in higher pH,. alkalinity and dissolved
.inorganic carbon concentrations in the spring treated water relative to lake
treated water (Tables 17 and 18). As mentioned previously, both increased
pH and dissolved inorganic carbon concentrations theoretically result in
lower copper solubility and may have contributed to the apparent difference
(Figure 56).
First-flush tapwater concentrations of lead in treated lake water were
always below the MCL for lead (Figure 69). Likewise, untreated tapwater
was also below the MCL for all observations. However, relatively few untreated
samples were collected and concentrations were generally close to the 0.05
mg Pb/L standard.
175
-------
1 10°
j=
2
<
cc
K
2 _2
2
0
U
§ io-3
LU
-J
JO'4
1 1 1 1 1
• Field Observed Value
a a MINEQL Calculated Results
^HL
o
D
" MCL
• "
• 0
' "Jl
' " » "
— —
1 1 1 I 1
4 5 6 7 89 1C
PH
Figure 72. A comparison of measured lead concentrations from first flush tai
water derived from CaCCL treatment and calculated values from the
chemical equilibrium model MINEQL as a function of pH. The maximum
contaminant level (MCL) for lead is indicated.
-------
REFERENCES
Anderson, R. and D. Berry. 1981. Regulating Corrosive water. Wat. Resour.
Res., 17:1571-1577.
ASTM, 1972. Manual on Test Sieving Methods, ASTM-STP447A, Philadelphia, Pa.
Baes, C. F. and R. E. Mesmer. 1976. The Hydrolysis of Cations. John Wiley,
NY. 458 p.
Beevers, D. G., E. Erskine, and M. Robertson. 1976. Blood Lead and hyper-
Tension. The Lancet 2:1-3.
Ball, J. W., D. K. Nordstrom and E. A. Jenne. 1980. Additional and Revised
Thermochemical Data and Computer Code for WATEQ2-A Computerized Chemical Model
for Trace and Major Element Speciation and Mineral Equilibria of Natural Waters,
Resour. Invest., WRI 78-116, U. S. Geological Survey.
Barr, A. J., J. H. Goodnight, J. P. Sail and J. T. Helwis. 1976. A User's
Guide to SAS76. SAS Institute, Raleigh, NC.
Barton, P., T. Vatanatham. 1975. Kinetics of Limestone Neutralization of
Acid Waters. Environ. Sci. and Tech., 10:262-266.
Berner, R. A., J. W. Morse, 1974. Dissolution Kinetics of Calcium Carbonate
in Sea Water: Theory of Calcite Dissolution. Am. J. Sci., 274:108-134.
Bjerle, I., G. Rochelle. 1982. Limestone Dissolution in Acid Lakes. Vatten,
38:156-163.
Blackwell, W. W. 1984. Chemical Process Design on a Programmable Calculator.
McGraw-Hill Inc., pp. 370-371.
Boynton, R. S. 1980. Chemistry and Technology of Lime and Limestone. Second
Edition, John Wiley and Sons.
Cameron, W. R. and R. C. Wunderlich. 1976. Trace Element Toxicity Associated
with a Public Water Supply. In: D. D. Hemphill (ed.). Trace Substances
in Environmental Health-X. Columbia, MO.
Campbell, B. C., A. D. Beattie, M. R. Moore, A. Goldbert and A. G. Reid.
1977. Renal Insufficiency Associated with Excessive Lead Exposure. British
Medical J. 1:482-485.
177
-------
Chu, I., J. Khalil. 1953. Mass Transfer in Fluidized Beds. Chem. Eng. Prog.,
49:141-149.
Dangel, R. A. 1976. Study of Corrosion Products in the Seattle Water Department
Tolt Distribution System. USEPA Rep. EPA-67012-75-036.
Doull, J., C. D. Klaassen anfd M. G. Andur (eds.). 1980. Toxicology: The
Basic Science of Poisons. 2nd ed. Macmillan, NY 778 p.
Driscoll, C. T., J. White, G. Schafran, J. Rendall. 1982. CaCO- Neutralization
of Acidified Surface Waters. J. of ASCE-Environ. Eng. Div., 108:NO. EE6.
Eden, G. E., G. A. Truesdale. 1950. Treatment of Waste Water from the Pickling
of Steel. J. of Iron and Steel Institute, pp 281.
Edwards, M. F., and J. F. Richardson, 1968. Gas Dispersion in Packed Beds.
Chem. Eng. Sci., 23:109.
Erga, 0., S. G. Terjesen. 1956. Kinetics of the Heterogeneous Reaction of
Calcium Bicarbonate Formation. Acta Chem. Scand., 10:872-874.
Faust, S. D., and 0. M. Aly. 1981. Chemistry of Natural Waters. Ann Arbor
Science. 400 p.
F.eithnecht, W. and P. Schindler. 1963. Solubility Constants of Metal Oxides,
Metal Hydroxides and Metal Hydroxide Salts in Aqueous Solution. Pure and
Applied Chemistry 6:134-197.
Frear, G. L., and J. Johnson. 1929. The Solubility of CaCO in Certain Aqueous
Solutions at 25 C. J. Am. Chem. Soc., 51:2082-2093.
Galloway, R. E., and J. F. Colville. 1970. Treatment of Spent Pickling Plant
Liquors. Management of Water in the Iron and Steel Industry. Iron and Steel
Institute Publication.
Gehm, H. W. 1944. Neutralization of Acid Waste Waters with an Up-Flow Expanded
Limestone Bed. Sewage Work Journal, 16:104-120.
Gortikova, V. M., and L. I. Panteeva. 1937. Kinetics of Solvation of Calcium
Carbonate. Journal of General Chemistry-USSR, 7:56-64.
Gran, G.. 1952. Determination of the Equivalence Point in Potentiometric
Titrations. Intern. Cong. Anal. Chem., 77:661-671.
Haddad, M. 1983. Neutralization of Corrosive Waters Using a Packed Bed of
Crushed Limestone. M.S. Thesis, Syracuse University.
Hem, J. D. and W. H. Durum. 1973. Solubility and Occurrence of Lead in Sur-
face Water. J. Am. Water Works Assoc., 65:562-568.
Hem, J. D. 1976. Geochemical Controls on Lead Concentrations in Stream Water
and Sediments. Geochim. Cosmochim. Acta. 40:599-609.
178
-------
Herrera, C. E., J. F. Ferguson and M. M. Benjamin. 1982. Evaluating the
Potential for Contaminating Drinking Water from the Corrosion of Tin-Antimony
Solder. J. Am. Water Works Assoc., 74:368-375.
Hilburn, R. D. 1983. Modelling Copper Corrosion in Water with Low Conductivity
by Using Electrochemical Techniques. J. Am. Water Works Assoc. 75:149-154.
Hoak, R. D., C. J. Lewis, and W. W. Hodge. 1945. Treatment of Spent Pickling
Liquors with Limestone and Lime. Ind. and Eng. Chem., 37:553-559.
Hoak, R. D., C. J. Lewis, C. J. Sindlinger, and B. Klein. 1947. Lime Treatment
of Waste Pickle Liquor. Ind. and Eng. Chem., 39: No. 2.
Hoak, R. D., C. J. Lewis, W. W. Hodge. 1944. Basicity Factors of Limestone
and Lime - Evaluation as Neutralizing Agents. Ind. and Eng. Chem., 36:
No. 3.
Hudson, H. E, and F. W. Gilcreas. 1976. Health and Economic Aspects of Water
Hardness and Corrosiveness. J. Am. Water Works Assoc., 68:201-204.
Jacobs, H. L. 1947. Acid Neutralization. Chem. Eng. Progress, 43:247-254.
Jarrett, R. E. 1966. Porous Limestone Barriers for Neutralization of Acid
Streams. M. S. Thesis, Pennsylvania State University at University Park.
Karalekas, P. C., C. R. Ryan and F. B. Taylor. 1983. Control of Lead, Copper
and Iron Pipe Corrosion in Boston. J. Am. Water Works Assoc., 75:92-94.
Kaye, C. A. 1957. The Effect of Solvent Motion on Limestone Solution. J.
Geol., 65:34-36.
King, C. V., and C. L. Liu. 1933. The Rate of Solution of Marble in Dilute
Acid. J. Am. Chem. Soc., 55:1928-1940.
Levenspiel, 0., and W. K. Smith. 1957. Notes on the Diffusion-Type Model
for the Longitudinal Mixing of Fluids in Flow. Chem. Eng. Sci., 7:227-233.
Levenspiel, 0. 1972. Chemical Reaction Engineering. Wiley, NY.
Levenspiel, 0., and K. B. Bishchoff. 1963. Patterns of Flow in Chemical
Process Vessels. Advances in Chemical Engineering, 4:95-198.
Lund, K. H. C. Folger, C; C. McCune, and J. W. Ault. 1975. The Dissolution
of Calcite in Hydrochloric Acid. Chem. Eng. Sci. 30:825-835.
Maessen, 0., B. Freedman and R. McCurdy. 1985. Metal Mobilization in Home
Well Water Systems in Nova Scotia. J. Am. Water Works Assoc., 77:73-80.
Meranger, J. C., T. R. Khan, C. Vario, R. Jackson, and W. C. Li. 1983. Lake
Water Acidity and the Quality of Pumped Cottage Water in Selected Areas of
Northern Ontario. Intern. J. Environ. Anal. Chem., 15:185-212.
179
-------
Mesmer, R. E. and C. F. Baes. 1974. The Hydrolysis of Cations: A Critical
Review of Hydrolytic Species and their Stability Constants in Aqueous Solution.
Oak Ridge National Laboratory, ORNL-NSF-EATC-3. Part III.
Miadokova, M. and H. Bednarova. 1968. Einfluss der lonen und Organischer
Molekule auf die Kinetic der Kalcitauflosung in der Bronwasserstoffsaure.
Acta Fac. Rerum., Nat. Univ. comenianae - Chimia, 12:35-44.
Mihok, E. A., et al. 1968. Mine Water Research - the Limestone Nautralization
Process. Report No,. 7191, U. S. Bureau of Mines, Pittsburgh, PA.
Morse, J. W. and R. A. Berner, 1972. Dissolution Kinetics of Calcium Carbonate
in Sea Water II: A Kinetic Origin for the Lysocline. Am. J. Sci., 272:840-851.
Morse, J. W. 1974. Dissolution Kinetics of Calcium Carbonate in Sea Water
III: A New Method for CaCO. Reaction Kinetics. Am. J. Sci., 274:97-107.
Morse, J. W. 1978. Dissolution Kinetics of Calcium Carbonate in Sea Water
IV: The Near Equilibrium Dissolution Kinetics of CaC03-Rich Deep Sea Sediments.
Am. J. Sci., 278:344-353.
Morse, J. W. 1974. Dissolution Kinetics of Calcium Carbonate in Sea Water
V: Effects of Natural Inhibitors and the Position of the Chemical Lysocline.
Am. J. Sci., 274:638-647.
Murray, J. A., et al. 1954. Shrinkage of High Calcium Limestones During Burn-
ing. J. A. Ceram. Soc., 37:323-328.
National Academy of Science. 1977. Drinking Water and Health. Report of
the National Research Council. Washington, DC. 939 p.
Nestaas, I., and S. J. Terjessen. 1969. The Inhibiting Effect of Scandium
Ions upon the Dissolution of Calcium Carbonate. Acta. Chem. Scand., 23:2519-
2531.
Nierode, D. E. , and B. B. Williams. 1971. Characteristics of Acid Reaction
in Limestone Formation. Soc. Petrol. Journal, Trans. 251:406-418.
Patterson, J. W., H. E. Allen and J. J. Scala. 1977. Carbonate Precipitation
for Heavy Metal Pollutants. J. Water Pollut. Con. Fed., 49:2397-2410.
Patterson, J. W. , and J. O'Brien. 1979. Control of Lead Corrosion. J.
Am. Water Works Assoc., 71:264-271.
Pearson, F. H., and A. J. McDonnel. 1975. Limestone Barriers to Neutralize
Acidic Streams. ASCE J. of Environ. Eng. Div., 101: No. EE3.
Pearson, F. H., and A. J. McDonnel. 1975. Use of Crushed Limestone to Neu-
tralize Acid Wastes. J. Environ. Eng. Div. ASCE, 101:No. EE1.
Plummer, L. N., T. M. L. Wigley, and D. L. Parkhurst. 1978. The Kinetics
of Calcite Dissolution in CO Water Systems at 5 to 60 C and 0 to 1.0 atm.
C02. Am. J. Sci., 278:179-216.
180
-------
Plummer, L. N. , and T. M. L. Wigley. 1976. The Dissolution of Calcite in
CO Saturated Solutions at 25 C and 1 Atm. Total Pressure. Geochemica et
Cosmochemica Acta, 40:191-202.
Plummer, L. N., and E. Bussenberg. 1982. The Solubility of Calcite, Arogonite,
and Veterite in CO -HO solutions between 0 and 90 C and an Evaluation of
the Aqueous Model for the System CaCO.,-CO -H 0. Geochemica et Cosmochemica
Acta, 46:1011-1040. J
Quintin, M. 1937. Sur 1'hydrolyse du benzene sulfonate de cuivre. Compt.
Rend., 204:968.
Reidl, A. L. 1947. Limestone Used to Neutralize Acid Waste. Chem. Engineering,
100-101.
Rickard, D., and E. L. Sjoberg. 1983. Mixed Kinetic Control of Calcite Dis-
solution Rates. Am. J. Sci., 283:815-830.
Roberts, P. V., P. Cornel, and R. Scott. 1985. External Mass-Transfer Rate
in Fixed-Bed Adsorption. Journal of Environmental Engineering, ASCE, 111:891-905.
Robinson, R. A., and R. H. Stokes. 1959. Electrolyte Solutions. Butterworth,
London.
SAS User's Guide-Statistics. 1982. SAS Institute Inc., Gary, NC.
Schock, M. R. 1980. Response of Lead Solubility to Dissolved Carbonate
in Drinking Water. J. Am. Water Works Assoc., 73:695-704.
Schock, M. R. 1984. Treatment of Water Quality Adjustment to Attain MCL's
in Metallic Potable Water Plumbing Systems. Proceedings: Seminars on Plumbing
Materials and Drinking Water Quality, USEPA, Environmental Research Center,
Cincinnati, OH. May 16-17.
Schock, M. R. and M. C. Gardels. 1983. Plumbosolvency Reduction by High
pH and Low Carbonate-solubility Relationships, J. Am. Water Works Assoc.
75:87-91.
Sillen, L. G. and A. E. Martell. 1964. Stability Constants of Metal-ion
Complexes. The Chemical Society Special Publication 17, London. 754 p.
Sjoberg, E. L. 1976. A Fundamental Equation for Calcite Dissolution Kinetics.
Geochemica et Cosmochemica Acta, 40:441-447.
Sjoberg, E. L., and D. Rickard. 1983. The Influence of Experimental Design
on the Rate of Calcite Dissolution. Geochemica et Cosmochemica Acta, 47:
2281-2285.
Sjoberg, E. L., and D. Rickard. 1984. Calcite Dissolution Kinetics: Surface
Speciation and the Origin of Variable pH Dependence. Chem. Geol., 42:119-136.
Sjoberg, E. L. and D. Rickard. 1984. Temperature Dependence of Calcite Dis-
solution Kinetics between 1 and 62 C at pH 2.7 to 8.4 in Aqueous Solutions.
Geochemica et Cosmochimica Acta, 48:485-493.
181
-------
Slavin, W. 1968. Atomic Absorption Spectroscopy. John Wiley Interscience,
New York.
Small, H., T. S. Stevens and W. C. Bauman. 1975. Novel Ion Exchange Chromato-
graphic Method Using Conductimetric Detection. Anal. Chem. 47:1801-1809.
Smith, R. M. and A. E. Kartell. 1976. Critical Stability Constants
Vol. 4. Inorganic Complexes. Plenum Press, NY. 257 p.
Snoeyink, N., and D. Jenkins. 1980. Water Chemistry. John Wiley and Sons,
NY.
Spivakovskii, V. B. and G. V. Makovskaya. 1968. Copper Hydroxide Chlorides,
Hydroxide and Hydroxo-complexes. A New Version of the Method of Three Vari-
ables. Russ. J. Inorg. Chem., 13:815.
Stainton, M. P. 1973. A Syringe Gas-stripping Procedure for Gas-chromato-
graphy Determination of Dissolved Inorganic and Organic Carbon in Freshwater
and Carbonate in Sediments. J. Fish. Res. Bd., Canada, 30:1441-1445 .
Standard Methods for the Examination of Water and Wastewater. 1985. 16th
ed. American Public Health Association, Washington, DC.
Stutnm, W. and J. J. Morgan. 1981. Aquatic Chemistry, Wiley-Interscience,
New York, NY.
Sverdrup, H., and I. Bjerle. 1982. The Calcite Utilization Efficiency and
the Long Term Effect on Alkalinity in Several Swedish Liming Projects. Lund
Inst. Publication No. S-220.
Terjesen, S. G., 0. Erga, G. Thorsen, and A. Ve. 1961. Phase Boundary Processes
as Rate Determining Steps in Reactions Between Solids and Liquids. Chem.
Eng. Sci., 14:277-289.
Topelmann, H. 1929. J. Prakt. Chem. 121:320.
Treweek, G. P., J. Glicker, B. Chow and M. Sprinkler. 1985. Pilot-plant
Simulation of Corrosion in Domestic Pipe Materials. J. Am. Water Works Assoc.,
77:74-82.
USEPA. 1983. Methods for Chemical Analysis of Water and Wastes. Environmental
Monitoring and Support Laboratory, Cincinnati, OH, EPA-600/4-79-02G, 350.1-1
to 350.1-6.
Vaceta, J. 1976. Adsorption of Lead and Copper on Quartz from Aqueous
Solutions: Influence of pH Ionic Strength and Complexing Ligands. PhD thesis
California Institute of Technology. 206 p.
Vaillencourt, G. W. 1981. Crushed Limestone Neutralization of Dilute Acidified
Adirondack Surface Waters. M. S. Thesis, Cornell University.
182
-------
Vatanatham, T. 1975. Kinetics of Limestone Neutralization of Mine Drainage.
M.S. Thesis, Penn. State University, University Park, PA.
Volpocili, V. , C. L. Santoro, and P. Ciambeli. 1981. Neutralization of
Acid Water Containing Aluminum Ions by Means of Limestone. Chem. Eng. Journal,
21:29-38.
Wagman, D. D., W. H. Evans, V. B. Parker, I. Halow, S. M. Bailey, and R. H.
Schumm. 1968. Selected Values of Chemical Thermodynamic Properties. National
Bureau of Standards Technical Note 270-3 to 270-6.
Waldbott, G. L. 1978. Health Effects of Environmental Pollutants. The C.
V. Mosby Company, St. Louis.
Wehner, J. F., and R. H. Wilhelm. 1956. Boundary Conditions of Flow Reactors.
Chem. Eng. Sci., 6:89-93.
Wentzler, T. H., and F. F. Apian. 1972. Kinetics of Limestone Dissolution
by Acidic Wastewater. Environmental Control AIME, NY.
Wentzler, T. H. 1971. A Study of the Interaction of Limestone in Acid Solution.
M.S. Thesis, Penn. State University, University Park, PA.
Westall, J. C., J. L. Zachary, and F. M. Morel. 1976. MINEQL-A Computer
Program for the Calculation of Chemical Equilibrium Composition of Aqueous
Systems. TN-18 Parson Laboratory, MIT, Cambridge, MA.
Weyl, P. K. 1958. The Solution Kinetics of Calcite. J. Geol. 66:163-176.
Wilhelm, R. H. 1962. A Priori Design of Chemical Reactors. Pure Applied
Chemistry, 5:403-409.
183
-------
Appendix A
Chemical Equilibrium Model
Used in Contactor Design Calculations
184
-------
INTRODUCTION
Determination of the limestone contactor effluent chemistry requires know-
ledge of the chemical equilibrium conditions in the solution which is immediately
adjacent to the limestone surface (see Figure 2 and Eq. 25).
The equilibrium water chemistry at the limestone surface was determined
for two cases:
1. When a complete chemical analysis of the raw water is available,
and,
2. When only a partial knowledge of the chemical composition of the
raw water is available.
Three operational conditions were also considered:
a. Closed system: The contactor and the contactor effluent are closed
to the atmosphere and therefore there is no exchange of carbon dioxide
between the solution and the atmosphere.
b. Open system: The water in the contactor is continuously in equilibrium
with atmospheric carbon dioxide.
c. Closed/Open system: The water in the contactor is closed to the
atmosphere but the effluent is open to the atmosphere.
The three operational conditions are illustrated schematically in Figure
A.I.
In the description of the computational procedure which follows the know-
ledge of the raw water chemistry and the operational conditions which pertain
to a given procedure are designated by a number and a letter, e.g., "la" indi-
cates that a complete chemical analysis of the raw water is available and the
system is closed to the atmosphere.
The solute species Ca"1"4. H2C03, C03=, H* and OH" in the solution which
is immediately adjacent to the limestone particle surface are unknown. To
define the solution composition and to determine the unknown species the following
equations were used:
- Charge balance equation:
Z,- C,- =0 (A.I)
i=l X X
where Z^ and C^ are the charge and molar concentration of specie (i).
185
-------
Influent
Limestone
-*-Treated Water
(a) CLOSED SYSTEM
Influent C02
31
Limestone
-^.Treated Water
(b) OPEN SYSTEM
Influent
C02
Limestone
•*-Treoted Water
(C) OPEN/CLOSED SYSTEM
Figure A.I. Operational conditions used in the
chemical equilibrium modelling.
186
-------
- Mass action expressions for the deprotonation of carbonic acid:
{H+HHCCT }
{H2C03> (A'2)
{H+}{C032~}
(A-3)
where {i} is the activity of specie (i).
- Solubility product expression for CaC03(s):
Ksp = {Ca2+}{C032-} (A. 4)
- Ion product expression for water:
Kw = {H+MOH"} (A. 5)
- Henry's law expression for carbon dioxide dissolved in water:
H2C03*
KH = ~~~
where pC02 is the partial pressure of carbon dioxide.
-'Mass balance equations:
DIG = [H2C03*] + [HC03~] + [C032~] (A. 7)
where
[H2C03*] = DOC x a0, (A. 8.1)
[H2C03~] = DIG x als (A. 8. 2)
[C032"] = DIG x o2, (A. 8. 3)
DIG is the dissolved inorganic carbon concentration and ao, a^ and a2 are the
ionization fractions for the carbonate system (Stumm and Morgan, 1981):
Kal KalKa2 -1
Kal [H+]
(A.9.2)
(A.9.3)
KalKa2 Ka2
For a dilute acidic water flowing into the contactor, equation (A.I) becomes:
187
-------
2[Ca2+] + Cc + [H+] = [HC03-] + 2 [C03~2] + Ca + [OH~] (A.10)
where Cc is the total concentration of non-calcium and hydrogen ion cations,
in'equivalents per liter, Ca is the total concentration of non-inorganic carbon
and hydroxyl ion anions in equivalents per liter, and the brackets denote molar
concentration.
As water flows through the contactor CaC03 is dissolved from the limestone
and the calcium and DIG concentrations increase, i.e.,
CbL = Cbo + S (A.11.1)
and
DIG = DIC0 + S (A.11.2)
where CbL and S are the molar concentrations of calcium ion and calcium carbonate
dissolved from the limestone at an axial location, L, in the contactor bed,
Cbo is the calcium concentration in the influent and DICO is the influent DIG
concentration.
With the substitution of Eqs. (A.8.2), (A.8.3) and (A.11) in the solubility
product equation, (A.4), and charge balance equation, (A.10), the following
expressions are obtained:
2[Cbo + S] + Cc [H+] = (DIC0 + S) (ax + 2a2) + Ca+ [OH"] (A.13)
{Cbo + S} {(DIC0 + S) a2} = Ksp (A.14)
or
Cv- + DICn Cho + DICn 2 Ksp ^
) - -=* r)] (A.15)
where y2 *-s tne activity coefficient for divalent ions, in this case the Ca2+
and the C032~ ions.
Computational Procedure
The equilibrium calculations assume that the influent water is dilute,
i.e., the ionic strength, I, is less than 0.01 and negligible complexing of
ions occurs.
The equilibrium calcium concentration, Ceq, was determined for each set
of raw water chemical conditions and temperature using an algorithm in which
the pH is systematically varied to find the point at which both the charge
188
-------
balance, equation (A.13), and the solubility product relationship, equation
(A.14) are satisfied.
The search procedure was conducted using three computational loops:
-First loop: the pH interval 6 to 10.5 was searched in steps of 0.25
pH units and the point (pH^) at which equations (A.13) and (A.14) were
satisfied was found.
-Second loop: the pH interval (pH^ + 0.30) was searched in steps of 0.05
pH units and the point (pH2) at which equations (A.13) and (A.14) were
satisfied was found.
-Third loop: the pH interval (pH2 + 0.06) was searched in steps of 0.01
pH units and the point (pH3) at which equations (A.13) and (A.14) were
satisfied was found. At this point:
PH3 = pHeq
ceq ' cbo + s U.16)
DICeq = DIC0 + S
In the above calculations the following were considered:
-Equations derived by Plummer and Bussenburg (1982) were used to calculate
the equilibrium constants Ka^, Ka2, and KJJ at infinite dilution as a function
of temperature. Plummer's equations are given in Table (A.I)..
-The effective CaCC>3 solubility product (Ksp 20°) of !-9 x 10~9 at 20°c
(Section 5) was corrected for temperature using the following relationship
(Snoeyink and Jenkins, 1980):
Ksp = Ksp 20° {exp [- | (i - 2^3)]} (A.17)
where Ksp is the CaC03 solubility product at temperature, T. Values for the
enthalpy, H, and the Boltzmann constant, R, were taken from Snoeyink and Jenkins,
•| = 1484.5 (degree Kelvin)
K
Values of the equilibrium constants, Ka^, Ka2 and KJJ and the effective
CaC03 solubility product, Ksp, (Equation A.17) for a range of temperature (1
to 25°C) are presented in Table A.2.
189
-------
Table A-l Equations Used to Calculate the Equilibrium Constants,
^al» Ka2 and KJJ as a Function of Temperature
(Plummer and Bussenburg, 1982)
log Kal = -356.3094 - (0.0609196 x T) + (2.834.37/T)
+ (126.8339 x log T) - (168491/T2)
log Ka2 = -107.8871 - (0.03252849 x T) + (515179/T)
+ (38.92561 x log T) - (563713.9/T2)
log KH = 108.3865 + (0.0198507 x T) + (669365/T2)
- (6919.53/T) - (40.45154 x log T)
where, T, is in degrees Kelvin
190
-------
-At each pH in the search procedure the ionic strength, I, and the activity
coefficients ,Yi> were calculated using
I = l/2( E Zi2 Ct) (A. 18)
and
Log Yi = -A Zt2 1% for I < 10~2-3 (A. 19)
- AZi2 fr_
Log Yi = f^ J3£ for I < 10'1 (A. 20)
where A = 0.509.
The calculations were made using a computer program written in APL. Out-
lines of the program calculations are given below for conditions la, Ib, Ic,
2a, 2b and 2c.
l.a Closed-to-the-Atmosphere System (Complete Influent Water Chemistry
is Known).
After the water chemical composition and temperature are input, the program
is used to compute the temperature corrected values of the equilibrium constants
Kai and Ka2 (see Table A. 2) and the effective CaC03 solubility product Ksp
(Equation A. 17) .
-lonization fractions for the carbonate and bicarbonate ions are then
estimated for the first pH value in the interval being searched and the carbonate
and bicarbonate concentrations are calculated using equations (A. 9. 2), (A. 9. 3),
(A. 8. 2) and (A. 8. 3).
-The ionic strength, I, is estimated using Eq. A. 18 and accordingly the
activity coefficients for monovalent, YI> and divalent, Y2' i°ns are calculated
using Eq. A. 19 or A. 20. With the known activity coefficients, the equilibrium
constants, Ka^ and Ka£ are corrected for ionic strength as follows:
Kai
K'al = -4 (A. 21)
al Z
, =
-------
equation, Eq. A. 13. A quantity, DEL, defined as the difference between the
left and the right side of the charge balance equation is then calculated:
DEL = {2[Cbo + S] + Cc + [Hi +]} - {((DIC0 + S)(a]+ 2a2)) + Ca + [0%]} (A. 23)
-The program then repeats the above calculations using the next pH in
the interval. The pH in the search interval at which DEL is a minimum is the
point where the solubility product and the charge balance equations (Eqs. A. 13
and A. 14) are essentially satisified. In the first loop the pH at the point
where DEL is a minimum is pH^.
After pEi is obtained the second loop begins. The calculations in the
second and third loop are the same as those in the first loop except, as noted,
smaller pH intervals are searched and smaller pH increments are used in the
search across each interval.
To use the contactor design equations the calcium concentration in the
contactor effluent C^LJ must be determined for the case when the effluent is
not in equilibrium with the limestone, i.e., pH < pHeq and C^L^ Geq. Usually
a target effluent pH is known and one must then calculate the corresponding
effluent calcium concentration.
The magnitude of C^L f°r a given effluent pH is determined using the charge
balance equation, Eq. A. 13. The target effluent pH is used to determine a^,
ct2> [IT1"] and [OH~] and these quantities are used with C^Q, DICO, Cc and Ca
to solve Eq. A. 13 for the quantity, S. The desired effluent calcium concentra-
tion is equal to C^o + S. Note that this value of S is less. than the equili-
brium value from Eq. A. 15.
l.b - Open-to-the-Atmosphere System
For an open to the atmosphere system the computational procedure was the
same as that used for a closed-to-the-atmosphere system except that the value
of the dissolved inorganic carbon concentration in equations A. 13 and A. 14
was estimated at each pH using equations A. 8.1 and A. 6. Combining equations
A. 8 and A. 9 yields,
KH
DIG =- (A'24>
192
-------
Table A-2 Values of Kai, Ka2, KJJ and Ksp(CaC03) for a Range of Temperatures.
The Equations of Plummer and Bussenburg (1982) were used to
Calculate these Quantities.
T°C
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
.17
18
19
20
21
22
23
24
25
log Kal
-6.56
-6.55
-6.54
-6.53
-6.51
-6.50
-6.49
-6.48
-6.47
-6.46
-6.45
-6.44
-6.43
-6.42
-6.41
-6.41
-6.40
-6.39
-6.38
-6.38
-6.37
-6.36
-6.36
-6.35
-6.35
log Ka2
-10.61
-10.59
-10.58
-10.56
-10.55
-10.54
-10.52
-10.51
-10.50
-10.48
-10.47
-10.46
-10.45
-10.44
-10.42
-10.41
-10.40
-10.39
-10.38
-10.37
-10.36
-10.35
-10.34
-10.33
-10.32
log KH
-1.12
-1.14
-1.15
-1.17
-1.19
-1.20
-1.22
-1.23
-1.25
-1.26
-1.28
-1.29
-1.31
-1.32
-1.34
-1.35
-1.36
-1.38
-1.39
-1.40
-1.41
-1.43
-1.44
-1.45
-1.46
log Ksp
-8.56
-8.57
-8.58
-8.59
-8.60
-8.61
-8.61
-8.62
-8.63
-8.64
-8.65
-8.65
-8.66
-8.67
-8.68
-8.69
-8.69
-8.70
-8.71
-8.72
-8.72
-8.73
-8.74
-8.75
-8.75
193
-------
An equation derived by Plummer and Bussenburg (1982) for determining Henry's
Law constant for carbon dioxide (see Table A.I) was used with a partial pressure
of atmospheric C02 of 10"3-5.
l.c - Closed/Open System
The closed/open system calculation involved the pH interval search procedure
and Eq. A.23 with the following substitutions;
S = 0
cbo " cbL
and from Eq. A.24,
KH pC02
DIC0 + S = V
o
C^L is the calcium concentration in the contactor effluent. Eq. A. 15 is omitted
from the pH interval search calculations because the effluent is not in contact
with solid CaCC>3.
2 - A procedure for the case when there is limited information on the chemistry
of the raw water
The availability of a well equipped laboratory and trained technical per-
sonnel in a small water supply system may be limited. In order to proceed
with the determination of the chemical equilibrium conditions at the limestone
surface, knowledge of the total anion, Ca, and cation, Cc concentrations and
their effects on the total ionic strength is necessary to estimate the activity
coefficients for individual ions. A procedure was developed for use when only
the measured specific conductance, K,^ initial calcium concentration, C^Q,
initial pH, pHo, and alkalinity are known.
An equation relating the portion of the total ionic strength contributed
by Cc and Ca ions, IAB» to t*18 corresponding specific conductivity, K^BJ was
derived using data from the analysis of water from 34 lakes in the Adirondack
Region of New York State. The equation is given by:
IAB = constant x KAJJ- (A.25)
The complete chemical analyses for these lakes were obtained from the
results of a survey conducted by the U.S. EPA (Kanciruck et al. 1985). The
data for the 34 lakes were chosen at random from a list of over 100 lakes.
194
-------
The MINEQL chemical equilibrium program (Westall et al. 1976) was used
to calculate the total ionic strength, I, for each of the 34 sets of data.
The' total component concentrations and temperature for each lake were entered
in the MINEQL program. The values of the ionic strength obtained from MINEQL
for the 34 lakes ranged from 2 x 10"4 to 9 x 10"4 M.
The contributions of Ca"1""1", H"1", OH", HC03", and C03= to the total ionic
strength I, was estimated using:
I' = 1/2 (4[Ca++] + [H+] + [HC03-] + 4[C032-] + [OH']) . (A.26)
The ionic strength attributable to Cc and Ca was determined by calculating
the difference between the total ionic strength, I, and I', i.e.,
IAB = i - i'- (A.27)
The specific conductance attributable to Ca and Cc ions, K^g, was estimated
for each lake by computing the difference between the measured specific conduc-
tance, KJJJ, and the sum of the specific conductances attributable to Ca"1"1", IT1",
HC03-, C032-, and OH", i.e.,
KAB = Km - Kl (A-28)
where
K! = [ca4^] Aea++ + [H+] AH+ +[Hco3~] xHC03_ +
[C03-2] ACo3 + [OH] AOH- (A.29)
and, A, is the specific ionic conductance in water at 25°C, in micromhos/cm.
The values of the specific conductance used in the analysis were taken
from Robinson et al. (1959) and are listed in Table A.3. The values of I,
KIIP IAB> KAB for the 34 lakes are listed in Table A.4.
TABLE A.3 Individual ion specific conductance
Ion Specific Conductance, A
H+ 349.8
HC03" 44.5
C03= 69.3
Ca^ 59.5
OH" 198.3
To determine the value of the constant in Equation A.25 a nonlinear least
squares procedure which produces least squares or weighted least squares esti-
195
-------
mates of the parameters of the model was used (SAS 1982). This procedure uses
the modified Gauss-Newton iterative method. The analysis gave the following
equation:
IAB = 1.31 x ID'5 KAB (r2 = 0.55) (A.30)
The low value of R2 might have resulted from errors in the pH and/or DIG measure-
ments which were used in estimating IAB (see equations A. 26 and A. 27).
The computational procedure for conditions 2a, 2b and 2c was the same
search algorithm as was used for conditions la, Ib and Ic. The only difference
between the two procedures is in the determination of the ionic strength of
the solution at each pH. Determination of the ionic strength of the solution
at each pH when limited information is available on the ionic constituents
of the raw water can be summarized as follows:
-According to the charge balance equation (A. 10):
CAB = ca * cc = (2[CboJ + [H+] - (DlCo(ai + 2
-------
Table A.4 Ionic Strength and Specific Conductivity for
34 Adirondack Region Lakes
Ref. #
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
• 17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
1 x 104
[M]
3.81
1.97
4.23
2.21
2.16
2.63
2.98
3.10
2.3
2.94
3.08
3.41
2.80
4.14
6.47
7.18
7.09
6.76
7.61
8.36
6.00
6.96
8.71
5.84
5.48
4.77
2.47
2.99
5.36
4.44
4.51
4.83
6.11
7.34
KAB
micromhos/cm
23.70
22.70
26.00
28.00
33.70
16.60
17.60
22.40
24.50
27.10
24.10
21.00
19.90
22.60
33.40
40.40
50.50
41.10
44.00
54.80
33.40
44.90
81.30
34.80
18.00
24.10
20.90
27.00
29.00
19.06
24.20
32.80
34.20
24.7
IAB * io4
[M]
3.14
1.41
2.63
1.81
1.48
1.79
1.89
1.94
1.78
1.99
2.08
2.42
1.91
3.32
3.87
4.28
5.30
3.33
4.59
4.82
3.32
3.83
4.87
3.61
4.78
3.15
1.67
3.08
3.13
3.27
2.90
3.06
3.18
6.34
KAB
micromhos/cm
16.36
13.85
19.20
17.35
17.03
13.44
13.16
18.17
15.26
17.82
18.89
16.90
15.86
16.63
20.71
26.56
43.68
23.64
39.87
38.97
20.95
30.09
63.26
25.00
13.08
16.60
16.40
18.15
19.09
14.65
17.70
24.97
21.18
14.67
197
-------
For Condition 2.c,
Equation A.33 was used to calculate the chemistry of the contactor effluent
(closed system) and then equation A.34 was used for the condition when the
contactor effluent is opened to the atmosphere.
For all conditions, once I^B and I' are known, the total ionic strength
of the solution can be estimated using equation A.27,
i = IAB + i1 (A.35)
and the computational procedure for the three conditions (2.a, 2.b and 2.c)
proceeds in the same manner as was described for the case when the detailed
chemistry (Ca and Cc) is known.
198
-------
Appendix B
Dissolution Rate Data
from Column Experiments
199
-------
O
o
Superficial
Run Velocity
Number Column (cm/min)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
5.5
11.0
16.5
22.0
27.5
5.5
11.0
16.5
22.0
27.5
27.5
41.2
55.0
72.0
5.5
2
Influent Water Characteristics
PH
4.19
4.19
4.19
4.08
4.08
3.92
4.00
4.00
3.92
4.00
4.34
4.50
4.50
4.50
4.50
Calcium
(mg Ca/L)
0
0
0
0
0
3.0
1.7
1.7
3.0
4.3
0.3
0
0
0.3
0
Dissolved In-
organic Carbon
(mg C/L)
0.1
0.1
0.1
0.2
0.2
0.3
0.1
0.1
0.3
0.2
0.1
0.1
0.1
0.1
0.2
Water Tem-
perature
°C
16
16
16
16
16
16
16
16
16
16
10
10
.10
10
10
Overall Dissolution
Rate Constant,
K x 103 (cm/min)
o
35
54
61
37
54
37
22
44
51
62
46
54
54
69
18
See Figure for limestone particle diameter and sphericity and bed porosity
Background electrolyte concentration was 20 mg NaCl/L
-------
Run
Humber
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
Co lumn
A
A
A
A
A
A
A
A
A
A
A
A
A
A
A
Superficial
Velocity
(cro/min)
5.5
5.5
16.5
5.5
16.5
27.5
27.5
16.5
5.5
54.8
38.4
21.9
5.5
55.0
38.4
PH
3.89
3.90
3.90
3.91
3.91
3.91
3.89
3.89
3.89
5.45
5.45
5.45
5.45
4.00
4.00
Influent
Calcium^
(mg Ca/L)
3.2
5.2
5.2
0.1
0.1
0.1
0.2
0.2
0.2
0
0
0
0
0.2
0. 2
Water Characterisi
Dissolved In-
organic Carbon
-------
10
o
Run
Number
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
Column
A
A
D
D
D
D
D
D
D
D
D
D
D
C
C
Superficial
Velocity
(cm/mln)
22.0
5.5
8.8
6.1
3.5
0.9
8.8
6.1
3.5
0.9
8.8
6.1
3.5
54.8
38.4
PH
4.00
4.00
5.99
5.99
5.48
5.48
3.86
3.86
3.88
3.98
3.41
3.41
3.56
6.12
5.12
Influent
Calcium .
(mg Ca/L)
0.2
0.2
0.4
0
0
0
0.4
0.4
0.7
0.3
0.0
0.0
0.3
2.4
2.4
Water Cha racterisf
Dissolved In-
organic Carbon
(mg C/L)
0
0
0
0
0
0
0
0
0.1
0.2
0.2
0.2
0.3
0.2
0.2
tics
Water Tem-
perature
°C
10
10
9
9
9
9
9
9
9
9
9
9
9
10
10
Overall Dissolution
Rate Constant,
K x 103(cin/min)
o
38
32
35
15
11
7
19
17
11
6
21
18
20
105
52
See Figure for limestone particle diameter and sphericity and bed porosity
Background electrolyte concentration was 20 mg NaCl/L
-------
to
o
u>
Superficial
Run Velocity
lumber Column (cm/min)
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
C
C
C
C
C
C
C
C
C
C
B
B
B
B
B
21.9
5.5
54.8
38.4
21.9
5.5
54.8
38.4
21.9
5.5
54.8
38.4
22.0
5.5
55.0
2
Influent Water Characteristics
PH
6.12
6.12
4.02
4.02
4.02
4.38
3.53
3.53
3.53
3.53
5.45
5.45
5.45
5.45
4.00
Calcium
(mg Ca/L)
2.4
2.4
0.5
0.4
0
0
0.2
0.2
0.2
0.2
0
0
0
0
0.2
Dissolved In-
organic Carbon
(mg C/L)
0.2
0.2
0.2
0.2
0.2
0.2
0.2
0.2
0.2
0.2
0.3
0.3
0.3
0.3
0
"Water Tem-
perature
°C
10
10
10
10
10
10
10
10
10
10
10
10
10
10
10
Overall Dissolution
Rate Constant,
K x 103{cm/mln)
42
26
116
78
54
23
63
40
27
12
126
52
25
19
150
See Figure for limestone particle diameter and sphericity and bed porosity
2
-------
10
O
Superficial
Run , Velocity
Number Column (cm/min)
61
62
63
64
B
B
B
B
38.4
21.9
5.5
5.5
2
Influent Water Characteristics
PH
4.00
4.00
4.00
3.51
Calcium
(mg Ca/L)
0.2
0.2
0.2
0.3
Dissolved In-
organic Carbon
(mg C/L)
0
0
0
0
X-
Water Tem-
perature
°C
10
10
10
10
Overall Dissolution
Rate Constant,
K x 103(cm/min)
o
70
45
45
35
See Figure for limestone particle diameter and sphericity and bed porosity
Background electrolyte concentration was 20 mg NaCl/L.
-------
APPENDIX C
Estimates of Limestone Contactor Cost
205
-------
The Culligan contactor unit (see Figure 8) with 100 Ib (45 kg) of Cullneu
medium (2ft3 (57 L) of medium) costs $672 installed (March 1986). A 50 Ib
(23 kg) bag of Cullneu costs $50.40. Culligan recommends that the unit be
used with a flow rate of less than 5 gpm (0.3 L/s) and that the medium be back-
washed periodically. The piping supplied with the unit enables one to backwash
using the influent flow. Culligan also suggests that the Cullneu medium be
replenished by the addition of small amounts ("handfulls") at frequent intervals
(monthly).
The box contactor, depicted in Figure 7, was constructed by graduate stu-
dents at Syracuse University. The materials used in its construction, (plywood,
acrylic plastic, fiberglass, etc.) were purchased for approximately $800.
About 80 man-hours of labor were required. The unit contained about 800 Ib
(363 kg) of limestone. The empty box weighed approximately 400 Ibs (182 kg)
and therefore installation of the box contactor in the mountain-side spring
was a very time-consuming laborious process.
The least expensive approach involves the purchase of a fiberglass pressure
vessel and filling it with crushed limestone. This is what was done in the
case of the wound-fiberglass column (Column 1, Figure 8). It is recommended
that the limestone be analyzed to determine amounts of chemical contaminants
and CaC03 purity before it is used. The cost of limestone is negligible
( 'v,$0.01/lb, $0.02/kg) compared to the cost of a container. The cost of fiber-
glass pressure vessels is given in Table C.I. Depending on the size of the
unit the cost ranges from $3 to $7/L ($85 to $198/ft3)capacity.
206
-------
TABLE C.I Cost of Fiberglass Pressure Vessels
Vessel Dimensions Approximate Approximate Cost
Volume Diameter Length Cost (March 1984) Dollars/Liter
14L 15 cm 46 cm $ 92 6.6
28L 20 cm 100 cm $137 4.9
57L 20 cm 132 cm $198 3.5
100L 33 cm 137 cm $296 3.0
142L 36 cm 165 cm $410 2.9
207
------- |